GNGTS 2015 - Atti del 34° Convegno Nazionale

GNGTS 2015 S essione 1.1 33 faults and inhibits faulting along thrusts (Carminati et al. , 2004), the two types of faulting are asymmetric in terms of geological and mechanical behavior (Doglioni et al. , 2011). We model a fault cross-cutting the brittle upper crust and the ductile lower crust. In the brittle layer the fault is assumed to have stick–slip behavior, whereas the lower ductile crust is inferred to deform in a steady-state shear. Therefore, the brittle–ductile transition (BDT) separates two layers with different strain rates and structural styles. This contrasting behavior determines a stress gradient at the BDT that is eventually dissipated during the earthquake. During the interseismic period, along a normal fault it should form a dilated hinge at and above the BDT. Conversely, an over-compressed volume should rather develop above a thrust plane at the BDT. On a normal fault the earthquake is associated with the coseismic closure of the dilated fractures generated in the stretched hangingwall during the interseismic period. In addition to the shear stress overcoming the friction of the fault, the brittle fault moves when the weight of the hangingwall exceeds the strength of the dilated band above the BDT. On a thrust fault, the seismic event is instead associated with the sudden dilation of the previously over-compressed volume in the hangingwall above the BDT, a mechanism requiring much more energy because it acts against gravity. In both cases, the deeper the BDT, the larger the involved volume, and the bigger the related magnitude. We tested two scenarios with two examples from L’Aquila 2009 (Italy) and Chi-Chi 1999 (Taiwan) events. GPS data, energy dissipation and strain rate analysis support these contrasting evolutions. Our model also predicts, consistently with data, that the interseismic strain rate is lower along the faultsegment more prone to seismic activation (Doglioni et al. , 2011). Earthquakes deliver in few seconds the elastic energy accumulated in hundreds of years. Where and when will be the next earthquake remains a difficult task due to the chaotic behavior of seismicity and the present lack of available tools to measure the threshold of the crustal strength. However, the analysis of the background strain rate in Italy and the comparison with seismicity shows that larger earthquakes occur with higher probability in areas of lower strain rate.We present a statistical study in which a relationship linking the earthquake size (magnitude) and the total strain rate (SR) is found. We combine the information provided by the Gutenberg– Richter law (GR) of earthquake occurrence and the probability density distribution of SR in the Italian area. Following a Bayesian approach, we found a simple family of exponential decrease curves describing the probability that an event of a given size occurs within a given class of SR. This approach relies on the evidence that elastic energy accumulates in those areas where faults are locked and the SR is lower. Therefore, in tectonically active areas, SR lows are more prone to release larger amount of energy with respect to adjacent zones characterised by higher strain rates. The SR map of Italy, compared with 5 years seismicity supports this result and may become a powerful tool for identifying the areas more prone to the next earthquakes (Riguzzi et al. , 2012). We find that geodetic strain rate (SR) integrated with the knowledge of active faults points out that hazardous seismic areas are those with lower SR, where active faults are possibly approaching the end of seismic cycle. SR values estimated from GPS velocities at epicentral areas of large historical earthquakes in Italy decrease with increasing elapsed time, thus highlighting faults more prone to reactivation. We have modelled an exponential decrease relationship between SR and the time elapsed since the last largest earthquake, differencing historical earthquakes according to their fault rupture style. Then, we have estimated the characteristic times of relaxation by a non-linear inversion, showing that events with thrust mechanism exhibit a characteristic time (~990 yr) about three times larger than those with normal mechanism. Assuming standard rigidity and viscosity values we can infer an average recurrence time of about 600 yr for normal faults and about 2000 yr for thrust faults (Riguzzi et al. , 2012). The fault activation (fault on) interrupts the enduring fault locking (fault off) and marks the end of a seismic cycle in which the brittle-ductile transition (BDT) acts as a sort of switch.