Prediction of landslide velocity and displacement from groundwater level changes considering the shear rate-dependent friction of sliding zone soil

In reactivated slow-moving landslides, the prediction of velocity and displacement is a crucial issue for understanding landslide kinematics and realizing warning systems. A variety of data-based numerical models have been used to predict landslide displacement; however, kinematics-based models, which consider the shear behaviors of soils within the sliding zone that fundamentally control landslide movement, have been relatively limited. In the present research, the Tangjiao landslide in the Three Gorges Reservoir (TGR) in China was taken as a case study. The monitored ground displacement from March 2007 to September 2016 and deformation signs observed in the field survey show that the Tangjiao landslide experienced slow movement with the ground displacement increasing stepwise in response to seasonal rainfall and periodic fluctuation of the reservoir water level. We performed a rate-stepped continuous ring shear test on two specimens remolded from the sliding zone soil at shear rates ranging from 0.01 to 10 mm/s. The test results show that the residual strengths of both specimens show a nonmonotonic shear rate dependency, i.e., a weak negative shear rate dependency (fitted by an exponential law) followed by a significant positive dependency (fitted by a linear law). The transition from a negative shear rate dependency to a positive shear rate dependency may be attributed to the change in shear mode when the shear rate exceeds a critical value. This complex shear rate-dependent residual behavior regulates the movement of the Tangjiao landslide in the reactivated state. On this basis, a kinematics-based model for predicting landslide velocity and displacement was proposed by taking into account the shear rate-dependent friction of the sliding zone soil. This model can reproduce the velocity and displacement of the Tangjiao landslide from the change in groundwater level (i.e., stress perturbations in this paper). The results suggest that by using this model, the well-investigated material parameters can enhance the prediction reliability and accuracy of landslide behavior. Our proposed approach, with the ability to reliably predict landslide velocity and displacement, such as the Tangjiao landslide in this work, has important implications for predicting movement trends and developing early warning systems for reactivated slow-moving landslides, especially translational landslides under similar geological conditions.