Potential Recognition of Flash Flood Disasters in Small Watersheds in China’s Southwestern Mountainous Areas Considering Source Supply Conditions*

In the field of flash flood disaster prevention study, great attention has been paid to the role of heavy rainfall and flooding, yet the coupling of flooding and sediment caused by silt deposition is largely neglected. This study revised the impact factor system that considers the spatial heterogeneity in the loose solid material storage, using Aba Prefecture as the target area. The nonlinear relationship was extracted between the susceptibility of flash floods and environmental factors conducive to disasters such as loose deposit sources. This study employed a sensitivity-ensemble learning coupling algorithm to perform early potential recognition of flash flood risk in small watersheds of the southwestern mountainous areas. Based on digital maps for early potential recognition of flash flood risk, the potential volume of landslide sources in typical high-risk watersheds was estimated through the Transient Rainfall Infiltration and Grid-Based Regional Slope-Stability Model (Trigrs) with slope as the basic unit. This study established a data-physics dual-driven method for the potential recognition of flash flood disasters that is adaptable to different spatial scales and explored the impact of landslide sediment yield on flash floods, as well as the macro- and micro-mechanisms of flash flood disaster formation in typical small mountainous watersheds. The results indicate that: (1) The coupling disaster caused by flash floods and sediment in Aba Prefecture exhibits a certain degree of spatial clustering, with some impact factors demonstrating similar sensitivity patterns towards disaster occurrence. (2) High risks of flash flood disasters are identified in parts of eastern, central-southern, and a small portion of the northwestern watersheds in Aba Prefecture. These highly prone areas are primarily concentrated along topographic fault zones, which are relatively close to the high-frequency areas of accumulated loose deposit sources, where a larger probability of flooding-sediment coupling disaster can be found. (3) Compared with the results of the flash flood risk survey and assessment, the Certainty Factor-Adaptive Boosting (CF-AdaBoost) coupling algorithm identifies a coverage of 40.7% for high-susceptibility areas and a coverage of 12.4% for very low-susceptibility areas. Compared to the Flood Frequency Index Approach (FFIA), the coverage of high-susceptibility areas increases by 24.2%, while the coverage of very low-susceptibility areas decreases by 5.1%. (4) Post-processing the results of the slope stability model through the division of slope units estimates that the area of high-risk slopes in the Shouxi River basin during the intense rainfall event on August 20th is 4.99×10 ⁷ m ² . The key source supply areas are primarily concentrated in the upper and middle reaches of the basin, with an estimated sediment transport capacity of 1.219×10 ⁷ m³ for each flood event.