Validated Synthetic Datasets for Treaty Verification

Nuclear treaty verification efforts will require a broad menu of negotiation options that can maintain treaty objectives while not releasing sensitive information about the items or introduce facility risk all while still being robust to potential cheating scenarios. The Enhanced Modeling for Arms Control Scenarios (EMACS) program, part of the Priority Research Objectives for Arms Control Technology Innovation, Verification and Evaluation (PROACTIVE) venture, seeks to aid in preparations for just such an eventuality through generating validated and physics-based synthetic datasets. The purpose of this work is to counter the requirement for large experimental campaigns, which are a difficult task within this application space. These datasets, based on experimental data and Monte Carlo N-Particle Transport Code (MCNP), are composed of source terms and detector response functions. The two may then be coupled with induced systematic and statistical variations representing a range of realistic scenarios that can in turn be used to benchmark detectors and algorithms that are still under development. Alterations could include items such as varying source strength, measurement times, as well as background rates and compositions, all without the need to obtain more data experimentally. The bulk of the work to date has occurred using data from the Spiral Zero experiment, wherein a treaty verification exercise was simulated. The exercises were performed with increases steps of complexity to demonstrate the relationship between confidence and sensitivity concerns as related to treaty accountable objects. The data from this experiment were in turn used to validate MCNP models which were used for synthetic data generation.