The Southern Ocean Limb of the Global Deep ... - AMS Journals

JANUARY 2001

SLOYAN AND RINTOUL

143

The Southern Ocean Limb of the Global Deep Overturning Circulation* BERNADETTE M. SLOYAN Alfred Wegener Institute for Polar and Marine Research, Bremerhaven, Germany

STEPHEN R. RINTOUL Antarctic CRC, University of Tasmania, and CSIRO Division of Marine Research, Hobart, Tasmania, Australia (Manuscript received 22 June 1999, in final form 30 March 2000) ABSTRACT Nine hydrographic sections are combined in an inverse box model of the Southern Ocean south of ;128S. The inclusion of independent diapycnal flux unknowns for each property and air–sea (heat, freshwater, and momentum) fluxes make it possible to estimate the three-dimensional deep water circulation. The authors find a vigorous 50 3 10 6 m 3 s21 deep overturning circulation that is dominated by an equatorward flow of Lower Circumpolar Deep Water and Antarctic Bottom Water and poleward flow of upper deep water including Indian and Pacific Deep Water below 1500 dbar. In the subtropical Indian and Pacific Oceans the deep overturning cell is essentially isolated from the thermocline and intermediate waters of the subtropical gyre. The southward flowing upper deep water shoals south of the Antarctic Circumpolar Current, where air–sea fluxes convert outcropping upper deep water to Antarctic surface water and drive a net air–sea transformation of 34 3 10 6 m 3 s21 to lighter intermediate water. It is the outcropping of upper deep water and transformation by air–sea fluxes that connects the deep and intermediate circulation cells. The significant poleward transport of relatively light (i.e., above all topography at the latitude of Drake Passage) upper deep water, as required here to balance lateral and diapycnal divergence and air–sea exchange, provides observational evidence that advection by standing and transient eddies carries significant meridional transport in the Southern Ocean.

1. Introduction The overturning circulation refers to the flow of dense deep and bottom waters away from their sources and the compensating return flow of less dense upper-ocean water. The sources of deep and bottom water are limited to a few high latitude sites in the present-day ocean. In the North Atlantic, dense overflow water spills across numerous sills in the Greenland–Scotland Ridge. Subsequent mixing of this dense water in the northern Atlantic results in a prominent high salinity, relatively cold water mass, the North Atlantic Deep Water (NADW), that penetrates southward into the South Atlantic (Dickson and Brown 1994). Bottom waters are also produced at specific sites around the Antarctic continent, including the Weddell Sea, Ross Sea, and Ade´lie Land (Gill

* Alfred Wegener Institute for Polar and Marine Research Contribution Number 1782. Corresponding author address: Dr. Bernadette Sloyan, NOAA/ PMEL/OCRD Building 3, 7600 Sand Point Way NE, Seattle, WA 98115-6349. E-mail: [email protected]

q 2001 American Meteorological Society

1973; Carmack 1990; Fahrbach et al. 1994; Rintoul 1998; Wong et al. 1998). Newly formed Antarctic Bottom Water (AABW) mixes with overlying deep water as it spreads northward and fills the abyssal basins of the Atlantic, Indian, and Pacific Oceans (Orsi et al. 1993; Mantyla and Reid 1995). Much of the focus to date has been on the size and path of the NADW component of the deep circulation (Gordon 1986; Rintoul 1991; Saunders and King 1995; Macdonald and Wunsch 1996). On the other hand, Gordon (1975) estimated a total production rate of AABW of 35 3 10 6 m 3 s21 to 40 3 10 6 m 3 s21 , roughly twice that of NADW. Despite the fact that this early estimate, based on a simple heat budget calculation for the abyssal ocean, suggested a significant Southern Ocean role in the deep overturning circulation, attention has remained fixed on NADW. More recently, a number of studies have renewed interest in the Southern Ocean’s contribution to the global overturning circulation, including its influence on NADW production (Toggweiler and Samuels 1993a; Rahmstorf and England 1997; Goodman 1998; Toggweiler and Samuels 1998), and its role in closing the NADW cell by converting deep water to intermediate water (Doö¨s and Coward 1997). The sinking of NADW and AABW, and subsequent

144

JOURNAL OF PHYSICAL OCEANOGRAPHY

spreading into the abyssal basins, forms the lower limb of the overturning circulation. The upper-ocean return limb of the overturning flow is also important, particularly for estimates of the meridional transport of heat, salt, and other properties. Estimates of property transports to date have usually been confined to a single section or single basin (Rintoul 1991; Toole and Warren 1993; Saunders and King 1995; Robbins and Toole 1997; Tsimplis et al. 1998). Notable exceptions include the global analyses of Schmitz (1995), Macdonald and Wunsch (1996), Schmitz (1996a,b), and MacDonald (1998). Schmitz combined a large number of published transport estimates from regional studies in an effort to map out the large-scale circulation. Perhaps surprisingly, he found that the individual transport estimates were largely consistent with each other, allowing the construction of a global circulation scheme. Macdonald and Wunsch (1996) and Macdonald (1998) derived a global circulation scheme using an inverse model similar in spirit, but different in several important respects, to the model described here. Our picture of the overturning circulation in the Southern Ocean has changed little from that developed by Sverdrup and Deacon in the 1930s. They traced property extrema along meridional sections across the Southern Ocean and inferred a poleward transport of deep water, balanced by an equatorward transport split between denser bottom water and less dense intermediate water. To date, this schematic view of the Southern Ocean overturning circulation has yet to be quantified using observations, although a number of recent numerical modeling studies have focused on this issue. An important element in this circulation scheme is the diapycnal fluxes, driven by air–sea buoyancy forcing and interior mixing, which are needed to transfer water between density layers and close the overturning cell. These diapycnal transports must be included and quantified in any complete description of the overturning circulation in the Southern Ocean and its link to lower latitudes. Here we use an inverse model and hydrographic data to determine the three-dimensional circulation of the ocean south of about 128S. Two novel features of the model make it particularly appropriate for this task. By including independent unknowns for the diapycnal flux of each property, the model can successfully determine net diapycnal fluxes, as verified by applying the inverse model to output from a numerical model (McIntosh and Rintoul 1997; Sloyan and Rintoul 2000). In addition, the model explicitly includes air–sea fluxes of heat, freshwater, and momentum (Ekman) and the water mass transformations they drive. The inverse solution provides an estimate of the three-dimensional circulation of the Southern Ocean (with error bars), which is consistent with the simple model dynamics, the observed hydrography, and climatological estimates of air–sea fluxes (as corrected by the model). From this solution, we can determine the strength of the overturning circulation and quantify the relative contribu-

VOLUME 31

tions of air–sea fluxes and interior mixing to the diapycnal fluxes that close the cell. The purpose of this study is to describe the circulation of deep and bottom waters in each sector of the Southern Ocean. From these an estimate of the overall Southern Ocean overturning circulation is derived. Companion papers present other results from the inverse model: a discussion of the interior diapycnal fluxes is given in Sloyan and Rintoul (2000) and the formation and circulation of mode and intermediate waters is given in Sloyan and Rintoul (2001) The paper is set out in the following way. In section 2 we provide details of the hydrographic data used, a priori assumptions, and give a brief description of the design of the inverse model, including the treatment of air–sea forcing. The net mass, heat, and salt fluxes across each section are described in section 3. Section 4 discusses the deep circulation in each basin. The results from individual basins are drawn together in section 5 to provide an integrated description of the overall overturning circulation in the Southern Ocean. The conclusions are summarized in section 6. 2. Inverse model description a. Hydrographic data, domain, and layers Nine hydrographic sections are used to define six ‘‘boxes’’ in the Southern Hemisphere oceans, as shown in Fig. 1. The sections used are recent (1984–94) high quality hydrographic sections, apart from the Ind18 section, which was occupied in 1976 (Table 1). The sections across the Pacific, in Drake Passage, and south of Australia are part of the World Ocean Circulation Experiment (WOCE) dataset. Twenty-three layers (Table 2) are chosen to span the water masses in the model domain. The layers are defined by neutral density surfaces calculated using the Jackett and McDougall (1997) algorithm. Table 2 also gives the average potential temperature and salinity of each layer at each section. [Note that cold, fresh Antarctic Surface Water (AASW) lowers the average temperature and salinity in layers 13 to 15 at the Southern Ocean choke point sections, relative to the lower latitude sections.] The temperature and salinity data are used to determine the baroclinic or relative velocity between adjacent stations along the hydrographic sections. Numerous methods have been used to estimate the property transports in the bottom triangles below the deepest common depth of adjacent stations (Wunsch 1996). In this study the horizontal density gradient at the deepest common depth is held constant through the bottom triangle area of each station pair. The associated property transport is then included in the deepest layer at each station pair. The Ekman property fluxes are calculated normal to the station pairs along the hydrographic section. The depth of the Ekman layer is assumed to be 60 dbar for

JANUARY 2001

145

SLOYAN AND RINTOUL

FIG. 1. Position of hydrographic sections and box regions used in the inverse model. The 3500-m isobath and the major bathymetric features are shown. In the Atlantic sector Br: Brazil Basin, An: Angola Basin, Ar: Argentine Basin, Ca: Cape Basin, and WAB: Weddell abyssal plain; in the Indian sector Na: Natal Basin, Mo: Mozambique Basin, Mas: Mascarene Basin, Mad: Madagascar Basin, CI: Central Indian basin, WA: West Australian Basin, Pe: Perth Basin, En: Enderby Basin, Cr: Crozet Basin, SA: South Australian Basin, and AA: Australian–Antarctic Basin; and in the Pacific sector Ta: Tasman Basin, CP: Campbell Plateau, SWP: Southwest Pacific Basin, Ch: Chile Basin, and SEP: Southeast Pacific Basin.

all sections and the Ekman flux is calculated using Hellerman and Rosenstein (1983) annual mean wind stress. The Ekman property fluxes are calculated as a weighted mean from the sea surface to 60 dbar assuming an exponential decay of the Ekman velocity. The Ekman property fluxes are then added to the top layer of each station pair. Across the meridional sections (SAVE2, SAVE4, Ind18, Ind23, and Pac32) the initial reference level is the boundary between the northward flowing AABW and southward flowing North Atlantic Deep Water (NADW), Indian Deep Water (IDW), and Pacific Deep Water (PDW). In the Argentine basin of SAVE4 and the three Southern Ocean choke point sections (DrakeP, SAfrica, and SAust) the initial reference level is taken to be the deepest common depth at each station pair. b. Model constraints, weighting, and error estimates A box inverse model, following that of Wunsch (1978), is designed. Mass, heat, and salt are conserved in all layers while the silica constraint is imposed as a

total box constraint. Two novel aspects of the inverse model are the use of independent diapycnal flux unknowns for each property [mass (w*m ), heat (w*h ), and salt (w*s )], and the explicit inclusion of air–sea fluxes of heat, freshwater, and momentum (wind) and the water mass transformation they drive. The inclusion of independent diapycnal fluxes for each property represents the net diapycnal flux that results from all mixing processes that act to transfer mass, heat, or salt between water masses in the ocean interior. (Here ‘‘interior’’ is taken to mean the entire ocean beneath the sea surface. Diapycnal fluxes across isopycnals outcropping in the surface mixed layer, for example, are included in the interior diapycnal fluxes.) These mixing processes include advection, diffusion, cabbeling, and eddy fluxes. McIntosh and Rintoul (1997) and Sloyan and Rintoul (2000) show that inverse models using this representation of diapycnal fluxes can reproduce the ‘‘true’’ diapycnal fluxes when tested using output from a numerical model. In contrast, inverse models that use a single unknown w* (so that the diapycnal mass, heat, and salt fluxes are given by w*, w*T, w*S, where the

TABLE 1. The hydrographic sections (year and austral season) used in this study and abbreviated name used in the text. Sections SAVE2 128S SAVE4 308–478S Drake Passage AJAX—08E AJAX—Weddell Sea Indian 188S Indian 328S WOCE-SR3 1408E WOCE-P6 328S

Year (season)

Name

Reference

1988 (summer) 1989 (summer) 1990 (summer) 1984 (summer) 1984 (summer) 1976 (winter) 1987 (summer) 1994 (summer) 1992 (winter)

SAVE2 SAVE4 DrakeP SAfrica Wedsea Ind18 Ind32 SAust Pac32

SAVE (1992a) SAVE (1992b) Roether et al. (1993) Whitworth and Nowlin (1987) Whitworth and Nowlin (1987) Warren (1981) Toole and Warren (1993) Rosenberg et al. (1997) Tsimplis et al. (1998)

146


VOLUME 31

TABLE 2. The 22 neutral density surfaces (g n ) that bound the 23 layers and the water masses they define at the hydrographic sections. Also shown are the layer-average potential temperature (upper, 8C) and salinity (lower, psu) at each station. Asterisks indicate no layer present in section. Sections Layer 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

gn

SAVE2

SAVE4

DrakeP

SAfrica

Wedsea

Ind18

Ind32

SAust

Pac32

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

Top 22.0 22.0 22.5 22.5 23.0 23.0 23.5 23.5 24.0 24.0 24.5 24.5 25.0 25.0 25.5 25.5 26.0

26.59 36.307 26.53 36.872 24.52 36.766 22.77 36.603 19.86 36.217

20.68 35.449 20.12 35.659 17.528 35.489

26.0 26.5 26.5 26.8 26.8 27.0 27.0 27.2 27.2 27.4

15.93 36.626 12.75 35.195 10.54 34.944 8.40 34.724 6.00 34.519

13.64 35.009 10.42 34.661 8.54 34.532 6.50 34.382 4.39 34.254

8.72 34.055 7.77 34.062 5.94 34.052 4.44 34.048 2.86 34.059

12.77 34.859 9.56 34.479 7.86 34.404 6.00 34.293 3.37 34.140

27.4 27.6 27.6 27.8 27.8 28.0 28.0 28.1 28.1 28.2

4.31 34.472 3.87 34.652 3.57 34.895 2.73 34.906 2.09 34.857

3.18 34.302 2.78 34.485 2.67 34.727 2.23 34.809 1.52 34.766

1.55 34.146 1.86 34.387 2.06 34.639 1.66 34.721 1.12 34.720

28.2 28.3 28.3 28.4 28.4 28.5 28.5 Bottom

0.95 34.739

0.60 34.692 0.26 34.660

0.69 34.708

*

*

*

*

*

*

*

* * *

*

Water mass

26.10 34.858 25.01 34.844 23.65 34.984 22.15 35.110 20.73 35.220 19.26 35.375 17.61 35.457

22.49 35.294 20.95 35.441 19.72 35.645 18.13 35.696 15.61 35.554 12.65 35.170 9.90 34.777 7.71 34.567 5.61 34.432

12.28 34.880 9.97 34.544 7.01 34.273 5.06 34.155 2.88 34.075

13.82 34.978 10.99 34.753 8.70 34.549 6.79 34.406 5.22 34.346

Intermediate water (SAMW AAIW)

1.399 33.725 1.33 33.872

15.34 35.432 12.81 35.215 10.30 35.892 8.12 34.644 6.33 34.574

1.81 34.154 1.08 34.335 0.95 34.560 1.30 34.709 0.98 34.718

0.48 34.030 0.07 34.220 0.27 34.490 0.60 34.628 0.55 34.675

5.15 34.631 3.98 34.679 2.66 34.720 1.82 34.732 1.32 34.721

4.05 34.437 3.20 34.550 2.53 34.691 1.93 34.747 1.31 34.727

1.79 34.174 1.80 34.390 1.91 34.634 1.64 34.729 1.08 34.720

4.01 34.408 3.07 34.521 2.16 34.638 1.71 34.688 1.23 34.712

Deep water (IDW PWD NADW CDW)

0.50 34.693 20.27 34.658 20.44 34.652

0.12 34.670 20.20 34.658 20.45 34.647

1.02 34.714

0.82 34.701

1.06 34.708

*

*

0.38 34.691 20.03 34.676

*

*

*

*

*

*

*

*

*

*

20.89 35.522 19.37 35.418 17.27 35.337

* * * * * *

overbar represents the average on an isopycnal surface) (e.g., Wunsch 1984; Macdonald 1998), or which integrate the residuals in the mass conservation equations to infer the diapycnal flux (e.g., Wunsch et al. 1983; Roemmich and McCallister 1989), are unlikely to accurately represent the diapycnal fluxes (McDougall 1987; McIntosh and Rintoul 1997; Sloyan and Rintoul 2000).

