Atmos Chem Phys 19 4541ndash4559 2019httpsdoiorg105194acp-19-4541-2019copy Author(s) 2019 This work is distributed underthe Creative Commons Attribution 40 License
Accounting for the vertical distribution of emissions inatmospheric CO2 simulationsDominik Brunner1 Gerrit Kuhlmann1 Julia Marshall2 Valentin Cleacutement34 Oliver Fuhrer4 Greacutegoire Broquet5Armin Loumlscher6 and Yasjka Meijer6
1Empa Swiss Federal Laboratories for Materials Science and Technology Uumlberlandstrasse 129 Duumlbendorf Switzerland2Max Planck Institute for Biogeochemistry (MPI-BGC) Jena Germany3Center for Climate Systems Modelling (C2SM) ETH Zuumlrich Zuumlrich Switzerland4Federal Institute of Meteorology and Climatology MeteoSwiss Kloten Switzerland5Laboratoire des Sciences du Climat et de lrsquoEnvironnement CEA-CNRS-UVSQ Universiteacute Paris Saclay91191 Gif-sur-Yvette CEDEX France6European Space Agency (ESA) ESTEC Noordwijk the Netherlands
Received 10 September 2018 ndash Discussion started 4 December 2018Revised 1 March 2019 ndash Accepted 18 March 2019 ndash Published 5 April 2019
Abstract Inverse modeling of anthropogenic and biosphericCO2 fluxes from ground-based and satellite observations crit-ically depends on the accuracy of atmospheric transport sim-ulations Previous studies emphasized the impact of errors insimulated winds and vertical mixing in the planetary bound-ary layer whereas the potential importance of releasing emis-sions not only at the surface but distributing them in the verti-cal was largely neglected Accounting for elevated emissionsmay be critical since more than 50 of CO2 in Europe isemitted by large point sources such as power plants and in-dustrial facilities In this study we conduct high-resolutionatmospheric simulations of CO2 with the mesoscale Consor-tium for Small-scale Modeling model extended with a mod-ule for the simulation of greenhouse gases (COSMO-GHG)over a domain covering the city of Berlin and several coal-fired power plants in eastern Germany Poland and Czech Re-public By including separate tracers for anthropogenic CO2emitted only at the surface or according to realistic source-dependent profiles we find that releasing CO2 only at thesurface overestimates near-surface CO2 concentrations in theafternoon on average by 14 in summer and 43 in win-ter over the selected model domain Differences in column-averaged dry air mole XCO2 fractions are smaller between5 in winter and 8 in summer suggesting smaller yet non-negligible sensitivities for inversion modeling studies assim-ilating satellite rather than surface observations The results
suggest that the traditional approach of emitting CO2 onlyat the surface is problematic and that a proper allocation ofemissions in the vertical deserves as much attention as an ac-curate simulation of atmospheric transport
1 Introduction
Reliably predicting future atmospheric concentrations ofCO2 the most important long-lived greenhouse gas re-quires a profound understanding of the global carbon cyclethe contributions from anthropogenic and natural fluxes andtheir sensitivity to climate change and political and soci-etal drivers An important tool for advancing our knowledgeof the carbon cycle is the integration of CO2 observationswith atmospheric transport simulations in an inverse model-ing framework Global inverse modeling systems helped tobetter constrain the terrestrial carbon budget to allocate theglobal land sink to different continents and ecosystems andto assess interannual variability and the sensitivity to climatevariations (eg Bousquet et al 2000 Chevallier et al 2010Peylin et al 2013 van der Laan-Luijkx et al 2017 Roumlden-beck et al 2018) Mesoscale inverse modeling systems as-similating observations from dense regional in situ networksare increasingly being used to study biospheric fluxes at the
Published by Copernicus Publications on behalf of the European Geosciences Union
4542 D Brunner et al Atmospheric CO2 simulations with vertical emissions
regional scale (eg Sarrat et al 2009 Goeckede et al 2010Broquet et al 2011 Meesters et al 2012)
The Paris Climate Agreement adopted in 2015 (UnitedNations Framework Convention on Climate Change 2016)which requires each signing partner nation to accurately re-port its greenhouse gas (GHG) emissions and to reduce emis-sions in the future following its Nationally Determined Con-tribution has boosted the interest of the atmospheric sciencecommunity in quantifying not only natural fluxes but alsoanthropogenic emissions of CO2 ldquoTop-downrdquo inverse esti-mation of anthropogenic emissions from atmospheric obser-vations has in fact been proposed as an independent methodto complement the traditional ldquobottom-uprdquo collection of na-tional emission inventories (Nisbet and Weiss 2010) Sev-eral initiatives at the national and international level such asthe World Meteorological Organization (WMO)rsquos IntegratedGlobal Greenhouse Gas Information System (IG3IS) havebeen launched to advance and harmonize inverse methodswith the long-term vision to establish these methods as apolicy-relevant verification and support tool A summary ofthe state of the science in inverse modeling for verificationof greenhouse gas inventories has recently been presented byBergamaschi et al (2018a)
Because of the great challenge of accurately estimat-ing CO2 fluxes and distinguishing between the biosphericand anthropogenic contributions the scientific community iscalling for a globally integrated carbon observation and anal-ysis system This system should build on a substantially ex-panded ground-based and satellite observation capacity andshould integrate observations and bottom-up data with atmo-spheric transport modeling in an inverse framework (Ciaiset al 2014) The individual components and necessary stepstowards an operational European system for quantifying an-thropogenic CO2 emissions were outlined in two recent re-ports to the European Commission (Ciais et al 2015 Pintyet al 2017)
A central component of this system is a constellation ofCO2 satellites with imaging capability similar to the Car-bonSat concept proposed by Bovensmann et al (2010) Thesatellites are currently planned to be implemented in theframework of the European Earth observation Copernicusprogram and are commonly referred to as Sentinel-CO2These satellites should be able to quantify the CO2 emissionsfrom large sources such as power plants and cities during sin-gle overpasses Corresponding observing system simulationexperiments (OSSEs) were presented by Pillai et al (2016)and Broquet et al (2018)
In order to establish the requirements for the constella-tion of Sentinel-CO2 satellites the European Space Agency(ESA) has launched several scientific support studies includ-ing the use of satellite measurements of auxiliary reactivetrace gases for fossil fuel carbon dioxide emission estima-tion (SMARTCARB) a project that focused on the poten-tial of complementary satellite measurements of NO2 or COto improve the quantification of CO2 emissions from cities
and power plants The present study makes use of the high-resolution atmospheric transport simulations conducted inSMARTCARB to address a specific topic beyond the mainfocus of the project The core results of the SMARTCARBstudy will be presented elsewhere
Estimating CO2 fluxes by inverse modeling requires ac-curate simulation of observed atmospheric CO2 concentra-tions Systematic model biases are particularly critical asthey tend to directly translate into biased flux estimates Sev-eral studies investigated the impact of uncertainties in atmo-spheric transport on simulated CO2 concentrations and howthey could be accounted for in an inverse modeling frame-work (eg Lin and Gerbig 2005 Chan et al 2008 Lau-vaux et al 2009) A particular focus was placed on potentialbiases introduced by errors in vertical transport in the plane-tary boundary layer (PBL) (Gerbig et al 2008 Kretschmeret al 2012 2014) Uncertainties associated with bottom-upCO2 emissions were related to uncertainties in the horizon-tal gridding (Hogue et al 2016) or the representation of thetemporal variability (Liu et al 2017)
Uncertainties associated with the vertical placement ofemissions conversely have been largely ignored so far Infact in the vast majority of CO2 atmospheric transport andinverse modeling studies CO2 emissions were released ex-clusively at the surface (eg Sarrat et al 2009 Lauvauxet al 2009 Broquet et al 2011 Ganshin et al 2012Meesters et al 2012 Pillai et al 2016 Lauvaux et al 2016Liu et al 2017 Graven et al 2018 Fischer et al 2018)In Lagrangian models such as the Stochastic Time-InvertedLagrangian Transport model (STILT) (Lin et al 2003) orFLEXible PARTicle dispersion model (FLEXPART) (Stohlet al 2005) which are often used in backward adjoint modefor inverse modeling particles are typically sampled over afixed vertical depth above the surface or relative to the heightof planetary boundary layer to derive source sensitivitiesSimilar to the release of emissions at the surface in Eulerianmodels this ignores the potentially different sensitivities toemissions from elevated sources
These approaches in Eulerian and Lagrangian models arequestionable given the fact that a large proportion of CO2 isreleased from point sources such as power plants well abovethe surface Smoke stacks are in fact designed to minimizethe impact on concentrations at the ground This is partic-ularly important during situations with stable inversions inwinter or on clear nights However even in well-mixed con-ditions an elevated release may lead to a faster dilution anddifferent propagation of the signal due to wind speeds anddirections changing with altitude
In the air quality modeling community the importance ofvertically distributing emissions has been recognized muchearlier (eg Houyoux et al 2002) and is now well estab-lished especially for species such as SO2 that are primar-ily emitted from power plants and industrial sources (Bieseret al 2011 Mailler et al 2013 Karamchandani et al 2014Guevara et al 2014) Accounting for plume rise has also
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4543
been demonstrated to be critical for biomass burning emis-sions (Achtemeier et al 2011) The main goal of this studyis to demonstrate that this is also critical for the simulationof atmospheric CO2 concentrations
We employ very-high-resolution (11 kmtimes 11 km) simu-lations for the year 2015 conducted over a 750 kmtimes 650 kmwide domain centered in the city of Berlin Germany Thesimulations included separate tracers representing CO2 emit-ted only at the surface and CO2 emitted according to source-dependent vertical profiles As the domain covered severallarge coal-fired power plants we also investigate the im-pact of dynamically accounting for plume rise versus apply-ing static vertical emission profiles We focus on domain-averaged statistics rather than on individual plumes andon near-surface concentrations in the afternoon and totalcolumns at satellite overpass time as typically used in inversemodeling Conditions with a well-mixed and fully developedboundary layer during daytime are expected to minimize themismatch between model and observations
2 Data and methods
21 COSMO-GHG model
The Consortium for Small-scale Modeling (COSMO) is alimited-area non-hydrostatic numerical weather prediction(NWP) model developed by the German weather service to-gether with a consortium of seven European weather services(Baldauf et al 2011) In addition to operational weather pre-diction the model is applied widely for climate and air pollu-tion research in various modified and extended versions (egRockel et al 2008 Davin et al 2011 Vogel et al 2009Zubler et al 2011) making COSMO a versatile communitymodel
In the project CarboCount-CH (Oney et al 2015)COSMO was extended with a module for the simulation ofgreenhouse gases hereafter referred to as COSMO-GHGThe extension was built on a newly developed tracer module(Roches and Fuhrer 2012) which replaced the previous non-uniform treatment of meteorological tracers in the model Asimilar independently developed COSMO-based CO2 modelsystem was presented by Uebel and Bott (2018)
COSMO-GHG was first applied by Liu et al (2017) tostudy the spatiotemporal patterns of fossil fuel CO2 in Eu-rope They concluded that fossil fuel CO2 accounts for morethan half of the total (anthropogenic plus biospheric) tempo-ral variations in atmospheric CO2 over large parts of EuropeFurthermore they evaluated the model against observationsdemonstrating that the simulated fossil fuel variations favor-ably agreed with observations of fossil fuel CO2 derived fromconcurrent measurements of CO and 14CO2
COSMO is the first operational NWP model worldwidethat is running on hardware accelerated using graphical pro-cessing units (GPUs) (Fuhrer et al 2014) This highly effi-
cient code is operationally used by the Swiss weather serviceMeteoSwiss and has been applied for decadal convection-resolving climate simulations (Leutwyler et al 2017) In theframework of SMARTCARB the modules of the GHG ex-tension were ported to the GPU version following the con-cept of OpenACC compiler directives which is a high-levelapproach to offload compute-intensive parts to a GPU ac-celerator (Lapillonne and Fuhrer 2014) Benchmark testsshowed that the GPU version achieved a speedup by a factorof 6 which allowed for an increase in spatial and time res-olution and which greatly reduced the computational (andenergy) costs of the simulations conducted in this study
22 Model domain and setup
The model simulations were conducted in the framework ofthe project SMARTCARB which aimed at studying the po-tential benefit of adding an NO2 or CO channel to the instru-ment package of a future CO2 satellite mission with respectto quantifying CO2 emissions from strong localized sourcessuch as cities and power plants For this purpose a modeldomain was selected covering the city of Berlin as well asseveral large power plants in Germany Poland and CzechRepublic The domain extended approximately 750 km in theeastndashwest and 650 km in the southndashnorth directions to coverat least a complete 250 km wide satellite swath on either sideof the city Berlin was selected not only because it is oneof the largest cities in Europe but also because it is ratherisolated and because it has already been investigated in pre-vious CO2 modeling and observation studies (Pillai et al2016 Hase et al 2015) Simulations were conducted for thecomplete year 2015 at 11 kmtimes 11 km horizontal resolutionand 60 vertical levels with a top at 24 km The lowest modellayer had a thickness of 20 m The vertical allocation of theemissions relative to this vertical resolution of the model willbe described in Sect 24 The high horizontal resolution wasselected to comply with the small pixel size of 2 kmtimes 2 kmof the planned CO2 imaging satellite keeping in mind thatthe effective resolution of Eulerian transport models is muchlower than the spacing of the model grid (Kent et al 2014)
The simulations included a total of 50 different passivelytransported tracers of CO2 CO and NOx The tracers repre-sented different sources release times or release altitudes andincluded background tracers constrained at the lateral bound-aries by global-scale models Two additional tracers for bio-spheric fluxes (respiration and photosynthesis) were includedfor CO2 and four additional tracers with varying e-foldinglifetimes of 2 4 12 and 24 h for NOx In this study how-ever we focus on only a few of these tracers notably on thefollowing four anthropogenic CO2 emission tracers
ndash CO2_VERT all anthropogenic emissions in the modeldomain released according to source-specific verticalprofiles
4544 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ndash CO2_SURF same as CO2_VERT but all emissions re-leased at the surface ie into the lowest model layer
ndash CO2_PP-PR emissions from the six largest powerplants with explicit plume rise calculations
ndash CO2_PP-EMEP emissions from the six largest powerplants released according to a fixed vertical emissionprofile
The simulations were nested into the operational EuropeanCOSMO-7 analyses of MeteoSwiss which provided the lat-eral boundary conditions for meteorological variables (tem-perature pressure wind humidity clouds) at 7 km horizon-tal and hourly temporal resolution Because of the relativelysmall domain these boundary conditions provided a strongconstraint for the meteorology within the domain For CO2lateral boundary conditions were obtained from a global free-running high-resolution CO2 simulation (T1279 137 levelssim 15 km horizontal resolution) provided by the EuropeanCentre for Medium-Range Weather Forecasts (ECMWF)through the European Earth observation Copernicus program(Agustiacute-Panareda et al 2014) The simulation was indirectlyconstrained by in situ data by using a global climatology ofoptimized biospheric fluxes computed with the CO2 assimi-lation system of Chevallier et al (2010)
23 Emissions and biospheric fluxes
Anthropogenic emissions were obtained by merging theNetherlands Organisation for Applied Scientific ResearchEuropean Monitoring Atmospheric Composition and Cli-mate version 3 (TNOMACC-3) inventory with a detailed in-ventory available for the city of Berlin Dedicated municipalinventories can accurately account for the specific conditionsand activities in a city and may therefore significantly devi-ate from inventories obtained by simple downscaling fromlarger-scale inventories (Timmermans et al 2013 Gatelyand Hutyra 2017)
Version 3 of the TNOMACC inventory was used whichhas an improved representation of point sources as comparedto version 2 described in Kuenen et al (2014) It has a nomi-nal resolution of 116times 18 (approximately 7 kmtimes 7 km)but additionally reports the emissions from strong pointsources at their exact location Emissions are provided sepa-rately for different source categories following the Standard-ized Nomenclature for Air Pollutants (SNAP) classification(European Environment Agency 2002) The emissions of theyear 2011 were taken which was the most recent year avail-able from the inventory
The city of Berlin has developed a very detailed inven-tory for about 30 air pollutants and greenhouse gases forseven major source categories The inventory was providedin a Geographic Information System (GIS) format as a col-lection of shape files representing individual area point andline sources The latest available year was 2012 In our simu-lations separate tracers were included for three broad source
groups traffic industry and heating The attribution of the 10SNAP categories in TNOMACC-3 and the seven source cat-egories in the Berlin inventory to these groups are describedin detail in the final report of the SMARTCARB project(Kuhlmann et al 2019)
Both inventories were projected onto the COSMO modelgrid (rotated latitudendashlongitude grid with the North Poleat 43 N and 10W 001times 001 resolution) using mass-conservative methods Point sources were placed into theproper COSMO grid cell As a last step the two inventorieswere merged using a mask for the city of Berlin The mergedemission e per grid cell was estimated as e = (1minusf )eTNO+
eBerlin where f is the fraction of the grid cell area coveredby Berlin
