Atmos. Chem. Phys., 19, 7347–7376, 2019 https://doi.org/10.5194/acp-19-7347-2019 © Author(s) 2019. This work is distributed under the Creative Commons Attribution 4.0 License. Modelling CO2 weather – why horizontal resolution matters Anna Agustí-Panareda1 , Michail Diamantakis1 , Sébastien Massart1 , Frédéric Chevallier2 , Joaquín Muñoz-Sabater1 , Jérôme Barré1 , Roger Curcoll3 , Richard Engelen1 , Bavo Langerock4 , Rachel M. Law5 , Zoë Loh5 , Josep Anton Morguí3 , Mark Parrington1 , Vincent-Henri Peuch1 , Michel Ramonet2 , Coleen Roehl6 , Alex T. Vermeulen7 , Thorsten Warneke8 , and Debra Wunch9 1 European Centre for Medium-Range Weather Forecasts, Shinfield Park, Reading, RG2 9AX, UK 2 Laboratoire des Sciences du Climat et de l’Environnement, LSCE/IPSL, CEA-CNRS-UVSQ, Université Paris-Saclay, 91198 Gif-sur-Yvette, France 3 Environmental Science and Technology Institute, Universitat Autònoma de Barcelona, ICTA-UAB, Bellaterra, Spain 4 Royal Belgian Institute for Space Aeronomy (BIRA-IASB), Brussels, Belgium 5 CSIRO Oceans and Atmosphere, PMB 1, Aspendale, Victoria 3195, Australia 6 California Institute of Technology, Pasadena, California, USA 7 ICOS ERIC Carbon Portal, Sölvegatan 12, 22362 Lund, Sweden 8 University of Bremen, Institute of Environmental Physics, Otto-Hahn-Allee 1, 28359 Bremen, Germany 9 University of Toronto, Department of Physics, Toronto, Ontario, Canada Correspondence: Anna Agustí-Panareda (anna.agusti-panareda@ecmwf.int) Received: 20 February 2019 – Discussion started: 27 February 2019 Revised: 7 May 2019 – Accepted: 8 May 2019 – Published: 4 June 2019 Abstract. Climate change mitigation efforts require informa- ulated wind and CO2 . Mountain stations display the largest tion on the current greenhouse gas atmospheric concentra- improvements at high resolution as they directly benefit from tions and their sources and sinks. Carbon dioxide (CO2 ) is the more realistic orography. In addition, the CO2 spatial gra- the most abundant anthropogenic greenhouse gas. Its vari- dients are generally improved with increasing resolution for ability in the atmosphere is modulated by the synergy be- both stations near the surface and those observing the to- tween weather and CO2 surface fluxes, often referred to as tal column, as the overall inter-station error is also reduced CO2 weather. It is interpreted with the help of global or re- in magnitude. However, close to emission hotspots, the high gional numerical transport models, with horizontal resolu- resolution can also lead to a deterioration of the simulation tions ranging from a few hundreds of kilometres to a few skill, highlighting uncertainties in the high-resolution fluxes kilometres. Changes in the model horizontal resolution affect that are more diffuse at lower resolutions. not only atmospheric transport but also the representation We conclude that increasing horizontal resolution matters of topography and surface CO2 fluxes. This paper assesses for modelling CO2 weather because it has the potential to the impact of horizontal resolution on the simulated atmo- bring together improvements in the surface representation of spheric CO2 variability with a numerical weather prediction both winds and CO2 fluxes, as well as an expected reduction model. The simulations are performed using the Copernicus in numerical errors of transport. Modelling applications like Atmosphere Monitoring Service (CAMS) CO2 forecasting atmospheric inversion systems to estimate surface fluxes will system at different resolutions from 9 to 80 km and are eval- only be able to benefit fully from upgrades in horizontal res- uated using in situ atmospheric surface measurements and olution if the topography, winds and prior flux distribution atmospheric column-mean observations of CO2 , as well as are also upgraded accordingly. It is clear from the results that radiosonde and SYNOP observations of the winds. an additional increase in resolution might reduce errors even The results indicate that both diurnal and day-to-day vari- further. However, the horizontal resolution sensitivity tests ability of atmospheric CO2 are generally better represented at indicate that the change in the CO2 and wind modelling error high resolution, as shown by a reduction in the errors in sim- Published by Copernicus Publications on behalf of the European Geosciences Union. 7348 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather with resolution is not linear, making it difficult to quantify in order to improve the forecast skill. But until now, global the improvement beyond the tested resolutions. tracer transport models have generally used lower resolution Finally, we show that the high-resolution simulations are than NWP models, as chemical transport models including useful for the assessment of the small-scale variability of chemistry and/or long window data assimilation cannot af- CO2 which cannot be represented in coarser-resolution mod- ford such computational expense. els. These representativeness errors need to be considered Observations of atmospheric CO2 are used in data assim- when assimilating in situ data and high-resolution satellite ilation systems based on tracer transport models to produce data such as Greenhouse gases Observing Satellite (GOSAT), optimal estimates of atmospheric CO2 concentrations (e.g. Orbiting Carbon Observatory-2 (OCO-2), the Chinese Car- Massart et al., 2016) or model parameters and CO2 fluxes bon Dioxide Observation Satellite Mission (TanSat) and fu- in atmospheric inversion systems (e.g. Rayner et al., 2005; ture missions such as the Geostationary Carbon Observatory Chevallier et al., 2010). If tracer transport models cannot rep- (GeoCarb) and the Sentinel satellite constellation for CO2 . resent the synoptic variability accurately, then the resulting For these reasons, the high-resolution CO2 simulations pro- errors when comparing the tracer from the model with obser- vided by the CAMS in real time can be useful to estimate vations will prevent these observations from being used ef- such small-scale variability in real time, as well as provid- fectively in the data assimilation systems (e.g. Brooks et al., ing boundary conditions for regional modelling studies and 2012). The model–observation mismatch caused by differ- supporting field experiments. ences in the resolution of the tracer transport model – includ- ing both the resolution of the meteorological fields and the resolution of the fluxes on the model grid – and the resolution of the observation footprint is also known as representative- 1 Introduction ness error. Failure to properly account for representativeness errors in data assimilation will lead to errors in the optimised Over synoptic weather timescales of hours to days and spatial parameters, whether atmospheric concentrations, model pa- scales less than 1000 km, the assumption that atmospheric rameters or surface fluxes. CO2 is well-mixed into a homogeneous background does not Several studies have investigated the spatial representa- hold, as shown by the observed variability at baseline in situ tiveness errors of CO2 (Miller et al., 2007; van der Molen stations (e.g. Halter and Harris, 1983). CO2 weather is de- and Dolman, 2007; Corbin et al., 2008; Tolk et al., 2008) fined here as the atmospheric CO2 variability at timescales by analysing the CO2 distribution within model grid cells, and spatial scales of weather systems (Parazoo et al., 2011) based on nested high-resolution simulations on limited do- as depicted in Fig. 1. It reflects a complex combination of an- mains over Europe, North America and South America for thropogenic and natural CO2 fluxes near the Earth’s surface certain months or by statistical parameterisation of CO2 co- and transport by weather systems in the atmosphere (Geels variances based on lower-resolution simulations (Alkhaled et al., 2004; Patra et al., 2008). This synergy of CO2 fluxes et al., 2008). The importance of high resolution over complex and weather results in intricate atmospheric CO2 patterns terrain has also been demonstrated on regional scales, e.g. in of positive and negative anomalies, collocated with weather Europe (van der Molen and Dolman, 2007; Ahmadov et al., variations on top of the well-mixed CO2 background that 2009; Pillai et al., 2011) and in North America (Lin et al., varies slowly on timescales of weeks to years (Keeling et al., 2017; Hedelius et al., 2017) using very high resolution sim- 1976). ulations (down to 1 km). However, other studies with coarser Modelling the synoptic-scale transport that modulates the global tracer transport models have compared CO2 simula- CO2 weather is crucial for interpreting the variability of tions with a range of resolutions from a few degrees down to surface CO2 concentrations from in situ observations (Law 0.5◦ without finding significant improvements with respect et al., 2010) and column-averaged CO2 from satellite and to observations (Lin et al., 2018; Remaud et al., 2018). ground-based observations (Corbin et al., 2008) and for fore- The full impact of horizontal resolution on the simu- casting CO2 from 1 to 10 d ahead (Agustí-Panareda et al., lated tracer variability depends on the resolution of trans- 2014; Tang et al., 2018) in order to examine the predictive port and emissions/biogenic fluxes (e.g. Vogel et al., 2013) skill of the models. Tracer transport models use the numer- as well as the resolution of the topography and the winds ical schemes and meteorological information of numerical (e.g. Sekiyama et al., 2015) in the model. In this study the weather prediction (NWP) to simulate the tracer variability full sensitivity of CO2 synoptic variability to the model hori- in the atmosphere. Increasing the horizontal resolution asso- zonal resolution (including all the aspects mentioned above) ciated with the grid spacing of tracer transport models has the is investigated by quantifying the change in model error with benefit of reducing the numerical errors in tracer simulations, horizontal resolution at observing stations. Three main ques- leading to convergence of the transport solution from differ- tions are addressed: ent transport schemes (Prather et al., 2008). NWP models for weather forecasting have been doubling the global hori- zontal resolution approximately every 8 years (Wedi, 2014) Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7349 Figure 1. Snapshots of column-averaged CO2 (XCO2 ) (ppm) above (in reds) and below (in greens) the global mean on 15 January (a) and 15 July (b) at 12:00 UTC from the Copernicus Atmosphere Monitoring Service global CO2 forecast at high horizontal resolution (∼ 9 km). 1. What is the sensitivity of the modelled atmospheric CO2 observations and tools used to assess the importance of hori- variability at diurnal and synoptic timescales to hori- zontal resolution for simulating atmospheric CO2 variability zonal resolution? related to weather is presented in Sect. 2. Section 3 shows the impact of horizontal resolution on the error of simulated 2. How is horizontal resolution affecting the medium- horizontal winds (Sect. 3.1) and atmospheric CO2 (Sect. 3.2 range (1–10 d) forecast error growth of atmospheric and 3.3). The results of the sensitivity to horizontal resolution CO2 ? are explained in the context of the small-scale variability in Sect. 3.4. The diagnostics of small-scale variability provide 3. What are the typical CO2 representativeness errors in an estimate of the expected representativeness errors for CO2 models with horizontal resolutions of 1◦ × 1◦ , currently simulations with coarser horizontal resolutions. Finally, an considered as high resolution in tracer transport models, example of an urban site is shown in Sect. 3.5, where the im- and where and when are these representativeness errors pact of horizontal resolution is positive in January and neg- largest? ative in July. The implications of the results for CO2 fore- casting and atmospheric inversion systems are discussed in The model simulations use the operational Copernicus At- Sect. 4, with a summary of the main findings on why and mosphere Monitoring Service (CAMS) global CO2 forecast- where horizontal resolution matters. ing system (Agustí-Panareda et al., 2014; Massart et al., 2016) which is based on the Integrated Forecasting Sys- tem (IFS) model of the European Centre for Medium-Range Weather Forecasts (ECMWF). They are performed over a range of resolutions currently used operationally in NWP from 9 to 80 km. A detailed description of the simulations, www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7350 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 2 Methodology the surface. However, the temporal coverage of TCCON ob- servations is limited to clear-sky and sunny conditions, which 2.1 Observations means there are generally more gaps in the TCCON data than in near-surface in situ data. Continuous in situ observations near the surface and column- averaged observations from the Total Carbon Column Ob- 2.2 Global atmospheric CO2 model serving Network (TCCON) provide the reference for atmo- spheric CO2 variability. Figure 2 shows the spatial distribu- The model used in this study is the Integrated Fore- tion of the CO2 observing stations used in this study. Hourly casting System (IFS), the same model used in NWP near-surface CO2 observations are provided by 51 in situ sta- at ECMWF and in the CAMS atmospheric composi- tions operated by various organisations throughout the period tion analysis and forecasting system to issue 5 d CO2 of the simulations: data from 44 stations are taken from the and CH4 forecasts (https://atmosphere.copernicus.eu/charts/ cooperative GLOBALVIEWplus (ObsPack , 2015) data set, cams/carbon-dioxide-forecasts, last access: 30 May 2019), and additional data have been obtained from three additional as well as reactive gases and aerosol forecasts relevant for stations from CSIRO in Australia and Antarctica and four air quality (Flemming et al., 2015; Morcrette et al., 2009). stations from the ClimaDat network (Morguí et al., 2013, The IFS model version is CY43r1, the operational weather 2017) over the Iberian Peninsula. The cooperative GLOB- forecast model at ECMWF from 22 November 2016 to ALVIEWplus (ObsPack , 2015) data set is coordinated by 10 July 2017. A full evaluation of this model cycle can be NOAA, and it comprises data collected by various insti- found in Haiden et al. (2017). The high horizontal resolu- tutions and laboratories including AEMET, AGH, CSIRO, tion is based on a cubic octahedral reduced Gaussian – called ECCC, ECN, EMPA,FMI, HMS, LSCE, NCAR, NOAA, hereafter octahedral – grid (Holm et al., 2016). The imple- JMA, NIWA, SAWS, TU, UBA-SCHAU, UEA, UHEI-IUP mentation of the octahedral grid has allowed a substantial and UR (see Tables A1 and A2 for full list of stations with increase in the grid point resolution from 16 km to approxi- their organisations and associated references). No selection mately 9 km, without having to increase the spectral resolu- criteria are applied to the stations from the GLOBALVIEW- tion of the model (Malardel et al., 2016). The 9kmEXP sim- plus ObsPack (2015), CSIRO and ClimaDat data sets, other ulation comprises up to 904 million model grid points, 137 than availability of hourly data for January and July 2014. levels and a time step of 7.5 min. Most stations are on the World Meteorological Organi- The tracer transport is modelled by three different numeri- zation (WMO) CO2 scale, although the inter-calibration of cal schemes to represent (i) the resolved advection of CO2 by standard gases is not critical for this study because the fo- the winds, and the sub-grid-scale (ii) convection and (iii) tur- cus is on the relative difference between the high- and low- bulent mixing processes that need to be parameterised. The resolution simulations to quantify the sensitivity of modelled tracer advection is computed using a semi-implicit semi- CO2 to horizontal resolution in the model. The distribution Lagrangian scheme (Temperton et al., 2001; Diamantakis of the stations is not homogeneous over the globe. How- and Magnusson, 2016) which is an unconditionally stable ever, there is a wide variety of locations that sample syn- method for the integration of the transport equations and optic variability on various types of terrain including many for the fast terms associated with gravity waves. Semi- coastal, mountain, continental and oceanic sites over differ- Lagrangian advection schemes have small dispersion and ent continents on both hemispheres. Wind observations from phase speed errors despite using long time steps (Staniforth around 400 radiosondes stations and all the operational 10 m and Côté, 1991). In practice, these properties mean that the SYNOP stations around the globe are used to evaluate the time step is limited only by the local truncation error and sensitivity of wind errors to the model horizontal resolution not by numerical stability bounds. The semi-Lagrangian at different atmospheric levels in the troposphere. advection scheme in the IFS is not mass-conserving. Thus, a Total column observations from 18 TCCON Fourier trans- mass fixer is required to ensure mass conservation at every form spectrometers (FTSs) (Wunch et al., 2011) available in time step (Agustí-Panareda et al., 2017; Diamantakis and January and July 2014 – shown as red triangles in Fig. 2 – are Agustí-Panareda, 2017). The turbulent mixing scheme is de- also used to evaluate the variability of the column-averaged scribed in Beljaars and Viterbo (1998), Koehler et al. (2011) dry-air mole fraction of CO2 – hereafter referred to as XCO2 and Sandu et al. (2013). The convection scheme is based on – (Table A3). These TCCON observations are retrieved from Tiedtke (1989) (see Bechtold et al., 2008, 2014, for further direct solar near-infrared spectra (http://www.tccon.caltech. details). Full documentation of the IFS can be found at https: edu, last access: 30 May 2019), and they provide a ground //www.ecmwf.int/en/forecasts/documentation-and-support/ reference to the GOSAT (Kuze et al., 2009), OCO-2 (Crisp changes-ecmwf-model/ifs-documentation (last access: et al., 2017) and TanSat (Yang et al., 2018) satellite obser- 30 May 2019). vations (e.g. Inoue et al., 2016; Wunch et al., 2017). Total The CO2 surface fluxes from the ocean, biomass burning column averages are less sensitive to the uncertainties asso- and anthropogenic emissions are prescribed using invento- ciated with vertical mixing than the CO2 abundances near ries or climatologies, while the biogenic fluxes over land Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7351 Figure 2. Map of in situ (blue squares) and TCCON (red triangles) stations. Detailed information on each station is provided in Tables A1 and A3. are modelled online (see Table 1). The anthropogenic CO2 used in tracer transport models and the typically higher reso- emissions come from the EDGAR v4.2 FT2010 inventory lutions of operational weather forecasts models. For instance, for 2012 (last year with gridded emissions). They are ex- the deterministic weather forecast at ECMWF currently runs trapolated in time to the year of the simulation with coun- at 9 km resolution, and it was the global forecasting system try trends provided by the EDGAR database (http://edgar. with the highest resolution in the world when it was intro- jrc.ec.europa.eu, last access: 30 May 2019). The biogenic duced on 8 March 2016 (Holm et al., 2016). CO2 emissions from land vegetation are modelled with the The octahedral grid is used for all simulations, except for A-gs photosynthesis scheme and an empirical model to sim- the lowest-resolution simulation at 80 km which uses a re- ulate the ecosystem respiration fluxes which are integrated duced linear Gaussian grid as in the ERA-Interim and CAMS in the CHTESSEL land surface model of the IFS (Boussetta Reanalysis (Inness et al., 2019). The time steps are also de- et al., 2013). The fluxes have been evaluated with FLUXNET pendent on the horizontal resolution and range from 7.5 to data and compared to different models (e.g. CASA and OR- 45 min. As described in Sect. 2.2, the semi-implicit semi- CHIDEE) showing a comparable performance on synoptic to Lagrangian method used in the IFS is free from stability seasonal scales (Balzarolo et al., 2014). An online bias cor- restrictions. Thus, the model uses the longest possible time rection scheme (Agustí-Panareda et al., 2016) is applied to step that provides the most accurate result for each spatial the modelled gross primary production (GPP) and ecosys- resolution. This is selected through experimentation and val- tem respiration (Reco) fluxes to correct for biases in the net idation, but a rule of thumb is that as the horizontal reso- ecosystem exchange (NEE) budget on a timescale of 10 d lution increases, the time step decreases to keep the mean compared to a climatology of optimised fluxes (Chevallier Courant–Friedrichs–Lewy (CFL) number constant. This typ- et al., 2010). Figure S1 in the Supplement shows the monthly ically leads to much longer time steps than Eulerian models mean NEE for the highest and lowest resolutions. for which their time step is restricted by the typical CFL sta- The atmospheric tracer transport and CO2 biogenic fluxes bility limit (i.e. the maximum CFL number being less than are two of the largest contributors to the synoptic variabil- 1). ity of atmospheric CO2 globally (Geels et al., 2004; Agustí- All the simulation experiments are conducted for a win- Panareda et al., 2014). Thus, the modelling of these two com- ter and a summer month, in January and July 2014, as we ponents online in the IFS allows us to investigate the full im- expect that winter and summer periods will show markedly pact of the resolution coming from the winds and the tracer different variability patterns in CO2 . Figure 3 shows the con- transport, as well as the fluxes. figuration of the simulations. A 10 d forecast is performed at 00:00 UTC each day of the month. The meteorological 2.3 Global atmospheric CO2 simulations initial conditions of each forecast come from the ECMWF operational NWP analysis (Rabier et al., 2000), whereas the atmospheric CO2 tracer is initialised with the previous 1 d A set of global simulations are performed at several resolu- forecast, which means CO2 is essentially free-running, as in tions from 9 to 80 km (Table 2) to investigate the impact of Agustí-Panareda et al. (2014). The first initial conditions for horizontal resolution on the modelled CO2 variability at di- CO2 on 1 January and 1 July 2014 are extracted from the urnal and synoptic scales. These are the resolutions that are CAMS CO2 analysis (Massart et al., 2016). NWP analysis of currently used operationally in global meteorological reanal- meteorological fields is one of the main elements determin- ysis – e.g. ERA-Interim at 80 km (Dee et al., 2011) – widely www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7352 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Table 1. Datasets and models of CO2 fluxes used in simulations listed in Table 2. Flux type Source Temporal Resolution Reference resolution (lat × long.) Anthropogenic EDGAR v4.2 FT2010 Annual mean 0.1◦ × 0.1◦ Olivier and Janssens-Maenhout (2015) Biomass burning GFAS Daily mean 0.1◦ × 0.1◦ Kaiser et al. (2012) Ocean Takahashi climatology Monthly mean 4.0◦ × 5.0◦ Takahashi et al. (2009) NEE CHTESSEL Adapted to Adapted to Boussetta et al. (2013), model time step model resolution Agustí-Panareda et al. (2016) Table 2. List of simulations with different resolutions given by dif- 2.4 Diagnostics for model evaluation ferent model grids. All simulations use 137 vertical model levels. All the experiments have been performed in January and July 2014 The focus of this paper is on assessing the skill of the model using the same CO2 surface fluxes (see Table 1). in simulating CO2 weather with short-term variability over a period of a month. For this purpose, the root mean square Experiment Model Model grid Model error resolution time step v N u 9kmEXP 9 km Tco1279 7.5 min u1 X 16kmEXP 16 km Tco639 12 min RMSE = t (mi − oi )2 , (1) N i=1 25kmEXP 25 km Tco399 15 min 40kmEXP 40 km Tco255 20 min 80kmEXP 80 km Tl255 45 min the systematic error or bias N 1 X µ= (mi − oi ) (2) ing the quality of the tracer transport (Locatelli et al., 2013; N i=1 Polavarapu et al., 2016). Keeping the meteorological fields close to the analysis by having a sequence of 1 d forecasts and the random error ensures the tracer transport is as realistic as possible. There- v N u fore, the sequence of 1 d forecasts is used as the standard u1 X STDE = t (mi − oi − µ)2 (3) (cyclic forecast) configuration for the simulations at differ- N i=1 ent resolutions. The extension to the 10 d forecasts allows us to assess the of the modelled CO2 dry molar fraction (m) are computed impact of errors in the meteorological fields – which grow with respect to N hourly observations (o) at each observing during the forecast – on the CO2 simulations. There are 10 site. The standard deviation of the site error – also known realisations of CO2 for each day, one for each forecast lead as inter-station error – is used as an indicator of the spatial time (Fig. 3). Each forecast lead time is evaluated separately variability of the error e (e.g. RMSE, µ) between the M ob- in order to estimate the error growth during the forecast. serving sites: For consistency in the evaluation of the different forecast v lead times, the periods from 10 January to 10 February and u u 1 X M 10 July to 10 August are used in the validation diagnostics. σe = t (es − e)2 , (4) The simulations also include an additional CO2 tracer M − 1 s=1 which is only transported (i.e. does not respond to CO2 sur- face fluxes) during the forecast. We refer to this tracer as where e is the mean error of all sites. It reflects the skill of NFX. This tracer is still initialised with the standard CO2 the model in representing spatial gradients between the sites. at the beginning of each forecast. The difference between the The Pearson’s correlation coefficient is also used to assess NFX CO2 and the standard CO2 tracers can provide insight the skill of the model in simulating the diurnal and synoptic into the sensitivity to local flux at different horizontal resolu- variability at the sites. tions. Similarly, the change in the error of the simulation with The model is sampled in the horizontal by taking the near- resolution for both the standard and the NFX tracers can be est grid point to the station over land. This approach is widely used as an indicator of transport versus local flux influence used in model evaluation (Patra et al., 2008) as it allows as- in the assessment of the impact of horizontal resolution. sessment of the model directly at grid point scale. At coastal locations, coarse-resolution models can find a better fit to ob- servations by sampling the nearest ocean grid point as land grid points tend to overestimate the diurnal cycle (Law et al., Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7353 Figure 3. Schematic of CO2 simulations with cyclic forecast configuration with 10 d forecasts initialised every day from 1st of the month to the 10th day of the following month. Initial conditions are depicted by arrows (see legend), and the period of evaluation in which several forecast lead times can be compared is delimited by the red dashed line. The standard CO2 simulations are composed by a series of 1 d forecasts as shown by the green rectangles. 2010). For this reason, the sampling protocol for observa- that the sampling at a.m.s.l. greatly underestimates the am- tions in the atmospheric inversion system moves some sta- plitude of the diurnal cycle, as the sensitivity to local fluxes tions offshore (Gurney et al., 2003). However, coastal sites is reduced at higher levels above the ground. Since most can be influenced by both ocean and land, which means that low-resolution models used in atmospheric inversions tend to they will have contrasting periods sampling baseline air as- use the model sampling a.m.s.l. at mountain sites (e.g. Wang sociated with low variability and periods with land and local et al., 2018), a comparison of the two approaches (a.g.l. and influences associated with high CO2 variability (Patra et al., a.m.s.l.) is provided in the Supplement. At the TCCON sta- 2008). In this study we have chosen to consistently sample tions, the model profile is processed with the TCCON aver- the nearest land point over land because we are interested aging kernel and prior, as described in Appendix A of Mas- in assessing the capabilities of the model to represent both sart et al. (2016). baseline and local influences. The temporal sampling is per- Atmospheric CO2 variability is subject to local- or small- formed with a linear interpolation from the 3-hourly archived scale influences (< 100 km) associated with complex topog- model fields to the observation time as in Agustí-Panareda raphy, coastal boundaries, local fluxes and mesoscale atmo- et al. (2014). spheric flow (Lin, 2007). Most models used in carbon cy- At the surface stations, the model is also interpolated to the cle studies are unable to represent such local variability. The altitude of the sampling height above ground level (a.g.l.). resulting representativeness errors reflect the sub-grid-scale This ensures the same model levels are used for the dif- variability associated with the coarse resolution of the mod- ferent horizontal resolutions. The model has hybrid coordi- els (e.g. Tolk et al., 2008). At high resolution it is possible nates that follow the terrain close to the surface. Selecting to estimate part of this sub-grid-scale variability of coarser the model level at the station height above mean sea level models. In order to do that, the 9kmEXP simulation is inter- (a.m.s.l.) would imply the use of different model levels for polated onto a 0.1◦ × 0.1◦ regular lat–long. grid, and subse- different resolutions when the orographic height varies be- quently it is sampled for each time zone (computed hourly tween the horizontal resolutions. It would therefore lead to along longitude) at 13:00 ± 0.5 h local time. This temporal comparisons of CO2 in the planetary boundary layer and sampling at 13:00 is consistent with the GOSAT (http://www. free troposphere at mountain sites where the low-resolution gosat.nies.go.jp/en/, last access: 30 May 2019) and OCO-2 model underestimates the orographic height. Lin et al. (2017) (http://oco.jpl.nasa.gov/mission, last access: 30 May 2019) tested both approaches at several mountain sites. They found overpass time. Thus, it provides a more relevant estimate www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7354 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather of the potential representativeness error for lower-resolution sessed with the error of the CO2 simulations with respect inversion systems, which use daytime surface in situ data to hourly mean observations. The change in the RMSE with and satellite data (e.g. Chevallier et al., 2010, 2014). The horizontal resolution based on the surface CO2 and XCO2 representativeness error is estimated by computing the stan- observations (see Sect. 2.1) is shown in Figs. 5 to 7. dard deviation of the CO2 dry molar fraction at 0.1◦ resolu- At the surface there is