Dealing with non-detect values in time-series measurements of radionuclide concentration in the marine environment.

The attention of scientists in the field of environmental radioactivity is drawn to statistical methods recommended by Dennis Helsel for dealing with datasets including measurements that fall below the detection limits, as often encountered in environmental monitoring programmes. The methods are described by Helsel in his book entitled "Nondetects and Data Analysis: Statistics for Censored Environmental Data" (John Wiley and Sons, New York, 2005, 250p). These methods are applied to a data subset (using data from France) of the Radioactive Substance Committee (OSPAR commission for the protection of the marine environment of the North-East Atlantic), corresponding to time-series measurements of Cs-137 concentration in seaweed in the vicinity of the Areva NC reprocessing plant at La Hague, which is used as an illustrative example. Despite the presence of 163 non-detect values out of 514 measurements, it is possible to estimate descriptive parameters and perform statistical tests to compare concentration levels between two periods of time. Finally, evidence is obtained for an overall decreasing trend with time. The benefits of these statistical methods for data analysis are discussed.