Coviello, V.; Arattano, M.; Comiti, F.; Macconi, P.; Marchi, L., 2019, Seismic characterization of debris flows_ Insights into energy radiation and implications for warning, Journal of Geophysical Research Earth Surface 124 (2019): 1440–1463. doi_10.1029/2018JF004683,
URL: http://www.cnr.it/prodotto/i/404942

Debris flows represent a major hazard in mountainous areas, due to their rapid motion along steep channels and to the transport of large sediment volumes, including large boulders. In this paper, we present data of channelized debris flows characterized by different velocities and sediment concentrations recorded in an instrumented channel reach of the Gadria basin (eastern Italian Alps). From the analysis of the seismic energy produced by the interaction of solid particles with channel boundaries, we show that (i) the peak amplitudes are representative of the kinetic energy of each surge and (ii) most energy transfer occurs during the passage of the surge fronts. Then, we propose a debris flow detection algorithm based on the amplitude information gathered from a linear array of geophones installed along the channel. The short time average over long time average ratio of the seismic signal is used to early detect the debris flow occurrence in a continuous stream of seismic data. The algorithm recognizes moving, long-lasting sources of ground vibration (i.e., debris flows) and filters out different seismic sources (i.e., anthropic noise, earthquakes, and rockfalls). The alarm is triggered when the short time average/long time average threshold is exceeded on two geophones, progressively with time from upstream to downstream. The algorithm is employed in the early warning system installed for research purposes at Gadria. Complementary data (rainfalls, flow stage measurements, and video recordings) permitted a detailed event characterization and alarm validation. During three monitoring seasons, all debris flows were successfully detected, with the alarm lasting for their entire duration, and no false positives were produced.

Data from https://intranet.cnr.it/people/