optimum of the mid-Holocene (6000 years BP) and
the last glacial maximum (LGM, 20 000 years BP).
Most of this work to date has been with atmosphere-only models, but a coupled simulation of the
mid-Holocene has been performed (Hewitt and
Mitchell, 1998). This suggested that ocean dynamics and mixed-layer processes played a role in the
enhanced monsoon rainfall shown by palaeodata.
No coupled model simulations of the LGM have
yet been performed, largely because of the long
time scales required for the ocean circulation to
adjust to the very different state at that time
(including lower sea level and extensive ice sheets).
2.3.6 Coupled model simulation of future
climates
One of the main uses of global coupled models is
for projection of possible future climate change
due to anthropogenic emissions of CO 2 and other
greenhouse gases. This is a vast subject and we
only touch here on a few topics. For a comprehensive review the reader is referred to recent assessment reports of the Intergovernmental Panel on
Climate Change (IPCC, 1995; a new assessment
report (IPCC, 2001) is in preparation at the time
of writing and due for publication in mid-2001).
2.3.6.1 Climate sensitivity and transient
response
The response of the atmosphere to changes in
radiatively active species is determined by a number
of complex feedbacks, especially those involving
water vapour and clouds. The net result of these
feedbacks in a model is often expressed as the ‘climate sensitivity’ of the model, defined as the change
in global mean surface temperature at equilibrium,
in response to a doubling of atmospheric CO 2
concentration. The equilibrium response is usually
determined by running the atmosphere model coupled to a ‘slab’ ocean mixed layer. There is considerable variation in the climate sensitivity between
models, with the current generation of models
having sensitivities between 1.5 and 4.5°C (IPCC,
1995). This reflects uncertainty in the water vapour
and cloud feedbacks. Although the equilibrium climate response is an important benchmark of models, it does not necessarily give a good guide to the
transient response of climate over the next century
to a particular scenario of greenhouse gas increase.
To understand the transient response, the uptake of
heat by the ocean must be taken into account (Section 2.3.5.4), and coupled models are necessary.
Figure 2.3.6b (see Plate 2.3.6b, p. 76) shows the
zonal mean temperature difference between the
decade around 2050 and the decade around 2000,
when the HadCM3 model is forced with a scenario of increasing greenhouse gases. The broad
pattern of deeper penetration of heat at high latitudes is as expected, and similar to the CFC-11
uptake in Fig. 2.3.6a (see Plate 2.3.6a, p. 76), but
beyond this the similarity between the two fields
ends. Not only are the surface flux fields driving
the CFC and temperature changes very different,
but in the greenhouse gas run the ventilation
process changes in response to the changing climate. For example, by 2050 the Labrador Sea convection would have collapsed in the greenhouse
gas run (Wood et al., 1999), and this would contribute to the relatively shallow penetration of heat
anomalies when compared with the CFC. Thus
transient tracer observations give us a valuable
constraint on models by telling us about ventilation processes in the present climate, but they cannot tell us about ventilation in a changed climate.
This means it would be difficult to tune a ‘slab’
mixed-layer model to give reliable projections of
transient climate change.
2.3.6.2 Stability of the thermohaline
circulation
Most climate models suggest that the climate
response to increasing greenhouse gases includes a
warming of surface air temperature and an increase
in high-latitude precipitation. Both these effects
might be expected to inhibit subpolar convection
and so weaken the THC, and indeed most models
show a THC weakening when forced with increasing greenhouse gases. This weakening probably
has an important effect in moderating the climate
response, particularly over northern and western
Europe. However, there is great uncertainty over
the magnitude of the response. For example, in
experiments where CO 2 was increased to four
times preindustrial values, the THC weakened by
about 25% in HadCM3 (Wood et al., 1999), but
collapsed completely in the GFDL model (Manabe
and Stouffer, 1993). Further uncertainty arises
because of the possibility of hysteresis behaviour
of the THC, which could lead to sudden (decadal
time scale) state transitions similar to that which is
believed to have occurred in the Younger Dryas
2.3 Coupled Ocean–Atmosphere Models
93
Wood and Bryan
the last glacial maximum (LGM, 20 000 years BP).
