The National Satellite Meteorological Center (NSMC)/CMA global sea surface temperature (SST) data are derived from
measurements made by the Visible and Infrared Radiometer (VIRR) on board the FY-3 series polar orbiting satellites.
Quality controlled in situ data from iQUAM (STAR/NESDIS/NOAA) is used in FY-3B/C VIRR matching procedure.
The monthly matchup database (MDB) is created from FY-3C VIRR measurements paired with coincident SST
measurements from buoys since November 2013. The satellite sensor’s brightness temperature and buoy SST pairs are
included in the MDB if they are coincident within 3km in space and 1 hour in time. Least-Square Regression is used for
estimating the first-guess coefficient and SST residuals. Outliers are removed using Median±2STD, and the final
coefficients of robust regression are estimated. A set of SST regression formalisms are tested base on NOAA-
19/AVHRR 2010 MDB. The test shows that, for daytime split-window nonlinear SST (NLSST) is the best, for nighttime
triple-window MCSST (TCSST) is the best, which is agree with STAR/NESDIS’s. The same regression analysis method
also used on FY-3C/VIRR MDB. Compare with the three daytime SST algorithms and five nighttime SST algorithms,
the best algorithm to retrieve FY-3C/VIRR SST for daytime is NLSST and for nighttime is TCSST. Compare with the
coefficients of nighttime algorithm TCSST, it shows that for FY-3B/C VIRR SST, the contribution of 3.7μm band is
smaller than split-window bands. The performance of 3.7μm band of FY-3C/VIRR is better than FY-3B/VIRR, but
worse than NOAA-19/AVHRR.