Prog. Oceanog.Vol. 26, pp. 1-73, 1991. 0079 - 6611/91 $0.00 + .

50
Printed in Great Britain. All fights reserved. © 1991 Pergamon Press pie

Upper-level circulation in the South Atlantic Ocean

RAY G. P~-rwtSON and LOTHAR Sa~AMMA

lnstitut fiir Meereskunde an der Universitiit Kiel, Diisternbrooker Weg 20, 2300 Kiel 1, F.R.G.

Abstract - In this paper we present a literature survey of the South Atlantic's climate and its

oceanic upper-layer circulation and meridional beat transport. The opening section deals with climate
and is focused upon those elements having greatest oceanic relevance, i.e., distributions of atmospheric
sea level pressure, the wind fields they produce, and the net surface energy fluxes. The various
geostrophic currents comprising the upper-level general circulation are then reviewed in a manner
organized around the subtropical gyre, beginning off southern Africa with the Agulhas Current
Retroflection and then progressing to the Benguela Current, the equatorial current system and
circulation in the Angola Basin, the large-scale variability and interannual warmings at low latitudes,
the Brazil Current, the South Atlantic Cmrent, and finally to the Antarctic Circumpolar Current system
in which the Falkland (Malvinas) Current is included. A summary of estimates of the meridional heat
transport at various latitudes in the South Atlantic ends the survey.

CONTENTS

1. Introduction 2
2. Climatic Elements 2
3. Subtropical and Equatorial Circulation 11
3.1. Agulhas Current Retroflection 11
3.2. Benguela Cmrent 16
3.3. Equatorial Cttrrents 18
3.3.1. Components of the system 18
3.3.2. Angola Basin circulation 26
3.3.3. Large-scale variability and interannual warmings 29
3.4. Brazil Current 33
3.5. South Atlantic Current 39
3.5.1. Subtropical Front 40
3.5.2. Geostrophie flow 41
4. Antarctic Circumpolar Ctmrent 42
4.1. Zonation 42
4.2. Volume transport 51
5. Meridional Heat Transport 59
6. Outlook 61
7. Acknowledgements 62
8. References 63

d~O 26:1-^
2 R.G. P~rv_a~soNand L. STRAMMA

1. INTRODUCTION

The circulation patterns and diversity of water masses in the South Atlantic Ocean are
extraordinary in many important respects, as the early results of the Meteor expedition of the
1920s clearly demonstrated (WOST, 1935; D~AN'r, 1936). Since then, the South Atlantic has been
the subject of increasing levels of scientific interest and research, especially of late, from which
substantial amounts of new knowledge have come. This is particularly true for the near-surface,
wind-driven circulation which is much better understood than its deep, thermohaline-driven
counterpart. Presently, as there has been much learned about the South Atlantic's upper-level
currents in recent years, it seems timely that the published observations concerning them be drawn
together into a single picture. Although the available information is still far less than what is
needed for a complete, detailed picture, it is nonetheless adequate for assembling a reasonably
coherent overview. The purpose of this paper, for the region of the Atlantic extending from the
South Equatorial Current to the Antarctic Circumpolar Current, is to provide such an overview.
The large-scale, upper-level geostrophic circulation in the South Atlantic Ocean is schemati-
cally depicted in Fig.1. This figure is intended to serve as a reference for the following
discussions. The same holds true of Fig.2, which shows the major bathymetric and other physical
features of the South Atlantic. Referring back to Fig. 1, the upper-level circulation is dominated
by a system ofgyres and by the equatorial and circumpolar current systems. Being that these major
systems are all wind-driven, we start by considering a few of the basic climatological elements
of the South Atlantic. Next come descriptions of the upper-level circulation beginning with the
Agulhas Current Retroflection and progressing anti-clockwise around the basin to the Antarctic
Circumpolar Current. We then summarize the meridional heat transport in the South Atlantic and
conclude with a brief outlook for further studies.

2. CLIMATIC ELEMENTS

All weather observations made at sea from the year 1961 and later have been collected and
evaluated by eight nations, with each being responsible for a particular ocean area (WMO, 1963;
1977). Observations from the South Atlantic are handled by the Federal Republic of Germany,
and these have been used by HOFLICH(1984) to construct a climatology for this ocean. His is the
chief source of information for this section, with the exception of wind stress fields which are
obtained from other sources.
Shown in Fig.3 are the climatic averages of atmospberic sea level pressure for the months of
January and July. As can be inferred, the climate of much of the South Atlantic Ocean is
dominated by a semi-permanent high pressure system in the subtropics, one that is similar to those
found over other ocean basins. In the southern summer, the South Atlantic subtropical high is
centered near 32°S 5°W with a central pressure of just over 1021 mbar (1 mbar = 100Pa). In winter,
the center of this high increases in pressure to about 1025mbar and moves some 800km toward
the northwest, to approximately 27°S 10°W. Although the central pressure is highest in southern
winter, the pressure differences between the center of the high and the coast lines of South
America and southern Africa are greater in summer, which are direct results of the thermal lows
which develop over each continent during summer. Consequently, the along-shore winds offthe
west coast of southern Africa and the associated upwellings are stronger in summer. And because
the center of the high resides over the eastern basin of the South Atlantic, the trade winds near
southern Africa are generally stronger than those near Brazil. Farther south, in the mid- and high-
latitudes, the isobars are nearly zonal in the mean for each month, producing the westerly winds
which extend southward to the circumpolar belt of low pressure centered near 65°S.
 Upper-level circulation in the South Atlantic Ocean 3

tea

~tJ
"d

o~

O
b t/)
tD
4 R.G. Pm'~SON and L. STZAMMA

O ¢M

U/ W
,.O

¢D
¢D

cq

O
¢D
 Upper-level circulation in the South Atlantic Ocean 5

70 ° 60 ° 50 ° 40 ° 30 ° 20 ° 10 ° 0 • 10 ° 20 °

/0- ~.0" 50" 40 ° 30 ° 20 ° 10 ° 0 o 10 ° 20 o

(i
/u bu- .b(J- 40" 30" 20 ° 1Q ° 0 ° 10 ° 20 °

FEe. 3. Mean atmospheric sea level pressure (mbar) for (a) January, and (13)July (adapted
from HO~_JcH, 1984).
6 R.G. Pm'~.~SONand L. S'm.~,~v,A

The surface wind patterns during January and July over the South Atlantic are shown in Fig.4
with wind roses in 10 ° squares. Clearly seen is a relative steadiness in each season, in both
direction and speed, of the trade winds near Africa and in the tropics. Most observations from
those regions, by far, are of winds weaker than Beaufort force 5 (less than 1 l m s-l). In contrast,
there is much greater variability in the higher latitudes where westerly winds prevail, in both speed
and direction, owing to the eastward translation of troughs of low pressure extending northward
from the subpolar lows. Within this band of westerly winds, the highest velocities are found
generally to the south and southwest of Africa, which is illustrated by the annual averages of
surface wind stress components in Fig.5. The highest values in the South Atlantic, greater than
0.2Pa, occur near 45°S 15°E. These result from the compression of isobars between the high
pressure ridge lying over the warm Agulhas Current (Fig.3) and the subpolar low pressure belt
displaced north by the zonal asymmetry of Antarctica east of the WeddeU Sea. This compression
of isobars also leads to the greatest values of wind stress curl in the annual mean, both positive
and negative, to occur south and southwest of Africa (Fig.6) with absolute values in each regime
exceeding 2x 10-7 Pa m -~.
The climatology of ~ and LEE (1981) was used to produce the annual-mean field of wind
stress curl in Fig.6 (S. PATIERSON,personal communication, 1985), and it shows a line of zero curl
extending from the southem tip of Africa to just north of the eastern promontory of Brazil. The
orientation of this line corresponds well with the separation of the Benguela Current from the
southern end of Africa and with its northwestward extension into the South Equatorial Current
(Fig. 1). North of the line are weakly negative values of wind stress curl, which are consistent with
there being a weak cyclonic gyre in the Angola Basin. The poleward limit to the subtropical gyre,
however, does not coincide with a line of zero wind curl; instead, the Subtropical Front better
corresponds in position with a maximum in curl oriented zonally across the basin near 40°S, close
to which should be a maximum in Ekman-layer convergence. But just how accurately we know
the distributions of wind stress curl in the South Atlantic is questionable to say the least. The
HELLEgUANand ROSENSTEIN(1983) wind stress climatology was used by GORDONand BOSLEY
(1990) tO present another field of the annually-averaged curl of wind stress, and in their field a
zonal maximum lies closer to 30°S in the central portions of the basin. This disparity in the
position of the wind curl ridge is the most obvious difference between the two climatologies, and
is likely a result of the poor density of atmospheric measurements from regions south of the major
shipping lanes. The same problem exists with the interpretation of the line of zero wind curl
extending zonally across the basin at latitudes of roughly 45°-50°S. In the central and eastem
South Atlantic it corresponds rather well with the position of the Subantarctic Front, which is
poorly-understood in terms of the dynamics responsible for it. Near the western boundary though,
in a region of complex bottom topography and flow pattems, there is little correlation between
the positions of these two features.
The fluxes of heat energy across the air-sea interface are strongly dependant upon season, as
is shown in Fig.7. Depicted in this figure are fields of net oceanic heat gain in January and July
due to the combined fluxes of sensible, latent, and radiative energy. Dominating these fields are
the latent and radiational fluxes, which for the most part are opposed in direction owing to strong
evaporation over the tropical and subtropical regions. Sensible heat exchanges are normally of
secondary importance. During the southern summer, the greatest heat gain (>200W m -2) occurs
in the region of cool, upwelled water off the western coast of southern Africa. Though the along-
shore winds and evaporation in that region are strongest then, sparse cloudiness and high solar
elevation have prevailing effects. Large heat gains also exist in the cool regions of the Falkland
Current and the Antarctic Circumpolar Current. The only region of heat loss (~50W m -2) found
south of the equator during summer, as a consequence of strong evaporation, is near the warm
Agulhas Current.
 Upper-level circulation in the South Atlantic Ocean 7

70* 60 ~ 50" 40* 30" 20" 10" O* I0" 20"

" ~ 3 tii!!iiRiiiii:.::::iiif~i:.~;~ii::ilRiiii~iii
iiiiii~!:.:i::. i i . : ~-~

(a)
70 ° 60* 50 ° 40 ° 30* 20 ° 10" 0 ° 10" 20"

70" 60" 50" 40" 30" 20" 10" 0" 10" 20"

it!J!ii:2ieRi~i
i: i2:siiiU
ii i:i:iii r:ii i:ii ,~ % ~ '~ -

iiii~i~:i: . ~ ~ , ~!i!iiiiiiiiiiiiiiiim!iiiii~t~

(b)
70 = 60" 50" ,40" 310= 20 = 10" 0 ° 10 = 2,0 =

I~Q. 4. Wind roses for 10 ° squares for (a) January, and (b) July (HOocH, 1984). Lengths o f the arms
indicate the mean relative frequencies (according to the scale in upper left) o f eight wind directions
divided into four groups o f wind force. From the center outward they are Beaufort 1-3 (0.5-5.5m s
1), 4-5 (5.5-11m s-I), 6-7 (11-17m s-t), and >8 (>17m s-l). Numbers within the circles are percent
occurrences o f calm or light and variable winds.
8 R.G. P~rexsoN and L. S ~

60" ~ 40" 30 ° 20* IO* 0 too" 20* 30"E

GO' 50* 40* 20"

°~ ) *-'

(b)

FIG.5. Annual averages of (a) x-component and (b) y-componentof wind s ~ s s ( 10"2Pa;B ~ , 1988).
Upper-level circulation in the South Atlantic Ocean 9

r~

;)
10 R.G. P~rmtsON and L. S ~

70* 60* 50" 40* 30* 20* 10* n* 1~* ~n*

(a)
~u OU :gU- 40- 30" 20 ° 10" O* 10" 20*

I
70* 60*
"
50*
0 -
40* 30*
"
20*
-75 --
10" 0* 10" 20*

0° " ~

0 10"

20 20°

30 -2
30 °

40°~ ~10/0~;~/Il ~ . ~ ~ ~ ~% ~ :~0"

70* 60* 50* 40* 30* 20 ° ']0" O* 10" 20*

Fie. 7. Net heat gain by the sea (W m 2) for (a) January, and (b) July (HOFLICH,1984).
 Upper-level circulation in the South Atlantic Ocean 11

The highest sea surface temperatures in the Atlantic are found at the position of the atmospheric
intertropical convergence zone (HOFLICI-I,1974; LAMB,1977), a zone which moves from just north
of the equator in January to some 10°-15 ° farther north in July. As a result of much reduced
cloudiness in July, combined with an upwelling of cool water which acts to reduce the flux of
latent heat, the equatorial region experiences heat gains approaching 100W m -2 (Fig.7b). Heat
gains of nearly this magnitude also occur in the upwelling region inshore of the Benguela Current,
and although the contours suggest a continuity between these two regimes, they are in fact separate
features. Of greater magnitudes during the southern winter though, are the heat losses in regions
of strong evaporation, which are supplemented by losses of sensible heat on the order of 50W m-
2 in the southern Antarctic Circumpolar Current and 30W m -2 off southern Africa and on the
Patagonian shelf. The most notable net energy fluxes during southern winter are the values
approaching -250W m a in the Agulhas Current region south of Africa where strong and cold
winds blow over warm, subtropical water. Large heat losses resulting primarily from evaporation
also occur over much of the central South Atlantic during winter when the sea surface is warmer
than the overlying air. In the Falkland Current region, the air-sea temperature difference tends
toward zero and the fluxes of sensible and latent heat are minimized. These combine with a north-
south gradient in the radiation balance to produce relatively weak fluxes of about -25W m -2near
40°S 57°W (Fig.7b).

3. SUBTROPICALAND EQUATORIALCIRCULATION

In this section we give an overview of the various upper-level currents comprising the South
Atlantic subtropical gyre and those equatorward. A convenient starting point is the Agulhas
Current Retroflection off southern Africa. Although it occupies just a small fraction of the South
Atlantic, the Agulhas Retroflection is nonetheless important for the inter-ocean exchanges of
water masses taking place there. We then proceed to the Benguela Current, the equatorial currents
and circulation in the Angola Basin, the large-scale and interannual variability in the lower
latitudes, and then around the rest of the subtropical gyre with the Brazil and South Atlantic
currents.

3.1. Agulhas Current Retroflection

The Agulhas Current is the western boundary current of the southernmost of two cells
comprising the subtropical gyre in the Indian Ocean (WYRTra, 1971; GORDON,Ltrrna-IARMSand
GRONDLINGH,1987), a current lying along the southeastern continental shelf of Africa where it is
dynamically similar to its Gul fStream and Kuroshio Current counterparts (DIm~cn, 1936; PEARCE,
1977). The southern Agulhas Current flows southwestward as a narrow jet along a steep
continental slope and is normally pinned to within 10-15km of its mean position at latitudes of
28.5°-34°S (GRt3-~DLI~GH, 1983). Large meanders can sometimes occur within this region
(PEARCE, 1977), which according to satellite observations are solitary events (the "Natal pulse")
extending an average of 170km offshore with downstream propagation rates of approximately
21 cm s-I; these rates decrease downstream to 5cm s ~ as the continental shelf broadens near 34°S
(LtrrJEI:ARMS and ROBERTS,1988). The current then separates from the coast and continues
southwestward along the Agulhas Bank (see Fig.8 for a schematic representation) where
numerous meanders, plumes and eddies exist (LtrrmnARMS, CATZELand VALErCrn,m, 1989). The
growth rate of the Agulhas Current from Durban (30°S) to the middle Agulhas Bank has been
estimated from quasi-synoptic data by GORDON, LtrrJEHARMSand GRONDLINGH(1987) as being
12 R.G. P~ImCSONand L. S'mAMMA

2.7Sv per 100kin (1 sverdrup = 1061113S1),less than half the 6Sv per 100km obtained by GRONDLINGH
(1980) from historical data between Port Edward (32°S) and the southern Agulhas Bank. The
maximum volume transport of the current is attained in the vicinity of the Agulhas Bank, where
depth-integrated values relative to the bottom are in the range of 95Sv (GoRoON,LtrrJErlARMSand
GRONDLINGH,1987) to 136Sv (JACOBSand GEORGI,1977). These values are significantly higher
than the 40-60Sv obtained by applying the Sverdrup relation to the curl of wind stress along 32°S
in the Indian Ocean (Fig.10 and Table 3 in Hm.L~RMANand ROSF2~SaXIN,1983), leading GORDON,
LtrrmHARMSand GRt3-NDLINGH(1987) to conclude that the Agulhas Current experiences significant
enhancement within a recirculation cell.
For the region offDurban (30°S), PEARCE(1977) defined the core of the Agulhas Current to
be where surface velocities exceed 100cm s-1. On the basis of directly-measured upper-ocean
velocity profiles, he found the current core on average to be 34krn wide with a mean peak speed
of 136cm s~; the greatest individual peak speed was 245cm s-1,in good agreement with maximum
velocities of 250cm s-1indicated by ship drifts (BARLOW,193 l) and 260cm s-1by satellite-tracked
drifting buoys (GRONDLINGH,1977; 1978). A directly-measured surface velocity in excess of
200cm s-1was also obtained by GRt3"NI)LINGH(1980, his Fig. 2) offPort Edward (-31.5°S). Speeds
of this magnitude, though, have not been observed with geostrophic shears alone; relative to
1500dbar on the eastern side of the Agulhas Bank near 36°S (where the peak speeds are expected
to be less than farther north because of the southward broadening of the current), GORDON,
LtrrJEHARMSand GRONDLINGH(1987) obtained their greatest geostrophic surface speed of 110cm
s-1. Observations from an array of current meters moored south of Africa for a two-year period
showed velocities off the Agulhas Bank at depths of 200m and more of 80-100cm s-:, with most
of the variability occurring in the period band of 80-120 days (LtrrrEN, 1988). Although these
current meter records revealed significant variability at the longest periods resolved, historical
data have yet to show any clear seasonality in this poleward flow (PEARCEand GRONDLINGH,1982).
Near 36°S the Agulhas Current leaves the continental shelf, develops oscillations of increasing
amplitude (HARRISand BANG,1974), and usually in the region of 16°-20°E retroflects back toward
the Indian Ocean (LtrrJEaAR~Sand VANBALLEOOOYEN,1988) as the Agulhas Return Current. The
retroflection loop encloses a pool of Indian Ocean surface water south of Africa whose
temperature is more than 5°C wanner than South Atlantic surface water at similar latitude
(GoROON, 1985), but the loop has on one occasion been observed to contain a cold-core eddy which
transported 64Sv of water (relative to 2400dbar) in a direction opposed to that of the surrounding
anticyclonic ridge (BE)~Err, 1988). Downstream of the retroflection, bathymetry in the vicinity
of the Agulhas Plateau produces an equatorward topographic Rossby wave in the return flow,
which has been observed with hydrographic and XBT (expendable bathythermograph) data
(BANG, 1970; DARaYSHmE, 1972; HARMSand VAN FOREEST, 1978; GORDON,LLrrJEnAR~Sand
GRO~-OLINGH,1987), in satellite infrared (IR) images (HARRIS,LFX;ECraSand VAn FOREF.ST,1978;
LtrrJEHARMS,1981; L t r r ~ s and VAn BALLEC,OOVEN, 1988), and in the trajectories of satellite-
tracked drifting buoys (GROr~LInOH,1978; Fig.6d in H o ~ , 1985). Infrequently, the core of
the Agulhas Return Current can pass over the Agulhas Plateau (BEr~mrr, 1988).
The Agulhas Current is highly baroclinic, as more than four-fifths of its transport occurs in the
upper 1000m (DuNcAN, 1970). Consistent with the high baroclinicity are water mass distributions
showing Indian Ocean water to be restricted to the upper 1500-2000m of the retroflection, beneath
which are waters from the Atlantic and Southern Oceans (GORDON,LtrrnmAm~Sand GROr~L~GH,
1987; BE~r~rr, 1988). But not all of the Indian Ocean water carried into the retroflection turns
back with the Agulhas Return Current (R~'~LL, 1832; Dxm'PaCri,1935; TAFT,1963; SHANNON,1966;
BANG,1973); a small portion of it leaks into the Atlantic. It is not known how much of this leakage
 Upper-level circulation in the South Atlantic Ocean 13

5°E 10 ° 20 ° 30 ° 35°E
25°S
• . . . .

, L0deritz
• .. '.': :'?:
- . . . . • • - . . . .

: ::i:i
:.:.. : • • . i.. 30 °

:.:.. - • . . ° "..'.::

• : '. . :" ..'. 'po.-.

".. " " .' - " .'- " .' : . ". Elizabeth.!
Capetown • .'- : - : : : : ,: "." : ~

35 °
©
SOUTH A T L A N T I C ~ C U R R E N T

AGULHAS

40 °
~ ..'.i'::ii::iiii? ' " q

taidg ~ .....:.:.::.:.::'
•" " "
Agulhas
~ee~ e° eddy

- - 1000 m
ANTARCTIC CIRCUMPOLAR CURRENT
ii:i:i:::! ! < 3 0 0 0 m

45°S

Fio. 8. Schematic representation of the Agulhas Current system (adapted from LtrrmHARMSand VAN
BALLEC,OOYm~,1988 andLtrrmmu~MS,1989). Open arrows indicatethe general direction of the surface
geostrophic currents.

occurs on a continuous basis, but according to satellite IR images there are times, apparently not
common, when an unbroken extension of the Agulhas Current can round the southern end of
Africa and directly feed Indian Ocean water into the Benguela Current (SHANNON,LtrrmHARMSand
ACENBA6, 1989). Estimates for the rate of this interocean transfer in the upper 1500m or less range
from 3 to 14Sv (Table 1), or a tenth or less of the total Agulhas Current transport. But all of these
have been computed from data obtained on surveys crossing at least one Agulhas eddy. Most
recently, GORDON and HAXBY (1990) have estimated that just the process of Agulhas eddies
moving into the Atlantic can account for a time-averaged transfer of 10-15Sv in the layer above
the Antarctic Intermediate Water (nominally in the upper 1000m). GORDON(1985; 1986) has
proposed that this transfer of warm and salty Indian Ocean water into the Atlantic is an important
part of the global thermohaline circulation which must exist to balance the export of North
Atlantic Deep Water from the Atlantic Ocean by way of the Antarctic Circumpolar Current.
Results from an inverse calculation, however, suggest that a large net transfer according to this
"warm water route" hypothesis is inconsistent with the known circulation and that the "cold water
route" (northward flow of intermediate and bottom waters from the Antarctic) forms the more
important closure to the meridional thermohaline circulation (RINTOUL, 1990).
Within the region just south of Africa are levels of eddy kinetic energy, as estimated from
14 R.G. Pnrv_~souand L. S'mAMMA

TABLE1. Estimates of transfer of Indian Ocean water to the Atlantic from the Agulhas Current
Retroflection

Transfer
Author Data Reference (10qn3s' ) Remarks

HARRISand VANF O R ~ AfricanaH, Mar 1969 1000dbar 5 estimate uncertain

(1978)

GORDON(1985) Knorr, Nov-Dec1983 1500dbar 15

GORDON,LtrrmnARMSand Knorr,Nov-Dec 1983 1500dbar 10

GRONDLINGH(1987)

BENNE'rr(1988) Knott, Nov-Dec 1983 1500dbar 6.3 <9°C (300-800dbar)

Knorr, Jan-Feb 1984 1500dbar 9.6 "
Washington, Feb-Mar 1985 1500dbar 2.8 "

STRAMMAand P~rva~soN Knorr,Nov 1983 -3500m 8 in the upper 1000m

(1990)

GORDONand HAXBY(1990) Geosat altimeter census 10-15" above the AAIWb

"Time-averagedrate, due only to the migration of Agulhas eddies into the South Atlantic
bAntarctic Intermediate Water (~800m depth)

drifter trajectories, which are higher than anywhere else in the southern hemisphere. Depending
on the computational method used, these eddy energies range from approximately 1000cm2s 2
(JOHNSON, 1989) to 4000cm2s -1 (PAT'IERSON, 1985). Using a global data set of ship drift obser-
vations, W'rRTm, MAGAARDand HAGER(1976) found the eddy energies in the Agulhas Retroflection
region to be comparable to eddy energies in western boundary current regimes of the northern
hemisphere. Seasat altimetry gives similar results (CrmNEY, MARSH and BECKLEY, 1983). Con-
tributing to this region's high eddy kinetic energies, according to satellite IR images (LtrrmnAR~S
and VAN BALLEC,OOVEN, 1988), are an average of nine penetrations per year of the Agulhas
Retroflection into the Atlantic, each followed by the shedding of an eddy; the newly-shed eddies
average 320km in diameter and drift away toward the west at about 12cm s-'. Altimeter obser-
vations from GEOSAT, however, reveal a slower production rate of Agulhas eddies, only about
5 per year, and a slower mean drift of 5-8cm s -1 toward the northwest once they have crossed the
Walvis Ridge (GORDONand HAXBY, 1990). A sea-going survey has recently documented the
actual shedding of one of these eddies from the retroflection loop (LtrrmaARMS and GORDON,1987).
The transport around that eddy was 40Sv, while an older Agulhas eddy farther to the northwest
carried 35Sv (GORDON,LtrrmUaRMS and GRONDLINGH,1987). Depending on radial distance from
the center, surface speeds around Agulhas eddies range from 20 to 80cm s-' (OLSoN and EVANS,
1986). The sea surface at the crest of one such eddy stood 30-40cm above that of the general
surroundings (GORDONand HAXBY, 1990).
The eventual fate of Agulhas eddies is only now coming to light. Using thermocline
topography combined with a 2-layer diagnostic model, OLSONand EVANS (1986) concluded that
Agulhas eddies are among the most energetic in the world and should have lifetimes of 5-10 years
with potential importance to the structure of the South Atlantic subtropical gyre. McCARal~-EVand
 Upper-levelcirculation in the South Atlantic Ocean 15