an overall substantial reduction of tion within the coarser grid boxes of 1◦ × 1◦ over the whole RMSE between 80 and 9 km (i.e. between 1.8 and 3.5 ppm globe: for hourly data) which is clearly not linear (Fig. 5a and b). v The RMSE difference between the 80kmEXP and 40km- u u 1 X n EXP simulations or the 40kmEXP and 25kmEXP simula- σCO2 = t (mj − m)2 , (5) n − 1 j =1 tions is not as large as the difference between the higher- resolution simulations (e.g. the 25kmEXP and 16kmEXP or n 1 the 16kmEXP and 9kmEXP simulations). This is particularly mj ; n is the number of 0.1◦ resolution grid P where m = n j =1 pronounced for the daily maximum CO2 occurring usually cells within the coarser grid cell of 1◦ × 1◦ ; m is the CO2 dry at night-time, which is generally controlled by local fluxes molar fraction at 0.1◦ resolution; and m is the average within and small-scale transport of tracers, and therefore it is more the coarser grid cell. sensitive to resolution. The daily maximum values are gen- erally much better captured at 9 km resolution compared to 80 km, with a reduction in the RMSE of around 2.5 ppm in 3 Results January and 6 ppm in July. Indeed, there are large differ- ences between the RMSE of the daily maximum and mini- 3.1 Impact of horizontal resolution on winds mum CO2 values. As expected, daily minimum values that The accuracy of the winds is a crucial aspect of the CO2 emerge during daytime have a smaller RMSE. This is be- transport quality, as winds drive the advection of CO2 across cause during daytime the minimum CO2 values are influ- the resolved gradients in the model. In this section we in- enced by the larger-scale fluxes and tracer transport which vestigate the benefit of increasing the resolution from 80 to are less sensitive to high resolution. The reduction in RMSE 9 km on the RMSE of the zonal and meridional components of the daily minimum CO2 is therefore smaller than for the of the wind. We investigate the changes in the global wind er- daily maximum, but it is still considerable, with an RMSE ror with model resolution based on 12-hourly radiosonde ob- decrease of around 0.75 ppm from 80 to 9 km resolutions in servations which measure the horizontal wind components both January and July. These differences reflect the ability throughout the troposphere. Figure 4 shows there is a con- of the model to represent the diurnal cycle. The 9kmEXP sistent and significant RMSE reduction of the vector wind simulation clearly shows a general improvement in the CO2 for the 1 d forecast with resolution. The impact of resolu- diurnal cycle near the surface, with smaller differences in the tion – quantified here by the difference in RMSE between RMSE of the two daily extremes. The largest RMSE reduc- the 80kmEXP and 9kmEXP simulations – is largest near the tion comes from mountain sites (over 1000 m a.m.s.l.), rang- surface at 850 and 1000 hPa, with a RMSE reduction ranging ing between 6 and 10 ppm for hourly CO2 (Fig. 6a and b), between 0.2 and 0.6 m s−1 . This is equivalent to a reduction compared to the lowland sites, which can see improvements in RMSE of around 15 % near the surface. In the middle and between 0.5 and 2 ppm for hourly CO2 RMSE near the sur- upper troposphere (500 and 200 hPa) there is a consistent but face (Fig. 6c and d). smaller RMSE reduction, ranging between 0.1 and 0.2 m s−1 . In general there is also a notable reduction in the spread of The RMSE reduction extends throughout the 10 d fore- the RMSE at the different sites with resolution, as shown by cast for the two components of the wind between 1000 and the σ RMSE values below the panels in Figs. 5 and 6. This 850 hPa with values around 0.4 m s−1 , and it is consistent in implies that the spatial gradients between stations are better both the Northern and Southern Hemisphere and the trop- represented at higher resolutions. The global mean correla- ics (not shown). The results are also in agreement with the tion coefficient also increases with resolution from 0.47 to RMSE with respect to 10 m wind speed from SYNOP obser- 0.56 in January and 0.51 to 0.59 in July for the hourly CO2 , vations, with a mean RMSE reduction over the global domain with consistently higher correlations for the daily mean, min- of 0.34 m s−1 . The reduction of the mean error is smaller than imum and maximum CO2 at higher resolution. the RMSE (< 0.2 m s−1 ) throughout the troposphere, which As expected, the sensitivity to the strategy of sampling the means the largest component of the wind error is random. model level at observing stations is generally small over low- lands but large over mountains, particularly at low resolution 3.2 Impact of horizontal resolution on CO2 diurnal (Fig. S2). At mountain sites, the model level at the real sta- and synoptic variability tion height above mean sea level is predominantly in the free troposphere, and therefore it has a small sensitivity to the The sensitivity of the model skill at hourly and daily local fluxes and flow, whereas taking the model level with timescales to the horizontal resolution of the model is as- respect to the model ground generally exhibits larger errors Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7355 Figure 4. Mean RMSE of vector wind (m s−1 ) at different model resolutions in (a) January and in (b) July for around 400 radiosonde stations over the globe. Different colours represent different pressure levels (see legend). All the model simulations are based on the standard 1 d forecast configuration shown in Fig. 3. Note that the number of data at the 1000 hPa level might be slightly smaller than at the other pressure levels, as the observations at 1000 hPa are not available when the surface pressure is lower than 1000 hPa. Figure 5. Mean RMSE of near-surface CO2 (ppm) (a, b) and Pearson’s correlation coefficient R (c, d) at different model resolutions in January (a, c) and July (b, d) for all 51 stations (see Table A1). The standard deviation of the plotted variable from each station is shown by the numbers below the horizontal resolution for each temporal resolution (hourly, daily mean, daily min and daily max). All the model simulations are based on the 1 d forecast. Note that different scales are used in each panel. www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7356 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Figure 6. Mean RMSE of near-surface CO2 (ppm) at different model resolutions in January (a, c) and July (b, d) for (a, b) 37 lowland stations (below 1000 m a.m.s.l.) and (c, d) 12 mountain stations (1000 m a.m.s.l., excluding bao and spo, as listed in Table A1). The standard deviation of the plotted variable from each station is shown by the numbers below the horizontal resolution for each temporal resolution (hourly, daily mean, daily min and daily max). All the model simulations are based on the 1 d forecast. Note that different scales are used in each panel. associated with local influences in the boundary layer. The tions – associated with its sensitivity to large-scale flux pat- difference between the two sampling strategies in the RMSE terns – (Keppel-Aleks et al., 2011, 2012) is causing most TC- and correlation coefficients becomes smaller at high resolu- CON sites to be less sensitive to horizontal resolution. How- tion (Fig. S3). This reflects an improvement in the capability ever, there is a large variation in RMSE between sites (see σ of the model to represent the flow and fluxes around complex RMSE in Figs. 7 and S9), which is reduced at high resolu- topography at higher horizontal resolution. tion. In particular, the TCCON site at Pasadena (California, The XCO2 RMSE at the TCCON sites during daytime USA), located near the anthropogenic emission hotspot of also displays a general decrease with resolution (Fig. 7), with the megacity of Los Angeles, stands out (Fig. S9). The im- differences of the order of 0.1 ppm from 80 to 9 km resolu- provement associated with high resolution at Pasadena is in- tions and increases in the correlation coefficients (r) of up to deed remarkable in January (i.e. approximately 2 ppm RMSE 0.05. In boreal summer, the XCO2 daily minimum has the reduction). A more detailed study for Pasadena is provided in largest/smallest RMSE/r because it reflects the uncertainty Sect. 3.5. associated with modelled photosynthesis and negative XCO2 The change of RMSE with resolution is partly associated anomalies, whereas in boreal winter, the XCO2 daily maxi- with the improvement in the transport and also the represen- mum has the largest RMSE because ecosystem respiration tation of the local fluxes at higher resolutions. Figure 8 shows associated with positive XCO2 anomalies is the dominant that when the fluxes are switched off during the 1 d forecast, process at most TCCON sites. It is likely that the larger foot- there is still an improvement with resolution at most sites, print of XCO2 (Belikov et al., 2017) at most TCCON sta- but the magnitude of the error reduction is smaller (see sym- Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7357 Figure 7. Mean RMSE of XCO2 (ppm) and Pearson’s correlation coefficient R with respect to observations from 18 TCCON stations (see Table A3) at different model resolutions in (a) January and (b) July. The standard deviation of the plotted variable from each station is shown by the numbers below the horizontal resolution for each temporal resolution (hourly, daily mean, daily min and daily max). All the model simulations are based on the 1 d forecast. Note that different scales are used in each panel. bols to the right of the dashed line). This is very clear for deviation of the bias at each station – i.e. the inter-station a large number of mountain sites and TCCON sites affected bias – is reduced at 9 km with respect to 80 km, as shown by by anthropogenic emissions such as Pasadena (USA) in Jan- the shaded area in Figs. S4 and S5. The largest decrease in uary and Saga (Japan) in July. However, there are also some the inter-station bias between 80kmEXP and 9kmEXP sim- sites and months in which the impact of resolution is better ulations occurs in January, when it is almost halved near the without fluxes than with fluxes (e.g. Pasadena in July). This surface. The errors at the individual observing stations are would indicate that in this case the errors in the fluxes are the listed in Tables S1, S2, S3 and S4. main cause of the deterioration in RMSE with resolution. The overall global error statistics of the 9kmEXP and 3.3 Impact of horizontal resolution on CO2 forecast 80kmEXP simulations including the systematic (or bias) er- error growth ror and the standard (or random) error are shown in Table 3. The reduction in RMSE at 9 km is associated with a decrease In 10 d the global mean RMSE of CO2 forecast at the in in the magnitude of the CO2 biases on average of 1.5 to situ surface stations grows by around 1.4 ppm in January and 2 ppm near the surface and up to 0.2 ppm for XCO2 and a around 1 ppm in July (Fig. 9). It is worth noting that this error general reduction in the CO2 random error of 1 to 1.5 ppm growth is smaller in magnitude than the impact of increas- near the surface and 0.1 ppm for XCO2 (Figs. S4 and S5). ing horizontal resolution from 80 to 9 km. Namely, the 10 d The biases depend largely on the bias of the CO2 initial con- forecast at 9 km is better than the 1 d forecast at 80 km near ditions, as well as the biases of the fluxes and tracer transport. the surface. At the TCCON sites the XCO2 RMSE grows What is important in this sensitivity study is that the standard on average between 0.2 and 0.5 ppm in 10 d (Fig. 10). The forecast RMSE growth for near-surface CO2 and XCO2 does www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7358 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Figure 8. Sensitivity of horizontal resolution impact to surface CO2 fluxes at (a, b) surface stations and (c, d) TCCON stations in January and July, as shown by the difference in RMSE between different CO2 tracers, i.e. the standard tracer (RMSE9 km –RMSE80 km ) and the CO2 tracer with just transport (RMSENFX9km –RMSENFX80km ) in the y axis and x axis respectively. The symbols that are close to the dashed line correspond to stations that have a small sensitivity to local fluxes, while at the stations associated with symbols that are located above/below the dashed line there is a negative/positive contribution of the local fluxes at high resolution. The further from the dashed line, the larger the contribution of the local fluxes. The stations located along the y axis are mainly impacted by local fluxes. The surface stations in (a, b) are depicted with different symbols depending on whether they are classified as mountain, continental, coastal or remote (see Table A1), while TCCON stations that are strongly influenced by fluxes are labelled with station names. Note that different scales are used in each panel. not appear to be linear, with a slow growth until day 4 and a the overestimation (e.g. biogenic fluxes responding to rapid faster increase from day 5 onwards. The CO2 RMSE growth adjustments in meteorology after analysis re-initialisation at at 80 km is slightly faster than at 9 km. In summary, the gain 00:00 UTC or issues with the tracer transport associated with in skill from horizontal resolution is maintained throughout the short spin-up period), but these are beyond the scope of the 10 d forecast. Thus, the results suggests that the horizon- this study. tal resolution has a small but positive impact on the short- The near-surface CO2 RMSE increase during the forecast and medium-range forecast skill for CO2 and XCO2 . appears to come mostly from an increase in random error As expected, the RMSE in July is largest because of in January and from both mean and random error in July the high uncertainty associated with the modelled biogenic (Fig. S4), whereas for XCO2 , both mean and random errors fluxes at synoptic scales which influence the variability at contribute equally to the forecast RMSE growth in January continental sites (Agustí-Panareda et al., 2014). There is also and July (Fig. S5). This is probably linked to the distribu- a larger uncertainty in the meteorology driving the tracer tion of the stations, as most in situ stations are located in the transport during summer compared to winter (Haiden et al., Northern Hemisphere, whereas TCCON stations are more 2017). The fact that the forecast RMSE for day 1 is larger equally distributed in both hemispheres, and thus, the mean than for day 2 in July is associated with a sporadic overes- error at all stations does not show differences between sum- timation of daily maximum CO2 peaks at sites influenced mer and winter conditions. by strong local fluxes. There are several potential causes of Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7359 Table 3. Surface CO2 and XCO2 mean statistics for bias, STDE and RMSE of all stations and the standard deviation of inter-station statistics (in brackets and bold font) from the highest- and lowest-resolution simulations, i.e. 9kmEXP and 80kmEXP respectively. All the SFC CO2 stations used in January and July are listed in Tables S1 and S2; the TCCON stations used in the XCO2 statistics are listed in Tables S3 and S4. Data Period Temporal Bias STDE RMSE resolution 9kmEXP 80kmEXP 9kmEXP 80kmEXP 9kmEXP 80kmEXP Hourly mean −0.09 (2.25) 1.73 (4.70) 2.70 (2.25) 3.78 (3.28) 3.21 (2.66) 4.96 (5.04) Daily mean −0.06 (2.35) 1.76 (4.71) 1.76 (1.43) 2.59 (2.21 2.43 (2.17) 4.08 (4.48) SFC CO2 January Daily min −0.30 (1.48) 0.62 (2.44) 1.41 (1.09) 1.82 (1.38) 1.87 (1.39) 2.59 (2.18) Daily max 0.02 (4.43) 2.66 (7.82) 3.68 (4.00) 4.73 (4.56) 4.77 (5.12) 7.12 (7.76) Hourly mean 0.96 (5.76) 2.67 (9.64) 6.70 (6.94) 9.64 (11.93) 7.90 (8.02) 11.56 (14.18) Daily mean 1.14 (6.55) 2.84 (9.80) 4.08 (5.46) 5.52 (8.30) 5.91 (7.44) 8.55 (11.38) SFC CO2 July Daily min 1.48 (4.17) 0.86 (4.82) 3.70 (4.40) 4.11 (3.74) 4.87 (5.36) 5.59 (4.83) Daily max −0.44 (14.09) 5.07 (23.05) 9.10 (11.91) 12.20 (19.57) 13.22 (15.71) 19.63 (26.44) Hourly mean −0.49 (0.74) −0.38 (0.92) 0.58 (0.26) 0.69 (0.54) 1.02 (0.37) 1.12 (0.68) Daily mean −0.51 (0.77) −0.38 (0.99) 0.47 (0.23) 0.58 (0.48) 0.97 (0.40) 1.09 (0.68) XCO2 January Daily min 0.60 (1.09) 0.75 (1.18) 0.92 (0.51) 1.03 (0.60) 1.36 (0.87) 1.53 (1.0) Daily max −1.60 (1.01) −1.49 (1.23) 1.00 (0.77) 1.13 (0.91) 1.99 (1.1) 2.11 (1.15) Hourly mean 1.01 (0.74) 1.04 (0.92) 0.71 (0.32) 0.74 (0.29) 1.28 (0.57) 1.35 (0.63) Daily mean 0.99 (0.77) 1.03 (0.99) 0.56 (0.28) 0.59 (0.27) 1.18 (0.56) 1.25 (0.61) XCO2 July Daily min 2.09 (1.09) 2.18 (1.18) 1.05 (0.67) 1.07 (0.69) 2.37 (1.00) 2.46 (1.06) Daily max −0.01 (1.01) −0.07 (1.23) 0.92 (0.48) 0.87 (0.45) 1.21 (0.56) 1.15 (0.46) Figure 9. Mean RMSE of near-surface CO2 (ppm) at different forecast lead times for the 9kmEXP (red) and 80kmEXP (blue) in (a) January and (b) July. The errors are computed with respect to hourly continuous in situ surface measurements from 51 stations (see Table A1). The error standard deviation between the different stations is shown with the shaded area: red for 9kmEXP, blue for 80kmEXP and grey for overlap. Note that different scales are used in each panel. 