Observations and Numerical Models
273
tracer age is considerably greater. Panels c and d show the potential vorticity in
the two cases, and the three regions of the flowing the subtropical gyre are
again easily observed.
Although the overall results of the numerical experiment lend support to
the theory, there are significant differences between the numerical results and
the analytical theory. In the analytical theory, in which diffusion of density or
potential vorticity is absent in the interior, isolines of q on each density surface
coincide with streamlines and thread their way from the outcrop line to the
western boundary or wrap up in the pool regions and become homogenized.
The numerical experiments are diffusive, and the turbulent flux of density,
particularly, no longer permits exact conservation of potential vorticity. As a
consequence, the potential vorticity isolines are shaped as drawn-out tongues
rather than completed pathways, and the streamlines of the flow cross the
tongues as a manifestation of the lack of potential vorticity conservation. In
the numerical experiments this is particularly striking in the fine-resolution case
with its active eddy field. In the coarse-resolution calculation the potential
vorticity in the western pool is variable rather than uniform as a consequence
of nonadiabatic effects in the model. The eddy-resolving model shows large
regions of nearly uniform potential vorticity near the boundary between the
ventilated and pool regions. Cox suggested that strong and rapid eddy mixing
of potential vorticity across the pool boundary with the ventilated region,
rather than slow homogenization within the pool region is the mechanism in
the numerical experiment for the production of large regions of nearly uniform
q. The model produces regions of large potential vorticity near both the eastern
and western boundaries separated by low potential vorticity water, produced
by convection, that feeds the ventilated region from the outcrop line. The
strong eddy mixing rubs out the tongue of low potential vorticity over a large
part of the basin and yields an enormous region of uniform q connecting the
eastern and western zones.
Thus the numerical, primitive equation models present yet an additional
mechanism for the production of regions of uniform potential vorticity beyond
those described by the homogenization theories of Chapter 3 or the quasigeostrophic numerical models. To what extent the numerical models accurately
represent the effect of the eddy field on the general circulation is unclear.
The numerical models emphasize (perhaps even overemphasize) what is
also evident in the data, namely that potential vorticity is not exactly
conserved. Nonadiabatic effects are clearly important in the flow. However, the
data, for example, Talley's maps of salinity and potential vorticity, show the
tracer tongues more drawn-out and less fragile to eddy mixing than the numerical experiments would seem to indicate.
Overall the theories prove remarkable in capturing the flow structure and
the density field of the interior circulation, at least in the subtropical gyre. The
western boundary current, which is not included in the analytical theories, does
not seem fundamentally to upset the interior dynamics either in the
observations or in the numerical models of the general circulation.
273
tracer age is considerably greater. Panels c and d show the potential vorticity in
the two cases, and the three regions of the flowing the subtropical gyre are
again easily observed.
Although the overall results of the numerical experiment lend support to
the theory, there are significant differences between the numerical results and
the analytical theory. In the analytical theory, in which diffusion of density or
potential vorticity is absent in the interior, isolines of q on each density surface
coincide with streamlines and thread their way from the outcrop line to the
western boundary or wrap up in the pool regions and become homogenized.
The numerical experiments are diffusive, and the turbulent flux of density,
particularly, no longer permits exact conservation of potential vorticity. As a
consequence, the potential vorticity isolines are shaped as drawn-out tongues
rather than completed pathways, and the streamlines of the flow cross the
tongues as a manifestation of the lack of potential vorticity conservation. In
the numerical experiments this is particularly striking in the fine-resolution case
with its active eddy field. In the coarse-resolution calculation the potential
vorticity in the western pool is variable rather than uniform as a consequence
of nonadiabatic effects in the model. The eddy-resolving model shows large
regions of nearly uniform potential vorticity near the boundary between the
ventilated and pool regions. Cox suggested that strong and rapid eddy mixing
of potential vorticity across the pool boundary with the ventilated region,
rather than slow homogenization within the pool region is the mechanism in
the numerical experiment for the production of large regions of nearly uniform
q. The model produces regions of large potential vorticity near both the eastern
and western boundaries separated by low potential vorticity water, produced
by convection, that feeds the ventilated region from the outcrop line. The
strong eddy mixing rubs out the tongue of low potential vorticity over a large
part of the basin and yields an enormous region of uniform q connecting the
eastern and western zones.
Thus the numerical, primitive equation models present yet an additional
mechanism for the production of regions of uniform potential vorticity beyond
those described by the homogenization theories of Chapter 3 or the quasigeostrophic numerical models. To what extent the numerical models accurately
represent the effect of the eddy field on the general circulation is unclear.
The numerical models emphasize (perhaps even overemphasize) what is
also evident in the data, namely that potential vorticity is not exactly
conserved. Nonadiabatic effects are clearly important in the flow. However, the
data, for example, Talley's maps of salinity and potential vorticity, show the
tracer tongues more drawn-out and less fragile to eddy mixing than the numerical experiments would seem to indicate.
Overall the theories prove remarkable in capturing the flow structure and
the density field of the interior circulation, at least in the subtropical gyre. The
western boundary current, which is not included in the analytical theories, does
not seem fundamentally to upset the interior dynamics either in the
observations or in the numerical models of the general circulation.