Observational Requirements for Modeling of Global...
41
therefore warm more quickly than the western warm pool as greenhouse gas concentrations
increase. Furthermore, to the extent that highly reflective anvil clouds in the west dampen
the rate of warming of the ocean surface there, any increase in anvil ice water content might
result in a smaller perturbation of the surface energy budget in the west than in the east.
Finally, preferential increases in evaporative cooling in the warmer west Pacific and in downward
longwave heating in the less opaque atmospheric boundary layer over the east Pacific might
also lead to a stronger warming trend in the eastern ocean.
Several of these effects are nonlinear: If the longitudinal SST gradient decreases, the Walker
cell will weaken, thus reducing both the trade winds that maintain the ocean mixed layer depth
gradient and the convergence that supplies water to the cumulus anvils. The latter implies that
the climate change in tropical cloud forcing will depend not only on the thermodynamics of
clouds and convection but also on the regional pattern of SST change. In idealized experiments
with prescribed SST changes in the GISS GCM, the global climate sensitivity is largely determined by changes in tropical anvil clouds (Del Genio et al., 1996). When the SST changes are
applied uniformly, maintaining the current SST gradient, large increases in anvil cloud cover
and ice water content occur, substantially reducing the global mean sunlight absorbed by the
planet and generating a net negative cloud feedback and very low climate sensitivity. However,
if the same basin-average SST change is applied so as to reduce the longitudinal gradient in
the warmer climate, the increase in anvil cloudiness and reflectance is smaller, and net positive
cloud feedback and relatively high global climate sensitivity result.
The latter scenario appears to be more relevant to actual greenhouse climate change: In a
doubled CO2 experiment with the most recent version of the GISS atmospheric GCM coupled
to a mixed layer ocean with fixed ocean heat transports, the tropical Pacific SST gradient
weakens by O.5-l.0°C across the basin, and the deep convection pattern shifts eastward in a
manner reminiscent of ENSO (Fig. 2.6). The climate sensitivity of this model is 3.5°C, despite
parameterized detrainment of cumulus condensate into anvil clouds and interactive cloud radiative properties. Thus, the concept of a tropical cirrus "thermostat" limiting the magnitude
of climate change does not appear to be valid, because changes in clouds and dynamics interact
to lessen the impact of local SST increases.
One direct implication of ocean dynamics concerns the latitudinal profile of greenhouse warming. In equilibrium, polar amplification is predicted by all mixed-layer models. But if high
latitude rainfall and ice melting were to increase sufficiently to slow down the North Atlantic
thermohaline circulation, the climate system might make a transition to a different state (Broecker, 1991), and the North Atlantic could conceivably cool for several decades. Thus, while the
rest of the globe was beginning to experience warming, western Europe, which depends on the
Gulf Stream to maintain a comfortable climate, might actually cool temporarily instead. Even
without a transition, if land warms faster than ocean, a tendency for oceanic high pressure
anomalies might result. The associated clockwise flow might cause western continental regions
to warm more slowly than their eastern counterparts (Hansen et al., 1988).
Another aspect of ENSO is the teleconnection patterns that propagate thousands of kilometers
from the forcing region in the tropical Pacific (Horel and Wallace, 1981). These are known
to create regional climate anomalies in the Northern midlatitudes. The specific teleconnection
pattern differs for each ENSO depending on the magnitude and location of the peak SST
anomaly. If the spatial pattern of SST change in a greenhouse warming resembles ENSO, then
the remote effect in midlatitudes might well be partly determined by the relevant teleconnection
pattern, i.e., one manifestation of regional climate change might be an increase in the frequency
of occurrence of spatial anomalies characteristic of the warm phase of ENSO. Since clouds as a
general rule prefer low pressure, one might expect regional cloud cover increases and decreases
to correlate to some extent with regional low and high pressure anomalies, respectively.
41
therefore warm more quickly than the western warm pool as greenhouse gas concentrations
increase. Furthermore, to the extent that highly reflective anvil clouds in the west dampen
the rate of warming of the ocean surface there, any increase in anvil ice water content might
result in a smaller perturbation of the surface energy budget in the west than in the east.
Finally, preferential increases in evaporative cooling in the warmer west Pacific and in downward
longwave heating in the less opaque atmospheric boundary layer over the east Pacific might
also lead to a stronger warming trend in the eastern ocean.
Several of these effects are nonlinear: If the longitudinal SST gradient decreases, the Walker
cell will weaken, thus reducing both the trade winds that maintain the ocean mixed layer depth
gradient and the convergence that supplies water to the cumulus anvils. The latter implies that
the climate change in tropical cloud forcing will depend not only on the thermodynamics of
clouds and convection but also on the regional pattern of SST change. In idealized experiments
with prescribed SST changes in the GISS GCM, the global climate sensitivity is largely determined by changes in tropical anvil clouds (Del Genio et al., 1996). When the SST changes are
applied uniformly, maintaining the current SST gradient, large increases in anvil cloud cover
and ice water content occur, substantially reducing the global mean sunlight absorbed by the
planet and generating a net negative cloud feedback and very low climate sensitivity. However,
if the same basin-average SST change is applied so as to reduce the longitudinal gradient in
the warmer climate, the increase in anvil cloudiness and reflectance is smaller, and net positive
cloud feedback and relatively high global climate sensitivity result.
The latter scenario appears to be more relevant to actual greenhouse climate change: In a
doubled CO2 experiment with the most recent version of the GISS atmospheric GCM coupled
to a mixed layer ocean with fixed ocean heat transports, the tropical Pacific SST gradient
weakens by O.5-l.0°C across the basin, and the deep convection pattern shifts eastward in a
manner reminiscent of ENSO (Fig. 2.6). The climate sensitivity of this model is 3.5°C, despite
parameterized detrainment of cumulus condensate into anvil clouds and interactive cloud radiative properties. Thus, the concept of a tropical cirrus "thermostat" limiting the magnitude
of climate change does not appear to be valid, because changes in clouds and dynamics interact
to lessen the impact of local SST increases.
One direct implication of ocean dynamics concerns the latitudinal profile of greenhouse warming. In equilibrium, polar amplification is predicted by all mixed-layer models. But if high
latitude rainfall and ice melting were to increase sufficiently to slow down the North Atlantic
thermohaline circulation, the climate system might make a transition to a different state (Broecker, 1991), and the North Atlantic could conceivably cool for several decades. Thus, while the
rest of the globe was beginning to experience warming, western Europe, which depends on the
Gulf Stream to maintain a comfortable climate, might actually cool temporarily instead. Even
without a transition, if land warms faster than ocean, a tendency for oceanic high pressure
anomalies might result. The associated clockwise flow might cause western continental regions
to warm more slowly than their eastern counterparts (Hansen et al., 1988).
Another aspect of ENSO is the teleconnection patterns that propagate thousands of kilometers
from the forcing region in the tropical Pacific (Horel and Wallace, 1981). These are known
to create regional climate anomalies in the Northern midlatitudes. The specific teleconnection
pattern differs for each ENSO depending on the magnitude and location of the peak SST
anomaly. If the spatial pattern of SST change in a greenhouse warming resembles ENSO, then
the remote effect in midlatitudes might well be partly determined by the relevant teleconnection
pattern, i.e., one manifestation of regional climate change might be an increase in the frequency
of occurrence of spatial anomalies characteristic of the warm phase of ENSO. Since clouds as a
general rule prefer low pressure, one might expect regional cloud cover increases and decreases
to correlate to some extent with regional low and high pressure anomalies, respectively.