also some support for a Wacongne–Pacanowski
cell from an ADCP current meter record (Reppin
et al., 1999). These authors find no annual cycle in
equatorial flow at the top of their record (25 m),
but their Plate 1b appears to show a seasonal cycle
of the correct sign just below 25 m, contaminated
by substantial 16-day wave activity.
McPhaden (1982a) found almost no mean
seasonal cycle in meridional flow in the Gan data.
Annual mean meridional velocity had the right
sign, but the magnitude was about four times too
large to match the Sverdrup relation. Residual
variability had a flat spectrum. These results may
reflect a distortion of flow around Gan.
Equatorial surface heat budget
Adequate representation of the surface mixedlayer heat budget is crucial to representing SST
variations. Despite the strong semiannual signal in
equatorial currents, McPhaden (1982a) found that
annual frequencies dominated the thermal structure at Gan. McPhaden (1982b) examined the surface heat budget with the Gan data, and found
that a simple balance, basically equating rate of
change of mixed-layer heat content to the anomaly
of observed heat flux, worked surprisingly well in
accounting for mixed-layer temperature changes.
However, Molinari et al. (1986) compared surface
heat fluxes with the rate of change of SST over
much of the Indian Ocean, and found that such
behaviour was rare. They found that net surface
heat fluxes accounted for 80% or more of the
observed changes in seasonal heat content of the
mixed layer in only 11% of the ocean area north
of the equator and west of 80°E. The comparable
figure for 0–20°S was 36%. Rao and Sivakumar
(1999) examined the mixed-layer heat budget for
the Indian Ocean north of the equator and west of
80°E. They include horizontal advection by taking
the scalar product of ship’s drift current with SST
gradient multiplied by the mixed-layer depth, and
obtain entrainment velocity following McPhaden
and Hayes (1991). Climatological salinity variations are also included. The result accounts quite
well for SST change in December–May, when SST
warms up to 30°C (a major factor in monsoon
onset). Advective processes – and salinity effects –
play important roles in most regions, though near
Gan (73°E) advective terms are small, in agreement with McPhaden (1982b). Loschnigg and
Webster (2000) also conclude that advection plays
an important role in SST development before
monsoon onset.
Western boundary current flows
Mass balance requires that an equal and opposite
subsurface flow must balance the net surface
meridional Ekman flow out of the northern Indian
Ocean. In order to dissipate its vorticity, this subsurface flow must cross the equator in the western
boundary current. According to 2 years of current
meter data, the northward annual mean thermocline flow across the equator in the western
boundary current is about 10 Sv (Schott et al.,
1990; referred to below as SSF), which is close to
the magnitude expected from this argument. It all
occurs above 500 m, and has a transport-weighted
mean temperature of 15.4°C. SSF note that if the
compensating surface Ekman flow has a temperature of 28°C, it ‘would result in a southward heat
transport of 93.610
14 W, which is not too far
off the air–sea flux results’.
This direct, wind-driven process is an important
one for carrying heat southward out of the Indian
Ocean; but it may not be the only one. A special
feature of the ‘Somali Current System’ (SCS) is that
it reverses seasonally. This raises the possibility
that ‘eddy fluxes’, on seasonal or other time scales,
may also be important. SSF note that in winter (December–February) the equatorial western
boundary currents themselves have almost zero
cross-equatorial flow, but 5 Sv of southward flow
in the top 100 m nearly compensates the 4 Sv of
northward flow in 100–400 m. They find a southward heat flow of about 310
14 W at this time,
associated with this cell near the western boundary.
Does this observed wintertime cell near the
western boundary contribute to the annual mean
heat transport out of the Indian Ocean? It seems
reasonable that it does. It will be argued below
that, to first approximation, most of the thermocline water crosses the equator northeastwards in
one winter, in the western boundary; is upwelled
and warmed the following summer; and returns
across the equator the next winter, at a higher,
warmer level. (This level need not be the surface
mixed layer.) In such a situation there will be a net
southward flow of heat on annual mean, which is
strictly internal to the reversing SCS and does not
depend on the meridional Ekman transports.
