Interaction Between Sea Ice And Atmospheric Circulation In The Southern Ocean

Climatic Considerations

The annual and interannual variability in the amount of sea ice is one of the most significant factors in the energy balance of the Southern Hemisphere atmosphere and ocean. Continental Antarctica, covering an area of approximately 14 x 106km2, is cold and dry with a high albedo that is virtually unchanging from year to year. The areal extent of sea ice, on the other hand, undergoes average seasonal changes from approximately 3 x 106 km2 in February to 20 x 106 km2 in September, covering at its maximum over 8% of the surface area of the hemisphere. This seasonal variation by more than the land area of Antarctica affects the high-latitude energy balance not only because of the large albedo difference between ice and open ocean, but also because the ice cover is a barrier to energy exchange between the atmosphere and ocean. Latent and sensible heat fluxes over open water can be up to two orders of magnitude greater than over multiyear ice (Weller, 1980) so that it is important to know the fractional coverage by polynyas and open leads within the area enclosed by the maximum extent ice-line. Recent analysis has shown that Southern Hemisphere sea ice is considerably more open than previously thought (Zwally et al., 1979) so that the ocean heat loss is correspondingly higher (by as much as a factor of 6 according to Weller, 1982).

The amount of sea ice present in a particular season can also vary considerably from one year to another. These interannual fluctuations produce large changes in the polar energy balance, which can feed back into sea ice amounts. For example, a delay in the onset of the spring thaw can mean more ice than usual is present in summer at the time of maximum solar radiation and this reduces the amount of oceanic heating which can occur in that season.

There are two other physical processes (in addition to surface-atmosphere heat exchange and the albedo effect) by which sea ice can affect the climate system directly. The first is through thermal inertia of the ice pack, whereby changes in the temperature of the overlying atmosphere are delayed due to the release of latent heat as the ice freezes in autumn and the uptake of heat during spring melting. The other effect is through the northward movement of sea ice (see Keys, this volume, for a more detailed discussion of icebergs). This movement produces a nett equatorward transport of fresh water (affecting the salinity of the oceans) and a nett poleward transport of heat.

The presence of sea ice also affects the poleward heat flux indirectly through modifying the local atmospheric stability and the large-scale north-south temper ature gradient; these factors influence the generation of high latitude depressions and their subsequent motion. A complicating factor in understanding the sea ice-atmosphere interactions is that wind stresses generated over the ice pack by travelling storms can affect the position of the maximum ice extent line. Observational studies have so far not resolved this cause and effect problem satisfactorily. We will consider these research efforts further in a later section, and turn our attention first to the seasonal growth and decay of the Antarctic sea ice zone.

Sea Ice Climatology

Present evidence from relatively limited satellite data suggests that at any given longitude there can be large variations in sea ice amount from year to year. However, decreases at one longitude are frequently compensated by increases at another (Zwally et al., 1979) so that interannual variations in total maximum ice extent may not be very great. The seasonal variation in the zonally averaged position of the boundary between Antarctic pack-ice and the open ocean, as shown previously in Fig. 2.11, is therefore fairly well established. Minimum ice extent occurs in February and March, and maximum extent in September and October. The growth and northward advance of the sea ice occurs rather more slowly than the subsequent decay.

Monthly changes in sea ice extent show considerable regional variability in any given year, as well as large differences between years. This variability is greatest in the longitudes of the Antarctic coastal embayments. The regional pattern of ice coverage for three particular dates during 1974 is illustrated in Fig. 2.15. Rapid ice growth has occurred preferentially in the Ross and Weddell Seas between February and June. From June to September, however, there was little further expansion of the ice-ocean boundary at these longitudes: ice tended to be advected downstream by the Antarctic Circumpolar Current, resulting in a more symmetrical distribution about the Pole. Cavalieri and Parkinson (1981) highlighted these pattern changes by Fourier analysis of the latitude of the ice edge. The early-season pattern of February is primarily wave 2, and reflects the shape of the Antarctic continent; the rapid ice-growth phase in the major embayments shows up as a large wave 3 component, and the maximum ice-extent phase when there is greatest symmetry about the Pole shows wave number 1 to be dominant at that time. Such seasonal changes in the high latitude forcing may have important consequences for the standing long-wave patterns in the atmosphere, and hence seasonal forecasting.

Relationships Between Sea Ice and Circulation Features

There is a problem of causality in deciphering relationships between amounts of sea ice and atmospheric circulation features: the ice edge, the position of sea surface temperature gradients, tracks of developing vortices and atmospheric pressure patterns all seem to be correlated but it is not clear whether the physical chain should begin with the atmosphere or the ocean. At the sea-ice boundary,

Fig. 2.15. Antarctic sea ice extent for three selected times during 1974, showing examples of dominant wave patterns: 3 February (wave 2 dominant), 15 June (wave 3), and 28 September (wave 1) (from Cavalieri and Parkinson, 1981).

latitudinal temperature gradients are intensified, which may therefore be expected to augment cyclone development downstream (and also equatorward). However, a high frequency of depression centres in a particular location near the ice edge can produce an ice advance to the west of the low centre (and likewise ice to the east may be advected southwards, or at least restrained from advancing). Strong winds around cyclone centres also serve to break up thin ice.

