GNGTS 2018 - 37° Convegno Nazionale

140 GNGTS 2018 S essione 1.1 It consists of two separate routines: PEGRN, which calculates the Green functions needed for the strain and stress computations, and PECMP, which computes the poroelastic variables (such as deformation, stress tensor, pore pressure, and Darcy flow) at receiver locations and at selected time steps. During the simulation, each fault can be activated, i.e. undergoing slip steps, at different time instants. In the postseismic stage, the faults are assumed as locked. In our simulations for both the May 20 and May 29 events, we use the coseismic slip distribution models, the faults geometries, and the elastic layering proposed by Nespoli et al. (2017). The four layers of the poroelastic half-space have different diffusivity values representing a decrease of permeability over depth as proposed by Ingebritsen and Manning (2010). Results and Discussion. At the surface, the modeled poroelastic, postseismic horizontal displacement reaches the maximum absolute value of 1 cm about 10 days after the May 20 mainshock. This duration is related to the draining time of the shallowest layers of our model, since at this time scale the deeper layers have a second-order influence on surface deformation. After 1 year, the modeled postseismic poroelastic displacement field is in accordance with the IC3 (Nespoli et al. , 2018), so we interpreted this component as the poroelastic contribution to surface displacement, that is mainly confined in the near field. This interpretation is mainly supported by GPS data since InSAR time series can be temporally too sparse to sample the related transient displacement signal. In particular, at point 1, the modeled poroelastic Line Of Sight (LOS) displacement explains about 50% of the total measured displacement: here the afterslip significantly contributed to the total postseismic displacement. Positive postseismic poroelastic CFF changes, spatially correlated with the regions seismically activated, are estimated starting from the May 20 mainshock to 30 days of simulations, nonetheless, they are small (~10 kPa). Finally, we found an important fluid flow along the faults’ strike direction (Fig. 2), especially when assuming a heterogeneous slip distribution. This means that, in order to simulate the evolution of pore-pressure induced by earthquakes, a 3D fluid flow modeling can be very important. References Albano M., Barba S., Solaro G., Pepe A., Bignami C., Moro M., Saroli M. and Stramondo S.; 2017: Aftershocks, groundwater changes and postseismic ground displacements related to pore pressure gradients: Insights from the 2012 Emilia-Romagna earthquake . J. Geophys. Res. Solid Earth, 122 , doi:10.1002/2017JB014009 . Chan K., Lee T. W. and Sejnowski T.; 2003: Variational Bayesian learning of ICA with missing data. Neural Comput., 15 (8), 1991–2011. Fig. 2 - Coseismic pore-pressure distribution at 10 km depth. Yellow and orange lines: borders of the Ferrara and Mirandola faults emerging above the panels. Black arrows: fluid flow streamlines (logarithmic scale). Stars represent the May 20 and May 29 hypocenters.