In the oceanic component of climate models, ocean current patterns are represented
significantly better in models of higher resolution in large part because ocean
current systems (including mesoscale eddies), ocean variability (including ENSO
events), and the thermohaline circulation (and other vertical mixing processes)
and topography which greatly influence the ocean circulation, can be better
represented. Improved resolution and understanding of the important facets of
coupling in both atmosphere and ocean components of global climate models have
also been proven to reduce flux imbalance problems arising in the coupling of
the oceanic and the atmospheric components. However, it must still be noted
that uncertainties associated with clouds still cause problems in the computation
of surface fluxes. With the availability of computer power, a central impediment
to the gain in model accuracy is being reduced; however, there is still a long
way to go before many of the important processes are explicitly resolved by
the numerical grid. In addition there continues to be a necessary “concomitant”
increase in resources for process studies and for diagnosis as computer power
increases. It must still be remembered that the system presents chaotic characteristics
that can only be evaluated through an analysis of ensembles statistics, and
these ensembles must be generated by running suites of models under varied initial
and forcing conditions.
In a few model calculations, a large rate of increase in the radiative forcing
of the planet is enough to cause the ocean’s global thermohaline circulation
almost to disappear, though in some experiments it reappears given sufficiently
long integration times (see Chapter 7, Section 7.3.7
and Chapter 9, 9.3.4.3). This circulation is important
because in the present climate it is responsible for a large portion of the
heat transport from the tropics to higher latitudes, and it plays an important
role in the oceanic uptake of CO2. Palaeo-oceanographic investigations suggest
that aspects of longer-term climate change are associated with changes in the
ocean’s thermohaline circulation. We need appropriate observations of the
thermohaline circulation, and its natural variations, to compare with model
simulations (see Chapter 9, Section 9.3.4.3; see also
Chapter 7, Section 7.6 and Chapter 8,
Section 8.5.2.2).
The coming decade will be important for ocean circulation in the context of
climate. A particularly exciting development is the potential for assimilating
synoptic ocean observations (e.g., the US/French ocean TOPography satellite
altimeter EXperiment (TOPEX-POSEIDON) and Argo) into ocean general circulation
models. Key questions, such as how well do the ocean models capture the inferred
heat flux or tracer distributions, are central to the use of these models in
climate studies. The effort of comparing models with data, as the direct path
for model rejection and model improvement, is central to increasing our understanding
of the system.
There is increasing evidence that there is a decline in the extent and thickness
of Arctic sea ice in the summer that appears to be connected with the observed
recent Arctic warming (see Chapter 2, Section 2.2.5.2;
Chapter 7, Box 7.1, and Chapter
8, Section 8.5.3; see also Chapter 7, Section 7.5.2
for a general discussion on the role of sea ice in the climate system as well
as recent advances in modelling sea ice).
It is not known whether these changes reflect anthropogenic warming transmitted
either from the atmosphere or the ocean or whether they mostly reflect a major
mode of multi-decadal variability. Some of this pattern of warming has been
attributed to recent trends in the Arctic Oscillation (see Section
2.6); however, how the anthropogenic signal is imprinted on the natural
patterns of climate variability remains a central question. What does seem clear
is that the changes in Arctic sea ice are significant, and there is a positive
feedback that could be triggered by declines in sea-ice extent through changes
in the planetary albedo. If the Arctic shifted from being a bright summer object
to a less bright summer object, then this would be an important positive feedback
on a warming pattern (see the “left loop” in Chapter
7, Figure 7.6).
In addition to these recently available observations, there have been several
models (Commonwealth Scientific and Industrial Research Organisation (Australia)
(CSIRO) – Gordon and O’Farrell, 1997; Department of Energy (USA) Parallel
Climate Model (DOE PCM) – Washington et al., 2000; National Center for
Atmospheric Research (USA) Climate System Model (NCAR CSM) – Weatherly
et al., 1998; see also Chapter 7, Section 7.5.2 and Chapter
8, Section 8.5.3) that have improved their sea ice representation since
the SAR. These improvements include simulation of open water within the ice
pack, snow cover upon the ice, and sea ice dynamics. The incorporation of sophisticated
sea ice components in climate models provides a framework for testing and calibrating
these models with observations. Further, as the formulation of sea ice dynamics
becomes more realistic, the validity of spatial patterns of the simulated wind
stress over the polar oceans is becoming an issue in Atmosphere-Ocean General
Circulation Model (AOGCM) simulations. Hence, improvements, such as the above-mentioned
data, in the observational database will become increasingly relevant to climate
model development. In addition, satellite observations have recently been used
to determine sea-ice velocity (Emery et al., 1997) and melt season (Smith, 1998).
New field programmes are under way with the explicit goal of improving the
accuracy of model simulations of sea ice and polar climate (see Randall et al.,
1998, for a review). In order to improve model representations and validation,
it will be essential to enhance the observations over the Arctic including ocean,
atmosphere, and sea ice state variables. This will help provide more reliable
projections for a region of the world where significant changes are expected.
The refinement of sea-ice models along with enhanced observations reduces the uncertainty associated with ice processes. (See Chapter 7, Section 7.5 and Chapter 8, Section 8.5.3 for more discussion and evaluation of model performance; for some open issues see Chapter 9, Section 9.4.) This progress is important, and efforts are needed to expand upon it and, as stated, to improve the observational basis significantly.
Other reports in this collection |