Abstract: Coastal sea surface temperature (SST) supports operational applications such as red-tide monitoring, fishery-resource assessment, marine-disaster forecasting and data assimilation, for which product accuracy and time-series stability are decisive. HY-1C and HY-1D carry the same Chinese Ocean Colour and Temperature Scanner (COCTS) and are a crucial source of China's independent SST remote sensing. However, owing to on-orbit instrument drift and anomalies in the cloud/quality threshold tests, HY-1C SST had long exhibited a systematic cold bias varying with SST, latitude and season, and the operational retrieval algorithm underwent an upgrade in October 2025. To quantify the improvement independently, we use 2025 matchups with 32 buoys over the Chinese coastal seas, adopt European Centre for Medium-Range Weather Forecasts Reanalysis V5 (ERA5) SST as an independent reference, and apply a matched-temperature method that compares pre-upgrade (January–September) and post-upgrade (October–December) errors based on the matched-temperature method, thereby removing the seasonal cycle from the assessment. After the upgrade of the operational retrieval algorithm, the root mean square error (RMSE) reached 1.35 °C and 1.22 °C for HY-1C and HY-1D, respectively, leaving no substantial gap between the two satellites, and the robust standard deviation of the clear-sky core fell below 1 °C; the accuracy nevertheless still trails that of infrared sensors equipped with mid-infrared channels. At equal SST values the robust standard deviation decreased in every interval and the ERA5 cross-validation reproduced the buoy results, confirming that the improvement originated from the algorithm, not from seasonal variations. A statistical post-processing correction removed only part of the systematic bias and proved far less effective than the source-level upgrade. Since the post-upgrade data cover only autumn and winter, the retrieval accuracy under high-SST conditions still requires further validation with summer observations.

Key words: sea surface temperature, HY-1 satellite, coastal remote sensing, matched-temperature method, radiometric drift, split-window retrieval