3.4 Impact of horizontal resolution on CO2 small-scale ability at the surface in the 9kmEXP compared to the 80km- variability EXP is also illustrated in Fig. S6. Larger values than 10 ppm can be found over most land areas at night-time (Fig. 12). The sensitivity of the RMSE to resolution is generally as- These values are likely to be underestimated, since we ex- sociated with regions that are affected by small-scale vari- pect horizontal gradients to become steeper as the resolution ability that cannot be properly represented by typical global increases, the point sources associated with anthropogenic tracer transport models (Law et al., 2008; Patra et al., 2008). activities become stronger at the grid cell scale and part of Figure 11 shows that the mean small-scale variability, given the sub-grid-scale flow is resolved. by the standard deviation within 1◦ × 1◦ grid box, can be as Coastal sites and mountain sites have a typical sub-grid- large as 10 ppm near emission hotspots at the surface dur- scale variability of around 5 ppm during daytime. This vari- ing daytime. The representation of the CO2 small-scale vari- ability varies from January to July, depending on meteoro- www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7360 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Figure 10. Same as Fig. 9 but for XCO2 (ppm). The errors are computed with respect to hourly TCCON observations from 18 TCCON stations (see Table A3). Figure 11. Monthly mean surface CO2 small-scale variability (σ ) within 1◦ × 1◦ grid boxes (ppm) at 13:00 local time in (a) January and (b) July. Grey shading denotes σ < 1.0. Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7361 Figure 12. Monthly mean surface CO2 small-scale variability (σ ) within 1◦ × 1◦ grid boxes (ppm) at 01:00 local time in (a) January and (b) July. Grey shading denotes σ < 1.0. logical conditions (e.g. stagnant or windy conditions) and steeper gradients throughout the troposphere – as illustrated the magnitude and sign of fluxes (e.g. biogenic activity shift- in Fig. 1b – that contribute to the enhanced sub-grid-scale ing northwards in Northern Hemisphere summer). Over land, variability in summer compared to winter. Over the ocean, the patterns of sub-grid-scale variability of surface and total the small-scale variability of XCO2 ranges between 0.1 and column are consistent (Figs. 11 and 13), as both are subject 0.3 ppm, with lower values in the winter and higher values in to surface heterogeneity in terms of topography and fluxes. the summer. In the Northern Hemisphere summer, the values However, there is a difference in magnitude because the vari- over the ocean and over the land are comparable, whereas ability of the total column average is much smaller than the near the surface, the mean sub-grid-scale variability is an variability at the surface. order of magnitude smaller over the ocean than over land. XCO2 has a maximum standard deviation of 1 ppm near This is because over land the surface fluxes dominate the surface flux hotspots and typically less than 0.5 ppm in most gradients resulting in the steepest gradients being near the regions (Fig. 13), which is consistent with other estimates surface, while over the ocean, the transport associated with from regional studies (Corbin et al., 2008; Pillai et al., 2010) the weather systems creates steep CO2 gradients in the free (see also Fig. S7 for a visual illustration). The differences troposphere. Therefore, column-averaged CO2 is much more in the small-scale XCO2 variability between day and night likely to be influenced by sub-grid-scale variability associ- appear to be small. Interestingly, the small-scale variability ated with weather systems than by surface CO2 fluxes over of XCO2 is much larger in summer than in winter (both in the ocean. the Northern and Southern Hemisphere). During the growing season, negative CO2 anomalies associated with plant photo- synthesis and positive anomalies associated with ecosystem respiration and anthropogenic emissions combine to create www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7362 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Figure 13. Monthly mean XCO2 small-scale variability (σ ) within 1◦ × 1◦ grid boxes (ppm) at 13:00 local time in (a) January and (b) July. Grey shading denotes σ < 0.1. 3.5 Example of horizontal resolution impact at an ary layer (PBL) (Fig. S12) and therefore in XCO2 . The CO2 urban site budget of the anthropogenic emissions used at 9 and 80 km is the same. However, the instantaneous values of the emis- Although the winds, the topography and the spatial hetero- sions per square metre are much higher at 9 than at 80 km, geneity of the fluxes are generally better represented at high representing some of the steep gradients and heterogeneous horizontal resolution, there can still be a deterioration in the distribution of fossil fuel emissions within the LA basin, with RMSE scores at sites where the local influence is strong and higher emissions in downtown LA and lower emissions in the emissions/biogenic fluxes have large errors in the model. Pasadena (e.g. Feng et al., 2016). At 80 km, Pasadena and In this section we present an example of such a case at the downtown LA are in the same model grid box, which means Caltech TCCON site in Pasadena (California, USA; see Ta- this gradient cannot be represented. In addition to the influ- ble A3) with XCO2 under clear-sky and daylight conditions. ence of anthropogenic emissions, the seasonal variation of The variability of the simulated XCO2 exhibits a substantial the winds is very pronounced in Pasadena, with a large con- improvement with high resolution in winter and an equally trast in the origin or air masses between winter and summer considerable deterioration in summer (Fig. 15). Thus, it illus- (Verhulst et al., 2017). trates some of the challenges associated with urban regions. In winter, air masses originate from various directions: Pasadena is located 14 km north-east of the megacity of from the prevailing westerly and southerly winds, bring- Los Angeles (LA) with a large local anthropogenic emission ing and accumulating polluted air from the LA megacity, to influence (Wunch et al., 2009; Newman et al., 2016). The northerly and easterly flow, characterised by cleaner air with XCO2 variability in the model is also mainly explained by lower CO2 values from the surrounding desert and moun- the local anthropogenic emissions (Figs. S10 and S11) pro- tains (Newman et al., 2016). Persistent low wind conditions ducing very large CO2 enhancements in the planetary bound- lead to a large accumulation of CO2 in the LA basin as it Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7363 Figure 14. Monthly mean XCO2 small-scale variability (σ ) within 1◦ × 1◦ grid boxes (ppm) at 01:00 local time in (a) January and (b) July. Grey shading denotes σ < 0.1. remains trapped by the mountains. These episodes result in larger at 9 than at 80 km (Fig. S11). This suggests an overesti- large enhancements in XCO2 (Hedelius et al., 2017) and high mation of the hotspot emissions over the LA basin in the tem- CO2 anomalies over periods of a few days (e.g. 26 to 30 Jan- porally extrapolated EDGAR inventory, which is smoothed uary in Fig. 15a). In those stagnant conditions, the 9kmEXP and less noticeable at lower resolution. There are many rea- simulation is in much closer agreement with the observed sons why the anthropogenic emissions used in the model can XCO2 peaks than the 80kmEXP simulation, which overesti- be overestimated, including the temporal extrapolation based mates the XCO2 anomalies. This is because at 80 km reso- on country-scale scaling factors and the use of annual con- lution there is an effectively uniform emission for the whole stant emissions in EDGAR v4.2 FT2010 instead of season- LA basin. Note that the CO2 and XCO2 small-scale variabil- ally varying emissions (Corbin et al., 2010). ity around LA appears to be larger in winter than in sum- Differences in the sampling location (centre of grid is 3 mer (Figs. 11 and 14). Without preserving the sharp gradient and 34 km from station location at 9 and 80 km respectively) in emissions between Pasadena and downtown LA, the CO2 and orography (15 m below and 46 m below the station height accumulation is overestimated in Pasadena. at 9 and 80 km respectively), as well as differences in flow The atmospheric circulation in summer is mainly con- and local biogenic fluxes can also play a role in explaining trolled by the sea–mountain breeze (Lu and Turco, 1994). the differences between the simulations at 80 and 9 km res- Daytime advection of anthropogenic CO2 -rich air from the olutions. The results are consistent with previous studies by city of LA results in XCO2 peaking in the afternoon before Feng et al. (2016) and Hedelius et al. (2018). They found that it is vented over the mountains (Newman et al., 2013, 2016). uncertainties in the fluxes and their high-resolution represen- The overestimation in the summer XCO2 peaks at 9 km likely tation in the LA basin are as important as the atmospheric reflects an overestimation of the emissions in downtown LA. tracer transport in the representation of the CO2 enhancement The enhancement of CO2 from anthropogenic emissions is and its variability in Pasadena. www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7364 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Figure 15. Hourly XCO2 (ppm) at TCCON site of Pasadena (CA, USA) in (a) January and (b) July from 80kmEXP (blue) and 9kmEXP (red) simulations. Hourly observations are shown by black circles. Triangles represent the model XCO2 after smoothing with TCCON averaging kernel and prior. The bias (µ), standard error (STDE) and Pearson’s correlation coefficient (R) from 80kmEXP (blue) and 9kmEXP (red) are shown at the top of each panel. This example at Pasadena highlights the importance of – to 80 km, which corresponds to the ERA-Interim Reanal- horizontal resolution in representing local gradients of CO2 ysis resolution, widely used by many offline tracer trans- fluxes in order to reduce the atmospheric CO2 representa- port models. The conclusions to the three main questions ad- tiveness error. It emphasises that the impact of increasing dressed in the paper are summarised below. horizontal resolution is not only to reduce the error of at- mospheric CO2 simulations but to enhance the sensitivity of 1. What is the sensitivity of the modelled atmospheric CO2 the modelled atmospheric CO2 variability to the CO2 fluxes variability at diurnal and synoptic timescales to hori- in urban regions characterised by emission hotspots. There- zonal resolution? fore horizontal resolution is crucial for atmospheric inversion The high horizontal resolution of 9 km leads to a general systems that aim to estimate anthropogenic emissions. improvement in the simulated variability of hourly near- surface and column-averaged atmospheric CO2 com- pared to the resolution of 80 km. This is shown by a 4 Discussion and conclusions reduction in the mean RMSE of around 1.8 ppm in win- ter and 3.5 ppm in summer (equivalent to 33 % error re- This paper addresses the importance of horizontal resolution duction) and 0.1 ppm (i.e. around 10 % error reduction) in the representation of CO2 variability at diurnal and synop- at in situ and TCCON sites respectively, which is asso- tic scales, referred to here as CO2 weather. The CO2 simula- ciated with a reduction of both the mean and random tions performed with the ECMWF IFS model allow the com- errors in the model. The inter-station variability is also bined impact of horizontal resolution associated with (i) the generally improved in the 9kmEXP simulation for near- online modelling of the winds, (ii) the numerical tracer trans- surface and column-averaged CO2 in January and July, port model and (iii) the spatial–temporal distribution of CO2 with the standard deviation of station biases reduced up fluxes over land to be quantified. The assessment is done by to 50 % compared to the 80kmEXP simulation in Jan- comparing the model errors at various horizontal resolutions uary for near-surface CO2 . with respect to a wide range of observations with hourly res- Column-averaged CO2 is not as sensitive to horizontal olution and distributed around the globe. The horizontal res- resolution as near-surface CO2 because it has a larger olution of the model ranges from 9 km – as in current opera- footprint or area of flux influence, except for sites like tional high-resolution weather and CO2 forecasts at ECMWF Pasadena which are close to CO2 emission hotspots. Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7365 Similarly, minimum daily values of atmospheric CO2 from day 1 to day 10, implying a long-lived improve- are less sensitive to the horizontal resolution than max- ment in the prediction skill. The RMSE growth is small imum daily values because their footprint tends to be from days 1 to 4, namely less than 0.5 ppm near the sur- larger in size. face CO2 and less than 0.05 ppm for XCO2 . Over the This study also shows that the RMSE reduction with 10 d there is an increase in RMSE of 1 to 1.5 ppm at horizontal resolution is not linear. This implies that the surface and 0.1 to 0.5 ppm for the total column. This results from sensitivity studies exploring the impact error growth is not linear. For example, in July the er- of resolution based on coarse simulations which show ror of the 1 d forecast is worse than the 2 d forecast, small sensitivity to horizontal resolution cannot be ex- with a slower error increase during the 2 to 4 d fore- trapolated to higher horizontal resolutions. These results cast and a generally faster error increase from day 5 to are consistent with the findings of the Lin et al. (2017) day 10 in the forecast. This incoherent change in the er- study based on a wider range of model resolutions from ror evolution at the beginning of the forecast is likely ∼ 100 down to 1 km and observations at three moun- linked to the strong influence of the biogenic surface tain sites. The reduction in model error associated with fluxes, which respond very fast to changes in tempera- the increase of horizontal resolution to 9 km emanates ture, moisture and radiation forcing in the model. Incon- from four different well-known and connected aspects, sistencies between the initial conditions from the anal- as listed below. ysis and the model forecast can cause spin-up adjust- ments which may lead to a degradation of the 1 d fore- a. Better accuracy of the horizontal winds is inte- cast. gral to the reduction of error. The strength of the winds determines the observed CO2 variability – Generally, the improvement of forecast skill with in- i.e. the detected CO2 enhancement – close to emis- creased horizontal resolution is most pronounced in Jan- sion hotspots like in urban regions (Newman et al., uary, when at 9 km resolution the skill of the 10 d fore- 2013; Xueref-Remy et al., 2018). Therefore, the er- cast is better or equal to the accuracy of the 1 d forecast ror in the wind will affect the value of the enhanced at 80 km both near the surface and for the column av- CO2 as much as the error in the fluxes. In this con- erage CO2 . It is likely that the skill of the 10 d forecast text, for example, a wind speed error reduction of to represent variability of CO2 during summer condi- 0.5 m s−1 – as shown in Sect. 3.1 – across a gradi- tions is hampered by the growing errors in the surface ent of 10 ppm per degree – typical of urban areas as biogenic fluxes during the forecast, as they can be an shown in Sect. 3.4 – throughout a 6 h period can re- important contributor to synoptic variability in the sum- sult in a CO2 error reduction of around 1 ppm. Un- mer (Agustí-Panareda et al., 2014). certainty in the winds has been shown to be one of 3. Where and when are the typical representativeness er- the largest contributors to the uncertainty in the es- rors associated with unresolved small-scale variability timated fluxes over urban areas (e.g. Hedelius et al., largest? 2018). b An overall reduction of the numerical error associ- During daytime, the CO2 small-scale variability of the ated with lower spatial and temporal truncation er- 9 km resolution forecast ranges from 1 to 10 ppm at rors leads to a reduction in tracer advection errors the surface and is an order of magnitude smaller (0.1 (Prather et al., 2008). to 1 ppm) for the total column average. It points to the areas associated with small-scale gradients where hori- c. There is a general improvement in the horizontal zontal resolution matters: coastal boundaries and moun- and vertical sampling at the station locations in the tain regions have typical values of 5 ppm per degree, and model associated with a more realistic representa- CO2 flux hotspots have the highest variability of up to tion of orography and coastal boundaries. 10 ppm per degree. During night-time, the small-scale d. A more realistic representation of CO2 flux distri- variability tends to be larger than 10 ppm over most ar- bution at the surface is key. High resolution gives an eas near the surface, whereas that of column-averaged increased capability to represent small-scale sharp CO2 shows small differences between day and night. gradients associated with complex topographical boundaries at coastal and mountainous terrain sites, The high horizontal resolution gives us an insight into as well as the presence of strong local surface fluxes the areas with high sensitivity to uncertainty associated of CO2 such as anthropogenic emission hotspots. with both local tracer transport and fluxes. It is in these areas where improvements in the tracer transport and 2. How is the horizontal resolution affecting the forecast increased understanding of the heterogeneity and com- error growth of atmospheric CO2 ? plexity of the surface will be crucial in future model The horizontal resolution has a consistent positive im- developments. Since these areas are close to emission pact on the error reduction at all forecast lead times, hotspots, it is clear that in order to monitor CO2 emis- www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7366 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather sions, particularly from cities and power stations such Currently, the precision of XCO2 from satellite obser- as in the new Carbon Human Emission project (http: vations is around 1.0 to 1.5 ppm for ACOS-GOSAT data //www.che-project.eu, last access: 30 May 2019), it is (O’Dell et al., 2012) and OCO-2 data (Wunch et al., 2017). paramount to invest in high horizontal resolution mod- However, if tracer transport models cannot represent their els. variability accurately in space and time, all the efforts to re- Interesting differences are found between surface and duce the errors from the satellite retrievals of CO2 will not column-averaged variability. Near the surface the vari- be fruitful in their attempt to reduce the uncertainty in the ability is most pronounced close to emission hotspots estimation of surface fluxes. This is because relatively small and complex terrain. For column-averaged CO2 the sub- differences in atmospheric mixing ratios are associated with grid-scale variability is also substantial over the ocean significant differences in surface fluxes (Houweling et al., downstream from emissions. This emphasises the im- 2010; Ott et al., 2015). The benefits of high resolution in in- portance of the transport influence on XCO2 variabil- version systems will also need to be balanced with the costs ity. Small-scale variability is also found to be more pro- of running a model at such high resolution. nounced in summer than in winter, as biogenic CO2 Finally, the CAMS high-resolution forecast running cur- fluxes of opposite sign in summer enhance the CO2 gra- rently at 9 km resolution can provide benchmarks for other dients in the atmosphere. simulations using coarser grids or offline meteorology (Yu et al., 2018). Both CAMS analysis and high-resolution In summary, this paper has shown that model simulations forecasts are freely available to users (https://atmosphere. using the CAMS CO2 forecasting system at 9 km resolution copernicus.eu, last access: 30 May 2019). Potential applica- can provide a more accurate representation of tracer transport tions include the estimation of representativeness errors and and the local influences of surface fluxes than at lower resolu- data selection screening of observations from satellites and tions ranging from 80 to 16 km, resulting in an overall better in situ stations in data assimilation systems, spatial colloca- representation of the atmospheric CO2 variability at diurnal tion of XCO2 from satellite and TCCON data for validation and synoptic timescales. However, at higher horizontal reso- purposes (e.g. Guerlet et al., 2013) or as boundary conditions lution there is also higher sensitivity of atmospheric CO2 to for high-resolution simulations and/or inversions at regional CO2 flux errors, as emissions and biogenic flux hotspots are scales. not diffused over large areas like in lower-resolution mod- els. Thus, higher-resolution models also risk deterioration in the forecast RMSE, e.g. near emission hotspots associated Data availability. The data are accessible by contacting the corre- with larger errors. With the enhancement of the model un- sponding author (anna.agusti-panareda@ecmwf.int). certainty at high resolution, the prospect of further increasing the horizontal resolution needs to be carefully balanced with improvements in the most uncertain model processes. The impact of horizontal resolution on the accuracy of the winds highlights that errors in the wind need to be consid- ered as an important source of uncertainty both in the atmo- spheric CO2 analysis and forecast and in the inversion sys- tems (Polavarapu et al., 2016). The findings in this study also suggest that increasing horizontal resolution up to kilometric scales in atmospheric data assimilation and inversion systems would allow the use of more in situ and high-resolution satel- lite observations close to strong sources and sinks and over complex terrain. Lin et al. (2017) found that a minimum hor- izontal resolution of 4 km is required to simulate a realistic diurnal cycle of CO2 at mountain sites. Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7367 Appendix A Table A1. Continuous in situ stations (surface and tower) used to evaluate synoptic variability. NA denotes references that are not available. The full names for the abbreviations of the network organisations are provided in Table A2. Station Lat–long. Altitude Intake Network Reference Type ID (m a.m.s.l.) height (m a.g.l.) alt 82.45◦ N, 62.51◦ W 200 10 ECCC Worthy et al. (2003) remote brw 71.32◦ N, 156.61◦ W 11 16 NOAA Peterson et al. (1986) coastal cby 69.01◦ N, 105.05◦ W 35 12 ECCC NA continental inu 68.32◦ N, 133.53◦ W 113 10 ECCC Worthy et al. (2003) continental pal 67.97◦ N, 24.12◦ E 560 5 FMI Hatakka et al. (2003) continental bck 62.80◦ N, 116.05◦ W 179 60 ECCC NA continental chl 58.75◦ N, 94.07◦ W 29 60 ECCC Worthy et al. (2003) coastal llb 54.95◦ N, 112.45◦ W 540 10 ECCC Worthy et al. (2003) continental etl 54.35◦ N, 104.98◦ W 492 105 ECCC Worthy et al. (2003) continental mhd 53.33◦ N, 9.90◦ W 5 24 LSCE Ramonet et al. (2010) coastal wao 52.95◦ N, 1.12◦ E 20 10 UEA Wilson (2013) coastal ces 51.97◦ N, 4.93◦ E −1 200 ECN Vermeulen et al. (2011) continental est 51.66◦ N, 110.21◦ W 707 3 ECCC Worthy et al. (2003) continental fsd 49.88◦ N, 81.57◦ W 210 40 ECCC Worthy et al. (2003) continental cps 49.82◦ N, 74.98◦ W 381 8 ECCC Worthy et al. (2003) continental esp 49.38◦ N, 126.54◦ W 7 40 ECCC Worthy et al. (2003) coastal kas 49.23◦ N, 19.98◦ E 1989 5 AGH Necki et al. (2003), Rozanski et al. (2014) mountain ssl 47.92◦ N, 7.92◦ E 1205 12 UBA-SCHAU Schmidt et al. (2003) mountain hun 46.95◦ N, 16.65◦ E 248 115 HMS Haszpra et al. (2001) continental jfj 46.55◦ N, 7.99◦ E 3570 10 EMPA Schibig et al. (2015) mountain lef 45.95◦ N, 90.27◦ W 472 396 NOAA Andrews et al. (2014) continental puy 45.77◦ N, 2.97◦ E 1465 10 LSCE Lopez et al. (2015) mountain amt 45.03◦ N, 68.68◦ W 53 107 NOAA Andrews et al. (2014) continental egb 44.23◦ N, 79.78◦ W 251 3 ECCC Worthy et al. (2003) continental wsa 43.93◦ N, 60.02◦ W 5 25 ECCC Worthy et al. (2003) remote vac 42.88◦ N, 3.21◦ W 1086 20 ClimaDat Morguí et al. (2013) mountain tpd 42.62◦ N, 80.55◦ W 231 35 ECCC Worthy et al. (2003) continental dec 40.74◦ N, 0.79◦ E 1 10 ClimaDat Morguí et al. (2013) coastal hdp 40.56◦ N, 111.65◦ W 3351 17.7 NCAR Stephens et al. (2011) mountain spl 40.45◦ N, 106.73◦ W 3210 9.1 NCAR Stephens et al. (2011) mountain gic 40.35◦ N, 5.18◦ W 1436 20 ClimaDat Morguí et al. (2013) mountain nwr 40.05◦ N, 105.59◦ W 3523 3.5 NCAR Stephens et al. (2011) mountain bao 40.05◦ N, 105.0◦ W 1584 300 NOAA Andrews et al. (2014) continental ryo 39.03◦ N, 141.82◦ E 260 20 JMA Tsutsumi et al. (2005) coastal snp 38.62◦ N, 78.35◦ W 1008 17 NOAA Andrews et al. (2014) mountain wgc 38.27◦ N, 121.49◦ W 0 483 NOAA Andrews et al. (2014) coastal sgc 36.70◦ N, 5.38◦ W 850 20 ClimaDat Morguí et al. (2013) continental sct 33.41◦ N, 81.83◦ W 115 305 NOAA Andrews et al. (2014) continental wkt 31.31◦ N, 97.33◦ W 251 457 NOAA Andrews et al. (2014) continental izo 28.31◦ N, 16.50◦ W 2373 13 AEMET Gomez-Pelaez and Ramos (2011) mountain yon 24.47◦ N, 123.02◦ E 30 20 JMA Tsutsumi et al. (2005) coastal mnm 24.28◦ N, 153.98◦ E 8 20 JMA Tsutsumi et al. (2005) remote mlo 19.54◦ N, 155.58◦ W 3397 40 NOAA Thoning et al. (1989) mountain smo 14.25◦ S, 170.56◦ W 42 10 NOAA Halter et al. (1988) remote cpt 34.35◦ S, 18.49◦ E 230 30 SAWS Brunke et al. (2004) coastal ams 37.80◦ S, 77.54◦ E 55 20 LSCE Gaudry et al. (1991) remote cgo 40.68◦ S, 144.69◦ E 94 70 CSIRO Francey et al. (2003) coastal mqa 54.50◦ S, 158.94◦ E 6 10 CSIRO Stavert et al. (2018) remote cya 66.28◦ S, 110.52◦ E 47 7 CSIRO Loh et al. (2017) remote syo 69.01◦ S, 39.59◦ E 14 8 TU NA remote spo 89.98◦ S, 24.80◦ W 2810 10 NOAA Conway and Thoning (1990) remote www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7368 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Table A2. Organisations associated with observing stations. Abbreviation Organisation AEMET Izaña Atmospheric Research Center, Meteorological State Agency of Spain AGH AGH University of Science and Technology, Krakow, Poland BIRA-IASB Royal Belgian Institute for Space Aeronomy, Brussels, Belgium Caltech California Institute of Technology ClimaDat Land, Atmosphere and Oceans Laboratory at the Institut Català de Ciències del Clima (2010–2016); at Institut de Ciència i Tecnologia Ambientals, Universitat Autònoma de Barcelona (since 2017) CSIRO Commonwealth Scientific and Industrial Research Organization, Oceans & Atmosphere ECCC Environment and Climate Change Canada ECN Energy Research Centre of the Netherlands EMPA Swiss Federal Laboratories for Materials Science and Technology FMI Finish Meteorological Institute HMS Hungarian Meteorological Service KIT Karlsruhe Institute of Technology LSCE Laboratoire des Sciences du Climat et de l’Environnement MPI-BGC Max Planck Institute for Biogeochemistry NASA National Aeronautics and Space Administration JAXA Japan Aerospace Exploration Agency JMA Japan Meteorological Agency NIES National Institute for Environmental Studies NIWA National Institute of Water and Atmospheric NCAR National Center For Atmospheric Research NOAA NOAA Global Monitoring Division SAWS South African Weather Service TU Tohoku University UBA-SCHAU Umweltbundesamt, Schauinsland station UBremen-IUP Institute of Environmental Physics, Universität Bremen UEA University of East Anglia UHEI-IUP University of Heidelberg, Institut für Umweltphysik UOW University of Wollongong UR Université de La Réunion Table A3. TCCON stations Station ID Latitude–longitude Altitude (m) N data Jan N data Jul Organisation Reference bialystok01 53.23◦ N, 23.02◦ E 180 15 68 UBremen-IUP Deutscher et al. (2014) bremen01 53.10◦ N, 8.85◦ E 27 8 44 UBremen-IUP Notholt et al. (2014a) Karlsruhe 49.10◦ N, 8.44◦ E 116 33 90 KIT Hase et al. (2014) orleans01 47.97◦ N, 2.11◦ E 130 67 16 UBremen-IUP Warneke et al. (2014) garmisch01 47.48◦ N, 11.06◦ E 740 33 90 KIT Sussmann and Rettinger (2014) parkfalls01 45.94◦ N, 90.27◦ W 440 28 168 Caltech Wennberg et al. (2014a) rikubetsu01 43.46◦ N, 143.77◦ E 30 21 9 NIES Morino et al. (2017a) lamont01 36.60◦ N, 97.49◦ W 320 129 299 Caltech Wennberg et al. (2016) tsukuba02 36.05◦ N, 140.12◦ E 30 111 120 NIES Morino et al. (2014b) edwards01 34.96◦ N, 117.88◦ W 699 191 316 NASA Iraci et al. (2016) pasadena01 34.14◦ N, 118.13◦ W 230 160 302 Caltech Wennberg et al. (2014b) saga01 33.24◦ N, 130.29◦ E 7 30 30 JAXA Shiomi et al. (2014) izana01 28.30◦ N, 16.48◦ W 2370 43 18 AEMET/KIT Blumenstock et al. (2014) ascension01 7.92◦ S, 14.33◦ W 10 153 158 MPI-BGC Feist et al. (2014) darwin01 12.43◦ S, 130.89◦ E 30 34 264 UOW Griffith et al. (2014a) reunion01 20.90◦ S, 55.49◦ E 87 150 136 BIRA-IASB/UR De Mazière et al. (2014) wollongong01 34.41◦ S, 150.88◦ E 30 157 96 UOW Griffith et al. (2014b) lauder02 45.04◦ S, 169.68◦ E 370 104 86 NIWA Sherlock et al. (2014) Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7369 Supplement. The supplement related to this article is available Financial support. This research has been supported by the Coper- online at: https://doi.org/10.5194/acp-19-7347-2019-supplement. nicus Atmospheric Monitoring Service and by the European Com- mission (CHE project, grant no. 776186). Author contributions. The simulations were performed by AAP. The coding of the mass fixer required for the high-resolution trans- Review statement. This paper was edited by Christoph Gerbig and port in the IFS was done by MD. The concept and ideas to design reviewed by two anonymous referees. the high-resolution simulations were devised by FC, AAP, MD, SM and JMS in discussion with RE and VHP. RML, ZL, JAM and RC provided additional observations at crucial sites and guidance on the evaluation of the simulations. CR and DW provided data and input on the interpretation of the model evaluation at the TCCON site of References Pasadena. The CO2 validation tools have been developed by SM and AAP. The paper was prepared by AAP with input and feedback Agustí-Panareda, A., Massart, S., Chevallier, F., Boussetta, S., Bal- from MD, SM, FC, JMS, JB, RE, BL, RML, ZL, JAM, MP, VHP, samo, G., Beljaars, A., Ciais, P., Deutscher, N. M., Engelen, MR, CR, ATV, TW and DW. R., Jones, L., Kivi, R., Paris, J.