Identifying Suitable Digital Elevation Models and Deriving Features for Landslide Assessment in Idukki District, Kerala, India

This study compares the vertical accuracy of different digital elevation models (DEMs), such as Cartosat-I, ASTER-GDEM, SRTM-GL1, ALOS3D30 and FABDEM with a resolution of 30 m, to the toposheet-derived 264 spot heights of Idukki district, Kerala, India, obtained from the Survey of India. We quantitatively assess the vertical accuracy of these DEMs by analysing their accuracy against randomly selected topographic map spot heights. The study also validates the accuracy of the DEMs by evaluating the vertical accuracy separately for different elevation classes representing varying terrain characteristics of the Idukki district. Statistical measures are used to evaluate the performance of the DEMs. The results of the study show that FABDEM exhibits an RMSE of 41.79 m, which is lower than that of other models. The study utilizes FABDEM to derive a set of 12 geomorphological and hydrogeological features, including slope, aspect, elevation, profile curvature, plan curvature, distance to road, relative relief, ruggedness index, drainage density, height above near drainage, wetness index and stream power index. The characteristics of various parameters are analysed. The uniqueness of this study lies in its utilization of geomorphological and hydrogeological features derived from FABDEM that directly impact the susceptibility of landslides in the region. The study identifies that a combination of these dynamic and static parameters, which vary with elevation classes, plays a significant role in determining landslide occurrence in this region.