JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. C2, 3051, doi:10.1029/2001JC000887, 2003

6. Discussion

[43]   It can be seen in Figure 16 that equivalent cross-sectional areas for brine pockets, brine tubes, and mirabilite crystals have comparable magnitudes at -15°C, meaning that no single constituent dominates light scattering at this temperature. As pointed out earlier, tubes contained about 90% of the brine in our samples (see Figure 8), leaving only 10% distributed among pockets. However, Figure 8 also shows that tubes account for only 75% of the total cross-sectional area attributable to brine inclusions. Relative to the tubes, the pockets produce about 3.5 times more equivalent cross-sectional area per unit volume. It is clear from Figure 16 that tubes provide the largest contribution to of any constituent in the ice at -15°C, accounting for roughly half the total in our samples. It is likely that this fraction would be somewhat greater in the higher brine volume sample analyzed by Perovich and Gow [1996] due to the population of larger brine tubes. There were many pockets in the ice whose size was at the lower limits of our 0.01 mm resolution and it seems likely that there were even smaller pockets present in the ice. Experimental evidence indicates that large numbers of such small brine pockets do exist in natural sea ice (H. Eicken, personal communication, 2000). To estimate the potential effects of such pockets on radiative transfer in the ice, we extrapolated the size distribution shown in Figure 5 down to 0.001 mm which produced 200 pockets per mm³ for 0.001 < l < 0.01 mm. Calculations indicate that the cross-sectional area of these very small pockets would contribute less than 0.002 to the total and hence we expect their contribution to be negligible. While the relative importance and size distribution of pockets and tubes reported here should be generally applicable to interior first-year ice, it is not known to what extent these results can be applied to other types of sea ice, e.g. multiyear ice.

[44]   At -15°C, the cross-sectional area for all observed bubbles was an order of magnitude smaller than that for brine pockets, and was only 1% of the total . This is primarily due to the fact that only 5% of all brine inclusions were observed to contain a bubble. Brine tubes generally did contain single bubbles, and these were typically the largest bubbles observed in the ice. Bubbles were rarely seen in brine inclusions when l < 0.03 mm. While the observed number density of bubbles was considerably smaller than that of brine inclusions, it was still forty times larger than previously reported bubble number densities. Because small bubbles should be more readily visible than brine or mirabilite inclusions of equal size due to the strong refractive index contrast between gas and brine, these observations suggest that there may be a minimum size required to form a stable bubble. Bubble nucleation should be successful for bubbles with radii sufficiently large that the surface tension is smaller than the internal tension of the brine. Bubbles with diameter less than 0.001 mm have been observed in small, low-temperature, high-salinity fluid inclusions, where the high salinity of the liquid yields small surface tension [Roedder, 1984].

[45]   Since the total volume occupied by mirabilite crystals is small (only about 3% of the brine volume at -15°C) there was no direct way to determine a size distribution for these crystals, as was done for the brine and gas inclusions. Because the salts precipitate in response to freezing equilibrium, and because their size is generally small compared to the brine inclusions, their presence is tightly tied to the chemical properties of sea ice, and only loosely tied to the inclusion size distributions for tubes, pockets, and gas bubbles. While the observations indicate that mirabilite crystals too large to form in smaller inclusions do sometimes occur in larger brine tubes, most of the precipitated salt crystals do not appear to grow larger than 0.01 mm. In general, we believe the precipitation patterns of crystals to be independent not only of processes occurring at the growth interface, but also of the age and type of the ice. Rather, crystal characteristics are likely to depend primarily on the ice temperature and the chemical composition of the included brine.

[46]   Figure 16 shows three distinct temperature regimes: (1) T < -22.9°C, (2) -22.9° T < -8.2°C, and (3) T -8.2°C. Below -22.9°C, the total is increasingly dominated by the precipitation of hydrohalite crystals. While we do not yet have detailed information on the size and number distributions of these crystals, it is clear from visual observation that their presence causes the ice to become highly scattering at these temperatures [Light, 1995]. The cross-sectional area for hydrohalite shown in Figure 16 was predicted using an effective crystal edge length of 0.01 mm at all temperatures. It is still unclear whether hydrohalite crystals tend to nucleate on inclusion walls or within the brine, whether the crystals become incorporated into the ice as the inclusions shrink, or what conditions promote the formation of slush-filled tubes. Furthermore, it is not clear to what extent the brine may become supersaturated with respect to hydrohalite. Regardless of the exact details, the precipitation of hydrohalite is clearly the major factor determining the total cross-sectional area in cold ice, although the role of scattering by individual crystals, in contrast to scattering created by an ice hydrohalite slush has not been investigated.

