Project Title: Participation in the 3-D Model Verification Study
Investigators: Dr. Kim B. Olsen, (805) 893 7394, kbolsen@quake.crustal.ucsb.edu
Institution: Institute for Crustal Studies, UC Santa Barbara

Problems Addressed

1)How will the Los Angeles basin sediments amplify long-period seismic waves from moderate and large earthquakes in southern California?

2)Validating 3-D ground motion modeling codes.

3)Lens-effect in Santa Monica?

Results

1)This problem is addressed in chapter 4 of the Phase III report (Olsen, 1997). Here, we have used 3D/1D 0-0.4 Hz peak velocity ratios to construct site amplification maps for the Los Angeles basin for nine earthquake scenarios (M 6.75 earthquakes on the Palos Verdes, Elysian Park, Santa Monica, and Newport-Inglewood faults, approximations to the M 6.7 1994 Northridge, the M 6.4 1933 Long Beach, and the M 5.9 1987 Whittier-Narrows events, and two M 7.75 earthquakes on a 170-km long stretch of the San Andreas fault). The individual scenarios show amplification ratios up to an order of magnitude (Fig. 1). The distribution of mean peak velocity ratios, as calculated from the nine scenarios, has a maximum of 4.1 (Fig. 2). In general, both the individual scenarios and their mean values show that the largest amplification occurs above the deepest parts of the basin. The largest mean amplification is furthermore associated with relatively small uncertainties. The largest uncertainties of the mean amplification above the basin (log std > 0.6) are associated with sites located in the southern and southeastern part (Fig. 2). For the nine scenario earthquakes, the amplification tends to increase with distance from the causative fault to the basin structure. It is possible that this pattern is due to stronger amplification of the more complex wavefield impinging onto the basin structure from the distant earthquakes than that of the predominant body waves from nearby earthquakes. The amplification is caused by a combination of effects from the 3D basin structure and differences in impedance between the 1D and 3D models. The impedance difference effects accounts for a factor of 1.7 above the deepest part of the basin. After correction for the impedance effects, the maximum amplification averaged over sites above similar basin structure for the scenario earthquakes is 2.4, associated with sites located above the deeper part of the basin. Durations are sign)ficantly increased by the 3-D basin structure (Fig. 3). Compared to the smaller earthquakes, the M 7.75 San Andreas events generate extended durations over a much larger area, particularly above and beyond the southern, eastern, and western parts of the basin. The largest 3D -1D durations are obtained for the Santa Monica (60 s) and San Andreas (50/55 s) earthquakes, for sites located above the deepest part of the basin. In order to examine the accuracy of the 3D modeling obtained in this study, we compare the 0-0.4 Hz peak velocities from the Northridge simulation at 62 sites in the LA basin to those computed from data (Fig. 4). Most of the peak velocities and durations are fit within a factor of 2 for this frequency range. The model tends to underpredict the peak ground motion closer to the fault (< 40 km) but generally overpredicts the peak velocities farther away (> 40 km). In light of the simple approximation to the source and limitations of the basin model, the results are satisfactory.

The report was submitted to a review committee in late October, and the committee and the authors of the report met for discussion at USC mid November. The comments was generally positive, in particular for the chapter (4) on 3-D basin effects. Recommendations were made to test the validity of the 1-D impedance corrections.
Movies and high-resolution images of the results can be found on the web on: http://quake.crustal.ucsb.edu/~kbolsen/LA3D.html

2)A workshop was held in connection with the 1997 SCEC Annual Meeting in Costa Mesa to compare simulation results from participating groups using a selected volume of the San Fernando basin from the 3-D model by Magistrale et al. The most popular method was based on the 2-4 velocity-stress staggered-grid scheme (UCSB, 2x Woodward-Clyde, UCB, and Caltech). CMU used a irregular-grid finite-element method, and S-cubed used a variable-grid finite-difference method. The results by CMU, Caltech, and S-cubed generally differed strongly from each other and from the remaining groups. The results from UCSB, 2x Woodward-Clyde, and UCB showed many similarities, in particular in peak velocities, duration, and waveform of the initial arrivals. However, the results from these 4 groups differed by up to a factor of two for the later arrivals. Comparisons of time histories at adjacent grid points suggests that the use of only 5 grid points per minimum wavelength (approximately used by all groups) is insufficient to accurately model the ground motion in the LA basin model. First-cut comparisons of point-source synthetics to data for modeling results with and without attenuation indicated that finite Q may play a more important role in long-period ground motion than previously expected.

Images of the results can be found on: http://crustal.ucsb.edu/~kbolsen/3dex.html

3)The 1994, M 6.7 Northridge, California, earthquake, and many of its aftershocks showed anomalously high amplification of the ground motion in Santa Monica. Gao et al. (1996) proposed from analysis of Northridge aftershocks that the amplification was due to focusing from a lensshaped boundary between the Los Angeles basin sediments and the underlying bedrock. They furthermore explained the generation and amplification of a strong secondary phase in the vertical records of the (mostly deeper) aftershocks, arriving less than 1 second after the direct P arrival, by focusing at the lens. We have used finite-difference simulations of 10-Hz P-SV and SH waves to analyze possible causes for the ground motion amplification in the Santa Monica area, California, observed for seismic waves incident from north. Our P-SV and SH-wave simulations of a 17-km deep Northridge aftershock with epicenter 30 km north of Santa Monica showed that focusing at a lens-shaped sediment/bedrock interface below Santa Monica and in particular, caustics generated at the intersection of this interface and the Santa Monica fault can cause factor-of-four amplifications within the area that was heavily damaged during the 1994 M 6.7 Northridge earthquake (see Fig. 5 and 6). The strongest phase arriving 5-6 seconds after the direct P waves is identified as the S wave focused by the lens-shaped boundary of our bedrock/basin model. Similarly, we identify the observed large-amplitude phase arriving less than one second into the aftershock records as the direct P wave amplified by focusing.

References

Alex, C.M., and K.B. Olsen (1997). Lens-effect in Santa Monica? In: Proceedings from SCEC 1997 Annual Meeting, October 4-7, 1997, Costa Mesa, California, p. 52.

Alex, C.M., and K.B. Olsen (1997). Lens-effect in Santa Monica? Submitted to Geophys. Res. Lett.

Gao, S., H. Liu, P.M. Davis, and L. Knopoff (1996). Localized amplification of seismic waves and correlation with damage due to the Northridge earthquake, Bull. Seis. Soc. Am. 86, S209-230.

Olsen, K.B. (1997). Site class)fication and site-specific amplification for basin effects, in Probabilistic seismic hazard in Southern California: uncertainties due to assumptions and models ("Phase 3 report"), Southern California earthquake Center (SCEC), in review.