Polarimetric Decomposition Over Glacier Ice Using Long-wavelength Airborne PolSAR

Abstract

In recent years there has been increased interest in using SAR to study and monitor glaciers and ice sheets for glaciological and climate change research. However, due to the medium's complexity, SAR backscattering from ice remains poorly understood, including the relative importance of scattering from surface and volume layers, and dependencies on frequency and glacier zone. Extreme weather conditions can result in quickly changing surface conditions influencing backscatter signatures, while leaving the underlying volume of interest unchanged. Surface and volume components must thus be separated in order to infer information regarding the properties of the ice volume. This paper describes a three-component scattering model to decompose polarimetric synthetic aperture radar (PolSAR) images of glacier ice. Total backscatter is modeled as the incoherent summation of surface, volume and sastrugi (wind-induced feature) components. The proposed model adapts and extends the Freeman and Durden decomposition for an ice volume scenario in which the volume is a dielectric medium. Forms of the model for both random and oriented volumes are considered and a new oriented sastrugi component is introduced which is able to explain backscatter behavior between different winter scenes. Validation is performed with airborne PolSAR data at L- and P-band collected using DLR's E-SAR system over the Austfonna ice cap in Svalbard, Norway as part of the ICESAR campaign.