Geoscience Reference
In-Depth Information
for NWP are therefore adopted and applied. Because the data delivered by data
assimilation is influenced by the nature of the GCM used, this means there are
discontinuities in the derived global data sets associated with model changes. For
this reason, some major modeling centers have deliberately created model-
calculated data sets by applying consistent data assimilation procedures to
historical observational data using the same recent model to provide more
consistent data series. These data are called
reanalysis data
. Thus,
when GCMs are
used to create model-calculated reanalysis data, a consistent data assimilation
procedure is used with the same version of the GCM to calculated globally available
data sets of weather variables from historical data records.
As experience in the use of GCMs in NWP grew, confidence also grew that
although their ability to predict actual weather decayed after several days, their
ability to make reasonable predictions of the
statistics of weather
, i.e., climate, at a
particular place was retained. This confidence sparked a whole new branch of
atmospheric science, namely
climate prediction
. When GCMs are used in this way,
they are used to predict a sequence of weather events which, it is assumed, is
representative of those that will occur, or that would occur given prescribed
changes in the factors that influence weather. Among the factors that may change
future climate are the concentration of gases that are involved in the absorption
and emission of radiation in the atmosphere and the nature of the vegetation
covering land areas of the globe.
Thus, climate prediction involves running GCMs for longer periods than weather
predictions, at least for seasons and commonly for several years, decades, centuries,
and even millennia. An initial global field of weather variables must still be provided
when a GCM is used for climate prediction and different patterns of simulated
weather sequences will result for different initiation fields. For this reason, climate
predictions are now usually made with several different initiation fields selected
(perhaps randomly) to be typical from those made available by operational NWP.
The resulting suite of predictions given by a GCM with different initiation fields is
called an
ensemble
. The mean of such an ensemble might be adopted as the average
climate prediction for the GCM, and the variability of the ensemble considered a
measure of the GCM dependent accuracy of the prediction. Thus,
when GCMs are
used for climate prediction the purpose is to simulate the average statistics of weather
across the globe over long periods and to quantify the changes induced in these statistics
in response to prescribed changes in the influences that determine weather.
How do General Circulation Models work?
Sequence of operations
The operational sequence used in running GCMs is illustrated in Fig. 8.2. As
described in the previous section, the first step is to define an initial state for all of
the variables whose evolution is being simulated using 4DDA.
