Dam Monitoring - Optimization of Sidelobe Suppression in Radar Images during StaMPS Persistent Scatterer Interferometry Processing

Kurzfassung

Dams are critical infrastructure and require continuous monitoring as structural failures can have catastrophic consequences. Persistent Scatterer Interferometry (PSI) is a satellite-based technique capable of detecting millimeter-scale surface deformation by analyzing time series data from multiple Synthetic Aperture Radar (SAR) acquisitions. Sentinel-1 (S-1) provides freely available SAR data with a 6 to 12 day repeat cycle, enabling regular and cost-effective monitoring. PSI is particularly effective in urban environments, where stable structures, such as dams, act as Persistent Scatterers (PS), maintaining phase coherence over time. The use of active Corner Reflectors (CRs) further enhances the phase signal for PSI analysis, providing a perfectly stable reference point. However, their strong signal can introduce sidelobes into the SAR data. These artifacts may obscure important features and lead to false PS selections, making effective sidelobe suppression techniques crucial for reliable dam monitoring. Therefore, this study investigates sidelobe and noise suppression in the context of PSI-based monitoring of the Sorpe dam in the Sauerland region of Germany. Several filtering techniques are tested, including apodization filters for sidelobe suppression (Hamming, Dual Apodization (DA), and Spatially Variant Apodization (SVA)), a median speckle filter, and the Goldstein filter for phase noise reduction. These methods are integrated into the traditional SentiNel Application Platform to Stanford Method for Persistent Scatterers (SNAP2StaMPS) and Stanford Method for Persistent Scatterers (StaMPS) workflows to process a S-1 time series from May to December 2023. Furthermore, an amplitude-based approach for sidelobe suppression is developed by applying filtering exclusively to the amplitude and recombining it with the original phase using the Cartesian representations of I (In-phase) and Q (Quadrature). The effectiveness of each method is evaluated based on its impact on sidelobe suppression, the retention of reliable PS, and phase quality. In addition, the performance of the SVA method is further assessed under different sampling conditions by testing various sampling ratios, including Nyquist sampling, an integer-2 ratio, and a non-sampled configuration with native sampling ratios of 1.35 in range and 1.61 in azimuth, based on Sentinel-1 (S-1) Single Look Complex (SLC) data acquired in Interferometric Wide Swath Mode 2 (IW2). Among all tested methods, only the SVA filter effectively suppresses sidelobes while retaining reliable PS points. When combined with the amplitude-based approach, SVA preserves the original phase, while still achieving sidelobe suppression. The Hamming and DA filters show only minimal improvement, and the median filter introduces distortion and leads to random PS selection, while Goldstein filtering does not suppress sidelobes. Nyquist sampling provides the most distinguishable mainlobe–sidelobe separation and achieves the lowest sidelobe levels at -30 decibel (dB) in the azimuth direction. While integer-2 sampling does not reduce sidelobes, the unsampled dataset also performs better in azimuth but exhibits higher sidelobe levels at -22 dB. Overall, SVA combined with amplitude-based processing provides robust sidelobe suppression and a phase-preserving solution to improve PSI-based dam monitoring, while smapling affects SVA performance. The developed amplitude-based SVA approach offers a practical enhancement to existing PSI workflows and demonstrates the potential for accurate and reliable deformation analysis of critical infrastructure using freely available S-1 data.