The ALOS-2 data were provided by Japan aerospace exploration agency (JAXA) under the 4th research announcement program (principal investigator Torbjørn Eltoft, ALOS PI-number 1199). The Radarsat-2 data were provided under the norwegian-canadian Radarsat agreement 2015. TerraSAR-X data were provided by DLR through TerraSAR-X AO OCE2582. The study was supported by norges forskningsr˚ad (NFR) through the project: centre for integrated remote sensing and forecasting for arctic operations (CIRFA) (NFR project number 237906).
Backscatter reduction due to damping of gravity waves in frazil or grease ice slicks
It is well known that slicks of frazil or grease ice can cause a low radar backscat-ter compared to open wabackscat-ter roughened by wind or many types of consolidated sea ice. A detailed understanding of this reduction of the radar backscatter is important for characterisation, classification and parameter retrieval of sea ice from radar remote sensing data. It is also critical for interpreting oil spills in ice infested waters, since these also may cause a low radar backscatter and may be misinterpreted as ice slicks (or vice versa). In this chapter, this is examined in more detail. The general approach mimics the one used for slicks of oil presented in, for instance, Alpers & H¨uhnerfuss (1989) and Gade et al. (1998).
The chapter begins with a brief motivation and background on the ice types of consideration. Wave dispersion is then discussed, specifically for open water and for slicks of ice. The model by Keller (1998) (referred to as the Keller model) is considered for the latter, which treats an ice slick as a viscous layer and provide the corresponding dispersion relation for gravity waves. The model is solved numerically and compared to the well known dispersion relation for open water. Next, the action balance equation is considered for linking the dispersion relation to the wave spectrum. The source terms of the action balance equation are examined. The term for wind input and the one for viscous dissipation are assumed dominant and examined in detail. Lastly, the wave spectrum is coupled with the SPM in order to relate the dispersion relation to a scattering band ratio.
The ratio predicts the reduction of the radar cross section (RCS) due to an ice 171
slick, relative to some reference wavelength. Choosing the reference wavelength to correspond to L-band radar waves, the ratio is used to interpret observations from chapter 6.
The analysis concludes that under certain conditions, ice slicks have strong impact on the spectral behaviour of the radar backscatter. According to the numerical solution of the Keller model, the damping due to ice slicks is generally very strong at wavelengths relevant to radar remote sensing. Consequently, the wind speed presumably needs to be high in order to cause significant roughness of the ice slick surface and thus detectable amounts of Bragg scattering.
While the approach presented in this study is essentially the same as the one used for oil slicks in for example Gade et al. (1998), this study provide the first attempt (to my knowledge) to quantitatively describe the spectral backscatter characteristics of frazil or grease ice slicks. It is moreover different from Gade et al. (1998) in that a band ratio (ratio of the RCS from a slick at one wavelength relative to another) is here considered instead of a damping ratio (ratio of the RCS from a slick relative to that of open water) as considered in Gade et al.
(1998) among others.
7.1 Motivation and objectives
Frazil and grease ice are under turbulent conditions the earliest ice types in the ice formation process (see section 3.3.1 for further details). Their characteristics and abundances are therefore important to monitor in order to better understand and model sea ice growth and ocean-atmosphere interactions. As an example, Smedsrud & Martin (2015) show that by incorporating grease ice into a basin-scale sea ice-ocean model, the closing of sea ice leads is delayed and the heat loss from the ocean to the atmosphere is significantly increased in fall and winter.
With increasing human activity in the Arctic Ocean in terms of fishing, ship-ping and petroleum exploration or production, slicks of mineral oil at the sea surface are likely to become much more frequent. Radar remote sensing has proven to be extremely useful in detecting such slicks, since they typically re-duces the radar backscatter significantly compared to open water (Brekke &
Solberg 2005). It is well known that slicks of frazil or grease ice also can cause a reduction in backscatter. In figure 7.1, an example of a SAR image containing streaks of grease ice is shown as an example (dark stripes perpendicular to the highlighted transect). In ice infested waters, misclassification between slicks of oil and ice is therefore a potential problem, thus a detailed understanding of their characteristics is of great interest.
In this chapter, the reduction of the radar backscatter due to slicks of frazil or grease ice is examined in more detail. In particular, the damping characteristics as a function of radar wavelength is considered. This will also yield insights to
Figure 7.1: Example of grease ice in a SAR image. These can be seen as dark stripes perpendicular to the highlighted transect in the middle of the image.
The slicks were validated from co-incident optical data. The image cover the Terra Nova Bay Polynya in Antarctica and is acquired by ALOS PALSAR (L-band) on 10 September in 2009. The image is provided as a processed joint photographic experts group (JPEG) file by Wolfgang Dierking through the ESA project ALO.3545.
observations made for open leads mentioned in the chapter 6, which is discussed at the end of this chapter (section 7.4.5).