Performance evaluation and sensitivity analysis of ASTER and SRTM (30M) DEM-derived terrain variables in landslide susceptibility assessment: A case from the western ghats

Abstract

Performance and sensitivity of freely available equal resolution space-borne digital elevation model derivatives in landslide susceptibility analysis were carried out in a selected part of the Western Ghats, India. ASTER and SRTM digital elevation models having a 30-m resolution were used to derive the terrain variables such as slope, aspect, relative relief, slope length and steepness, curvature, landform and stream networks. Most of the variables showed spatial variability in distribution pattern, which affects the results of geo-environmental processes analysed. Sensitivity and performance of each variable derived from the digital elevation models were assessed by preparing landslide susceptibility index (LSI) maps using the Information Value (InfoVal) technique and were validated through receiver operator characteristics (ROC) curve analysis. LSI maps generated point towards the capability of the SRTM digital elevation model to correctly generate the terrain variables than the ASTER elevation surface, by giving the accuracy of LSI maps greater than those produced using the ASTER-derived parameters (0.77 and 0.72 for SRTM; 0.67 and 0.65 for ASTER). The results of the present study suggest that the SRTM digital elevation data are more sensitive and suitable for terrain analysis and earth surface process modelling than the ASTER elevation data sets, although both possess equal resolutions.