GNGTS 2016 - Atti del 35° Convegno Nazionale

86 GNGTS 2016 S essione A matrice In Fig. 1 are reported some examples of the interferograms produced during the sequence. Each fringe corresponds to a ground displacement equal to half of the sensor radar wavelength (shown in Tab.1). The LoS ground displacement was obtained from the unwrapping of the interferometric fringes. The confirm the surface subsidence observed in past normal fault earthquakes in the Apennine (e.g., the Mw 6.3 L’Aquila earthquake, 2009), extending for about 25 km along the N-NE direction. The maximum displacement value was about -20 cm in correspondence of the area of Accumoli. The analysis of the LoS displacement maps, allowed to detect some local surface phenomena, probably related to slope instability; they were especially well depicted in the co-seismic deformation map from CSK data, which provide higher spatial resolution and accuracy. Thanks to special support from ESA, it was possible to fully exploit the Sentinel-1 constellation capacities. ESA provided data from Sentinel-1B satellite even though it is still in the commissioning phase and therefore not fully operational. It was thus possible to generate “cross-sensor” interferograms with a 6-day temporal baseline (instead of the ordinary 12 days) consequently doubling the observation temporal sampling. Source modelling. The InSAR data used in the inversion procedure were about 19500 measurements obtained by sub-sampling 5 unwrapped interferograms (those marked with * in Tab. 1) and 107 CGPS site displacement measurements obtained by the INGV-CNT Working Group “GPS Geodesy” (2016). The following modelling procedure was adopted: the geocoded displacement maps were sampled over a double resolution grid, with a point every 500 m in the area with significant deformation and every 2000 m outside. Then, the datasets were modeled through an analytical elastic dislocation model (Okada, 1985), with the fault plane geometry and rake estimated using a non-linear inversion algorithm (without any external constraint) and the slip distribution obtained through a damped linear least square inversion, with a positivity constrain. A single and double source were tested. In Tab. 2 we summarize the inversion results for both hypotheses. Conclusions. The single and double fault models show very similar slip distributions (Fig. 2) and the fit with the data is similar (Tab. 2). Though the majority of fault slip appears to be located below a depth of about 4 km and there is limited slip towards the surface, if the fault plane is geometrically extended to intersect the surface, its trace runs parallel and very close (within ± 800 m) to the trace of the Gorzano-Laga-Vettore fault system (Fig. 2), suggesting that this is the fault responsible for the mainshock. For the two faults model the same pattern can be observed for the South fault, while the trace of the north fault turns toward NE and should emerge about 3 km to the east of Mt. Vettore. The models are also coherent with the aftershock distribution, especially for the northern segment, although the relocated seismicity seems to highlight geometrical complexities not fully recovered by the geodetic models. Both models show that, moving southward from the Tab. 2 – Parameters of the modelled seismic sources. Note: The above values of length, width and depth of each fault are related to the area with larger slip value. Model Observed Geodetic Length Max Depth Strike Dip Rake Max RMS Mw Magnitude width Slip Single fault 6.0 (+ 5.3) 6.2 –21 km –9 km –1500 m 164° 46° -73° 120 cm 1.4 cm Double fault- 6.0 (+ 5.3) 6.2 –8 km –8 km –3000 m 175° 39° -65° 140 cm 1.4 cm North Double fault- –12 km –5 km –2500 m 165° 51° -70° 130 cm South