Geography Reference
In-Depth Information
10.8
NUMERICAL SIMULATION OF THE GENERAL CIRCULATION
The previous section discussed the role of laboratory experiments in contribut-
ing to a qualitative understanding of the general circulation of the atmosphere.
Although laboratory experiments can elucidate most of the gross features of the
general circulation, there are many details that cannot possibly be duplicated in
the laboratory. For example, the possible long-term climatic effects of aerosols
and trace gases that are added to the atmosphere as a result of human activities
could not possibly be predicted on the basis of laboratory experiments. As another
example, the influence of an ice-free Arctic Ocean on the global climate would also
be extremely difficult to simulate in the laboratory. Because all the conditions of
the atmosphere cannot be duplicated in the laboratory, the only practical manner in
which to quantitatively simulate the present climate, or to predict possible climate
modifications resulting from intentional or unintentional human intervention, is
by numerical simulation with the aid of supercomputers.
Atmospheric general circulation models (AGCMs) are similar to large-scale
numerical weather prediction models (see Chapter 13) in that they attempt to
explicitly simulate synoptic-scale weather disturbances. However, whereas weather
prediction is an
initial value
problem, which requires that the evolution of the
flow be computed from a specified initial state, general circulation modeling is a
boundary value
problem in which the average circulation is computed for specified
external forcing conditions. In many AGCMs the sea surface temperature is treated
as a specified forcing. In reality, of course, there are strong interactions between
the atmosphere and the ocean: the winds drive currents, which influence the sea
surface temperature distribution, which in turn influences the atmosphere.
Although the dynamical equations used in general circulation models are the
same as those used in short-range numerical prediction models, general circulation
models are generally more complex, as in simulations of time scales beyond a
few days physical processes that are unimportant for the short-term evolution may
become crucial. Thus, parameterizations are needed for physical processes such as
surface heat and moisture fluxes, moist convection, turbulent mixing, and radiation.
However, because a forecast of the flow evolution from a specific initial state is not
needed, the problems associated with specifying initial data in forecast models do
not arise in AGCMs. In recent years, as numerical prediction has been extended
into the middle range of 1-2 weeks, more physical processes have been included in
the forecast models, and the differences between global weather forecast models
and general circulation models have become smaller.
10.8.1
The Development of AGCMs
The success of the quasi-geostrophic model in short-range prediction suggests
that for simulating the gross features of the general circulation such a model
