Analysis of Arctic Sea Ice Anomalies in a Coupled Model Control Simulation - Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications - page 195

Wallace , 1998], is shown in Figure 4e. The time series of
the AO index is closely related to the NAO [ Deser , 2000]
and describes 24% of the variance in the monthly North-
ern Hemisphere extratropical sea level pressure field over
the 350-year period computed. From Figures 4a and 4e, it
is seen that the event for simulation year 451 coincides with
a brief period of extreme positive values in the AO index in
February and march of that year. The AO index for February
of simulation year 451 is the largest monthly value in over a
200-year period of the model simulation. This corresponds
to an increase in wintertime ice velocities in the lincoln
Sea and through Fram Strait (see Plate 4a). For reference,
Plate 4a may be compared with the average ice velocity field
shown in Plate 2. Plate 4a also indicates anomalous export
through the Barents Sea. However, the total ice volume for
the Arctic during this event is already less than average val-
ues. For example, regions of ice thickness greater than 2.5
m cover a much smaller region of the central Arctic dur-
ing simulation year 451 than in the averaged depiction. This
may also be seen in the time series of ice volume shown in
Figure 4b. In this time series, the total ice volume anomaly
for the Arctic decreased from a long-term maximum at the
beginning of model year 441 to a negative anomaly of ap-
proximately 3700 km 3 in simulation year 451. It may then
be inferred that the Arctic ice volume had been precondi-
tioned to thinner ice in the years prior to 451 and that the
extraordinary minimum ice extent of 4.2 × 10 6 km 2 for that
September is the result of both dynamic and thermodynamic
processes. Using the long-term average annual extent of sea
ice, the corresponding ice thickness anomaly is equivalent to
about 0.5 m (see Figure 4b).
In Figure 4c, the time series of wintertime ice export
through Fram Strait is shown. The export is computed on the
ccSm3 native ocean grid. For monthly values (not shown),
ice export through the Fram Strait corresponds closely with
the AO index, and very large export values of greater than
800 km 3 month -1 correspond to the large values of the Feb-
ruary and march AO index for the year 451. However, on a
seasonal average, the Fram Strait ice export was not excep-
tional for the year 451, as seen in Figure 4c. Figure 4c shows
that the ice volume preconditioning occurred in two periods
of anomalous export and melting. In the export time series,
these 3- to 5-year periods occurred in simulation years cen-
tered on years 444 and 450. It is concluded that the ice extent
anomaly of 451 is the result of both thermodynamic and dy-
namic processes occurring in a preconditioned system.
In Figure 4a, it is seen that the ice extent recovers consider-
ably in the following summer. Negative ice extent anomalies
for the years 452-454 range from 0.4 × 10 6 to 0.8 × 10 6 km 2 ,
and by year 455 the ice extent anomaly is positive. This con-
trasts with the ice volume anomaly, which averages greater

than (negative) 2600 km 3 over the period 451-464. Subse-
quent recovery over simulation years 465-472 is associated
with decreases in winter ice export and sea ice melt (Figures
4c and 4d). Sea ice melt is the sum of top, lateral, and basal
terms and is averaged over the central Arctic Ocean pole-
ward of 80°N in the Atlantic Sector (90°W to 105°e) and
poleward of 70°N elsewhere. The melt curve in Figure 4d
shows a great deal of year-to-year variability but typically
lower values during this period of recovery in ice volume.
In terms of ice motion, the recovery period immediately
after the ice extent anomaly is dramatic. Plate 4b indicates
a highly anomalous velocity pattern with surface ice every-
where drifting from the Fram Strait to the Bering Strait in
the presence of an expanded Siberian High. Simulation year
453 sees a return to a more normal circulation pattern with
the return of a dominant Beaufort Gyre in the Canada Basin
(Plate 4c). The contrast in velocity patterns between sea ice
extent depletion and recovery years emphasizes the role of
the Beaufort Gyre in maintaining the cover during the sum-
mer months. Oceanic heat transports (not shown) computed
for the major passages in the North Atlantic do not show
significant variability during this event.
Arctic summertime conditions during simulation year 451
are dominated by the sea level pressure anomaly patterns
in the prior winter and concurrent summer. Pan-Arctic av-
erages of near-surface air temperature and precipitation for
the June-September months of year 451 are found to be
near climatology values. In Figure 5, the winter and sum-
mer fields of sea level pressure anomaly are contoured, and
hatched areas indicate statistical significance. The patterns
emphasize the role of the wintertime circulation, with low-
pressure anomalies in excess of 6 hPa over a large region
of the central Arctic Ocean. In summer, high pressure is lo-
cated over the canadian Arctic Archipelago; however, the
Arctic Ocean is again shaded by negative sea level pressure
anomalies. The event may be seen as more analogous to the
September 2002 event as described by Serreze et al. [2003],
in which persistent low pressure dispersed the ice pack, as
opposed to more recent events where high pressure over the
Arctic Ocean (and increased downwelling shortwave radia-
tion absorbed at the surface) played a more dominant role.
As the summer air temperatures are found to be near clima-
tology and there is no significant melt signal, it would seem
plausible that the event was dynamically induced during a
period of low sea ice concentration.

3.2. Simulation Year 490 Event

As described previously, the spatial pattern of the Septem-
ber sea ice minimum event for simulation year 490 is shown
in Plate 3b. Similar to the year 451 event, year 490 sea ice