92
ANNE-MARIE TREGUIER
adding a parameterization of double diffusion can have a demonstrable
positive impact on the solutions.
3.
Bottom boundary layer and topographic
effects
3.1
Bottom friction
Two-dimensional geostrophic turbulence has the property that energy cascades towards large scales, and enstrophy (the relative vorticity
squared) cascades towards small scales (Batchelor, 1969). Thus, if eddies
have their energy source at a scale close to the internal Rossby radius,
nonlinear interactions tend to transfer this energy to larger scales where
it must be dissipated. Viscous bottom drag can provide the energy sink
which is required to equilibrate the flow. It is thus necessary to include a
parameterization of bottom drag in eddy-resolving models. The strength
of the bottom drag can have an influence on the spatial organisation of
the flow because it affects the baroclinic instability of eastwards jets
(Riviere et al., 2004).
Another interesting effect of bottom drag happens in overflow regions.
In a rotating fluid, and under the hydrostatic approximation, dense
plumes have a strong tendency to follow isobaths rather than plunging.
A high bottom friction makes the flow less geostrophic and increases the
rate of descent of the plumes (Stratford and Haines, 2000).
Despite such important dynamical effects, there does not seem to be
any study documenting the effect of bottom drag in high resolution basin
flows, especially in the presence of bottom topography. Most models use
a quadratic bottom drag with constant coefficient (Table. 2).
3.2
Effects of overflows
Overflows are currents from marginal or semi-enclosed seas into the
main ocean basins, through sills or along continental slopes. They set
the properties of many water masses (Price and Yang, 1998). A major
difficulty in modelling overflows is that many of them are subgrid scale
physics for a given choice of resolution: the width of the strait or channel
is narrower than the grid size. Note that straits can be one grid pointwide on a staggered ”C” grid such as used in the OPA code, but two
grid points are necessary to allow throughflow on a ”B” grid with no-slip
boundary condition (see Haidvogel and Beckmann, 1999, for a definition
of staggered grids). The modeller may decide to open too wide a strait.
In that case, the transport may be too large due to the exaggerated
cross-section. G. Madec (personal communication) decreases the grid
ANNE-MARIE TREGUIER
adding a parameterization of double diffusion can have a demonstrable
positive impact on the solutions.
3.
Bottom boundary layer and topographic
effects
3.1
Bottom friction
Two-dimensional geostrophic turbulence has the property that energy cascades towards large scales, and enstrophy (the relative vorticity
squared) cascades towards small scales (Batchelor, 1969). Thus, if eddies
have their energy source at a scale close to the internal Rossby radius,
nonlinear interactions tend to transfer this energy to larger scales where
it must be dissipated. Viscous bottom drag can provide the energy sink
which is required to equilibrate the flow. It is thus necessary to include a
parameterization of bottom drag in eddy-resolving models. The strength
of the bottom drag can have an influence on the spatial organisation of
the flow because it affects the baroclinic instability of eastwards jets
(Riviere et al., 2004).
Another interesting effect of bottom drag happens in overflow regions.
In a rotating fluid, and under the hydrostatic approximation, dense
plumes have a strong tendency to follow isobaths rather than plunging.
A high bottom friction makes the flow less geostrophic and increases the
rate of descent of the plumes (Stratford and Haines, 2000).
Despite such important dynamical effects, there does not seem to be
any study documenting the effect of bottom drag in high resolution basin
flows, especially in the presence of bottom topography. Most models use
a quadratic bottom drag with constant coefficient (Table. 2).
3.2
Effects of overflows
Overflows are currents from marginal or semi-enclosed seas into the
main ocean basins, through sills or along continental slopes. They set
the properties of many water masses (Price and Yang, 1998). A major
difficulty in modelling overflows is that many of them are subgrid scale
physics for a given choice of resolution: the width of the strait or channel
is narrower than the grid size. Note that straits can be one grid pointwide on a staggered ”C” grid such as used in the OPA code, but two
grid points are necessary to allow throughflow on a ”B” grid with no-slip
boundary condition (see Haidvogel and Beckmann, 1999, for a definition
of staggered grids). The modeller may decide to open too wide a strait.
In that case, the transport may be too large due to the exaggerated
cross-section. G. Madec (personal communication) decreases the grid