Construction of a Dynamic Analysis and Monitoring–Early-Warning Model for Debris Flow Evolution Based on COMSOL Simulation

A frequent and sudden two-phase (solid–liquid) geological hazard in mountainous areas, the evolution of debris flows involves the coupling of multiple physical fields, making monitoring and early warning particularly challenging. To accurately reveal the dynamic patterns of debris flow evolution and improve early-warning accuracy, this study focused on the Ni Chang Valley area in Shimian County, Ya’an City, Sichuan Province. Based on the COMSOL Multiphysics coupling simulation platform, a multiphysics bidirectionally strongly coupled numerical model was proposed and constructed, integrating the SPH (smoothed particle hydrodynamics) meshless particle method, FLO-2D shallow-water dynamics, and the MassFlow full-process simulation approach. Using COMSOL as a unified framework, this model employs MassFlow’s deep-integration, continuous medium method to simulate rainfall triggering and material source activation, FLO-2D’s shallow-water equations to describe macroscopic flow-deposition processes, and SPH’s mesh-free particle method to accurately capture large deformations and free-surface flow. The model fully reproduces the entire dynamic chain of debris flow processes, from rainfall triggering and soil mobilization to fluid transport and channel deposition. The reliability and accuracy of the model were verified by comparing it with field measurements from the 20 September 2022 historical debris flow event at Ni Chang Valley. Quantitative analysis indicates that when the viscosity coefficient increases from 0.1 Pa·s to 100 Pa·s, the flow velocity decreases by approximately 47% and the flow depth increases by approximately 62%. When the yield stress increases from 1 Pa to 100 Pa, the deposition area shrinks from 269,900 m2 to approximately 109,000 m2, a reduction of about 60%. Combining the results of the dynamic analysis, daily maximum temperature, daily precipitation, moisture content, mud-water level, and ground surface displacement were selected as core monitoring indicators. The analytic hierarchy process (AHP) was used to determine the weights of each indicator, and a data- and physics-driven weighted summation model for debris flow monitoring and early warning was constructed to achieve a five-level debris flow monitoring and early-warning system. Historical disaster cases demonstrate that this early-warning model can provide advance predictions of debris flow disasters up to 2 h and 40 min in advance. The warning lead time is sufficient, the grading logic is clear, and the model is capable of accurately capturing precursor information on disasters.