Abstract:The sea surface temperature (SST) was retrieved by nonlinear algorithm for Visible and Infra Red Radiometer (VIRR) onboard the Chinese Fengyun 3A (FY 3A) polar orbiting meteorological satellite. In this paper, the matchup dataset was created by the SST measurements from global ship observation and FY 3A/VIRR data in 2010, and the coefficients for the nonlinear SST (NLSST) equation applicable to FY 3A/VIRR data were derived by using multiple linear regression, which could be used to retrieve accurate SST products for FY 3A/VIRR data. An independent matchup dataset was used to assess the accuracy of NLSST algorithm by linear model using a robust least absolute deviation method, and the result showed that the biases were 0.05℃ and -0.05℃ for daytime and nighttime, respectively. The absolute deviation was less than 0.5℃ and the standard deviation was less than 0.65℃. The VIRR SST was calculated by the SST algorithm presented in this paper to contrast with official MODIS SST products, showing that there was a good correlation between VIRR SST and MODIS SST. All these have indicated that the SST algorithm realized in this paper could provide reliable VIRR SST products to ocean and climate variability studies.