A Novel Measure for Categorization and Optimal Phase History Retrieval of Distributed Scatterers for InSAR Applications

Time-series interferometric synthetic aperture radar (InSAR) methods are developed in recent times to retrieve deformation signal from two types of radar targets, i.e., persistent scatterers (PSs) having stable phase history and distributed scatterers (DS) with moderate temporal coherence. Recent algorithms (SqueeSAR, CAESAR, and PD-PSInSAR) claim to jointly process PS and DS. SqueeSAR algorithm applies a threshold based on the statistically homogenous pixels (SHP), which prevents only the PS pixels having single-bounce scatterer from inclusion in DS processing, but the remaining types of PS pixels cannot be preserved. The included PS pixels lose their coherent information during phase history optimization in DS processing. In this paper, a new measure, the similar time-series interferometric phase (STIP) pixel is used for the identification of brotherhood pixels, which improves the classification of interferometric pixels and coherence estimation. The application of STIP instead of SHP significantly improves the phase optimization efficiency of DS pixels while retaining valuable information of PS more effectively. Furthermore, to optimize the interferometric phase of DS pixels, a least squares model is developed, which is more time efficient as compared to the Maximum likelihood estimators of the conventional DS processing techniques. Results on SAR data acquired over specific sites of Uttarakhand, India, have shown superior performance of the proposed methodology over the standalone PS-InSAR, SqueeSAR, and PD-SAR.