Most of this work to date has been with atmosphere-only models, but a coupled simulation of the
mid-Holocene has been performed (Hewitt and
Mitchell, 1998). This suggested that ocean dynamics and mixed-layer processes played a role in the
enhanced monsoon rainfall shown by palaeodata.
No coupled model simulations of the LGM have
yet been performed, largely because of the long
time scales required for the ocean circulation to
adjust to the very different state at that time
(including lower sea level and extensive ice sheets).
2.3.6 Coupled model simulation of future
climates
One of the main uses of global coupled models is
for projection of possible future climate change
due to anthropogenic emissions of CO 2 and other
greenhouse gases. This is a vast subject and we
only touch here on a few topics. For a comprehensive review the reader is referred to recent assessment reports of the Intergovernmental Panel on
Climate Change (IPCC, 1995; a new assessment
report (IPCC, 2001) is in preparation at the time
of writing and due for publication in mid-2001).
2.3.6.1 Climate sensitivity and transient
response
The response of the atmosphere to changes in
radiatively active species is determined by a number
of complex feedbacks, especially those involving
water vapour and clouds. The net result of these
feedbacks in a model is often expressed as the ‘climate sensitivity’ of the model, defined as the change
in global mean surface temperature at equilibrium,
in response to a doubling of atmospheric CO 2
concentration. The equilibrium response is usually
determined by running the atmosphere model coupled to a ‘slab’ ocean mixed layer. There is considerable variation in the climate sensitivity between
models, with the current generation of models
having sensitivities between 1.5 and 4.5°C (IPCC,
1995). This reflects uncertainty in the water vapour
and cloud feedbacks. Although the equilibrium climate response is an important benchmark of models, it does not necessarily give a good guide to the
transient response of climate over the next century
to a particular scenario of greenhouse gas increase.
To understand the transient response, the uptake of
heat by the ocean must be taken into account (Section 2.3.5.4), and coupled models are necessary.
Figure 2.3.6b (see Plate 2.3.6b, p. 76) shows the
zonal mean temperature difference between the
decade around 2050 and the decade around 2000,
when the HadCM3 model is forced with a scenario of increasing greenhouse gases. The broad
pattern of deeper penetration of heat at high latitudes is as expected, and similar to the CFC-11
uptake in Fig. 2.3.6a (see Plate 2.3.6a, p. 76), but
beyond this the similarity between the two fields
ends. Not only are the surface flux fields driving
the CFC and temperature changes very different,
but in the greenhouse gas run the ventilation
process changes in response to the changing climate. For example, by 2050 the Labrador Sea convection would have collapsed in the greenhouse
gas run (Wood et al., 1999), and this would contribute to the relatively shallow penetration of heat
anomalies when compared with the CFC. Thus
transient tracer observations give us a valuable
constraint on models by telling us about ventilation processes in the present climate, but they cannot tell us about ventilation in a changed climate.
This means it would be difficult to tune a ‘slab’
mixed-layer model to give reliable projections of
transient climate change.
2.3.6.2 Stability of the thermohaline
circulation
Most climate models suggest that the climate
response to increasing greenhouse gases includes a
warming of surface air temperature and an increase
in high-latitude precipitation. Both these effects
might be expected to inhibit subpolar convection
and so weaken the THC, and indeed most models
show a THC weakening when forced with increasing greenhouse gases. This weakening probably
has an important effect in moderating the climate
response, particularly over northern and western
Europe. However, there is great uncertainty over
the magnitude of the response. For example, in
experiments where CO 2 was increased to four
times preindustrial values, the THC weakened by
about 25% in HadCM3 (Wood et al., 1999), but
collapsed completely in the GFDL model (Manabe
and Stouffer, 1993). Further uncertainty arises
because of the possibility of hysteresis behaviour
of the THC, which could lead to sudden (decadal
time scale) state transitions similar to that which is
believed to have occurred in the Younger Dryas
2.3 Coupled Ocean–Atmosphere Models
93
Wood and Bryan