Geoscience Reference
In-Depth Information
observations made north of Severnaya Zemlya in the 1970s
and 1980s estimated upward heat fluxes of 4 to 6 W m -2
which would affect the overlying sea ice cover [ Walsh et
al. , 2007]. The mechanisms through which warm Atlantic
and Pacific water inflows into the Arctic may affect the sea
ice cover have been identified as known gaps in the present
understanding (Arctic Research Consortium of the United
States, Arctic observation integration workshops report, 55
pp., Study of environ. Arct. change Proj. Off., Fairbanks,
Alaska, 2008, available at http://www.arcus.org/SeArcH/
meetings/2008/aow/report.php).
A goal of this study is to examine the relative roles of the
various climate components in coupled model simulations
of the Arctic climate. Numerical general circulation models
represent an integration of our knowledge of the processes
that are associated with the Arctic climate system and that
are essential for prognostic assessments. These processes
are simulated in a manner that is largely absent of a presup-
posed prioritization and is physically consistent in conserv-
ing mass, energy, and momentum. Global, coupled models
also incorporate remote forcings from outside the Arctic Ba-
sin. Understanding the output of a climate model may shed
additional light on the forcing mechanisms that are affecting
the contemporary Arctic. Currently, there is great interest
in prognostic simulations of the 21st century because of the
intermodel variability in the timing and conditions that are
associated with the transition from perennial to seasonal ice
cover. For example, Holland et al. [2006a] found a multi-
staged transition in which different mechanisms dominated
at different periods of time, producing a stair-stepped time
series of late summer ice concentration. These mechanisms
may be viewed to occur as a result of increased interannual
variability that is superimposed on a forced signal as the sea
ice thins [ Holland et al. , this volume].
The approach here is to examine a control simulation from
a state-of-the-art climate model using constant 1990 trace gas
forcing over an extended period of time. extended control
simulations on the order of 1000 years enable a comparison
of the relation between variables over a period time of time
that is significantly longer than the observational record. In
particular, we focus on a 350-year segment of the time se-
ries in which possible spin-up effects from the beginning of
the simulation may be discounted. Over this time period, the
average conditions closely resemble those found in observa-
tions of the contemporary Arctic climate, both in the annual
mean and in the seasonal cycle. However, this simulation
shows drastic declines in sea ice extent occurring 3 to 4 times
in the time series. These episodic events are comparable to
the recent dramatic decrease in ice extent observed in Sep-
tember 2007, and yet the simulated ice recovers to “normal”
sea ice extent conditions within a few years. This contrasts
with anthropogenically forced simulations of the 21st cen-
tury from the same climate model, which produces virtually
ice-free summer Arctic conditions around 2040-2055 from
which the system never recovers [ Holland et al. , 2006a]. An
understanding of the mechanisms that are responsible for the
decline and recovery of the sea ice cover in control simula-
tions when compared with those of the forced simulations is
of particular interest and is timely. Indeed this gives a dif-
ferent perspective on the question of what fraction of the
observed recent change in sea ice extent is forced and what
fraction is coming from natural variability. Some questions
of interest to be examined in this study are as follows:
· Do these events of substantial sea ice cover loss all
occur in a similar manner?
· How do anomalous events evolve? Does precondition-
ing play a role?
· Do these anomalies in ice cover coincide with extreme
conditions in atmospheric or oceanic circulation?
· How rapidly does the ice pack recover?
This paper is divided as follows: Section 2 describes the
Community Climate System Model, version 3 (CCSM3),
which is used in this study. A brief evaluation of this model
in comparison to the contemporary Arctic climate is also
presented in this section. A description of the mechanisms
responsible for the decline and recovery in Northern Hemi-
sphere sea ice cover is given in section 3. Finally, a discus-
sion of the findings of this study and the implications of these
results for the contemporary Arctic is given in section 4.
2. THe cOmmuNITy clImATe SySTem mODel,
verSION 3
The CCSM3 is a fully coupled climate model composed
of atmosphere, ocean, sea ice, and land surface components
developed by the National Center for Atmospheric Research
(NCAR) [ Collins et al. , 2006a]. The model was released
in may 2004 and was a participating model of the Fourth
IPCC Assessment [ Meehl et al. , 2007]. The model has in-
corporated a number of updates from the previous version
that was described by Kiehl and Gent [2004] and includes
an improved representation of cloud process and the mixed
ocean layer [ Holland et al. , 2006b]. The atmospheric com-
ponent is referred to as the Community Atmosphere Model,
or cAm3, and is based on an eulerian spectral dynamical
core [ Collins et al. , 2006b]. Standard simulations use a con-
figuration with 26 hybrid sigma levels in the vertical and
horizontal resolution with triangular truncation of 85 wave
numbers. This is equivalent to an equatorial grid spacing
of about 1.4°. The orography used in cAm3 is shown in
Search WWH ::




Custom Search