Mitigation of atmospheric effects in deformation monitoring using dual-polarization MTInSAR and improved wavelet correlation analysis

Abstract

Multitemporal interferometric synthetic aperture radar (MTInSAR) has been widely used in ground deformation monitoring. The accuracy of MTInSAR is often limited by the atmospheric phase screen (APS), even though many studies focus on atmospheric correction. One efficient approach utilizes wavelet correlation analysis (WCA) to separate APS based on the polarimetric independence of InSAR signals. However, the traditional WCA-based method fails to separate the APS when the deformation signal existed. To address this issue, we propose a novel MTInSAR approach that utilizes dual-polarization data and an improved WCA to effectively separate the deformation and APS. In this method, we generate a series of zero temporal baseline interferograms (ZTBIs) with original interferograms to reduce the effect of deformation on the WCA. A hybrid indicator is then designed to determine the optimal wavelet decomposition scale in WCA, which enables more accurately separation of APS phase component of the ZTBI. The APS for the original interferogram is retrieved using the least squares from ZTBI. The APS-compensated interferograms can be then used to accurately estimate the ground deformation. The effectiveness and improvement of the proposed method when compared with the traditional methods is demonstrated with experiments using a Sentinel-1 polarized SAR dataset over the Santa Ana Basin, Los Angeles. The results demonstrate that the proposed method can effectively mitigate the effects of APS on the InSAR-derived deformation, with an root-mean-square-error less than 4.4 mm compared with GPS measurement. This represents an enhancement of 22.8% and 45.7% over the traditional small baseline subset and generic atmospheric correction online service based results, respectively.