202
periods. These fluctuations in basin mean temperature are symptomatic
of the deep- decoupling type of variability. An examination of the details
of the circulation changes, however, shows that there are elements of the
pure decadal variability present as well. This kind of variability will be
discussed next.
We can "tune in" the pure decadal variability in the cold boundary
condition experiment described above by halving the surface temperature
restoring coefficient. Recall that this moves the model in the direction of
stability. The result is a regular oscillation with about a 20 year period
(Fig. 17 a). Figure 17b shows the details of the oscillation. The velocity
vectors at 175 m depth are overlain by contours of vertically averaged
density. The first panel shows a zonally oriented baroclinic jet impinging
upon the eastern boundary between 30° Nand 50° N. A portion of this jet
splits off and propagates in the Kelvin wave direction along the northern
boundary. Light water is left on the northern boundary in the wake of this
disturbance, reversing the direction of the zonal flow along the northern
boundary.
Similar propagating features can be found in the mixed boundary condition decadal variability studied by Weaver and Sarachik (1991) and Yin
and Sarachik (1995). The mechanism for the variability also appears to
be the same as that for internal decadal variability found under a fixed
flux boundary condition for a single buoyancy variable (Winton, 1995c).
Apparently, the mixed boundary condition variability does not involve the
interplay of heat and salt nor either of the advective and convective instabilities described above. It is rather a product of the adjustment of
baroclinic currents impinging on weakly stratified coasts to the no normal
flow boundary condition. This adjustment spins up deep warm and cold
features on the boundary that, upon reaching sufficient amplitude, propagate slowly in the Kelvin wave direction (Winton, 1995c). The property
shared by the mixed boundary condition and fixed flux boundary condition experiments that permits this variability to occur is weak damping.
In the mixed boundary condition case, a halo cline shuts off convection and
the timescale for modification of the water column reverts from the short
timescale for the surface boundary condition (about 10 years) to a much
longer diffusive timescale. The propagation is particularly clear in the case
presented here because of the uniform coverage of the high-latitude ocean
by the halocline.
8 Mechanism and Predictability
We can define three types of predictability. Deterministic predictability is
best known from our experience with weather prediction and is limited to
periods. These fluctuations in basin mean temperature are symptomatic
of the deep- decoupling type of variability. An examination of the details
of the circulation changes, however, shows that there are elements of the
pure decadal variability present as well. This kind of variability will be
discussed next.
We can "tune in" the pure decadal variability in the cold boundary
condition experiment described above by halving the surface temperature
restoring coefficient. Recall that this moves the model in the direction of
stability. The result is a regular oscillation with about a 20 year period
(Fig. 17 a). Figure 17b shows the details of the oscillation. The velocity
vectors at 175 m depth are overlain by contours of vertically averaged
density. The first panel shows a zonally oriented baroclinic jet impinging
upon the eastern boundary between 30° Nand 50° N. A portion of this jet
splits off and propagates in the Kelvin wave direction along the northern
boundary. Light water is left on the northern boundary in the wake of this
disturbance, reversing the direction of the zonal flow along the northern
boundary.
Similar propagating features can be found in the mixed boundary condition decadal variability studied by Weaver and Sarachik (1991) and Yin
and Sarachik (1995). The mechanism for the variability also appears to
be the same as that for internal decadal variability found under a fixed
flux boundary condition for a single buoyancy variable (Winton, 1995c).
Apparently, the mixed boundary condition variability does not involve the
interplay of heat and salt nor either of the advective and convective instabilities described above. It is rather a product of the adjustment of
baroclinic currents impinging on weakly stratified coasts to the no normal
flow boundary condition. This adjustment spins up deep warm and cold
features on the boundary that, upon reaching sufficient amplitude, propagate slowly in the Kelvin wave direction (Winton, 1995c). The property
shared by the mixed boundary condition and fixed flux boundary condition experiments that permits this variability to occur is weak damping.
In the mixed boundary condition case, a halo cline shuts off convection and
the timescale for modification of the water column reverts from the short
timescale for the surface boundary condition (about 10 years) to a much
longer diffusive timescale. The propagation is particularly clear in the case
presented here because of the uniform coverage of the high-latitude ocean
by the halocline.
8 Mechanism and Predictability
We can define three types of predictability. Deterministic predictability is
best known from our experience with weather prediction and is limited to