The Role of Natural Versus Forced Change in Future Rapid Summer Arctic Ice Loss - Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications - page 130

Climate models project that decreasing Arctic ice cover
will continue into the foreseeable future. An analysis of the
models participating in the Intergovernmental Panel on Cli-
mate Change fourth assessment report shows that all models
simulate reduced Arctic ice cover throughout the 21st cen-
tury, with about 50% of them reaching seasonally ice-free
conditions by 2100 [ Arzel et al ., 2006; Zhang and Walsh ,
2006]. As reported by Holland et al . [2006a] (hereinafter re-
ferred to as HBT), future reductions in September (summer
minimum) ice cover occur quite abruptly in some models,
with periods of relative stability followed by very rapid ice
loss. Model simulations run with a middle-range forcing sce-
nario (Special Report on emissions Scenario (SReS) A1B)
suggest that in the future (generally from 2015 to 2050) short-
timescale (5-10 years) September ice loss could occur 3-4
times faster than what has been observed over comparable
length time periods through 2005. When observed conditions
from 2007 are included, the simulated ice loss over an abrupt
event as identified by HBT (see section 3.2) is only from 1.1
to 2.2 times faster (depending on the event) then a compara-
ble length of the observed record. So, while the observed ice
cover does not yet exhibit abrupt ice loss as defined by HBT,
the observed changes through 2007 are approaching those
levels. It is notable that model integrations do not simulate
these rates of ice loss until 2015 at the earliest.
HBT found that in the Community Climate System Model,
version 3 (CCSM3), there are three factors that contribute
to rapid September ice retreat. These include a thinning of
the sea ice to a more vulnerable state, pulse-like increases
in ocean heat transport to the Arctic that appear to trigger
the rapid change in sea ice extent, and the positive surface
albedo feedback which accelerates the retreat. One outstand-
ing question from the HBT study is whether the events iden-
tified in CCSM3 occur because of the ice cover reaching
an unstable threshold (“tipping point”) in which the sea ice
then rapidly switches to a new, stable (seasonally ice free)
equilibrium. Abrupt climate change is often defined in the
context of this type of tipping point behavior [e.g., National
Research Council , 2002]. Here we use a broader definition
of “abrupt change” in relation to summer sea ice loss, indi-
cating a change that is fast relative to the forcing but may not
constitute a threshold response.
Previous studies using relatively simple models suggest
that a threshold instability could exist in the transition to
year-round Arctic ice-free conditions [e.g., North , 1984;
Gildor and Tziperman , 2001] and that this may also occur
in some GCM simulations [ Winton , 2006]. It is possible that
a transition to seasonally ice-free conditions could also re-
sult from such “tipping point” behavior. Alternatively, the
abrupt ice loss simulated by CCSM3 may represent the in-
teraction of large intrinsic Arctic variability with increasing

forced change because of rising greenhouse gas concentra-
tions. This could occur in the absence of an instability in the
ice cover and could have implications for the predictability
of such events. The result of an abruptly changing ice cover
would be the same, however, with considerable impacts
on the socioeconomics, climate, and biological systems in
the Arctic. Here we examine the possibility of a critical ice
state that leads to the abrupt September ice retreat present
in CCSM3 integrations and the role of natural versus forced
change in determining simulated rapid ice loss events. The
implications for potential predictability of these events are
also discussed.

2. MODeL INTeGRATIONS

The Community Climate System Model, version 3 (CCSM3)
is a state-of-the-art fully coupled climate model, which includes
atmosphere, ocean, land, and sea ice components [ Collins
et al ., 2006a]. For the primary integrations considered here,
the atmosphere model (CAM3) [ Collins et al ., 2006b] is run
at T85 resolution (approximately 1.4°) with 26 vertical lev-
els. The ocean model [ Smith and Gent , 2004] includes an
isopycnal transport parameterization [ Gent and McWilliams ,
1990] and a surface boundary layer formulation following
Large et al . [1994]. The dynamic-thermodynamic sea ice
model [ Briegleb et al ., 2004; Holland et al ., 2006b] uses
the elastic-viscous-plastic rheology [ Hunke and Dukowicz ,
1997], a subgrid-scale ice thickness distribution [ Thorndike
et al ., 1975], and the thermodynamics of Bitz and Lipscomb
[1999]. Both the ice and ocean models use a nominally 1°
resolution grid in which the North Pole is displaced into
Greenland. The land component [ Bonan et al ., 2002] in-
cludes a subgrid mosaic of plant functional types and land
cover types based on satellite observations. It uses the same
spatial grid as the atmospheric model.
We discuss 20th and 21st century CCSM3 simulations
that were performed for the Intergovernmental Panel on Cli-
mate Change, fourth assessment report (IPCC-AR4) [ Inter-
governmental Panel on Climate Change , 2007]. These will
be referred to as the IPCC-AR4 CCSM3 integrations. eight
20th to 21st century ensemble members are available. The
20th century runs use external forcings based on the ob-
served record and chemical transport models. These include
variations in sulfates, solar input, volcanic forcing, ozone,
a number of greenhouse gases, halocarbons and black car-
bon. They were initialized from different years of a multi-
century preindustrial (year is 1870) control integration. For
the 21st century simulations, we focus on the middle-range
SReS A1B forcing which reaches approximately 720 ppm
CO 2 levels by 2100 [ Intergovernmental Panel on Climate
Change , 2001].