A simple robust two-scale phase component inversion scheme for persistent scatterer interferometry (dual-scale PSI)

Abstract

We have developed a new persistent scatterer interferometry (dual-scale PSI) method with a simpler and more robust phase component inversion than that of the iterative linear modeling used in many commercial and scientific PSI solutions. The new method can measure displacements down to a millimetre, also for smaller interferometric stacks (~20 scenes) and strongly nonlinear motion histories. A key feature is a split of the phase signal into small and large spatial scale contributions. This is carried out via a priori creation of a “pseudo-atmospheric” stack using a strong nonlinear filter at the large spatial scale of about five times the atmospheric correlation length (~15 km). We assume the external digital elevation model used for initial flattening unbiased at the large spatial scale. The true atmospheric phase can then be separated from the large-scale surface displacement through temporal low-pass filtering. We estimate and subtract the height error from the small spatial scale phase component assuming only piece-wise “smoothness” of the displacement time series. The result is the (model-free) nonlinear displacement at small spatial scale. The simplicity of the method allows fast, automated throughput of stacks consisting of full-sized scenes; a single iteration is usually sufficient to achieve reliable results. We demonstrate the performance on a RADARSAT-1 stack over Los Angeles and compare with global positioning system (GPS) records.