Geoscience Reference
In-Depth Information
Initiation
i.e. using 4DDA to define initial values of state variables globally
Solving the dynamics
i.e. calculating updates values of the state variables by solving
conservation laws assuming fixed values of divergences
Run time
outputs
Calculating the physics
i.e. calculating updates values of divergences for each volume
element assuming fixed state variables
Smoothing divergences
i.e. applying smoothing to maintain model stability
Figure 8.2 Sequence of
operations during a GCM run.
Stop and output
The equations coded in the GCM are then applied sequentially in two groups at
each time step. During the first set of calculations the processes that control changes
in modeled variables within each simulated volume (such as radiation divergence,
phase changes, and the input or loss of energy, momentum, and mass at the top and
bottom of each atmospheric column) are held constant. The conservation laws and
ideal gas law are applied to compute how energy, mass, and momentum are
re-distributed between the many volume elements over the time step and how
the equivalent meteorological variables describing the state of the atmosphere, the
state variables , are altered as a result. Applying this set of equations with fixed flux
divergences at each grid point is sometimes called solving the dynamics .
In the next step in GCM operation the state variables are held constant and the
processes that give rise to internal changes in state variables (the divergence terms)
are calculated in anticipation of their application during the next time step. Making
these calculations is sometimes called calculating the physics although some of the
processes described may actually be chemical or biological in nature. If the effect
of changing influences on weather are being investigated, imposed changes in, for
example, the concentration of radiatively active gases or the representation of
surface vegetation are imposed while calculating the physics. In the real world,
equations involved in solving the dynamics and solving the physics apply simulta-
neously rather than in sequence, and the need to apply them in sequence in a
GCM run is a compromise which can give rise to model instability. For this reason
it is usually necessary to apply some form of smoothing procedure to the
divergences calculated during the physics calculations before the next time step.
This two stage sequence of calculations, i.e., first solving the dynamics and then
calculating physics, is then repeated successively as the GCM runs forward in time
Search WWH ::




Custom Search