with one another, and in principle, it is never possible to
know which scales should be considered and how to represent the neglected scales in a simplified way. A typical
example is the formation of clouds and precipitation. Let’s
take the example of convective clouds (of the cumulonimbus
type), whose core is a rising column of moist air in which the
water vapor condenses as it rises. This ascending column
mixes with the surrounding drier air, and this mixing
depends on many factors (for example, the intensity of the
upward thermal current and wind shear). In order to take
these mixtures into account, it is first necessary to know
precisely the vertical profile of the atmospheric variables in
the vicinity of the column. The turbulent exchanges between
the column and its environment must also be calculated, as
well as the coupling between these turbulent exchanges and
the formation or dissipation of rain drops, hail or snowflakes.
This requires a modeling on a very small scale (a few
hundred nanometers to a few meters), which is not possible
with global models. Therefore, a simplified model must be
developed based only on large scale variables which reproduce the effect of unresolved small scale processes. This type
of modeling, called parameterization, is based on important
simplifications that nevertheless require a thorough physical
analysis and an in-depth understanding of the phenomena.
The aim is to obtain a simplified model that is not only as
accurate as possible, but also justified and well understood.
There are many parameterizations in a climate model
which can be directly related with the atmospheric circulation (for example gravity waves and orographic effects), the
calculation of radiative exchanges, deep convection, or
boundary layer phenomena. Many of these parameterizations play a key role in the water cycle, the formation of
clouds and their radiative properties, precipitation, heat and
water fluxes on the surface of continents or oceans, among
others. All these phenomena have an impact on the simulation of the current climate and as there are strong interactions between them, it is often very difficult to identify the
precise role of each of the parameterizations on the simulation of these phenomena, and in particular to understand
why some of them are poorly simulated.
Parameterizations also play a very important role in the
climate response to different forcings, and in the simulation
of past and future climate changes. One example is the
simulation of clouds. Clouds exert two opposing effects on
the terrestrial radiative balance: on the one hand, they reflect
part of the solar radiation, and on the other, they absorb
infrared radiation and thus contribute to the greenhouse
effect. The relative importance of these two effects depends
on many factors, in particular, cloud altitude. Over the past
twenty years, we have learned that, on average, the first
effect outweighs the second, and therefore that clouds have a
cooling effect on the climate, especially low-lying clouds,
because they have little impact on infrared radiation.
However, this does not explain the role that clouds could
play in global warming. Depending on how their properties
change, clouds may attenuate or, on the contrary, amplify,
global warming. The physical formation mechanisms
involve so many processes and spatial scales (from a
micrometer to a thousand kilometers), and their radiative
properties depend on so many factors that it is impossible to
conclude on the basis of theory, simple reasoning or analysis
of available observations, how they will evolve in the future.
Simulation of the Current Climate and Recent
Changes
Analysis of simulated climate combined to the comparison
of model results with observations is an important step to
establish the reliability of climate models. The aim is to
evaluate not only the mean climate, but also the climate
variability at different time scales (from a few days to a few
decades) as well as recent climate changes. Unless otherwise
indicated, this section presents the simulated climate characteristics from the twenty AOGCMs that contributed to the
preparation of the Fifth IPCC Assessment Report (IPCC
2013).
Mean Climate
The difference in mean insolation is what causes the temperature difference between the equator and the poles. This
difference is the main driver of atmospheric and oceanic
circulations, which act to reduce the equator-pole temperature gradient. It is also influenced by the presence of clouds,
reflective surfaces (snow, glaciers, sea ice), large mountain
ranges and the topography of the ocean. Latitudinal temperature variations are thus a key criterion for evaluation of
climate models. All the models simulate this strong
equator-pole gradient: the simulated temperature is 25 °C at
the equator, −20 °C at the North Pole and −40 °C at the
South Pole, which agrees with observations. However, the
models also show significant biases over Antarctica,
Greenland, and over large mountain ranges in general, such
as the Himalayas. These are due to an approximate representation of the topography (due to the limited spatial resolution of the models) and a poor representation of turbulent
exchanges under conditions of strong thermal stability. Over
the oceans, there is a warm bias on the eastern coasts caused
by a poor representation of stratus clouds observed in these
regions.