Figure 1 presents a map of the CO2 inventory of Berlinin the original format and after projection onto the COSMOgrid The rasterized map reveals 22 strong CO2 point sourceswhich together account for 41 of the total emissions in thecity The two horizontal stripes correspond to the paths ofairplanes taking off and landing at the two main airports
In order to calculate hourly emissions as input for themodel simulations temporal scaling factors were applied de-scribing diurnal day-of-week and seasonal variations Thesame temporal profiles were used as in Liu et al (2017)which are based on factors originally developed for air pol-lution modeling (Builtjes et al 2003) These profiles dependon SNAP category and except for diurnal profiles on coun-try Diurnal profiles were matched to the local time of Ger-many
Biospheric CO2 fluxes due to gross photosynthetic pro-duction and respiration were computed at the resolution ofthe COSMO model using the Vegetation Photosynthesis andRespiration (VPRM) model (Mahadevan et al 2008) Thisdiagnostic model is based on meteorological input (2 m tem-perature and downward shortwave radiation at the surface)along with the Enhanced Vegetation Index (EVI) and LandSurface Water Index (LSWI) calculated from MODIS satel-lite reflectances These indices were available as an 8-dayproduct (MOD09A1 V006) at 500 m spatial resolution Veg-etation classes were determined from the 1 km SYNMAPland cover map (Jung et al 2006) Model parameter valuesdescribing the fluxes from different vegetation types weretaken from a previous study (Kountouris et al 2018) inwhich they were optimized using data from 47 Europeaneddy covariance measurement sites for the year 2007 Themodel was run offline for the entire year 2015 driven by thehighest resolution ECMWF meteorological data available
24 Vertical allocation of emissions and plume rise
Emissions were distributed in the vertical using source-specific profiles developed for the European Monitoring andEvaluation Program (EMEP) (see eg Bieser et al 2011)For the purpose of the present study the number of verticallayers was increased from 7 to 10 to enable a finer allocation
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4545
Figure 1 (a) Schematic of CO2 emissions in the Berlin inventory with point line and area sources from different emission categories Pointline and area sources are in different units (in kgsminus1 for point kgsminus1 mminus1 for line and kgsminus1 mminus2 for area sources) each being coloredseparately between minimum and maximum values (b) Total CO2 emissions re-projected and rasterized onto the COSMO model grid
to the model layers of COSMO-GHG Specifically the layers4ndash90 m and 90ndash170 m of the original EMEP profiles were di-vided into three and two sublayers respectively The attribu-tion of emissions to these sublayers was done in a rough wayfor example by placing a larger proportion of emissions fromldquocombustion in manufacturing industryrdquo into the upper partsbut distributing emissions from ldquonon-industrial (residential)combustionrdquo rather evenly over the three layers between 4and 90 m considering that industrial sources are likely toemit at higher altitudes The profile for SNAP 9 (waste) wasmodified in two ways (i) by moving 10 from higher layersto the lowest layer to account for CO2 emissions from land-fills and waste water treatment plants and (ii) by moving thelarge fractions originally placed into the layers 170ndash310 m(40 ) and 310ndash470 m (35 ) into lower layers between 90and 310 m This latter modification was made following thestudy of Pregger and Friedrich (2009) which showed that theemission-weighted height of waste incinerator stacks in Ger-many is only about 100 m and that 90 of the correspondingemissions including plume rise are expected to occur below300 m
The modified EMEP profiles are presented in Table 2 andillustrated in Fig 2a They were only applied to point sourcessuch as power plants and large industrial facilities A differ-ent set of profiles with lower average emission heights wasapplied to area sources as these represent much weaker andmore dispersed sources The corresponding profiles are pre-sented in Table 3 and Fig 2b The motivation for this dis-tinction is best illustrated for SNAP category 2 residentialand other non-industrial heating In the case of point sourcesthese are large heating facilities such as combined heat andpower plants with tall stacks In the case of area sources incontrast these are mostly private heating systems releasingCO2 through chimneys at roof level This is reflected in ourprofiles for SNAP 2 area sources being limited to the layersbetween 4 and 60 m above surface
As noted by Bieser et al (2011) the EMEP profiles arebased on very limited information originally collected forthe city of Zagreb Croatia They therefore proposed a dif-ferent set of profiles based on plume rise calculations for alarge number of point sources across Europe Their study in-dicated that the vertical placement of emissions in the EMEPprofiles is too high for combustion processes (SNAP 1 2 3and 9) but too low for production processes (SNAP 4 and5) For SNAP 1 (power plants) for example they proposeda median release height of about 300 m whereas the medianin the EMEP profiles is about 400 m
Although we did not apply the profiles of Bieser et al(2011) our modification of SNAP 9 and the distinction be-tween point and area sources effectively reduces the emis-sion heights of CO2 compared to the standard EMEP pro-files Furthermore for the 22 largest point sources in Berlinas well as the six largest power plants in the model domainin Germany (Jaumlnschwalde Lippendorf Boxberg SchwarzePumpe) and Poland (Turoacutew Patnoacutew) the static EMEPSNAP 1 profiles were replaced by dynamic plume rise simu-lations for each hour of the year
The effective emission height can be much higher thanthe geometric height of a stack because of the momentumand buoyancy of the flue gas In general plume rise dependson stack geometry (height and diameter) flue gas proper-ties (temperature humidity exit velocity) and meteorolog-ical conditions (wind speed atmospheric stability) Plumerise and the vertical extent of the plumes were calculatedusing the empirical equations recommended by the Asso-ciation of German Engineers (VDI ndash Fachbereich Umwelt-meteorologie 1985) which are based on the original workof Briggs (1984) Hourly profiles of wind and temperaturefor the year 2015 were extracted from the COSMO-7 analy-ses of MeteoSwiss at the locations of the individual stacksFor power plants outside Berlin typical stack and flue gasparameters were mainly taken from published statistics forGermany (Pregger and Friedrich 2009) since these param-
4546 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 2 Vertical emission profiles applied for (a) point sources and (b) arealine sources The alternating gray and white backgroundsdenote the vertical layers
Table 1 Stack parameters used for plume rise calculation for the six largest power plants in the model domain
Name Longitude Latitude Stack Effluent Volume fluxb
a Average parameters by fuel and plant capacity taken from Pregger and Friedrich (2009) b httpsdewikipediaorgwikiKT1textbackslashT1textquotedblrightprotectT1textbraceleftuprotectT1textbracerighthlturm(last access 28 August 2018)
eters are not publicly available (see Table 1) For stacks inBerlin parameters were included in the emission inventoryprovided by the city of Berlin
Two complicating factors were not considered in thepresent study The European standards for large combustionplants (httpdataeuropaeuelidec_impl20171442oj lastaccess 26 March 2019) require the flue gas to be cleaned forsulfur and nitrogen oxides As a consequence of the chemicalwashing process the temperature of the flue gas is reduced
to a level where it can no longer be released via a classi-cal smoke stack (Busch et al 2002) In order to avoid re-heating the flue gas is often directed to the cooling towerand mixed into its moist buoyant air stream This is truefor the major German power plants in the domain whereasthe power plants Turoacutew and Patnoacutew in Poland were to thebest of our knowledge still equipped with smoke stacks in2015 Plume rise computations are much more complex inthe case of cooling towers due to latent heat release and
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4547
Table 2 Vertical emission profiles for point sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground)
Table 3 Vertical emission profiles for area sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground) Emissions in SNAP 1 are released exclusively from point sources
the interaction with moisture in the ambient air (Schatzmannand Policastro 1984) The additional release of latent heatmay enhance plume rise by 20 to 100 compared to adry plume (Hanna 1972) Second large power plants suchas Jaumlnschwalde are equipped with multiple cooling towers ashort distance from each other Their plumes tend to interactin a way that enhances plume rise an effect that additionallydepends on the alignment of the towers relative to the flowdirection (Bornoff and Mokhtarzadeh-Dehghan 2001) Ne-glecting these effects may thus underestimate true plume risein our calculations
3 Results
31 Plume rise at power plants
An example for the results of the plume rise simulations arepresented for Jaumlnschwalde the largest power plant in the do-main Figure 3a shows the hourly evolution of the plume cen-
ter heights during the year 2015 Plumes typically rise 100 to400 m above the top of the cooling tower but occasionallymuch further when both winds and atmospheric stability arelow Plume rise varies strongly from day to day and shows apronounced seasonal cycle Plumes rise on average to 360 mabove ground in summer but only to 250 m in winter Dueto the diurnal evolution of the PBL plume rise also varieswith the hour of the day except during winter In summerthe amplitude of this diurnal variability is about 150 m witha broad minimum at night from 2100 to 0800 LT (1900 to0600 UTC) and a peak in the early afternoon
Figure 4 compares the histograms of plume rise at Jaumln-schwalde in summer and winter to the standard EMEPSNAP 1 profile of Fig 2 In agreement with Bieser et al(2011) the computed profiles tend to place emissions sig-nificantly lower compared to EMEP even in summer Me-dian and mean effective emission heights in 2015 were 266and 310 m above ground respectively For the smaller powerplant (Lippendorf) these numbers were 187 and 210 m Sim-ulated plume rise for these two power plants was thus at least
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4542 D Brunner et al Atmospheric CO2 simulations with vertical emissions
regional scale (eg Sarrat et al 2009 Goeckede et al 2010Broquet et al 2011 Meesters et al 2012)
The Paris Climate Agreement adopted in 2015 (UnitedNations Framework Convention on Climate Change 2016)which requires each signing partner nation to accurately re-port its greenhouse gas (GHG) emissions and to reduce emis-sions in the future following its Nationally Determined Con-tribution has boosted the interest of the atmospheric sciencecommunity in quantifying not only natural fluxes but alsoanthropogenic emissions of CO2 ldquoTop-downrdquo inverse esti-mation of anthropogenic emissions from atmospheric obser-vations has in fact been proposed as an independent methodto complement the traditional ldquobottom-uprdquo collection of na-tional emission inventories (Nisbet and Weiss 2010) Sev-eral initiatives at the national and international level such asthe World Meteorological Organization (WMO)rsquos IntegratedGlobal Greenhouse Gas Information System (IG3IS) havebeen launched to advance and harmonize inverse methodswith the long-term vision to establish these methods as apolicy-relevant verification and support tool A summary ofthe state of the science in inverse modeling for verificationof greenhouse gas inventories has recently been presented byBergamaschi et al (2018a)
Because of the great challenge of accurately estimat-ing CO2 fluxes and distinguishing between the biosphericand anthropogenic contributions the scientific community iscalling for a globally integrated carbon observation and anal-ysis system This system should build on a substantially ex-panded ground-based and satellite observation capacity andshould integrate observations and bottom-up data with atmo-spheric transport modeling in an inverse framework (Ciaiset al 2014) The individual components and necessary stepstowards an operational European system for quantifying an-thropogenic CO2 emissions were outlined in two recent re-ports to the European Commission (Ciais et al 2015 Pintyet al 2017)
A central component of this system is a constellation ofCO2 satellites with imaging capability similar to the Car-bonSat concept proposed by Bovensmann et al (2010) Thesatellites are currently planned to be implemented in theframework of the European Earth observation Copernicusprogram and are commonly referred to as Sentinel-CO2These satellites should be able to quantify the CO2 emissionsfrom large sources such as power plants and cities during sin-gle overpasses Corresponding observing system simulationexperiments (OSSEs) were presented by Pillai et al (2016)and Broquet et al (2018)
In order to establish the requirements for the constella-tion of Sentinel-CO2 satellites the European Space Agency(ESA) has launched several scientific support studies includ-ing the use of satellite measurements of auxiliary reactivetrace gases for fossil fuel carbon dioxide emission estima-tion (SMARTCARB) a project that focused on the poten-tial of complementary satellite measurements of NO2 or COto improve the quantification of CO2 emissions from cities
and power plants The present study makes use of the high-resolution atmospheric transport simulations conducted inSMARTCARB to address a specific topic beyond the mainfocus of the project The core results of the SMARTCARBstudy will be presented elsewhere
Estimating CO2 fluxes by inverse modeling requires ac-curate simulation of observed atmospheric CO2 concentra-tions Systematic model biases are particularly critical asthey tend to directly translate into biased flux estimates Sev-eral studies investigated the impact of uncertainties in atmo-spheric transport on simulated CO2 concentrations and howthey could be accounted for in an inverse modeling frame-work (eg Lin and Gerbig 2005 Chan et al 2008 Lau-vaux et al 2009) A particular focus was placed on potentialbiases introduced by errors in vertical transport in the plane-tary boundary layer (PBL) (Gerbig et al 2008 Kretschmeret al 2012 2014) Uncertainties associated with bottom-upCO2 emissions were related to uncertainties in the horizon-tal gridding (Hogue et al 2016) or the representation of thetemporal variability (Liu et al 2017)
Uncertainties associated with the vertical placement ofemissions conversely have been largely ignored so far Infact in the vast majority of CO2 atmospheric transport andinverse modeling studies CO2 emissions were released ex-clusively at the surface (eg Sarrat et al 2009 Lauvauxet al 2009 Broquet et al 2011 Ganshin et al 2012Meesters et al 2012 Pillai et al 2016 Lauvaux et al 2016Liu et al 2017 Graven et al 2018 Fischer et al 2018)In Lagrangian models such as the Stochastic Time-InvertedLagrangian Transport model (STILT) (Lin et al 2003) orFLEXible PARTicle dispersion model (FLEXPART) (Stohlet al 2005) which are often used in backward adjoint modefor inverse modeling particles are typically sampled over afixed vertical depth above the surface or relative to the heightof planetary boundary layer to derive source sensitivitiesSimilar to the release of emissions at the surface in Eulerianmodels this ignores the potentially different sensitivities toemissions from elevated sources
These approaches in Eulerian and Lagrangian models arequestionable given the fact that a large proportion of CO2 isreleased from point sources such as power plants well abovethe surface Smoke stacks are in fact designed to minimizethe impact on concentrations at the ground This is partic-ularly important during situations with stable inversions inwinter or on clear nights However even in well-mixed con-ditions an elevated release may lead to a faster dilution anddifferent propagation of the signal due to wind speeds anddirections changing with altitude
In the air quality modeling community the importance ofvertically distributing emissions has been recognized muchearlier (eg Houyoux et al 2002) and is now well estab-lished especially for species such as SO2 that are primar-ily emitted from power plants and industrial sources (Bieseret al 2011 Mailler et al 2013 Karamchandani et al 2014Guevara et al 2014) Accounting for plume rise has also
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4543
been demonstrated to be critical for biomass burning emis-sions (Achtemeier et al 2011) The main goal of this studyis to demonstrate that this is also critical for the simulationof atmospheric CO2 concentrations
We employ very-high-resolution (11 kmtimes 11 km) simu-lations for the year 2015 conducted over a 750 kmtimes 650 kmwide domain centered in the city of Berlin Germany Thesimulations included separate tracers representing CO2 emit-ted only at the surface and CO2 emitted according to source-dependent vertical profiles As the domain covered severallarge coal-fired power plants we also investigate the im-pact of dynamically accounting for plume rise versus apply-ing static vertical emission profiles We focus on domain-averaged statistics rather than on individual plumes andon near-surface concentrations in the afternoon and totalcolumns at satellite overpass time as typically used in inversemodeling Conditions with a well-mixed and fully developedboundary layer during daytime are expected to minimize themismatch between model and observations
2 Data and methods
21 COSMO-GHG model
The Consortium for Small-scale Modeling (COSMO) is alimited-area non-hydrostatic numerical weather prediction(NWP) model developed by the German weather service to-gether with a consortium of seven European weather services(Baldauf et al 2011) In addition to operational weather pre-diction the model is applied widely for climate and air pollu-tion research in various modified and extended versions (egRockel et al 2008 Davin et al 2011 Vogel et al 2009Zubler et al 2011) making COSMO a versatile communitymodel
In the project CarboCount-CH (Oney et al 2015)COSMO was extended with a module for the simulation ofgreenhouse gases hereafter referred to as COSMO-GHGThe extension was built on a newly developed tracer module(Roches and Fuhrer 2012) which replaced the previous non-uniform treatment of meteorological tracers in the model Asimilar independently developed COSMO-based CO2 modelsystem was presented by Uebel and Bott (2018)
COSMO-GHG was first applied by Liu et al (2017) tostudy the spatiotemporal patterns of fossil fuel CO2 in Eu-rope They concluded that fossil fuel CO2 accounts for morethan half of the total (anthropogenic plus biospheric) tempo-ral variations in atmospheric CO2 over large parts of EuropeFurthermore they evaluated the model against observationsdemonstrating that the simulated fossil fuel variations favor-ably agreed with observations of fossil fuel CO2 derived fromconcurrent measurements of CO and 14CO2
