Identification of the significant parameters in spatial prediction of landslide hazard

Abstract

Abstract Landslides are the most commonly occurring natural hazard in the hilly regions of the world. Tehri Garhwal in the Uttarakhand State of India is one such region where several landslide events have been reported. Several researchers have performed landslide susceptibility mapping (LSM) studies for the Tehri region. However, these studies lack consistency in selecting landslide-causing parameters for the susceptibility analysis and mapping. The variability in selecting parameters for the same region by various researchers has made it difficult to compare the models’ prediction accuracies. Hence, this study presents a scientific method to identify the most significant landslide-causing parameters for an enhanced LSM analysis. The selected combination of parameters was further validated on the two landslide-prone test sites with similar terrain conditions. To achieve these objectives, first, the landslide inventory map of 332 historical landslide events was prepared for the Tehri region. Second, the statistical relevance of 21 landslide-causing parameters for predicting landslide susceptibility was investigated using multicollinearity analysis considering Pearson correlation and the artificial neural network (ANN) model’s sensitivity analysis. Out of 21 parameters considered for the Tehri region, 11 were found to be significant for LSM and achieved the prediction accuracy of 0.93 area under curve (AUC) value. Third, the relevance of these 11 parameters in predicting the landslide susceptibility was checked for the two test sites of the Himalayan region. For this purpose, these parameters and their hierarchy were imported into the analytical hierarchy process (AHP) framework for predicting the LSM of the Tehri region and two landslide-prone sites, namely the Chamba and Bhuntar sites of Himachal Pradesh. The AHP-based LSM for Chamba, Bhuntar, and Tehri regions achieved a prediction accuracy of 0.86, 0.82, and 0.89 AUC values. Thus, this study recommends using the derived 11 landslide parameters and their hierarchy for carrying out LSM in the Himalayan region.