Journal of Marine Research, 48, 109-144, 1990
The North Atlantic Current and Subarctic
Intermediate Water
by Michel Arhan I
ABSTRACT
The region where the North Atlantic Current crosses the ocean around SON is studied from a
watermass point of view and using the ventilated thermocline model of Luyten et al. (1983).
Both approaches focus on the part played by the Subarctic Intermediate Water which, owing to
its pronounced thermohaline anomalies stands out as a clue to the complex oceanic circulation in
this area. Vertical mixing of this watermass with the overlying North Atlantic Central Water
creates a fresher variety of central water found between the two major branches of the North
Atlantic Current. Using the ventilated thermocline model several tracks are explored to try to
reproduce the subduction of Subarctic Intermediate Water, its movement into the subtropical
region, and quick return to the western boundary.
1. Introduction
The transition region between the North Atlantic subtropical and subpolar gyres,
which we may roughly ascribe to the latitudinal band 40N-55N, has always been of
great interest to oceanographers. The presence of the North Atlantic Current (NAC)
which feeds the eastern North Atlantic Ocean at European latitudes with warm water
of Gulf Stream origin stimulated many studies; yet, because of the great complexity of
the area, our knowledge of it remains uncertain.
The analysis of hydrographic data led Dietrich (1969) to the scheme of a NAC
crossing the ocean but splitting up into several branches over the Mid-Atlantic ridge.
Worthington (1962, 1976) arrived at the different conclusion that the NAC almost
completely recirculates west of the ridge in a northern anticyclonic gyre within the
Newfoundland Basin. Also on the basis of hydrographic data, McCartney and Talley
(1982), tracing the current from the subpolar mode water it transports, observed both a
western "compact recirculation" of part of it, and passage of the remainder into the
eastern basin.
Numerous satellite-tracked
drifting buoys in the Gulf Stream region (Richardson,
1983) and in the Newfoundland Basin (Krauss and Meincke, 1982; Krauss and Kase,
1984; Krauss, 1986; Krauss et al., 1987) have provided in recent years another means
of investigation of that region. Although some buoy tracks reported in the above papers
I. IFREMER,
Centre de Brest, B. P. 70, 29280 Plouzane, France.
109
Journal of Marine Research
110
[48, I
suggested western recirculation and many others illustrated crossing over the ridge,
neither Worthington's nor Dietrich's permanent circulation schemes could be recognized in the extreme tangling of trajectories. What stood out from these Lagrangian
experiments was the high mesoscale variability associated with the NAC meanders
and eddies.
At this stage it seems useful to turn back to basic descriptions and rationalization of
the area. Extracting simple and permanent flow patterns from such a variability is no
easy task, yet two ways seem promising to this end.
One is again the hydrographic approach. The NAC branches are associated with
strong water mass contrasts. As the permanent presence of some watermasses in the
region may be taken for granted, a closer study of these watermass-branch
relationships should help to establish the permanent character of some branches. This can now
be done in more detail than before on the basis of new hydrographic sections carried
out in the area, using CTD casts at a finer horizontal resolution.
The other approach is theoretical and based on the model of "the ventilated
thermocline"
(Luyten et al.. 1983) and its developments (Luyten and Stommel,
1986a). The question of the intergyre communication arose naturally in this model: the
line of vanishing Ekman suction, when zonal, is an impassable boundary for the
geostrophic flow in Sverdrupian homogeneous models, but not in layered models where
exchanges may occur through internal modes. Pedlosky (1984) and Schopp and Arhan
(1986) for the wind-forced case, Luyten and Stommel (1986b) for the wind and
buoyancy forced case, described such communicating solutions. Although these results
are based on linear vorticity conservation and clear evidence exists of higher order
dynamics in the NAC system, they provide us with a simple theoretical framework
suitable for a study of some basic features of the circulation in this region.
Both approaches are used in this paper. In the second section we study the
relationship between NAC branches and watermasses using quasi-meridional
CTD
sections carried out in 1983-1984
Franco-German
experimental
primary importance
on both sides of the Mid-Atlantic
program
TOPOGULF.
in this region of strong thermohaline
Ridge during the
Mixing processes must be of
contrasts. They are studied in
the third section, which reviews possible mixing mechanisms to account for the
observed modification of some watermass properties. The fourth section makes use of
the ventilated thermocline model, both wind and buoyancy forced. The purpose here is
not a comprehensive representation of this latitudinal band, but more modestly to see
which ingredients in the model help in reproducing some particular features of the
observed circulation.
Throughout the paper special attention is given to the Subarctic Intermediate Water
(SAIW),2 a watermass having temperatures between 4 and 7°C and salinities below
2. This has been the usual terminology for this watermass since Iselin (1936). although Wiist (1935)
prefered the designation "North Atlantic Intermediate Water."
1990]
Arhan: Subarctic Intermediate Water
I]1
34.9 psu in its formation region north of the northern NAC branch. Wiist (1935),
Iselin (1936) and Bubnov (1968) described the SAIW, its subduction and subsequent
southward spreading which they all observed to be limited to the northwest of the
Azores plateau. This was confirmed by Worthington (1976) who found most of the
SAIW in the region of his northern subtropical gyre. However, our knowledge of this
watermass has not yet come to the point where questions like its formation process, its
basic circulation pattern, or exchanges of properties with the surrounding waters can
be answered. These issues are discussed below.
2. Description of the North Atlantic Current System
The TOPOGULF
hydrographic data were reported by the TOPOGULF
Group
(1986) and a watermass study covering latitudes from 24N to 53N has already been
carried out from this data set by Harvey and Arhan (1988, hereafter HA). Sy (1988)
from the same data estimated the eastward transports associated with the NAC and
discussed the influence of the current on the distribution of Mediterranean
Water
(MW). As permanency of the NAC branches is at issue in this paper, maps from
historical data are first presented for comparison with the TOPOGULF patterns.
a. The branches of the North Atlantic Current
Figure 1 presents the 300 m potential temperature and 100/1500 m dynamic height
fields in the Newfoundland Basin from a set of hydrographic sections gathered in 1958
and 1964. These sections, which form a we1l-known star-shaped array centered around
46N-35W, are part of the basic data set on which our knowledge of the NAC was built
(Worthington, 1976; McCartney and Ta1ley, 1982). Despite the nonsynopticity of the
data we assume these property fields to be a correct representation of the gross features
of this area. Comparison with the TOPOGULF data wi1l strengthen confidence in this
assertion.
The sharp quasi-meridional temperature front associated with the NAC along the
Grand Banks continental slope (Fig. 1a) turns eastward at about 51 N. Isotherms from
5°C to 8°C keep their zonal direction farther east along this latitude but those from 9°C
to 11°C pursue their anticyclonic rotation until they make an eastward bend around
47N. The latter is suggestive of recirculation within the Newfoundland Basin but the
eventual eastward turning of a1l isotherms in two bunches suggests two eastward
branches. Recirculation stands out more clearly on the dynamic height field, yet does
not appear to be a completed process. The water flowing south between 35W and 38W
abruptly turns east along the southern temperature front of Figure la. The northern
front around 5] N is less pronounced than the southern one in terms of dynamic height.
For simplicity we refer in the fo1lowing to the recirculating pattern as Worthington's
northern gyre, although the real flow may not possess a1l the characteristics Worthington ascribed to his gyre. In particular we accept that eastward flowing water escapes
from it.
Journal of Marine Research
112
W50
W40
W35
W30
W25
[48, I
W2~18
N55
N55
N50
N50
N45
N45
N40
N40
N35
N35
W50
W45
W40
W35
W30
W25
W2~18
W50
W45
W40
W35
W3D
W25
W2~18
N55
N55
N50
N50
N45
N45
N40
N40
N35
N35
W50
W45
W40·
W35
W30
Figure I. Potential temperature at 300 m (OC, a) and dynamic height 100 m/1500 m (dynamic
meters, b) in the Newfoundland Basin from a historical composite set of data gathered from
R/V Discovery in August 1958, R/V Atlantis II from January to April 1964, and R/V Baffin
in July 1964. Subdomains I, 2 and 3 discussed in the text are reported (bold) on Figure Ia, and
positions of the three moorings laid at TOPOGULF site A on Figure Ib. Time se!'iesdisplayed
on Figure 3 are from position AI. The topographic contour is 3000 m.
The TOPOGULF
25W-35W
about
sections (Fig. 2) carried out in the approximate
twenty
confirmed the multi-branch
years later did not sample
character
longitudinal
the recirculation
band
region but
of the NAC farther east. The southern tempera-
ture front was present near 48N during the 1983 survey (Fig. 2a), again associated
with a sharp dynamic height gradient (Fig. 2b). The survey that year did not extend
sufficiently far to the north to observe the northern front. Both temperature
fronts are
Arhan: Subarctic Intermediate Water
1990]
W35
W30
W25 W22
W35
W30
113
W25 W22
N53
N53
N50
N50
N45
N45
N40
N4G
W35
W35
W30
W30
W25 W22
W25 W22
W35
W35
W3G
W25 W22
W30
W25 W22
N54
N54
N50
N50
N45
N45
N44
N44
W35
W30
W25 W22
W35
W30
W25 W22
Figure 2. Potential temperature at 300 m (DC) and dynamic height 100 m/ISOO m (dynamic
meters) for the TOPOGULF CTD arrays carried out in 19113from R/V Poseidon and in 1984
from R/V Meteor. (a) ()300(1983), (b) DH'OO/IsOO (1983), (c) ()300(1984), (d) DHIoo/1500(1984).
