This study implements new site-specific data and improved process-based transport model for 26 elements (Ac, Ag, Am, Ca, Cl, Cm, Cs, Ho, I, Nb, Ni, Np, Pa, Pb, Pd, Po, Pu, Ra, Se, Sm, Sn, Sr, Tc, Th, U, Zr), and validates model predictions with site measurements and literature data. The model was applied in the safety assessment of a planned nuclear waste repository in Forsmark, Oregrundsgrepen (Baltic Sea). Radionuclide transport models are central in radiological risk assessments to predict radionuclide concentrations in biota and doses to humans. Usually concentration ratios (CRs), the ratio of the measured radionuclide concentration in an organism to the concentration in water, drive such models. However, CRs vary with space and time and CR estimates for many organisms are lacking. In the model used in this study, radionuclides were assumed to follow the circulation of organic matter in the ecosystem and regulated by radionuclide-specific mechanisms and metabolic rates of the organisms. Most input parameters were represented by log-normally distributed probability density functions (PDFs) to account for parameter uncertainty. Generally, modelled CRs for grazers, benthos, zooplankton and fish for the 26 elements were in good agreement with site-specific measurements. The uncertainty was reduced when the model was parameterized with site data, and modelled CRs were most similar to measured values for particle reactive elements and for primary consumers. This study clearly demonstrated that it is necessary to validate models with more than just a few elements (e.g. Cs, Sr) in order to make them robust. The use of PDFs as input parameters, rather than averages or best estimates, enabled the estimation of the probable range of modelled CR values for the organism groups, an improvement over models that only estimate means. Using a mechanistic model that is constrained by ecological processes enables (i) the evaluation of the relative importance of food and water uptake pathways and processes such as assimilation and excretion, (ii) the possibility to extrapolate within element groups (a common requirement in many risk assessments when initial model parameters are scarce) and (iii) predictions of radionuclide uptake in the ecosystem after changes in ecosystem structure or environmental conditions. These features are important for the longterm (>1000 year) risk assessments that need to be considered for a deep nuclear waste repository.
2014. Vol. 133, 48-59 p.
Probabilistic simulations, Concentration ratio, Baltic Sea, Bioaccumulation, Risk assessment