WRF-Chem CO2 simulation over a medium sized city: An evaluation across grey-zone resolutions

Metropolitan areas are known to be anthropogenic hot spots of Greenhouse Gas (GHG) fluxes. To track the effectiveness of climate mitigation policies and emission reduction objectives, large metropolitan areas like Munich and Paris regions are currently being instrumented with dense atmospheric GHG networks, further assimilated in inversion systems with high-resolution inventories, also complementing the data collected by remote sensing instruments on the ground and in space. To study medium-sized cities, where a large fraction of the global population lives, spaceborne measurements often fail to quantify fossil fuel emissions since the atmospheric signatures are below the detection threshold of current instruments. For the past two years (2022 and 2023), two large-scale campaigns of the MAGIC initiative led by CNRS and CNES (https://magic.aeris-data.fr) have been taking place in Reims, France, a city with a population of 300,000 inhabitants (207 hab./km2) located to the East of Paris (approx. 100 km away). During these two intensive measurement campaigns, a wide range of ground-based instruments have been deployed around the city to measure CO2 concentrations, in addition to instrumented balloons and aircraft. The goal of these campaigns was to evaluate CO2 emissions from the area and to assess the detection capabilities of current satellite instruments.In our study, we simulated the atmospheric CO2 mixing ratios using the Weather Research Forecast model coupled to a chemistry transport model (WRF-Chem) at 4 horizontal resolutions (9 km, 3 km, 1 km, and 333 m). Typically, mesoscale models are used for resolutions coarser than 1 km while microscale Large-Eddy Simulation models (LES) are used for resolutions finer than 100m. In between, i.e. the grey-zone, turbulent motions are not resolved explicitly but high resolutions might offer a better representation of fine plume structures. Here, we present the results of a multi-scale multi-instrument comparison between the model and the observations to characterize the model performances and the ability of the model to reproduce the observed variations in concentrations. We found that the detectability of the various CO2 plumes remains challenging. First, the strength of the anthropogenic signals from the city remains low compared to gradients from nearby sources, whether industrial or metropolitan, hence making the city plume hard to study. We also showed that improvements in the modelling of CO2 plumes were not significant between the 1 and 0.3 km horizontal resolution scales, thus suggesting that LES models might be better suited for such studies.