Reducing Uncertainty in Ground Motion Prediction Equations by Understanding Path Effects

Valerie J. Sahakian, Annemarie S. Baltay, & Tom C. Hanks

Published August 1, 2016, SCEC Contribution #6441, 2016 SCEC Annual Meeting Poster #175

The differences between recorded ground motions and median predictions made by Ground Motion Prediction Equations (GMPEs) can be widely scattered, leading to large model standard deviations which can result in large ground motion amplitudes at low probabilities of exceedance. As such, reducing uncertainty in GMPEs is integral in decreasing the variability in predicted ground motions at low probabilities. Our approach for reducing uncertainty in the predictions is to understand the separate effects of source-, path-, and site-specific information. The knowable, repeatable parts of these effects can thus be identified and removed from populations of residuals, which results in location-specific GMPEs. These terms can then be associated with other geophysical information. This study focuses on correlating the path term to material properties along each recording’s raypath.

We employ a small database of approximately 3000 events recorded on the Anza seismic network, as shown in Baltay et al. (submitted), with magnitudes ranging from M 1 to M 5.4, with the majority of the events in the range 1 < M < 3. The Anza network has been in operation since ~1981, resulting in redundancy in source-to-station paths. We show various models for a regional GMPE, resulting from an inversion of all of the data. Next, we decompose the residuals for the preferred model into source, site, and path components, and focus on the path term. We examine two tomographic models of the Southern California region, the joint body wave and ambient noise model of Fang et al. (2016) and Allam et al. (2014). We raytrace through the Fang et al. (2016) model, and discuss various indices that can be computed from both models representing the effects of seismic attenuation, velocity structure on the raypath, and how they correlate to the computed path term. Finally, we describe a plan for moving forward with robust estimation of the effects of material properties on the path term, and how to incorporate this new path-specific knowledge into GMPE’s. We will discuss the implications on the resulting reduction in uncertainty and hence hazard level for low-probability earthquakes.

Key Words
GMPE's, Path Term, Anza network

Citation
Sahakian, V. J., Baltay, A. S., & Hanks, T. C. (2016, 08). Reducing Uncertainty in Ground Motion Prediction Equations by Understanding Path Effects. Poster Presentation at 2016 SCEC Annual Meeting.

Related Projects & Working Groups
Seismology