A recent paper by Francis & Hunter provides an interesting discussion about reasons for the recent decline in the Arctic sea-ice extent, based on new satellite observations. One common proposition about sea ice is that it involves a positive feed-back because the ice affects the planetary albedo (how the planet reflects the sunlight back to space before the energy enters the ‘climate system’). Yet, there is more to the story, as the ice acts more-or-less like an insulating lid on top of the sea. There are subtle effects such as the planet losing more heat from the open sea than from ice-covered region (some of this heat is absorbed by the atmosphere, but climates over ice-covered regions are of more continental winter character: dry and cold). The oceanic heat loss depends of course on the sea surface temperature (SST). Open water also is a source of humidity, as opposed to sea-ice (because its cold, not because its dry), but the atmospheric humidity is also influenced by the moisture transport associated with the wind (moisture advection). Francis & Hunter found a positive correlation between lack of ice and the downward long-wave radiation, something they attributed primarily to cloudiness. Hence, clouds play a role, both in terms of influencing the albedo as well as trapping out-going heat. Francis & Hunter suggest that the changes in the long-wave radiation is stronger than the clouds’ modulation of the direct sunlight.
In the past (IPCC TAR), sea-ice models were notorious for difficulties in providing realistic description. Part of the problem, however, may also be the coupling between the ocean and the atmosphere component. I do not know if these aspects are improved in the upcoming IPCC report, however, Francis & Hunter propose several factors that may affect the sea ice edge position, such as oceanic influences, river discharge, non-linear effects, temperature advection and wind, and find that they play different roles in different locations.
The polar regions have white nights during summer with 24-hour sunshine (albeit at an angle) and 24-hour darkness during winter. The albedo feed-back of sea-ice can only be active during summer when there is sunlight to be reflected. The sea-ice extend also varies with seasons, with more open sea during summer and more ice covered area during winter. Furthermore, sea ice is pushed around by surface winds. Thus the effect of any change in the sea-ice extent may differ for different seasons. Less sea-ice during winter may cause more heat loss from open sea – however, at one point when the temperature drops sufficiently, ice will start to form again. Open sea may provide more favourable conditions for polar lows (storms). Cooling in the surface layer and mixing from wind exposure may furthermore affect oceanic currents and perhaps the deep water formation. During summer, the extra shortwave absorbtion is likely to dominate.
Another interesting question is: How does the loss of sea-ice affect the ‘planetary heat engine’, when more heat escapes from high northern latitudes? If heat loss in the Arctic is enhanced as sea-ice retreats and uncovers a sea surface with higher skin temperature? This may be more than compensated for by a reduction in the albedo locally, as well as increased heat transport in the ocean/atmosphere, or enhanced AGW. It is plausible that a ‘cloud-enhanced greenhouse effect’ in the Arctic may cancel part of the oceanic heat loss, and this effect is consistent with the findings of Francis & Hunter. Thus, enhanced cloudiness and associated increase in the downward long-wave radiation may act as an additional positive feed-back for the Arctic surface, in addition to the albedo-effect, whereas the additional heat loss acts like a negative feed-back. Although subtle, Francis & Hunter’s findings may have quite important implications for the planetary heat budget.
Positive feed-back processes taking place on a local scale, may give rise to larger local variations. In addition to the effect of wind-driven sea-ice, such feed-backs may explain the large local temperature anomalies and large natural variations in the Arctic. Another aspect, that sometimes also gets neglected, is the fact that the polar regions represent fairly ‘small’ regions in terms of degrees of freedom. Furthermore, temperature fluctuations tend to be fairly coherent over large parts (there is an anti-correlation in the see-saw structure associated with the NAO, however). Thus, profiles of zonal mean values (the mean values of the measurements taken at a constant latitude ring around the planet) involves comparing long stretched areas near the equator to broad short regions near the poles (if the surface area is compensated for), but due to spatial correlation, this implies that the bands near the equator involve many more degrees of freedom than those near the poles. Such latitudinal profiles of zonal means are analogous to comparing mean values of different sample sizes (a bit like comparing daily values to monthly and annual means). The implication is that events such as the early century warming are not as significant as when the a warming of similar magnitude is seen in the zonal mean profile at lower latitudes.