A distribution-free test for anomalous gamma-ray spectra

• We develop a new non-parametric test for detecting anomalous gamma-ray spectra.
• The proposed test has good empirical power for detecting weak signals.
• It can serve as an effective tripwire for invoking more thorough scrutiny of the source.

Gamma-ray spectra are increasingly acquired in monitoring cross-border traffic, or in an area search for lost or orphan special nuclear material (SNM). The signal in such data is generally weak, resulting in poorly resolved spectra, thereby making it hard to detect the presence of SNM. We develop a new test for detecting anomalous spectra by characterizing the complete shape change in a spectrum from background radiation; the proposed method may serve as a tripwire for routine screening for SNM. We show that, with increasing detection time, the limiting distribution of the test is given by some functional of the Brownian bridge. The efficacy of the proposed method is illustrated by simulations.