Potential recognition of flash flood disasters in China’s southwestern mountainous areas considering source supply conditions

This study presents a multi‑scale method for recognising flash flood disaster potential in the mountainous regions of southwestern China, placing disaster risk reduction at the forefront. It integrates terrain, geological, soil, vegetation, hydrological and human‑activity factors, while explicitly incorporating loose sediment source conditions—an element often overlooked in traditional assessments. The research combines a Certainty Factor–AdaBoost data‑driven model with the TRIGRS physics‑based slope‑stability model to map susceptibility across 5,250 watersheds in Aba Prefecture and to estimate landslide‑derived sediment volumes in high‑risk basins. 

The study highlights recommendations for improving flash‑flood risk governance, emphasising the need to integrate sediment‑source heterogeneity into national early‑warning and assessment systems. It encourages the use of multi‑source meteorological datasets, refined slope‑unit modelling and remote‑sensing products to better identify key sediment‑supply zones. The authors suggest expanding the framework to data‑scarce regions through emerging technologies such as InSAR and transfer‑learning approaches, enabling more accurate, proactive and locally adaptable disaster‑risk‑reduction strategies.