OCEANOLOGICA ACTA- VOL.

14 - N°6

Canal de Mozambique
Niveau moyen
Vent
Flux méridien
The Mozambique channel Tourbillon anticyclonique

revisited Mozambique Channel

Mean sea leve!
Wind
Meridian flow
Andcyclonic gyre

Jean-René DONGUY, Bernard PITON

Centre ORSTOM de Brest, B.P. 70, 29280 Plouzané, France.

Received 10/06/91, in revised form 15/10/91, accepted 16/10/91.

ABSTRACT The meridian flow through the Mozambique Channel is investigated by the diffe-
rence between the sea level on each side of the channel and by the influence of the
wind components measured at two low islands. From these data, it is inferred that
the resulting variations in meridian flows in the northern and southern
Mozambique Channel are roughly in phase and that the meridian transport is sub-
jected to a seasonal cycle. Moreover, three unpublished croises (1974-1975) in the
Mozambique Channel show an anticyclonic gyre in the northern part {12°-16°S)
and a large south ward flow across the narrows of the channel. According to the
results of the croises, this northern anticyclonic gyre is submitted to seasonal and
also year-to-year variations.
Oceanologica Acta, 1991. 14, 6, 549-558.

RÉSUMÉ Le canal de Mozambique revisité

Le flux méridien à travers le Canal de Mozambique est étudié en utilisant la diffé-

rence entre les niveaux moyens des deux côtés du Canal et en estimant l'influence
des composantes du vent mesuré sur deux îles basses situées dans le Canal. Les
flux méridiens dans le nord et le sud du Canal de Mozambique sont à peu près en
phase et le transport méridien est soumis à un cycle saisonnier. Trois croisières
inédites (1974-1975) dans le Canal de Mozambique révèlent dans la partie nord
{12°-16°S) un tourbillon anticyclonique soumis à des variations saisonnières et
interannuelles et un fort flux portant au sud à l'endroit le plus étroit du Canal.
Oceanologica Acta, 1991. 14, 6, 549-558.

INTRODUCTION Indonesian straits, lndian Ocean in the 10-15° S latitude

belt, southward flow in the Mozambique Channel (Fig. 1 a)
According to Clowes and Deacon (1935), the North and entry into the South Atlantic by a branch of the
Atlantic Deep Water is exported to the lndian and Pacifie Agulhas current. This hypothesis relies on the existence of
Ocean by the Antarctic Circumpolar Current. The compen- a southward surface current between Madagascar and
sating flow would come within the warm water in the ther- Africa. This current, usually called Mozambique current, is
mocline layer (Gordon, 1986). The itinerary of such a indicated on Figure 1 a from Saetre (1985). However,
water would be: Pacifie to Indian Ocean through the reliable evidence of this current is not compelling and the

0399-1784/91/06 549 10/$ 3.00/© Gauthier-Villars 549

J.-R. DONGUY, B. PITON

30°E 35° 3QOE

50'1------L--~~~~-i----~~----L-----+ 50~-----L---,-L-----i----~~----+

s (a) s (b)

200

..
G

)/
. ~/(

:((~'
Sç'
9~~::_\
Figure 1

Suiface circulation in the Mozambique Channel: a) historical scheme from Saetre (1985); b) scheme submitted by Soares (1975 ).
Meteorological and tide gauge stations are' located on this map: G, Glorieuses Islands; JN, Juan da Nova; E, Europa; MO,
Moçambique Island; NB, Nosy-Bé; MA, Maputo; T, Tulear; c) scheme submitted by Saetre and Da Silva (1984) during the southern
summer; d) same as c) except during the southern winter.

Circulation de surface dans le canal de Mozambique: a) schéma historique, d'après Saetre (1985); b) schéma proposé par Soares
(1975). Position des stations météorologiques et des marégraphes : G, Iles Glorieuses; JN, Juan da Nova ; E, Europa; MO, Ile
Moçambique; NB, Nosy-Bé; MA, Maputo; T, Tuléar; c) schéma proposé par Saetre et Da Silva (1984) en été austral; d) même
schéma que c) mais en hiver austral.

southward flow through the Mozambique Channel seems Table 1

more complicated than this schematic pattern, as illustrated
Location of the tide gauge stations in latitude and longitude and the
on Figures 1 b, 1 c and 1 d from Soares (1975) and Saetre period of records available.
and Da Silva ( 1984). Our intention here is to provide new
evidence of this flow and its seasonal variability. Position des marégraphes en latitude et longitude et période
d'enregistrement disponible.

