Arctic sea ice microstructure observations relevant to microwave scattering

ABSTRACT. Sea ice microstructure characteristics relevant to ice microwave scattering were studied during SIMMS'91 field experiment in Resolute Bay in May/June 1991. Thin sections of the top 300 mm layer of first - year and multi - year ice (from hummocks and melt ponds) were prepared and examined. Analysis is based mostly on qualitative observations, although statistics on bubble dimensions and geometry were obtained from digital analysis of thin section photographs. First - year ice featured mostly an oriented columnar grain structure. Both spherical and needle - shaped brine pockets were observed. In multi - year ice, a variety of grain structures and inclusion patterns were observed from the same floe. Hummock and melt pond ice are different in terms of grain structure and air bubble contents. Air bubbles in hummock ice are highly random and interconnected, especially near the surface. At lower depths, they retain simpler shapes and become oriented parallel to the dominant grain growth direction. In melt pond ice, two types of air inclusions whose typical dimensions differ by an order of magnitude were observed. Significant spatial variability of multi - year ice microstructure within a single floe is demonstrated.

Key words: sea ice, ice microstructure, ice microwave scattering, SAR ice signatures, air bubbles in sea ice

INTRODUCTION

Sea ice is a heterogeneous mixture of pure ice crystals, gas (air) bubbles and brine pockets in either liquid or solid forms. Several studies on microstructure of laboratory - made saline ice have been conducted to understand its formation and growth processes (Weeks and Assur, 1963; Weeks and Lofgren, 1967; Cox and Weeks, 1975). Only a few studies, however, have been conducted to relate the microstructure of natural first - year ice to the growth rate and the meteorological and oceanographical conditions (Nakawo and Sinha, 1981, 1984). Most studies focus on the fabric of crystalline structure in relation to the ice age and the region of origin.

Peyton (1966) studied the structure of columnar ice 0.5 m or more beneath the surface of ice along the Alaskan coast and found that it was characterized by a strongly c - axis orientation in the horizontal plane. Strongly preferred c - axis orientation was also noticed by Sinha (1977) to occur within 20 mm below the surface in Strathcona Sound, Baffin Island, Canada. Photographsof highly oriented grains and intragranular salt inclusions are presented in the same reference. Weeks and Gow (1978) noted oriented columnar ice in the northern part of Alaska in land - fast ice and concluded that the alignment was related to the dominant orientation of water current under the ice cover. Sinha (1984) found the same structure in first - year and multi - year ice in Crozier Channel in the Beaufort Sea region. About 70% of the first - year ice adjacent to the east coast in Mould Bay (Canadian Western Arctic) was found to be vertically - oriented frazil ice with the c - axis of crystals randomly oriented in the horizontal plane (Sinha, 1986). This was attributed to the strong westerly wind during the formation of the ice cover. Tucker et al., (1987) reported that about 75% of first - year ice in Fram Strait (between the East Greenland coast and Svalbard) consisted of columnar grains and the remaining 25% consisted of granular ice. Eicken and Lange (1989) studied three wind - controlled ice formation regimes in the Weddell Sea, where columnar ice predominated in undeformed ice (formed at low wind speeds), while granular ice predominated in heavily ridged and rafted ice (formed under high wind speeds). The lack of ice crystal orientation in the Sea of Bothnia as compared to ice in the Arctic Ocean was attributed by Weeks et al., (1990) to the weaker or more variable sea current. Few studies have been carried out on characteristics of inclusions in ice (Poe et al., 1974; Bjerkelund et al., 1985) -- a subject of direct relevance to ice microwave remote sensing.

Sea ice monitoring for operational and scientific applications has progressed through the use of advanced remote sensing tools, particularly in the microwave band. Both active and passive sensors have been used. The prime sensor within the category of active sensors is the Synthetic Aperture Radar (SAR) (Ulaby et al., 1986). Interpretation of sea ice SAR images depends on understanding the radar backscattering mechanism. Extensive data on radar backscattering from ice at different frequencies are presented in the literature (e.g., Onstott, 1992). Backscattering is influenced by three factors: the ice surface roughness, the average dielectric constant of ice, and the microstructural properties of ice in terms of crystal structure and, more importantly, inclusions in ice. The third factor is the subject of this study.

Data on ice microstructural properties relevant to ice microwave remote sensing have become more available recently. Tucker et al., (1991) presented results from surface - based active and passive microwave measurements, made in conjunction with physical and structural properties of ice. They showed that microwave signatures are affected by brine (liquid content) as well as gaseous void distribution. Eicken et al., (1990) demonstrated the feasibility of quantitative texture analysis of ice microstructure images. Samples of different ice grain textures were distinguished based on the slope of the histogram of a texture parameter. Perovich and Gow (1991) calculated the correlation function between a template representing the typical appearance of brine pockets and the images of actual brine pockets in first - year ice. Statistics derived from the correlation function were ascribed to geometrical characteristics of brine pockets.

From the operational and scientific viewpoints, first - year and multi - year ice are among the most important ice types. For these types, the dominant inclusions are brine pockets and air bubbles respectively. Air bubbles exist at depths above the water level. The saline - free nature of the upper layer of multi - year ice allows more microwave energy to penetrate and interact with air bubbles. This interaction invokes volume scattering, which, together with surface scattering, forms the total backscatter received by a microwave remote sensor. Both experimental studies and microwave scattering model results (Gogineni et al., 1990) show that volume scattering from first - year ice is not significant, particularly at small incidence angles of the incoming radiation. Modeling volume scattering in multi - year ice depends on information on bubble characteristics. At present, this information is scarce in the literature. Attempts to measure bubble dimensions are presented in Poe et al., (1974) and Bjerkelund et al., (1985). The latter reference shows that 83% of the spherical bubbles in a horizontal section from completely desalinated second - year ice have a diameter less than 1 mm. More detailed study is required to explore the variability of bubble characteristics in relation to the topography of ice (i.e., hummock and melt pond), as well as the grain structures that commonly exist in each type. A requirement for in situ data to characterize inclusions that can be incorporated in microwave scattering models has been identified in the literature (e.g., Winebrenner et al., 1989).

The present study was carried out to characterize air bubbles and brine pockets in relation to grain structures in natural sea ice in the Arctic. The study focused on differences between crystalline structure in hummocks and melt ponds. Results are based mostly on qualitative analysis, although limited quantitative information on air bubbles in multi - year ice is obtained from digital analysis of thin sections. The main objective of the study is to further define the parameters contributing to the microwave volume scattering. Hence, a better capability to interpret radar ice signatures observed in SAR images can be developed.

METHODS

The present study was the second in a series conducted under the Sea Ice Monitoring and Modeling Site (SIMMS) program. It was designed to collect baseline geophysical data from snow - covered sea ice and relate them to ice signatures from SAR (Barber, 1991).

The field study was conducted during 21 May to 4 June, 1991 in Resolute Passage off Barrow Strait, Northwest Territories, Canada (Fig. 1). This period signals the early melt season when the snow cover becomes wet but is not yet flooded (Livingstone et al., 1987). In addition to ice microstructure data, in situ data such as thickness, surface roughness distribution, temperature and salinity profiles were collected on snow and ice. Some of these data are presented in other papers in this issue. Full documentation of the data are included in Reddan et al., 1991.