RESUMEN
This paper presents results of testing models for the radioactive contamination of river water and bottom sediments by (90)Sr, (137)Cs and (239,240)Pu. The scenario for the model testing was based on data from the Techa River (Southern Urals, Russia), which was contaminated as a result of discharges of liquid radioactive waste into the river. The endpoints of the scenario were model predictions of the activity concentrations of (90)Sr, (137)Cs and (239,240)Pu in water and bottom sediments along the Techa River in 1996. Calculations for the Techa scenario were performed by six participant teams from France (model CASTEAUR), Italy (model MARTE), Russia (models TRANSFER-2, CASSANDRA, GIDRO-W) and Ukraine (model RIVTOX), all using different models. As a whole, the radionuclide predictions for (90)Sr in water for all considered models, (137)Cs for MARTE and TRANSFER-2, and (239,240)Pu for TRANSFER-2 and CASSANDRA can be considered sufficiently reliable, whereas the prediction for sediments should be considered cautiously. At the same time the CASTEAUR and RIVTOX models estimate the activity concentrations of (137)Cs and (239,240)Pu in water more reliably than in bottom sediments. The models MARTE ((239,240)Pu) and CASSANDRA ((137)Cs) evaluated the activity concentrations of radionuclides in sediments with about the same agreement with observations as for water. For (90)Sr and (137)Cs the agreement between empirical data and model predictions was good, but not for all the observations of (239,240)Pu in the river water-bottom sediment system. The modelling of (239,240)Pu distribution proved difficult because, in contrast to (137)Cs and (90)Sr, most of models have not been previously tested or validated for plutonium.