Uncertainty analysis on early health effects modeling in accident consequence codes

The uncertainty analysis on early health effect modeling is part of the Joint CEC/USNRC Accident Consequence Code Uncertainty Analysis using Expert Judgment. For a complete overview on the uncertainty analysis on the early health effects module of the Accident Consequence Codes, see [1]. This article focuses on the determination of a joint distribution on the uncertain code input parameters describing risk on mortality for certain causes. The aim of uncertainty analysis is to determine a joint distribution on the target variables, i.e. uncertain code input parameters. An element of the methodology adopted-in the joint effort is to query the experts only on quantities which are physically observable, potential measurable and to which the expert can relate. Such variables are called elicitation variables. Target variables may be elicitation variables, but there may be cases where target variables are unsuitable to be elicitation variables. The elicitation variables together with additional information are presented to the expert in such a way that the expert can envision easily an experiment which would measure the elicitation variables. For each elicitation variables, the experts were asked to provide the 5%, 50%, 95% quantile points of their distribution.