A Simple Physically-Based Distributed Translational Landslide Model

Abstract

Detailed landslide susceptibility mapping (LSM) requires a skillful landslide model. Considering that translational landslide is the most type of landslides occurred in the world, a well-behaved translational model is sought. This study presents a simple physically-based distributed translational landslide model. In this model, the incident of landslide is detected from the value of factor of safety (FoS) which is computed based on Mohr–Coulomb failure criterion. In here, FoS is calculated as the ratio of shear strength and shear stress. The lower the FoS, the higher the possibility of a landslide to occur. The model input data consists of soil cohesion ${\varvec{c}}$ (kg/cm²), soil specific weight ${\varvec{\gamma}}$ (g/cm³), depth of surface of rupture (m), slope of surface of rupture ${\varvec{\beta}}$ (degree) and friction angle $\boldsymbol{\varphi }$ (degree). Application of the model was performed in Sirampog and Kandang Serang, two subdistricts in Western Central Java that underwent the most frequent landslides in the region. Model validation was conducted by comparing the values of FoS of unsaturated and saturated soils and identifying FoS in the sites where landslide events recorded. Several goodness of fit indices to measure the model performance are accuracy (ACC), success index (SI), average index (AI) and distance to perfect classification (D2PC). Under unsaturated condition, the result shows that the number of grids having FoS less than 1 are 0% and 0.6% for Sirampog and Kandang Serang respectively, indicating no landslide occurrence. When the soil gets saturated, 17.6% and 36% of area have FoS less than 1 for Sirampog and Kandang Serang respectively. This shows that the landslide occurred in this region is rainfall-induced landslide. Overall, the model shows a good performance with ACC, SI, AI, D2PC values are 0.82, 0.58, 0.54, 0 and 0.64, 0.49, 0.49, 0 for Sirampog and Kandang Serang respectively.