COSMO is the first operational NWP model worldwidethat is running on hardware accelerated using graphical pro-cessing units (GPUs) (Fuhrer et al 2014) This highly effi-
cient code is operationally used by the Swiss weather serviceMeteoSwiss and has been applied for decadal convection-resolving climate simulations (Leutwyler et al 2017) In theframework of SMARTCARB the modules of the GHG ex-tension were ported to the GPU version following the con-cept of OpenACC compiler directives which is a high-levelapproach to offload compute-intensive parts to a GPU ac-celerator (Lapillonne and Fuhrer 2014) Benchmark testsshowed that the GPU version achieved a speedup by a factorof 6 which allowed for an increase in spatial and time res-olution and which greatly reduced the computational (andenergy) costs of the simulations conducted in this study
22 Model domain and setup
The model simulations were conducted in the framework ofthe project SMARTCARB which aimed at studying the po-tential benefit of adding an NO2 or CO channel to the instru-ment package of a future CO2 satellite mission with respectto quantifying CO2 emissions from strong localized sourcessuch as cities and power plants For this purpose a modeldomain was selected covering the city of Berlin as well asseveral large power plants in Germany Poland and CzechRepublic The domain extended approximately 750 km in theeastndashwest and 650 km in the southndashnorth directions to coverat least a complete 250 km wide satellite swath on either sideof the city Berlin was selected not only because it is oneof the largest cities in Europe but also because it is ratherisolated and because it has already been investigated in pre-vious CO2 modeling and observation studies (Pillai et al2016 Hase et al 2015) Simulations were conducted for thecomplete year 2015 at 11 kmtimes 11 km horizontal resolutionand 60 vertical levels with a top at 24 km The lowest modellayer had a thickness of 20 m The vertical allocation of theemissions relative to this vertical resolution of the model willbe described in Sect 24 The high horizontal resolution wasselected to comply with the small pixel size of 2 kmtimes 2 kmof the planned CO2 imaging satellite keeping in mind thatthe effective resolution of Eulerian transport models is muchlower than the spacing of the model grid (Kent et al 2014)
The simulations included a total of 50 different passivelytransported tracers of CO2 CO and NOx The tracers repre-sented different sources release times or release altitudes andincluded background tracers constrained at the lateral bound-aries by global-scale models Two additional tracers for bio-spheric fluxes (respiration and photosynthesis) were includedfor CO2 and four additional tracers with varying e-foldinglifetimes of 2 4 12 and 24 h for NOx In this study how-ever we focus on only a few of these tracers notably on thefollowing four anthropogenic CO2 emission tracers
ndash CO2_VERT all anthropogenic emissions in the modeldomain released according to source-specific verticalprofiles
4544 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ndash CO2_SURF same as CO2_VERT but all emissions re-leased at the surface ie into the lowest model layer
ndash CO2_PP-PR emissions from the six largest powerplants with explicit plume rise calculations
ndash CO2_PP-EMEP emissions from the six largest powerplants released according to a fixed vertical emissionprofile
The simulations were nested into the operational EuropeanCOSMO-7 analyses of MeteoSwiss which provided the lat-eral boundary conditions for meteorological variables (tem-perature pressure wind humidity clouds) at 7 km horizon-tal and hourly temporal resolution Because of the relativelysmall domain these boundary conditions provided a strongconstraint for the meteorology within the domain For CO2lateral boundary conditions were obtained from a global free-running high-resolution CO2 simulation (T1279 137 levelssim 15 km horizontal resolution) provided by the EuropeanCentre for Medium-Range Weather Forecasts (ECMWF)through the European Earth observation Copernicus program(Agustiacute-Panareda et al 2014) The simulation was indirectlyconstrained by in situ data by using a global climatology ofoptimized biospheric fluxes computed with the CO2 assimi-lation system of Chevallier et al (2010)
23 Emissions and biospheric fluxes
Anthropogenic emissions were obtained by merging theNetherlands Organisation for Applied Scientific ResearchEuropean Monitoring Atmospheric Composition and Cli-mate version 3 (TNOMACC-3) inventory with a detailed in-ventory available for the city of Berlin Dedicated municipalinventories can accurately account for the specific conditionsand activities in a city and may therefore significantly devi-ate from inventories obtained by simple downscaling fromlarger-scale inventories (Timmermans et al 2013 Gatelyand Hutyra 2017)
Version 3 of the TNOMACC inventory was used whichhas an improved representation of point sources as comparedto version 2 described in Kuenen et al (2014) It has a nomi-nal resolution of 116times 18 (approximately 7 kmtimes 7 km)but additionally reports the emissions from strong pointsources at their exact location Emissions are provided sepa-rately for different source categories following the Standard-ized Nomenclature for Air Pollutants (SNAP) classification(European Environment Agency 2002) The emissions of theyear 2011 were taken which was the most recent year avail-able from the inventory
The city of Berlin has developed a very detailed inven-tory for about 30 air pollutants and greenhouse gases forseven major source categories The inventory was providedin a Geographic Information System (GIS) format as a col-lection of shape files representing individual area point andline sources The latest available year was 2012 In our simu-lations separate tracers were included for three broad source
groups traffic industry and heating The attribution of the 10SNAP categories in TNOMACC-3 and the seven source cat-egories in the Berlin inventory to these groups are describedin detail in the final report of the SMARTCARB project(Kuhlmann et al 2019)
Both inventories were projected onto the COSMO modelgrid (rotated latitudendashlongitude grid with the North Poleat 43 N and 10W 001times 001 resolution) using mass-conservative methods Point sources were placed into theproper COSMO grid cell As a last step the two inventorieswere merged using a mask for the city of Berlin The mergedemission e per grid cell was estimated as e = (1minusf )eTNO+
eBerlin where f is the fraction of the grid cell area coveredby Berlin
Figure 1 presents a map of the CO2 inventory of Berlinin the original format and after projection onto the COSMOgrid The rasterized map reveals 22 strong CO2 point sourceswhich together account for 41 of the total emissions in thecity The two horizontal stripes correspond to the paths ofairplanes taking off and landing at the two main airports
In order to calculate hourly emissions as input for themodel simulations temporal scaling factors were applied de-scribing diurnal day-of-week and seasonal variations Thesame temporal profiles were used as in Liu et al (2017)which are based on factors originally developed for air pol-lution modeling (Builtjes et al 2003) These profiles dependon SNAP category and except for diurnal profiles on coun-try Diurnal profiles were matched to the local time of Ger-many
Biospheric CO2 fluxes due to gross photosynthetic pro-duction and respiration were computed at the resolution ofthe COSMO model using the Vegetation Photosynthesis andRespiration (VPRM) model (Mahadevan et al 2008) Thisdiagnostic model is based on meteorological input (2 m tem-perature and downward shortwave radiation at the surface)along with the Enhanced Vegetation Index (EVI) and LandSurface Water Index (LSWI) calculated from MODIS satel-lite reflectances These indices were available as an 8-dayproduct (MOD09A1 V006) at 500 m spatial resolution Veg-etation classes were determined from the 1 km SYNMAPland cover map (Jung et al 2006) Model parameter valuesdescribing the fluxes from different vegetation types weretaken from a previous study (Kountouris et al 2018) inwhich they were optimized using data from 47 Europeaneddy covariance measurement sites for the year 2007 Themodel was run offline for the entire year 2015 driven by thehighest resolution ECMWF meteorological data available
24 Vertical allocation of emissions and plume rise
Emissions were distributed in the vertical using source-specific profiles developed for the European Monitoring andEvaluation Program (EMEP) (see eg Bieser et al 2011)For the purpose of the present study the number of verticallayers was increased from 7 to 10 to enable a finer allocation
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4545
Figure 1 (a) Schematic of CO2 emissions in the Berlin inventory with point line and area sources from different emission categories Pointline and area sources are in different units (in kgsminus1 for point kgsminus1 mminus1 for line and kgsminus1 mminus2 for area sources) each being coloredseparately between minimum and maximum values (b) Total CO2 emissions re-projected and rasterized onto the COSMO model grid
to the model layers of COSMO-GHG Specifically the layers4ndash90 m and 90ndash170 m of the original EMEP profiles were di-vided into three and two sublayers respectively The attribu-tion of emissions to these sublayers was done in a rough wayfor example by placing a larger proportion of emissions fromldquocombustion in manufacturing industryrdquo into the upper partsbut distributing emissions from ldquonon-industrial (residential)combustionrdquo rather evenly over the three layers between 4and 90 m considering that industrial sources are likely toemit at higher altitudes The profile for SNAP 9 (waste) wasmodified in two ways (i) by moving 10 from higher layersto the lowest layer to account for CO2 emissions from land-fills and waste water treatment plants and (ii) by moving thelarge fractions originally placed into the layers 170ndash310 m(40 ) and 310ndash470 m (35 ) into lower layers between 90and 310 m This latter modification was made following thestudy of Pregger and Friedrich (2009) which showed that theemission-weighted height of waste incinerator stacks in Ger-many is only about 100 m and that 90 of the correspondingemissions including plume rise are expected to occur below300 m
The modified EMEP profiles are presented in Table 2 andillustrated in Fig 2a They were only applied to point sourcessuch as power plants and large industrial facilities A differ-ent set of profiles with lower average emission heights wasapplied to area sources as these represent much weaker andmore dispersed sources The corresponding profiles are pre-sented in Table 3 and Fig 2b The motivation for this dis-tinction is best illustrated for SNAP category 2 residentialand other non-industrial heating In the case of point sourcesthese are large heating facilities such as combined heat andpower plants with tall stacks In the case of area sources incontrast these are mostly private heating systems releasingCO2 through chimneys at roof level This is reflected in ourprofiles for SNAP 2 area sources being limited to the layersbetween 4 and 60 m above surface
As noted by Bieser et al (2011) the EMEP profiles arebased on very limited information originally collected forthe city of Zagreb Croatia They therefore proposed a dif-ferent set of profiles based on plume rise calculations for alarge number of point sources across Europe Their study in-dicated that the vertical placement of emissions in the EMEPprofiles is too high for combustion processes (SNAP 1 2 3and 9) but too low for production processes (SNAP 4 and5) For SNAP 1 (power plants) for example they proposeda median release height of about 300 m whereas the medianin the EMEP profiles is about 400 m
Although we did not apply the profiles of Bieser et al(2011) our modification of SNAP 9 and the distinction be-tween point and area sources effectively reduces the emis-sion heights of CO2 compared to the standard EMEP pro-files Furthermore for the 22 largest point sources in Berlinas well as the six largest power plants in the model domainin Germany (Jaumlnschwalde Lippendorf Boxberg SchwarzePumpe) and Poland (Turoacutew Patnoacutew) the static EMEPSNAP 1 profiles were replaced by dynamic plume rise simu-lations for each hour of the year
The effective emission height can be much higher thanthe geometric height of a stack because of the momentumand buoyancy of the flue gas In general plume rise dependson stack geometry (height and diameter) flue gas proper-ties (temperature humidity exit velocity) and meteorolog-ical conditions (wind speed atmospheric stability) Plumerise and the vertical extent of the plumes were calculatedusing the empirical equations recommended by the Asso-ciation of German Engineers (VDI ndash Fachbereich Umwelt-meteorologie 1985) which are based on the original workof Briggs (1984) Hourly profiles of wind and temperaturefor the year 2015 were extracted from the COSMO-7 analy-ses of MeteoSwiss at the locations of the individual stacksFor power plants outside Berlin typical stack and flue gasparameters were mainly taken from published statistics forGermany (Pregger and Friedrich 2009) since these param-
4546 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 2 Vertical emission profiles applied for (a) point sources and (b) arealine sources The alternating gray and white backgroundsdenote the vertical layers
Table 1 Stack parameters used for plume rise calculation for the six largest power plants in the model domain
Name Longitude Latitude Stack Effluent Volume fluxb
a Average parameters by fuel and plant capacity taken from Pregger and Friedrich (2009) b httpsdewikipediaorgwikiKT1textbackslashT1textquotedblrightprotectT1textbraceleftuprotectT1textbracerighthlturm(last access 28 August 2018)
eters are not publicly available (see Table 1) For stacks inBerlin parameters were included in the emission inventoryprovided by the city of Berlin
Two complicating factors were not considered in thepresent study The European standards for large combustionplants (httpdataeuropaeuelidec_impl20171442oj lastaccess 26 March 2019) require the flue gas to be cleaned forsulfur and nitrogen oxides As a consequence of the chemicalwashing process the temperature of the flue gas is reduced
to a level where it can no longer be released via a classi-cal smoke stack (Busch et al 2002) In order to avoid re-heating the flue gas is often directed to the cooling towerand mixed into its moist buoyant air stream This is truefor the major German power plants in the domain whereasthe power plants Turoacutew and Patnoacutew in Poland were to thebest of our knowledge still equipped with smoke stacks in2015 Plume rise computations are much more complex inthe case of cooling towers due to latent heat release and
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4547
Table 2 Vertical emission profiles for point sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground)
Table 3 Vertical emission profiles for area sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground) Emissions in SNAP 1 are released exclusively from point sources
the interaction with moisture in the ambient air (Schatzmannand Policastro 1984) The additional release of latent heatmay enhance plume rise by 20 to 100 compared to adry plume (Hanna 1972) Second large power plants suchas Jaumlnschwalde are equipped with multiple cooling towers ashort distance from each other Their plumes tend to interactin a way that enhances plume rise an effect that additionallydepends on the alignment of the towers relative to the flowdirection (Bornoff and Mokhtarzadeh-Dehghan 2001) Ne-glecting these effects may thus underestimate true plume risein our calculations
3 Results
31 Plume rise at power plants
An example for the results of the plume rise simulations arepresented for Jaumlnschwalde the largest power plant in the do-main Figure 3a shows the hourly evolution of the plume cen-
ter heights during the year 2015 Plumes typically rise 100 to400 m above the top of the cooling tower but occasionallymuch further when both winds and atmospheric stability arelow Plume rise varies strongly from day to day and shows apronounced seasonal cycle Plumes rise on average to 360 mabove ground in summer but only to 250 m in winter Dueto the diurnal evolution of the PBL plume rise also varieswith the hour of the day except during winter In summerthe amplitude of this diurnal variability is about 150 m witha broad minimum at night from 2100 to 0800 LT (1900 to0600 UTC) and a peak in the early afternoon
Figure 4 compares the histograms of plume rise at Jaumln-schwalde in summer and winter to the standard EMEPSNAP 1 profile of Fig 2 In agreement with Bieser et al(2011) the computed profiles tend to place emissions sig-nificantly lower compared to EMEP even in summer Me-dian and mean effective emission heights in 2015 were 266and 310 m above ground respectively For the smaller powerplant (Lippendorf) these numbers were 187 and 210 m Sim-ulated plume rise for these two power plants was thus at least
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4543
been demonstrated to be critical for biomass burning emis-sions (Achtemeier et al 2011) The main goal of this studyis to demonstrate that this is also critical for the simulationof atmospheric CO2 concentrations
We employ very-high-resolution (11 kmtimes 11 km) simu-lations for the year 2015 conducted over a 750 kmtimes 650 kmwide domain centered in the city of Berlin Germany Thesimulations included separate tracers representing CO2 emit-ted only at the surface and CO2 emitted according to source-dependent vertical profiles As the domain covered severallarge coal-fired power plants we also investigate the im-pact of dynamically accounting for plume rise versus apply-ing static vertical emission profiles We focus on domain-averaged statistics rather than on individual plumes andon near-surface concentrations in the afternoon and totalcolumns at satellite overpass time as typically used in inversemodeling Conditions with a well-mixed and fully developedboundary layer during daytime are expected to minimize themismatch between model and observations
2 Data and methods
21 COSMO-GHG model
The Consortium for Small-scale Modeling (COSMO) is alimited-area non-hydrostatic numerical weather prediction(NWP) model developed by the German weather service to-gether with a consortium of seven European weather services(Baldauf et al 2011) In addition to operational weather pre-diction the model is applied widely for climate and air pollu-tion research in various modified and extended versions (egRockel et al 2008 Davin et al 2011 Vogel et al 2009Zubler et al 2011) making COSMO a versatile communitymodel
