JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 91, NO. C4, PAGES 5037-5046, APRIL 15, 1986
Interocean Exchange of Thermocline Water
ARNOLD L. GORDON
Lamont-Doherty Geolo•Tical Observatory of Columbia University, Palisades, New York
Formation of North Atlantic Deep Water (NADW) represents a transfer of upper layer water to
abyssal depths at a rate of 15 to 20 x 10 6 m3/s. NADW spreads throughout the Atlantic Ocean and is
exported to the Indian and Pacific Oceans by the Antarctic Circumpolar Current and deep western
boundary currents. Naturally, there must be a compensating flow of upper layer water toward the
northern North Atlantic to feed NADW production. It is proposed that this return flow is accomplished
primarily within the ocean's warm water thermocline layer. In this way the main thermoclines of the
ocean are linked as they participate in a thermohaline-driven global scale circulation cell associated with
NADW formation. The path of the return flow of warm water is as follows: Pacific to Indian flow within
the Indonesian Seas, advection across the Indian Ocean in the 10ø-15øS latitude belt, southward transfer
in the Mozambique Channel, entry into the South Atlantic by a branch of the Agulhas Current that does
not complete the retroflection pattern, northward advection within the subtropical gyre of the South
Atlantic (which on balance with the southward flux of colder North Atlantic Deep Water supports the
northward oceanic heat flux characteristic of the South Atlantic), and cross-equatorial flow into the
western North Atlantic. The magnitude of the return flow increases along its path as more NADW is
incorporated into the upper layer of the ocean. Additionally, the water mass characteristics of the return
flow are gradually altered by regional ocean-atmosphere interaction and mixing processes. Within the
Indonesian seas there is evidence of strong vertical mixing across the thermocline. The cold water route,
Pacific to Atlantic transport of Subantarctic water within the Drake Passage, is of secondary importance,
amounting to perhaps 25% of the warm water route transport. The continuity or vigor of the warm
water route is vulnerable to change not only as the thermohaline forcing in the northern North Atlantic
varies but also as the larger-scale wind-driven criculation factors vary. The interocean links within the
Indonesian seas and at the Agulhas retroflection may be particularly responsive to such variability.
Changes in the warn: water route continuity may in turn influence formation characteristics of NADW.
INTRODUCTION
Warm salty water spreads into the northern North Atlantic,
where it is cooled primarily by evaporation. Ironically, this is
a consequence of its anomalously high temperature relative to
the atmosphere [Warren, 1983]. This in turn maintains rela-
tively high salinity and density despite an abundance of pre-
cipitation. The cooled salty water sinks to the deep ocean,
marking the formation of North Atlantic Deep Water
(NADW) [Warren, 1981; Killworth, 1983]. The NADW from
the two northern sites (Labrador Sea and the Greenland Sea-
Norwegian Sea overflow), with a mean temperature and salini-
ty of approximately 2øC and 34.93%0 (the northern component
defined by Broecker et al., [1976] and Broecker and Peng
[1982]), spreads to the south within the deep western bound-
ary current, being joined by the saltier outflow from the Medi-
terranean Sea. The NADW water mass influences most of the
global ocean [Reid and Lynn, 1971]. Warren [1981], reviewing
the estimates of NADW formation rate, arrives at a number of
14 Sv (1 Sv = 106 m3/s). Broecker [1979], using radiocarbon
data, suggests a formation rate of greater than 20 Sv. The two
northern components account for over 90% of the NADW
volume flux.
The process of NADW formation is self-perpetuating in
that as the surface layer water sinks and is exported south-
ward within the deep layer, more upper layer water is drawn
into the northern North Atlantic. This in turn drives the high
evaporation rates continuing the NADW formation process.
Whether a random event initialized the circulation or whether
it is a response to the changing distribution of the land masses
and associated large-scale circulation remains a tantalizing
problem in oceanography. The thermohaline circulation pat-
Copyright 1986 by the American Geophysical Union.
