Similarity and Diversity of Debris Flow Footprints in Seismic Records

The ability of seismic instruments to monitor catastrophic channelized flows (e.g., bedload transport, debris flows, glacial lake outburst floods, and lahars) is becoming of interest to scientists and practitioners. However, using debris flows as an example, the variability in catchment geology, event properties, and seismic instrument configurations complicates the development of event detectors that can be transferred between sites without major adjustments of parameters and thresholds.

In this work, we built a global debris flow seismic data catalog comprising more than seventy events from three regions (Europe, China, and the USA). The collected events from nine catchments represent rainfall-triggered debris flows originating from diverse environmental contexts, such as post-fire catchments, post-earthquake catchments, and high-erosion catchments. We analyzed the similarities and differences among these events using dimensionless amplitude damping fitting. Furthermore, we evaluated the performance of a pre-trained machine learning detector applied to our event catalog to assess the feasibility of a generalized early warning approach. Our results will reveal the key signatures of debris flow footprints in seismic records within complex areas, which will guide the design of next-generation event detectors and warning systems. At the same time, the differences will guide us to customize the warning thresholds based on local site conditions and stakeholder interests. This study thus provides a foundation for affordable, seismic-data-driven early warning systems for debris flows and other channelized flows.