Assimilating FY-4B satellite aerosol data to improve PM₂.₅ and surface shortwave radiation prediction

Aerosols significantly impact solar radiation forecasts, but substantial uncertainties persist in numerical models due to inadequate aerosol representation. This study develops a data assimilation framework integrating hourly surface-level particulate matter (PM₂.₅) retrievals from the FY-4B geostationary satellite into the WRF-Chem model with solar radiation diagnostics via the Gridpoint Statistical Interpolation (GSI) three-dimensional variational (3DVAR) system. Assimilation experiments conducted over central and eastern China in November 2022 demonstrate marked improvements in PM₂.₅ forecasts, with correlation coefficients increasing from 0.39 to 0.82, and root mean square error (RMSE) decreasing by approximately 45%. Improved aerosol initial conditions significantly reduce uncertainties in surface downward shortwave radiation (SWDOWN) predictions, lowering midday bias by over 50% and RMSE by roughly 40% across the domain. Consistent forecast enhancements were verified through spatiotemporal analyses across various pollution levels. These results highlight the practical value of assimilating hourly FY-4B PM₂.₅ retrievals for simultaneously improving air quality and solar radiation forecasts. The proposed assimilation approach offers a robust, replicable solution for near-real-time operational forecasting, thereby supporting photovoltaic energy planning and effective air quality management.