Multi-failure numerical analysis of rainfall-triggered landslides at the basin-scale

Abstract

Climate change and rapidly growing urbanization demand advanced, physically based approaches for mapping landslide-prone regions. Yet, most existing models rely on simplified assumptions regarding subsurface water flow and slope stability analysis. This paper introduces a novel fully numerical framework (MFNA-3D) that advances basin-scale simulation of rainfall-induced slope failures through an innovative multi-failure algorithm based on Numerical Limit Analysis (NLA). The proposed method employs a one-way coupling strategy that links a three-dimensional solution of Richards equation for transient, variably saturated subsurface flow with a newly developed NLA-based stability model tailored for large-scale slope stability assessment. The NLA-based algorithm enables the efficient delineation of multiple and simultaneous failure zones with irregular geometries and the quantification of failure volumes, without requiring a priori assumptions about their shape or location. MFNA-3D improves the physical realism of basin-scale landslide simulations by capturing the effects of antecedent and extreme rainfall on pore-water pressure and stability evolution. The methodology was applied to a landslide-prone tropical basin in the state of Rio de Janeiro, Brazil, incorporating mesh refinement and parametric sensitivity analyses. Quantitative validation was conducted by comparing the Fs maps obtained from the proposed approach with those generated using the infinite slope method, against mapped landslide scars. Results confirm the model’s capability to reproduce complex landslide behavior at the basin scale. This study positions MFNA-3D as a scalable and physically grounded numerical tool for advanced landslide hazard mapping.