Methodology for Incorporation of 3-D Simulation Results into Non-Ergodic Ground-Motion Models for Central California

Kathryn E. Wooddell, & Norman A. Abrahamson

Submitted August 15, 2017, SCEC Contribution #7730, 2017 SCEC Annual Meeting Poster #021

In the last 10 years, the large increase in the number of recorded ground motions has shown that ergodic GMPEs that combine data from around the world into a single average ground-motion model do not work well for a specific earthquake scenario and site location. In particular, about 75% of the variance in ergodic GMPEs, which is treated as aleatory variability, is from systematic and repeatable source, path, and site effects. Use of nonergodic GMPEs will reduce the aleatory variability and result in corresponding changes in the median ground motions for a particular scenario and site that will have a large effect on seismic hazard at long return periods where the GMPE dominates the uncertainty.

Using the NGA-W2 database of recorded ground motions in California and Nevada, an empirically based nonergodic ground-motion model including epistemic uncertainty has been developed for California (Abrahamson et al, 2017). The model uses nonergodic GMPEs that include spatially variable coefficients for the constant term and the geometrical spreading term that depend on the coordinates of the site and the source following Landwehr et al (2016). The constant term is separated into the site term and the source term. The GMPEs also include an anisotropic linear distance scaling term that depends on the path from the source to the site. Epistemic uncertainty in the nonergodic terms is included based on the density of available recordings in the site and source region.

In many parts of California, the density of the recorded ground motions remains sparse leading to large epistemic uncertainties in the nonergodic terms of the ground-motion model. To supplement the empirical datasets used for nonergodic GMPE development, the SCEC Central California Seismic Project (CCSP) is developing 3-D velocity models and conducting suites of 3-D numerical simulations in central California. We present the framework for using the CCSP 3-D simulations to update the empirical nonergodic GMPE in central CA using Bayesian updating. One issue for using 3-D simulations from finite ruptures is converting the path effects from finite rupture to a path effect in the GMPE which uses rupture distance. Based on Villani and Abrahamson (2015), we use the path effect for closest point on the rupture to parameterize the simulation results into the GMPE. The goal is to have a methodology that can update the logic tree weights of nonergodic GMPES for central California on a regular basis.

Key Words
nonergodic, nonergodic GMPE, ground-motion models

Citation
Wooddell, K. E., & Abrahamson, N. A. (2017, 08). Methodology for Incorporation of 3-D Simulation Results into Non-Ergodic Ground-Motion Models for Central California. Poster Presentation at 2017 SCEC Annual Meeting.


Related Projects & Working Groups
Special Projects