The choice of atmospheric dispersion model and meteorological sampling scheme for use in accident consequence assessment

This paper briefly reviews the applications for which the different dispersion models contained in COSYMA are appropriate. Some studies have been carried out at NRPB and KfK in which results of accident consequence assessments using different dispersion models have been compared. These studies are summarised and guidance on the choice of model for particular applications is given. The studies carried out at NRPB used an extensively modified version of RIMPUFF and the version of MARC from which COSGAP was derived, and considered not only the difference between these models but also the importance of trajectory modelling rather than assuming a linear plume trajectory. The studies considered collective health effects from a large, short release. The results obtained with the different models or assumptions are not significantly different. The studies carried out at KfK have compared results obtained with MUSEMET used as a linear and a trajectory model, and have compared results from MUSEMET with those from RIMPUFF. The release considered was a few percent of the reactor inventory spread over a period of a few hours. These studies have considered total health effects in the population and also the areas above certain levels of contamination or dose. They show that predictions of the areas are more sensitive to the modelling assumptions than are those of numbers of effects. However they suggest that the choice of dispersion model is not important provided the release phases are described adequately, and that material released at different times is not assumed to travel in the same direction