Garhwal Himalayas is one of the landslide-prone regions of India, which is characterized by frequent landslide occurrences causing damage to property, natural resources, and lives. In this study, the historical landslide records were associated with natural factors (climate, physiography, soil, geology, and vegetation factors) and anthropogenic factors (population density, livestock population, distance from settlements, and distance from road) using the random forest model. On the basis of variable importance value, it was found that the distance to road, vegetation (NDVI mean), slope, distance to water, and distance to the settlement were the most important factors in the model development. It indicated that developmental activities were one of the major factors responsible for landslides in the region. The results showed that 4655.94 km2 was susceptible to landslide hazards, out of which 38.29% area was under high and very high susceptibility class. The results validation using AUC (0.97), kappa coefficient (0.8), and TSS (0.84) showed that the model performance was excellent in predicting the landslides. The results provide information about the sensitive areas which are prone to landslides in the region. The study highlights the significance of scientific development and plantation of deep-rooted trees and grasses in landslide-prone areas.
|