THE GEOMETRY OF SAR IMAGES FOR GEOCODING AND STEREO APPLICATIONS

The geometry of Synthetic Aperature Radar (SAR) needs to be properly understood in order to match the images to a map or to extract topographic data. This paper concentrates on the distortions introduced into an image because of the relief of the ground and the way in which the topography can be extracted if two overlapping images are available. A mathematical model can be determined to describe a single image, and if a terrain model is available this can be used with the mathematical model to geocode the image. Information on the way in which the SAR data is recorded and processed is essential for producing a correct model. A number of methods have been developed and tested and these are described and assessed in the paper. Accurate geocoding is not possible without terrain data and if such data are not available, only corrections to allow for the ellipsoidal earth can be applied. A possible solution to this problem is to produce the elevation model from stereoscopic SAR or, if this is not possible, from a stereo pair made up of a SAR image and an image from another sensor such as SPOT. The geometric aspects of this are considered, together with some of the practical problems which may arise when working with stereo images of this type.