Comparison of hybrid sea surface temperature (SST) with empirical regression SST in the seas around Korea

For the retrieval of satellite-derived sea surface temperature (SST), an empirical regression algorithm has been widely used so far despite its local bias. Recently, as fast radiative transfer model (RTM) enable near real-time simulations of clear-sky BTs, a hybrid SST algorithm has been suggested, which is based on regression between the incremental values (satellite-observed BT minus first-guess BT, and in-situ SST minus first-guess SST) and scaling procedures. In this study, the hybrid SST was retrieved from Communication, Ocean and Meteorological Satellite (COMS) Meteorological Imager (MI) data and compared to regression SSTs in the seas around Korea. Comparison with in-situ SST measurements showed that the hybrid SST ensured higher accuracy, especially in nighttime. As nighttime RMSE compared to in-situ SST was improved from 0.88°C to 0.38°C, RMSE for whole matchups was also improved from 0.84°C to 0.45°C for the period of May 2014.