InSAR Phase Unwrapping Error Correction for Rapid Repeat Wetland Water-Level Change Measurements

Abstract:

As part of NASA’s Delta-X project to study the processes controlling soil accretion and retention in deltas, the Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) is scheduled to perform radar surveys over Louisiana’s coastal wetlands to provide water level change measurements within areas with emergent vegetation. Interferometric synthetic aperture radar (InSAR) will be used to measure spatially-extensive water-level change, including in areas difficult to reach and where no tide-station can be installed. The measurements will help constrain hydrology and sediment transport models. To be able of accurately measure water-level changes within the wetlands using InSAR, it is necessary to correctly phase-unwrap the interferograms, which is particularly challenging in coastal wetlands because 1) the landscape is naturally separated into islands, 2) tidal dynamics often drive large and rapid changes in water level and 3) riverine flooding can overtop vegetation. Areas of temporal decorrelation along with high phase gradients are commonly found in the interferograms, and can introduce the incorrect number of phase cycles (2pi) when computing phase unwrapping. In this study, we present a combined algorithm to correct phase unwrapping errors in InSAR for water-level measurements incorporating spatial bridging between islands and phase closure between interferograms. The dataset used consists of 12 UAVSAR L-band SAR images covering the regions of the Wax Lake and Atchafalaya River deltas. The acquisitions were taken on October 16-17, 2016, with images acquired approximately 30 minutes apart on each day. Our method consists of implementing an iteratively combined phase unwrap correction in space and time domain, correcting first for interferograms with shortest temporal baseline and increasing the interferometric network by one order in each iteration. Time-series analysis of the corrected interferometric stack is performed to retrieve the dynamics of water-level change throughout the wetlands. Preliminary time-series results show water-level decrease of up to 6cm during low tide and increase of up to 28 cm during high tide, displaying the largest change in the regions near the deltas and closely matching the water-level change observed from tide-gauge stations located nearby.