Comparing Sentinel-5P TROPOMI NO2 column observations with the CAMS regional air quality ensemble

The Sentinel-5P TROPOspheric Monitoring Instrument (TROPOMI) instrument, launched in October 2017, provides unique observations of atmospheric trace gases at a high resolution of about 5gkm, with near-daily global coverage, resolving individual sources like thermal powerplants, industrial complexes, fires, medium-scale towns, roads, and shipping routes. Even though Sentinel-5P (S5P) is a global mission, these datasets are especially well suited to test high-resolution regional-scale air quality (AQ) models and provide valuable input for emission inversion systems. In Europe, the Copernicus Atmosphere Monitoring Service (CAMS) has implemented an operational regional AQ forecasting capability based on an ensemble of several European models, available at a resolution of 0.1g. In this paper, we present comparisons between TROPOMI observations of nitrogen dioxide (NO2) and the CAMS AQ forecasts and analyses of NO2. We discuss the different ways of making these comparisons and present quantitative results in the form of maps for individual days, summer and winter months, and a time series for European subregions and cities between May 2018 and March 2021. The CAMS regional products generally capture the fine-scale daily and averaged features observed by TROPOMI in much detail. In summer, the comparison shows a close agreement between TROPOMI and the CAMS ensemble NO2 tropospheric columns with a relative difference of up to 15g% for most European cities. In winter, however, we find a significant discrepancy in the column amounts over much of Europe, with relative differences up to 50g%. The possible causes for these differences are discussed, focusing on the possible impact of retrieval and modeling errors. Apart from comparisons with the CAMS ensemble, we also present results for comparisons with the individual CAMS models for selected months. Furthermore, we demonstrate the importance of the free tropospheric contribution to the estimation of the tropospheric column and thus include profile information from the CAMS configuration of the ECMWF's (European Centre for Medium-Range Weather Forecasts) global integrated model above 3gkm altitude in the comparisons. We also show that replacing the global 1g a priori information in the retrieval by the regional 0.1g resolution profiles of CAMS leads to significant changes in the TROPOMI-retrieved tropospheric column, with typical increases at the emission hotspots up to 30g% and smaller increases or decreases elsewhere. As a spinoff, we present a new TROPOMI NO2 level 2 (L2) data product for Europe, based on the replacement of the original TM5-MP generated global a priori profile by the regional CAMS ensemble profile. This European NO2 product is compared with ground-based remote sensing measurements of six Pandora instruments of the Pandonia Global Network and nine Multi-AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) instruments. As compared to the standard S5P tropospheric NO2 column data, the overall bias of the new product for all except two stations is 5g% to 12g% smaller, owing to a reduction in the multiplicative bias. Compared to the CAMS tropospheric NO2 columns, dispersion and correlation parameters with respect to the standard data are, however, superior.