Background Rainfall induced landslide is a common kind of mountain hazard in China, and its mass in near failure state makes it more susceptible to further erosion and may become the main source of debris flow. The traditional slope stability analysis models mainly focus on slope safety assessment, but they are not up to the calculation of the Potential Gravity Erosion (PGE) induced by rainfall triggered landslide. For the absence of such methods, simulation and prediction of PGE is still an urgent work in Soil and Water Conservation.

Methods An idealized 3D slope model was built in COMSOL to quantify the PGE. The transient pore water pressure and solid deformation development of slope soil under rainfall scenario were simulated in the fluid-solid coupling framework. Furtherly, both the global and local stability of slope were analyzed by using the strength reduction method (SRD) and defining a Local Factor of Safety (LFS) scalar field, and the volume integration of PGE amount quantified based on the LFS. Finally, the practicability of this work was evaluated and confirmed by comparing its slide surface geometries with SRD.

Results 1) The pore water pressure (PWP) was linearly distributed with the elevation from +100 kPa to -100 kPa in initial state, and decreased significantly with rainfall at surface layer. After 5 d of rainfall, the infiltration depth was about 3.5 m with a transient saturation depth in 2 m. The effective stress increased with depth and parallel distributed to the slope surface in the initial state, and decreased significantly in surface layer and increased slightly in the deep layer during rainfall. The range of average effective stress in 5 d of rainfall changed from 204-16.9 kPa to 206-1.86 kPa. 2) In the early stage of rainfall, the LFS of slope was generally greater than 1, and then gradually decreased on the slope surface with a trend spreading from shallow to deep layer. The shallow straight failure surface (depth < 1 m) appeared in 2 d and subsequently evolved into a deep arc-shaped sliding surface. The simulated data of PWP, effective principal stress and LFS were consistent in time and space dimensions, confirming the objective knowledge that rainfall affects slope stability by changing effective stress. 3) In the calculation case, the PGE appeared at 1 d, increasing slowly at first and then rapidly. After 5 d of rainfall, it reached to 230 m³.

Conclusions The LFS method can achieve the sliding bed profile that is more consistent with SRD with more advantages in computational efficiency and conservatism results. The quantification of PGE can be well realized as illustrated in this work by following steps: 1) field terrain and soil hydrology survey; 2) establishment of seepage-stress coupling model; 3) extraction of LFS equivalent profile; 4) geometric cutting and volume integral.