Geodetically Accurate InSAR Data Processor

Abstract

We present a new interferometric synthetic aperture radar (InSAR) processing approach that capitalizes on the precise orbit tracking that is available with modern radar satellites. Our method uses an accurate orbit information along with motion-compensation techniques to propagate the radar echoes to positions along a noninertial virtual orbit frame in which the location and focusing equations are particularly simple, so that images are focused without requiring autofocus techniques and are computed efficiently. Motion compensation requires two additional focus correction phase terms that are implemented in the frequency domain. If the images from an interferometric pair or stack are all computed along the same reference orbit, flat-Earth topographic correction is not needed, and image coregistration is simplified, obviating many difficulties that are often encountered in InSAR processing. We process several data sets collected by the ALOS PALSAR instrument and find that the geodetic accuracy of the radar images is 10-20 m, with up to 20 m of additional image distortion needed to align 100 km × 100 km scenes with reference digital elevation models. We validated the accuracy by using both known radar corner reflector locations and by the registration of the interferograms with digital maps. The topography-corrected interferograms are free from all geometric phase terms, and they clearly show the geophysical observables of crustal deformation, atmospheric phase, and ionospheric phase.