Subdomains 2S and 2N are reported on Figure 2c. The topographic contour is 3000 m.
again visible around 48N and 52N on the map from the 1984 survey, with a third
additional one comprising isotherms between 8°C and lOoC running in a southwestnortheast direction. Dynamic height gradients are weaker across this front. This survey
shows the front at 48N turning southward at about 23W after crossing the ridge. Even
if present in 1983, this direction change could not be observed because the survey was
limited to the region west of the front.
Because they are present in the TOPOGULF
fields and the composite map of
Figure la the temperature fronts at 47-48N and 51-52N are thought to be permanent
features of the region. In the following we name them the "southern"
and "northern"
Journal of Marine Research
114
[48, 1
PeDS!
450
400
35D
:[vv
T C'C)
f'-v/\ ~
V
V
\
Vecm s-')
20
N
20
o
o
-20
-20
-~~,-~-~,
~~-~,-~-~~-~,-~-~,
1
JUL
1
SEP
1
NOV
1
JAN
B3
1
MAR
1
MAY
-~
1
JUL
84
Figure 3_ Time series of pressure, temperature, and velocity vector at nominal level 350 m of the
TOPOGULFmooring AI (see position on Fig. lb).
fronts and the associated currents "southern" and "northern" NAC branches.
divide the region between 2SW and 3SW into three subdomains:
They
Region 1: South of the southern branch the potential temperature at 300 m is greater
than 11°C (Fig. 1) or 12°C (Fig. 2). HA showed this domain to be that of "pure"
North Atlantic Central Water (NACW).
Region 2: The intermediate domain is characterized by a quasi-constant temperature
(8-9°C) in Figure la, but crossed by a third temperature front in Figure 2c (called
"third front" in the following). This front (absent on Fig. 2a) is not permanent
and
will be discussed below.
Region 3: North of the northern branch the potential temperature
SoC: this is the domain of "pure" SAIW.
at 300 m is less than
The exact positions of the two main NAC branches meander around 48N and 52N.
Analysis of the results from a current meter array deployed along 48N during
TOPOGULF led Arhan et 01. (1989) to conclude that the amplitude of the oscillations
of the southern branch around its central latitude is of the order of 150 km. It is this
relatively low value which allows recognition of the branches on different surveys. The
existence of fronts and their variability
is best illustrated
by these current
meter
measurements. We show in Figure 3 a thirteen month time series of temperature,
velocity, and pressure at the nominal depth of 350 m and position 47°57N, 34°01 W
close to the center of the hydrological
time series is somewhat perturbed
array of Figure 1. Although
the temperature
by vertical excursions of the instruments,
passages
1990]
Arhan: Subarctic
Intermediate
Water
115
of the fronts are easily detected. What is remarkable is the tendency of the temperature
to stay constant between the sharp gradients. We illustrate these levels on the figure by
horizontal segments at 12°C, 8.5°C and 5.5°C. The presence in this record of three
preferred temperature levels is additional support for the existence of two basic frontal
structures in the region. At the beginning and end of the record the temperature stays
around 12°C for several months. Referring to the temperature map at 300 m on Figure
la, the southern NAC branch at these periods must be situated to the north of the
mooring position, and the current meter therefore immersed in NACW. The first
period at 12°C is followed by a two-month period at 8-9°C. The current meter is then
in region 2 of Figure 1. We also observe in mid-Maya 10- day period at 5-6°C, when
the current meter was in SAIW. Such brief periods occurred occasionally in three
other time series at 350 m at neighboring positions (Fig. 1). They are more likely
indicative of eddies detached from the northern front or filaments of SAIW extending
southward from this front, than of a displacement of the front itself.
b. The watermasses
This subsection develops in greater detail that part of HA's study relative to the
upper and intermediate watermasses of the region north of 40N. We identify the quasimeridional TOPOGULF
sections from their positions relative to the Mid-Atlantic
Ridge, east or west, followed by the year of realization (for example E83, W84 ... ).
During the 1984 survey, region 2 situated between the main NAC branches was itself
subdivided by the third front into two parts which we ca1l2S and 2N (Fig. 2c).
Figure 4 presents the () - S scatter diagrams in each region from the 1984 survey.
Also reported on the diagrams are:
• A best fit straight line adjusted to the lower part (lO°C < ()< 14°C) of the
NACW temperature range, using all TOPOGULF stations (including the South
Azores surveys) containing NACW. Following HA we call the water showing this
() - S relationship "pure NACW."
• Another segment from the base of the NACW to the () - S point characteristic of
Labrador Sea Water (LSW). This line distinguishes the waters primarily influenced by SAIW (to the left) from those primarily influenced by MW (to the
right).
• The enclosed region which HA proposed as the "pure" SAIW domain.
The diagram of region 1 (Fig. 4d) exhibits a tight () - S relationship above 10°C,
characteristic of pure NACW. At densities higher than 27.3 both influences of MW
and SAIW are visible. A total absence of NACW stands out in region 3 (Fig. 4a). HA
placed the upper limit of the SAIW domain at the isopycnal27.3 approximately 100 m
deep in this region. Here points up to 20 m depth are reported, which show the
existence of a seasonal thermocline with temperature reaching 14°C at salinities as low
as 34.5 psu. As pure SAIW is found directly below this thermocline, region 3 is that of
Journal of Marine Research
116
o
.f)
lOCI
lOCI
10
10
6
6
2
2
34.5
[48, 1
35.0
35.5
S (P.S.U.)
o
34.5
35.0
35.(j
S (P.S.UI
35.0
35.5
S IP.SW
e
('CI
10
10
6
6
2
2
34.5
35.0
35.5
S
(P.S.U.I
34.5
Figure 4. Meridional evolution of the () - S diagram. Points at a 20 db interval arc reported for
p> 20 db, for stations in regions 3 (a), 2N (b), 2S (c) and 1 (d).
pure and outcropping 8AIW. Both diagrams of region 2 (Fig. 4b, 4c) confirm its
intermediate character. In the density range (19 < 27.3 S increases with 0 as in NACW
but the cloud of points is not so tight as in region 1, and displaced to the left of the pure
NACW line. This water is called "modified NACW" by HA and appears as a mixture
ofNACW with fresher water. The perturbation from pure NACW is more pronounced
in region 2N than 28. At densities greater than 27.3 the water of dominant arctic
influence found below the modified NACW is called "subducted and mixed 8AIW." It
is present in both subdomains 28 and 2N. On the other hand the Mediterranean
influence is observed only south of the third front: this seems to be the watermass
signature of this front.
To better assess the relation between watermasses and frontal features we now
1990]
Arhan: Subarctic Intermediate Water
SI.Nb. ~ ~
117
",
•...
Nl'IO-
,.....N """
St Nb.!,~
o
1200
26.90
27.20
~.~\
[
/1, 32.25
32.35
5CPSUI
4SN
4QN
SON
Lahlude
Figure 5. Vertical sections of potential density I1p (a), salinity anomaly M (b), salinity (c), and
large-scale potential vorticity q (d). The upper boundary is at the depth 50 m in plots (b), (c)
and (d), drawn with I1p as vertical coordinate. Vertical smoothing was performed to eliminate
features of small vertical scale, using a square filter of width 30 m on Sand 6S. and 50 m on q.
Domains - 0.05 < M < 0.05 and q < 7.510-11 m-I S-I at 118 < 27.3 are shaded for recognition
of pure NACW and subpolar mode water. Domain M < 0 at 27.3 < 118,1 < 32.25 is also shaded
for recognition of SAIW.
discuss the three quasi-meridional CTD vertical sections carried out during
TOPOGULF: E83, E84, and W83+84. The latter, though consisting of sections from
two different years joined end to end, exhibits both major fronts of the area and may be
regarded as synoptic for our purpose. Figures 5, 6 and 7 present for each line the section
of density versus pressure (a) down to 1600 db, and sections versus density of salinity
anomaly (b), salinity (c), and large scale potential vorticity (d). The salinity anomaly is
the difference liS of the observed salinity from the salinity at the same density on the
dividing solid 8 - S line of Figure 4. Potential vorticity q was computed from the
Brunt-Vai'siilii frequency N, q = fN2jg (fis the Coriolis parameter and g the acceleration of gravity). The display is limited to the density range (11 < 32.35. Isopycnals
(10 = 27.3 and u] = 32.25 are shown on the sections. The former is an approximate
Journal of Marine Research
118
St.Nb.
£
[48,1
~
n
n
N
N
N
26.00
@
NB
SI.Nb.E
o
26.60
0'0
400
UJ
a:
27.20
=>
l2
800
UJ
l27.50
a:
Q.
32,'5
0',
32.25
[
32.35
26.00
26.00
26.30
26.30
26.60
0'0
26.90
26.90
27.20
27.20
27.50
27.50
15
32,15
r32.
CT,
l32.25
<1,
32.25
[
3235
32.35
Latilude
45N
SON
Lalitude
45N
50N
Figure 6. As in Figure 5 but for section E84.
lower boundary of the pure NACW and upper boundary of the SAIW intrusion (Figs.
6c, 7c). The latter was chosen by HA as the lower boundary of SAIW.
Upper density range (0'0 < 27.3). Except for three patches of weak salinity anomalies in
region 2 of Figure 6b all three liS sections confirm the southern NAC branch to be the
northern limit of pure NACW defined as I liS I < 0.05. Because of the curvature of the
S diagram in NACW, which Schmitt (1981) attributed to double-diffusion, liS
may be higher than 0.05 in the upper NACW. This is observed around 0'0 = 26.85 at
the southern end of section W83 + 84. Values of liS lower than - 0.05 are also found in
the uppermost part of regions 1, which corresponds to the seasonal thermocline. North
of the southern NAC branch with slight southward extensions at densitities 27.05 <
0'0 < 27.3 on W83+84 and E83 is the modified NACW characterized by -0.5
<
o-
Arhan: Subarctic Intermediate Water
1990]
a"
26
1
119
60
21.50
32,'5
(1,
32.25
[
,32 35
.
