Constraining families of dynamic models using geological, geodetic and strong ground motion data.

The dense coverage and high quality of geodetic and seismic data allow to image intriguing details of earthquake rupture kinematics, however the mechanical viability of such models is not guaranteed. Fully dynamic models are needed to provide a physics-based understanding of how earthquakes start, propagate, and stop. Earthquake dynamic rupture simulations couple the non-linear interaction of fault yielding and sliding behavior to seismic wave propagation. Using modern numerical methods and computing infrastructure allows for realistic 3D dynamic rupture scenarios of complex, multi-fault earthquakes. Although there have been significant improvements in forward modeling scenarios, fully dynamic inversion of recorded moderate-to-large events are still sparse and rely on simplifying the forward problem. Dynamic earthquake inversion models can also be affected by parameter trade-offs, the a-priori choice of the constitutive law, the limited frequency bandwidth considered, among many other factors. In this work we develop a systematic approach to constrain spontaneous dynamic rupture models based on a given kinematic model, allowing us to evaluate the dynamic consistency of the latter while matching seismic and geodetic observations. We consider the well-recorded 2016 Mw6.5 Norcia normal faulting event as a case study. Specifically, we design and analyze "families" of complex multi-fault dynamic models, each recovering main kinematic characteristics but varying in terms of their initial dynamic parameters which determine frictional strength and stress. We then examine the consistency of friction coefficients (static and dynamic) with the mechanical properties of the rocks where the earthquake occurred. Our approach permits to validate the viability of published kinematic models and classify dynamic rupture scenarios that match observations and geological constraints and our results are an important contribution toward a new generation of dynamic earthquake inversion modeling.