<#56#> Schmitz and McCartney<#56#>
[1993] review the state of our knowledge of the North Atlantic mean
circulation and construct flow patterns for the various parts of the water
column [see also <#57#> Reid,<#57#> 1994]. Their version of the upper layer
(Figure~1a,<#841#>#tex2html_wrap1802#<#841#> temperatures above 7#tex2html_wrap_inline1756# C)
updates many previous ones going back to <#59#> Worthington<#59#> [1976] and <#60#>
Stommel<#60#> [1957]. With respect to the former it repudiates assertions that
the ocean is not in geostrophic balance to leading order (i.e. horizontal
pressure gradients approximately balance the Coriolis force) and that the
subtropical gyre, the anticyclonic or clockwise part of the circulation
between 20#tex2html_wrap_inline1758# N and 40#tex2html_wrap_inline1760# N, is completely separated from the
subpolar gyre to the east of Newfoundland.
Within this scheme one of the best known numbers is the 30~Sv (1~Sv =
10#tex2html_wrap_inline1762#~m#tex2html_wrap_inline1764#/sec) of water transported by the Florida Current. This has
been determined from a variety of techniques including direct sampling of
the velocity field using shipboard techniques [e.g., <#61#> Schmitz and
Richardson,<#61#> 1968; <#62#> Leaman et al.,<#62#> 1987], voltage measurements
between ends of communications cables across the Florida Strait (see <#63#>
Larsen<#63#> [1992] for a comprehensive review) and pressure differences across
the Strait [<#64#> Mayer and Maul,<#64#> 1991]. Figure~2<#842#>#tex2html_wrap1804#<#842#>
from <#66#> Larsen<#66#> [1992] is a ten-year record of estimated volume
transport between Jupiter Inlet (Florida) and Settlement Point (Bahamas).
There is significant variability, much of it contained in an annual cycle,
with little evidence for any long-term trend over the period of
measurement.
This 30~Sv coming north in the Florida Current was once
believed to be the return flow of that water driven south and west by the
wind stress acting over the ocean interior [<#67#> Leetmaa et al.,<#67#> 1977].
However <#843#> <#68#>Wunsch<#68#> and Roemmich<#843#> [1985] questioned the
applicability of the Sverdrup relation (which relates the curl of the wind
stress to the depth-averaged flow) and <#69#> Schmitz and Richardson<#69#> [1991,
see also <#70#> Schmitz et al.,<#70#> 1993] demonstrate that approximately 13~Sv
originates in the South Atlantic and cannot be accounted for by the action
of North Atlantic winds. A resolution of this difficulty is given by <#71#>
Schmitz et al.<#71#> [1992]: previous estimates of the wind-driven component
failed to account for that part returned to the north, east of the
Bahamas, in the Antilles Current. Figure~1 shows about 12~Sv here,
although long-term current meter observations east of the Bahamas from the
STACS program (<#72#> Lee et al.<#72#> [1990]; T. N. Lee, W. E. Johns, R. J.
Zantopp, and E. Fillenbaum, Moored observations of western boundary
current variability and thermohaline circulation at 26.5#tex2html_wrap_inline1766# N in the
subtropical North Atlantic, submitted to <#871#> J.~Phys. Oceanogr.<#871#> , 1994; hereinafter referred
to as submitted manuscript, 1994) indicate a somewhat smaller northward
transport of 5~Sv. This flow is concentrated in the thermocline
(Figure~3),<#844#>#tex2html_wrap1806#<#844#> and appears to be primarily a
recirculation of 18#tex2html_wrap_inline1768# C water formed south of the Gulf Stream.
There is also evidence for a smaller anticyclonic gyre off the Bahamas
embedded in the southern reaches of the Gulf Stream recirculation
(Figure~4,<#845#>#tex2html_wrap1808#<#845#> <#75#> Olson et al.<#75#> [1984], <#76#> Lee
et al.<#76#> [1990]). Part of the long-standing uncertainty about the
existence of a mean ``Antilles'' Current off the Bahamas probably has to
do with the structure of this flow; it is predominantly a thermocline
rather than surface flow, and does not have a strong signature in surface
dynamic topography maps.
A summary of the subtropical western boundary circulation
(T. N. Lee et al., submitted manuscript, 1994) shows a total transport of 32~Sv in the Florida
Current at 27#tex2html_wrap_inline1770# N (of which 2~Sv is added north of Miami through
Northwest Providence Channel) and 5~Sv in the Antilles current, leading to
a total transport of 37~Sv. As 13~Sv is of South Atlantic origin, 24~Sv is
left to be balanced by southward wind-driven flow in the interior
Atlantic. The Sverdrup transport across 27#tex2html_wrap_inline1772# N is estimated at 26~Sv
giving closure to within uncertainty.