40,.,
50N
Lahlude
3235
"ON
"5N
Lahtude
Figure7. As in Figure 5 but for sectionW83+84.
M < -0.05. North of the northern branch on E84 and W83+84, values of I:1S are
lower than -0.5. This is water from the seasonal thermocline above the pure SAIW,
which we call Subarctic Surface Water. Because of nonuniform intervals between
isopleths the gradient of t:lS is not so apparent at the northern front as it is at the
southern one.
South of the southern branch and within the isolated NACW bubbles to the north on
E84 a potential vorticity minimum exists, centered at (16 '" 26.9 on W83 + 84 and (10 '"
27.0 on the eastern sections, at temperatures around 12-13°C. In region 2 another
minimum is present east of the ridge within the modified NACW around (10 = 27.2 and
= lOoC. These are two varieties of McCartney and Talley's (1982) subpolar mode
water (HA).
o
Intermediate density range (27.3 < (16,1 < 32.25). Outcropping of SAIW north of the
northern branch and its subduction south of it is best observed on the s!llinity sections
of Figures 6 and 7. On W83 + 84 pure outcropping SAIW is also present at station 206
south of the branch, probably indicative of an eddy having detached from the front.
The separation between MW and SAIW stands out from the I:1S contours (the domain
120
Journal of Marine Research
[48, 1
of dominant subarctic influence is shaded). It takes different forms on the different
sections. On W83 + 84 and E83 it occurs south of the southern branch, respectively
between 41°30'N and 44°30'N, and between 45N and 47N, the upper SAIW being
found farther south than the lower. In both cases the transition is associated with a
progressive slope (over two to three degrees of latitude) of the isopycnals. On E84 the
transition is sharper but occurs at two different latitudes according to density: in the
vicinity of the southern branch at densities 27.3 < ()8 <27.5, and at 51 N, where the
third front is observed, at 27.5 < 0'8,1 < 32.25.
In all cases transition from MW to SAIW in the intermediate density range is
associated with additional meridional density gradients. Sy (1988) also observed the
branch around 51 N (our third branch) on E84 to be the northern boundary of MW,
but considered it the southernmost branch. The above watermass analysis leads to the
slightly different conclusion that the southern branch is at "" 47N, and deeper MW
may be present north of it.
Though we have been able to place on the sections unequivocal boundaries between
SAIW and MW, a significant thermohaline mesoscale variability exists in the intermediate density range, most visible on the eastern salinity sections. An example is the
parcel of SAIW found at station 231 south of the third front on E84. Mesoscale density
structures are associated with such features which suggest the boundary between MW
and SAIW in this region to be highly unstable.
A potential vorticity maximum (~15 10-11 m-I S-I) exists throughout the sections,
at 0'8 "" 27.45 on E83 and E84 (Figs. 5d, 6d), and 0'8 "" 27.3 on W83 +84 (Fig. 7d). This
maximum is connected in the north to the extremely high values of the upper
outcropping SAIW. However, as it extends south of the southern limit of SAIW, it
should probably not be linked to this watermass. What is certain, on the other hand, is
that no trace of a minimum exists, i.e. the SAIW, though having its origin at high
latitudes, is not a mode water formed through convective processes. Additional
elements to the question of the SAIW formation are given in Section 4.
Iselin (1936) gave a description of the NAC system on the basis of a meridional
section at 30W carried out by R.V. Atlantis from 53N to 37N. He already pointed out
the two main branches which he placed "between latitudes 51 °27'N and 49°23'N" and
at "about latitude 46°30'N". His figures (Fig. 29 and 30) showing the temperature and
salinity contours exhibit several features discussed above, like the freshening of the
NACW in region 2. Transition between subducted SAIW and MW occurs near the
southern branch. Iselin describes the subduction of SAIW at the northern front but
observes that in crossing the front this watermass
"is so subjected to turbulence
very low salinities are soon absorbed and disappear."
that its
This is one aspect of the strong
mixing activity of this region which we discuss in the next section.
3. How is the modified North Atlantic Central Water formed?
The presence of a modified version of NACW between the two main NAC branches
raises the question of the formation of this watermass.
The description
above suggests
1990]
Arhan: Subarctic Intermediate Water
121
that it owes its birth to the mixing of pure NACW with fresher water which in this
region can only be of subarctic origin. In this section we review the several possibilities,
namely:
• diapycnal mixing with the subducted SAIW below,
• lateral mixing at the northern front with outcropping
• vertical mixing with a thin layer of Subarctic
NACW.
a. Diapycna/ mixing across
lIe =
SAIW,
Surface Water spreading above the
27.3
This mechanism could account for both observed loss of salt of NACW and gain of
salt of the SAIW below. It is supported by the observation that subducted SAIW is
always accompanied, above lIe = 27.3, by modified NACW. This is true in region 2
where all the overlying NACW is modified, but also in the northern part of region I
when the upper subducted SAIW extends south of the southern branch (in E83 and
W83+84). There some modified NACW (~<
-0.05) is present at the base of the
upper density range below pure NACW, at stations 171 to 174 on E83 (Fig. 5b) and
stations 128 to 135 on W83+84 (Fig. 7b).
Diapycnal mixing between NACW and SAIW could result from shear instability or
from double-diffusion, for which conditions appear favorable. A good indicator of
double-diffusion is the density ratio Rp = exTzf{3Sz, where ex and {3are the thermal
expansion and haline contraction coefficients in the equation of state of the seawater.
Schmitt (1979, 1981) showed that layers with 1 < Rp < 2 are favorable to saltfingering and that a typical value in the NACW is Rp "" 1.9. Figure 8a presents the
(J - Rp diagram of station 228 in region 2 : values as low as 1.6 are observed at potential
temperatures between 9°C and lOoC, at the boundary between NACW and subducted
SAIW. Double-diffusive convection in the salt-fingering regime generates thermohaline staircases, which are the most obvious signature of this mechanism. No real
staircase was observed in the TOPOGULF data from this region but irregular steps
were present at some stations, which could be remnants of staircases. An example is
presented in Figure 8b for another station in region 2. Irregular steps 20 to 40 m high
exist at the boundary between NACW and SAIW, in the temperature and density
ranges 8° .$ 0 .$ 9°C and 27.25 .$ (1es 27.35.
To illustrate the effects of diapycnal mixing between the two watermasses a simple
model was used, similar to that developed by Schmitt (1981) to study the influence of
double diffusion on the shape of the () - S curve in central waters. We assume that at
time t = 0 a layer of pure SAIW of thickness (D - h) has just subducted below a layer
of pure NACW of thickness h. and study the subsequent evolution of the 0 - S
diagram
under the effect of diapycnal
mixing, both turbulent
and double diffusive.
Advection is ignored in the model, though a relative flow between the two watermasses
certainly exists in the real ocean. The advective terms would be of primary importance
7.6
I
B.4
B.O
'
,
35.10
0
[48, I
Journal of Marine Research
122
•
B.B
I
9.2
9.6 O("C)
'I
35.00
~
35.30 SIPS UI
P
Sf 228
fOC)
IdBI
Sf 223
580
12
10
I
8
I
I
I
6
660
I
I
-.q:::=I
I
4
®
2
700
,
r-
0
740
2
3
4
Rp
Figure 8. Signs for double diffusion in region 2 at the interface between NACW and subducted
SAIW. (a) Diagram 0 - Rp at station 228 on section E84. Vertical derivatives in the
expression of Rp are computed using f:.z = 50 m. (b) Irregular thermohaline steps around rJo =
27.3 at station 233 on section E84.
in a steady state tracer balance. But our only purpose here is to reproduce qualitatively
the deformation of the () - S curves observed on Figures 4b and 4c.
The initial interface is at density (J8 = 27.3 (Fig. 9a). Above and below we assume
that the () - S relationships are linear in both watermasses:
• S = 0.13 () + 34.05, in th~ pure NACW,
i.e. the equation of the best fit straight
line of Figure 4 .
• S
=
-0.16
()
+
35.64 in the pure SAIW. This is an average line of the domain
shown in Figure 4.
Mean vertical
O.003°C m
-I
temperature
gradients
Tz = a.Ol°C m-I in NACW
in SAIW were estimated from the TOPOGULF
and
T. =
profiles.
The equations are:
(1)
where A is the turbulent
ties for Sand
diffusivity, and As and AT equivalent salt-fingering
diffusivi-
T, dependent on Rp and different from each other. Salt fingering causes
downward fluxes
F T and Fs related by aFT
=
rf3Fs, where r is a flux ratio assumed
Arhan: Subarctic Intermediate Water
1990]
T ~-~-~-~-~-.-,
123
~---------.-,
(oe)
12
/
.........
~ (Jo.27.3
;:;../
10
,.. -"
8
,,/'
,-
/'
/'
6/~
4
toO
2
t.l
@
@)
12
/
10
8
6
,.-
,(
"
4
t.6
t.2
2
34.5
(
35.0
35.5
34.5
35.0
35.5
S (psu)
Figure 9. Evolution of the T - S diagram in region 2 assuming that turbulent mixing (A = 10-4
m2 S~I) and double diffusion (ADD = 10-3 m2 S-I, 'Y = .6) are active at the interface between
NACW and subducted SAIW. The time (1) is expressed in months.
equal to'Y = .6. In terms of equivalent diffusivities this gives:
AT = 'YAs /
Rp.
We use for As (Rp) the same expression as Schmitt
double-diffusive fluxes for Rp > 2:
(1981) which ensures vanishing
with Rc = 1.7 and n = 32. We impose no flux at the upper and lower boundaries.
Figure 9 presents the results obtained with h = 300 m, D - h = 500 m, A = 10-4 m2
S-I, ADD = 10-3 m2 S-I.
