Objectives On July 20, 2024, an extreme rainstorm struck Hanyuan County, Ya'an City, Sichuan Province, China, delivering a peak hourly rainfall intensity of 53.8 mm and a cumulative precipitation of 107.5 mm within 12 hours. This event triggered a large-scale mountain flood debris flow in the Xiaogou watershed. The disaster filled and broke three check dams. Downstream drainage channels, bridges, houses, and infrastructure were severely damaged. Local residents' lives and property faced major threats.

Methods To reveal the erosion‑sedimentation characteristics and dynamic evolution of this debris flow, the Xiaogou watershed is taken as the research object. The small baseline subset interferometric synthetic aperture radar (SBAS‑InSAR) technique and the OpenLISEM numerical model are used. Unmanned aerial vehicle (UAV) aerial images, high‑resolution remote sensing interpretations, and field survey data are also combined. First, 22 Sentinel‑1A ascending images and 24 descending images from 2022 to 2024 are processed. The SBAS‑InSAR technique provides time‑series surface deformation information. Erosion and accumulation areas are identified. A coherence‑point statistical method is used to estimate the erosion rate and annual erosion volume. Second, UAV aerial photography and field surveys identify detailed sediment source distribution. The total volume of loose deposits is calculated as 1.26×10⁶ m³. Key sediment parameters are obtained through laboratory grain‑size and mechanical tests. The OpenLISEM model, which couples rainfall, surface runoff, debris flow initiation, erosion, transport, and deposition processes within a physically based framework, is then applied to simulate the entire dynamic evolution. The model's accuracy is validated against the measured deposition footprint and accumulation volume of the “7·20” event, achieving an overlapping accuracy coefficient Ω of 1.80, indicative of high reliability. Simulations are subsequently performed for rainfall return periods of 20, 50, and 100 a.

Results The average erosion rate in the Xiaogou watershed is 2 588 m³/(km²·a), with an annual erosion volume of 1.63×10⁴ m³/a. Under rainfall frequencies with return periods of 20, 50, and 100 a, the maximum flow velocities of the mountain flood debris flow in Xiaogou are 18.00 m/s, 17.78 m/s, and 17.64 m/s, respectively, while the maximum sediment depths are 6.38 m, 6.69 m, and 6.89 m, respectively. Under the 50 a and 100 a return periods, a significant amount of solid material is discharged, causing substantial impacts on the structures at the mouth of the gully. Part of the solid material rushes out of the gully into the river channel. It could raise the riverbed and cause semi-blockage of the Baiyan River. Consequently, a significant threat is posed to the downstream industrial park.

Conclusions The integration of SBAS‑InSAR for pre‑disaster erosion‑sedimentation characterization and the OpenLISEM model for multi‑process dynamic simulation provides a robust technical pathway for analyzing debris flow evolution from source to deposition. The results help improve the early warning capability of mountain flood debris flow in the southwestern mountainous regions and offer a sound scientific foundation for post-disaster emergency response and disaster prevention strategies.