* * *

Surface/thermocline water

19.25 34.906 18.13 35.016 16.26 35.054

*

Bottom water (AABW WSDW WSBW RSBW ALBW)

Exchange of heat and freshwater with the atmosphere results in net buoyancy forces that can transfer fluid from one density layer to another (Walin 1982). To date, inverse methods have not explicitly included these effects on the mass, heat, or salt conservation equations. Inverse models usually either downweight the conservation requirement in layers that outcrop and interact with the overlying atmosphere or only consider con-

JANUARY 2001

servation below the thermocline (Metzl et al. 1990; Rintoul 1991; Macdonald 1998). Rather than throw away the information contained in the conservation constraints for outcropping layers, it is desirable to include the effects of air–sea interaction explicitly. This is particularly so at high latitudes, where many layers outcrop. More importantly, the diapycnal fluxes driven by air– sea interaction are a fundamental part of the three-dimensional circulation of the ocean. An inverse model that seeks to determine this circulation should include these processes in a physically consistent manner. The effect of air–sea fluxes of heat, freshwater, and momentum can be included explicitly in the model following the strategy of Walin (1982), Tziperman (1988), Schmitt et al. (1989), and Speer and Tziperman (1992). The mass flux ( f m ) due to air–sea fluxes of heat and freshwater, and Ekman transport is fm 5

147

SLOYAN AND RINTOUL

2aH dr E 1 rbQS 2 r Q 1 . Cp dr

E EE dt

year

dA f md (r 2 r9),

(1)

(2)

area

where the d function samples the density flux at surface r. The mass convergence in a layer bounded by isopycnals r and r 1 dr is C m (r)dr 5 2 [F m (r 1 dr) 2 F m (r)].

(3)

In a similar manner, the convergence of heat and salt in density layers due to air–sea fluxes of heat, freshwater, and momentum can be found. The heat flux ( f h ) across outcropping density surfaces is fh 5

dr EuDe 2H 1 ; Cp r dr

n i, j

i,j

n51

i, j

1 rbQ i,n j S

p

]

2

2 Q i,n j 1 E i,n j P(r 2 r9)

O Dt O DA 12

Fh (r) 5

n51

i, j

3

i, j

[

(6)

]

Hi,nj 1 E i,n j uDei,j P(r 2 r9) Cp r (7)

O Dt O DA O [E S 12

Fs (r) 5

12

n i,j

i, j

i, j

De ij

]P(r 2 r9).

(8)

n51

The climatological data used for the air–sea fluxes are: heat flux (H) COADS by da Silva et al. (1994); wind stress (E) Hellerman and Rosenstein (1983) and; freshwater (Q) from GASP by Budd et al. (1995). We use Levitus and Boyer (1994) to find the outcropping position of the neutral density surfaces for each month, the weighted-mean Ekman layer temperature and salinity, and the annual mean depth and property concentration of the neutral density surfaces found within each box. The datasets used to calculate the air–sea fluxes contain errors, including errors in the bulk formulae used and a lack of oceanographic and meteorological observations, especially at high latitudes and over subtropical regions away from commercial shipping routes (Speer and Tziperman 1992; Barnier et al. 1995). As a result, we treat these climatologies as an initial guess in the inverse model and use the inverse solution to determine corrections to the climatological datasets, which make them consistent with the hydrographic data and the model physics. A complete description of the method, the derived corrections to the air–sea fluxes, and the water mass formation driven by the air–sea fluxes can be found in Sloyan (1997). The conservation equations, including the diapycnal flux unknowns for each property and the air–sea fluxdriven transformations, take the form

O Dx E N

(4)

h m11

j

j51

while the salt flux ( f s ) is fs 5

O Dt O DA [D1r 12aCH 12

Fm (r) 5

n51

In (1) r is the sea surface density, a thermal expansion coefficient, b saline contraction coefficient, H heat flux, Q freshwater flux (evaporation 2 precipitation 2 runoff ), S sea surface salinity, E the Ekman transport, and C p specific heat. Each of these terms is a function of time and position. The first two terms of (1) represent the effect of heat and freshwater fluxes on density, while the last two terms are direct input/loss of freshwater and the horizontal Ekman transport, respectively. The total mass flux across an isopycnal with density r, for one year, as a function of sea surface density is Fm (r) 5

mations across outcropping neutral density surfaces are summed over a discrete interval surrounding the outcropping density (Speer and Tziperman 1992). The delta function is now a boxcar (P) of width Dr:

(r Cj (y r 1 b) j dp) 1 E j r Cj 1 (w*A r C) m c

hm

2 (w*A r C) m11 1 (Fc 1 F*) c c m 2 (Fc 1 F*) c m11 dr ESDe , dr

(5)

where u De and S De are the weighted mean temperature and salinity over the Ekman layer (D e ). The mass, heat, and salt fluxes across each outcropping layer are calculated from climatological datasets. As the data are discontinuous, the property transfor-

5 0.

(9)

In (9) Dx j is the station spacing at pair j, C j is the property value/unit mass at pair j. The baroclinic or relative velocity (y rj ) is determined from the hydrographic data, E j C j is the Ekman property flux at pair j, and F c is the total flux across an outcropping isopycnal driven by buoyancy forcing. The subsequent system of

148


VOLUME 31

TABLE 3. A priori constraints (1ve–north/eastward) and uncertainties applied to the inverse model. The inverse model conserves mass, heat, and salt in all layers while silica is conserved in each box as a whole. Constraint 20.862 3 10 6 m 3 s21 462 3 10 6 m 3 s21 062 3 10 6 m 3 s21 1665 W m22 6065 3 10 6 m 3 s21 662 3 10 6 m 3 s21 1165 3 10 6 m 3 s21 Silica [O(1000 kmol s21 )]

Net southward transport at SAVE2 (Coachman and Aagaard 1988) Northward bottom water transport in Brazil Basin (Hogg et al. 1982; Speer and Zenk 1993) Net transport across Wedsea Heat loss over Weddell Sea (Gordon and Huber 1990; Fahrbach et al. 1994) Northward Malvinas Current (Peterson and Stramma 1991; Peterson 1992) Northward transport AABW across Argentine Basin (Whitworth et al. 1991) Northward transport LCDW/AABW coincident with WOCE PCM-9 (Pac32) (Whitworth et al. 1999) Conservation all regions (Tre`guer et al. 1995)

simultaneous equations is solved for the unknown reference level velocities (b), the diapycnal fluxes for mass, heat, and salt (w*m , w*h , w*s ) and the corrections to the air–sea climatologies (F*c ) (9). The importance of appropriate column and row weighting has been discussed by McIntosh and Rintoul (1997). The rows are weighted by the property norm and the columns are weighted by the area norm. A scaling factor is also applied to the reference velocity unknowns, the diapycnal property flux unknowns, and air– sea climatology corrections such that the condition number (the ratio of largest eigenvalue to smallest eigenvalue) is minimized. The salt conservation equations are expressed as salt anomaly, S 2 35.00. Errors due to the combination of nonsynoptic hydrographic sections lead us to choose a rank where the data residual norms are of O[1 to 2 (310 6 m 3 s21 )]. Table 3 contains information on the a priori constraints applied to the inverse model. Peterson and Whitworth (1989) and Peterson (1992) find large bottom velocities beneath the Malvinas Current that increase the transport substantially above that estimated from geostrophy relative to a deep reference level. This information is included by setting the initial choice of the reference velocity at the first six station pairs adjacent to the South American coast to 0.132 m s21 (13.2 cm s21 ), comparable to observations from near-bottom current meters in this region (Peterson and Whitworth 1989; Peterson 1992). Mooring array PCM-9 was deployed at 328S in the Pacific Ocean along a 1000-km line east of the Tonga–Kermadec Ridge (Whitworth et al. 1999). The mooring transport constraint is applied to deep and bottom water below neutral density surface g n 5 28.0, across the hydrographic stations that span the longitude range of the mooring array (1818E– 192.58E). Therefore, only 1000 km of the 6000-km wide Southwest Pacific basin is constrained. No prior constraint is placed on the transport of the Antarctic Circumpolar Current (ACC). Total silica conservation in each box is a weak constraint (large box imbalance is assumed), reflecting the uncertainty in the silica flux. The error bars presented are the sum of the noise and null space errors, which are calculated following Wunsch et al. (1983) and Rintoul (1991). These errors

represent the formal errors of the inverse method—the error associated with determining the barotropic velocity, and diapycnal fluxes. These formal errors do not take into account errors due to the asynoptic data, solution sensitivity to the first guess, and ocean variability. As the formal errors do not include these other error sources they are unlikely to reflect the true uncertainty in the estimates. In recognition of the inadequacy of the formal error bars, Macdonald and Wunsch (1996) and Macdonald (1998), for example, chose to increase the uncertainty of all their heat flux estimates by a somewhat arbitrary 60.25 PW [based on Holfort (1994) error analysis in the South Atlantic]. They believed this value more accurately accounted for the effect of (the unmeasured) oceanic variability. The value used is much larger than the formal error bar provided by the inverse method. Here we have chosen to present the estimates with the formal error bars so that it is clear what the error bars represent. If a good estimate exists of the uncertainty introduced by error sources not included in the formal error bars (e.g., Holfort’s value of 0.25 PW for the South Atlantic), this value should be added to the error bars shown here. Ultimately better assessments of the uncertainty in ocean flux estimates—however made—require measurements of ocean variability (and forcing), which do not yet exist for most of the ocean. One exception is the WOCE SR3 section south of Australia, which was occupied six times between 1991 and 1996, including each season. Sloyan (1997) tested the sensitivity of the inverse solution by rerunning the model using a realization of the SR3 section from a different year and different season (winter 1995). Overall, the changes were small (layer transport changes of ,1 3 10 6 m 3 s21 in almost all layers, and usually ,0.5 3 10 6 m 3 s21 , except at the SR3 section itself ). The main difference from the solution described below is an increase of 5 3 10 6 m 3 s21 in the eastward flow south of Australia, which is balanced by increased northward flow in the Pacific, increased Indonesian Throughflow, and increased southward flow in the Indian Ocean. The net heat flux across roughly 308–408S (SAVE4, Ind32, and Pac32) is almost unchanged (difference of 0.02 PW), despite the increase in strength of the circum-Australia

JANUARY 2001

149

SLOYAN AND RINTOUL

FIG. 2. Integrated fluxes of mass (bold: 3 10 6 m 3 s21 ), heat (dashed: PW, relative to O8C), and salt (solid: 3 10 9 kg s21 ) across each hydrographic section.

flow. The pattern and magnitude of the lateral and diapycnal fluxes—in particular, the strength and structure of the deep overturning circulation, the subject of this paper—are similar in both models. While using different realizations of a single section clearly does not provide a complete assessment of the impact of oceanic variability on our circulation estimates, this sensitivity test and those described in the appendix supports the assertion that robust estimates of the large-scale circulation and fluxes can be made from models such as the one used here. 3. Net meridional and zonal fluxes In the following discussion we give a brief description of the net mass, heat, and salt fluxes across each section. The overall circulation generally agrees with our current understanding of the circulation of the Southern Ocean and adjacent basins. We also describe the corrections to the climatological air–sea heat and freshwater flux fields required to ensure consistency between the forcing fields and the hydrography. a. Mass transport The integrated mass, heat, and salt fluxes across the hydrographic sections are shown in Fig. 2 and Table 4. The individual layer mass fluxes for each section are given in Table 5.

In the Southern Ocean the Antarctic Circumpolar Current dominates the three choke point sections with an eastward mass flux of 134.961 3 10 6 m 3 s21 at Drake Passage increasing to 145.661.2 3 10 6 m 3 s21 south of Australia. Across the Atlantic SAVE2 and SAVE4 sections there is a small southward mass flux of 20.7762.9 3 10 6 m 3 s21 and 20.8062.7 3 10 6 m 3 s21 . This corresponds to the leakage of Pacific water through Bering Strait and into the North Atlantic Ocean. In the Indian Ocean there is a southward mass flux at 188S and 328S of 210.463.1 3 10 6 m 3 s21 and 210.362.3 3 10 6 m 3 s21 , respectively. This southward mass flux through the Indian Ocean is within reasonable agreement of recent estimates of the size of the Indonesian Throughflow (Cresswell et al. 1993; Meyers et al. 1995). The net southward flux from the Indian Ocean results in a net increase in flow south of Australia, accounting for the mass flux increase between south of Africa and Australia. At 328S in the Pacific Ocean there is a northward mass flux of 7.462 3 10 6 m 3 s21 . [The small mass imbalances in each box result from choosing a solution (rank) where the layer residual norms are of O(1 to 2 3 10 6 m 3 s21 ). The imbalance does not represent a convergence of freshwater. The freshwater convergence for each box is given by the air–sea freshwater flux and is discussed below.] The mass fluxes across each of the sections (Fig. 2) are in reasonable agreement with previous studies (Rintoul 1991; Toole and Warren 1993;

TABLE 4. Total section property fluxes (1ve–northward/eastward). Heat fluxes are relative to 08C and converted to PW (1 PW 5 244 3 10 6 8C m 3 s21 ). Units: mass 3 10 6 m 3 s21 ; heat PW 3 1015W; salt 3 10 6 kg s21 ; freshwater 3 10 9 kg s21 ; silica kmol s21. Section SAVE2 SAVE4 Drake P S. Africa Weddell Sea Ind18 Ind32 S. Aust. Pac32

Mass 20.77 20.80 134.9 135.4 20.35 210.4 210.3 145.6 7.4

6 6 6 6 6 6 6 6 6

Heat 2.9 2.7 0.75 2.7 2.5 3.1 2.3 1.0 1.8

0.42 0.28 1.41 1.19 20.05 21.27 20.87 1.83 0.13

6 6 6 6 6 6 6 6 6

Salt 0.05 0.04 0.01 0.03 0.01 0.05 0.06 0.01 0.04

228.3 227.6 4785.9 4795.1 213.7 2371.7 2369.1 5164.4 261.1

6 6 6 6 6 6 6 6 6

Freshwater 29.5 55.5 14.1 51.7 57.3 56.3 43.9 18.8 43.7

20.85 20.83 134.2 134.5 20.35 210.3 210.2 144.7 7.32

6 6 6 6 6 6 6 6 6

3.0 2.9 0.83 2.8 2.6 3.0 2.4 1.2 2.0

Silica 124 295 9071 8450 226 261 425 8017 821

6 6 6 6 6 6 6 6 6

194 339 73 407 339 679 316 99 283

0.01 0.04 0.06 0.08 0.07 0.10 0.13 0.14 0.18 0.30 0.40 0.47 1.15 2.03 1.48 0.62

* * * * 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 * * *

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

20.24 21.48 20.79 2.55 3.25 1.97 2.32 1.95 1.79 1.73 0.94 20.49 29.79 29.69 1.79 3.40

SAVE2

Layers

0.12 0.94 0.59 0.88 3.36 2.48 1.28 20.39 21.53 21.96 23.23 210.70 0.74 5.49 1.11

* * * * * * 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 * * 0.01 0.03 0.04 0.14 0.16 0.24 0.20 0.47 0.48 0.56 1.24 1.40 1.14 1.22 0.51

SAVE4

1.13 1.40 2.30 9.34 18.51 19.29 19.48 30.45 15.98 12.90 4.18

* * * * * * * * * 6 6 6 6 6 6 6 6 6 6 6 * * * 0.01 0.01 0.01 0.07 0.11 0.11 0.13 0.32 0.32 0.38 0.42

DrakeP

0.05 21.01 20.35 20.36 1.85 4.61 6.62 15.31 18.89 19.33 33.19 22.54 18.25 1.47 24.31 20.67

* * * * * * 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 * 0.00 0.01 0.01 0.06 0.15 0.13 0.15 0.32 0.35 0.45 0.88 1.25 1.31 1.17 1.16 0.45

SAfrica

0.41 20.15 20.21 0.66 21.48 21.24 22.34 27.47 1.72 9.74

* * * * * * * * * * * * 6 6 6 6 6 6 6 6 6 6 * 0.01 0.03 0.04 0.06 0.20 0.26 0.53 1.62 1.82 0.55

Wedsea

21.66 211.73 0.50 0.03 0.71 0.08 0.12 0.33 20.04 0.80 2.04 20.84 21.70 21.36 21.24 23.21 5.64 1.11

* * 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 * * *

Ind18

0.03 0.08 0.05 0.06 0.06 0.06 0.08 0.11 0.13 0.22 0.37 0.21 0.41 0.61 1.02 1.36 2.48 0.04 20.20 20.11 23.23 22.03 21.53 2.33 21.11 23.67 24.35 28.52 25.67 25.61 8.71 4.23 10.40

* * * * * 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 * * *

Ind32

TABLE 5. Layer mass (310 6 m 3 s21 ) fluxes for each section (1ve–northward/eastward).

