Reports focus on the possibility a record minimum for Arctic sea ice in September, but a major loss during the early summer months is climatologically more important
"Well, it's not really good timing to write about global warming when the summer feels cold and rainy," a journalist told me last week. Hence, at least here in Germany, there hasn't been much reporting about the recent evolution of Arctic sea ice – despite the fact that Arctic sea ice extent in July, for example, was the lowest ever recorded for that month throughout the entire satellite record. Sea-ice extent in August was also extremely low, second only to August 2007 (Fig. 1). Whether or not we're in for a new September record, the next weeks will show.
Figure 1: Evolution of Arctic sea-ice extent in July and August from 1979 until 2011. (NSIDC)
A rainy summer might be one reason for an apparent lack of public attention with respect to the ongoing sea-ice loss. Another reason, however, is possibly the fact that we scientists have failed to make sufficiently clear that a major loss of sea ice during the early summer months is climatologically more important than a record minimum in September. This importance of sea-ice evolution during the early summer months is directly related to the role of sea ice as an efficient cooling machine: Because of its high albedo (reflectivity), sea ice reflects most of the incoming sunlight and helps to keep the Arctic cold throughout summer. The relative importance of this cooling is largest when days are long and the input of solar radiation is at its maximum, which happens at the beginning of summer. If, like this year, sea-ice extent becomes very low already at that time, solar radiation is efficiently absorbed throughout all summer by the unusually large areas of open water within the Arctic Ocean. Hence, rather than being reflected by the sea ice that used to cover these areas, the solar radiation warms the ocean there and thus provides a heat source that can efficiently melt the remaining sea ice from below. In turn, additional areas of open water are formed that lead to even more absorption of solar radiation. This feedback loop, which is often referred to as the ice-albedo feedback, also delays the formation of new sea ice in autumn because of the accompanying surplus in oceanic heat storage.
Measurements from ice buoys show that indeed melting at the bottom of the sea ice has increased significantly in recent years. While field experiments that were carried out in the 20th century showed unambiguously that surface melting used to be the dominant mechanism for the thinning of Arctic sea ice, now in larger and larger areas melting at the underside of the ice is almost equally important. Such melting from below is particularly efficient since the temperature at the ice-ocean interface is fixed by the phase equilibrium that must be maintained there. Hence, any heat provided by the ocean to this interface will lead to thinning of the ice in summer and to slower ice growth in winter. At the surface, the ice temperature is not fixed as long as the ice isn't melting, and heat input from the atmosphere can in part be compensated for by a change in surface temperature and an accompanying change in outgoing long-wave radiation at the ice surface.
In addition to these climatological reasons, there is another reason for why a public focus on just the September sea-ice extent is possibly misleading: Such focus might give the impression that sea-ice extent is stable in other seasons but summer. That this is not the case becomes obvious from the graphical distribution of extreme sea-ice extent for each individual month that is shown in Figure 2. The figure shows in red the years with the five lowest values of sea-ice extent for a certain month and in blue the years with the five highest values. A retreat of sea ice throughout the entire year is obvious. In fact, the sea-ice extent for every month since June 2010 has been among the five lowest values ever recorded by satellites.
Figure 2: Distribution of record minima and record maxima of Arctic sea-ice extent (NSIDC). The years with the five lowest values of sea ice extent for a certain month are marked in red, those with the five highest values of sea-ice extent are marked in blue. The darkness of the color indicates the ranking: the darkest red marks the lowest value, the darkest blue the highest.
Such widespread loss of Arctic sea ice has sometimes given rise to the concern that the total loss of Arctic sea ice at least during summer time can no longer be avoided. In this context, usually the ice-albedo feedback is mentioned, since it provides a mechanism that can in principle lead to a so-called "tipping point" beyond which the loss of the remaining sea ice becomes unstoppable. However, recent research shows that this scenario is too pessimistic. For example, in a paper published in Geophysical Research Letters in January 2011, Tietsche et al.  used climate model simulations to examine the evolution of Arctic sea ice after an extreme loss event. In their model simulations, they artificially removed all Arctic sea ice at the beginning of June for selected years and examined if the ice would recover from such extreme event.
Their main result is shown in Fig. 3: It took only about two years after each complete sea-ice removal until the ice had recovered to roughly the extent it had before the removal. Hence, sea ice extent is primarily defined by the prevailing climate conditions; the ice-albedo feedback mechanism is, in isolation, too weak to stabilize a very low sea-ice cover. In examining the mechanisms behind this finding, Tietsche et al. found that unusually large amounts of heat indeed accumulate in the ocean during the ice-free summer. However, this heat is efficiently released to the cold atmosphere already during the following autumn and winter. Once that heat release has cooled the ocean to its freezing temperature, sea ice forms again. Because this ice is initially very thin, the efficient release of heat from the ocean continues for some time, causing a rapid growth of the new sea ice. Much of this ice then survives the following summer, and sea-ice conditions can quickly return to those before the artificial perturbation.
Figure 3: Evolution of September sea-ice extent in coupled climate model simulations. The blue curve shows the evolution of the unperturbed sea-ice extent for the A1B scenario, with the gray shading showing the ensemble spread of three model runs. For the red curves, sea ice was artificially removed at the beginning of June in 1980, 2000, 2020, 2040 and 2060 within the climate model simulations. For all these perturbations, sea-ice extent recovered rapidly to the unperturbed extent. A similar result was found for sea-ice volume.
The finding that the long-term evolution of Arctic sea ice is primarily governed by the prevailing climate conditions implies that the loss of Arctic sea ice can still be slowed down and eventually stopped if an efficient reduction of CO2 emissions were to become reality soon. Last week, however, it became obvious once more how unlikely such scenario is: On 30th August, Exxon announced a deal with Rosneft, the Russian state oil company. As part of this deal, Exxon will invest more than US$2 billion to support Rosneft in the exploitation of oil reserves in the Kara Sea, which is part of the Arctic Ocean north of Siberia. One requirement for the success of this deal: a further retreat of Arctic sea ice. Given that climate model simulations indeed all project such further retreat of Arctic sea ice, it seems that at least to some degree, managers of big oil companies have started to make business decisions based on climate-model simulations. That may be good news. Or not.
• This article is in part based on a German article that was published at Klimalounge.