Role of Data

DOE-STD-1023-95

Role of Data

A wide range of earth sciences (geologic, geophysical, and seismologic) data are considered

when conducting a seismic hazard analysis. DOE-STD-1022-94 discusses the manner in

which these types of data are used to characterize seismic sources and to evaluate ground

motions. The extent to which particular data sets have been gathered in the site region and

immediate site vicinity will have a direct impact on the uncertainties in the seismic hazard

analysis. In cases where significant uncertainties exist regarding seismic sources or site-

specific conditions important to ground motions, additional data may need to be gathered to

reduce uncertainties in the site-specific seismic hazard analysis.

Because the development of the basic inputs to seismic hazard analysis requires interpretations

of data to develop models and parameter values, there is commonly a considerable range of

possible interpretations for any particular data set. For example, for a site in the eastern

United States, experts will make variable use of available geophysical data, tectonic

information, and historical seismicity data to define the configurations of seismic source zones.

Likewise, the available data pertinent to earthquake recurrence rates and maximum magnitudes

for the seismic sources will likely allow a range of permissible interpretations.

Seismic hazard analysts should take great care that the models and parameters are consistent

with the data, which include all physical information (geophysical, geological, and geotechnical

data, etc.) and historical data (earthquake catalogs). Models and hypotheses seemingly in

disagreement with data (for example, a recurrence model that predicts recurrence rates several

times higher than the empirical data) should be explained. All models and information provided

should be thoroughly documented so that an independent party could review the study and

understand the manner in which the data have been used to support the seismic hazard

interpretations.

Uncertainty in Hazard

Probabilistic seismic hazard analysis incorporates the random variability in the location, size,

and ground motions associated with future earthquakes. In addition to this random variability,

there is also a component of uncertainty related to lack of knowledge of the models and

parameters that characterize the seismic hazard. For example, alternative seismic source maps

could be developed, uncertainties in recurrence parameters can be quantified, and alternative

ground motion attenuation relationships can be identified. These uncertainties result in a

distribution of seismic hazard curves, from which the median (50th percentile) or mean seismic

hazard curve may be selected. The mean seismic hazard curve is usually quite sensitive to

uncertainties and, therefore, full inclusion of uncertainties in the seismic hazard analysis is

necessary.

Two equally-permissible approaches can be used to quantify and propagate uncertainties in

models and parameter values: the logic tree approach and the Monte Carlo simulation

approach (see SSHAC, 1997). In the logic tree approach, alternative models and alternative

parameter values are identified and a relative weight is assigned to each alternative that

A-3