Perhaps the most dramatic new element of the general
circulation picture to emerge is the prevalence of smaller scale
recirculations, the aforementioned cell to the east of the Bahamas being
one example (Figure~1). Others include one to the south of the Gulf Stream
between 55#tex2html_wrap_inline1774# W and 75#tex2html_wrap_inline1776# W [<#77#> Worthington,<#77#> 1976] and
another to the north of the Gulf Stream, referred to as the Northern
Recirculation Gyre [<#78#> Hogg et al.,<#78#> 1986].
The discovery of the recirculation phenomenon has prompted
considerable theoretical work. Conjectures center on the notions that they
are either inertial (e.g., require no forcing, <#79#> Cessi<#79#> [1990], <#80#>
Marshall and Nurser<#80#> [1986], <#81#> Hogg and Stommel<#81#> [1985]) or that they
are forced through rectification of time-dependent motions [e.g., <#82#>
Holland and Rhines,<#82#> 1980; <#83#> Cessi et al.,<#83#> 1987; <#84#> Hogg,<#84#> 1988;
<#85#> Malanotte-Rizzoli et al.,<#85#> 1994]. With respect to the Northern
Recirculation Gyre, some progress has been made. <#86#> Hogg<#86#> [1993] has
shown that the eddy stresses are sufficient to drive realistic
recirculations, in a depth-averaged sense. Within the context of the
two-layer model, it is the weighted sum of the two layers that is made to
recirculate but the balance for the lower layer is much more inertial.
A necessary consequence of these recirculations is that the
downstream transport of the Gulf Stream is not constant but increases
steadily from the 30~Sv in the Florida Current to as much as 150~Sv
downstream of Cape Hatteras. Classical transport estimates have been
based on poorly supported guesses at where to place a reference level for
geostrophic calculations but in recent years this increase has been well
documented by long-term, direct measurement programs. This began with the
``Pegasus'' program near 73#tex2html_wrap_inline1778# W [<#87#> Halkin and Rossby,<#87#> 1985]. Some
16~velocity sections obtained over a 32-month period, when averaged in a
coordinate system aligned with the instantaneous Gulf Stream, gave a
downstream transport of 88~Sv. Farther downstream, large arrays of moored
current meters have given #tex2html_wrap_inline1780#~Sv at 68#tex2html_wrap_inline1782# W [<#88#> Johns et al.,<#88#>
1995] and approximately 150~Sv at both 60#tex2html_wrap_inline1784# W and 55#tex2html_wrap_inline1786# W
[<#89#> Hogg,<#89#>~1992].
These transport numbers are definition-dependent and meant
to quantify downstream transport in a coordinate system which moves with
the Gulf Stream itself. This is referred to as the average ``synoptic''
transport as it is closest to that inferred from shipboard surveys.
Estimates of the time-mean flow from moorings yield a Gulf Stream which is
much broader and weaker as meandering blurs the focus
(Figure~5)<#846#>#tex2html_wrap1810#<#846#> and, because of the reverse
recirculations, the transport estimates are lower, 88~Sv at 68#tex2html_wrap_inline1788# W
[<#91#> Johns et al.,<#91#> 1995] and 93~Sv at 55#tex2html_wrap_inline1790# W [<#92#> Hogg,<#92#>~1992].
Gulf Stream transport is traditionally separated into two
components, a ``baroclinic'' part referenced to some depth (typically
1000~m or the bottom) that is determined entirely from internal density
variations and a ``barotropic'' part arising from variations in sea
surface elevation. The former changes little downstream of Hatteras.
Referenced to 1000~m it is about 50~Sv all the way to 55#tex2html_wrap_inline1792# W [<#93#>
Johns et al.,<#93#> 1995; <#94#> Hogg,<#94#> 1992; <#95#> Hall and Fofonoff,<#95#> 1993]
reflecting the small change in the structure of the thermocline in this
region. Thus the transport change is mainly in the barotropic part
(Figure~6)<#847#>#tex2html_wrap1812#<#847#> which about doubles in size mostly
through a quasibarotropic inflow from the Northern Recirculation Gyre.
This inflow is not uniform [<#97#> Johns et al.,<#97#> 1995] but occurs
preferentially in regions where the Stream is cutting across isobaths most
sharply, consistent with an argument put forth by <#98#> Hogg and
Stommel<#98#>~[1985] based on the conservation of angular momentum (or
potential vorticity).
