Baroclinic Transport Variability of the Antarctic Circumpolar Current South of Australia (WOCE Repeat Section SR3) Stephenr

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 106, NO. C2, PAGES 2815-2832, FEBRUARY 15, 2001

Baroclinic transport variability of the Antarctic Circumpolar Current south of Australia (WOCE repeat section SR3) StephenR. Rintoul and SergueiSokolov CSIRO Marine Research and Antarctic Cooperative Research Centre Hobart, Tasmania, Australia

Abstract. Baroclinic transport variability of the Antarctic Circumpolar Current (ACC) near 140øEis estimatedfrom six occupationsof a repeat sectionoccupied as part of the World OceanCirculation Experiment (WOCE sectionSR3). The meantop-to-bottom volume transport is 147+10 Sv (mean+l standarddeviation), relative to a deep referencelevel consistentwith water mass properties and float trajectories. The location and transport of the main fronts of the ACC are relatively steady: the Subantarctic Front carries 105+7 Sv at a mean latitude between 51o and 52øS;the northern branch of the Polar Front carries 5+5 Sv to the east between 53ø and 54øS;the southern Polar Front carries 24+3 Sv eastward at 59øS; and two cores of the southern ACC front at 62ø and 64øS carry 18+3 and ll+3 Sv, respectively. The variability in net property transports is largely due to variability of currents north of the ACC, in particular, an outflow of 8+13 Sv of water from the Tasman Sea and a deep anticyclonicrecirculation carrying 22+8 Sv in the Subantarctic Zone. Variability of net baroclinic volume transport is similar in magnitude to that measuredat Drake Passage.In density layers, transport variability is small in deeplayers, but significant(range of 4 to 16 Sv) in the SubantarcticMode Water. Variability of eastwardheat transportacross SR3 is significant(range of 139øCSv, or 0.57x 10•5 W, relativeto 0øC)and large relative to meridionalheat flux in the Southern Hemispheresubtropical gyres. Heat transport changesare primarily due to variations in the westward flow of relatively warm water acrossthe northern end of the section.Weak (strong)westward flow and large (small) eastwardheat flux coincideswith equatorward(poleward) displacements of the latitude of zero wind stress curl.

1. Introduction a more seriouschallenge. Early attempts to measurethe barotropicflow using short-term current meter records The Antarctic CircumpolarCurrent (ACC) provides led to wildly differenttransport estimates(see Peter- the primary means by which mass, heat, salt, and other son[1988] for a review). The discrepancybetween these properties are transferred between the ocean basins. early direct velocity measurementsinspired a concerted The interbasin connectionprovided by the ACC is there- effort during the International SouthernOcean Studies fore a key element in the global overturning circulation. (ISOS) experimentin Drake Passage.High-resolution The vigorous interbasin exchangeaccomplished by the hydrographicsections revealed the bandednature of the ACC also opens the possibility of oceanic teleconnec- ACC and explained the earlier current meter results: tions, where anomalies formed in one basin may be car- an instrument deployed in one of the strong narrow ried around the globe to influence climate at remote fronts would give vastly different resultsfrom one de- locations [e.g., White and Peterson,1996; White and ployedin the weak flow betweenthe jets. During ISOS, Cherry, 1998]. current meter mooringsand pressuregauges were de- For these reasonsthe transport of the ACC has been ployed across the passageand maintained for a num- a subject of intenseoceanographic interest for decades. ber of years. The result of this dedicated effort was an While calculation of the geostrophictransport relative estimate of the mean absolute transport of the ACC: to a deep reference level from measurements of den- 1344-13Sv [ Whirworth,1983; Whirworthand Peterson, sity was straightforward, the barotropic flow presented 1985].This numberis oftentaken, alongwith the transport of the Gulf Stream through the Florida Straits, as Copyright2001 by the AmericanGeophysical Union. one of the few "knowns"in the oceanographer'scanon. The ISOS measurements also revealed the variability Paper number 2000JC900107 of the ACC. Deep pressuregauges rnoored on either side 0148-0227/01/2000JC900107509.00 of the passageprovided a 4 year record of the pressure

2815 2816 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC

Table 1. Occupationsof the WOCE SR3 Repe;•tHydrographic Line

Cruise Date Number of Stations Reference

AU9101 October 1991 24 Rintoul and Bullister [1999] AU9309 March 1993 47 Rosenberget al. [1995a] AU9407 January 1994 53 Rosenberget al. [1995b] AU9404 January 1995 51 Rosenberget al. [1996] AU9501 July 1995 54 Rosenberget al. [1997] AU9601 September 1996 57 Rosenbergct al. [1997]

differenceacross the passage,which could be relateddi- moorings(consisting of temperatureand/or conductiv- rectlyto changesin net transportthrough the passage ity sensorsat 6-7 levelsmoored on the 2500m isobath) [Whitworthand Peterson, 1985]. The pressurerecords on either side of the passageprovided a i year record showedthat the transportvaried by about 15-20 Sv on of baroclinic transport variability. While the overall timescalesof a few days [Whirworth,1983]. Transport consistencyin hydrographicstructure revealed on dif-

