[S-SS07] Environmental Seismology: from deep earth to surface process

[SSS07-P10] Debris flow process reconstruction and monitoring early warning using seismic signal characteristics

*he ren wang¹, yan yan¹ (1.Southwest Jiaotong University)

Keywords:Debris flow, Seismic signal, Reconstruction, Monitoring and early warning

Debris flows are among the most dangerous geological hazards, as they move quickly and have great destructive power, often resulting in significant casualties and damage to infrastructure. The development of monitoring and early warning methods has proven to be an effective strategy for mitigating the effects of debris flows. With the advances in environmental seismology, the utilization of seismic signals generated during debris flows has been shown to be feasible for monitoring and early warning. In this study, we established a monitoring system in the Minjiang River Basin in the Wenchuan region of China, combining seismic motion as a core component with precipitation and video monitoring. Several debris flows were successfully monitored by the system. By analyzing seismic data from several debris flows, we investigated the relationships between debris flow velocity, discharge, particle size and the characteristics of seismic signals. Verification was based on precipitation and video data, which enabled the reconstruction and monitoring of the debris flow movement process. Based on this, we applied three machine learning algorithms - Benford's Law, Random Forest and the Impulse Method - to make early warning predictions based on seismic signal characteristics. The reliability of these three methods was analyzed. This research provides a framework for using seismic signals for debris flow inversion, monitoring and early warning.