Land surface temperature retrieval from SDGSAT-1: assessment of different retrieval algorithms with different atmospheric reanalysis data

Abstract

Land surface temperature (LST) is an important parameter, with significant implications for climate change, urban heat island effects, and agricultural drought monitoring. The thermal infrared payload of SDGSAT-1 has three thermal infrared channels with a spatial resolution of 30 m, which is significantly finer than many existing thermal infrared satellites. This study aims to evaluate the performance of different LST retrieval methods using SDGSAT-1 thermal infrared data in conjunction with different atmospheric reanalysis datasets. The methods examined include the single channel (SC), the split window (SW) and the temperature emissivity separation (TES) method. Since the SC and TES methods require atmospheric profile data for surface temperature inversion, they were compared using two atmospheric reanalysis datasets: MERRA2 and ERA5. To compare the accuracy of the surface temperature inversion results, site data from the Heihe River Basin Observation Network were used for validation. The results indicate that the SW algorithm achieved the highest retrieval accuracy, followed by the TES algorithm, with the SC algorithm showing lower accuracy. Additionally, ERA5 outperformed MERRA2. Seasonal analysis revealed that retrieval results were generally more accurate in autumn than in summer. This study provides theoretical support for the development of LST products based on SDGSAT-1 satellite data.