Several technical issues have been addressed successfully
by recent experience with moored arrays near the Gulf Stream. Firstly, the
use of fairing and large syntactic foam spheres for buoyancy so reduces
the drag that moorings can be usefully maintained in strong WBCs. With
instruments as shallow as 400~m from the surface in water depths greater than
4000~m the upper instruments dip downward less than 100~m on average [<#99#>
Tarbell et al.,<#99#> 1993]. Furthermore, <#100#> Hogg<#100#> [1991] has shown that this
residual ``mooring motion'' can be corrected for using the assumption that
there is a fixed structure, within WBCs, that is profiled as an instrument
is dragged downward in strong flows and as the current meanders across the
mooring. A related approach is used by <#101#> Hall and Bryden<#101#> [1985] and
refined by <#102#> Hogg<#102#> [1992] to compute the cross-stream structure and,
consequently, the integrated transport of the Stream from a single mooring
well instrumented in the vertical. <#103#> Johns et al.<#103#> [1995] compare this
technique with a more accurate one incorporating measurements from a full
cross-stream array and show that it does very well away from the edges of
the jet. Also, the use of Acoustic Doppler Current Profilers (ADCPs)
looking upward from the top of such moorings permits measurement of the
velocity structure to within tens of meters of the surface [<#104#> Johns et
al.,<#104#>~1995].
<#105#> Schmitz and Richardson's<#105#> [1991] landmark work on the
sources of the Florida Current prompted a renewed effort toward
understanding the pathways of upper
ocean thermohaline flow in the
Atlantic. This flow represents a substantial perturbation to the
wind-driven ocean circulation and therefore plays a fundamental role in
determining the structure of WBCs through the whole of the Atlantic [<#106#>
Onken,<#106#> 1994]. The interaction of this flow with the wind-driven
circulation in the tropical Atlantic was recently investigated in the
WESTRAX program [<#107#> Brown et al.,<#107#> 1992]. The North Brazil Current (NBC)
is the major WBC in the equatorial Atlantic and both closes the
anticyclonic wind-driven gyre straddling the equator and carries
thermohaline flow northward across the equator. Near 5--10#tex2html_wrap_inline1794# S the
annual mean NBC transport is approximately 24~Sv [<#108#> Stramma et al.,<#108#>
1990], increasing to 32~Sv at the equator [<#109#> Schott et al.,<#109#> 1993], and
decreasing again to about 23~Sv near 4#tex2html_wrap_inline1796# N [<#110#> Johns et al.,<#110#>
1993] where 3--5~Sv is found over the broad, shallow Brazilian shelf [<#111#>
Candela et al.,<#111#> 1992]. At each of these locations the mean transport is
15--20~Sv larger than that predicted from Sverdrup theory, indicating a
substantial thermohaline component in line with the accepted magnitude
[<#112#> Schmitz and Richardson,<#112#> 1991; <#113#> Hall and Bryden,<#113#> 1982; <#114#>
Roemmich and Wunsch,<#114#>~1985]. (Note that in Figure~1a the transport
schematic for the tropical region equatorward of 15#tex2html_wrap_inline1798# N is intended
to represent only the cross-equatorial thermohaline flow and not
recirculating gyre flows.)
The continuation of this flow northward along the western
boundary is interrupted by the cyclonic wind-driven tropical gyre centered
near 10#tex2html_wrap_inline1800# N, sandwiched between the anticyclonic equatorial and
subtropical gyres [<#115#> Mayer and Weisberg,<#115#> 1993]. According to Sverdrup
theory this gyre should have a southward boundary current with a mean
strength of 9~Sv, for which there is no evidence (op cit); rather,
shipdrifts and surface drifters indicate northward flow at the surface,
the so-called ``Guyana'' Current [<#116#> Richardson and Walsh,<#116#> 1986; <#117#> Richardson
and Reverdin,<#117#> 1987]. Plausibly, the thermohaline flow simply overpowers
the weaker wind-driven flow, leading to a net northward transport of 5~Sv.
Approximately 10~Sv of thermohaline flow from the
South Atlantic would then flow into the interior and supply the northward
interior Sverdrup transport in this gyre (Figure~1).
In reality the western boundary dynamics in this region are
a great deal more complicated than this simple reasoning might suggest.
There are large seasonal changes in the circulation associated with the
annual migration of the Intertropical Convergence Zone (ITCZ), and the
amount of water transported along the western boundary probably varies
considerably over the course of the year [<#118#> Philander and Pacanowski,<#118#>
1984; <#119#> Mayer and Weisberg,<#119#> 1993]. This transport is accomplished in
part by anticyclonic eddies that break off from the NBC where it
``retroflects'' into the interior at the rate of 3/year mainly in late
fall and winter [<#120#> Johns et al.,<#120#> 1990; <#121#> Didden and Schott,<#121#> 1993;
<#122#> Richardson et al.,<#122#> 1994]. The annualized mean northward transport
by these eddies of approximately 3~Sv is believed to be an important route
for water mass exchange between the North and South Atlantic. <#123#>
Richardson et al.<#123#> [1994] hypothesize that the Guyana Current is a
rectified signature of these eddies. Some evidence exists for an
undercurrent flowing equatorward beneath the Guyana current [<#124#> Schott
and Böning,<#124#> 1991; <#125#> Wilson et al.,<#125#> 1994] that may account for part
of the Sverdrup requirement of a southward wind-driven current closing the
tropical gyre. Further study in this region will probably be necessary to
understand the opposing influences of the wind and thermohaline forcing
that determine the net meridional transports.