GNGTS 2017 - 36° Convegno Nazionale

650 GNGTS 2017 S essione 3.2 (2) (Bard and Bouchon, 1985), where h v f S 4/ 0 = and v S is the low-frequency S-wave velocity. If the horizontal extent of the basin is much larger than the layer thickness, we are in the 1- D approximation and the equations simplify as follows: f SH = f SV = f 0 . In the case of several soft layers over bedrock, the S-wave velocity is the average obtained with the time-average equation. The H/V spectra may exhibit not only a peak at frequency f 0 , but also a trough (minimum) at a higher frequency f 1 . Konno and Ohmachi (1998) found that the H/V spectra tend to show the peak/trough structure when a site has soft surface soils and they investigated the variability of the ratio f 1 /f 0 reporting a value of ∼ 2. Recent studies (e.g. Tuan et al. , 2011) concluded that the peak/trough structure indicates a high Poisson ratio in the surface soil, and a high impedance contrast at the underlying bedrock interface. Intrinsic attenuation and bedrock deformability affect the amplitude and frequency of the resonance peaks. Carcione et al. (2017) studied the effects of anelasticity on the S-wave amplification function and found that while attenuation (quantified by the Q factor) controls the damping of the higher modes, it does not affect significantly the peak locations. On the other hand, the impedance contrast α between the layer and the bedrock strongly affects the resonance frequencies. In particular, when α ≤ 1 (soft over stiff medium) the first maximum of the S-wave amplification function is located at the fundamental frequency f 0 . Conversely, α ≥ 1 (stiff over soft medium) gives a minimum at f 0 and the first maximum is located at f 1 = 2 f 0 . Therefore, bedrock elasticity (i.e. deformation) must be considered to obtain reliable estimations of the layer thickness. In the present work, the H/V spectra are obtained using the free software GEOPSY (http:// www.geopsy.org – SESAME Project). The software performs a statistical analysis of the recorded wavefield in the frequency domain, by computing the amplitude spectra of the three components in a number of selectable time windows. The width of these windows depends on the target frequency band and on the record length. For the computation of the H/V ratio, the amplitude spectra of the horizontal components are combined using vector summation. An anti- triggering algorithm was applied to the time series in order to include in the computation only signals with quasi-stationary amplitudes. We adopted the ellipticity inversion to determine the basement velocities and to verify whether a rigid or soft bedrock are present beneath the ice (Picotti et al ., 2017). GEOPSY includes a package (Dinver) for ellipticity inversion, based on the continuous wavelet transform (Fäh et al. , 2001). Dinver is a general environment for ellipticity and surface wave dispersion curves inversion based on a hybridMonte Carlo optimization scheme. Currently it implements amodified version of the neighborhood algorithm (Wathelet, 2008) to drive the inversion to converge to a minimum misfit solution. For each generated model, the algorithm computes the misfit between the theoretical fundamental-mode Rayleigh wave ellipticity and the ellipticity curve obtained from the time-frequency analysis of the H/V spectrograms (Wathelet et al. , 2004). Examples. In order to verify the reliability of the HVSR technique in different glacier settings we selected the following five target Alpine chain glaciers (Fig. 1): the Pian di Neve and Lobbia glaciers in the Adamello massif (Italy), the Forni and La Mare glaciers in the Ortles- Cevedale massif (Italy) and the Aletschgletscher in the Bernese Oberland Alps (Switzerland). We also analyzed passive seismic data previously acquired on the Whillans Ice Stream (WIS), a fast flowing ice stream in West Antarctica (Horgan et al ., 2012; Picotti et al ., 2015). Four geophysical methodologies have been employed in this work, i.e., GPR, geoelectric, active multichannel and passive seismic methods. Several exploration campaigns were conducted on the five target glaciers between April 2013 and October 2015 (Fig. 1). In August 2013, October 2013, 2014 and 2015, passive and active seismic data were acquired on the Pian di