Integrating Split-Window and Temperature-Emissivity Separation Algorithms for Hourly Land Surface Temperature Retrieval From GOES-16 ABI Data

Land surface temperature (LST) serves as a critical biophysical parameter characterizing global-scale surface energy partitioning and water exchange processes. The Split-Window (SW) and Temperature-Emissivity Separation (TES) methods are the most widely used LST retrieval methods, but they rely on accurate land surface emissivity and atmospheric correction as prior knowledge, respectively. While the SW-TES hybrid method mitigates these limitations, atmospheric correction errors in the SW stage can accumulate into the TES stage, thereby affecting the accuracy of LST retrieval. To address this limitation, this study develops an improved SW-TES algorithm that incorporates total column water vapor as a correction factor for surface-leaving radiance calculations. Based on GOES-16 ABI data, we utilized the improved algorithm to retrieve hourly LST estimates for the contiguous United States in 2020 and validated the results against the GOES-16 LST product and ground-based measurements from the Surface Radiation Budget Network (SURFRAD). The retrieved LST exhibits high consistency with the GOES-16 LST product, demonstrating a root mean square error (RMSE) of 1.48 K and a bias of 0.90 K. Based on SURFRAD ground observations, the improved SW-TES algorithm achieves an average RMSE of 2.27 K and a bias of 0.26 K across seven validation sites. In comparison, the GOES-16 LST product shows an RMSE of 2.86 K and a bias of –1.21 K at the same sites. Furthermore, the algorithm maintains superior stability under high temperature and high water vapor content conditions.