-D., Peuch, V.-H., Sherlock, V., Vermeulen, A. T., Wennberg, P. O., and Wunch, D.: Forecast- ing global atmospheric CO2 , Atmos. Chem. Phys., 14, 11959– 11983, https://doi.org/10.5194/acp-14-11959-2014, 2014. Competing interests. The authors declare that they have no conflict Agustí-Panareda, A., Massart, S., Chevallier, F., Balsamo, G., Bous- of interest. setta, S., Dutra, E., and Beljaars, A.: A biogenic CO2 flux adjustment scheme for the mitigation of large-scale biases in global atmospheric CO2 analyses and forecasts, Atmos. Chem. Acknowledgements. This research was generated using Copernicus Phys., 16, 10399–10418, https://doi.org/10.5194/acp-16-10399- Atmosphere Monitoring Service (2018) information. Anna Agustí- 2016, 2016. Panareda has been partly funded by the CHE project. The CHE Agustí-Panareda, A., Diamantakis, M., Bayona, V., Klappenbach, project has received funding from the European Union’s Horizon F., and Butz, A.: Improving the inter-hemispheric gradient of to- 2020 Research and Innovation programme under grant agreement tal column atmospheric CO2 and CH4 in simulations with the no. 776186. Frédéric Chevallier received funding from the Coper- ECMWF semi-Lagrangian atmospheric global model, Geosci. nicus Atmosphere Monitoring Service, implemented by the Euro- Model Dev., 10, 1–18, https://doi.org/10.5194/gmd-10-1-2017, pean Centre for Medium-Range Weather Forecasts (ECMWF) on 2017. behalf of the European Commission. Thanks are given to all the Ahmadov, R., Gerbig, C., Kretschmer, R., Körner, S., Rödenbeck, station principal investigators from the numerous individual sta- C., Bousquet, P., and Ramonet, M.: Comparing high resolution tions and networks like NOAA, ICOS, AEMET, AGH, CSIRO, WRF-VPRM simulations and two global CO2 transport mod- ECCC, ECN, EMPA, FMI, HMS, LSCE, NCAR, JMA, MPI- els with coastal tower measurements of CO2 , Biogeosciences, BGC, NIWA, SAWS, TU, UBA-SCHAU, UEA, UHEI-IUP, UR, 6, 807–817, https://doi.org/10.5194/bg-6-807-2009, 2009. UBremen-IUP, BIRA-IASB, Caltech, KIT, NASA, JAXA, NIES Alkhaled, A. A., Michalak, A. M., and Kawa, S. R.: Us- and UOW that contributed their CO2 observations to the cooper- ing CO2 spatial variability to quantify representation errors ative GLOBALVIEWplus (ObsPack , 2015) product and TCCON of satellite CO2 retrievals, Geophys. Res. Lett., 35, L16813, which are fundamental for the evaluation of the model simula- https://doi.org/10.1029/2008GL034528, 2008. tions. The ClimaDat Network has received funding from the “la Andrews, A. E., Kofler, J. D., Trudeau, M. E., Williams, J. C., Neff, Caixa” Foundation, under agreement 2010-002624. We are grate- D. H., Masarie, K. A., Chao, D. Y., Kitzis, D. R., Novelli, P. C., ful to many colleagues at ECMWF for their support and fruitful Zhao, C. L., Dlugokencky, E. J., Lang, P. M., Crotwell, M. J., discussions, particularly to Gabor Radnoti, Thomas Haiden and Fischer, M. L., Parker, M. J., Lee, J. T., Baumann, D. D., Desai, Martin Janouseck for their technical support in the evaluation of A. R., Stanier, C. O., De Wekker, S. F. J., Wolfe, D. E., Munger, the winds; Miha Razinger for his technical support in the produc- J. W., and Tans, P. P.: CO2 , CO, and CH4 measurements from tall tion of Fig. 1; Johannes Flemming and Zak Kipling for their sup- towers in the NOAA Earth System Research Laboratory’s Global port in the implementation of the additional CO2 tracers in the IFS Greenhouse Gas Reference Network: instrumentation, uncer- model; Sylvie Malardel for her support and discussions on the gen- tainty analysis, and recommendations for future high-accuracy eral aspects of atmospheric tracer modelling in the IFS; and Gian- greenhouse gas monitoring efforts, Atmos. Meas. Tech., 7, 647– paolo Balsamo, Souhail Boussetta, Zak Kipling and Johannes Flem- 687, https://doi.org/10.5194/amt-7-647-2014, 2014. ming for their technical support in the implementation of a bug fix in Bechtold, P., Köhler, M., Jung, T., Doblas-Reyes, F., Leut- the CTESSEL model of biogenic emissions. Many thanks are given becher, M., Rodwell, M., Vitart, F., and Balsamo, G.: Advances to Paul Wennberg (Caltech) for his advice on the use of the TCCON in simulating atmospheric variability with the ECMWF model: data; Martin Krol (Wageningen University) for his suggestions on From synoptic to decadal time-scales, Q. J. Roy. Meteor. Soc., the evaluation of the daily maximum CO2 ; and Britton Stephens 134, 1337–1351, 2008. (NCAR) for his comments on the mountain site evaluation which Bechtold, P., Semane, N., Lopez, P., Chaboureau, J.-P., Beljaars, A., helped improve the description of the vertical sampling strategy and and Bormann, N.: Representing equilibrium and nonequilibrium emphasise the importance of high resolution at mountain sites. convection in large-scale models, J. Atmos. Sci., 71, 734–753, 2014. www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7370 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Belikov, D. A., Maksyutov, S., Ganshin, A., Zhuravlev, R., Conway, T. J. and Thoning, K. W.: Short-term variations of atmo- Deutscher, N. M., Wunch, D., Feist, D. G., Morino, I., Parker, spheric carbon dioxide at the South Pole, Antarct. J. US, 236– R. J., Strong, K., Yoshida, Y., Bril, A., Oshchepkov, S., Boesch, 238, 1990. H., Dubey, M. K., Griffith, D., Hewson, W., Kivi, R., Mendonca, Corbin, K., Denning, A. S., Wang, J.-W., Lu, L., and J., Notholt, J., Schneider, M., Sussmann, R., Velazco, V. A., and Baker, I. T.: Possible representation errors in inversions of Aoki, S.: Study of the footprints of short-term variation in XCO2 satellite CO2 retrievals, J. Geophys. Res., 113, D02301, observed by TCCON sites using NIES and FLEXPART atmo- https://doi.org/10.1029/2007JD008716, 2008. spheric transport models, Atmos. Chem. Phys., 17, 143–157, Corbin, K. D., Denning, A. S., and Gurney, K. R.: The space https://doi.org/10.5194/acp-17-143-2017, 2017. and time impacts on U.S. regional atmospheric CO2 concen- Balzarolo, M., Boussetta, S., Balsamo, G., Beljaars, A., Maignan, trations from a high resolution fossil fuel CO2 emissions in- F., Calvet, J.-C., Lafont, S., Barbu, A., Poulter, B., Cheval- ventory, Tellus B, 62, 506–511, https://doi.org/10.1111/j.1600- lier, F., Szczypta, C., and Papale, D.: Evaluating the poten- 0889.2010.00480.x, 2010. tial of large-scale simulations to predict carbon fluxes of ter- Crisp, D., Pollock, H. R., Rosenberg, R., Chapsky, L., Lee, R. A. restrial ecosystems over a European Eddy Covariance network, M., Oyafuso, F. A., Frankenberg, C., O’Dell, C. W., Bruegge, C. Biogeosciences, 11, 2661–2678, https://doi.org/10.5194/bg-11- J., Doran, G. B., Eldering, A., Fisher, B. M., Fu, D., Gunson, M. 2661-2014, 2014. R., Mandrake, L., Osterman, G. B., Schwandner, F. M., Sun, K., Beljaars, A. and Viterbo, P.: The role of the boundary layer in a nu- Taylor, T. E., Wennberg, P. O., and Wunch, D.: The on-orbit per- merical weather prediction model, in Clear and cloudy boundary formance of the Orbiting Carbon Observatory-2 (OCO-2) instru- layers, Royal Netherlands Academy of Arts and Sciences, North ment and its radiometrically calibrated products, Atmos. Meas. Holland Publishers, Amsterdam, 1998. Tech., 10, 59–81, https://doi.org/10.5194/amt-10-59-2017, 2017. Blumenstock, T., Hase, F., Schneider, M., García, O. E., and Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, Sepúlveda, E.: TCCON data from Izana, Tenerife, Spain, Re- P., Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., lease GGG2014R1, TCCON data archive, hosted by Caltech- Bauer, P., Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bid- DATA, California Institute of Technology, Pasadena, CA, USA, lot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, https://doi.org/10.14291/tccon.ggg2014.izana01.R1, 2014. A. J., Haimberger, L., Healy, S. B., Hersbach, H., Hólm, E. V., Boussetta, S., Balsamo, G., Beljaars, A., Agustí-Panareda, A., Isaksen, L., Kallberg, P., Köhler, M., Matricardi, M., McNally, Calvet, J.-C., Jacobs, C., van den Hurk, B., Viterbo, P., La- A. P., Monge-Sanz, B. M., Morcrette, J.-J., Park, B.-K., Peubey, font, S., Dutra, E., Jarlan, L., Balzarolo, M., Papale, D., C., de Rosnay, P., Tavolato, C., Thépaut, J.-N., and Vitart, F.: The and van der Werf, G.: Natural carbon dioxide exchanges in ERA-Interim re-analysis: Configuration and performance of the the ECMWF Integrated Forecasting System: Implementation data assimilation system, Q. J. Roy. Meteor. Soc., 137, 553— and offline validation, J. Geophys. Res.-Atmos., 118, 1–24, 597, https://doi.org/10.1002/qj.828, 2011. https://doi.org/10.1002/jgrd.50488, 2013. De Mazière, M., Sha, M. K., Desmet, F., Hermans, C., Scolas, F., Brooks, B.-G. J., Desai, A. R., Stephens, B. B., Bowling, D. R., Kumps, N., Metzger, J.-M., Duflot, V., and Cammas, J.-P.: Burns, S. P., Watt, A. S., Heck, S. L., and Sweeney, C.: Assess- TCCON data from Reunion Island (La Reunion), France, ing filtering of mountaintop CO2 mole fractions for application Release GGG2014R0, TCCON data archive, hosted by Caltech- to inverse models of biosphere-atmosphere carbon exchange, At- DATA, California Institute of Technology, Pasadena, CA, USA, mos. Chem. Phys., 12, 2099–2115, https://doi.org/10.5194/acp- https://doi.org/10.14291/tccon.ggg2014.reunion01.R0/1149288, 12-2099-2012, 2012. 2014. Brunke, E.-G., Labuschagne, C., Parker, B., Scheel, H. E., and Deutscher, N., Notholt, J., Messerschmidt, J., Weinzierl, C., Whittlestone, S.: Baseline air mass selection at Cape Point, South Warneke, T., Petri, C., Grupe, P., and Katrynski, K.: TC- Africa: Application of 222 Rn and other filter criteria to CO2 , At- CON data from Bialystok, Poland, Release GGG2014R1, mos. Environ., 38, 5693–5702, 2004. TCCON data archive, hosted by CaltechDATA, Cal- Chevallier, F., Ciais, P., Conway, T. J., Aalto, T., Ander- ifornia Institute of Technology, Pasadena, CA, USA, son, B. E., Bousquet, P., Brunke, E. G., Ciattaglia, L., https://doi.org/10.14291/tccon.ggg2014.bialystok01.R1/1183984, Esaki, Y., Fröhlich, M., Gomez, A., Gomez-Pelaez, A.J., 2014. Haszpra, L., Krummel, P. B., Langenfelds, R. L., Leuen- Diamantakis, M. and Agustí-Panareda, A.: A positive definite berger, M., Machida, T., Maignan, F., Matsueda, H., Morguí, J. tracer mass fixer for high resolution weather and atmospheric A., Mukai, H., Nakazawa, T., Peylin, P., Ramonet, M., Riv- composition forecasts, ECMWF Technical Memorandum, No. ier, L., Sawa, Y., Schmidt, M., Steele, L. P., Vay, S. A., Ver- 819, available at: https://www.ecmwf.int/en/elibrary/17914- meulen, A. T., Wofsy, S., and Worthy, D.: CO2 surface fluxes positive-definite-tracer-mass-fixer-high-resolution-weather-and- at grid point scale estimated from a global 21 year reanalysis atmospheric (last access: 30 May 2019), 2017. of atmospheric measurements, J. Geophys. Res., 115, D21307, Diamantakis, M. and Magnusson, L.: Sensitivity of the ECMWF https://doi.org/10.1029/2010JD013887, 2010. model to semi-Lagrangian departure point iterations, Mon. Chevallier, F., Palmer, P. I., Feng, L., Boesch, H., O’Dell, C. Weather Rev., 144, 3233–3250, https://doi.org/10.1175/MWR- W., and Bousquet, P.: Toward robust and consistent regional D-15-0432.1, 2016. CO2 fluxestimates from in situ and spaceborne measurements Feist, D. G., Arnold, S. G., John, N., Geibel, M. C., and of atmospheric CO2 , Geophys. Res. Lett., 41, 1065–1070, Christoph, M.: TCCON data from Ascension Island (SH), https://doi.org/10.1002/2013GL058772, 2014. Release GGG2014R0, TCCON data archive, hosted by Caltech- DATA, California Institute of Technology, Pasadena, CA, USA, Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7371 https://doi.org/10.14291/tccon.ggg2014.ascension01.R0/1149285, https://doi.org/10.14291/tccon.ggg2014.wollongong01.R0/1149291, 2014. 2014. Feng, S., Lauvaux, T., Newman, S., Rao, P., Ahmadov, R., Deng, Guerlet, S., Butz, A., Schepers, D., Basu, S., Hasekamp, O.P., A., Díaz-Isaac, L. I., Duren, R. M., Fischer, M. L., Gerbig, Kuze, A., Yokota, T., Blavier, J.-F., Deutscher, N. M., Grif- C., Gurney, K. R., Huang, J., Jeong, S., Li, Z., Miller, C. E., fith, D. W. T., Hase, F., Kyro, E., Morino, I., Sherlock, V., Suss- O’Keeffe, D., Patarasuk, R., Sander, S. P., Song, Y., Wong, K. W., mann, R., Galli, A., and Aben, I.: Impact of aerosol and thin cir- and Yung, Y. L.: Los Angeles megacity: a high-resolution land– rus on retrieving and validating XCO2 from GOSAT shortwave atmosphere modelling system for urban CO2 emissions, At- infrared measurements, J. Geophys. Res.-Atmos., 118, 4887– mos. Chem. Phys., 16, 9019–9045, https://doi.org/10.5194/acp- 4905, https://doi.org/10.1002/jgrd.50332, 2013. 16-9019-2016, 2016. Gurney, K. R., Law, R. M., Denning, A. S., Rayner, P. J., Baker, D., Flemming, J., Huijnen, V., Arteta, J., Bechtold, P., Beljaars, A., Bousquet, P., Bruhwiler, L., Chen, Y.-H., Ciais, P., Fan, S., Fung, Blechschmidt, A.-M., Diamantakis, M., Engelen, R. J., Gaudel, I. Y., Gloor, M., Heimann, M., Higuchi, K., John, J., Kowal- A., Inness, A., Jones, L., Josse, B., Katragkou, E., Marecal, czyk, E., Maki, T., Maksyutov, S., Peylin, P., Prather, M., Pak, V., Peuch, V.-H., Richter, A., Schultz, M. G., Stein, O., and B. C., Sarmiento, J., Taguchi, S., Takahashi, T., and Yuen, C.-W.: Tsikerdekis, A.: Tropospheric chemistry in the Integrated Fore- TransCom 3 inversion intercomparison: 1. Annual mean control casting System of ECMWF, Geosci. Model Dev., 8, 975–1003, results and sensitivity to transport and prior flux information, Tel- https://doi.org/10.5194/gmd-8-975-2015, 2015. lus B, 55, 555–579, 2003. Francey, R. J., Steele, L. P., Spencer, D. A., Langenfelds, R. L., Law, Haiden, T., Janousek, M., Bidlot, J.