Permafrost is one of the crucial components of the cryosphere, covering about 25% of the global continental area. The active layer thickness (ALT), as the main site for heat and water exchange between permafrost and the external atmosphere, its changes significantly impact the carbon cycle, hydrological processes, ecosystems, and the safety of engineering structures in cold regions. This study constructs a Stefan CatBoost-ET (SCE) model through machine learning and Blending integration, leveraging multi-source remote sensing data, the Stefan equation, and measured ALT data to focus on the ALT in the Qinghai-Tibet Plateau (QTP). Additionally, the SCE model was verified via ten-fold cross-validation (MAE: 20.713 cm, RMSE: 32.680 cm, R²: 0.873, and MAPE: 0.104), and its inversion of QTP’s ALT data from 1958 to 2022 revealed 1998 as a key turning point with a slow growth rate of 0.25 cm/a before 1998 and a significantly increased rate of 1.26 cm/a afterward. Finally, based on multiple model input factor analysis methods (SHAP, Pearson correlation, and Random Forest Importance), the study analyzed the ranking of key factors influencing ALT changes. Meanwhile, the importance of Stefan equation results in SCE model is verified. The research results of this paper have positive implications for eco-hydrology in the QTP region, and also provide valuable references for simulating the ALT of permafrost.

As a crucial component of the cryosphere, snow exhibits high sensitivity to climate change and exerts a significant influence on the hydrological cycle and ecological framework. Snow depth is one of the important information about snow, so studying high-resolution snow depth data is of great significance for regional climate, hydrology, and disasters. Utilizing passive microwave data from AMSR-2 and optical remote sensing data from SSE mod FSC, along with DEM and Land Cover Type datasets, the Extreme Gradient Boosting (XGBoost) model is employed to leverage the advantages of multi-source remote sensing information. Subsequently, The spatial-environmental-XGBoost (SE-XGB) model is constructed for the downscaling retrieval of snow depth. In the Arid Region of Northwest China, the inversion results of SE-XGB model were validated against measured snow depth at the weather station. The validation yields an R of 0.912, RMSE of 2.65 cm, and MAE of 0.81 cm. The snow depth estimated by the SE-XGB algorithm exhibits excellent agreement with the measured data. This study validated the snow inversion accuracy of four major land cover types, showing that Bare Land had the highest accuracy. In contrast, the inversion accuracies for Cultivated Land, Urban Land, and Grass Land exhibit roughly comparable performance, yet all three demonstrate significantly lower accuracy than Bare Land. The validation regions selected for this study include northern Xinjiang and Northeast China. At the same time, the SE-XGB algorithm is verified in the northern Xinjiang Line 1 and line 2 of China snow survey. The SE-XGB algorithm exhibits enhanced accuracy in terms of RMSE and MAE, as well as significantly reduced production time, when compared to the down-scale snow depth inversion algorithm Spatial Dynamic Downscaling (SDD) algorithm employed by similar products. The SE-XGB algorithm employed in this study exhibits remarkable performance in discriminating snow and achieves high precision in retrieving snow depth. This leads to a more detailed spatial distribution of snow information, effectively eliminating the large area block structure, and providing valuable references for the development of global downscaled snow depth datasets.