Sea Ice–Albedo Feedback and Nonlinear Arctic Climate Change - Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications - page 123

The change of atmospheric heat transport convergence
with polar temperature in Figure 2 shows that, in the GCMs
as well as the EBM, enhancement of polar warming by
atmospheric transport at low temperatures gives way to a
damping impact at higher temperatures. However, the GCM
transition occurs at much lower temperature, perhaps partly
because of its having a polar albedo response at lower tem-
peratures. Other mechanisms, not present in the EBM, can
significantly impact poleward transport in the GCMs, for ex-
ample, the enhancement of latent heat transport with temper-
ature [ Alexeev , 2003; Held and Soden , 2006]. As the ice-free
state is approached, the damping effect of the atmospheric
heat transport change is much larger than longwave damping
in the EBM as in the GCMs.

Arctic Ocean becomes more convective, are also found in
this experiment (not shown).
Our main interest in the experiment is to assess the sta-
bility of the Arctic ice and its causes. The net surface heat
flux change in the ice-covered regions is the result of two
competing changes: (1) increased shortwave absorption due
to lowered albedo and (2) increased longwave and turbulent
heat loss due to increased surface temperature. The net up-
ward heat flux change of 24 W m −2 (Figure 4) indicates that
the second change is dominant, so the ice is stable and would
grow back at an initial rate of 2.5 m a −1 . At the top of the
atmosphere, the extra shortwave absorption is only partly
balanced by increased OLR. Most of the damping influence
comes from a reduction of heat transport into the ice-covered
region by the atmosphere. This reduced heat transport, in
turn, is mainly supported by a reduction in surface heat flux
from the adjacent ice-free ocean, particularly in the North
Atlantic (Plate 8).
The AMIP experiment fixes SSTs implicitly assuming
that the ocean has an infinite heat capacity. This assumption
may be reasonable here since the near-ice regions that are
experiencing large heat flux changes are occupied by deep
wintertime mixed layers with much larger heat capacity than
the sea ice or the shallow atmospheric layer that interacts
with it. Nonetheless, it is useful to relax this assumption by
performing a similar experiment in a fully coupled climate
model, a developmental version of GFdl CM2.1. In this
100-year experiment, we force the ice region by lowering
the ice albedos. Figure 5 shows changes in climate model
heat fluxes as in Figure 4. As expected, there is an increase
in shortwave absorption at the top of the atmosphere, and
as in the AMIP case, it is only partially offset by a local
OlR change. Again, the main balancing effect is from a re-

6. TETHERING EFFECT OF HEAT TRANSPORT

The previous section shows that atmospheric heat trans-
port plays an important but complicated role in polar climate
change: initially forcing the region to warm at a greater-
than-global rate but eventually becoming a cooling influence
at higher temperatures. Held and Soden [2006] show that the
latent heat component of the transport scales up in a warming
climate according, roughly, to the Clausius-Clapeyron rela-
tionship. This increase drives an increase of the total trans-
port toward the North Pole in spite of polar amplification.
However, it is possible that, even in the early warming, part
of the transport is helping to maintain the very constant polar
amplifications seen in Plate 2. To expose this moderating
role, we perform two diagnostic experiments that force only
the polar regions and examine the damping mechanisms.
The first is a modification of the Atmospheric Model Inter-
comparison Project (AMIP) experiment: an atmospheric
model run with specified sea surface temperatures and sea
ice cover. We perform a twin to this experiment where the
sea ice boundary condition is replaced with seawater freez-
ing temperature and albedo. The experiment is done with the
atmospheric component of the Geophysical Fluid dynamics
laboratory (GFdl) CM2.1 climate model. A similar exper-
iment for the DJF season was performed earlier by Royer
et al . [1990]. The impact on atmospheric temperatures and
winds in the current experiment are in general agreement
with those found by Royer et al. Plate 7 shows that there is
an intense warming of the lower polar atmosphere, mainly
confined below the 0°C potential temperature contour of the
control, “ice-in,” simulation. This is consistent with the re-
gionally limited response of the GCMs to transition to ice-
free conditions shown in Figure 3. Other features found by
Royer et al., an equatorward shift of the jet, redistribution
of sea level high pressure away from the central Arctic to
adjacent land regions, and a reduction of cloud cover as the

Figure 4. Difference in atmospheric heat fluxes over ice-covered
and ice-free regions of the Northern Hemisphere between “ice-out”
and “ice-in” AMIP runs. The “ice-covered” region is defined by
the annual mean ice concentration of the “ice-in” experiment. All
fluxes are in units of W m −2 ice-covered region.