-R., Ferranti, L., R. M., Krummel, P. B., Fraser, P. J., Etheridge, D. M., Derek, N., Prates, F., Vitart, F., Bauer, P., and Richardson, D.: Coram, S. A., Cooper, L. N., Allison, C. E., Porter, L., and Baly, Evaluation of ECMWF forecasts, including 2016–2017 S.: The CSIRO (Australia) measurement of greenhouse gases in upgrades, ECMWF Technical Memorandum, No. 817, the global atmosphere, report of the 11th WMO/IAEA Meeting ECMWF, available at: https://www.ecmwf.int/en/elibrary/ of Experts on Carbon Dioxide Concentration and Related Tracer 17913-evaluation-ecmwf-forecasts-includuing-2016-2017-upgrades Measurement Techniques, Tokyo, Japan, September 2001, edited (last access: 30 May 2019), 2017. by: Toru, S. and Kazuto, S., World Meteorological Organization Halter, B. and Harris, J. M.: On the variability of atmospheric car- Global Atmosphere Watch, 97–111, 2003. bon dioxide concentration at Barrow, Alaska, during winter, J. Gaudry, A., Monfray, P., Polian, G., Bonsang, G., Ardouin, B., Je- Geophys. Res., 88, 6858–6864, 1983. gou, A., and Lambert, G.: Non-seasonnal variations of atmo- Halter, B. C., Harris, J. M., and Conway, T. J.: Component signals in spheric CO2 concentrations at Amsterdam Island, Tellus B, 43, the record of atmospheric carbon dioxide concentration at Amer- 136–143, 1991. ican Samoa, J. Geophys. Res., 93, 15914–15918, 1988. Geels, C., Doney, S. C., Dargaville, R., Brandt, J., and Chris- Hase, F., Blumenstock, T., Dohe, S., Groß, J., and Kiel, M.: tensen, J. H.: Investigating the sources of synoptic variability TCCON data from Karlsruhe, Germany, Release GGG2014R1, in atmospheric CO2 measurements over the Northern Hemi- TCCON data archive, hosted by CaltechDATA, Cal- sphere continents: a regional model study, Tellus B, 56, 35–50, ifornia Institute of Technology, Pasadena, CA, USA, https://doi.org/10.1111/j.1600-0889.2004.00084.x, 2004. https://doi.org/10.14291/tccon.ggg2014.karlsruhe01.R1/1182416, Gomez-Pelaez, A. J. and Ramos, R.: Improvements in the Car- 2014. bon Dioxide and Methane Continuous Measurement Programs Haszpra, L., Barcza, Z., Bakwin, P. S., Berger, B. W., Davis, K. at Izana Global GAW Station (Spain) during 2007–2009, in: J., and Weidinger, T.: Measuring system for the long-term mon- GAW report (No. 194) of the 15th WMO/IAEA Meeting of Ex- itoring of biosphere/atmosphere exchange of carbon dioxide, J. perts on Carbon Dioxide, Other Greenhouse Gases, and Related Geophys. Res., 106, 3057–3070, 2001. Tracer Measurement Techniques (Jena, Germany; 7–10 Septem- Hatakka, J., Aalto, T., Aaltonen, V., Aurela, M., Hakola, H., Komp- ber 2009), edited by: Brand, W. A., World Meteorological Orga- pula, M., Laurila, T., Lihavainen, H., Paatero, J., Salminen, K., nization (TD No. 1553), 133–138, 2011. and Viisanen, Y.: Overview of the atmospheric research activi- Griffith, D. W. T., Deutscher, N., Velazco, V. A., Wennberg, P. O., ties and results at Pallas GAW station, Boreal Environ. Res., 8, Yavin, Y., Keppel Aleks, G., Washenfelder, R., Toon, G. C., 365–384, 2003. Blavier, J.-F., Murphy, C., Jones, N., Kettlewell, G., Connor, B., Hedelius, J. K., Feng, S., Roehl, C. M., Wunch, D., Hill- Macatangay, R., Roehl, C., Ryczek, M., Glowacki, J., Cul- yard, P. W., Podolske, J. R., Iraci, L. T., Patarasuk, R., gan, T., and Bryant, G.: TCCON data from Darwin, Australia, Rao, P., O’Keeffe, D., Gurney, K. R., Lauvaux, T., and Release GGG2014R0, TCCON data archive, hosted by Caltech- Wennberg, P. O.: Emissions and topographic effects on col- DATA, California Institute of Technology, Pasadena, CA, USA, umn CO2 (XCO2 ) variations, with a focus on the Southern https://doi.org/10.14291/tccon.ggg2014.darwin01.R0/1149290, California Megacity, J. Geophys. Res.-Atmos., 122, 7200–7215, 2014. https://doi.org/10.1002/2017JD026455, 2017. Griffith, D. W. T., Velazco, V. A., Deutscher, N., Mur- Hedelius, J. K., Liu, J., Oda, T., Maksyutov, S., Roehl, C. M., Iraci, phy, C., Jones, N., Wilson, S., Macatangay, R., Ket- L. T., Podolske, J. R., Hillyard, P. W., Liang, J., Gurney, K. R., tlewell, G., Buchholz, R. R., and Riggenbach, M.: TCCON Wunch, D., and Wennberg, P. O.: Southern California megac- data from Wollongong, Australia, Release GGG2014R0, ity CO2 , CH4 , and CO flux estimates using ground- and space- TCCON data archive, hosted by CaltechDATA, Cal- based remote sensing and a Lagrangian model, Atmos. Chem. ifornia Institute of Technology, Pasadena, CA, USA, Phys., 18, 16271–16291, https://doi.org/10.5194/acp-18-16271- 2018, 2018. www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7372 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Holm, E., Forbes, R., Lang, S., Magnusson, L., and Malardel, S.: Wofsy, S. C.: The imprint of surface fluxes and transport on vari- New model cycle brings higher resolution, ECMWF ations in total column carbon dioxide, Biogeosciences, 9, 875– Newsletter, No. 147, available at: https://www.ecmwf.int/ 891, https://doi.org/10.5194/bg-9-875-2012, 2012. en/elibrary/16299-newsletter-no-147-spring-2016 (last access: Koehler, M., Ahlgrimm, M., and Beljaars, A.: Unified treatment of 30 May 2019), 2016. dry convective and stratocumulus-topped boundary layers in the Houweling, S., Aben, I., Breon, F.-M., Chevallier, F., Deutscher, ecmwf model, Q. J. Roy. Meteor. Soc., 137, 43–57, 2011. N., Engelen, R., Gerbig, C., Griffith, D., Hungershoefer, Kuze, A., Suto, H., Nakajima, M., and Hamazaki, T.: Thermal and K., Macatangay, R., Marshall, J., Notholt, J., Peters, W., near infrared sensor for carbon observation Fourier-transform and Serrar, S.: The importance of transport model uncer- spectrometer on the Greenhouse Gases Observing Satellite for tainties for the estimation of CO2 sources and sinks using greenhouse gases monitoring, Appl. Optics, 48, 6716–6733, satellite measurements, Atmos. Chem. Phys., 10, 9981–9992, 2009. https://doi.org/10.5194/acp-10-9981-2010, 2010. Law, R. M., Peters, W., Rödenbeck, C., Aulagnier, C., Baker, I., Inness, A., Ades, M., Agustí-Panareda, A., Barré, J., Benedic- Bergmann, D. J., Bousquet, P., Brandt, J., Bruhwiler, L., tow, A., Blechschmidt, A.-M., Dominguez, J. J., Engelen, R., Cameron-Smith, P. J., Christensen, J. H., Delage, F., Den- Eskes, H., Flemming, J., Huijnen, V., Jones, L., Kipling, Z., ning, A. S., Fan, S., Geels, C., Houweling, S., Imasu, R., Massart, S., Parrington, M., Peuch, V.-H., Razinger, M., Remy, Karstens, U., Kawa, S. R., Kleist, J., Krol, M. C., Lin, S.- S., Schulz, M., and Suttie, M.: The CAMS reanalysis of at- J., Lokupitiya, R., Maki, T., Maksyutov, S., Niwa, Y., On- mospheric composition, Atmos. Chem. Phys., 19, 3515–3556, ishi, R., Parazoo, N., Patra, P. K., Pieterse, G., Rivier, L., https://doi.org/10.5194/acp-19-3515-2019, 2019. Satoh, M., Serrar, S., Taguchi, S., Takigawa, M., Vautard, R., Inoue, M., Morino, I., Uchino, O., Nakatsuru, T., Yoshida, Y., Vermeulen, A. T., and Zhu, Z.: TransCom model simulations Yokota, T., Wunch, D., Wennberg, P. O., Roehl, C. M., Grif- of hourly atmospheric CO2 : Experimental overview and diurnal fith, D. W. T., Velazco, V. A., Deutscher, N. M., Warneke, T., cycle results for 2002, Global Biogeochem. Cy., 22, GB3009, Notholt, J., Robinson, J., Sherlock, V., Hase, F., Blumenstock, https://doi.org/10.1029/2007GB003050, 2008. T., Rettinger, M., Sussmann, R., Kyrö, E., Kivi, R., Shiomi, K., Law, R. M., Steele, L. P., Drummel, P. B., and Zahorowski, W.: Kawakami, S., De Mazière, M., Arnold, S. G., Feist, D. G., Bar- Synoptic variations in atmospheric CO2 at Cape Grim: a model row, E. A., Barney, J., Dubey, M., Schneider, M., Iraci, L. T., intercomparison, Tellus B, 62, 810–820, 2010. Podolske, J. R., Hillyard, P. W., Machida, T., Sawa, Y., Tsuboi, Lin, J. C., Mallia, D. V., Wu, D., and Stephens, B. B.: How K., Matsueda, H., Sweeney, C., Tans, P. P., Andrews, A. E., Bi- can mountaintop CO2 observations be used to constrain re- raud, S. C., Fukuyama, Y., Pittman, J. V., Kort, E. A., and Tanaka, gional carbon fluxes?, Atmos. Chem. Phys., 17, 5561–5581, T.: Bias corrections of GOSAT SWIR XCO2 and XCH4 with TC- https://doi.org/10.5194/acp-17-5561-2017, 2017. CON data and their evaluation using aircraft measurement data, Lin, X., Ciais, P., Bousquet, P., Ramonet, M., Yin, Y., Balkanski, Atmos. Meas. Tech., 9, 3491–3512, https://doi.org/10.5194/amt- Y., Cozic, A., Delmotte, M., Evangeliou, N., Indira, N. K., Lo- 9-3491-2016, 2016. catelli, R., Peng, S., Piao, S., Saunois, M., Swathi, P. S., Wang, Iraci, L. T., Podolske, J., Hillyard, P. W., Roehl, C., Wennberg, P. R., Yver-Kwok, C., Tiwari, Y. K., and Zhou, L.: Simulating CH4 O., Blavier, J.-F., Allen, N., Wunch, D., Osterman, G. B., and CO2 over South and East Asia using the zoomed chemistry and Albertson, R., TCCON data from Edwards (US), Release transport model LMDz-INCA, Atmos. Chem. Phys., 18, 9475– GGG2014R1, TCCON data archive, hosted by Caltech- 9497, https://doi.org/10.5194/acp-18-9475-2018, 2018. DATA, California Institute of Technology, Pasadena, CA, USA, Lin, Y.-L.: Mesoscale Dynamics, Cambridge University Press, https://doi.org/10.14291/tccon.ggg2014.edwards01.R1/1255068, Cambridge, https://doi.org/10.1017/CBO9780511619649, 2007. 2016. Locatelli, R., Bousquet, P., Chevallier, F., Fortems-Cheney, A., Kaiser, J. W., Heil, A., Andreae, M. O., Benedetti, A., Chubarova, Szopa, S., Saunois, M., Agusti-Panareda, A., Bergmann, D., N., Jones, L., Morcrette, J.-J., Razinger, M., Schultz, M. G., Bian, H., Cameron-Smith, P., Chipperfield, M. P., Gloor, E., Suttie, M., and van der Werf, G. R.: Biomass burning emis- Houweling, S., Kawa, S. R., Krol, M., Patra, P. K., Prinn, R. sions estimated with a global fire assimilation system based G., Rigby, M., Saito, R., and Wilson, C.: Impact of transport on observed fire radiative power, Biogeosciences, 9, 527–554, model errors on the global and regional methane emissions es- https://doi.org/10.5194/bg-9-527-2012, 2012. timated by inverse modelling, Atmos. Chem. Phys., 13, 9917– Keeling, C. D., Bacastow, R. B., Bainbridge, A. E., Ekdahl, C. 9937, https://doi.org/10.5194/acp-13-9917-2013, 2013. A., Guenther, P. R., Waterman, L. S., and Chin, J. F.: Atmo- Loh, Z. M., Law, R. M., Ziehn, T., van der Schoot, M. V., Krummel, spheric carbon dioxide variations at Mauna Loa Observatory, P. B., Steele, L. P., Etheridge, D. M., Spencer, D. A., Gregory, Hawaii, Tellus, 28, 538–551, https://doi.org/10.1111/j.2153- R. L., Langenfelds, R. L., Stavert, A. R., and Thornton, D. P.: 3490.1976.tb00701.x, 1976. The Australian Greenhouse Gas Observation Network: Current Keppel-Aleks, G., Wennberg, P. O., and Schneider, T.: Sources status and vision for the future. 10th International Carbon of variations in total column carbon dioxide, Atmos. Chem. Dioxide Conference (ICDC10), 21–25 August 2017, Interlaken, Phys., 11, 3581–3593, https://doi.org/10.5194/acp-11-3581- Switzerland, available at: http://www.icdc10.unibe.ch/unibe/ 2011, 2011. portal/fak_naturwis/micro_icdc10/content/e342182/e604227/ Keppel-Aleks, G., Wennberg, P. O., Washenfelder, R. A., Wunch, e604229/files623284/Loh_Zoe.pdf (last access: 30 May 2019), D., Schneider, T., Toon, G. C., Andres, R. J., Blavier, J.-F., Con- 2017. nor, B., Davis, K. J., Desai, A. R., Messerschmidt, J., Notholt, Lopez, M., Schmidt, M., Ramonet, M., Bonne, J.-L., Colomb, A., J., Roehl, C. M., Sherlock, V., Stephens, B. B., Vay, S. A., and Kazan, V., Laj, P., and Pichon, J.-M.: Three years of semi- Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7373 continuous greenhouse gas measurements at the Puy de Dôme Morino, I., Matsuzaki, T., and Shishime, A.: TCCON data from station (central France), Atmos. Meas. Tech., 8, 3941–3958, Tsukuba, Ibaraki, Japan, 125HR, Release GGG2014R1, https://doi.org/10.5194/amt-8-3941-2015, 2015. TCCON data archive, hosted by CaltechDATA, Cal- Lu, R. and Turco, R. P.: Air pollutant transport in a coastal envi- ifornia Institute of Technology, Pasadena, CA, USA, ronment. Part I: Two dimensional simulations of sea-breeze and https://doi.org/10.14291/tccon.ggg2014.tsukuba02.R1/1241486, mountain effects, J. Atmos. Sci., 51, 2285–2308, 1994. 2014b. Malardel, S., Wedi, N., Deconinck, W., Diamantakis, M., Kühn- Necki, J., Schmidt, M., Rozanski, K., Zimnoch, M., Korus, A., lein, C., Mozdzynski, G., Hamrud, M., and Smolarkiewicz, P.: Lasa, J., Graul, R., and Levin, I.: Six-year record of atmospheric A new grid for the IFS, ECMWF Newsletter No. 146, carbon dioxide and methane at a high-altitude mountain site in available at: https://www.ecmwf.int/sites/default/files/elibrary/ Poland, Tellus B, 55, 94–104, 2003. 2016/15041-newsletter-no-146-winter-201516.pdf (last access: Newman, S., Jeong, S., Fischer, M. L., Xu, X., Haman, C. L., Lefer, 30 May 2019), 2016. B., Alvarez, S., Rappenglueck, B., Kort, E. A., Andrews, A. E., Massart, S., Agustí-Panareda, A., Heymann, J., Buchwitz, M., Peischl, J., Gurney, K. R., Miller, C. E., and Yung, Y. L.: Diur- Chevallier, F., Reuter, M., Hilker, M., Burrows, J. P., Deutscher, nal tracking of anthropogenic CO2 emissions in the Los Ange- N. M., Feist, D. G., Hase, F., Sussmann, R., Desmet, F., Dubey, les basin megacity during spring 2010, Atmos. Chem. Phys., 13, M. K., Griffith, D. W. T., Kivi, R., Petri, C., Schneider, M., 4359–4372, https://doi.org/10.5194/acp-13-4359-2013, 2013. and Velazco, V. A.: Ability of the 4-D-Var analysis of the Newman, S., Xu, X., Gurney, K. R., Hsu, Y. K., Li, K. F., Jiang, GOSAT BESD XCO2 retrievals to characterize atmospheric CO2 X., Keeling, R., Feng, S., O’Keefe, D., Patarasuk, R., Wong, K. at large and synoptic scales, Atmos. Chem. Phys., 16, 1653– W., Rao, P., Fischer, M. L., and Yung, Y. L.: Toward consis- 1671, https://doi.org/10.5194/acp-16-1653-2016, 2016. tency between trends in bottom-up CO2 emissions and top-down Miller, C. E., Crisp, D., DeCola, P. L., Olsen, S. C., Rander- atmospheric measurements in the Los Angeles megacity, At- son, J. T., Michalak, A. M., Alkhaled, A., Rayner, P., Ja- mos. Chem. Phys., 16, 3843–3863, https://doi.org/10.5194/acp- cob, D. J., Suntharalingam, P., Jones, D. B. A., Denning, A. 16-3843-2016, 2016. S., Nicholls, M. E., Doney, S. C., Pawson, S., Boesch, H., Notholt, J., Petri, C., Warneke, T., Deutscher, N., Buschmann, M., Connor, B. J., Fung, I. Y., O’Brien, D., Salawitch, R. J., Weinzierl, C., Macatangay, R., and Grupe,P.: TCCON Sander, S. P., Sen, B., Tans, P., Toon, G. C., Wennberg, P. data from Bremen, Germany, Release GGG2014R0, O., Wofsy, S. C., Yung, Y. L., and Law, R. M.: Precision re- TCCON data archive, hosted by CaltechDATA, Cal- quirements for space-based XCO2 data, J. Geophys. Res., 112, ifornia Institute of Technology, Pasadena, CA, USA, D10314, https://doi.org/10.1029/2006JD007659, 2007. https://doi.org/10.14291/tccon.ggg2014.bremen01.R0/1149275, Morcrette, J.-J., Boucher, O., Jones, L., Salmond, D., Bech- 2014a. told, P., Beljaars, A., Benedetti, A., Bonet, A., Kaiser, J. ObsPack: Cooperative Global Atmospheric Data Integra- W., Razinger, M., Schulz, M., Serrar, S., Simmons, A. J., tion Project, Multi-laboratory compilation of atmo- Sofiev, M., Suttie, M., Tompkins, A. M., and Untch, A.: spheric carbon dioxide data for the period 1968–2014, Aerosol analysis and forecast in the European Centre for obspack_co2_1_GLOBALVIEWplus_v1.0_2015-07-30, Medium-Range Weather Forecasts Integrated Forecast Sys- NOAA Earth System Research Laboratory, Global Monitoring tem: forward modeling, J. Geophys. Res., 114, D06206, Division, https://doi.org/10.15138/G3RP42, 2015. https://doi.org/10.1029/2008JD011235, 2009. O’Dell, C. W., Connor, B., Bösch, H., O’Brien, D., Frankenberg, Morguí, J. A., Àgueda, A., Batet, O., Curcoll, R., Ealo, M., Grossi, C., Castano, R., Christi, M., Eldering, D., Fisher, B., Gunson, M., C., Occhipinti, P., Sanchez-Garcia, L., Arias, R., and Rodo, McDuffie, J., Miller, C. E., Natraj, V., Oyafuso, F., Polonsky, I., X.: ClimaDat: A long-term network to study at different scales Smyth, M., Taylor, T., Toon, G. C., Wennberg, P. O., and Wunch, climatic processes and interactions between climatic compart- D.: The ACOS CO2 retrieval algorithm – Part 1: Description and ments, Geophys. Res. Abstr., EGU13-10265, EGU General As- validation against synthetic observations, Atmos. Meas. Tech., 5, sembly 2013, Vienna, Austria, 2013. 99–121, https://doi.org/10.5194/amt-5-99-2012, 2012. Morguí, J. A., Curcoll, R., and Agustí-Panareda, A.: Using the Olivier, J. and Janssens-Maenhout, G.: CO2 Emissions from CLIMADAT atmospheric CO2 dataset to evaluate (CAMS) the Fuel Combustion – 2012 Edition, IEA CO2 report 2012, performance of a transport model over the Iberian Peninsula, Part III, Greenhouse-Gas Emissions, ISBN 978-92-64-17475- 10th International Carbon Dioxide Conference (ICDC10) Theme 7, available at: https://edgar.jrc.ec.europa.eu/overview.php?v= 1.2: The contemporary carbon cycle – Emerging approaches CO2ts1990-2014 (last access: 30 May 2019), 2015. and novel observation techniques, 21–25 August 2017, In- Ott, L. E., Pawson, S., Collatz, G. J., Gregg, W. W., Men- terlaken, Switzerland, available at: http://www.icdc10.unibe. emenlis, D., Brix, H., Rousseaux, C. S., Bowman, K. W., ch/unibe/portal/fak_naturwis/micro_icdc10/content/e342182/ Liu, J., Eldering, A., Gunson, M. R., and Kawa, S. R.: As- e588294/e591299/ICDC10_Abstracts_14082017.pdf (last sessing the magnitude of CO2 flux uncertainty in atmospheric access: 30 May 2019), 2017. CO2 records using products from NASA’s Carbon Monitor- Morino, I., Yokozeki, Y., Matzuzaki, T., and Shishime, A.: ing Flux Pilot Project, J. Geophys. Res.