Control and verification of nuclear site declarations


Project concept and background

The overall project concept is to contribute with experience and examples to the International Atomic Energy Agency (IAEA) how the use of an open information source such as commercial high-resolution satellite images could benefit the process of control of nuclear site declarations. This project shall use Swedish facilities as a test to investigate how declarations supplied to the Swedish Nuclear Power Inspectorate (SKI) can be verified. The results can be used by IAEA on an international level.

By comparing nuclear sites with VHR satellite images collected over a time sequence it will be possible to independently detect and follow small re-constructions of buildings, infrastructure changes or other activities. This is very important for detecting clandestine or undeclared extensions. With the possibilities of 3D modelling of nuclear sites it may also be possible to detect extension of buildings in height. The height is needed to verify the number of stories declared and determine if the building size is reasonable in relation to alleged activity.

The Swedish Nuclear Power Inspectorate, SKI, is the governmental authority responsible for the collection, aggregation, verification and delivery of site declarations from the Swedish nuclear facilities to IAEA, and it is in the interest of SKI to perform this duty as effective as possible. One possible way is to use satellite data.

To develop a method for site declarations /control a joint project is formed between SKI and Metria Miljöanalys where Metria is the experts on satellite image analysis and SKI on site declarations.

Project goals

During 2004 the project aims to achieve the following main goals:

An operational method utilising VHR (e.g. QuickBird, IKONOS or SPOT-5) data for a more effective verification of nuclear facility declarations;

Demonstrate to IAEA the use of VHR satellite imagery in the process of control and verification of nuclear site declarations. Results from this project will be transferred to IAEA through the Swedish support program;

Increase the consistency and accuracy of Swedish declarations.

To achieve these main goals it will probably be necessary to develop the following procedures and technology:

Multi-temporal change detection techniques and methods for object/pattern recognition adapted for VHR image data acquired under various lighting and viewing conditions;

3D techniques to detect changes in building heights and for creating 3D models over nuclear sites which can facilitate inspections;

Streamlined image based production methods integrated and adjusted to the normal routines of SKI for verification of site declarations.


Approach and methodology

The first part of the proposed project will be directed towards methodology development of multi-temporal change detection and 3D-techniques for recognising and monitoring structural changes within nuclear facilities. This could also be used for planning of routine inspections and/or render the inspections of nuclear sites more effective. Most probably a visual examination of changes in combination with image processing techniques will be the most practicable approach.

The second part of the project will be the demonstration of the image-based technology within the normal workflow of SKI and IAEA. We believe it is essential that the introduction of imagery is streamlined, and that the project is managed so as to result in specified products or services. Important elements to consider will be format specifications, guidelines to power operators, concept for reporting, etc. These elements shall be compatible to existing requests and guidelines at IAEA.

Output results from the process will be in the form of reports describing suspected changes of facilities, or other non-normal activities that have not been reported earlier. These changes will thereafter be subject for further investigation.

Time table

The project will start January 2004. New VHR satellite data will be acquired. Data from May can be used and thereafter the availability of good data will have an influence on the time schedule but the project is planned to finish at end 2004.


Senast uppdaterad: 26 maj 2009