A real-time statistical alarm method for mobile gamma spectrometry-Combining counts of pulses with spectral distribution of pulses

A well-founded decision needs to take into account as much information from a sample as possible. In gamma spectrometry, the number of photons and their energy are the two quantities readily accessible to the physicist and both should be used in order to increase the power of a statistical test. While the problem of counts of pulses has been much studied the problem of spectral distribution of pulses has been generally overlooked. This work presents a statistical test combining tests on count rate and tests on spectral distribution. The proposed method is shown to have an acceptable false positive rate and, when compared with two other test statistics found in the literature, greater power. (C) 2012 Elsevier B.V. All rights reserved