Owing to the high temperature and salinity gradients across the initial interface the
S curve evolves very quickly at the beginning. After one month the upper part of
the SAIW segment has been significantly eroded toward higher salinities and the lower
NACW has lost approximately 0.1 psu along isopycnals (Fig. 9b). The upper NACW
is still pure, a situation comparable to that observed above the SAIW present in region
I of E83 and W83 + 84. This suggests that at the time when these sections were carried
o-
out SAIW had just progressed farther south and "modification"
of the NACW above it
was not yet completed. After two months the whole NACW is modified. The process is
[48, I
Journal of Marine Research
124
still going on after six months but the amplitude of the salinity anomaly !::S in the
NACW does not increase any further. Due to the different diffusivities for Tand S the
diagram has rotated, the upper water losing density to the benefit of the lower. This
behavior, discussed by Schmitt (I 981), is qualitatively similar to that observed on the
diagrams of region 2 in Figure 4. It appears when double diffusion dominates turbulent
mixing (here ADDIA = 10). Rotation of the diagram was slower on other numerical
tests carried out with equivalent influence of both mixing processes. It was not observed
with turbulent mixing alone. Because advection is ignored, there is no hope of attaining
any steady state with this model. However we observe a certain stabilization of the
anomaly 6.S, a consequence of the weakening of the double-diffusive fluxes when Rp
increases due to the rotation of the diagram. South of the third front on E84 the salinity
anomaly of the observed modified NACW had values between -0.1 and -0.2 psu
(Figs 4, 6), i.e. comparable to what is predicted by the model. North of it, however, the
anomaly was more intense ( -0.5 < (j.S < -0.2psu),suggestingthatothermixingmechan isms may be at work there.
Given the ratio ADD lA, the time needed to reach a nearly constant (j.S depends on
the magnitude of ADD' A few months were required here with Schmitt's value ADD =
10-3 m2 S-l, but this time scale is of course very uncertain as are ADD and the estimates
of double-diffusive salt fluxes themselves (Herbert, 1988).
b. Lateral mixing at the northern front
Schmitt and Georgi (1982) and Georgi and Schmitt (I983) studied the mixing
mechanisms in this region on the basis of CTD data along a section from the Azores to
Flemish Cap, and four dives of the profiler SCIMP (Self-Contained
Imaging Micro
Profiler) allowing optical recognition of mixing situations. The section crossed the
NAC at a location where it is still flowing northward offshore of the Grand Banks,
around 46°30'N and 42W, and far west of the TOPOGULF area. These authors show
the double-diffusive
vertical
interleaving
active at the front to be responsible
mixing there than in the gyre interior.
The resulting
for stronger
net cross-frontal
ex-
changes were estimated from Joyce's (1977) model to have equivalent lateral diffusivities around 10 m2 S-I.
Although the TOPOGULF CTD array was not designed for such mixing studies, the
two stations which happened to be within the northern branch of the NAC on either
side of the ridge confirm that the same mechanism was at work there, about one
thousand kilometers northeast from where Schmitt and Georgi (I982) observed it.
Intrusions of warm and salty water into the colder-fresher subarctic water stand out at
station 212 on Figure lOa, both above and below (Yo = 27.3. Above this isopycnal
interleaving involves the modified NACW and Subarctic Surface Water while below it
occurs between the subducted-mixed
thermohaline
indicative
and pure versions of SAIW. Figure lOb shows the
profiles of the uppermost
of salt-fingering.
intrusion.
This mechanism
Steps are apparent
involves the Subarctic
at its base,
Surface
Water
Arhan: Subarctic Intermediate Water
1990]
125
Sl 212
10
5
1
34.5
35.0
7.5
8.0
50
P
(dB)
S
35.5
8.5
e
PSU
34.6
34.7
34.8
34.9 S(P.S.u)
®
100
150
Figure 10. Signs for double-diffusive interleaving at the front associated with the northern NAC
branch. Station 212 is on section W83+84. (a) 0 - S diagram showing the intrusions. (b)
Thermohaline profiles showing irregular steps at the base of one intrusion.
adjacent to the front in the seasonal thermocline of the outcropping SAIW, best
recognized on Figure 4a. As this water, like NACW, has densities lower than (Jo =
27.3, mixing here has an isopycnal component in the upper density range, hence is
basically different from the purely diapycnal mixing across (Jo = 27.3 discussed above.
These exchanges at (Jo < 27.3 are able to enhance the freshening of the northernmost
modified NACW, as observed on the 8 - S diagram of region 2N (Fig. 4b), not on that
of region 2S (Fig. 4c).
As double-diffusive interleaving is expected to occur at water mass fronts of widths
typically 5 km it must also be active at the rim of eddies detached from the northern
branch and carrying parcels of SAIW southward. Such a structure was observed at
station 206 (Fig. 7c), and although no well-defined intrusive feature was present at the
neighboring stations, probably because of the large sampling interval, eddies certainly
provide a means of carrying the effects of double-diffusive interleaving farther south.
Schmitt and Georgi (1982) suggest that while double-diffusion is responsible for
most of the vertical mixing near the front, shear instability
might be dominant away
Journal of Marine Research
126
[48,1
from it. This seems to contradict the discussion of Section 3a above in which double
diffusion was invoked as an important vertical mixing process between subducted
SAIW and NACW. However, Schmitt and Georgi's conclusion about the gyre interior
mainly rests on one SCIMP dive at a station near the Azores ("" 40N, 29W) which
seems devoid of SAIW. What was suggested above is that double diffusion may be
important in that portion of the gyre interior where SAIW is present below NACW.
There the transition from warm-salty to cold-fresh water causes a lowering of Rp
toward 1, favoring salt-fingering. This is what was reproduced in the simple model of
where Rp was initially equal to 1 at density fro = 27.3. Actually, the only level at their
gyre interior station where Schmitt and Georgi infer salt-fingering is around 380 m and
fro"" 27 .3, at a place where evidence of mixing was given by the SCIMP, Rp was lower
than 2, and thermohaline steps were observed.
c. Vertical mixing of North Atlantic Central Water with Subarctic Surface Water
Figure 11 presents expanded views of the upper and northern part of the salinity
sections (vs pressure) E84 and W84. A layer approximately 50 m thick of extremely
fresh Subarctic Surface Water (some salinity values are lower than 34.0 psu) is seen
spreading southward above the modified NACW over an area which seems to coincide
with region 2N of Figure 2c. Mixing with this water could be another cause of the
enhanced freshening of the modified NACW north of the third front. However, as this
surface layer constitutes the seasonal thermocline of the area it occupies (data of
Figure 11 are from September), this third mixing possibility cannot be permanent, as
were the other ones discussed above. This layer may be regarded as the lightest part of
the Subarctic Surface Water present north of the front. Due to its low salinity it has
become lighter than the modified NACW when heated at the surface and its spreading
over this watermass was interpreted by Iselin (1936), who also observed it, as
"probably a good example of the transfer of surface water to the right of the wind
direction." Though shear instability at the thermocline could induce vertical property
transfers in the summer period, mixing is certainly most efficient in autumn when the
seasonal thermocline disappears. Then, assuming that a 50 m thick layer of Subarctic
Surface Water of mean salinity 34.7 psu completely mixes with a 300 m layer of
modified NACW of salinity 35.1 psu, a freshening of the latter by about 0.06 psu would
result. Though significant, this value is still weaker than the salinity anomalies
observed in region 2N (-0.5 .:S 6.S .:S -0.2). Double-diffusive
must be the prevailing mixing mechanism there.
4. Subduction of the Subarctic Intermediate
Figure 12 showing two temperature
subduction
Water
maps from Dietrich's atlas (1969) illustrates the
of SAIW. At 150 m the outcropping
an eastward-pointing
interleaving at the front
version of this watermass
tongue of cold water north of the northern
appears as
NAC branch and
Arhan: Subarctic Intermediate Water
1990]
127
lD
N
N
lD
M
8T. NB.
N
.
0
LU
a:
~
200
en
en 400
LU
a:
CL
600
z
o
LAT.
If)
lD
81. NB.
0
0
N
N
N
0
LU
a:
~
200
en
C/)
LU
400
a:
CL
600
Z"
LAT.
o
If)
Figure 11. Expanded salinity sections E84 (a) and W84 (b) showing the spreading of Subarctic
Surface Water over the modified NACW. Note the changes in intervals between isohalines at
34.9 and 35.0.
adjacent to it. At 600 m the northern NAC branch is still visible from the 5°C and 6°C
isotherms. The subducted SAIW flows southwa'rd west of about 25W, its westernmost
part then turning southwestward, encircling an area of warmer water. This flow takes
place in the recirculation region of Worthington's (1976) northern gyre (west of ~32W,
between 40N and SON), a confirmation that the flow direction inferred here from the
shape of the 8°C isotherm is correct. Entrapment of warmer water to the west may also
be related to the presence at the same location of a saline Core on a salinity map at 6°C
shown by Worthington (his Fig. 23). These thermohaline contrasts make the SAIW a
particularly
good tracer of the recirculation
of the northern
gyre in Worthington's
128
[48, I
Journal of Marine Research
3Q"w
2Q"w
Figure 12. Temperature maps from Dietrich's atlas showing the eastward pointing tongue of
outcropped SAIW (a) and the southward spreading of the subducted SAIW (b). Lines
F w (X) = 0 and L(X, F) = 0 from Figure 13 are reported on Figure 12a. Three stars on
Figure 12b indicate the southern limits of SAIW influence on the TOPOGlJLF sections.
