Spatial statistics of the Clark County Parcel Map, trial geotechnical models, and effects on earthquake ground motions in Las Vegas Valley

William H. Savran, John N. Louie, Brady Flinchum, Satish K. Pullammanappallil, & Aasha Pancha

Submitted 2012, SCEC Contribution #1536

When deterministically modeling the propagation of seismic waves, shallow shear-wave velocity plays a crucial role in predicting measures of shaking intensity such as peak ground velocity (PGV) and duration. The Clark County Parcel Map provides us with a data set of >10,000 geotechnical velocities in and around Las Vegas Valley, measured with SeisOpt® ReMi™ by Optim SDS. This is an unprecedented level of geotechnical detail. Las Vegas Valley is a geologic basin having similar geologic properties to some areas of Southern California. We analyze elementary spatial statistical properties of the Parcel Map, and calculate its spatial variability. We then analyze these spatial statistics from the PGV maps computed from two geotechnical models that incorporate the Parcel Map as input. Plotting a histogram of the Parcel Map’s 30-meter depth-averaged shear velocity (Vs30) values shows the data to approximately fit a bimodal normal distribution with µ1 = 400 m/s, σ1 = 76 m/s, µ2 = 790 m/s, σ2 = 149 m/s, and p = 0.49, where µ is the mean, σ is standard deviation, and p is the probability mixing factor for the bimodal distribution. Based on plots of spatial power spectra, the Parcel Map appears to be fractal between 0.1 and 10 cycles/km spatial frequency, or 0.1 to 10 km wavelengths. The 1-d spatial spectra exhibit the same fractal dimension in the N-S and the E-W directions, indicating isotropic scale invariance for the 2-d spatial spectra. We configured finite-difference wave propagation models at 0.5 Hz with LLNL’s E3D code, utilizing the Parcel Map as input, to compute a PGV map of the shaking intensity expected from a scenario earthquake (Black Hills M6.5). The resulting PGV map is fractal over the same spatial frequencies as the Vs30 maps associated with their respective models. The fractal dimension is systematically lower in all of the PGV maps as opposed to the Vs30 maps, showing that the PGV maps are richer in lower spatial frequencies. This is potentially caused by seismic waves averaging through spatial heterogeneities as they propagate. Finally, we develop a method to produce a comprehensive and adaptable Vs30 geotechnical model containing the Parcel Map overlain on stochastically generated Vs30 values. This model preserves the spatial statistics across the entire modeled map, and implements the deterministic features discovered by the Parcel Map.

Citation
Savran, W. H., Louie, J. N., Flinchum, B., Pullammanappallil, S. K., & Pancha, A. (2012). Spatial statistics of the Clark County Parcel Map, trial geotechnical models, and effects on earthquake ground motions in Las Vegas Valley. Presentation at 2012 Symposium on Engineering Geology and Geotechnical Engineering.