-Atmos., 120, 734–765, TCCON data from Rikubetsu, Hokkaido, Japan, Release https://doi.org/10.1002/2014JD022411, 2015. GGG2014R1, TCCON data archive, hosted by Caltech- Parazoo, N. C., Denning, A. S., Berry, J. A., Wolf, A., Randall, D. DATA, California Institute of Technology, Pasadena, CA, USA, A., Kawa, S. R., Pauluis, S. R., and Doney, S. C.: Moist synoptic https://doi.org/10.14291/tccon.ggg2014.rikubetsu01.R1/1242265, transport of CO2 along the mid-latitude storm track, Geophys. 2017a. www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7374 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Res. Lett., 38, L09804, https://doi.org/10.1029/2011GL047238, tion of CO2 transport, Geosci. Model Dev., 11, 4489–4513, 2011. https://doi.org/10.5194/gmd-11-4489-2018, 2018. Patra, P. K., Law, R. M., Peters, W., Rödenbeck, C., Takigawa, Rozanski, K., Necki, J., Chmura, L., Sliwka, I., Zimnoch, M., M., Aulagnier, C., Baker, I., Bergmann, D. J., Bousquet, P., Bielewski, J., Galkowski, M.,Bartyzel J., and Rosiek, J.: An- Brandt, J., Bruhwiler, L., Cameron-Smith, P. J., Christensen, J. thropogenic changes of CO2 , CH4 , N2 O, CFCl3, CF2Cl2, H., Delage, F., Denning, A. S., Fan, S., Geels, C., Houwel- CCl2FCClF2, CHCl3, CH3CCl3, CCl4, SF6 and SF5CF3 mix- ing, S., Imasu, R., Karstens, U., Kawa, S. R., Kleist, J., ing ratios in the atmosphere over southern Poland, Geol. Q., 58, Krol, M. C., Lin, S.-J., Lokupitiya, R., Maki, T., Maksyu- 673–684, 2014. tov, S., Niwa, Y., Onishi, R., Parazoo, N., Pieterse, G., Riv- Sandu, I., Beljaars, A., Bechtold, P., Mauritsen, T., and Balsamo, G.: ier, L., Satoh, M., Serrar, S., Taguchi, S., Vautard, R., Ver- Why is it so difficult to represent stably stratified conditions in meulen, A. T., and Zhu, Z.: Transcom model simulations of numerical weather prediction (NWP) models?, J. Adv. Model. hourly atmospheric CO2 : analysis of synoptic scale variations Earth Sy., 5, 1–17, https://doi.org/10.1002/jame.20013, 2013. for the period 2002–2003, Global Biogeochem. Cy., 22, GB4013, Schibig, M. F., Steinbacher, M., Buchmann, B., van der Laan- https://doi.org/10.1029/2007GB003081, 2008. Luijkx, I. T., van der Laan, S., Ranjan, S., and Leuenberger, Peterson, J. T., Komhyr, W. D., Waterman, L. S., Gammon, R. H., M. C.: Comparison of continuous in situ CO2 observations at Thoning, K. W., and Conway, T. J.: Atmospheric CO2 variations Jungfraujoch using two different measurement techniques, At- at Barrow, Alaska, 1973–1982, J. Atmos. Chem., 4, 491–510, mos. Meas. Tech., 8, 57–68, https://doi.org/10.5194/amt-8-57- 1986. 2015, 2015. Pillai, D., Gerbig, C., Marshall, J., Ahmadov, R., Kretschmer, Schmidt, M., Graul, R., Sartorius, H., and Levin, L.: The Schauins- R., Koch, T., and Karstens, U.: High resolution model- land CO2 record: 30 years of continental observations and their ing of CO2 over Europe: implications for representation er- implications for the variability of the European CO2 budget, J. rors of satellite retrievals, Atmos. Chem. Phys., 10, 83–94, Geophys. Res.-Atmos., 108, 4619–4626, 2003. https://doi.org/10.5194/acp-10-83-2010, 2010. Sekiyama, T., Kunii, M., Kajino, M., and Shimbori, T.: Horizon- Pillai, D., Gerbig, C., Ahmadov, R., Rödenbeck, C., Kretschmer, tal Resolution Dependence of Atmospheric Simulations of the R., Koch, T., Thompson, R., Neininger, B., and Lavrié, J. V.: Fukushima Nuclear Accident Using 15-km, 3-km, and 500- High-resolution simulations of atmospheric CO2 over complex m Grid Models, J. Meteorol. Soc. Jpn. Ser. II, 93, 49–64, terrain – representing the Ochsenkopf mountain tall tower, At- https://doi.org/10.2151/jmsj.2015-002, 2015. mos. Chem. Phys., 11, 7445–7464, https://doi.org/10.5194/acp- Sherlock, V., Connor, B., Robinson, J., Shiona, H., Smale, D., and 11-7445-2011, 2011. Pollard, D.: TCCON data from Lauder, New Zealand, 120HR, Polavarapu, S. M., Neish, M., Tanguay, M., Girard, C., de Grandpré, Release GGG2014R0, TCCON data archive, hosted by Caltech- J., Semeniuk, K., Gravel, S., Ren, S., Roche, S., Chan, D., and DATA, California Institute of Technology, Pasadena, CA, USA, Strong, K.: Greenhouse gas simulations with a coupled meteo- https://doi.org/10.14291/tccon.ggg2014.lauder01.R0/1149293, rological and transport model: the predictability of CO2 , Atmos. 2014. Chem. Phys., 16, 12005–12038, https://doi.org/10.5194/acp-16- Shiomi, K., Kawakami, S., Ohyama, H., Arai, K., Oku- 12005-2016, 2016. mura, H., Taura, C., Fukamachi, T., and Sakashita, M.: Prather, M. J., Zhu, X., Strahan, S. E., Steenrod, S. D., and TCCON data from Saga, Japan, Release GGG2014R0, Rodriguez, S. M.: Quantifying errors in trace species trans- TCCON data archive, hosted by CaltechDATA, Cal- port modeling, P. Natl. Acad. Sci. USA, 105, 19617–19621, ifornia Institute of Technology, Pasadena, CA, USA, https://doi.org/10.1073/pnas.0806541106, 2008. https://doi.org/10.14291/tccon.ggg2014.saga01.R0/1149283, Rabier, F., Järvinen, H., Klinker, E., Mahfouf, J.-F., and Sim- 2014. mons, A.: The ECMWF operational implementation of four- Staniforth, A. and Côté, J.: Semi-Lagrangian integration dimensional variational assimilation. part I: Experimental results schemes for atmospheric models – a review, Mon. Weather with simplified physics, Q. J. Roy. Meteor. Soc., 126, 1143– Rev., 119, 2206–2223, https://doi.org/10.1175/1520- 1170, 2000. 0493(1991)119<2206:SLISFA>2.0.CO;2, 1991. Ramonet, M., Ciais, P., Aalto, T., Aulagnier, C., Chevallier, F., Stavert, A. R., Law, R. M., van der Schoot, M., Langenfelds, R. Cipriano, D., Conway, T. J., Haszpra, L., Kazan, V., Mein- L., Spencer, D. A., Krummel, P. B., Chambers, S. D., Williams, hardt, F., Paris, J.-D., Schmidt, M., Simmonds, P., Xueref- A. G., Werczynski, S., Francey, R. J., and Howden, R. T.: The Remy, I., and Necki, J.: A recent build-up of atmospheric Macquarie Island (LoFlo2G) high-precision continuous atmo- CO2 over Europe. Part 1: observed signals and possible ex- spheric carbon dioxide record, Atmos. Meas. Tech., 12, 1103– planations, Tellus B, 62, 1–13, https://doi.org/10.1111/j.1600- 1121, https://doi.org/10.5194/amt-12-1103-2019, 2019. 0889.2009.00442.x, 2010. Stephens, B. B., Miles, N. L., Richardson, S. J., Watt, A. S., Rayner, P. J., Scholze, M., Knorr, W., Kaminski, T., Gier- and Davis, K. J.: Atmospheric CO2 monitoring with single- ing, R., and Widmann, H.: Two decades of terrestrial cell NDIR-based analyzers, Atmos. Meas. Tech., 4, 2737–2748, carbon fluxes from a carbon cycle data assimilation sys- https://doi.org/10.5194/amt-4-2737-2011, 2011. tem (CCDAS), Global Biogeochem. Cy., 19, GB2026, Sussmann, R. and Rettinger, M.: TCCON data from https://doi.org/10.1029/2004GB002254, 2005. Garmisch, Germany, Release GGG2014R0, TC- Remaud, M., Chevallier, F., Cozic, A., Lin, X., and Bous- CON data archive, hosted by CaltechDATA, Cali- quet, P.: On the impact of recent developments of the fornia Institute of Technology, Pasadena, CA, USA, LMDz atmospheric general circulation model on the simula- Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/ A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather 7375 https://doi.org/10.14291/tccon.ggg2014.garmisch01.R0/1149299, port modeling and regional fossil fuel CO2 observations? Tellus 2014. B, 65, 18681, https://doi.org/10.3402/tellusb.v65i0.18681, 2013. Takahashi, T., Sutherland, S. C., Wanninkhof, R., Sweeney, C., Wang, J. S., Kawa, S. R., Collatz, G. J., Sasakawa, M., Gatti, L. V., Feely, R. A., Chipman, D. W., Hales, B., Friederich, G., Machida, T., Liu, Y., and Manyin, M. E.: A global synthesis in- Chavez, F., Sabine, C., Watson, A., Bakker, D. C. E., Schus- version analysis of recent variability in CO2 fluxes using GOSAT ter, U., Metzl, N., Yoshikawa-Inoue, H., Ishii, M., Midorikawa, and in situ observations, Atmos. Chem. Phys., 18, 11097–11124, T., Nojiri, Y., Kortzinger, A., Steinhoff, T., Hoppema, M., Olafs- https://doi.org/10.5194/acp-18-11097-2018, 2018. son, J., Arnarson, T. S., Tilbrook, B., Johannessen, T., Olsen, A., Warneke, T., Messerschmidt, J., Notholt, J., Weinzierl, C., Bellerby, R., Wong, C. S., Delille, B., Bates, N. R., and de Deutscher, N., Petri, C., Grupe, P., Vuillemin, C., Truong, Baar, H. J. W.: Climatological mean and decadal changes in F., Schmidt, M., Ramonet, M., and Parmentier, E.: TC- surface ocean pCO2 , and net sea-air CO2 flux over the global CON data from Orleans, France, Release GGG2014R0, oceans, Deep-Sea Res. Pt. II, 56, 554–577, 2009. TCCON data archive, hosted by CaltechDATA, Cal- Tang, W., Arellano, A. F., DiGangi, J. P., Choi, Y., Diskin, G. ifornia Institute of Technology, Pasadena, CA, USA, S., Agustí-Panareda, A., Parrington, M., Massart, S., Gaubert, https://doi.org/10.14291/tccon.ggg2014.orleans01.R0/1149276, B., Lee, Y., Kim, D., Jung, J., Hong, J., Hong, J.-W., Kanaya, 2014. Y., Lee, M., Stauffer, R. M., Thompson, A. M., Flynn, J. H., Wedi, N. P.: Increasing horizontal resolution in numer- and Woo, J.-H.: Evaluating high-resolution forecasts of atmo- ical weather prediction and climate simulations: illu- spheric CO and CO2 from a global prediction system during sion or panacea?, Philos. T. R. Soc. A, 372, 20130289, KORUS-AQ field campaign, Atmos. Chem. Phys., 18, 11007– https://doi.org/10.1098/rsta.2013.0289, 2014. 11030, https://doi.org/10.5194/acp-18-11007-2018, 2018. Wennberg, P. O., Wunch, D., Roehl, C., Blavier, J.-F., Toon, G. Temperton, C., Hortal, M., and Simmons, A.: A two-time-level C., Allen, N., Dowell, P., Teske, K., Martin, C., and Mar- semi-Lagrangian global spectral model, Q. J. Roy. Meteor. Soc., tin, J.: TCCON data from Lamont, Oklahoma, USA, Release 127, 111–126, 2001. GGG2014R1, TCCON data archive, hosted by CaltechDATA, Thoning, K. W., Tans, P. P., and Komhyr, W. D.: Atmospheric California Institute of Technology, Pasadena, CA, USA, carbon dioxide at Mauna Loa Observatory, 2. Analysis of the https://doi.org/10.14291/tccon.ggg2014.lamont01.R1/1255070, NOAA/GMCC data, 1974–1985, J. Geophys. Res., 94, 8549– 2016. 8565, 1989. Wennberg, P. O., Roehl, C., Wunch, D., Toon, G. C., Blavier, J.-F., Tiedtke, M.: A comprehensive mass flux scheme for cumulus pa- Washenfelder, R., Keppel-Aleks, G., Allen, N., and Ayers, rameterization in large-scale models, Mon. Weather Rev., 117, J.: TCCON data from Park Falls, Wisconsin, USA, Release 1779–1800, 1989. GGG2014R1, TCCON data archive, hosted by Caltech- Tolk, L. F., Meesters, A. G. C. A., Dolman, A. J., and Peters, W.: DATA, California Institute of Technology, Pasadena, CA, USA, Modelling representation errors of atmospheric CO2 mixing ra- https://doi.org/10.14291/tccon.ggg2014.parkfalls01.R0/1149161, tios at a regional scale, Atmos. Chem. Phys., 8, 6587–6596, 2014a. https://doi.org/10.5194/acp-8-6587-2008, 2008. Wennberg, P. O., Wunch, D., Roehl, C., Blavier, J.-F., Toon, G. Tsutsumi, Y., Matsueda, H., and Nishioka, S.: Consistency C., and Allen, N.: TCCON data from California Insti- of the CO2 primary standards in JMA, 12th WMO/IAEA tute of Technology, Pasadena, California, USA, Release meeting of experts on carbon dioxide concentration and re- GGG2014R1, TCCON data archive, hosted by Caltech- lated tracers measurement techniques (Toronto, Canada, 15– DATA, California Institute of Technology, Pasadena, CA, USA, 18 September 2003), Global Atmosphere Watch Report No. 161 https://doi.org/10.14291/tccon.ggg2014.pasadena01.R1/1182415, (WMO/TD-No.1275), 23–31, 2005. 2014b. van der Molen, M. K. and Dolman, A. J.: Regional car- Wilson, P. A.: Insight into the Carbon Cycle from Continuous Mea- bon fluxes and the effect of topography on the variabil- surements of Oxygen and Carbon Dioxide at Weybourne Atmo- ity of atmospheric CO2 , J. Geophys. Res., 112, D01104, spheric Observatory, UK, PhD thesis, University of East Anglia, https://doi.org/10.1029/2006JD007649, 2007. Norwich, UK, 2013. Verhulst, K. R., Karion, A., Kim, J., Salameh, P. K., Keeling, R. F., Worthy, D. E., Higuchi, K., and Chan, D.: North American influ- Newman, S., Miller, J., Sloop, C., Pongetti, T., Rao, P., Wong, C., ence on atmospheric carbon dioxide data collected at Sable Is- Hopkins, F. M., Yadav, V., Weiss, R. F., Duren, R. M., and Miller, land, Canada, Tellus B, 55, 105–114, 2003. C. E.: Carbon dioxide and methane measurements from the Los Wunch, D., Wennberg, P. O., Toon, G. C., Keppel-Aleks, G., Angeles Megacity Carbon Project – Part 1: calibration, urban en- and Yavin, Y. G.: Emissions of greenhouse gases from a hancements, and uncertainty estimates, Atmos. Chem. Phys., 17, North American megacity, Geophys. Res. Lett., 36, L15810, 8313–8341, https://doi.org/10.5194/acp-17-8313-2017, 2017. https://doi.org/10.1029/2009GL039825, 2009. Vermeulen, A. T., Hensen, A., Popa, M. E., van den Bulk, W. C. Wunch, D., Toon, G. C., Blavier, J.-F. L., Washenfelder, R. M., and Jongejan, P. A. C.: Greenhouse gas observations from A., Notholt, J., Connor, B., Griffith, D. W. T., Sher- Cabauw Tall Tower (1992–2010), Atmos. Meas. Tech., 4, 617– lock, V., and Wennberg, P. O.: The total carbon column 644, https://doi.org/10.5194/amt-4-617-2011, 2011. observing network, Philos. T. R. Soc. A, 369, 2087–2112, Vogel, F. R., Thiruchittampalam, B., Theloke, J., Kretschmer, R., https://doi.org/10.1098/rsta.2010.0240, 2011. Gerbig, C., Hammer, S., and Levin, I.: Can we assess a fine- Wunch, D., Wennberg, P. O., Osterman, G., Fisher, B., Naylor, B., grained emission model using high-resolution atmospheric trans- Roehl, C. M., O’Dell, C., Mandrake, L., Viatte, C., Kiel, M., Griffith, D. W. T., Deutscher, N. M., Velazco, V. A., Notholt, J., www.atmos-chem-phys.net/19/7347/2019/ Atmos. Chem. Phys., 19, 7347–7376, 2019 7376 A. Agustí-Panareda et al.: Impact of horizontal resolution on CO2 weather Warneke, T., Petri, C., De Maziere, M., Sha, M. K., Sussmann, Yang, D. X., Liu, Y., Cai, Z. N., Chen, X., Yao, L., and R., Rettinger, M., Pollard, D., Robinson, J., Morino, I., Uchino, Lu, D. R.: First global carbon dioxide maps produced O., Hase, F., Blumenstock, T., Feist, D. G., Arnold, S. G., Strong, from TanSat measurements, Adv. Atmos. Sci., 35, 621–623, K., Mendonca, J., Kivi, R., Heikkinen, P., Iraci, L., Podolske, https://doi.org/10.1007/s00376-018-7312-6, 2018. J., Hillyard, P. W., Kawakami, S., Dubey, M. K., Parker, H. A., Yu, K., Keller, C. A., Jacob, D. J., Molod, A. M., Eastham, S. D., Sepulveda, E., García, O. E., Te, Y., Jeseck, P., Gunson, M. R., and Long, M. S.: Errors and improvements in the use of archived Crisp, D., and Eldering, A.: Comparisons of the Orbiting Carbon meteorological data for chemical transport modeling: an analy- Observatory-2 (OCO-2) XCO2 measurements with TCCON, At- sis using GEOS-Chem v11-01 driven by GEOS-5 meteorology, mos. Meas. Tech., 10, 2209–2238, https://doi.org/10.5194/amt- Geosci. Model Dev., 11, 305–319, https://doi.org/10.5194/gmd- 10-2209-2017, 2017. 11-305-2018, 2018. Xueref-Remy, I., Dieudonné, E., Vuillemin, C., Lopez, M., Lac, C., Schmidt, M., Delmotte, M., Chevallier, F., Ravetta, F., Per- russel, O., Ciais, P., Bréon, F.-M., Broquet, G., Ramonet, M., Spain, T. G., and Ampe, C.: Diurnal, synoptic and seasonal variability of atmospheric CO2 in the Paris megacity area, At- mos. Chem. Phys., 18, 3335–3362, https://doi.org/10.5194/acp- 18-3335-2018, 2018. Atmos. Chem. Phys., 19, 7347–7376, 2019 www.atmos-chem-phys.net/19/7347/2019/