Towards big SAR data era: An efficient Sentinel-1 Near-Real-Time InSAR processing workflow with an emphasis on co-registration and phase unwrapping

• Published in: ISPRS journal of photogrammetry and remote sensing Vol. 188; pp. 286 - 300
• Main Authors: Ma, Zhangfeng, Liu, Jihong, Aoki, Yosuke, Wei, Shengji, Liu, Xiaojie, Cui, Yan, Hu, Jia, Zhou, Cheng, Qin, Shuhong, Huang, Teng, Li, Zhen
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.06.2022
• Subjects: Along-track Deformation; Co-registration; InSAR; Near-Real-Time; Phase Unwrapping; Sentinel-1; InSAR; Phase Unwrapping; Co-registration; Along-track Deformation; Near-Real-Time; Sentinel-1
• ISSN: 0924-2716; 1872-8235
• DOI: 10.1016/j.isprsjprs.2022.04.013

Sentinel-1 Near-Real-Time (NRT) processing embraces great potential in its applications to a wide range of research topics. Towards higher quality NRT products, it is crucial to avoid data reprocessing and to dynamically update InSAR time series. Although state-of-the-art techniques, enabled by high-performance computers and high-capacity storage platforms, make the highly multi-looked data available online quickly, the technique and procedure for efficiently obtaining relatively high resolution NRT data for a specified region are still lacking. Here we propose a workflow in a review way towards efficient and high-resolution Sentinel-1 NRT InSAR processing. In addition, we are committed to addressing issues that are often overlooked but indispensable in NRT data processing: co-registration and phase unwrapping. Through this workflow, not only NRT results in the radar Line-of-Sight (LOS) direction can be obtained, but also along-track NRT observations in overlap regions of images can be achieved. Unlike deriving LOS time series displacements which requires additional operations after co-registration, along-track displacements, a by-product of co-registration, can be directly obtained without further processing. We test and validate the proposed workflow in three regions with representative geomorphological processes including volcanic activity, city subsidence, and landslides. Experimental results demonstrate that our workflow can perform NRT monitoring at a relatively high resolution and a relatively low computational and storage “burden”. The time efficiency is increased by ∼ 65% and the storage space is saved by ∼ 21.8%.