An integrated approach to develop a slope susceptibility map based on a GIS-based approach, soft computing technique and finite element formulation of the bound theorems

Abstract

Listen

Translate article

Although both GIS-based approach and soft computing have been extensively explored in predicting and mapping the landscape in terms of slope failure based on landslide susceptibility, there is always a question raised which is how robust landslide susceptibility estimates are? In this study, we strive to bridge the gap by introducing geotechnical modeling using lower and upper-bound solutions of limit analysis as benchmarking or comparing the techniques. Similar to other landslide or slope susceptibility mapping studies, this study commenced with developing the slope susceptibility map using analytic hierarchy process (AHP) and fuzzy logic (GIS-based approach and soft computing respectively) and later followed by a comparison of the results using upper and lower bound solutions. The results of the simulations denoted both AHP and fuzzy logic present closely related results in highly susceptible areas while differing greatly in both low and moderate susceptible. Following that six sections of the study area divided into three categories low, moderate, and highly susceptible were used for validation using upper and lower bound solutions. The finite element solutions have indeed confirmed that the areas classified as high, moderate, and low susceptible are benchmarked with upper bound solutions, however, they slightly differed from the lower bound solution. It is argued that if correct mechanical and physical properties of the slopes are used the AHP and Fuzzy logic can present reasonable results in terms of slope susceptibility. Owing to that the sharp drop in the AHP and fuzzy logic is believed to be controlled by the algorithms among the models. It is therefore recommended that further studies may be conducted in understanding a sharp distinction between the GIS-based approach and soft computing tools.