GNGTS 2017 - 36° Convegno Nazionale

GNGTS 2017 S essione 3.3 689 localized and marked lateral variations, P- and S-wave tomographic interpretations may results inadequate to effectively delineate location and depth of these sharp anomalies with reliable detail. Since surface waves (SWs) propagate parallel to the ground surface, their analysis can be used as an additional tool for the detection of these lateral heterogeneities. Several SW based methods have been developed to locate the sharp lateral variation and to estimate its embedment depth. For location purposes, the energy computation on the recorded seismograms can be a useful tool to highlight energy concentrations due to back-reflection of energy at the lateral variation or to energy trapping within the discontinuity (Nasseri-Moghaddam et al. , 2005; Bergamo and Socco, 2014; Colombero et al. , 2017). Analogously, the computation of the energy decay exponent (γ) can provide a global estimation of the discontinuity location. In particular, disregarding the effect of intrinsic attenuation and compensating the recorded traces for SW geometrical spreading, γ is expected to be zero in a homogeneous medium, and thus strong deviations from this value can be interpreted as the effect of either energy concentrations (γ<0) or energy decays (γ>0) induced by reflections caused by the lateral variation. To additionally retrieve a depth estimation, techniques based on the evaluation of the attenuation coefficient (α), describing the trend of energy decay as a function of frequency, can be adopted. The value of α is indeed related to the mechanical properties of the subsoil affected by the propagation in different frequency components of Rayleigh waves. When sharp heterogeneities are present in the subsoil the value of α is strongly influenced by the reflection of energy at the interfaces, and hence, abrupt variations can be interpreted as the effect of lateral discontinuities in the subsurface. Since the depth of penetration of Rayleigh waves increases as their frequency decreases, only low frequencies of the incident wave are expected to have enough energy below the discontinuity to be transmitted. Particularly, identifying the threshold between higher frequencies that undergo backward reflection or are trapped within the discontinuity and undisturbed lower frequencies can provide the embedment depth of the discontinuity. Generally, this depth could be roughly estimated as 1/3 of the wavelength, and thus the ratio between the mean Rayleigh-wave velocity (V R ) and three times the identified cut-off frequency. Both γ and α methods were originally designed for single-fold data, and later extended to multifold data by Bergamo and Socco (2014), to improve the reliability and interpretation of the results. In the same work, single-fold autospectrum plots (Zerwer et al. , 2005), displaying the energy content of seismogram as a function of offset and frequency, were compared to the previous methods for the simultaneous achievement of both discontinuity location and depth estimation purposes. Finally, when the location is known, the Transmitted- over-Incident (T/I) spectral ratio technique (Hévin et al. , 1998; Bièvre et al. , 2012; Colombero et al. , 2017) can help to further constrain the discontinuity depth. From the ratio between the Fourier spectra of transmitted traces (after the discontinuity) and incident traces (before the discontinuity), the cut-off frequency can be determined as well. In this work, these five SW based procedures (energy, γ, α, autospectrum and T/I spectral ratio methods) are simultaneously applied to a synthetic dataset, for a cross-comparison of the results in terms of both location and depth estimation of a shallow discontinuity. Novel attempts to reach a single subsurface imaging from the processing of multifold data are presented. Methods. A 2D finite-element-model was built in Comsol Multiphysics to retrieve a set of synthetic seismograms for testing the different SW based methods. Large model dimensions were chosen to simulate a half-space configuration (Fig. 1a). To avoid strong wave reflections at the borders of the model, low-reflecting boundaries were applied at the bottom and lateral sides of the domain and the bottom corner points were fixed to zero displacement. The upper surface was left free; in its central part a synthetic array of 72 geophones with 0.5-m spacing was simulated. A rectangular heterogeneity (7-m wide and 2-m deep) was built in the center of the array, between geophones G29-G30 and G43-G44 (Fig. 1b). A marked contrast in physical and mechanical properties between the material inside and outside the box was assigned, as summarized in Tab. 1. For both materials, Rayleigh damping was introduced in the model.