GNGTS 2013 - Atti del 32° Convegno Nazionale

The complete procedure consisted of three different inversions in which the grid spacing was progressively shrinked from 12x12x4 km 3 to 3x3x1 km 3 . The smallest grid space has been chosen based on the spatial resolution as inferred from specific checkerboard-type resolution tests. As a quantitative measure of the model resolution, the semblance between the true and recovered checkerboard anomalies has been computed which allowed to delineate the optimal resolved regions. For both the parameterizations 6x6x2 km 3 and 3x3x1 km 3 , the minimum grid size used in our study, the extent of the resolved areas increases with depth according to the earthquakes location and the best resolution is obtained for depths ranging between 4 and 14 km for the first parameterization and 4-10 km for the latter (Fig. 2). Results. Both the P- and S-wave velocity models indicate the presence of a strong lateral variation in the seismic velocity along a direction orthogonal to the Apeninic chain, in the 4-8 km depth range (Fig. 2). This variation defines two geological formation domains which are characterized by a relatively low (3.5-4.8 km/s) and high (5.2-6.5 km/s) P-velocity, respectively. The zone where the sharpest lateral transition occurs in the NE direction, is well correlated with the location of the NW-SE oriented, primary normal fault associated with the 1980, Ms 6.9 earthquake, which cuts at SW the outcrops of the carbonatic Campanian platform, and separates at NE the older Mesozoic limestone formations from the younger Pliocene-Quaternary basin deposits. The spatial distribution of seismicity delineates at south-west the possible border of the Irpinia master fault, while at northeast it shows a more diffused pattern due to the presence of a system of highly organized, sub-parallel normal faults as it has been inferred from the fault mechanisms and the coherent orientation of the tensional axes, consistently with a NE-SW dominant extensional stress regime (De Matteis et al. 2012). We have extracted two 1D P-wave velocity models from the final 3D model obtained by using the 3x3x1km 3 parameterization, in correspondence of two deep wells (Fig. 3a). We note Fig. 3 – Earthquake locations and comparison of 1D borehole and tomography P-velocity models on- and off- fault zone. (a) Location of selected microearthquakes and stations used for the tomographic analysis. Sites of exploration wells San Gregorio Magno (SGM) and San Fele (SF) are also shown, together with trace of cross- section of Fig. 1B. (b) Comparison of the tomographic (continuous line) and sonic-log (dashed line) 1D P-velocity models at the borehole sites SGM and SF. (c) Vp / Vs ratio vs. depth measured at on-fault SGM (red line) and off-fault SF (blue) sites. (d) Histogram of number of events vs. depth for the earthquakes analyzed in this study. 9 GNGTS 2013 S essione 1.1