- © 2014 by the Seismological Society of America
The inclusion of epistemic uncertainties, generally via logic trees (Kulkarni et al., 1984), within probabilistic seismic‐hazard assessments (PSHAs) is becoming standard for all types of studies (commercial, governmental, or research; site specific, national, regional, or global). Consequently many studies publish expected ground motions for a given annual frequency of exceedance (AFE) or return period derived from the hazard curves for the mean, median, and various fractiles (percentiles). The spread of these values represents the uncertainty captured in the results (the greater the spread the higher the uncertainty). For example, Figure 1 shows the distribution of AFE for a peak ground acceleration (PGA) of 0.25g obtained in the study for the Washington Nuclear Plant (WNP)‐2 nuclear power plant (Hanford Reservation, Washington State) reported by Kulkarni et al. (1984). Distributions of ground‐motion levels for a given AFE are now most commonly reported in recent PSHAs rather than distributions of AFEs for a certain ground‐motion level.
Woo (2002) calls for the epistemic uncertainty to be overlaid on seismic hazard maps, although this is rarely, if ever, done. Giardini et al. (2004, their fig. 34) present the relative uncertainty in the Swiss National Seismic Hazard Map showing that parts of the map are associated with considerable uncertainty (more than 40%) because of doubts over the seismic source zones and b‐values. A recent detailed study of epistemic uncertainties in a PSHA is by Bradley et al. (2012), who rank the impact of various uncertainties on hazard results for two New Zealand cities (Wellington and Christchurch). There are, however, no studies to our knowledge …