Effect of spatially variable saturated hydraulic conductivity with non-stationary characteristics on the stability of reservoir landslides

Conventional landslide analyses, based on deterministic methods or simple random variable methods, cannot consider the spatial variability and the depth-dependent nature of saturated hydraulic conductivity (K-s) in accumulative landslides. However, studies show that these conditions exist objectively and have a large impact on the seepage and deformation of landslides. Thus, the non-stationary characteristics of K-s need to be properly accounted for. In this paper, the Baishuihe landslide is taken as a case study. First, based on surface nuclear magnetic resonance (surface-NMR) technology, spatial variability and the depth-dependent nature of the K-s are obtained. Then, a non-stationary random field model is established to model the trend and fluctuating components of the K-s individually, and a deterministic model is established to model the trend of the K-s with depth. Finally, numerical simulations of the stochastic model and deterministic model under reservoir water fluctuation conditions are carried out by the non-intrusive stochastic finite element method, and the seepage-deformation and stability characteristics of the landslide are analyzed. The results show that compared with that in the deterministic model, the change in pore pressure in the stochastic model obviously lags, during reservoir water drawdown, the displacement of the trailing edge in the stochastic model is larger than that in the deterministic model, with increasing reservoir water deceleration, the hysteresis of the pore pressure in the stochastic model is more notable, and the deformation is larger and the stability is worse in the stochastic model. It is demonstrated that ignoring the spatial variability of the K-s will overestimate the stability of the landslide.