Advanced LST Retrieval Algorithms for SDGSAT-1: Enhanced Split-Window and Three-Channel Methods Under Varied Atmospheric Water Vapor Conditions

Land surface temperature (LST) is a key indicator of thermal dynamics and environmental change, with critical applications in evapotranspiration (ET) estimation, urban heat monitoring, and drought assessment. The Sustainable Development Science Satellite 1 (SDGSAT-1), with its three thermal infrared bands, offers significant potential for high-resolution LST retrieval, yet lacks established algorithms and calibration parameters. This study presents a comprehensive calibration and validation approach for LST retrieval from SDGSAT-1 Thermal Infrared Spectrometer (TIS) data using both the split-window (SW) and three-channel (TC) methods, applicable under varied atmospheric conditions, day and night. In this process, the atmospheric profile data from the Thermodynamic Initial Guess Retrieval (TIGR) dataset and observation data from the University of Wyoming were used to build LST retrieval models.
Validation was performed with in situ LST measurements from sites in China and the SURFRAD network in North America. The models achieved high accuracy, with root-mean-squared errors (RMSEs) of 2.507 K (daytime) and 2.272 K (nighttime) for the SW method, and 2.847 K (daytime) and 1.923 K (nighttime) for the TC method. Models using University of Wyoming data outperformed those with TIGR2000 profiles, underscoring the value of accurate atmospheric profiles. The proposed models are robust across varying atmospheric water vapor content and surface conditions. In this study, we provide an innovative LST retrieval solution for SDGSAT-1 TIS data, enabling high-precision temperature prediction and shedding light on climate change patterns and trends. Overall, we propose practical tools to achieve the UN Sustainable Development Goals, bolstering our understanding of climate change while advancing sustainability globally.