Development of rainfall threshold equation and bayesian probabilistic analysis for landslide prediction: A case study of Shimla, Northwestern Himalaya, India

Landslides are the most dangerous and recurring disaster in mountainous regions. The majority of landslides in the Himalayas are triggered during the monsoon season as oversaturated slopes get destabilized by incessant rainfall. While numerous global efforts have explored the relationship between landslides and rainfall, the specific rainfall thresholds for landslide occurrence in Himachal Pradesh have not been extensively studied. This research aims to develop an early warning system based on rainfall thresholds for Shimla, Himachal Pradesh, which is the state capital and a major tourist destination in India. The Rainfall Intensity-Duration (ID) threshold for landslides is developed for the Shimla area using the past 31 years (1990–2020) landslide data and daily rainfall data, using differential evolution optimization method. The threshold equation (I ​= ​7.20∗D^−0.26) was derived to define the conditions under which landslides are most likely to occur. Bayesian probability analysis was applied to assess the likelihood of landslides under varying rainfall intensities and durations. The results show an increase in probability from 0.06 for 1 day of rainfall duration to 1 for 10 day of rainfall duration. The influence of antecedent rainfall on landslide occurrence is analyzed using cumulative rainfall data for periods of 3, 7, 10, 15, 20, and 30 days prior to the landslide. The results indicate that 110 ​mm of antecedent rainfall over 30 days is sufficient to trigger landslides in the study area. These findings provide a framework for early warning systems and risk management strategies in Shimla, Himachal Pradesh.