ESR12: Debris Flow Monitoring with Distributed Acoustic Sensing
Debris flows pose a severe threat to mountain communities worldwide. Precipitation-based event forecasts tend to be too inaccurate to provide reliable alerts to the population and critical infrastructure. In contrast, seismic measurements and theoretical source models have recently improved and may soon provide accurate detections and real-time observations of dynamic characteristics of debris flows. Although seismic sensor coverage is rapidly increasing, many affected torrent catchments and potential runout paths of debris flows are still under-instrumented limiting our detection capabilities of destructive events. Here, we propose to use fibreoptic networks as a new means to detect and monitor debris flow activity. Recent advances in Distributed Acoustic Sensing (DAS) allow to turn kilometre-long fibreoptic cables into thousands of seismic sensing points, which have been shown to produce clear records of Alpine mass movements including rock falls, snow avalanches and debris flows. Near several Swiss debris flow torrents, we will read out fibreoptic cables, some of which have been installed for conventional communication purposes but follow the torrent stretch over significant distances. Using conventional array seismology and innovative machine learning algorithms, we will develop reliable detection methods, which leverage the vast number of seismic sensors along the cable.
With this project we will monitor torrential processes at unprecedented spatial resolution. Furthermore, we will develop a new approach for debris flow monitoring, which relies on pre-installed fibreoptic infrastructure and thus does not require its own instrumentation. In many Alpine regions, fibreoptic coverage is dense and typically installed along train lines, highways and in tourist areas, which are particularly vulnerable to natural hazards and in particular debris flows. Over two summers we will record an estimated 10-15 debris flows at Illgraben and Val Greva, both located in Switzerland, which serves as an ideal basis for detection algorithm development.
Link to apply: https://apply.refline.ch/273855/1410/pub/1/index.html
Keywords: Natural hazards, distributed acoustic sensing, debris flows, machine learning, signal classification.
Supervisors: Fabian Walter (WSL, Switzerland), Stuart Lane (Uni Lausanne, Switzerland)
Co-supervisor: Lina Polvi-Sjörberg, University of Umeå, Sweden
Host institute: Swiss Federal Institute for Forest, Snow and Landscape Research, WSL, Switzerland
Collaborators: Seismology and Wave Physics Group at ETH Zürich, Switzerland (Andreas Fichtner)