Evaluating the impact of aerosol treatments in TROPOMI NO2 retrievals on top-down NOx emission inversions over Eastern China

Accurate quantification of anthropogenic nitrogen oxides (NOx) emissions is essential for understanding complex regional air pollution and evaluating emission reduction policies. Top-down inversion has become an important method for updating emission inventories; however, comprehensive assessments of how aerosol treatment differences in satellite inversion algorithms affect inferred emission estimates remain limited. This study utilized the GEOS-Chem adjoint model to assimilate both the operational Sentinel-5P/TROPOMI (S5P-OFFL) and Peking University POMINO NO2 products to optimize anthropogenic NOx emissions over eastern China in January and July 2023. The January VCD difference between POMINO and S5P-OFFL increased with aerosol optical depth (R2 = 0.51), whereas no relationship occurred in July (R2 < 0.01), indicating stronger aerosol-related retrieval effects in winter. Against MEIC-2023, the January S5P-OFFL inversion reduced the regional emission NMB to 30.2%, while POMINO preserved the grid-scale pattern more closely, increasing R from 0.67 to 0.80. In July, POMINO reduced the regional NMB by 18.7 percentage points to 10.0%, with similar spatial correlations for the two inversions. The POMINO-driven January posterior agreed more closely with OMI, achieving R = 0.85 and NMB =� 25.1%; the corresponding S5P-OFFL values were 0.81 and � 36.4%. POMINO also better reproduced winter NO2 clustering. A common-weight sensitivity experiment reduced the positive emission adjustments in the POMINO inversion, demonstrating the influence of observation errors and regularization. Explicit aerosol treatment therefore preserved useful NO2 information under high aerosol loading, while observation weighting, regularization, and model structure jointly controlled posterior emission magnitude.