Enhancing Ground-Based SAR Monitoring With PCA-Based Geometry Transformation for Improved Phase Unwrapping

Abstract

Ground-based synthetic aperture radar (GB-SAR) systems are most often used for landslide and open-pit mine monitoring due to their high temporal sampling and spatial coverage. For infrastructure monitoring, it has not yet attained widespread adoption for this purpose, mainly due to the complex imaging geometry and related challenges for phase unwrapping. In case of vertical structures, the GB-SAR projection geometry induces strong layover and foreshortening that may be difficult to handle during phase unwrapping. In this letter, we present an approach based on principal component analysis to transform the GB-SAR interferograms into a suitable geometry, to ease the phase unwrapping, making it more efficient and more robust against unwrapping errors. The method is tested on a distorted imaging scenario at the Enguri Dam in Georgia. The results show a strong improvement of the phase unwrapping and encourage the usage of this method in the case of interferometric analysis of strongly distorted SAR images. Depending on the scenario, the subsequent required filtering may remove local deformation patterns, but considerably increases the consistency of the global displacement pattern. Two displacement maps after correction for atmospheric and repositioning influences are presented and compared with a numerical simulation based on a model calibrated with the plumblines inside the dam. The comparison shows an overall good agreement between numerical simulations and the displacement maps.