Characterization and identification of partially correlated persistent scatterers for InSAR remote sensing

Abstract

Interferometric synthetic aperture radar (InSAR) is an effective tool for measuring temporal changes in the Earth's surface and producing high accuracy, wide coverage images of crustal deformation fields. Decorrelation due to spatial and temporal baseline is a major limiting factor in estimating the deformation signal, but may be ameliorated by using persistent scatterer (PS) techniques. Phase unwrapping and subsequent deformation estimation on the spatially sparse PS network depend largely on the accurate selection of PS pixels and the density of the network. Many additional pixels can be added to the PS list if we are able to identify those in which a dominant scatterer exhibits partial, rather than complete, correlation across all radar scenes. In this work, we discuss and compare statistical methods to model, characterize, and select partially correlated PS pixels.