GIS-based landslide susceptibility zonation and comparative analysis using analytical hierarchy process and conventional weighting-based multivariate statistical methods in the Lachung River Basin, North Sikkim

In this work, an extensive study of landslide susceptibility zonation (LSZ) has been performed in Lachung River Basin owing to its landslide-prone nature. The LSZ has been framed as a comparative study of two different multi-criteria analysis methods, namely the Analytical Hierarchy Process (AHP) and Conventional Weighting (CW) method. The causative factors of landslides (geology, land use and land cover, rainfall, drainage density, normalised difference vegetation index, slope, elevation, aspect) have been duly considered as spatial thematic data layers. Furthermore, subsequent calculation and superimposition of all these data layers are performed on the GIS environment. Both the AHP and CW method gave their optimum predictions, which indicates that highly susceptible zones cover the maximum area of the river basin. On the other side, the very high susceptible zones are mainly observed along the riverbank, which is barren in nature, and the barren steep-sloped terrain already experiences a number of landslides. Similarly, the very low and low susceptible zones have been observed in the areas covered by glaciers and perpetual snow. Finally, the predictions of the AHP and CW methods have been validated with field data of the location of existing landslide points and two different statistical techniques (ROC curve, landslide density) that offer a clear estimation of the comparative merits/demerits of these methods. However, the CW method claims more acceptability as a predictive measure of LSZ in the present study area for portraying the ground truth more accurately than the widely accepted AHP method for the same purpose.