A new Persistent Scatterer InSAR method based on phase decomposition, with application to subsidence in greater Houston area

Abstract:

A Phase-Decomposition-based Persistent Scatterer InSAR (PD-PSInSAR) method is developed in this study to improve the coherence and spatial density of targets. The general idea of conventional PSInSAR is to find and analyze pointwise stable persistent scatterers (PS). In order to improve the PS network density, distributed scatterers (DS) has also been utilized in several advanced PSInSAR techniques. Unlike these techniques which assumes that a DS involves many independent small scatterers sharing the same scatterering mechanism, this study considers additional two general cases: (1) a DS that contains many small scatterers sharing two or more different scatterering mechanisms, (2) two or more dominant scatterers with different scatterering mechanisms that exist within the same resolution pixel.

DSs with multiple scatterering mechanisms can occur in rural areas and some urban areas, especially with low spatial resolution. Extracting information from DSs with multiple scatterering mechanisms is difficult for the existing DS algorithms because of the interference between different scatterering mechanisms. The new PD-PSInSAR method is developed to overcome this limit by using Eigen-decomposition to estimate the phases corresponding to the different scatterering mechanisms, and then implement these estimated phases in conventional PSInSAR process. Therefore, the interference between different scatterering mechanisms becomes mitigated and the obtained phases are expected to have better coherence.

This PD-PSInSAR technique is used to estimate the land deformation over the greater Houston area using 25 ENVISAT ASAR data spanning from July 2004 to June 2010. The deformation map reveals significant subsidence up to approximately 2 cm/year over north and northwestern part of greater Houston. Comparison between the conventional PSInSAR and PD-PSInSAR method verifies that the proposed method can detect more PSs and provide better coherences.