Shallow landslide initiation susceptibility mapping by GIS-based weights-of-evidence analysis of multi-class spatial data-sets: a case study from the natural sloping terrain of Western Ghats, India

Abstract

The study aims to prepare a landslide susceptibility zonation (LSZ) map and quantitative validation of the result at a scale of 1:50,000 for the upland catchment of river Meenachil, Kottayam district, Kerala, India by recognizing and mapping the palaeoslide locations and the associated terrain attributes. A bivariate statistical method, known as the weights of evidence modeling, which offers a flexible way of testing the importance of input factors used for landslide susceptibility assessment, were used for the analysis. Fourteen terrain parameters namely, geomorphology, drainage density, soil type, soil thickness, land use, Normalized Difference Vegetation Index (NDVI), slope, aspect, relative relief, slope length, profile curvature, plan curvature, flowpath length and topographic wetness index were generated and its weights were determined. The conditional independence of the evidential themes was assessed by ‘Omnibus test’ and those themes which found to be conditionally independent were integrated to produce the final LSZ map. The results shows that areas having slope less than 16° is classified as stable areas since the active geomorphic processes are absent there, areas covered by side-slope plateau and denudational hills constitutes the very high and high susceptible region and is contributed by the slope between 16° and 45°, facing southwest, west and northwest directions, with extended influence from relative relief, slope length, flowpath length and compound topographic index. The reliability and accuracy of the LSZ map was assessed by the receiver operating characteristic curve analysis and the model shows a prediction accuracy of 89.2% (area under the curve = 0.892). In order to enhance the utility of the results of the present study, a panchayath-wise (the lowest administrative body) susceptibility map was produced and it shows that, among these panchayaths, highly susceptible areas are found in Munnilavu, Talanad, Tikoy and Poonjar Thekkekara. Suggestions are also drawn to optimize the hazard assessment for the region having high landslide susceptibility.