7.2¾

1.71 Tasman •Sea 2

Figure 1. Locationof theWOCE SR3 section (dots) and the SURVOsTRAL high density XBT line (diamonds).Contours are dynamicheight at 500m relativeto 3000m fromthe stationdata set(not the mapped fields) of Gouretskiand Janeke [1998]; the field is mapped using a biharmonic method[Sokolov and Rintoul, 1999]. Depths less than 3000 m areshaded. Trajectories of several AutonomousLagrangian Circulation Explorer (ALACE) floats [Davis et al., 1992]are shown. The floatsare deployedat approximately950 dbar. The deploymentlocation is indicatedwith a +. The lengthof eachsegment represents the drift over27 days. RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC 2817 ferent transectsacross Drake Passage [e.g., Reid and in unitsof kg m-a, all salinityvalues are on the prac- Nowlin Jr., 1971] has contributedto an emphasison tical salinity scale, and all temperatures are potential the barotropic component of the variability, the baro- referenced to the sea surface. The emphasisof this pa- clinic variability from the dynamicheight mooringsis of per is on the baroclinic transport variability. Readers comparable magnitude to the absolute transport vari- interested in a more complete discussionof the water ability inferred from the pressuregauges, although the masses and circulation at SR3 are referred to Rintoul former varies on longer timescales. and Bullister [1999]. Yaremchuk½t al. [thisissue] pro- Most of what we know about the mean flow and vari- vide a different perspective on the mean flow and variability of the ACC is based on these ISOS results in ability at SR3, based on results of a variational model. Drake Passage. No similar measurementprogram had Figure 1 showsthe location of the SR3 section, the been carried out at other longitudes until the World mean dynamic topography at 500 m relative to 3000 m OceanCirculation Experiment (WOCE). DuringWOCE and the bathymetry in the area. The strongest flow of the transport of the ACC was monitored at the South- the ACC is found on the north side of the Southeast In- ern ocean "chokepoints," using repeat conductivity- dian Ridge. Near 140øE the ridge and the current turn temperature-depth(CTD) and expendablebathyther- to the southeast. The South Tasman Rise lies north of mograph(XBT) sections,pressure gauges, moored in- the ACC and reaches sufficiently close to the sea sur- struments, and remote sensing.In this paper we present face (<1500 m depth) to providea significantobstacle results from six occupationsof the WOCE repeat hy- to flow in this region. Streamlinesassociated with the drographicsection (SR3) acrossthe Australian choke- eastwardflow south of the ridge and west of 140øE turn point section. (The use of the term chokepoint,while to the north upon encounteringthe shoalingtopogra- common,is perhaps misleadingfor sectionsas long as phy of the ridge and then turn back to the southeast thosesouth of Africa (m4000km) and Australia(m2500 again oncethey crossthe ridge. km).) We describethe primary circulation features between 3. Results Tasmania and Antarctica and quantify their baroclinic 3.1. Mean Density Structure transport variability. Variability in the net fluxes of mass, heat, salt, and nutrients is documented and re- To provide a context for the transport discussionto lated to changesin specificcurrent branchessouth of follow, Figure 2 showsa sectionof neutral density along Australia. In particular, the interbasin exchange of SR3 constructed from a mean of the six repeat sections. properties south of Australia is shown to be sensitive Density surfacesgenerally slope upward to the south to variations in the strength of westward flow observed across the section. At the northern end of the section acrossthe northern end of SR3. These changesare the overall trend of the neutral surfacesis reversed, and shown to be linked to meridional shifts in the latitude density surfaces shoal to the north to form a shallow of zero wind stress curl. The results are compared to isopycnalbowl between 50øS and 44øS. A thick pycnos- earlier measurementsfrom Drake Passage. Finally, we tad with density between 26.9 and 27.0 fills the upper discussthe implications of the observedtransport vari- part (150-600m) of the bowl. This is the local variety ability for interbasin exchangeand for the representa- of SubantarcticMode Water (SAMW) formedby deep rivenessof single occupationsof hydrographiclines, as convection in winter. obtained during the WOCE "one-time" survey. The horizontal density gradientsextend from the sea surfaceto the seafloor. Assuminga deep referencelevel 2. Data (the choiceof referencelevel is consideredin more detail below),the meanbaroclinic structure illustrated in Fig- The WOCE SR3 section was occupied six times dur- ure 2 suggeststhe following circulation: westwardflow ing the 5 years between October 1991 and September north of 46øS, a broad band of strong eastwardflow be- 1996, coveringeach season of the year (Table 1). In tween 500 and 54øS, weak flow between 540 and 58øS, each case, CTD stations were occupied to within 10- and a broad band of moderate eastward flow between 15 m of the seafloorwith a nominal station spacingof 580 and 66øS. 56 km, with tighter station spacingin regionsof steep The mean field of course smooths some features of the dynamic or bottom topography. The first occupation flow revealed in the individual sections. To illustrate in October 1991 had coarser spatial resolution in some the variability of the baroclinic structure in a compact locations because of extreme weather and sea ice con- way, dynamic height at the sea surfacerelative to 2000 ditions. In addition, the sections do not exactly co- m is shownfor each of the sections(Figure 3). The incide at the southern end, where the ship was often greatest variability is found at the northern end of the required to divert from the nominal cruise track to find section. For example, the relatively shallow isopycnal a way through the sea ice. Details concerningthe CTD bowl between 500 and 44øS in the mean field is the calibration, bottle analyses, problems encountered, and result of averagingsections with deep bowls (October samplingcarried out on each cruisecan be found in a se- 1991and January1995) with sectionswith flatter isopy- ries of technicalreports [Rosenberg½t al., 1995a, 1995b, cnals(January 1994, July 1995, and September1996). 1996, 1997]. In the following,all densityvalues quoted The southward decreasein dynamic height between 500 are neutral density% [Jackerrand McDougall,1997] and 54øS is relatively steep in September 1996 and Jan- 2818 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC

MEAN¾n on WOCE SR3 •r• .9 5OO

1000 27.5

1500 '-27.8

2000

2500

3OOO

28.1

3500

4000

45 50 55 60 65 LATITUDE (S)

Figure 2. Meansection of neutraldensity anomaly (kg m-a) fromsix occupations of SR3.

5 i i i i i

ß , . Sep 1996 Jul 1995 4.5 ' ' ' , ...... Jan 1995 • - '/ x " , .... Jan 1994 - - - Marl993

4...... iß ...... '" '4, ...... '...... : Oct1991

E '- 3.5 ...... : ...... "'""• ..... :...... ":'...'•..,,,.:.•..'w ...... -,,,...... • ,,:' '., : : •: E

•o 3 ...... '...... "•'.....

ß -•'" 2.5...... ! •'•' • -'•' •,':%...... ""'""...... ' ..... ",', • ':...... i ...... "• ,, *'11,l,,i s _o• • • , • ,,,,, <: 2 ...... '...... /•':'' '• ...... •','.,',''•..... '• ,';,,,;''...... '.'''

1.5 ...... : ...... :' '•--,•...... •'',• :• ...... •' %.... :...... ' .....

0.5 I I 45 5O 55 6O 65 LATITUDE (S)

Figure 3. Dynamicheight (dyn m) at the seasurface relative to 2000 rn on six occupationsof SR3. Each curve is offset by 0.5 dyn m. RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC 2819 uary 1994 and more gradual in July 1995. Two sections Westward flow over the upper continental slope, as- (January1994 and September1996) show a dip in dy- sociatedwith the Antarctic Slope Front (ASF) [Ja- namic height north of the ACC, indicating the presence cobs,1991; Whirworth½t al., 1998], is found at many of a cyclonicmeander or eddy pinched off from the Sub- locationsaround the continent, including at SR3. Cur- antarcticFront (SAF). rent meter measurements in the eastern Weddell Sea indicate strong westwardbottom velocitiesassociated 3.2. Choice of Reference Level with this flow [Fahrbachet al., 1994]. Upper ocean A deep reference level is usually assumed for the velocitiesinferred from ice drift ITchernia and Jean- ACC, based primarily on evidence from direct velocity nin, 1983;H½il and Allison, 1999],surface drifters, and measurementsin Drake Passage.Current meter records acousticDoppler current profiler measurements [Bind- there show that the ACC reachesto great depth and, off ½t al., 2000] are westwardacross the southernend more quantitatively, that referencing geostrophic cal- of the SR3 section. For those sections that extended as culations to velocity measurements across the passage far as the continental shelf a reference level was chosen gives similar transports to estimatesmade assuminga to give westwardflow at all depthsin the ASF. level of no motion at the deepest common depth at each The velocity at the deepest common depth is held stationpair [Whirworthet al., 1982; Whirworth,1983]. constant through the "bottom triangle" beneath the Rintoul and Bullistcv[1999] show that a near-bottom deepest common depth at each station pair. Because reference level is consistent with water mass properties the reference level is close to the bottom, the transport and float trajectoriesat SR3 (samplesof the latter are estimates are insensitive to the treatment of the bottom shownin Figure 1), exceptat the southernend of the triangles. section. There they argue that Adelie Land Bottom While the water properties and ALACE trajecto- Water [Rintoul, 1998] and Ross Sea Bottom Water ries provide some guidance on a reasonablechoice of formed to the east of SR3 must flow west on aver- reference level, the reader should keep in mind that age, requiring a referencelevel between newly formed all the results shown are for the baroclinic transport Antarctic Bottom Water (AABW) and older Circum- alone. (Throughout the paper, baroclinicis used to polar Deep Water (CDW) above. They took a refer- refer to velocity and transport relative to a reference ence level coincident with the 0 = -0.2øC isotherm, level; barotropic is used to refer to the velocity at the which in this region also approximately coincideswith referencelevel). Direct velocitymeasurements in other the 34.682 isohaline and the %=28.318 surface. We parts of the Southern Ocean suggestthat significant adopt the same strategy here. barotropic flows exist. The geostrophictransport rela-

WOCE SR3 - OPTIMAL RL

200 ! i i ! Sep 1996 ...... Jul 1995 180 ...... Jan 1995 Jan 1994 Mar 1993 160 Oct 1991

140 o.

