Annual Report, 1997

Analysis of Long Period Ground Motion Variability Related to Uncertainty in the Los Angeles Region 3D Velocity Model

Robert W. Graves, Woodward-Clyde Federal Services, Pasadena

David J. Wald, U.S. Geological Suruey, Pasadena

INTRODUCTION

One of the objectives of SCEC is to develop integrated 3D velocity and subsurface structure models of the Los Angeles basin region. While there has been important progress toward this goal, there still remains significant uncertainty in the current models and hence it is necessary to validate these models and quantify the uncertainty they introduce when predicting "scenario" earthquake ground motions. We have calculated ground motions in the LA region for the Landers earthquake using the proposed 3D velocity models of Graves (1996), Hauksson and Haase (1996), and Magistrale et al. (1996). We have extended this analysis to include simulations of the 1987 Whittier Narrows and 1994 Northridge earthquakes.

The goals of this work are 1) to understand the variability in predicted ground motion response related to uncertainty in the 3D velocity models, 2) to evaluate which features of the models are well resolved through the modeling of recorded data, and 3) to refine these models in order to reduce the uncertainty in ground motion prediction for future earthquakes.

New directions on the part of SCEC in terms of prioritizing the development of a multidisciplinary 3D earth model of the Los Angeles region have slightly shifted our focus. Most recently, rather than testing the earlier model of Magistrale et al. (1996) on with the Northridge and Whittier data, we have focussed on the latest 3D model, Model "A", proposed by Magistrale (1997), again using the Landers earthquake strong motion recordings made in the Los Angeles region. This work was done specifically for directing the discussion at the SCEC 3D Velocity Model Workshop in November, 1997.

UPDATED LOS ANGELES BASIN MODEL VALIDATION

The updated model of Magistrale (1997) has a deeper basin root in the San Fernando valley than the previous model, as seen in the north-northwest to south-southeast crosssection A-A' (Figure 1), which runs through the deepest portion of the valley. This deep structure strongly affects the amplification pattern in the northern San Fernando valley (Figure 2), where peak amplitudes exceed those above the deepest portion of the Los Angeles basin. Although the current Landers data set does not provide much adequate constraint there, the data further south, in the central part of the valley (stations NRG and SVG in Figure 2), indicate that the geologically based model mismatches the waveform observations, particularly in comparison to the Graves model. This requires reconciliation between the more complex, geologically-based model of Magistrale, with the simpler, seismologicallybased model of Graves.

In addition to analysis of the Northridge and Whittier mainshock data we have examined important USGS recordings of the 1992 Landers earthquake, and we have persuaded the USGS to digitize these data with full durations (though they are not yet available). These sites are important additions to the current validation setup described in Wald and Graves
(1998). Of note, two additional stations in the San Fernando valley (JFP and SVA) are critical for testing major mod)fications in the 3D model proposed by Magistrale (1997).

Interestingly, there is some improvement in the amplification pattern in the San Gabriel valley in the new Magistrale (1997) model '`A" (Figure 2), yet only subtle changes in the San Gabriel valley portion of the model were made (Magistrale, 1997, personal communication).

NORTHRIDGE AND WHITTIER MODELING

We have collected and processed the uncorrected, unfiltered ("Volume 1") data for the Northridge earthquake from the USGS, CDMG and USC in order to recover and model longer-period wave propagation effects. Figure 3 shows the ground velocities bandpass filtered from 0.15 to 1 sec. To enhance the comparison of the basin effects, the amplitudes have been scaled to correct for distance (approximately) based on a regression of the Northridge velocity data. The pattern of amplification can be more easily identified with a comparison of the distance-corrected peak amplitudes (Figure 3). Note the strong amplification in the northwestern portion of the Los Angeles basin and the San Gabriel valley, but only moderate relative amplification in the deep Los Angeles basin.

We have set up finite difference models for the Northridge and Whittier Narrows mainshocks. The model dimensions, shown in Figure 3, cover the SF, SG, and northern LA basins for the Northridge event, and the Whittier Narrows simulation covers the LA, SG, and southern SF basins.

CONCLUSIONS

One of the goals of this study was to emphasize the importance of calibration and validation of 3D earth models prior to use in ground motion estimation and prediction studies. We illustrate this is most suitably done with waveform modeling. This study has already contributed to this goal as new mod)fications to existing models of the Los Angeles basin have taken these findings into consideration (e.g., Magistrale, 1997), and validation of updated models has been considered a more critical aspect to their overall development.

References

Graves, R. W., 1996a. Simulating realistic earthquake ground motions in regions of deep sedimentary basins, Proceedings of the 11th World Conf. on Earthquake Engineering, Acapulco.

Hauksson., H., and J. Haase, 1997. Three-dimensional Vp and Vp/Vs in the Los Angeles basin and central Transverse Ranges, California, J. Geophys. Res., 102, 5423-5454.

Magistrale, H., K. McLaughlin, and S. Day, 1996. A geology based 3-d velocity model of the
Los Angeles basin sediments, San Andreas fault, Bull. Seism. Soc. Am., 86, 1161-1166.

Magistrale, H., 1997. Geology based 3D seismic velocity models of populated southern California basins, EOS, 78.

Wald, D. J., and R. W. Graves, 1998. The Seismic Response of the Los Angeles Basin, California, submitted to Bull. Seism. Soc. Am.