Paper number 6C0064.
0148-0227/86/006C-0064505.00
tern in the meridional plane associated with the NADW for-
mation is one of a negative estuary [Stommel, 1956; Reid,
1961; Worthington, 1981; Gordon and Piola, 1983]: upper
layer water moves to the north, while deeper water moves to
the south. This pattern is clearly seen in the results of the
inverse solutions for the Atlantic Ocean using the Internation-
al Geophysical Year (IGY) set of zonal hydrographic sections
[Roeromich, 1980, 1983; Fu, 1981; Wunsch, 1984; Roeromich
and Wunsch, 1985]. Comparison of the IGY data with recent
data sets indicates the suspected stability of the thermohaline
circulation, though the distribution of the transport within the
various density strata does vary [Roemmich and Wunsch
1985].
The choice for the separation between the two layers varies,
but only slightly, among authors. Gordon and Piola [1983],
noting that the salty characteristic of NADW is incorporated
into the Antarctic circumpolar belt below the ao density of
27.6 [Georgi, 1981], place water less dense than that value
within the upper layer (this includes the thermocline and inter-
mediate water). The density interval from ao of 27.7 to if2 of
36.82 divides the upper and lower layers in the inverse method
approaches [Roemmich, 180; Roemmich and Wunsch, 1985].
McCartney and Talley's [1984] separation between northward
and southward flow falls near 4øC, which coincides more or
less with the 27.7 ao.
Broecker and Peng [1982] in their discussion of the upper
layer feed for NADW point out that the upper layer water
must have about the same nutrient concentrations as NADW,
since there is no significant source or sink of nutrients in the
North Atlantic. In their Table 7-2 the nutrient concentrations
for the various components of NADW are listed. Character-
istic PO4, NO3, and SiO 2 concentrations are 1, 15, and 12
mol/kg, respectively. Inspection of the GEOSECS data [Bain-
bridge, 1981] indicates that within the central North Atlantic
these values are associated with temperatures between 11 ø and
13øC and salinity of approximately 35.55%0, corresponding to
5037
5038 GORDON: INTEROCEAN EXCHANGE OF THERMOCLINE WATER
a ao of 27.0. Thus the feed water is derived from within the
main thermocline, well above the Antarctic Intermediate
Water (AAIW) layer.
The magnitude of the thermohaline circulation for the At-
lantic Ocean has been estimated by various means. The in-
verse method approach of Roemmich and Wunsch [1985]
yields for the IGY and 1981 transatlantic sections at 24 ø and
36øN an average of 17 Sv for the northward flow of upper
layer water (surface component of 11.3 Sv and intermediate
water component of 5.8 Sv), which balances the southward
flowing deep water (19.9 Sv, 2.9 Sv of which is returning Ant-
arctic Bottom Water). This estimate is close to the 15- to
16-Sv values given by Roemmich [1980] and Hall and Bryden
[-1982]. These estimates do not take into account the Pacific
to Atlantic transfer of about 1 Sv in the Bering Strait [Coach-
man et al., 1975], so the southward deep flow is expected to be
slightly above the northward upper layer flow. Box models
based on temperature and salinity yield similar results. Gordon
and Piola [1983] find that gradual increase in salinity of the
northward flowing Atlantic Ocean upper layer water can be
justified with the Baumgartner and Reichel [1975] estimates of
fresh water exchange with the atmosphere if the magnitude of
the thermohaline circulation cell is near 20 Sv. McCartney and
Talley [1984] investigate the thermal field, finding southward
flow across 50øN of 14.1 Sv of cold water. Their feed of upper
layer water into the NADW formation region is about 11.5øC,
similar to the temperature horizon necessary to meet the nu-
trient requirements.