.o • 120

o o• 100 z

"' 80

z_ 60

-20 45 50 55 60 65 LATITUDE (S)

Figure 4. Cumulativevolume transport (Sv) integratedfrom southto north acrosssix occupationsof SR3 relativeto a bestguess reference level (seetext). 282o RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC

WOCE SF{3 - OPTIMAL RL 7 I I I

ß SAF SF i I i i TAS CCR; PF SF I ' , SAZ , .,• • PF , , ,

Sep 1996 6 1 •: SF , TA$ CCR PF SF • SAZ • • PF • •' • • ß i i i i

Jul 1995

SAF SF i i TAS • , SAZ PF i i i i i

Jan 1995

SAF SF • I TAS CCRß PF SF I ' , SAZ, ,'I•iB PF ' ' ' ' •: ' . Jan1994 •<3

I T:AS ' i__ ! PFi SF

Mar 1993 2

ß i

T.AS • • SAZ PF I I I

Oct 1991

10 Sv

0 500' 1000 15oo 2000 2500 km 0 [ I i I I i I 45 50 55 60 65 70 LATITUDE(S)

Figure 5. Transport per unit width for six occupationsof SR3. Area under each curve is proportional to transport; the scale bar equals 10 Sv. Eastward flow is positive. Curves are offset by 1 Sv km-•. The locationof major frontsand circulationfeatures are shownabove each plot: TAS, Tasman outflow; SAZ, Subantarctic Zone recirculation; CCR, cold core ring or meander; SAF, Subantarctic Front; PF, Polar Front; SF, southern ACC front. The brackets indicate the limits used to sum up the transports of individual current branches. tive to any choiceof referencelevel may not give a good above)has a meanof 147Sv and a rangeof 135-158Sv indication of the absolute transport. (Figure4). (Transportsrelative to otherreference level choicesare discussedin section4). The 1991 section 3.3. Vertically Integrated Baroclinic Transport does not extend as far south as the later sections; the The distribution of vertically integrated transport origin for the cumulative transport curve for this sec- with latitude is illustrated in Figures4 and 5. (The tion is taken to be the mean of the other five sections at reader may also find it useful to refer to the schematic the latitude of the southernmost station pair in 1991. summary of the circulation in Figure 6, which is dis- The net transport south of 55øS is 50 Sv to the east cussedin section3.7.) The net baroclinictransport (Figure4). Threeweak transport maxima occur within (relative to the "best guess"reference level described this broadband of eastwardflow (Figure5). The trans- RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC 2821

