To illustrate methods for incorporating uncertainty into seismic-hazard analyses, we describe the characterization of earthquake sources used in a seismic-hazard analysis for a site in western Washington. A simple and effective tool for incorporating uncertainty into seismic-hazard analysis is called a logic tree; one was used in this study to include the range of possible characteristics of the Cascadia subduction zone seismic sources. At present, considerable uncertainty exists regarding the earthquake potential of the Cascadia zone, particularly due to the completely aseismic nature of the interface between the Juan de Fuca and North America plates. A key issue is whether this aseismic behavior is merely a function of our short period of historical observation (i.e., interseismic quiescence) or representative of longer-term behavior (i.e., aseismic subduction). To develop a complete seismic-source characterization spanning the range of interpretations regarding the earthquake potential of Cascadia, a group of 14 experts was selected based on their experience with the Cascadia subduction zone. These experts assessed source characteristics, including subduction-zone geometry, potential seismic sources, probability that each potential source is active, expected locations and dimensions of rupture, maximum earthquake magnitude, earthquake recurrence models, geologic recurrence intervals, plate convergence rate, and seismic coupling.

In general, all of the experts identified two distinct subduction-zone earthquake sources: an “intra-slab” source representing earthquakes within the subducting Juan de Fuca plate, and the plate interface between the Juan de Fuca and North American plates. An analysis of uncertainties given by individual experts, as well as their differences of opinion, shows that some characteristics of subduction-zone sources are relatively agreed upon, while other characteristics are highly uncertain. For example, it was generally believed that the intra-slab source is seismogenic and that the future locations of earthquakes within the slab will follow the spatial pattern represented by historical seismicity. In contrast, a wide range of assessments was given for the probability that the plate interface is seismogenic and for the amount of seismic coupling between the plates. In making these assessments, the experts relied on a wide range of data sets, including analogies to other subduction zones, thermal-mechanical models of subduction, and geophysical imaging of the Cascadia subduction zone.

The study demonstrates the viability of using expert opinion to assess seismic hazards and illustrates one approach to obtaining a realistic estimate of the uncertainty in the seismic hazard.