Streten and Pike (1980) examined various characteristics of sea ice extent and the associated atmospheric circulation over the period 1972-77. These authors looked for temporal relationships between ice extent and the strength of the westerlies to the north. Averaged over the whole period, the zonally-averaged westerlies exhibit semi-annual oscillations in strength (Fig. 2.4), and thus display no relation to sea ice extent (Fig. 2.11) over the normal seasonal cycle. There is also no clear correlation between zonally-averaged ice extent and the strength of the westerlies in the preceding or succeeding months. However, on examining interannual variability of the westerlies, they found a greater variability during the time of maximum ice extent: i.e., the westerlies were most variable in October-November, and most consistent year-to-year in February.

Streten and Pike (1980) also showed that the shape of the ice boundary seemed to be related to the location of persistent pressure minima in the Circumpolar

Trough. The mean ice edge in October (near maximum extent) followed the basic continental shape, but was perturbed so that it was located at lower latitudes close to, or just westward of, those longitudes having a high frequency of low pressure centres. However, this association broke down in Weddell Sea longitudes, which led Streten and Pike to argue that regional (i.e., restricted area) rather than hemispheric studies were needed when relating anomalies of ice extent to atmospheric circulation features. Regional relationships are particularly appropriate when looking at interannual fluctuations because of the previously mentioned tendency for an ice anomaly at one longitude to be compensated by one of opposite sign downstream.

In one such regional study by Streten and Pike (1980), interannual variations in the strength of the winter westerlies north of the Ross Sea (160°E-150°W) were found to correlate positively with the amount of sea ice in the following spring. Stronger westerlies in the winter of 1975 were followed by more ice in spring; weaker westerlies in 1977 were followed by less ice. The authors' explanation of this correlation was that weak winds were indicative of a weaker cyclonic gyre in the Ross embayment, and therefore less northward transport of ice. Once again, however, we must emphasize that studies which postulate causative relationships between aspects of polar climate based on short-period correlations must be treated with caution.

Many studies have attempted to relate variations in high latitude cyclogenesis to fluctuations in sea ice extent (Schwerdtfeger and Kachelhoffer, 1973; Streten and Pike, 1980; Carleton, 1981a, b; Cavalieri and Parkinson, 1981). There is no clear relation between the seasonal cycles of mean sea ice extent and cyclonic activity, because of the complicating influence of the semi-annual oscillation (Fig. 2.11) and strong ocean-atmosphere coupling at the oceanic Polar Front (Fig. 2.10). Greater success has been achieved in looking at fluctuations on very short (a few days) or very long (interannual) timescales, where again it is necessary to consider longitudinal variations.

Examining 3-day averages during the time of maximum ice growth in the Ross and Weddell Seas, Cavalieri and Parkinson (1981) found areas of rapid ice growth lying to the west of intense cyclone centres. They attributed this to both equator-ward ice transport and in situ freezing as a result of advection of cold air from more southerly latitudes. During the months of maximum ice decay, the association between low centres and the regions of ice retreat was not as marked.

Carleton (1981b) compared circulation differences in 1974 and 1976, the two most highly contrasting years in the 1973-77 period in terms of cyclone activity. The greatest total cyclone activity (except in very high latitudes) occurred in 1974, which was also the year with the greatest ice extent. However, the lowest cyclone activity occurred in 1976, an intermediate year as far as ice extent was concerned. Carleton also presented evidence that longitudinal variations in ice extent were due to variations in high latitude cyclone activity, arguing on the same lines as Cavalieri and Parkinson (1981) and Streten and Pike (1980) above. However, a re-analysis of Carleton's data (Fig. 2.16) can be used to argue the reverse relationship: namely, that differences in ice extent between 1974 and 1976 were responsible for the observed longitudinal differences in cyclone frequency.

The upper curve of Fig. 2.16 shows the latitude difference in the sea ice margin between 1976 and 1974. Negative values indicate those longitudes where there was less sea ice in 1976. The lower curve shows the difference in vortex frequencies between 1976 and 1974, negative values indicating those longitudes where the local vortex frequency in 1976 (expressed as a percentage of the 1976 hemispheric mean) was less than the 1974 local vortex frequency. The two curves of Fig. 2.16 look very similar, apart from a phase shift. The maximum correlation (r = +0.70) occurs for a 50 degree eastward displacement of the ice extent curve. Again, there is the question of which is cause and effect but, in this case, it seems rather unlikely that a cyclone (see Fig. 2.16 for typical scale) could affect ice growth 50 degrees westward of the low centre. It is much more probable that an expanded ice line increases the latitudinal sea-surface temperature gradient locally, thus generating more incipient disturbances that reach their maximum intensity some distance downstream.

Budd (1982) and Carleton (1981b) have also found that, on an interannual timescale, the latitude of maximum cyclone frequency shifts north or south according to the position of the sea ice margin. Comparing 1975 and 1977, for example, the sea ice was up to 5 degrees further south in 1977 in the region south of eastern Australia, when there was a corresponding southward shift in the distribution of cyclone tracks (Budd, 1982).

Fig. 2.16. Longitudinal variation of (a) 1976-74 difference in latitude of sea ice margin, and (b) 1976-74 difference in cyclone percentage frequency, over winter June-September period (adapted from Figs 2 and 5 of Carleton, 1981b). The cyclone frequency data have been smoothed before plotting.
0 0

Post a comment