Real-time airborne gamma-ray background estimation using NASVD with MLE and radiation transport for calibration

Abstract

a b s t r a c tHelicopter-mounted gamma-ray detectors can provide law enforcement officials the means to quickly and accurately detect, identify, and locate radiological threats over a wide geographical area. The ability to accurately distinguish radiological threat-generated gamma-ray signatures from background gamma radiation in real time is essential in order to realize this potential. This problem is non-trivial, especially in urban environments for which the background may change very rapidly during flight. This exacerbates the challenge of estimating background due to the poor counting statistics inherent in real-time airborne gamma-ray spectroscopy measurements. To address this challenge, we have developed a new technique for real-time estimation of background gamma radiation from aerial measurements without the need for human analyst intervention. The method can be calibrated using radiation transport simulations along with data from previous flights over areas for which the isotopic composition need not be known. Over the examined measured and simulated data sets, the method generated accurate background estimates even in the presence of a strong, 60 Co source. The potential to track large and abrupt changes in background spectral shape and magnitude was demonstrated. The method can be implemented fairly easily in most modern computing languages and environments.

Modeling Aerial Gamma-Ray Backgrounds Using Non-negative Matrix Factorization

“…This application highlights another distinction between FSA, NMF, and NASVD -typically, FSA is used to decompose spectra that have not yet been measured, while NASVD is used to decompose only previously measured spectra (although see ref. [24] for a real-time application of NASVD). NMF has been presented here retrospectively like NASVD, but both NMF and NASVD components can be fit to future measurements under the assumption that the components derived from the training set are also effective at reducing the dimensionality of future data.…”

Survey of Machine Learning Approaches in Radiation Data Analytics Pertained to Nuclear Security

“…To that end, maximum likelihood estimation (MLE) has been proposed in [43] to individually estimate the count rate in every channel of an airborne gamma radiation detector. Furthermore, a methodology that initially uses data learning to estimate the Poisson process parameters, and subsequently applies the Poisson model to estimating background counts in spectral peaks was introduced in [44].…”

Airborne Gamma-Ray Spectroscopy for Modeling Cosmic Radiation and Effective Dose in the Lower Atmosphere

“…The improvement of AGRS data quality evolved side by side with the integration of geological data via increasingly refined statistical and geostatistical methods: the combined effect of both aspects allowed to go beyond traditional mineral exploration and lead to the investigation of new multidisciplinary fields, like landslide monitoring (Baroň et al, 2013), peat thickness 1 arXiv:1712.04777v1 [physics.app-ph] 13 Dec 2017 estimation (Keaney et al, 2013), prediction models for trees' growth (Mohamedou et al, 2014), radioelement distribution in weathered materials (Iza et al, 2016) and precision agriculture (Söderström et al, 2016). Concurrently, the potentialities of the AGRS technique in the sector of homeland security have been widely explored in terms of feasibility of real-time identification of anthropogenic radionuclides on top of the natural background (Cardarelli et al, 2015;Detwiler et al, 2015;Kock et al, 2012;Kulisek et al, 2015;Kock et al, 2014) and of merging and comparing results from multi-regional AGRS campaigns performed in the framework of intercomparison exercises (Bucher et al, 2009). In the light of environmental contamination assessment, the detection of anthropogenic radionuclides emitting low energy gamma-rays (e.g.…”

“…estimation (Keaney et al, 2013), prediction models for trees' growth (Mohamedou et al, 2014), radioelement distribution in weathered materials (Iza et al, 2016) and precision agriculture (Söderström et al, 2016). Concurrently, the potentialities of the AGRS technique in the sector of homeland security have been widely explored in terms of feasibility of real-time identification of anthropogenic radionuclides on top of the natural background (Cardarelli et al, 2015;Detwiler et al, 2015;Kock et al, 2012;Kulisek et al, 2015;Kock et al, 2014) and of merging and comparing results from multi-regional AGRS campaigns performed in the framework of intercomparison exercises (Bucher et al, 2009). In the light of environmental contamination assessment, the detection of anthropogenic radionuclides emitting low energy gamma-rays (e.g.…”