Probabilistic back analysis of rainfall induced landslide-A case study of Malin landslide, India

On 30th July 2014, a devastating landslide resulted in the burial of a village called Malin in western India and led to about 160 deaths. The catastrophic failure was triggered by heavy rainfall over the last 3 days. Site investigation was conducted to obtain representative information about the area. To understand the relevant physics, which initiated landslide, seepage and back analysis of the unsaturated slope was performed. Transient seepage analysis was carried out using finite difference method to simulate the slope behavior during rainfall infiltration. A coupled fluid flow analysis was employed to capture the fluid/mechanical interaction in unsaturated soils accurately. The results show that antecedent rainfall played a role in slope instability; rainfall intensity and duration also have a big impact on slope instability. This paper presents a methodology to back analyse the slope so as to identify the mechanisms responsible for landslide initiation. Back analysis is performed using a probabilistic method based on Bayesian analysis. Probabilistic methods can back-analyse numerous sets of uncertain slope stability parameters. The uncertain parameters are imported as random variables in the analysis and are described by their probability distributions. Bayesian analysis updates these distributions based on the observed slope behavior. The method evaluates the reduction in matric suction which initiated landslide in Malin slope. The results show that the decrease in matric suction of about 100% in addition to the development of positive pore pressures is the main mechanism which has triggered the landslide. This amount of reduction in matric suction and the development of positive pore pressures decreases the mobilized shear strength in soil below the threshold value required to maintain equilibrium in the slope and hence the slope failed. (C) 2016 Elsevier B.V. All rights reserved.