GNGTS 2023 - Atti del 41° Convegno Nazionale

Session 3.3 ______ ___ GNGTS 2023 Estimation of S- and P-wave velocity models from surface-wave data F. Khosro Anjom 1 , F. Adler 2 , L.V. Socco 1 1 Department of Environment, Land and Infrastructure Engineering, Politecnico di Torino, Italy 2 TotalEnergies, CSTJF, Pau, France Introduction Surface-methods are commonly used for near-surface characterizations. Dispersion curves (DCs) of surface-waves are known to be highly sensitive to S-wave velocity (VS), but less sensitive to P-wave velocity (VP). As a result, common surface-wave approaches solely focus on VS model estimation and take into account a priori VP or Poisson’s ratio in the inversion algorithms. Recently, the Wavelength-Depth (W/D) data transform method is introduced that is shown to be sensitive to both VS and VP (Socco et al, 2017; Socco and Comina, 2017). This relationship exploits the correlation between surface-wave wavelength and skin depth and is composed of wavelength-depth couples with same phase velocity and time average VS. The W/D relationship can be used to directly estimate the time-average VS from DCs (Socco et al, 2017). Socco and Comina (2017) showed the relationship is highly sensitive to Poisson’s ratio and it can be used to directly estimate the time-average VP. Khosro Anjom et al. (2019) created a W/D workflow to estimate interval VS and VP model in laterally varying sites without any prior information. The method provides the Poisson’s ratio that can be used a priori in scheme of laterally constrained inversion (LCI) and surface wave tomography (SWT). Here, we show the application of W/D method to estimate VS and VP models using real data collected from a stiff site characterized by significant lateral variations and elevation contrast. The data is acquired using the so-called carpet recording (Lys et al., 2018), in which the receivers are spread over a regular grid and sources are limited to logistically accessible locations, creating irregular source-receiver outline. We also apply the LCI (Socco et al., 2009) and SWT (Khosro Anjom et al., 2021) to the data set to estimate both VS and VP model, considering the estimated Poisson’s ratio from W/D method as prior information in their algorithm. We then quantitatively compare the calculated velocity models from the three methods, namely, W/D, LCI and SWT. Method The only inputs of the W/D method are the multi-channel DCs. We consider the recordings from receivers that are spread over a square area, and we compute the spectrum using phase shift method. To increase the signal-to-noise ratio, we stack the spectra computed for the same