As NADW spreads into the South Atlantic and then east-
ward with the Antarctic Circumpolar Current into the Indian
and Pacific oceans, the negative estuary circulation pattern is
expected to extend out of the North Atlantic into the rest of
the world ocean [see Broecker and Peng, 1982, Figures 1-13;
Piola and Gordon, 1984]. The objective of this paper is to
explore the general form of this global scale thermohaline
circulation cell and demonstrate support for the hypothesis
that upper layer return flow is accomplished primarily within
the main thermoclines of the ocean.
The NADW upwells within the world ocean, returning
water to the upper layer within the Antarctic region and into
the thermocline. Antarctic and thermocline upwelling may be
coupled in that deep water upwelling around Antarctica con-
tributes to the formation of Antarctic Intermediate Water,
which then spreads below the thermocline and upwells into
the thermocline. There are two routes by which the upper
layer water can return to the Atlantic Ocean, though they are
not mutually exclusive: the cold water route within the Drake
Passage, in which AAIW and Subantarctic Mode Water
(SAMW) pass into the South Atlantic [Georgi, 1979; Piola and
Georgi, 1982; McCartney, 1977], and the warm water route, in
which Indian Ocean thermocline water is introduced to the
South Atlantic south of Africa (Gordon, 1985). It is proposed
that the warm water route is the more important.
WARM WATER ROUTE
As was mentioned above, most of the northward flow
within the upper layer in the North Atlantic resides in the
warmer water above the intermediate stratum [Broecker and
Pen.q, 1982: Roemmich and Wunsch, 1985]. A simple demon-
stration that this is also the case in the South Atlantic can be
made on the basis of the equatorward oceanic heat flux
characteristic of the South Atlantic. Hastenrath [1982] deter-
mines that 69 x 1013 W pass to the north across 30øS. Invok-
ing the concept that the wind-driven Brazil Current within the
thermocline is weakened by the thermohaline-driven circu-
lation pattern [Stommel, 1957] as some South Atlantic ther-
mocline water is transferred into the North Atlantic, it is pos-
sible to determine the mean temperature of the northward
moving upper layer water across 30øS.
The mass and heat flux equation across 30øS are
Mass
V• = [V•z + I/n] -- V•s (la)
Heat
FiT i = [ Vbz rbz --[- [7n Tn] --[- Q œ (lb)
where V• is the mass flux of the northward moving water, with
a temperature of T•, within the interior of the South Atlantic;
V•z is the mass flux of the Brazil Current within the upper
layer, with a transport averaged temperature of T•z; Vn is the
mass flux of NADW across 30øS, with a temperature of Tn;
is the mass flux of water through the Bering Strait; and Qs is
the northward oceanic heat flux across 30øS.
A production rate of 20 Sv for NADW, with uniform up-
welling north of 30øS, yields 16 Sv for Vn; Tn is taken as 2øC.
V•., is taken as 1.5 Sv [Coachman et al., 1979]. The transport of
the Brazil Current in the upper layer increases from approxi-
mately 6.5 Sv across 19ø-23øS to 17.1 Sv across 38øS [Gordon
and Greengrove, 1986]. An intermediate value of 10 Sv is taken
for V•=. The combined transport of the Brazil Current and
NADW flow across 30øS is 26 Sv, which approximately bal-
ances the Sverdrup interior transport across 30øS, given as 30
Sv by Hellerman and Rosenstein [1983]. The value used for T•z
is 18øC, as was determined from the distribution of the meridi-
onal component of the volume transport in temperature-
salinity space given by Miranda and Castro Filho [1981] for
the Brazil Current at 19øS. Using the Hastenrath [1982] Qœ,
50% of Q•, and 5% of Q•, equations (la) and (lb) are solved,
resulting in a T• of 15.4øC, 12.0øC and 9øC, respectively (Figure
1).