Current meter records show that the southward
winter flow across the equator does not penetrate
4.3 The Tropical Ocean Circulation
239
Godfrey, Johnson, McPhaden, Reverdin and Wijffels
cell from an ADCP current meter record (Reppin
et al., 1999). These authors find no annual cycle in
equatorial flow at the top of their record (25 m),
but their Plate 1b appears to show a seasonal cycle
of the correct sign just below 25 m, contaminated
by substantial 16-day wave activity.
McPhaden (1982a) found almost no mean
seasonal cycle in meridional flow in the Gan data.
Annual mean meridional velocity had the right
sign, but the magnitude was about four times too
large to match the Sverdrup relation. Residual
variability had a flat spectrum. These results may
reflect a distortion of flow around Gan.
Equatorial surface heat budget
Adequate representation of the surface mixedlayer heat budget is crucial to representing SST
variations. Despite the strong semiannual signal in
equatorial currents, McPhaden (1982a) found that
annual frequencies dominated the thermal structure at Gan. McPhaden (1982b) examined the surface heat budget with the Gan data, and found
that a simple balance, basically equating rate of
change of mixed-layer heat content to the anomaly
of observed heat flux, worked surprisingly well in
accounting for mixed-layer temperature changes.
However, Molinari et al. (1986) compared surface
heat fluxes with the rate of change of SST over
much of the Indian Ocean, and found that such
behaviour was rare. They found that net surface
heat fluxes accounted for 80% or more of the
observed changes in seasonal heat content of the
mixed layer in only 11% of the ocean area north
of the equator and west of 80°E. The comparable
figure for 0–20°S was 36%. Rao and Sivakumar
(1999) examined the mixed-layer heat budget for
the Indian Ocean north of the equator and west of
80°E. They include horizontal advection by taking
the scalar product of ship’s drift current with SST
gradient multiplied by the mixed-layer depth, and
obtain entrainment velocity following McPhaden
and Hayes (1991). Climatological salinity variations are also included. The result accounts quite
well for SST change in December–May, when SST
warms up to 30°C (a major factor in monsoon
onset). Advective processes – and salinity effects –
play important roles in most regions, though near
Gan (73°E) advective terms are small, in agreement with McPhaden (1982b). Loschnigg and
Webster (2000) also conclude that advection plays
an important role in SST development before
monsoon onset.
Western boundary current flows
Mass balance requires that an equal and opposite
subsurface flow must balance the net surface
meridional Ekman flow out of the northern Indian
Ocean. In order to dissipate its vorticity, this subsurface flow must cross the equator in the western
boundary current. According to 2 years of current
meter data, the northward annual mean thermocline flow across the equator in the western
boundary current is about 10 Sv (Schott et al.,
1990; referred to below as SSF), which is close to
the magnitude expected from this argument. It all
occurs above 500 m, and has a transport-weighted
mean temperature of 15.4°C. SSF note that if the
compensating surface Ekman flow has a temperature of 28°C, it ‘would result in a southward heat
transport of 93.610
14 W, which is not too far
off the air–sea flux results’.
This direct, wind-driven process is an important
one for carrying heat southward out of the Indian
Ocean; but it may not be the only one. A special
feature of the ‘Somali Current System’ (SCS) is that
it reverses seasonally. This raises the possibility
that ‘eddy fluxes’, on seasonal or other time scales,
may also be important. SSF note that in winter (December–February) the equatorial western
boundary currents themselves have almost zero
cross-equatorial flow, but 5 Sv of southward flow
in the top 100 m nearly compensates the 4 Sv of
northward flow in 100–400 m. They find a southward heat flow of about 310
14 W at this time,
associated with this cell near the western boundary.
Does this observed wintertime cell near the
western boundary contribute to the annual mean
heat transport out of the Indian Ocean? It seems
reasonable that it does. It will be argued below
that, to first approximation, most of the thermocline water crosses the equator northeastwards in
one winter, in the western boundary; is upwelled
and warmed the following summer; and returns
across the equator the next winter, at a higher,
warmer level. (This level need not be the surface
mixed layer.) In such a situation there will be a net
southward flow of heat on annual mean, which is
strictly internal to the reversing SCS and does not
depend on the meridional Ekman transports.
Current meter records show that the southward
winter flow across the equator does not penetrate
4.3 The Tropical Ocean Circulation
239
Godfrey, Johnson, McPhaden, Reverdin and Wijffels