Forecasting step-like reservoir landslide using a physical-mechanical-numerical framework

The complex creep behavior and dynamic evolution of step-like reservoir landslides pose significant challenges for landslide displacement forecasting. To address these challenges, a novel Physical-Mechanical-Numerical (PMN) framework is proposed for predicting displacements in step-like reservoir landslides. This framework integrates physical-mechanical mechanisms with data-driven updating, improving the interpretability of predicted displacements and mechanical parameters in landslide geological models. The PMN framework comprises two phases: simulation and forecasting. In the simulation phase, a Seepage-mechanical-deformation (SMD) block model is developed, incorporating real-time fluctuations in reservoir water levels and rainfall, to simulate landslide displacement. In the forecasting phase, Bayesian updating and Markov Chain Monte Carlo (MCMC) methods are employed to invert, correct, and update key geological parameters, linking them to monitoring data to enhance efficiency and accuracy. The performance of the PMN framework is validated using the Baijiabao landslide in the Three Gorges Reservoir area. Furthermore, compared to traditional Back-Propagation Neural Network (BPNN), Long Short-Term Memory (LSTM), and Least Squares Support Vector Machine (LSSVM) models, the PMN framework effectively addresses forecasting challenges and provides novel insights into the interpretability of key geological parameters and monitoring data, enhancing the understanding of step-like displacement characteristics. It is expected to become a promising approach for long-term forecasting of step-like reservoir landslides.