Modeling Ocean Circulation
41
for an ocean circulation model. While an isopycnal model is in principle the most
accurate architecture, the practical implementation is actually quite complicated due
to the complex density equation of sea water. Two decades after these two new
architectures were originally introduced, the full potential of either one has not been
completely realized for simulating ocean circulation and climate. Much more effort
is needed in this area.
In the 1980s the National Center for Atmospheric Research became a major
laboratory for ocean circulation modeling. GFDL alumni Bill Holland and later Bert
Semtner and Frank Bryan joined the NCAR staff. Peter Rhines and Francis Bretherton,
while not directly involved in large-scale ocean modeling, had an important influence.
Bill Holland and Frank Bryan led an effort to follow up early studies (Holland and
Lin, 1975) of the North Atlantic with geostrophic models with more ambitious, “eddy
permitting” calculations using the primitive equations in a version of Cox’s model
(Bryan and Holland, 1989). At about the same time, Bert Semtner with Bob Chervin
(Semtner and Chervin, 1992) carried out similar “eddy permitting” calculations for
the entire World Ocean in the most ambitious application of computers up to that
time. The North Atlantic studies of Holland, and the World Ocean study of Semtner
and Chervin set a pattern, which have been continued up to the present with higher
and higher spatial resolution as more capable computers become available.
SUMMARY
This account only covers three decades, stopping at the nineties and is focused on
Princeton and the Weather Bureau. By the 1990s ocean modeling became a truly
international undertaking. Ocean and ocean–climate modeling was being carried out
in so many places in this country and abroad that no one person can give a proper
account. For recent developments in ocean circulation modeling, the reader is referred
to a many-authored summary by Griffies et al. (2000).
As impressive as the recent progress in ocean modeling is, there might be questions in the atmospheric community as to why it has taken so long to reach this point.
In the parlance of applied mathematics, the ocean circulation is a classic “stiff” system
relative to the atmosphere. This means that the ocean has a wide range of spatial and
temporal time-scales, one or two orders of magnitude greater than the atmosphere.
Resolving all these space and time scales entails a great computational burden. Ocean
modeling is as critically dependent on advances in computing technology as on the
imagination and skill of modelers. As mighty as today’s supercomputers seem to be,
ocean modelers are still waiting for computers powerful enough to allow them to
model the coupled ocean–atmosphere system in all its detail. Much of the present
account dwells on the technical aspects of developing successive generations of numerical models and the spawning of new applications of the models to climate and
geochemistry. Since the ocean is a complicated continuum, it is very hard to develop
quantitative theories for its behavior and relationships to actual data. The power of
41
for an ocean circulation model. While an isopycnal model is in principle the most
accurate architecture, the practical implementation is actually quite complicated due
to the complex density equation of sea water. Two decades after these two new
architectures were originally introduced, the full potential of either one has not been
completely realized for simulating ocean circulation and climate. Much more effort
is needed in this area.
In the 1980s the National Center for Atmospheric Research became a major
laboratory for ocean circulation modeling. GFDL alumni Bill Holland and later Bert
Semtner and Frank Bryan joined the NCAR staff. Peter Rhines and Francis Bretherton,
while not directly involved in large-scale ocean modeling, had an important influence.
Bill Holland and Frank Bryan led an effort to follow up early studies (Holland and
Lin, 1975) of the North Atlantic with geostrophic models with more ambitious, “eddy
permitting” calculations using the primitive equations in a version of Cox’s model
(Bryan and Holland, 1989). At about the same time, Bert Semtner with Bob Chervin
(Semtner and Chervin, 1992) carried out similar “eddy permitting” calculations for
the entire World Ocean in the most ambitious application of computers up to that
time. The North Atlantic studies of Holland, and the World Ocean study of Semtner
and Chervin set a pattern, which have been continued up to the present with higher
and higher spatial resolution as more capable computers become available.
SUMMARY
This account only covers three decades, stopping at the nineties and is focused on
Princeton and the Weather Bureau. By the 1990s ocean modeling became a truly
international undertaking. Ocean and ocean–climate modeling was being carried out
in so many places in this country and abroad that no one person can give a proper
account. For recent developments in ocean circulation modeling, the reader is referred
to a many-authored summary by Griffies et al. (2000).
As impressive as the recent progress in ocean modeling is, there might be questions in the atmospheric community as to why it has taken so long to reach this point.
In the parlance of applied mathematics, the ocean circulation is a classic “stiff” system
relative to the atmosphere. This means that the ocean has a wide range of spatial and
temporal time-scales, one or two orders of magnitude greater than the atmosphere.
Resolving all these space and time scales entails a great computational burden. Ocean
modeling is as critically dependent on advances in computing technology as on the
imagination and skill of modelers. As mighty as today’s supercomputers seem to be,
ocean modelers are still waiting for computers powerful enough to allow them to
model the coupled ocean–atmosphere system in all its detail. Much of the present
account dwells on the technical aspects of developing successive generations of numerical models and the spawning of new applications of the models to climate and
geochemistry. Since the ocean is a complicated continuum, it is very hard to develop
quantitative theories for its behavior and relationships to actual data. The power of