0.00 0.00 0.03 0.12 0.41 0.46 0.37 0.36 0.26 0.44 0.54 0.92 1.11 1.19 0.85

0.02 3.75 14.74 12.38 15.61 15.88 16.37 28.18 21.72 13.12 5.40 21.56

* * * * * * * * * 6 6 6 6 6 6 6 6 6 6 6 6 * *

SAust

0.01 0.04 0.17 0.09 0.09 0.13 0.16 0.40 0.44 0.51 0.46 0.30

21.24 0.28 9.07 0.43 20.73 21.20 1.57 21.61 23.55 28.42 213.46 20.37 18.56 8.02

* * * * * * 6 6 6 6 6 6 6 6 6 6 6 6 6 6 * * *

Pac32

0.05 0.07 0.12 0.14 0.11 0.13 0.33 0.27 0.25 0.41 1.13 0.88 0.77 0.28

150 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 31

JANUARY 2001

SLOYAN AND RINTOUL

Saunders and King 1995; Macdonald and Wunsch 1996; Robbins and Toole 1997; Rintoul and Bullister 1999). b. Western boundary current transports The western boundary currents are reproduced in the inverse solution. In the Atlantic, at 128S (SAVE2) the weak, shallow beginnings of the Brazil Current (22.560.4 3 10 6 m 3 s21 ) are found. In the Argentine Basin the strong northward Malvinas Current (57.763.2 3 10 6 m 3 s21 ), resulting from the bottom velocities imposed there in the initial guess, is maintained. The Malvinas constraint concentrates the southward NADW transport to the Brazil Basin at SAVE2, and increases the southeastward flow of intermediate and deep water over the mid-Atlantic Ridge at SAVE4 (Sloyan 1997). Adjustments to the circulation in the Atlantic sector in response to the Malvinas constraint also affect the Indian and Pacific Oceans, resulting in a 2 3 10 6 m 3 s21 decrease in the net southward, eastward, and northward transport across the Indian Ocean, south of Australia and at 328S in the Pacific, respectively. In the Indian Ocean at 188S the East Madagascar Current transport is 228.165 3 10 6 m 3 s21 ; previous estimates vary from 241 3 10 6 m 3 s21 to 220 3 10 6 m 3 s21 (Lutjeharms et al. 1981; Schott et al. 1988; Swallow et al. 1988). At 328S in the Indian Ocean the Agulhas Current transport is 248.366 3 10 6 m 3 s21 within 400 km of the African coast. This is composed of 252.3 3 10 6 m 3 s21 southward flow above 2000 dbar and 4 3 10 6 m 3 s21 northward below 2000 dbar. Using the same section Toole and Warren (1993) estimate a topto-bottom transport of 279.5 3 10 6 m 3 s21 (280.7 3 10 6 m 3 s21 southward above 2000 dbar and 1.2 3 10 6 m 3 s21 northward below 2000 dbar) and a net southward transport of 285 3 10 6 m 3 s21 was found within 400 km of the African coast. At the same 328S section, Robbins and Toole (1997) and Macdonald (1998) estimate an Agulhas Current of 290 3 10 6 m 3 s21 and 29362 3 10 6 m 3 s21 , respectively. Estimates from a Lowered Acoustic Doppler Current Profiler (ADCP) within 230 km of the African coast show a volume transport of 271 3 10 6 m 3 s21 (Beal and Bryden 1997). The LADCP data show a V-shaped zero velocity surface (ZVS) with a maximum depth of 2500 m some 80 km offshore (Beal and Bryden 1997). Inshore there is evidence of a northward undercurrent close to the continental slope. In the Natal Basin Toole and Warren (1993) use a zero velocity surface of 2000 m, which increases to 3000 m in the deepest part of the Natal Basin. [Robbins and Toole (1997) and Macdonald (1998) employ the Toole and Warren (1993) zero velocity surface.] Beal and Bryden (1997) find that the difference between their Agulhas transport and that of Toole and Warren (1993) is due to their use of the V-shaped ZVS. In this study all ZVSs are neutral density surfaces that may deepen or shoal across an ocean basin. In the Natal Basin the ZVS is neutral density surface

151

g n 5 28.0; at station pairs where this layer is not found a deepest common ZVS is used. Here g n 5 28.0 shows a similar, although not as sharp, V-shaped ZVS to that of Beal and Bryden (1997)—it is above 2000 dbars adjacent to the African coast; in the middle of the Natal Basin it reaches a maximum depth of 2500, and then shoals slightly before ‘‘grounding’’ at the Mozambique Ridge. The different ZVS employed in this study may partly explain the smaller Agulhas Current compared to those of Toole and Warren (1993), Robbins and Toole (1997), and Macdonald (1998). A more significant factor that affects the strength of the Agulhas Current is the difference in the vertical extent of the deep overturning circulation between our studies and those previously quoted. In Toole and Warren (1993), Robbins and Toole (1997), and Macdonald (1998) northward flowing deep and bottom water is balanced by southward flow of intermediate and thermocline water in the Agulhas Current. Our deep overturning cell is confined to below 1500 dbars and the balancing southward flow occurs east of 908E, as discussed below. In the Pacific Ocean at 328S we find a relatively weak East Australian Current of 221.362 3 10 6 m 3 s21 , which agrees with recent estimates (Ridgway and Godfrey 1994; Macdonald 1998). c. Heat transport All heat fluxes are temperature fluxes, relative to 08C, and scaled to PW, where 1 PW ([31015 W) 5 244 3 10 6 8 C m 3 s21 . Across SAVE4 (nominally 378S) there is a northward heat flux of 0.28 6 0.04 PW. This estimate agrees with that of Rintoul (1991) but is slightly smaller than more recent estimates of 0.49 6 0.25 PW by Macdonald (1998) and 0.5 6 0.1 PW by Saunders and King (1995). The northward heat flux at SAVE2 is 0.42 6 0.05 PW. A heat input of 0.14 PW (positive heat flux represents heat gain by the ocean) in the subtropical Atlantic results from the COADS heat fluxes and the inverse corrections (;10%) (Fig. 3). This heat input is larger than Talley (1984) but similar to estimates of Trenberth and Solomon (1994). The inverse model of Macdonald (1998) implies an even larger heat input in the subtropical Atlantic, the result of weak heat loss (20.16 6 0.25 PW) between 278 and 238S and strong heat gain between 238 and 118S (0.56 6 0.25 PW). In the subtropical Indian Ocean the heat flux is 21.27 6 0.05 PW at 188S and 20.87 6 0.06 PW at 328S. The large southward heat flux at 188S is the result of both heat gained by the ocean in the northern Indian Ocean and heat entering from the Pacific Ocean via the Indonesian Throughflow. At the same sections Macdonald (1998) estimates a southward flux of 21.45 6 0.30 PW and 21.30 6 0.28 PW, respectively, while at 328S Toole and Warren (1993) estimate a southward flux of 21.67 PW. These previous heat flux estimates are significantly larger than found in this study, although all studies have a similar sized Indonesian Throughflow of ;10 3 10 6

152


VOLUME 31

FIG. 3. COADS heat fluxes, correction to the fluxes obtained from the inverse solution, and corrected heat fluxes for outcropping layers in the (a) subtropical Atlantic (Box I), (b) Southern Ocean Atlantic sector (Box II), (c) subtropical Indian (Box IV), (d) Southern Ocean Indian sector (Box V), and (e) Southern Ocean Pacific sector (Box VI).

m 3 s21 . Our smaller southward heat flux results from the reduced vertical extent of the deep and bottom overturning cell when compared to the previous studies, as discussed in detail in the following section. In the subtropical Indian Ocean between 188 and 328S the air–sea fluxes (COADS plus inverse correction) result in an oceanic heat loss of 20.40 PW. This is similar to estimates of Trenberth and Solomon (1994) but larger

than estimates of Talley (1984) and the 20.15 6 0.30 PW found by Macdonald (1998). At 328S in the Pacific Ocean there is a northward heat flux of 0.13 6 0.04 PW. Macdonald (1998) estimates a northward heat flux of 0.26 6 0.28 PW at 438S and a small southward heat flux of 20.04 6 0.30 PW at 288S. Wunsch et al. (1983), using the same 438S and 288S section as that of Macdonald (1998), estimate a south-

JANUARY 2001

SLOYAN AND RINTOUL

ward flux of 20.03 6 0.35 PW and 20.18 6 0.22 PW, respectively. [Note, however, that Wunsch et al. (1983) set the throughflow to zero.] As the circulation of the Indian and Pacific Oceans are linked via the Indonesian Throughflow it is more appropriate to consider the combined heat flux of these oceans. At 328S the southward Indian–Pacific heat flux is 20.73 6 0.10 PW. Including the heat flux at SAVE4 provides us with an estimate of the net meridional heat flux at nominally 358S (SAVE4, Ind32, and Pac32) of 20.46 6 0.14 PW. A similar calculation by Macdonald (1998) at 308S gives an estimated southward heat flux of 20.9 6 0.3 PW. Across the Southern Ocean choke point sections the eastward heat fluxes are 1.41 6 0.01 PW at Drake Passage, 1.19 6 0.03 PW south of Africa, and 1.83 6 0.01 PW south of Australia. [The reader is reminded that the errors quoted represent only the formal error and therefore may not reflect the true uncertainty in the estimates (see end of section 2b).] In the Southern Ocean Atlantic sector there is also a southward flux of 20.05 6 0.01 PW into the Weddell Sea. The heat fluxes at Drake Passage and south of Africa are similar to estimates of Macdonald (1998) while south of Australia there is a 0.10 PW difference in the estimated heat flux. The inverse corrected air–sea heat fluxes show that the Atlantic sector (Box II) gains 10.11 PW from the atmosphere. This heat gain and the lateral heat flux through Drake Passage balance the ocean heat export south of Africa, into the Weddell Sea, and northward across SAVE4. In the Southern Ocean Indian sector the corrected air–sea fluxes indicate a heat loss from the ocean to the atmosphere of 20.22 PW. This heat loss is more than compensated by the southward heat flux across 328S in the Indian Ocean. In the Pacific sector the ocean loses 20.31 PW to the atmosphere. The ocean–atmosphere heat loss and the northward ocean heat flux at 328S balance the difference between the eastward heat flux south of Australia and Drake Passage. The air–sea heat fluxes in the Southern Ocean Indian and Pacific sectors in this study are similar to those found by Georgi and Toole (1982). The similarities in the estimated air–sea heat flux result in part from the use of the same observation data between the two studies; that is, COADS is derived from the heat flux estimates used by Georgi and Toole (1982). In the Atlantic sector Georgi and Toole (1982) estimate an ocean to atmosphere heat loss of 20.16 PW. This ocean heat loss is estimated from the integration of air–sea heat fluxes of Taylor et al. (1978) and Gordon (1981) south of 408S. The difference in the air–sea heat flux in the Southern Ocean Atlantic sector between our study and Georgi and Toole (1982) results in part from our use of more recent estimates suggesting smaller heat loss [216 W m22 (Gordon and Huber 1990; Fahrbach et al. 1994)] over the Weddell Sea. In the Southern Ocean Macdonald (1998) estimates a larger air–sea ocean heat gain in the Atlantic sector

153

of 10.3 6 0.1 PW, a larger heat loss in the Indian sector of 20.7 6 0.4 PW, and a smaller heat loss in the Pacific sector of 20.1 6 0.3 PW. The most significant difference in the estimated air–sea heat flux between Macdonald (1998) and this study occurs in the Southern Ocean Indian sector. This difference may be linked to the difference in southward heat flux at 328S in the Indian Ocean. Macdonald (1998) has a large southward ocean heat flux across 328S in the Indian Ocean. To reconcile this large heat input into the Southern Ocean Indian sector with the heat divergence in the ACC between Australia and Africa, a large air–sea ocean heat loss is required. In our model the smaller southward heat flux across 328S is compatible with the ACC heat divergence between Australia and Africa and the COADS heat flux (Fig. 3). The most significant difference between this study and the earlier study of Georgi and Toole (1982) is the calculation of ocean heat flux across the three choke point sections. Georgi and Toole (1982) assume that the ACC mass transport is constant at 127 3 10 6 m 3 s21 at all choke points, and effectively solve a heat balance assuming no Indonesian Throughflow. In our model the ACC transport at Drake Passage and south of Africa is approximately 13562 3 10 6 m 3 s21 while south of Australia the ACC transport increases to 14561 3 10 6 m 3 s21 as required to offset the 1063 3 10 6 m 3 s21 Indonesian Throughflow. The differences in mass transport result in significantly different ocean heat fluxes at the choke point sections, and hence in the meridional heat fluxes across 408S inferred from the choke point heat transport divergences. d. Salt, freshwater, and silica transport The net salt flux in the Atlantic at SAVE2 and SAVE4 is 228.3629.5 3 10 6 kg s21 and 227.6655.5 3 10 6 kg s21 , respectively. This salt flux is in reasonable agreement with the salt flux of 26.7 3 10 6 kg s21 from the Pacific Ocean to the Atlantic through Bering Strait (Wijffels et al. 1992). (The net salt fluxes include the salt flux associated with a nonzero mass flux at a section and the freshwater divergence in the box.) The import of salt from the subtropical Atlantic increases the salt flux south of Africa relative to Drake Passage. In the subtropical Indian Ocean there is a net southward salt flux associated with the input of Pacific water into the Indian Ocean via the Indonesian Throughflow (Table 4). The input of salt from the Indian Ocean results in a larger eastward salt flux south of Australia than south of Africa. The import of salt into the Southern Ocean in the Atlantic and Indian sector is balanced by a northward flux of salt in the Pacific Ocean across 328S. The air–sea freshwater fluxes from inverse-corrected GASP data (Fig. 4) show that the subtropical Atlantic (Box 1) and subtropical Indian (Box IV) Oceans lose 20.34 3 10 9 kg s21 and 20.33 3 10 9 kg s21 of freshwater via evaporation while the Southern Ocean Atlantic

154


VOLUME 31

FIG. 4. GASP freshwater fluxes, correction to the fluxes obtained from the inverse solution, and the corrected freshwater flux for outcropping layers of (a) subtropical Atlantic (Box I), (b) Southern Ocean Atlantic sector (Box II), (c) subtropical Indian (Box IV), (d) Southern Ocean Indian sector (Box V), and (e) Southern Ocean Pacific sector (Box VI).

(Box II), Indian (Box V), and Pacific (Box VI) sectors gain 0.01 3 10 9 kg s21 , 0.20 3 10 9 kg s21 , and 0.33 3 10 9 kg s21 of freshwater, respectively. The freshwater convergences for each region are in good agreement with the integrated Baumgartner and Reichel (1975) freshwater fluxes (Wijffels et al. 1992). Perhaps surprisingly, only small corrections to the air–sea freshwater flux are needed to make the air–sea fluxes and

hydrography consistent with the model constraints (Fig. 4). The silica flux is dominated by large eastward fluxes at the Southern Ocean choke point sections (Table 4). The net northward silica flux at 328S in the Indian Ocean of 424.7 6 316.4 kmol s21 is significantly smaller than the northward flux of 1532 kmol s21 estimated by Toole and Warren (1993), but is larger than the estimate of 21

JANUARY 2001

SLOYAN AND RINTOUL

kmol s21 by Robbins and Toole (1997). Independent estimates of the global silica budget suggest that the Southern Ocean is a net sink for silica (DeMaster 1981; Tre`guer et al. 1995). Therefore, the size of the northward silica flux at 328S in the Pacific of 821.4 6 283 kmol s21 and the large net meridional flux of silica across 308–408S are of concern and led to additional experiments as described in the Appendix. The transports of mass and other properties in the standard model and the silica experiments are the same within the given errors. A close examination of the fluxes in individual layers reveals how the solution responds to the requirement that silica be more exactly conserved. The relatively tight silica constraints in experiments 2 and 3 (appendix) are met by converting intermediate water (IW) in the Indian and Pacific regions to upper deep water (UDW), and producing an eastward flow of extreme AABW south of Africa and Australia. The conversion of low silica, high oxygen IW to high silica, low oxygen intermediate deep water (IDW) within the Indian Ocean is hard to reconcile with our understanding of the circulation and tracer patterns in this basin. Recent studies also support westward flow of extreme AABW south of Australia and Africa (Frew et al. 1995; Rintoul 1998), as found in the standard model. Therefore we consider the solutions with the tight silica constraints (experiments 2 and 3) to be unlikely. The changes required to meet the weaker constraints of experiment 1 [fluxes constrained to be 0 6 200 kmol s21 and total silica conserved to within O(500 kmol s21 )] are more modest, and the resulting circulation is plausible given our understanding. In any case, the inclusion of tighter constraints on silica does not change the main conclusions of the study: the deep overturning circulation is increased when silica is more tightly constrained, and the role of air–sea fluxes in converting UDW to IW is unchanged. 4. Deep overturning circulation: Individual basins In the following sections we present a detailed analysis of the deep overturning circulation in the individual ocean basins, including both lateral and diapycnal fluxes. The flux estimates are compared to previous studies, in particular the global analyses of Schmitz (1995, 1996b) and Macdonald (1998). The multiplicity of water mass definitions in the literature, many of them regionally specific, can create confusion. In the following, we specify the layers to which each water mass corresponds; the neutral densities defining each layer and their average temperature and salinity are listed in Table 2. Note, for example, that bottom water here refers to water denser than g n 5 28.2, a slightly broader definition than the common criterion of water with potential temperature less than 08C.

