S31C-4422:
Distance-dependent Ground motion variability from source models of the 1992 Landers earthquake and synthetic rupture models.
Wednesday, 17 December 2014
Jagdish Chandra Vyas, Paul Martin Mai and Martin Galis, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia
Abstract:
We investigate near field ground motion variability due to five different kinematic rupture models for 1992 Landers earthquake. The ground motion simulations are accomplished by solving the elasto-dynamic equations of motion using a generalized finite-difference method (Ely et al., 2008). Simulated waveforms are calibrated against near-field strong-motion recordings. We then analyze a large data-set of ground-motions from 2000 sites, binned with respect to distance and azimuth to compute mean and standard deviation of peak ground velocity (PGV). We consider 1D-velocity structures as used in the source inversions, and honor the geometrical complexity due to fault segmentation. Our simulations reveal that ground motion variability is reduced as the distance from the fault increases. Variability in the kinematic sources has considerable impact on the resulting shaking variability, although the five source models considered are derived by inversion of seismic and/or geodetic data. Simulated mean PGV and its standard deviation are comparable to empirical estimates using the ground-motion prediction equation (GMPE) of Boore and Atkinson (2008). In addition, we find that intra-event ground motion variability is large in fault-parallel and small in fault-normal direction. We also compare our simulations with and without Spudich and Chiou (2008) directivity corrections, applied to Boore and Atkinson (2008), which shows that slip heterogeneity controls near-field ground-motion variability. We further investigate the effect of slip heterogeneity by considering eleven (ten heterogeneous and one uniform) synthetic rupture models. Heterogeneous slip models are generated following the algorithm of Mai and Beroza (2002) for different correlation lengths and Hurst exponents. We then examine synthetic seismograms calculated at 1500 stations for the eleven rupture models, confirming that the distance decay of ground motion variability is due to slip heterogeneity.