Domains 5 < T < 7°C (a) and 5 < T < SoC (b) are shaded for recognition of the outcropping
and subducted SAIW.
lower-thermocline layer, as illustrated on Figure 12b. Three stars on this figure mark
the southernmost limits of subducted SAIW on the three TOPOGULF
sections of
Figures 5, 6 and 7. Expressed in terms of salinity anomaly, not temperature, these
limits are somewhat south of the 8°C boundary used on this map. Yet they show that
the greater southward spreading in the western basin observed at TOPOGULF and
pointed out by Bubnov (1968) is consistent with recirculation of most of this water west
of the ridge. Figure 12a also shows the line of vanishing Ekman suction, determined
from Bunker's (1980) mean wind stress field. SAIW is formed north of that line, then
crosses it while spreading southward.
In this section we use the model of the "ventilated
thermocline"
some features present in these maps. This simple Sverdrupian
represent
all the complexities
of the circulation
to try to reproduce
model cannot, of course,
in this region, like the multi-front
system described in Section 2. However, as it is particularly suited to the study of
watermass subduction and cross-gyre communication, it should be helpful in trying to
understand
the observed behavior of SAIW and the existence of the northern
gyre
itself.
a. The model (Luyten et al., 1983; Luyten and Stommel, 1986a)
Two layers of thicknesses h, H - h and densities PI' P2' containing
the NACW and
SAIW, are in motion over a third resting layer of density P3' The horizontal
compo-
1990]
Arhan: Subarctic
Intermediate
Water
129
nents of the geostrophic velocity are
U1
1'2(
I
= -
Hy
1'1)
+ 1'2 hy ,
(2)
1'2
I Hy,
U2 = -
where I'i = g( PH I - Pj) I Po, andf, g and Po are respectively the Coriolis parameter, the
acceleration of gravity, and a mean density. The linear potential vorticity equation is
expressed in each layer on a t1-plane:
ht1Vl
-Wi
-
f
+ WH
= -We
(3)
(H - h)t1V2
I
Wi -
= -WH
Here we is the Ekman suction and the vertical velocity
W
at the upper interface has
been partitioned into Wi = -VI' 'ilh, and wH = W - Wi' associated
isopycnal fluxes. In the following we use the nondimensional horizontal
X
= xla
Coriolis
with crosscoordinates
and F = Ilia where a is the zonal width of a rectangular ocean basin andlo the
parameter at a central latitude A.a. In Eq. (3) we express the velocity
components as functions of the interfacial depths and follow the procedure of Luyten
and Stommel (1986a) to obtain the characteristic equation:
which expresses that Hvaries
as
dH
h 1'1
= - - - WH,
ds
H 1'2
-
dX _
Uc -
dF
ds
aLlaF
_(3_ [
_
ds -
along characteristics
-
I~Fa 1'2
1'1 h(H - h)]
+F
Vc
=
H
and L is the Sverdrup function
h are assumed known (HE' hE) along the eastern boundary of the
ocean. H is computed from (5) in the area spanned by characteristics
boundary.
(5)
H
FWe
=
where s is the distance along characteristics
In practice, Hand
H
defined by
The depth
of the upper
interface
is then obtained
issued from that
from the zonally
Journal of Marine Research
130
[48, 1
Figure
13. Fields of Ekman pumping (we' positive upward) and surface heat flux into the ocean
determined from Bunker's (1980) fields of mean wind stress and heat flux. The line of
vanishing L(X, F) is also reported (dashed).
(Q)
integrated Sverdrup relation:
H2
+ 1'1 h2 _- H2E+
1'2
1'1 h2
E+ 2L
1'2
.
(6)
Integration in regions not reached by characteristics from the east must be carried out
from the western boundary. Another relation involving hand/or H must be specified
there to complement Eq. (6) and determine the depths of both interfaces at this
boundary (HM!' hw)'
Had the characteristic equation (4) been written for h, the rate of variation along the
same characteristic would have been:
dh
ds --
If wH
=
(7)
0 the second equation in system (3) reduces to
which expresses that potential vorticity in layer 2 is exactly conserved. This is the case
of the original model of Luyten et al. (1983).
b. Wind and buoyancy forcing in the region
Figure 13a shows the distribution of the vertical velocity we at the base of the Ekman
layer. The line of vanishing we is that reported on Figure 12. Another important line to
be discussed below is that defined by L(X, F) = 0 which, due to the southwestnortheast inclination of the line of vanishing we is distinct from it.
1990]
Arhan: Subarctic Intermediate Water
131
The interfacial flux wH appearing in the above equations is related to the divergence
of the diffusive density fluxes through the heat equation:
(8)
where we distinguish the divergences of horizontal and vertical fluxes. The great heat
losses to the atmosphere prevailing in the region (Fig. 13b) causing a convergence of
vertical density fluxes at the upper levels would, if alone, require a negative WH to
equilibrate them. But other causes of buoyancy fluxes may exist, whose effects on WH
are more uncertain. If double diffusion is active as suggested above, it is associated with
negative values of Fpv. Depending on their vertical gradients these diffusive fluxes may
require vertical velocities of either sign. As salt-fingering also causes lateral fluxes
through interleaving at fronts, it may also influence the term V H • FpH of the above
equation. The eddy action could also be sensitive through this term. Whether the latter
mechanisms are able to reverse the sign of WH is unsure. Estimates of double-diffusive
fluxes are uncertain, and their divergence is still more; the divergence of horizontal
fluxes is certainly not better known. We consider in the following that negative wH are
most likely in the region, but accept, in one of the examples below, the possibility for it
to be positive, particularly near the outcropping line. Positive values of WH east of 30W
along 48N were indeed inferred by Arhan et al. (1988) from an analysis of the
TOPOGULF current meter data using the same Sverdrupian dynamics.
As buoyancy forcing appears in this model in terms of interfacial fluxes, it can only
be present in those regions where two moving layers exist. In the present application
there is no way to reproduce buoyancy forcing north of the outcropping line where
layer 2 is the only moving layer. This is a deficiency of the following representations of
the flow in this area, since diffusive buoyancy fluxes, either vertical or lateral, certainly
exist north of the northern N AC branch.
c. The problem of intergyre communication
We call the latitude where We vanishes Fw(X). Possibilities
latitude are reviewed by setting we = 0 in Eqs. (4) and (5) .
of flow across this
• If wH = 0 and the second term in Eq. (4) does not vanish, this equation shows the
characteristics
to also be isolines for H, i.e. streamlines for layer 2. As Eq. (5)
shows ve' hence V2, to vanish with We' flow may exist across F = F w (X) only if this
line is not zonal. The real line of vanishing we is indeed not zonal west of "",25W,
and this mechanism which was already discussed in Talley (1985) and Schopp
and Arhan (1986) could be relevant here .
• If wH = 0 and the line F = Fw (X) is zonal, the only possibility for Hx to be
different from zero on that line is that its coefficient in Eq. (4) be itself equal to
Journal of Marine Research
132
[48, I
zero. This is the internal mode solution discussed by Pedlosky (1984) and Schopp
and Arhan (1986).
"*
• If W H
a no such special solution is required for H to vary along F = F w (X). As
buoyancy forcing is certainly active in the region, this is the most natural
mechanism for intergyre communication. It was studied by Luyten and Stommel
(1986b).
We discuss below the capacity of these several mechanisms to account for the flow of
SAIW across the line F = Fw(X), illustrated on Figure 12. We successively consider
cases with (i) a non zonal line of vanishing We' (ii) a positive WH, and (iii) a negative
WH' The part played by the internal mode of Pedlosky (1984) appears in the latter case.
d. Case with wind forcing only and a nonzonalline F = F w (X)
We use Ao = 47N as the reference latitude and consider a rectangular ocean model
of zonal width a = 2500 km; i.e., about the real width of the North Atlantic from lOW
to 43W at this latitude. We only consider latitudes between 21N and 61N corresponding to 0.5 ~ F ~ 1.2. This basin is driven by a sinusoidal Ekman pumping
We = -Wo
sin [?r(F - Fw(X)
+ 1)]
(9)
with Wo = 10-6 m S-I. In this first example we choose to represent the SW-NE sloping
of the line of vanishing we west of about 26W (or X = 0.05). Limitation of the slope to
the western part of the ocean could be the reason why SAIW hardly enters the eastern
basin:
F w(X)
{Fw(X)
Luyten and Stommel
wH
~
= AX + B, A = .2, B = 1, for
X~
= FWE = 1.1
X> .5
for
(1986a) showed that subduction
.5
can only occur where we -
O. As wH equals zero here and we want the SAIW to subduct from the region of
positive we' our only possibility
is that the outcropping
line FI (X)
be identical
to
Fw(X).
This may be regarded as a coarse approximation to reality, as the northern
NAC branch also shows over the whole ocean width a slope toward northeast and twice
intersects the line F
=
F w (X) on Figure 12a.
The model was first run by setting HE = 1000 m, hE = 0, and mean densities in the
three layers equal to PI
that it is the outcropping
isopleth H
=
= 1.0272,
P2 = 1.02745, and PJ = 1.02775. Figure 14a shows
water situated south of the line L(X, F) = (identical to the
1000 m) which subducts
a
and enters the region of negative
Ekman
pumping. The water issued from the western boundary north of that line remains in the
subpolar gyre where it circulates cyclonically. As already noted by Talley (1985) it is
the line L(X, F) = 0 rather than that of vanishing
subtropical
We
which separates the subpolar and
gyres. Both these lines having been drawn on Figure
12a we observe,
Arhan: Subarctic Intermediate Water
1990]
133
.9
.7
®
.5
o
.4
.8
X
Figure 14. Wind-forced cases with the line F w(X) inclined in the western half of the basin. (a)
Isolines of H (in meters) giving the flow direction in layer 2, for HE = 1000 m, hE = 0 m, 1'1 =
2.5 \0-3 m s-2, 1'2 = 3.10-3 m S-2. (b) Isolines of H for HE = 2000 m, hE = 0 m, 1'1 = 5 10-3
m s-2, 1'2 = 1.510-3 m S-2. (c) Isolines of H + blh2)h giving the flow direction in layer I, for
case (b). Boundaries of the subducted SAIW are represented by dashed lines. (d) Isolines of H
for a case similar to (b) with a 3 106 m3 S-I source of MW added at the eastern boundary.