155

a. Subtropical Atlantic About 20 3 10 6 m 3 s21 of North Atlantic Deep Water (layer 17 to 18) enters the subtropical (South) Atlantic across SAVE2 at 128S (Fig. 5). The southward flow of NADW occurs principally in the western Brazil Basin (21765 3 10 6 m 3 s21 ) with only a small southward flux in the eastern basin over the Guinea Rise (2363 3 10 6 m 3 s21 ). Below NADW there is a northward flux of 562 3 10 6 m 3 s21 of Lower Circumpolar Deep Water (LCDW, layer 19) and Antarctic Bottom Water (AABW, layer 20). The magnitude of NADW flowing south and its principal confinement to the western basin agrees with previous studies (McCartney and Curry 1993; Dickson and Brown 1994). At 238S, Macdonald (1998) estimates a southward flux of 220 3 10 6 m 3 s21 NADW and a northward flux of 6 3 10 6 m 3 s21 AABW, although the southward flow of NADW is evenly split between the western and eastern basins. The southern boundary of the subtropical Atlantic region (Box I) is SAVE4. Because SAVE4 is not strictly zonal (Fig. 1), it is useful to divide discussion of the deep fluxes into the western Argentine Basin, the MidAtlantic Ridge and Angola Basin, and the eastern Cape Basin. In the Argentine Basin 3662 3 10 6 m 3 s21 of Upper Circumpolar Deep water (UCDW layers 15 to 17) and 1762 3 10 6 m 3 s21 LCDW (layer 18 and 19) move northward into the subtropical Atlantic (Fig. 5). Below these layers there is a 662 3 10 6 m 3 s21 northward flux of AABW (layer 20) and colder Weddell Sea Deep Water (WSDW, layer 21). Southeastward flow of deep water (layers 15–19) occurs over the mid-Atlantic ridge and in the Angola Basin, with 4363 3 10 6 m 3 s21 of UCDW/NADW and 2663 3 10 6 m 3 s21 of NADW/LCDW. A small southward flux of NADW (261 3 10 6 m 3 s21 ) is also found in the Cape Basin, while northward penetration of dense AABW into the Cape Basin is restricted by the Walvis ridge. This flow structure is clearly evident when looking at the cumulative layer fluxes (west to east) across SAVE4 (Fig. 6). The recirculating southward–northward flow of UCDW/NADW (15–17) and NADW/ LCDW (18–19) between 558 and 408W in the Argentine Basin is associated with a salinity maximum ($34.8 psu). Northward fluxes of UCDW, LCDW, and AABW occur in the eastern part of the Argentine Basin. The northward flow of UCDW and LCDW is more than compensated by returning southward flow of these waters and a net input of NADW over the mid-Atlantic Ridge and in the Angola Basin. Saunders and King (1995) show a similar cumulative southward flux of deep water between 148W and 68E, although they find more NADW moving southward in the Cape Basin than suggested in this study. [Differences in the fluxes in the deep layers result in part from the inclusion of our lightest deep water (15) in intermediate water by Saunders and King (1995).] The northward flow in the western basin, southeastward flow over the ridge, and small

156


VOLUME 31

southward flow in the eastern basin agrees with that shown by Reid (1989, his Figs. 22–25). The saline NADW flowing to the south at SAVE2 is sandwiched between fresher UCDW and Antarctic Intermediate Water above and fresher LCDW and AABW below. Sloyan and Rintoul (2000) show that the diapycnal fluxes between these water masses along with isopycnal mixing between deep water varieties can explain the changes in temperature and salinity in these layers in the subtropical Atlantic. The mixing of AABW with overlying NADW/LCDW results in the conversion of 362 3 10 6 m 3 s21 of AABW to NADW/LCDW (Fig. 5). This mixing results in the warmer, saltier AABW and LCDW observed to move northward across SAVE2 into the North Atlantic, and the cooler, fresher NADW at SAVE4 relative to SAVE2. Across the upper boundary of NADW (layer 17) there is a salt flux into less saline UCDW and intermediate water [Antarctic Intermediate Water (AAIW) and Subantarctic mode water (SAMW)]. This salt flux drives an interior 863 3 10 6 m 3 s21 mass flux of NADW (layer 17) into overlying UCDW (layer 15 and 16) of which 463 3 10 6 m 3 s21 is further modified into AAIW and SAMW (layers 10 to 14) (Fig. 5). The mass and salt flux from NADW into overlying UCDW, AAIW, and SAMW results in a salinity increase of these water masses in the subtropical Atlantic and an erosion of the NADW salinity maximum. In the subtropical South Atlantic Macdonald and Wunsch (1996) find a net upwelling of 1 3 10 6 m 3 s21 across the 3.58C isotherm. This isotherm is slightly above the salinity maximum core (;2000 dbar) at 128S but in the eastern (Angola and Cape) basins at 308S it is in the salinity minimum AAIW water. The interpretation of this upward flux in terms of mixing of specific water masses is not clear due to the isotherm (in the South Atlantic) intersecting two water masses. Schmitz (1996b) estimates 3 3 10 6 m 3 s21 of AABW is converted to less dense NADW/LCDW, similar to our estimate (see his Fig. II-155, between ‘‘Agulhas Leakage’’ and the equator). He finds 8 3 10 6 m 3 s21 of denser NADW upwells into lighter UCDW/NADW, again similar to our study (863 3 10 6 m 3 s21 flux of NADW/ LCDW to UCDW). The main difference between our study and Schmitz (1996b) is that we estimate approximately half of the diapycnal flux between NADW and UCDW is eventually converted to intermediate water by interior mixing, while Schmitz (1996b) has no conversion of NADW to intermediate water in the subtropical South Atlantic. FIG. 5. Lateral and diapycnal volume fluxes (310 6 m 3 s21 ) in bottom and deep layers for the Atlantic region (Box I and II). (top) The volume flux driven by air–sea buoyancy forcing acting on the lightest outcropping layer (15). Diapycnal fluxes due to interior mixing are indicated by curly arrows, solid arrow shows flux across lower interface, and dashed arrow flux across upper interface. Upward diapycnal arrows represent upwelling and downward arrows downwelling across the interface. Error estimates for the lateral and diapycnal fluxes are given in the text.

b. Atlantic sector of the Southern Ocean The ACC carries bottom and deep water masses eastward across the Atlantic sector (Box II) of the Southern Ocean (Fig. 5). At Drake Passage 6961 3 10 6 m 3 s21 of UCDW (layers 15 to 17), 2962 3 10 6 m 3 s21 of LCDW (layers 18 and 19), and 461 3 10 6 m 3 s21 of

JANUARY 2001

SLOYAN AND RINTOUL

157

FIG. 6. Cumulative flux in AABW (open circle), NADW/LCDW (open square), and UCDW/NADW (open diamond) across SAVE4 from the western Argentine Basin to the eastern Cape Basin. Positive is northward. Lower plot is the salinity (solid: psu) distribution along the section, the bounding neutral density (dot-dash:kg m 23 ) of the deep water masses: UCDW/NADW g n , 28.0, NADW/LCDW 28.0 , g n , 28.2 and, AABW g n . 28.2. Southward mass flux is shaded.

AABW (layer 20) enter the region. Across SAVE4 there is a net import of 1763 3 10 6 m 3 s21 of modified NADW (layer 15–19) into the Southern Ocean and an export of 662 3 10 6 m 3 s21 of AABW into the Argentine Basin. Across the Weddell Sea section an overturning circulation carries a southward flux of 21163 3 10 6 m 3 s21 of LCDW/AABW/UCDW balanced by a northward flux of 1161 3 10 6 m 3 s21 of denser WSDW/ WSBW (Weddell Sea Bottom Water). The net exchange of LCDW/AABW with WSDW/WSBW in the Weddell Sea agrees with a recent estimate of Fahrbach et al. (1994). South of Africa there is a net import of 361 3 10 6 m 3 s21 of extreme AABW (layer 21) from the Indian Ocean. Import of extreme bottom water into the Atlantic sector from the Indian Ocean is supported by evidence for AABW formation off Enderby Land (Jacobs and Georgi 1977; Mantisi et al. 1991), and recent studies showing bottom water formed along the Ade´lie coast flows westward through the Princess Elizabeth Trough to enter the Atlantic (Frew et al. 1995; Rintoul 1998). The exchange of AABW, CDW, and NADW between the Atlantic sector of the Southern Ocean and neighboring basins results in changes to the volume flux and

temperature and salinity characteristics of these deep water masses between Drake Passage and south of Africa (Sloyan and Rintoul 2000). The 1465 3 10 6 m 3 s21 of extreme bottom water (layers 21 to 23) entering the Atlantic from the Weddell Sea and Indian sector mixes with LCDW entering through Drake Passage to produce warmer, saltier AABW (layer 20). The modified AABW spreads northward into the Argentine Basin and replaces water in this layer lost to the Weddell Sea. Onethird, or 562 3 10 6 m 3 s21 of AABW (layers 20 to 23), is eventually converted to LCDW (Fig. 5), resulting in cooler, fresher NADW/LCDW exiting the basin south of Africa. There is only a small interior diapycnal flux of 162 3 10 6 m 3 s21 of UCDW/NADW to denser NADW/LCDW. In the Southern Ocean Atlantic sector the deep water isopycnals shoal and the UCDW/NADW (layers 15–17) outcrops. The outcropping of upper deep waters exposes them to air–sea fluxes (wind, heat, and freshwater), which modify their temperature and salinity properties resulting in formation of cold, fresh AASW. The buoyancy forcing drives a conversion of 8.4 3 10 6 m 3 s21 of AASW to lighter water across outcropping surface

158


15 (Fig. 5). This diapycnal transport driven by air–sea fluxes is partly compensated by an interior diapycnal flux (462 3 10 6 m 3 s21 ) of intermediate water (layer 14) into the lightest deep water layer (15). The mixing of NADW with LCDW and conversion of UCDW to intermediate water presented here differs from the circulation scheme proposed by Schmitz (1996b) (see his Fig. II-155, south of the Agulhas Leakage). He indicates a conversion of NADW/LCDW (10 3 10 6 m 3 s21 ) to intermediate water, with the rest of the southward flowing LNADW (4 3 10 6 m 3 s21 ) being converted to AABW. He also shows 8 3 10 6 m 3 s21 of UNADW flowing southward into the Southern Ocean of which 4 3 10 6 m 3 s21 is converted to intermediate water and the remainder is converted to AABW. In contrast we find that NADW spreads south along density surfaces to mix with LCDW entering the basin through Drake Passage, thereby reinforcing the salinity maximum of LCDW leaving the basin south of Africa. The injection of NADW into CDW balances the LCDW lost in the production of WSDW/WSBW and subsequent entrainment to form AABW, as well as the transformation of UCDW to intermediate water driven by air– sea fluxes. If NADW were to be directly converted to intermediate water in the Atlantic sector of the Southern Ocean, it must be accomplished by interior mixing processes, as these layers do not outcrop in this sector. However, the interior diapycnal fluxes found here show a net conversion of AABW to LCDW, rather than NADW/LCDW to intermediate water. The main impact of this study on the diagram of Schmitz (1996b) is to remove the 10 3 10 6 m 3 s21 (light blue in his figure) conversion of NADW to intermediate water. Instead, the NADW exported from the Atlantic increases the volume flux of LCDW leaving the basin south of Africa over that entering through Drake Passage. [The divergence of LCDW in the Atlantic sector of the ACC is balanced by a convergence of SAMW, as in Rintoul (1991) and Sloyan and Rintoul (2001).] Across the 3.58C isotherm Macdonald and Wunsch (1996) estimate a 9 3 10 6 m 3 s21 downwelling. In the Southern Ocean Atlantic sector this isotherm is essentially the boundary of UCDW/NADW and intermediate water. This large downwelling from intermediate water to deep water disagrees with the net conversion of AASW (air–sea modified upper deep water) to intermediate water found in this study. c. Subtropical Indian Ocean Across 188S, in the subtropical Indian Ocean (Box IV) a small amount of AABW moves northward in the western Mascarene Basin (161 3 10 6 m 3 s21 ) while a net transport of 663 3 10 6 m 3 s21 of LCDW (layer 19) is achieved by northward flow in the eastern Western Australian Basin and overlying the bottom water in the Mascarene Basin (Fig. 7). The northward flow of Antarctic origin water is balanced by 763 3 10 6 m 3 s21 of

VOLUME 31

Indian Deep Water returning southward above CDW/ AABW in the Madagascar and Central Indian Basin. (The 361 3 10 6 m 3 s21 southward flow in layer 18 occurs in the central Indian Basin). The inflow of water of Antarctic origin (layers 20 and 19) in the Masacarene and West Australian Basin agrees with water mass distributions on potential density surfaces shown by Mantyla and Reid (1995). At 188S, 1058E there is an increase in the northward transport of LCDW (Fig. 7), associated with a decrease in the temperature and increase in density of LCDW (below 3000 db) in comparison with water directly to the west [see Plates 1 and 3 of Warren (1981)]. The deep water below 3000 dbar at 1058E has similar properties to deep water found at the western boundary of the West Australian Basin. The cold LCDW at 1058E must be fed by a southern source. At 1058E the eastern Indian Ocean is directly open to the South Australian Basin through the Perth Basin (Fig. 1). The northward flow of LCDW near 1058 could be influenced by topographic features at 208S (see Fig. 1, unlabeled topographic features), which rise to above 3000 dbar and a string of topographic features (not shown) farther south that rise above 4000 dbar. These topographic features may enhance/maintain direct northward flow of AABW/LCDW near 1058E in the Perth and West Australian Basins. This suggests that there are two deep western boundary currents in the eastern Indian Ocean: adjacent the Ninetyeast Ridge, and along 1058E. Farther south at 328S the Indian Ocean is divided into numerous basins by ridges and plateaus. The effect of topography is to guide and restrict the 2364 3 10 6 m 3 s21 northward flow of AABW (layer 20, 1062 3 10 6 m 3 s21 ) and LCDW (layer 18 and 19, 1362 3 10 6 m 3 s21 ) into the western Natal, Mozambique, Madagascar, Crozet, and Perth Basins (Fig. 8). The northward transport of Antarctic bottom and lower deep water is partially balanced by 2062 3 10 6 m 3 s21 of southward IDW/UCDW (layers 17–15). The southward flow of IDW/UCDW occurs east of 808E in the Central Indian Basin, over the Broken Plateau, and in the Perth Basin, although there is some southward flow in the western (Natal and Madagascar) basins. The northward flow of Antarctic origin bottom and deep water in the numerous western boundary currents agrees with previous analysis of water properties (Gordon and Tchernia 1972; Toole and Warren 1993; Mantyla and Reid 1995). The southward flow of IDW between 808E to Australia is associated with an oxygen minimum and nutrient (phosphate and nitrate) maximum. These property characteristics are derived from the north. Much of the southward flow of IDW occurs over and adjacent to the Broken Ridge and in the eastern Perth Basin. Toole and Warren (1993), using the same 328S section, find a net northward transport of water below 2000 dbar in the Perth Basin, but note that strong southward transport of deep water is found in the eastern part of the basin. They also find weak transport over

