points. Traditional in-situ observations require the
deployment of instruments at sea, a process that
requires enormous logistical planning and is subject to all the uncertainties and difficulty of work
in a hostile environment. Each ocean data set,
therefore, represents a considerable investment of
time and effort on the part of the investigators
involved, and to extract the maximum value from
the data set much work is required after the observations have been made. A single research cruise
provides a detailed view of the state of the ocean
at a particular place and time.
Climate modellers, on the other hand, begin
with a global view. As discussed in Section 2.3.3,
the large-scale, long-term mean transports of heat
and water by the ocean and atmosphere are a fundamental feature of the climate system that models
must attempt to reproduce. This means that climate modellers have a particular interest in the
development of integrated, consistent global syntheses of a large number of individual data sets, to
produce estimates of these quantities. The use of
the vast WOCE data set to produce such a global
picture for the 1990s, with reduced uncertainties
compared with our previous knowledge, will be an
important test of the success of WOCE (see especially Chapters 6.1, 6.2 and 7.1).
The above does not mean that climate modellers are interested only in large-scale, basin mean
transports; climate change and variability are
effected by many smaller-scale mechanisms. For
example, the response of the thermohaline circulation to changes in surface forcing is likely to be
carried from the deep convection regions to the
rest of the ocean via sill overflows and boundary
processes (Böning and Semtner, Chapter 2.2).
Improved knowledge of the dynamical and thermodynamical processes by which the ocean achieves its
large-scale transports is needed so that we can
both diagnose any deficiencies in model transports
and increase confidence that our models will
respond correctly to any applied perturbations
(e.g. greenhouse gas increase).
The nature of decadal climate variability and
change means that the representativeness of the
existing ocean database (especially the WOCE
‘snapshot’) of the long-term mean state of the
ocean is an important question. Ultimately this can
only be addressed by a long-term programme of
repeated observations (see Chapters 7.3 and 7.4). In
the meantime, we must rely on models, combined
with judicious use of the historical database, in
order to increase our understanding of decadal
changes in the ocean. For example, Wong et al.
(1999) interpret a large amount of historical
hydrographic data to show changes in the properties of intermediate water masses during the
second half of the twentieth century. Quantitatively
similar changes are seen in the HadCM3 model by
Banks et al. (2000), who suggest that the observed
freshening of Subantarctic Mode Water is most
likely a result of anthropogenic climate change,
whereas changes seen in other water masses may
be due to internal climate variability. While such
results must be treated cautiously, as the models
still have many deficiencies, studies such as these
have the potential to provide useful input to the
design of long-term observing networks.
2.3.8 Summary and future developments
Considerable progress has been made in the development of coupled climate models during the
WOCE period. In 1990, at the start of WOCE, only
a few coupled models were in existence, and only a
few runs had been made. During the intervening
decade, in parallel with the WOCE observational
programme, many groups have developed models;
in 1996, control runs from 18 coupled models were
contributed to the first phase of the Coupled Model
Intercomparison Project (CMIP). Model performance has improved, and models are now emerging
that produce stable, credible simulations of many
aspects of present climate. A particular success is
that some models can now reproduce the observed
large-scale heat balances of the climate system,
without the need for unphysical flux adjustments.
There remain many aspects of the model simulations that are inconsistent with or untested
against observations, and the ocean components of
coupled models still suffer from rather coarse resolution. However, it must be remembered that the
models are largely used to study climate variability
and change, and that a large error in, say, the
properties of a particular deep water mass, may
not have a large impact on atmospheric climate on
the decadal to century time scale. In fact the extent
and mechanisms by which the ocean is playing an
active role in climate on these time scales is an
active area of current research.
Future model developments are likely to include a
move to higher resolution (including eddy-permitting
2.3 Coupled Ocean–Atmosphere Models
95
Wood and Bryan
deployment of instruments at sea, a process that
requires enormous logistical planning and is subject to all the uncertainties and difficulty of work
in a hostile environment. Each ocean data set,
therefore, represents a considerable investment of
time and effort on the part of the investigators
involved, and to extract the maximum value from
the data set much work is required after the observations have been made. A single research cruise
provides a detailed view of the state of the ocean
at a particular place and time.
Climate modellers, on the other hand, begin
with a global view. As discussed in Section 2.3.3,
the large-scale, long-term mean transports of heat
and water by the ocean and atmosphere are a fundamental feature of the climate system that models
must attempt to reproduce. This means that climate modellers have a particular interest in the
development of integrated, consistent global syntheses of a large number of individual data sets, to
produce estimates of these quantities. The use of
the vast WOCE data set to produce such a global
picture for the 1990s, with reduced uncertainties
compared with our previous knowledge, will be an
important test of the success of WOCE (see especially Chapters 6.1, 6.2 and 7.1).
The above does not mean that climate modellers are interested only in large-scale, basin mean
transports; climate change and variability are
effected by many smaller-scale mechanisms. For
example, the response of the thermohaline circulation to changes in surface forcing is likely to be
carried from the deep convection regions to the
rest of the ocean via sill overflows and boundary
processes (Böning and Semtner, Chapter 2.2).
Improved knowledge of the dynamical and thermodynamical processes by which the ocean achieves its
large-scale transports is needed so that we can
both diagnose any deficiencies in model transports
and increase confidence that our models will
respond correctly to any applied perturbations
(e.g. greenhouse gas increase).
The nature of decadal climate variability and
change means that the representativeness of the
existing ocean database (especially the WOCE
‘snapshot’) of the long-term mean state of the
ocean is an important question. Ultimately this can
only be addressed by a long-term programme of
repeated observations (see Chapters 7.3 and 7.4). In
the meantime, we must rely on models, combined
with judicious use of the historical database, in
order to increase our understanding of decadal
changes in the ocean. For example, Wong et al.
(1999) interpret a large amount of historical
hydrographic data to show changes in the properties of intermediate water masses during the
second half of the twentieth century. Quantitatively
similar changes are seen in the HadCM3 model by
Banks et al. (2000), who suggest that the observed
freshening of Subantarctic Mode Water is most
likely a result of anthropogenic climate change,
whereas changes seen in other water masses may
be due to internal climate variability. While such
results must be treated cautiously, as the models
still have many deficiencies, studies such as these
have the potential to provide useful input to the
design of long-term observing networks.
2.3.8 Summary and future developments
Considerable progress has been made in the development of coupled climate models during the
WOCE period. In 1990, at the start of WOCE, only
a few coupled models were in existence, and only a
few runs had been made. During the intervening
decade, in parallel with the WOCE observational
programme, many groups have developed models;
in 1996, control runs from 18 coupled models were
contributed to the first phase of the Coupled Model
Intercomparison Project (CMIP). Model performance has improved, and models are now emerging
that produce stable, credible simulations of many
aspects of present climate. A particular success is
that some models can now reproduce the observed
large-scale heat balances of the climate system,
without the need for unphysical flux adjustments.
There remain many aspects of the model simulations that are inconsistent with or untested
against observations, and the ocean components of
coupled models still suffer from rather coarse resolution. However, it must be remembered that the
models are largely used to study climate variability
and change, and that a large error in, say, the
properties of a particular deep water mass, may
not have a large impact on atmospheric climate on
the decadal to century time scale. In fact the extent
and mechanisms by which the ocean is playing an
active role in climate on these time scales is an
active area of current research.
Future model developments are likely to include a
move to higher resolution (including eddy-permitting
2.3 Coupled Ocean–Atmosphere Models
95
Wood and Bryan