port peak between 580 and 60øS correspondsto the have similar •9-S characteristics to stations in the west- southernbranch of the Polar Front (PF) [Rintouland ward limb of the recirculation between 460 and 48øS Bullister,1999], a featuredefined by a deepeningof the (red curves). These profilestherefore suggest that a temperatureminimum layer to the north (and at some pool of AAIW recirculates north of the ACC. longitudesby a seasurface temperature (SST) gradi- Stations north of 46øS on the same occupation of SR3 ent, hencesometimes called the "surfaceexpression" of are distinctlywarmer and saltier (greencurves in Plate the PF). The maximanear 620 and 64øScoincide with 1) than stationsto the south. Typical valuescharac- ironts whose characteristics are similar to the "south- terizing the salinity minimum water are •9:4.5ø-5.0øC ern ACC front" (SACCF) of Orsi et al. [1995]. At and S:34.44. Geostrophiccalculations suggest this wa- SR3 the SACCF appearsto split into two branchesor ter is also flowing to the west. The water mass prop- to meander acrossthe section, perhapsin responseto erties are consistent with this interpretation: the the topographicbump seennear 63øSin Figure 2. The curves north of 46øS on SR3 overlap the envelope of "southernboundary" of the ACC (SB) [Orsi e! al., •9-S curves observed at 43øS in the southern Tasman 1995]sometimes merges with the SACCF (e.g.,July Sea (magentacurves in Plate 1), suggestingthe warm 1995, Figure 5) and sometimesappears as a distinct salty variety of AAIW is carried from the Tasman Sea weakcore of eastwardflow (e.g., January1995). The and flows south and west across SR3. Property distri- latitude and transport of the fronts south of 55øS are butions below the AAIW suggest the Tasman outflow remarkably similar on each of the six sections. extendsto the seafloor [Rintoul and Bullister, 1999]. The cumulativetransport increasesby about 100-120 The envelope of •9-S curves at 43øS in the Tasman Sv between550 and 48øS (Figure 4). The rapidincrease Sea is broad. The AAIW there is a mixture of cold, in transport reflectsthe contribution of the SAF and the fresh, high oxygenAAIW entering the Tasman Seafrom northern branch of the PF. Each section shows a num- the south, and warm, salty, low-oxygenAAIW entering ber of transportmaxima that correspondto multiple the Tasman Sea from the east, north of New Zealand coresof the SAF and the northernPF (Figure5). The [Wyrtki, 1962; $okolovand Rintoul, 2000]. The prop- picture of the ACC as consistingof a few well-defined erties of the AAIW entering the basin from the east frontsseparated by regionsof uniformwater mass prop- are illustrated by •9-S curves from the East Australia erties,developed from DrakePassage experience [e.g., Current at 30øS(black curvesin Plate 1). Nowlin and Clifford,1982], does not apply to the ACC The dynamic topography map in Figure i illustrates at SR3. the connectionbetween the flow at SR3 and the regional Each of the curves in Figure 4 slopesdown to the circulation. The 1.71 dyn m contour loops through the north at the northern end of SR3, correspondingto southern Tasman Sea before turning back to the west, westwardflow. The strengthof the westwardflow varies consistent with weak outflow from the Tasman Sea. betweensections: in January1995 the net transportbe- This patternagrees with observationsat 155øE(WOCE tween the cumulative transport maximum at 48øS and P11S), whichshow sharp property fronts near 38øS,in- the Tasmaniacoast is 50 Sv to the west;in July 1995 dicating that AAIW entering from the south reachesno the correspondingnumber is about10 Sv (Figure4). farther north beforeturning back to the south [$okolov In particular, the cumulativetransport curvesfor two and Rintoul, 2000]. More confinedanticyclonic recir- cruises(March 1993 and January1995) diverge from culations are evident immediately north of the ACC to the other four north of this latitude. Flow across both the east and west of the South Tasman Rise. Part of sectionsis strongly westwardnorth of 46øS, between the recirculation to the west of the South Tasman Rise, the South Tasman Rise and the Tasmanian continental which is partially crossedby SR3, continues westward slope(Figure 5). As a result,the net baroclinictrans- acrossthe Great Australian Bight. port across the section at these times is about 20 Sv Both branches of westward flow also appear in float lower than the other sections. trajectories(Figure 1)[seealso Davis, 1998]. Two floats deployed in the southwest Tasman Sea moved south 3.4. Origin of Westward Flow South of Tasmania and westward through the gap between Tasmania and the South Tasman Rise; a third float deployed in the Rintoul and Bullister [1999] identify two coresof gap also moved westward. No floats moved eastward westward flow across the October 1991 section: west- through this gap. The Tasman Sea floats also support ward flow associatedwith an anticyclonicrecirculation the notion that the southern Tasman Sea is generally in the SubantarcticZone (SAZ) andan outflowof water characterized by sluggish and variable flow, with lit- from the Tasman Sea. The source of the westward flow tle net inflow from the south extending beyond about acrossSR3 can be most clearly illustratedby compar- 38øS.Three floats deployedalong SR3 west of the South ing O-S curvesat the depth of the salinity minimum of Tasman Rise carried out an anticyclonic loop between the AntarcticIntermediate Water (AAIW). For exam- about 1390 and 146øE, 490 and 45øS, consistent with ple, Plate i shows 0-S curves from the northern end of the idea that the SAZ recirculation on SR3 is at least the January 1995 section. Stations south of 46øS fall in part a local feature, as inferred from the water mass in a family of curveswith typical salinityminimum wa- properties. The large spatial scale and relatively long ter properties of about •9=4øC, S:34.35. Stations in transit times of the anticyclonicloops (typically more the eastwardflow between 480 and 50øS (blue curves) than 150 daysto completea circuit) suggestthe floats 2822 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC are tracing a permanentbut variablepart of the gen- sity layers lighter than 27.0 showsthe same pattern as eral circulation,rather than a transient eddy. While noted above for deeper layers: sections with low to- the numberof float trajectoriesin the regionis small, tal transport have less water flowing east in the light and inadequte to provide a quantitative referenceve- layers, while sections with high total transport have locity for transport calculations,all of the trajectories larger transport in light layers: the mean of the two are consistent with the sense of flow inferred from water low-transportsections (March 1993and January1995) mass properties and a deep referencelevel. is 9.0 Sv; the mean of the two high-transport sections The westward flows across the northern end of SR3 is 17.4 Sv. vary with time (Figure 5). The January1995 section illustrated in Plate 1 is an exampleof a time when both 3.6. Temperature, Salts and Nutrient the SAZ recirculation and the Tasman outflow are well Transports developed.In contrast,in January 1994and July 1995 The westward flow south of Tasmania is negatively there waslittle TasmanSea water present,and the flow correlated with net eastward transport acrossSR3: the northof 46øSwas weakly westward (January 1994) or sectionswith strong westward flow at the northern end eastward(July 1995). The transportof the SAZ recir- of the section(March 1993and January1995, Figure 5) culation varies from a maximum of 35 Sv in October alsohave low net eastwardtransport (Figure 4). The 1991 to a minimum of 12 Sv in March 1993. impact of strong westwardflow on the net transport of properties other than volume can be even more dra- 3.5. Transport in Density Layers matic for properties whose concentration varies with Anotherimportant aspectof the variabilityat SR3 is latitude. the changein transport of individual water masses.Are The net transport of temperature, salt, nutrients and periods of relatively high baroclinic transport associ- oxygen is included in Table 2. The range of temper- ated with a top-to-bottomincrease in the ACC affecting ature flux is 139øCSv, or 0.57 PW0c (where we use all layers, or is an increasein net transport associated PW0c to represent the conversionof temperature flux with larger transports of particular layers? to units of heat flux relative to 0øC obtained by mul- Plate 2 shows the transport acrosseach SR3 section tiplying the temperature flux by the density and spe- in neutral density layers. The net transport of interme- cific heat; becausethis is not strictly a heat flux when diate, deep, and bottom water is very steady: transport there is a nonzeromass flux (e.g., a changeof temper- of the AAIW and deeper layers varies about the mean ature units to K would give a differentnumber), we for each layer by < 4-2 Sv. Nevertheless,the two low- use the subscript to indicate the temperature reference transportsections (March 1993 and January1995, red used). Changesin baroclinicheat transportmay not in- and magentacurves in Plate 2), aresystematically lower dicate changesin the absolute heat transport between in transport in all density layersfrom 26.9 to 28.3, while the basinsif the barotropic flow can carry enoughheat the converseis true for the high-transport July 1995 and to compensatethe baroclinic variations. Such compen- September 1996 sections. sation would require that the barotropic heat trans- The greatest transport variability occurs in layers port vary on a similar timescale,with similar magnitude lighter than 7,• = 27.0. In particular, transport of but oppositesign, to the observedbaroclinic variations. SAMW (7,•: 26.9-27.0) variesfrom 4 to 15.5Sv. Part While the mechanism that would link the barotropic of this variability is seasonal: in winter, there is little and baroclinic variability in this way is not obvious to water on SR3 lighter than 26.8; in summer, warming us, we cannot rule out the possibility in the absenceof convertsSAMW to lighter densities. Therefore we ex- measurementsof the barotropic flow. pect larger transports in the 26.8-26.9 density classin Nutrient and oxygenfluxes across a singlesection are winter, and larger transports in lighter layers in sum- difficult to interpret in isolation, and there are few esti- mer, as seen in Plate 2. mates with which to compare our results. While nutri- The seasonalconversion of water from one density ent fluxes are not often quoted in papers such as this, classto another, however,is not the whole story. The the nutrient fluxes are straightforward to compute and net transport across each section summed over den- the nutrient budgetsprovide valuable constraintson the

Table 2. Baroclinic Property Fluxes AcrossSR3

Cruise Volume, Temperature, Heat, Salt, Nitrate, Phosphate, Silicate, Oxygen Sv øC PW0c kmol s-• kmol s-x kmol s-x kmol s-x kmol s-x