JANUARY 2001

SLOYAN AND RINTOUL

159

FIG. 7. Cumulative flux in AABW (open circle), LCDW (open square), and IDW (open diamond) across Indian 188S from the western Mascarene Basin to the eastern West Australian Basin. Positive is northward. Lower plot is the oxygen (solid: mmol kg21 ) distribution along the section, the bounding neutral density (dot-dash: kg m 23 ) of the deep water masses: IDW g n , 28.1, LCDW 28.1 , g n , 28.2, AABW g n . 28.2. Southward mass flux is shaded.

the Broken Plateau below 1000 dbar. Robbins and Toole (1997) find a small net southward transport of IDW (their layer 5) across the entire subtropical basin. Talley and Baringer (1997) find southward transport above 2000 dbar between 1008 and 103.58E and east of 1098E. South of Australia at 1158E Hufford et al. (1997) finds an eastward flowing deep boundary current of 11 3 10 6 m 3 s21 to 19 3 10 6 m 3 s21 , which has water properties similar to the southward flowing IDW in the Perth Basin. The southward flow of IDW/UCDW (layers 15–17) across 328S is supplied by an upward diapycnal flux of 2063 3 10 6 m 3 s21 of LCDW to IDW/UCDW between 328 and 188S and southward flow (461 3 10 6 m 3 s21 ) of IDW (layers 17–15) across 188S (Fig. 9). The net upward diapycnal flux of LCDW to IDW/UCDW results from modification of LCDW, AABW (962 3 10 6 m 3 s21 ), and IDW (layer 18) to less dense IDW/UCDW. Most of the upwelled bottom and lower deep water remains in the UCDW and IDW returning to the Southern Ocean sector and only a small portion (462 3 10 6 m 3 s21 ) is converted further into intermediate and thermocline water masses (Fig. 9). This means that the large (2363 3 10 6 m 3 s21 ) overturning Indian Ocean cir-

culation found in this study is effectively balanced below 1500 dbar. Estimates of the strength of the deep overturning circulation in the Indian Ocean vary considerably (Table 6). Schmitz (1996b) provides two scenarios of the overturning circulation (Fig. II-130 and Fig. II-156), the former based on 13 3 10 6 m 3 s21 overturning from Robbins and Toole (1997) while the latter is similar to the circulation scheme in Schmitz (1995). The difference of 11 3 10 6 m 3 s21 between the southward flow of IDW/UCDW is balanced by a difference in the northward flux of bottom and lower deep water across 328S in the two scenarios. Opposed to the hydrographic studies is the estimate of a small (4 3 10 6 m 3 s21 ) deep inflow from a primitive equation model that assimilates the Levitus and Boyer (1994) climatological temperature and salinity fields (Lee and Marotzke 1997). This climatology provides a smoothed temperature and salinity field that may not adequately resolve the deep western boundary currents. As shown in Fig. 8 it is at the western boundaries that strong northward flow of LCDW and AABW is found—not resolving these features will likely lead to smaller northward fluxes of these

160


VOLUME 31

FIG. 8. Cumulative flux in AABW (open circle), LCDW (open square), and IDW/UCDW (open diamond) across Indian 328S from the western Natal Basin to the eastern Perth Basin. Positive is northward. Lower plot is the oxygen (solid: mmol kg21 ) distribution along the section, the bounding neutral density (dot-dash: kg m 23 ) of the deep water masses: IDW/UCDW g n , 28.0, LCDW 28.0 , g n , 28.2, and AABW g n . 28.2. Southward mass flux is shaded.

water masses and a small overturning circulation, as also noted by Robbins and Toole (1997). MERIDIONAL

FLUXES

The Toole and Warren (1993) overturning estimate results in a large net northward silica flux (Table 6). This large silica flux from the Southern Ocean is difficult to reconcile with global silica budgets, which suggest that the Southern Ocean is a net sink for silica (DeMaster 1981; Tre`guer et al. 1995). In Robbins and Toole (1997) the small northward silica flux is achieved by reducing the northward flux of Antarctic origin bottom and deep water. The present study still maintains a large overturning circulation but has a much reduced silica flux compared to Toole and Warren (1993). The reduction of the northward silica flux, while still maintaining a large overturning circulation, is achieved by a significant deepening of the returning southward flow. Figure 10 shows a comparison of the overturning scheme of Toole and Warren (1993), Robbins and Toole (1997), and this study. (Note that the discrepancy in the deepest layer between our results and the other studies is largely due to different layer definitions; the sum over their

deepest two layers covers a similar density range to our layers 19 and 20, and the sum of the fluxes in these layers is similar in each of the studies.) In this study the inflowing AABW and LCDW is balanced below 1500 dbar by southward flow of UCDW and IDW, while in the previous studies the northward flow of AABW and LCDW is largely balanced by intermediate and thermocline water. Indian Deep Water and UCDW have a significantly higher silica concentration than intermediate and thermocline water, resulting in the smaller northward silica flux found here compared to Toole and Warren (1993). The difference in the vertical structure of the overturning circulation affects not only the silica balance but also the heat flux. In this study the heat flux in the Indian Ocean at 188S is 21.27 6 0.05 PW. Macdonald (1998) estimates a southward heat flux of 21.45 6 0.3 PW, with a similar-sized Indonesian Throughflow (10.4610 3 10 6 m 3 s21 ). The large southward heat flux at 188S results from both the input of warm Pacific water via the Indonesian Throughflow and water warmed in the northern Indian Ocean where heat is gained from the atmosphere. We can partition the southward heat flux at 188S into its two sources, assuming that there is little

JANUARY 2001

SLOYAN AND RINTOUL

161

calculation for Macdonald and Wunsch (1996) partitions the heat flux into 1 PW from the Pacific and 0.45 PW from the northern Indian Ocean. That is, the discrepancy in the meridional heat flux at 188S is due to differences in the implied heat gain by the ocean in the northern Indian Ocean. Using residual techniques, Trenberth and Solomon (1994) estimate an Indian Ocean heat gain north of 208S of approximately 0.8 PW, larger than either ocean inversion but closer to Macdonald and Wunsch (1996). At 328S the heat flux is 20.87 6 0.06 PW. The heat flux divergence of 0.40 PW between 188 and 328S represents the ocean heat loss to the atmosphere, which is explicitly included in this model (Fig. 3). Other estimates of the southward heat flux at 328S, with size of Indonesian Throughflow given in parentheses, are shown in Table 6. This study and those quoted all have similar sized Indonesian Throughflow thus discounting the effect of the throughflow on the differences between the heat fluxes at 328s. The heat flux differences in these studies stem from both the estimated size of the overturning circulation and, more importantly, its vertical extent. Both the Toole and Warren (1993) and Macdonald and Wunsch (1996) overturning circulation extends over the complete water column from 4000 to 500 dbar. In this study the overturning circulation is restricted to below 1500 dbar. The difference in temperature of the upper, southward flowing limb of the overturning circulation accounts for the difference in the heat flux estimates. Hastenrath and Greischar (1993), using bulk formula, estimate an annual mean oceanic southward heat flux of 20.43 PW at 308S. They point out that this estimate does not include Pacific heat input via the Indonesian Throughflow, although they hypothesize that as the integration area includes the South China Sea this may compensate for the noninclusion of the throughflow. This heat flux estimate is much smaller than the oceanographic estimates and suggests that neglecting the heat flux associated with the Indonesian Throughflow seriously affects the estimated heat flux at 328S. FIG. 9. Lateral and diapycnal volume fluxes (310 6 m 3 s21 ) in bottom and deep layers for the Indian region (Box IV and V). (top) The volume flux driven by air–sea buoyancy forcing acting on the lightest outcropping layer (15). Diapycnal fluxes due to interior mixing are indicated by curly arrows, solid arrow shows flux across lower interface, and dashed arrow flux across upper interface. Upward diapycnal arrows represent upwelling and downward arrows downwelling across the interface. At 188S total northward flow of AABW and LCDW and smaller southward flow of layer 18 are shown separately. Error estimates for the lateral and diapycnal fluxes are given in the text.

heat loss from Indonesian Throughflow water between its input into the Indian Ocean and 188S. The division of the heat flux into Pacific origin and north Indian origin is then 1 PW (assuming a throughflow of 10 3 10 6 m 3 s21 at 248C) and 0.27 PW, respectively. A similar

d. Indian sector of the Southern Ocean In the Southern Ocean Indian sector the ACC dominates with strong eastward transports of CDW south of Africa (11264 3 10 6 m 3 s21 ) and south of Australia (9663 3 10 6 m 3 s21 ) (Fig. 9). In the bottom water class 561 3 10 6 m 3 s21 of generic warmer, saltier AABW (layer 20) moves eastward south of Australia, while there is a westward flux of denser ‘‘new’’ AABW (Ross Sea and Ade´lie Land Bottom Water, layers 21 and 23) south of Australia and Africa. The westward flow of dense AABW suggests production of AABW along the continental margin between 1408E and 08. Recent analysis suggests AABW formed along the Ade´lie coast contributes up to 25% of the global inventory of AABW (Rintoul 1998). Bottom and deep water flow northward

162


VOLUME 31

TABLE 6. Comparison of the estimated strength of the Indian Ocean overturning circulation (310 6 m 3 s21 ) at 328S between Toole and Warren (1993: T&W), Robbins and Toole (1997: R&T), Macdonald (1998: Macd), Schmitz (1996b), and this study. Schmitz (1996b) provides two overturning schemes: Schmitzpa (his figures Fig. II-130) and Schmitzpb (Fig. II-156). For this study we show results from the standard model and silica sensitivity experiment 1 (appendix). Also given is the net silica (kmol s 21 ) flux and heat (PW) flux, with the size of the Indonesian Throughflow (310 6 m 3 s21 ) given in parentheses. The balance of the overturning circulation (310 6 m 3 s21 ) between northward transport of AABW and LCDW and southward transport of UCDW/IDW, intermediate water (IW), and thermocline water (TW) is shown. Robbins and Toole (1997) provide the overturning balance for their study and Toole and Warren (1993). Macdonald (1998) use a different deep water classification, which is given in parentheses (BW: bottom water, DW: deep water). Columns are blank where estimate is not given. Positive is northward. Indian 328S This study

Overturning Silica Heat Flux (Indo) AABW/LCDW UCDW/IDW TW/IW

T&W

R&T

27 1532 21.67 (6.7) 25 26 220

12 6 3 21 13 27 26

Macd 17 6 5 21.30 5 12 217

6 6 6 6

0.28 (10) 2 (BW) 4 (DW) 5

in the Indian sector to fill the abyssal basins of the Indian Ocean. As AABW moves eastward and northward, its property characteristics are modified due to mixing with overlying LCDW. The involvement of LCDW in the production and modification of AABW results in a net diapycnal flux of 1763 3 10 6 m 3 s21 of LCDW to AABW. The LCDW lost in both the formation of and subsequent entrainment into AABW is replaced by a

Schmitzpa Schmitzpb 13

24

13 24 29

24 215 29

Standard 23 424 20.87 23 220 24

6 6 6 6 6 6

3 316 0.06 (10 6 3) 3 2 3

Expt 1 21 51 20.85 21 216 25

6 6 6 6 6 6

4 192 0.06 4 3 2

downward diapycnal flux of 2564 3 10 6 m 3 s21 of UCDW. The downwelling of slightly fresher, less oxygenated UCDW (IDW) into LCDW results in the dilution of the salinity maximum and reduced oxygen concentration of the LCDW leaving the Indian Basin south of Australia relative to that entering south of Africa (Sloyan and Rintoul 2000). In addition to the diapycnal fluxes resulting from mixing in the ocean interior, the shoaling of the upper deep layers near the Antarctic continent exposes them to air– sea fluxes, which change the temperature and salinity characteristics to AASW and result in air–sea flux driven transformation of 18 3 10 6 m 3 s21 of AASW to lighter water masses (Fig. 9). Some of this water is balanced by interior diapycnal mixing (1263 3 10 6 m 3 s21 ) of lower intermediate water to UCDW (layer 15). Sloyan and Rintoul (2001) discuss the circulation and interactions of interior and air–sea diapycnal fluxes between AASW, intermediate, and thermocline waters. For the Southern Ocean Indian sector, Schmitz (1996b) first scenario (overturning strength 13 3 10 6 m 3 s21 ) lacks details of the circulation south of the Polar Front zone and there is no conversion of UCDW to intermediate water. The second scenario (24 3 10 6 m 3 s21 overturning) has a conversion of 5 3 10 6 m 3 s21 of deep water into intermediate water, which is smaller than the 18 3 10 6 m 3 s21 found in this study. e. Subtropical Pacific

FIG. 10. Comparison of the Indian Ocean overturning circulation at 328S between Toole and Warren (1993) (dashed), Robbins and Toole (1997) (thin solid), and this study (thick solid).

Across 328S there is northward flow of 860.5 3 10 6 m 3 s21 of AABW (layer 20) and 1862 3 10 6 m 3 s21 of LCDW (layer 18 and 19), which results in a 2663 3 10 6 m 3 s21 net northward flux of Antarctic origin water into the Pacific Ocean (Fig. 11). The northward a priori transport constraint of 11 3 10 6 m 3 s21 of lower deep water (layer 18 and 19) and bottom water (layers 21–23) between 1818E and 192.58E is maintained (Fig. 12). [From the same current mooring array (PCM-9)

JANUARY 2001

SLOYAN AND RINTOUL

FIG. 11. Lateral and diapycnal volume fluxes (310 6 m 3 s21 ) in bottom and deep layers for the Pacific region (Box VI). (top) The volume flux driven by air–sea buoyancy forcing acting on the lightest outcropping layer (15). Diapycnal fluxes due to interior mixing are indicated by curly arrows, solid arrow shows flux across lower interface, and dashed arrow flux across upper interface. Upward diapycnal arrows represent upwelling and downward arrows downwelling across the interface. South of Australia the flux in AABW shows the principal eastward flow of generic AABW and smaller eastward flow of newly formed AABW. Error estimates for the lateral and diapycnal fluxes are given in the text.

Moore and Wilkin (1998) estimate a northward flux of AABW and LCDW of 20.4 3 10 6 m 3 s21 below 2000 dbar and 17.5 3 10 6 m 3 s21 below 3000 dbar. The transport estimate of Moore and Wilkin (1998) is larger than that of Whitworth et al. (1999), as their analysis takes into account the vertical shear of adjacent current moor-

163

ings to fill missing current records.] Farther northward transport of AABW is shown immediately east of 192.58E, while northward flow of LCDW continues to accumulate across much of the southwest Pacific Basin (Fig. 12). Reid (1986) also shows a broad northward flow in his adjusted steric height maps at 2000 and 3000 dbar between 1908 and 2308E. Broad northward flow is also shown in a recent analysis of the Pacific 328S section (Wijffels et al. 2000, manuscript submitted to J. Geophys. Res.). The northward flow of AABW and LCDW in this model is significantly larger than previous estimates, which vary from between 20 3 10 6 m 3 s21 and 7 3 10 6 m 3 s21 (Warren 1981; Wunsch et al. 1983; Macdonald 1998; Tsimplis et al. 1998). The northward transport of AABW and LCDW is balanced by return flow below 1500 dbar of 22563 3 10 6 m 3 s21 of Pacific Deep Water (PDW) and UCDW. The balance between northward flowing CDW/AABW and middepth southward flowing PDW/UCDW agrees with previous studies (Reid 1973; Wunsch et al. 1983; Toggweiler and Samuels 1993b; Wijffels 1993). Macdonald (1998) estimates northward transport of deep and bottom water of 865 3 10 6 m 3 s21 at 188S and 966 3 10 6 m 3 s21 at 288S. These estimates are clearly much smaller than our estimate of 2663 3 10 6 m 3 s21 at 328S. Macdonald (1998) discusses difficulties with reconciling her deep Pacific circulation with that of previous studies. At similar latitudes, her northward transport of deep water at 108 and 248N is only half that estimated by Wijffels (1993) and Roemmich and McCallister (1989), respectively. More importantly, she finds at 108N the northward transport of abyssal water is balanced by southward thermocline transport, rather than by PDW. This deep to thermocline Pacific overturning between 108 and 248N disagrees with other studies that show a deep to middepth overturning circulation in the Pacific Ocean (Toggweiler and Samuels 1993b; Wijffels 1993). The deep water transport discrepancy between this study and that of Macdonald (1998) is apparently not due to the fact that we impose an a priori constraint on the northward transport of deep and bottom water adjacent to the Tonga–Kermedac Ridge, as Sloyan (1997) shows that models with no deep water transport constraint have a similar or slightly larger northward transport of deep and bottom water. A more likely explanation for the discrepancy in Pacific deep water transports is the different representations of diapycnal fluxes used in the two studies, as discussed in section 5. At 118S the northward flow of LCDW/AABW is restricted to deep passages in the western basin, namely the Samoan Passage, Robbie Ridge, and on the eastern flank of the Manihiki Plateau. Current moorings in these restrictions give a mean (17 month) northward transport of 10.661.7 3 10 6 m 3 s21 for water colder than 1.18C (Roemmich et al. 1996). Below ;18C we estimate a northward flow of 860.5 3 10 6 m 3 s21 at 328S. This

