Added value of multitemporal polarimetric UAVSAR data for permanent scatterers detection

Abstract

In the last decades, differential synthetic aperture radar (SAR) interferometric (InSAR) (DInSAR) techniques have been used to estimate the Earth's surface deformation with high resolution. In this paper, we present an approach for increasing the quantity of permanent scattered pixels. These pixels are selected for DInSAR processing based on polarimetric information prepared by new sensors. The objective of this paper is then to test existing algorithms that confirm the contribution of polarimetric data for improving persistent scatterers (PS) detection. These algorithms are formulated based on two different selection criteria: amplitude dispersion index and mean coherence. Different approaches are analyzed to optimize both selection criteria in terms of pixels' quantity and density and finally their results are quantitatively compared. Experimental results with exploiting quad-pol UAVSAR data set over an urban area in CA, provide the expected improvement. Comparing the number of PSs between quad-pol with dual-pol and single-pol cases illustrate remarkable improvement in both selection criteria. For quad-pol case, we achieve an increase of 50% and 60% with respect to dual-pol and single-pol data, respectively, when using average coherence and over 6 times more for amplitude dispersion index. The results of our study demonstrates the added value of polarimetric SAR observations (dual pol and quad-pol) for improved permanent scatterers detection monitored areas.