Physics-based and data-driven long-term evolution of a landslide: From inversion to prediction
Citation
Du, W., Fu, X., Sheng, Q., Chen, J., Zhou, Y., Zheng, S. (2025). Physics-based and data-driven long-term evolution of a landslide: From inversion to prediction. Engineering Geology, 350: 107960. Link to paper
Abstract
This research explores data-driven inversion of current disaster conditions, physics-driven prediction of disaster evolution, and data fusion, taking the Taihe landslide as a case study. A disaster-scenario digital twin model is developed that extends traditional numerical simulation by integrating data and physics-driven approaches to create a simulation framework with temporal generalization capability. Long-term stability simulations integrate the strength evolution of geotechnical materials to incorporate time-span characteristics in the landslide twin model for long-term behavior. A simulation framework with time generalization capability is constructed adopting the Material Point Method (MPM), and coupled with the evolution of properties of geotechnical materials, multi-perspective and multi-dimensional scenario analyses can reveal the mechanisms underpinning landslide instability.