eddy variability (e.g. Fu and Smith, 1996) to see
the improvements. One feature that remains to be
reproduced effectively in the global models is the
high variability near the African coast south of
Madagascar, where a significant underestimate of
variability may be related to inadequate meridional overturning as a result of poor initial conditions or surface forcing in the Southern Ocean.
Additional analyses of the global simulations,
including comparison with drifter data and discussion of eddy length scales, are found in Fu and
Smith (1996), Stammer et al. (1996), McClean
et al. (1997), Maltrud et al. (1998) and Saunders
et al. (1999).
Figure 2.2.9, from Smith et al. (2000), depicts
frequency–wavenumber spectra for the Gulf Stream
region for an Atlantic model with 0.1° equatorial
grid spacing, a model with 28 km equatorial grid
spacing (i.e. the ‘1/6°’ global model), and the same
fields from T/P; another plot of the same quantity
with 44 km equatorial grid spacing (i.e. from the
‘1/4°’ global model) is given in Stammer et al.
(1996). These clearly show the overall improvement obtained. In addition, typical deficits of previous Atlantic simulations, such as a failure to
reproduce the variability in the eastern basin associated with the Azores Current (e.g. Beckmann
et al., 1994a,b), troublesome aspects of separation
of the Gulf Stream (e.g. Chao et al., 1996) and
incorrect pathways of the North Atlantic Current in
the Newfoundland Basin (e.g. Willebrand et al.,
2000), are remedied as grid size reaches 1/10°. In
each of these areas the improvement in the simulation of variability accompanies an improvement in
the simulation of the mean flow, suggesting a significant regime transition in capturing ocean dynamics
between 0.1° and 0.2° resolution.
The close correspondence between eddy kinetic
energy distributions and mean baroclinic currents
seen in both altimeter data and model results,
and also the estimates of energy conversion rates
from intensive field programmes, for example the
SYNOP (Synoptic Ocean Prediction) study for the
Gulf Stream (Cronin and Watts, 1996), strongly
suggest baroclinic instability to be a main generation mechanism of mesoscale eddies. Another
mechanism contributing to the spectrum of intraseasonal variability of ocean currents is wind forcing.
Theory suggests that the extratropical ocean’s
response to forcing by wind fluctuations of annual
period or less should mainly be barotropic (Willebrand et al., 1980). A significant excitation of
large-scale, barotropic variability at higher latitudes, notably for the eastern South Pacific and
northern North Pacific, has been demonstrated by
various altimeter and model analyses, e.g. Chao
and Fu (1995), Fu and Smith (1996) and Fukumori
et al. (1998). The question of a direct contribution
of variable wind forcing to the observed distribution of mesoscale eddy energy has long been under
debate, with some support coming from indications of a seasonal modulation of energy levels in
some regions of the northeastern North Atlantic
(Dickson et al., 1982; White and Heywood, 1995).
Recent model results show a similar seasonal
cycle, with a winter maximum in eddy energy, as a
result of forcing by high-frequency wind fields
(Stammer et al., 2000a). However, a significant
wind impact on the generation of eddy energy
appears to be confined to areas with very low
values in their background eddy variability, i.e.
away from the major frontal zones.
As noted above, for the foreseeable future it will
not be possible to undertake extensive climate
simulations with explicitly resolving ocean eddies.
Representation of eddy effects in coarse-resolution,
z-coordinate models has traditionally been rather
crude, using simple downgradient diffusion terms
for momentum and tracers. In the last few years,
2.2 Modelling of Thermohaline and Wind-Driven Circulation
71
Böning and Semtner
80°S
4 0
0
4 0 °N
0
2
4
6
8
10
12
14
16
18
20
rms SSH (cm)
Fig. 2.2.8 Meridional profiles of the variance of SSH in
the western Pacific (averaged between 150°E and 160°E).
The bold solid lines are from T/P, the dashed lines are
from a global model run (OCCAM) at 1/4° resolution,
and the thin solid lines from a run at 1/8° resolution.
