Innovative hybrid algorithm for simultaneous land surface temperature and emissivity retrieval: Case study with SDGSAT-1 data

The split-window (SW) and temperature-and-emissivity separation (TES) algorithms have been widely used for land surface temperature (LST) estimation from thermal infrared (TIR) observations for various missions. However, the SW algorithm requires prior estimates of land surface emissivity (LSE). The TES algorithm encompasses an atmospheric correction module, which increases the complexity and uncertainty of operational LST retrieval. To address this, we proposed a split-window-driven temperature-and-emissivity separation (SWDTES) algorithm in this study to estimate LST and LSE simultaneously without the need of atmospheric correction by combining the respective advantages of SW and TES. The inputs to the SWDTES algorithm are largely simplified, which only requires atmospheric water vapor content (AWVC) apart from the top-of-atmosphere TIR radiance. The developed SWDTES algorithm was applied to the high spatial resolution Thermal Infrared Spectrometer (TIS) data from the newly launched Sustainable Development Science Satellite-1 (SDGSAT-1) mission, and its performance was assessed using the MODIS data and ground measurements. The cross validation shows that the correlation coefficient (r), bias and root mean square error (RMSE) between MODIS-converted LSE and retrieved LSE using the SWDTES algorithm for the nighttime case is 0.904, −0.033 and 0.038 for band 1; 0.677, −0.008 and 0.014 for band 2; and 0.576, −0.000 and 0.008 for band 3, indicating a good consistency between the two LSE estimates. In addition, the evaluation using ground measurements shows that the r, bias and RMSE between the in-situ LST and retrieved LST using the SWDTES algorithm are 0.99, −0.67 K and 2.10 K, respectively. Compared to the OSW and TES algorithms, the SWDTES algorithm reduces the RMSE by 0.34 K and 0.90 K, respectively, indicating an improvement in LST retrieval accuracy. We conclude that the proposed SWDTES algorithm can achieve high-accuracy and high-resolution LST retrieval from the SDGSAT-1 mission, supporting fine-scale applications in energy, water, and carbon cycle modeling.