Physics‐Informed Deep Learning for Revealing the Evolutionary Characteristics of Landslides Induced by Rainfall Process

Rainfall-induced landslides involve complex interactions of multi-physical fields between seepage and stress fields across varying scales, posing challenges for conventional numerical models in achieving accuracy and generalization. We propose a Physical Information-Driven and Guided Multimodal Multi-task Neural Network (PMNN), which explicitly integrates fundamental geomechanical equations, including the Van Genuchten seepage model and Bishop's effective stress equation, while incorporates hydromechanical governing laws, such as mass conservation equation and Darcy's law, within the loss function to enforce physical constraints. Experiments on 800 numerically simulated and 55 model tested rainfall scenarios demonstrate that the internal physics embedding leads to over a twofold improvement in multi-physical fields prediction accuracy, while external physical constraints effectively reduce stability prediction errors. Furthermore, our results reveal distinct triggering mechanisms: high-intensity rainfall rapidly saturates shallow soils, forming a “shallow critical” zone and triggering shallow landslides, whereas prolonged low-intensity rainfall promotes deeper infiltration fronts, leading to deep-seated slope failures.