SBAS-InSAR deformation reconstruction based on low-rank matrix completion in southern Zibo, China

Abstract

As an advanced form of differential interferometric synthetic aperture radar (DInSAR), small baseline subset (SBAS)-interferometric synthetic aperture radar (InSAR) boasts large space coverage and high precision and provides an effective monitoring technique for ground deformations. Maintaining high coherence through the monitoring process is a common method for SBAS-InSAR to select ground stable points. Usually, such points are obtained with small temporal and spatial baselines. However, it is unlikely to maintain high coherence in non-urban and non-rock landscapes, due to the change of surface cover or vegetation growth. This change is common in long-term surface deformation. Even with a high average coherence, an interferogram still contains many low coherence areas, where it is difficult to implement phase unwrapping or collect sufficient deformation information. This often results in an incomplete spatial image of the deformation pattern, which hinders the interpretation and research of surface deformation. What is worse, surface deformation might not be detected at all in incoherence areas. To solve these problems, this paper proposes a low-rank matrix completion method to complete the low coherence areas, making the interferograms complete. Then, the general SBAS method was followed to extract the deformed time series. The proposed strategy was applied to the time series monitoring of surface deformation of a 250-km2 area in southern Zibo, China, from January 2, 2016, to March 22, 2018. Compared with the traditional SBAS method, the proposed method can obtain more information of surface deformation. The complete deformation information of the area is obtained.