38
Kirk Bryan
Since these calculations took months to carry out with the computers we had available,
this result was extremely frustrating. It was only much later that we realized we were
tilting at windmills. From a modern perspective it was obvious that the models we
were using did not have enough resolution to give a realistic simulation of climate in
a coupled model, without the rigid specification of upper ocean boundary conditions
of the Bryan and Lewis (1979) simulation.
THE 1980s: THE ADOLESCENCE OF OGCMs
We continued to have trouble getting our global coupled model to give realistic simulations. However, we were able to get interesting results for climate models in more
idealized geometries. For the more constrained case of symmetry across the equator Manabe and I were able to calculate equilibrium climates over a wide range of
atmospheric CO 2 levels. For colder climates the thermohaline circulation decreased
in strength, but for warmer climates the thermohaline circulation maintained or even
increased in strength in spite of a reduced meridional temperature gradient. The key
factor was found to be the increase of the thermal expansion coefficient in the ocean
with increasing temperature. Larger thermal expansion allowed north–south density
gradients to be maintained in the warmer climates in spite of reduced meridional
temperature gradients (Manabe and Bryan, 1985). Two other results in idealized geometry anticipated more recent studies of global warming with more general models.
Mike Spelman and I (Bryan and Spelman, 1985) showed that global warming in a
simple coupled climate model would lead to an eventual collapse of the thermohaline
circulation. In a second calculation it was shown that if one hemisphere had a higher
ratio of land compared to ocean, the response to greenhouse warming would be much
greater in the hemisphere with more land. It will soon be possible to test these early
predictions against observations.
Claes Rooth (Figure 3.6) spent a sabbatical year in Princeton in the 1980s. With
his broad interests Claes interacted with almost everyone in the laboratory. His most
important contribution at Princeton was to question the “uniqueness” of our ocean
model solutions. This was something we had not considered. He suggested that the
same upper boundary conditions on an ocean model might allow more than one type
of ocean circulation, an idea going back to Stommel’s (1961) seminal paper on a box
model of the thermohaline circulation. Claes helped to inspire an oft-cited study by
Frank Bryan (1986) using an ocean circulation model in a simple basin extending to
both hemispheres. In spite of identical upper boundary conditions for a model ocean
in the two hemispheres, the preferred numerical solution was asymmetric across the
equator. This gave Manabe and I a clue to what was going wrong in our coupled
ocean–atmosphere models. The ocean-only and the atmosphere-only components of
the model were able to simulate observations, but the looser boundary conditions
in the coupled model allowed other, totally unrealistic solutions. The resolution of
the ocean models was still too crude to provide realistic heat balances at the ocean
Kirk Bryan
Since these calculations took months to carry out with the computers we had available,
this result was extremely frustrating. It was only much later that we realized we were
tilting at windmills. From a modern perspective it was obvious that the models we
were using did not have enough resolution to give a realistic simulation of climate in
a coupled model, without the rigid specification of upper ocean boundary conditions
of the Bryan and Lewis (1979) simulation.
THE 1980s: THE ADOLESCENCE OF OGCMs
We continued to have trouble getting our global coupled model to give realistic simulations. However, we were able to get interesting results for climate models in more
idealized geometries. For the more constrained case of symmetry across the equator Manabe and I were able to calculate equilibrium climates over a wide range of
atmospheric CO 2 levels. For colder climates the thermohaline circulation decreased
in strength, but for warmer climates the thermohaline circulation maintained or even
increased in strength in spite of a reduced meridional temperature gradient. The key
factor was found to be the increase of the thermal expansion coefficient in the ocean
with increasing temperature. Larger thermal expansion allowed north–south density
gradients to be maintained in the warmer climates in spite of reduced meridional
temperature gradients (Manabe and Bryan, 1985). Two other results in idealized geometry anticipated more recent studies of global warming with more general models.
Mike Spelman and I (Bryan and Spelman, 1985) showed that global warming in a
simple coupled climate model would lead to an eventual collapse of the thermohaline
circulation. In a second calculation it was shown that if one hemisphere had a higher
ratio of land compared to ocean, the response to greenhouse warming would be much
greater in the hemisphere with more land. It will soon be possible to test these early
predictions against observations.
Claes Rooth (Figure 3.6) spent a sabbatical year in Princeton in the 1980s. With
his broad interests Claes interacted with almost everyone in the laboratory. His most
important contribution at Princeton was to question the “uniqueness” of our ocean
model solutions. This was something we had not considered. He suggested that the
same upper boundary conditions on an ocean model might allow more than one type
of ocean circulation, an idea going back to Stommel’s (1961) seminal paper on a box
model of the thermohaline circulation. Claes helped to inspire an oft-cited study by
Frank Bryan (1986) using an ocean circulation model in a simple basin extending to
both hemispheres. In spite of identical upper boundary conditions for a model ocean
in the two hemispheres, the preferred numerical solution was asymmetric across the
equator. This gave Manabe and I a clue to what was going wrong in our coupled
ocean–atmosphere models. The ocean-only and the atmosphere-only components of
the model were able to simulate observations, but the looser boundary conditions
in the coupled model allowed other, totally unrealistic solutions. The resolution of
the ocean models was still too crude to provide realistic heat balances at the ocean