Abstract
One major problem with nuclear security measurements involves source identification inthe presence of low signal-to-background ratio. This scenario iscommon to several applications, ranging from radiation identification atportal monitors to radiation source search with unmanned vehicles. In this context of identification of a large variety of sources, including natural and medical sources, sensitive sources of particular interest, but also potentially new/unknown sources for which no reference measurement is available, statistical methods are particularly appealing for their ability to capture the random nature of the measurements. Among them, Bayesian methods form a generic framework allowing for uncertainty quantification and propagation, which is of prime interest for detection (of known and unknown sources), classification, and quantification of smuggled nuclear and radiological materials. We demonstratethe use of Bayesian models for the identificationof mixed gamma sources, measured with organic scintillatorswithinshort acquisition times. We alsocompare the estimation performance using two different materials: liquid EJ-309 and stilbene crystal.
Included in
Purdue Conference on Active Nonproliferation
One major problem with nuclear security measurements involves source identification inthe presence of low signal-to-background ratio. This scenario iscommon to several applications, ranging from radiation identification atportal monitors to radiation source search with unmanned vehicles. In this context of identification of a large variety of sources, including natural and medical sources, sensitive sources of particular interest, but also potentially new/unknown sources for which no reference measurement is available, statistical methods are particularly appealing for their ability to capture the random nature of the measurements. Among them, Bayesian methods form a generic framework allowing for uncertainty quantification and propagation, which is of prime interest for detection (of known and unknown sources), classification, and quantification of smuggled nuclear and radiological materials. We demonstratethe use of Bayesian models for the identificationof mixed gamma sources, measured with organic scintillatorswithinshort acquisition times. We alsocompare the estimation performance using two different materials: liquid EJ-309 and stilbene crystal.