Combing soil spatial variation and weakening of the groundwater fluctuation zone for the probabilistic stability analysis of a riverside landslide in the Three Gorges Reservoir area

DOI: https://doi.org/10.1007/s10346-022-02023-1

IF: 6.153

2023-01-15

Landslides

Abstract:Some properties of landslide soils are generally recognized as inherently spatially variable because of the heterogeneity of natural geological deposits. Fluctuations in water levels of the Three Gorges Reservoir cause the depth of the groundwater table at landslide toes to change, resulting in fluctuations in the soil water content and significant soil degradation. The spatial variability and temporal weakening of soil properties should be incorporated into the landslide stability analysis. More importantly, rational deformation prediction and stability analysis of landslide numerical models require an advanced soil constitutive model. Herein, taking the Tangjiao landslide as a case study, the statistical characteristics of shear strength parameters were studied based on valuable soil test data. Then, the spatial variability of these parameters was modeled as a random field for the sliding mass. A hypoplastic constitutive model for clay was used to simulate the landslide deformation over 6 years caused by precipitation and changes in the reservoir water level. In addition, soil degradation induced by the fluctuating groundwater level was accounted for in key model parameters on the basis of experimental results. Eventually, the non-intrusive random finite element method was used to compute the landslide deformation and stability for the random field model. Landslide simulation of the deterministic model ignoring the spatial variation of soil parameters was also performed. Simulation results indicate that the difference in the landslide safety factor between the deterministic and random field models is up to 0.14 for the leading edge and up to 0.12 for the trailing edge. Random field models predict greater deformation and less stability than the deterministic model, suggesting that they are more conservative in this specific case. This research can serve as a useful reference for probabilistic stability analyses of riverside landslides considering soil spatial and temporal variability, which may be quantified more precisely in future research based on multi-source data inversion.

geosciences, multidisciplinary,engineering, geological