An integrated approach for debris-flow seismic monitoring: amplitude, impulses and frequencies analysis at Gadria basin (Italy)

Seismic monitoring has been profusely employed worldwide to detect vibrations induced by slope deformation and/or landslide detachment and propagation. The analysis of the seismic signal may provide, in fact, relevant information on the dynamics of unstable slopes. As an example, it may allow the identification of precursors of collapse before slope failure occurs and the estimation of volume and propagation velocity of rock-avalanches, rock-slides and debris-flows. In particular, the monitoring of several characteristics of debris-flows can be efficiently performed through the use of seismic devices. The passage of a debris-flow wave, in fact, induces strong ground vibrations that can be easily and clearly recorded by different types of ground vibration sensors (accelerometers, velocimeters, microphones). Since the monitoring of debris flow is fundamental for studying their propagation and hazard implications, many kind of sensors have been tested and employed to measure the parameters that might be relevant for debris-flow investigation and study. Itakura et al. (2005) provided an extended review on this topic. However, ground vibration sensors provide an important advantage, in comparison with other devices, since they can be installed at a safe distance from the channel bed and do not interfere with the passage of the debris flow. This overcomes an important shortcoming of other types of sensors, like ultrasonic gauges, videocameras or speedometers, which need to be hung over the channel and thus are more prone to be damaged by the flow during the event. The recording of ground vibrations produced by debris flows presents however some difficulties and problems that need to be addressed and solved, such as the large amount of data detected by the sensors that need to be safely recorded. The output signal of the most commonly employed seismic sensors (velocimeters) is in fact a voltage proportional to the ground oscillation velocity. The typical frequencies of this signal usually ranges from 10 to 80 Hz and since the acquisition device needs to operate at a sampling frequency greater than the Nyquist sampling rate, usually a precautionary sampling rate greater than 100 Hz is adopted. This might be a problem when the device used for data recording is a standard data-logger, because of its limited storage capability. To solve the problem it is common to implement at least two different recording frequencies: a lower (no-event mode, NEM) recording frequency (usually 1Hz) employed to record the data during the periods when no debris flow is taking place and a higher (event mode, EM) recording frequency that is adopted when a debris flow occurs. For this purpose a threshold value has to be used that is associated to an algorithm that checks the variations of the signal recorde at low frequency to identify when it overcomes the threshold and switch the recording from NEM to EM. Two different techniques of transformation have been applied so far to the original ground velocity signal measured by the geophone to obtain a lower recording frequency (usually 1Hz) and reduce the amount of recorded data: (i) the transformation of the velocity signal into amplitude (Arattano, 1999) and (ii) the transformation into impulses (Abancò et al., 2012). The availability of high frequency monitoring data improves considerably the detection of debris-flows and the chances of their correct identification, as it occurs for other types of mass movements. Suriñach et al. (2005), as an example, pointed out that the spectrogram for a station that is approached by a sliding mass exhibits a triangular time/frequency signature, due to an increase over time in the higher-frequency constituents, that can be reliably identified. Recently, also Feng (2012) was ab