164


VOLUME 31

FIG. 12. Cumulative flux in AABW (open circle), LCDW (open square), and PDW/UCDW (open diamond) across Pacific 328S from the western Tasman Basin to the eastern Chile Basin. Positive is northward. Lower plot is the oxygen (solid: mmol kg21 ) distribution along the section, the bounding neutral density (dot-dash: kg m 23 ) of the deep water masses: PDW/UCDW g n , 28.0, LCDW 28.0 , g n , 28.2, and AABW g n . 28.2. Southward mass flux is shaded.

suggests that there is little conversion of AABW to less extreme LCDW between 328S and 118S. f. Pacific sector of the Southern Ocean The ACC transports 6061 3 10 6 m 3 s21 of UCDW (layers 15–17) and 3662 3 10 6 m 3 s21 of LCDW (layers 18 and 19) to the east south of Australia (Fig. 11). During its transit of the Pacific, the transport of UCDW increases to 6961 3 10 6 m 3 s21 and LCDW decreases to 2962 3 10 6 m 3 s21 on leaving the basin at Drake Passage. The Pacific sector also exports 260.5 3 10 6 m 3 s21 of extreme AABW (layer 21) into the Indian sector, while 561 3 10 6 m 3 s21 of warmer AABW (20) enters the Pacific from the Indian sector. At Drake Passage there is also a small (461 3 10 6 m 3 s21 ) eastward flux of warmer, saltier AABW (layer 20). The southward flow of PDW/UCDW and northward flow of LCDW/AABW across 328S result in the observed changes of transport in these layers between Australia and Drake Passage. Sloyan and Rintoul (2000) show that the temperature and salinity characteristics of these layers are also substantially modified by diapycnal fluxes in the Pacific sector of the Southern Ocean. A

downward diapycnal flux of 2465 3 10 6 m 3 s21 of UCDW into LCDW replaces both the net export of LCDW to the north and east, and the 1165 3 10 6 m 3 s21 of LCDW involved in the formation of AABW in this sector. The 1165 3 10 6 m 3 s21 includes LCDW consumed in the process of forming dense water over the continental shelf, as well as subsequent entrainment of LCDW into newly formed AABW to form the warmer, saltier, less dense AABW that moves northward in the Pacific Ocean and through Drake Passage. The interior diapycnal flux of UCDW to LCDW is not the only diapycnal process acting on UCDW. UCDW shoals and outcrops south of the Polar Front. Air–sea fluxes convert outcropping UCDW to AASW by cooling and freshening and drive a net conversion of 8 3 10 6 m 3 s21 of water in this density class to lighter intermediate waters. This air–sea flux driven transformation is more than compensated by interior mixing converting 1564 3 10 6 m 3 s21 of lower intermediate water into the upper deep water (layer 15). Schmitz (1996b) details the large-scale meridional circulation for the Pacific Ocean [see his Fig. II-69, the portion from just north of the Subantarctic Frontal Zone (SFZ) to Antarctica]. He estimates that the 17 3 10 6 m 3

JANUARY 2001

SLOYAN AND RINTOUL

s21 of LCDW/AABW moving northward into the Pacific Ocean is balanced by a return flow in upper deep layers. In the Southern Ocean the UCDW is principally converted into denser LCDW and AABW. This circulation path is similar to that found in this study, although our overturning circulation is larger (2663 3 10 6 m 3 s21 ). Schmitz (1996b) also indicates 2 3 10 6 m 3 s21 of UCDW is converted to intermediate water and 10 3 10 6 m 3 s21 of Antarctic surface water is converted into intermediate water between the SFZ and Polar Frontal Zone (PFZ). Schmitz inferred such a transformation on the basis of published estimates of layer fluxes at lower latitudes, but could not determine the mechanism responsible for the diapycnal fluxes. Here we are able to explicitly identify the combination of air–sea buoyancy and momentum forcing and interior diapycnal mixing responsible for the water mass conversions observed. We estimate a net 8 3 10 6 m 3 s21 transformation of AASW to lighter intermediate water by air–sea fluxes, which is effectively balanced by interior diapycnal fluxes from lower intermediate water to the upper layer (15) of UCDW. Macdonald and Wunsch (1996) indicate a downwelling of 9 3 10 6 m 3 s21 across the 3.58C isotherm in the Southern Ocean Pacific sector. In this study we estimate interior mixing of 1564 3 10 6 m 3 s21 of intermediate water to upper deep water (layer 15). As stated above, this interior flux of intermediate water to deep water is partially offset by the 8 3 10 6 m 3 s21 air–sea flux of AASW to intermediate water. 5. The Southern Ocean contribution to the deep overturning circulation The overturning circulations in each basin, and their zonal integral, are illustrated in Fig. 13 by summing over layers corresponding to five broadly defined water masses. The export of 17 3 10 6 m 3 s21 of NADW from the Atlantic is balanced by a northward flow of thermocline water (TW) and intermediate water (IW) (10 3 10 6 m 3 s21 ) and bottom water (BW) (6 3 10 6 m 3 s21 ) across SAVE4 (308–508S). As a result of upwelling in the subtropical Atlantic, at SAVE2 (128S) the upper layers carry a larger fraction of the northward flow required to balance the export of NADW (14 3 10 6 m 3 s21 out of 18 3 10 6 m 3 s21 ). In the Atlantic sector of the Southern Ocean, 11 3 10 6 m 3 s21 of BW is exported from the Weddell Sea. [The net northward flow of 4 3 10 6 m 3 s21 of BW from the Weddell Sea is a balance between southward flowing CDW and warmer AABW (layer 20, indicated by thin arrow in Fig. 13) and newly formed Weddell Sea Deep and Bottom Water (layers 21–23).] Comparing fluxes in density layers shows that more IW enters the Atlantic through Drake Passage, and more LDW leaves the basin south of Africa (boxed numbers in Fig. 13). [The small net transports of intermediate water at 328S in the Indian and Pacific Oceans are the difference between much larger gross fluxes to

165

the north and south (31 3 10 6 m 3 s21 and 25 3 10 6 m 3 s21 , respectively). The intermediate circulation of the model is discussed in Sloyan and Rintoul (20001).] The subtropical Indian Ocean is dominated by a strong, deep overturning circulation. Inflow of 23 3 10 6 m 3 s21 of LDW and BW is largely balanced by outflow of 20 3 10 6 m 3 s21 of UDW (IDW in this case). Most of the upwelling required to close this circulation occurs between 328 and 188S. A net southward flow of 14 3 10 6 m 3 s21 in the upper layers balances the 10 3 10 6 m 3 s21 Indonesian Throughflow and 4 3 10 6 m 3 s21 of UDW converted to IW in the subtropical Indian Ocean. The ACC carries more IW (18 3 10 6 m 3 s21 ) out of the Indian basin south of Australia than enters south of Africa; the reverse is true for UDW (210 3 10 6 m 3 s21 ) and LDW (26 3 10 6 m 3 s21 ). The Indian Ocean export of BW (7 3 10 6 m 3 s21 ) is achieved by westward transport of ‘‘new’’ dense bottom water south of Africa and an eastward transport of warmer bottom water south of Australia. That is, the Indian basin as a whole imports UDW, LDW, and throughflow water, and exports IW and BW. The strong, deep overturning (26 3 10 6 m 3 s21 ) in the Pacific Ocean at 328S is similar to that of the Indian Ocean. The inflow of IW in the eastern subtropical Pacific is nearly compensated by outflow in this density class in the western Pacific, so the net transport of IW across 328S is small. The throughflow is largely fed by 8 3 10 6 m 3 s21 of TW flowing northward across 328S. The Pacific sector of the Southern Ocean imports IW and LDW from the Indian sector, and exports UDW to the Atlantic sector. The zonal sum of the fluxes in density layers is completely dominated by the strong deep overturning cells in the Indian and Pacific Oceans. Roughly 50 3 10 6 m 3 s21 of LDW and BW are exported to lower latitudes, balanced by an equal poleward flux of slightly less dense UDW. The deep overturning cell is closed below about g n 5 27.4, or a depth of about 1500 m. [The total transport of deep water entering and leaving the Southern Ocean is even larger due to partial compensation in the zonal sum by flows of opposite sign in different basins in each layer. The total southward flow across roughly 308–408S is 62612 3 10 6 m 3 s21 (1765, Atlantic; 2062, Indian; 2663, Pacific). The total northward flow is 5669 3 10 6 m 3 s21 (762, Atlantic; 2364, Indian; 2663, Pacific).] The buoyancy (heat) input needed to drive the deep upwelling of this model is estimated from the temperature change required to convert AABW/LCDW to UCDW (1.58C) and to convert UDW/NADW/IDW to intermediate water (28–38C) in the Atlantic and Indian Oceans. In the subtropical Atlantic 3 3 10 6 m 3 s21 of AABW is converted to LCDW (DT 1.58C), which requires a buoyancy input of 0.02 PW. The conversion of 4 3 10 6 m 3 s21 of UCDW/NADW to intermediate water results in a buoyancy input of 0.04 PW. In the subtropical Indian Ocean the 20 3 10 6 m 3 s21 conversion of

166


VOLUME 31

FIG. 13. A schematic five-layer view of the overturning circulation (310 6 m 3 s21 ) in each Southern Hemisphere basin, and the zonal sum for SW: surface water, TW: thermocline water, IW: intermediate water, UDW: upper deep water, LDW: lower deep water, and BW: bottom water. The neutral density surfaces used to define each layer are shown. Air–sea flux-driven diapycnal fluxes are shown by bold, dashed arrows at the sea surface. Diapycnal fluxes due to interior mixing are indicated by thin dashed arrows. Fluxes in boxes represent the net convergence (1ve) or divergence (2ve) of a particular water mass in that sector of the Southern Ocean due to meridional and diapycnal fluxes; mass is conserved by a compensating divergence in zonal transport of the ACC. Two-headed arrows in intermediate water highlight that the net flux is the difference of nearly balancing northward and southward fluxes. Small [2–3 (310 6 m 3 s21 )] imbalances result from choosing a solution where layer residual norms are of O[1–2 (310 6 m 3 s21 )] and rounding to nearest whole number.

AABW/LCDW to UCDW requires an input of 0.12 PW, while the conversion of UCDW to intermediate water requires a further 0.05 PW input. North of 328S in the Pacific the 26 3 10 6 m 3 s21 conversion of AABW/

LCDW to UCDW/PDW requires an input of 0.16 PW. The 50 3 10 6 m 3 s21 overturning circulation requires a net input of 0.39 PW. The buoyancy input that drives the deep upwelling

JANUARY 2001

SLOYAN AND RINTOUL

in the subtropical oceans is supplied by lateral advection and diapycnal mixing. The heat required to convert AABW/LCDW to UCDW is supplied by warm deep water flowing southward across the northern boundary of the subtropical gyre. Diapycnal mixing between this warm upper deep water and the underlying cold water provides the buoyancy input required to convert LCDW to UCDW/IDW/PDW. Munk and Wunsch (1998) suggest that tides and wind provide the energy source for the mechanical mixing, with surface buoyancy fluxes and hydrothermal heating playing a secondary role. Enhanced mixing over rough topography (Polzin et al. 1997) is also likely to play a part in driving the deep mixing that the large-scale budgets require. Further conversion of UCDW to intermediate water in the subtropical Atlantic and Indian Ocean is driven by downward diapycnal heat fluxes from overlying thermocline and intermediate water (Sloyan and Rintoul 2000). The vigorous deep overturning circulation requires similarly vigorous diapycnal fluxes. The transformation of water from one density class to another is accomplished both by air–sea fluxes of heat and freshwater (hence buoyancy) and momentum (Ekman), and by interior diapycnal mixing. Air–sea fluxes convert 34 3 10 6 m 3 s21 of UDW to IW in the Southern Ocean, with more than half of this occurring in the Indian sector. In a steady-state ocean the conversion of UDW to IW by air–sea fluxes must be compensated by interior mixing when the global extent of the isopycnal is considered (Tziperman 1988). In this model we find that the compensating flux occurs in the Southern Hemisphere oceans south of 308–408S where interior mixing converts 31 3 10 6 m 3 s21 of IW to denser UDW, although it is not balanced locally within each Southern Ocean sector. [Recall that ‘‘interior’’ in this context refers to the entire ocean beneath the sea surface.] As shown in Fig. 13, within each sector the transformation and interior mixing contributions to the flux of IW and UDW are not balanced: more IW is produced by air–sea fluxes than is consumed by mixing in the Atlantic and Indian sectors, while the opposite is true in the Pacific. In addition, there is large O(80 3 10 6 m 3 s21 ) export of new (cold, fresh) IW to the subtropics, which is nearly balanced by a return flow of modified (warm, salty) IW from the subtropics (Sloyan and Rintoul 2001). Intermediate water gains heat and salt along its circumpolar and subtropical circulation path. The interior compensation of slightly warmer, saltier IW for cold, fresh AASW results in a transfer of heat and salt south of the Polar Front (Sloyan and Rintoul 2001). In the Southern Ocean, 50 3 10 6 m 3 s21 of UDW is converted to denser LDW; 27 3 10 6 m 3 s21 is converted to even denser BW. Because our air–sea flux climatologies do not include the intense buoyancy loss taking place near the Antarctic margin, our model cannot distinguish the relative contributions of air–sea fluxes and interior mixing to this transformation. (It should also be kept in mind that in Fig. 13 the layer called ‘‘BW’’

167

includes all water denser than g 5 28.2; that is, this layer includes newly formed AABW and LDW entrained as new AABW sinks from the Antarctic shelf and spreads northward. Our results should not be interpreted to mean that 27 3 10 6 m 3 s21 of dense water sinks from the Antarctic shelf.) The 50 3 10 6 m 3 s21 of LDW and BW exported to the lower latitudes is converted to UDW by interior diapycnal mixing, largely in the Indian and Pacific Oceans. The substantial diapycnal fluxes in both the Southern Ocean and subtropical latitudes can explain the changes in water mass properties observed on isopycnals (Sloyan and Rintoul 2000), as well as the changes in transport described here. Our results are similar to those of Schmitz (1995, 1996b), and the numerous observational studies on which his synthesis was based. The net overturning circulation across about 408S inferred from Schmitz (1995, plate 7) consists of a southward deep-water transport of 52 3 10 6 m 3 s21 , balanced by a northward transport of 48 3 10 6 m 3 s21 of bottom water and 4 3 10 6 m 3 s21 of intermediate water. The updated version of this figure in Schmitz (1996b, Fig. II-156) shows a deep overturning circulation of similar strength: 48 3 10 6 m 3 s21 of bottom water flowing north across 408S, roughly balanced by 53 3 10 6 m 3 s21 of deep water flowing south. The results discussed here differ in several important respects from the global inversion of Macdonald (1998). The zonally integrated overturning circulation in deep water classes is 32 3 10 6 m 3 s21 in her model. The Indian Ocean overturning is weaker and extends much higher in the water column, and the overturning in the Pacific is much weaker in her model than found here. Part of the reason for the differences in the two models may lie in the representation of the diapycnal fluxes. McIntosh and Rintoul (1997) and Sloyan and Rintoul (2000) find that use of a single unknown for the diapycnal flux across each interface, as in Macdonald (1998), fails to reproduce the known diapycnal fluxes when such a model is applied to output from a numerical model. The choice of weights may also play a role: Macdonald (1998) found that the deep transports in the Pacific increased in a model that was weighted to better resolve the diapycnal fluxes. In this study the deep overturning circulation is completely dominated by a CDW–BW cell. The volume transports involved in the NADW cell are relatively small in comparison. This suggests that the common practice of equating the global overturning (or thermohaline) circulation with the NADW cell is misleading. Goodman (1998) reached a similar conclusion on the basis of a coarse resolution ocean general circulation model (OGCM). Our results also provide some insight into the pathways involved in the NADW cell. In order to compensate the sinking of dense water in the North Atlantic, somewhere deep water must be converted to less dense intermediate or thermocline water to close the circulation. We do not find any evidence for upwelling from

