69 4. OVERALL SUMMARY AND CONCLUSIONS Two marine dispersion scenarios were studied within the frame of MODARIA Working Group 10. The scenarios simulated dispersion of radionuclides in the Baltic Sea following the Chornobyl accident; and dispersion of radionuclides in the Pacific Ocean following the Fukushima Daiichi NPP accident. For the Baltic Sea radiological scenario, 137Cs dispersion was simulated by four models, which were either box or hydrodynamic models. In all cases, radionuclide transfer between water and suspended matter and bed sediments were included. Two main aspects are considered when comparing the results from each model: (1) the significant conceptual, numerical and parametric di?erences between models; (2) the complexity of the Baltic Sea hydrodynamics. Despite these two aspects, radionuclide dispersion results from each model were relatively close, including for bed sediments. The observed temporal trends of 137Cs activity concentrations taken from the HELCOM database are generally well reproduced by all models. The results of this study suggest that some processes do not significantly influence radionuclide transport within the Baltic Sea including winter ice cover and, surprisingly, water stratification by the halocline and thermocline. For the Pacific Ocean radiological scenario, 137Cs dispersion was simulated with six models. Box models were not used as they are unsuitable for the highly dynamic oceanographic conditions found off the Pacific Coast of Japan. A relatively good agreement between models could only be achieved after harmonization of model inputs, whereby all models were executed with the same hydrodynamic fields, the same parameters for describing water–sediment interactions, the same bathymetry and the same horizontal and vertical di?usion coe?cients. A step-by-step reduction of variability between models was achieved through this harmonization process. It was found that the different water currents from the different hydrodynamic models were the main factor producing variability between models. Where each model was executed with its own water circulation and model parameterization (Exercise 4b), there was in general a very good agreement in model–model and model–data comparisons. The results of Exercise 4b are influenced by a previous knowledge of measured data. Thus, the final component of the study was not a genuine blind model test. Two marine environments were studied: a semi-enclosed basin (Baltic Sea radiological scenario) and a highly dynamic system (Pacific Ocean radiological scenario). The description of hydrodynamics had a more significant impact on model results in the highly dynamic system. In the case of the Baltic Sea radiological scenario, results were in good agreement despite the di?erent hydrodynamic modelling approaches and simplifications applied. In the case of the Pacific Ocean radiological scenario, even similar hydrodynamic models led to di?ering current fields which, in turn, lead to very di?erent radionuclide dispersion patterns. Given the intensity and variability of currents in this area, as well as the presence of unsteady eddies due to current convergence, small di?erences in the hydrodynamics produced di?erent dispersion patterns and these di?erences tended to be amplified with time. This highlights the inherent di?culties in developing operational modelling systems for emergency decision support in this type of highly dynamic marine environment, i.e. the output of the system will be very dependent on the hydrodynamic model which has been used for the