GNGTS 2019 - Atti del 38° Convegno Nazionale

GNGTS 2019 S essione 1.1 5 A UNIFIED NUMERICAL MODEL FOR THE SIMULATION OF THE SEISMIC CYCLE IN DIP-SLIP ENVIRONMENTS: EXAMPLES FROM ITALY M. Albano 1 , S. Barba 1 , C. Bignami 1 , C. Doglioni 1 , E. Carminati 2 , M. Saroli 3 , M. Moro 1 , S. Stramondo 1 , S. Samsonov 4 1 Istituto Nazionale di Geofisica e Vulcanologia, Roma 2 Dipartimento di Scienze della Terra, ‘Sapienza’ Università di Roma 3 Dipartimento di Ingegneria Civile e Meccanica, Università degli Studi di Cassino e del Lazio Meridionale, Cassino, FR 4 Natural Resources Canada, Ottawa, ON KA E, Canada Introduction. According to the concept of the seismic cycle, earthquakes are the result of the strain accumulation in the earth’s crust over a variable decade to millennial period, i.e., the interseismic stage, followed by a sudden stress release at a crustal discontinuity, i.e., the coseismic stage, finally evolving in a postseismic stage (Scholz, 2019). Commonly, the seismic cycle is modelled with analytical and numerical approaches. Quasi- static analytical methods simulate the interseismic coupling, the coseismic dislocation and the postseismic relaxation independently and assuming an elastic, viscoelastic or poroelastic half-space. Often, these models impose the slip on single or multiple planar sources to infer fault geometry, slip distribution and regional deformations in order to fit the available geodetic or seismological measurements, often regardless of the magnitude and orientation of the interseismic gravitational and tectonic forces (Anderlini et al. , 2016; Atzori et al. , 2009). Numerical approaches allow simulating complex geometries in heterogeneous media and at different modelling scales, assuming elastic, viscoelastic, or elasto-viscoplastic constitutive laws. However, such models often impose the slip on the fault plane to simulate the observed coseismic dislocation or the propagation of the seismic waves (Trasatti et al. , 2011), or they adopt ad-hoc boundary conditions to investigate the interseismic stress accumulation or the postseismic relaxation for specific cases (Carminati and Vadacca, 2010). We contribute to the understanding of the seismic cycle associated to a single fault segment by developing a numerical model to simulate the long-term crustal interseismic deformation, the coseismic brittle episodic dislocation, and the postseismic relaxation of the upper crust within a unified environment for both normal and reverse fault events. This model is developed to simulate typical extensional and compressional earthquakes in Italy (Fig. 1a) and includes the forces acting during the interseismic period, i.e., the lithostatic load and the horizontal stress field (Finocchio et al. , 2016). We adjusted the setup of our model to simulate the interseismic, coseismic and postseismic phases for two major seismic events in Italy, the 2009, M w 6.1 L’Aquila normal fault earthquake (Fig.1b) and the 2012, M w 5.9 Emilia-Romagna reverse fault earthquake (Fig. 1c). The results of our analysis, compared with geodetic and InSAR data from the literature, show that the proposed numerical model is able to reproduce the seismic cycle associated with the investigated events. Methods. We developed a first-order model that combines poroelasticity, discrete discontinuities simulating detachments and faults, tectonic forces, and gravity. To this purpose, we built two plane-strain numerical models by exploiting the finite element commercial code MSC Marc 2018 (MSC Software Corporation, 2018). The 2D models (Sections A, and B Fig. 1a and Fig. 2) extend 220 km horizontally and to a depth of 40 km (Fig. 2a, b) (Finocchio et al. , 2016) and are almost orthogonal to the strike of the earthquake causative faults. Discontinuities in the model’s mesh are introduced to simulate a steadily shearing fault in the interseismic phase (segment n°1 in Fig. 2) and the earthquake causative fault in the coseismic phase (segment n°2 in Fig. 2), whose dip and length are defined according to the retrieved fault geometry from previous inversions of geodetic and satellite data (Atzori et al. , 2009; Pezzo et al. , 2013). Mesh discontinuities are modelled using a contact interface, where nodes are doubled so that the upper and lower parts of the domain can move relative to each other. Unlike common, quasi-static, analytical and numerical modelisations, no forces or displacements are imposed on the fault’s