Deterministic High-Frequency Ground Motions from Simulations of Dynamic Rupture along Rough Faults

Kyle Withers, Kim Olsen, Zheqiang Shi, Rumi Takedatsu, and Steve Day
Dept. of Geological Sciences, San Diego State University

The accuracy of the earthquake source description is a major limitation in high-frequency (>1 Hz) deterministic ground motion prediction, which is critical for performance-based design by building engineers. We address this issue by an attempt to quantify the contributions to high-frequency ground motion from both small-scale fault geometry and media complexity and perform validation against recent Next Generation Attenuation (NGA) relations. Specifically, we compute the ground motion synthetics using dynamic rupture propagation along a rough fault imbedded in a velocity structure with heterogeneities described by a statistical model. First, simulations of dynamic rupture are carried out using a support operator method (SORD, Shi and Day [2013]), in which the assumed fault roughness follows a self-similar fractal distribution with wavelength scales spanning three orders of magnitude from ~102 m to ~105 m. The rupture irregularity caused by fault roughness generates high-frequency accelerations with near-flat power spectra up to almost 10 Hz. Next, we perform wave propagation simulations using the moment- rate time histories from the dynamic rupture simulation as a kinematic source to extend the ground motions to farther distances from the fault with a highly scalable fourth-order staggered-grid finite difference method (AWP-ODC) with a uniform grid spacing of 20 m. The latter wave propagation simulations use a characteristic 1D rock model with and without small-scale heterogeneities and intrinsic attenuation, for a total of four models. We then compare our ground motion results (e.g., distance and period dependence of peak ground accelerations and peak spectral accelerations) with the empirical curves given by recent Next Generation Attenuation (NGA) relations, for both the median and the standard deviation. Finally, the high-frequency ground motions from the rough-fault simulations are compared to those obtained by a hybrid deterministic–stochastic technique (Mai et al, 2010) in order to understand the limitations of high-frequency generation using the latter method.