Integrating Seismic, Infrasound, and Radiation Data to Characterize Movements of Nuclear Materials

Monitoring and characterizing the movements of nuclear material is an important task in understanding the operations of a nuclear facility for non-proliferation. A facility’s normal cadence can indicate its operations, e.g., unirradiated material goes from building A to B and irradiated material goes from building B to A, and deviations from the normal cadence can be an indication of diversion or misuse. We explore how diverse data streams can be integrated to characterize movements of nuclear materials using data from networks of seismic, infrasound, and radiation detectors deployed at Oak Ridge National Laboratory’s (ORNL) High Flux Isotope Reactor (HFIR) and Radiochemical Engineering Development Center (REDC) complex. Sources of detector-specific background, inconsistent data from multiple detectors, and limited labeled data present complications to this task. We will present strategies for multi-modal data integration to address these challenges, while incorporating the most probable route from the radiation data, vehicle information from seismic and infrasound data, and likely nuclear material present from the radiation data, resulting in a data-driven characterization of the contents and movements of nuclear material transfers.