Orphan source detection in mobile gamma-ray spectrometry - Improved techniques for background assessment

Sammanfattning: Hazardous radioactive sources out of regulatory control are referred to as orphan sources. Focussing on gamma-emitting orphan sources, this thesis describes methods that can be used in mobile gamma-ray spectrometry to perform real-time discrimination between an actual source signal and the signal due to the natural radiation background. Two categories of methods for orphan source detection are presented: visualisation and statistical hypothesis testing. Visualisation leaves the interpretation to the operator, whereas the hypothesis testing aims at answering the question: is there a signal or not? The methods are shown to work well for purely stochastic background data. The visualisation method gives a homogeneous representation of large data sets, which enhances the capability to discern significant positive deviations. The hypothesis testing method is shown to exhibit false alarms at the expected rate. Furthermore, by taking into account the spectral distribution in a hypothesis test, the power of the test was increased when compared with reference methods from the literature. Systematic background radiation variations can be disclosed by using data from airborne gamma-ray spectrometry (AGS) surveys. Validation of AGS data at the ground level by the use of terrestrial surveys yielded a strong positive correlation between the two data sets for background activity of uranium and thorium. By deriving a step function from AGS data, the systematic background variations were successfully modelled at the ground level. This approach was shown to reduce the number of excess false positives in an area with strong background gradients. In addition to random and systematic background radiation fluctuations, germanium spectrometers can have problems with microphonic noise due to the vibrations that inevitably will occur in mobile gamma-ray spectrometry. The additional noise can be expected to degrade the resolution of the detector. The level of resolution degradation was found to be 0-19%, depending on the speed of the vehicle.