In the project CarboCount-CH (Oney et al 2015)COSMO was extended with a module for the simulation ofgreenhouse gases hereafter referred to as COSMO-GHGThe extension was built on a newly developed tracer module(Roches and Fuhrer 2012) which replaced the previous non-uniform treatment of meteorological tracers in the model Asimilar independently developed COSMO-based CO2 modelsystem was presented by Uebel and Bott (2018)
COSMO-GHG was first applied by Liu et al (2017) tostudy the spatiotemporal patterns of fossil fuel CO2 in Eu-rope They concluded that fossil fuel CO2 accounts for morethan half of the total (anthropogenic plus biospheric) tempo-ral variations in atmospheric CO2 over large parts of EuropeFurthermore they evaluated the model against observationsdemonstrating that the simulated fossil fuel variations favor-ably agreed with observations of fossil fuel CO2 derived fromconcurrent measurements of CO and 14CO2
COSMO is the first operational NWP model worldwidethat is running on hardware accelerated using graphical pro-cessing units (GPUs) (Fuhrer et al 2014) This highly effi-
cient code is operationally used by the Swiss weather serviceMeteoSwiss and has been applied for decadal convection-resolving climate simulations (Leutwyler et al 2017) In theframework of SMARTCARB the modules of the GHG ex-tension were ported to the GPU version following the con-cept of OpenACC compiler directives which is a high-levelapproach to offload compute-intensive parts to a GPU ac-celerator (Lapillonne and Fuhrer 2014) Benchmark testsshowed that the GPU version achieved a speedup by a factorof 6 which allowed for an increase in spatial and time res-olution and which greatly reduced the computational (andenergy) costs of the simulations conducted in this study
22 Model domain and setup
The model simulations were conducted in the framework ofthe project SMARTCARB which aimed at studying the po-tential benefit of adding an NO2 or CO channel to the instru-ment package of a future CO2 satellite mission with respectto quantifying CO2 emissions from strong localized sourcessuch as cities and power plants For this purpose a modeldomain was selected covering the city of Berlin as well asseveral large power plants in Germany Poland and CzechRepublic The domain extended approximately 750 km in theeastndashwest and 650 km in the southndashnorth directions to coverat least a complete 250 km wide satellite swath on either sideof the city Berlin was selected not only because it is oneof the largest cities in Europe but also because it is ratherisolated and because it has already been investigated in pre-vious CO2 modeling and observation studies (Pillai et al2016 Hase et al 2015) Simulations were conducted for thecomplete year 2015 at 11 kmtimes 11 km horizontal resolutionand 60 vertical levels with a top at 24 km The lowest modellayer had a thickness of 20 m The vertical allocation of theemissions relative to this vertical resolution of the model willbe described in Sect 24 The high horizontal resolution wasselected to comply with the small pixel size of 2 kmtimes 2 kmof the planned CO2 imaging satellite keeping in mind thatthe effective resolution of Eulerian transport models is muchlower than the spacing of the model grid (Kent et al 2014)
The simulations included a total of 50 different passivelytransported tracers of CO2 CO and NOx The tracers repre-sented different sources release times or release altitudes andincluded background tracers constrained at the lateral bound-aries by global-scale models Two additional tracers for bio-spheric fluxes (respiration and photosynthesis) were includedfor CO2 and four additional tracers with varying e-foldinglifetimes of 2 4 12 and 24 h for NOx In this study how-ever we focus on only a few of these tracers notably on thefollowing four anthropogenic CO2 emission tracers
ndash CO2_VERT all anthropogenic emissions in the modeldomain released according to source-specific verticalprofiles
4544 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ndash CO2_SURF same as CO2_VERT but all emissions re-leased at the surface ie into the lowest model layer
ndash CO2_PP-PR emissions from the six largest powerplants with explicit plume rise calculations
ndash CO2_PP-EMEP emissions from the six largest powerplants released according to a fixed vertical emissionprofile
The simulations were nested into the operational EuropeanCOSMO-7 analyses of MeteoSwiss which provided the lat-eral boundary conditions for meteorological variables (tem-perature pressure wind humidity clouds) at 7 km horizon-tal and hourly temporal resolution Because of the relativelysmall domain these boundary conditions provided a strongconstraint for the meteorology within the domain For CO2lateral boundary conditions were obtained from a global free-running high-resolution CO2 simulation (T1279 137 levelssim 15 km horizontal resolution) provided by the EuropeanCentre for Medium-Range Weather Forecasts (ECMWF)through the European Earth observation Copernicus program(Agustiacute-Panareda et al 2014) The simulation was indirectlyconstrained by in situ data by using a global climatology ofoptimized biospheric fluxes computed with the CO2 assimi-lation system of Chevallier et al (2010)
23 Emissions and biospheric fluxes
Anthropogenic emissions were obtained by merging theNetherlands Organisation for Applied Scientific ResearchEuropean Monitoring Atmospheric Composition and Cli-mate version 3 (TNOMACC-3) inventory with a detailed in-ventory available for the city of Berlin Dedicated municipalinventories can accurately account for the specific conditionsand activities in a city and may therefore significantly devi-ate from inventories obtained by simple downscaling fromlarger-scale inventories (Timmermans et al 2013 Gatelyand Hutyra 2017)
Version 3 of the TNOMACC inventory was used whichhas an improved representation of point sources as comparedto version 2 described in Kuenen et al (2014) It has a nomi-nal resolution of 116times 18 (approximately 7 kmtimes 7 km)but additionally reports the emissions from strong pointsources at their exact location Emissions are provided sepa-rately for different source categories following the Standard-ized Nomenclature for Air Pollutants (SNAP) classification(European Environment Agency 2002) The emissions of theyear 2011 were taken which was the most recent year avail-able from the inventory
The city of Berlin has developed a very detailed inven-tory for about 30 air pollutants and greenhouse gases forseven major source categories The inventory was providedin a Geographic Information System (GIS) format as a col-lection of shape files representing individual area point andline sources The latest available year was 2012 In our simu-lations separate tracers were included for three broad source
groups traffic industry and heating The attribution of the 10SNAP categories in TNOMACC-3 and the seven source cat-egories in the Berlin inventory to these groups are describedin detail in the final report of the SMARTCARB project(Kuhlmann et al 2019)
Both inventories were projected onto the COSMO modelgrid (rotated latitudendashlongitude grid with the North Poleat 43 N and 10W 001times 001 resolution) using mass-conservative methods Point sources were placed into theproper COSMO grid cell As a last step the two inventorieswere merged using a mask for the city of Berlin The mergedemission e per grid cell was estimated as e = (1minusf )eTNO+
eBerlin where f is the fraction of the grid cell area coveredby Berlin
Figure 1 presents a map of the CO2 inventory of Berlinin the original format and after projection onto the COSMOgrid The rasterized map reveals 22 strong CO2 point sourceswhich together account for 41 of the total emissions in thecity The two horizontal stripes correspond to the paths ofairplanes taking off and landing at the two main airports
In order to calculate hourly emissions as input for themodel simulations temporal scaling factors were applied de-scribing diurnal day-of-week and seasonal variations Thesame temporal profiles were used as in Liu et al (2017)which are based on factors originally developed for air pol-lution modeling (Builtjes et al 2003) These profiles dependon SNAP category and except for diurnal profiles on coun-try Diurnal profiles were matched to the local time of Ger-many
Biospheric CO2 fluxes due to gross photosynthetic pro-duction and respiration were computed at the resolution ofthe COSMO model using the Vegetation Photosynthesis andRespiration (VPRM) model (Mahadevan et al 2008) Thisdiagnostic model is based on meteorological input (2 m tem-perature and downward shortwave radiation at the surface)along with the Enhanced Vegetation Index (EVI) and LandSurface Water Index (LSWI) calculated from MODIS satel-lite reflectances These indices were available as an 8-dayproduct (MOD09A1 V006) at 500 m spatial resolution Veg-etation classes were determined from the 1 km SYNMAPland cover map (Jung et al 2006) Model parameter valuesdescribing the fluxes from different vegetation types weretaken from a previous study (Kountouris et al 2018) inwhich they were optimized using data from 47 Europeaneddy covariance measurement sites for the year 2007 Themodel was run offline for the entire year 2015 driven by thehighest resolution ECMWF meteorological data available
24 Vertical allocation of emissions and plume rise
Emissions were distributed in the vertical using source-specific profiles developed for the European Monitoring andEvaluation Program (EMEP) (see eg Bieser et al 2011)For the purpose of the present study the number of verticallayers was increased from 7 to 10 to enable a finer allocation
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4545
Figure 1 (a) Schematic of CO2 emissions in the Berlin inventory with point line and area sources from different emission categories Pointline and area sources are in different units (in kgsminus1 for point kgsminus1 mminus1 for line and kgsminus1 mminus2 for area sources) each being coloredseparately between minimum and maximum values (b) Total CO2 emissions re-projected and rasterized onto the COSMO model grid
to the model layers of COSMO-GHG Specifically the layers4ndash90 m and 90ndash170 m of the original EMEP profiles were di-vided into three and two sublayers respectively The attribu-tion of emissions to these sublayers was done in a rough wayfor example by placing a larger proportion of emissions fromldquocombustion in manufacturing industryrdquo into the upper partsbut distributing emissions from ldquonon-industrial (residential)combustionrdquo rather evenly over the three layers between 4and 90 m considering that industrial sources are likely toemit at higher altitudes The profile for SNAP 9 (waste) wasmodified in two ways (i) by moving 10 from higher layersto the lowest layer to account for CO2 emissions from land-fills and waste water treatment plants and (ii) by moving thelarge fractions originally placed into the layers 170ndash310 m(40 ) and 310ndash470 m (35 ) into lower layers between 90and 310 m This latter modification was made following thestudy of Pregger and Friedrich (2009) which showed that theemission-weighted height of waste incinerator stacks in Ger-many is only about 100 m and that 90 of the correspondingemissions including plume rise are expected to occur below300 m
The modified EMEP profiles are presented in Table 2 andillustrated in Fig 2a They were only applied to point sourcessuch as power plants and large industrial facilities A differ-ent set of profiles with lower average emission heights wasapplied to area sources as these represent much weaker andmore dispersed sources The corresponding profiles are pre-sented in Table 3 and Fig 2b The motivation for this dis-tinction is best illustrated for SNAP category 2 residentialand other non-industrial heating In the case of point sourcesthese are large heating facilities such as combined heat andpower plants with tall stacks In the case of area sources incontrast these are mostly private heating systems releasingCO2 through chimneys at roof level This is reflected in ourprofiles for SNAP 2 area sources being limited to the layersbetween 4 and 60 m above surface
As noted by Bieser et al (2011) the EMEP profiles arebased on very limited information originally collected forthe city of Zagreb Croatia They therefore proposed a dif-ferent set of profiles based on plume rise calculations for alarge number of point sources across Europe Their study in-dicated that the vertical placement of emissions in the EMEPprofiles is too high for combustion processes (SNAP 1 2 3and 9) but too low for production processes (SNAP 4 and5) For SNAP 1 (power plants) for example they proposeda median release height of about 300 m whereas the medianin the EMEP profiles is about 400 m
Although we did not apply the profiles of Bieser et al(2011) our modification of SNAP 9 and the distinction be-tween point and area sources effectively reduces the emis-sion heights of CO2 compared to the standard EMEP pro-files Furthermore for the 22 largest point sources in Berlinas well as the six largest power plants in the model domainin Germany (Jaumlnschwalde Lippendorf Boxberg SchwarzePumpe) and Poland (Turoacutew Patnoacutew) the static EMEPSNAP 1 profiles were replaced by dynamic plume rise simu-lations for each hour of the year
The effective emission height can be much higher thanthe geometric height of a stack because of the momentumand buoyancy of the flue gas In general plume rise dependson stack geometry (height and diameter) flue gas proper-ties (temperature humidity exit velocity) and meteorolog-ical conditions (wind speed atmospheric stability) Plumerise and the vertical extent of the plumes were calculatedusing the empirical equations recommended by the Asso-ciation of German Engineers (VDI ndash Fachbereich Umwelt-meteorologie 1985) which are based on the original workof Briggs (1984) Hourly profiles of wind and temperaturefor the year 2015 were extracted from the COSMO-7 analy-ses of MeteoSwiss at the locations of the individual stacksFor power plants outside Berlin typical stack and flue gasparameters were mainly taken from published statistics forGermany (Pregger and Friedrich 2009) since these param-
4546 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 2 Vertical emission profiles applied for (a) point sources and (b) arealine sources The alternating gray and white backgroundsdenote the vertical layers
Table 1 Stack parameters used for plume rise calculation for the six largest power plants in the model domain
Name Longitude Latitude Stack Effluent Volume fluxb
a Average parameters by fuel and plant capacity taken from Pregger and Friedrich (2009) b httpsdewikipediaorgwikiKT1textbackslashT1textquotedblrightprotectT1textbraceleftuprotectT1textbracerighthlturm(last access 28 August 2018)
eters are not publicly available (see Table 1) For stacks inBerlin parameters were included in the emission inventoryprovided by the city of Berlin
Two complicating factors were not considered in thepresent study The European standards for large combustionplants (httpdataeuropaeuelidec_impl20171442oj lastaccess 26 March 2019) require the flue gas to be cleaned forsulfur and nitrogen oxides As a consequence of the chemicalwashing process the temperature of the flue gas is reduced
to a level where it can no longer be released via a classi-cal smoke stack (Busch et al 2002) In order to avoid re-heating the flue gas is often directed to the cooling towerand mixed into its moist buoyant air stream This is truefor the major German power plants in the domain whereasthe power plants Turoacutew and Patnoacutew in Poland were to thebest of our knowledge still equipped with smoke stacks in2015 Plume rise computations are much more complex inthe case of cooling towers due to latent heat release and
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4547
Table 2 Vertical emission profiles for point sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground)
Table 3 Vertical emission profiles for area sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground) Emissions in SNAP 1 are released exclusively from point sources
the interaction with moisture in the ambient air (Schatzmannand Policastro 1984) The additional release of latent heatmay enhance plume rise by 20 to 100 compared to adry plume (Hanna 1972) Second large power plants suchas Jaumlnschwalde are equipped with multiple cooling towers ashort distance from each other Their plumes tend to interactin a way that enhances plume rise an effect that additionallydepends on the alignment of the towers relative to the flowdirection (Bornoff and Mokhtarzadeh-Dehghan 2001) Ne-glecting these effects may thus underestimate true plume risein our calculations
3 Results
31 Plume rise at power plants
An example for the results of the plume rise simulations arepresented for Jaumlnschwalde the largest power plant in the do-main Figure 3a shows the hourly evolution of the plume cen-
ter heights during the year 2015 Plumes typically rise 100 to400 m above the top of the cooling tower but occasionallymuch further when both winds and atmospheric stability arelow Plume rise varies strongly from day to day and shows apronounced seasonal cycle Plumes rise on average to 360 mabove ground in summer but only to 250 m in winter Dueto the diurnal evolution of the PBL plume rise also varieswith the hour of the day except during winter In summerthe amplitude of this diurnal variability is about 150 m witha broad minimum at night from 2100 to 0800 LT (1900 to0600 UTC) and a peak in the early afternoon
Figure 4 compares the histograms of plume rise at Jaumln-schwalde in summer and winter to the standard EMEPSNAP 1 profile of Fig 2 In agreement with Bieser et al(2011) the computed profiles tend to place emissions sig-nificantly lower compared to EMEP even in summer Me-dian and mean effective emission heights in 2015 were 266and 310 m above ground respectively For the smaller powerplant (Lippendorf) these numbers were 187 and 210 m Sim-ulated plume rise for these two power plants was thus at least
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4544 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ndash CO2_SURF same as CO2_VERT but all emissions re-leased at the surface ie into the lowest model layer
ndash CO2_PP-PR emissions from the six largest powerplants with explicit plume rise calculations
ndash CO2_PP-EMEP emissions from the six largest powerplants released according to a fixed vertical emissionprofile
The simulations were nested into the operational EuropeanCOSMO-7 analyses of MeteoSwiss which provided the lat-eral boundary conditions for meteorological variables (tem-perature pressure wind humidity clouds) at 7 km horizon-tal and hourly temporal resolution Because of the relativelysmall domain these boundary conditions provided a strongconstraint for the meteorology within the domain For CO2lateral boundary conditions were obtained from a global free-running high-resolution CO2 simulation (T1279 137 levelssim 15 km horizontal resolution) provided by the EuropeanCentre for Medium-Range Weather Forecasts (ECMWF)through the European Earth observation Copernicus program(Agustiacute-Panareda et al 2014) The simulation was indirectlyconstrained by in situ data by using a global climatology ofoptimized biospheric fluxes computed with the CO2 assimi-lation system of Chevallier et al (2010)
