178
ocean and conversely, when something changes in the atmosphere and is
reflected in the surface fluxes of heat and momentum, the ocean responds
which then adjusts the atmosphere ad infinitum. The system is mutually
and inextricable coupled.
It has become common, however, for mostly practical reasons, to force
the ocean with simplified boundary conditions and examine the consequent
variability with the understanding that these simplified boundary conditions do not have the full dynamics of the coupled system. The most
common forcing is "restoring boundary conditions", on both temperature
and salinity, diagnosing the resulting fresh water fluxes, and then forcing with restoring boundary conditions on temperature and constant flux
boundary conditions on salinity, the so called "mixed" boundary conditions
introduced by F. Bryan (1986). The argument is given that these mixed
boundary conditions are more appropriate than purely restoring boundary conditions since the temperature of the atmosphere does respond to
SST while precipitation in the atmosphere does not respond to sea surface
salinity.
There have been raging discussions about the suitability of mixed boundary conditions for the correct simulation of the ocean. We will take the
point of view that there is no alternative to coupling the ocean to a real
atmosphere, but that simplicity is its own reward: to the extent that the
simplified boundary conditions gives the correct fluxes, we will consider
those boundary conditions reasonable. When variability is found under
mixed boundary conditions, it will be taken as smoke: whether or not
there is fire will be left to more complex coupled atmosphere-ocean models.
It should be pointed out that in choosing a simple restoring boundary
condition and then diagnosing the fresh water flux to get mixed boundary conditions, the restoring time is crucial. As Willebrand (1993) and
Marotzke (1994) have pointed out, the restoring time is spatially scale dependent and therefore choosing a single restoring time as an artifice implies
choosing a single spatial scale and therefore a single process of interest. We
will consider that the most important process is the cooling of water in the
Gulf Stream extension to become dense enough to sink in the GreenlandIceland-Norwegian (GIN) seas and perhaps also in the Labrador sea. The
horizontal scale is of order 1000 km and the appropriate restoring time scale
is about 50 days. If the time scale is made long enough, the water cannot
lose enough heat through surface fluxes and arrives at its final destination
ocean and conversely, when something changes in the atmosphere and is
reflected in the surface fluxes of heat and momentum, the ocean responds
which then adjusts the atmosphere ad infinitum. The system is mutually
and inextricable coupled.
It has become common, however, for mostly practical reasons, to force
the ocean with simplified boundary conditions and examine the consequent
variability with the understanding that these simplified boundary conditions do not have the full dynamics of the coupled system. The most
common forcing is "restoring boundary conditions", on both temperature
and salinity, diagnosing the resulting fresh water fluxes, and then forcing with restoring boundary conditions on temperature and constant flux
boundary conditions on salinity, the so called "mixed" boundary conditions
introduced by F. Bryan (1986). The argument is given that these mixed
boundary conditions are more appropriate than purely restoring boundary conditions since the temperature of the atmosphere does respond to
SST while precipitation in the atmosphere does not respond to sea surface
salinity.
There have been raging discussions about the suitability of mixed boundary conditions for the correct simulation of the ocean. We will take the
point of view that there is no alternative to coupling the ocean to a real
atmosphere, but that simplicity is its own reward: to the extent that the
simplified boundary conditions gives the correct fluxes, we will consider
those boundary conditions reasonable. When variability is found under
mixed boundary conditions, it will be taken as smoke: whether or not
there is fire will be left to more complex coupled atmosphere-ocean models.
It should be pointed out that in choosing a simple restoring boundary
condition and then diagnosing the fresh water flux to get mixed boundary conditions, the restoring time is crucial. As Willebrand (1993) and
Marotzke (1994) have pointed out, the restoring time is spatially scale dependent and therefore choosing a single restoring time as an artifice implies
choosing a single spatial scale and therefore a single process of interest. We
will consider that the most important process is the cooling of water in the
Gulf Stream extension to become dense enough to sink in the GreenlandIceland-Norwegian (GIN) seas and perhaps also in the Labrador sea. The
horizontal scale is of order 1000 km and the appropriate restoring time scale
is about 50 days. If the time scale is made long enough, the water cannot
lose enough heat through surface fluxes and arrives at its final destination