Sea Surface Temperature Derived From FY-4A/AGRI

• Published in: IEEE journal of selected topics in applied earth observations and remote sensing 2024, Vol.17, p.14237-14247
• Main Authors: Yang, Chang, Guan, Lei
• Format: Article
• Language: English
• Subjects: Advanced Geosynchronous Radiation Imager (AGRI); Algorithms; Approximation algorithms; Atmospheric modeling; Bias; Brightness temperature; Clouds; Daytime; Fengyun-4A; latitude segmentation; multichannel algorithm; Night; Night-time; Nighttime; Orbits; Radiometry; retrieval coefficients; Root-mean-square errors; Satellite broadcasting; Satellites; Sea surface temperature; sea surface temperature (SST); Standard deviation; Surface radiation temperature; Water vapor; Water vapour; Wavelength
• ISSN: 1939-1404; 2151-1535
• DOI: 10.1109/JSTARS.2024.3439881

In this study, we develop an improved algorithm for retrieving sea surface temperature (SST) from the Advanced Geosynchronous Radiation Imager (AGRI) on the Fengyun-4A satellite (FY-4A). First, we use a multichannel nonlinear SST algorithm that combines data from the 3.7, 8.5, 10.7, and 12 µm channels during the nighttime, while during the daytime it combines data from the three long-wavelength bands centered at 8.5, 10.7, and 12 µm. Second, to minimize the impact of water vapor and obtain more accurate SST, we provide different retrieval coefficients obtained from the in situ SST and the observed brightness temperature for different latitude regions and different time periods using two-thirds of the matchups from January 2019 to December 2021. We validate the retrieved FY-4A/AGRI SST and operational FY-4A/AGRI SST by comparing them with in situ SST using one-third of the matchups from January 2019 to December 2021. Compared with the in situ data, the full-disk retrieved AGRI SST has the bias, median, standard deviation (STD), robust standard deviation (RSD), and root mean square error (RMSE) of 0.01, 0.03, 0.59, 0.52, and 0.59 K in the daytime, respectively. In nighttime, the bias, median, STD, RSD, and RMSE are 0.02, 0.05, 0.63, 0.55, and 0.63 K, respectively. Our analyses of the results further demonstrate that the improved algorithm significantly improves the accuracy compared with the operational AGRI SST, correcting the large bias in the temporal and spatial scales and effectively accounting for the effect of water vapor and satellite zenith angle.