DATA Latitude Longitude · Record time

Nosy-Bé 13° 24 s 48° 18 E Jan.1958-June 1967

The meridional flow through the entire channel can be Moçambique
investigated by the difference between the sea level on Island W02S 40°44E Jan. 1963-Dec. 1964
each side of the Mozambique Channel. To illustrate the Tulear 23° 22 s 43° 40E March 1963-Dec. 1967
variability of this flow, we shall rely on sea-level records Maputo 25° 59 s 32° 34E Jan. 1961-Dec. 1967
by tide gauges kindly provided by EPSHOM (Brest) and

550
 THE MOZAMBIQUE CHANNEL REVISITED

Table 2

Sequence of events connected to the monsoon in the North and in the South Mozambique Channel.

Séquence des événements rattachés à la mousson au nord et au sud du canal de Mozambique.

NORTH MOZAMBIQUE CHANNEL

M A M J J A s 0 N D J F
SW monsoon NEmonsoon

Sea leve! 1 Madagascar Low High

Africa Low High
Oyre Anticyclonic Anticyclonic
Flow anomal y Northward Southward

SOUTH MOZAMBIQUE CHANNEL

M A M J J A s 0 N D J F
SWmonsoon NEmonsoon

Sea leve! 1 Madagascar Low High

Africa Low High
Oyre weak changes
Flow anomaly North ward South ward

PSMSL (Birkenhead), at four locations in the Mozambique French oceanographie cruises were numerous in the
Channel (Nosy-Bé and Tulear in the East, Moçambique Mozambique Channel from 1965 to 1975, due to the pre-
Island arid Maputo in the West; Tab. l; locations on Fig. 1 sence of the RN Vauban at the Centre ORSTOM de Nosy-
b) in 1963 and 1964. This data set is representative of coas- Bé (Madagascar). This vesse! carried out severa1 cruises in
tai areas (Fig. 2 and Tab. 2) on both sides of the the area not only with classic hydrographie measurements
Mozambique Channel. but also with direct surface current measurements using a
Geomagnetic Electro Kinetograph (GEK), the accuracy of
The wind was recorded between 1970 and 1975 at two 1ow which has been estimated at± 0.04 rn s-l in velocity and at
islands located in the northem and southem Mozambique ± 8° in direction (Piton and Poulain, 1974). As the measu-
Channel: Glorieuses (11 °30 S, 47°22 E), Europa (22°21 S, rements reached only 550 rn depth, the geostrophic flow
40°21 E). The meteorological stations located on these . has been calculated relatively to 500 dbar. It is known that
islands are operated by the French Meteorological Service ; the near surface circulation extends deeper (Lutjeharms,
measurements are made every three hours. Wind observa- 1976) and also that the seasona1 variabi1ity does penetrate
tions at Juan de Nova located in the centre of the channel deeper (Donguy and Piton, 1969). The scarce distribution
(17°03' S, 42°43' E) started only in 1974. of the hydrographie stations is not able to resolve all the

Cbl 1963 1964

C•J 1963 1964 J FMAMJ JASONDiJ FMAMJJ A SONO
J FMAM J jAsoN oiJ FMÀMJ JAS OND

+20 cm
Figure2

1963-1964 sea leve/ and· se'a leve/

difference (in cm): a) at Nosy~Bé (up)
and Moçambique Island (down); (b) at
Tulear (up) and Maputo (down). ln the
lower case, sea leve/ difference and
direction of the inferredf/ow.

Niveau moyen et différence de niveau

moyen en 1963-1964 :·a) à Nosy-Bé et
à l'île Moçambique ; b) à Tuléar et à
Maputo. En bas, différence de niveau
moyen et direction du flux.

