Remote Sensing for Persistent Monitoring of Nuclear Power Plants

All nuclear power plant operators have a duty to inform the international community in case of potential damages or incidents at their plants with transboundary effects. This duty is a paramount, such that neighboring nations can take the appropriate actions to mitigate the effects of the potential nuclear fallout. History unfortunately shows that this duty may be neglected. This leads to a need for independent verification of nuclear power plant health. Remote sensing technologies present a promising avenue to achieve indications of nuclear power plant distress. New advances within Machine Learning methodologies for Remote Sensing presents an ability to automatically monitor nuclear power plants for changes or damages, which could raise concern. The goal is to achieve persistent, automatic, and global monitoring of nuclear power plants, for nuclear fallout early warning.

This study uncovers how new and existing remote sensing methodologies can be leveraged to detect changes and damages at nuclear power plants. This study includes existing and repurposed fire detection, structural change detection, and flood detection Machine Learning methodologies. Combined with new research on measuring steam generation from cooling towers, and temperature changes in cooling water reservoirs. This study is based on a large body of data from nuclear power plants from optical and SAR remote sensing payloads. The study also leverages existing, and open-source data from various natural disasters which are transferrable to the nuclear power plant monitoring task.