Forecasting Landslides Using Mobility Functions: A Case Study from Idukki District, India

Abstract

Catastrophic landslides and associated destructions are increasing every year, because of the change in climatic conditions and land-use patterns. The ecologically sensitive zones of Western Ghats are highly susceptible to landslides and require scientific attention in developing an efficient early warning system. Definition of empirical rainfall thresholds on local, regional or global scales is the most commonly followed method of forecasting rainfall induced landslides. The limitations associated with such thresholds demands for better forecasting performance, incorporating the effect of physical processes in the initiation of landslides. This study is an attempt to forecast landslides in Idukki district, using mobility functions. The function separates the impossible and certain mobilisation parts and forecasts whether landslides can occur or not. Based on the critical value of mobility function, two different warning levels are proposed for four different reference areas in the district. The study shows that the model is 97% efficient in smaller areas with uniform topographical and geological conditions, and the performance is reduced as the area becomes larger, with varying topographical and geological properties. The model proves to be an effective landslide forecasting tool that can be integrated with a rainfall forecasting system, to develop an early warning system for the region.