551
J.-R. DONGUY, B. PITON

scales of the circulation within the channel. However, the tions are obvious in sea level at both stations. These are
qualitative agreement between the geostrophic estimate partly related to the sea surface temperature variations
and the absolute surface current from GEK measurements associated with the seasonally varying air-sea exchanges
suggests that the essence of the near-surface variability in (Colborn, 1975). The wind field also presents an annual
the dynamic height field has been captured. variability for both meridian and zonal components.
Piton ( 1989) has presented the hydrographie sections (0-600 Hastenrath and Lamb ( 1979) do not report zonal change of
rn) and the current measurements as weil as the wind recor- the air-sea flux of beat in the channel but the space resolu-
ded at three meteorological stations operated on low islands tion of the data in the atlas is not sufficient to ascertain that
of the Mozambique Channel by the French Meteorological the only cause of the variability of the zonal sea leve! slope
Service: Glorieuses (11 o 30' S, 47° 22' E), Juan de Nova is the intensity of meridian flow.
(17° 03' S, 42° 43' E) and Europa (22° 21' S, 40° 21' E)
(locations on Fig. 1 b ). Three emises will be considered : SWmonsoon
two during the warm season (Mozambique emise Febmary-
March 1974 and Juan de Nova 4 emise , March 1975) and In 1963 and 1964 along the coast of Madagascar (Nosy Bé),
one during the cool season (Europa emise, June-July 1974). the sea leve! is low in this season and rather high at the
Both geostrophic currents and absolu te surface current mea- African coast (Moçambique Island), particularly during
surements from GEK were collected. September-October 1963 (Fig. 2 a). The sea leve! difference
between Nosy Bé and Moçambique Island indicates a north-
ward surface flow relative to the mean which is probably
MERIDIAN CIRCULATION IN THE MOZAMBIQUE located close to Madagascar under the form of a coastal flow
CHANNEL associated with upwelling. Off the African coast, the flow is
usually southward (Donguy and Piton, 1969); high sea leve!
is an index of weak current. The wind at Glorieuses Island
Conditions in the northern Mozambique Channel
(Fig. 3) has a1so a northward component which is consistent
with the flow and also with the coastal current off
The conditions in the northem Mozambique Channel (north Madagascar. A weak south ward current in the west of the
of l7°S, where the channel is narrowest) are frrst considered: Mozambique Channel and a coastal northward flow in the
Sea leve! data at Nosy Bé and Moçambique Island (Fig. 2 east infer the existence of an anticyclonic eddy. Harris ( 1972)
a) and the wind field at Glorieuses Island (north of indicates that an anticyclonic eddy was indeed present in
Mozambique Channel; Fig. 3) are compared. Annual varia- September-October 1964.

N CrTlfs N S GLORIEUSES •---..

400
NS EUROPA
·- N
3oo~Gion.euses
150
N-S cm;s
.

0
1
-1~

-350
s '--...._,_~,..._,.,....._,,--~--,.,...
N 707172737475

&eo~Europo
N·S cm;1
340 1

-340
,, li
'

-200
(al -680

Scm;5 ~~~~~~~~~~~~~~~~~~~~~~
JAN FEB MAR APR MAY JUN JUL ALG SEP OCT NOV DEC
s ~72,-L-,-73,--L-7-4'-7,--15
..,7,.,.0-'-=71
Figure 3

Evolution at Glorieuses and

E cm;5
E W GLORIEUSES •---• Europa Islands of the: a)
200 /\ • EW EUROPA -
E meridian; and b) zonal
1
1 1
1
380 WGiorieuses
E-W cm;8 compone nt of the wind, two
1 1
1 1 190 1 weeks data averaged, from
100 ) \_.., 0
1 1 1970-/976 measurements.
0 / ' 1 Right: anomalies relative to
-190
'\ average.
-lOO
1
,. -:580

\ 1 w '-=-'-::,-L-,-:-'-=,--L-~,--1 Évolution aux Iles Glorieuses

0. \ ;)~-------·~"! .. . . . / E 701172737475
-200 ~.P 420 ~Europo
E·W cny1
et Europa de la composante
du vent méridienne (a) et
-300 v . . . . .____,
1o
•
/
1
'---• - \'.(
~
,
\
-..

