annual mean. Most annual mean heat flux climatologies show a minimum near the COARE region.
Thus the average over the whole of Wijffels’ western Pacific domain would probably be substantially greater than 25 W m
92
.
Apart from the question of the magnitude of
the surface heat flux into the western equatorial
Pacific, a further question is raised by Wijffels’
study: how is the heat mixed down, to change
South Pacific subthermocline water into the upperlayer water that flows out with the Indonesian
Throughflow? Two processes can be postulated.
The first involves the New Guinea Coastal Current
(NGCC; Lindstrom et al., 1990). On annual mean
this western boundary current travels northwestward toward the equator; but in austral summer,
the surface flow is downwind (southeastward),
against the mean northwestward flow. A strong
shear zone therefore exists at this time in the
NGCC, and low Richardson numbers are observed
below the mixed layer, suggesting mixing there
(S. Cresswell, personal communication). Second,
Ffield and Gordon (1996) suggested that tidal
mixing raised vertical eddy diffusivities by a factor
of about 10 throughout the Indonesian Archipelago
over the usual values of about 10
95 m
2 s
91
. This
implies that surface heat fluxes of order 40 W m
92
are carried through both the surface and the base
of the mixed layer, to the thermocline waters
below. It is quite plausible that tidally enhanced
eddy diffusivities (perhaps not as large as in Indonesia) are widely distributed in the west Pacific, near
reef complexes; e.g. Pinkel et al. (1997).
Thermohaline phenomena
In the Warm Pool regions of the western Pacific and
eastern Indian Ocean, the net heat flux is not as
large as in the east Pacific; but rainfall is heavy.
Heat and fresh water contribute about equally to
the surface buoyancy flux in this region. Under these
conditions the true mixed layer is often defined by
a halocline; short-wave penetration through this
halocline gently warms the water just beneath (e.g.
Lewis et al., 1990), resulting in a nearly isothermal
region below the mixed layer. This is named the
‘barrier layer’, because the combination of low eddy
diffusivity and low temperature gradient within it
imply low net turbulent heat fluxes (Godfrey and
Lindstrom, 1989; Lukas and Lindstrom, 1991).
Lukas (1988) suggested that the barrier layer in the
central Pacific might be maintained dynamically,
by subduction of warm salty water at the eastern
edge of the Warm Pool. Picaut et al. (1996)
describe observations of a salinity front in the central equatorial Pacific, at which just such a subduction phenomenon is thought to occur. The front
moves east and west by over 1000 km. This behaviour has been simulated successfully in a numerical
model, and salty water does indeed subduct at the
model front to continue westward a few tens of
metres below the surface (Vialard and Delecluse,
1998a,b). In the model, the non-linear advective
movement of the front is tightly linked to wind
driving associated with the El Niño (P. Delecluse,
personal communication).
The net result of these effects is that to the
west of the mid-Pacific salinity front, wind-driven
entrainment from the base of the mixed layer can
slightly warm the mixed layer, rather than cooling
it as in the eastern Pacific. This process (along
with low wind speeds and evaporation) may be the
reason that SST remains high in the Warm Pool
region, despite the high cloud cover and low insolation in these regions. Thus there may be a new
positive feedback mechanism associated with
ENSO. An ENSO event spreads the fresh, Warm
Pool eastward; SST warms to above 28°C, because
of ‘Barrier-Layer Shutoff of Entrainment’ (Lukas,
1988); this SST increase permits deep convection
to occur, and the resulting heavy precipitation perpetuates the barrier layer. Support for this positive
feedback mechanism comes from a basin-scale
analysis of interannual variations in surface layer
hydrography and precipitation in the tropical
Pacific (Ando and McPhaden, 1997).
Eddy diffusion
Pacanowski and Philander (1981) found that the
inclusion of a strong Richardson number dependence in their parameterizations of vertical viscosity and diffusivity isolated most dissipation to the
shear zone between the EUC and the surface current. This greatly improved their model’s ability to
reproduce observed equatorial density and current
structures. Direct observations (Peters et al., 1988)
showed the Richardson number dependence to
be still stronger than Pacanowski and Philander
assumed. Kunze et al. (1990) developed a parameterization of dissipation rates, to be applied only in
unstable conditions (Ri:0.25); Peters et al. (1995)
and Polzin (1996) found this formula agreed with
their dissipation data, to within a factor of 2.
