68 discharge channels of the Fukushima Daiichi NPP. This source term implies a total release of 3.5 PBq of 137Cs between 26 March and 30 June 2011. For atmospheric deposition of 137Cs on the sea surface, two atmospheric dispersion models were applied and their outputs compared. Since both outputs were similar, the average between the models was calculated by KAERI over 3 hour intervals. Exercise 4 was carried out with these realistic source terms and consisted of two parts: calculations were performed with all common parameters and water circulation (as in Exercise 3) and, in addition, each model was run with its own configuration. This method allowed for a broad range of model expert estimation to be obtained and such a range may be considered an uncertainty estimation. At a general level, model results were in good agreement with observations; although 137Cs concentrations in surface water tended to be underestimated. It was also found that some models perform better for some target variables than for others. There was, however, agreement between all models in the sense that contamination of sediment extends over a banded area along the coast. It may therefore be concluded that the dispersion models used are effective tools, although very sensitive to water circulation description in highly dynamic environments, such as the Fukushima coastal waters. If models for supporting decision making after emergencies are specially designed, great attention needs to be given to the forcing of the dispersion model by water circulation in this type of environment.