WOODGATE-JONES(1990) detected an anticyclonic eddy in a hydrographic section near the eastem

side of the mid-Atlantic ridge at 5°W 23°S, and surmised that it had most probably separated from
the Agulhas Retroflection two winters earlier and had subsequently advected along with the large-
scale circulation of the subtropical gyre. GEOSAT altimeter data are confirming that this can
happen, and are providing evidence that Agulhas eddies can remain intact all the way across the
South Atlantic to the region of the Brazil Current (GoRDoNand HAXBY,1990). It appears that most
Agulhas eddies find their way into the South Atlantic subtropical gyre, but some of them can
migrate south into the Antarctic Circumpolar Current. The first indication for this was given by
DEACON (1937, p.75) when he implied that an observation of subtropical water at a position
normally south of the Subtropical Front was one of Agulhas Current water, but CLOW~ (1950)
argued it to be an isolated patch of warm Atlantic water instead. A disruption of the Subtropical
Front southwest of Cape Town in 1964 was reported by DtmcAn (1968) as being the effect of an
Agulhas eddy, one that also contained subtropical Atlantic water (VIssER, 1969). Other warm
eddies poleward of the Subtropical Front south of Africa have been observed (LurmHARMSand
VALEm't~, 1988a, 1988b), with at least one being composed of Agulhas Current water (LtrrJEHARMS,
1988).
Until recently, the dynamics controlling the Agulhas Current Retroflection have not received
much attention. Using a two-layer, wind-driven analytic model of the World Ocean without
bottom topography, VEROmS(1973) concluded that if the curl of wind stress in the Indian Ocean
were weaker, the Agulhas Current would be able to round southern Africa and supply large
amounts of Indian Ocean water to the Atlantic. VEROraS(1978) then included thermal driving in
his model and observed the retroflection to result from a surfacing of the thermocline on the
western side of the Agulhas. Effects of bottom topography on the course of the current were first
addressed by DARBVSHmE(1972), and again by LUrJEHARMSand VAN BALLEGOOVEN(1984), both
with an inertial jet model originally proposed for the Gulf Stream by WARREN(1963) and later
modified by ROBINSONand NItLER(1967). LtrrJEHARMSand VAN BALLEC,OOYEN(1984) deduced that
high bottom velocities (>5cm s -I) are required for the current to penetrate into the Atlantic and that
the location of retroflection critically depends on volume transport. Using fixed surface and
bottom velocities, but with different velocity profiles, they obtained paths in which no retroflec-
tion occurred with low transport and increasingly earlier retroflection with higher transport rates
(their Fig. 3c). But Ou and DE RUIrrER (1986) questioned whether bottom topography is really the
main steering mechanism of the retroflection in light of the persistent circulation pattern observed
and there being no substantial topographic feature in the retroflection area. They formulated a
reduced-gravity two-layer analytic model incorporating inertia (only in the upper layer) and a
curved coastline and were able to reproduce the retroflection with roughly the right dimensions.
They concluded that the current path is primarily controlled by volume transport (inertia) and the
beta effect. They also saw that if the volume transport is reduced by just 10% the modelled current
will continue around the continent without leaving the coast. DE RuIYrERand BOtrORA(1985) and
BOUDRAand DE RtnrrER (1986) further included temporal variability and friction in a wind-driven
quasi-isopycnic numerical model (BLEcKand BOUDRA, 1981) tO explain the retroflection in terms
of a change in the vorticity balance as the current leaves the South African coast. BOtn~RAand
CHASSIGr,~T (1988) and CHASSlGnETand BOLrORA(1988) used the same model with increased
realism; the former finding an increased importance in the divergent component of flow in the
retroflection, while the latter found the shape of Africa and the inertia and baroclinicity of the
current to be of primary importance for eddy formation, though the roles of barotropic and
baroclinic instabilities were unclear. The necessary conditions for each of these hydrodynamic
instabilities are met by the observed density field (BENNETr, 1988).
16 R.G. l~mm~SONand L. S ~

Although the models which have been used to study the Agulhas Current are often quite
different from one another, an aspect which seems to recur is that a weak Agulhas Current would
not retroflect, but would continue around southern Africa into the Atlantic. As Ou and DE RtarmR
(1986) put it, could it be that (a) the Agulhas Current never weakens sufficiently to go around the
African continent, or (b) other dynamical constraints such as those imposed by large-scale
circulation outside the immediate retroflection area prohibit this occurrence? It now appears,
however, that this occurrence is not so completely prohibited. During 1986 there was an abnormal
intrusion of Agulhas Current water into the southern Benguela Current that persisted for much of
the year (SHANNON,LtrrmnARMS and A6m,mA6, 1989). This intrusion took the form of a nearly
continuous stream of surface water (as shown by a satellite IR image in their Fig. 1) from around
the Cape of Good Hope to as far north as 31°S. While this warm intrusion of Indian Ocean water
was cut off in early December 1986 by the appearance of Subantarctic Surface Water from the
south, it demonstrates that part of the Agulhas Current can at times round the continent's southern
terminus as a continuous stream.

3.2. Benguela Current

The Benguela Current is the eastern boundary current of the South Atlantic subtropical gyre.
VEROrnS (1973) developed an analytic two-layer global model driven by wind, and remarked that
eastern boundary currents are required for geostrophically balanced circulation and set the stage
for the intense upwelling often observed. But he also attributed a different function to the
Benguela Current; it exits at least in part because of the abrupt adjustment of the thermocline depth
at the latitude where subtropical and denser waters meet, giving rise to a flow (almost a source
f o w ) that can be sufficiently intense for inertial processes to be important, as in western boundary
currents.
According to charts by Dm~Arrr(1961), Rein, NowL~r and Parze.~T (1977), FU (1981), TsucmvA
(1985), REID (1989) and S ~ and P~:r~_a~SON(1989), the Benguela Current begins as a
northward flow off the Cape of Good Hope before bending toward the northwest to separate from
the African coast at around 30°S while widening rapidly (Fig.9). In addition to being fed primarily
by the South Atlantic Current, the southern current band of the subtropical gyre ( S ~ and
PEa'ERSON, 1990), the Benguela Current can also receive Agulhas Current water as well as
Subantarctic Surface Water, the latter coming from perturbations in the Subtropical Front
(SHANNON, LtrrJEnARMS and AGEI~raA6, 1989). In terms of its vertical dimensions, the Benguela
Current is modest; REID(1989) has shown southward geostrophic flow all along the southern coast
of eastern Africa at 1500dbar (his Fig.20) and essentially no depth-integrated net flow within the
Cape Basin (his Fig.43).
Atmospheric conditions in the Benguela Current region are strongly influenced by the semi-
permanent high pressure system over the subtropical South Atlantic and by a low pressure system
that develops over southern Africa during the austral summer (Fig.3). The prevailing winds in
the region are from the south and southeast (Fig.4), which in turn drive an offshore surface drift
and coastal upwelling of cold, nutrient-rich water. This upwelling exists, with much spatial and
temporal variability, all along the southwestern coast of Africa from Cape Point (34.35°S) to Cape
Frio (I 8.4°S), supporting one of the world's richest fisheries (GRt~Pmts, 1981). The main upwelling
occurs along the southern portion of the coast during the southern summer and moves north during
winter (JonES, 1971), hut the maximum compression of atmospheric isobars along the coast
occurs in spring, and with it comes the most vigorous upwelling (ScrUpLE, 1968). The sharpest
fronts and most intense northward jets along the shelf-edge, however, may occur during summer.
SHANNON(1966) and BAN6 (1973) have found that vestiges of Agulhas Current water are almost
always present in the region offshore of the southern upwelling regime in summer, warm water
 Upper-level circulation in the South Atlantic Ocean 17

O* 10" E 20*

Fio. 9. Field of geostrophictransport of surface water (less dense than t~0= 27.05kg m-a,nominally
in the upper 600m)relative to the t~offi27.75kg m3 isopycnal(1700-2000mdepth) from Sr~,MMAand
PmeKSON(1989). Each flow line represents 3Sv. Also shown are the 1000m isobath (light line),
selected 2000m isobaths (dashed lines), and areas less than 3000m deep (shading).

are almost always present in the region offshore of the southern upwelling regime in summer,
warm water which would act to intensify the frontal jet. Because of it being an important fisheries
environment, the upwelling region around Cape Columbine (33°S) and the frontal jet which
develops there have received much study. As summarized by NELSONand Htrrca-i~Gs (1983), this
northward jet lies over the steeply shelving 200-300m isobaths, has typical speeds in excess of
50cm s -1, and is 20-30km wide. Farther south, BANG and ANDREWS(1974) obtained direct
measurements of the current in the upwelling region near Cape Town (34°S) in January 1973 and
observed an intense frontal jet, of similar dimensions to that off Cape Columbine, having a
maximum northward velocity of 120cm s -~ and a volume transport of about 7Sv.
The upwelling regime offsouthwestem Africa, and the associated frontal currents, have been
reviewed in detail by Nk'LSONand HLrrcn~GS (1983 ), SHANNON(1985), and C I - I A P ~ and SI-IAr~ON
(1985), and we do not attempt to duplicate those efforts. But with the many studies in this region,
there has developed an inconsistency in what the term 'Benguela Current' stands for. Tradition-
ally, beginning with its inlroduction in the late 1800s, this term has been used in reference to the
large-scale northward flow at the eastern side of the South Atlantic subtropical gyre ( K R ~ ,
18 R.G. ~'rm~sor~and L. S ~

1882; SWRDRta', JOHNSONand FLe~m~G,1942). But drawing from the locally evolved terminology,
HARTand CLrRRm(1960) suggested that usage of this term be reserved only for the region of cool
upwelled water and northward jets near the coast, proposing that the adjacent circulation of
subtropical water be called the 'Southeast Trade Wind Drift.' Adoption of this meaning for
'Benguela Current' can be found in a host of papers, such as DARaYSrn~ (1963), STAtlER (1964),
SHANNON(1966), SHANNONand VAN RuswiJcr. (1969), and NELSONand HtrrcmNGs (1983). More
recently, though, LtrrJEHARMSand STOCk'TON(1987) and LtrrmnaRMS and VALmcrir~ (1987) have
left out the term 'Benguela Current' altogether in descriptions of the coastal upwelling off
southwestern Africa, consistent with the large-scale view of the Benguela Current forming the
eastern limb of the subtropical gyre.
As compared with the upwelling processes along the southwestern coast of Africa, the offshore
portion of the Benguela Current, as the eastern limb of the South Atlantic subtropical gyre, has
been little studied. SVERDRtrP,JOHNSONand FLErCnNG(1942) estimated the northward geostrophic
transport of this current at 30°S as being 16Sv relative to a depth of about 1200m. STRA.~nAand
PErERSON(1989) noted that a northward transport of 18.7Sv in the upper 600m in the eastern basin
of the South Atlantic (east of 13°W) can be extracted from Table 7 in Wt3sr (1957), which were
based on observations made by the Meteor along 28°S. Again using these Meteor data plus an
inshore station at 25°S, WOOSTERand REIn (1963) calculated a Benguela Current transport of
about 15Sv relative to 1000dbar. Applying inverse techniques on the Meteor and IGY (Inter-
national Geophysical Year, 1957-1958) data, Fu (1981) derived a field of geostrophic transport
(his Fig.9a) showing the Benguela Current as carrying approximately 20Sv of surface water (less
dense than Go = 26.8kg m -3, nominally in the upper 500m) across 32°S, whereupon the current
widens considerably toward the northwest. Carded along with this northwestward flow is a 13°C
thennostad which is formed in the thermocline of the eastern South Atlantic subtropical gyre by
vertical mixing and with a possible contribution coming from the Indian Ocean (TsocHrYA, 1986).
Most recently, SrRAIVa~ and PET~SON (1989) used historical data to compute the offshore
geostrophic transport of the Benguela Current. They found that at 32°S the current is located near
the African coast with a northward transport of about 21Sv in the upper layer (Fig.9). Near 30°S,
the Benguela Current begins to separate from the eastern boundary as most of the flow turns
toward the northwest. This northwestward flow carries about 18Sv of surface water across 28°S,
whereupon it turns mainly toward the west to flow over a relatively deep segment of the Walvis
Ridge south of the Valdivia Bank. A surface current with northward speeds of 10cm s -1was observed
to flow along the western side of the Valdivia Bank, while another northward surface current of
about 20cm s -t was seen some 300km west of the bank. About 3Sv of surface flow do not leave
the Cape Basin south of the Valdivia Bank, but drift north out of the Cape Basin as a wide, sluggish
flow into the Angola Basin.

3.3. Equatorial Currents

3.3.1. Components of the system. Several multi-institutional research programs have been
conducted in the equatorial oceans in recent years, which for the Atlantic have included the GARP
(Global Atmospheric Research Program) Atlantic Tropical Experiment (GATE), the First GARP
Global Experiment (FGGE), the Seasonal Equatorial Atlantic (SEQUAL) program, and Pro-
gramme Franqais Octan Climat Atlantique Equatorial (FOCAL). These have led to considerable
improvements in the description and theory of equatorial processes and have added a wealth of
information to the literature; assimilating all of it is well beyond the scope of this paper. Instead,
we focus in this section on the basic upper-level structure and seasonal and interannual variability
 Upper-level circulationin the South Atlantic Ocean 19

in the lower latitudes of the South Atlantic. For more comprehensive reviews concerning
observation, theory, and modelling, the reader is referred to articles by ~ E R (1973), MooRE
and PHILANDER(1977), LEm'MAA,McCREARYand MOORE(1981), SmOLERand PItB..ANDER(1982),
SmDLER(1983), CANEand SARAC'Hm(1983), KNOXand ANDERSON(1985), and ERmSENand KATZ
(1987).
Ship drift data have long been used to describe surface currents in the tropical Atlantic, as, for
example, in the early work of RENNELL(1832). Classically, the Benguela Current has been por-
trayed as feeding into a broad South Equatorial Current (SEC) that flows westward to the eastern
promontory of South America, Cabo de S~o Roque, whereupon it bifurcates into the North Brazil
Current (sometimes called the North Brazilian Coastal Current) and the Brazil Current. This
bifurcation can also be seen in the trajectories of satellite-tracked drifting buoys (MOLINAm, 1983;
REVERDtNand McPHAOEN,1986). The North Brazil Current is the stronger of the two and continues
into the northern hemisphere, partly accounting for the net northward transport of upper-level
water from the South Atlantic to the North. This net transfer has been known for more than a
century and a half; RENNELL(1832) Was quite convinced of it, but was at a loss for a satisfactory
explanation of how it could be. SVERDROP,JOHNSONand FLEMING(1942) estimated the northward,
cross-equatorial transport of surface water as being 6Sv accompanied by 2Sv of intermediate
water and 1Sv of bottom water, all being compensated by a southward transport of 9Sv of deep
water. Using zonal sections along 24°N, 8°N, 8°S, and 24°S to find a balance between geostrophic
and Ekman flows, ROEMMICH(1983) computed a net upper-ocean northward transport of 10Sv
across each section. In the upper ocean between 8°N and 8°S he found a geostrophic convergence
of 25Sv compensated by an Ekman divergence of the same magnitude.
The equatorial current system is marked by variability occurring over wide ranges of time and
space scales. Being that the Coriolis parameter vanishes in the presence of density stratification,
the equatorial ocean can respond strongly to changes in the overlying wind fields. ERIKSENand
KATZ (1987) have pointed out that given a mean thermohaline state, equatorial variability is
essentially driven by atmospheric forcing. And because of the low-latitude reduction of time
scales and associated increases in space scales for baroclinic disturbances, this variability is
dominated by wave-like motions, whereas eddy dynamics are more important at latitudes higher
than about 5 ° (SmDLER, 1983). Fluctuations of current velocity in response to seasonal changes
in equatorial wind stress fields should therefore be mainly zonal along the equatorial wave guide,
and this has been statistically observed in ship drift data (RICHARDSONand WALSH, 1986). Also
from ship-drift data, RICHARDSONand WALSH(1986) have derived maps of mean surface velocities
in 2 ° latitude by 5 ° longitude boxes, which for the austral summer (February) and winter (August),
together with the annually-averaged speeds, are shown in Fig. 10. In their maps, the austral winter
strengthening of the North Equatorial Countercurrent (NECC), which is fed by the North Brazil
Current (PHILANDERand PACANOWSga,1986a,b) is clear, while in each season the SEC has strong
westward flow (about 30cm s-') near the equator and weaker westward flow (10-15cm s -I) in a
broad band south of 10°S. RICHARDSONand WALSH(1986) noted that the stronger portion of the
SEC, that near the equator, has a banded structure in the annual mean; in the central part of the
basin there is a maximum at 2°N 22.5°W and another at 4°S 17.5°W.
The same data have been used by ARNAULT(1987), and Fig. 11 shows her maps of the annual
means of surface currents in 2 ° latitude by 4 ° longitude boxes in the tropical Atlantic from ship
drifts, together with those from geostrophic computations relative to 500dbar and from compu-
tations of Ekman drift. Large differences appear between the ship drifts and surface geostrophic
currents, but, except at the equator where Coriolis accelerations vanish, the differences are well
accounted for by the Ekman drift which has larger amplitude in most regions than the geostrophic
20 R.G. F~,m~sor~ and L. S T p , . ~ ,

20.~ 70"W 60" 50" 40 ° 30" 20 ° 10" O" IO'E

10

IC

(al
20%,

20 o,,

1(

(b
20 ° _

20"N

10"

0°

10"

(c)
20"S
70"W 60" ,~D" 40 ° 30" 20* 10" O* IO'E

FIo. 10. Mean surface velocities from ship drift observations in 2 ° latitude by 5 ° longitude boxes for
(a) February, (b) August, and (c) annually-averaged speed in cm s"1(RIcna~soN and WALSn, 1986).
 Upper-level circulation in the South Atlantic Ocean 21

component. For the band of 2°-8°S, she found westward ship drift at all longitudes and all months,
in agreement with the Ekman Drift, whereas east of 8°W the surface geostrophic flow is eastward
in all months except June to September. It is interesting to note that in those months where the
surface geostrophic flow is no longer eastward in the Gulf of Guinea and northern Angola Basin,
there is also no Angola Dome present in the thermocline, a small-scale feature described in the
following section. MERLE and ARNAULT(1985) found the maximum variability of dynamic to-
pography (relative to 500m) in the tropical Atlantic to occur in the Gulf of Guinea and north of
the equator beneath the intertropical convergence zone (in the region of the NECC), whereas the
seasonal variability is small in the entire western and central South Atlantic north of 20°S.
The system of upper-level currents near the equator comprises a complex array of westward
currents and eastward countercurrents. A revised version of DEFANT'S (1936) generalization of
equatorial currents in the Atlantic is contained in the paper by KNOLL,ZENK and BAUER (1982).
While many of the three-dimensional details of how the bands are arranged across the width of
the Atlantic and how water is transferred from one band to another are not yet worked out, there
are still some generalizations worth making, but with the caveat that space-time variations will
often produce a picture quite unlike the idealized, southern winter representation shown in Fig. 12.
As illustrated, the NECC lies roughly between 3 ° and 10°N and is the northern boundary to the
multiple-banded SEC; both of these currents are at their strongest and are at their most northerly
positions during the austral winter.
As seen in Fig. 10, the SEC in the central Atlantic has velocity maxima just north and south of
the equator. The relative minimum at the equator corresponds in position with that of the
Equatorial Undercurrent (EUC), a narrow jet having a half-width of about 1.5 ° latitude and
eastward velocities often well over 100cm s -1in the upwelled thermocline a hundred meters or less
from the sea surface. METCALF,VOORHISand STALCUP(1962) and Prm.ANDER(1973) have sum-
marized the discovery of this relatively saline and oxygenated current. On the basis of
temperature-oxygen relationships, METCALFand STALCtrP(1967) found that most of the water in
the EUC comes from the South Atlantic thermocline by way of the North Brazil Current with very
little coming from the North Atlantic. This is also shown by the analysis of historical data by
TSUCHIYA(1986) and by the general circulation model of PHILANDERand PACANOWSKI(1986b). In
the thermohaline fields, this jet is prominent as a local vertical spreading of the thermocline and
as a high salinity core with values of 36.5 or more in the western Atlantic as opposed to 35 to 35.5
in the surrounding water (Mm~ALF, VOORmSand STALCUP,1962). As an average at 22°W, FAHRBACH,
MEINCKEand SY (1986) found the salinity maximum of the EUC to be located 30km south of the
velocity maximum.
In general, the EUC is most intense in the western Atlantic and weakens as it flows east, with
observations made during GATE (DOB~G,HISARD,KATZ, MEINCKE,MILLER,MOROSHION,PHILAN-
DER,RmNIKOV,VOIGTand WFaSBERG,1975) showing that the core of the current meanders between
0°50'N and 0°50'S over time scales of 2-3 weeks (refer to Figs. 13a-d for a half-cycle at 28°W),
consistent with a long (2600km), unstable wave propagating westward with a phase speed of 1.9m
s -1. Along with the meandering, the EUC also displays a pulsating behaviour; the maximum core
velocities at 28 °W varied from around 60cm s-1to more than 120cm s -1, and the volume transports
obtained by integrating the velocities within the 20cm s -1 isotachs ranged from 4 to 15Sv. The
eastward weakening of the EUC is illustrated in Figs. 13e-g, where the maximum core velocities
decreased from over 80cm s -~ at 12°W to only half as large at 4°W, and to just over 30cm s -1 at
4.5°E. A feature to note in the easternmost section is that in addition to the EUC core near the
equator there was another, and larger one centered near 1.5°S. This may have been an early sign
of part of the current turning into the southern hemisphere as it approaches the eastern boundary.
Near the eastern boundary, at 6.5°E, RINKEL,SUNDand NEUMAr~ (1966) observed the high salinity
22 R.G. Pnr~a~soN and L. STRAMMA

~0- to* so- 50" 40- 30- zo" ~oow o- ~O-E zo"
3o ° ~ ~ ~ ~ 1 ~ - ~ , . ~ ~~ ~ ~ , - . ~ ~
-~
~),', :'"
'~:::::::::: ..::i:i:i:!:!
~"'~ o.,~ 1

~ ~ ~ ~ *" ~ ,,.. ~ ===================================================================

0 N ~ ~ .............. ..,...,., . . . . . . . . . . . . . . . -..

~ ~ . ' . " : i : ~ "~'-'-- " - - ~ ~ ======================

0 o ~-:.:.::;~-~_~.,-----,--. ,--~__,--_._,---.. ,----"-.r-.. -.-.. ,.. 3 "'/_::::!:!:::::::t~
i~r:~:::: :::::::::::::::::::::::::::::::::::::::::: : l , ~ p ~ - ~ - - ~ - ~ ~ ~ ,----- "----I,---- ~
~ --, e,.:.:.:.:.:.:.:.:~:l~
:::::::::::::::::::::::::::::::::::::::::::::::: ,---~-~ ,-._ ,... ~:!:::::::~
~::::::::::::::::::::::::::::::::::::::::::::~....... ~ ' ~ ~ ._ - .__ . . . . ~. ~ ~i:::::::::::~d
0 -.1- "-'." "-','.','.'.'.'.'.'.'.'.'.'. ..................... ' I1~ ~ ~ .....

( a ) ~1.......
~:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::~:,-if~'~:::
. . . . .;;;;;;;,'fff,',',',,',",',',;;;;;;;
. . . . . . . . . . . .
: ~
l. l .m . ~. . ~.
~- 6.--n~,--,._'"~ .-'- *-*J--,,..
~ r' '
,,.. ,,--" ",,. "~ '~,t
'11 T ] ~ l l ~ n ~ l I . . . . . . . . . . . . . .~
......
':::':':
. . . . . ~:,:.:.:.:i}~

.... ~. . . . . k. . . . . I .... tri..=l. .... L .... ;. . . . . I .... k---- I

-,~ , ~ ~ , ===================== u AL M E,

.- ," Z .-" Z " ; , , ~ " ~ ~:--:.iii!i!~iiiii!!i:::::::::::::::::.:.:.:.:.:.:.:.:.:.:.:.:-:.:.:iR

o N~ ~.~ ' ~ ~ : ~ ............:.:.:.:.:.:.:...:...:.......-~

0 ° ! ii!:!:iiii!iiii:ii!!!}!i i i ! ~ i ~ ~ iiii
2. • : • : : : : :':':':':';'F:':':.F:.2.: . . . . . . ' 2':'i-;-2.2. -
==================================================================================. ~ ~ ",, ":';'F:.:.:.

0 S ..'.v.i.'.!.:.:.:.:.:.'i'iii'ii%:.:..:.:. . . . . . . . . . . . - , . . . . . . . . . .:.:.:::
Z

30"

\ " \ -;'FF'.FF:.I.'.

