A Framework for Estimating All-Weather Land Surface Temperature and Sea Surface Temperature

Earth’s surface temperature (EST), encompassing both land surface temperature (LST) and sea surface temperature (SST), serves as a crucial indicator of climate change. This study introduces a groundbreaking framework for the daily estimation of Earth’s Surface Temperature (EST), integrating reanalysis data with thermal infrared remote sensing data merging (RTM) techniques and employing machine learning methods. The spatial distribution of the generated all-weather EST aligns effectively with MODIS EST, showcasing its capability to recover EST values in cloudy regions and estimate missing values in orbital gap areas. Validation results for all-weather LST and SST demonstrate commendable accuracy, with minimal variations observed under both clear-sky and cloudy conditions. The Root Mean Square Error (RMSE) for LST ranges from 1.69 to 2.84 K, while for SST, it spans from 0.38 °C to 0.59 °C. The framework exhibits adaptability to diverse weather conditions, maintaining consistent relative trends across different geographical locations. In summary, this innovative approach provides a robust solution for generating all-weather ESTs, effectively addressing challenges associated with conventional Thermal Infrared (TIR) data.