GNGTS 2016 - Atti del 35° Convegno Nazionale

578 GNGTS 2016 S essione 3.2 on the quality and characteristics of recorded data. GPR processing was performed through commercial software GRED HD. The first processing step was to estimate the propagation velocity through the clearest hyperbola in the radargram. In this way we estimated a propagation velocity of 5.40 cm/ns. Then a band-pass filter, 20-200 MHz, and a SOIL SAMPLE effect were activated to eliminate the initial part of the profile corresponding to the air propagation, and to define the start time. Processing continued by applying the Smoothed Gain, to improve the signal sharpness. Finally, with the purpose of correctly converting from time-distance to depth- distance section, the migration process was applied in the domain of MIGRATION_TD time. The Profile01 shows a very shallow reflector with high amplitude reflected phases at an average depth of 5.5 m (Fig. 2). The sharp amplitudes and the comparison with stratigraphic data suggest that these phases are generated by the reflection from the top of carbonate rocks. The profile also shows five horizontal sub-layers within the pyroclastic layer, which can be interpreted through the core drilling stratigraphy. The Profile02, arranged parallel to the first one and spread to a distance of 10 m shows the same situation of the first profile, clearly marking the passage from pyroclastic sediments to carbonate deposits, at an average depth of 5.5 m. This interpretation is also supported by a sharp seismic velocity change, as shown in the following MASW analysis. MASW profiles were positioned at the same location of GPR acquisition, crossing the drilling location and the HVSR measurement station (Fig. 1). We used a RAS-24 seismograph, 24 bit, 117db @ 2 ms, equipped with vertical geophones of 4.5 Hz and 10Hz. The source consisted of an 8 kg hammer. We acquired 13 recordings (shots), along the same line, obtaining profiles with a length of 22 m, spacing of 2 m. Recordings were obtained with different offsets, and, for the same offset, geophones of 4.5 Hz and 10 Hz were used. The chosen profile has an offset of 4 m from the first geophone, and 10 Hz geophones. Processing was performed with the Geopsy software (http://www.geopsy.org/ ), a powerful open source software for processing of a wide variety of geophysical data. The first step was to import the files containing the field data, typically in SEG2 format, ensuring the acquisition parameters used for the imported files. The result of processing is the conversion of the MASW time-distance sections into a frequency-wave number spectrum. The assumption is that the most energetic part in the signals is composed by Rayleigh waves. So, the manual picking of the maxima in the spectrum allows to define the Rayleigh wave dispersion function. The obtained dispersion curve was then inverted to obtain the subsoil velocity 1D model, though the Dinver software (http://www.geopsy.org/ ). This inversion algorithm is based on a stochastic direct search method for finding models of acceptable data fit inside a multidimensional parameter space. The operator has the task of acting in order to reach the solution closest to the true subsoil structure. In doing this, constrains from geophysical, other than stratigraphic data methods, are advantageous. In our study case, data obtained from direct survey, HVSR and GPR data were crucial, in order to define a suitable initial parameter space (Fig. 3a). We started from a two layers initial model, formed by: a first layer with a linear increment of P-wave velocity (100-400 m/s), S wave velocity (80-300 m/s), Poisson’s ratio (0.3-0.4), and density (1500-2000 kg/m 3 ); a second layer with uniform parameters, which are left varying among models in the ranges: (400-1500 m/s) for the V P , (200-700 m/s) for the V S , (0.2-0.4) for the Poisson ratio, (1800-220 kg/m 3 ) for the density. The obtained best-fit solution is very close to the subsoil structure estimated by the GPR method and by the direct survey (Fig. 3b): a pyroclastic cover layer subdivided into 5 sub-layer overlying the carbonate basement. This paper reports results from measures carried out by using different geophysical methodologies in order to assess their applicability in estimating the near-surface Earth structure. The MASW method is the most popular techniques to derive the S-wave profile of the subsoil, in the hypothesis of a 1D layer structure. HVSR is definitely the cheapest and the most practical method. It allows a punctual inspection of the subsoil structure, in the hypothesis of a 1Dmedium with a strong impedance contrast at depth. It has identified the site’s fundamental resonance