Using the dynamics of faulting to explore radiated seismic energy and ground motion

Using a newly developed finite-element scheme we study the dynamics of faulting in a realistic geologic environment. The finite-element scheme uses one integration point and an hourglass control scheme. The versatility of the finite element scheme allows us to take into account the complex fault geometry, heterogeneous velocity structure and surface topography in ground-motion simulations. In a kinematic simulation of the 1812 Wrightwood earthquake, we show that large-scale surface topography in Southern California, in particular the San Gabriel Mountains, can shield the Los Angeles Basin from surface waves generated on the San Andreas Fault. We also show that the one-point integration scheme in the finite-element method can both easily implement the Perfectly Matched Layer absorbing boundary and incorporate material attenuation.; We dynamically simulate the 1994 Mw 6.7 Northridge earthquake and the 2004 Mw 6.0 Parkfield earthquake. Our dynamic rupture models constrained by a large amount of near-source ground motions show that the seismic radiation field is highly heterogeneous with large portions of radiated energy concentrated in the forward direction of rupture propagation due to directivity. Because of the heterogeneous distribution of energy flux, it is difficult to correct for directivity in teleseismic and regional estimates of radiated energy that are based on a point-source assumption, leading to bias in energy estimates. We demonstrate the key role of the static field in evaluating the distribution of radiated energy density on a surface. Because of the non-negligible static work on the fault, the radiated energy cannot be mapped onto the fault itself. Mapping the radiated energy onto a distant surface (e.g., a sphere in the far field) is valid only when the surface is far enough from the fault such that the static field is negligible on that surface.