AU9101 149 419 1.87 5144 4512 332 8931 33786 AU9309 137 369 1.65 4705 4285 286 8091 29879 AU9407 149 431 1.93 5132 4448 309 8363 33918 AU9404 135 338 1.51 4653 4195 290 8103 30710 AU9501 158 477 2.13 5457 4596 322 8414 34851 AU9601 156 472 2.11 5379 4654 307 8714 34663 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC 2823 circulation [e.g.,Robbins and Took, 1997]. The useof the barotropicflow to simultaneouslycompensate the such constraintsrequires knowledgeof the variability baroclinicvariability in both temperatureand nutrient of the fluxes. Table 2 shows the changesin the baro- fluxes. For example,a barotropicflow to the east across clinic transports of nitrate, phosphate, and silica are the southern half of the section and to the west across large relative to the meridionalfluxes entering the sub- the northern half would decrease the eastward temper- tropical gyres (which are usuallyassumed to be close ature flux while increasingthe eastwardnutrient fluxes. to zero, given that there is no atmosphericpathway for nutrients and storagein the sedimentsat low latitudes 3.7. Transport Summary is believedto be small). More study is requiredto un- Figure6 summarizesthe meanbaroclinic transport derstand the implications of the nutrient flux variabil- and its variability south of Australia. Six major cur- ity, and we encourageother authorsto publish nutrient rent branchesare identified,although the boundarybe- flux estimates. One possibility is that variations in the tween each current branch or front is somewhat arbi- barotropic transport might compensateall or part of trary. Traditionaldefinitions of ACC fronts(e.g., the the baroclinic transport variability, as noted above for northernmostextent of a particularisotherm) generally temperature. However, given that the vertically aver- coincidewith a singlestation pair at the core of the aged temperature tends to increasetoward the north front, whilethe shearand transportassociated with the while the vertically averaged nutrient concentrations front may spanseveral station pairs. Becausethe focus tend to decreaseto the north, it might be difficult for here is on transport, we have relied on extrema and

7000, •40OE 150øE lflOøE • •zoo•

Figure 6. Schematicsummary of main CircubLtionfeatures on SR3' EAC, extensionof East Australian Current; TAS, Tasman outflow; SAZ, SubantarcticZone recirculation; SAF, Sub- antarctic Front; PF, Polar Front; SF, southernACC front; SB, southernboundary of the ACC; ASF, AntarcticSlope Front. Numbersgive top-to-bottom transports in Sv (mean +1 standard deviation). See text and Figure 5 for a descriptionof the integration limits for each current branch or front. 2824 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC zero crossingsin the transport versuslatitude curve for the southeastafter crossingSR3, staying roughly paral- each section, as well as water properties, to define the lel to the SoutheastIndian Ridge. Float trajectories are limits of each front (see Figure 5). By usinga phe- consistent with this shift to the south, and individual nomenonologicaldefinition (rather than, for example,a floats crossthe Macquarie Ridge through one of three band of latitudes) to definethe fronts we avoid intro- gaps, at 520, 540, or 56øS. ducingtransport variability causedby shifts in latitude The northern branch of the PF is found between 530 or orientation of the fronts. The transports in Figure 6 and 54øS on all six occupationsof SR3. It carries 54-5 are integrated over the entire water column; the uncer- Sv to the east. West of SR3, this front lies farther south tainties associated with each branch are 4-1 standard (between550 and 58øS [see, e.g., Gordonand Molinelli, deviation. The schematic streamlines are drawn to be 1982, Plate 141]); as the flow encountersthe shoaling consistentwith climatologicaldynamic height (Figure topography of the Southeast Indian Ridge, the front 1), float trajectories [Davis,1998], and additionalhy- shiftsequatorward, as might be expectedfor weak, deep drographicand XBT data from the region, as described reachingflow conservingpotential vorticity. below. Between 530 and 58øS, the flow across SR3 is weak Water originating in the Tasman Sea can be clearly and variable and the isopleths of all properties are rel- identified by its water properties, as described in sec- atively flat. This is consistentwith flow nearly aligned tion 3.4. Summing the transport over station pairs with with the sectionthrough this latitude band, as expected Tasman Sea water results in a westward flow of 84-13 from climatologicaldynamic height (Figure 1), trajec- Sv. As shown in Figures 1 and 6, part of the northward tories, and as indicated by the 50 of latitude northward inflow to the Tasman Sea returns to the south over the deflection of isotherms in this latitude band on the SUR- TasmanJan continental slope and continues west across VOSTRAL XBT section relative to their location on SR3. The large standard deviation indicates the vari- SR3 [Rintoul et al., 1997]. able nature of this flow. (Note that not all the SAMW These data sourcessuggest the southern PF also de- and AAIW crossingSR3 turns north into the Tasman flects to the north when encountering the ridge. Be- Sea; some continues to the east and enters the Pacific cause the ridge is further east at these latitudes, the southof New Zealand.) southern branch of the PF is not deflected north until To quantify the transport and variability of the re- after crossingSR3. Both the location(59øS) and the circulation identified in the SAZ, we define its northern transport(244-3 Sv) of the southernPF are remarkably edge to be the southern limit of Tasman Sea-influenced steady at SR3. water and then integrate south to the point where the South of 60øS, the transport of the southern ACC senseof flow reverses(i.e., the bottom of the isopycnal front at 62øS and 64øS is also remarkably steady: 18 4-3 bowlnorth of the SAF). This givesthe westwardflow on Sv and 11 4-3 Sv, respectively. The southern boundary the northern limb of the recirculation. An equal amount of the ACC (SB) and the AntarcticSlope Front (ASF) of the eastward flow south of this point is attributed to are not very well resolvedby the SR3 repeats.(The sea the recirculation; the remainder is taken to be part of ice over the continental slope at this longitude is often the SAF, as described below. deformedand mobile, making CTD operationsdifficult, The recirculation reaches from top-to-bottom, in- evenin summer.)This may contributeto the relatively volvesa substantial transport, and is somewhatvariable high variability of the transport estimatesthere. The in time (224-8 Sv). The zonal extent of the recircula- transport of the SB is 2.64-2.4 Sv, while the westward tion is not well defined by the hydrographic data. Since flow associated with the ASF is 1.44-0.9 Sv. it is deep reaching, it is likely to be confined to the east by the relatively shallow South Tasman Rise. Hy- 4. Discussion drographicsections to the east and west of SR3 (e.g., 132øE [Schodlocket al., 1997]and 155øE(WOCE sec- The results summarized above concernonly the baro- tion PllA) [Rosenberget al., 1995a])do not showthe clinic transport. Substantialbarotropic flows likely oc- deep isopycnalbowl north of the SAF that is so promi- cur in the Southern Ocean. For example, shipboard and nent on SR3, in agreement with Figure 1 which shows loweredacoustic Doppler current profiler (ADCP) mea- a tight recirculation confined between 135øE and the surementshave shown that significantbarotropic veloc- South Tasman Rise, embedded within a weaker anticy- ities exist in someACC fronts [Heywoodet al., 1999;K. clonic circulation that extends across the Great Aus- A. Donohueet al., Absolutegeostrophic velocity within tralian Bightø the Subantarctic Front in the Pacific Ocean, submitted The SAF is defined operationally to extend from the to Journal of GeophysicalResearch, 2000; hereinafter southern limit of the recirculation to the transport min- referred to as K. A. Donohue, submitted manuscript, imum betweenthe SAF and the PF (wherethe location 2000]. However,ADCP data is not yet sufficientlyac- of the P F is defined in the traditional way as the north- curate to providea referencefor transport calculations ernmost extent of temperature minimum water cooler acrosslong sections[K. A. Donohueet al., submitted than 2øC near 200 m depth [Botnikov,1963]). The manuscript,2000]. Quantitativedirect estimates of the mean transport of the SAF is 105 Sv and is surprisingly barotropic contribution would require sustained,spa- stablewith time (standarddeviation of 7 Sv). The dy- tially coherent,and highly accurate velocity measure- namic height map in Figure 1 showsthe SAF turns to ments acrossthe 2500 km span of the section, a very 12

10-

....--31.B" '"

•' // •,( EastAuetrali"a Current (30S) ....• • •.•.•..Tasman (43S) -,. .•,,,•-• SR3:north of46•S_ ..-- SR3:48-50S '-- -- -• _% -•

• SR3:4..6-.48S' .... . '-,,,•

i iii I i i ii iiiiiiiiiiii i i iiiiiiiiii i ii 34.3 34.4 34.5 34.6 34.7 34.8 salinity Plate 1. Potential temperature-salinity profiles from stations at the northern end of SR3 and from the TasmanSea. (The TasmanSea data at 430 and 30øScome from a 1989 R/V Franklin cruiseFR8910; data are availablefrom CSIRO Marine Research.) Dashedcurves are density relative to 1000 dbar. The salinity minimum of the AAIW is warmer and saltier at the northern end of SR3 and similar in properties to AAIW in the southern Tasman Sea.

