PNNL

Abstract

This work aims to solve some of the technical and logistical challenges inherent in performing On Site Inspection activities under the authority of the Comprehensive Nuclear-Test-Ban Treaty. Inspectors require equipment that can reliably identify the radionuclide signatures of nuclear test explosions amid a background of environmental contamination. Detection of these radiation anomalies by mobile search teams in the air or on the ground can narrow the search field and target specific areas for more detailed inspection or sampling. The need to protect confidential information of the inspected State Party, especially regarding past nuclear testing activities, suggests that full access to measured gamma-ray spectra should be limited. Spectral blinding techniques---in which only a fraction of the information derived from the spectra is displayed and stored---have the potential to meet the needs of both the OSI team and the State Party. In this paper, we describe one such algorithm that we have developed for identifying anomalous spectra from handheld, mobile, or aerial sensors. The algorithm avoids potential sensitivities by reducing the gamma-ray spectrum into a single number that is displayed and stored. A high value indicates that the spectrum is anomalous. The proposed technique does not rely on identifying specific radionuclides, operates well in the presence of high background variability, and can be configured to ignore specific spectral components. In previous work, the algorithm has proven very effective in classifying gamma-ray spectra as anomalous or not, even with poor statistical information. We performed a limited simulation of an airborne search scenario to demonstrate the potential algorithm for OSI missions. The technique successfully detected an injected source of interest whose count rate was an order of magnitude below background levels. We also configured the algorithm to ignore 137Cs as irrelevant to the mission. The resulting alarm metrics were unaffected by the presence of injected 137Cs contamination.