\ ~ ' . ' - : : : : : : : 1 ~
~ ~ ~ ~ : ~ : : ~ t ~
{C~ g ~ u ~ ~ ~ ~ I E I - -
" ' 80* 70* 60" ,50* 40" 50" 20" IO'W C)" IO'*E 20"

FIo. 11. Surface fields of annual mean (a) ship drift, Co) geostrophic velocity relative to 500dbar, and
(c) Ekrnan drift, all in 2 ° latitude by 4 ° longitude boxes (ARNAtmT, 1987).
 Upper-level circulation in the South Atlantic Ocean 23

I 0 *IV

oo III1

=Ecc lj --> EUC

C
/5*S ~~ " - ~ NEUC

sEuc
)OOm
SECC

200m 0 I00
I | i I I I I ' ~ ' I
cm s -I

l~o. 12. Idealized representation of austral winter currents at 0, 100, and 200m depth in the near-
equatorial region of the central Atlantic. Abbreviated terms are: North Equatorial Cotmtercurrent
(NECC), South Equatorial Current (SEC), South Equatorial Countercurrent (SECC),
Equatorial Undercurrent (EUC), North Equatorial Undercurrent (NEUC), and South Equatorial
Undercurrent (SEUC).

core of the EUC to bend southeast before terminating 150krn from the African coast near 1°20'S.
But on the basis of T-S curves, WAtrrrIv (1977) has traced EUC water to as far south as 13°S. And
most recently, WACONOr~ and PrroN (1990) have surmised that the EUC is probably continuous
with the Gabon-Congo Undercurrent along the African coast at 1°-6°S, which in turn supplies
water to the Angola Current farther south. This will be discussed in the following section.
The existence of the EUC has traditionally been thought of as being a direct result of westward
pressure gradients associated with a zonal tilt of the sea surface: surface water accumulates in the
west by the frictional action of the wind, and subsurface water returns to the east under the action
of the pressure gradient force (i.e., ARTrltm 1960). Because of the change in sign of the Coriolis
parameter at the equator, the westward winds lead to surface divergence and upward motion along
the equator which are compensated by convergence at depth. The three-dimensional circulation
involving the EUC has been addressed by many investigators, as PmLAm~F_~(1973) has docu-
mented in detail. Recently, BRVDENand B ~ Y (1985) studied the equatorial circulation in the
Pacific with a diagnostic model and pointed out that the vertical motion occurs largely on the zonal
plane in which eastward-moving fluid elements rise along upward sloping isopycnals. Because
cross-isopycnal mixing was not needed for the vertical motion, PEOLOSm"(1987) Was prompted
24 R.G. Pure_xsoNand L. STRAMMA

I°N 0° 1°S 2ON 0o 2°S Om

Om

200
om t ii!ii!;i~iiiii!i!iii~ 60ii:!:i: 0 r n
~!~iiiii!.!:i:

0 rn ........................... •............................ !!iiii!!!!iso!~i::~::!::!:: I |

200

0rn

0m

-(g) ~ ~ l I

200 I°N 0° 1°S 2ON 0o 2os 200

FIG. 13. Meddional sections of directly-measured eastward velocity (cm s"~) in the upper 200m il-
lustrating the Equatorial Undercurrent (shading denotes westward flow). Interval between solid
contours is 20cm s-~. The fields on the left were all measured at 28°W in the year 1974 from R/V lselin
during (a) July 26-28, (b) July 30- August 1, (c) August 2-3, and (d) August 4-6 (adapted from Dt.~o,
HISARD,KATZ,MI~CKE, Mn., ~ , MoROSl-rg~,~ m , RISNIKOV,VOlOTand WE]SBERO,1975, by
permission fromNature, Vol. 257, pp. 531-534. Copyright for original: 1975 Macmillan Magazines
Ltd). The fields on the right were measured during April-June 1964 from MS Prof. A. Penck a t (e)
12°W, (f) 4°W, and (g) 4.5°E (adapted from S ~ and VGIOT,1966).

to extend a simple argument based on the conservation of potential vorticity (FoFoNOFF and
MONTC~MERY, 1955) to explain the EUC. He developed an inviscid, non-linear, two-layer inertiai
model in which the Bernoulli function was conserved along the equator and where the far-field,
extra-equatorial solution was matched with that of the ventilated thermocline model o f LtrrrEN,
PEDLOSKYand STOMMEL(1983). But this purely non-frictional, vorticity-conserving model was
unsatisfactory in that it produced an eastward accelerating current which is not observed. Using
the numerical simulation of PHII.ArCOERand PACANOWSKI(1984) tO compute zonal m o m e n t u m
balances in the equatorial Atlantic, WACONGNS(1989) found that only near the western boundary
 Upper-level circulation in the South Atlantic Ocean 25

does the subsurface eastward pressure gradient force exceed westward frictional forces, thus
giving rise to an eastward current beneath westward surface flow; away from the western
boundary friction dominates and the net force is westward, causing the undercurrent to weaken
as it flows east. Dissipation in the form of cross-isopycnal entrainment of EUC water into the
upper mixed layer was added by P ~ L O S ~ (1988) to his earlier model to get an eastward-
decelerating EUC, and in the third paper of the series, P~LOSKY and SA~LSON (1989) added surface
wind stress. This final model makes the three-dimensional linkage between the structure of the
EUC and the equatorial winds, the extra-equatorial thermocline, and the conservation of potential
vorticity and Bernoulli function. But this equatorially-symmetric solution is not completely
applicable to the Atlantic EUC which gets most of its water from the South Atlantic therm0cline.
Distinct from the thermocline EUC are deeper countercurrents just beneath the thermocline.
In the central Atlantic, at 15 ° and 20°W, ~ (1968) observed a narrow, eastward geostrophic
current centered near 4°S having maximum velocities of 30-60cm s-1 (relative to 600dbar) near
150m depth. West of 25°W, COCI-mAN~,KELLY and OLLn~G (1979) observed the presence of
subthermocline countercurrents having maximum eastward geostrophic velocities of 40-50cm s
1 (relative to 800dbar) centered near 4.5°N and 4.5°S; in the south the maximum velocities were
found near 200m depth and were largely covered by westward flow at the surface. On average,
the southern current was 209km wide and transported 15Sv. MOLINARI,VorrtrRmz and DUNCAN
(1981) proposed the names North Equatorial Undercurrent (NEUC) and South Equatorial
Undercurrent (SEUC) for these subthermocline eastward flows following a convention already
in use in the Pacific. In the western and central Atlantic they found the SEUC to be typically
located between 3 ° and 5°S with an average width of about 100km and a core depth of 150 to
200m. The average geostrophic transport of this current in the western Atlantic was also 15Sv
(relative to 1000m). Using additional data at longitudes of25°-28°W, MOLmARI(1982) observed
the EUC transport to range from 5 to 23Sv, with maximum core velocities of about 40cm s -a
(relative to 1000m) near 200m depth; the overlying surface flow was at times eastward and at
others westward. The source of the SEUC is the region off the coast of Brazil (CoCHRANE,KELLY
and OLLING, 1979), and it appears that this current extends all the way across the Atlantic to feed
water into the northern Angola Basin (Vorrtnum, 1981; Vowopam and ~ B L A N V , 1982).
MOLINARI(1982) observed the SEUC to lie between two branches of the SEC, the northern one
being between l°S and the SEUC with a geostrophic transport ranging from 6 to 23Sv (relative
to 1000m) toward the west, while the other was just south of the SEUC with a comparable
westward transport of 7-26Sv. South of the second branch, at 7°-9°S, another distinct eastward
flow was observed, the South Equatorial Countercurrent (SECC). As with the SEUC, its
maximum geostrophic velocities were beneath the surface (10cm s -1 near 275m depth), but was
typically accompanied by eastward flow at the sea surface. Its geostrophic transport was
estimated to range between 3 and 7Sv relative to 1000m. The SECC was first observed by R~m
(1964a,b) near 14°W between 5 ° and 12.5°S in a section made in July 1963 in which maximum
eastward velocities of 9cm s -~were found at depths of 200-250m, also accompanied by eastward
flow at the surface. As will be discussed in Section 3.3.3, anomalous conditions were being
experienced in the tropical Atlantic at that time which may have given rise to a stronger than
normal SECC. As with the SEUC, it appears that the SECC extends all the way across the ocean
and turns southward in the eastern Angola Basin (MoRosHrIN, BtraNOV and BULATOV, 1970;
SHANNON, BOYV, BRUNOm'rand TAUrCroN-CLAm:, 1986).
From the thermostad investigation of TSUCHrYA(1986) and the inverse computations of Ftr
(1981), it is evident that the main westward current of the South Atlantic subtropical gyre lies
generally to the south of 10°S and reaches that latitude only near the coast of Brazil. MOLINAm
26 R.G. Pm'ERSOr~and L. S ~

(1982) estimated that the southern branch of the SEC (south of the SECC) transports only 3Sv
relative to 1000m, but the sections he used extended to just 11 °S, thus leaving the northern limb
of the subtropical gyre largely unsampled. Using historical data, STI~MMA, IKEDAand PewVaSON
(1990) found tile southern branch of the SEC to carry about 16Sv in the upper 500m (relative to
about 1200m) across 30°W (Fig. 14). They observed the SEC to bifurcate near 10°S in the months
of February and March, whereupon approximately 12Sv turned north into the North Brazil
Current and only 4Sv south into the Brazil Current. It is not known if there is any seasonal
variability in the location and character of the SEC bifurcation, as the available data are inadequate
for addressing this.
3.3.2. Angola Basin circulation. The northward surface drift out of the Cape Basin, noted at
the end of Section 3.2, together with the flow of cool, upwelled water along the African coast,
reaches to about 15 °- 17°S, depending on season, before coming into contact with warm surface
waters in the Angola Basin (SHANNON,AGEr,mA6 and BuYs, 1987). A sharp thermal front, first
observed by HARTand CORmE(1960), develops there and marks the southern boundary to surface
circulation in the Angola Basin. This front, the Angola-Benguela Front, is well developed in the
upper 50m, and can be detected in the salinity field to depths of at least 200m (SHANNON,AGEm3AO
and BuYs, 1987; GORDONand BOSLEY, 1990).
According to MAZF.IKA(1967), a feature which stands out in the upper-layer temperature
distribution during southern summer, but not winter, is a doming of shallow isotherms centered
off the Angola coast near 10°S 9°E. This "Angola Dome" is analogous to a similar feature, the
"Guinea Dome," which occurs during northern summer off the African coast near 10°N. These
features, and their seasonality, are depicted in Fig.15. MAZEIKA(1967) observed that the Angola
Dome is evident in historical means at depths of 20m to 150m, but can not be detected in fields
of sea surface temperature. Being that the Angola Basin was sampled in the IGY during the
southern summer, this thermal dome can be seen in the temperature section made by the Crawford
at 8°S, and to some extent in the section at 16°S (FurlJSaXR, 1960). Also according to MAZEIKA
(1967), salinities are 0.3-0.5 lower and oxygen values 2-3ml 1-1less within the Angola Dome as
compared with surrounding waters on equal isopycnic surfaces, and that upwelling is probably
involved in its formation (such upwelling would not be a part of the coastal upwelling regime
found farther south, which at this time of year is most vigorous near the southern end of Africa).
Vorrtmmz (1981) has presented evidence that although the Angola Dome is a seasonal feature in
the thermocline, it is probably permanent beneath the thermocline. According to him, the SEUC
(roughly 200m deep) induces a permanent subthermocline thermal dome when it turns poleward
near the eastern boundary, and that the thermocline portion of the dome corresponds to a seasonal
uplifting of the deeper structure when low atmospheric pressure and favorable wind stress curl
cause upwelling. Vorltlm~ and HI~.~I~ArCD(1982) have presented additional evidence for this, and
WACONGm~and PITON(1990) considered the southward turn of the SEUC to be a source of water
for a highly variable poleward undercurrent they observed in current meter measurements taken
at the continental shelf break at 1°-6°S. This undercurrent, which WACONOmSand PrroN (1990)
referred to as the Gabon-Congo Undercurrent, was observed to have an average velocity of 8cm
s -1 toward the south, and they conjectured that it feeds into the Angola Current farther south.
The first quasi-synoptic hydrographic survey of the Angola Basin was made by the Kurchatov
from 5 April to 9 June 1968, during the southern autumn. Measurements from that cruise were
used by MOROSrlK~, BtmNov and Bta~ATOV(1970) to observe a cyclonic geostrophic gyre less than
1000kin across (according to their figures) and centered near 13°S 4°E, more than 600km
southwest of the thermal dome. The gyre extended from just beneath a very thin (10-20m) wind-
 Upper-level circulation in the South Atlantic Ocean 27

40 35 °W 30

. . . . . . . . J

E
Vitoria-Trindade Ridge

40 35 °W 30

FIo. 14. Schematic representation of the geostrophic flow field in the upper 500m based on historical
hydrographic data within the region defined by the dashed box. Solid flow lines represent a transport
of 2Sv, disconnected arrows a transport of 1Sv. The location of where the South Equatorial Current
bifurcates is based on data obtained in the months of February and March. Reproduced from Sa'~,,~tzv~,
IK~A and Prom<soN (1990).
28 R.G. PL'mRSONand L. S ~

30 ° W 20 ° I0 ° 0° -I 0 °- E 9.00
N

IO o

3o

I0 o

S
20 °
N

I0 o

0o

I0 o

S
20 ° ,

FiG. 15. Thermocline domes in the eastern Atlantic revealed by average distributions of temperature
(°C) based on bathythermograph and hydrographic station data from the years 1905-1963 for (a)
austral winter (July-September) at 50m depth and Co)auslral summer (January-April) at 20m depth
(adapted from ~ , 1967 by Vornm~, 1981).

driven surface layer to about 300m depth, and produced subsurface southward velocities of up to
50cm s -1 in a narrow coastal current which they referred to as the Angola Current. As V o ~
( 1981) noted, the Angola Dome observed by MAZmXA(1967) and the cyclonic gyre described by
MOROSHKIN,BUBNOVand BULATOV(1970) are two different phenomena. By inverting historical
data to estimate the absolute geostrophic circulation in the South Atlantic, ReID (1989) has also
found cyclonic motion north of the seaward extension of the Benguela Current, and his maps of
adjusted steric height show this gyre to be more of a basin-wide feature centered at roughly 10°W
and extending downward to at least 1500dbar as a recognizable, closed circulation. GORDONand
BOSLEY (1990) have subsequently observed a cyclonic gyre in a collection of station data taken
in the region during a number of hydrographic surveys in 1983-1984 (Fig.16). They found the
 Upper-level circulationin the South Atlantic Ocean 29

gyre to be centered at approximately the same location as did MOROSmON,BtrBNOVand Bta.ATOV

(1970), but Fig.3a in GORDONand BOSLEY (1990) indicate that it is a larger feature, perhaps
2000km across. Additionally, GORDONand BOSL~ (1990) observed the sea surface at the gyre's
center to be about 8 dynamic centimeters (relative to 1500dbar) lower than that of the surrounding
region, and that its baroclinic signature is confined mainly to the upper 300dbar. Surface
geostrophic speeds around the gyre were 2-4cm s -1, and the associated volume transport was
typically 5Sv in the upper 300dbar relative to 1500dbar. They also pointed out that the wind-
driven surface layer has larger velocities which effectively mask the gyre from being observed
with ship-drift data. Although the curl of wind stress is negative in the northeastern South Atlantic
(Fig.6), BOSLEY(1990) has argued, on the basis of a linear, two-layer model with wind forcing and
upwelling, that Sverdrup dynamics alone do not fully account for the gyre.
As opposed to the Angola Dome, which appears to be bounded to the north by the SEUC, the
larger cyclonic gyre is often considered to have as its northern boundary the SECC (MoROSHra~,
BUBNOVand BULATOV,1970; SHANNON,AGENBAGand BuYs, 1987; GORDONand BOSLEY, 1990),
which together with the EUC and SEUC probably feeds into the Angola Current. But the
southward flow in the region of the Angola Current appears to be significantly deeper than the
equatorial currents coming in from the north, at least at times. Using a composite of historical data,
KmWAN (1963) performed an objective, isentropic analysis showing southward flow off Angola
at depths of 300-1000m. His data in the Angola Basin consisted of both the Meteor and IGY
measurements. Fu (1981) applied inverse techniques on the IGY data and obtained similar results
off Angola at 8°S, i.e., southward flow from the surface down to the Upper North Atlantic Deep
Water. This, however, was not observed in the Meteor data at the same location, by neither Fu
(1981) nor WOST (1957). Those Meteor stations were occupied in late August 1926, whereas the
IGY stations, made from the Crawford, came in early March 1957, indicating that significant
temporal (perhaps seasonal) variability in the deeper flow occurs in the eastern Angola Basin.
REm's (1989) analysis of historical data show southward flow in the eastern Angola Basin to at
least 3000dbar, while GORDONand BOSLEY(1990) found only weak evidence of North Atlantic
Deep Water moving south in the region.
3.3.3. Large-scale variability and interannual warmings. The focus of this section is the
interannual variability that occurs in the tropical Atlantic, and to some extent the annual changes
as well, both of which have been likened to the El Nifio/Southern Oscillation (ENSO) phenomenon
in the Pacific. Here we will only mention a few of the more salient features of the Pacific ENSO
events; for more details regarding the meteorological and oceanographic aspects of these, the
reader is referred to papers by RASMUSSONand WALLACE(1983), CANE (1983) and KNOX and
ANDERSON(1985).
In the equatorial Pacific, the thermally-driven Walker circulation dominates the lower 10km
or so of the atmosphere. Under normal conditions, this circulation is characterized by rising
motion and precipitation in the far western Pacific and Indonesia, eastward flow in the upper
troposphere, subsidence of dry air in the eastern Pacific, and westward flow at the sea surface
which supports a zonal tilt in sea level. During a typical ENSO event, the intertropical
convergence zone (ITCZ), a narrow, zonal band of low-level convergence, shifts south and
consequently the zonal pressure gradient at the surface along the equator diminishes. The trade
winds near the date line and at locations farther west thereby undergo a relaxation, or even shift
to eastward flow. Packets of equatorially-trapped downwelling Kelvin waves are then excited,
which move eastward across the basin, and in a few months, arrive at the Ecuador-Peru coast
causing rises in both sea level and sea surface temperature (SST). The zones of heavy rains that
30 R.G. Pmv_~soNand L. STRAMMA

5°W O0 5°E 10" 15

5 ° '1 -f "1 1 1 5°S
205

D288

10 ° 10 °

12

15 ° 15 °
509
266

m 0 I

20 ° # 28 20 °
271 219 237

) 10 Sv
+ KNORR (AJAX)
o OCEANUS ~ 1500 db reference
• WILKES (NAVOCEANO) ~ 3 0 0 0 d b reference
25 ° ~1 2 5 ° S
5°W 0° 5°E 10 ° 15 °

1~o. 16. Map of the transport between the surface and 300dbar referenced to 1500dbar. Values using
a 1500dbar reference level are solid, while those using a 3000dbar reference level are open. The
eastern stations do not extend to 3000dbar. Station dates: 132, 154 ( O c e a n u s ) - March 1983; 15, 28
(Knorr) - October 1983; 205, 212, 219, 237 ( W i l k e s ) - May 1984; 266, 271 ( W i l k e s ) - June 1984; 288
(Wilkes) - July 1984; 307, 309 ( W i l k e s ) - November 1984. From GOROONand BOSLEY(1990).

normally occur over the far western Pacific and Indonesia move east to normally arid regions,
leaving the western Pacific in drought.
There is no close analogy in the equatorial Atlantic (Fig.3) to the zonal atmospheric pressure
gradient in the Pacific, and consequently, there is no preferred region of heavy precipitation such
as that which normally occurs in Indonesia. Instead, a band o f precipitation is associated with the
ITCZ which extends across the Atlantic and seasonally migrates north and south between the
equator and about 15°N. It reaches its southernmost latitude in March and April, bringing a rainy
 Upper-levelcirculation in the South Atlantic Ocean 31

season to northeastem Brazil. In some years the rains fail, and such years of drought are known
as SEcas. CAVIEDES (1973) noted that El Nifio events in the Pacific and S~cas in northeastem
Brazil occasionally occur at the same time, a connection that was investigated by H A s ~ r n and
HELLER(1977). On the basis of maps of general atmospheric circulation in the tropics, they found
a strong negative linkage between rainfall in northeastern Brazil and SST along the Ecuador-Peru
coast, and that SScas characteristically occur when the South Atlantic subtropical high expands
equatorward and the ITCZ moves farther north. During those periods, the South Atlantic trade
winds become stronger than normal and anomalously cold surface water appears in much of the
equatorial Atlantic. They also suggested that an inverse relationship exists between long-term
atmospheric pressure variations over the eastern South Pacific and South Atlantic oceans, which
was in turn confh'med by COVEYand H A S ~ r n (1978).
As opposed to the tropical Pacific, where interannual variations in SST exceed weak annual
signals (see for example Fig.1 in B ~ , 1969), the tropical Atlantic experiences annual
changes in SST of 6°-8°C (HASrEnRATHand LAMB, 1977) that are substantially greater than the
interannual (MERLE, 1980). The restriction of the surface signature of equatorial upwelling in the
Atlantic to the northern summer season is the primary reason. Also during the northern summer,
seasonal upwelling occurs in the northern Gulf of Guinea, but apparently not as a result of local
winds or oceanic circulation (HooaHTON, 1976). Analogous to theories concerning the interan-
nual E1Nifio events in the Pacific where downwelling Kelvin waves are involved, MOORE,HISARD,
McCREARY, MERLE,O'BRIEN, PICAULX,VERSa~'TE and WtmscrI (1978) proposed that the annual
upwelling in the Gulf of Guinea results from wind excitation of an equatorially trapped upwelling
Kelvin wave in the western Atlantic which propagates eastward into the region. According to a
linear, two-layer numerical model formulated by O' BRmN,ADAMECand MOORE(1978) and ADA~mC
and O'BRmN (1978), an increase in westward wind stress over the western equatorial Atlantic
excites a disturbance that propagates eastward along the equator at internal gravity wave speeds
to later produce strong upwelling in the Gulf of Guinea; local winds could not account for this
upwelling. An adjustment phenomenon similar to the one believed to cause the interannual E1
Nifio in the Pacific is thus part of the annual cycle in the Atlantic (/VIERLE,1980), and one which
can be enhanced in the Gulf of Guinea by a surfacing of the thermocline (MERLE,Fmtrx and HISARD,
1980). Other numerical studies have produced similar results (for example, McCREARY, PICAtrLT
and MOORE, 1984), and recent in situ observations also provide support for the idea that the
thermocline shoalings in the Gulf of Guinea are remotely forced by winds farther west in the
equatorial Atlantic (HouGHTONand COLIN,1986; HOUGHTON,1989).
The intensity of upwelling in the Gulf of Guinea decreases in certain years and is occasionally
accompanied by large positive anomalies (>2°C) in SST linked to anomalous southerly positions
of the ITCZ (MERLE, 1980). One such warming occurred in 1968, coinciding with intensifications
of the North Equatorial Countercurrent and the Equatorial Undercurrent, southward movements
of the ITCZ, and abnormally high levels of rain fall in the Gulf of Guinea; all these phenomena
indicated that a real E1 Nifio-like response had occurred in the eastern tropical Atlantic (HISARD,
1980). This response was preceeded by abnormal wind forcing off the northern coast of Brazil,
consistent with observations of historical non-seasonal variability of SST in the Gulf of Guinea
being highly and inversely correlated (with a one-month lag) with non-seasonal fluctuations of
the trade winds in the western equatorial Atlantic (SERvA~, PICAULTand MERLE,1982). Although
the factors responsible for anomalous variations in the trade winds in the western Atlantic are not
understood, Hmsx and HASXENRAaTJ(1983) have deduced that such variations constitute an early
link in a long and complex chain of events culminating in the appearance of abnormally warm
32 R.G. Pmm~sonand L. Sav,AuY,A

surface water off the coast of Angola and anomalously large amounts of coastal rainfall. They
found that the anomalous relaxation of trade winds in the western equatorial Atlantic accounts for
23% of the variance of SST off the Angola coast, whereas local wind forcing accounts for only
9%.
During the southern summer, when the thermocline dome is present in the Angola Basin
(Section 3.3.2) and the most vigorous coastal upwelling in the Benguela Current region is
displaced southward (Section 3.2), warm and salty waters from the Angola Basin penetrate
poleward to depths of 50m or more to regions offshore of northern and central Namibia (O'ToOL~,
1980; SHAr~ON, 1985). However, major southward intrusions of warm and saline water into the
upelling regions south of 15°S, comparable to those observed during Pacific El Nifio events off
westem South America, are relatively rare. SnAr~rON,BOVD, BRUNDRrrand TAUNTON-CLARK
(1986) argued that from the early 1950s only two such intrusions have occurred, those being in
1963 and 1984. They also documented evidence for an El Nifio-like event in 1934 and an
unusually strong warm anomaly in 1950. However, using historical sea surface temperatures in
the southeastern Atlantic, TAUNTON-CLARKand SnAr~ON (1988) found evidence for a greater
frequency of such events, more on the order of every ten years or so.
The strongest Pacific E1 Nifio event of this century was that of 1982-83 (CANE, 1983; HOREL,
KOUSKYand KAGANO,1986). In 1983 there were extreme dry conditions in northeastern Brazil
(LArdS, PEPPLERand HASTENRATrI,1986; Fig.4 in RASMUSSO~and WALLACE,1983), anomalously
strong trade winds and high surface pressures in the tropical Atlantic and low SSTs in the
equatorial regions ( H o ~ , Kouszv and KAGANO,1986), a continuation of unusually vigorous
upwelling along the coastal regions of the northern Benguela Current that began in 1982
(SHANNON,BOYD,BRUNDR1Tand TAUNTON-CLARK,1986), and severe drought in southern Africa
(TAuNTON-CLARKand SHANNON,1988). Also occurring during the austral summer of 1982-83 was
a warming in the southern portions of the upwelling regime of the Benguela Current (SHANNON,
BOYD,BRUNDRrrand TAUNToN-CLARK,1986). This might be compared with Fig.3 in the paper by
CovEy and HASTEnRArn(1978), which shows that when averaged over ten Pacific El Nifio years
higher than normal SSTs are found during late summer along the southwest African coast.
With the following year, 1984, came the most intense warm event the tropical Atlantic had
experienced for two decades. The fields of SST from June 1983 and June 1984 shown in Fig.17
illustrate the magnitude of this most recent extreme event. Surface flow in the equatorial Atlantic
slowed to near zero westward velocities in Febrnary-April 1984 ~ I N and McPnADEN,1986),
and anomalously warm surface waters penetrated into the northern upwelling regime of the
Benguela Current system on a scale observed only once before (SHANNON,BOYD,BRtn~Rrr and
TAUN'I~N-CLARK,1986), in the middle of 1963 as described by SXAnDERand DE DECKER(1969).
During the 1984 event, positive anomalies of almost 3°C were observed at the coast near 23°S,
but in the southern regions of the upwelling regime the effects were minimal; the higher SSTs were
accompanied by an unusually large southward shift of the ITCZ (also a feature of Pacific ENSO
events) and heavy rains in northeastern Brazil and the coastal zones of equatorial and southwestern
Africa. However, there was no analogy to the eastward displacement of the region of heavy
precipitation typically observed during Pacific E1 Nifios (Prm.AnDm% 1986).

3.4. Brazil Current

A schematic view of the Brazil Current is given in Fig. 18 which depicts many of the features
 Upper-level circulation in the South Atlantic Ocean 33

20eN

.o.= .... z~,.. I " ~ ~ # : o ' " ' " AF.,c.

IO=N

"'; \ \ "~ "~ ""~J" f :::.. ABIDdAN ". . . . . . " i . .

'"';':'" . ~- . ' "~ _/" " : . " !.: d."