/
rt'
1
210

0
'
zonale (b), par moyenne de
deux semaines, à partir de

·-
mesures faites de 1970 à
-400 '•-"'
1 /
,/ -•1a l11
1976. A droite : anomalies
(b) \,.,,./ w par rapport à la moyenne.
Wcmn~~~~~~~~~L-~~~~~~~-=L-~--~
70717271747&
JAN FEB MAR APR MAY JUN JUL ALN SEP OCT NOV DEC

552
 THE MOZAMBIQUE CHANNEL REVISITED

NEmonsoon (Sv, 106 m3 s-1) flow relative to 600 dbar found by

Nehring et al. (1984).
Along the coast, in the northem Mozambique Channel, the
sea level (Fig. 2 a) is low at the coast of Africa where In the south of the channel, at Europa Island, SE trade
upwelling occurs ; it is high at the coast of Madagascar. winds are prevailing throughout the year without apparent
The sea level difference between Nosy Bé and influence on the seasonal variations of the resulting flow.
Moçambique Island indicates a maximum southward flow In conclusion, it seems that the regime of the surface cir-
which, at this season, is mostly concentrated along the culation in the northem Mozambique Channel is as fol-
African continental shelf, according to Nehring et al. lows (Tab. 2):
(1984). Off the coast of Madagascar, the flow is usually
·from March to November, i.e. from the end of the NE mon-
northward (Donguy and Piton, 1969) and weak as the sea
level is high. A strong south ward flow in the west of the soon to the end of the SW monsoon, an anticyclonic gyre
Mozambique Channel and a weak northward flow in the prevails. The zonally averaged flow anomaly is northward;
east infer the existence of an anticyclonic eddy • from November to March, an anticyclonic gyre prevails
(Lutjeharms, 1976). also, but the resulting flow anomaly is southward.
Col born ( 1975), by analy sis of the thermal structure in his- In the southem Mozambique Channel, the flow anomaly is
torical data of the Mozambique Channel, shows that the northward from March to August and southward from
seasonal changes of the thermal structure are consistent August to February.
with the sea leve! observed at Moçambi.que Island (west of
It is important to note that the resulting flows in the nor-
the Channel): a shallow thermocline occurs during the NE
them and southem Mozambique Channel are roughly in
monsoon at the time of low sea level. Moreover, the
phase, with a net seasonal cycle of the meridian transport
monthly mean topography of the 20° C isotherm in the nor-
through the channel. These features are to be compared
them Mozambique Channel (Quadfasel, 1982) is consistent
with the circulation observed during severa! unpublished
with Colbom's analysis. On the other hand, Donguy and
cruises in the Mozambique Channel from classical hydro-
Piton ( 1969) present for Nos y Bé (east of the channel) a
graphie measurements and direct absolu te current measure-
different seasonal cycle of temperature with the lowest
ments with GEK.
thermocline during the NE monsoon.
Based on the wind field by Cadet and Dielh (1984 ),
Ekman pumping has been calculated for the same period UNPUBLISHED CRUISES IN THE MOZAMBIQUE
as the sea level record ( 1963-1964) at locations off the
CHANNEL
sea level gauges at Nosy Bé and Moçambique Island, but
the values are not consistent with the sea level seasonal
Mozambique cruise (22 February-28 March 1974; Fig. 4)
cycle. This discrepancy may be explained by the diffe-
rences of the conditions induced by boundary currents
and those induced by the gyre-scale circulation in the In the northem part of the Mozambique Channel, an anti-
open ocean.We now consider the conditions of the sou- cyclonic eddy clea~ly appears, as already suggested by
them Mozambique Channel: Donguy and Piton (1969), Lutjeharms (1976) and confir-
med by Nehring et al. (1984). This eddy is fed in the north
by the westward flowing South Equatorial Current. The
Conditions in the southern Mozambique Channel South Equatorial Current accelerates as it passes the Cap
d'Ambre, north of Madagascar, reaching a velocity of 1.5
SWmonsoon rn s-1 measured both by GEK and calculated from dyna-
mic heights and transporting 23 Sv from the surface to 300
During the greater part of the SW monsoon, the sea lev el is rn depth [ with a reference at 500 dbar; Piton, 1989
low on each side of the Mozambique Channel (Tulear and (Swallow et al., 1988 found 7 to 18 Sv with a reference
Maputo; Fig. 2 b) and also in the centre according to the leve! at 1100 dbar)]. At 12° S, the velocity decreases from
isotherm depths (Colbom, 1975). So, the change in the 1.3 rn s-1 at 45° E to 0.5 rn s- 1 in the vicinity of Africa and
anticyclonic component of the circulation is not obvious. the current separates into two coastal branches, one flo-
However, the sea leve! difference between Tulear and wing northwestward and carrying 11 Sv up to 300 rn
Maputo indicates first a surface north ward flow between depth, the other called the Mozambique Current south-
March and August and then a southward flow relative to ward and carrying 15 Sv. At 17° S, the current crosses the
the mean until the end of January. channel reaching 2 rn s- 1 according to GEK measurements
with a transport of 19.4 Sv according to geostrophic esti-
NEmonsoon mate referred at 500 dbar. At the Madagascar shelf, the
current again separates into two branches : a southeast-
During the NE monsoon, the sea level is high on each ward one (16 Sv) and a northward one (9 Sv), this latter as
side of the Mozambique Channel (Tulear and Maputo; a part of the anticyclonic eddy extending over the northem
Fig. 2 b) and also in the centre according to the isotherm Mozambique Channel. Due to the lack of measurements
depth (Colbom, 1975). The sea leve! difference between between Madagascar and Comoro Islands it is not possible
Tulear and Maputo indicates a southward flow relative to to estimate how much flow from the anticyclonic eddy
the mean in accordance with a southward 15 sverdrup recirculates into the South Equatorial Current.

