Basin ZX Maps for use in the USGS National Seismic Hazard Model for the Western United States

Oliver S. Boyd, Eric M. Thompson, Allison Shumway, Morgan P. Moschetti, William J. Stephenson, & Sanaz Rezaeian

Submitted August 15, 2017, SCEC Contribution #7706, 2017 SCEC Annual Meeting Poster #274

Seismic hazard assessments depend on an accurate prediction of ground motion, which depends on properties of the earthquake source, propagation path, and site amplification. Here, we focus on long-period site response that is sensitive to basin depth for which several western United States (WUS) ground motion prediction equations (GMPEs) utilize the depth to 1.0 and 2.5 km/s shear wavespeed, termed Z1.0 and Z2.5 respectively, as a proxy. We produce maps for these basin depth terms by extracting information from published 3-D velocity models from across the WUS.

Some of the velocity models are not resolved well enough at low velocities to provide reliable estimates of the depth to 1 km/s, but may have reasonable depths to 2.5 km/s according to the Z2.5(VS30) relation of Campbell and Bozorgnia (2014; CB14). In such cases, we estimate Z1.0 from Z2.5 and the ratio of the Z1.0(VS30) relation of Chiou and Youngs (2014; CY14) to the Z2.5(VS30) relation of CB14. VS30 values are obtained from the USGS global VS30 model (Worden and others, 2015; https://earthquake.usgs.gov/data/vs30/). If the depth to 2.5 km/s is also unreasonably small, Z1.0 and Z2.5 values are set to 1/5th of the values resulting from the CY14 and CB14 relations with the application of the USGS VS30 map. The fraction 1/5th is an arbitrary choice selected to produce lower bound estimates of Z1.0 and Z2.5.

We assess the benefit of the resulting maps of Z1.0 and Z2.5 to improve estimates of long-period ground motions in the USGS National Seismic Hazard Model by evaluating the ground motion models with and without these basin estimates. We compare the predicted values to observed ground motions in the NGA-West 2 data set and compute the bias, inter-event standard deviation, and intra-event standard deviation of the residuals. Application of ZX values derived from relatively high-resolution local and regional 3-D velocity models in California can reduce the variance of intra-event residuals by 40% at 5-second period relative to default, VS30-based, ZX values, which is not unexpected given that some of those models were used to produce the GMPEs that we are evaluating. Lower resolution national-scale velocity models can give positive variance reduction relative to default values without an increase in bias suggesting that in the absence of local models, it can be reasonable to apply ZX values derived from lower resolution national models.

Key Words
Basin, site response, ground motion prediction equation

Citation
Boyd, O. S., Thompson, E. M., Shumway, A., Moschetti, M. P., Stephenson, W. J., & Rezaeian, S. (2017, 08). Basin ZX Maps for use in the USGS National Seismic Hazard Model for the Western United States. Poster Presentation at 2017 SCEC Annual Meeting.


Related Projects & Working Groups
Ground Motions