GNGTS 2013 - Atti del 32° Convegno Nazionale

In this area, two distinct seismicity trends, one following the Penninic front to the west and one following the Austro-Alpine front to the east, converge delimiting an almost aseismic corridor (Giglia et al. , 1996). Following the large-scale zonation of Delacou et al. (2004) and Barani et al. (2010), the Dora Maira massif can be ascribed to the internal sector of the Western Alps, a continuous zone of extension characterized by low-to-moderate but relatively frequent earthquakes with hypocenters down to approximately 15-20km depth. Analyzing focal mechanism solutions and Global Positioning System (GPS) data shows that extension (i.e., T -axes) is virtually perpendicular to the structural trend of the Alps (e.g., Eva and Solarino, 1998; Calais et al. , 2002; Nocquet and Calais, 2003, 2004), following a radial pattern (e.g., Frechet, 1978; Nicolas et al. , 1990; Champagnac et al. , 2004; Delacou et al. , 2004). The focal mechanisms of the two major events of the swarm, which were computed by applying the first onset methodology through the FPFIT program (Reasenberg and Oppenheimer, 1985), agree with the regional seismotectonics, which, as stated above, is characterized by a prevalent extensional-transtensional regime. In particular, the orientation of the T -axis is in agreement with the direction of extension (approximately E-W) in this sector of the Alps. Regarding the second seismic episode considered in this study, the Lunigiana-Garfagnana area represents the western border of the northern Apennines. This area is characterized by a complex structural setting which is related to the evolution of the northern Apennines, including three major episodes: 1) consumption of oceanic crust driven by a west-dipping slab, 2) post-collisional evolution, and 3) rotation of the Corsica-Sardinia block with opening of the Tyrrhenian sea (Cattaneo et al. , 1983; Ponziani et al. , 1995). The structures derived from these main events are related both to compressive and extensive forces. Compression, which is associated to the collision between Africa and Europe, is responsible for the emplacement of different tectonic units belonging to different paleogeographic domanis (the Liguride allochton formed by oceanic crust, the Tuscan unit, and the Umbria-Marche-Romagna unit) (Castaldini et al. , 1998). Extensional stresses are due to the roll-back of the subducting Adria-Ionian lithosphere (Negredo et al. , 1999), which created tectonic depressions such as the Lunigiana and Garfagnana basins (Bartolini and Bortolotti, 1971). The Lunigiana and Garfagnana basins are two NW-SE-striking asymmetric grabens, originated in the hanging wall of regional low-angle detachment faults, sloping gradually beneath the Apennine chain (Brozzetti et al. , 2007). The Lunigiana sequence took place between the Lunigiana and the Garfagnana basins, activating a fault system with anti-apenninic direction. The focal mechanism of the major event ( Ml = 5.2), as derived from the INGV moment tensor solution database (http://ingvterremoti. wordpress.com/ ), shows a strike-slip rupture process confirming that the fault responsible of the main shock is NE-SW oriented. Methodology. The waveform similarity analysis is used in this work to identify and characterize the seismogenic structures activated during the two seismic episodes. This technique is used to identify groups of events (i.e., earthquake families) characterized by similar locations, source mechanisms, and by similar propagation patterns. The waveform similarity analysis is performed on signals recorded along the vertical component by the seismic stations nearest to the area affected by the seismic events. More precisely, DOI and PZZ stations (Fig. 1) are considered for the Sampeyre swarm whereas EQUI and FIVI stations (Fig. 2) for the Lunigiana sequence. The cross-correlation analysis is performed on selected subsets of events with magnitude greater than or equal to 1.0, gap in azimuth coverage (i.e., maximum azimuthal distance between two nearby-distance seismic stations) less than or equal to 200°, horizontal and vertical location error less than 5 km. First, all waveforms are filtered by a band-pass filter to reduce the bias of noise and high frequency wiggles as well as minor waveform dissimilarities due to differences in magnitude, focal mechanisms, and small-scale heterogeneities. Then, seismogram cross-correlation is performed on pairs of signals including both the P-wave and S-wave onsets by using the normalized cross-correlation function (e.g., Augliera et al. , 1995; Cattaneo et al. , 1997, 1999). Multiplets are defined as groups of events 20 GNGTS 2013 S essione 1.1