553
J.-R. DONGUY, B. PITON

Figure 4 40"E 45"E

Cruise Mozambique (22 February-28

March 1974 ). Surface circulation
inferred from surface dynamic heights lOOS
relative to 500 decibars and from
GEK measurements. Note that the
current velocity scale for GEK
measurements is different from
Figures 5 and 6.

Croisière Mozambique (22 février-28

mars 1974). Circulation de surface
d'après les hauteurs dynamiques
relatives à 500 décibars et d'après les
mesures de GEK. Noter que l'échelle
AFRICA
, ... - - ..... ,Il '
1 150 \1
des vitesses mesurées au GEK est 1> \ 1
différente de celle des figures 5 et 6. 1
1
'
11
1 1 1

MADAGASCAR

MOZAMBIQUE
CHANNEL

zo•s

The salient feature of this cruise is the strong

flow crossing the Mozambique Channel
through the narrows between 15° and 17°S.
This feature is consistent with the wind regi-
me prevailing during the cruise. In the north
of the channel (Iles Glorieuses) the wind was
mainly ENE. In the centre (Ile Juan de
Nova), the wind was variable: first NE, then
NW and finally ESE. In the south (Ile
Europa), easterly wind prevailed. The wind
conditions in the channel were rather favou-
rable to a south ward Ekman transport of the
surface waters, but the velocity of the wind
was small (2 rn s· 1) and consequently Ekman
transport calculated from the wind measured
at Juan de Nova Island was weak (0.5 Sv).

Juan de Nova cruise (7-21 March 1975;

Fig. 5)
Figure 5
Cruise Juan de Nova IV (7-21 March 1975). Surface circulation inferredfrom surface dynamicAn anticyclonic eddy centred on 16° 30 Sis
heights relative to 500 decibars and from GEK measurements. present as it was in 1974 during the
Croisière Juan de Nova IV (7-21 mars 1975). Circulation de surface d'après les hauteurs~ozambique croise but with a smaller velo-
dynamiques relatives à 500 décibars et d'après les mesures de GEK. City (al ways less than 1 rn s·l ). South of

