Seismic Signal Classification of Snow Avalanches using Distributed Acoustic Sensing in Grasdalen, Western Norway

Abstract. We show the usage of Distributed Acoustic Sensing for analyzing seismic signals from snow avalanche events. For three winter seasons we continuously recorded seismic data using Distributed Acoustic Sensing (DAS) on a section of a standard telecommunication fiber along a mountain road in Grasdalen, western Norway. Multiple snow avalanche events were registered, alongside various other signals such as road traffic and detonations from remote avalanche triggering. We describe signal characteristics of natural and manually triggered avalanche events and present a comparison with other observed signals in both time and frequency domain. Our frequency analysis shows that avalanche signals are most visible between 20–50 Hz. For larger avalanches, we observe weak low-frequency precursor signals, which correspond to the avalanche’s approach. The more prominent high-amplitude signals appear to be produced by the snow masses impacting stopping cones or the steep terrain near the road. In one natural avalanche event, we interpret distinct spike signals as likely corresponding to the stopping snow mass, based on similar findings from previous studies using geophones. Automatic detection, tested with simple STA/LTA thresholds in the 20–50 Hz range, presents challenges due to false positives from road traffic. Further refinement and testing are required to improve detection reliability in this complex environment. Our study represents an initial exploration into the application of Distributed Acoustic Sensing for snow avalanche detection, showcasing its potential as an effective monitoring tool for long road networks in mountainous regions.