WOCE SR3 - OPTIMAL RL

26- i i :: :: Sep!1996 I ! ! i i Jul1995 • i i ! i Jani995

26.5 - .1•i •--J i i i i Jan1994 ...... I :...... :...... :...... ; ...... Mar-:.1-993 ...... I • Oct1991 III B• SAMW 03 27 z

z 27.5 I :! I U.'CDW 28- iii] ...... ' I II III' II I"I;! ^^BW 28.5 -5 0 5 10 15 20 25 30 TRANSPORT (Sv) Plate 2. Variability of transport in neutral de]tsity layers for six occupationsof SR3. SAMW, Subantarctic Mode Water; AAIW, Antarctic Intermediate Water; UCDW, Upper Circumpolar Deep Water; LCDW, Lower Circumpolar Deep Water; and AABW, Antarctic Bottom Water. 2826 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC

LATITUDE OF ZERO WIND CURL 2O 2O

'' - 0

-10 • • -40 1990 1991 1992 1993 1994 1995 1996 1997 1998

Figure 7. Deviationof the latitudeof zerowind stresscurl from its meanvalue of 53øS(solid line, in degreesof latitude). The curl of monthlywind stressvalues from the National Centers for EnvironmentalPrediction (NCEP) reanalysis[Kalnay et al., 1996]is averagedbetween 1300 and 150øE and shown for the period 1980-1998. Solid dots indicate transport of Tasman Sea originwater acrossthe northernend of SR3 for the six occupationsof the section(in Sv, scale on right-handside of plot). Positivevalues are towardthe east. challengingtask (e.g., a 0.1 cms- • biaswould carry 10 Figure 7 showsan eight year record of the latitude Sv acrossSR3). The resultsof three recentinverse cal- of zero wind stress curl south of Tasmania. The mean culations [ Yaremchuket al., this issue;Sloyan and Rin- latitude is about 53øS. The median position is 2ø-3øof toul, 2000a; Ganachaud,1999] providesome evidence latitude farther south, reflecting the fact that the zero that the barotropic contribution to the net property curl line is most often near 55øS, with occasionallarge fluxes acrossSR3 may be relatively small. The models equatorward displacementsthat usually persist for a differ substantially, but all agree that only small ref- few months. erence level velocities are needed to satisfy the model The solid dots in Figure 7 show the transport of constraints, and their estimates of net property fluxes the Tasman outflow across the six occupations of SR3. are similar to those found here. While the number of sections is small, a relationship appears to exist between meridional shifts of the zero 4.1. Relationship Between Wind Stress Curl curl line and westward transport; poleward shifts of the and Tasman Outflow zero curl line (hencepoleward shifts of the subtropical We have shown that the transport variability across regime)correspond to largeTasman outflow. When the SR3 is dominated by changes at the northern end of zero curl line shifts north to about the latitude of Tas- the section. The heat transport variability, in partic- mania (43øS), the transport is near zero or eastward. ular, reflects variations in the amount of warm water Figure 7 suggeststhat four of the six occupationsof flowing west between Tasmania and the South Tasman SR3 were made during periods when the zero curl line Rise. A strong Tasman outflow can be thought of as was anomalously far north. Hence our estimate of the a poleward shift of the subtropical regime. This view mean transport of the Tasman outflow is likely biased suggestslooking for links between meridional shifts in low. On the basis of the mean position of the zero curl wind stresscurl and the strength of the Tasmanoutflow. line the mean transport of tire outflow is likely to be In closed basins the zero of the wind stress curl marks about 20 Sv. The strength of the Tasman outflow is the boundary between the cyclonicsubpolar gyres and correlated with the net eastward heat flux across the the anticyclonic subtropical gyres. While the zero curl section, so our mean heat flux estimate may be biased line in the zonally unbounded Southern Ocean doesnot high. coincidewith gyre boundariesin the same sense,it still Figure 7 suggestsmeridional shifts in wind stresscurl marks the transitionfrom subtropical(Ekman pump- are likely to play an important role in regulating the ing) to subpolar(Ekman suction) regimes. variability in interbasin exchangesouth of Tasmania. RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC 2827

De Ruijter and Boudra [1985] used an eddy-resolving Another perspectiveon the differencein baroclinic barotropic model to demonstratea similar relationship structurebetween Drake Passageand SR3 can be gained between poleward shifts of the latitude of zero wind by comparingdynamic heights at either sideof the pas- stress curl and enhancedinterbasin exchangesouth of sages.Transport moorings deployed for i year in Drake Africa. De Ruijter and Boudra speculatedthat variabil- Passagegave mean valuesof 1.374-0.05and 0.584-0.02 ity in the inflow of warm Agulhas water to the Atlantic dyn m at the northern and southern sides (for the may have a significant impact on climate in that re- dynamic height anomaly at 500 m relative to 2500 gion; westwardflow of relatively warm water south of m) [Whitworth,1983]. The correspondingmean val- Tasmania may influencesouthern Australian climate in ues from six occupationsof SR3 are 1.424-0.07 and a similar manner. 0.504-0.005dyn m, respectively.That is, the dynamic height is both higheron the high (equatorward)side and lower on the low (poleward)side of the ACC at 4.2. Comparison to Drake Passage SR3. The much longer SR3 section extends 120 far- How do the baroclinic transport estimates south of ther north and 3øfarther south than Drake Passageand Australia compare to Drake Passage, the only other crosseslighter water north of the SubtropicalFront and location where repeat measurements have allowed the denser Antarctic water in the south. As found in Drake transport variability of the ACC to be assessed? Nowlin Passage,the variability of dynamicheight anomaliesat and Clifford[1982] calculated baroclinic tranport above the northern end of SR3 is much higher than at the and relative to 2500 m for five sections across Drake southern end. Passageoccupied in 1975-1976 and found remarkably The mean baroclinic transport acrossSR3 relative to constant values: a mean of 87 Sv and a range of only the best guessreference level is 1474-9.7Sv. The vari- 4-1 Sv. Whirworth[1980] found slightly smallerval- ability is very similar to the standarddeviation of abso- ues, with a wider range, and noted an apparent seasonal lute transportat Drake Passage(10.5 Sv) estimatedby difference: the mean of 16 summer sections was 794-13 Whirworthand Peterson[1985]. The best estimateof Sv and of 6 winter sections was 714-15 Sv. A i year the mean transport at Drake Passageis 134 Sv (125 Sv record of transport above and relative to 2500 m from above2500 m [Whitworthand Peterson,1985] plus 9 transport mooringson either side of the passagegave a Sv below2500 m [Whitworth,1983]), resulting in an ap- mean value of 87 Sv, with a standard deviation of 5.5 parent convergenceof 13 Sv in the Pacific. To conserve Sv [Whitworth,1983]. massin the Pacific basin as a whole, the differencein ab- The transport above and relative to 2500 m south of solute transport between SR3 and Drake Passagemust Tasmania is substantially higher than that observedin be balancedby outflow through the Indonesianpassages Drake Passage. The mean of six CTD sections along and the BeringStraits (ignoringsmall masstransports SR3 is 107 Sv, with a rangeof 95-117 Sv (Table 3). A due to evaporation,precipitation and runoff). Esti- similar pattern is seenwhen transports relative to alter- mates of the transport of the IndonesianThroughflow native reference levels between 2000 m and the bottom rangefrom-2 to 22 Sv, with a best guessof the mean are compared: the baroclinic transport south of Tas- likely to be about 10 Sv [Cresswell½t al., 1993; Mey- mania is substantially higher in each case. At SR3 the ers ½tal., 1995]. About i Sv is believedto flow out of baroclinic transport variations above 2500 m are not the Pacificthrough the BeringStraits [Coachmanand compensatedby variations of opposite sign at greater Aagaard,1988]. Given the uncertaintyof the estimates depth. The transport above and relative to 2500 m is (e.g.,of the barotropicflow at SR3 and at the Indone- a nearly constantfraction (664-1.4%) of the transport sian throughflowand the uncertaintyin the Drake Pas- relative to the bottom for each of the six SR3 sections. sageabsolute transports), the closeagreement is likely The transport variations result from shifts in the den- fortuitous. sity field that affect the entire water column, near one The use of pressuregauges to monitor variability or both endpoints of the section. in absolute transport assumesthat the variability is

