Integrated remote sensing and dynamic simulation for railway debris-flow hazard assessment and mitigation in alpine canyons

Railway bridges in alpine canyon environments are highly exposed to debris-flow impacts. However, existing regional- or large-watershed-scale studies commonly use relatively coarse spatial units and treat source susceptibility assessment and runout simulation as separate tasks, limiting site-specific hazard evaluation for critical railway infrastructure. This study develops a fine-scale, historical-event-constrained framework for the Bujueyida Gully, which is crossed by the Teke Middle Bridge on the Chengdu-Kunming Railway， Southwest China. Multi-temporal remote-sensing imagery, field observations, and terrain-derived data were integrated with Random Forest (RF) modelling and SHapley Additive exPlanations (SHAP) analysis to assess source-area susceptibility and interpret the dominant conditioning factors. Observations from the 2019 debris-flow event were used to calibrate key rheological parameters based on the Voellmy equation through particle swarm optimization (PSO), and the calibrated parameters were further applied in MassFlow to simulate debris-flow processes before and after mitigation under a 100-year return-period design scenario. The results show that high- and moderatesusceptibility raster cells account for 89.23% of all raster cells within the Bujueyida Gully catchment, while highand moderate-hazard raster cells account for 93.15% of all raster cells within the MassFlow-simulated impact area. The proposed drainage-channel system reduces the high-hazard area by 26.26% relative to the pre-mitigation condition, although residual high-hazard conditions remain near the railway bridges. The principal methodological contribution of this study is a traceable workflow linking fine-scale source-area interpretation, historical-eventbased parameter calibration, dynamic simulation, and bridge-scale mitigation pre-evaluation, thereby providing a practical basis for refined hazard assessment and engineering decision-making along railway corridors in alpine canyon regions.