44 provided for the tracer dispersion exercise and radionuclide concentration for the surface water, bottom water and sediment is provided for the 137Cs exercises. FIG. 25. Computational grid used by all models in Exercise 3. Water depths are given in m. 3.4.2. Tracer dispersion Results for the tracer experiment are summarized in Figure 26 and are within the same order of magnitude for all models at points P1 and P2 (note that now a linear scale has been used for the y axis instead of a logarithmic one as previously). Thus, in general, the use of the same bathymetry and di?usion coe?cients has slightly improved the agreement between models. Nevertheless, as can be seen from Exercises 1 and 2, the main factor in producing model discrepancies is water circulation. In other words, the agreement improvement is higher from Exercise 1 to 2 than from Exercise 2 to 3. In the case of P3, results are similar to those of Exercise 2, where the agreement between models is relatively good for both the arrival of the signal and the calculated concentrations. The exception is the result of NTUA model which produces an instantaneous arrival of the signal and concentrations significantly higher than the other models. 3.4.3. 137Cs dispersion and no SPM in the water column Calculated time series for 137Cs at points P1–P3 are presented in Figures 27–29. It can be seen that, for instance, at P2 the use of the same parameters has improved model agreement. Model results are, in general, within the same order of magnitude (note that logarithm scale is not used). In the case of sediment, the JAEA model produces lower concentrations than the KAERI