GNGTS 2013 - Atti del 32° Convegno Nazionale

source. An S E value equal to zero represents an isolated source, and an increase in S E indicates an increased influence of the other faults. Thus, the S E is an intrinsic and static propriety of each single source due to the fault system geometry. Through this approach, one can obtain synthetic earthquake catalogues that contain the slip per event and time of occurrence for all sources in the fault system and compute the inter-event times, the mean recurrence times and the relative coefficients of variation (CV). We vary and SRV lt to simulate various earthquakes catalogues to investigate, using statistical approaches, the relationships among the geometric and energetic parameters of an active fault system, the stress interaction and the CV. Case study. In the simulation presented herein, we adopt a model containing nine active normal faults. The fault network used in this study is presented in Fig. 1a. That network contains both vertical and dipping faults to accommodate a combination of strike-slip and dip-slip events. The fault system (Fig. 1a) extends approximately 80 km in the strike direction and approximately 50 km perpendicular to the strike direction. The long-term fault-slip rates (ranging from 0.4 mm/a to 1.0 mm/a), the characteristic magnitude (ranging from 6.3 to 6.7) and the fault lengths (ranging from 20 to 35 km) are a simplification of the actual mapped fault system in central Italy (Boncio et al. , 2004; Peruzza et al. , 2011). All fault dips are taken as 60°, and the thickness of the seismogenic layer is assumed to be 15 km, which are both consistent with the available geological, seismological, geodetic and rheological data for central Italy (Boncio et al. , 2009). Results and discussion. The 240 as-produced synthetic earthquake catalogues contain a large number of events (approximately 500-1500), with magnitudes ranging from M w 5.4 to M w 7.0, over periods of 10 5 years. The previously described model setup is used to simulate long earthquake sequences on a system of nine faults characterised by normal faulting. Fig. 2a shows a window of the seismic histories for the nine sources over a period of 20 ka, assuming a and SRV lt equal to 0.02 bar/a and 0.5, respectively Fig. 2b depicts histograms of the inter-event times. The inter-event time distributions exhibit a wide variety of behaviours, from peaked around a single mean recurrence interval (sources A, B, and F), to double-peaked (sources C and G) to widely scattered (sources D, E, H and I). We computed highest CVs for sources A, B, E and F, with sources C and G returning the lowest CV values, confirming these different behaviours. We demonstrate that the shape of the histograms appears to be controlled by the geometries of the sources and the fault system and by the efficacy of the stress transfer in modulating the earthquake recurrence intervals. Decreases (unloading) or increases (loading) in the stress due to the spatial relationship between the slipping source and the location of the receiver (Fig. 1e – in the shadow zone, i.e., across the strike, or in the positive stress zone, i.e., along the strike) influence the recurrence intervals. For example, we noted long recurrence intervals for sources C and G, located where the receiver stress decreases from a source across the strike and the two sources are the largest in the system. Therefore, the effect of ΔCFS is reduced for two reasons: a) the mean ΔCFS is low after a neighbour source ruptures; b) their rupture threshold (correlated with their characteristic magnitudes) is higher than those for smaller sources such as A and I. This finding implies that an earthquake occurring on source H contributes more to the delay in an event on source I than on source G; similarly, an earthquake on source D has a lower impact on the loading variable of source G with respect to the slip of source F on source I. Conversely, source E, located in the centre of the fault system, receives positive and negative ΔCFS from all surrounding sources, resulting in the highest CV. Therefore, the recurrence distributions for an individual source within a fault system depend on its position within the fault system. These characteristics indicate that the gross fault system geometry plays a primary role in establishing the stress evolution characteristics that control earthquake recurrence. We quantify the influence of the position of a fault embedded in a fault system through the S E coefficient (described in the methods section). On first glance, the highest S E returns the highest CV (sources B and E), with the lowest CVs occurring in sources C and G, which have the lowest S E values. 161 GNGTS 2013 S essione 2.1