Quantized compression of SAR data: Bounds on signal fidelity, InSAR PS candidates identification and surface motion accuracy

Abstract

Satellite radar imaging has been used as a remote sensing tool for studying Earth’s surface. High spatial resolution achieved by Synthetic Aperture Radar (SAR) allows to classify landcover and, using radar interferometry, to measure topography and surface deformation at millimeter scale. However, the handling of voluminous SAR data has presented a significant challenge over the past decade, leading to a technological barrier for researchers to study regional and global scale problems. A possible solution will be to explore data compression algorithms. Despite of the importance of SAR data compression, limited research has focused on downsizing user-level SAR images, and none of the existing studies have explicitly explored the impact of compression on interferometric (InSAR) processing.This study investigates compression algorithms to downsize Sentinel-1 single-look complex (SLC) images by a factor of 2–4 within signal noise levels. The performance of the compressed images was evaluated in the context of PS-InSAR processing, specifically on persistent scatterer (PS) candidate selection and displacement time-series estimation. We observe that even when images were compressed by a factor of 4, more than 98 % of the PS candidates were identical to those selected from the original uncompressed images. In terms of linear velocity, the discrepancy is <2.5 mm/yr at a 99 % confidence interval for the identical PS set. This work provides evidence that, in the context of PS-InSAR processing, Sentinel-1 radar data can be compressed by a factor of 4, leading to possible wider participation in InSAR research and applications.

Keywords

• Sentinel-1 SAR data compression
• Environmentally-responsible Big data processing
• InSAR precision
• Climate change
• Pearl River region deformation mapping