23 Emissions and biospheric fluxes
Anthropogenic emissions were obtained by merging theNetherlands Organisation for Applied Scientific ResearchEuropean Monitoring Atmospheric Composition and Cli-mate version 3 (TNOMACC-3) inventory with a detailed in-ventory available for the city of Berlin Dedicated municipalinventories can accurately account for the specific conditionsand activities in a city and may therefore significantly devi-ate from inventories obtained by simple downscaling fromlarger-scale inventories (Timmermans et al 2013 Gatelyand Hutyra 2017)
Version 3 of the TNOMACC inventory was used whichhas an improved representation of point sources as comparedto version 2 described in Kuenen et al (2014) It has a nomi-nal resolution of 116times 18 (approximately 7 kmtimes 7 km)but additionally reports the emissions from strong pointsources at their exact location Emissions are provided sepa-rately for different source categories following the Standard-ized Nomenclature for Air Pollutants (SNAP) classification(European Environment Agency 2002) The emissions of theyear 2011 were taken which was the most recent year avail-able from the inventory
The city of Berlin has developed a very detailed inven-tory for about 30 air pollutants and greenhouse gases forseven major source categories The inventory was providedin a Geographic Information System (GIS) format as a col-lection of shape files representing individual area point andline sources The latest available year was 2012 In our simu-lations separate tracers were included for three broad source
groups traffic industry and heating The attribution of the 10SNAP categories in TNOMACC-3 and the seven source cat-egories in the Berlin inventory to these groups are describedin detail in the final report of the SMARTCARB project(Kuhlmann et al 2019)
Both inventories were projected onto the COSMO modelgrid (rotated latitudendashlongitude grid with the North Poleat 43 N and 10W 001times 001 resolution) using mass-conservative methods Point sources were placed into theproper COSMO grid cell As a last step the two inventorieswere merged using a mask for the city of Berlin The mergedemission e per grid cell was estimated as e = (1minusf )eTNO+
eBerlin where f is the fraction of the grid cell area coveredby Berlin
Figure 1 presents a map of the CO2 inventory of Berlinin the original format and after projection onto the COSMOgrid The rasterized map reveals 22 strong CO2 point sourceswhich together account for 41 of the total emissions in thecity The two horizontal stripes correspond to the paths ofairplanes taking off and landing at the two main airports
In order to calculate hourly emissions as input for themodel simulations temporal scaling factors were applied de-scribing diurnal day-of-week and seasonal variations Thesame temporal profiles were used as in Liu et al (2017)which are based on factors originally developed for air pol-lution modeling (Builtjes et al 2003) These profiles dependon SNAP category and except for diurnal profiles on coun-try Diurnal profiles were matched to the local time of Ger-many
Biospheric CO2 fluxes due to gross photosynthetic pro-duction and respiration were computed at the resolution ofthe COSMO model using the Vegetation Photosynthesis andRespiration (VPRM) model (Mahadevan et al 2008) Thisdiagnostic model is based on meteorological input (2 m tem-perature and downward shortwave radiation at the surface)along with the Enhanced Vegetation Index (EVI) and LandSurface Water Index (LSWI) calculated from MODIS satel-lite reflectances These indices were available as an 8-dayproduct (MOD09A1 V006) at 500 m spatial resolution Veg-etation classes were determined from the 1 km SYNMAPland cover map (Jung et al 2006) Model parameter valuesdescribing the fluxes from different vegetation types weretaken from a previous study (Kountouris et al 2018) inwhich they were optimized using data from 47 Europeaneddy covariance measurement sites for the year 2007 Themodel was run offline for the entire year 2015 driven by thehighest resolution ECMWF meteorological data available
24 Vertical allocation of emissions and plume rise
Emissions were distributed in the vertical using source-specific profiles developed for the European Monitoring andEvaluation Program (EMEP) (see eg Bieser et al 2011)For the purpose of the present study the number of verticallayers was increased from 7 to 10 to enable a finer allocation
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4545
Figure 1 (a) Schematic of CO2 emissions in the Berlin inventory with point line and area sources from different emission categories Pointline and area sources are in different units (in kgsminus1 for point kgsminus1 mminus1 for line and kgsminus1 mminus2 for area sources) each being coloredseparately between minimum and maximum values (b) Total CO2 emissions re-projected and rasterized onto the COSMO model grid
to the model layers of COSMO-GHG Specifically the layers4ndash90 m and 90ndash170 m of the original EMEP profiles were di-vided into three and two sublayers respectively The attribu-tion of emissions to these sublayers was done in a rough wayfor example by placing a larger proportion of emissions fromldquocombustion in manufacturing industryrdquo into the upper partsbut distributing emissions from ldquonon-industrial (residential)combustionrdquo rather evenly over the three layers between 4and 90 m considering that industrial sources are likely toemit at higher altitudes The profile for SNAP 9 (waste) wasmodified in two ways (i) by moving 10 from higher layersto the lowest layer to account for CO2 emissions from land-fills and waste water treatment plants and (ii) by moving thelarge fractions originally placed into the layers 170ndash310 m(40 ) and 310ndash470 m (35 ) into lower layers between 90and 310 m This latter modification was made following thestudy of Pregger and Friedrich (2009) which showed that theemission-weighted height of waste incinerator stacks in Ger-many is only about 100 m and that 90 of the correspondingemissions including plume rise are expected to occur below300 m
The modified EMEP profiles are presented in Table 2 andillustrated in Fig 2a They were only applied to point sourcessuch as power plants and large industrial facilities A differ-ent set of profiles with lower average emission heights wasapplied to area sources as these represent much weaker andmore dispersed sources The corresponding profiles are pre-sented in Table 3 and Fig 2b The motivation for this dis-tinction is best illustrated for SNAP category 2 residentialand other non-industrial heating In the case of point sourcesthese are large heating facilities such as combined heat andpower plants with tall stacks In the case of area sources incontrast these are mostly private heating systems releasingCO2 through chimneys at roof level This is reflected in ourprofiles for SNAP 2 area sources being limited to the layersbetween 4 and 60 m above surface
As noted by Bieser et al (2011) the EMEP profiles arebased on very limited information originally collected forthe city of Zagreb Croatia They therefore proposed a dif-ferent set of profiles based on plume rise calculations for alarge number of point sources across Europe Their study in-dicated that the vertical placement of emissions in the EMEPprofiles is too high for combustion processes (SNAP 1 2 3and 9) but too low for production processes (SNAP 4 and5) For SNAP 1 (power plants) for example they proposeda median release height of about 300 m whereas the medianin the EMEP profiles is about 400 m
Although we did not apply the profiles of Bieser et al(2011) our modification of SNAP 9 and the distinction be-tween point and area sources effectively reduces the emis-sion heights of CO2 compared to the standard EMEP pro-files Furthermore for the 22 largest point sources in Berlinas well as the six largest power plants in the model domainin Germany (Jaumlnschwalde Lippendorf Boxberg SchwarzePumpe) and Poland (Turoacutew Patnoacutew) the static EMEPSNAP 1 profiles were replaced by dynamic plume rise simu-lations for each hour of the year
The effective emission height can be much higher thanthe geometric height of a stack because of the momentumand buoyancy of the flue gas In general plume rise dependson stack geometry (height and diameter) flue gas proper-ties (temperature humidity exit velocity) and meteorolog-ical conditions (wind speed atmospheric stability) Plumerise and the vertical extent of the plumes were calculatedusing the empirical equations recommended by the Asso-ciation of German Engineers (VDI ndash Fachbereich Umwelt-meteorologie 1985) which are based on the original workof Briggs (1984) Hourly profiles of wind and temperaturefor the year 2015 were extracted from the COSMO-7 analy-ses of MeteoSwiss at the locations of the individual stacksFor power plants outside Berlin typical stack and flue gasparameters were mainly taken from published statistics forGermany (Pregger and Friedrich 2009) since these param-
4546 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 2 Vertical emission profiles applied for (a) point sources and (b) arealine sources The alternating gray and white backgroundsdenote the vertical layers
Table 1 Stack parameters used for plume rise calculation for the six largest power plants in the model domain
Name Longitude Latitude Stack Effluent Volume fluxb
a Average parameters by fuel and plant capacity taken from Pregger and Friedrich (2009) b httpsdewikipediaorgwikiKT1textbackslashT1textquotedblrightprotectT1textbraceleftuprotectT1textbracerighthlturm(last access 28 August 2018)
eters are not publicly available (see Table 1) For stacks inBerlin parameters were included in the emission inventoryprovided by the city of Berlin
Two complicating factors were not considered in thepresent study The European standards for large combustionplants (httpdataeuropaeuelidec_impl20171442oj lastaccess 26 March 2019) require the flue gas to be cleaned forsulfur and nitrogen oxides As a consequence of the chemicalwashing process the temperature of the flue gas is reduced
to a level where it can no longer be released via a classi-cal smoke stack (Busch et al 2002) In order to avoid re-heating the flue gas is often directed to the cooling towerand mixed into its moist buoyant air stream This is truefor the major German power plants in the domain whereasthe power plants Turoacutew and Patnoacutew in Poland were to thebest of our knowledge still equipped with smoke stacks in2015 Plume rise computations are much more complex inthe case of cooling towers due to latent heat release and
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4547
Table 2 Vertical emission profiles for point sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground)
Table 3 Vertical emission profiles for area sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground) Emissions in SNAP 1 are released exclusively from point sources
the interaction with moisture in the ambient air (Schatzmannand Policastro 1984) The additional release of latent heatmay enhance plume rise by 20 to 100 compared to adry plume (Hanna 1972) Second large power plants suchas Jaumlnschwalde are equipped with multiple cooling towers ashort distance from each other Their plumes tend to interactin a way that enhances plume rise an effect that additionallydepends on the alignment of the towers relative to the flowdirection (Bornoff and Mokhtarzadeh-Dehghan 2001) Ne-glecting these effects may thus underestimate true plume risein our calculations
3 Results
31 Plume rise at power plants
An example for the results of the plume rise simulations arepresented for Jaumlnschwalde the largest power plant in the do-main Figure 3a shows the hourly evolution of the plume cen-
ter heights during the year 2015 Plumes typically rise 100 to400 m above the top of the cooling tower but occasionallymuch further when both winds and atmospheric stability arelow Plume rise varies strongly from day to day and shows apronounced seasonal cycle Plumes rise on average to 360 mabove ground in summer but only to 250 m in winter Dueto the diurnal evolution of the PBL plume rise also varieswith the hour of the day except during winter In summerthe amplitude of this diurnal variability is about 150 m witha broad minimum at night from 2100 to 0800 LT (1900 to0600 UTC) and a peak in the early afternoon
Figure 4 compares the histograms of plume rise at Jaumln-schwalde in summer and winter to the standard EMEPSNAP 1 profile of Fig 2 In agreement with Bieser et al(2011) the computed profiles tend to place emissions sig-nificantly lower compared to EMEP even in summer Me-dian and mean effective emission heights in 2015 were 266and 310 m above ground respectively For the smaller powerplant (Lippendorf) these numbers were 187 and 210 m Sim-ulated plume rise for these two power plants was thus at least
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4545
Figure 1 (a) Schematic of CO2 emissions in the Berlin inventory with point line and area sources from different emission categories Pointline and area sources are in different units (in kgsminus1 for point kgsminus1 mminus1 for line and kgsminus1 mminus2 for area sources) each being coloredseparately between minimum and maximum values (b) Total CO2 emissions re-projected and rasterized onto the COSMO model grid
to the model layers of COSMO-GHG Specifically the layers4ndash90 m and 90ndash170 m of the original EMEP profiles were di-vided into three and two sublayers respectively The attribu-tion of emissions to these sublayers was done in a rough wayfor example by placing a larger proportion of emissions fromldquocombustion in manufacturing industryrdquo into the upper partsbut distributing emissions from ldquonon-industrial (residential)combustionrdquo rather evenly over the three layers between 4and 90 m considering that industrial sources are likely toemit at higher altitudes The profile for SNAP 9 (waste) wasmodified in two ways (i) by moving 10 from higher layersto the lowest layer to account for CO2 emissions from land-fills and waste water treatment plants and (ii) by moving thelarge fractions originally placed into the layers 170ndash310 m(40 ) and 310ndash470 m (35 ) into lower layers between 90and 310 m This latter modification was made following thestudy of Pregger and Friedrich (2009) which showed that theemission-weighted height of waste incinerator stacks in Ger-many is only about 100 m and that 90 of the correspondingemissions including plume rise are expected to occur below300 m
The modified EMEP profiles are presented in Table 2 andillustrated in Fig 2a They were only applied to point sourcessuch as power plants and large industrial facilities A differ-ent set of profiles with lower average emission heights wasapplied to area sources as these represent much weaker andmore dispersed sources The corresponding profiles are pre-sented in Table 3 and Fig 2b The motivation for this dis-tinction is best illustrated for SNAP category 2 residentialand other non-industrial heating In the case of point sourcesthese are large heating facilities such as combined heat andpower plants with tall stacks In the case of area sources incontrast these are mostly private heating systems releasingCO2 through chimneys at roof level This is reflected in ourprofiles for SNAP 2 area sources being limited to the layersbetween 4 and 60 m above surface
As noted by Bieser et al (2011) the EMEP profiles arebased on very limited information originally collected forthe city of Zagreb Croatia They therefore proposed a dif-ferent set of profiles based on plume rise calculations for alarge number of point sources across Europe Their study in-dicated that the vertical placement of emissions in the EMEPprofiles is too high for combustion processes (SNAP 1 2 3and 9) but too low for production processes (SNAP 4 and5) For SNAP 1 (power plants) for example they proposeda median release height of about 300 m whereas the medianin the EMEP profiles is about 400 m
Although we did not apply the profiles of Bieser et al(2011) our modification of SNAP 9 and the distinction be-tween point and area sources effectively reduces the emis-sion heights of CO2 compared to the standard EMEP pro-files Furthermore for the 22 largest point sources in Berlinas well as the six largest power plants in the model domainin Germany (Jaumlnschwalde Lippendorf Boxberg SchwarzePumpe) and Poland (Turoacutew Patnoacutew) the static EMEPSNAP 1 profiles were replaced by dynamic plume rise simu-lations for each hour of the year
The effective emission height can be much higher thanthe geometric height of a stack because of the momentumand buoyancy of the flue gas In general plume rise dependson stack geometry (height and diameter) flue gas proper-ties (temperature humidity exit velocity) and meteorolog-ical conditions (wind speed atmospheric stability) Plumerise and the vertical extent of the plumes were calculatedusing the empirical equations recommended by the Asso-ciation of German Engineers (VDI ndash Fachbereich Umwelt-meteorologie 1985) which are based on the original workof Briggs (1984) Hourly profiles of wind and temperaturefor the year 2015 were extracted from the COSMO-7 analy-ses of MeteoSwiss at the locations of the individual stacksFor power plants outside Berlin typical stack and flue gasparameters were mainly taken from published statistics forGermany (Pregger and Friedrich 2009) since these param-
4546 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 2 Vertical emission profiles applied for (a) point sources and (b) arealine sources The alternating gray and white backgroundsdenote the vertical layers
Table 1 Stack parameters used for plume rise calculation for the six largest power plants in the model domain
Name Longitude Latitude Stack Effluent Volume fluxb
a Average parameters by fuel and plant capacity taken from Pregger and Friedrich (2009) b httpsdewikipediaorgwikiKT1textbackslashT1textquotedblrightprotectT1textbraceleftuprotectT1textbracerighthlturm(last access 28 August 2018)
eters are not publicly available (see Table 1) For stacks inBerlin parameters were included in the emission inventoryprovided by the city of Berlin
Two complicating factors were not considered in thepresent study The European standards for large combustionplants (httpdataeuropaeuelidec_impl20171442oj lastaccess 26 March 2019) require the flue gas to be cleaned forsulfur and nitrogen oxides As a consequence of the chemicalwashing process the temperature of the flue gas is reduced
to a level where it can no longer be released via a classi-cal smoke stack (Busch et al 2002) In order to avoid re-heating the flue gas is often directed to the cooling towerand mixed into its moist buoyant air stream This is truefor the major German power plants in the domain whereasthe power plants Turoacutew and Patnoacutew in Poland were to thebest of our knowledge still equipped with smoke stacks in2015 Plume rise computations are much more complex inthe case of cooling towers due to latent heat release and
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4547
Table 2 Vertical emission profiles for point sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground)
