The assumption that each exposure outdoors or on deck was at a random location results in too low an upper bound. It is far more likely that a participant’s assigned duties placed him repeatedly at particular places on a ship or island, and it is unlikely that he was in entirely different areas on the same day. If, for example, the participant was a member of a decontamination crew, his duty location might more likely be in a high-activity area than in an area with an exposure rate reflected by the mean of all the survey data. Even if a participant was not a crewmember, there is no reason to believe that his topside or island activities would place him in completely random locations during the first few days, when most of the dose would be incurred. For personnel billeted on islands, even though the indoor dose is reduced by an assumed 50% shielding factor, it is not negligible, and the indoor location was likely to be repeated rather than random, although his outdoor locations might have been more varied.

The upper-bound gamma doses from exposure to fallout at the Pacific test sites are generally based on a default 50% coefficient of variation (CV) in the mean exposure rate on an island or ship. However, the 50% CV estimate is itself highly uncertain for most events, and that uncertainty is not considered in estimating upper bounds. The CV estimate includes a component representing the variance due to the measurement itself,5 so it does not just represent the variability with location.

Even accepting the 50% CV estimate as reasonable, or a modified value for ships with few data or for which better information is available, a more reasonable upper-bound estimate can be obtained by assuming reasonable uncertainty distributions of parameter values and estimating the total uncertainty stochastically with a Monte Carlo calculation. In contrast with the method used in the NTPR program, we make the pessimistic assumption that the participant was exposed at the same indoor and outdoor locations for the entire period of exposure.

The uncertainty due to time topside on a ship (or outdoors on a residence island) versus inside and the shielding factor, uncertainty in the decay rate, and uncertainty in converting free-in-air exposure to dose must also be considered. Let

I = ∫{[E(t0) × SF × IN] + [E(t0) × OUT]} × t−x × FBE dt.

The integral is over the period of exposure t1 measured from the time of the test to some time T when the veteran left the test site area, and t0 is the time of the measurements of exposure rate. (If the period of exposure is not continuous, the integral can be written as a sum of integrals.) In the equation, I is the integrated dose, E(t0) is the mean of the survey exposure rates measured at time t0, SF is a


The uncertainty (precision) in survey meter readings is given as about ±10 to 25% for most instruments. However, the bias is presumed to be small on the basis of an assumption that the instruments were calibrated properly (see Brady and Nelson, 1985).

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement