GNGTS 2019 - Atti del 38° Convegno Nazionale

GNGTS 2019 S essione 3.2 697 of couples of receivers and their common, in-line, sources are cross-correlated and the path- average DCs are extracted from the amplitude maxima of the cross-correlation matrix. The reference DCs are used as indicators of the expected phase velocities at each frequency, to guarantee that the correct amplitude maxima are picked. If many records are available for the same receiver couples, their cross-correlation matrices are stacked before picking the DC, to increase S/N ratio. Data coverage provides an indication of the reliability of the result, and it is improved with the increasing number and spatial distribution of the paths. The path-average DCs are inverted with the tomographic inversion scheme of Boiero (2009). This method discretizes the subsurface into a grid, and a 1D V S model is obtained at each grid position. The use of lateral constraints to relate the V S of all models, according to a priori knowledge, is possible. The process is fast since it avoids the building of phase-velocity maps and estimates directly V S . For the estimation of statics, the inverted pseudo-2D V S model is transformed into V Sz . Then, the pseudo-2D V Pz model is computed at each model point position j, as: (2) Finally, the one-way time (static shift) at datum depth z d is obtained at each position j from the V Pz model, as: (3) Results. The synthetic V S model is shown in Fig. 1. The density values of the layers are equal to 2000 Kg/m 3 , 2100 Kg/m 3 and 2200 Kg/m 3 and their V P equal to 180 m/s, 240 m/s and 350 m/s. At position 60 m, the thickness of the second layer decreases from 8 m to 3 m, separating the model into two laterally-homogeneous areas (noted as “area 1” and “area 2” in Fig. 1). The simulated dataset consists of 101 receivers, spaced at 1 m, and 25 shots, spaced at 5 m (black dots and red circles in Fig. 1, respectively). The simulated record length was 2 s and the sampling interval was 1 ms. Fig. 1 - Synthetic model. The colors indicate V S , given by the color bar. The black dots and the red asterisks represent, respectively, the positions of the simulated receivers and shots.As “area 1” and “area 2”, the two separate homogeneous areas are denoted. We extracted two reference DCs (one for each homogeneous area), which are shown in Fig. 2a. For their extraction, we applied the Phase-Shift method (Park et al. , 1998) using all the recordings of each area. To obtain the reference V Sz profiles, used in the W/D method, we inverted the reference DCs using a Monte Carlo algorithm (Socco and Boiero, 2008). In Fig. 2b, we show the estimated W/D of each area and in Fig. 2c, the retrieved apparent Poisson’s ratio. To extract the path-average DCs, we selected all the receiver couples and stacked the cross- multiplication matrices of all possible shots. The DCs were picked automatically, as the spectral