compared to the typical timeframe in climate evolution. Thus,
in parallel with the development of general circulation models,
simpler models, more adapted to the study of paleoclimates,
have been developed. The aim was to represent from the
outset the slow-moving components of the climate system, the
ocean and the ice caps, in order to study long-term climate
change (i.e. for time scales in excess of a thousand years). To
develop these models, representation of the rapid components
of the climate system, particularly of the atmosphere, has to be
simplified. In fact, the term ‘simple model’ is misleading
because it refers above all to models which are more efficient
in their use of computing time. Developing a model of this
type is not necessarily ‘simple’ because one cannot simply
retain the basic equations of atmospheric dynamics. The
saving of computation time is generally achieved by establishing parameters for the transport of heat and moisture by the
stationary and/or transient waves (such as, for example, by the
depressions of the mid-latitudes), in other words, by trying to
represent the effect of these phenomena without explicitly
calculating them. This makes it possible to extend the duration
of a time step and to use a coarser spatial resolution. The
number of degrees of freedom of these models lies between
conceptual models (around ten) and general circulation models (*10
5
–10
6
) and they are called ‘EMICs’, ‘Earth system
Models of Intermediate Complexity’ (Claussen et al. 2002). In
fact, the terminology was created long after the development
of the first models in this category. It emerged at a time when
the modelers who specialized in these models decided to join
together to define the specificity of their models compared
with others. These models are characterized by a more complete representation of the climate system than the ‘simple’
ocean-atmosphere models and by a relatively short computing
time compared to the general circulation models, characteristics which allow the evolution of the climate system to be
studied over long time scales and many different scenarios to
be explored. There are many EMICs, corresponding to the
many different ways the representation of the climate system
can be ‘simplified’. It should be noted that some models have
been developed by ‘downgrading’ a general circulation model,
i.e. by reducing its vertical and horizontal resolutions. These
are the most complex models in the EMIC category.
As with general circulation models, climate models of
intermediate complexity can be used to obtain realistic climate simulations or to study the sensitivity to certain forcings, processes or feedbacks of the represented system. In
the following sections, we give examples of the use of
EMICs, both for long-term simulations and for studies
requiring numerous experiments. Experiments of this type
could not have been carried out using general circulation
models given current capacity of computing power. This
shows the complementarity of the two types of models, one
type being useful for its ‘efficiency’, the other for the spatial
and temporal detail in its representation.
Examples of Long-Term Simulations and Studies
of Sensitivity to Forcings
One of the first models of intermediate complexity of the
climate system is the one developed by the Catholic
University of Louvain-la-Neuve. This model includes simplified representations (by latitude and vertically, for the
northern hemisphere) of the atmosphere, the ocean, sea ice
and the polar ice sheet. It was developed specifically to study
the glacial-interglacial cycles, as demonstrated by the first
simulations of Gallée et al. (1992). Since then, Berger et al.
(1998) and Loutre and Berger (2000) have taken up this
model and carried out sensitivity experiments to identify the
respective roles of orbital variations and greenhouse gases in
the last glacial-interglacial cycle.
Figure 25.6 shows a selection of the results of these two
articles in terms of volume of ice (top) and temperature in
the northern hemisphere (bottom). The continuous lines
represent the results of the model forced by both the variations in CO 2 recorded in the ice cores (Jouzel et al. 1993)
and by the variations in the orbital parameters as calculated
by Berger (1978). The initial state of the model is an interglacial state, with no ice sheet in the northern hemisphere.
Over the last 200,000 years, the model simulates two major
glaciations, with a complete freeze-up occurring in steps and
a complete deglaciation following the glacial maximum. The
coldest temperatures are of course simulated for these glacial
maxima. During the last interglacial and the last glaciation
(between 126,000 and 80,000 years), despite a high level of
recorded variability, temperatures remain sufficiently high so
that the frozen-over periods last no longer than 15,000 years.
