Validation of Broadband Ground Motion from Dynamic Rupture Simulations: towards better characterizing seismic hazard for engineering applications

In areas where observed ground motion data is lacking (e.g. near-source), seismicity is infrequent, or where geologic structures complicate seismic wave propagation, simulations provide an approach to improving the accuracy of ground motion predictions for seismic hazard analyses. Dynamic ruptures are a way to model an earthquake rupture by governing physical equations from initially prescribed stress conditions and a friction law imposed along the fault interface. This method avoids the potential pitfalls of other methods, such as stochastically generating the high-frequency portion of ground motion (as in hybrid techniques) and physically inconsistent ruptures (as is possible in kinematic techniques).

Here, we form a group of dynamic rupture modelers and generate a suite of dynamic rupture simulations at frequencies relevant to engineering applications (initially focusing on frequencies up to 5 Hz), each using his preferred code (previously verified as part of the SCEC/USGS Rupture Verification Project: Harris et. al, 2018). In this first year of work, we limit our interest to a confined magnitude and distance range, including spectral accelerations at a range of periods for Mw ~6-7 earthquakes. We pursue two main routes of source generation: (1) imposing stochastic conditions along a planar fault with heterogeneous stress or friction conditions and (2) fractal rough-faults, where homogenous background stress conditions introduce initial heterogeneous stress along the fault. In addition, by varying regionally imposed stress conditions and hypocenter locations, we sample a range of earthquake rupture conditions.

The resulting broadband ground motions are pooled and validated by comparing trends of period/distance with empirical models from leading GMPEs. We follow a similar approach to the SCEC Broadband Simulation Platform (Goulet et al., 2015), where pseudospectral acceleration from simulations at various magnitudes and distances was compared with GMPEs. Additionally, we analyze the synthetic ground motion variability (isolating in terms of both intra- and inter-event) as a function of both distance and period. Finally, we keep track of other metrics that are useful in helping constrain results and ensure agreement with observations, such as displacement along the surface trace of the fault.