Description of a complex, rainfall-induced landslide within a multi-stage three-dimensional model

The mechanical processes involved in movements of earth or rock masses under the effect of gravity—a landslide—may include several phases where the failure of one portion of terrain can lead to the instability of the surrounding parts. An earlier landslide might have follow-up landslides until the landscape finds the equilibrium state. Recurring landslides are often recorded in landslide inventories but the information is seldom exploited in physical landslide modeling. Here, we study the landslide mechanism using Scoops3D—a three-dimensional, physically based landslide model. The program employs the three-dimensional column limit equilibrium and a digital elevation model to perform a slope stability analysis. Scoops3D evaluates the stability of rotational, spherical slip surfaces encompassing many grid cells, and finds the least-stable sliding surface throughout the entire digital landscape. The program generates two important outputs, namely a factor of safety map and a terrain map showing the new topographical conditions of the site with unstable areas removed. To define the final predicted landslide boundary, we have run Scoops3D repeatedly to assess the site’s stability using a newly produced terrain profile until the model predicts as stable all of the grid cells within the entire landscape. We compared the method's prediction with the actual sliding scar that took place on August 05, 2019, following a historical rainstorm in Sapa-Vietnam. Results for modified success rate, a performance metric, show that with reliable input data, the approach can predict the evolution of landslides with improved results compared to the traditional method using Scoops3D.