Sea Ice (Continuation)

Accurate high resolution and near-real time determination of sea ice conditions is necessary for safe navigation in sea ice infested waters and refinements of climate models, but it has not been achieved to date. In particular, distinguishing ice from open water in low sea ice concentration areas is challenging and the result may depend on the frequency band of the sensor used for the retrieval. Here an algorithm for ice/water classification of C- and L-band dual polarization synthetic aperture radar (SAR) data is presented. The algorithm is based on backscatter intensities in co- and cross-polarization and autocorrelation as a texture feature and separates sea ice, open water and thin ice/calm water. The mapping between image features and ice/water classification is made with a neural network. The area of interest is Fram Strait located between Greenland and Svalbard, where most of the ice is exported from the Arctic.

A comparison of the two different frequencies is made in order to investigate the potential to improve classification results and to increase the temporal resolution of highly dynamic areas with multi frequency data.  

Accurate ice/water maps for both frequencies with a resolution of about 200 m are produced by the algorithm and the results of the two frequencies agree in 86% of the cases. The major deviation is introduced by the different sensitivity for thin ice/calm water, which has been allocated a separate class. Thin ice is discriminated more easily in L-band due to its longer penetration depth and reduced influence of small scale surface roughness. Differences are also pronounced in the marginal ice zone, where sea ice motion occur between the acquisitions of the different frequencies. Furthermore a larger contrast between open water and sea ice in the cross-polarization channel of C-band facilitates discrimination compared to L-band.

C-band reliably reproduces the outline of the ice edge, while L-band has its strengths in detecting thin ice/calm water areas within the ice pack.

The classification results show good agreement with lower resolution ice/water maps derived from met.no ice-charts and radiometer data from AMSR-2. Variations are found in the marginal ice zone where generalization of the ice charts and lower accuracy of ice concentration from radiometer data introduce deviations. Differences are also found within the ice pack where the presented algorithm discriminates areas of thin ice while these features might be omitted in the creation of ice/water maps by thresholding.

Usage of dual frequency SAR data could thus be beneficial for improving spatial and temporal resolution of ice cover information for ship navigation and modelling of ice extent and ocean-atmosphere interactions. 

Melt ponds are a common feature on Arctic sea ice. They are linked to the sea ice surface albedo and transmittance of energy to the ocean from the atmosphere and thus constitute an important process to parameterize in Arctic climate models and simulations. This paper presents a first attempt to retrieve the melt pond fraction from compact polarization (CP) SAR imagery, which has wider swath and shorter revisit time than the quad-polarization systems, e.g. RADARSAT-2. The co-polarization (co-pol) ratio has been verified to provide estimates of melt pond fractions. However, it is a challenge to link CP parameters and the co-pol ratio. The theoretical possibility of making this linkage with the CP parameter C₂₂/C₁₁(the ratio between the elements of the covariance matrix of CP SAR) for melt pond detection and monitoring is presented with the tilted-Bragg scattering model for the ocean surface [Valenzuela, 1978]. Our empirical transformed formulation denoted the ‘compact polarization and quad-pol’ (‘CPQD’) model, is proposed. Our methodology is based on the existing VV/HH melt pond fraction retrieval method by Scharien et al. [2014]. The tilted-Bragg scattering model suggests the theoretical potential of C₂₂/C₁₁ from CP mode data for melt pond observations. Thus, the linkage between C₂₂/C₁₁,from CP mode data, and VV/HH, from quad-pol data, can be established with our collocated number of quad-pol observations of open ocean water. We used two data sets. Our CPQD is firstly based on 2062 RS-2 quad-pol SAR images, collocated with in situ measurements. Secondly, we compared the retrieved melt pond fraction with CP parameters simulated from quad-pol SAR data with results retrieved from the co-pol ratio from quad-pol SAR observations acquired during the Arctic-Ice Covered Ecosystem in a Rapidly Changing Environment (Arctic-Ice) field project conducted from May to June 2012.  The results are shown to be comparable for observed melt pond measurements in spatial and temporal distributions.  Thus, the utility of CP mode SAR for melt pond fraction estimation on first year level ice is established. 

References:

Li, H., Perrie, W., Li, Q., & Hou, Y. (2017) Estimation of melt pond fractions on first year sea ice using compact polarization SAR. Journal of Geophysical Research: Oceans, 122, 8145–8166. https://doi.org/10.1002/2017JC013248

Scharien, R. K., Landy, J., and Barber, D. G. (2014)  First-year sea ice melt pond fraction estimation from dual polarisation C-band SAR – Part 1: In situ observations, The Cryosphere, 8, 2147–2162, doi:10.5194/tc-8-2147-2014.

Valenzuela,G. R. (1978) Theories for the interaction of electromagnetic and ocean waves- A review, Boundary-Layer Meteorol., vol. 13, no. 1–4, pp. 61–85