From Saunders et al. (1999).
the improvements. One feature that remains to be
reproduced effectively in the global models is the
high variability near the African coast south of
Madagascar, where a significant underestimate of
variability may be related to inadequate meridional overturning as a result of poor initial conditions or surface forcing in the Southern Ocean.
Additional analyses of the global simulations,
including comparison with drifter data and discussion of eddy length scales, are found in Fu and
Smith (1996), Stammer et al. (1996), McClean
et al. (1997), Maltrud et al. (1998) and Saunders
et al. (1999).
Figure 2.2.9, from Smith et al. (2000), depicts
frequency–wavenumber spectra for the Gulf Stream
region for an Atlantic model with 0.1° equatorial
grid spacing, a model with 28 km equatorial grid
spacing (i.e. the ‘1/6°’ global model), and the same
fields from T/P; another plot of the same quantity
with 44 km equatorial grid spacing (i.e. from the
‘1/4°’ global model) is given in Stammer et al.
(1996). These clearly show the overall improvement obtained. In addition, typical deficits of previous Atlantic simulations, such as a failure to
reproduce the variability in the eastern basin associated with the Azores Current (e.g. Beckmann
et al., 1994a,b), troublesome aspects of separation
of the Gulf Stream (e.g. Chao et al., 1996) and
incorrect pathways of the North Atlantic Current in
the Newfoundland Basin (e.g. Willebrand et al.,
2000), are remedied as grid size reaches 1/10°. In
each of these areas the improvement in the simulation of variability accompanies an improvement in
the simulation of the mean flow, suggesting a significant regime transition in capturing ocean dynamics
between 0.1° and 0.2° resolution.
The close correspondence between eddy kinetic
energy distributions and mean baroclinic currents
seen in both altimeter data and model results,
and also the estimates of energy conversion rates
from intensive field programmes, for example the
SYNOP (Synoptic Ocean Prediction) study for the
Gulf Stream (Cronin and Watts, 1996), strongly
suggest baroclinic instability to be a main generation mechanism of mesoscale eddies. Another
mechanism contributing to the spectrum of intraseasonal variability of ocean currents is wind forcing.
Theory suggests that the extratropical ocean’s
response to forcing by wind fluctuations of annual
period or less should mainly be barotropic (Willebrand et al., 1980). A significant excitation of
large-scale, barotropic variability at higher latitudes, notably for the eastern South Pacific and
northern North Pacific, has been demonstrated by
various altimeter and model analyses, e.g. Chao
and Fu (1995), Fu and Smith (1996) and Fukumori
et al. (1998). The question of a direct contribution
of variable wind forcing to the observed distribution of mesoscale eddy energy has long been under
debate, with some support coming from indications of a seasonal modulation of energy levels in
some regions of the northeastern North Atlantic
(Dickson et al., 1982; White and Heywood, 1995).
Recent model results show a similar seasonal
cycle, with a winter maximum in eddy energy, as a
result of forcing by high-frequency wind fields
(Stammer et al., 2000a). However, a significant
wind impact on the generation of eddy energy
appears to be confined to areas with very low
values in their background eddy variability, i.e.
away from the major frontal zones.
As noted above, for the foreseeable future it will
not be possible to undertake extensive climate
simulations with explicitly resolving ocean eddies.
Representation of eddy effects in coarse-resolution,
z-coordinate models has traditionally been rather
crude, using simple downgradient diffusion terms
for momentum and tracers. In the last few years,
2.2 Modelling of Thermohaline and Wind-Driven Circulation
71
Böning and Semtner
80°S
4 0
0
4 0 °N
0
2
4
6
8
10
12
14
16
18
20
rms SSH (cm)
Fig. 2.2.8 Meridional profiles of the variance of SSH in
the western Pacific (averaged between 150°E and 160°E).
The bold solid lines are from T/P, the dashed lines are
from a global model run (OCCAM) at 1/4° resolution,
and the thin solid lines from a run at 1/8° resolution.
From Saunders et al. (1999).