AUTOMATIC PRODUCTION OF A EUROPEAN ORTHOIMAGE COVERAGE WITHIN THE GMES LAND FAST TRACK SERVICE USING SPOT 4/5 AND IRS-P6 LISS III DATA

Since the mid 1980s an European Land Cover dataset has been regularly produced for land cover changes, land cover map (CORINE), high resolution forest layer and built-up areas including soil sealing. Within the GMES (Global Monitoring for Environment and Security) Fast Track Land Service 2006-2008 a new dataset of orthorectified satellite images has to be produced covering the EU25 and neighbouring countries (total 38 countries). On behalf of ESA/ESRIN the DLR established an automatic processing chain to orthorectify about 3800 satellite images (two European coverages) within a time frame of 5 months including quality control and creation of a consistent GCP database. High resolution satellite images from SPOT 4 (20m GSD), SPOT 5 (10m GSD) and IRS-P6 LISS III (23m GSD) serve as input to derive orthoimages in European Map Projection with 25m resolution and National Map Projection for each country with 20m resolution and with an overall accuracy better than 20m RMSE in each direction with respect to the European Land Cover dataset Image2000 (EU25) and USGS ETM+ Land Cover dataset (neighbouring countries). For SPOT4/5 the Line-of-Sight vector is derived from continues measurements of the state vectors and attitude parameters as well as the calibrated camera model provided by SpotImage. For IRS-P6 LISS III the RPCs (Rational Polynomial Coefficients) serve as input, which is provided by Euromap (Universal Sensor Model). Further input is the European wide digital elevation model (DEM) from SRTM-C band Version2 of NASA, improved by using inputs from MONAPRO and SRTM-X band DEM within a fusion process. In order to achieve the required accuracy of 20m RMSE ground control points (GCPs) are automatically extracted via image matching between the Image2000 / USGS Land Cover dataset and the new satellite scenes. From these GCPs corrections of the exterior orientation for SPOT4/5 and of the RPC for IRS-P6 LISS III (affine transformation) are derived. The paper describes the background, the requirement specifications and the methodology of the automatic orthorectification chain as well as its limitations in problematic cases. Quality assessment is based on automatically extracted ICPs, from which mean RMSE values for each scene and whole countries are derived or from which residual plots are produced. The paper also describes the H/W infrastructure established for this demanding task as well as the S/W environment, which is based on a mySQL database, to administrate the huge amount of data to organize the parallel processing.