UC Berkeley

Regional slope stability analyses can provide a system-level assessment of landslide susceptibility. This study aims to develop a regional back-analysis model for rainfall-induced landslides to compute ground shear strength that can be used for landslide prediction. Most commonly, mechanistic-based models for regional slope stability consist of simplified one-dimensional (1D) analyses compared to more realistic three-dimensional (3D) procedures. A pseudo-3D methodology is applied on 788 landslides over two 2.5 km2 areas in the plutonic rocks of the Utuado Batholith. First, an infinite slope stability triggering model is applied on a 1 m-resolution digital elevation model (DEM) to identify the triggering grid cells. A spatial 3D projection procedure is then implemented to generate 3D landslides with defined location, area, and volume. Two inversion procedures are considered herein: inversion against landslide area density and inversion against individual landslides. Unlike inversion against landslide area density, inversion against individual landslides generated good match in landslides’ locations and sizes and resulted in lower scatter in regional strength estimates. The average back-calculated cohesion and friction angle were c′ = 1 kPa, ϕ′ = 57° and c′ = 2 kPa, ϕ′= 56° for the first and second study areas, respectively. Inversion against landslides’ volumes in addition to their areas and centroid locations reduced the uncertainty in the triggering depth and modeled shallower landslides, which better matched the mapped landslides.