Validation of Sea Surface Temperature Reanalysis Product against In Situ Observations in the East Sea

Abstract

This study evaluates the performance of sea surface temperature (SST) reanalysis data against in situ observations at eight representative hydrographic stations in Vietnamese coastal and offshore waters during 2003–2022. Monthly mean SST values were compared using statistical indicators, including the Pearson correlation coefficient, Nash–Sutcliffe efficiency, mean absolute error, and mean bias. The ability of the reanalysis product to reproduce characteristic SST values, regional SST patterns, long-term warming tendency, and ENSO-related error variations was also examined. The results show that the reanalysis data reproduced monthly SST variability and seasonal cycles well at all stations, with correlation coefficients ranging from 0.869 to 0.984 and NSE values from 0.738 to 0.954. The best agreement was found at Hon Dau and Con Co, while larger absolute errors occurred at northern and north-central stations, especially Bai Chay and Sam Son, mainly due to stronger seasonal amplitude, winter cooling, and coastal influences. Southern and offshore stations generally showed smaller errors and weaker systematic bias. The regional annual mean SST fields revealed a persistent north–south thermal contrast and a warming trend of approximately 0.034 °C yr⁻¹ during 2003–2022. ENSO-based analysis showed that deseasonalized anomaly errors were highest during La Niña conditions, while bias remained small across all ENSO phases. These findings indicate that SST reanalysis data are suitable for regional SST assessment and marine climate studies, although local bias correction is recommended for quantitative applications in coastal areas with strong seasonal variability.