0o

t00~ ANDA IOOS

20"S
60"W 40*W 200W O~W 20°E

20"N
/~ ~×~ 2 , i

IO=N

O*

~NOA tOeS

i , i = 20=S
60"W 40*W 20°W O*W 20°E

FIo. 17. Sea surface temperatures (°C) in the tropical Atlantic in (a) June 1983 and (b) June 1984
( ~ , 1986; reprinted by permission from Nature, Vol. 322, pp. 236-238. Copyright1986
Macmillan Magazines Ltd.).

discussed below. One feature already commented upon (end of Section 3.3.1) is that most of the
water carried toward Brazil by the southern SEC enters the North Brazil Current, and that only
about 4Sv turn south near 10°S to feed the incipient Brazil Current. Because much of the transport
in the northern limb of the South Atlantic subtropical gyre is lost to the northern hemisphere and
to equatorial countercurrents, the northern Brazil Current is conspicuously weak when compared
with other western boundary currents. And the Brazil Current transport remains relatively small
over most of its southward course. S x R ~ , IXW_a~Aand Pur~.a<soN (1990) found no systematic
JPO 26:1-0
34 R.G. Fmr.KSONand L. SaXAMMA

strengthening of the current between the latitudes of 10 ° and 20°S, and of the several estimates
of Brazil Current transport from the region of 19 ° to 25°S all are 11Sv or less (Table 2).
A problem encountered when estimating the transport of the Brazil Current in its more
northerly reaches is that it is shallow and closely confined to the continental shelf. At 19°S,
MIRAr~A and CASTROFILHO(1982) calculated a maximum southward surface velocity of 72cm s
relative to about 500m, finding that over 50% of the current transport of 6.5Sv was contained
in the upper 200m. In the region of a seamount chain at 20.5°S, EVANS,SmNom~ and MmArCOA
(1983) observed maximum surface speeds of 50-60cm s -~relative to 500m as the current flowed
through a passage nearly 2000m deep between the most inshore of the banks, which rise to within
200m of the sea surface. Using direct measurements from a Pegasus profiler near Cabo Frio
(23°S), EVANSand SmNORI~ (1985) obtained evidence that nearly half of the current's southward
transport of 11Sv was inshore of the 200m isobath. Satellite IR images of the region of 21°-35°S
(GARFrm.n, 1988) show that the inshore edge of the Brazil Current is confined to regions less than
2000m deep, with its mean position near the 200m isobath also indicating that a significant portion
of the current transport occurs over the continental shelf. As STRAMMA(1989) pointed out, this
may be why Fu (1981 ), who by using inverse techniques on the Meteor and IGY data, could find
only a very marginal Brazil Current at 24°S, and none at 16°S.
Another feature revealed by satellite images along the Brazilian coast (GARFIELD,1988) is the
occurrence of a semi-permanent offshore meander in the vicinity of Cabo Sao Tome (22°S) to
Cabo Frio (23°S), which may be related to the local upwelling described by MASCARENAS,MIRANDA
and ROCK (1971), IKEDA, MIRANDAand ROCK (1974) and MA~taOCCA, MIRANDA and SIaNORIm
(1979). Although this feature was not evident in the survey by EVANS, SmNORINIand MIRANDA
(1983), its influence can be seen in the distributions of temperature at 50m depth and surface
dynamic topography in the data of SIGNOR~I (1978).
AS the Brazil Current flows south from 24°S, it intensifies at a rate of about 5% per 100km
(GORDONand GREEN6ROVE,1986), a growth rate similar to that of the Gulf Stream though the actual
transport values here are considerably less. South of about 30°S, this downstream intensification
appears to be linked to a recirculation cell which has been observed with hydrographic
measurements (LENz, 1975; REID, NOWL~ and PATZERT, 1977; TSUCHIYA, 1985; GORDON and
GREENGROVE,1986; STRAMMA,1989) and satellite IR imagery (BRowN, EVANS,OLSONand PODESTA,
1985) supplemented with drifter trajectories (OLSON,PODESTA,EVANSand BROWN,1988). Because
of the downstream growth of the Brazil Current, GOROONand GREENCROVE(1986) surmised that
the recirculation cell carries about 12Sv in the upper 1400m, and by closing off a region with
hydrographic sections STRAMMA(1989) estimated the geostrophic transport of the cell as being
7.5Sv in the upper 800m (relative to 1300-1600m), bringing the southward transport along the
western boundary to nearly 18Sv at 33°S. Evidence for another recirculation cell in the western
South Atlantic, north of 30°S, was found by TsuCmYA (1985) in historical data, and also with
historical data REIO (1989, his Fig.7) has observed a single, large recirculation cell at the surface
extending from 20 ° to 40°S in the western Atlantic. The poleward end of the cell fuels further
growth for the Brazil Current, which attains transport values at 38°S of 19-22Sv relative to 1400-
1500m (GORDONand GREENGROVE,1986; GORDON, 1989; GARZOHand GARRA~O, 1989), where it
then encounters the northward flow of the Falkland (Malvinas) Current and turns offshore.
All of the estimates discussed above for the geostrophic transport of the Brazil Current have
been obtained by assuming levels-of-no-motion at shallow or intermediate depths. This
procedure has not been seriously questioned for areas well north of where the Brazil Current turns
 Upper-level circulation in the South Atlantic Ocean 35

60°W 50 ° 40 ° 30°W
Oos

10 °

20 °

30 °

40°S

l~o. 18. Schematic illustration of the Brazil Current region. Abbreviated terms are: North Brazilian
Current (NBC), South Equatorial Undercurrent (SEUC), South Equatorial Countercurrent (SECC),
South Equatorial Current (SEC), Subantarctic Front (SAF), Brazil Current Front (BCF), and
Subtropical Front (STF). Bathymetric contours are at 1000m intervals; depths less than 3000m are
shaded.
36 R.G. ~ S O N and L. S ~

TABLE2. Estimates of the southward transport of the Brazil CkLrrent obtained from in situ
measurements. 1Sv = 105#s ~

Latitude Reference Transport

Author Data (°S) (m) (Sv) Remarks

S ~ , IramA 13 historical sections 10-20 -500 0.8-6.0 average value near 4Sv
and PmY_~soN(1990)
MmmCDA and CASTRO Besnard, Sep-Oct 1967 19 -500 6.5
Fmno (1982)
EVANS, Sior~oRml and Besnard, Apr 1982 19 500 5.3 XBT' data + T/S b
/VlmAr~DA(1983) 20.5 500 3.8
1000 6.8
21.7 500 4.4 XBT data + T/S
23 -25 500 4.1
1000 7.8
IGY, Apr-Jun 1959 32 1000 13
EVAt~Sand SIONORINI Oceanus, Apr 1983 23 DM c 11 6Sv offshore of 200m
(1985) isobath, 5Sv inshore
SIONOR~rt(1978) unspecified, Jul 1973 22-25 600 6.8-9.4
S ~ (1989) Oceanus, Feb 1983 23 600 10.2
1300 11.0
Melville, Nov 1972 24 600 9.6
1300 10.2
Atlantis, Apr 1959 32 800-1600 19.2 5.9Sv in northward offshore
countercurrent
Melville, Nov 1972 33 800-1600 17.5 includes an estimated 5.3Sv
on the shell
Fu (1981, Table 6) Meteor, 1925-1926 --28 variable 20 inverse calculation with
intermediate and upper deep
water contributions
IGY, 1958-1959 -30 variable 27
EVANS and M A S C ~ S 31 DM ° 17 personal comment quoted by
GORDON(1989)
McCART~IEY and ZEM~A, Washington, 27 600 9
(1988), and ZEMBAand Sep-Oct 1984 37 3000 76 includes contributions from
McCARTm~Y (1988) intermediate and deep
waters
GARZOLIand BIAr~cm inverted echo sounders, 37.5-38 800 10 (mean)
(1987) Nov 1984- Jun 1985 23 (max)
GARZOLIand GAI~,A~O inverted echo sounders, 37.5-38 800 11 (mean)
(1989) Jtln 1985 - Mar 1986 800 18 (max) includes part of the Falkland
Current return
1400 22 (max) includes part of the Falkland
CtLrrent return
Balda, Jllli 1985 37.5-38 800 8.5
Balda, Mar 1986 37.5-38 800 5.8 may be small due to an
inshore northward flow
GORDONand GP.E~6ROVE Atlantis 11, 38 1400 19.0
(1986) Dec 1979 -Jan 1980 42 1400 20.2 includes part of the Falkland
current return
GORDON(1989) Washington, Oct 1984 38 1500 21.5
43 1500 22.5
P i n . s o N (1990) Atlantis I1, 38 variable 70 includes contributions from
Dec 1979 - Jan 1980 intermediate and deep
(~3000) waters
SCH~Am'DA (1980) Haeckel, Jun-Jul 1978 42 800 18

'Expendable bathythermograph
bTemperature/salinity relations
CDirect measurements from a Pegasus profiler
 Upper-level circulationin the South Atlantic Ocean 37

offshore, as it is fairly well accepted that the current is relatively shallow and remains close to the
coast. But for the southern Brazil Current this has not been so unanimously accepted. On the basis
of water mass characteristics (REID,NOWL~ and PATZERT,1977; Gm~NrROVE, 1986; Pu-rva~SONand
WI-IrrWORTH,1989; and REID, 1989), it appears that the southward flow in the region of the southern
Brazil Current extends from the sea surface down through the North Atlantic Deep Water, to
depths of 3000m or more. Taking this into consideration, McCARTNEYand ZEMBA(1988) and
ZEIVmAand McCARTNEY(1988) have estimated the total southward geostrophic transport across
37°S in the region of the offshore-turning Brazil Current as 76Sv relative to 3000m. Pu-l'va~SON
(1990) obtained a comparable value of 70Sv at 38°S using a similar deep reference level.
Satellite IR images taken over a period of about 3 years (July 1984 to June 1987) show that the
point at which the Brazil Current separates from the continental shelf (1000m isobath) varies from
33 ° to 38°S, with the average being near 36°S and probably not related to the local winds (OLSON,
PODESTA,EVANSand BROWN,1988). This variability is corroborated by time-series of the dynamic
height of the sea surface relative to 800dbar obtained from an array of inverted echo sounders
moored in the confluence zone during 1984-1986 (GARZOLIand GARRA~O, 1989). Though the
available data are not yet sufficient to statistically establish whether there is any seasonality in the
point of separation, an analysis of satellite IR images taken during the period of January 1980 to
March 1981 indicates that the front separating Brazil Current water from subantarctic water is
farther offshore (and hence the separation point farther north) during the austral winter and spring
than it is during summer and fall (GoDoI, 1982). Further work based on 1R images by OLSON,PODESTA,
EvANs and BROWN (1988) also reveals a more northerly separation point during winter than
summer. The mechanisms responsible for this apparent seasonality are not known, but one might
speculate on the relations of this with the seasonality elsewhere in the South Atlantic. As
compared with the austral summer situations, those in winter (pointed out in previous sections)
include the subtropical high pressure system being centered farther north with a higher central
pressure, while the South Equatorial Current is also farther north and is stronger. In addition, the
zero-line of wind stress curl which extends across the southern South Atlantic in winter lies
approximately 5 ° north of its summer position (Fig.8 in OLSON,PODESTA,EVANSand BRoWN, 1988),
all of these together suggest that the subtropical gyre may shift northward during the winter, with
one manifestation being the apparent seasonality in the separation point of the Brazil Current.
After separating from the western boundary, the Brazil Current continues to flow in a general
southward direction for some distance together with the return flow of the Falkland Current; the
Brazil Current component of this confluence was estimated by GORDON(1989) as being 22.5Sv
relative to 1500m at 43°S, which is near the most southerly positions attainable by the current
before it turns east and north. The southern limit to the warm water bounded by the Brazil Current
has been observed in satellite IR images to fluctuate between 38 ° and 46°S over time scales of
about two months (L~Ecrds and GORDON, 1982). The bi-monthly southward excursions of the
Brazil Current are analogous to the semi-periodic westward penetrations of the Agulhas Current
into the Atlantic which occur over similar time scales (Section 3.1).
Gom~gN (1981) has noted that enhanced ocean-to-atmosphere heat fluxes should be expected
in the southward extension of thermocline water, and by using temperature and salinity
measurements from the region he gave evidence that convective processes OCCUlTing in the
poleward loop of the Brazil Current are important in ventilating the main thermocline farther
north. Figure 19 provides an example of the poleward loop of the Brazil Current with a salinity
field at 100m depth constructed from observations made during four different research cruises.
The contours can be loosely regarded as being parallel to the flow. The salty (>36) core of the
38 R.G. Pm~RsoN and L. S ~

Brazil Current is observed near the continental slope north of about 38°S, whereupon the current
separates from the western boundary at its confluence with the northward-flowing Falkland
Current to lose heat and salt through tubulent mixing with the cooler and fresher water of
subantarctic origins. Also shown in the figure are the positions of the Subantarctic and Polar fronts
according to P~:r~a~SONand WmawoRa~a (1989) (to be discussed in Section 4.1), poleward of which
are increasingly cold and fresh upper-layer water masses. The confluence of subtropical and
subpolar waters extends southward to approximately 46°S, before the poleward-flowing thermocline
water executes a large anti-cyclonic return to the north. Because this return flow appears to be
closing back on its source flow near 42°S, a maximum southward excursion of the Brazil Current
had probably taken place. As with the westward penetrations of the Agulhas Current, maximum
southward extensions of the Brazil Current are followed by the shedding of warm-core, but less
energetic eddies which are lost to the Subantarctic Zone of the Antarctic Circumpolar Current. In
comparison with the large (320km diameter), single eddies shed from each pulse of the Agulhas
Retroflection, those formed during northward retreats of the Brazil Current, as observed in
satellite IR images (LFx~ECraSand GORDON,1982), are smaller (averaging about 150km across) and
more numerous (formation rate of about one per week), though their azimuthal surface speeds
nearing 80cm s-1 (OLSON,GORDONand HOOK~a~,1985; GORDON, 1989) are similar. Unfortunately,
the relatively coarse station grid in Fig.19 does not allow for such a pattern to be well resolved.
A volume transport of about 20Sv around a newly-formed Brazil Current eddy near 44°S was
obtained by GORDON(1989), which is just half that of an Agulhas equivalent.
Aside from the transport estimates pertaining to the total southward flow, which contain
contributions from the deep circulation (Table 2), all other estimates of the Brazil Current
transport based on hydrographic data are considerably less than those predicted from fields of
wind stress. An analytic wind-driven model with simplified geometry was used by VEa~o~s
(1973) to calculate a maximum Brazil Current transport of 38Sv at 36°S, whereas H ~ and
ROSENS~n~ (1983) used the homogeneous Sverdrup relation to obtain an annual mean transport
of 30Sv at 30°S. Including bottom topography and a thermohaline component in a wind-driven
numerical model, Mm~_OR, Mw~aoso and Kwro (1982, their Fig. 11 ) obtained, without comment,
a transport for the Brazil Current of more than 60Sv. This value compares with the 52Sv estimated
by VEROmS (1978) from his wind-driven analytic model modified to include thermohaline
forcing. V ~ o ~ s (1973) thought that intense mixing at the Brazil-Falkland Confluence Zone
might be one factor contributing to the weakness of the Brazil Current, whereas another factor to
consider is whether or not the Benguela Current absorbs some of the transport from the interior.
But SrOM~W_L(1957) had earlier proposed that the Brazil Current is weaker than expected from
the wind fields because of an opposing effect of the thermohaline circulation, which in the North
Atlantic involves the formation and sinking of North Atlantic Deep Water. His view is supported
by a salinity budget for the Atlantic (GORDONand PIot~, 1983) in which the formation of North
Atlantic Deep Water requires a net transfer of thermocline water from the South Atlantic to the
North, together with net northward fluxes of intermediate and bottom waters (Ri~rrotn~, 1990).
These results are consistent with observations showing that the surface circulation of the South
Atlantic subtropical gyre is not a closed system, but is one where most of the transport of its
northern limb is lost to the equatorial countercurrents and to the northern hemisphere (STP~,~tA,
I ~ A and Pzrm~SON, 1990). Without this loss, the Brazil Current would no doubt be significantly
stronger. GORDONand GREENGROVE(1986) have pointed out that the deficit in the Brazil Current
transport relative to Sverdrup-type solutions might be compensated by the deep western boundary
current which carries North Atlantic Deep Water toward the south.
 Upper-level circulation in the South Atlantic Ocean 39

600W 50 ° 40°
35°S

40 °

45*

50 °

FIo. 19. Distribution of salinity at 100-m depth in the southwestern South Atlantic as constructed from
bottle samples taken during the following four cruises: Atlantis H cruise 107 leg m (triangles); Atlantis
//cruise 107 leg X (squares); Thomas Washington Marathon Expedition leg 8 (circles); and Robert
D. Conrad cruise 28-03 (crosses) ( S ~ and PL'maSON,1990). Also indicated are the Subantarctie
Zone (SAZ), Polar Frontal Zone (PIZZ),Subantarctic Front (SAF), and Polar Front (PF) after Pm'e.aSON
and WrtrrwoRrn (1989).
3.5. South Atlantic Current

The circulation o f the South Atlantic subtropical grye is closed in the south by the South
Atlantic Current (SAC). This current is analogous to the North Atlantic Current (for information
see KRAUSS, 1986) in that it is the seaward extension of that part o f the western boundary current
which does not enter the subtropical recirculation near the American shelf. The S A C is distinct
from the Antarctic Circumpolar Current, and is separated from it by the Subtropical Front (STF)
(STRAMMAand P~u~-~SON, 1990).
40 R.G. ~ u and L. S'm,,~M~

3.5.1. SubtropicalFront. The STF was identified by K R ~ L (1882) as being a sharp dis-
continuity in surface temperature which he found to stretch across most of the South Atlantic from
its origins in the Brazil-Falkland Confluence Zone. DEACON(1933) later studied this discontinuity
in greater detail, referring to it as the Subtropical Convergence, and observed sudden changes in
surface temperature and salinity of at least 4°C and 0.5, respectively, as the discontinuity is
crossed. He traced its position across the South Atlantic, finding that it lies for the most part 2 °
or 3 ° north of 40°S except near 30°W and south of Africa where it is south of that latitude.
According to DEACON(1937), the surface waters just north of the front have minimum temperatures
ranging from 11.5°C in winter to 14.5°C in summer, and are often warmer by 5°C or more, and
that they have salinities of at least 34.9. Progressing northward from the front, both mean
temperature and mean salinity continue to increase, until temperatures of nearly 28°C are reached
at the equator during southern summer (H6FLICn, 1984) and annual-mean salinities in excess of
37.25 are reached near 15°S (TcrmRmA, 1980).
DEACON'S (1933; 1937) generalizations of the temperature and salinity changes taking place
across the discontinuity have been widely accepted, though the name he used for this frontal
feature has been an occassional point of contention. Early on, B 0 ~ r m (1936) and others held
that the term Subtropical Convergence should be reserved for a wind-driven surface convergence
at 30°-35°S, but this objection was not widely endorsed and the terminology used by DEACONhas
been almost universally followed until late. In locating thermostads beneath the seasonal
thermocline around the southern hemisphere, MCCARTr~'V(1977) envisaged a continuity of those
in the southwestern Atlantic with others found in the Indian and Pacific oceans. The thermostads
in the Atlantic were found to be just north of DEACON'S convergence, whereas other but colder
thermostads were found north of the Subantarctic Front in the other sectors of the Antarctic
Circumpolar Current. Because of the apparent continuity of all the thermostads observed, and,
in retrospect, probably because there was very little information about the Subantarctic Front in
the South Atlantic downstream of Drake Passage, MCCART~-~ (1977) argued that a single,
circumpolar front resides along the poleward sides of these deeply-mixed layers, and he referred
to it as the Subantarctic Front. The Subantarctic Front is now thought to be circumpolar in extent,
but it is not connected with what has been popularly called the Subtropical Convergence, a feature
which does not go through Drake Passage. But other authors (RoDin% 1986; 1989; OLSon, PODESTA,
EVANS and BROWN, 1988) have followed MCCARTNEV'S(1977) South Atlantic terminology, and
while these are exceptions to the rule, the results they have presented need to be integrated into
the whole and thus the inconsistencies in terminology clarified. Finally, to avoid implications
regarding the type of motion involved, the term Subtropical Convergence has recently been
modified by a few authors to Subtropical Front (for example, HOFMANN, 1985; Wm'rwoRTI~and
NOWL~, 1987), and we adopt this terminology.
Because the STF is the poleward boundary to the warm and salty thermocline waters of the
subtropical gyre, and because no subtropical waters are found in Drake Passage, the STF by
definition must get its start at the western boundary with the confluence of the Brazil and Falkland
currents. But the points at which the surface fronts associated with these two currents separate
from the western boundary (the 1000m isobath) are usually not coincident. On the basis of satellite
IR images, OLSON,PODESTA,EVANSand BROWN(1988) observed an intervening zone up to 300km
wide populated with eddies which separates the two seaward-turning fronts and which can be
traced as a continuous feature into the ocean interior. Inspection of their Fig.7 reveals that when
averaged over a period of nearly three years the seaward extension of Brazil Current water
(identified by the 21°C isotherm) reaches 40°-42°W at a latitude of about 37°S, and that
subantarctic water (identified by the 16°C isotherm) reaches the same longitudes at 41°-42°S.
 Upper-levelcirculation in the South Atlantic Ocean 41

Figure 1 in the paper by GORDONand GREENGROVE(1986) shows that during the early austral
summer the 16°C isotherm is representative of the wannest subantarctic waters found in the
confluence zone, and should therefore be a close indicator of the STF. The seaward locations for
the two fronts as revealed from the satellite survey agree well with the positions of two fronts
observed earlier by RODEN(1986) in a closely-spaced hydrographic survey (shown as the open
circles in Fig. 19). But the surface temperatures in the fronts observed at sea were about 5°C less
than in the satellite survey, presumably because the ship-board measurements were made in the
austral spring (October-November), when sea surface temperatures are near their annual minima;
whereas the identifiers used in the satellite survey remained unchanged with season. The sea-
going survey measured temperatures and salinities at the southern front which were consistent
with the classical definitions of the STF, butin both papers (RoDEN, 1986; OI,SON,POD~STA,EVANS
and BROWn, 1988), and in a third (RooEN, 1989), this feature was referred to as the Subantarctic
Front.
The salinity and temperature changes across the STF compensate one another in the density
field, resulting in weak baroclinic shears of horizontal velocity in the upper mixed layer (RoDEN,
1986; WrIrrwoRTn and NowLn~, 1987). But this is not the case with the northern of the two fronts,
which RODEN(1986) referred to as the Brazil Current Front (BCF). According to him, the BCF
is characterized by sharp thermohaline gradients in the upper 500m and large vertical shears of
horizontal velocity. As can be inferred from Fig.19, however, the two fronts do not remain
distinctly separated once they first diverge from one another near the western boundary. In the
west, the STF extends southward to around 46°S with the poleward extension of thermocline
water, whereas if the 35.4 isohaline is used as a rough indicator, the BCF appears to remain north
of 42°S. There are then apair of regions where meanders bring the two fronts into close proximity
before they separate perhaps for the final time near 42°W. The mean course of the BCF is not
established, but it is likely to continue around the recirculation cell of very warm and salty surface
waters in the Argentine Basin similar to the way the upper-level flow fields in the region have been
depicted by REID, NOWL~ and PAa'ZERT(1977) and S ~ (1989). Circumstantial evidence for
this is that no front resembling the BCF has been documented in the central and eastern portions
of the subtropical gyre.
3.5.2. Geostrophic flow. The zonal current associated with the STF, the South Atlantic
Current, is understood in only the most general ways. S ~ and Pm~RSON (1990) observed
the SAC to be an identifiable eastward current band flowing mainly along, and just to the north
of, the STF. The surface speeds of the SAC are not large when compared with those of the Brazil
or Agulhas currents or those associated with the fronts within the circumpolar current, which is
a result of the opposing effects of the distributions of temperature and salinity in the density field
near the STF. Surface velocities of the SAC are apparently at their largest in the Argentine Basin,
where they are on the order of20cm s-1(RoDEN, 1986) tO 27cm s-1( S ~ and PmeKSON, 1990),
before decreasing to values only half as large in the Cape Basin (WiarrwoRrn and NOWLn~, 1987;
S ~ and Pm~tsor~, 1990). Volume transport of this current follows the same trend; in the
Argentine Basin it is typically 30Sv in the upper 1000m relative to about 3000m, reaching values
as high as 37Sv, whereas in the Cape Basin it is only about 15Sv ( S ~ and Purua~SON, 1990).
The transport values of 30Sv and more for the SAC in the Argentine Basin are comparable to,
and even larger than, those of the Brazil (without deep contributions) and Benguela currents. But
details concerning the eastward attrition of the SAC are not available. As S ~ and Pu~-~soN
(1990) have pointed out, the hydrographic data base near 40°S between 25 °W and the Greenwich
Meridian is extremely paltry, making it impossible to judge what effects the mid-ocean ridge
might have on the transport of the SAC. In the western basin the deeper waters move in the same
42 R.G. P~rmtsoNand L. SaXAMUA

general direction as the near-surface waters (REID, NOWLm and PATZ~T, 1977), whereas in the
eastern basin there seems to be very little zonal flow at all at large depth beneath the STF
(Wm-~WORrn and NOWL~, 1987; SaXAMMA and Pm~SON, 1990). Given similar baroclinic
structures in the upper layers, the transport of the SAC should be less in the eastern basin than in
the west, because of the differences in the deeper flow, and this is evidently the case. But the
available data are too few to test whether there is a marked reduction as the current flows over the
mid-ocean ridge, or whether the flow gradually stems off to the interior of the subtropical gyre.
As the weakened SAC flows east, it enters the eddy-filled region of the Agulhas Current
Retroflection and is strongly influenced by it. Instead of continuing east with the STF south of
Africa, the SAC turns north to feed the Benguela Current. Why the SAC should separate from
the STF is not entirely clear, or even if it does so all the time, but it seems that the saltiness of the
Indian Ocean Central Water is of primary importance. In the region southwest of a newly-shed
Agulhas eddy, STRAMMAand Pm'vatSON(1990) calculated a volume transport at the STF of nearly
10Sv in the upper 1000m toward the northwest, which is in the opposite direction relative to the
front as is the usual case. The cause of this was deemed to be the presence of very strong horizontal
salinity gradients which over-compensated the thermal effects in the density field, and this was
completely consistent with requirements for mass balance in the region. Shown in Fig.20 is their
field of geostrophic volume transport in the upper 1000m in a region southwest of southern Africa.
The flow into the region across these two quasi-synoptic sections amounts to about 26Sv, with
the only possible outlet being toward the north. The 17Sv flow of the SAC near the prime meridian
can not continue east with the STF, but must turn north, which would be consistent with the
transport of the Benguela Current across a line closing the triangular region to the north as
determined by STRAMMAand Pm'V.KSON(1989). Being joined by about 8Sv of Agulhas Current
water, this turn of the SAC into the Benguela Current provides the southeastern closure of the
South Atlantic subtropical gyre.

