Earthquake rupture and ground motion modeling of the 2016 Mw 6.2 Amatrice and Mw 6.5 Norcia, Central Italy earthquakes constrained by Bayesian dynamic finite-fault inversion

The complex evolution of earthquake rupture during the 2016 Mw 6.2 Amatrice and Mw 6.5 Norcia, Central Italy, earthquakes, was recorded by a uniquely dense seismic network providing an opportunity to better understand the processes that control earthquake dynamics, strong ground motion, and earthquake interactions. Here we use initial stress and friction conditions that are constrained by Bayesian dynamic source inversion (Gallovic et al., 2019a,b) to inform complex 3D dynamic rupture scenarios. The best-fitting forward models are chosen out of ~10^6 simulations. Such constrained, highly heterogeneous dynamic models fit very well with observations. Using SeisSol (www.seissol.org), we take into account non-planar (e.g., listric) fault geometry, topography, inelastic off-fault yielding, and surface rupture effects that can not be accounted for in the highly efficient yet simple forward models of the dynamic source inversion. We systematically investigate the impact of source complexity on rupture dynamics and ground motions up to 5 Hz. We find that in our Amatrice scenario: (1) In the extreme near field, fault listricity reduces synthetic ground motions by up to 50 percent on the foot-wall The hanging-wall shaking increases by up to 150 percent within 10 km from the fault due to wave-focusing effects; (2) In the frequency range under consideration, topography shielding de-amplifies PGV by up to 50 percent; (3) Off-fault plastic strain localization correlates with aftershock distribution and may indicate smaller-scale fault zone complexity. In the Norcia scenario, we find that normal stress changes due to free-surface effects weaken the shallow fault by up to 50 percent, causing the rupture front to leap ahead. Our results suggest that geometrical fault complexity is important for seismic hazard assessment adjacent to dipping faults but difficult to identify in kinematic source inversion.