Table 3 Vertical emission profiles for area sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground) Emissions in SNAP 1 are released exclusively from point sources
the interaction with moisture in the ambient air (Schatzmannand Policastro 1984) The additional release of latent heatmay enhance plume rise by 20 to 100 compared to adry plume (Hanna 1972) Second large power plants suchas Jaumlnschwalde are equipped with multiple cooling towers ashort distance from each other Their plumes tend to interactin a way that enhances plume rise an effect that additionallydepends on the alignment of the towers relative to the flowdirection (Bornoff and Mokhtarzadeh-Dehghan 2001) Ne-glecting these effects may thus underestimate true plume risein our calculations
3 Results
31 Plume rise at power plants
An example for the results of the plume rise simulations arepresented for Jaumlnschwalde the largest power plant in the do-main Figure 3a shows the hourly evolution of the plume cen-
ter heights during the year 2015 Plumes typically rise 100 to400 m above the top of the cooling tower but occasionallymuch further when both winds and atmospheric stability arelow Plume rise varies strongly from day to day and shows apronounced seasonal cycle Plumes rise on average to 360 mabove ground in summer but only to 250 m in winter Dueto the diurnal evolution of the PBL plume rise also varieswith the hour of the day except during winter In summerthe amplitude of this diurnal variability is about 150 m witha broad minimum at night from 2100 to 0800 LT (1900 to0600 UTC) and a peak in the early afternoon
Figure 4 compares the histograms of plume rise at Jaumln-schwalde in summer and winter to the standard EMEPSNAP 1 profile of Fig 2 In agreement with Bieser et al(2011) the computed profiles tend to place emissions sig-nificantly lower compared to EMEP even in summer Me-dian and mean effective emission heights in 2015 were 266and 310 m above ground respectively For the smaller powerplant (Lippendorf) these numbers were 187 and 210 m Sim-ulated plume rise for these two power plants was thus at least
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4546 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 2 Vertical emission profiles applied for (a) point sources and (b) arealine sources The alternating gray and white backgroundsdenote the vertical layers
Table 1 Stack parameters used for plume rise calculation for the six largest power plants in the model domain
Name Longitude Latitude Stack Effluent Volume fluxb
a Average parameters by fuel and plant capacity taken from Pregger and Friedrich (2009) b httpsdewikipediaorgwikiKT1textbackslashT1textquotedblrightprotectT1textbraceleftuprotectT1textbracerighthlturm(last access 28 August 2018)
eters are not publicly available (see Table 1) For stacks inBerlin parameters were included in the emission inventoryprovided by the city of Berlin
Two complicating factors were not considered in thepresent study The European standards for large combustionplants (httpdataeuropaeuelidec_impl20171442oj lastaccess 26 March 2019) require the flue gas to be cleaned forsulfur and nitrogen oxides As a consequence of the chemicalwashing process the temperature of the flue gas is reduced
to a level where it can no longer be released via a classi-cal smoke stack (Busch et al 2002) In order to avoid re-heating the flue gas is often directed to the cooling towerand mixed into its moist buoyant air stream This is truefor the major German power plants in the domain whereasthe power plants Turoacutew and Patnoacutew in Poland were to thebest of our knowledge still equipped with smoke stacks in2015 Plume rise computations are much more complex inthe case of cooling towers due to latent heat release and
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4547
Table 2 Vertical emission profiles for point sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground)
Table 3 Vertical emission profiles for area sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground) Emissions in SNAP 1 are released exclusively from point sources
the interaction with moisture in the ambient air (Schatzmannand Policastro 1984) The additional release of latent heatmay enhance plume rise by 20 to 100 compared to adry plume (Hanna 1972) Second large power plants suchas Jaumlnschwalde are equipped with multiple cooling towers ashort distance from each other Their plumes tend to interactin a way that enhances plume rise an effect that additionallydepends on the alignment of the towers relative to the flowdirection (Bornoff and Mokhtarzadeh-Dehghan 2001) Ne-glecting these effects may thus underestimate true plume risein our calculations
3 Results
31 Plume rise at power plants
An example for the results of the plume rise simulations arepresented for Jaumlnschwalde the largest power plant in the do-main Figure 3a shows the hourly evolution of the plume cen-
ter heights during the year 2015 Plumes typically rise 100 to400 m above the top of the cooling tower but occasionallymuch further when both winds and atmospheric stability arelow Plume rise varies strongly from day to day and shows apronounced seasonal cycle Plumes rise on average to 360 mabove ground in summer but only to 250 m in winter Dueto the diurnal evolution of the PBL plume rise also varieswith the hour of the day except during winter In summerthe amplitude of this diurnal variability is about 150 m witha broad minimum at night from 2100 to 0800 LT (1900 to0600 UTC) and a peak in the early afternoon
Figure 4 compares the histograms of plume rise at Jaumln-schwalde in summer and winter to the standard EMEPSNAP 1 profile of Fig 2 In agreement with Bieser et al(2011) the computed profiles tend to place emissions sig-nificantly lower compared to EMEP even in summer Me-dian and mean effective emission heights in 2015 were 266and 310 m above ground respectively For the smaller powerplant (Lippendorf) these numbers were 187 and 210 m Sim-ulated plume rise for these two power plants was thus at least
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4547
Table 2 Vertical emission profiles for point sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground)
Table 3 Vertical emission profiles for area sources for different SNAP categories as fractions of the total emitted mass per vertical layerLayers are denoted by their lower and upper limits (meters above ground) Emissions in SNAP 1 are released exclusively from point sources
the interaction with moisture in the ambient air (Schatzmannand Policastro 1984) The additional release of latent heatmay enhance plume rise by 20 to 100 compared to adry plume (Hanna 1972) Second large power plants suchas Jaumlnschwalde are equipped with multiple cooling towers ashort distance from each other Their plumes tend to interactin a way that enhances plume rise an effect that additionallydepends on the alignment of the towers relative to the flowdirection (Bornoff and Mokhtarzadeh-Dehghan 2001) Ne-glecting these effects may thus underestimate true plume risein our calculations
3 Results
31 Plume rise at power plants
An example for the results of the plume rise simulations arepresented for Jaumlnschwalde the largest power plant in the do-main Figure 3a shows the hourly evolution of the plume cen-
ter heights during the year 2015 Plumes typically rise 100 to400 m above the top of the cooling tower but occasionallymuch further when both winds and atmospheric stability arelow Plume rise varies strongly from day to day and shows apronounced seasonal cycle Plumes rise on average to 360 mabove ground in summer but only to 250 m in winter Dueto the diurnal evolution of the PBL plume rise also varieswith the hour of the day except during winter In summerthe amplitude of this diurnal variability is about 150 m witha broad minimum at night from 2100 to 0800 LT (1900 to0600 UTC) and a peak in the early afternoon
Figure 4 compares the histograms of plume rise at Jaumln-schwalde in summer and winter to the standard EMEPSNAP 1 profile of Fig 2 In agreement with Bieser et al(2011) the computed profiles tend to place emissions sig-nificantly lower compared to EMEP even in summer Me-dian and mean effective emission heights in 2015 were 266and 310 m above ground respectively For the smaller powerplant (Lippendorf) these numbers were 187 and 210 m Sim-ulated plume rise for these two power plants was thus at least
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4548 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 3 Effective CO2 emission heights at the power plant in Jaumlnschwalde Germany based on plume rise calculations (a) Hourly plumerise in 2015 The orange line is a 30-day moving average The black solid line denotes the height of the cooling tower (120 m) (b) Meanannual winter (DJF) and summer (JJA) diurnal cycles
Figure 4 Normalized histogram of plume altitude above groundat the Jaumlnschwalde power plant in winter (blue) and summer (red)The blue line shows the EMEP standard profile for power plants forcomparison Areas below the curves integrated along the verticalaxis are normalized to 1
100 m lower compared to the EMEP profile The large frac-tion of emissions placed above 500 m in the EMEP profileseems particularly unrealistic Our plume rise calculationsare more consistent with the profiles recommended by Bieseret al (2011)
32 Emission profiles over Berlin
The main emission sources of CO2 in Berlin are ldquotrafficrdquoldquoprivate households and public buildingsrdquo ldquoindustryrdquo ldquotrade
and othersrdquo and the ldquotransformation sectorrdquo which includeslarge facilities for heat and electricity production In 2015traffic only accounted for 29 of all emissions householdsand industry for 27 (Amt fuumlr Statistik Berlin-Brandenburg2018) By far the largest single source was the transformationsector with a share of 42 of the total As a result a signifi-cant fraction of emissions is released through stacks
Figure 5 shows the monthly mean vertical profiles of CO2emissions from Berlin in a winter month (February) and asummer month (August) as used in our simulations The pro-files reflect the superposition of emissions from different sec-tors with different vertical profiles (see Fig 2) and differentseasonal contributions In February for example residentialheating is an important source between 4 and 60 m aboveground whereas emissions in these layers are almost com-pletely absent in August Although in both months the pro-files have a pronounced peak at the surface 36 of CO2is released above 90 m in February with that share rising to58 in August when residential heating is low These num-bers suggest that a proper vertical allocation of emissions isimportant even in cities
The vertical profiles of the emissions of CO and NOx areoverlaid in Fig 5 for comparison since coincident measure-ments of NO2 or CO may be used by a future CO2 satellitefor emission quantification Both CO and NOx have a morepronounced peak at the surface due to the larger proportionof traffic emissions Similar to CO2 a large (albeit smaller)fraction of NOx is emitted well above ground by large pointsources Interestingly these sources emit very little CO sug-gesting efficient combustion or cleaning of the exhaust Thefraction of CO released above 90 m is below 5 in sharpcontrast to CO2
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4549
Figure 5 Monthly mean vertical emission profiles of CO2 CO and NOx over Berlin in (a) February and (b) August
33 CO2 at the surface
Maps of the monthly mean afternoon CO2 dry air mole frac-tions (hereafter referred to as ldquoconcentrationsrdquo) in the low-est model layer (0ndash20 m) from all anthropogenic emissionswithin the model domain are presented in Fig 6 for Jan-uary and July respectively Afternoon values were selectedsince current inversion systems usually assimilate only ob-servations in the afternoon when vertical concentration gra-dients are smallest (eg Peters et al 2010) Figure 6a and cshow the results for the tracer CO2_VERT with verticallydistributed emissions while Fig 6b and d show the tracerCO2_SURF with all emissions concentrated at the surfaceFor all power plants labeled in the figure plume rise was ex-plicitly simulated for the tracer CO2_VERT
The concentrations are generally higher in January than inJuly due to higher emissions and reduced vertical mixing inwinter The concentrations are also significantly higher whenall CO2 is released at the surface The main reason for thesedifferences are power plants and other point sources whichstand out prominently in the maps for the tracer CO2_SURFbut not for CO2_VERT
As shown in Fig 7 the differences are much larger in win-ter than in summer In summer power plant emissions aremixed efficiently over the depth of the afternoon PBL Since
this mixing is not instantaneous differences are noticeableclose to the sources but fade out rapidly with increasing dis-tance In winter conversely when vertical mixing is weakthe differences between the two tracers remain well above1 ppm over distances of a few tens of kilometers downstreamoccasionally over 100 km or more
Domain-averaged monthly mean diurnal cycles of thetwo tracers are presented in Fig 8 for the months of Jan-uary and July Consistent with the maps the concentra-tions of CO2_SURF are significantly higher than those ofCO2_VERT This difference is around 16 ppm in Januaryand almost constant during the day The variations duringthe day appear to be dominated by the diurnal cycle of theemissions rather than by the dynamics of the PBL
In summer the differences are generally smaller and ex-hibit a pronounced diurnal cycle Differences are about1 ppm at night and almost vanish (about 01 ppm) in the af-ternoon Due to the low PBL at night the concentrations in-crease overnight despite relatively low emissions This in-crease is much more pronounced for tracer CO2_SURFwhich is susceptible to surface emissions from point sourcesthat do not stop at night The tracer CO2_VERT only showsa marked increase during the early morning hours whentraffic increases and the PBL is still low Emissions frompoint sources on the other hand are likely released above
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4550 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 6 Mean afternoon (1400ndash1600 LT) near-surface CO2 in (a b) January and (c d) July contributed by all anthropogenic sources inthe domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b d) Tracer CO2_SURF released only at the surface
Figure 7 Difference in mean afternoon (1400ndash1600 LT) near-surface CO2 between tracers CO2_SURF and CO2_VERT in (a) January and(b) July
the nocturnal PBL leading to marked differences betweenCO2_VERT and CO2_SURF at night
Statistics of the afternoon concentrations of the two CO2tracers are summarized in Table 4 in terms of mean val-ues and different percentiles of the frequency distributionDomain-averaged CO2 concentrations are 43 higher inJanuary when all CO2 is released at the surface compared to
when emissions are distributed vertically The differences arelarger for the high percentiles suggesting that backgroundvalues are less affected than peak values This is understand-able as vertical mixing tends to reduce the differences withincreasing distance from the sources In summer the differ-ences are generally much smaller but as suggested in Fig 8this is only true for afternoon concentrations Mean differ-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4551
Figure 8 Mean diurnal cycles of the tracers CO2_VERT andCO2_SURF in January and July
ences in the afternoon are of the order of 14 Again higherpercentiles tend to show larger differences The impact onobservations from tall tower networks measuring CO2 some100 to 300 m above the surface (Bakwin et al 1995 An-drews et al 2014) will likely be somewhat smaller than sug-gested by the numbers above especially in winter when theatmosphere is less well mixed
34 Column-averaged dry air mole XCO2 fractions
As shown in the previous section the choice of vertical al-location of the emissions has a large impact on ground-level concentrations Since the tracers CO2_VERT andCO2_SURF are based on the same mass of CO2 emitted intothe atmosphere and only differ by the vertical placement ofthis mass differences in column-averaged dry air mole frac-tions (ratio of moles of CO2 to moles of dry air in the verti-cal column XCO2) are expected to be small However sincewind speeds tend to increase with altitude CO2 emitted athigher levels is more likely to be transported away from thesources and leave the model domain more rapidly
Maps of XCO2 and of the differences between the twotracers are presented in Figs 9 and 10 in the same way as forthe surface concentrations Instead of presenting mean after-noon values the figures show the situation at 1130 LT (theaverage of output at 1000 and 1100 UTC) corresponding tothe expected overpass time of the planned European satel-lites Note that we did not account for daylight savings timein the diurnal cycles of emissions but assumed a constant off-set of +1 h between local time in the domain and UTC Dif-ferences in XCO2 between the two tracers are indeed muchsmaller than the differences in surface concentrations sug-gesting that total column observations are much less sensi-tive to the vertical placement of emissions However differ-ences are not negligible especially in winter (Fig 10) Thelargest differences in January are seen in the northern parts
of Czech Republic which is the main coal-mining region ofthe country featuring a large number of power plants Sinceplume rise was not explicitly calculated CO2_VERT emis-sions from these power plants followed the EMEP profileswhich tend to place emissions too high as mentioned earlierDifferences over the power plants with explicit plume rise aresmaller but not negligible with values around 1 ppm close tothe sources
Domain-averaged mean diurnal cycles are presented inFig 11 and overall statistics in Table 4 Similar to the resultsfor near-surface concentrations the differences are larger inwinter than in summer In contrast to the situation at the sur-face the differences in XCO2 remain fairly constant overthe day not only in winter but also in summer Relative dif-ferences in mean XCO2 are only 77 in January (com-pared to 43 at the surface) and 48 in July (compared to14 at the surface) Note that the differences in the columnsare related to the synthetic nature of our model experimentsince no anthropogenic CO2 emissions are advecting into thedomain from sources outside Nevertheless the analysis re-veals significant differences in the contributions from sourceswithin the domain which is the information used by any re-gional inverse modeling system Consistent with the findingsfor near-surface concentrations the differences tend to in-crease for higher percentiles reaching about 10 at the 95thpercentile (see last column in Table 4)