4. ANTARCTICCIRCUMPOLAR CURRENT

In this section we give a short overview of the observed zonation and volume transport of the
Antarctic Circumpolar Current (ACC) in the South Atlantic. For an in-depth treatment of
mesoscale variability, meridional exchanges, and theories and models for the current as a whole,
the reader is referred to the comprehensive review by NOWLtS and KL~CK (1986). Additional
information concerning deep water masses and stratification of the ACC can be found in the
papers by SmVERS and NOWLIN (1984) (Drake Passage), Pmm~soN and WHrrwoRra (1989)
(southwestern Atlantic), and WrirrwoRTrI and NOWLIN(1987) (Greenwich Meridian).

4.1. Zonation

The ACC has been studied in greatest detail in Drake Passage, particularly during the mid-
1970s to the early 1980s with the implementation of the International Southern Ocean Studies
(ISOS) program. Results from the ISOS investigations have shown the ACC to be strongly
banded in the velocity field, which is illustrated in Fig.21. Shown in the figure are vertically-
averaged geostrophic speeds in the upper 2500m (relative to 2500m) normal to a line of stations
spanning Drake Passage. The zonal jets and bands of relative quiescence for which the ACC is
now well known are clear. Near the terminus of South America is the Subantarctic Zone (SAZ),
which, at locations other than Drake Passage, is bounded to the north by the STF. Southward from
the SAZ are the Subantarctic Front (SAF), Polar Frontal Zone (PFZ), Polar Front (PF), Antarctic
Zone (AZ), Continental Water Boundary (CWB), and Continental Zone (CZ). At the narrow
 Upper-level circulation in the South Atlantic Ocean 43

0 5 10 15 °E 20

~~'::ii i iosi l.
30

Jl
I I I I ::i:i:i:i:i:i:i:i

oS

35

40 ~?'A" 40

45 - 45

I I I I
0 5 10 15 °E 20

PIG.20. Field of geostrophicvolume Izansport(Sv) in the upper 1000mrelativeto the potential density
surface of ~o = 45.87kg m-3in the region enclosed by two sections made by the R/V Knorr in October
and November 1983 ( S ~ and Purm~sor~,1990). Except for the coastal stations, the reference
level lies in the depth range of 3350-3730m.

fronts, the vertically-averaged through-passage speeds are in the range of 12-17cm s -t, and, except
near the South American continent where the landmass may act to constrict the flow, there are
much weaker speeds in the water mass zones. NowLn~ and C i a ~ o ~ (1982) have observed typical
upper-level geostrophic speeds (relative to 2500m) in Drake Passage to be 30-45cm s -1at the SAF
and PF, and 15-30cm s -1 at the CWB. They also found that the three fronts account for three-
quarters of the through-passage geostrophic transport, but, while being on the order of 40-60km
wide apiece, together occupy only one-fifth of the width of the passage.
First evidence for the banding of the ACC came with observations by ScHorr (1902) of sharp
changes in surface temperature and salinity occurring between Antarctic and subantarctic waters
southwest of Africa and north of Kerguelan. ~ h a o u s (1923) traced the approximate course of
that hydrographic boundary from the eastern Pacific to the mid Indian Ocean, a course later
referred to by ScaoTr (1926) as the "Meinardus Line," by I3~arrr (1928) and WOsT (1928) as the
"Oceanic Polar Front," by DEACON( 1933) as the "Antarctic Convergence," and by W~mTra (1960)
as the "Antarctic Polar Front." The simplified term "Polar Front" (PF) has come into widespread
usage as it makes no implications about relative motions at the boundary, which have been thought
to be possibly divergent (WEa~R, 1959).
Hydrographically, the PF is the northern boundary to cold (-1.5 ° to 2°C) near-surface water
formed by winter cooling (MosBY, 1934). Beneath this Antarctic Surface Water is the relatively-
warm and homogeneous Circumpolar Deep Water. During summer, the Antarctic Surface Water
44 R.G. Pm~JzsoN and L. SaXAMMA

68°W 66 ° 64 ° 62 ° 60 ° 58°W

~6 °

: iiii ::¸

58 °

~ 0 ~ ~ "~: :::

!iii!ii!ii!iii!~:.
'~iiiiiiiii~:.
60 °

:IE

0 5 10 15 c m s" 1 ::::--~
"~i\~:~...:: .........:~-il;i 62°S

- - lO00 m
.:.:.;.:.:.:.;.:.:,:.
i?~:!:i:!:~:~:i:!:i
< 3000 /

FIo. 21. Vertically-averaged geostrophic speeds in the upper 2 5 0 ( 0 , relative to 2500m, normal
to a line of stations occupied by R/V T h o m p s o n during 28 February - 3 March 1976 (adapted from
P~rmtsoN, NowLn~ and Wm'rwoarn, 1982).
 Upper-levelcirculation in the South Atlantic Ocean 45

is overlain by a seasonally warmed surface layer, thereby leading to the appearance of an upper-
layer temperature minimum. The temperature minimum is embedded in a strong halocline in the
upper 200m and is generally colder than 0°C, except near or at the PF where it warms and deepens
northward along inclined isopycnals. When subsurface temperature measurements were avail-
able to them, DEACON(1933) and MACKIrrrOSH(1946) fixed the position of the PF as to where the
temperature minimum sinks below 200m, which was modified by GORDON(1967) to where the
temperature minimum makes a rapid northward descent. NowLm, WHrrWORTrIand Pn.LSBtmY
(1977) used the northern extent of the 0°C isotherm associated with the temperature-minimum
layer as an indicator for the location of the PF, a variation from the widely-used northern extent
of the 2°C isotherm near 200m depth (BoTNmOV, 1963). Although the various indicators provide
locations for the PF that are often in close agreement, P~rm~SONand WHrrWORWI(1989) argued
in favor of the criterion proposed by DEAcoN (1933). But these subsurface indicators yield
positions for the PF which are for the most part not identical with those of the most intense surface
temperature gradients. In the region south of Africa, LtrrmHARMSand VAimmams (1984) have found
that in 75% of the cases the subsurface expression for the PF is north of the surface front, with the
distance between the two being as large as 300km.
Additional banding of the ACC was first noted by DEAcoN (1937) when he observed internal
isopleths of temperature and salinity rising toward the south across the current in a series of steps;
but an explicit account of a second front in the ACC was not given until BtmLIN~ (1961 ) observed
two types of subantarctic water between the Polar Front and Subtropical Front in the region from
southern New Zealand to Macquarie Island. He called the boundary between the two subantarctic
water masses the "Australasian Subantarctic Front," and observed that it extends from the sea
surface to depths of more than 1000m. OSTAPO~(1962) soon after applied a frictional, diagnostic
model to the zonal geostrophic velocities obtained from a line of widely-spaced hydrographic
stations along the Greenwich Meridian to predict the existence of a narrow zone of upward motion
some 5 ° or 6 ° latitude north of the Polar Front. He called the northern feature the "Subantarctic
Front" and drew comparisons between it and the front described by BtmLm6 (1961). Also south
of New Zealand, GORDON(1967) found the Polar Front to occupy a complex zone of mixing 2 °-
4 ° latitude wide. Arguing that the Polar Front is probably not a distinct line, GORDON,TAYLORand
GFX)RG1(1974) used the term "Polar Front Zone," the northern boundary to which was near
BURLn~G'S (1961) Australasian Subantarctic Front. GORI~N, GEORGIand TAYLOR(1977) found
similar complexity in defining the Polar Front in the Scotia Sea, maintaining that it would be better
considered a zone, and not a line, in that region as well. Their Polar Front Zone was described
as being limited to the north by a subantarctic boundary and to the south by an Antarctic one, with
each being near an axis of the ACC. Consistent with this were findings by NOWLIN,WHrrWORrH
and PILLSmmY (1977) within Drake Passage of a vertically-coherent band of relatively large
eastward velocity at each of these boundaries of the "Polar Frontal Zone" (PFZ), a region of
transition between Antarctic and Subantarctic surface waters. This eddy-populated region of
transition was studied from south of Australia to Drake Passage by EMERY(1977) and was seen
to be a continuous feature of variable width. He suggested calling the southern boundary, which
was the same as had been tradionally known as the Polar Front, the "Antarctic Front," and
proposed that the northern boundary, the same as that described by BtmLINO (1961), be simply
called the "Subantarctic Front" (SAF). This meaning for the SAF, i.e., that it is the hydrographic
boundary between the PFZ and the SAZ, has since been adopted by investigators working in
Drake Passage (WHrrWORTH,1980; P~Iv_a~SON,NOWLINand WHrrWORTH,1982; NOWLINandCLIFFORD,
1982; and several others), in the southwestern Atlantic (Pm-va~SONand WHrrWORTH, 1989), at the
Greenwich Meridian (WmTWORTH and NOWLIN, 1987), south of Africa (LtrrJEnAl~S and
46 R.G. P~I~ISONand L. SrRA.MMA

VALErzrn~, 1984), SOuth of New Zealand (I-I~a-a, 1981), and along the entire length of the ACC
as well (CLn=FO~, 1983; HO~>rN, 1985). AS was previously discussed (section 3.5.1), this term
has been used on occasion in a different sense, i.e., for what we have called the STF in the
southwestern Atlantic.
A third front exists within Drake Passage, the Continental Water Boundary. This term was
introduced by StaYERS and EMERY(1978) tO demarcate the northern limit of a cold water mass
(colder than about 0°C) near the South Shetland Islands having a subsurface isothermal layer
extending from about 150m depth to more than 500m. This feature was earlier described by
DEACON(1933) as separating water of Bellingshausen Sea origin (west of Drake Passage) from
that of the Weddell Sea. The influence of Bellingshausen Sea water decreases toward the east in
southern Drake Passage (GORDONand NOWLIN,1978), but recent work by SMrm (1989) has shown
that the Continental Water Boundary remains a continuous feature into the southern Scotia Sea
where it forms the northern limit to the Weddell-Scotia Confluence. GORDON(1967) described
the confluence as being a zone of discontinuity between deep water masses originating from the
southeast Pacific Ocean and Weddell Sea. He suggested that it might extend as far east as 30°E.
SMITH(1989) referred to the northern boundary of the Weddell-Scotia Confluence as the Scotia
Front, and the poleward boundary in the northwestern Weddell Sea as the Weddell Front (refer
to Fig.l). He also presented evidence that in addition to it being an extension of the Continental
Water Boundary from Drake Passage, the Scotia Front is also continuous with the ACC-Weddell
Gyre Boundary observed at the Greenwich Meridian by WHrrwoRrH and Nowt.iN (1987). We do
not describe this frontal feature in any detail here. For further information concerning the
hydrography of this boundary, and of the deep-reaching vertical convection and mixing processes
occurring near it, the reader is referred to DEACONand MooP~Y (1975), DEACONand FOSTER(1977),
PATTERSONand SmWRS (1980), and BERSCHand BF~Om (1986) and SMITH(1989).
The spatial orientation of the Polar Front in the South Atlantic was described by DEAcoN(1933)
and again in more detail by MACKINTOSH(1946). In both cases, the positions of the front were
established on the basis of sea surface temperature, and when available, the location at which the
temperature-minimum of Antarctic Surface Water sinks below 200m depth. Shown in Fig.22 is
the well-known map constructed by MAcrarcrosn (1946). As he pointed out, the Polar Front
executes a large S-shaped turn between the Falkland Islands and South Georgia, much like that
drawn by DEACON(1933). According to MACr,JtcrOSH'S Table 9, most of the frontal positions in
that region were based on surface temperature, and it is in that region where DEACON(1933) found
the two methods for fixing the Polar Front to agree the least. He noted that while the surface
conditions there are often subantarctic, and probably only temporary, Antarctic water appears
only 100 m below the surface. GORDON,GEORGIand TAYLOR(1977) confirmed the great com-
plexity of the region with a hydrographic and XBT survey of the western Scotia Sea. They found
their Polar Front Zone as describing a course similar to the envelope of PF positions depicted in
MACKINTOSH'Smap. More recently, P u t , s o N and WHrrWORrH(1989) used observations from two
cruises in the southern Argentine and northern Georgia basins to find the PF at 49°-50°S between
35 ° and 40°W (that portion of Fig. 1 is based on their observations), which is on the order of 200km
north of the course largely based on surface temperatures depicted by MACr~CrOSH. The dif-
ference is probably caused by subantarctic water overriding Antarctic water. Shown in Fig.23 is
a vertical section of temperature along a short line of closely-spaced XBT stations made in the
southern Argentine Basin. As depicted in the inset, the section began on the Falkland Ridge and
extended north across another narrow ridge and into the Argentine Basin. The temperature-
minimum of Antarctic water is clear at the stations south of the crest of the northern ridge, whereas
the surface temperature front normally associated with the PF is somewhere south of the section
 Upper-level circulationin the South Atlantic Ocean 47

and well south of the subsurface front. The coincidence of the subsurface expression of the PF
and the crest of the northern ridge was a recurring observation, as was the close proximity of the
PF with the SAF in the region (Purv.,tson and WmTWORTn, 1989). At locations farther east in the
South Atlantic, the two fronts are separate features. The PF has recently been found in much the
same positions as described by DEAcon (1933) and MACKINTOSH(1946); at both the Greenwich
Meridian (WHrrWORTnand NOWLIN, 1987) and south of Africa (LtrrnmaRMS, 1985) the PF is near
50°S.
The course of the SAF has been studied to a much lesser extent than that of the PF, which is
partly because the SAF is not always clear in the surface temperature fields; it is better identified
with upper-layer salinities. North of the front, the SAZ is characterized by weak vertical gradients
of salinity in the upper layer that overlay a weak maximum, which in turn is found above the
salinity minimum of the Antarctic Intermediate Water. Well north of the SAF, the salinity
minimum of Antarctic Intermediate Water typically resides near the potential density surface of
G0 = 27.2kg m -3 at depths of 500-1000m. In the southern SAZ, this minimum rises rapidly
southward along tilted isopycnals into the SAF, and approaches or reaches the sea surface within
the PFZ. The rapid northward descent of the salinity minimum of Antarctic Intermediate Water
is also evident where BORLm6(1961, his Figs.9 and 11) found his Australasian SAF south of New
Zealand, though he identified the front there more in terms of absolute values of upper-level
salinity, such as those less than 34.5 being south of the front,
Although the SAF is less recognizable in the temperature field than in the salinity, it seems to
coincide well with enhanced subsurface horizontal temperature gradients between the 3 ° and 5°C
isotherms (SIEVERS and EMERY, 1978; LtrrJErIARMS and VALE~rrIr~, 1984) and Can be located
reasonably well by the 4°C isotherm at 200m depth in the southwestern Atlantic (Pure_arsonand
WHrrWORTH,1989). Shown in Fig.24 is a field of the depth of the 4°C isotherm as observed during
late winter conditions in the region of the Falkland Current. On the basis of this field alone, the
SAF appears to turn sharply northward east of Drake Passage, following the Patagonian shelf to
the Brazil-Falkland Confluence Zone at about 40°S whereupon it retroflects back toward the south
before making an eastward turn in the southern Argentine Basin. Vertical sections of salinity,
oxygen, and nutrients confirm such a course, not only in the survey depicted, but in another during
summer conditions in the western Argentine Basin (Pme~SON and Wm'rwoRTn, 1989). Tem-
perature-salinity relations and horizontal distributions of density also show the thermohaline
structure within the cyclonic trough of the Falkland Current as being similar to the water mass
zonation observed in northern Drake Passage (PIOLAand GORDOn, 1989).
Transport estimates for the Falkland Current have been relatively few. ZYRYAnOVand SEV~OV
(1979) used historical density fields in a multi-level, diagnostic numerical model to calculate
depth-integrated transports of the Falkland Current at 45°S to be 32Sv during the southern
summer and 40Sv in winter, noting that a large cyclonic flow field exists to the east of the
northward-flowing current. Using a reference level of 1400m, GORDONand GmW~6ROVE(1986)
obtained northward transports of 9.8 and 11.4Sv at 42°S and 46°S, respectively. But because their
surface geostrophic velocities (17cm s-x) in the Falkland Current were only half those of a pair of
surface drifters that traversed the same region, they concluded that non-zero bottom velocities
must exist and that their estimates represent lower limits. PtOLAand BIANCm(1990) used a number
of hydrographic sections made in the years since 1980 and found the Falkland Current transport
to range only between 10 and 12Sv relative to 1000m. This result indicates that the intemal
distribution of density in the current does not change appreciably, however, the absolute Falkland
Current transport is probably quite a lot higher. Characteristics of water masses at depth indicate
that the northward flow off the Patagonian shelf extends from the sea surface to the bottom (REID,
48 R.G. P~r~,son and L. S ~

/ / \\

~i'/ ~. / / ! li i:,il)

,.-.,

~, s L Iii
t

\ ~,:j,,m,:,~ t-s , "~:>i

i

8 I \ ",.@,l ~+

<-,
• ~, "~.~ /
 Upper-level circulation in the South Atlantic Ocean 49

(m) qldeQ
0 0 0 0 0
O 0 0 0 0
0

Q~
o ~ .~
C

o~

C2~
o ~

J~

~ O°o,
%

J ~ 26:1-C
50 R.G. Pro'aRSONand L. STgA~WA

70°W 60 50 40
35°S

40

45

50

55

60

~o. 24. Depth of the 4°C isotherm (m) observed from the R/V Atlantis I1 during the austral winter
of 1980 (adapted from P~[~.gsoN and WmTWORTU,1989). The course of the Subantarctic Front (SAF)
is approximated by the 200m contour, while the Polar Front (PF) is depicted with the heaviest line.
The hatched area represents central water bounded by the Brazil Current (BC).
 Upper-levelcirculationin the South Atlantic Ocean 51

NOWLn~and PAT~.RT,1977; GREENGROVE,1986; Pm r_a~SONand WHrrwoRxn, 1989; REID, 1989), which

makes a compelling case that a significant barotropic component exists in the Falkland Current
which can not be accounted for by choosing an arbitrary reference level. By using a quasi-
synoptic, full-depth hydrographic survey enclosing the Brazil-Falkland Confluence Zone,
PmeRSON (1990) estimated that the depth-integrated absolute northward transport of the Falkland
Current is approximately 70Sv at 42°S. This estimate was arrived atas a residual quantity required
to get mass balances in several isopycnic layers once the Brazil Current transport had been solved
on the basis of water mass characteristics. This estimate is much larger than any previous ones
for the Falkland Current, but was not thought to be unreasonable in light of the eastward flow from
the SAF and northwards in Drake Passage turning north over the Falkland Plateau (~2500m sill
depth) and into the Falkland Current subsequently to override abyssal flow also moving north. The
deeper flow in Drake Passage associated with the SAF, continues east into the Georgia Basin
before entering the Argentine Basin through deep breaks in the bathymetry (WruawoRaT-Iand NOWL~,
1990).
Just north of the SAF in the southeastern Pacific, the coldest variety of Subantarctic Mode
Water (SAMW) is formed in late winter by deep (400-600m) convection and is advected through
Drake Passage into the western South Atlantic (McCARTa~EY, 1977). But how much of this gets
into the South Atlantic is not known. In studying upper-level modal salinity and density
characteristics around Antarctica, Pioi~ and GEORGX(1982) concluded that relatively little of the
southeast Pacific SAMW is advected through Drake Passage. Qualitatively, this is consistent with
Fig.24 in which only a few isolated pools of nearly isothermal (-4.5°C) and isohaline (34.0-34.2)
water (reaching from the surface to as deep as 700m) are seen to occur both north and south of
the SAF. SAMW might be supplied to the South Atlantic only through episodic events.
At the poleward end of the eddy field generated by the confluence of the Brazil and Falkland
Currents is a surface temperature front, characterized by temperatures of 10 °- 12°C at 30m depth;
this was described by IKEDA,SIEDLERand ZWIERZ(1989) as the Falkland Escarpment Front. They
found it to be very closely pinned to the southern edge of the Argentine Basin (during the austral
summer) between 50 ° and 55°W in several north-south XBT sections made near the western
boundary. This locates their front at nearly the same place as where the SAF turns eastward after
returning to the south from the Brazil-Falkland Confluence Zone, but the relation between the
two fronts is unclear as the spatial extent and temporal permanence of the Falkland Escarpment
Front await further investigations.
Eastward from the narrow neck formed by its northward and return branches in the southwestern
Argentine Basin, PmeRSON and WmawORXri (1989) observed the SAF to coincide with the
northern side of the Falkland Plateau, and that in the region of the Ewing Bank the SAF can at times
merge with the PF. The combined horizontal density gradients of the two fronts were observed
to produce an intense eastward jet having surface geostrophic speeds of 80cm s -1 and more, or
twice the speeds associated with the individual fronts in Drake Passage. North of the Islas Orcadas
Rise and to the east, the two fronts were observed once again to be distinct. At both the Greenwich
Meridian (WmaWORTHand Nowise, 1987) and south of Africa ( L t r r ~ s , 1985), the SAF lies
some 500km north of the Polar Front.

4.2. Volume transport

The volume transport of the ACC has been a point of contention since the first estimate was
given by CLOWES(1933). He argued that the eastward flow of the ACC through Drake Passage
extends to great depth and that no water from the Atlantic passes westward into the Pacific. On
the basis of hydrographic stations made during 1929-1930 from ships of the Discovery expedi-
52 R.G. PmegsoN and L. STRAm~

tions, he arrived at a v o l u m e transport of 110Sv, or about 4 times that of the Florida Current.
Several other estimates have been made since then based on a variety of techniques and these are
summarized in Table 3. As listed, the estimates of volume transport through Drake Passage
obtained from in situ measurements range from more than 200Sv eastward, to flow in the opposite
direction. Much of the disparity comes from using the equivalent barotropic assumption and in
adjusting geostrophic shears to direct measurements in a streaky and highly variable current
system.

