GNGTS 2013 - Atti del 32° Convegno Nazionale

extension-rate measured by geodetic data, which is of the order of ~2-3mm/yr oriented NE- SW (see Fig. 1). In particular we test three different scenarios in which we consider as fault boundary: 1) the Alto Tiberina LANF, 2) the antithetic high-angle normal faults and 3) both faults. To estimate the best model solution we compute for each inversion the reduced chi squared of data and we use the Fisher test (Stein and Gordon, 1984) to evaluate the acceptance between n and n+1 plate models, i.e. to asses if more complex models are justified by the data. Tab. 1 reports the results of our tests, together with the corresponding slip-rates obtained in each inversion. As we can see from Tab. 1, the reduced chi-squared values are lower assuming geometry 3, for which also the F-test is positive. The corresponding fault slip-rates obtained from each inversion are representative of the attempt of inversions to reproduce the horizontal tectonic extension by mean elevated slip-rates on faults, which are higher than those proposed on literature (Collettini et al. , 2003; Pucci et al. , 2003), but on geometry 1 we obtain the same slip-rate as on Hreinsdóttir and Bennett (2010). Using two fault systems as plate boundary we obtain lower down-dip slip-rates more in agreement with geological information, giving a total horizontal extension comparable with geodetic signal. Considering thus the result here obtained with the numerous information proving a very likely activity of both faults, we could infer that the tectonic extension on this sector of Apennines should be accommodated by at least these two major fault systems. Tab. 1 – Reduced chi-squared values computed for the whole GPS dataset (tot) and for a selected set of stations (sel) located close to the northern sector of Umbria-Marche Apennines, for each inversion, performed with different setting geometries: 1 – only AT fault as block boundary; 2 – only antithetic faults as block boundary; 3 – considering both fault systems; the sixth and seventh columns report inferred down-dip fault slip-rates from elastic block modeling and in the last one is computed the corresponding horizontal slip-rate on extensional direction. Geometry Chi2rid (tot) Chi2rid (sel) ATF S.R. (mm/yr) GuF S.R. (mm/yr) Hor. S.R. (mm/yr) 1 - ATF 9.49 8.46 -2.4 - -2.3 2 - GuF 9.51 7.89 - -2.8 -2.1 3 - ATF + GuF 9.36 7.29 -1.5 -1.3 -2.4 Discussion and interseismic coupling on the ATF plane. Our block-modeling analysis suggests that on northern sector of the Umbria-Marche Apennines both the Alto Tiberina LANF and the antithetic, west-dipping, high-angle normal fault, here defined by the Gubbio fault, accommodate the tectonic extension measured by GPS stations. Nevertheless looking a velocity cross section about normal to the strike of major faults (see Fig. 2), we observe that a group of GPS sites, located between the two fault systems, show a systematic “flattening” of the velocity gradient, which is not well modeled by the three inversions discussed above. We tried to understand if the gradient can be better explained using different fault parameters for ATF and for this purpose we performed a series of inversions varying systematically locking depth and dip of the fault (Mastrolembo Ventura, 2012). Evaluating for each inversion the corresponding reduced chi-squared, we found a minimum for the ATF parameters that are close to the initial values (10 km of locking depth instead of 12 km). This result suggests that the ATF could have a significant elastic contribution on the observed geodetic gradient. The approach used so far considers faults as rectangular planes. To evaluate amodel including variable, non-uniform slip-deficit on the ATF, we generate a curved surface, meshed with triangular patches, using the GMSH software (Geuzaine and Remacle, 2009), and following the depth contour lines provided by faults isobaths from Mirabella et al. (2011). Moreover we modified the original code to invert for the slip-deficit distribution using a linear least-square 150 GNGTS 2013 S essione 1.2