168


deep water through the thermocline in low latitudes. Rather, inflowing NADW (modified by mixing in the ACC) enters the Indian and Pacific Oceans, upwells, and is converted to slightly less dense deep water before being returned to the Southern Ocean. These results are consistent with the radiocarbon and OGCM results discussed by Toggweiler and Samuels (1993b) and Toggweiler and Samuels (1998). The TW/IW and DW circulations are largely isolated from each other in the subtropics, in contrast to the traditional assumption that uniform upwelling balances the sources of deep water, an assumption that underlies a number of conceptual models of the deep circulation (e.g., Stommel and Arons 1960; Munk 1966; Gordon 1986). The deep water does interact with intermediate water in the Southern Ocean, where the deep layers outcrop. More specifically, the UDW exported from the Indian and Pacific Oceans is sufficiently light to outcrop at high latitudes. The NADW itself is mostly too dense to outcrop. At the outcrop, air–sea fluxes of heat and freshwater convert UCDW to AASW and ultimately to IW. The NADW cell is closed by first being converted to UDW in the Indian and Pacific Oceans, followed by conversion of UDW to IW by air–sea exchange, and a return flow of IW to the Atlantic. That it is UCDW, rather than denser LCDW/NADW, which upwells at high latitude and is converted to lighter water by air–sea buoyancy fluxes has important implications for the Southern Ocean overturning, as well as the NADW cell. The continuity constraints and air–sea flux climatology require a significant net southward flux of UCDW at 308–408S. UCDW occupies a depth range of 1500–2500 dbar between 308 and 408S. As this layer shoals to the south across the latitude of the Drake Passage gap, it lies at depths both blocked and unblocked by topography. The poleward transport of UCDW above topography required to balance the equatorward Ekman transport and air–sea transformation to lighter layers must therefore occur in an eddy-driven transport dynamically balanced by divergence of interfacial form stress. The eddy-driven transport can occur at all scales, from the large-scale standing eddies to small-scale mesoscale eddies. The large-scale standing eddies and mesoscale eddies result in northward or southward movement of the ACC and the vertical displacement of particular water masses along a latitude band. The possibility that part of the UCDW southward flow is accomplished by eddies is supported by observations that significant meridional gradients of potential vorticity exist in the UDW layer (Speer et al. 2000). OGCMs that include a parameterization of eddy advection (Gent and McWilliams 1990) show a maximum in the eddyinduced transport in the ACC (Gent et al. 1995), where eddies carry UCDW poleward. Similarly, Guilyardi et al. (1999) use a coupled atmosphere–ocean model to show that poleward advection of UCDW by eddies supplies the upwelling required to feed the water mass transformation driven by air–sea buoyancy fluxes. That

VOLUME 31

the three-dimensional circulation derived here includes a large poleward transport of relatively light (i.e., above topography) UCDW provides observational evidence that advection by standing and transient eddies is an essential physical process in the Southern Ocean. 6. Conclusions A box inverse model is used to determine the circulation of the Southern Hemisphere oceans, including diapycnal fluxes due to interior mixing and water mass transformation driven by air–sea buoyancy fluxes. The fluxes due to interior mixing and water mass transformation are large terms in the layer balances. A consistent description of the three-dimensional ocean circulation must include an adequate representation of both contributors to the diapycnal flux. The section-integrated fluxes and transports of major currents are generally consistent with earlier estimates, but some significant discrepancies with earlier work are found. In the Atlantic, the export of 17 3 10 6 m 3 s21 of NADW is balanced primarily by inflow of intermediate and bottom water. The deep circulation of both the Indian and Pacific Oceans is dominated by strong, deep overturning cells. In the Indian, 23 3 10 6 m 3 s21 of northward flowing lower deep and bottom water are converted to upper deep water, which leaves the basin to the south below 1500 dbar. The restricted vertical extent of the deep overturning is in contrast to the overturning cells found by Toole and Warren (1993), Robbins and Toole (1997), and Macdonald (1998), where the northward flow of deep water was balanced by southward flow of thermocline and intermediate water. As a result of the small difference in temperature between northward and southward flowing limbs in the deep Indian overturning, the meridional heat flux at 328S found here is smaller than that found by the earlier authors. The circulation of the deep Pacific consists of a deep overturning cell similar in strength and vertical extent to that of the Indian Ocean. The deep circulation is essentially isolated from the thermocline circulation in the subtropics. The zonal integral of the meridional flow across roughly 308–408S reveals a vigorous 50 3 10 6 m 3 s21 deep overturning circulation consisting of poleward flow of UCDW (including IDW and PDW) and equatorward flow of denser abyssal water. The cell is closed at high latitude by conversion of UCDW to LCDW/BW by interior mixing and/or buoyancy loss to the atmosphere, and at low latitudes by interior mixing converting abyssal water to lighter deep water. NADW exported from the Atlantic Ocean makes a relatively small contribution (in terms of mass transport) to the global deep overturning circulation. The UCDW also participates in a counterrotating overturning cell south of 308–408S. The explicit inclusion, in this model, of air–sea fluxes and the transformation they drive shows that UCDW upwells at high

JANUARY 2001

169

SLOYAN AND RINTOUL

latitude and is driven north in the Ekman layer. Air–sea fluxes convert upwelling UCDW to cooler, fresher Antarctic Surface Water (AASW), and transform 34 3 10 6 m 3 s21 of AASW to lighter intermediate water. Some of the intermediate water formed in this way flows north in the Atlantic to close the NADW overturning cell. The Ekman transport is largely compensated by a return flow at slightly greater density, rather than by poleward flow in density layers below topography. The significant meridional transports in density layers above topography suggest that advection by standing and transient eddies plays a central role in the Southern Ocean overturning circulation. This study has shown that diapycnal transports driven by interior mixing and air–sea buoyancy fluxes are a key part of the large-scale circulation of the Southern Ocean, and that they can be determined with a suitably designed inverse model. The bulk, area-averaged estimates of diapycnal transport due to mixing in the ocean interior provided by such an inverse model complement the small, but growing, number of direct observations of ocean mixing now available. The dataset used here is limited, and improved estimates of large-scale lateral and diapycnal fluxes will undoubtedly result from applying similar inverse modeling techniques to the WOCE dataset, with its much more complete spatial and temporal coverage, additional constraints (e.g., tracers, floats), and repeat measurements with which to assess the impact of oceanic variability. Acknowledgments. The majority of this work was done while B. Sloyan was a student at the Institute of Antarctic and Southern Ocean Studies, University of Tasmania, Hobart, Australia. We thank Phil Morgan for writing the Dobox code. This work is supported by Environment Australia through the National Greenhouse Research Program. This study contributes to the World Ocean Circulation Experiment. APPENDIX Sensitivity Tests and the Silica Constraint In our ‘‘standard model’’ a weak silica constraint [box conservations O(1000 kmol s21 ) and no flux constraint at 308–408S] is applied to the inverse model. Although

this reduces the silica flux across 328S in the Indian Ocean compared to Toole and Warren (1993), the meridional silica flux across 308–408S is still large. In order to investigate the sensitivity of the circulation and the silica constraint a series of additional experiments were performed. The experiments differ from the standard model by the tightness of the net box silica conservation and the addition of a silica flux constraint at SAVE4, Ind32, and Pac32. In experiment 1 the meridional silica fluxes across the Atlantic, Indian, and Pacific are 0 6 200 kmol s21 and total box conservation is O(300–500 kmol s21 ). In experiment 2 the meridional silica flux constraints across the Atlantic, Indian, and Pacific are tightened to 0 6 150 kmol s21 and the box conservation is O(100–200 kmol s21 ). Experiment 3 constrains the meridional silica fluxes across the Atlantic, Indian, and Pacific to 0 6 25 kmol s21 and box conservation of O(30–50 kmol s21 ). In all experiments a mass imbalance of O(1–2 3 10 6 m 3 s21 ) is maintained. Table A1 shows that the net mass flux in the additional silica experiments and the standard model are the same within the given error. (As stated, the errors are only the formal errors of the inverse method. As such they are likely to underestimate the true uncertainty in the estimates.) The property fluxes (not shown) show a similar agreement between the standard model and silica experiments. The silica experiments maintain the deep northward transport (43616 3 10 6 m 3 s21 , expt 1) across 308–408S. As in the standard model the silica experiments also show that the northward transport of deep and bottom water is essentially balanced by southward transport of UDW (53610 3 10 6 m 3 s21 , expt 1). The air–sea conversion of 34 3 10 6 m 3 s21 of UDW to IW is present in all experiments, although there are differences in the interior diapycnal flux between AABW, LCDW, and UCDW in the Southern Ocean sector between the standard and silica experiments. These differences result from changes in the transport and direction of AABW south of Australia and Africa (discussed below). Although the mass and property fluxes are similar between the models, the silica flux at SAVE4, Ind32, and Pac32 are significantly reduced (Table A2). The differences in the meridional silica flux result from small changes in the overturning circulation in the subtropical

TABLE A1. Comparison of the total mass flux (310 6 m 3 s21 ) between the standard model and the three silica experiments. Standard SAVE2 SAVE4 Drake P S. Africa Weddell Sea Ind18 Ind32 S. Aust Pac32

20.77 20.80 134.9 135.4 20.35 210.4 210.3 145.6 7.4

6 6 6 6 6 6 6 6 6

2.9 2.7 0.8 2.7 2.5 3.1 2.3 1.0 1.8

Expt 1 20.82 20.3 135.7 136.6 20.01 210.0 210.4 147.3 9.2

6 6 6 6 6 6 6 6 6

Expt 2 1.2 1.6 0.6 2.2 1 3.6 3.4 1.2 2.0

20.81 20.3 133.5 134.5 20.01 212.4 212.6 147.6 11.0

6 6 6 6 6 6 6 6 6

Expt 3 0.4 1.6 0.6 2.4 1 3.1 3.0 1.1 2.7

20.80 20.2 132.7 133.7 20.01 213.2 213.4 147.6 11.6

6 6 6 6 6 6 6 6 6

0.4 1.6 1.0 2.6 1 2.8 2.7 1.4 2.4

170


VOLUME 31

TABLE A2. Comparison of the total silica flux (kmol s 21 ) between the standard model and the three silica experiments. Standard SAVE2 SAVE4 Drake P S. Africa Weddell Sea Ind18 Ind32 S. Aust Pac32

124 295 9071 8450 226 261 425 8017 821

6 6 6 6 6 6 6 6 6

194 339 73 407 339 679 316 99 283

Expt 1 258 225 9070 8968 2 395 51 8581 25

6 6 6 6 6 6 6 6 6

basins. Figures A1 and A2 show the overturning circulation for the Indian Ocean north of 188S and the Pacific Ocean north of 328S for the standard model and the additional silica experiments. At 188S in the Indian Ocean the standard model and experiment 1 show that northward transport of lower deep and bottom water is balanced by southward transport of upper deep water. The small northward transport of intermediate water is converted to surface water. In experiments 2 and 3 the southward transport of upper deep water is balanced by northward transport of lower deep and bottom water and intermediate water. In these models the silica flux at 188S in the Indian Ocean is reduced by the conversion of low silica concentration intermediate water to higher silica concentration upper deep water. The standard model and experiment 1 show the greatest similarity with respect to individual water mass fluxes. Difference in the layer fluxes between these models occur in the Indian and Pacific Oceans and in the transport of extreme AABW (layer 21 and 22) and generic AABW (layer 20) south of Africa and Australia. Across 328S in the Indian Ocean, experiment 1 has a smaller net northward flux of bottom (463 3 10 6 m 3 s21 ) and larger northward transport of deep water (1765 3 10 6 m 3 s21 ), which results in a small decrease in northward transport of bottom and deep water when compared to

Expt 2

133 149 658 662 54 373 192 633 190

247 226 8818 8791 21 106 38 8679 4

6 6 6 6 6 6 6 6 6

113 109 644 636 53 216 140 627 138

Expt 3 27 22 8726 8717 21 8 7 8716 3

6 6 6 6 6 6 6 6 6

37 25 622 623 37 51 25 622 25

the standard model. The redistribution and slight reduction of the northward deep and bottom transport results in a decrease of the silica flux across 328S over that of the standard model. Across 328S in the Pacific Ocean, experiment 1 has a small northward flux of IW (3.062.8 3 10 6 m 3 s21 ), which is balanced by an increase in the southward return flow of PDW (Fig. A2). The large deep overturning circulation in the Pacific Ocean is maintained in experiment 1. However, the conversion of low silica concentration intermediate water to higher silica concentration PDW increases the southward transport of PDW and reduced the net silica flux across 328S. South of Australia and Africa experiment 1 reverses (now eastward) and reduces the transport of extreme AABW (layer 21–23) and increase the eastward transport of generic AABW (layer 20) over that of the standard model. The increased eastward transport of AABW increases the eastward silica flux. Given 1) the sensitivity of the silica flux to relatively small changes in the overturning and horizontal circulations; 2) the uncertainty as to how tightly the silica constraints should apply, given present understanding of the silica cycle; and 3) that the conversion of high oxygen, low silica IW to low oxygen, high silica UDW in the subtropical gyres is inconsistent with tracer patterns, we choose the standard model as our preferred

FIG. A1. Comparison of the overturning cell (310 6 m 3 s21 ) north of 188S in the Indian Ocean between the standard model, expt 1, expt 2, and expt 3 for SW: surface water, TW: thermocline water, IW: intermediate water, IDW: Indian Deep Water, LDW: lower deep water, and BW: bottom water.

JANUARY 2001

SLOYAN AND RINTOUL

171

FIG. A2. Comparison of the overturning cell (310 6 m 3 s21 ) north of 328S in the Pacific Ocean between the standard model, expt 1, expt 2, and expt 3 for TW: thermocline water, IW: intermediate water, PDW: Pacific Deep Water, LDW: lower deep water, and BW: bottom water.

solution. In any case, the most significant conclusions of this analysis regarding the role of the Southern Ocean in the global overturning circulation are not sensitive to the weight imposed on the silica constraint. REFERENCES Barnier, B., L. Siefridt, and P. Marchesiello, 1995: Thermal forcing for a global ocean circulation using a three year climatology of ECMWF analysis. J. Mar. Syst., 6, 363–380. Baumgartner, A., and E. Reichel, 1975: The World Water Balance. Elsevier, 179 pp. Beal, L. M., and H. L. Bryden, 1997: Observation of an Agulhas Undercurrent. Deep-Sea Res., 44, 1715–1724. Budd, W. F., P. A. Reid, and L. J. Minty, 1995: Antarctic moisture flux and net accumulation from global atmospheric analyses. Ann. Glaciol., 21, 149–156. Carmack, E. C., 1990: Large scale physical oceanography of polar oceans. Polar Oceanography Part A: Physical Science, W. O. Smith Jr., Ed., Academic Press, 171–222. Coachman, L., and K. Aagaard, 1988: Transport through Bering Strait: Annual and interannual variability. J. Geophys. Res., 93, 15 535–15 539. Cresswell, G. R., A. Frische, J. Peterson, and D. Quadfasel, 1993: Circulation in the Timor Sea. J. Geophys. Res., 98, 14 379– 14 389. da Silva, A. M., C. C. Young, and S. Levitus, 1994: Atlas of Surface Marine Data 1994. Vol. 1: Algorithms and Procedures, NESDIS Tech. Rep. 6, U.S. Department of Commerce, 83 pp. DeMaster, D. J., 1981: The supply and accumulation of silica in the marine environment. Geochim. Comsochim. Acta, 45, 1715– 1732. Dickson, R. R., and J. Brown, 1994: The production of North Atlantic Deep Water: Sources, rates and pathways. J. Geophys. Res., 99, 12 319–12 341. Doö¨s, K., and A. Coward, 1997: The Southern Ocean as the major Upwelling zone of North Atlantic Deep Water. Int. WOCE Newsl., 27, 3–4. Fahrbach, E., G. Rohardt, M. Schro¨der, and V. Strass, 1994: Transport and structure of the Weddell Sea. Ann. Geophys., 12, 840–855. Frew, R. D., K. Heywood, and P. F. Dennis, 1995: Oxygen isotope study of water masses in the Princess Elizabeth Trough. Mar. Chem., 49, 141–153. Gent, P. R., and J. C. McWilliams, 1990: Isopycnal mixing in ocean circulation models. J. Phys. Oceanogr., 20, 150–155. , J. Willebrand, T. J. McDougall, and J. C. McWilliams, 1995: Parameterizing eddy-induced tracer transports in ocean circulation models. J. Phys. Oceanogr., 25, 463–474.

