Box 7.1: Sea ice and climate change.
Sea ice in the Arctic and around Antarctica responds directly to climate
change and may, if properly monitored, become increas-ingly important
for detecting climate change. Although sea ice covers only about 5% of
the Earth's surface, its extent and thickness have important influences
in the coupled atmosphere-ocean system. Increasing the understanding of
these processes and representing them more realistically in climate models
is important for making more reliable climate change projections.
Several processes associated with sea ice are climatically relevant. The
sea-ice albedo effect is an important contributor to the amplification
of projected warming at high latitudes. Albedo decreases if the extent
of sea ice is reduced and more ocean surface is exposed, resulting in
increased heat absorption and hence warming. Melting of snow and the formation
of melt-water ponds also reduces albedo and alters the radiation balance.
Changes in sea-ice thickness and lead (open water) fraction modify the
heat transfer from the ocean: thinner sea ice and more leads result in
enhanced heat loss from the exposed ocean thus further warming the atmosphere.
Changes in cloud cover may influence how large this effect really is.
A principal mechanism for dense water formation in the ocean around Antarctica
and in shelf regions of the Arctic is the rejection of brine as sea water
Changes in sea-ice formation alter the properties and formation rates
of ocean deep water and therefore have an influence on the water mass
structure that reaches far beyond the area of sea ice. Finally, ice export
from the Arctic represents an important southward flux of fresh water
which influences the density structure of the upper ocean in the Nordic,
Labrador and Irminger Seas.
7.1, Figure 1: Observed and modelled variations of annual averages
of Northern Hemisphere sea-ice extent (10 6 km 2 ). Observed data
for 1901 to 1998 are denoted by open circles (Chapman and Walsh,
1993, revised and updated) and for 1978 to 1998 by open triangles
(Parkinson et al., 1999, updated). The modelled sea-ice extents
are from the GFDL and Hadley Centre climate model runs forced by
observed CO2 and aerosols. Modelled data are smoothed
by a polynomial fit. Sea-ice extent in these models was determined
as the area which had a thickness exceeding 2 cm. This criterion
was determined to yield the best agreement with the observed mean
during 1953 to 1998; this choice also reproduces the seasonal cycle
realistically. Figure from Vinnikov et al. (1999)
Observations of sea ice:
Observations of sea-ice extent and concentration (the fraction of local
area covered by ice) are based primarily on satellite data available since
the late 1970s. Sea-ice thickness is also important in assessing possible
changes in the amount of sea ice; however, thickness observations are
more difficult to make. For the Arctic, thickness data come primarily
from sonar measure-ments from submarines and a few oceanographic moorings.
Although limited, the observations indicate statistically significant
decreases in ice extent and thickness over the past few decades, with
Arctic sea-ice extent declining at a rate of about 3% per decade since
the late 1970s. Sea-ice retreat in the Arctic spring and summer over the
last few decades is consistent with an increase in spring temperature,
and to a lesser extent, summer temperatures at high latitudes. Thickness
data show a near 40% decrease in the summer minimum thickness of Arctic
sea ice over approximately the last 30 years. Estimates using independent
methods for the winter, but over a much shorter period, also suggest thickness
reductions, but at a markedly slower rate.
However, due to limited sampling, uncertainties are difficult to estimate,
and the influence of decadal to multi-decadal variability cannot yet be
While Arctic sea-ice extent and thickness have clearly decreased in the
last 20 years, changes in Antarctic sea-ice extent have been insignificant.
The earlier part of the data set indicates somewhat greater ice extents
in the early 1970s, and indirect evidence from historical records also
points to more northerly sea-ice margins in the 1930s and 1940s. Warming
over much of
Antarctica has only been about 0.5°C over the last 50 years with the
notable exception of the Antarctic Peninsula where temper-atures have
increased by about 2°C for reasons that remain unclear.
Sea-ice modelling and projection:
Sea ice is particularly difficult to simulate in climate models because
it is influenced directly by both the atmosphere (temper-ature, radiation,
wind) and the ocean (heat transport and mixing, and surface currents),
and because many of the relevant processes require high grid resolution
or must be parametrized. Recent coupled climate models include a sea-ice
component that incorporates openings in the ice, often in conjunction
with ice dynamics (motion and deformation). Furthermore, updated parametrizations
of snow ageing and associated albedo changes and multi-layer formulations
of heat conduction through the ice and overlying snow cover are being
implemented in some models. Although many thermodynamic processes are
crudely approximated, it is unclear how these approximations contribute
to errors in climate model simulations. Sea-ice dynamics is important
in determining local ice thickness and export of sea ice from the formation
areas, but despite the rather mature status of physically based sea-ice
dynamics models, only a few of the current coupled climate models include
such a component. Coupled model simulations of the seasonal cycle of sea-ice
coverage in both hemispheres exhibit large deviations from the limited
observational data base, as illustrated in Chapter 8, and current research
is aimed at improving model performance.
Coupled model projections of the future distribution of sea ice differ
quantitatively from one to another as shown in Chapter
9. However, they agree that sea-ice extent and thickness will decline
over the 21st century as the climate warms. Box 7.1,
Figure 1 illustrates this with annual mean Arctic ice extent results
from two coupled models. The simulations of ice extent decline over the
past 30 years are in good agreement with the observations, lending confidence
to the subsequent projections which show a substantial decrease of Arctic
sea-ice cover leading to roughly 20% reduction in annual mean Arctic sea-ice
extent by the year 2050.