2014 GSA Annual Meeting in Vancouver, British Columbia (19–22 October 2014)

Paper No. 237-2
Presentation Time: 1:25 PM

PREDICTING STRONG GROUND MOTIONS FROM M9 CASCADIA EARTHQUAKES USING 3D SIMULATIONS


FRANKEL, Arthur D., U.S. Geological Survey, University of Washington, Box 351310, Seattle, WA 98195 and STEPHENSON, William J., U.S. Geol Survey, Box 25046 MS-966, Denver, CO 80225, afrankel@usgs.gov

We have produced synthetic broadband (0-10 Hz) strong-motion records of M9 Cascadia earthquakes using 3D ground-motion simulations. We developed a composite rupture model for M9 Cascadia earthquakes based on modeling the strong motion records of the 2010 M8.8 Maule, Chile and 2011 M9.0 Tohoku, Japan earthquakes. We performed finite-difference simulations of ground motions for Cascadia earthquakes up to 1 Hz with a 3D velocity model derived from regional and local tomographic studies that used active and passive sources. The model includes the 3D geometry of the subducting slab and the Seattle, Tacoma, Portland, and Tualatin sedimentary basins.

The Cascadia rupture model consists of two parts. The first portion is the background slip that has a rise time of about 15 s and peak slip of about 20 m. This portion of the source model accounts for most of the coseismic slip and generates the ground motions at periods greater than about 5 s. The second part of the rupture process is a set of high stress drop M8 sub-events (asperities) that have slip rise times of about 2 s. These sub-events produce most of the strong motions for periods shorter than 5 s.

We modeled the observed strong motions of the Maule earthquake using a similar composite rupture model. Our procedure combines deterministic seismograms at long periods (> 1 s) with stochastic seismograms at short periods (< 1 s). We found that four sub-events with M7.9-8.2 were needed to duplicate the spectral accelerations (0.1 to 5 s period) and long durations of the observed recordings. These sub-events were required to have high stress drops of 200-350 bars to match the observed short-period response spectra and were located along the deeper half of the rupture zone.

The broadband synthetic seismograms for M9 Cascadia earthquakes have long durations of shaking and large spectral peaks at long periods (> 1 s). We have considered a variety of locations for the M8 sub-events along the rupture zone. A critical result from the simulations is the large amplification at periods of 3-5 s caused by the Seattle and Tacoma basins. In the simulations, this amplification reaches about a factor of five compared to sites outside the basins. It is essential to quantify this amplification to accurately evaluate the vulnerability of high-rise buildings to ground shaking from great Cascadia earthquakes.