New PEER Report 2019/03: "Ground-Motion Directivity Modeling for Seismic Hazard Applications"

May 16, 2019

PEER has just published Report No. 2019/03: "Ground-Motion Directivity Modeling for Seismic Hazard Applications." It was authored by Jennifer L. Donahue, Jonathan P. Stewart, Nicolas Gregor and Yousef Bozorgnia. Review Panel: Jonathan D. Bray, Stephen A. Mahin, I. M. Idriss, Robert W. Graves and Tom Shantz.

Visit the PEER publications page to download a free color pdf of the document.

Executive Summary:

We reviewed five models for modifying the natural log mean and within-event standard deviation of ground-motion models (GMMs) to account for directivity effects in the near-fault environment. We found broad consistency for strike–slip ruptures, with positive and negative directivity effects for cases of rupture towards and away from a site of interest, respectively. We found substantial divergence among directivity models for reverse slip, with some providing maximum directivity for sites positioned to experience the peak alignment of rupture direction with the fault-slip direction (this occurs in the up–dip direction), while others optimize directivity based on the amount of fault rupture towards the site (even if the azimuth of rupture propagation does not align with the fault-slip direction).

We found four of five NGA-West2 GMMs to be centered on a condition of null directivity. Therefore, we consider those GMMs suitable for use in combination with similarly centered directivity models.

We present two deterministic methods for adjusting ground-motion hazard results for the effects of directivity: one modifies ground motions for a specified hazard level based on location-specific (relative to fault) changes in mean and standard deviation; and the second produces a directivity-compatible conditional mean spectra. We also provide recommendations for incorporating directivity effects into calculations within the hazard integral by either (1) modifying the mean and standard deviation of ground motion to approximately account for the effect of variable hypocenter location; (2) integrating over a location-specific distribution of the directivity parameter, which also indirectly accounts for variable hypocenter location; or (3) integrating directly over alternate hypocenter locations.