Localization of seismic events produced by avalanches using multiple signal classification

Abstract

Snow avalanches generate seismic signals as many other mass movements. Therefore, detection of avalanches by seismic monitoring systems is highly relevant to assess the avalanche danger level. Apart from estimating the overall activity, information on the location of individual events is also important. We deployed a seismic monitoring system consisting of seven vertical geophones at a site above Davos, Switzerland. During a major snowfall on 2017 March 9 and 10, we visually identified 19 events in the seismic data that we considered as avalanches. For all these events, we calculated the backazimuth angles as well as the apparent velocity of the incoming wavefields using two different array processing techniques: a commonly used beam-forming approach and multiple signal classification (MUSIC). One of the avalanches released directly above the array and the path was well documented in a field survey. We analysed this event in more detail and compared both array processing techniques. Using the MUSIC method we were able to reproduce the avalanche path with reasonable accuracy, whereas the beam-forming method only provided a rough direction. The results obtained using MUSIC suggest that with our seismic array mainly infrasonic waves were recorded during the event. From the remaining 18 events, 15 had a clear path and were classified as avalanches. The other three events were dismissed as falsely classified avalanches. By comparing the results for these 15 avalanches events with our field survey, we were able to map 11 avalanches within a distance of 3 km of the array.