GREENHOUSE GASES EFFECTS 437
When the coupled model runs were carried forward from
1990 to 2050, increasing the CO 2 by 1% p.a. compound,
the effect of aerosols was to reduce the global greenhouse
warming from the 0.3C/decade shown in Figure 5 to only
0.2C/decade, and to largely offset it in highly polluted
regions.
More reliable estimates of the effects of aerosols on
climate must await much better observational data on the
sources, concentration, size, chemical composition and spa-
tial distribution of both natural and anthropogenic aerosols,
including strongly absorbing carbonaceous particles, and
dusts, and on the difference between droplet concentrations
and sizes in clean maritime and polluted continental clouds.
These data will be difficult and expensive to acquire; mean-
while, we are likely to have too many theories and computa-
tions chasing too few observations.
When aerosol particles are injected into the stratosphere,
they remain there for much longer periods and become
much more uniformly distributed than in the troposphere.
Concentrations remained consistently high in the 1980s rela-
tive to earlier decades, largely due to the El Chichon volca-
nic eruption in 1982. Recently they were much enhanced by
the Mt. Pinatubo eruption in June 1992 which injected about
20 million tons of SO 2 directly into the stratosphere. The con-
sequent reduction in radiative forcing at the top of the atmo-
sphere, estimated at 4 W/m^2 , would have required the surface
to cool by about 1C in order to restore equilibrium. The fact
that the temperature fell by only 0.3–0.5C during the follow-
ing two years may be due partly to absorption of radiation and
infra-red emission by the aerosols, to the thermal lag of the
oceans and to other negative feedbacks in the system.SEA-LEVEL RISEA potentially important consequence of greenhouse warm-
ing is the melting of sea-ice and ice sheets on land, only
the latter resulting in a rise in sea level. The sea level will
also rise as the ocean waters expand in response to the addi-
tional warming. Estimates of these consequences involve
large uncertainties because of the lack of observations and
understanding of the mass balance and dynamics of glaciers
and ice sheets. These uncertainties are compounded by the
uncertainty in the predicted increases in surface temperature
due to greenhouse warming.
Over the past 100 years, the sea level is estimated to have
risen about 10 cm. Thermal expansion of the ocean waters
has probably been responsible for 4 cm of this rise, melting
of mountain glaciers for 4 cm, and melting of the Greenland
ice sheet for 2.5 cm. Glaciologists believe that there has
been little, if any, overall melting of the Antarctic ice sheet
because the air temperatures are too low.
If air and surface temperatures were to increase because
of greenhouse warming, thermal expansion of the oceans
and melting of mountain glaciers are likely to continue to
make the largest contribution to sea-level rise. We have seen
that globally coupled atmosphere deep ocean models predict
that, when the atmospheric concentrations of carbon dioxideapproaches double the present value, the average surface air
temperature will increase by 0.3C/decade. The best esti-
mate of the corresponding rise in seal-level is 4 cm/decade,
about half resulting from expansion of ocean waters and half
from melting of landbased ice. These estimates, which may
conceivably be in error by a factor of two either way, imply
serious consequences for low-laying, highly populated areas
such as Bangladesh, but they are very much smaller than the
wildly exaggerated values that have appeared in the media.UNCERTAINTIES IN MODEL PREDICTIONSIn summarising the current state of knowledge and under-
standing of the likely magnitude, timing and impacts of
enhanced greenhouse warming, it is virtually certain that
the troposphere is warming very slowly in response to the
continually increasing concentrations of carbon dioxide and
other greenhouse gases, but the signal is as yet barely detect-
able above the large natural climate variations, probably
because it is being delayed by the large thermal inertia of
the oceans and has also been masked by the cooling effect of
man-made aerosols.
Predictions of the increase in globally averaged tem-
perature that may result from a doubling of carbon dioxide
have recently converged towards lower values ranging from
1.3C to 2.3C, based on coupled models with a deep ocean.
However, this trend may be deceptive because only a small
number of 70–100 year simulations have been published and
considerable problems and uncertainties remain, both in the
atmospheric physics and in the ocean dynamics. These arise
largely from the sensitivity of the models to the simulation
of clouds and their interaction with the radiation fields, the
uncertainty as to how well they simulate the ocean circu-
lation, and the necessity to adjust the ocean surface fluxes
in order to ensure that the ocean temperature and salinity
remain close to present-day climatology and that the control
model climate does not drift during long runs. Long-term
drift in the climate of the Southern Hemisphere arises from
an imbalance in the heat budget of the Antarctic leading
to a spurious slow-melting of the ice. This has now been
corrected and changes in the pack-ice are now included.
Another important defect of current low resolution ocean
models is that they do not capture narrow features such as
the Gulf Stream and Kuroshio currents and the regions of
strong upwelling off South Africa and South America, all of
which play an important role in heat transport. Some tests
with higher resolution (1.25 1.25) in the UKMO model
improved this situation but only partially. The fact that cur-
rent models have only limited success in simulating and
predicting such a spectacular event as the El Nino is also
evidence of defects in the treatment of atmosphere-ocean
interactions. There is also a need for an improved represen-
tation of atmospheric boundary layer.
Even if the various models agree quite well on the
globally-averaged effects, they show larger differences on
regional and sub-regional scales, which are politically and
economically more relevant. Further improvements in modelC007_002_r03.indd 437C007_002_r03.indd 437 11/18/2005 10:28:02 AM11/18/2005 10:28:02 AM