GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 2.2 263 between the two stations, exploiting the surface waves dispersive property. In practice, this method highlights the coherent portion of the signal acquired in the stations by multiplying the two temporal series shifted by different time intervals. The resulted function shows a maximum at the time shift corresponding to travel time of a wave group from one station to the other and provide an estimate of the Green function between the couple of stations (e.g., Derode et al., 2003; Snieder 2004; Wapenaar 2004). In order to extract information on Rayleigh waves, the method is applied to the vertical component. The dispersion curves of group velocity are obtained from a FTAN analysis (Frequency-Time Analysis: Levshin et al. 1972), by picking points corresponding to the arrival time of wave energy maxima at different frequencies; this allows the determination of the velocity with which the energy of different frequency wavetrains propagates from one station to the other. One problem of the cross-correlation method is the possible presence of a dominant noise source misaligned with respect to the station couple, which can cause the calculation of an overestimated apparent velocity. For this reason, in the implementation of this technique, we tested the application of a direction correction based on the results of the HVIP analysis. Indeed, this technique identifies all the time windows in which Rayleigh waves are dominant in the noise wavefield and provides the azimuth of their propagation direction. Thus, for the calculation of cross correlation, only these time intervals were considered and the times ΔT g of the samples of the cross-correlation function were modified to take into account the effect of difference between the direction of wave propagation and that of the alignment of the couple of stations, according to the equation ,          (1) where α L is the azimuth of the line joining the two stations and α R is the azimuth of the Rayleigh wave propagation direction. The Rayleigh wave dispersion curves, obtained by cross-correlating the recording of the two station are a mean representation of the medium velocity between the two stations. To obtain more precise information about the medium around each station, we reconstructed more localized dispersion curves by subtracting, for each frequency, the arrival times at the two stations located before and after that of interest and dividing by their distance to calculate the localized group velocities. Rayleigh wave ellipticity curves, obtained by the HVIP method, and Rayleigh wave dispersion curves, obtained by the cross-correlation method, were then assumed as combined targets for a joint inversion. We use DINVER (Wathelet 2005), to search, for each station, 1D subsoil models consistent with both types of experimental data relative to the area around that station. Results and discussion. Figure 2b shows a cross section derived synthetizing the 1D subsoil models, calculated for all the stations, so to construct a 2D velocity model. The introduction of other constraints resulted in a better resolution in depth with respect to the previous cross section (Fig. 2a). It is possible to distinguish the presence of a surface layer with Vs velocities lower than 800 m/s and a higher velocity material (Vs > 1000 m/s), which likely corresponds to the local slaty bedrock (Wasowski et al., 2021). The depth of the bedrock between stations YJG4 and YJG8 coincides with the present level of the Yangjia gully streambed. This is probably related to the stiffness of the substratum, which is harder to erode by the stream respect to the landslide dammaterial. However, in the upstream part of the profile, in particular between stations YJG1 and YJG3, the higher velocity layer is located below the streambed. According to the analysis of remotely sensed data and field observations (Wasowski et al., 2021), this