ESA GNC Conference Papers Repository

Recent improvements of key technology required for Earth observation payloads make it now possible to consider high resolution Earth observation mission from the geostationary orbit within five to ten years. The significant benefits of these missions and in particular permanent visibility and very short revisit delays will soon become real. Nevertheless, now those main technological breakthroughs for the payload are achieved we have to focus on how to meet the very stringent pointing performances for the platform. From an angular point of view, a 20m resolution from GEO is equivalent to a 0.4m resolution from a 720km orbit. The main challenges are thus the on-board stability of the line of sight and the knowledge of the realised line of sight used for image products. The latter, also called image geolocation, was the subject of an R&T study « Accurate geolocation for Earth images from geostationary orbit » managed by CNES and conducted by Astrium for Cnes in 2010. The future missions taken as reference were GEO HR, a survey mission for agriculture and natural disaters with a 20m nadir resolution, and GeoOcapi, a survey mission of the European and North-African seas with a 250m nadir resolution. Even with these completely different resolutions the geolocation need is quite similar in both cases with an objective in the range of 50 to 100m (1s) on ground all contributors included. Consequently this means that an improvement of a factor 10 to 100 on the classical attitude and orbit estimation performances is required. The study approach proceeded in two steps: Review of solutions for orbital position estimation and for attitude estimation with a preliminary assessment of the performances; this has led to the selection of promising architectures that could answer to the geolocation needs. Then detailed definition of the geolocation architectures and detailed analysis of their performances. For that purpose, we rely on an in-house Kalman filter prototyping toolbox. Pending the reference mission and the on-board/on-ground sharing specification Astrium has proposed three different architectures for restitution that will be described in this paper, all achieving the required geolocation need. The first architecture is based on ground control points use in an Image Navigation & Registration onground filter. The other two are based on star sensing: using star measurement through the optical line of sight of the payload therefore minimizing the measurement error and the thermoelastic distorsion. All these results demonstrate the feasibility of 50-to-100m performance with architectures that are compatible with a launch before 2020.