Data Science and Mass Movement Seismology

Project lead

Fabian Walter

Deputy

Alexandre Badoux

Project staff

Fabian Walter

Project duration

2022 - 2024

Cooperation Financing

Towards the Next Generation of Debris Flow Warning

Driven by the force of gravity, sediment erosion and deposition change the surface of our planet. Much of this transport happens during catastrophic mass movement events. In high Alpine areas, debris flows are a common mass movement type, which form an important part of the sediment cascade. Debris-flow warning requires rapid event detection and an observation-based understanding of sediment dynamics. However, both are hampered by limited access to the torrent head where instrumentation is needed to monitor sediment production and flow initiation. We propose to monitor mass movement formation using seismic data. Complicated seismic signatures still inhibit automatic signal classification needed for automatic alarm systems. To tackle this challenge, we propose innovative machine learning approaches which mine seismic data sets in debris-flow prone terrain. This will result in unprecedented warning times - before events cause destruction - with little new instrumentation to be deployed.

We have integrated the seismic detection algorithm into an Internet of Things (IoT) platform that manages and links incoming data streams from environmental sensors (see Figure). A pilot test at Illgraben in Switzerland’s Canton Valais, has shown how the occurrence of debris flow activity can be rapidly detected and visualized on the platform. Other data sources show terrain changes mapped with an autonomous drone and areas affected by the debris flow runout.