Stochastic ecohydrological-geotechnical modeling of long-term slope stability

Abstract

Rainfall-triggered landslides are the result of the complex water-related processes occurring in the atmosphere-soil-vegetation system. In this paper, we present a methodology to assess long-term slope stability from rainfall, soil, and vegetation parameters. Our approach couples the stochastic ecohydrological model developed by Rodriguez-Iturbe et al. (Proc R Soc A: Math Phys Eng Sci 455(1990): 3789–3805, 1999) and Laio et al. (Adv Water Resour 24(7):707–723, 2001b) and the unsaturated flow equation. In this way, the stochastic nature of the rainfall process propagates through the system, finally reaching the FS. Then, synthetic time series generated by numerical modeling are used to infer the FS statistical properties. The study of a hypothetical case study showed that the long-term slope stability is highly sensitive to the ratio of the amount of water that can be evapotranspirated and the amount of water the soil can infiltrate and less sensitive to the ratio of the saturated hydraulic coefficient and the long-term rainfall rate. A scaling analysis evidenced the existence of statistically significant (R2 ≈ 1) exponential relationships between ecohydrological dimensionless numbers and slope failure probability. The method presented here is a useful tool for landscape planning.