Global surface snowfall rate estimation is crucial for hydrological and meteorological applications but is still a challenging task. We present a novel approach to comprehensively consider passive microwave, infrared and physical constraints using deep neural networks with attention module for retrieving surface snowfall rate, namely PCSSR-DNNWA. PCSSR-DNNWA outperforms traditional approaches in predicting surface snowfall rate with CC ~ 0.75, ME ~ -0.03 mm/h, and RMSE ~ 0.21 mm/h. In addition, we found that graupel water path (GWP) is of vital importance with largest contributions in retrieving surface snowfall rate. Integrating the physical constraints, PCSSR-DNNWA paves a new avenue for retrieving satellite-borne surface snowfall rate by intelligently considering the varying importance of the multiple predictors, resulting in increased accuracy, interpretability, and computational efficiency.

The upper Tarim River basin is supporting about 50 million people by melting the glaciers and snow, which are highly vulnerable and sensitive to climate change. Therefore, assessing the relative effects of climate change on runoff of this region is essential not only for understanding the mechanism of hydrological response over the mountainous areas in Southern Xinjiang but also for local water resources management. This study quantitatively investigated the climate change in the mountainous area of the upper Tarim River basin, using the up-to-date ‘ground-truth’ precipitation and temperature data, the APHRODITE (1961–2010, 0.25°) data; analyzed the potential connections between runoff data, observed at Alar station, and the key climatological variables; and discussed the regression models on simulating the runoff based on precipitation and temperature data. The main findings of this study are: (1) both annual precipitation and temperature generally increases at rates of 0.85 mm/year and 0.25 ℃/10a, respectively, while the runoff data measured at Alar station shows fluctuating decreasing trends; (2) there are significant spatial differences in the temporal trends of precipitation, for example, the larger increasing rates of precipitation occurs in the Karakoram Mountains, while the larger decreasing rates happens in northwestern of Kashgar County; (3) the decreasing trends of temperature mainly occurs in the Kashgar County and its surrounding areas in Summer; (4) seasonal correlations in trends of precipitation and temperature are more significant than those at monthly and annual scale; and (5) the regression model in simulating the runoff in the upper Tarim River basin based on Radial Basis Function (RBF) is better than that based on least-squares method, with the predictive values based on RBF models significantly better (Correlation coefficient, CC, ~ 0.85) than those by least-squares models (CC ~ 0.75). These findings will provide valuable information to inform environmental scientists and planners on the climate change issues in the upper Tarim River basin of Southern Xinjiang, China, under a semiarid-arid cl