Reduction Of Ground Control Requirements For Spot Imagery By Pass Processing

Pass processing is a method to determine models for the geomet­ ric correction of satellite imagery. Most systems use a method we will call single scene processing in which the precision model is determined by marking ground control points (GCPs) within the area of the desired output scene. With pass processing, the cor­ rection model can be determined from GCPs located anywhere in within the same pass (orbit) as the desired output scene. The goal of pass processing is to achieve an accuracy for a given number of GCPs that is comparable to that obtained with single scene processing. Since the GCPs are distributed over many scenes the average GCP requirement per scene is greatly reduced. Moreover, pass processing reduces the sensitivity to the location of the GCPs and allows for the correction of scenes which contain no GCPs at all. This paper reports on the results of a study to extend the pre­ vious work on pass processing for Landsat imagery to SPOT and to measure the accuracy that can be achieved under various scenarios for the distribution and number of GCPs. The results showed that the goals of pass processing have been achieved for SPOT: accuracies comparable to single scene pro­ cessing were obtained using an average of 1/4 GCP per scene; large areas with no ground control were corrected.