The two-phase Material Point Method (MPM) model effectively handles soil-water interactions in landslide-tsunami simulation, but most models are restricted to simplified soil constitutive laws. While the MPM with a solid-fluid phase transition constitutive model has been proposed, its extension to fully three-dimensional (3D) full-scale landslide-tsunami conditions remains unexplored. To address this gap, this study develops a 3D two-phase MPM model that incorporates a solid-fluid phase-transition constitutive model and applies it to the Huangtian landslide for comprehensive analysis. Key contributions include: (i) establishing and validating a 3D phase-transition MPM model, enhanced with Graphics Processing Unit (GPU) acceleration for computational efficiency; and (ii) applying the MPM model to simulate the Huangtian landslide. In particular, this work conducts a comprehensive qualitative and quantitative analysis. Key findings include: (a) the simulation results show good agreement with experimental data and existing simulation solutions, validating the model's effectiveness; (b) the MPM simulation of the Huangtian landslide predicts a sliding velocity of 25 m/s and a maximum free surface elevation of 31 m, both in good agreement with field surveys and some existing analytical and numerical solutions; and (c) the energy transfer mechanisms of the Huangtian landslide are revealed: the onset of water kinetic energy coincides with the landslide’s peak kinetic energy, occurring at approximately 25 s, while the water’s peak kinetic energy lags behind the landslide’s by 15–20 s. This study offers an effective numerical platform for evaluating landslides and the generated waves.
Rock slopes may accumulate progressive deformation before collapse, but the amount that can be accommodated depends on geological structure and cannot be treated as a universal warning threshold. This study examines the effect of weak-plane orientation on pre-collapse deformation capacity using controlled gypsum-plate experiments, distinct element method (DEM) simulations, and a comparison with 24 natural slopes in Taiwan. Average strain quantifies slope deformation and is defined as the average displacement normalized by the length of the sliding body along the slope direction; critical strain is the average strain at the operationally defined last stable state before a through-going failure surface and coherent collapse. The physical records and DEM simulations produced the same critical-strain ranking, from lowest to highest: Cataclinal 40°, Cataclinal 60°, Anaclinal 30°, and Anaclinal 60°. In the selected field cases, the mean scarp ratio was higher for anaclinal slopes (27.6 ± 9.2
Red beds are widely distributed in the southwestern region of China and are often referred to as areas prone to landslides. Taking the Wushanping landslide that occurred in Fengjie County, Chongqing, China, in 2014 as an example, this study used field methods, fluid–solid coupling numerical simulation, and ring shear tests to investigate the instability mechanism of red bed rotational landslides under heavy rainfall. The results show that (1) the red bed sliding zone soil underwent shear liquefaction under the infiltration of prolonged rainfall, and the significant reduction in shear strength led to the instability of the sliding mass. Ring shear tests revealed that the red bed sliding zone soil undergoes shear failure after approaching a limit pore water pressure of 160 kPa under the tested conditions. (2) Heavy rainfall rapidly infiltrates along the vertically developed fracture zones of synclinal structures, causing the water level within the sliding mass to rise quickly. The high pore pressure in the front edge blocks of the slope cannot be discharged in time, and the high hydrostatic pressure at the rear and the reduction of effective stress cause the front edge blocks to undergo shear failure, ultimately leading to the instability of the landslide. The aquifer structure in synclinal cores, Jurassic red bed strata, and short-duration intense rainfall constitute the primary controlling factors for such hazards. This study can provide a theoretical basis for the research on disaster prevention and mitigation of similar landslide disasters in red bed areas.
Frequent strong earthquakes in Taiwan’s steep mountains trigger massive landslides, but the relative roles of tectonic setting and topography in shaping their spatial distribution remain poorly understood. This study examines the dominant controls and spatial variability of landslides triggered by two representative earthquakes—the 1999 Mw 7.6 Chi–Chi and the 2024 Mw 7.4 Hualien events—and conducts a cross-event probabilistic assessment for rapid hazard assessment in Taiwan. A unified dataset integrating two event-based landslide inventories and ten conditioning factors was used to construct three logistic regression (LR) models: a Chi–Chi-specific, a Hualien-specific, and a hybrid cross-event model. The SHapley Additive exPlanations (SHAP) method was used to quantify feature importance and characterize nonlinear response patterns of seismic, topographic, and hydrological variables. The results demonstrate that the occurrence of landslides in both events is dominated by the interaction of strong ground motion and rugged topography, with their relative contributions differing according to local tectonic settings. The Chi–Chi earthquake was characterized by near-fault shaking–induced failures concentrated within 10–30 km of the Chelungpu fault, while the Hualien event showed greater dependence on slope aspect, relief, and moisture conditions. In terms of probability mapping, single-event models yield the highest accuracy under local seismic conditions but exhibit systematic bias when transferred to other earthquakes, whereas the hybrid model provides more spatially consistent and robust predictions across the two studied events. This study presents an interpretable probabilistic method for evaluating seismic landslide hazards in tectonically complex regions, with the hybrid model providing a useful basis for near-real-time landslide probability mapping in Taiwan and potentially in regions with similar conditions.
Ice avalanches from hanging glaciers represent an important high-mountain hazard in the Himalayas, where steep valley topography and climatic changes can contribute to the destabilisation of hanging ice masses. This study presents a scenario-based assessment of hanging-glacier avalanche dynamics and downstream exposure for three hanging glacier sites surrounding Badrinath Township in the central Himalayas. Potential release areas and ice volumes were calculated using remote-sensing data and model-based approaches, and avalanche dynamics were simulated using the Rapid Mass Movement Simulation (RAMMS) model. The simulations were conducted under low-, medium-, and high-mobility conditions by varying the Coulomb friction (μ) and turbulent friction (ξ) within terrain- and elevation-based ranges. The results indicate that flow behaviour was primarily influenced by release volume, glacier slope, and downstream terrain, with larger release volumes and steeper slopes generally associated with greater flow intensity and runout. The high-mobility condition consistently produced the most severe flow characteristics. However, the differences between the low- and high-mobility conditions remained relatively small. The low-to-high-mobility conditions resulted in changes of 2.3–6.7
