Developing hillslope hydrology based probabilistic rainfall threshold for shallow landslides

Developing rainfall thresholds is crucial for predicting rainfall-induced landslides. Typically, such thresholds are established by correlating past landslide events with corresponding rainfall records. However, this process particularly becomes challenging in regions with limited data. To address this issue, a methodology is designed in this study, which combines both transient seepage and stability analysis, followed by threshold development. This method first identifies the cumulative rainfall (E) and rainfall duration (D) parameters for the critical stability condition (i.e., Factor of Safety ≤ 1). Subsequently, this information is used to derive the E–D threshold line using the frequentist probabilistic approach. The methodology is applied to six shallow landslides located along the National Highway–10 in the Kalimpong region of Darjeeling Himalaya, India. At a 5% exceedance probability (T₅), the threshold is estimated as T₅ = 26.92D^0.44 for a duration range of 1 < D (hour) < 54. The comparison of the derived threshold has been found in good agreement with previous attempts, along with the reliability assessment for past recorded landslides. Hence, it is envisaged that the presented methodology will be applicable for developing rainfall thresholds in other data-scarce landslide-prone areas.