GIS-driven multi-phase simulation framework for assessing rainfall-triggered landslides using SPH-FDM techniques

Abstract

Rainfall-induced landslides are critical geohazards that jeopardise infrastructure and human safety, emphasising the need for precise predictive models to enable effective management and mitigation strategies. This study introduces a GIS-enabled, multi-phase numerical framework that integrates smoothed particle hydrodynamics (SPH) for modelling landslide initiation and the finite difference method (FDM) for analysing post-failure mass flow dynamics. The SPH-based landslide initiation model (LIM) simulates rainfall infiltration and transient seepage effects on slope stability to identify potential failure zones. Subsequently, the FDM-based landslide propagation model (LPM) evaluates the kinematic behaviour of the failed material, providing detailed insights into post-failure mechanics. The framework was validated using benchmark scenarios to confirm its accuracy and robustness. It was then applied to a case study near a hydropower structure, where cumulative rainfall of 282 mm over six days resulted in significant deformation in approximately 7% of the 0.35 km² study area. Depth of failure analysis estimated a release volume of 1.35 \(\times \) 10⁴ m³, with the displaced mass reaching a maximum height of 10.6 m and a peak velocity of 30.1 m/s in narrow gullies. This integrated framework significantly advances the understanding of landslide processes in complex terrains and offers a computationally efficient tool for hazard assessment and infrastructure resilience planning. Future research should prioritise incorporating obstacle–flow interactions within the framework to optimise the design of protective measures and enhance disaster mitigation strategies.