Isolines H = 1000 m in (a) and H = 2000 m in (b) and (d) define the line L(X, F) = O.
surprisingly enough, that it is between them, as in the model, that outcropped SAIW is
found. This suggests that the inclination of the line of vanishing we may be an
important factor governing the amount of SAIW present at the surface of the ocean.
This simple model is consistent with what was already inferred from Figure 12a, that
Journal of Marine Research
134
[48, I
the SAIW directly issues from the western boundary current of the subpolar gyre, i.e.
the Labrador Current. It has, indeed, the thermohaline characteristics of the cold and
fresh water transported southward by this current. The direction of the western
boundary currents is not reproduced by the model, and we have to accept that the
Labrador Current overshoots the line L(X, F) = O. This behavior is observed in the
real ocean. Such an origin for the SAIW differs from that of mode waters formed
through thermal convection in the ocean interior and is consistent with the observation
of high potential vorticity values in this watermass.
Considering the behavior of the subducted SAIW, however, the model is less
satisfactory. While the SAIW is expected to re-enter the western boundary between
35N and 40N at 43W, from Figure 12b, the model predicts a far greater southward
spreading to 20N. The only way to force the water of layer 2 to re-enter the western
boundary at higher latitudes is by increasing H, i.e. admitting that the water below
1000 m is not at rest. This we do by now assuming that layer 2 contains the SAIW and
the LSW, and modifying its depth and density accordingly to be HE = 2000 m and
P2 = 1.0277, with P3 = 1.02785. The flow now re-enters the western boundary in a
narrow region around 35N, or F = 0.8 (Fig. 14b). As was observed on Figure 12b it
encircles in the west a small region which we here assume at rest but must contain
water of subtropical properties, being adjacent to the subtropical western boundary
layer. At the opposite side subducted SAIW reaches the eastern boundary, although its
subduction was limited to X < .5.
We use this example to illustrate (Fig. 14c) the consequences on layer I of the flow of
subducted SAIW. The southward transport of SAIW, where it exists, causes through
the Sverdrup constraint a northward component in layer I. The direction change
observed at the western boundary of subducted SAIW is reminiscent of, though weaker
than, the one observed at about 35W on Figure I b. Streamlines west of that boundary
form an uncompleted northern gyre, open in the south. Comparison with Figure 14b
shows that closed streamlines within Worthington's northern gyre are more likely to be
found at intermediate
Streamlines
directly
levels where subducted
3600 m and 3800 m on Figure
enter the ocean interior
subtropical
recirculation
SAIW is present, than at the surface.
14c have a different
above subducted
SAIW.
behavior as they
They define a southern
limited to the western half of the ocean. The northernmost
streamlines on the same figure are clearly not satisfactory.
vanishes with we at the subduction
Moreover, the outcropping
The slope of the interface hx
line where in the real ocean we have a front.
line was imposed at the constant latitude F WE in the eastern
half of the ocean, so no water is seen flowing northward there contrary to reality.
Because of our choice of an outcropping line F] (X) the only water in motion in layer
2 is the water having subducted across the oblique part of F w(X) at longitudes X < .5.
Reality is quite different, as the SAIW and LSW flowing from the northwest certainly
interact with the MW issuing from the eastern boundary. Such interactions may
modify the circulation
pattern of these watermasses.
effects from this oversimplified
To give a hint of their possible
model we place in the third example
(Fig. 14d) an
Arhan: Subarctic Intermediate Water
1990]
F
F
1.1
1.1
FWE
FWE
.9
.9
135
3
~
WH
10-6ms-1
0
0
:t
.7
.7
.5
.5
@
a
a
.8
.4
.4
.8
X
Figure 15. Case with a positive WHo (a) Isolines of H + ('YJ'Y2)h (in meters) showing the flow
direction in layer I. (b) Isolines of H showing the flow direction in layer 2. The buoyancy
forcing WH is assumed nonzero, with the reported meridional distribution, in the shaded area
where subducted SAIW is present.
outflow of MW (3 106 m3s-1) at the eastern boundary between 34N and 40N. This
causes a more direct re-entry of the subducted SAIW into the western boundary. The
effect, though non-negligible, must in reality be far stronger, as part of the MW flows
northward, which it does not in the model, and comes in contact with SAIW.
e. Case with a positive
WH
We now assume that the line F
=
F w(X) is zonal throughout
the ocean at the
latitude F WE = 1.05 (i\ '" SON) and that flow across it is buoyancy driven. This line
being a characteristic, the sign of the zonal characteristic velocity, Uc gives the sign of
H x' hence the flow direction across this latitude. Luyten and Stommel (I 986a) call the
location where Uc changes sign at F = F WE' the Rossby Repellor. Examination of Eq. 5
shows that a southward flow of SAIW in layer 2, which we are looking for, may be
obtained either by a positive wH west of the Rossby Repellor where Uc > 0 or a negative
WH east of that point, where Uc < O. We successively study the two possibilities in this
subsection and the next one.
Positive values of WH, provided that they are higher than we' allow subduction of
SAIW in the subpolar gyre. One of the constraints imposed on the solution presented in
Figure IS is that subduction occurs in the western half of the basin. We considered that
positive
WH
values existed only over the area occupied by the subducted SAIW. the rest
of the ocean being devoid of buoyancy forcing. This distribution
was chosen for the
Journal of Marine Research
136
[48, I
simplifying purpose of having only the SAIW in motion in layer 2. It is also consistent
with the idea that positive wH in this region, probably due to dynamically active mixing
between watermasses, must have a distribution related to that of watermasses. As
SAIW crosses the line F = F WE west of about 25W (or X = .6) the Rossby Repellor
must be situated east of this longitude. The position of the Rossby Repellor is obtained
by setting the quantity in brackets equal to zero in the second of Eqs. 5. Replacing
aLI by its expression as function of We and using relation (9) we find the longitude of
the Rossby Repellor:
aF
(10)
As no buoyancy forcing exists in this example to change the interfacial depths east of
XR, hR and HR at the Rossby Repellor are equal to hEand HE' Coming back to a layer 2
containing only SAIW we have HE = 1000 m and 1'1 = 2.5 10-3 m S-2. Also, the
outcropping line needs no longer intersect the eastern boundary so that a nonzero value
of hE may be used, hE = 400 m, close to the thickness of the NACW. This places the
Rossby Repellor at X = 0.91, well to the east of where we want SAIW to cross the line
F= FWE'
The procedure followed to construct the solution is described in greater detail in the
Appendix. Briefly, another imposed parameter was the latitude at which SAIW
re-enters the western boundary (;\ ~ 35N, or F ~ .8), and the amplitude of the
buoyancy forcing was taken wH =10-6 m S-I, except north of FWE (Fig. 15) where it
had to be increased in order to have a subduction line with a positive slope dFjdX over
the domain 0 ~ X ~ 0.5.
Again, a part of the water from the subpolar western boundary current in layer 2
subducts (Fig. 15b) north of F WE and enters the subtropical region, while the
remainder circulates cyclonically in the subpolar domain. No subducted SAIW exists
east of X = .6 along F = F WE' This is a picture closer to reality than were those of
Figure 14 where SAIW reached the eastern boundary. The width of the southward
flowing tongue of SAIW is controlled by the intensity of the buoyancy forcing. The
value 1O-6m S-I, comparable to that of the maximum wind forcing, produces the
required
narrowness
observed on Figure 12b. A northward
flow is now necessary at
F = FWE in layer 1 (Fig. 15a) to keep the zero meridional Sverdrup transport where
SAIW is flowing southward.
northward
The model also reproduces
this feature, and the ensuing
flow of warm water in the eastern subpolar gyre. The rate of variation of h
along characteristics at the subduction is from (7) dhlds = -we + WHo The high wH
values imposed at the subduction zone act to maintain a sharp interfacial slope which is
reminiscent of the front associated with the northern NAC branch.
f
Case with a negative wH
Luyten and Stommel (l986b)
showed how a region of negative buoyancy forcing to
the east of the Rossby Repellor also generates
a southward
flow in layer 2. Provided
Arhan: Subarctic
1990]
Intermediate
Water
137
Z
1m,
h
2000
4000
5000
o
.4
.8
X
Figure 16. Case with a negative wH over the entire ocean width at F WE' (a) Function XR = f (hR)
along F WE' The variation of h along characteristics issued from the eastern and western
boundaries is also shown. (b) Profiles of h, Hand H + C"I,h2)h along F WE'
that the Rossby Repellor can be placed at about 35W (XR "" .3) this could explain the
flow of SAIW east of this longitude across/WE' However, a negative wH cannot account
for the subduction line in the subpolar region.
In the different cases studied by these authdrs as in those described above the Rossby
Repellor was situated either east or west of the region of nonzero wH, but not in it. In
this subsection we discuss the case where a negative wH is present throughout the width
of the ocean at F WE' and only consider the solution along this longitude. Figure 13b
suggests that such a distribution is expected if wH is to account for the heat losses to the
atmosphere.
