Application of a three-component scattering model over snow-covered first-year sea ice using polarimetric C-band SAR data

Abstract

In this study we examine the utility of a three-component scattering model to quantify the sensitivity of radar incidence angle over snow-covered landfast first-year sea ice (FYI) during the late winter season. This three-component scattering model is based on (1) surface scattering contributed from the snow-covered FYI (smooth-ice (SI), rough-ice (RI), and deformed-ice (DI) types); (2) volume scattering contributed from snow layers which consist of enlarged snow grains, elevated brine volume, and preferential orientation of snow grains relative to radar look direction, as well as the underlying sea ice; and (3) double-bounce scattering contributed from ice ridges and ice fragments. This study uses RADARSAT-2 C-band polarimetric synthetic aperture radar (POLSAR) data acquired on 15 and 18 May 2009 for Hudson Bay, near Churchill, during late winter with surface air temperatures ≤−8°C at two different incidence angles (29° and 39°). The three-component scattering model is used to discriminate between snow-covered smooth, rough, and deformed FYI. The model shows enhanced discrimination at an incidence angle of 29°, compared with an incidence angle of 39°. The model is then used to quantify the sensitivity of radar incidence angle to each of the three scattering contributors. The results show that the relative fraction of surface scattering dominates for all three FYI types (SI ≈ 77.3%; RI ≈ 66.0%; and DI ≈ 61.1%) at 29° and decreases with increasing incidence angle and surface roughness. Volume scattering is found to be the second dominant mechanism (SI ≈ 19.1%, RI ≈ 32.2%, and DI ≈ 37.4% at 29° and SI ≈ 28.3%, RI ≈ 41.0%, and DI ≈ 49.5% at 39°) over snow-covered FYI and it increases with incidence angle and surface roughness. The double-bounce scattering contribution is low for all FYI types at both incidence angles.

Acknowledgements

Special thanks to Dr T. Geldsetzer for his technical support that helped improve the manuscript. The authors are grateful to the staff of the Churchill Northern Studies Centre (CNSC): Dr Randy Scharien, Mark Fuller, Melissa Peters, Peter Bezeau, Adrienne Cunnings, J.V. Gill, and Johnson Chan, for their logistical support and field assistance. The Science Operational Application Research (SOAR) programme of the Canadian Space Agency is acknowledged for providing the RADARSAT-2 data. PCI Geomatica and the Canadian Ice Service of Environment Canada are recognized for their technical and software support. RADARSAT-2 data and products © 2009 MacDonald, Dettwiler and Associates Ltd. – All rights reserved.

Funding

Dr J.J. Yackel is funded by an NSERC Discovery Grant, the Canadian Space Agency SOAR-E for RADARSAT-2 data, and a research grant provided by CNSC, ArcticNET, and NSERC Networks of Centres of Excellence.