Quantifying Uncertainty

DOE-STD-1023-95

expresses the relative credibility of that alternative in light of the available data. Elements of

the logic tree are sequenced to provide for a logical progression in the assessment from general

elements to more specific elements. In the simulation approach, uncertainties in inputs are

characterized by continuous distributions, and multiple simulations are run to sample from the

distributions. Both approaches have common application in seismic hazard analysis and lead

to reliable estimates of mean hazard.

Studies of the results of past hazard analyses have shown that care needs to be taken in

accounting for the possible correlation of uncertainties in some input parameters. For

example, the correlations in uncertainties in a-values and b-values in earthquake recurrence

relationships need to be considered (SSHAC, 1997).

Quantifying Uncertainty

Detailed recommendations for the application of expert elicitation and peer review are

contained in SSHAC (1997).

Because the models and parameters of seismic hazard analysis are not known with certainty,

hazard assessments should be designed to quantify not only the central tendencies but also

uncertainties. Two approaches are acceptable for characterizing and quantifying uncertainties

in PSHA: elicitation of multiple experts and peer review (the approaches can be used

separately or together). In the first approach, the judgments of multiple experts are elicited

regarding the elements of seismic sources and ground motion attenuation. The goal is to assess

the uncertainties for any given expert and the range of diversity of interpretations among

multiple experts.

A second approach to quantifying uncertainties consists of a single analyst or contractor (such

as a consulting company) conducting a seismic hazard analysis and subjecting the study to

peer review by an independent panel of experts. The peer review should include review of the

process as well as the inputs. The hazard analyst should strive to incorporate the range of

scientific interpretations and the peer reviewers should ensure that all reasonable

interpretations have been considered. Multiple cycles of peer review, focusing on particular

components of the analysis, are often needed to allow for modification and updating of the

inputs.

An important aspect of uncertainty characterization is documentation. Regardless of whether

the expert elicitation or the peer review procedure is used, the technical basis for all

assessments must be documented in a form suitable for third party review. For example, a

seismic source map must be supported by a written description of the basis for the source

boundaries in terms of evaluations of geologic, geophysical, and seismicity data. Likewise, the

basis for alternative source maps must be documented. One purpose of the documentation is

to provide a mechanism to examine the impact that new data and interpretations may have on

the interpretations as new studies are conducted or new findings are made. For example, a

potentially important consideration might be the occurrence of a moderate to large earthquake

in the region of a site after the seismic hazard analysis has been completed. The location of the

event and its magnitude can be compared with the sources considered in the analysis and the

A-4