TABLe3. Estimates of the eastward volume transport of the Antarctic Circumpolar Current through
Drake Passage obtained from in situ measurements (after Prrratsor~, 1988a)

Transport
Author Data Reference (10~3s -1) Remarks

CLOWNS(1933) Discovery expeditions, 3500dbar I10 non-sysnoptic

1929-1930
SVIKRDRUP,JOHNSONand composite 3000dbar 90 transport function
FLm¢~o (1942)
KORT(1959) Ob, Jun 1958 3000m 134 not verifiable"b
bottom 165 not verifiable~b
EsraN (1959) Ob, Jun 1958 3500m 141 not verifiableb
Os'rAPOFF(1960) Discovery 11, Apr 1930 variable 0 Defantc+ EBA
OSTAPOr~(1951) Discovery 11, Apr 1930 variable -1 Defant + EBA
2000dbar 33
3000dbar 85
Ob, Jun 1958 variable 9 Defant + EBA
2000dbar 34
3000dbar 86
VOROB'VL~Vand GmoYsH composite, summer variable 119 Defant
(1965) composite, winter variable 91 Defant
GORDON(1967) Ob, Jun 1958 variable 218 T/S + EBA
REIDand NOWL~(1971) Washington, Jan 1 9 6 9 3000dbar 103
3500dbar 113
bottom 113
variable 237 relative to 6 CMRs
300m above bottom
Discovery 11, Apr 1 9 3 0 3000dbar 84
3500dbar 92
bottom 92
Discovery H, Mar 1 9 3 4 3000dbar 88
3500dbar 105
bottom 117
Ob, Jun 1958 3000dbar 87
3500dbar 108
bottom 113
FOSTER(1972) Hudson, Feb 1970 variable -5 Defant + EBA
-15 relative to 12 CMRs
from 4 moorings
 Upper-level circulation in the South Atlantic Ocean 53

Transport
Author Data Reference (10t'm3s"1) Remarks

NOWLIN, WHrrWORTHand Melville, Feb-Mar 1975 3000dbar 95 5-section average

Pa.LSat~Y (1977) variable 124 relative to CMRs,
several moorings
Thompson, Mar 1976 3000dbar 110
Hudson, Feb 1970 3000dbar 75
BRYDEN and PILLSBURY ISOS, Mar-Oct 1975 ~2700m mean:39a barotropic only,
(1977) range:220 relative to 6 CMRs
WHrrWORTH (1980) 16 summer sections 2500dbar 79 ± 13
2500dbar 81 ± 8 transport function
6 winter sections 2500dbar 71 ± 15
2500dbar 74 ± 15 u~nspert function
LtrrJEHARMS (1982) composite 3000m ~200 trans.fun.(45°S)
3000m -160 Irans.fun.(60°S)
NOWL~ and CLI~ORD Thompson, Feb 1976 2500m 88
(1982) Thompson, Feb-Mar 1976 2500m 88
Melville, Feb-Mar 1975 2500m 87
Melville, Mar 1975 2500m 86
Atlantis H, Jan-Feb 1980 2500m 88
WHrrWORTH, NOWL~ and Melville, Jan-Feb 1977 3000dbar 75
WORLEV (1982) Melville, Jan-Feb 1979 3000dbar 110
Yelcho, Apr-May 1979 3000dbar 102
Atlantis H, Jan-Feb 1980 3000dbar 105
Melville, Jan-Feb 1979 variable 117 relative to CMRs,
several moorings
Yelcho, Apr-May 1979 variable 144 relative to CMRs,
several moorings
Atlantis H, Jan-Feb 1980 variable 137 relative to CMRs,
several moorings
WHITWORTH(1983) ISOS, speed at 500m mean: 12" in upper 2500m
Jan 1979-Feb 1980 range: 61
WmTWORTH and 500m pressures, mean: 123 inupper 2500m, from
PL~SON (1985) Jan 1977-Feb 1980 and a regression
Mar 1981-Mar 1982 range: 63 model using across-
passage pressure
differences only

EBA equivalent barotropic assumption

-

T/S temperature/salinity relation, used to identify deep countercurrent in southern Drake Passage (COP,DON,
-

1966)
CMR current meter record
-

ISOS - International Southern Ocean Studies

• Noted by OSTAPOPF(1961)
b Noted by REIO and NOWLm (1971)
c Defant's method used for determining reference level
Authors suggest adding 100Sv for total transport
° Revised value given by WHrrWORTHand t'ETEaSON
54 R.G. Pm~a<sor~and L. S ' m . ~

Resolution of the banding of the ACC was a key goal of the ISOS program, which culminated
with the deployment of an extensive array of moored instruments across Drake Passage during
the year 1979. In addition to much information regarding the space and time scales of current
variability (see NowLm and KLINCr, 1986 for review), it was learned that while most of the ACC
transport is in the baroclinic field, the transport fluctuations are mainly barotropic (WmTWORTn
and Purm<SON, 1985). This is consistent with REIDand NowLn~ (1971) who had earlier pointed out
that the internal pressure field is relatively steady and that the various hydrographic sections yield
similar transports when treated in the same manner. Referring to Table 3, Wm'lWORTH(1980)
observed the baroclinlc volume transport relative to 2500dbar across 16 summer sections to
average about 80Sv with standard deviations of 10%-15% of the mean value. In five summer
sections, NOWL~ and CLn~ORD (1982) obtained transports relative to 2500m ranging from 86 to
just 88Sv. Relative to 3000dbar, there are 12 independent and verifiable, quasi-synoptic estimates
that average 93Sv and range from 75 to 110Sv. Relative to the bottom, there are four such
estimates averaging 109Sv and ranging from 92 to 117Sv.
The baroclinic component of variability is comparatively small, which allowed WrnaWORTH
and Pb~ re<SON(1985) to use multi-year records of bottom pressure at each side of Drake Passage
to estimate variations in volume transport over longer time intervals than otherwise possible.
Their time series of volume transport in the upper 2500m is shown in Fig.25, obtained from across-
passage differences of bottom pressure at 500m depth. The mean value, 123Sv, was set equal to
that of a time series of net transport through the passage during 1979 which was computed with
additional information (WmawOaT,, 1983; WnrrWORTHand P~r~_a~SON,1985). In Fig.25 it is of
particular interest to note that transport fluctuations of nearly half the mean value can occur over
periods as short as two weeks (July 1978 and June-July 1981), and that there is no interannual
phase relationship in the seasonal variations. Much longer time series are needed for a better
understanding of transport fluctuations occurring over seasonal and longer time scales, but
unfortunately, an attempt to extend the bottom pressure series through the use of coastal sea level
records spanning two decades and more from each side of Drake Passage was not successful
(PETERSON, 1988b).
There have been only a few attempts to relate transport fluctuations of the ACC with the
overlying wind fields. VAN LOON (1972) adjusted monthly mean sea level differences across
Drake Passage with atmospheric sea level pressure, finding that the semi-annual oscillation in the
slope of the sea surface is inconsistent in phase with that of the local eastward surface winds.
WEARS and BAKER(1980), however, obtained high coefficients of correlation (0.6) between two
years of bottom pressure differences across Drake Passage at 500m depth, smoothed with a 28-
day running mean filter, and zonally-averaged eastward wind stress in the latitude band of 40 °-
65°S. As they pointed out, their high correlations were entirely the result of the agreement
between bottom pressure at the southern side of the passage and the wind stress fields. Pressure
at the northern side was poorly correlated with wind. CHELTON(1982) noted that their high
correlations might have resulted merely from both series containing strong seasonal signalL
Using a yearlong time series of transport through Drake Passage based on records obtained from
an extensive array of moored instruments, WmaWORTn(1983) found only a narrow band of periods
(16-24 days) where transport and zonally-averaged eastward wind stress were coherent at the 95%
confidence level.
The spectral composition of each the baroclinic and barotropic components of transport
variability at Drake Passage was analyzed by ~ S O N (1988a). It was found that transport
variability occurring over subseasonal time scales, as measured by moored instruments at each
 Upper-level circulation in the South Atlantic Ocean 55

• , • , ,j~¥, ) i • ,

81-82,'~

/i
..79- 80 ~. :"""
", ......... ~:: H....~:

16C
.,: ,,,,,,.,.,.,,,.,.,.,.,.,.,.,.,.,.,.,.,.,.,.,.~.,..., ,.,..,.,.,.,.,.,.,.,.,.
0/ 1"/ "
1,50
L . I , , , I ' i i

140

E
~ao

0 120
m
¢
el
I)O

100

Jan July Jan July Jan July Jan Jan July Jan
1977 1978 1979 1980 1981 1982

F~o,25. Volume transport of the ACC throughthe upper 2500mof Drake Passagefzomacross-passage
differences of bottom pressure at 500m depth for January 1977 to February 1980 and for March 1981
to March 1982. The fight curve shows 10-day low-pass filtered transport, which is smoothed with a
90-day low pass filter (heavy curve) to illustrate the seasonal variablity. The inset shows yearlong
segments of the 90-day low-passed series (WarrwoRTHand P~re~soN, 1985).

side of the passage, has important components arising from the lunar fortnightly and monthly tides
and baroclinic activity at the northern side of the passage, processes that are independant of wind.
Although the time series of dynamic height at the two sides of the passage (which provided
information on the baroclinic variability) were only a year long apiece, it was thought that most
of the variability over seasonal time scales is barotropic. At the northern side of the passage the
seasonal signal is largely annual, but not phase-locked, whereas at the southern side it has both
annual and semi-annual components.
A nonlinear analytical model was presented by ~ (1982) which also shows that the
seasonal transport fluctuations are barotropic, and that in the mean the Sverdrup balance does not
hold for the ACC. This latter point seems to be the consensus of opinion (though STOM]~mL(1957)
has argued that the ACC does not really occupy a zonal channel in which there are no Sverdrup
solutions, but that there are partial mefidional boundaries which could lead to much of the A C C ' s
flow being Sverdrup-like). Many models have been formulated in which the ACC is driven by
the eastward winds (see NOWLINand KLI~CK, 1986), and a recent example is the work by JOHNSON
and BRYDm~ (1989) where the mean zonal momentum imparted by wind was seen as being
transported downward by eddy form drag to be removed at great depth by bottom pressure drag.
Over seasonal time scales, the eastward wind stress in the Southern Ocean varies with a
strongly semi-annual character, whereas bottom pressure at both sides of Drake Passage have
important annual components. Cuff of wind stress, on the other hand, changes from being mainly
56 R.G. Pmmtsou and L. S ~

semi-annual at the higher latitudes to mainly annual north of about 45°S. Pmv_~soN(1988a) supposed
that these differences in the seasonal variability of curl could conceivably lead to redistributions
of mass through a Sverdrup-like response and therefore to changes in the slope of the sea surface
across the ACC. The multi-year bottom pressure series at Drake Passage were then compared with
zonally-averaged values of wind stress curl north and south of the ACC, with the results showing
that the annual cycle in the strengths of the subtropical highs is consistent with the annual signals
in bottom pressure at both sides of the passage, whereas effects from the strong semi-annual
variability of the subpolar lows are restricted to the southern side of the ACC. Fig.26 shows 3-
year long comparisons between the zonally-averaged curl of wind stress in a band along the
northern side of the ACC (40°-46°S) and bottom pressure (inverted) at the northern side of the
passage, and between curl along the southern side of the ACC (60°-66°S) and bottom pressure at
the southern side of the passage. The qualitative agreement is fairly good, except for a phase
problem at the northern side during 1978. It was noted that if the ACC were closed along one side
and subjected to the changes in curl along the other, the pure Sverdrup transport zonally-integrated
along the current would be 2-3 orders of magnitude larger than needed to account for the observed
changes in bottom pressure, indicating that the ACC's Sverdrup-like response is very small.
Also, a tacit assumption made in this type of analysis is that the seasonal fluctuations of ACC
transport are zonally coherent, which is a very poorly understood problem. The only information
about this comes from two studies using satellite altimeter data: Fu and CBELTON(1984) used three
months of Seasat data to infer a general increase in the eastward surface velocity between 40 ° and
65°S, and Crt~TON, SCHLAX,Wrrc~R and RlCrtMAN(1990) used empirical orthogonal functions on
26 months of Geosat data to find only a very weak zonal coherency in the lowest modes (primarily
the annual and semi-annual cycles). But the latter study did not extend poleward of 60°S, thus
leaving the question of net flux from one basin to another unresolved. Additionalinsituandsatellite
data are obviously needed.
Quasi-synoptic estimates of volume transport of the ACC in other regions of the South Atlantic
have been few in number and are influenced by other flow regimes bordering the current. Between
South Africa and Antarctica, JACOBSand GEORGI(1977) obtained a net eastward transport of 129Sv
relative to the deepest observations, whereas GEORGIand TOOLE(1982) otained 140Sv for the ACC
(also relative to the deepest observations). WHITWORTBand NOWLIN(1987) used a section ex-
tending from Cape Town to the Greenwich Meridian at about 40°S and then southward to the
Antarctic continent to estimate the net eastward transport as 142Sv relative to the bottom. Results
from their computations are shown in Fig.27. The highest surface speeds, 26 and 23cm s-~, are
found at the SAF and PF, respectively, as are the greatest volume transports between station pairs.
The STF, crossed at the north in the upper panel of Fig.27, has surface speeds of 13cm s -~and virtually
no zonal flow beneath 2000m depth (refer to section 3.5 for discussions of the STF). The other
major front in this section is the ACC-Weddell Gyre Boundary, which has a surface speed of 12cm
s -~just south of the mid-ocean ridge near 55°S.
Finally, non-synoptic data have been used to compute the baroclinic transport function relative
to 3000m for the ACC as a whole. Inspection of Fig. 163 in SVFmDRtlP,JOHNSONand FLEMING(1942)
shows a net eastward flow south of Africa of about 90Sv, whereas LtrrmuaRMS (1982) obtained
a value of approximately 300Sv, one that does not include the westward flow of the the Agulhas
Current (estimated by JACOBSand GEonoI, 1977 as being 136Sv) nor does it include any westward
flow near the Antarctic continent. The disparity among these few estimates are large and point
to a clear need for additional measurements in the region.
 Upper-level circulation in the South Atlantic Ocean 57

1977 1978 1979

12 July Jan. July Jan. July Jan. July
...... i -) iil; ..... ' ...... . . . . ' .. . . .. . . . . . . . ' .. . . . . , . . . .. . . . . . . . . . , . . . . . .. . . . . . . , ........... 15
. . . . . . ~. A cI~L 4 0 - 4 6 3
i,
el

',t I ,' .... /" ,

o a'~ o- • r ~ Z
}:"""
i',' ~ l
:/ i½'/ i /1" "
/ oo
-6
L '"', ;
:,
,",
',°
0.0

--12 I -.05

12 .... w . . . . . . ,--- ,- .o-d- --.- -- ,- .-I. -- ,- .- .- -- i- --,. -. ,. -. i- .. w- -- ,--- i- --.- - - . - - - i - -- ,. -. ,- -"

0.0

~URL,'60~r66S
, ,, . /', . .;
V ~ - , .,^ . • . . . . . .
-.1

, ,.., , ,, -'-'/1 . , ,, I

', /' , ,I ' ', , ,, k, ,' ' , ,

-.2 ~
';
~, .
~j
v . . VLU-
.
,,
-
.
,'
.... ,,
i i
/,l
f
,,,
f
:
-6
%'

. J . . . I . . . . . . . ,... I . . . , . . . , . . . I . . . ,... ,... I . . . , . . . , . . . I . . . ,... ,... I . . . , . . . , . . . I... ,... ,..

-12
July Jan, July Jon. July Jan, July
1 9 7 7 1 9 ? 8 ! 9"19

~ o . 26. (Top) Time series of zonally-averaged curl of wind s~ess (dashed curve) in the latitude band
40°-46°S, together with bottom pressure at 500m depth (solid curve, inverted, mean removed) from
the northern Drake Passage. (Bottom) Time series of zonally averaged curl of wind stress (dashed
curve) in the latitude band 60°-66°S, together with bottom pressure at 500m depth (solid curve, mean
removed) from the southern Drake Passage. All series are smoothed with a 60-day low pass filter
(Pro ~,~sos, 1988a).

JPO Z6: l - O
58 R.G. ~ r ~ and L. Sa'RAMMA

60 65 70 75 80 85 90 94
h~, i I,~,~,,.l~l ~ l r ~ l r ~ , l ")
160

~120
E
%
112
0

~40
w

40*S 50* 60* 70*S

Fle. 27. (Top) Geostrophic speed (cm s"1,westward is shaded) and Coottom) integrated geostrophic
transport ( 1(PmS s-~),both relative to the deepest common sample depths for station pairs occupied by
R/V Knorr during October 1983 (stations 45-49) and January to February 1984 (stations 60-94)
(adapted from Wrm,woRrn andNowLm, 1987). Note that the northem ends ofthese two panels donot
follow the same lines (see inset).
 Upper-level circulation in the South Atlantic Ocean 59

5. MERIDIONAL HEAT TRANSPORT

Because the South Atlantic is the major conduit for water mass exchanges between the
globally-important sources for deep water masses in the North Atlantic and the rest of the World
Ocean, it has a unique pattern of heat transport which is directed toward the equator at mid- and
low-latitudes, contrary to an intuitive point of view where heat transport would be expected to be
directed in a symmetric fashion away from the equator. Although we do not discuss the deep
circulation here in any detail, it nonetheless seems appropriate to give a brief summary of
estimates of meridional heat transport in the South Atlantic.
According to ROE~nVnCH(1983), approximately 10Sv of warm upper-ocean water is trans-
ported northward across the equator from the South Atlantic, which is compensated by a
comparable southward transport of North Atlantic Deep Water. SXrBRDRtrP,JOHNSONand FLEMING
(1942, pg. 629) discussed this balance, which GORDON(1986) identified as being an important part
of a global-scale thermohaline circulation cell. With an inverse model, however, RncrouL (1990)
calculated that an export of 17Sv of deep water from the North Atlantic is balanced by an
equatorward return flow equally split between the surface layers and the intermediate and bottom
water. The South Atlantic can thus be seen as the link between the North Atlantic and the other
oceans in the global thermohaline circulation.
Although it is generally accepted that the South Atlantic has a northward heat flux, the exact
magnitudes are difficult to compute owing to large uncertainties in the computational methods
and to the unknown sizes ofinterannual and seasonal variability south of about 20°S. With respect
to computational methods, there are two classes that utilize observed quantities, the direct and
indirect methods. For the direct method, absolute velocities must be known, both the Ekman and
geostrophic components. Monthly mean values of meridional Ekman heat fluxes were computed
by LEvrros (1987) for the individual ocean basins. Poleward of 30°S in the South Atlantic, the
zonally-averaged meridional fluxes are less than 0.5PW, which are small when compared with
the Ekman heat fluxes north of 30°S, which are directed toward the south. A seasonal signal is
evident in the latitude band of 0°-15°S, with the largest changes occurring at 4.5°S; during May
to October there is a maximum southward Ekman heat flux of 2.5PW, whereas during January to
March it reduces to about 1.5PW. The latitude band of 15 °-30°S is marked by a relatively constant
Ekman heat flux throughout the year.
For indirect computations, there are two fundamentally different approaches. The classical
method has been to consider air-sea interactions and temporal changes of heat storage in the water
column. The energy fluxes across the sea surface are summarized in section 2 of this paper. The
annual cycle of heat storage for the individual ocean basins was computed by LEvrrcs (1984) on
the basis of climatological monthly-mean temperature fields. One of the results is that at mid-
latitudes the annual cycle of oceanic heat storage lags that of the incoming solar radiation by about
three months. In the mid-latitude South Atlantic, maximum heat storage occurs in March and
minimum storage in September. For a 5 ° wide latitude belt centered at 27.5°S, the heat stored in
the upper 275m of the South Atlantic is 72.2 x 102°J in excess of the annual mean during February
to April, whereas there is a deficit of 70.9 x 102°J during August to October. Using a larger data
set, HsnmG, NEWELLand HOUGnTSY(1989) also computed monthly means of heat storage in the
Atlantic Ocean north of 20°S, finding good agreement with the results of LEvrros (1984).
The other indirect method uses incoming and outgoing atmospheric radiation combined with
atmospheric heat transport to derive the oceanic component as a residual. This method has
revealed that meridional heat fluxes in both the ocean and atmosphere (northern hemisphere) lie
in the range of 0.1-1PW, with the ocean contributing more in the tropics and the atmosphere more
60 R.G. Pm'm~soN and L. S ~

¢ Bryan 1962 + Hsiung 1985 Sarmiento 1986

© Lamb 1981 • Fu 1981 • Hastenrath 1982
Bennett 1978 x Roemmich 1983 • Bunker 1988
~, Georgi,Toole 1982 • Rintoul 1990 ........ Philander,
Pacanowski 1986b

1.5
PW <>

1.0 • ..."" "......- +

• • ." +
A• ~ • x .-"
..
0.5

.¢
m+ ~ + +•
I I I I
60 ° 40 ° 20 ° S 0° 20 ° N-

Fio. 28. Total northward heat transport (PW) of the South and tropical Atlantic Ocean. Symbols
correspond to the indicated sources.

TAnI~ 4. Estimates of northward heat transport (PW) in the South Atlantic Ocean derived from (s)
indirect computations using surface estimates, (h) direct computations using hydrographic data to
solve for velocities, and (m) from numerical circulation models. The values attributed to S
(1986) and PmLZa,m ~ and PAct~ows~a (1986b) are adopted from figures shown in their papers. Those
for Fo (1981) are mean values of results given for each section

Author Method Latitude (°S)

0 8 15-16 20-21 24 28 30 32 60

H s a ~ o (1985) s 0.54 0.15 0.04

~ a T a (1982) s 0.98 0.69 0.60
Bum¢~ (1988) s 0.54
(1981) s 1.02
Bm~aqm-r(1978) h 0.65 0.68
BRYAN(1962) (Meteor) h 0.63
BRYAN(1962) (1GY) h 1.30 0.33
Fu (1981) (Meteor) h 0.41 0.86 0.54 0.83
Fu (1981) (IGY) h 0.18 0.73 0.54 0.77
Ro~t]~cH (1983) h 0.74 0.61
RncrouL (1990) h 0.15
S,~.tcmmcro (1986) m 0.70 0.38 0.37 0.40 0.44 0.27
P m T , . ~ , and
PAc.~ows~a (1986b) m 0.86 0.64
 Upper-level circulationin the South Atlantic Ocean 61

in the higher latitudes (OoRr and VonDm~ HAAR, 1976). Lastly, oceanic circulation models are also
used to estimate heat fluxes.
The various estimates of northward heat transport in the South Atlantic Ocean are summarized
in Table 4 and in Fig.28. As can be seen, all the estimates are positive (northward), but with large
differences in magnitude among them. The disparities likely result from a combination of large
seasonal variations (Hsnm6, NL~CCELLand Hou6m~v, 1989) and computational uncertainties. For
example, by using different widths for the western boundary current, B m ~ : r r (1978) obtained
northward heat transports across 32°S that varied from 0.16 to 0.68PW. Also, Table 4 shows that
even when the same Meteor and IGY sections are used, investigators using different methods
obtain widely varying estimates.
Using a general circulation model of the tropical Atlantic OCEAN,PhilAm~R ANDPacanowski
(1986b) found little seasonal variability of heat transport across 5°S. However, the northward heat
flux across 12°S was seen to vary from 0.4PW in August to 0.8PW in February. They also
computed heat flux across the sea surface in their model finding clear seasonality at latitudes
higher than about 10 °. At 12°S, there was a 25W m-2 gain by the ocean in February (austral
summer) and an equal loss in Aocust. Hsiung, NmVELLAnd Houghtby (1989) investigated sea-
sonal variations in heat transport between 50°N and 20°S with oceanic observations and indirect
computations. Fig.29 shows their results for the latitude band of 20°N-20°S. With the exception
of southward heat transport in the tropical South Atlantic for about three months in the austral
summer, all of their other monthly values are northward. But contrary to the model rESULTSof
Philar~OER ANDPacanowski (1986b), Hsmng, NLWELLAnd Houghtby (1989) found large seasonal
variations of heat transport across 5°S, ranging from a southward transport of 1.7PW in January
to a northward value of 1.6PW in March. Figure 29 also shows direct estimates from a number
of sources, which agree rather well on the whole with the indirect computAa~ONSOFHslung, NEWIK.L
And Houghtby (1989). Therefore, much of the scatter in Fig.28 is probably a result of seasonal
variablity.

6. OUTLOOK

In this paper we have attempted to provide a moderately detailed review of the upper-level
circulation in the South Atlantic Ocean. Although portions of the system of currents comprising
this circulation are fairly well understood, such as the Agulhas Current Retroflection, parts of the
Brazil Current, and the Antarctic Circumpolar Current within Drake Passage, there are still large
regions of the South Atlantic that have yet to be measured in detail, either spatially or through time.
We have only a vague idea of how bottom topography influences the course of the Benguela
Current, only very recent indications that the cyclonic geostrophic gyre in the Angola Basin is a
permanent feature, no solid understanding of how the equatorial currents feed into that gyre nor
of how water is exchanged between the individual current bands of the equatorial system. The
seasonal variability of the equatorial current system is becoming better understood, but how this,
and the atmospheric seasonality translates into the Brazil Current is not known at all. How do the
Agulhas eddies moving across the South Atlantic affect the Brazil Current? We have very little
information concerning the flow field in the interior regions of the subtropical gyre, and only
recently has there been a description of the current band that closes the gyre in the south. Does
the South Atlantic Current always turn completely north into the Benguela Current, or does a
significant amount of subtropical water ever flow directly from the Atlantic into the Indian Ocean?
How much thermocline water is lost to the Antarctic Circumpolar Current through the southward
shedding of Brazil Current eddies into the Subantarctic Zone?
62 R.G. Pin,arsoNand L. S~aMA

Dec Jan Feb Mar Apr May dun dul Aug Sep Oct Nov Dec Jan
' ' ' ' " '
oN

0
0 -05 f \ \

• \ / x // / / 10

°~V,'~,XiillllI IxO.~l(F)x" ' 1 fi / \ ] .1 I-!'5;~ os

2oIl-, °,\ i!,IIIIl fs B! , i II ,/ , ,I I / ,(M °,20
Dec Jan Feb Mar Apr May dun dul Aug Sep Oct Nov Dec Jan

FIo. 29. Monthly zonal averages of northwardheat transport (PW) for the Atlantic Ocean between
20°N and 20°S (adapted from I-Isnmo,Nmvo.tLand HouGrrrav,1989). Also shown are estimates from
the direct method of computation denoted with letters corresponding to their source: B -BRYAN
(1962), R - Roedbnc'n (1983), F - Fo (1981).

There are equally important uncertainties with the circumpolar current in much of the South
Atlantic. The northward flow of the Falkland Current appears to be bottom-reaching, and when
transport estimates beating this consideration are made they come up to be much larger than had
usually been thought• How much of this water continues north without returning to the
circumpolar current, and how would this affect the meridional heat and salt fluxes in the South
Atlantic and the overall balance in the global thermohaline circulation? Is there any seasonality
in the strength of the Falkland Current, as might be anticipated by the seasonal transport
fluctuations of the Antarctic Circumpolar Current? How would this relate to the apparent
seasonality in the separation of the Brazil Current? What happens to the water between the major
fronts of the ACC when the fronts come together? There must certainly be important meridional
exchanges in such occurrences.
These are only a few of the questions concerning the upper-level circulation in the South
Atlantic that might be raised, and a host of others exist for the deep circulation as well. It is clear
that much work remains to be done, and with improvements in our knowledge of circulation in
the South Atlantic, there will come improved input to coupled atmosphere-ocean models, which
would lead to better predictions of future climatic states. Field experiments planned for the South
Atlantic during the forthcoming World Ocean Circulation Experiment and other activities will no
doubt provide the basis for filling large gaps in our understanding of this extraordinary ocean.