The annual variation in solar radiation is the strongest
energy ‘disruption’, apart from the diurnal cycle, to which
the surface of the Earth is subjected. The observed seasonal
temperature cycle is generally well reproduced by the
models: it is higher at high latitudes than at low ones (30 °C
vs. 5 °C, essentially reflecting the seasonal amplitude of
31 From the Climates of the Past to the Climates …
461
know which scales should be considered and how to represent the neglected scales in a simplified way. A typical
example is the formation of clouds and precipitation. Let’s
take the example of convective clouds (of the cumulonimbus
type), whose core is a rising column of moist air in which the
water vapor condenses as it rises. This ascending column
mixes with the surrounding drier air, and this mixing
depends on many factors (for example, the intensity of the
upward thermal current and wind shear). In order to take
these mixtures into account, it is first necessary to know
precisely the vertical profile of the atmospheric variables in
the vicinity of the column. The turbulent exchanges between
the column and its environment must also be calculated, as
well as the coupling between these turbulent exchanges and
the formation or dissipation of rain drops, hail or snowflakes.
This requires a modeling on a very small scale (a few
hundred nanometers to a few meters), which is not possible
with global models. Therefore, a simplified model must be
developed based only on large scale variables which reproduce the effect of unresolved small scale processes. This type
of modeling, called parameterization, is based on important
simplifications that nevertheless require a thorough physical
analysis and an in-depth understanding of the phenomena.
The aim is to obtain a simplified model that is not only as
accurate as possible, but also justified and well understood.
There are many parameterizations in a climate model
which can be directly related with the atmospheric circulation (for example gravity waves and orographic effects), the
calculation of radiative exchanges, deep convection, or
boundary layer phenomena. Many of these parameterizations play a key role in the water cycle, the formation of
clouds and their radiative properties, precipitation, heat and
water fluxes on the surface of continents or oceans, among
others. All these phenomena have an impact on the simulation of the current climate and as there are strong interactions between them, it is often very difficult to identify the
precise role of each of the parameterizations on the simulation of these phenomena, and in particular to understand
why some of them are poorly simulated.
Parameterizations also play a very important role in the
climate response to different forcings, and in the simulation
of past and future climate changes. One example is the
simulation of clouds. Clouds exert two opposing effects on
the terrestrial radiative balance: on the one hand, they reflect
part of the solar radiation, and on the other, they absorb
infrared radiation and thus contribute to the greenhouse
effect. The relative importance of these two effects depends
on many factors, in particular, cloud altitude. Over the past
twenty years, we have learned that, on average, the first
effect outweighs the second, and therefore that clouds have a
cooling effect on the climate, especially low-lying clouds,
because they have little impact on infrared radiation.
However, this does not explain the role that clouds could
play in global warming. Depending on how their properties
change, clouds may attenuate or, on the contrary, amplify,
global warming. The physical formation mechanisms
involve so many processes and spatial scales (from a
micrometer to a thousand kilometers), and their radiative
properties depend on so many factors that it is impossible to
conclude on the basis of theory, simple reasoning or analysis
of available observations, how they will evolve in the future.
Simulation of the Current Climate and Recent
Changes
Analysis of simulated climate combined to the comparison
of model results with observations is an important step to
establish the reliability of climate models. The aim is to
evaluate not only the mean climate, but also the climate
variability at different time scales (from a few days to a few
decades) as well as recent climate changes. Unless otherwise
indicated, this section presents the simulated climate characteristics from the twenty AOGCMs that contributed to the
preparation of the Fifth IPCC Assessment Report (IPCC
2013).
Mean Climate
The difference in mean insolation is what causes the temperature difference between the equator and the poles. This
difference is the main driver of atmospheric and oceanic
circulations, which act to reduce the equator-pole temperature gradient. It is also influenced by the presence of clouds,
reflective surfaces (snow, glaciers, sea ice), large mountain
ranges and the topography of the ocean. Latitudinal temperature variations are thus a key criterion for evaluation of
climate models. All the models simulate this strong
equator-pole gradient: the simulated temperature is 25 °C at
the equator, −20 °C at the North Pole and −40 °C at the
South Pole, which agrees with observations. However, the
models also show significant biases over Antarctica,
Greenland, and over large mountain ranges in general, such
as the Himalayas. These are due to an approximate representation of the topography (due to the limited spatial resolution of the models) and a poor representation of turbulent
exchanges under conditions of strong thermal stability. Over
the oceans, there is a warm bias on the eastern coasts caused
by a poor representation of stratus clouds observed in these
regions.
The annual variation in solar radiation is the strongest
energy ‘disruption’, apart from the diurnal cycle, to which
the surface of the Earth is subjected. The observed seasonal
temperature cycle is generally well reproduced by the
models: it is higher at high latitudes than at low ones (30 °C
vs. 5 °C, essentially reflecting the seasonal amplitude of
31 From the Climates of the Past to the Climates …
461