A 1 km daily soil moisture dataset over the Qinghai-Tibet Plateau (2001-2010)

Soil moisture is the key variable of water and energy cycle, but the long-term, high-resolution soil moisture data with high accuracy is still relatively lacking on the Qinghai-Tibet Plateau. Therefore, we provide the 1 km seamless daily soil moisture data over the Qinghai-Tibet Plateau during 2001-2010 (named as BTCH). Firstly, several predictors including the vegetation index (NDVI and EVI), land surface temperature (LST), evapotranspiration, precipitation, topography (DEM, aspect, slope, TWI), soil properties and three soil moisture related indices (SWCI SIWSI VSDI) were utilized. Five machine/deep learning methods including the artificial neural network (ANN), convolutional neural network (CNN), residual neural networks (ResNet), the long short-term memory network (LSTM) and XGBoost were trained for each year taking the ESA CCI soil moisture data as target. Then the Bayesian three-cornered hat method was adopted for model integration and the final dataset was generated. Evaluaion against four in-situ measurement networks shows that the BTCH dataset has relativey high accuracy both as station and network scales with mean unbiased RMSE values of 0.048 m3/m3 and 0.034 m3/m3. The dataset can be used for various regional hydrological, meteorological, ecological analysis and modeling.