Although differences in monthly mean XCO2 values arerelatively small the differences can be very large at a givenlocation and time as illustrated in Fig 12 for 2 July 2015The figure shows the differences in XCO2 between two CO2tracers representing emissions from the largest power plantsin the domain The two tracers were released either usingexplicit plume rise simulations (CO2_PP-PR) or accordingto EMEP SNAP 1 profiles (CO2_PP-EMEP) No power-plant-only tracer was simulated with emissions at the sur-face Plume rise was rather moderate (to about 340 m aboveground) at this time due to pronounced easterly winds Thered (positive) parts in the figure correspond to plumes pro-duced by CO2_PP-PR whereas the blue parts (negative) cor-respond to the tracer CO2_PP-EMEP
Due to wind directions changing with altitude and dueto the different emission heights of the tracers the plumesare transported in different directions Spiraling wind direc-tions are typical of the boundary layer where winds nearthe surface have an ageostrophic cross-isobaric componentdue to surface friction while winds become increasinglygeostrophic at higher levels This is known as the Ekmanspiral The plumes of CO2_PP-EMEP show a stronger lat-eral dispersion because the tracer is released over a large ver-tical extent (blue line in Fig 2) The vertical cross-sectiontransecting the plumes about 15 to 30 km downwind of thesources suggests that both tracers are partially mixed overthe full depth of the boundary layer at this distance but thatthe centers of the plumes are clearly higher above the surfacefor the tracer CO2_PP-EMEP than for CO2_PP-PR
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4552 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 9 Column-averaged dry air mole fractions (XCO2) at 1130 LT in (a b) January and (c d) July contributed by all anthropogenicsources in the domain (a c) Tracer CO2_VERT released according to realistic vertical profiles (b c) Tracer CO2_SURF released only atthe surface
Figure 10 Difference in 1130 LT column-averaged dry air mole fractions (XCO2) between tracers CO2_SURF and CO2_VERT in (a) Jan-uary and (b) July
4 Discussion
The results for near-surface concentrations revealed a strongsensitivity to the vertical placement of emissions especiallyin winter Similarly large sensitivities were reported for airpollution simulations By conducting a set of five 1-year
European-scale model simulations differing only in the ver-tical allocation of emissions Mailler et al (2013) found thatground-based concentrations of SO2 an air pollutant primar-ily released by point sources increased on average by about70 when reducing all emission heights by a factor of 4The changes were less significant (about 15 ) for NO2 due
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4553
Table 4 Statistics of near-surface CO2 and column-averaged dry air mole XCO2 fractions in the model domain for the tracers CO2_VERTand CO2_SURF Differences are presented in terms of (CO2_SURF-CO2_VERT)CO2_VERT
Tracer Month Hour Mean Median 25 75 95 (ppm) (ppm) (ppm) (ppm) (ppm)
Figure 11 Mean diurnal cycles of column-averaged dry air molefractions (XCO2) of the tracers CO2_VERT and CO2_SURF inJanuary and July
to the larger contribution from traffic emissions Sensitivitiesof a similar magnitude were reported by Guevara et al (2014)for simulations over Spain when replacing the EMEP profilesfor power plants by more realistic plume rise calculations
The results of the present study may be considered asupper limits of the sensitivity for two reasons First of allour simulations covered a region in Europe with a par-ticularly high density of coal-fired power plants Simula-tions over other regions such as France where electricityis mostly produced by nuclear power would likely haveyielded lower sensitivities Nevertheless averaged over Eu-rope large point sources are responsible for at least half ofall anthropogenic CO2 emissions (52 in 2011) according tothe TNOMACC-3 inventory which underscores the generalimportance of properly allocating their emissions vertically
Second for most point sources in our domain the stan-dard EMEP profiles were applied which tend to place emis-sions too high in the atmosphere as suggested consistentlyby Bieser et al (2011) Guevara et al (2014) Mailler et al(2013) and by our own comparison with explicit plumerise simulations Several improvements were already imple-mented in the present study each of them leading to a re-duction of the effective emission heights and likely to a morerealistic representation as compared to EMEP This includeda modification of SNAP 9 profiles the distinction betweenpoint and area sources and the explicit computation of plumerise for the largest sources in the domain Due to a lack ofrepresentative studies these modifications were somewhatarbitrary More studies like Bieser et al (2011) and Preg-ger and Friedrich (2009) are needed but should not only tar-get large point sources but also emissions from the remain-ing 48 of emissions including residential heating eventhough their vertical placement will be less critical Explicitplume rise computations for all large point sources as per-formed in some air quality models (Karamchandani et al2014) would be the best alternative to using static profilesbut this adds significant complexity to the model and indi-vidual stack and flue gas parameters are not publicly avail-able in Europe (Pregger and Friedrich 2009)
None of the studies mentioned above considered the issueof interaction between multiple plumes and latent heat re-lease in cooling tower plumes which tend to enhance plumerise The SNAP 1 profiles recommended by Bieser et al(2011) and the simulations conducted here are only represen-tative of isolated stacks which is not consistent with any ofthe German coal-fired power plants in our domain Compre-hensive models for plume rise from single and multiple inter-acting cooling tower plumes have been presented by Schatz-mann and Policastro (1984) and Policastro et al (1994) butthey seem not to be widely applied although the code of
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4554 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Figure 12 Difference on 2 July 2015 1100 LT between power plantCO2 tracer with explicit plume rise (CO2_PP-PR) and releasedaccording to standard EMEP SNAP 1 profile (CO2_PP-EMEP)(a) Map of the difference in column-averaged dry air mole XCO2fractions (b) Vertical cross-section of the difference in CO2 alongthe transect indicated in panel (a) The length of this cross-sectionis about 100 km
Schatzmann and Policastro (1984) is available through theAssociation of German Engineers (VDI httpswwwvdideindexphpid=4791 last access 26 March 2019) Mostplume rise studies date back 20 to 50 years and moderncomputational fluid dynamics simulations are lacking withthe exception of the study of Bornoff and Mokhtarzadeh-Dehghan (2001) on the interaction of plumes from two ad-jacent cooling towers
Mean afternoon differences at the surface in January be-tween the tracers CO2_VERT and CO2_SURF are of the or-der of 15 ppm which is close to 50 of the total anthro-pogenic CO2 signal due to emissions within the model do-main In summer the relative differences in the afternoon aremuch smaller due to more efficient vertical mixing but still aslarge as 14 Inaccurate vertical placement of the emissionsmay thus lead to biases of a magnitude comparable to othererror sources reported in the literature Random uncertain-ties around 30 ndash50 of the regional biospheric CO2 signal
have been estimated for example for model errors in PBLmixing (Gerbig et al 2008) and in wind speed and direc-tion (Lin and Gerbig 2005) Systematic biases in simulatednear-surface wind speeds of mesoscale weather predictionmodels like COSMO or WRF are typically on the order of05ndash1 m sminus1 or about 10 ndash20 of the mean wind speeds(Jimeacutenez and Dudhia 2012 Brunner et al 2015 Bagleyet al 2018) which may translate into similar biases in near-surface concentrations Systematic differences in emissionestimates using different regional transport and inverse mod-eling systems were reported to be around 20 ndash40 (Huet al 2014 Bergamaschi et al 2018b) Our analysis fo-cused on the situation in the lowest model layer at 0ndash20 mHowever the recommended strategy for CO2 monitoring isto sample from tall towers well above the surface At higherelevations the model sensitivity to the vertical placement ofemissions is likely smaller which is another benefit of talltower measurements in addition to their greater spatial rep-resentativeness
Mean differences in total column XCO2 between the trac-ers CO2_VERT and CO2_SURF are much smaller around8 in winter and 5 in summer However differences arelarger at the highest percentiles (about 10 at the 95 per-centile) which are more relevant for satellite missions likeSentinel-CO2 designed to image individual plumes Further-more the vertical placement of emissions has a significant ef-fect on the speed and direction of individual plumes suggest-ing that an accurate vertical placement is a critical require-ment for inverse modeling of power plant emissions fromsatellite observations Irrespective of a correct vertical place-ment of emissions an appropriate simulation of power plantplumes will remain a great challenge for any mesoscale at-mospheric transport model Current approaches for estimat-ing power plant emissions are circumventing this problem bydirectly matching the ldquoobservedrdquo wind direction defined bythe location of the plume eg by fitting a Gaussian plumemodel (Krings et al 2018 Nassar et al 2017) Howeverthese methods also need to make a realistic assumption aboutthe height of the plume since the mean advection speed ofthe plume needs to be estimated from a simulated or observedvertical wind profile
5 Conclusions
We investigated the sensitivity of model-simulated near-surface and total column CO2 concentrations to a realisticvertical allocation of anthropogenic emissions as opposed tothe traditional approach of emitting CO2 only at the surfaceThe study was conducted using kilometer-scale atmospherictransport simulations for the year 2015 for a domain cov-ering the city of Berlin and numerous power plants in Ger-many Poland and Czech Republic More than 50 of CO2in Europe is emitted from large point sources mostly throughstacks and cooling towers suggesting that a proper represen-
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4555
tation of these sources in the vertical dimension may be crit-ical Our results indeed confirm a strong sensitivity of near-surface afternoon concentrations a regional CO2 tracer re-leased in the model domain only at the surface was on aver-age 43 higher in winter and 14 higher in summer thana tracer released according to realistic vertical profiles Sincemeasurements of CO2 are often taken from towers some 100to 300 m above the surface (Bakwin et al 1995) the impacton actual ground-based observations will likely be somewhatsmaller
Differences were smaller but not negligible (5 ndash8 ) fortotal column XCO2 suggesting that the assimilation of satel-lite observations is less sensitive to the vertical placementof emissions than the assimilation of ground-based observa-tions Individual plumes as imaged by future CO2 satelliteshowever may propagate more rapidly and in different direc-tions when using realistic vertical profiles instead of releas-ing CO2 only at the surface
Plume rise was explicitly simulated for the six largestpower plants in the domain and for the 22 largest pointsources in Berlin Power plant plumes typically rose be-tween 100 and 400 m above the top of the stacks Plumerise showed significant seasonal and diurnal variability thatwould be missed when applying static vertical profiles Sim-ulated plume rise was on average more than 100 m lower thansuggested by the frequently used EMEP profile for powerplant emissions An accurate vertical placement of emissionsis not only critical for power plants but may also be relevantfor the simulation of city plumes In the case of Berlin forexample more than 35 of CO2 is released through stackspresumably more than 90 m above surface
We strongly recommend the representation of CO2 emis-sions in all three dimensions in regional atmospheric trans-port and inverse modeling studies The specific impact on themodel results will depend on factors such as vertical modelresolution and boundary layer scheme but in general cannotbe expected to be negligible Current gridded emission in-ventories only provide information in two dimensions Infor-mation on the source-specific vertical allocation of emissionsas used in this study conversely is still sparse and should re-ceive more attention in the future
Data availability Column-averaged dry air mole fractions ofall simulated tracers are available both as 2-D fields andas synthetic level 2 satellite products through ESA The to-tal three-dimensional model output amounts to 75 TB and isarchived at Empa servers Selected fields or time periods canbe made available upon request The TNOMACC-3 inven-tory is available through Copernicus (httpdrdsijrceceuropaeudatasettno-macc-iii-european-anthropogenic-emissions last ac-cess 26 March 2019) The emissions inventory of Berlin was kindlyprovided by Andreas Kerschbaumer (Senatsverwaltung Berlin) andis available for research upon request The global CO2 simulationwhich provided the lateral boundary conditions was conducted inthe framework of Copernicus Atmosphere Monitoring Service and
can be retrieved from ECMWFrsquos archive as experiment gf39 streamlwda class rd
Author contributions DB led the project SMARTCARB devel-oped the idea of the present study and wrote the manuscript withinput from all co-authors GK designed the model experimentsconducted all simulations and contributed several figures JM con-tributed the VPRM flux data required for the simulations VC portedthe COSMO-GHG extension to GPUs OF surveyed and supportedthe porting to GPUs and the setup of the model on the supercom-puter at CSCS GB followed the project as external advisor and con-tributed critical input to an earlier version of the manuscript YMand AL accompanied the study as ESA project and technical offi-cers respectively and provided critical input and reviews during allphases of the project
Competing interests The authors declare that they have no conflictof interest
Acknowledgements This study was conducted in the context ofthe project SMARTCARB funded by the European Space Agency(ESA) under contract no 400011959916NLFFmg The views ex-pressed here can in no way be taken to reflect the official opinionof ESA The work was supported by a grant from the Swiss Na-tional Supercomputing Centre (CSCS) under project ID d73 Weacknowledge CSCS the Swiss Federal Office for Meteorology andClimatology (MeteoSwiss) and ETH Zurich for their contributionsto the development of the GPU-accelerated version of COSMO Wewould like to thank Richard Engelen and Anna Agusti-Panareda(ECMWF) for providing support and access to global CO2 modelsimulation fields which were generated using the Copernicus Atmo-sphere Monitoring Service Information (2017) The TNOMACC-3 emissions inventory and temporal emission profiles were kindlyprovided by Hugo Denier van der Gon (TNO the Netherlands) Fi-nally we are very grateful to Andreas Kerschbaumer (Senatsver-waltung Berlin) for providing the emission inventory of Berlin andadditional material and for being available for discussions on itsproper usage
Review statement This paper was edited by Michael Pitts and re-viewed by three anonymous referees
References
Achtemeier G L Goodrick S A Liu Y Garcia-MenendezF Hu Y and Odman M T Modeling Smoke Plume-Rise and Dispersion from Southern United States Pre-scribed Burns with Daysmoke Atmosphere 2 358ndash388httpsdoiorg103390atmos2030358 2011
Agustiacute-Panareda A Massart S Chevallier F Boussetta S Bal-samo G Beljaars A Ciais P Deutscher N M EngelenR Jones L Kivi R Paris J-D Peuch V-H Sherlock VVermeulen A T Wennberg P O and Wunch D Forecast-
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4556 D Brunner et al Atmospheric CO2 simulations with vertical emissions
ing global atmospheric CO2 Atmos Chem Phys 14 11959ndash11983 httpsdoiorg105194acp-14-11959-2014 2014
Amt fuumlr Statistik Berlin-Brandenburg Statistischer BerichtEnergie- und CO2-Bilanz in Berlin 2015 Tech rep PotsdamGermany available at httpwwwstatistik-berlin-brandenburgde (last access 26 March 2019) 2018
Andrews A E Kofler J D Trudeau M E Williams J C NeffD H Masarie K A Chao D Y Kitzis D R Novelli P CZhao C L Dlugokencky E J Lang P M Crotwell M JFischer M L Parker M J Lee J T Baumann D D DesaiA R Stanier C O De Wekker S F J Wolfe D E MungerJ W and Tans P P CO2 CO and CH4 measurements from talltowers in the NOAA Earth System Research Laboratoryrsquos GlobalGreenhouse Gas Reference Network instrumentation uncer-tainty analysis and recommendations for future high-accuracygreenhouse gas monitoring efforts Atmos Meas Tech 7 647ndash687 httpsdoiorg105194amt-7-647-2014 2014
Bagley J E Jeong S Cui X Newman S Zhang J PriestC Campos-Pineda M Andrews A E Bianco L Lloyd MLareau N Clements C and Fischer M L Assessment of anatmospheric transport model for annual inverse estimates of Cal-ifornia greenhouse gas emissions J Geophys Res-Atmos 1221901ndash1918 httpsdoiorg1010022016JD025361 2018
Bakwin P S Tans P P Zhao C Ussler W and Quesnell EMeasurements of carbon dioxide on a very tall tower Tellus B47 535ndash549 1995
Baldauf M Seifert A Foumlrstner J Majewski D Raschendor-fer M and Reinhardt T Operational Convective-Scale Nu-merical Weather Prediction with the COSMO Model Descrip-tion and Sensitivities Mon Weather Rev 139 3887ndash3905httpsdoiorg101175MWR-D-10-050131 2011
Bergamaschi P Danila A Weiss R Ciais P Thompson RBrunner D Levin I Meijer Y Chevallier F Janssens-Maenhout G Bovensmann H Crisp D Basu S Dlugo-kencky E Engelen R Gerbig C Guumlnther D Hammer SHenne S Houweling S Karstens U Kort E Maione MManning A Miller J Montzka S Pandey S Peters WPeylin P Pinty B Ramonet M Reimann S Roumlckmann TSchmidt M Strogies M Sussams J Tarasova O van Aar-denne J Vermeulen A and Vogel F Atmospheric monitor-ing and inverse modelling for verification of greenhouse gas in-ventories no EUR 29276 EN in JRC Science for policy re-port Publications Office of the European Union Luxembourghttpsdoiorg102760759928 2018a
Bergamaschi P Karstens U Manning A J Saunois M Tsu-ruta A Berchet A Vermeulen A T Arnold T Janssens-Maenhout G Hammer S Levin I Schmidt M RamonetM Lopez M Lavric J Aalto T Chen H Feist D G Ger-big C Haszpra L Hermansen O Manca G Moncrieff JMeinhardt F Necki J Galkowski M OrsquoDoherty S Para-monova N Scheeren H A Steinbacher M and Dlugo-kencky E Inverse modelling of European CH4 emissions dur-ing 2006ndash2012 using different inverse models and reassessedatmospheric observations Atmos Chem Phys 18 901ndash920httpsdoiorg105194acp-18-901-2018 2018b
Bieser J Aulinger A Matthias V Quante M and van derGon H D Vertical emission profiles for Europe based onplume rise calculations Environ Pollut 159 2935ndash2946httpsdoiorg101016jenvpol201104030 2011
Bornoff R and Mokhtarzadeh-Dehghan M A numerical studyof interacting buoyant cooling-tower plumes Atmos Environ35 589ndash598 httpsdoiorg101016S1352-2310(00)00296-X2001
Bousquet P Peylin P Ciais P Le Queacutereacute C Friedlingstein Pand Tans P P Regional changes in carbon dioxide fluxes of landand oceans since 1980 Science 290 1342ndash1347 2000
