Geoscience Reference
In-Depth Information
If the simulated abrupt transitions are the manifestation of
a “tipping point” and result from reaching a certain thresh-
old, then it is likely that they would be preceded by a similar
“critical” ice state. Plate 3 shows a sample of the May aver-
aged ice conditions in the year that an abrupt ice loss event is
initiated and the region of ice loss over that event. The May
ice thickness is shown because this generally represents the
conditions at the beginning of the ice melt season. In Plate
3, we only consider the five ice loss events that are initiated
between 2020 and 2030. These spatial maps indicate that
the ice lost during different abrupt events varies in its initial
thickness and the region of loss. The fractional coverage of
thin ice in this region also varies (not shown). Quantitative
measures of these properties (e.g., the ice thickness in the
region where ice is lost during the abrupt event, the ice con-
centration in this region, and the concentration of thin ice,
etc.) have been analyzed and found to differ considerably at
the initiation of the different events. The properties over the
abrupt ice loss region at the start of the events are instead
closely related to the year in which the event begins which
varies from 2012 to 2045. earlier occurring abrupt ice loss
events are generally initiated with thicker and more extensive
sea ice than those events which occur later in the 21st cen-
tury. Although the ice conditions vary at the initiation of
different events, it is clear that an abrupt ice loss event is
only possible if the ice has thinned adequately. No events
are present during the 20th century simulations because of a
relatively thick ice cover.
The May ice thickness 5 years prior to the time at which
a model grid cell transitions from September ice-covered
to September ice-free conditions varies regionally but is
broadly similar across the different ensemble members
(Plate 4). In particular, the thickness 5 years prior to ice-free
conditions in the Beaufort, Chukchi, and east Siberian seas
Figure 1.
Annual cycle of ice extent, defined as the area with
greater then 15% ice concentration, averaged from 1980 to 1999
for the IPCC-AR4 CCSM3 ensemble mean (solid line) and the ob-
servations (dotted line). The observed values are from the National
Snow and Ice Data Center sea ice index [
Fetterer et al.
, 2002],
which are obtained from the passive microwave satellite data. The
vertical lines indicate the interannual standard deviation for each
month.
observed conditions through 2007 are considered, however,
the simulated trends over these events are only modestly
larger (1.1-2.2 times larger, depending on the event) than the
observed ice loss. While the observed ice cover has yet to ex-
hibit abrupt ice loss as defined above, with the extremely low
ice conditions in 2007 the observed record is approaching the
rates of change of the simulated abrupt events. In the model
integrations, these events typically do not occur until after
2020, with the earliest event starting in 2012 (Plate 2g).
Table 1.
Trends and Standard Deviation in the 1979-2007 September Ice extent From Observations and
the IPCC-AR4 CCSM3 ensemble Members
a
Standard Deviation (10
6
km
2
)
Raw Time Series
Trend (10
6
km
2
a
−1
)
Detrended Time Series
Observations
−0.07 (−0.06)
0.80 (0.65)
0.51 (0.42)
Run 1 (b.eS01)
−0.08
0.80
0.43
Run 2 (a)
−0.08
0.79
0.45
Run 3 (b)
−0.02
0.46
0.45
Run 4 (c)
−0.04
0.56
0.44
Run 5 (d)
−0.06
0.65
0.42
Run 6 (e)
−0.07
0.69
0.33
Run 7 (f.eS01)
−0.02
0.49
0.46
Run 8 (g.eS01)
−0.06
0.74
0.50
a
For the observations, the numbers in parentheses show values for the 1979-2006 time period. The letters
in parentheses indicate the given CCSM3 case name and eS01 refers to simulations that were run on the
earth Simulator machine.
Search WWH ::
Custom Search