A Robust Interferometric Phase Extraction Algorithm With Hierarchical PS Sequence Construction for GNSS-Based InBSAR

Abstract

Global Navigation Satellite System (GNSS)-based interferometric bistatic synthetic aperture radar (InBSAR) enables deformation monitoring using on-orbit navigation satellites as transmitters and a ground-fixed receiver. However, the low spatial resolution, poor signal-to-noise ratio (SNR), and inherent asymmetric bistatic geometry of GNSS-based InBSAR cause focus position offsets and intermittent missing observations of persistent scatterer (PS), resulting in fragmentation of PS sequences as well as decreasing the availability of interferometric phase. This article presents a robust phase extraction algorithm for GNSS-based InBSAR by introducing a completion strategy. First, preselected PSs (prePSs) are identified and associated under temporal, spatial, and coherence constraints to construct reliable PS candidate links. Second, in order to alleviate the PS sequence fragmentation, the candidate links are organized into disjoint and temporally ordered PS sequences through a hierarchical optimization with one-to-one constraints. Third, interferometric phase is extracted hierarchically by enhancing long sequences with neighboring short sequences as auxiliary observations and by recovering short sequences under local reference constraints with reliability gating and adaptive regularization. Experiments using real data from three BeiDou satellites show that the proposed algorithm improves the mean coherence of PS sequences from 0.917 to 0.964, increases the mean number of valid differential interferometric pairs from 367 to 518, and also reduces the RMSE of deformation estimation from 12.31 to 5.57 mm.