Using (10) with our set of parameters, a value of XR around .3 can only be attained
by assuming as in the second and third examples of subsection 3d that layer 2 contains
both the SAIW and the LSW (HE = 2000 m, P2 = 1.0277, P3 = 1.02785). The exact
position of the Rossby Repellor, however, cannot be determined "a priori," because
nonzero values of WH between XE = I and XR now cause hR and HR to differ from hE
and HE' Using (6) to eliminate HR in Eq. (IO),XR is expressed as a function of hR only:
XR
=
I -
~'YlhR 3
7r/oawoF WE
[(H~ +
'YI
'Y2
h~ _
'YI
h1)1/2
-
hR]'
(II)
')'2
This relation was already discussed by Pedlosky (I984) and Schopp and Arhan (I986).
If valid over a finite ocean width, it constitutes the internal mode solution proposed by
these authors. Figure 16a displays the curve of Eq. (II) and the path followed from the
eastern boundary to reach a Rossby Repellor (RE) at X "" 0.35 on this curve, having
made choices of hE = 800 m and wH = -0.2 1O-6m S-I. The negative buoyancy
forcing causes h to diminish westward along F WE from (7). As one approaches the
Rossby Repellor the zonal interfacial slope dh/dX = wH/uc becomes infinite because of
138
Journal of Marine Research
[48, I
the vanishing Uc' Due to this infinite slope only the lower of the two solutions hR of Eq.
(II), situated on the upper part of the curve, can be attained.
Now considering the region to the west of RE, h diminishes eastward along the
characteristic from the western boundary. This characteristic, of positive uc' cannot
reach RE• If it could the slope dhjdX would similarly be infinite at RE, and the curve
h(X) would have to cross the region of negative uc' The only way we can complete the
solution west of RE is by letting Uc be zero from RE to the westernmost point Rw of the
curve ofEq. (11), which itself can be reached by the characteristic from the: west. Thus,
if a negative buoyancy forcing is active over the entire width of the ocean, a finite band
of longitude (Rw RE) must exist, in which the internal mode solution of Pedlosky
(1984) is valid, with Uc = 0 and a northward flow in layer 2.
Figure 16b shows the variations along FwEof h, H, and H + ('Ylh2)h which governs
the flow in the upper layer. Singularities exist at Rwand RE• Assuming that dissipation
there may develop internal boundary layers which make the solution acc;eptable, the
observed doming of the h curve at X = 0.35 is similar to the pattern present around
33W on the 48N IGY section (McCartney and Talley, 1982, their Fig. II). Water in
the upper layer is flowing southward west of this longitude as in Worthington's
northern gyre, and northward east of it. Such a vertical picture is consistent with the
map of Figure I b, but H + ('Y1I'Y2) h being greater at the western boundary than at the
eastern, more water is flowing southward west of the crest than northward east of it.
This suggests that part of the water at the upper levels is able to completely recirculate
within the northern gyre. The opposite zonal slopes of hand H east of the crest are also
observed in reality. However, west of the crest in the Newfoundland Basin isopycnal
slopes of the real ocean keep the same sign from surface to bottom.
5. Summary and discussion
a. Formation of Worthington's northern gyre
The inertial-frictional models of wind-driven subtropical oceans published by Bryan
(1963) and Veron is (1966) already showed the formation of a distinct gyre in the
northwestern corner of the basin provided that the nonlinearity was sufficiently strong.
Section 4 above describes a different formation process as it shows that the northern
gyre can be interpreted as a pure Sverdrup recirculation in the ocean interior. A
generally acknowledged advantage of inertial recirculation cells is to increase the
transport in western boundary layers toward more realistic values. That aspect was not
considered here.
We discussed in Section 4 the several possibilities offered by the ventilated thermocline model for cross-gyre communication. But there is another one which this model
cannot reproduce, namely communication through the western boundary layer. Csanady (1986) pointed out the importance of the Ekman flow entering the subtropical
domain across the lines of vanishing we at its northern and southern boundaries. This
inflow is of the order of 10 to 20106 m3 S-I for the North Atlantic. The compensating
]990]
Arhan: Subarctic Intermediate Water
139
outflow can only occur in the western boundary layer, and Csanady showed that it is
directed northward from the subtropical to the subpolar domain. Of particular interest
is the case when the western boundary current detaches from the coast and becomes an
interior jet flowing northeastward
into the subpolar region. Models of a detached
current have been much used since Parsons (1969) proposed a simple explanation for
this process. Veronis (1973) extended the model to the subpolar region and his results
for the Atlantic Ocean (his Fig. 17 and 18) also reveal anticyclonic recirculation in the
lower layer, the main subtropical gyre being of course in the upper layer. This is a
feature shared by the solutions in Figures 14 and 15 above. However Veron is' northern
gyre lies completely south of the zonal line of vanishing we' and no subduction of the
lower layer water occurs within it. The numerical experiments of Huang (1987), also
based on the original Parsons'model,
provide (his Fig. 2), flow patterns basically
similar to Veron is' solution, with some improvement on the above two points. The
northern gyre now captures some outcropped water north of the gyre boundary and
recirculates it after subduction in the subtropical region. Such a picture is qualitatively
similar to what was obtained with buoyancy forcing in Figure 15. What differs is that
the northern gyre in the solutions of Veronis (1973) and Huang (1987) is directly tied
to the presence of the separated Gulf Stream: the southwestward return flow in the
lower layer acts as a countercurrent of the strong interior current in the upper layer.
Models of a detached western boundary current are certainly relevant to the region
south of the Grand Banks to represent the Gulf Stream separation near Cape Hatteras.
But their applicability to the reigon west of the Grand Banks where the northern gyre is
observed is less obvious. There the NAC is flowing along the continental slope before
turning eastward at about 51 N as described in Section 2 (Fig. 1). There is no trace of a
Gulf Stream extension above the southward directed tongue of SAIW.
In Section 4 we investigated the role of other potentially important mechanisms
which may generate a northern gyre having the observed characteristics. An inclined
line of vanishing Ekman pumping makes the mass transfers across it easier. But
buoyancy forcing which is very active in this region is certainly a major contributor of
these cross-gyre exchanges. A positive wH at the northern NAC front is required for the
SAIW to subduct in the subpolar region. On the other hand if thermal fluxes at the
surface dominate WH and force it to negative values, then the Rossby Repellor of
Luyten and Stommel (1986a) should be placed at about 35W at the eastern limit of the
southward recirculating warm water within Worthington's northern gyre. East of this
point it is the cold subducted SAIW which is flowing southward as a rim of the gyre. In
this case the internal mode solution of Pedlosky (1984) was shown to complement the
buoyancy driven baroclinic flow to produce the required cross-gyre exchanges.
b. Watermasses and the North Atlantic Current
Section
4 focussed on modeling
the recirculation
associated
with Worthington's
northern gyre in the western half of the basin. However the description
suggests that a different
circulation
regime exists farther
of Section 2
east, with two branches
Journal of Marine Research
140
[48, I
escaping from the gyre somewhere above the western flank of the Mid-Atlantic Ridge.
This double-branch pattern was already suggested in HA, and further evidence of it
was provided here.
The transition between the two regimes in the upper layer is illustrated by the
temperature map of Figure I. How it occurs at the intermediate levels is not so
straightforward.
On the one hand we argued in Section 4 that SAIW completely
recirculates within the northern gyre, on the other hand we showed in Section 2 from
the TOPOGULF data that farther east subducted SAIW is found between the two
NAC branches (region 2). Figure 12b indeed shows the two branches characterized by
isotherms 5-6°C and 7-lOoC at 600 m, and subducted SAIW between them at
longitudes around 27W. Closer examination of this figure may help to solve the puzzle.
It seems that only a part of the subducted SAIW recirculates within the northern gyre
west of about 32W, while the remaining proceeds farther east between the two fronts.
Such a scheme would be consistent with several observations of small amounts of
subducted SAIW around 20W in the eastern basin (Harvey, 1982).
There are indications, both from the observations and the model, that the SAIW
directly originates in the Labrador Current. As the water transported by this current is
extremely fresh that would readily explain the low salinities of the SAIW. Another
cause is often invoked for the low salinities of subpolar intermediate waters, namely the
excess of precipitation over evaporation (P-E) prevailing at these latitudes. But
Schmitt et al. (1989) showed P-E to be approximately uniform (between 25 and 50
em/year) along SON across the Atlantic Ocean. As pure SAIW is found almost
exclusively in the western basin (Fig. 12a), one should perhaps favor the hypothesis of
an origin in the Labrador Current.
The quick erosion of the thermohaline anomalies of SAIW was pointed out by the
few oceanographers
who studied
this watermass.
Lateral
adjacent to it in the southeast is certainly important.
possibility,
namely
vertical
explain the freshening
mixing
with the MW
We studied in Section 3 another
mixing with the overlying
of the latter into a "modified"
NACW,
which could also
version of NACW
also found
between the NAC branches. Double diffusion was evoked as a possible mechanism
for
the vertical mixing. Conditions are favorable and the observed rotation of the 0 - S
diagram could be a consequence of salt-fingering.
Central
Water is known to be more saline in the eastern than the western North
Atlantic, at European latitudes. This was often attributed
Pollard and Pu (1985) rather suggest that greater
basin is mixing downward
the saltier surface
exchanges between NACW
and subducted
to the presence of MW but
winter convection in the eastern
waters. The above-discussed
vertical
SAIW provide another possible explana-
tion if, as suggested, SAIW present in the northwest is able to pump down salt from the
NACW.
Such downward
fluxes are more likely than upward fluxes above the MW,
and would similarly generate the relative high salinity anomaly of the eastern Central
Water.
1990]
Arhan: Subarctic Intermediate Water
o
.2
.4
.6
.8
141
1.0
X
Figure AI. Field of characteristics used to build the solution with positive wH in Subsection 4e.
