401
at a depth of 2-3 km manage to slowly heat up the subsurface Southern
Ocean until it is destabilised. Convection then establishes fairly quickly
all around Antarctica and reaches full strength within less than 100 years.
During the following 200 years convection slowly decreases again due to
the action of net precipitation in that area. The result is an asymmetric
oscillation pattern. Pierce et al. (1995) show that this oscillation is only
possible when a nonlinear equation of state is used.
The above mechanism is similar to the flushes found in many other models
(Marotzke 1989; Wright and Stocker 1991; Weaver and Sarachik 1991b;
Winton and Sarachik 1993) whose characteristic feature is a decoupling of
the lower ocean from the surface. In contrast to these earlier studies where
an entire basin had to be destabilized by diffusion and hence evolved on
much longer time scales of 0(10 3 - 10 4 yrs), destabilisation here is due to
advection of warmer NADW, i.e. an "efficient" process with a time scale
of a few hundred years.
3.6 Other Mechanisms
Internal variability on the interdecadal time scale is found in simple ocean
models when a thermodynamic sea ice component is included. Formation of
a sea ice cover releases salt into the water column and induces convection.
By altering density gradients this increases the thermohaline circulation
advecting more heat northward which then melts the sea ice cover. Melting
increases the surface density which tends to decrease the thermohaline
circulation leading to a net cooling. This feedback loop was found in a
2-dimensional ocean circulation model coupled to a thermodynamic sea
ice model (Yang and Neelin 1993). Zhang et al. (1995) showed that this
process is also present in a 3-dimensional OGCM coupled to an ice model
and found the destabilizing influence of brine rejection less important than
that of slow heating below the ice cover due to thermal insulation.
In the previous section mainly oceanic processes localized in or related to
the Atlantic have been discussed. However, there are also recent model
examples of interdecadal cycles in the Pacific and in the atmosphere alone.
James and James (1992) find in their atmospheric circulation model variability on time scales of 10 to 40 years and associate them with changes in
the structure of the subtropical and mid-latitude jets.
Von Storch (1994), on the other hand, uses a coupled AjOGCM and identifies two types of low-frequency variability. The atmospheric fluctuations
at a depth of 2-3 km manage to slowly heat up the subsurface Southern
Ocean until it is destabilised. Convection then establishes fairly quickly
all around Antarctica and reaches full strength within less than 100 years.
During the following 200 years convection slowly decreases again due to
the action of net precipitation in that area. The result is an asymmetric
oscillation pattern. Pierce et al. (1995) show that this oscillation is only
possible when a nonlinear equation of state is used.
The above mechanism is similar to the flushes found in many other models
(Marotzke 1989; Wright and Stocker 1991; Weaver and Sarachik 1991b;
Winton and Sarachik 1993) whose characteristic feature is a decoupling of
the lower ocean from the surface. In contrast to these earlier studies where
an entire basin had to be destabilized by diffusion and hence evolved on
much longer time scales of 0(10 3 - 10 4 yrs), destabilisation here is due to
advection of warmer NADW, i.e. an "efficient" process with a time scale
of a few hundred years.
3.6 Other Mechanisms
Internal variability on the interdecadal time scale is found in simple ocean
models when a thermodynamic sea ice component is included. Formation of
a sea ice cover releases salt into the water column and induces convection.
By altering density gradients this increases the thermohaline circulation
advecting more heat northward which then melts the sea ice cover. Melting
increases the surface density which tends to decrease the thermohaline
circulation leading to a net cooling. This feedback loop was found in a
2-dimensional ocean circulation model coupled to a thermodynamic sea
ice model (Yang and Neelin 1993). Zhang et al. (1995) showed that this
process is also present in a 3-dimensional OGCM coupled to an ice model
and found the destabilizing influence of brine rejection less important than
that of slow heating below the ice cover due to thermal insulation.
In the previous section mainly oceanic processes localized in or related to
the Atlantic have been discussed. However, there are also recent model
examples of interdecadal cycles in the Pacific and in the atmosphere alone.
James and James (1992) find in their atmospheric circulation model variability on time scales of 10 to 40 years and associate them with changes in
the structure of the subtropical and mid-latitude jets.
Von Storch (1994), on the other hand, uses a coupled AjOGCM and identifies two types of low-frequency variability. The atmospheric fluctuations