GNGTS 2018 - 37° Convegno Nazionale

GNGTS 2018 S essione 1.1 57 energy and first calculate the rotated spectra on the unsaturated acceleration traces for the entire 3-s-time window of the seismogram. Results of this analysis, compared to spectral ratios analysis on noise samples recorded at the same station (GdL-INGV, 2017) reveal that azimuthal distribution of seismic energy on the horizontal components may be influenced by local site effects. Therefore, we perform azimuthal spectral analysis for smaller time-windows to separate the energy contribution of different seismic phases and assess the polarization of the P-wave from particle motion (Fig. 1). We observe clear positive onsets on all three components and a strongly linearly polarized P-phase motion, with north-northeast–south-southwest horizontal particle motion suggesting a back azimuth of ~215° and an incidence angle of ~20° with respect to the vertical direction. Compared to the location reported by ISIDe (2016), the epicentral distance with respect to station IOCA is very similar, while the back-azimuth is rotated towards SW by 20°. Moment tensor solutions. A number of discrepant focal mechanism solutions have been proposed for the Ischia earthquake, based on time domain regional moment tensor inversion (Fig. 2). We consider the following three reference solutions: (1) Regional Centroid Moment Tensor (RCMT; Pondrelli et al. , 2006), Time Domain Moment Tensor (TDMT; Scognamiglio et al. , 2009), and a method developed by the Saint Louis University (SLU; Herrmann et al. , 2011). The large discrepancy among the proposed focal mechanisms can be attributed to two main factors: the observation capability and the seismic source depth. Following the method described in Cesca et al. , (2013), we performed spectral and waveform-based moment tensor inversions to determine the seismic source geometry, by assuming a pure double couple and a full moment tensor model and using the on-shore stations of the Italian Seismic Network (ISN) located at regional distances (Fig. 2). In comparison to former inversions, we use a greater number of seismic stations (up to 14 stations) to reduce the azimuthal gap (down to ~200°). Thanks to the improvement in stations’ geometry and the fit of high frequency data we can better resolve the centroid depth and the moment tensor. We first invert for a double couple (DC), obtaining a Mw 3.9 normal fault with a best fit solution at a depth of 8 km. This solution is in good agreement with the one calculated by TDMT (Fig. 2). We additionally perform a full moment tensor inversion, to assess the presence and robustness of isotropic and CLVD components. Comparative results of full MT and DC inversions (Fig. 3) demonstrate a large improvement of spectral and waveform fit, when a very shallow MT solution is chosen, at a depth of 2 - 4 km. The best fitting MT solution, for a depth of 4 km, is characterized by a significant negative isotropic component of 36% (contraction), a negative CLVD of 26% and a normal faulting DC component of 38%. The seismic moment amounts to 2.30*10 15 Nm, corresponding to a moment magnitude of Mw 4.1. The ~0.2 increase in Mw- magnitude, in comparison to the best DC solution and other reference solutions, can be mostly attributed to the non-DC term. Fig. 2 - Comparison of moment tensor solutions by: RCMT (green, best DC source), SLU (blue, best DC source), TDMT (purple, DC component of the best deviatoric MT) and this work (red, best DC source).