FY-3C Global Daily 1km Seamless Land Surface Temperature Dataset

All-weather, seamless, high-spatiotemporal-resolution land surface temperature (LST) is a key parameter for modelling land surface processes and monitoring global climate change. Thermal infrared (TIR) data from the Fengyun-3C (FY-3C) satellite suffer from significant data gaps due to interference from clouds and rain, while passive microwave products face challenges of low spatial resolution and gaps in orbital coverage. To address these limitations, this study proposes an all-weather LST generation framework integrating microwave-optical synergy and machine learning based on FY-3C observations. The framework incorporates ERA5 reanalysis data as physical constraints to reconstruct a spatiotemporally continuous microwave physical background field. By synergistically leveraging the all-weather capability of the Microwave Radiation Imager (MWRI) and the high spatial resolution of the Visible and InfraRed Radiometer (VIRR), the framework employs the eXtreme Gradient Boosting (XGBoost) algorithm and Geographically Weighted Regression (GWR) to extract nonlinear mappings between high-resolution vegetation (NDVI), topography (DEM), and the microwave background. This approach achieves the generation of global seamless daily LST at a 1 km resolution. Independent validation using in-situ data from the Heihe River Basin and the AmeriFlux network demonstrates that the model exhibits high physical fidelity and robustness across heterogeneous surfaces, including wetlands, deserts, and forests. The validation yields an all-weather root-mean-square error of approximately 5.8 K and correlation coefficients exceeding 0.83. This research significantly enhances the application potential of domestic Fengyun satellites in global environmental monitoring.