Multivariate Copula-based Framework for Stochastic Analysis of Landslide Runout Distance

The increasing frequency of landslide disasters worldwide highlights the importance of accurate prediction of post-failure runout distances for risk management. However, the prediction of runout distances of gravitational mass flow remains challenging due to the inherent complex heterogeneity of geomaterials and its inter-correlated spatially varying mechanical properties. To address this challenge, this study proposes a novel multivariate stochastic method based on the combination of the Generalized Interpolation Material Point (GIMP) method and the multivariate copula-based approach. The method considers geotechnical uncertainties by simulating copula-based cross-correlated random fields from sparse field data and incorporating them into the GIMP analysis through Monte Carlo simulation (MCS). Two slope cases with similar geometries but different sources of probability information are presented to illustrate the effectiveness of the proposed method. Results show that considering the influence of multivariate random fields can significantly affect the post-failure analysis of landslides. Both slope cases show about over 40% of all MCS samples exceed the deterministic case, which indicates that the current deterministic analysis notably underestimates the risk induced by large runout distances of landslides. This original study demonstrates the necessity to take interdependency spatial heterogeneity into consideration for post-failure modeling of landslides.