A Case for the Use of 3-D Attenuation Models in Ground Motion and Seismic Hazard Assessment

Attenuation relationships that are used to characterize estimated ground motion often ignore details of the earth’s highly-variable three-dimensional velocity and attenuation structure. Increasingly available attenuation models can be used to refine the expected ground motions. First, some tests are performed to look at the effect of the variability in several parameters, such as crustal attenuation, upper mantle attenuation and crustal thickness. A concrete example is then provided using the results of a recent crust and upper mantle attenuation model of the Middle East. We find 30-40% variations in 1 Hz spectral accelerations simply from the variations of the same event recorded in different directions. Since overall regional variability is expected to be even higher, this effect seems significant enough to be accounted for in strong ground motion estimates and seismic hazard assessment. This has the potential to account for some of the smaller scale amplitude variations not considered using broadly applied 1-D attenuation relationships. Introduction and Previous Work Methods to assess seismic hazard require attenuation relationships that can estimate earthquake strong ground motions from parameters that characterize the earthquake source, the propagation path, and local site conditions. The attenuation relationships that are used to predict the ground motions from postulated events are critical to the overall quality of the assessment. The extensive amount of work on this problem (see McGuire, 2008 for a review of the history of seismic hazard assessment) has tended to focus on regressing empirical strong ground motion data to equations of a form that tries to capture as much of the physics of the problem as possible. For example, some critical input parameters are earthquake magnitude (usually moment magnitude), some estimate of distance (epicentral distance, Joyner-Boore distance, distance to closest point on fault, etc.), site conditions (soil, soft rock, hard rock), frequency content, etc. More sophisticated analysis can also include fault type (normal, thrust, strike-slip), 3-D basin effects, sub-crustal or subduction zone earthquakes, etc. Much of this, in fact, has been the focus of Next Generation Attenuation (NGA) models and associated NGA workshops (see http://earthquake.usgs.gov/hazards/about/workshops/nga_Workshop.php and talks therein). One thing that hasn’t been considered in great detail is accounting for some of the large amplitude small-scale variations in the earth’s seismic attenuation. It is, of course, quite true that there are variations in the attenuation relationships from region to region (e.g. western United States vs. eastern/central United States). There can