Space-time Bayesian analysis of the environmental impact of a dismissing nuclear power plant.

The assessment of the radiological impact of decommissioning activities at a nuclear power plant requires a detailed analysis of the distribution of radionuclides in the environment surrounding it. The present work concerns data of three campaigns carried out during the last twenty years in the plain of the Garigliano river surrounding the Garigliano Nuclear Power Plant (GNPP), which is located in Southern Italy and shut down in 1979. Moreover, some data from surveys held in the eighties, across the Chernobyl accident, have been taken in account. The results for the soil samples, in particular for Cs and U specific activity, were analyzed for their extension in space and in time. Some of the problems related to the classical analysis of environmental radiological data (non-normal distribution of the values, small number of sample points, multiple comparison and presence of values lesser than the minimum detectable activity) have been overcome with the use of Bayesian methods. The scope of the paper is threefold: (1) to introduce the data of the last campaign held in the Garigliano plain; (2) to insert these data in a larger spatio-temporal frame; (3) to show how the Bayesian approach can be applied to radiological environmental surveys, stressing out its advantages over other approaches, using the data of the campaigns. The results show that radionuclides specific activity in soil is dominated by the natural sources with the contribution of the atmospheric fallout. A detailed study was performed on the Cs data to evaluate both their statistical distribution and the trend over the space and the time. It results that (i) no new contribution there was in the last decades, (ii) specific activity values of the area surrounding the GNPP are consistent with those obtained in other farther areas, (iii) the effective depletion half-life factor for Cs is much lower than the half-life of the radionuclide.