Statistics of velocity structure in the Los Angeles Area

William H. Savran, Kim B. Olsen, & Bo H. Jacobsen

Published September 2013, SCEC Contribution #1892

With recent advancements in supercomputing, the highest frequency of 3D deterministic earthquake simulations has reached 5-10Hz, the range required by engineers for structural design. However, realistic ground motion estimation at such frequencies assumes complexity of both the earthquake source and surrounding medium on the order of 10-100 m. While such small-scale heterogeneities may significantly affect earthquake ground motions in geologic basins, they are not included in state-of-the-art 3D velocity models, and have no prospect of characterization by direct measurements. Instead, we propose to include statistical models of the small-scale crustal heterogeneities in the 3D velocity models. As a first step toward this goal, we have collected readily available near-surface velocity data, including several datasets. We used direct and indirect Vs30 values as well as shallow and deep downhole velocity logs from oil exploration and geotechnical surveys. We analyze this data using a moving-window variogram method to estimate statistical parameters from the data set assuming a fractal distribution. Our preliminary results show Hurst exponents in the range of 0.05 to 0.3 and vertical correlation lengths of approximately 90-100m. The next step is to validate the observationally-constrained model of near-surface heterogeneities. We simulate 0-5 Hz visco-elastic wave propagation for the 2008 Mw5.4 Chino Hills, CA, earthquake in statistical models superimposed on the Southern California Earthquake Center (SCEC) Community Velocity Model (CVM) version 4, and compare the results to strong-motion data. We use a parallel staggered-finite-difference code (AWP-ODC) for the simulations in a 56 km x 40 km section of the Los Angeles basin with a minimum S-wave velocity of 200 m/s. We find amplification/de-amplification of peak ground velocities of approximately up to a factor of two, and that strong motion records provide useful constraints on the parameters of the statistical model. In addition, we find that widely used estimates of anelastic S-wave attenuation for the Los Angeles basin may need to be tuned, when the statistical model is included.

Savran, W. H., Olsen, K. B., & Jacobsen, B. H. (2013, 9). Statistics of velocity structure in the Los Angeles Area. Poster Presentation at SCEC Annual Meeting 2013.