A persistent scatterer method for retrieving accurate InSAR ground deformation map over vegetation-decorrelated areas

Abstract:

InSAR is commonly used to measure surface deformation between different radar passes at cm-scale accuracy and m-scale resolution. However, InSAR measurements are often decorrelated due to vegetation growth, which greatly limits high quality InSAR data coverage. Here we present an algorithm for retrieving InSAR deformation measurements over areas with significant vegetation decorrelation through the use of adaptive interpolation between persistent scatterer (PS) pixels, those points at which surface scattering properties do not change much over time and thus decorrelation artifacts are minimal. The interpolation filter restores phase continuity in space and greatly reduces errors in phase unwrapping. We apply this algorithm to process L-band ALOS interferograms acquired over the San Luis Valley, Colorado and the Tulare Basin, California. In both areas, groundwater extraction for irrigation results in land deformation that can be detected using InSAR. We show that the PS-based algorithm reduces the artifacts from vegetation decorrelation while preserving the deformation signature. The spatial sampling resolution achieved over agricultural fields is on the order of hundreds of meters, usually sufficient for groundwater studies. The improved InSAR data allow us further to reconstruct the SBAS ground deformation time series and transform the measured deformation to head levels using the skeletal storage coefficient and time delay constant inferred from a joint InSAR-well data analysis. The resulting InSAR-head and well-head measurements in the San Luis valley show good agreement with primary confined aquifer pumping activities. This case study demonstrates that high quality InSAR deformation data can be obtained over vegetation-decorrrelated region if processed correctly.