Assessing the precision and accuracy of cloud removal and satellite angle correction techniques for SST retrieval in the South Sea with MODIS

Abstract Cloud cover and satellite angle significantly impact sea surface temperature (SST) retrievals from remote sensing imagery, yet traditional methods often overlook these factors. This study takes Moderate Resolution Imaging Spectroradiometer remote sensing imagery from the northern South China Sea in winter as a case study to investigate the effects of cloud cover and satellite angle on SST retrieval. By employing adaptive cloud detection, cloud masking, and angle correction techniques, a SST retrieval model based on the split-window algorithm was established and compared with microwave and in-situ data. The research results indicate that: (1) The established cloud removal and satellite angle correction model for SST retrieval demonstrates high precision and accuracy, with an average error of less than 0.5 °C. (2) Combination a (cloud removal + angle correction), Combination b (non-cloud removal + angle correction), and Combination c (non-cloud removal+ non-angle correction) have an average error of −0.916 °C, −0.311 °C, −1.047 °C respectively. It is evident that the inversion results of the proposed model (Combination a) exhibit a higher level of agreement with the measured data compared to t Combination b and Combination c. (3) The average error of the microwave inversion data (combination x) is 0.560 °C; Furthermore, combination x shows a different temperature curve trend compared to the measured data, whereas combination a aligns more closely with the temperature curve trend of the measured data. Accurate retrieval of SST is of great significance for understanding large-scale oceanic circulation systems, boundary currents, eddies, and ocean currents, as well as studying seasonal and interannual variations in marine ecosystems.