Geoscience Reference
In-Depth Information
5
The equation of state for the atmosphere (p. 22).
roughness) and soil (moisture content) are typically
incorporated into the GCM. These are smoothed to be
representative of the average state of an entire grid cell
and therefore much regional detail is lost. Sea-ice extent
and sea-surface temperatures have often been specified
by a climatological average for each month in the past.
However, in recognition that the climate system is quite
interactive, the newest generation of models includes
some representation of an ocean which can react to
changes in the atmosphere above. Ocean models (Figure
8.2) include a so-called swamp ocean where sea-surface
temperatures are calculated through an energy budget
and no annual cycle is possible; a slab or mixed-layer
ocean, where storage and release of energy can take
place seasonally, and the most complex dynamic ocean
models, which solve appropriate equations for the ocean
circulation and thermodynamic state similar to 1-5
above and which are coupled to atmospheric models.
Such coupled models are referred to as atmosphere-
ocean general circulation models (AOGCMs). When
the global ocean is considered, seasonal freezing/
melting and the effects of sea ice on energy exchanges
and salinity must also be modelled. Therefore, dynamic
sea-ice models, which actively calculate the thickness
and extent of ice, are now replacing the specification
of climatological sea ice. Because of the century-long
timescale of deep ocean circulations, the use of a
dynamic ocean model requires large amounts of sim-
ulation time for the different model components to
equilibrate which greatly increases the cost of running
these models.
Because coupled AOGCMs are used in long-term
(century or millennium scale) simulations, an important
6
In addition, conservation equations for other atmos-
pheric constituents such as sulphur aerosols may be
applied in more complex models.
Model simulations of present-day and future climate
conditions involve iterating the model equations for
perhaps tens to hundreds of years of simulated time
depending on the question at hand. In order to solve
these coupled equations, additional processes such as
radiative transfer through the atmosphere with diurnal
and seasonal cycles, surface friction and energy trans-
fers and cloud formation and precipitation processes
must be accounted for. These are coupled in the manner
shown schematically in Figure 8.1. Beginning with a set
of initial atmospheric conditions usually derived from
observations, the equations are integrated forward in
time repeatedly using time steps of several minutes to
tens of minutes at a large number of grid points over the
earth and at many levels vertically in the atmosphere;
typically ten to twenty levels in the vertical is common.
The horizontal grid is usually of the order of several
degrees' latitude by several degrees' longitude near
the equator. Another, computationally faster, approach
is to represent the horizontal fields by a series of
two-dimensional sine and cosine functions (a spectral
model). A truncation level describes the number of two-
dimensional waves that are included. The truncation
procedure may be rhomboidal (
R
) or triangular (
T
);
R
15
(or
T
21) corresponds approximately to a 5° grid spacing,
R
30 (
T
42) to a 2.5° grid, and
T
102 to a 1° grid.
Realistic coastlines and mountains as well as
essential elements of the surface vegetation (albedo,
Figure 8.1
Schematic diagram of the
interactions among physical processes
in a general circulation model.
Source
: From Druyan
et al
. (1975), by per-
mission of the American Meteorological
Society.
