The Atmospheric Response to Realistic Reduced Summer Arctic Sea Ice Anomalies - Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications - page 104

prescribed as ice extent (ocean grid box is either completely
covered or totally ice free) or ice concentration (partial grid
box covered in ice allowed) based on monthly observations.
Fifty-one ensemble members were integrated from April to
October 1995 using climatological sea surface temperatures.
The control simulation was integrated with global climato-
logical sea ice extent and SSTs. The strongest response was
found during the month of August when the ice area is near-
ing its minimum for the year.
The Arctic displays a local thermal response with increased
surface heat fluxes (sensible plus latent plus longwave) into
the atmosphere, warmer SATs, and a weak decrease in
SLP. The atmospheric response is also characterized by an
anomalous high in sea level pressure in the North Pacific,
which is part of a northward expansion of the summertime
subtropical high. The atmospheric response with height is
equivalent barotropic, and the anomalous high increases in
amplitude with height and is significant at 200 hPa. There is
a significant decrease (increase) of precipitation along the
eastern (northwestern) part of the mean North Pacific storm
track, consistent with the 500-hPa geopotential height vari-
ances and 850-hPa transient eddy heat fluxes that indicate
enhanced storminess north of the mean storm track and a
decrease over the mean storm track in the North Pacific.
Additional climate experiments were conducted to deter-
mine the model sensitivity to the location of sea ice anom-
alies. When ice reduction is limited to only the Kara Sea,
the Laptev-East Siberian seas, or the Beaufort Sea the atmo-
spheric response patterns for SLP, geopotential height, and
precipitation are similar but weaker than when the sea ice is
reduced for all the seas, suggesting that the model is sensi-
tive to sea ice anomalies in all three regions. The area of the
significant response increases from the Kara to the Beau-
fort, which is closest to the North Pacific. These results are
analogous to a GCM study by Geisler et al. [1985] where
the model Pacific North American response pattern (magni-
tude) is insensitive (sensitive) to the longitude of the tropical
Pacific SST anomaly. The August 1995 experiment was re-
peated using ice concentration anomalies. The atmospheric
response is similar though it resembles a wave train in the
Pacific, similar to what Alexander et al. [2004] found for
the response during winter to sea ice concentration extremes
during winter of 1995-1996.
There has been increased interest recently in understand-
ing mechanisms that force and maintain the summertime
subtropical highs. In a zonal average, the subtropical highs
are strongest in winter when subsidence associated with the
Hadley circulation is most vigorous [ Grotjahn and Osman ,
2007]. However, the North Pacific (NP) high is strongest
during boreal summer [see Grotjahn , 2004, Figure 1] and
forms to the west of a region with strong thermal contrast

between the cool ocean water and the warm North American
landmass. Miyasaka and Nakamura [2005] employed a non-
linear spectral primitive equation model driven by zonally
asymmetric diabatic heating and demonstrated that the strong
surface thermal contrast can explain ~70% of the strength of
the subtropical high, consistent with ideas first proposed by
Hoskins [1996]. Grotjahn [2004] proposed that extratropi-
cal storms could provide forcing through transient eddies to
maintain the subtropical high. Grotjahn and Osman [2007,
Figure 2] present a conceptual picture of how ageostrophic
motions arising from developing storms converge at the jet
level, leading to sinking motion on the east side of the sub-
tropical high and low-level divergence and southward mo-
tion that strengthens the subtropical high. They demonstrate
that the variability of the NP high is dominated by midlati-
tude forcing during summer. Some of the features found
in a warm season SLP composite analysis of observations
by Grotjahn and Osman [2007] are qualitatively similar to
circulation anomalies forced by reduced sea ice in CCM3.
They find that SLP is weaker in parts of the Arctic Ocean
when the North Pacific high is stronger and a stronger North
Pacific high is associated with positive SLP anomalies on
the northern flank of the high.
The LBm analysis suggests that the far field response is
not forced directly by the Arctic ice but could rather be a
consequence of the local Arctic response, which acts to re-
duce the flow between the Arctic and the lower latitudes.
There may be some parallel with modeling studies of the
response to Antarctic sea ice extremes. Hudson and Hewit-
son [2001] have examined the response to realistic monthly
varying sea ice and SST anomalies in the Antarctic. They
found that where the sea ice has been reduced and ocean
exposed, the SAT increases and there is a strengthening
and a southward extension of the subtropical high-pressure
belt. Raphael [2003] found complementary results using the
NCAR CCSm.
The dominant mode of variability determined from Cn-
tle empirical orthogonal function (EOF1) of SLP in August
resembles the Arctic Oscillation. The model response to
reduced summer ice does not correspond to the dominant
mode for SLP or 500-hPa heights. Given results from previ-
ous studies we hypothesize that the reason that this occurs is
because the ice anomaly is located far from the storm tracks.
Honda et al. [1999] and Alexander et al. [2004] found that
wintertime North Pacific ice edge anomalies, located well
north of the average storm track, do not project on the domi-
nant modes of the GCM. In contrast, Deser et al . [2004],
Magnusdottir et al. [2004], and Alexander et al. [2004]
found that the GCM response to ice edge anomalies in the
North Atlantic during winter strongly project on the domi-
nant modes of variability. The storm track is located nearly