The new tool uses advanced machine learning techniques to automate processing of EO Products within a high-performance Cloud Computing environment. It will be made available as an easy to use Cloud-based service for New Space mission providers and Institutions.
The volume of EO data from satellites has increased 10x in the last 5-years due to modern satellites launched by New Space companies and Governments/Institutions. The total data generated is already on the petabyte scale and it is not feasible to perform quality assurance of such data volumes with only human interaction alone, or traditional image processing techniques. The automated tool seeks to address this challenge of analysing the huge volume of EO data products.
“Quality assurance is a critical step in the Earth Observation value chain. Yet with the drastically increasing volumes of acquired data, traditional approaches to quality assurance are no longer viable. This activity will explore how novel technologies, such as cloud processing and new machine learning approaches, can help making the process more efficient and ensuring that the best quality data products are delivered to the user community.”
Dr Patrick Griffiths, EO data engineer at ESA/ESRIN and technical officer for the project
The new activity, funded by the European Space Agency (ESA), seeks to build on previous internal development activities. Overall, it seeks to strengthen Telespazio UK’s world leading capabilities in EO data quality assurance and to provide an easy-to-use service to commercial organisations and institutions.
“We are excited to be applying innovative Artificial Intelligence technology to develop this tool. This new capability allows cost effective provision of quality control services to our customers and allows them to continue provide leading-edge EO data and information products to their users”.
Dr Geoff Busswell, Head of Sales at Telespazio UK
This satellite image sensed over the Tunisian section of the Sahara Desert, demonstrates the presence of the scan start anomaly. This anomaly, which occupies the left hand side of the product scene, originates from the fault where the data sampling start time within each mirror scan is delayed and so a near-consistent proportion of data is lost from each nominal scan line (Image: ESA/NASA/USGS)