Global assimilation of NRL MODIS aerosol optical thickness and its impact on aerosol direct radiative effect over a full year

Abstract

This study investigates the accuracy of aerosol optical thickness (AOT) forecasts and analyses during a whole year, by assimilating AOT retrievals from the Naval Research Laboratory (NRL) Moderate Resolution Imaging Spectroradiometer (MODIS) into the aerosol-coupled Non-hydrostatic ICosahedral Atmospheric Model. We explore the impact of data assimilation on aerosol direct radiative effect (DRE), taking into account the interactions between aerosol particles and radiation. Evaluation against the assimilated MODIS AOT data shows an improvement in the AOT fields. The root mean square error (RMSE) dropped from 0.027 in the free-run to 0.018 (a 33% reduction) for the forecast and to 0.017 (a 37% reduction) for the analysis, while the correlation coefficient rose from 0.640 (free-run) to 0.911 (forecast) and 0.986 (analysis), respectively. Furthermore, the most significant improvements were observed during the peak biomass burning period from August to October. This enhanced performance is further certified independently by Aerosol Robotic Network (AERONET) observations, which show a reduction in RMSE from 0.050 (free-run) to 0.038 (forecast) and 0.040 (analysis), alongside a marked rise in correlation coefficient to 0.810 and 0.884, respectively. The forecast DRE under clear-sky condition at the TOA is −2.69 ± 2.02 W/m² and at the surface is −4.04 ± 2.96 W/m². Under all-sky conditions, aerosol DRE are influenced by clouds, the forecast DRE under all-sky condition at the TOA is −1.45 ± 1.26 W/m² and at the surface is −2.74 ± 1.98 W/m².