The accurate calculation of ground motions for future large earthquakes depends upon knowledge of three-dimensional (3D) geologic structure and the earthquake source physics, as well as sufficient computational resources. The San Francisco Bay Area, and particularly the Loma Prieta region, is a unique area for study of 3D wave propagation due to the pronounced lateral velocity contrasts across the region's strike-slip faults. Understanding the effects of such long wavelength structure is doubly important when considering a source located close to such a contrast, as was the 1989 Loma Prieta, CA earthquake. The focus of this study is to attempt to reconcile estimated ground motions computed with a 3D velocity model and the published source model of Wald et al., 1991, for Loma Prieta with recorded strong motion observations. Additionally, the 3D finite difference code (Larsen et al., 1995) allows us to test the sensitivity to structures in the velocity model: also shown are results from a velocity model which does not include the San Andreas Fault (SAF), in order to examine the effects of this velocity contrast.
Figure 9.1 shows the spatial variation of peak horizontal velocity for the 3D model (left) and for the 3D model with the SAF removed (right). Both plots show strong directivity of energy to the north and northwest, an effect of the source used. Also, both plots show that the location of deeper alluvium in the 3D model in Santa Clara Valley has a profound effect on ground motions in that area: higher amplitudes in ground motion correspond to deep alluvial basins while a thin zone of lower amplitudes corresponds to a ridge in the bedrock/alluvium interface. Another area of deep alluvium in the 3D model, San Pablo Bay, shows elevated levels of ground motion in both runs. These results indicate the importance of including alluvial basins in 3D velocity models. Comparison of runs with and without the strong lateral velocity contrast of the SAF demonstrates the importance of including fault structures in 3D velocity models. While both runs show energy being directed generally to the northwest by the source, the run without the SAF shows that without a velocity contrast at the fault the energy would have propagated parallel to the fault. Instead, refraction by the SAF propagated more energy to the north and into Santa Clara Valley, and decreased ground motion right above the fault. These effects agree more closely with recorded strong motions in the area for Loma Prieta.
Special thanks to A. Lomax for access to his original basin structure. This research was supported by the USGS NEHRP program under USDI contract 1434-HQ-97-GR-03091. The computer time at LLNL was supported by the Advanced Computing Initiative in Science and Engineering (ACISE) program.
Larsen, S. and C.A. Schultz, Technical Report No. UCRL-MA-121792, 1995.
Wald, D., D.V. Helmberger, and H.T. Heaton, "Rupture model of the 1989 Loma Prieta earthquake from the inversion of strong-motion and broadband teleseismic data.", Bull. Seism. Soc. Am., 81, 1540, 1991.