Acknowledgments. The author was supported in this investigation by the IFREMER Grant
210110. The TOPOGULF hydrographic data used in the paper were gathered by the Institut fUr
Meereskunde of Kiel (R.F.A.) during cruises on R/V Poseidon in September 1983 and R/V
Meteor in August 1984 (Chief scientist Dr. J. Meincke). Useful discussions with J. Harvey and
the aid of K. Speer, I. Bodevin, J. Kervella and J. Le Gall in the preparation of the manuscript
are acknowledged.
APPENDIX
Procedure used to construct the solution with a positive
WH
The solution with wH> 0 presented on Figure 15 was constructed, rather artificially,
to reproduce some of the major features of the distribution of SAIW exhibited on
Figure 12:
• Subduction north of F WE in the western half of the basin.
• Southward spreading in the subtropical region as a narrow tongue and quick
return to the western boundary layer.
• Absence of subducted SAIW in the eastern North Atlantic east of about 20-25W.
The interfacial depths along the eastern boundary were chosen to be hE = 400 m and
HE = 1000 m, the approximate depths of the base of NACW and subducted SAIW.
The westernmost point of the subduction line (A) was placed (Fig. AI) on the western
boundary at the latitude FWE' There we have hA = 0 and HA = (H~ + ('Ylh2)h~)1/2.
142
[48,1
Journal of Marine Research
The eastern limit of this line (B) at Xs = .5 is characterized by hs = 0 and Hs = HE'
Its latitude is, from Eq. (6), solution of: 2 L(Xs, Fs) = - <'Ylh2)
The same relation
defining the northern part (BC) of the outcropping line, the longitude of point C at
h1.
Fe = 1.2 is readily obtained.
Along the characteristic
at F = F WE' numbered 1 on the figure, integration is
carried out eastward starting from A. The buoyancy forcing WH is assumed equal to the
maximum wind forcing Wo until the point D where H = HE is reached. East of this
point WH is set equal to zero. The eastern boundary of the subducted SAIW is then
defined as arcs of ellispes of equations:
and
where FE is chosen equal to .8 to force the subducted SAIW to re-enter the western
boundary north of about 35N.
Only layer 2 is in motion west of the outcropping line ABC: there characteristics
(nbs 2 to 5) and streamlines are identical and given by Eq. (6) with h set equal to zero.
East of the outcropping line in the subpolar region integration is carried out backward
from the latitude F = 1.2 along characteristics
6 to 10. The buoyancy forcing is
assumed zero east of BD, then given the meridional distribution displayed in Figure IS.
Points where h = 0 define the subduction line. In the subtropical region integration is
also carried out backward along characteristics
11 to 20 with again WH assumed zero
east of ED, then equal to Wo until H reaches the value HA• The line where this occurs is
the western boundary of the subducted SAIW, and WH is again assumed to vanish west
of it.
Luyten et al. (1983) showed a singularity
at outcropping
lines situated
above a
resting layer. This occurs here between Band C and at position A, where the interface
is vertical at reaching the surface. This singularity
is revealed by the two characteris-
tics arriving at C (from the one-layer regime, nb. 5, and from the two-layer regime, nb.
6) being slightly different.
The line F
by positive
=
WHo
Fw(X)
was supposed zonal in this example to bring out the part played
Had it been nonzonal as in Subsection 4d, the solution would have been
qualitatively similar, with the advantage of a better representation
of the gyres
common boundary. In that case, some characteristics issued from the subpolar western
boundary would enter the subtropical region.
REFERENCES
Arhan, M., A. Colin de Verdiere and H. Mercier. 1989. Direct observations of the mean
circulation at 48N in the Atlantic Ocean. J. Phys. Oceanogr, 19. 161-181.
Bryan, K. 1963. A numerical investigation of a non-linear model of a wind-driven ocean. J.
Atmos. Sci., 20.594-606.
1990]
Arhan: Subarctic Intermediate Water
143
Bubnov, V. A. 1968. Intermediate subarctic waters in the northern part of the Atlantic Ocean.
Okeanologia, 19,136-153. [English translation (NOOTrans 545), U. S. Naval Oceanographic
Office, Washington, DC 1973].
Bunker, A. F. 1980. Trends of variables and energy fluxes over the Atlantic Ocean from 1948 to
1972. Mon. Weather Rev., 108. 720-732.
Csanady, G. T. 1986. Cross-gyre transports. J. Phys. Oceanogr., 16. 1703-1711.
Dietrich, G. 1969. Atlas of the hydrography of the northern North Atlantic Ocean. Conseil pour
I'Exploration de la Mer, Service Hydrographique, Charlottenlund Slot, 357 pp.
Georgi, D. T. and R. W. Schmitt. 1983. Fine and microstructure observation on a hydrographic
section from the Azores to the Flemish Cap. J. Phys. Oceanogr.,13. 632-647.
Harvey, J. 1982. 0 - S relationship and water masses in the eastern North Atlantic. Deep-Sea
Res., 29, 1021-1033.
Harvey, J. and M. Arhan. 1988. The watermasses of the central North Atlantic in 1983/1984. J.
Phys. Oceanogr., 18,1855-1875.
Hebert, D. 1988. Estimates of salt-finger fluxes. Deep-Sea Res., 35. 1887-1901.
Huang, R. X. 1987. A three-layer model for wind-driven circulation in a subtropical-subpolar
basin. Part. II: the supercritical and hypercritical states. J. Phys. Oceanogr., 17, 679-697.
Iselin, C. 0'0. 1936. A study of the circulation of the western North Atlantic. Pap. Phys.
Oceanogr.,4, 101 pp.
Joyce, T. 1977. A note on the lateral mixing of water masses. J. Phys. Oceanogr., 7, 626-629.
Krauss, W. 1986. The North Atlantic Current. J. Geophys. Res., 91,5061-5074.
Krauss, W. E., E. Fahrbach, A. Aitsam, J. Elken and P. Koske. 1987. The North Atlantic
Current and its associated eddy field southeast of Flemish Cap. Deep-Sea Res., 34, 11631185.
Krauss, W. and R. Kase. 1984. Mean circulation and eddy kinetic energy in the eastern North
Atlantic. J. Geoph. Res., 89,3407-3415.
Krauss, W. and J. Meincke. 1982. Drifting buoy trajectories in the North Atlantic Current.
Nature, 296. 737-740.
Luyten, J. R., J. Pedlosky and H. Stommel. 1983. The ventilated thermocline. J. Phys.
Oceanogr., /3,292-309.
Luyten J. R. and H. Stommel. 1986a. Gyres driven by combined wind and buoyancy flux. J.
Phys. Oceanogr., 16.1551-1560.
--1986b.
Experiments with cross-gyre flow patterns on a Beta-plane. Deep-Sea Res., 33,
963-972.
McCartney, M. S. and L. D. Talley. 1982. The subpolar mode water of the North Atlantic
Ocean. J. Phys. Oceanogr., 12,1169-1188.
Parsons, A. T. 1969. A two-layer model of Gulf Stream separation. J. Fluid Mech., 39,511-528.
Pedlosky, J. 1984. Cross-gyre ventilation of the subtropical gyre: an internal mode in the
ventilated thermocline. J. Phys. Oceanogr., /4, 1172-1178.
Pollard, R. T. and S. Pu. 1985. Structure and circulation of the upper Atlantic Ocean northeast
of the Azores. Prog. Oceanogr., 14, 443-462.
Richardson, P. L. 1983. Eddy kinetic energy in the North Atlantic from surface drifters. J.
Geoph. Res., 88,4355-4367.
Schmitt, R. W. 1979. The growth rate of super-critical salt fingers. Deep-Sea Res., 26A, 23-40.
--1981.
Form of the temperature salinity relationship in the Central Water: evidence for
double-diffusive mixing. J. Phys. Oceanogr., 11, 1015-1026.
Schmitt, R. W. and D. T. Georgi. 1982. Finestructure and microstructure in the North Atlantic
Current. J. Mar. Res., 40 (Suppl.), 659-705.
Schmitt, R. W., P. S. Bogden and C. E. Dorman. 1989. Evaporation minus precipitation and
density fluxes for the North Atlantic. J. Phys. Oceanogr.,19. 1208-1221.
144
Journal of Marine Research
[48, 1
Schopp, R. and M. Arhan. 1986. A ventilated mid-depth circulation model for the eastern North
Atlantic. J. Phys. Oceanogr., 16,344-357.
Sy, A. 1988. Investigation of large-scale circulation patterns in the central North Atlantic: the
North Atlantic Current, the Azores Current, and the Medditerranean Water plume in the
area of the Mid-Atlantic Ridge. Deep-Sea Res., 35, 383-413.
Talley, L. D. 1985. Yentilation of the Subtropical North Pacific: the shallow salinity minimum.
J. Phys. Oceanogr., 15, 633-649.
Topogulf Group (The). 1986. TOPOGULF data report, Yol. 1: CTD, O2 and nutrients. Berichte
aus dem Institut flir Meereskunde, NR 154, Kiel, West Germany.
Veronis, G. 1966. Wind-driven ocean circulation. Part 2: numerical solutions of the non-linear
problem. Deep-Sea Res., 13, 31-55.
--1973.
Model of world ocean circulation. III - Thermally and wind driven. J. Mar. Res., 36,
1-44.
Worthington, L. Y. 1962. Evidence for a two gyre circulation system in the North Atlantic.
Deep-Sea Res., 9, 51-67.
--1976.
On the North Atlantic circulation. Johns Hopkins Oceanographic Studies, vol. 6,
The Johns Hopkins University Press, 110 pp.
Wiist, G. ]935. The strastosphere of the Atlantic Ocean. Scientific results of the German
Atlantic Expedition of the research vessel METEOR, 1925-1927. Yol. 6, section 1. English
translation, W. J. Emery, ed.
Received: 24 January, 1989; revised: 19 December, 1989