Bovensmann H Buchwitz M Burrows J P Reuter M KringsT Gerilowski K Schneising O Heymann J Tretner A andErzinger J A remote sensing technique for global monitoring ofpower plant CO2 emissions from space and related applicationsAtmos Meas Tech 3 781ndash811 httpsdoiorg105194amt-3-781-2010 2010
Briggs G A Plume rise and buoyancy effects atmo-spheric sciences and power production vol DOETIC-27601 (DE84005177) p 850 TN Technical Informa-tion Center US Dept of Energy Oak Ridge USAhttpsdoiorg1021726503687 1984
Broquet G Chevallier F Rayner P Aulagnier C Pison I Ra-monet M Schmidt M Vermeulen A T and Ciais P A Euro-pean summertime CO2 biogenic flux inversion at mesoscale fromcontinuous in situ mixing ratio measurements J Geophys Res-Atmos 116 D23303 httpsdoiorg1010292011JD0162022011
Broquet G Breacuteon F-M Renault E Buchwitz M ReuterM Bovensmann H Chevallier F Wu L and Ciais PThe potential of satellite spectro-imagery for monitoring CO2emissions from large cities Atmos Meas Tech 11 681ndash708httpsdoiorg105194amt-11-681-2018 2018
Brunner D Savage N Jorba O Eder B Giordano L BadiaA Balzarini A Baroacute R Bianconi R Chemel C Curci GForkel R Jimeacutenez-Guerrero P Hirtl M Hodzic A Hon-zak L Im U Knote C Makar P Manders-Groot A vanMeijgaard E Neal L Peacuterez J L Pirovano G Jose R SSchoumlder W Sokhi R S Syrakov D Torian A TuccellaP Werhahn J Wolke R Yahya K Zabkar R Zhang YHogrefe C and Galmarini S Comparative analysis of mete-orological performance of coupled chemistry-meteorology mod-els in the context of AQMEII phase 2 Atmos Environ 115470ndash498 httpsdoiorg101016jatmosenv201412032 2015
Builtjes P van Loon M Schaap M Teeuwisse S VisschedijkA and Bloos J Project on the modelling and verificationof ozone reduction strategies contribution of TNO-MEP TechRep TNO-report MEP-R2003166 TNO Netherlands Organisa-tion for applied scientific research Apeldoorn the Netherlands2003
Busch D Harte R Kraumltzig W B and Montag U New naturaldraft cooling tower of 200 m of height Eng Struct 24 1509ndash1521 httpsdoiorg101016S0141-0296(02)00082-2 2002
Chan D Ishizawa M Higuchi K Maksyutov S and ChenJ Seasonal CO2 rectifier effect and large-scale extratropicalatmospheric transport J Geophys Res-Atmos 113 D17309httpsdoiorg1010292007JD009443 2008
Chevallier F Ciais P Conway T J Aalto T Anderson B EBousquet P Brunke E G Ciattaglia L Esaki Y Froumlh-lich M Gomez A Gomez-Pelaez A J Haszpra L Krum-mel P B Langenfelds R L Leuenberger M MachidaT Maignan F Matsueda H Morguiacute J A Mukai HNakazawa T Peylin P Ramonet M Rivier L Sawa Y
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4557
Schmidt M Steele L P Vay S A Vermeulen A TWofsy S and Worthy D CO2 surface fluxes at grid pointscale estimated from a global 21 year reanalysis of at-mospheric measurements J Geophys Res 115 D21307httpsdoiorg1010292010JD013887 2010
Ciais P Dolman A J Bombelli A Duren R Peregon ARayner P J Miller C Gobron N Kinderman G Mar-land G Gruber N Chevallier F Andres R J BalsamoG Bopp L Breacuteon F-M Broquet G Dargaville R Bat-tin T J Borges A Bovensmann H Buchwitz M ButlerJ Canadell J G Cook R B DeFries R Engelen R Gur-ney K R Heinze C Heimann M Held A Henry M LawB Luyssaert S Miller J Moriyama T Moulin C My-neni R B Nussli C Obersteiner M Ojima D Pan YParis J-D Piao S L Poulter B Plummer S Quegan SRaymond P Reichstein M Rivier L Sabine C SchimelD Tarasova O Valentini R Wang R van der Werf GWickland D Williams M and Zehner C Current system-atic carbon-cycle observations and the need for implementinga policy-relevant carbon observing system Biogeosciences 113547ndash3602 httpsdoiorg105194bg-11-3547-2014 2014
Ciais P Denier Van der Gon H Engelen R Heimann MJanssens-Maenhout G Rayner P and Scholze M Towards aEuropean Operational Observing System to Monitor Fossil CO2emissions Tech rep European Commission B-1049 Brusselshttpsdoiorg102788350433 2015
Davin E L Stoumlckli R Jaeger E B Levis S and Senevi-ratne S I COSMO-CLM2 a new version of the COSMO-CLM model coupled to the Community Land Model Clim Dy-nam 37 1889ndash1907 httpsdoiorg101007s00382-011-1019-z 2011
European Environment Agency EMEPCORINAIR Emis-sion Inventory Guidebook ndash 3rd edition 2001 TechRep 302001 Copenhagen Denmark available at httpswwweeaeuropaeupublicationstechnical_report_2001_3(last access 26 March 2019) 2002
Fischer M L Parazoo N Brophy K Cui X Jeong S LiuJ Keeling R Taylor T E Gurney K Oda T and GravenH Simulating estimation of California fossil fuel and biospherecarbon dioxide exchanges combining in situ tower and satellitecolumn observations J Geophys Res-Atmos 122 3653ndash3671httpsdoiorg1010022016JD025617 2018
Fuhrer O Osuna C Lapillonne X Gysi T Cumming BBianco M Arteaga A and Schulthess T Towards a perfor-mance portable architecture agnostic implementation strategyfor weather and climate models Supercomputing Frontiers andInnovations 1 44ndash61 available at httpsuperfriorgsuperfriarticleview17 (last access 26 March 2019) 2014
Ganshin A Oda T Saito M Maksyutov S Valsala V AndresR J Fisher R E Lowry D Lukyanov A Matsueda H Nis-bet E G Rigby M Sawa Y Toumi R Tsuboi K VarlaginA and Zhuravlev R A global coupled Eulerian-Lagrangianmodel and 1times1 km CO2 surface flux dataset for high-resolutionatmospheric CO2 transport simulations Geosci Model Dev 5231ndash243 httpsdoiorg105194gmd-5-231-2012 2012
Gately C K and Hutyra L R Large Uncertainties in Urban-ScaleCarbon Emissions J Geophys Res-Atmos 122 11242ndash11260httpsdoiorg1010022017JD027359 2017
Gerbig C Koumlrner S and Lin J C Vertical mixingin atmospheric tracer transport models error characteriza-tion and propagation Atmos Chem Phys 8 591ndash602httpsdoiorg105194acp-8-591-2008 2008
Goeckede M Michalak A M Vickers D Turner D P and LawB E Atmospheric inverse modeling to constrain regional-scaleCO2 budgets at high spatial and temporal resolution J GeophysRes 115 1ndash23 httpsdoiorg1010292009JD012257 2010
Graven H Fischer M L Lueker T Jeong S GuildersonT P Keeling R F Bambha R Brophy K Callahan WCui X Frankenberg C Gurney K R LaFranchi B WLehman S J Michelsen H Miller J B Newman SPaplawsky W Parazoo N C Sloop C and Walker S JAssessing fossil fuel CO2 emissions in California using at-mospheric observations and models Environ Res Lett 13065007 httpsdoiorg1010881748-9326aabd43 2018
Guevara M Soret A Arevalo G Martinez F and BaldasanoJ M Implementation of plume rise and its impacts on emis-sions and air quality modelling Atmos Environ 99 618ndash629httpsdoiorg101016jatmosenv201410029 2014
Hanna S R Rise and Condensation of Large Cooling TowerPlumes J Appl Meteorol 11 793ndash799 1972
Hase F Frey M Blumenstock T Groszlig J Kiel M KohlheppR Mengistu Tsidu G Schaumlfer K Sha M K and Orphal JApplication of portable FTIR spectrometers for detecting green-house gas emissions of the major city Berlin Atmos MeasTech 8 3059ndash3068 httpsdoiorg105194amt-8-3059-20152015
Hogue S Marland E Andres R J Marland G and WoodardD Uncertainty in gridded CO2 emissions estimates Earthrsquos Fu-ture 4 225ndash239 httpsdoiorg1010022015EF000343 2016
Houyoux M Vukovich J Seppanen C and Brandmeyer J ESMOKE User Manual Tech rep MCNC Environmental Mod-eling Center College Park MD USA 2002
Hu L Montzka S A Miller J B Andrews A E LehmanS J Miller B R Thoning K Sweeney C Chen HGodwin D S Masarie K Bruhwiler L Fischer M LBiraud S C Torn M S Mountain M Nehrkorn TEluszkiewicz J Miller S Draxler R R Stein A F HallB D Elkins J W and Tans P P US emissions ofHFC-134a derived for 2008ndash2012 from an extensive flask-airsampling network J Geophys Res-Atmos 120 801ndash825httpsdoiorg1010022014JD022617 2014
Jimeacutenez P A and Dudhia J Improving the Representation of Re-solved and Unresolved Topographic Effects on Surface Windin the WRF Model J Appl Meteorol Clim 51 300ndash316httpsdoiorg101175JAMC-D-11-0841 2012
Jung M Henkel K Herold M and Churkina G Ex-ploiting synergies of global land cover products for car-bon cycle modeling Remote Sens Environ 101 534ndash553httpsdoiorg101016jrse200601020 2006
Karamchandani P Johnson J Yarwood G and Knipping EImplementation and application of sub-grid-scale plume treat-ment in the latest version of EPArsquos third-generation air qual-ity model CMAQ 501 J Air Waste Manage 64 453ndash467httpsdoiorg101080109622472013855152 2014
Kent J Whitehead J P Jablonowski C and Rood R BDetermining the effective resolution of advection schemes
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
4558 D Brunner et al Atmospheric CO2 simulations with vertical emissions
Part I Dispersion analysis J Comput Phys 278 485ndash496httpsdoiorg101016jjcp201401043 2014
Kountouris P Gerbig C Roumldenbeck C Karstens U Koch T Fand Heimann M Technical Note Atmospheric CO2 inversionson the mesoscale using data-driven prior uncertainties method-ology and system evaluation Atmos Chem Phys 18 3027ndash3045 httpsdoiorg105194acp-18-3027-2018 2018
Kretschmer R Gerbig C Karstens U and Koch F-T Errorcharacterization of CO2 vertical mixing in the atmospheric trans-port model WRF-VPRM Atmos Chem Phys 12 2441ndash2458httpsdoiorg105194acp-12-2441-2012 2012
Kretschmer R Gerbig C Karstens U Biavati G Ver-meulen A Vogel F Hammer S and Totsche K U Im-pact of optimized mixing heights on simulated regional atmo-spheric transport of CO2 Atmos Chem Phys 14 7149ndash7172httpsdoiorg105194acp-14-7149-2014 2014
Krings T Neininger B Gerilowski K Krautwurst S Buch-witz M Burrows J P Lindemann C Ruhtz T Schuumlt-temeyer D and Bovensmann H Airborne remote sensingand in situ measurements of atmospheric CO2 to quantifypoint source emissions Atmos Meas Tech 11 721ndash739httpsdoiorg105194amt-11-721-2018 2018
Kuenen J J P Visschedijk A J H Jozwicka M and De-nier van der Gon H A C TNO-MACC_II emission inven-tory a multi-year (2003ndash2009) consistent high-resolution Euro-pean emission inventory for air quality modelling Atmos ChemPhys 14 10963ndash10976 httpsdoiorg105194acp-14-10963-2014 2014
Kuhlmann G Cleacutement V Marschall J Fuhrer O BroquetG Schnadt-Poberaj C Loumlscher A Meijer Y and Brun-ner D SMARTCARB ndash Use of Satellite Measurements ofAuxiliary Reactive Trace Gases for Fossil Fuel Carbon Diox-ide Emission Estimation Final report of ESA study contractn400011959916NLFFmg Tech rep Empa Swiss FederalLaboratories for Materials Science and Technology Duumlben-dorf Switzerland available at httpswwwempachdocuments56101617885FR_Smartcarb_final_Jan2019pdf (last access26 March 2019) 2019
Lapillonne X and Fuhrer O Using Compiler Directives to PortLarge Scientific Applications to GPUs An Example from At-mospheric Science Parallel Processing Letters 24 1450003httpsdoiorg101142S0129626414500030 2014
Lauvaux T Pannekoucke O Sarrat C Chevallier F Ciais PNoilhan J and Rayner P J Structure of the transport uncer-tainty in mesoscale inversions of CO2 sources and sinks us-ing ensemble model simulations Biogeosciences 6 1089ndash1102httpsdoiorg105194bg-6-1089-2009 2009
Lauvaux T Miles N L Deng A Richardson S J Cambal-iza M O Davis K J Gaudet B Gurney K R Huang JOrsquoKeefe D Song Y Karion A Oda T Patarasuk R Razli-vanov I Sarmiento D Shepson P Sweeney C Turnbull Jand Wu K High-resolution atmospheric inversion of urban CO2emissions during the dormant season of the Indianapolis FluxExperiment (INFLUX) J Geophys Res-Atmos 121 5213ndash5236 httpsdoiorg1010022015JD024473 2016
Leutwyler D Luumlthi D Ban N Fuhrer O and Schaumlr C Eval-uation of the convection-resolving climate modeling approachon continental scales J Geophys Res-Atmos 122 5237ndash5258httpsdoiorg1010022016JD026013 2017
Lin J C and Gerbig C Accounting for the effect of transporterrors on tracer inversions Geophys Res Lett 32 L01802httpsdoiorg1010292004GL021127 2005
Lin J C Gerbig C Wofsy S C Andrews A E DaubeB C Davis K J and Grainger C A A near-field toolfor simulating the upstream influence of atmospheric ob-servations The Stochastic Time-Inverted Lagrangian Trans-port (STILT) model J Geophys Res-Atmos 108 4493httpsdoiorg1010292002JD003161 2003
Liu Y Gruber N and Brunner D Spatiotemporal patternsof the fossil-fuel CO2 signal in central Europe results froma high-resolution atmospheric transport model Atmos ChemPhys 17 14145ndash14169 httpsdoiorg105194acp-17-14145-2017 2017
Mahadevan P Wofsy S Matross D M Xiao X Dunn A LLin J C Gerbig C Munger J W Chow V Y and Got-tlieb E W A satellite-based biosphere parameterization for netecosystem CO2 exchange Vegetation Photosynthesis and Res-piration Model (VPRM) Global Biogeochem Cy 22 1ndash17httpsdoiorg1010292006GB002735 2008
Mailler S Khvorostyanov D and Menut L Impact of thevertical emission profiles on background gas-phase pollutionsimulated from the EMEP emissions over Europe AtmosChem Phys 13 5987ndash5998 httpsdoiorg105194acp-13-5987-2013 2013
Meesters A G C A Tolk L F Peters W Hutjes R W A Vel-linga O S Elbers J a Vermeulen A T van der Laan S Neu-bert R E M Meijer H A J and Dolman A J Inverse car-bon dioxide flux estimates for the Netherlands J Geophys Res-Atmos 117 D20306 httpsdoiorg1010292012JD0177972012
Nassar R Hill T G McLinden C A Wunch D Jones DB A and Crisp D Quantifying CO2 Emissions From Individ-ual Power Plants From Space Geophys Res Lett 44 10045ndash10053 httpsdoiorg1010022017GL074702 2017
Nisbet E and Weiss R Top-down versus bottom-up Science 3281241ndash1243 httpsdoiorg101126science1189936 2010
Oney B Henne S Gruber N Leuenberger M Bam-berger I Eugster W and Brunner D The CarboCountCH sites characterization of a dense greenhouse gas ob-servation network Atmos Chem Phys 15 11147ndash11164httpsdoiorg105194acp-15-11147-2015 2015
Peters W Krol M C van der Werf G R Houweling SJones C D Hughes J Schaefer K Masarie K A Jacob-son A R Miller J B Cho C H Ramonet M SchmidtM Ciattaglia L Apadula F Heltai D Meinhardt F DiSarra A G Piacentino S Sferlazzo D Aalto T HatakkaJ Stroumlm J Haszpra L Meijer H A J Van Der Laan SNeubert R E M Jordan A Rodoacute X Morguiacute J-A Ver-meulen A T Popa E Rozanski K Zimnoch M Man-ning A C Leuenberger M Uglietti C Dolman A J CiaisP Heimann M and Tans P P Seven years of recent Eu-ropean net terrestrial carbon dioxide exchange constrained byatmospheric observations Glob Change Biol 16 1317ndash1337httpsdoiorg101111j1365-2486200902078x 2010
Peylin P Law R M Gurney K R Chevallier F JacobsonA R Maki T Niwa Y Patra P K Peters W Rayner PJ Roumldenbeck C van der Laan-Luijkx I T and Zhang XGlobal atmospheric carbon budget results from an ensemble of
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
D Brunner et al Atmospheric CO2 simulations with vertical emissions 4559
atmospheric CO2 inversions Biogeosciences 10 6699ndash6720httpsdoiorg105194bg-10-6699-2013 2013
Pillai D Buchwitz M Gerbig C Koch T Reuter M Bovens-mann H Marshall J and Burrows J P Tracking city CO2emissions from space using a high-resolution inverse mod-elling approach a case study for Berlin Germany AtmosChem Phys 16 9591ndash9610 httpsdoiorg105194acp-16-9591-2016 2016
Pinty B Janssens-Maenhout G Dowell M Zunker H Brun-hes T Ciais P Dee D Denier van der Gon H Dol-man H Drinkwater M Engelen R Heimann M Holm-lund K Husband R Kentarchos A Meijer Y Palmer Pand Scholze M An Operational Anthropogenic CO2 EmissionsMonitoring amp Verification Support capacity ndash Baseline Require-ments Model Component and Functional Architecture TechRep EUR 28736 EN European Commission B-1049 Brusselshttpsdoiorg10276008644 2017
Policastro A Dunn W and Carhart R A model for seasonaland annual cooling tower impacts Atmos Environ 28 379ndash395 httpsdoiorg1010161352-2310(94)90118-X 1994
Pregger T and Friedrich R Effective pollutant emissionheights for atmospheric transport modelling based onreal-world information Environ Pollut 157 552ndash560httpsdoiorg101016jenvpol200809027 2009
Roches A and Fuhrer O Tracer module in the COSMOmodel Tech Rep 20 Consortium for Small-Scale Mod-elling (COSMO) Center for Climate Systems Mod-elling (C2SM) and MeteoSwiss Switzerland available athttpwwwcosmo-modelorgcontentmodeldocumentationtechReportsdefaulthtm (last access 26 March 2019) 2012
Rockel B Will A and Hense A The Regional ClimateModel COSMO-CLM (CCLM) Meteorol Z 17 347ndash348httpsdoiorg1011270941-294820080309 2008
Roumldenbeck C Zaehle S Keeling R and Heimann MHow does the terrestrial carbon exchange respond to inter-annual climatic variations A quantification based onatmospheric CO2 data Biogeosciences 15 2481ndash2498httpsdoiorg105194bg-15-2481-2018 2018
Sarrat C Noilhan J Lacarreacutere P Ceschia E Ciais P Dol-man A J Elbers J A Gerbig C Gioli B Lauvaux TMiglietta F Neininger B Ramonet M Vellinga O andBonnefond J M Mesoscale modelling of the CO2 interac-tions between the surface and the atmosphere applied to theApril 2007 CERES field experiment Biogeosciences 6 633ndash646 httpsdoiorg105194bg-6-633-2009 2009
Schatzmann M and Policastro A J An advanced integral modelfor cooling tower plume dispersion Atmos Environ 18 663ndash674 httpsdoiorg1010160004-6981(84)90253-1 1984
Stohl A Forster C Frank A Seibert P and Wotawa GTechnical note The Lagrangian particle dispersion modelFLEXPART version 62 Atmos Chem Phys 5 2461ndash2474httpsdoiorg105194acp-5-2461-2005 2005
Timmermans R van der Gon H D Kuenen J SegersA Honoreacute C Perrussel O Builtjes P and SchaapM Quantification of the urban air pollution increment andits dependency on the use of down-scaled and bottom-up city emission inventories Urban Climate 6 44ndash62httpsdoiorg101016juclim201310004 2013
Uebel M and Bott A Influence of complex terrain and an-thropogenic emissions on atmospheric CO2 patterns ndash a high-resolution numerical analysis Q J Roy Meteor Soc 144 34ndash47 httpsdoiorg101002qj3182 2018
United Nations Framework Convention on Climate ChangeReport of the Conference of the Parties on its twenty-firstsession held in Paris from 30 November to 13 Decem-ber 2015 Addendum Part two Action taken by the Con-ference of the Parties at its twenty-first session Tech RepFCCCCP201510Add1 United Nations Geneva Switzer-land available at httpunfcccintdocumentationdocumentsadvanced_searchitems6911phppriref=600008865 (lastaccess 26 March 2019) 2016
van der Laan-Luijkx I T van der Velde I R van der VeenE Tsuruta A Stanislawska K Babenhauserheide A ZhangH F Liu Y He W Chen H Masarie K A KrolM C and Peters W The CarbonTracker Data Assimila-tion Shell (CTDAS) v10 implementation and global car-bon balance 2001ndash2015 Geosci Model Dev 10 2785ndash2800httpsdoiorg105194gmd-10-2785-2017 2017
VDI ndash Fachbereich Umweltmeteorologie Dispersion of airpollutants in the atmosphere determination of plume riseTech Rep VDI 3782 Blatt 3 VDIDIN-Kommission Rein-haltung der Luft (KRdL) ndash Normenausschuss available athttpwwwvdieuncguidelinesvdi_3782_blatt_3-ausbreitung_von_luftverunreinigungen_in_der_atmosphaere_berechnung_der_abgasfahnenueberh (last access 26 March 2019)Richtlinie technische Regel can be ordered from BeuthPublishing Germany 1985
Vogel B Vogel H Baumlumer D Bangert M Lundgren K RinkeR and Stanelle T The comprehensive model system COSMO-ART ndash Radiative impact of aerosol on the state of the atmo-sphere on the regional scale Atmos Chem Phys 9 8661ndash8680httpsdoiorg105194acp-9-8661-2009 2009
Zubler E M Folini D Lohmann U Luumlthi D Muhlbauer APousse-Nottelmann S Schaumlr C and Wild M Implementa-tion and evaluation of aerosol and cloud microphysics in a re-gional climate model J Geophys Res-Atmos 116 D02211httpsdoiorg1010292010JD014572 2011