554
 THE MOZAMBIQUE CHANNEL REVISITED

around 18° S, the current becomes southwestward. The 0- circulation weaker in February-March 1975 than in
300 rn flow is also smaller than in 1974: the southward February-March 1974.
flow of the Mozambique Current is only 5.7 Sv at 15° S,
whereas the maximum flow through the channel is 9.4 Sv Europa cruise (17 June-13 Joly 1974; Fig. 6)
(17°S), but only 4.7 Sv are then going southward (18° S),
and the remaining flow north ward. In contrast with the pre- In the central part of the channel, a southeastward current
vions year, a low atmospheric pressure (less than 1 010 (0.8 rn s-1) is again present with a 0-300 rn flow carrying
mbar) prevailed over the channel during the earl y part of 14 Sv at 17° S, feeding a 5.7 Sv northward flow. At 18° S,
1975, associated with tropical storms or cyclones. In the the current is southwestward with a flow of 7.3 Sv. South
north of the channel (Iles Glorieuses) the wind was weak of 18° S, several convergent and divergent meanders are
and variable. In the centre (Ile Juan de Nova) the wind was observed, centred at 20°-21° S, at 24° Sand at about 26°-
SSW, whereas in the south (Ile Europa) a SE wind prevai- 300 S. The meteorological conditions during the Europa
1ed. A southward transport of the surface waters was not cruise are characteristic of the southern win ter: high atmos-
favoured by such heterogeneous winds. lt is interesting to pheric pressure, reaching as much as 1 017 rnbar in the
note that this unsettled wind regime was associated with a north and 1 022 mbar in the south. The wind, during the

A FR ICA

MADAGASCAR

1 m/s

Figure6

Cruise Europa (17 June-13 July 1974). Surface

circulation inferred from surface dynamic
heights relative to 500 decibars and from GEK
measurements.

Croisière Europa. Circulation de surface d'aprè'

les hauteurs dynamiques relatives à 500
décibars et d'après les mesures de GEK.

555
J.-R. DONGUY, B. PITON

croise, is related to the South Indian anticyclone. In the speed reaching 2 rn s·1, inducing a 0-300 rn transport of the
north of the channel (Iles Glorieuses), the wind is ESE, in order of 20 Sv. At 17-18°S, off the Madagascar continental
the centre (Ile Juan de Nova) and in the south (Ile Europa), shelf, the flow is divided into two branches. The northward
it isSE. one constitutes the eastern part of an anticyclonic gyre of
In summary, considering the results of the three croises car- variable size capable of extending into the entire northern
ried out by RN Vauban, it is possible to compare the south- Mozambique Channel, whereas the southern one consti-
ward flow from the surface to 300 rn depth at three loca- · tutes mainly the southward flow connecting the South
tions (Tab. 3) : Equatorial Current to the Agulhas Current through a sys-
tem including severa! southward meanders. The northern
• at 15°S along the African coast in the northern eddy gyre is submitted to seasonal and interannual variabilities
(Mozambique Current); which are mostly due to the wind. lt has been observed by
• at l7°S in the "narrows" of the Channel; severa! authors (Donguy and Piton, 1969; Duncan, 1970;
Harris, 1972; Soares, 1975; Lutjeharms, 1976;
·at !8°S along Madagascar in the southern eddy. Parfenovich, 1980; Piton et al., 1981) following the synthe-
Table 3
sis made by Saetre and Da Silva (1984). At !8°S, the sou-
Geostrophic flow in sverdrup relative to 500 decibars at three latitudes thward transport may reach 16 Sv but the mean transport in
in the Mozambique Channel during each cruise, the flow being the southern part of the channel is close to 10 Sv (Piton,
southward at l5°S, directed to the southeast at ll"S and westward at
l8°S.
1989), as suspected by Gordon (1986). Harris (1972)
considers a transport of 10 Sv through the narrows as
Flux géostrophique en sverdrups relatif à 500 décibars calculé à trois representative.
latitudes dans le canal de Mozambique lors de chaque croisière, le flux
portant au Sud à 15° S, au Sud-Est à 17° Set à l'Ouest à 18° S. However, seasonal and interannual variations in sea leve!
on each side of the northern Mozambique Channel (Fig. 2)
must induce changes in the southward flow. According to
Cruises 15° s 17° s 18° s
Swallow et al. ( 1988), in the South Equatorial Current, the
s SE w
largest depth responding to seasonal variation off Cap
Mozambique 15 19.4 16.6 d'Ambre is 200 m. For a 20 cm variation in sea leve!, the
Juan de Nova 5.7 9.4 4.7 calculation shows that the resulting variation in the south-
Europa 4.7 14 7.3 ward flow of the Mozambique Current from the surface to
200 rn depth at 14°-15°S is 10 Sv. This result is consistent
with the measurements noticed in Table 3: the southward
flow measured at !5°S varies from 4.7 to 15 Sv.
The main features and <fssociated questions are:
-The large (15-20 Sv) southward flow of the Mozambique
Current in the western boundary at 15° S, mainly during
the Mozambique croise. On the other hand, during a croise
of RN A. von Humboldt, also in Febroary-March (1980)
and at 15° S, the transport of the Mozambique Current was
larger than 15 Sv.
- The large southeastward flow across the narrows of the
channel at 17° S. This flow seems al ways greater than the
southward flow along the African coast at 15° S. This
raises the question whcther a recirculation occurs or whe-
ther the 500 dbar reference leve! is not appropriate near the
coast.
- The strong variability of the flow across the narrows of
the channel. This variability seems to be connected with
the wind and pressure conditions.