The warmest water in the Drake Passage is about 8øC
[Gordon and Molinelli, 1982-1, and the volumetric mode of the
AAIW/SAMW in the southwest Atlantic, which can be traced
into the South Atlantic, is 3.5øC [Georgi, 1979]. Thus even in
the extreme case in which the Hastenrath heat flux value is an
order of magnitude too high, the cold water route cannot be
the sole supplier of upper layer return to the NADW pro-
duction region. Assuming that the Hastenrath Qœ is roughly
correct, the bulk of the upper layer return flow must reside
within the thermocline. The only source for such water south
of 30øS is the Indian Ocean thermocline water within the
Agulhas Current.
The proposed global scale warm water route associated
with thermohaline-driven circulation cell is as follows (Figures
2a, 2b, and 2c): The upwelling NADW returns water to the
thermocline which was lost during production of NADW• The
primary site of upwelling is the southern ocean, where the
upwelled NADW eventually flows below the thermocline and
enters the thermocline as AAIW. The thermocline water is
returned to the North Atlantic by the following route:
1. The NADW introduced into the Pacific Ocean is trans-
ferred as North Pacific Central (thermocline) Water to the
Indian Ocean through the Indonesian seas.
2. The Pacific water crosses the Indian Ocean in the 10 ø-
15øS latitude belt, incorporating the saltier Indian Ocean ther-
mocline water.
3. The mix of Pacific Ocean and Indian Ocean water
passes southward within the Mozambique Channel, supplying
a small component of the Agulhas Current transport.
GORDON: INTEROCEAN EXCHANGE OF THERMOCLINE WATER 5039
for VeZ = 10 x 106 rn 3 sec
TSZ = 18 ø C
__ =2oC
O•VV
I o• Qf-345x10 w ß
....
RAKE PASSAGE I Of :34'5xlOlaW
•_ WATER I
]AAIW I I
IREASONABLE I
I•.• FOR I
•v.•
10 20 30 40 50 60 70
8O
VN 30os (10 6m3/sec)
Fig. 1. Relationship of the temperature of the water flowing to the north across 30øS (T•, see equation (1)) as is
required to balance the component of NADW advected southward across 30øS (•, see equation (1)). The relation is given
for three values of Q/(the northward heat flux across 30øS): the value given by Hastenrath [1982], half of that value, and
5% of that value. See equations (la) and (lb) for explanation of other values.
4. A branch of the Agulhas Current flows into the South
Atlantic and does not participate in the Agulhas retroflection,
which returns most of the Agulhas water to the Indian Ocean.
5. The warm water passes northward with the South At-
lantic subtropical gyre, crossing the equator to enter the upper
layer of the North Atlantic.
Assuming a 20-Sv production rate of NADW with uniform
upwelling into the world ocean (thus the Pacific receives 50%
of the deep water, with the Atlantic and Indian oceans split-
ting the rest equally [Piola and Gordon, 1984]) and assuming
that the cold water route is insignificant, it is possible to place
some numbers on the magnitude of the warm water route
(Figure 2a). The assumption of uniform upwelling (introduced
by Stornrnel and Arons [1960]) may not be strictly adhered to.
For example, if a larger proportion of NADW upwells in the
Atlantic, the transport values given for the Pacific-Indian and
Indian-Atlantic transfer would be reduced. Hence these num-
bers are not overly significant, since the assumptions are not
likely to be strictly followed, but they do provide some "ball
park" estimates which can be compared with other, indepen-
dent determinations.
The following section attempts to show supporting evidence
for key links of the proposed interocean warm water route.
EVIDENCE FOR THE WARM WATER ROUTE
lndonesian Through Flow
The most comprehensive physical oceanographic work on
the Indonesian seas is that of Wyrtki [1957, 1961]. On the
basis of water mass analysis he shows that Pacific water
spreads into the Indian Ocean down to depths of 1000 m.