Georgi, D. T., and J. M. Toole, 1982: The Antarctic Circumpolar Current and the oceanic heat and freshwater budgets. J. Mar. Res., 40, 183–197. Gill, A. E., 1973: Circulation and bottom water production in the Weddell Sea. Deep-Sea Res., 20, 111–140. Goodman, P. J., 1998: The role of North Atlantic Deep Water formation in an OGCM’s ventilation and thermohaline circulation. J. Phys. Oceanogr., 28, 1759–1785. Gordon, A. L., 1975: General ocean circulation. Numerical Models of Ocean Circulation, National Academy of Science, 39–53. , 1981: South Atlantic thermocline ventilation. Deep-Sea Res., 28, 1239–1264. , 1986: Interocean exchange of thermocline water. J. Geophys. Res., 91, 5037–5046. , and P. Tchernia, 1972: Waters of the continental margin off the Ade´lie Coast Antarctic. Antarctic Oceanology II: The Australian–New Zealand Sector, Antarctic Research Series, Vol. 19, D. E. Hayes, Ed., Amer. Geophys. Union, 59–69. , and B. A. Huber, 1990: Southern Ocean winter mixed layer. J. Geophys. Res., 95, 11 655–11 672. Guilyardi, E., G. Madec, and L. Terray, 1999: The role of lateral ocean physics in the upper ocean thermal balance of a coupled ocean–atmosphere GCM. Note du Poˆle de Mode´lisation 13, IPSL, 11 pp. Hastenrath, S., and L. Greischar, 1993: The monsoonal heat budget of the hydrosphere–atmosphere system in the Indian Ocean sector. J. Geophys. Res., 98, 6869–6881. Hellerman, S., and M. Rosenstein, 1983: Normal monthly wind stress over the world ocean with error estimates. J. Phys. Oceanogr., 13, 1093–1104. Hogg, N. G., P. Biscaye, W. Gardner, and W. J. Schmitz Jr., 1982: On the transport and modification of Antarctic Bottom Water in the Vema Channel. J. Mar. Res., 40, 231–263. Holfort, J., 1994: Großraümige Zirculation und meridionale Transporte im Su¨datlantik (English abstract). Ph.D. dissertation, Berichte aus dem Institut fur Meereskunde an der Christian-Albrechts-Universita¨t Kiel, No. 260, 95 pp. Hufford, G. E., M. S. McCartney, and K. A. Donohue, 1997: Northern boundary currents and adjacent recirculations off southwestern Australia. Geophys. Res. Lett., 24, 2797–2800. Jackett, D., and T. J. McDougall, 1997: A neutral density variable for the worlds oceans. J. Phys. Oceanogr., 27, 237–263. Jacobs, S. S., and D. T. Georgi, 1977: Observations on the southwest Indian/Antarctic Ocean. Deep-Sea Res., 24, 43–84. Lee, T., and J. Marotzke, 1997: Inferring meridional mass and heat transports of the Indian Ocean by fitting a general circulation model to climatological data. J. Geophys. Res., 102, 10 585– 10 602. Levitus, S., and T. Boyer, 1994: World Ocean Atlas 1994. Vol. 4:

172


Temperature, Tech. Rep. NOAA Atlas NESDIS 4, U.S. Department of Commerce, 117 pp. Lutjeharms, J. R. E., N. D. Bang, and C. P. Duncan, 1981: Characteristics of the current east and south of Madagascar. Deep-Sea Res., 28, 879–899. Macdonald, A. M., 1998: The global ocean circulation: A hydrographic estimate and regional analysis. Progress in Oceanography, Vol. 41, Pergamon, 281–382. , and C. Wunsch, 1996: An estimate of global ocean circulation and heat fluxes. Nature, 382, 436–439. Mantisi, F., C. Beauverger, A. Poisson, and N. Metzl, 1991: Clorofluoromethanes in the western Indian sector of the Southern Ocean and their relations with geochemical tracers. Mar. Chem., 35, 151–167. Mantyla, A. W., and J. L. Reid, 1995: On the origins of deep and bottom waters of the Indian Ocean. J. Geophys. Res., 100, 2417– 2439. McCartney, M. S., and R. A. Curry, 1993: Transequatorial flow of Antarctic Bottom Water in the western Atlantic Ocean: Abyssal geostrophy at the equator. J. Phys. Oceanogr., 23, 1264–1276. McDougall, T., 1987: Thermobaricity, cabbeling, and water-mass conservation. J. Geophys. Res., 92 (C5), 5448–5464. McIntosh, P. C., and S. R. Rintoul, 1997: Do box inverse models work? J. Phys. Oceanogr., 27, 291–208. Metzl, N., B. Moore, and S. R. Rintoul, 1990: Resolving the intermediate and deep advective flows in the Indian Ocean by using temperature, salinity, oxygen and phosphate data: The interplay of biogeochemical and geophysical tracers. Geochemical Variability in the Oceans, Ice and Sediments, Paleogeography, Palaeoclimatology, Palaeoecology, Vol. 89, L. P. Labeyrie and C. Jeandel, Eds., Elsevier Science, 81–111. Meyers, G., R. J. Bailey, and A. P. Worby, 1995: Geostrophic transport of Indonesian throughflow. Deep-Sea Res., 42, 1163–1174. Moore, M. I., and J. L. Wilkin, 1998: Variability in the South Pacific Deep western boundary Current from current observations and a high-resolution global model. J. Geophys. Res., 103, 5439– 5457. Munk, W. H., 1966: Abyssal recipes. Deep-Sea Res., 13, 707–730. , and K. Wunsch, 1998: Abyssal recipes. II: Energetics of tidal and wind mixing. Deep-Sea Res., 45, 1977–2010. Orsi, A. H., W. D. Nowlin Jr., and T. Whitworth III, 1993: On the circulation and stratification of the Weddell Gyre. Deep-Sea Res., 40, 169–203. Peterson, R., 1992: The boundary currents in the western Argentine basin. Deep-Sea Res., 39, 623–644. , and T. Whitworth III, 1989: The Subantarctic and Polar Fronts in relation to deep water masses through the southwestern Atlantic. J. Geophys. Res., 94, 10 817–10 838. , and L. Stramma, 1991: Upper-level circulation in the South Atlantic Ocean. Progress in Oceanography, Vol. 26, Pergamon, 1–73. Polzin, K. L., J. M. Toole, J. R. Ledwell, and R. W. Schmitt, 1997: Spatial variability of turbulent mixing in the abyssal ocean. Science, 276, 93–96. Rahmstorf, S., and M. H. England, 1997: Influence of Southern Hemisphere winds on North Atlantic Deep Water flow. J. Phys. Oceanogr., 27, 2040–2054. Reid, J. L., 1973: Transpacific hydrographic sections at Lats. 438S and 288S: The SCORPIO Expedition. III: Upper water and a note on southward flow at mid-depth. Deep-Sea Res., 20, 39–49. , 1986: On the total geostrophic circulation of the South Pacific Ocean: Flow patterns, tracers and transports. Progress in Oceanography, Vol. 16, Pergamon, 1–61. , 1989: On the total geostrophic circulation of the South Atlantic Ocean: Flow patterns, tracers and transports. Progress in Oceanography, Vol. 23, Pergamon, 149–244. Ridgway, K. R., and J. S. Godfrey, 1994: Mass and heat budgets in the East Australian current: A direct approach. J. Geophys. Res., 99, 3231–3248.

VOLUME 31

Rintoul, S. R., 1991: South Atlantic interbasin exchange. J. Geophys. Res., 96, 2675–2692. , 1998: On the origin and influence of Ade´lie land bottom water. Ocean, Ice and Atmosphere: Interactions at the Antarctic Continental Margin, Antarctic Research Series, Vol. 75, S. Jacobs and R. Wiess, Eds., Amer. Geophys. Union, 151–171. , and J. L. Bullister, 1999: A late winter hydrographic section from Tasmania to Antartica. Deep-Sea Res., 46, 1417–1454. Robbins, P. E., and J. M. Toole, 1997: The dissolved silica budget as a constraint on the meridional overturning circulation of the Indian Ocean. Deep-Sea Res., 44, 879–906. Roemmich, D., and T. McCallister, 1989: Large scale circulation of the North Pacific Ocean. Progress in Oceanography, Vol. 22, Pergamon, 171–204. , S. Hautala, and D. Rudnick, 1996: Northward abyssal transport through the Samoan passage and adjacent regions. J. Geophys. Res., 101, 14 039–14 055. Roether, W., R. Schlitzer, A. Putzka, P. Beining, K. Bulsiewicz, G. Rohardt, and F. Delahoyde, 1993: A chlorofluoromethane and hydrographic section across Drake Passage: Deep water ventilation and meridional property transport. J. Geophys. Res., 98, 14 423–14 435. Rosenberg, M., R. Eriksen, S. Bell, and S. Rintoul, 1997: Aurora Australis Marine Science Cruise AU9404—Oceanographic field measurements and analysis. Tech. Rep. 8, Cooperative Research Centre for Ant. and Southern Ocean Envirn., 53 pp. [Available from Antarctic CRC, GPO Box 252-80, Hobart, Tasmania 7001, Australia.] Saunders, P. M., and B. R. King, 1995: Oceanic fluxes on the WOCE A11 section. J. Phys. Oceanogr., 25, 1942–1957. SAVE, 1992a: South Atlantic Ventilation Experiment (SAVE), Chemistry, physical and CTD data Legs 1, 2 and 3. Tech. Rep. 92-9, Scripps Institution of Oceanography, University of California, San Diego, 729 pp. , 1992b: South Atlantic Ventilation Experiment (SAVE), Chemistry, physical and CTD data: Legs 4 and 5. Tech. Rep. 92-10, Scripps Institution of Oceanography, University of California, San Diego, 625 pp. Schmitt, R. W., P. S. Bogden, and C. E. Dorman, 1989: Evaportaion minus precipitation and density fluxes for the North Atlantic. J. Phys. Oceanogr., 19, 1208–1221. Schmitz, W. J., Jr., 1995: On the interbasin-scale thermohaline circulation. Rev. Geophys., 33, 151–173. , 1996a: On the World Ocean circulation. Vol. 1, Some global features/North Atlantic Circulation. Tech. Rep. WHOI-96-O3, Woods Hole Oceanographic Institute, 140 pp. , 1996b: On the World Ocean circulation. Volume II: The Pacific and Indian Oceans/A global update. Tech. Rep. WHOI-96-08, Woods Hole Oceanographic Institute, 237 pp. Schott, F., M. Fieux, J. Swallow, and R. Zantopp, 1988: The boundary currents east and north of Madagascar. 2: Direct measurements and model comparisons. J. Geophys. Res., 93, 4963–4774. Sloyan, B. M., 1997: The circulation of the Southern Ocean and the adjacent ocean basins determined by inverse methods. Ph.D. thesis, Institute of Antarctic and Southern Ocean Studies, University of Tasmania, 479 pp. [Available from IASOS University of Tasmania, GPO Box 252-77, Hobart, Tasmania 7001, Australia.] , and S. R. Rintoul, 2000: Estimates of area-averaged diapycnal fluxes from basin-scale budgets. J. Phys. Oceanogr., 30, 2320– 2341. , and , 2001: Circulation, renewal, and modification of Antarctic mode and intermediate water. J. Phys. Oceanogr., in press. Speer, K., and E. Tziperman, 1992: Rates of water mass formation in the North Atlantic Ocean. J. Phys. Oceanogr., 22, 93–104. , and W. Zenk, 1993: The flow of Antarctic Bottom Water into the Brazil Basin. J. Phys. Oceanogr., 23, 2667–2682. , S. R. Rintoul, and B. M. Sloyan, 2000: The diabatic Deacon cell. J. Phys. Oceanogr., 30, 3212–3222. Stommel, H., and A. B. Arons, 1960: On the abyssal circulation of

JANUARY 2001

SLOYAN AND RINTOUL

the worlds ocean. II: An idealized model of circulation pattern and amplitude in oceanic basins. Deep-Sea Res., 6, 140–154. Swallow, J., M. Fieux, and F. Schott, 1988: The boundary currents east and north of Madagascar. 1: Geostrophic currents and transports. J. Geophys. Res., 93, 4951–4962. Talley, L., 1984: Meridonal heat transport in the Pacific Ocean. J. Phys. Oceanogr., 14, 231–241. , and M. O. Baringer, 1997: Preliminary results from WOCE hydrographic sections at 808E and 328S in the central Indian Ocean. Geophys. Res. Lett., 24, 2789–2792. Taylor, H., A. L. Gordon, and E. Molinelli, 1978: Climate characteristics of the Antarctic Polar Front zone. J. Geophys. Res., 83, 4572–4578. Toggweiler, J. R., and B. Samuels, 1993a: Is the magnitude of the deep outflow from the Atlantic Ocean actually governed by Southern Hemisphere winds? The Global Carbon Cycle, NATO ASI Series I: Global Enviornmental Change, Vol. 15, M. Heimann, Ed., Springer-Verlag, 303–332. , and , 1993b: New radiocarbon constraints on the upwelling of abyssal water to the ocean’s surface. The Global Carbon Cycle, NATO ASI Series I: Global Enviornmental Change, Vol. 15, M. Heimann, Ed., Springer-Verlag, 333–363. , and , 1998: On the ocean’s large-scale circulation near limit of no vertical mixing. J. Phys. Oceanogr., 28, 1832–1852. Toole, J. M., and B. A. Warren, 1993: A hydrographic section across the subtropical South Indian Ocean. Deep-Sea Res., 40, 1973– 2019. Tre`guer, P., D. M. Nelson, A. J. Van Bennekorn, D. J. DeMaster, A. Leynaert, and B. Que`guiner, 1995: The Silica balance in the World Ocean: A reestimate. Science, 268, 375–379. Trenberth, K. E., and A. Solomon, 1994: The global heat balance: Heat transport in the atmosphere and ocean. Climate Dyn., 10, 107–134. Tsimplis, M. N., S. Bacon, and H. L. Bryden, 1998: The circulation of the subtropical South Pacific derived from hydrographic data. J. Geophys. Res., 103, 21 443–21 468. Tziperman, E., 1988: Calculating the time-mean oceanic general cir-

173

culation and mixing coefficient from hydrographic data. J. Phys. Oceanogr., 18, 519–525. Walin, G., 1982: On the relation between sea-surface heat flow and thermal circulation in the ocean. Tellus, 34, 187–195. Warren, B., 1981: Transindian hydrographic section at lat. 18 S: Property distribution and circulation in the South Indian Ocean. DeepSea Res., 28, 759–788. Whitworth, T., III, and W. D. Nowlin Jr., 1987: Water masses and currents of the Southern Ocean at the Greenwich Meridian. J. Geophys. Res., 92, 6462–6476. , , R. D. Pillsbury, and R. F. Weiss, 1991: Observations of the Antarctic Circumpolar Current and deep boundary current in the southwest Atlantic. J. Geophys. Res., 96, 15 105–15 118. , B. A. Warren, W. D. Nowlin Jr., S. B. Rutz, R. D. Pillsbury, and M. I. Moore, 1999: On the deep western-boundary current in the southwest Pacific basin. Progress in Oceanography, Vol. 43, Pergamon, 1–54. Wijffels, S., 1993: Exchanges between hemispheres and gyres: A direct approach to the mean circulation of the equatorial Pacific. Ph.D. thesis, Massachusetts Institute of Technology, Woods Hole Oceanographic Institution, 267 pp. , R. W. Schmitt, H. L. Bryden, and A. Stigebrandt, 1992: Transport of freshwater by the oceans. J. Phys. Oceanogr., 22, 155– 162. Wong, A. P. S., N. L. Bindoff, and A. Forbes, 1998: Ocean–ice shelf interactions and possible bottom water formation in Prydz Bay, Antarctica. Ocean, Ice and Atmosphere: Interactions at the Antarctic Continental Margin, Antarctic Research Series, Vol. 75, S. Jacobs and R. Wiess, Eds., Amer. Geophys. Union, 173–187. Wunsch, C., 1978: The North Atlantic general circulation west of 508W determined by inverse methods. Rev. Geophys. Space Phys., 16, 583–620. , 1984: An eclectic Atlantic Ocean circulation model. I: The meridonal flux of heat. J. Phys. Oceanogr., 14, 1712–1732. , 1996: The Ocean Circulation Inverse Problem. Cambridge University Press, 441 pp. , D. Hu, and B. Grant, 1983: Mass, heat, salt and nutrient fluxes in the South Pacific Ocean. J. Phys. Oceanogr., 13, 725–753.