Table 3. Baroclinic Volume Transports Across SR3 Relative to Different Reference Levels

Reference Level Cruise 2000 dbar, Sv 2500 dbar, Sv DCD, Sv BG, Sv

AU9101 87 107 160 149 AU9309 73 95 146 137 AU9407 87 109 165 149 AU9404 76 99 149 135 AU9501 91 115 169 158 AU9601 93 117 181 156

• DCD is the deepestcommon depth at each station pair and BG refers to the best guessreference level described in the text. 2828 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC predominantly barotropic. However, measurementsin basin exchangeof heat. As shownin Table 4, when Drake Passageindicate that the baroclinic variability, the GT82 calculation is updated to account for the while smaller than the barotropic variability, is not neg- throughflowand a more representativesection south ligible. During 1979, the baroclinic transport varied on of Australia, the results agree in sign with three re- longertime-scales, but the range(70-100 Sv) wascom- cent global or hemisphericinverse models. The mag- parableto that of the absolutetransport (105 to 140 nitudes are reasonably consistentwith two of the in- Sv) [Whirworth,1983, Figure 7]. The absolutetrans- verseestimates [Macdonald,1998; Slogtan and Rintoul, port above 2500 m had a standard deviation of 8.5 Sv; 2000b].The modelof Ganachaud[1999] has stronger the baroclinic transport above 2500 m had a standard heat gain in the Pacificand strongerheat lossin the deviationof 5.5 Sv [Whirworth,1983]. Observationsat Indian and Atlantic, as a result of a stronger Indone- SR3 suggestthe baroclinic variability is of comparable sian Throughflowand weakereastward heat transports magnitude to that at Drake Passage,so the assumption south of Africa and Australia. that the transport variability of the ACC is primarily The overall picture is of heat gain in the Pacific, barotropic may not be entirely valid. which is exported to the other basins. This pattern might at first glancebe taken as supportfor the pic- 4.3. Implications for Interbasin Exchange tureof the globalconveyor developed by Gordon[1986] Georgiand Toole[1982] (hereinafterreferred to as and Broecker[1991], where North Atlantic Deep Wa- GT82) usedhydrographic sections spanning the ACC ter (NADW) upwellsand is warmedin the Pacificand to estimate the exchangeof heat and freshwatersouth returns via the Indian Ocean as relatively warm inflow of Africa, New Zealand and America. From the diver- to the Atlantic. However,significant heat input in the gence they inferred the net air-sea exchangein indi- Pacific is required even if the net inflow to the Pacific vidual basins. Their calculation was an early exam- (requiredto balancethe outflowvia Indonesia)is made ple of the power of direct ocean transport estimates up of relatively warm upper layer water. For example, to constrain estimates of air-sea exchange,but a num- the warmestwater with significanteastward transport ber of factors contributed to introduce significant un- acrossSR3 is SAMW, with a temperatureof about 9øC. certainty in the inferred air-sea fluxes. Of most impor- A tieat input of 0.6 PW is requiredto convert10 Sv of tance, GT82 ignoredthe IndonesianThroughflow on the 9øC water to 24øC, the temperature of the Indonesian basisof the Wyrtki[1961] estimate of 1-2 Sv of through- Throughflow(similar to the net heat flux into the Pa- flow. More recent estimatessuggest the throughflowis cificestimated by the analysessummarized in Table4). closer to 10 Sv in the mean and perhaps higher. Be- If the net inflow into the Pacific was made up of much causethe mean temperatureof the throughflow(m 24 coolerdeep water, an evengreater heat input wouldbe øC [Took and Warren, 1993]) is so high relativeto required. that of the ACC, the existenceof a significantthrough- 4.4. Representativeness of the WOCE flow has a major impact on the inferred net air-sea One-Time Survey exchangein each basin. Second, in the absenceof a section south of Australia, GT82 used a section south Most of the WOCE global survey consistedof one- of New Zealand and assumed there was little net flow time occupationsof hydrographicsections. The repeat through the Tasman Sea. Third, in the absenceof di- sectionslike SR3 provide a meansof assessingthe pos- rectmeasurements ofthe barotropic flow, GT82 applied sible alias or bias introduced by the availability of only a uniform barotropic velocity at the African and New singlerealizations of a complex,temporally evolving cir- Zealand sectionsto make the volume transport there culation. equal that at Drake Passage. Finally, only at Drake Figure 6 showsthat the transport of the ACC it- Passagewere repeat sections available to determine a self does not vary much with time; variability in the representative mean field. net transport south of Australia dependsprimarily on In Table 4 we repeat the GT82 calculation including changesin the flow north of the ACC. Of most signif- an estimate of the throughflowcontribution and replac- icancefor the net transport of properties between the ing the New Zealand sectionwith the volume-transport- Indian and Pacific basins is the variability in the west- weighted temperature at SR3. Including the through- ward flow of relatively warm and salty water at the flow gives a very different impressionof the interbasin northern end of SR3. A strong Tasman outflow results flux and net air-sea heat exchange in the Indian and in a low net eastwardtransport of heat and salt, as well Pacific basins. Assuming a throughflowof 10 Sv with as volume, acrossthe section. a transport-weighted temperature of 24øC and using For example,the temperatureflux variesover a range the SR3 volume-transport-weightedtemperature, the of 139øC Sv, or 0.57 PW0c. These variationsin heat Indian basin is found to lose-0.4 PW of heat (rather flux are comparableto meridiohalheat fluxesat midlat- than gain 0.6 PW in GT82) and the Pacificbain gains itudes of the south Indian and Pacific basins. Changes 0.7 PW of heat (rather than losing-0.3PW). in the baroclinicheat transport south of Australia may Inverse models that use a variety of constraints to be compensatedin a number of ways: changesin the determine the distribution of the barotropic velocity heat storage,either localor basin-scale;changes in the (ratherthan assuminga constantvalue as in GT82) are zonalheat flux enteringthe Indian basinsouth of Africa probably a more effective way to determine the inter- or leavingthe Pacificthrough Drake Passage;changes RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC 2829

