A New Method for Reconstructing Decorrelated Insar Phase Measurements Over Densely Vegetated Natural Terrain

Phase decorrelation is a major limiting factor for Interferometric Synthetic Aperture Radar (InSAR) applications over densely vegetated terrain. In this study, we developed an algorithm to reconstruct spatially coherent InSAR phase measurements from severely decorrelated interferograms based on adaptive spatial filtering and the eigenvalue decomposition (EVD) of complex correlation matrix. At each pixel of interest, our method estimates a phase vector that best fits phase observations of all interferograms. We employed this algorithm to analyze C-band Sentinel-1 data over the Gulf Coast region and the Dhaka area in Bangladesh. In both cases, severe decorrelation artifacts are observed even in interferograms with short temporal baselines. We show that our algorithm can (1) minimize phase nonclosure of triplet combinations of any three individual SAR scenes; and (2) improve the consistency between independently processed InSAR and GPS observations.