Year
2017
Abstract
Recent developments in remote sensing and geographic information analysis offer the International Atomic Energy Agency new opportunities in using semi-automated techniques to monitor change continuously over nuclear facilities worldwide. Satellite image providers are launching constellations of small and nano satellites, with the aim of offering global, high spatial resolution images on a daily basis, and for selected areas, potentially as many as 70 images per day. In an era of imagery as a service, a robust change detection model is required to handle the anticipated vast volume of data. This paper describes the conceptual development of an automated procedure for analyzing large time-series of satellite data, including images from constellations of small satellites. A change detection model is developed to incorporate multiple spatial datasets, including digital elevation models extracted from satellite imagery, ancillary vector data from open sources or from previous image analysis, and medium and high spatial resolution satellite imagery time series data. The system compares new imagery, as it is ingested, to the existing archive, in order to identify changes in the context of the nuclear facility’s fuel cycle stage. If changes within a facility’s infrastructure are identified, an automated alert is generated that flags the image and prompts further automated and traditional manual evaluations by analysts. The proposed conceptual model is illustrated through the examples of uranium mining and nuclear power plants.