PolSAR-Ap: Exploitation of Fully Polarimetric Data for Glacier and Sea Ice Characterization
Parrella, Giuseppe1; Marino, A.2; Hajnsek, I.3
1German Aerospace Center (DLR), GERMANY; 2ETH Zurich, SWITZERLAND; 3ETH Zurich / DLR, SWITZERLAND

Monitoring glaciers and sea-ice is a primary task to understand and quantify climate change. At the same time, it is a major topic in remote sensing due to the difficulty of performing frequent in-situ expeditions [1, 2]. In this sense, microwave sensors like Synthetic Aperture Radars (SARs) have a great potential in cryospheric observations since they can operate in absence of solar illumination (i.e. during Polar nights) and with almost any weather conditions. Moreover, long-wavelength (e.g. L- band) SAR systems are capable to penetrate several meters deep into ice bodies. Hence, they are sensitive to the ice surface characteristics as well as to sub-surface structures (including in some cases the lower ice-water interface of sea-ice). Unfortunately, the description of the backscattering behavior of glacier and sea ice is particularly challenging. For this reason, many scientists moved toward systems able to increase the amount of information acquired. In this context, polarimetry plays a key role, because it is able to enhance the discrimination capability of the observed target, solving many ambiguities revealed in single polarization images [3].

The polarimetric response of a target is highly sensitive to its orientation along the Line of Sight LoS (i.e. the direction of propagation of the wave). Therefore, the same target with different orientations may give a substantially different response in SAR images. For the case of sea-ice, this in general will lead to wrong classification (since a different backscattering will be observed). By exploiting fully polarimetric data, it is possible to estimate the orientation of the target and remove its effects [3, 4]. Several methodologies were proposed for this correction; here the one based on the circular polarizations is exploited since it can be applied to distributed targets (coherency matrix) [4]. If the Pauli basis is exploited, any orientation along the LoS can be mathematically represented by a rotation matrix (along the first axis HH+VV). The circular co-polarization channel can be derived starting from the scattering matrix in linear basis. The phase of the coherence between the circular Right-Right and Left-Left channels keeps information regarding the orientation angle. The reason why some sea ice regions show orientation is not completely clear and can be due to orientation of the ice chunks, wind effects on the ice surface (e.g. sastrugi) or particles preferential orientations. Nevertheless, the correction allows solving this ambiguity and provides additional insights in the observed target.

Concerning glacier ice, this study addresses the modeling of possible scattering contributions to explain L- and P-band polarimetric SAR signatures of a subpolar glacier. The main focus is on the development of an innovative volume scattering model in order to simulate the backscatter contribution from different kinds of inclusions typically present in glacier ice (e.g. ice pipes and lenses, air bubbles, oriented crystals fabrics, etc.) [5]. The developed model allows the shape of the inclusions to vary, ranging from dipoles to spheres, as well as their main orientation and degree of randomness in a three dimensional space. For the case of oriented particles or propagation through an anisotropic medium, such as polar firn [6], differential propagation effects (i.e. different propagation velocities and losses) along the different polarizations have to be accounted [7]. The proposed model is finally completed by including the incidence angle dependency. Additional scattering components attributable to surface mechanisms (from the glacier surface) or to multiple interactions between the inclusions could also be included. Nevertheless they are believed to be much weaker and their characteristic polarimetric signatures do not contribute significantly to explain the ones observed in the data.

The dataset exploited for the sea-ice study was acquired during the ICESAR campaign in 2007 by the E-SAR airborne system of DLR (German Aerospace Center). The sea ice acquisitions were carried out in Svalbard over three different locations: Fram Strait, Storfjord and Barents Sea. In this analysis only L-band acquisitions are used, since they are the only one presenting quad-polarimetric data. After a first analysis of single channel images, three typologies of sea-ice seem present in the Storfjord site. However, the Pauli RGB image after the removal of the orientation angle reveals that there are only two typologies of ice, and two of them appear different for orientation effects. For the glacier-ice investigation, the available dataset includes fully polarimetric L- and P-band SAR data, acquired during the ICESAR campaign by the E-SAR sensor as well. The selected test site is the Austfonna ice-cap, located on the Nordaustlandet island of Svalbard.

[1] S. Sandven, Johannessen, O.M., and K Kloster, Sea Ice Monitoring by Remote Sensing, John Wiley & Sons, 2006.
[2] C.R. Jackson and J.R. Apel, Synthetic Aperture Radar Marine User's Manual: National Oceanic and Atmospheric Administration (NOAA), NOAA, 2004.
[3] S. R. Cloude, Polarisation: Applications in Remote Sensing, Oxford University Press, 2009.
[4] J.-S. Lee, D.L. Schuler, and T.L. Ainsworth, ''Polarimetric SAR data compensation for terrain azimuth slope variation,'' IEEE Transactions on Geoscience and Remote Sensing, vol. 38, 2000.
[5] W.S.B. Paterson, ''The Physics of Glaciers'', Butterworth Heinemann, Oxford, U.K., 1994.
[6] R.B. Alley,''Texture of Polar Firn for Remote Sensing'', Annals of Glaciology, vol. 9, pp. 1-4, 1987.
[7] S.R. Cloude, K.P. Papathanassiou, and W.M. Boerner, ''The Remote Sensing of Oriented Volume Scattering Using Polarimetric Radar Interferometry'', in Proc. of ISAP, pp. 549-552, Fukuoka, Japan, 2000.