Bengal. Reppin et al. (1999) show a cross-section
of annual mean zonal velocity at 80°30ЈE. There is
an eastward transport of about 6 Sv in the top
150 m, between 2°N and 2°S (assuming symmetry
about the equator). Reppin et al. note that this
average may not be representative, due to the
unseasonal winds of 1994. McPhaden’s (1982a)
equatorial data at Gan imply an eastward transport of about 10 Sv in the top 200 m if they are
typical of flows in 2°N–2°S. Donguy and Meyers
(1995) also show an annual mean transport of
about 15 Sv, between 2.5°N and 2.5°S. Thus, there
is agreement on substantial mean eastward flow
along the equator. McPhaden suggested that this
mean flow may be partly due to non-linear effects,
associated with Ekman convergence onto the
equator carrying eastward momentum down, in
the transition season (Cane, 1980; Philander and
Pacanowski, 1980).
The WJs and related semiannual flows may be
regarded as the near-surface manifestation of semiannual currents that extend to considerable depth.
Luyten and Roemmich (1982) analysed current
meter data at several equatorial moorings; their
zonal currents at 700 m showed a narrow semiannual frequency band. They found upward phase
propagation, indicative of downward group propagation, and suggested that the sharp dropoff of
semiannual energy below 200 m (e.g. compared to
semiannual currents at Gan, McPhaden, 1982a) is
due to the fact that semiannual Kelvin waves only
propagate this deep in a full crossing of the Indian
Ocean. Similarly, annual period flow in both
Kelvin and Rossby waves should only penetrate a
few hundred metres before reflection, perhaps
explaining the purity of the semiannual signal at
700 m. The velocities at that depth were primarily
due to semiannual Rossby waves, as borne out by
westward propagation of the semiannual signal.
LR remarked that the Indian Ocean at semiannual
frequency should behave like the Pacific does at
annual frequency, in agreement with the later findings of Kessler and McCreary (1993).
Cross-equatorial currents
The meridional currents across the equator, away
from the western boundary, are of particular interest since it is these that force oppositely directed
deeper, colder flow through mass conservation,
thereby transporting heat out of the Indian Ocean.
The movement of surface water across the equator
in the ocean interior might at first sight be
expected to take a long time to establish, because
the establishment of Sverdrup balance usually
takes months – the time required for Rossby waves
to cross the ocean. However, a zonal wind stress
(x)
:A(x, t)y that is linear in the distance y from
the equator has a non-divergent meridional Ekman
transport A/, with f:y the Coriolis parameter
and the water density. This Ekman transport is
also identical to the Sverdrup transport, i.e. the
vorticity added to the ocean through the curl of
this particular wind is exactly what is needed to
balance its Ekman transport, in its travel north or
south. Thus, for this wind, Sverdrup cross-equatorial flow is established on the time scales of Ekman
transports – i.e. only a few days or weeks. It is
apparent from Fig. 4.3.9 that a wind of this form
is a major contributor to the total wind stress, at
least in summer and winter and on annual mean,
in the near-equatorial Indian Ocean. On this argument, we would expect the Indian Ocean to
respond quite quickly to seasonal variations in
A(x,t); this behaviour is indeed seen in numerical
models.
A(x,t) also varies strongly in association with
Intraseasonal Oscillations (IO); as a result, crossequatorial transport also varies on intraseasonal
time scales. This may complicate analysis of
WOCE sections within 15° of the equator, since
the effect is seen (in one model) to dominate over
seasonal or longer-term variability, as shown in
Fig. 4.3.11 (from Loschnigg and Webster, 2000).
Much of the heat transport on intraseasonal time
scales may consist of a balance between Ekman
transport and a compensating barotropic flow,
with little consequence for the baroclinic flows
that can be deduced from WOCE sections. Nevertheless, it may be of interest to check whether baroclinic flows do develop in deep western boundary
currents on these short time scales, in models with
realistic bottom topography and adequate vertical
resolution.
