Interferometric synthetic aperture radar terrain elevation mapping from multiple observations

Abstract

All prior interferometric SAR imaging experiments to date dealt with pairwise processing. Simultaneous image collections from two antenna systems or two-pass single antenna collections are processed as interferometric pairs to extract corresponding pixel by pixel phase differences which encode terrain elevation height. The phase differences are wrapped values which must be unwrapped and scaled to yield terrain height. We propose two major classes of techniques that hold promise for robust multibaseline (multiple pair) interferometric SAR terrain elevation mapping. The first builds on the capability of a recently published method for robust weighted and unweighted least-squares phase unwrapping, while the second attacks the problem directly in a maximum likelihood (ML) formulation. We will provide several examples (actual and simulated SAR imagery) that illustrate the advantages and disadvantages of each method.<>