Evidential Belief Function (EBF) Model for Landslide Susceptibility Analysis in Idukki District, Kerala, India
Ch. N Vara Laxmi *
School of Spatial Information Technology, IST, JNTUK, Kakinada, Andhra Pradesh-533003, India.
K. Padma Kumari
School of Spatial Information Technology, IST, JNTUK, Kakinada, Andhra Pradesh-533003, India.
*Author to whom correspondence should be addressed.
Abstract
A Geo-statistical Dempster-Shafer Theory Evidential Belief Function (DST-EBF) model is selected for Landslide susceptibility evaluation. The aim of the study is to delineate zones prone to landslides within the Idukki district, Kerala using the EBF model and Geographical Information System (GIS) Technique. The objective is to integrate diverse datasets to assess and predict the landslide-prone areas, providing valuable insights for risk management and mitigation startgies. Topographical, Anthropogenic, and Geological factors are considered Landslide Conditioning Factors (LCFs), and Landslide Inventory data are used to establish and validate the Suscebtlity zones. Landslide inventory data is randomly divided into Training(70%) and Testing (30%) data. The resultant Landslide Susceptibility map is categorized into five zones using natural breaking classification method: Very low (20.96%), Low (29.71%), Medium (24.50%), High (18.63%), and Very high (6.20%), respectively, within the study area. The success rate and prediction rate were calculated using the AUC_ROC method and the EBF Model achieved the highest precision with a success rate of 0.935, and a prediction rate of 0.943 in the current study.
Keywords: Dempster- Shafer theory, landslide susceptibility, evidence belief function, AUC-ROC, GIS