Probability mapping of debris flows triggered by multiple mechanisms in the Himalayas

This study presents a comprehensive framework for mapping the probability of debris flows in the Himalayas by accounting for multiple triggering mechanisms—specifically, rainfall-triggered debris flow (RTDF), glacier debris flow (GDF), and glacial lake outburst debris flow (GLODF). Recognizing the limitations of past research that typically focused on single-trigger events, the authors developed an innovative indicator system and applied three hybrid machine learning models (ATF-SVC, ATF-RF, and ATF-XGB) to evaluate debris flow susceptibility across diverse Himalayan catchments. The approach integrates high-resolution geospatial, climatic, geological, and glacial data to better understand and predict the spatial distribution of each debris flow type.

The results revealed that over half (52.98%) of the Himalayan catchments are susceptible to at least one type of debris flow, with 2.03% at risk from all three types. RTDF-prone areas are concentrated in the Yarlung Zangbo and Indus River Valleys, GDF zones are mainly in the Western Himalayas, and GLODF risks are prominent along the central and eastern Himalayan ridge. The ATF-SVC model showed superior performance for RTDF and GLODF predictions, while ATF-RF was best for GDF. This study provides a robust method for multi-hazard risk assessment and offers essential insights for climate-informed disaster management in high-altitude environments.