N5A1  Instrumentation for Homeland and National Security, Algorithms and Simulations

Friday, Nov. 6  08:30-10:10  Town and Country

Session Chair:  William Pitts, Pacific Northwest National Laboratory, United States; Richard Lanza, Massachusetts Institute of Technology, United States

Show/Hide All Abstracts

(08:30) N5A1-1, Network Detection of Radiation Sources Using ROSD Localization

N. S. V. Rao1, C. Q. Wu2, M. L. Berry2, K. M. Grieme2, S. Sen1, R. R. Brooks3, C. Temples3

1Computer Science and Mathematics Division, Oak Ridge National Laboratory, Oak Ridge, TN, USA
2Department of Computer Science, University of Memphis, Memphis, TN, USA
3Department of Electrical and Computer Engineering, Clemson University, Clemson, SC, USA
Networks of radiation counters are increasingly being deployed in monitoring applications to provide faster and better detection performances than the individual detectors. Their performances, however, critically depend on the algorithms used to aggregate measurements from the individual detectors. Recently, localization-based algorithms have been developed for the network detection, which are in sharp contrast to the conventional “detect first and localize next” methods. We propose a localization-based radiation source detection algorithm using the Ratio of Squared Distance (ROSD) technique. This method overcomes the “no output” limitation of the previously proposed triangulation algorithm that may produce the imaginary roots or the real ‘phantom’ roots due to the underlying quadratic equations. We apply our ROSD-based detection algorithm to the measurements collected from the Domestic Nuclear Detection Office’s Intelligent Radiation Sensor Systems (IRSS) outdoor experiments. The datasets we used consist of 6 different runs of a Cs-137 source moving through a 4x4 grid of 16 NaI detectors covering a 42m X 42 m region. We compare the performance of our algorithm to the particle filter method, and for the baseline cases it provides better performances in terms of false alarm and detection rates.

(08:50) N5A1-2, Wavelet Analysis of High and Low Resolution Gamma-Ray Spectra: an Investigation of Peak Finding Techniques

C. I. Thompson, K. Vaughan, R. L. Turner

AWE, Reading, Berkshire, UK
For the detection of radiological materials, peak identification within gamma spectroscopy data is a useful tool. This paper discuses the application of wavelet analysis as a peak finding technique to high resolution data recorded by high purity germanium (HPGe) detectors. Whilst the wavelet method has been previously applied to high efficiency/low resolution data, this work considers the applicability of wavelet analysis to high resolution spectral datasets with varying levels of background. An algorithm has been developed that automatically locates wavelet transform modulus maxima (WTMM) lines; the method identifies local maxima and minima of the scalogram joining data points within the same potential. The approach was applied to analyse a range of experimental recorded radiation spectra, both from shielded and un-shielded sources. Peak locations were found by comparing peak widths with those expected based on the detector resolution function (DRF), alongside WTMM straightness and line length filter tests. Findings indicated that, when implemented in this approach, the wavelet analysis method was applicable to the identification of high-resolution photopeaks when concealed within varying levels of background. © British Crown Owned Copyright 2015/AWE

(09:10) N5A1-3, Methods, Techniques and Recent Results in Monte Carlo Simulation Validation for Sensitive Applications

M. G. Pia1, T. Basaglia2, M. Begalli3, C. Choi4, M. C. Han4, G. Hoff5, C. H. Kim4, H. S. Kim4, S. H. Kim4, P. Saracco1

1INFN Genova, Genova, Italy
2CERN, Geneva, Switzerland
3UERJ, Rio de Janeiro, Brazil
4Hanyang University, Seoul, Korea
5CAPES, Brasilia, Brazil
Monte Carlo simulation is a widely used instrument in several experimental domains, including sensitive ones such as homeland security and radiation protection, medical physics and medical imaging, and mission-critical ones, such as space science. For these applications objective assessment of the reliability of the simulation results is especially important. Nevertheless, extensive experience as reviewers and editors of scholarly literature in these fields shows that not only the assessment of the validity of simulation is often neglected or taken for granted in experimental practice, but also that the conceptual grounds for the validation of Monte Carlo simulation are far from established in the experimental community. Statistical analysis methods, which are part of common practice in experimental data analysis as instruments for quantifying objective assessments, are scarcely present in the literature concerning simulation validation in sensitive fields. This presentation reviews the epistemological grounds of simulation validation and its engineering framework in the context of IEEE Standard 1012 (Software Verification and Validation), highlighting the peculiarities of its application to nuclear instrumentation and medical imaging simulation software. It demonstrates validation methods and techniques applicable in everyday experimental practice through a series of relevant simulation scenarios, where recent results produced by the authors in the context of Geant4 validation are illustrated as concrete applications. Special emphasis is given to the validation of basic simulation observables in sensitive applications. Software tools enabling objective quantification of simulation validation are critically examined. Given the interdisciplinary character of the subject, its presentation intends to stimulate the discussion of methods, techniques and results of common interest to different experimental communities across the boundaries of a single discipline.

(09:30) N5A1-4, OSIRIS - Tests of a Measurement Restriction System for on-Site Inspections under the Comprehensive Nuclear-Test-Ban Treaty

B. Milbrath1, A. Caffrey2, A. Egger2, K. Krebs2, S. Padgett3, G. Warren1, N. Wimer3

1Pacific Northwest National Laboratory, Richland, WA, USA
2Idaho National Laboratory, Idaho Falls, ID, USA
3Lawrence Livermore National Laboratory, Livermore, CA, USA
Under the Comprehensive Nuclear-Test-Ban Treaty (CTBT), high resolution gamma spectroscopy measurements during an on-site inspection (OSI) may be restricted to only identify and quantify “relevant radionuclides” from fission rather than full spectral analysis. There are currently seventeen agreed upon, nongaseous CTBT OSI relevant radionuclides. To investigate implementing such a measurement restriction, we developed the On-Site Inspection RadioIsotopic Spectroscopy— OSIRIS —system. OSIRIS includes software filters that limit the display of spectral data to radioisotopic information relevant to CTBT OSIs, e.g., I-131. To evaluate OSIRIS or other measurement restriction approaches, three software and two hardware performance metrics were established. The software metrics are true positive identification of the relevant radionuclides, true negative identification of the relevant radionuclides, and peak area fidelity of the relevant radionuclides. The hardware metrics are energy-calibration accuracy and electronic-gain stability. For the software tests, a set of over 150 OSI-relevant spectra was produced via measurements and simulation. The test spectral compositions include non-nuclear-explosion scenarios, e.g., a severe nuclear reactor accident, and nuclear-explosion scenarios such as a vented underground nuclear test; and at many levels of activity relative to natural background. Compared to expert manual analyses of these test spectra, the OSIRIS analyses were over 95% correct for identification of treaty-relevant fission-product isotopes. Hardware tests utilized an ORTEC trans-SPEC high-purity germanium gamma-ray spectrometer that was also used in the CTBT Organization (CTBTO) Preparatory Commission’s Integrated Field Exercise 2014 – an OSI exercise conducted in Jordan over five weeks that involved over 200 participants.