Replacement of the water filling the deep basin of the Banda
Sea by sill overflow is an active feature within the Indonesian
seas. The displaced deep water is responsible for the low-
salinity Banda Intermediate Water of the Indian Ocean
[Rochford, 1966]. Above the 1000-m flow, Wyrtki traces some
eastward transfer, primarily by lateral mixing, of the Indian
02 minimum layer into the Banda Sea. At depths shallower
than approximately 500 m, flow is more substantial and is
directed into the Indian Ocean. This water, derived from the
Mindanao Current, contains subtropical salinity maximum
and intermediate salinity minimum water masses of the North
Pacific. "Thus in all layers with the exception of that of the
oxygen minimum, movements from the Pacific to the Indian
Ocean predominate" (page 114 of the Wyrtki [1961] study).
Water mass analysis using the gridded Levitus [1982] data
set (Figures 3, 4, and 5), supports the results of Wyrtki [1961]:
The water within the Banda Sea thermocline (10ø-20øC layer)
is similar to that of the North Pacific Central (thermocline)
Water, with only minor requirements for additional fresh
water (Figure 4). The Banda Sea thermocline water enters the
Indian Ocean primarily in the passages adjacent to Timor
island (Figure 3). The low-salinity thermocline within the east-
ern tropical Indian Ocean (Figure 3) is similar to the Banda
Sea/North Pacific thermocline water (Figure 5). It spreads
within the South Equatorial Current across the full width of
the Indian Ocean, as will be discussed below. Furthermore, it
is noted that within the 10ø-20øC layer of the Indonesian seas,
the vertical mixing (cross-isopycnal) coefficient must be large
(greater than 3 x 10 -4 me/s) to account for the modification
of the North Pacific thermocline temperature-salinity struc-
ture. The subtropical S maximum of the North Pacific ther-
mocline is destroyed as an isohaline thermocline forms in the
Banda Sea (Figure 4). The thermohaline structure is not con-
ducive to salt finger activity. The mixing is more likely a
product of interaction of the circulation with the irregular
topography of the region.
Intense vertical mixing implies substantial downward heat
flux within the Banda Sea thermocline. The vertical temper-
ature gradient is 10øC in 150 m (area 23 on Plate 205 of
Wyrtki [1971]); a vertical mixing coefficient of 4 x 10 -4 me/s
yields downward heat flux of 110 W/m e. The atmosphere to
ocean heat exchange for the region is approximately 90 W/m e
[Talley, 1984]. Thus as the water flows through the Indones-
ian seas, the upper layers of the thermocline would remain at
the same temperature or perhaps cool slightly, even though
5040 GORDON' INTEROCEAN EXCHANGE OF THERMOCLINE WATER
180øW 120 ø 60 ø 0 ø 60 ø 120 ø 180 ø E
b
\
\
b7
i
}øN
/b'
.•.
'? )o S
(•) UPWELLING
• SINKING
20"
0 o
DEEP WATER FLOW
"COLD"WATER TRANSFER
INTO ATLANTIC OCEAN
"WARM"UPPER LAYER FLOW
Fig. 2a. Global structure of the thermohaline circulation cell associated with NADW production. The warm water
route, shown by the solid arrows, marks the proposed path for return of upper layer water to the northern North Atlantic
as is required to maintain continuity with the formation and export of NADW. The circled values are volume flux in 106
m3/s which are expected for uniform upwelling of NADW with a production rate of 20 x 106 m3/s. These values assume
that the return within the cold water route, via the Drake Passage, is of minor significance.
the atmosphere to ocean heat flux is large, while the lower
thermocline would warm. This effect probably accounts for
the decrease of surface water temperature from over 28øC in
the western tropical Pacific to less than 28øC in the eastern
tropical Indian Ocean receiving the through flow (for example,
see the surface temperature maps of Pickard and Emery
[ 1982]) despite strong atmospheric heating of the ocean.
