A Framework for the Solution of Inverse Radiation Transport Problems

Radiation sensing applications for SNM detection, identification, and characterization all face the same fundamental problem: each to varying degrees must infer the presence, identity, and configuration of a radiation source given a set of radiation signatures. This is a problem of inverse radiation transport: given the outcome of a measurement, what source terms and transport medium caused that observation? This paper presents a framework for solving inverse radiation transport problems, describes its essential components, and illustrates its features and performance. The framework implements an implicit solution to the inverse transport problem using deterministic neutron, electron, and photon transport calculations embedded in a Levenberg-Marquardt nonlinear optimization solver. The solver finds the layer thicknesses of a one-dimensional transport model by minimizing the difference between the gamma spectrum calculated by deterministic transport and the measured gamma spectrum. The fit to the measured spectrum is a full-spectrum analysis-all spectral features are modeled, including photopeaks and continua from spontaneous and induced photon emissions. An example problem is solved by analyzing a high-resolution gamma spectrometry measurement of plutonium metal.