The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Articles | Volume XXXIX-B1
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XXXIX-B1, 225–229, 2012
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XXXIX-B1, 225–229, 2012

  24 Jul 2012

24 Jul 2012


M. Ravanbakhsh1, L.-W. Wang2, C. S. Fraser1, and A. Lewis2 M. Ravanbakhsh et al.
  • 1CRC for Spatial Information, Box 672, Carlton Sth, Vic 3053, Australia
  • 2Geoscience Australia, GPO Box 378, Canberra, ACT 2601, Australia

Keywords: satellite imagery, georeferencing, ALOS PRISM, orthorectification, long-strip adjustment

Abstract. The Australian Geographic Reference Image (AGRI) is a national satellite image mosaic that covers the vast Australian continent. Formed from 9560 ALOS PRISM images, AGRI provides a spatially correct reference image at 2.5 m resolution, with 1-pixel accuracy. The production of AGRI was made feasible by the development and implementation of a long-strip adjustment technique that facilitated a refinement in the georeferencing process of orbit and attitude parameters for orbital segments comprising 50 or more images. The strip of images is effectively treated as a single image. The ground control requirements for such full-pass georeferencing, which does not require the measurement of tie or pass points, amount to only 4–8 GCPs for the complete strip rather than 4 or more per image within the strip. Once the adjusted orbit parameters are obtained, the georeferencing and orthorectification process can revert to a fully automatic image-by-image computation. This paper first overviews AGRI and then describes the longstrip adjustment technique that made its production possible. Testing and validation are then discussed via the example of the georeferencing of a 1527 km single strip of 55 PRISM images. The testing phase verified that 1-pixel accuracy georeferencing could be achieved with the long-strip adjustment approach, and in the production of AGRI accuracy checks against 2460 checkpoints yielded an RMS discrepancy of close to 2.5 m and a 90% Circular Error (CEP90) of 5.5 m.