Wyrtki [1971] shows that the nutrient concentrations in the
Banda Sea on the 25.0ao surface (near the 18øC isotherm at
150 m) are 1.0, 15, and 23 mol/kg for PO4, NO3, and SiOn_,
respectively. Closer to the base of the thermocline, at the
26.6-ao surface (near the 10øC isotherm at 350 m), nutrients
are higher by somewhat less than a factor of 2: 1.8, 30, and 35
mol/kg for PO4, NO3, and SiOn_, respectively. Within the Pa-
cific Ocean's North Equatorial Current, which is the source
for the Mindanao Current [Tsuchiya, 1968] feeding the trans-
fer to the Indian Ocean, there is a strong nutrient-cline across
the 15 ø to 20øC layer [Wyrtki and Kilonsky, 1984]. Values of
0.5, 5, and 5 mol/kg near 18øC yield to 2.0, 25, 30 mol/kg near
the 10øC isotherm for PO,, NO3, and SiO2, respectively, a
factor of 5 increase. The breakdown of the nutrient-cline and
elevated nutrient concentrations in the upper layer of the
Banda Sea thermocline is likely to be a consequence of the
large vertical mixing coefficient within the Indonesian seas.
The high nutrient levels characteristic of the Indian Ocean
thermocline north of 15øS may be derived at least in part from
the vigorous mixing within the Indonesian seas.
The North Pacific origin for the water flow into the Indian
Ocean [Wyrtki, 1961] suggests that the NADW contribution
to the North Pacific thermocline is the chief supplier for this
link of the warm water route. The NADW entering the more
saline South Pacific thermocline may add some water to the
flow through, (associated with equatorial upwelling and ad-
vection) along the north coast of New Guinea (Figure 3), but
it appears to have a secondary impact on the Banda Sea
salinity. This leads to the question, where does most of the
z
HEAT HEAT u3 u3
HEAT z ,,,
OU3
zO • ea
NADW o
_ , <•
.e--SALTIER
FOEaATION_ • • •SALTIER • • __
• • •• THERMOCLINE I
ATLANTIC • AAIWl • INOIAN • I PACIFIC AAIW
I I TO ATLANTIC
,cc
I I
HEAT
It•H20
•> ACC
Fig. 2b. Schematic representation of the thermohaline circulation cell associated with NADW production and the
warm water route along line a-a' shown in Figure 2a. The upper layer water within the main thermocline begins its
passage to the North Atlantic in the Pacific as low-salinity water. It enters the Indian Ocean via the Indonesian seas,
where its salinity and volume flux increase by excess evaporation and further upwelling of NADW, respectively. The
thermocline water enters the Atlantic south of Africa and spreads to the northern Atlantic, continuing to increase in
salinity and volume flux.
GORDON: INTEROCEAN EXCHANGE OF THERMOCLINE WATER 5041
HzO HEAT HzO HEAT
-- t THER MOC•LIN•//(--. f
l
NADW S. MAX.
Fig. 2c. Schematic representation of the paths followed as
NADW upwells, becoming incorporated into the thermocline, along
line b-b' shown in Figure 2a. The main upwelling is expected to occur
in the Antarctic region, with eventual transfer to the thermocline as
Antarctic Intermediate Water.
NADW entering the South Pacific thermocline go ? This is not
the main topic of this paper, but it is suggested that it may
feed the cold water route via the Drake Passage. Under the
assumption of uniform NADW upwelling, this would account
for a maximum of 25% of the total return flow into the North
Atlantic Ocean.
There has been some work addressing the magnitude of the
Pacific to Indian Ocean transfer. Wyrtki [1961] determines
the current velocity from differences in dynamic heights. South
of Timor Island (Timor Sea) a single station pair indicates a
flux of 6.4 Sv. This flow is accomplished within the upper 300
m. However, in his summary of transports for the upper 150
to 200 m (his Table 12, p. 136), the resultant flow from the
Indonesian waters to the Indian Ocean ranges from a low of 1
Sv in the December to February period to 2.5 Sv in August,
with an annual average of 1.7 Sv.
