Polarimetric Change Detection for Wetlands

Abstract

An exciting development in wetland mapping and monitoring is the use of SAR polarimetry which uses both magnitude and phase of the backscattered radar signal for information extraction. This approach allows for the accurate delineation of flooded vegetation due to the double bounce scattering mechanism which the phase helps to identify unambiguously. Repeat pass polarimetric data is then used to monitor the temporal change in flooded vegetation. This information is useful for a variety of applications in wetland mapping and monitoring. This paper will present a novel Curvelet-based technique for the enhancement of polarimetric decomposition channels as well as temporal differences in these channels. Starting with the Freeman-Durden and the Cloude-Pottier polarimetric decomposition of Radarsat-2 data the Curvelet image enhancement and the Curvelet change detection are applied. The results are very promising although a validation by comparison with ground truth data still has to be done.