Dynamic coupling of TRIGRS and RAMMS models for physics-based prediction of rainfall-triggered shallow landslides: a case study of the 2023 yanghuachi landslide, Chongqing

Regional prediction of rainfall-triggered shallow landslides is critical for hazard mitigation, yet conventional qualitative methods relying on historical data fail to quantify dynamic landslide processes. This study establishes a dynamic prediction framework integrating the Transient Rainfall Infiltration and Grid-Based Slope-Stability (TRIGRS) model with the Rapid Mass Movement Simulation (RAMMS) model, while combining field investigations, remote sensing data, physical modeling, numerical simulations, and Kappa coefficient analysis. We analyzed landslide spatiotemporal patterns under varying rainfall conditions (intensity and duration) across evolutionary stages, and results demonstrate: (1) TRIGRS identified landslide-prone zones with 89.2% accuracy; (2) 75% spatial overlap between simulated deposits and field-mapped inundation areas; (3) A triphase failure sequence: infiltration-induced instability (0–2 h), progressive shear zone expansion (2–4 h), and rapid mass movement/deposition (6–8 h). A power-law relationship emerged between instability area (y) and cumulative rainfall (x): y = 0.028x ^0.36, establishing rainfall thresholds for landslide timing prediction in data-scarce western mountainous regions. This integrated framework advances mechanistic understanding of landslide dynamics and provides a quantitative basis for early warning systems in rainfall-prone mountainous areas.