DISCUSSION

The most characteristic features existing during the three

croises studied are an anticyclonic eddy in the northern ••• dlveroence
xxx convergence
channel and a weil established current flowing south-east-
ward in the central part of the Mozambique Channel (Fig. Figure 7
7). The origin of this current is the south part of the South
Equatorial Current which diverges off the Cap Delgado at Possible surface circulation scheme in the Mozambique Channel.
11 °S. The south ward branch, called the Mozambique Schéma possible de la circulation de surface dans le canal de
Current, flows along Africa, and crosses the channel with a · Mozambique.

556
 THE MOZAMBIQUE CHANNEL REVISITED

Circulation and hydrographie conditions are mostly depen-

dent on the thermohaline conditions. An important beat
content is stored seasonally as indicated by the high sea
level observed simultaneously at Maputo and Tulear (Tab.
2) and also in the center due to the great depth of the iso-
therms (Colbom, 1975).
Hydrographical observations resulting from the second
croise of RN Cdt Robert Giraud (Soares, 1975) during the
cold season (July-September 1960) and from a emise of
RN A. von Humboldt during the warm season (February-
March 1980; Nehring et al., 1984) show a common tran-
sect close to 39°E between 23°S and 26°S (Fig. 8). There
are large differences in the surface temperatures. The depth
m
of the 20°C isotherm is the same at 23°S in both seasons
100
but considerably deeper at 26°S during Febmary-March.
200 Moreover, there is a layer of waters warmer than 25°C (50
rn in Febmary-March) which is not present in July-
300
September (cold season). This greater beat content during
400 the warm season is also consistent with the high sea level
500 observed at the same time. On the other band, slowly
meandering surface waters are warmed and reach a tempe-
600 rature 1°C greater there than east of Madagascar (Soares,
T°C
1975). These features are associated with the seasonal pre-
sence of the Intertropical Convergence Zone of the wind
Figure 8
and its associated convective activity, which must result in
Vertical distribution of temperature approximately along 39°E a net heating of the ocean.
meridian, 23°-26°5, in February-March 1980 (top) and July-September
1960 (bottom).

Distribution verticale de la température située approximativement le

long du méridien 39° E, de 23 à 26° S, en février-mars 1980 (haut) et
CONCLUSION
juillet-septembre 1960 (bas).
Continuous sea leve! data from four tide gauges on each
In the central and southern Mozambique Channel, the side of the Mozambique Channel and wind data at two
available emises covering the entire area have been car- islands provide information on the variations in meridian
ried out only during the SW monsoon (the cold season; transport through the channel. In addition, data from three
Cdt Robert Giraud 1, 2, 4 and Almirante Lacerda). unpublished croises give more details on the surface circu-
However, other emises mainly devoted to biological lation, mainly in the northem part of the channel where an
research have covered parts of the channel during the NE anticyclonic gyre occurs. This gyre is monitored by wind,
monsoon and the warm season, particularly the African whereas in the southem part of the channel, the circulation
continental shelf. From this data set, Saetre and Da Silva is mostly thermohaline. The meridian flow is submitted to
(1984) show that another anticyclonic gyre is located in a seasonal cycle but the existence of a resulting southward
the central Mozambique Channel with a centre approxima- transport is not obvious.
tel y at 20°S, i. e. consistent with the circulation scheme
from geostrophic and absolute GEK measurements during
Aknowledgements
the Europa emise. In the same way, near the coast of
Madagascar, the strong southward flow observed south of
23°S during the Europa emise is in good agreement with The authors are grateful to Daniel Cadet for kindly pro-
the results of the other emises. viding the wind data, to EPSHOM (Brest) and PSMSL
During the NE monsoon (the warm season), it seems that . (Birkenhead) for providing sea level data in the
the variability of the circulation characterized by a south- Mozambique Channel and to Gilles Reverdin for his
ward flow is not influenced by wind (SE tracte winds). precious advice.

