The Atmosphere and Climate Modeling - Introduction to Carbon Capture and Sequestration - page 82

how an increasing number of these interactions have been included in the
subsequent climate models [2.2]. Early climate models only considered
sunshine, CO 2 , and rain. In these models, the cooling effects of clouds,
and many other effects were ignored. Over the years, more interactions, as
described in Figure 2.5.4 , have been included in the models. The most
recent additions are the coupling of the atmosphere with the carbon cycle
of the earth (see the next chapter) and a coupling with vegetation.
At this point it is important to note that improving climate modeling
is not merely a matter of increasing grid resolution by using bigger com-
puters and adding more interactions. For some phenomena, such as
cloud formation or convection of the atmosphere, we lack a suffi cient
fundamental understanding of the physics, and hence lack the equations
to include these phenomena in our climate models.
Climate models: a world without carbon emissions
Can these climate models accurately predict the climate? Before looking
at some actual predictions, we would like to emphasize that climate sci-
ence is built on very solid foundations. The equations underlying these
climate models are very well understood. Everybody who enters an air-
plane should realize that most of the testing of that airplane, including
imposition of some of the most extreme conditions, is done with the very
same equations that we use to predict the climate.
An important test of climate models is the comparison of model pre-
diction to historical data. For example, can these models correctly pre-
dict the seasonal variations of the climates in different parts of the world?
A more stringent test is whether these models can correctly predict the
response of the climate to major perturbations. For example, major vol-
canic eruptions such as that of Mt. Pinatubo in 1991 increase the amount
of dust in the atmosphere, which in turn leads to a faster increase in
nighttime compared to daytime temperatures, a larger degree of warming
in the Arctic, and small, short-term global cooling with subsequent
recovery.
The results of such a comparison between climate data and models
are illustrated in Figure 2.5.5 [2.2]. The typical protocol in this fi eld of
research is for multiple computer models hosted by different groups to
consider the same climate perturbation. Each of these models incorpo-
rates different computer code implementations and different assumptions.