7. ACKNOWLEDGEMENTS

This work has been supported by the Deutsche Forsehungsgemeinschaft grant Si 111/34-1 (RGP) and
by the Btmdesminister ffir Forschung und Technologie grant 07KF2128-12a (LS), both in Bonn,
Federal Republic of Germany. We would like to thankSophie Wacongne and Robert Molinari for their
helpful comments.
 Upper-level circulation in the South Atlantic Ocean 63

8. REFEREMCES

ADA~mC,D. and J.J. O'BRn~ (1978) The seasonal upwelling in the Gulf of Guinea due to remote forcing. Journal
of Physical Oceanography, 18, 1050-1060.
ARN^ULT,S. (1987) Tropical Atlantic geostrophic currents and ship drifts. JournalofGeophysicalResearch,92,5076-
5088.
ARxnt~, R.S. (1960) A review of the calculation of ocean currents at the equator. Deep-SeaResearch, 6, 287-297.
BANO, N.D. (1970) Dynamic interpretations of a detailed surface temperature chart of the Agulhas Current
retroflexion and fragmentation area. South African GeographicalJournal, 52, 67-76.
BANo,N.D. (1973)Characteristics of an intense ocean frontal system in the upwell regime west of Cape Town. Tellus,
25, 256-265.
B~G, N.D. and W.R.H. Ar~REWS(1974) Direct current measurements of a shelf-edge frontal jet in the Southern
Benguela system. Journal of Marine Research, 32,405-417.
Bam~ow, E.W. (1931) Cmrents in the western portion of the Indian Ocean. Parts I, II, III. Marine Observer,8,130-
132, 193-194, 254-259.
B~rL~rr, A.F. (1978) Poleward heat fluxes in the southern hemisphere oceans. Journal of Physical Oceanography,
8, 785-798.
B~r~n'r, S.L. (1988) Where Three OceansMeet: the Agulhas Retroflection Region. Ph.D. dissertation, WHOI-88-
51, Woods Hole Oceanographic Institution/Massachusetts Institute of Technology Joint Program in
Oceanography, Woods Hole, Massachusetts, 367 pp.
B~SCH, M. and G.A. B~a~a~ (1986) The surface expression of the western Weddell-Scotia Confluence.
Meeresforschung, 31, 58-69.
B ~ , J. (1969) Atmospheric teleconnection from the equatorial Pacific. MonthlyWeatherReview,97,163-172.
Bt~cK, R. and D.B. BOUDRA(1981) Initial testing of a numerical ocean circulation model using a hybrid (quasi-
isopycnic) vertical coordinate. Journal of Physical Oceanography, 11,755-770.
BOm,mo~, G. ( 1936) Temperatur, Salzgehalt trod Dichte an der Oberfl~che des Atlantischen Ozeans. Wissenschaftli-
che Ergebnisse der Deutschen Atlantischen Expedition Forschungs- trod Vermessungssehiff Meteor 1925-
1927, 5, 1-249.
Bost~Y, K.T. (1990) A simple model for the cyclonic gyre of the South Atlantic. Abstract in: Eos, 71(17), 544.
BorNmov, V.N. (1963) Geographical position of the Antarctic convergence zone in the Pacific Ocean. Soviet
Antarctic Information Bulletin, 4, 324-327.
BotrORA,D.B. and E.P. CH.~sIOr~'r (1988) Dynamics of Agulhas Retroflection and ring formation in a numerical
model. Part I: The vorticity balance. Journal of Physical Oceanography, 18, 280-303.
BOOVRA,D.B. and W.P.M. D~ RtJu trat (1986) The wind-driven circulation of the South Atlantic-Indian Ocean--I/.
Experiments using a multi-layer numerical model. Deep-SeaResearch, 33,447-482.
B~ows, O.B., R.H. EvANs, D.B. OLSONand G. Poom~rA (1985) A satellite infrared survey of the Brazil/Falklands
Confluence. Abstract in: Eos, 66(18), 291.
BRYAN,K. (1962) Measurements of meridional heat transport by ocean currents. Journalof GeophysicalResearch,
67, 3403-3414.
BRYDEN,H.L. and E.C. BRADY(1985) Diagnostic model of the three-dimensional circulation in the upper equatorial
Pacific Ocean. Journal of Physical Oceanography, 15, 1255-1273.
B~YO~, H.L. and R.D. PILLSBURY(1977) Variability of deep flow in Drake Passage from yearlong current
measurements. Journal of Physical Oceanography, 7, 803-810.
BUNKER,A.F. (1988) Surface energy fluxes of the South Atlantic Ocean. Monthly WeatherReview, 116, 809-823.
BtrRLm6, R.W. (1961) Hydrology of circumpolar waters south of New Zealand. Bulletin 143, New Zealand
Department of Scientific and Industrial Research, 66pp.
CANE,M.A. (1983) Oceanographic events during El Nifio. Science, 222, 1189-1195.
CANE,M.A. and E.S. S,~,ACHIK(1983) Equatorial oceanography. ReviewsofGeophysicsand SpacePhysics, 21,1137-
1148.
CAVmD~S,C.N. (1973) SScas and El Nifio: Two simultaneous elimatical hazards in South America. Proceedingsof
the Association of American Geographers, 5, a.a. a.9"
~ , P. and L.V. SI-a~otoN (1985) The Benguela ecosystem. 2. Chemistry and related processes. In: Ocea-
nography andMarine Biology, An Annual Review,23, M. BAm~s, editor, Aberdeen University Press, 183-
251.
CHASS~SL~'r,E.P. and D.B. BotrVRA(1988) Dynamics of Agulhas Retroflection and ring formation in a numerical
64 R.G. P~reaSONand L. S~aMA

model. Part H. Energetics and ring formation. Journal of Physical Oceanography, 18, 304-319.
CroUTON,D.B. (1982) Statistical reliability of the seasonal cycle: Comments on "Bottom pressure measuronumts
across the Antarctic Circumpolar Current and their relation to wind." Deep-SeaResearch,29, 1381-1388.
CmmTON,D.B., M.G. Scm,Ax,D.L. Wn-re~ and J.G. RXCHMAN(1990) Geosat Altimeter observations of the surface
circulation of the Southern Ocean. Journal of Geophysical Research (in press).
CBENEY,R.E., J.G. MARSHand B.D. BF.OO.~Y(1983) Global mesoscale variability from collinear tracks of SEASAT
altimeter data. Journal of Geophysical Research, 88, 4343-4354.
O.ARKE,A.J. (1982) The dynamics of large-scale, wind-driven variations in the Antarctic Circumpolar Current.
Journal of Physical Oceanography, 12, 1092-1105.
O.IFFORD,M.A. (1983) A descriptive study of the zonation of the Antarctic Circumpolar Current and its relation to
wind stress and ice cover. M.S. thesis, Texas A&M University, College Station, 93pp.
CLOWES,A.J. (1933) Influence of the Pacific on the circulation in the South-West Atlantic Ocean. Nature,London,
131, 189-191.
CLOWES,A.J. (1950) An introduction to the hydrology of South Afi'icanwaters. Investigational Report 12, Fisheries
and Marine Biological Survey Division, South Afi'ica, 28 pp., 20 charts.
COCtmANE,J.D., F.J. I¢d~LLY,JR. and C.R. OLLINO(1979) Subthermocline countercurrents in the western equatorial
Atlantic Ocean. Journal of Physical Oceanography, 9, 724-738.
CovEY,D.L. and S. ~ T H ( 1978) The PacificEl Nifiophenomenonandthe Atlanticcirculation. MonthlyWeather
Review, 106, 1280-1287.
DARnYSmRE,J. (1972) The effect of bottom topography on the Agulhas Current. Pure andAppliedGeophysics, 101,
208-220.
DARBYSI~m~,M. (1963) Computed surface currents off the Cape of Good Hope. Deep-SeaResearch, 10, 623-632.
DF.ACON,G.E.R. (1933) A general account of the hydrology of the South Atlantic Ocean. DiscoveryReports, 7,171-
238.
D~CON, G.E.R. (1937) The hydrology of the Southern Ocean. DiscoveryReports, 15, 3-122, 44 plates.
DEACON,G.E.R. and T.D. FOSTER(1977) The boundary between the Weddell Sea and Drake Passage currents. Deep-
Sea Research, 24, 505-510.
DEACON,G.E.R. and J.A. MOORU~(1975) The boundary region between currents from the WeddeU Sea and Drake
Passage. Deep-SeaResearch, 22, 265-268.
D~Awr, A. (1928) Die systematische Erforschung des Weltmeeres. In: Zeitschrifl derGesellschaftfiir Erdlmndezu
Berlin, Sonderbandzur Hundertjahrfeierder Gesellschafi, Berlin, 551 pp., 41 plates.
D~Awr, A. (1936) Schichtung und Zirkulatiun des Atlantischen Ozeans, Die Troposph~e. Wisscnschaftliche
Ergebuisse der Dentschen AtlantischenExpeditionForschungs-undVermessungsschiffMeteor 1925-1927,
6(1), 289-411.
D~A~rr, A. (1961) Physical Oceanography, Volume 1, Pergamon Press, 729 pp.
DE Rutted.a,W.F.M. and D.B. BOUDRA(1985) The wind-driven circulation in the South Atlantic-Indian Ocean. I.
Numerical experiments in a one-layer model. Deep-SeaResearch, 32, 557-574.
Dm'rmcH,G. (1935) Aufbanund Dynamikdes sfidlichenAgulhasstromgebietes. Ver6ffentlichungen des Instituts fiir
Meereskunde, Universit~t Berlin, A(27), 79 pp.
Dt~rracH, G. (1936) Aufbau und Bewegung yon Golfstrom und Aguihasstrom, eine vergleicbende Betrachtung.
Naturwissenschaften, 24, 225-230.
D0n~o,W., P. HISARD,E. KATZ,J. MEINCK~,L. Mn w~, K.V. MOROSHKIN,G. PHR.AND~,A.A. RmNIKOV,K. VOIOTand
R. W~SBERO(1975) Meanders and long waves in the equatorial Atlantic. Nature,London, 257,280-284.
DUNCAN,C.P. (1968) An eddy in the Subtropical Convergence southwest of South Africa. Journal of Geophysical
Research, 73, 531-534.
Dtn~CAN,C.P. (1970) The Agulhas Current. Ph.D. thesis, University of Hawaii, 76 pp.
EMERY, WJ. (1977) Antarctic Polar Frontal Zone from Australia to the Drake Passage. Journal of Physical
Oceanography, 7, 811-822.
Esncs~, C.C. and E.J. KA~Z(1987) Equatorial dynamics. Reviews of Geophysics, 25, 217-226.
E s ~ , L.L (1959) Contribution to the study of the water and thermal balance of Drake Passage (in Russian).
Information Byulletin Sovetskaia Antarkticheskaia ekspedifsiia, 12, 29-32 (English translation in: Soviet
Antarctic Expedition Information Bulletin, 2, 52-55, 1964).
EVANS,D.L. and S.S. SIONOR~(1985) Vertical structure of the Brazil Current. Nature, London, 315, 48-50.
EVANS,D.L., S.R. SioNOm~and L.B. MIRANDA(1983)A note on the transport of the Brazil Current. JournalofPhysical
Oceanography, 13, 1732-1738.
 Upper-level circulation in the South Atlantic Ocean 65

FAHRBACH,E., J. MI~NCKEand A. SY (1986) Observations of the horizontal separation of the salinity core and the
current core in the Atlantic Equatorial Undercurrent. Journal of Marine Research, 44, 763-779.
FANDRY,C. and R.D. PmLSBORY(1979) On the estimation of absolute geostrophic volume transport applied to the
Antarctic Circumpolar Current. Journal of Physical Oceanography, 9, 449-455.
FOFONO~,N.P. and R.B. MONTGOMERY(1955) The Equatorial Undercurrent in light of the votticity equation. Tellus,
7, 518-521.
FOSTER,L.A. (1972) Currentmeasurementsin the DrakePassage. M.S. thesis, Dalhousie University, Halifax, Nova
Scotia, 61 pp.
Fo, L.-L. (1981 ) The general circulation and meridional heat transport of the subtropical South Atlantic determined
by inverse methods. Journal of Physical Oceanography, 11, 1171-1193.
Fo, L.-L. and D.B. CHELTON(1984) Temporal variabilityof the Antarctic Circumpolar Current observed from satellite
altimetry. Science, 226, 343-346.
Ftronasr~ F.C. (1960) Atlantic Ocean atlas of temperature and salinity profiles and data from the International
Geophysical Year of 1957-1958. Atlas Series, Volume 1, Woods Hole Oceanographic Institution, Woods
Hole, Massachussetts, 209 pp.
G~mLD, N. (1988) Surface characteristics of the Brazil Current. Abstract in: Eos, 69(44), 1237.
GA~zota, S.L. and A. B t ~ c m (1987) Time-space variability of the local dynamics of the Malvinas-BrazilConfluence
as revealed by inverted echo sounders. Journal of Geophysical Research, 92, 1914-1922.
GARZOLI,S.L. and Z. G~autAr~o(1989) Transport, frontal motions and eddies at the Brazil-Malvinas Confluence.
Deep-Sea Research, 36, 681-703.
GBOROI,D.T. and J.M. TOOL~(1982) The Antarctic Circumpolar Current and the oceanic heat and freshwater budgets.
Journal of Marine Research, 40 (supplement), 183-197.
GoDoi, S.S. DE(1982) Esmdo das variagoes sazonals da Frenta Oce~fica Subtropical entre a Corrente do Brasil e
a Corrente das Malvinas, utilizartdo dados oceanoglTlficos e do sat6lite SMS-2. Disserta~ao de Mestrado
em Sensoriamento Remoto, Publica~ao no.INPE-2780-TDL/137, Instituto de Pesquisas Espaeiais, Sao
Jos6, Brasil, 328 pp.
GORVON,A.L. (1966) Potential temperature, oxygen and circulation of bottom water in the Southern Ocean. Deep-
Sea Research, 13, 1125-1138.
GORDON,A.L. (1967) Geostrophic transport through the Drake Passage. Science, 156, 1732-1734.
GORDON,A.L. (1981) South Atlantic thermocline ventilation. Deep-SeaResearch, 28, 1239-1264.
GORDON,A.L. (1985) Indian-Atlantictransfer of thermocline water at the Agulhas Retroflection. Science,227,1030-
1033.
GOROON,A.L. (1986) Interocean exchange of thermocline water. Journalof GeophysicalResearch, 91, 5037-5046.
GORDON,A.L. (1989) Brazil-Malvinas Confluence - 1984. Deep-SeaResearch, 36, 359-384.
GORDON,A.L. and K.T. Bost£Y (1990) Cyclonic gyre in the tropical South Atlantic. Deep-SeaResearch (in press).
GORDON,A.L., D.T. GBORO!and H.W. TAYLOR(1977) Antarctic Polar Front zone in the western Scotia Sea - summer
1975. Journal of Physical Oceanography, 7, 309-328.
GORDON,A.L. and C.L. GRm~OROV~(1986) Geostrophic circulation of the Brazil-Falkland confluence. Deep-Sea
Research, 33, 573-585.
GORDON,A.L. and W.F. I'D,xnY (1990) Agulhas eddies invade the South Atlantic: Evidence from Geosat altimeter
and shipboard conductivity-temperature-depthsurvey. Journalof GeophysicalResearch, 95, 3117-3125.
GoRoor~, A.L., J.R.E. L ~ s and M.L. GRONDLINGH(1987) Stratification and circulation at the Agulhas
Retroflection. Deep-SeaResearch, 34, 565-599.
GORDON,A.L. and W.D. NOWLm,JR. (1978) The basin waters of the Bransfield Strait. Journal of Physical Ocea-
nography, 8, 258-264.
GORDON,A.L. and A.R. PIOLA(1983) Atlantic Ocean upper layer salinity budget. Journal ofPhysical Oceanography,
13, 1293-1300.
GORDON,A.L., H.W. TAYLORand D.T. C_a~ROI(1974) Antarctic oceanographic zonation. In : Proceedingsof the
SCOR/SCAR Polar Oceans Conference, 5-11 May 1974, McGill University, Montreal, 682 pp.
GRm~GROV~,C.L. (1986) Thermohalinealteration of the South Atlantic pycnocline. Ph.D. dissertation, Lamont-
Doherty Geological Observatory of Columbia University, Palisades, New York, 211 pp.
GkwerrrlS,C.L. (1981) Upwelling processes in the southern Benguela region. SouthAfrican Journal of Science, 77,
99-100.
GRON-VLmOH,M.L. (1977) Drift observation from Nimbus VI satellite-tracked buoys in the southwestern Indian
ocean. Deep-SeaResearch, 24, 903-913.
66 R.G. Pe~reat~N and L. SaXAMMA

C~IYNDLt~OH,M.L. (1978) Drift of a satellite-tracked buoy in the southem Agulhas Current and Agulhas Return
Current. Deep-Sea Research, 25, 1209-1224.
GR0m~L~OH, M.L. (1980) On the volume transport of the Agulhas Current. Deep-Sea Research, 27, 557-563.
GR0WDL~6H,M.L. (1983) On the course of the Agulhas Cm'rent. South African Geographical Journal, 65, 49-57.
H_AN,Y.-J. and S.-W. Lm ( 1981 ) A new analysis of monthly mean wind stress over the global ocean. Report 26,
Climate Research Institute, Oregon State University, Corvallis, 148 pp.
HARRIS,T.F.W. and N.D. B ~ o (1974) Topographic Rossby waves in the Agulhas Current. South African Journal
of Science, 70, 212-214.
HARRIS,T.F.W., and D. VANFoR~s'r (1978) The Agulhas Current in March 1969. Deep-Sea Research, 25,549-561.
HARPaS,T.F.W., R. Let~V~xIsand D. VANFOR~S'r (1978) Satellite infra-red images in the Agulhas C'urrent system.
Deep-Sea Research, 25, 543-548.
HART,T.J. and R.I. CURRm(1960) The Benguela Current. Discovery Reports, 31, 123-298.
HAsrEm~Ttt, S. (1982) On meridional heat transport in the world ocean. Journal of Physical Oceanography, 12,922-
927.
HASTENRATH,S. and L. H~J~T~ (1977) Dynamics of climatic hazards in northeast Brazil. Quarterly Journal of the
Royal Meteorological Society, 103, 77-92.
HASmNRATI-I,S. and P.J. LAMB(1977) ClimaticAtlas of the TropicalAtlantic and Eastern Pacific Ocean. University
of Wisconsin Press, Madison, 15 pp., 97 charts.
HEAa~a,R.A. (1981) Oceanic fronts around southern New Zealand. Deep-Sea Research, 28,547-560.
H~.I ~ , S. and M. Ros~saxIN (1983) Normal monthly wind stress over the World Ocean with error estimates.
Journal of Physical Oceanography, 13, 1093-1104.
HntsT, A.C. and S. HASTEm~TH(1983) Atmosphere-ocean mechanisms of climate anomalies in the Angola-tropical
Atlantic sector. Journal of Physical Oceanography, 13, 1146-1157.
I-hSARO,P. (1980) Observation de r~ponses de type "El Nifio" dans l'Atlantique tropical oriental Golfe de Guin~e.
Oceanologica Acts, 3, 69-78.
H 6 ~ o t , O. (1974) The seasonal and secular variation of the meteorological parameters on both sides of the ITCZ
in the Atlantic Ocean. GATEReports, 2, Part VI, 36 pp.
Hov-Lictt,O. (1984) Climate of the South Atlantic Ocean. In: World Survey of Climatology, Volume 15, Climates of
the Oceans, H, VANLOON,editor, Elsevier, Amsterdam, 1-191.
HOFM.~, E.E. (1985) The large-scale horizontal structure of the Antarctic Circumpolar Current from FGGE drifters.
Journal of Geophysical Research, 90, 7087-7097.
Holu~, J.D., V.E. Kous~:Y and M.T. KAOASO(1986) Atmospbedc conditions in the Atlantic sector during 1983 and
1984. Nature, London, 322, 248-251.
HotJot-rroN, R.W. (1976) Circulation and hydrographic structure over the Ghana Continental Shelf during the 1974
upwelling. Journal of Physical Oceanography, 6, 910-924.
HOUQh~rON,R.W. (1989) Influence of local and remote forcing in the Gulf of Guine& Journal of Geophysical Re-
search, 94, 4816-4828.
HouorrroN, R.W. and C. COLIN(1986) Thermal structure along 4°W in the Gulf of Guinea during 1983-1984. Journal
of Geophysical Research, 91, 11727-11739.
HsItrSo, J. (1985) Estimates of global oceanic heat meridional transport. JournalofPhysical Oceanography, 15,1405-
1413.
Fishing, J., R.E. N ~ v - ~ and T. HouowraY (1989) The annual cycle of oceanic heat storage and oceanic meridional
heat transport. Quarterly Journal of the Royal Meteorological Society, 115, 1-28.
A, Y., L.B. MIRANDAand N.J. Rocx (1974) Observations on stages of upwelling in the region of Cabo Frio (Brazil)
as conducted by continuous surface temperature and salinity measurements. Boletim do Instituto
Oceanografico, Sao Paulo, 23, 33-46.
Ira,^, Y., G. S~DU~ and M. Zvn~,z (1989) On the variability of Southern Ocean front locations between southern
Brazil and the Antarctic Peninsula. Journal of Geophysical Research, 94, 4757-4762.
JACOBS, S.S. and D.T. GEOl~I (1977) Observations on the southwest Indian/Antarctic Ocean. In: A Voyage of
Discovery, M. AnG~, editor, Pergamon Press, 43-84.
Jom~soN, G.C. and H.L. BI~YI~E,~(1989) On the size of the Antaxctic Circumpolar Current. Deep-Sea Research, 36,
39-53.
JOHNSON,M.A. (1989) Southern Ocean surface characteristics from FGGE buoys. Journal of Physical Oceanog-
raphy, 19, 696-705.
Jo~w_s,P.G.W. ( 1971 ) The southern Benguela Current region in February, 1966: Part I, Chemical observations with
 Upper-level circulation in the South Atlantic Ocean 67

particular reference to upwelling. Deep-Sea Research, 18, 193-208.

I~WAN, A.D. (1963) Circulation of Antarctic Intermediate Water deduced through isentropic analysis. TAMU
reference 63-34F, Texas A&M University, College Station, 34 pp., 35 plates.
I~OLL, M., W. 7_a~NKand E. BAUER(1982) Some XBT-observations on the thermal structure of the Warmwassersph~e
in equatorial and lower latitudes of the eastern Atlantic. Deutsche Hydrographische Zeitschrift, 35, 73-81.
KNox, R.A. and D.L.T. AND~SON(1985) Recent advances in the study of low-latitude ocean circulation. Progress
in Oceanography, 14,259-317.
KORT, V.G. (1959) New data on the transport of Antarctic waters (in Russian). Information Byulletin Sovetskaia
Antarkticheskaia ekspedifsiia, 9, 31-34 (English translation in: Soviet Antarctic Expedition Information
Bulletin, 1, 358-361, 1964).
KRAt~SS,W. (1986) The North Atlantic Current. Journal of Geophysical Research, 91, 5061-5074.
KR~, O. (1882) Bemerkungen fiber die Meeresstr6mungen und Temperaturen in der Falklandsee.Aus dem Archiv
der Deutschen Seewarte, V, Hamburg, Nr. 2, 25 pp.
LAMB, P.J. (1977) On the surface climatology of the tropical Atlantic. Archivfiir Meteorologie, Geophysik und
Bioklimatologie, Serie B., 25, 21-31.
LAMB,P.J. (1981) Estimate of annual variation of Atlantic Ocean heat transport. Nature, London, 290, 766-768.
LAMB,P.J., R.A. P~I'LER and S. ~ T H (1986) Interannual variability in the tropical Atlantic. Nature, London,
322, 238-240.
L ~ . A , A., J.P. M ~ Y , JR. and D.W. MOORE(1981) Equatorial currents: measurements and theory. In:
Evolution of Physical Oceanography, B.A. WARRENand C. WVNSCH,editors, MIT Press, Cambridge,
Massachussetts, 184-196.
LFx~oos, R. and A.L. GORDON(1982) Satellite observations of the Brazil and Falkland currents - 1975 to 1976 and
1978. Deep-Sea Research, 29, 375-401.
~, W. (1975) Untersuchungen zur inneren hydrographischen Stmku~ des sfidlichen und mittleren Atlantiks (0-
2000m Tiefe) mit zoogeographischen Anmerkungen. Berichte tier deutschen wissenschffftlichenKommission
fiir Meeresforschung, 24, 1-22.
LLnrrrus, S. (1984) Annual cycle of temperature and heat storage in the World Ocean. Journal of Physical Ocea-
nography, 14, 727-746.
L~rros, S. ( 1987 ) Meridional Ekman heat flux es for the World Ocean and individual ocean basins. Journal of Physical
Oceanography, 17, 1484-1492.
LUTmnARMS,J.R.E. (1981) Features of the southern Agulhas Current circulation from satellite remote sensing. South
African Journal of Science, 77, 231-236.
LUTnmARMS,J.R.E. (1982) Baroclinic volume transport in the Southern Ocean. Journal of Physical Oceanography,
12, 3-7.
LtrrmHARMS, J.R.E. (1985) Location of frontal systems between Africa and Antarctica: some preliminary results.
Deep-Sea Research, 32, 1499-1509.
LtrrmUARmS,J.R.E. (1988) Meridional heat Uansport across the Sub-Tropical Convergence by a warm eddy. Nature,
London, 331,251-254.
LUTW.HARMS,J.R.E. (1989) The role of mesoscale turbulence in the Agulhas Current system. In: Mesoscale/Synoptic
Coherent Structures in Geophysical Turbulence, Proceedings of the 20th International Liege Colloquium
on Ocean Hydrodynamics, J.C.J. NIHOVLand B.M. JAMART,editors, Elsevier Oceanography Series, 50,
Amsterdam, 357-372.
L ~ s , J.R.E., R. CATZ~Land H.R. VALENTINE(1989) Eddies and other boundary phenomena of the Agulhas
Current. Continental Shelf Research, 9, 597-616.
LtrrmHARMS,J.R.E. and A.L. GORDON(1987) Shedding of an Agulhas ring observed at sea. Nature, London, 325,138-
140.
LtrrmHARMS,J.R.E. and H.R. ROBOTS(1988) The Natal pulse: An extreme transient on the Agulhas Current. Journal
of Geophysical Research, 93, 631-645.
LtrrmHARMS,J.R.E. and P.L. STOCKTON(1987) Kinematics of the upwelling front off southern Africa. South African
Journal of Marine Science, 5, 35-49.
LVTmHARMS,J.R.E. and H.R. V ~ (1984) Southern Ocean thermal fronts south of Africa. Deep-Sea Research,
31, 1461-1475.
LtrrmHARMS,J.R.E. and H.R. V ~ (1987) Water types and volumetric considerations of the south-east Atlantic
upweUing regime. South African Journal of Marine Science, 5, 63-71.
68 R.G. P~lm~SONand L. S ' m h m ~

