The Atmospheric Response to Realistic Reduced Summer Arctic Sea Ice Anomalies - Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications - page 99

dicating that the diabatic heating anomalies over the Arctic
are too shallow and weak to drive the large-scale response.
In other words, we argue for an indirect mechanism via the
storm track for the ice anomalies to impact the North Pacific
rather than impacting the flow directly.

of these regions and the induced climate anomalies are fairly
similar.
The precipitation response (not shown) patterns to the
partial ice anomalies are sensitive to the location of the ice
anomalies. The positive precipitation anomalies over eastern
Siberia are weakly evident in Sum95le and are significant in
Sum95be. The negative precipitation anomalies in the mean
model storm track zone are overall largest for Sum95be,
largest over south coastal Alaska for Sum95ke, and signifi-
cant for a limited area over the ocean for Sum95le. The sum
of the precipitation anomalies for the three partial ice experi-
ments is slightly larger than the precipitation anomalies for
Sum95e.

3.4. Partial Ice Reduction Sensitivity Experiments

To investigate the sensitivity of the atmospheric response
to the placement of the ice anomalies, three experiments
were conducted where CCM3 was forced with partial sea
ice anomalies from Sum95e ice conditions. The ice was re-
moved in the Kara (Sum95ke) (Plate 8a), Laptev-East Sibe-
rian (Sum95le) (Plate 8b), and Beaufort (Sum95be) (Plate
8c) seas. The integration and processing procedure for the
partial ice experiments was similar to one used for the full
anomaly case (Sum95e) to construct a 51 ensemble member
response. The largest positive net surface heat flux anom-
alies (not shown) into the atmosphere are located directly
over grid boxes where ice was removed and are identical to
the analogous anomalies from the Sum95e (Plate 2a) simula-
tion. Sum95ke and Sum95le display weak negative heat flux
anomalies over eastern Siberia, and Sum95be has significant
negative anomalies around 5 W m -2 .
Surface temperature and SAT responses to the partial
ice anomalies are characterized by warming in the vicinity
of the reduced ice anomaly, and the magnitudes are nearly
identical to those from the full ice experiment. Warm SAT
anomalies in far eastern Siberia-Bering region are signifi-
cant in Sum95le and Sum95be, with the Beaufort ice forcing
the largest response in eastern Siberia. The eastern Siberia
positive SAT anomalies are consistent with positive advec-
tion associated with the anomalous low (high) over the Sibe-
ria (North Pacific).
The atmospheric SLP and geopotential height responses
to the partial ice experiments resemble that of Sum95e. A
weak high over the Kara Sea, a weak low over east Siberia
stretching into the Chukchi Sea, and the anomalous high in
the North Pacific are all common features of the SLP and
geopotential height response patterns to partial ice anoma-
lies (Plates 8d-8i). The individual responses shown in Plate
8 are weaker than the response in Sum95e; however, the
sum of these three partial ice experiment response patterns
in Plate 8 for SLP and 500-hPa geopotential height is nearly
twice as strong as the response to the total ice anomaly
(Sum95e). Ice reductions in the Laptev-East Siberian and
Beaufort seas produce a statistically significant response in
the North Pacific. The anomalous low (SLP and 500-hPa
height) over east Siberia is stronger in the Beaufort par-
tial ice experiment than in Sum95e. This suggests that the
model atmosphere is sensitive to ice reductions in all three

3.5. Ice Concentration Experiments

The August 1995 experiment was repeated using ice con-
centration anomalies (Sum95c) (Plate 9a). The net surface
heat fluxes (not shown) and SAT were similar to Sum95e.
One feature different from the Sum95e ensemble average is
an area of significant negative surface air temperature anom-
aly between 120° and 150°E in eastern Siberia (Plate 9b).
The Sum95c SLP response has a weaker anomalous high in
the North Pacific and a stronger anomalous low in eastern
Siberia compared to the extent experiment. The atmospheric
response at 500 hPa is similar though it looks more like a
wave train in the Pacific (Plate 9c). During summer the con-
trast between using ice extent and concentration is small,
whereas the differences are larger in winter [ Alexander et al. ,
2004]. The area of open water is slightly larger during sum-
mer than winter (compare the difference between the two
time series in Figure 1 in 1995 with the difference for 1996
in Figure 1 of Alexander et al. [2004]), but the larger air-sea
temperature contrast in winter strongly influences the turbu-
lent heat fluxes. Anomalies for Sum95c are constructed by
taking the difference between the concentration experiment
(Sum95c) and an extent control (Cntle). Large ice anomaly
differences exist between the concentration control (Cntlc)
and Cntle, complicating the interpretation of the differences
when a concentration control is used, and thus the Cntlc ex-
periments are not used as a baseline here.

4. CONCLUSIONS

This study employs an atmospheric global climate model
(CCM 3.6) to examine the atmospheric response to observed
variations in sea ice during the summer of 1995, which had
the lowest ice extent during June-September in the Arctic
over the last ~30 years with the exception of 2007. (The
September ice minimum has been near or well below the
1995 levels since 2002 [ Stroeve et al. , 2008]). Sea ice was