SECTION 4 THE GLOBAL FLOW FIELD
220
Thus the average over the whole of Wijffels’ western Pacific domain would probably be substantially greater than 25 W m
92
.
Apart from the question of the magnitude of
the surface heat flux into the western equatorial
Pacific, a further question is raised by Wijffels’
study: how is the heat mixed down, to change
South Pacific subthermocline water into the upperlayer water that flows out with the Indonesian
Throughflow? Two processes can be postulated.
The first involves the New Guinea Coastal Current
(NGCC; Lindstrom et al., 1990). On annual mean
this western boundary current travels northwestward toward the equator; but in austral summer,
the surface flow is downwind (southeastward),
against the mean northwestward flow. A strong
shear zone therefore exists at this time in the
NGCC, and low Richardson numbers are observed
below the mixed layer, suggesting mixing there
(S. Cresswell, personal communication). Second,
Ffield and Gordon (1996) suggested that tidal
mixing raised vertical eddy diffusivities by a factor
of about 10 throughout the Indonesian Archipelago
over the usual values of about 10
95 m
2 s
91
. This
implies that surface heat fluxes of order 40 W m
92
are carried through both the surface and the base
of the mixed layer, to the thermocline waters
below. It is quite plausible that tidally enhanced
eddy diffusivities (perhaps not as large as in Indonesia) are widely distributed in the west Pacific, near
reef complexes; e.g. Pinkel et al. (1997).
Thermohaline phenomena
In the Warm Pool regions of the western Pacific and
eastern Indian Ocean, the net heat flux is not as
large as in the east Pacific; but rainfall is heavy.
Heat and fresh water contribute about equally to
the surface buoyancy flux in this region. Under these
conditions the true mixed layer is often defined by
a halocline; short-wave penetration through this
halocline gently warms the water just beneath (e.g.
Lewis et al., 1990), resulting in a nearly isothermal
region below the mixed layer. This is named the
‘barrier layer’, because the combination of low eddy
diffusivity and low temperature gradient within it
imply low net turbulent heat fluxes (Godfrey and
Lindstrom, 1989; Lukas and Lindstrom, 1991).
Lukas (1988) suggested that the barrier layer in the
central Pacific might be maintained dynamically,
by subduction of warm salty water at the eastern
edge of the Warm Pool. Picaut et al. (1996)
describe observations of a salinity front in the central equatorial Pacific, at which just such a subduction phenomenon is thought to occur. The front
moves east and west by over 1000 km. This behaviour has been simulated successfully in a numerical
model, and salty water does indeed subduct at the
model front to continue westward a few tens of
metres below the surface (Vialard and Delecluse,
1998a,b). In the model, the non-linear advective
movement of the front is tightly linked to wind
driving associated with the El Niño (P. Delecluse,
personal communication).
The net result of these effects is that to the
west of the mid-Pacific salinity front, wind-driven
entrainment from the base of the mixed layer can
slightly warm the mixed layer, rather than cooling
it as in the eastern Pacific. This process (along
with low wind speeds and evaporation) may be the
reason that SST remains high in the Warm Pool
region, despite the high cloud cover and low insolation in these regions. Thus there may be a new
positive feedback mechanism associated with
ENSO. An ENSO event spreads the fresh, Warm
Pool eastward; SST warms to above 28°C, because
of ‘Barrier-Layer Shutoff of Entrainment’ (Lukas,
1988); this SST increase permits deep convection
to occur, and the resulting heavy precipitation perpetuates the barrier layer. Support for this positive
feedback mechanism comes from a basin-scale
analysis of interannual variations in surface layer
hydrography and precipitation in the tropical
Pacific (Ando and McPhaden, 1997).
Eddy diffusion
Pacanowski and Philander (1981) found that the
inclusion of a strong Richardson number dependence in their parameterizations of vertical viscosity and diffusivity isolated most dissipation to the
shear zone between the EUC and the surface current. This greatly improved their model’s ability to
reproduce observed equatorial density and current
structures. Direct observations (Peters et al., 1988)
showed the Richardson number dependence to
be still stronger than Pacanowski and Philander
assumed. Kunze et al. (1990) developed a parameterization of dissipation rates, to be applied only in
unstable conditions (Ri:0.25); Peters et al. (1995)
and Polzin (1996) found this formula agreed with
their dissipation data, to within a factor of 2.
SECTION 4 THE GLOBAL FLOW FIELD
220