The red lines correspond to a simulation where the
insolation is constant and equal to the current insolation, and
where CO 2 levels vary in a similar way to the previous
simulation. It can be seen that the volume of ice increases to
about 35 Â 10
15 m
3 and remains at around 30 Â 10
15 m
3
during the rest of the simulation. Although the average
temperature of the northern hemisphere varies in line with
the greenhouse gas forcing, it remains too cold to bring
about deglaciation. The other curves are the result of simulations where CO 2 remains constant (fixed at 210 ppm for
the alternating dash-dot line and at 250 ppm for the dashed
curve) and where variations of the orbital parameters are
taken into account. This time, alternation between glacial
and interglacial periods is obtained, with maximum ice
volumes reached for the same periods as in the reference
simulation at values inversely related to the level of imposed
25 Modeling and Paleoclimatology
335
in parallel with the development of general circulation models,
simpler models, more adapted to the study of paleoclimates,
have been developed. The aim was to represent from the
outset the slow-moving components of the climate system, the
ocean and the ice caps, in order to study long-term climate
change (i.e. for time scales in excess of a thousand years). To
develop these models, representation of the rapid components
of the climate system, particularly of the atmosphere, has to be
simplified. In fact, the term ‘simple model’ is misleading
because it refers above all to models which are more efficient
in their use of computing time. Developing a model of this
type is not necessarily ‘simple’ because one cannot simply
retain the basic equations of atmospheric dynamics. The
saving of computation time is generally achieved by establishing parameters for the transport of heat and moisture by the
stationary and/or transient waves (such as, for example, by the
depressions of the mid-latitudes), in other words, by trying to
represent the effect of these phenomena without explicitly
calculating them. This makes it possible to extend the duration
of a time step and to use a coarser spatial resolution. The
number of degrees of freedom of these models lies between
conceptual models (around ten) and general circulation models (*10
5
–10
6
) and they are called ‘EMICs’, ‘Earth system
Models of Intermediate Complexity’ (Claussen et al. 2002). In
fact, the terminology was created long after the development
of the first models in this category. It emerged at a time when
the modelers who specialized in these models decided to join
together to define the specificity of their models compared
with others. These models are characterized by a more complete representation of the climate system than the ‘simple’
ocean-atmosphere models and by a relatively short computing
time compared to the general circulation models, characteristics which allow the evolution of the climate system to be
studied over long time scales and many different scenarios to
be explored. There are many EMICs, corresponding to the
many different ways the representation of the climate system
can be ‘simplified’. It should be noted that some models have
been developed by ‘downgrading’ a general circulation model,
i.e. by reducing its vertical and horizontal resolutions. These
are the most complex models in the EMIC category.
As with general circulation models, climate models of
intermediate complexity can be used to obtain realistic climate simulations or to study the sensitivity to certain forcings, processes or feedbacks of the represented system. In
the following sections, we give examples of the use of
EMICs, both for long-term simulations and for studies
requiring numerous experiments. Experiments of this type
could not have been carried out using general circulation
models given current capacity of computing power. This
shows the complementarity of the two types of models, one
type being useful for its ‘efficiency’, the other for the spatial
and temporal detail in its representation.
Examples of Long-Term Simulations and Studies
of Sensitivity to Forcings
One of the first models of intermediate complexity of the
climate system is the one developed by the Catholic
University of Louvain-la-Neuve. This model includes simplified representations (by latitude and vertically, for the
northern hemisphere) of the atmosphere, the ocean, sea ice
and the polar ice sheet. It was developed specifically to study
the glacial-interglacial cycles, as demonstrated by the first
simulations of Gallée et al. (1992). Since then, Berger et al.
(1998) and Loutre and Berger (2000) have taken up this
model and carried out sensitivity experiments to identify the
respective roles of orbital variations and greenhouse gases in
the last glacial-interglacial cycle.
Figure 25.6 shows a selection of the results of these two
articles in terms of volume of ice (top) and temperature in
the northern hemisphere (bottom). The continuous lines
represent the results of the model forced by both the variations in CO 2 recorded in the ice cores (Jouzel et al. 1993)
and by the variations in the orbital parameters as calculated
by Berger (1978). The initial state of the model is an interglacial state, with no ice sheet in the northern hemisphere.
Over the last 200,000 years, the model simulates two major
glaciations, with a complete freeze-up occurring in steps and
a complete deglaciation following the glacial maximum. The
coldest temperatures are of course simulated for these glacial
maxima. During the last interglacial and the last glaciation
(between 126,000 and 80,000 years), despite a high level of
recorded variability, temperatures remain sufficiently high so
that the frozen-over periods last no longer than 15,000 years.
The red lines correspond to a simulation where the
insolation is constant and equal to the current insolation, and
where CO 2 levels vary in a similar way to the previous
simulation. It can be seen that the volume of ice increases to
about 35 Â 10
15 m
3 and remains at around 30 Â 10
15 m
3
during the rest of the simulation. Although the average
temperature of the northern hemisphere varies in line with
the greenhouse gas forcing, it remains too cold to bring
about deglaciation. The other curves are the result of simulations where CO 2 remains constant (fixed at 210 ppm for
the alternating dash-dot line and at 250 ppm for the dashed
curve) and where variations of the orbital parameters are
taken into account. This time, alternation between glacial
and interglacial periods is obtained, with maximum ice
volumes reached for the same periods as in the reference
simulation at values inversely related to the level of imposed
25 Modeling and Paleoclimatology
335