REFERENCES

Cadet D.L. and B.C. Dielh (1984). lnterannual variability of surface Colborn J.G. (1975). The thermal structure of the lndian Ocean. The
fields over the Indian Ocean during recent decades. Mon. Weath. Rev., University Press of Hawaï, Honolulu.
112, 1921-1935. Donguy J.-R. and B. Piton (1969). Aperçu des conditions hydrolo-
Clowes A.J. and G.E.R. Deacon (1935). The deep water circulation giques de la partie nord du Canal de Mozambique. Cah. ORSTOM,
of the lndian Ocean. Nature, 136, 936-938. Sér. Océanogr., 3, 2, 2-26.

. 557
J.-R. DONGUY, B. PITON

Duncan C.P. (1970). The Agulhas Current. Ph. D. Dissertation, Piton B. and Poulain J.-F. (1974). Résultats des mesures de courants
University of Hawaii, 76 pp. superficiels effectués avec le N.O. Vauban dans le sud-ouest de
Gordon A.L. (1986). Interocean exchange of thermocline water. J. l'Océan Indien (1973-1974). Documents scientifiques, Mission ORS-
geophys. Res., 91, C4, 5037-5046. TOM de Nosy-Bé, n° 47,72 pp.
Harris T.F.W. (1972). Sources of the Agulhas Current in the spring Piton B., J.H. Pointeau and J.S. Ngoumbi (1981). Atlas hydrolo-
of 1964. Deep-Sea Res., 19, 633-650. gique du Canal de Mozambique (Océan Indien). Trav. Doc.
Hastenrath S. and P.J. Lamb (1979). Climatic Atlas of the lndian ORSTOM, 132,41 pp.
Ocean. University of Wisconsin Press, Madison, USA. Part 2. Quadfasel D.R. (1982). Low frequency variability of the 20° C iso-
Lutjeharms J.R.E. (1976). The Agulhas Current system during the therm topography in the western Equatorial Indian Ocean. J. geophys.
Northeast Monsoon season. J. phys. Oceanogr., 6, 655-670. Res., 87, C3, 1990-1996.
Nebring D., G. Art, G. Bu blitz, L. Gohs, F. Gosselck, E. Hagen, W. Saetre R. ( 1985). Surface currents in the Mozambique Channel.
Kaiser, E. Kijhner, N. Michelchen, L. Postel, R. Saestre, R. Deep-Sea Res., 32, 12, 1457-1467.
Schemainda, H. Siegel, P. Silva and G. Wolf (1984). The oceanolo-
gical conditions in the western part of the Mozambique Channel in Saetre R. and A.J. Da Silva (1984). The circulation of the
February-March 1980. Geodiit. geophys. Veroff., 4, 39, 163 pp. Mozambique Channel. Deep-Sea Res., 31, 5, 485-508.

Parfenovich S.S. (1980). Hydrological features of productivity in the Soares G. (1975). Contribution à l'étude de l'hydrologie et de la cir-
southwest part of the Indian Ocean. Trudy vses. nauchno-issled. lnst. culation du Canal de Mozambique en hiver austral. Thèse de Docto-
morsk. ryb. khoz. okeanogr. (VNIR0),145, 6-18 (in Russian). rat Jème cycle, Université de Paris VI, 89 pp.
Piton B. (1989). Quelques nouveaux aspects sur la circulation super- Swallow J., M. Fieux and F. Schott (1988). The boundary currents
ficielle dans le Canal de Mozambique (Océan Indien). Documents East and North of Madagascar. 1: Geostrophic Currents and
scientifiques, ORSTOM Brest, n° 54, 31 pp. Transports. J. geophys. Res., 93, CS, 4951-4962.

558