Retrieval of All-Sky Land Surface Temperature Considering Penetration Effect Using Spaceborne Thermal and Microwave Radiometry

Thermal infrared (TIR) remote sensing (RS) has been widely adopted for monitoring land surface temperature (LST). However, its application has been limited to cloud-free conditions, resulting in a need for LST retrieval methods that combine microwave (MW) and TIR channels. This is especially crucial in areas frequently covered by clouds. One limitation of the current LST retrieval methods is the absence of considering the penetration effect (PE) of MW, which leads to great uncertainty in barren and sparsely vegetated areas. To address this issue, this study proposes a new perspective that considers the PE to merge the LST retrieved from MW and TIR channels. The soil temperature integral equation is simplified based on the soil temperature and water content profiles. Consequently, a PE-based model is developed to convert the effective soil temperature into LST and merge the LST estimated from passive MW observations with those from moderate resolution imaging spectroradiometer (MODIS) LST products. The model considering PE performs better than the method that does not consider PE, as demonstrated by higher
$R$
and lower root-mean-square error (RMSE) values. The PE-based model is then applied to Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) data, and the estimated LST is found to fit well with the MODIS LST product (
$R$
= 0.91). Using this model, an all-sky LST is retrieved by merging passive MW observations and MODIS LST products. Validation of the model at eight ground-based stations over the Tibetan Plateau (TP) demonstrates its reasonable accuracy in both clear-sky and cloudy conditions.