An Adaptive, Statistical Multiscale Phase Unwrapping Approach to Process Large Swath Interferograms

This study investigates the potential of a statistical-based, adaptive approach to unwrapping sequences of differential synthetic aperture radar (SAR) interferograms that cover a large swath of the terrain. The proposed method adopts a multiscale decomposition strategy to identify efficiently and then process sets of coherent points at different spatial scales. The coherent point selection process is performed considering the statistical properties of the stack of wrapped multilooked SAR interferograms generated at various scales. Overall, the adopted procedure allows automatically recognizing the areas in large swath interferograms where significant and reliable phase changes occur while moving from neighboring spatial scales. Over these regions, multiscale phase unwrapping (PhU) operations are performed efficiently, with a computational improvement and without losing significant information. To this aim, the implementation of a conditioned space-time PhU scheme that operates sequentially at different spatial grids is detailed. Then, the unwrapped interferograms are inverted to generate ground displacement time series through advanced multitemporal interferometric SAR (MT-InSAR) approaches, recovering information at different scales (from local to regional/continental). Experimental results have been obtained by applying the developed scheme to large-swath SAR datasets collected at the C band by Sentinel-1 sensors. The results demonstrate the feasibility and soundness of the developed multiscale PhU method.