Modelling studies often show a shallow meridional overturning cell on the equator, confined to
the top 100 m or less. This model cell reverses sign
with the monsoon, and is confined within about 1°
of the equator. Because f is zero, the surface meridional wind
(y) drives a surface downwind flow
(and subsurface mass-balancing counterflow). This
surface flow is usually opposite in direction to
the equatorial Ekman transport, which as shown
4.3 The Tropical Ocean Circulation
237
Godfrey, Johnson, McPhaden, Reverdin and Wijffels
of annual mean zonal velocity at 80°30ЈE. There is
an eastward transport of about 6 Sv in the top
150 m, between 2°N and 2°S (assuming symmetry
about the equator). Reppin et al. note that this
average may not be representative, due to the
unseasonal winds of 1994. McPhaden’s (1982a)
equatorial data at Gan imply an eastward transport of about 10 Sv in the top 200 m if they are
typical of flows in 2°N–2°S. Donguy and Meyers
(1995) also show an annual mean transport of
about 15 Sv, between 2.5°N and 2.5°S. Thus, there
is agreement on substantial mean eastward flow
along the equator. McPhaden suggested that this
mean flow may be partly due to non-linear effects,
associated with Ekman convergence onto the
equator carrying eastward momentum down, in
the transition season (Cane, 1980; Philander and
Pacanowski, 1980).
The WJs and related semiannual flows may be
regarded as the near-surface manifestation of semiannual currents that extend to considerable depth.
Luyten and Roemmich (1982) analysed current
meter data at several equatorial moorings; their
zonal currents at 700 m showed a narrow semiannual frequency band. They found upward phase
propagation, indicative of downward group propagation, and suggested that the sharp dropoff of
semiannual energy below 200 m (e.g. compared to
semiannual currents at Gan, McPhaden, 1982a) is
due to the fact that semiannual Kelvin waves only
propagate this deep in a full crossing of the Indian
Ocean. Similarly, annual period flow in both
Kelvin and Rossby waves should only penetrate a
few hundred metres before reflection, perhaps
explaining the purity of the semiannual signal at
700 m. The velocities at that depth were primarily
due to semiannual Rossby waves, as borne out by
westward propagation of the semiannual signal.
LR remarked that the Indian Ocean at semiannual
frequency should behave like the Pacific does at
annual frequency, in agreement with the later findings of Kessler and McCreary (1993).
Cross-equatorial currents
The meridional currents across the equator, away
from the western boundary, are of particular interest since it is these that force oppositely directed
deeper, colder flow through mass conservation,
thereby transporting heat out of the Indian Ocean.
The movement of surface water across the equator
in the ocean interior might at first sight be
expected to take a long time to establish, because
the establishment of Sverdrup balance usually
takes months – the time required for Rossby waves
to cross the ocean. However, a zonal wind stress
(x)
:A(x, t)y that is linear in the distance y from
the equator has a non-divergent meridional Ekman
transport A/, with f:y the Coriolis parameter
and the water density. This Ekman transport is
also identical to the Sverdrup transport, i.e. the
vorticity added to the ocean through the curl of
this particular wind is exactly what is needed to
balance its Ekman transport, in its travel north or
south. Thus, for this wind, Sverdrup cross-equatorial flow is established on the time scales of Ekman
transports – i.e. only a few days or weeks. It is
apparent from Fig. 4.3.9 that a wind of this form
is a major contributor to the total wind stress, at
least in summer and winter and on annual mean,
in the near-equatorial Indian Ocean. On this argument, we would expect the Indian Ocean to
respond quite quickly to seasonal variations in
A(x,t); this behaviour is indeed seen in numerical
models.
A(x,t) also varies strongly in association with
Intraseasonal Oscillations (IO); as a result, crossequatorial transport also varies on intraseasonal
time scales. This may complicate analysis of
WOCE sections within 15° of the equator, since
the effect is seen (in one model) to dominate over
seasonal or longer-term variability, as shown in
Fig. 4.3.11 (from Loschnigg and Webster, 2000).
Much of the heat transport on intraseasonal time
scales may consist of a balance between Ekman
transport and a compensating barotropic flow,
with little consequence for the baroclinic flows
that can be deduced from WOCE sections. Nevertheless, it may be of interest to check whether baroclinic flows do develop in deep western boundary
currents on these short time scales, in models with
realistic bottom topography and adequate vertical
resolution.
Modelling studies often show a shallow meridional overturning cell on the equator, confined to
the top 100 m or less. This model cell reverses sign
with the monsoon, and is confined within about 1°
of the equator. Because f is zero, the surface meridional wind
(y) drives a surface downwind flow
(and subsurface mass-balancing counterflow). This
surface flow is usually opposite in direction to
the equatorial Ekman transport, which as shown
4.3 The Tropical Ocean Circulation
237
Godfrey, Johnson, McPhaden, Reverdin and Wijffels