FY-3C Global Daily 1 km All-Weather Land Surface Temperature Sample Dataset

All-weather land surface temperature (LST) at high spatio-temporal resolution is essential for environmental modelling. Thermal infrared observations from the Fengyun-3C (FY-3C) satellite are compromised by cloud cover, while passive microwave products remain limited by coarse spatial resolution and strip-orbit gaps. Here we present CAS-LST, a cascaded spatiotemporal reconstruction framework that generates global daily 1-km all-weather LST from FY-3C observations. The framework reconstructs a 25-km microwave LST background constrained by ERA5 reanalysis and corrects biases through cumulative distribution function matching. Geographically weighted regression downscales the corrected product before fusion with clear-sky thermal infrared observations to produce the final 1-km all-weather LST. Independent validation against Heihe River Basin and AmeriFlux networks shows robust performance across heterogeneous landscapes (RMSE ≈ 5.8 K; R² > 0.83), closely tracking diurnal fluctuations and seasonal freeze–thaw transitions. This work establishes FY-3C as a viable global all-weather source of LST and provides an open, reproducible dataset.