Application of novel framework approach for assessing rainfall induced future landslide hazard to world heritage sites in Indo-Nepal-Bhutan Himalayan region

Abstract

The present study focused on mapping of ‘landslide susceptibility (LS)’ and evaluating landslide risk on some important ‘world heritage sites (WHSs)’ in Indo-Nepal-Bhutan Himalayan. LS mapping was carried out using ‘eXtreme gradient boosting (XGBoost)’, ‘Random Forest (RF)’, and ‘convolutional neural network (CNN)’, considering 15 conditioning factors, including seismicity and rainfall. Since rainfall is the triggering factor of landslides, the future rainfall was estimated using four ‘Shared socioeconomic pathways (SSPs)’ scenarios of the ‘Climate Model Intercomparison Project-6 (CMIP-6)’ to identify the future LS and vulnerable WHSs of the Himalayan. The XGBoost is the robust model applied in future scenario-based LS assessments. The very high susceptibility zone has an increased tendency, about 13% to 31% area in the future scenarios where the predicted rainfall also increased, about 100 mm in 80 years. The findings of this study will aid strategy makers in conserving the heritage monument while also ensuring sustainability.