D9.61 – Guidance to select level of complexity

Models in radioecology, as in other fields, have several purposes, the most important being the prediction of the behavior of radionuclides in different ecosystems and the understanding of the processes driving that behavior. These models are often finally used for regulatory purposes by transforming the values to a limiting quantity, such as effective dose or absorbed dose to demonstrate the protection of humans or biota respectively. To account for the consideration that several processes are not perfectly known, a considerable overestimation of the predictions is normally included in the models. Moreover, the final estimations of doses are directly proportional to the estimations of activity concentrations in the environment. Obviously, for many applications, only models can be used for prognosis such as predicting future activity concentrations. For that reason, and to avoid undue restrictions caused by poor results of the models, improvement of models is desirable and a continuous effort in this direction is needed. In this report, an example of a methodology which can be used to systematically improve the models is presented by providing a conceptual overview of the system through the use of Interaction Matrices and Features, Events and Processes. For the developers and the end users of the models, objective indicators to show whether a model is improved or not are desirable. Adapted from other fields, notably from the meteorological sciences, a methodology combining quantitative and qualitative indicators is elaborated. These indicators are used together with measured data in the different ecosystems where the comparison is needed. Finally, in those sites included in the Territories Library Database where a compilation of measured data was included, a comparison of widely used models (usually simpler) with more advanced models (usually more complex) has been implemented. Specifically several models have been applied in the Norwegian Fen site (NORM), in the Belgian NORM site, in the Fukushima forests contaminated by the FDNPP 2011 accident and in the West Cumbrian beaches, contaminated by releases from the Sellafield reprocessing facility. In all the cases, a discussion on the models, together with a comparison of the indicators applied to each model used at every site, was included. Altogether, this report can be regarded as a methodology to improve and show objectively the improvement of models applied to real case studies of long-term situations where contamination exists (often referred to as legacy sites). Applications in different situations can be seen as examples of implementing this process.