Bayesian Inference for Source Term Estimation: Application to the International Monitoring System Radionuclide Network

Abstract

In recent years, there has been an enormous quantity of data obtained from the International Monitoring System radionuclide network for the verification of the Comprehensive Nuclear-Test-Ban Treaty. The complexity of the instruments deployed here, of the radionuclide sources, and of the myriad of scientific questions related to treaty verification lead invariably to complex inference problems (associated with source term estimation) that require the application of sophisticated statistical tools. In this report, we demonstrate that a rigorous and general framework for addressing these problems is through Bayesian probability theory, allowing the rational inference of the posterior probability distribution of the source parameters of interest given any prior information and available activity concentration measurements. The methodology is demonstrated by application to two different problems: namely, the emission rate profile reconstruction of a radioxenon release from the Fukushima Daiichi nuclear power plant and source reconstruction (location and emission rate) of a radioxenon release from the Chalk River Laboratories (CRL) medical isotope production facility. The sampling of the resulting posterior distribution of the source parameters is undertaken using two different Markov chain Monte Carlo techniques: namely, nested sampling and multiple-try differential evolution adaptive Metropolis sampling with a past archive. For the Fukushima nuclear power plant release, it is demonstrated that the limited temporal extent of the activity concentration time series obtained from the seven sampling sites cannot be used to constrain the emission rate profile at a later time. In particular, the Bayesian credible intervals for the reconstruction of the emission rate profile provide a quantitative indication of the uncertainty in this quantity,

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 2014
Accession Number
AD1004181

Entities

People

  • Alain Malo
  • Eugene Yee
  • Ian Hoffman
  • Kurt Ungar
  • Nils Ek
  • Pierre Bourgouin

Organizations

  • Defence Research and Development Canada

Tags

Communities of Interest

  • Biomedical
  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Accuracy
  • Bayesian Inference
  • Computational Science
  • Data Science
  • Detectors
  • Information Science
  • Markov Chains
  • Markov Processes
  • Measurement
  • Monte Carlo Method
  • North America
  • Nuclear Power Plants
  • Probability
  • Probability Distributions
  • Standards
  • Statistical Inference
  • Statistics

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Seismology
  • Statistical inference.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference