Improved hybrid algorithm for land surface temperature retrieval from Chinese GF-5B satellite

Land surface temperature (LST) is a key parameter in understanding surface–atmosphere energy exchanges. Conventional temperature and emissivity separation (TES) and split-window (SW) algorithms often suffer from limited accuracy when applied to surfaces with strong spectral contrast or complex atmospheric conditions. To overcome these issues, this study proposes a Maximum Emissivity Approximation OSTES (MEAOSTES) algorithm that integrates the generalized radiance-based split-window (GRSW) model with the optimized smoothing for temperature and emissivity separation (OSTES) framework. The MEAOSTES introduces a maximum-emissivity adjustment and iterative rollback mechanism to enhance the coupling between LST and emissivity retrievals. Furthermore, in this paper, the SW-TES algorithm is extended to the SW-OSTES algorithm, and the retrieval accuracy of different combination schemes of the SW and OSTES algorithms is discussed. Using data from the GF-5B satellite, which provides four thermal infrared channels at 40 m spatial resolution, the algorithms’ performance were evaluated against simulated datasets, in-situ measurements, and the MODIS MOD21 LST product. The results show that the MEAOSTES achieves the best overall accuracy, root-mean-square errors (RMSEs) of 0.92 K and 1.77 K for simulation and in-situ validation, outperforming both the OSTES and SW-OSTES approaches. The proposed algorithm effectively reduces sensitivity to atmospheric parameters and maintains consistent performance across surfaces with varying spectral contrast. These improvements demonstrate the robustness of MEAOSTES for high-resolution LST retrieval and provide insights for future hybrid algorithm development.