| Popis: |
Naturally occurring radioactive material in containerized cargo challenges the state of the art in national and international efforts to detect illicit nuclear and radiological material in transported containers. Current systems are being evaluated and new systems envisioned to provide the high probability of detection necessary to thwart potential threats, combined with extremely low nuisance and false alarm rates necessary to maintain the flow of commerce impacted by the enormous volume of commodities imported in shipping containers. Maintaining flow of commerce also means that inspection must be rapid, requiring relatively non-intrusive, indirect measurements of cargo from outside containers to the extent possible. With increasing information content in such measurements, it is natural to ask how the information might be combined to improve detection. Toward this end, we present an approach to estimating isotopic activity of naturally occurring radioactive material in cargo grouped by commodity type, combining container manifest data with radiography and gamma-ray spectroscopy aligned to location along the container. The heart of this approach is our statistical model of gamma-ray counts within peak regions of interest, which captures the effects of background suppression, counting noise, convolution of neighboring cargo contributions, and down-scattered photons to provide estimates of counts due to decay of specific radioisotopes in cargo alone. Coupled to that model, we use a mechanistic model of self-attenuated radiation flux to estimate the isotopic activity within cargo, segmented by location within each container, that produces those counts. We test our approach by applying it to a set of measurements taken at the Port of Seattle in 2006. This approach to synthesizing disparate available data streams and extraction of cargo characteristics, while relying on several simplifying assumptions and approximations, holds the potential to support improvement of detection systems using current capabilities and to enable simulation-based evaluation of new candidate detection systems. |
Full Text from ScienceDirect