LAUSR

Due to large uncertainties embedded in their geotechnical parameters, dynamic numerical models of landslide run-out distance can only be reliably applied to conduct post-failure back analyses, rather than accurate physical-based forward predictions. Therefore, it is important to quantify these uncertainties, and their effects, to improve the reliability of predictions. This paper proposes a run-out distance prediction methodology within a probabilistic framework to assess the exceedance probability of run-out distance for an individual landslide using a depth-integrated continuum method (Massflow) and a polynomial response surface method coupled with the first-order reliability method (RSM-FORM). The input parameters (i.e., friction angle and pore pressure ratio) are considered as random variables with reasonable probability density functions. The limit state function (LSF) is established based on a given threshold value of the run-out distance. The RSM is then adopted to establish a mathematical function that approximates the real LSF; and then, based on the fitted response surface, the FORM is employed to calculate the probability of the run-out distance exceeding the given threshold value. The 2015 Shenzhen landslide is considered as an example to illustrate the proposed methodology. The results show that our proposed methodology can successfully assess the exceedance probability of the run-out distance at different locations with acceptable computational errors. In addition, the computational efficiency is significantly improved compared to conventional Monte Carlo simulations. Moreover, according to the run-out distance-exceedance probability curve, the area potentially affected by the Shenzhen landslide can be classified into five categories (i.e., extremely high, high, medium, low, and extremely low) based on the different threshold values of exceedance probability (i.e., 50%, 10%, 1%, and 0.1%). The observed deposition area of the Shenzhen landslide lies within the extremely high and high hazard zones, demonstrating the feasibility of conducting landslide hazard zoning using the exceedance probability of the run-out distance. This study thus presents a practical procedure for enhancing the probabilistic run-out distance prediction of individual landslides and provides new ideas for conducting hazard zoning.