GNGTS 2014 - Atti del 33° Convegno Nazionale

complex deformation pattern and different tectonic domains. �� The area is therefore very peculiar from the tomographic point of view, with large lateral variation of velocity and Vp/Vs ratio, that is related to the high level of fracturing and, on the east, to the inhomogeneity of the medium. In addition, �� the upper crust is characterized by marked lateral and depth variations of the elastic moduli pattern, closely related to variability in rock mechanical properties. �� The Friuli region area is mainly characterized by compressive regime with different orientation of the faults, while west-Slovenian region by strike slip. Gentili and Gentile (2013) results are compared with the other achieved in the same area (heat flow estimates, gravimetric data, Vp and Vp/Vs tomography) trying to obtain a joint interpretation. The next section summarizes the method adopted for 3D S waves tomography adopted by Gentili and Gentile (2013). In the following, the results on reliability estimate are shown. Finally, the comparison with other data in the same area is discussed. The method. The attenuation is modelled in literature either using the parameter k (e.g. Anderson and Hough 1984), or using the quality factor Q , a dimensionless parameter introduced to quantify the fractional energy loss per cycle of oscillation as Q =2 πE / ΔE (Aki and Richards, 1980), where E is the energy and ΔE is the lost energy. Q and k are related; in particular Hough and Anderson (1988) modelled k as: (1) where Q I is the frequency-independent part of Q , and Q is parameterized as: Q ( f ) -1 = Q D ( f ) -1 + Q I -1 (2) Q D and Q I are respectively the frequency dependent and independent part of Q . In their work they use Q I ( z ) instead of Q I ( r ) because in their simplified model Q I depends only on the depth z (plane and parallel layers). However, their model can be generalized to r dependence (Hough and Anderson, 1988). �� Gentili and Gentile (2013) adopted Bianco et al. (2005) estimation, Q D ( f )=251 f 0.7 and assumed a negligible dependence of the geometrical spreading on frequency. They used 977 3D seismic traces corresponding to 156 events from 1994 to 2011 recorded by 28 seismic stations mainly belonging to Friuli and Veneto seismic network (FV), managed by Istituto Nazionale di Oceanografia e di Geofisica Sperimentale (OGS), but also recorded by Slovenian seismic network and two temporary networks. All the earthquakes have been re-located by a 3D crustal model using SIMULPS12 (Evans et al., 1994) tomography code obtaining horizontal and vertical location errors do not exceed 0.05 and 0.1 km, respectively (Bressan et al., 2009). The k value was estimated from the slope of the amplitude Fourier spectrum of acceleration data after correction for the frequency dependent part of the quality factor. The spectral band adopted for the analysis was determined by selecting the part of the spectrum where a linear decay was clearly evident. In order to avoid that the corner frequency was inside the selected frequency range, a threshold on earthquakes magnitudes was set to 3; k was computed both from the N-S and the E-W horizontal components of the signal �� as their �� weighted mean. The equation to be solved is: y = M x + e (3) where y is the vector of the differences between the real values of k and those obtained with the theoretical model, M is the matrix of the partial derivatives respect to QV , QV is the element wise multiplication of the Q quality factor 3D matrix and the V velocity 3D matrix, x is the variation of QV along the path (model parameters) and e is the vector of errors. It is important to remark that in Q inversion procedure the S waves velocity field is taken as input. It is, therefore, necessary to use a reliable velocity information in order not to bias results on Q inversion. The 3D inversion was done by an iterative damped least squares method using the program SIMUL2000. 146 GNGTS 2014 S essione 1.2