Estimation of all-weather land surface temperature through correcting cloud-shadowing bias simulated by hourly cloud information

Fusion of thermal infrared (TIR) and passive microwave (PMW) observations are important techniques for estimating all-weather land surface temperature (LST) of relatively higher resolutions. For such fusion techniques under cloudy weather, PMW observations are especially useful in correcting the cloud -shadowing bias of the real cloudy LST against a theoretically clear -sky one in the daytime. However, PMW observations have very limited transit times in a day, and cannot be fused with asynchronous TIR data as LST can vary significantly during the daytime. To overcome this flaw, we developed a model for simulating the variation of cloud -shadowing bias based on cloud total amount (CTA) information observed by geostationary satellites. As constant coefficients of the model were found to have dependence on space and time, they were dynamically fitted for each pixel in each month using cloud -shadowing bias calculated by TIR and synchronous PMW data in the early -afternoon (1:30 P. M.). The model was then used to provide middle -morning (10:30 A.M.) cloud -shadowing bias estimates, based on which, all-weather LST at 1 km were produced at that specific moment. Validation results show that reconstructed cloudy LST in the middle -morning, with no synchronous PMW observations, has its overall RMSE values against ground measurements and real remote sensing data under cloud between 2.4 and 4.0 K. This accuracy result is generally comparable to that obtained through fusing information from synchronous PMW data in the early -afternoon moment. Therefore, it is prospective to be further applied for generating high -resolution allweather LST with much shorter re -visit cycles than the satellite PMW data.