Probability for Type I errors in gamma-ray spectrometric measurements of drinking water samples

Abstract

In gamma-ray spectra, acquired in the absence of the sample, peaks occur which belong to the spectrometer background. When samples are measured, which contain radionuclides that appear in the background also and have activities near the detection level, the background contributes substantially to the peak areas. In the extreme case, when the contribution of the sample is much smaller than the contribution of the background, the peak area attributed to the radionuclide within a sample has the same probability of being positive or negative. Therefore, to interpret the results obtained from measurements of low-activity samples, the performance of the spectrum analysis procedure near the detection level must be known. To test the performance of the spectrum analysis procedure at low activities, the spectrometer background spectra were analyzed as if they had been water samples, prepared as dry residue obtained by evaporation of 50 L of water. The probabilities for false positives together with their decision thresholds are given for radionuclides appearing in the background spectra. For some of the radionuclides that do not appear in the background spectra, probabilities for false detection are given as well.