Ice Avalanche Hazard Assessment Using Multi-Source Geospatial Data and GWRF Model

With global warming, ice avalanche hazards have escalated, becoming one of the most severe cryospheric disasters on the Qinghai–Tibet Plateau. To date, research on ice avalanches remains limited, focusing largely on individual glaciers and lacking a regional-scale understanding of their overall activity. In this study, an ice avalanche inventory was established based on multi-source remote sensing data, and a geographically weighted random forest (GWRF) model was developed for regional-scale ice avalanche hazard assessment by integrating multiple topographic, climatic, and glacier-related conditioning factors while considering their spatially varying relationships with ice avalanche occurrence. The results indicate the following: (1) Ice avalanches are mainly distributed in the Yarlung Zangbo River Basin, predominantly occurring on north-facing slopes with gradients of 10–40° and elevations ranging from 4000 to 5000 m. (2) High-hazard zones were predominantly located in the southeast, whereas low-hazard zones were more extensively distributed in the northwest, presenting a spatial pattern of higher hazard in the east than in the west, and in the south than in the north. (3) The GWRF model outperforms the conventional RF model by more effectively capturing spatial heterogeneity, thereby improving the accuracy and stability of ice avalanche hazard prediction. This study provides a new perspective and methodological framework for cryospheric hazard assessment and supports improved risk management in high-mountain regions.