Application of index of entropy and Geospatial techniques for landslide prediction in Lunglei district, Mizoram, India

Abstract

The present study focuses on developing a landslide susceptibility zonation (LSZ) using GIS-based bivariate statistical model in the Lunglei district of Mizoram. Initially, 17 factors were selected after calculating the multicollinearity test for LSZ. A landslide inventory map was created based on 234 historic landslide events, which were randomly divided into training (70%) and testing (30%) datasets. Using the Index of Entropy (IOE) model, nine causative factors were identified as having significant weightage for LSZ: elevation, slope, aspect, curvature, normalized difference vegetation index, geomorphology, distance to road, distance to lineament, and distance to river. On the other hand, factors such as land use land cover, stream power index, terrain ruggedness index, terrain roughness, topographic wetness index, annual rainfall, topographic position index, and geology had negligible weightage. Based on the relative importance of the causative factors, two models were developed: scenario 1, which considered nine factors, and scenario 2, which considered all 17 factors. The results revealed that 16% and 14% of the district area were identified as very highly landslide prone in scenario 1 and scenario 2, respectively. The high susceptibility zone accounted for 26% and 25% of the area in scenario 1 and scenario 2, respectively. To assess the accuracy of the models, a receiver operating characteristic (ROC) curve and quality sum ratio method was performed using 30% of the testing landslide data and an equal number of non-landslide data points. The area under the curve (AUC) for scenario 1 and scenario 2 were 0.947 and 0.922, respectively, indicating higher efficiency for scenario 1. The quality sum ratios were 0.435 and 0.43 for scenario 1 and scenario 2, respectively. Based on these results, the LSZ mapping from scenario 1 is considered suitable for policymakers to address development and risk reduction associated with landslides.