[47]   The estimated magnitude of is fairly sensitive to our treatment of the salt crystals. Higher values of for the salts would result from smaller, more numerous crystals since the total mass of the precipitate is constrained by the salinity and temperature of the ice. The assumption that the effective crystal edge length is independent of temperature is likely an oversimplification. In fact, the imagery shows both examples of mirabilite crystal growth and additional crystal nucleation as the ice was cooled below -15°C (Figure 13). By assuming a constant crystal size, we preclude the possibility of crystal growth at temperatures below -15°C. If we were to permit individual crystals to grow as the ice cools, would increase less rapidly than the calculation shown in Figure 16. However, at temperatures below -15°C the change in _m would be small because few additional sulfate ions are available to precipitate, regardless of whether new crystals nucleate or existing crystals grow. If the crystal size at -15°C were required to be 0.01 mm, and crystal sizes were permitted to shrink upon warming, then values of _m would be larger than predicted in Figure 16 for -15°C < T < -8°C. This would presumeably reduce the sensitivity of _m to T in this temperature range. Since _m must decrease to zero at -8.2°C, at some point the sensitivity of _m to T must actually increase, depending on the relative rates of growth and nucleation at temperatures just below that of initial precipition. Furthermore, it is not known whether these hypothesized patterns of crystal growth and nucleation are the same for precipitation as for dissolution, and this may be a source of hysteresis in the temperature dependence of the microstructure.

[48]   At temperatures higher than -8.2°C, (T) is dominated by changes in the geometry of the brine inclusions. Values shown in Figure 16 are based strictly on observed and calculated changes in the size of individual tubes and pockets, without regard to possible merging among neighboring inclusions. A realistic structural-optical model, however, may need to take into account the effects of such merging [Grenfell, 1983]. How inclusions enlarge determines how l changes with temperature, but how inclusions merge affects how number densities change with temperature. Merging depends strongly on the spatial distribution of inclusions. Tightly packed inclusions were observed to begin merging at moderate temperatures (T < -8°C), while more widely spaced inclusions did not begin to merge until T reached -2 or -3°C. Between -15° and -5°C, number density was observed to decrease from 26 mm³ to 20 per mm³ (23%) for the mosaic in Figure 4. In another scene, number density decreased by 65% between -5° and -2°C. Such decreases in number density could reduce the predicted increases in (T). As a test, we examined changes in the cross-sectional area of the three brine inclusions that merged in Figure 14. If the inclusions were to have enlarged without merging, their cross-sectional area would have increased by a factor of 1.5 as they warmed from -8° to -4°C. Computing an equivalent cross-sectional area for the feature which formed at -4°C after the three pockets merged, indicated that was also 1.5 times the total at -8°C. In this particular example, the merged and unmerged realizations of the warmed structure turned out to be identical, suggesting that merging need not necessarily have a large impact on (T) during warming. It was not strictly fortuitous that the two scenarios produced identical results, as the equivalent spherical representation treats single tubes as strings of small spheres. In general, this indicates that merging of vertically stacked inclusions has minimial impact on (T), while lateral merging between inclusions may actually affect (T).

[49]   At intermediate temperatures between -23° and -8°C, the change in with temperature is surprisingly weak due to competing effects from the brine inclusions and mirabilite crystals. Warming causes brine inclusions to enlarge and increase their cross-sectional area while, at the same time, mirabilite crystals dissolve and decrease their cross-sectional area. These two effects tend to offset one another. In this regime, the microstructure and the chemistry are of nearly equal importance. Although changes in (T) are sensitive to the assumed crystal size, the estimates shown in Figure 16 assume that the size of the mirabilite crystals is constant at 0.01 mm. Reducing the average crystal size would result in a greater total surface area, a larger cross-sectional area for the mirabilite, and a steeper change in the total (T). A larger crystal size would reduce the temperature dependence of the total even further. The balance between the competing effects appears to be largely independent of salinity because both effects are predicted to strengthen as the salinity increases. Regardless, in this temperature regime, the mitigating influences of the mirabilite and brine volume yield a (T) with considerably less temperature dependence than either that of colder or warmer ice.

Citation: Light, B., G. A. Maykut, and T. C. Grenfell, Effects of temperature on the microstructure of first-year Arctic sea ice, J. Geophys. Res., 108(C2), 3051, doi:10.1029/2001JC000887, 2003.

Copyright 2003 by the American Geophysical Union.