Rectification of high resolution satellite data: evaluating accuracy for map updating

Since the first commercial high-resolution satellite was launched in the late 1999, the potential of metric accuracy has been widely demonstrated and increased by the 3D imageries offered by sensors such as IKONOS II, EROS and SPOT. This paper addresses the topic of the orthorectification of high resolution satellite data for map updating at scale of up to 1:5,000 and its dependency on DEM accuracy. In order to do so, EROS stereo-imagery, IKONOS II and QuickBird images were acquired over a large portion of the city of Bologna (northern Italy) and its surrounding area. Different orthoimages were generated from IKONOS and QuickBird scenes using DEMs with varying grid spacing and accuracy. Two DEMs were derived from digitalisation of technical maps with grid size of 40 metres and 10 metres respectively, whereas the photogrammetric processing of EROS stereo pair for DEM generation is currently being investigated. The PCI Geomatica software package was used for DEM management and orthorectification. A range of in bundle models is available for orthorectification including rigorous camera and satellite orbital model and rational function model with coefficients computed either from GCPs or metadata provided with the scenes. The accuracy achieved by combining a single satellite image with a DEM was evaluated using an external checkpoint dataset and compared. Thus a large number of Ground Control Points and checkpoints were acquired all over the area by a Real Time Kinematic GPS survey. Overall accuracy and problem encountered during the processing will be discussed.