GNGTS 2013 - Atti del 32° Convegno Nazionale

the same of Cormier (1982) with the difference that only frequency independent part of Q is considered. In the works on k estimation the term Q D is generally considered (Anderson and Hough, 1984, Castro et al. , 1997, Franceschina et al., 2006, Bressan et al. , 2007). Vice versa, when t* is estimated for attenuation tomography inversion, the frequency dependent term is usually neglected (Rietbrock 2001, Haberland and Rietbrock 2001, Olsen et al. , 2003, Eberhart- Phillips et al. , 2005). The opportunity to considering or not the Q dependence on frequency is debated. The reason for neglecting it is the difficulty in the evaluation of this term for the whole area under study and/or the lower dependence of the quality factor on frequency for higher frequencies. The latter reason is debated. Several different authors analysed the area under study in this paper and estimated the frequency dependence of Q . Console and Rovelli (1981) obtained a dependence of ~ f 1.1 in the frequency range 0.1-10 Hz, Castro et al. (1996) obtained a dependence of ~ f 1.01 in the frequency range 0.4-25 Hz, Malagnini et al. (2002) obtained a dependence of ~ f 0.55 in the frequency range 0.5-14 Hz. Estimates of k in the study area of this paper were performed by Franceschina et al. (2006) and Bressan et al. (2007) considering a term Q D =78 f 0.96 [according with Govoni et al. (1996) in the frequency range 1-25 Hz]. In a successive paper, Gentili and Franceschina (2011) considered the relation Q D ( f ) = 251 f 0.7 correspondingly to the total quality factor estimated for this area by Bianco et al. (2005) in the frequency range 0.5-16 Hz. In this work, for coherence with previous ones in the same area, we will not neglect the frequency dependence of Q . In particular, we will adopt Bianco et al. (2005) estimation, like Gentili and Franceschina (2011). Considering such correction does not affect too much the value of Q for small values of Q . For example, considering the value of Q D at a frequency in the middle of our spectral window (Eberhart – Phillips et al. , 2005), say 15 Hz, for Q I =20 we have Q =19.8. For very low attenuation, the effect of considering frequency dependence is larger: e.g. with Q I =1000 we have Q =626. For coherence with previous papers notation, we will call the frequency independent part of the quality factor for S waves simply Q S instead of Q IS . Attenuation parameter estimation. Assuming a negligible dependence of the geometrical spreading on frequency, we estimated the parameter by correcting the S-wave spectra for the frequency dependent part of the quality factor. Both the N-S and the E-W horizontal components of the signal were used. The S wave window was manually selected for both traces and the resulting data were tapered by a 5% cosine taper and padded with zeros before applying the FFT. The resulting spectra were smoothed by a sliding Hann window (Oppenheim and Schafer, 1999) of 0.5 Hz half-width. The spectral band adopted for the analysis was determined by selecting the part of the spectrum where a linear decay was clearly evident. In particular, we selected data in a frequency band [ f 1 , f 2 ]: f 2 is the frequency at which the noise level starts contaminating the signal and, in this study, ranged from 20 to 45 Hz. In order to determine f 1 , accordingly with Gentili and Franceschina (2011), we estimated k values for increasing values of the minimum frequency, we plotted the obtained k as function of f and selected k ( f 1 ) values in an observed range of stability of this function by a visual inspection of each plot; f 1 , generally depending on event size, ranged from 5 to 10 Hz. We adopted a threshold on the duration magnitude of analysed earthquakes of M D >3 in order to ensure the validity of the condition f 1 > f C , where f c is the corner frequency. In fact, accordingly with previous analysis in this area (Bressan et al. , 2007) for M D =3, f C 4.6 Hz. In order to fit the amplitude spectrum, we used the least absolute residual method. This method minimizes the absolute difference of the residuals rather than the squared differences, thus decreasing the influence of outliers on results. The k values estimated from the two horizontal components of each record were averaged by computing the corresponding weighted mean and standard deviation. In order to increase the robustness of the method, the weights were chosen as the inverse of the error of each single fit. Seismicity distribution and tectonics of the area. The area we analysed is sited in north- eastern Italy (Friuli Venezia Giulia region) and western Slovenia between the outer front of 70 GNGTS 2013 S essione 1.1