L ~ , J.R.E. and H.R. Vm.~NT~ (1988a) On mesoscale ocean eddies at the Agnlhas Plateau. South African
Journal of Science, 84, 194-200.
Ltmmu, gMs, J.R.E. and H.R. VhtamTm~ (1988b) Eddies at the Subtropical Convergence south of Africa. Journal of
Physical Oceanography, 18, 761-774.
Ltrrmri~,Ms, J.R.E. and R.C. VANBm.a.~.OOOYEN(1984) Topographic control in the Agulhas Current system. Deep-
Sea Research, 31, 1321-1337.
LtrrmHARMS,J.R.E. and R.C. VANB Ata.EC,OOYEN(1988) The retroflection of the Agulhas Cth-'rent Journal of Physical
Oceanography, 18, 1570-1583.
Lo'cmN, J. (1988) Agulhas eddy variability and mean flow fields. Abstract in: SAARIMeeting Report, May 24-26,
1988, Lamont-Doherty Geological Observatory of Columbia University, Palisades, New York, pg. 18.
LtrYTEN,J.R., J. PI~LOSI(Yand H. STOMMEL(1983) The ventilated thermocline. Journal of Physical Oceanography,
13, 292-309.
MACraNTOSH,N.A. (1946) The Antarctic Convergence and the distribution of surface temperature in Antarctic waters.
Discovery Reports, 23, 177-212.
MAOtaOCCA,A., L.B. MIRANDAand S.R. SIONOmNI(1979) Physical and chemical aspects of transient stages of the
upwelling at southwest of Cabo Frio fLat. 23°S - Long. 42°W). Boletim do Instituto Oceanografico, Sao
Paulo, 28, 37-46.
M-~CARm~.hS, A.S., JR., L.B. MIRANDAand N.J. ROCK(1971) A study of the oceanographic conditions in the region
of Cabo Frio. In: Fertility of the Sea, J.D. COSTLOW,JR., editor, Gordon and Breach, New York, 285-308.
M~z~r~, P.A. (1967) Thermal domes in the eastern tropical Atlantic Ocean. Limnology and Oceanography, 12,537-
539.
~ , P.A. (1968) Eastward flow within the equatorial current in the eastern South Atlantic. Journal of Geo-
physical Research, 73, 5819-5828.
McChRTNEY,M.S. (1977) Subantarctic Mode Water. In: A Voyage ofDiscovery, M. ANOEL,editor, Pergamon Press,
103-119.
McCAa~TNEy,M. and J. Zma3A (1988) Thermocline, intermediate and deep circulation in the southwestern South
Atlantic. Abstract in: SAAR1Meeting Report, May 24-26, 1988, Lamont-Doherty Geological Observatory
of Columbia University, Palisades, New York, 28-29.
McCAgTNEY,M.S. and M.E. WOODOATE-JoNF_S(1990) A deep reaching anticyclonic eddy in the subtropical gyre of
the eastern South Atlantic. Deep-Sea Research (in press).
McCRF_ARY,J.P., J. PICAULTand D.W. MooP~ (1984) Effect of annual remote forcing in the eastern tropical Atlantic.
Journal of Marine Research, 42, 45-81.
MEIN~DUS, W. (1923) Meteorologische Ergebnisse der Seefahrt des "Gauss" 1901-1903. Deutsche Stldpolar-
Expedition 1901-1903, IR. Band, Meteorologie I. Band, 1. H~f-te, HeR I1/, 437-579.
M~LOR, G.L., C.R. MeCaOSOand E. Kgro (1982) A diagnostic calculation of the general circulation of the Atlantic
Ocean. Deep-Sea Research, 29, 1171-1192.
MERLE,J. (1980) Variabilit6 thermique annuelle et interannuelle de l'oc6an Atlantique &luatorial Est. L'hypoth6se
d'un "El Niho" Atlantique. OceanologicaActa, 3, 209-220.
MERLE, J. and S. ARNAULT(1985) Seasonal variability of the surface dynamic topography in the tropical Atlantic
Ocean. Journal of Marine Research, 43, 267-288.
Mmum, J., M. Fmtrx and P. HtSARI)(1980) Annual signal and interannual anomalies of sea surface temperature in the
eastern equatorial Atlantic Ocean. Deep-Sea Research, GATE (Supplement II), 26, 77-101.
Mgrc~a.~, W.G. and M.C. STALCtrP(1967) Origin of the Atlantic Equatorial Undercurrent. Journal of Geophysical
Research, 72, 4959-4975.
MLzrCALF,W.G., A.D. VooRras and M.C. STALCUP(1962) The Atlantic Equatorial Undercurrent. Journal of Geo-
physical Research, 67, 2499-2508.
MaO.NOA,L.B. and B.M. CASTROF~LHO(1982) Geostrophic flow conditions of the Brazil Current at 19°S. Ciencia
lnteramericana, 22, 44 48.
MOIaNARLR.L. (1982) Observations of eastward currents in the tropical South Atlantic Ocean: 1978-1980. Journal
of Geophysical Research, 87, 9707-9714.
MOLtNARI,R.L. (1983) Observations of near-surface currents and temperature in the central and western tropical
Atlantic Ocean. Journal of Geophysical Research, 88, 4433-4438.
MotaN~, R.L., B. VorrtrRiEZand P. DUNO-N(1981 ) Observations in the subthermocline undercurrent of the equatorial
South Atlantic Ocean: 1978-1980. OceanologicaActa, 4, 451-456.
 Upper-level circulation in the South Atlantic Ocean 69

MOORE,D.W. and S.G.H. ~ E g (1977) Modeling of the tropical oceanic circulation. In: The Sea: Ideas and
Observationson Progressin theStudy oftheSeas,Volume6:MarineModeling, E.D. GOLDBEXO,I.N. McCAvE,
J.J. O'BRIEN, and J.H. S~mmv, editors, Wiley, Interscience, New York, 319-361.
MOORE,D., P. HISARD,J. McCREARY,J. MERe, J. O'BRIEN,J. PICAULT,J.-M. VI~,SaRAE~andC. WUNSCB(1978) Equatorial
adjustment in the eastern Atlantic. GeophysicalResearch Letters, 5, 637-640.
MOROSHma~,K.V., V.A. BtmNov and R.P. BtrLArOV(1970) Water circulation in the eastern South Atlantic Ocean.
Oceanology, 10(1), 27-34.
MossY, H. (1934) The waters of the Atlantic Antarctic Ocean. ScientificResults of the NorwegianAntarctic Ex-
pedition, 1927-1928, 11, 1-131.
NELSON,G. and L. Htrrca~os (1983) The Benguela upwelling area. Progress in Oceanography, 12, 333-356.
NowL~, W.D., JR. and M. CLIFFORD(1982) The kinematic and thermohaline zonatiou of the Antarctic Circumpolar
CkLrrent at Drake Passage. Journal of Marine Research, 40 (supplement), 481-507.
NOWLIN,W.D., JR. and J.M. Kt~CK (1986) The physics of the Antarctic Circumpolar Current. Reviewsof Geophysics
and Space Physics, 24, 469-491.
NowL~, W.D., JR., T. Wm'rwoRTHI11 and R.D. PmLSBtrgY(1977) Structure and transport of the Antarctic Circumpolar
Current at Drake Passage from short-term measurements. Journalof Physical Oceanography, 7, 788-802.
O'BRIEN,J.J., D. ADAMECand D.W. MOORE(1978) A simple model of upwelling in the Gulf Of Guinea. Geophysical
Research Letters, 5, 641-644.
OLSON,D.B. and R.H. EVANS(1986) Rings of the Aguihas Current. Deep-SeaResearch, 33, 27-42.
OLSON,D.B.,A. GORDOr;and S. Hooggg (1985)Dynamic description of Brazil/Falkland rings. Abstractin:Eos,66(18),
291.
OLSON,D.B., G.P. POD~TA,R.H. EVANSand O.B. BROWN(1988) Temporal variations in the separation of Brazil and
Malvinas currents. Deep-SeaResearch, 35, 1971 - 1990.
OORT,A.H. and T.H. VONDER~ (1976) On the observed annual cycle in the ocean-atmosphere heat balance over
the northern hemisphere. Journal of Physical Oceanography, 6, 781-799.
OSTAPOFF,F. (1960) On the mass transport through the Drake Passage. Journal of GeophysicalResearch, 65, 2861-
2868.
OSTAPOFF,F. (1961) A contribution to the problem of the Drake Passage circulation. Deep-SeaResearch, 8,111-120.
Osrta, oFF, F. (1962) On the frictionally induced transverse circulation of the Antarctic Circumpolar Current.
Deutsche Hydrographische Zeitschrift, 15, 103-113.
O'TOOLE,M.J. (1980) Seasonal distribution of temperature and salinity in the surface water off South West Africa,
1972-1974. InvestigationalReport 121, Sea Fisheries Institute, Department of Agriculture and Fisheries,
South Africa, 25 pp.
Otr, H.W. and W.P.M. DE Rtrtrr~ (1986) Separation of an inertial boundary current from a curved coastline. Journal
of Physical Oceanography, 16,280-289.
PATTERSON,S.L. (1985) Surface circulation and kinetic energy distributions in the southern hemisphere oceans from
FGGE drifting buoys. Journal of Physical Oceanography, 15, 865-884.
PA,'rm<SON,S.L. and H.A. SIEVERS(1980) The Weddell-Scotia Confluence. Journal of Physical Oceanography, I0,
1584-1610.
PEaRCE,A.F. (1977) Some features of the upper 500 m of the Agulhas CtLrrent. JournalofMarine Research, 35, 731-
751.
PEARCg,A.F. and M.L. GRONDt~OH(1982) IS there a seasonal variation in the Agulhas Current? Journal of Marine
Research, 40, 177-184.
P~DLOSKV,J. (1987) An inertial theory of the Equatorial Undercurrent. Journalof Physical Oceanography, 17,1978-
1985.
PEOLOSKY,J. (1988) Entrainment and the termination of the Equatorial Undercurrent. Journal of Physical Ocea-
nography, 18, 880-886.
PEDLOSKY,J. and R.M. SAMELSON(1989) Wind forcing and the zonal structure of the Equatorial Undercurrent. Journal
of Physical Oceanography, 19, 1244-1254.
PEaY_~SON,R.G. (1986) Abyssal Boundary Current Studies deployment cruise. AntarcticJournal of the UnitedStates,
21(5), 125-127.
PL:rmRSON,R.G. (1988a) On the transport of the Antarctic Circumpolar Current through Drake Passage and its relation
to wind. Journal of Geophysical Research, 93, 13993-14004.
Perr~soN, R.G. (1988b) Comparisons of sea level and bottom pressure measurements at Drake Passage. Journal of
Geophysical Research, 93, 12439-12448.
70 R.G. Pm'V_.KSONand L. STRAMMA

Pm'EKSON,R.G. (1990) On the volume transport in the southwestern South Atlantic Ocean. Abstract in: Eos, 71(17),
542.
Pm'v_xsoN, R.G. and T. WHrrwoRTHIZ[ (1989) The Subantarctic and Polar fronts in relation to deep water masses
through the Southwestern Atlantic. Journal of Geophysical Research, 94, 10817-10838.
PETmsoN, R.G., W.D. NOWLIN,JR. and T. Wm'rwoRTHHI (1982) Generation and evolution of a cyclonic ring at Drake
Passage in early 1979. Journal of Physical Oceanography, 12, 712-719.
PmLA~m, S.G.H. (1973) Equatorial Undercurrent: measurements and theories. Reviews of Geophysics and Space
Physics, 11, 513-570.
PHILANDER,S.G.H. (1986) Unusual conditions in the tropical Atlantic Ocean in 1984. Nature,London, 322,236-238.
PHILANDER,S.G.H. and R.C. PACAUOWSra(1984) Simulation of the seasonal cycle in the tropical Atlantic Ocean.
Geophysical Research Letters, 11,802-804.
PmLANDm, S.G.H. and R.C. PACANOWSKI(1986a) A model of the seasonal cycle in the tropical Atlantic Ocean.
Journal of Geophysical Research, 91, 14192-14206.
Pun.Am)m, S.G.H. and R.C. PACANOWSKI( 1986b) The mass and heat budget in a model of the tropical Atlantic Ocean.
Journal of Geophysical Research, 91, 14212-14220.
PIOLA,A.R. and A.A. BIANCm(1990) Geostrophic mass transports at the Brazil/Malvinas Confluence. Abstract in:
Eos, 71(17), 542.
PIOLA,A.R. and D.T. GEOROI(1982) Circumpolar properties of Antarctic Intermediate Water and Subantarctic Mode
Water. Deep-SeaResearch, 29, 687-711.
PIOLA,A.R. and A.L. GORDON(1989) Intermediate waters in the southwest South Atlantic. Deep-SeaResearch, 36,
1-16.
RASMUSSON,M. and J.M. WALLACE(1983) Meteorological aspects of the E1Nifio/Southem Oscillation. Science,222,
1195-1202.
REID,J.L., JR. (1964a) Evidence of a South Equatorial Counter Current in the Atlantic Ocean in July 1963. Nature,
London, 203, 182.
RE~, J.L., JR. (1964b) A transequatorial Atlantic oceanographic section in July 1963 compared with other Atlantic
and Pacific sections. Journal of Geophysical Research, 69, 5205-5215.
REID, J.L. (1989) On the total geostrophic circulation of the South Atlantic Ocean: Flow patterns, tracers and
transports. Progress in Oceanography, 23, 149-244.
REID,J.L. and W.D. NOWLIN,JR. ( 1971) Transport of water through the Drake Passage. Deep-SeaResearch, 18, 51-
64.
REID, J.L., W.D. NOWLIN,JR. and W.C. PATZERT(1977) On the characteristics and circulation of the southwestem
Atlantic Ocean. Journal of Physical Oceanography, 7, 62-91.
RENNELL,J. (1832) An investigation of the currents of the Atlantic Ocean, and of those which prevail between the
Indian Ocean and the Atlantic. J.G.&F. Rivington, London, 299 pp., 2 plates.
REV~Dm, G. and M.J. McPHADEN(1986) Near-surface current and temperature variability observed in the equatorial
Atlantic from drifting buoys. Journal of GeophysicalResearch, 91, 6569-6581.
IOCHARDSON,P.L. and D. WALSH(1986) Mapping climatological seasonal variations of surface currents in the tropical
Atlantic using ship drifts. Journal of Geophysical Research, 91, 10537-10550.
Rn~KEL,M.O., P. SUNDand G. NEUMAt~(1966) The location and termination area of the Equatorial Undercurrent in
the Gulf of Guinea based on observations during Equalant III. Journal of GeophysicalResearch, 71, 3893-
3901.
RINTOUL,S.R. (1990) South Atlantic interbasin basin. Journal of Geophysical Research, (in press).
ROBINSON,A.R. and P.P. NnLER(1967) The theory of flee inertial currents: I. Path and structure. Tellus, 19,269-291.
RODEN,G.I. (1986) Thermohaline fronts and baroclinic flow in the Argentine Basin during the austral spring of 1984.
Journal of Geophysical Research, 91, 5075-5093.
RODEN,G.I. (1989) The vertical thermohaline structure in the Argentine Basin. Journal of Geophysical Research,
94, 877-896.
ROEMMICH,D. (1983) The balance of geostrophic and Ekman transport in the tropical Atlantic Ocean. Journal of
Physical Oceanography, 13, 1534-1539.
SARVnENTO,J.L. (1986) On the North and tropical Atlantic heat balance. Journal of GeophysicalResearch, 91,11677 -
11689.
S e a , L , I.I. (1968) On the relation between the winds off South West Africa and the Benguela Current and Agulhas
Current penetration in the South Atlantic. DeutscheHydrographische Zeitschrift, 21,109-117.
 Upper-level circulation in the South Atlantic Ocean 71

S C H W A , R. (1980) Ergebnisse ozeanologischer Untersuchungen mit dem FFS "Ernst Haeckel" im Konver-
genzgebiet von Brasil- und Falklandstrom im Jtmi/Juli 1978. Beitrdge zur Meereskunde, 44/45, 109-121.
SCHOTr,G. (1902) Oceanographie mad maritime Meteorologie. Wissenschaftliche Ergebnisse derDeutschen Tiefsee-
Expedition Valdivia 1898-1899, 1,403 pp., 26 plates.
Scaorr, G. (1926) Geographie des Atlantischen Ozeans. C. Boysen, Hamburg, 368 pp., 28 plates.
S~VAm, J., J. PIcAtrLTand J. MERLE(1982) Evidence of remote forcing in the equatorial Atlantic Ocean. Journal of
Physical Oeanography, 12, 457-463.
SHANNON,L.V. (1966) Hydrology of the south and west coasts of South Africa. lnvestigational Report 58, Division
of Sea Fisheries, South Africa, 52 pp.
Srt~mor~, L.V. (1985) Description of the Ocean Colour and Upwelling Experiment. In: South African Ocean Colour
and Upwelling Experiment, L.V. SHANNON, editor, Sea Fisheries Research Institute, South Africa, 1-12.
SHANNON,L.V., J.J. AGENBAGand M.E.L. BuYs (1987) Large- and mesoscale features of the Angola-Benguela Front.
South African Journal of Marine Science, 5, 11-34.
SHANNON,L.V., A.J. BoYn, G.B. BRtr~pa'r and J. T A u r c r o N - ~ K (1986) On the existence of an El Nifio-type
phenomenon in the Benguela system. Journal of Marine Research, 44, 495-520.
SrIArmoN, L.V. and M. VANRDSWnCK(1969) Physical oceanography of the Walvis Ridge region, lnvestigational
Report 70, Division of Sea Fisheries, South Africa, 19 pp.
SHANNON,L.V., J.R.E. L ~ S and J.J. AOEI,mAO(1989) Episodic input of subantarctic water into the Benguela
region. Suid-Afrikaanse Tydskrif vir Wetenskap, 85,317-322.
SmDIa~, G. (1983) Tropical and equatorial regions. In: Eddies in Marine Science, A.R. ROBINSON,editor, Springer-
Verlag, 181-199.
SmnL~a~, G., and S.G.H. ~ (1982) Physics of the upper tropical ocean. In: The GARP Atlantic Tropical
Experiment (GATE) Monograph, WMO, Geneva, GARP Publication Series, 25, 219-235.
S~s, H.A. and W.J. EMERy(1978) Variability of the Antarctic Polar Frontal Zone in Drake Passage - summer
1976-1977. Journal of Geophysical Research, 83, 3010-3022.
Smvt~s, H.A. and W.D. Now~m, JR. (1984) The stratification and water masses in Drake Passage. Journal of
Geophysical Research, 89, 10489-10514.
SmNORI~a,S.R. (1978) On the circulation and the volume transport of the Brazil Current between the Cape of S~o
Tom6 and Guanabara Bay. Deep-Sea Research, 25, 481-490.
SMITH,S.G. (1989) On the Weddell-Scotia Confluence and the Scotia Front. M.S. thesis, Texas A&M University,
College Station, 85 pp.
STANDER,G.H. (1964) The Benguela Current off South West Africa. lnvestigational Report 2, Marine Research
Laboratory, South West Africa, 43 pp.
STANDER,G.H. and A.H.B. DE D E ~ (1969) Some physical and biological aspects of an oceanographic anomaly
off South West Africa in 1963. Investigational Report 81, Division of Sea Fisheries, South Africa, 46 pp.
STO~.L, H. (1957) A survey of ocean current theory. Deep-Sea Research, 4, 149-184.
ST~.MMA,L. (1989) The Brazil Current transport south of 23°S. Deep-Sea Research, 36,639-646.
S~,MMA, L. and R.G. Pmea~soN(1989) Geostrophic transport in the Benguela Current region. Journal of Physical
Oceanography, 19, 1440-1448.
SrRaM~, L. and R.G. Prr~-~soN (1990) The South Atlantic Current. Journal of Physical Oceanography, 20, 846-
859.
STV.AMMA,L., Y. IKEDAand R.G. Pmua~soN (1990) Geostrophic transport in the Brazil Current region north of 20°S.
Deep-Sea Research (in press).
STUP.M,M. and K. VOI6T (1966) Observations of the structure of the Equatorial Undercurrent in the Gulf of Guinea
in 1964. Journal of Geophysical Research, 71, 3105-3108.
SVlmDRUV,H.U., M.W. JOHNSONand R.H. Fgmcm~o(1942) The Oceans. Prentice Hall, 1087 pp.
TAFT,B.A. (1963) Distribution of salinity and dissolved oxygen on surfaces of uniform potential specific volume in
the South Atlantic, South Pacific, and Indian oceans. Journal of Marine Research, 21, 129-146 (text), 567-
579 (figures).
TAtnCroN-CI~RK,J. and L.V. SHaNNON( 1988) Annual and interannual variability in the south-east Atlantic during the
20th century. South African Journal of Marine Science, 6, 97-106.
T ~ , P. (1980) DescriptiveRegional Oceanography. Pergamon Marine Series Volume 3, Pergamon Press, Oxford,
253 pp., 19 plates.
TstrcrnarA, M. (1985) Evidence of a double-cell subtropical gyre in the South Atlantic Ocean. Journal of Marine
72 R.G. Pmm~SONand L. STmt~,tA

Research, 43, 57-65.

TsucmYA, M. (1986) Thermostads and circulation in the upper layer of the Atlantic Ocean. Progress in Oceanog-
raphy, 16, 235-267.
VANLOON,H. (1972) Half-yearly oscillations in the Drake Passage. Deep-Sea Research, 19, 525-527.
V~oms, G. (1973) Model of World Ocean circulation: I. Wind-driven,two-layer. Journal of Marine Research, 31,
228-288.
V~oms, G. (1978) Model of World Ocean circulation: III. Thermally and wind driven. Journal of Marine Research,
36, 1-44.
Vtssm, G.A. (1969) Hydrological observations in the south-east Atlantic Ocean. lnvestigational Report 77, Division
of Sea Fisheries, South Africa, 23 pp.
Vomrpa~z, B. (198 I) Le~ sous-courants &luatoriaux nord et sod et la formation des d6mes thermiques tropicaux.
Oceanologica Acta, 4, 497-506.
Vorrtntmz, B and A. ~ t a u ~ (1982) Comparaison des syst~mes productifs de l'Aflantique Tropical Est: dSmes
thermiques, upwellings cStiers et upweUing &luatorial. Rapports et Procds-Verbaux des Rdunions Conseil
International pour l'Exploration de la Met, 180, 114-130.
VOROB'Ym',V.N. and B.V. GmDYSH(1965) Magnitude of mass and heat transport through Drake Passage. Soviet
Antarctic Expedition Information Bulletin, 5, 425-428.
WACONGNE,S. ( 1989) Dynamicalregimes of a fully nonlinearstratified model of the Atlantic Equatorial Undercurrent.
Journal of Geophysical Research, 94, 4801-4815.
WACONG~rE,S. and B. PrroN (1990) Current measurements along the coasts of Gabon and Congo and relevance to the
larger scale circulation. Deep-Sea Research (submitted).
WAaRm~,B.A. (1963) Topographic influences on the path of the Gulf Stream. Tellus, 15, 167-183.
WAtrrI~, B. (1977) R~vision de la classification des eaux de surface du golfe de Guin6e (Berrit 1961). Cahiers
O.R.S.T.O.M., sdrie Ocdanographie, XV, 279-295.
W~,RN, R.B. and D.J. BAKER,JR. (1980) Bottom pressure measurements across the Antarctic Circumpolar Current
and their relation to the wind. Deep-Sea Research, 27, 875-888.
Wmo.Jm, H. (1959) The Antarctic Convergence - or Divergence? In: The Atmosphere and Sea in Motion, Rossby
Memorial Volume, B. BotaN, editor, Rockefeller Institute Press, New York, 107-120.
WmaWORTH,T., Ill (1980) Zonation and geostrophic flow of the Antarctic Circumpolar Current at Drake Passage.
Deep-Sea Research, 27, 497-507.
WarrWORTH,T., III (1983) Monitoring the transport of the Antarctic Circumpolar Current at Drake Passage. Journal
of Physical Oceanography, 13, 2045-2057.
Wrm'WORTH,T., HI and W.D. NOWtJN,JR. (1987) Water masses and currents of the Southern Ocean at the Greenwich
Meridian. Journal of Geophysical Research, 92, 6462-6476.
Wmawoaa'H, T., HI and W.D. Nowtm JR. (1990) The Antarctic Circumpolar Current and deep western boundary
current in the southwest Atlantic. Abstract in: Eos, 71(17), 544-545.
WmaWoR~, T., 111,W.D. Nowtm JR. and S.J. WoRtzr (1982) The net transport of the Antarctic Circumpolar Current
through Drake Passage. Journal of Physical Oceanography, 12, 960-971.
Wm'rwoRr~, T., HI and R.G. l='meaSON(1985) The volume transport of the Antarctic Circumpolar Current from
bottom pressure measurements. Journal of Physical Oceanography, 15, 810-816.
WOOSTER,W.S. and J.L. 17,fao,JR. (1963) Eastern boundary currents. In: The Sea, Volume 2, M.N. I-In~, editor,
Interscience, London, 253-280.
WORLDMETEOROLOGICALORGANIZATION(1963) Fourth World Meteorological Congress, Resolution 35. WMO No.
142, 79-80 and 157-166.
WORLD]~JLEq'EOROLOGICALORGANIZATION (1977) Guide to marine meteorological services. WMO 471, 155 pp.
W0ST,G. (1928) Der Ursprung der atlantischenTiefenwlisser. In: Zeitschrift der Gesellschaftfar Erdkunde zu Berlin,
Sonderband zur Hundertjahrfeier der Gesellschaft, Berlin, 551 pp., 41 plates.
W0ST, G. (1935) Schichtung und Zirkulation des Atlantischen Ozeans. Die Stratosphare. Wissenschaftliche
Ergebnisse der Dentschen Atlantischen Expedition auf dem Forschungs- und Vermessungsschiff "Meteor"
1925-1927, 6: Teil 1, Lieferung 2, 1-288.
WOsr, G. (1957) Stromgeschwindigkeiten und Strommengen in den Tiefen des Atlantischen Ozeans. Wissen-
schaftliche Ergebnisse der Deutschen Atlantisehen Expedition Forschungs- und Vermessungsschiff
Meteor 1925-1927, 6 (2), 261-420.
WYmxL K. (1960) The Antarctic Circumpolar Current and the Antarctic Polar Front. Deutsche Hydrographische
 Upper-level circulation in the South Atlantic Ocean 73

Zeitschrift, 3, 153-173.
WYRaya, K. (1971) Oceanographic Atlas of the International Indian Ocean Expedition. National Science Foun-
dation, Washington, D.C., 531 pp.
WYl~aya,K., L. MAG~JU~and J. I-L~o~ (1976) Eddy energy in the oceans. Journal of GeophysicalResearch, 81,2641 -
2646.
Z~mA, J.C. and M.S. McC~ma-~y (1988) Transport of the Brazil Current: it's bigger than we thought. Abstract in:
Eos, 69(44), 1237.
ZYRYANOV,V.N. and D.N. S~¢~ov (1979) Water circulation in the Falkland - Patagonia region and its seasonal
variation. Oceanology, 19, 518-522.