A study of uncertainties in radiological consequence modeling

Predicting radiological consequences to humans from radionuclide releases to the environment requires the use of computer models that usually involve a large number of model parameters. Most environmental model parameters, which include the values that quantify the relationships between various media such as the transfer of radionuclides between air, water, soil, vegetation, food, and human tissues, are not readily available on a site-specific basis. Even when they are available, because the values of these parameters vary from location to location and change with changing environmental conditions, the data cannot be precisely determined. In most present-day assessment situations, conservative assumptions are generally used in selecting model parameters. However, with the trend toward lowered regulatory dose/risk limits, the costs associated with this conservatism could be enormous. Concern for unnecessary cost burdens leads naturally to the need for a more accurate representation of the modeling results with the best estimates of the distributions, including the uncertainties of doses likely to be received by humans. This paper presents the results of a study that examined these uncertainties