Geoscientific Model Development | 2021
Variational regional inverse modeling of reactive species emissions with PYVAR-CHIMERE-v2019
Abstract
Abstract. Up-to-date and accurate emission inventories for air pollutants are\nessential for understanding their role in the formation of tropospheric\nozone and particulate matter at various temporal scales, for anticipating\npollution peaks and for identifying the key drivers that could help mitigate\ntheir concentrations. This paper describes the Bayesian variational inverse\nsystem PYVAR-CHIMERE, which is now adapted to the inversion of reactive\nspecies. Complementarily with bottom-up inventories, this system aims at\nupdating and improving the knowledge on the high spatiotemporal variability\nof emissions of air pollutants and their precursors. The system is designed\nto use any type of observations, such as satellite observations or surface\nstation measurements. The potential of PYVAR-CHIMERE is illustrated with\ninversions of both carbon monoxide (CO) and nitrogen oxides (NOx) emissions in Europe, using the MOPITT and\nOMI satellite observations, respectively. In these cases, local increments\non CO emissions can reach more than +50\u2009%, with increases located mainly\nover central and eastern Europe, except in the south of Poland, and\ndecreases located over Spain and Portugal. The illustrative cases for\nNOx emissions also lead to large local increments (>\u200950\u2009%), for example over industrial areas (e.g., over the Po Valley) and\nover the Netherlands. The good behavior of the inversion is shown through\nstatistics on the concentrations: the mean bias, RMSE, standard deviation,\nand correlation between the simulated and observed concentrations. For CO,\nthe mean bias is reduced by about 27\u2009% when using the posterior emissions,\nthe RMSE and the standard deviation are reduced by about 50\u2009%, and the\ncorrelation is strongly improved (0.74 when using the posterior emissions\nagainst 0.02); for NOx, the mean bias is reduced by about 24\u2009% and the\nRMSE and the standard deviation are reduced by about 7\u2009%, but the\ncorrelation is not improved. We reported strong non-linear relationships\nbetween NOx emissions and satellite NO2 columns, now requiring a\nfully comprehensive scientific study.\n