How can the reliability of radioecological assesment models be improved?

Assessment models are widely used to estimate the radiation exposure originating from routine or accidental releases of radionuclides into the environment. The major criterion for their quality is the reliability of their results. Radioecological models consist of three components: -the structure of the models arising from the number of compartments and their interelations. -the radioecological parameters, like transfer factors, deposition velocity, etc. and, -the assumptions on the living habit of man and on agricultural practices. Reviewing the international literature, it is obvious that assessment models in different countries have approximately the same structure and use similar values for radioecological parameters. Assumption on man's habits, however, may differ significantly according to the degree of conservatism of dose estimations. The aspect of the conservatism of assumptions is basically independent from model's structure and from the parameters and will not be discussed in this contribution. Internationally great effort is spent on the verification and validation of radioecological models, e.g within the framework of BIOMOVS, VAMP or BlOMASS. Comparison studies are initiated, intercomparison exercises for defined scenarios are carried out and radioecological parameters are compiled. It is demonstrated in this contribution that these efforts will only be of limited value for the improvement of assessment models. The major problem might arise from the fact, that the development of radioecological models is not always done in the right way. Usually ecosystems are divided into different compartments and afterwards the relevant parameters are estimated. It is demonstrated by examples that a better approach would be to optimize the structure of a radioecological model on the basis of the available knowledge on the behaviour of radionuclides in the environment, and therefore, consequently on the reliability of parameters. By this approach the existing differentiation between modelers and experimentalists would be abolished because modelling would be an integrating experimental and theoretical approach. (author)