PNNL

Abstract

Algorithms that estimate the location and magnitude of an atmospheric release using remotely sampled air concentrations typically involve a single chemical or radioactive isotope. A new Bayesian algorithm is presented that makes discrimination between possible types of releases (e.g., nuclear explosion, nuclear power plant, or medical isotope production facility) an integral part of the analysis for samples that contain multiple isotopes. Algorithm performance is demonstrated using synthetic data and correctly discriminated between most release-type hypotheses, with higher accuracy when data are available on three or more isotopes.