Accuracy Evaluation of the FY-4A AGRI Land Surface Temperature Product

Land surface temperature (LST) plays a key role in surface-atmosphere interactions and energy exchange processes and is an important parameter indispensable for earth science research. The LST accuracy retrieved from the Advanced Geosynchronous Radiation Imager (AGRI) onboard China's geostationary meteorological satellite FY-4A has not been well evaluated, which affects its further applications. In this article, the accuracy of AGRI land surface temperature products is evaluated by a direct verification method using land surface temperature data observed at meteorological stations in China. On this basis, the angle correction kernel model is used to perform angle correction for AGRI LST products by comparing the angle difference between AGRI and moderate-resolution imaging spectroradiometer (MODIS) sensor imaging moments, and MOD11A1 products in central China are selected to cross-validate the accuracy of AGRI LST products. The results show that the spatial and temporal distributions of AGRI land surface temperature and meteorological station observations are consistent, and the accuracy of AGRI LST differs somewhat in different seasons, with the lowest correlation of 0.68 and root-mean-square error (RMSE) of 10.92 K in Summer, and 0.89 and 6.89 K in Winter. The correlation between AGRI LST and MOD11A1 LST before angle correction is 0.64, and the RMSE is 5.45 K. After angle correction, the correlation increases to 0.90, and the RMSE decreases by 2.12 K. There are differences in the angle correction results for various land cover types and different terrains, and the accuracy of AGRI LST at the time of the ascending track (nighttime) is higher than that of the descending track (daytime). The overall results of direct verification and cross-validation indicate that FY-4A AGRI LST product has high accuracy and can accurately express the spatial and temporal distribution characteristics and variation patterns of land surface temperature.