Recently there have been a rash of estimates (based on a
variety of methods) of the Pacific to Indian transfer of water,
all of which are larger than Wyrtki's estimate (Table 1). The
average of all estimates is 9.2 Sv into the Indian Ocean.
Clearly, there is agreement that the water flow is toward the
Indian Ocean, but there is a wide range of estimates as to its
magnitude. The value proposed in this paper (8.5 Sv, Figure
2a) is close to the average estimate. The values would be
halved to about 4 Sv if the NADW entering the South Pacific
thermocline returns to the Atlantic by way of the Drake Pas-
sage. However, there is likely to be additional Pacific to
Indian transport which balances eastward transfer of (cooled)
thermocline water south of Australia. The value presented in
Figure 2a represents only that component involved in the
global scale NADW-thermocline circulation cell.
Trans-Indian Ocean
The low-salinity thermocline within the westward flowing
South Equatorial Current in the 10 ø to 15øS latitude belt of
the Indian Ocean (Figure 3) has temperature-salinity charac-
teristics similar to those of Banda Sea water emerging from
the Timor Sea region (Figures 5 and 6; also see Wyrtki
[1971], Plates 223, 231, and 237 for salinity on isopycnal sur-
faces within the thermocline and Plates 364 to 369 and 394 for
dynamic topography), attesting to its Pacific origin. The
characteristic of this feature stands in sharp contrast to the
more saline thermocline of the south Indian Ocean and Ara-
bian Sea; a lesser contrast is seen with the Bay of Bengal.
Inspection of the evolution of the temperature-salinity relation
in the downstream direction shows a steady increase of salini-
ty above the 10øC isotherm (Figures 3 and 6). The salinity
increases by approximately 0.5%0 throughout the thermocline
before reaching the Somali Basin in the western Indian Ocean.
It is likely that the salt is introduced by lateral (isopycnal)
mixing with the neighboring thermoclines. The salinity levels
within the neighboring thermoclines are maintained by the
regional excess of evaporation over precipitation [Baumgart-
net and Reichel, 1975]. In addition, NADW incorporated into
the Indian Ocean thermocline, via the AAIW route or directly
(Figure 2c), would be expected to swell the transport of the
warm water return flow during transit of the Indian Ocean.
The nutrient concentrations within the Indian Ocean ther-
mocline north of 15øS are significantly above those within the
south Indian Ocean thermocline [Spencer et al., 1982]. As was
mentioned above, similar levels are observed in the Banda
Sea, and it is suspected that the intense vertical mixing within
the Indonesian seas may, at least in part, be responsible. This
may also be true for the more or less isohaline thermocline of
the northern Indian Ocean. Thus it is possible that the Pacific
20øE
20øN
0 o
20 ø
40 ø 60 ø 80 ø I00 ø 120 ø 140 ø 160 ø E
' ?•!. '"':. ' ::•' ' ' '•. ' .':;" , '/ ' :; CZ"-" •'L- •, ....
..,,N•... / •--•/ / •. -• •- ., • .-•.. ;• .-,? • • • ._•
.) .---. ................ X::: C s w
- :/z • -•.•--' • •-• ...... • ..... •6'c '-. •:• •.?:] • -
. ./, .................. .-- ... .... .
•- • ............. ••••• :;.• ......... >-
.... ' • • • •... ..•I 20 oC
....... • ...... •[R• •• • 10-20•O LgYER
I I I •10• I • I I I I I I I I 40 • 8 • I00 • 120 • 140 • 160•E
Fig. 3. Average salinity in the 10 ø to 20øC layer of the main thermocline, prepared from the Levitus [1982] data set.
Intersections of the 20øC isotherm and the surface and of the 10øC isotherm and the sea floor are shown as thick dashed
lines. NPCW, SPCW, NICW, and SICW are the central or thermocline water of the North Pacific, South Pacific, north
Indian, and south Indian oceans, respectively.
20øN
0 o
20 ø
40øS