Coupling Hourly Land Surface Models With Spatiotemporal Variability Characteristics: A Study on FY-4B/AGRI Land Aerosol Retrieval Algorithm

摘要

The Fengyun-4B satellite is the first operational satellite of China’s new-generation geostationary meteorological satellite series. It is equipped with an advanced geostationary radiation imager (AGRI) that can detect aerosols. This method includes two aspects: 1) we constructed an hourly surface reflectance (SR) database and selected appropriate aerosol models based on seasonal variations. This approach enables the quantitative retrieval of aerosol properties over a variety of land surfaces, including dark pixels, urban areas, and bright desert surfaces, and 2) based on the spatiotemporal variability characteristics of aerosols in the past hour and within a 12-km radius, a quality control scheme was designed. This scheme was used to produce two datasets: the quality-controlled AODpure dataset and the original uncontrolled AODorig dataset. The aerosol products in 2023 were evaluated using Aerosol Robotic Network (AERONET) data. The results showed that the performance of AODpure was optimal during the summer and fall seasons, with the root-mean-square error (RMSE) less than 0.1, and more than 75% of the samples falling within the expected error (EE). However, due to the overestimation of low values in winter and underestimation of high values in spring, only 64.06% of the samples fell within the EE throughout the year. For the EE criterion, the number of samples of AODorig against AERONET was 10% lower than that of AODpure against AERONET. The diurnal variations of AODpure show more consistency with AERONET, proving the necessity of hourly SR. Both aerosol optical depth (AOD) datasets are capable of capturing dust events in spring and haze events in winter. Considering the cross-comparison, the quantitative ability of FY-4B/AGRI enhanced by this work is currently superior to that of Himawari-9/AHI.