InSAR Remote Sensing Over Decorrelating Terrains: Persistent Scattering Methods

Interferometric synthetic aperture radar, or InSAR, is a visual geodetic technique permitting detailed mapping of motion over wide areas. InSAR has been limited to regions without much vegetation, which shields the ground from the radar signals and contributes random motions to the observed deformation. The resulting "decorrelation" of the echoes precludes accurate displacement measurements in these areas. Decorrelation also occurs in interferograms with acquisitions separated too far in the sky. Yet certain points, denoted persistent scatterers, in a radar image are stable, do not decorrelate, and form a network of fiducial points that allow measurements in otherwise poor-quality interferograms. We have generalized an algorithm to find networks of stable points in natural terrain, rather than in urban areas, and applied the method to spaceborne satellite data. Using modern information theory to optimize persistent scatterer detection, we can now find many, many more such points than previously possible. We have applied this improved algorithm to the San Francisco Bay segments of the San Andreas and Hayward faults, and in both cases find that a large number of stable points are seen in the vegetated areas that have to date resisted InSAR analysis. Our method of integrating information theoretic estimation and detection theory to all parts of the method, improves the identification, filtering, and phase unwrapping of the observations. Identification of stable true-ground scattering points permits mapping of subtle surface motions and deformations and also of "bare-Earth" topography.