r.• r.• r.• o o o 2830 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC

in the meridional heat flux in the Indian and Pacific be correlated with meridional shifts of the wind stress subtropicalbasins; changes in the air-seaheat flux; or curl pattern. In density layers the deep flow is similar changesin the heat transport carried by the barotropic on each of the six occupations,but there are significant flow. We are not yet able to determine which combina- changesin the transport of the SAMW density class. tion of these factors is at work, but the magnitudeof The mean (baroclinic)transport across SR3 is about the heat transport variationssuggests the possibilityof 13 Sv larger than the present best estimate of the mean significant climate feedbacks. absolute transport through Drake Passage. The differ- Also of interest are the links betweenchanges in ence between the transport entering the Pacific south transport of individual density layers (Plate 2) and of Australia and that leaving the basin south of Amer- the large-scaleinterbasin circulations. The transport ica is of the right order to balance the outflow from the changesat depth are small (< -+-2Sv). The largest Pacific through Indonesia and the Bering Strait. How- transport changesoccur in the light layersat the north- ever, given the uncertainty in the barotropic transport ernend of thesection (SAMW andlighter layers), whose at both locations, the agreementis likely fortuitous. transport varies between about 9 and 17 Sv. The in- We have used the repeat measurements at SR3 to up- versemodel of $1oyanand Rintoul [2000a]shows that date estimates of interbasin exchangebetween the In- the SAMW participates in a circum-Australia circula- dian and Pacific, following the lead of Georgi and Took tion in the upperlayers of the Indian and Pacificbasins, [1982]. Taking into accountan IndonesianThrough- as follows: more SAMW enters the Pacific across SR3 flow of 10 Sv with a transport-weightedtemperature of than leavesthrough Drake Passage;the excessSAMW 24øC,the Pacificas a wholeis found to gain heat (0.7 feedsthe northwardflow, which (after substantialwa- PW) whilethe Indian andAtlantic basins lose heat (-0.4 ter massmodifications) ultimately suppliesthe Indone- and-0.3 PW, respectively).Including the Indonesian sian throughflow;throughflow water crossesthe Indian Throughflow reversesthe sign of the net heat exchange basinand is carriedsouth in the AgulhasCurrent; cool- in the Indian and Pacific basinsinferred by Georgi and ing and fresheningover the Agulhasextension converts Took [1982]. warm thermoclinewater suppliedby the throughflowto One justification for repeat hydrographyduring WOCE SAMW; and part of the SAMW formed in the Indian was the need to assessthe representativenessof the sector is exported back to the Pacific acrossSR3 to close WOCE "one-time" survey. The observedrange in east- the loop. Variations in the export of SAMW to the Pa- ward heat flux acrossSR3 is large relative to the merid- cificacross SR3 may be compensatedby changesin stor- ional transport of heat in the Southern Hemispheresub- age, or by changesalong this circuitousflow path. On tropical gyres. On the basisof a single repeat sectionit the basisof repeatsof a singlesection we can only iden- is not possibleto determine the impact of these heat and tify this intriguing mode of interannualoceanic variabil- volumetransport anomalieson basin heat budgetsor re- ity. Further studieswith the completeWOCE data set gional climate. The basin-widesynthesis of WOCE ob- will provide additional insight into the impact of the servationswill provide someinsight, although account- variationsfound on SR3. In this regard it is of inter- ing for variability suchas that observedon SR3 presents est that when the inverse model of $loyan and [gin- one of the more formidableand important challengesto foul [2000a]is run with differentrealizations of SR3, such analyses. An expanded array of sustained, broad- the largest changein the solution is an increaseor de- scaleoceanographic observations (e.g., from profiling creasein the strength of the IndonesianThroughflow floats), additionalrepeat sections, and directmeasure- and a compensatingchange in the transport of SAMW ments of ocean currents are also needed to determine at SR3. the climate significanceof the variability observedin the Southern Ocean.

5. Conclusions Acknowledgments. We thank the officers and crew of the RSV Aurora A ustralis for their contributions to this We have described the major circulation features work. Mark Rosenbergensured the high quality of the CTD south of Australia and quantified the baroclinic trans- data. John Church played an important role in securing port variability on the basis of six occupations of the support for this program. Nathan Bindoff acted as Princi- WOCE SR3 sectionnear 140øE(Figure 6). The mean pal Investigator on two of the SR3 transects. Russ Davis top-to-bottombaroclinic transport (relative to a best supplied the ALACE floats and provided the data prior to guessreference level discussed in section3.2) is 147+10 publication. We thank John Toole for his contribution to the Sv. The transport of the individual fronts of the ACC discussionof interbasin exchange. Karen Heywood and an anonymous reviewer provided many useful suggestionsthat is remarkably steady with time. Variations in the net helped improve the manuscript. The support of the Aus- transport of mass and heat across SR3 are linked to tralian National Antarctic ResearchExpeditions (ANARE) variations in the westward flow of relatively warm wa- and EnvironmentAustralia thrbugh the National Green- ter across the northern end of the section. While it house Research Program is gratefully acknowledged. This is true that the ACC is the primary means by which paper is a contribution to the World Ocean Circulation Ex- water is exchanged between basins, the variability in periment. interbasin exchangesouth of Australia is dominated by changesin flow north of the ACC. While the number of References sectionsis small,the strengthof the westwardflow (and Bindoff, N. L., M. J. Warner, and M. A. Rosenberg,On the hencethe net heat flux acrossthe section)appears to circulation of the waters over the Antarctic continental RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC 2831

slope and rise between 800 and 150øE, Deep Sea Res., polar Current through Drake Passageand its relation to Part II, 4{7, 2299-2326, 2000. wind, J. Geophys. Res., 93, 13,993-14,004, 1988. Botnikov, V. N., Geographicalposition of the Antarctic con- Reid, J., and W. D. Nowlin Jr., Transport of water through vergencezone in the Pacific Ocean, Soy. Antarct. Exped. Drake Passage,Deep-Sea Res. Oceanog.Abstr., 18, 51-64, In/. Bull., Engl. Trans., •, 324-327, 1963. 1971. Broecker, W. J., The great ocean conveyor, Oceanography, Rintoul, S. R., On the origin and influence of Adelie Land 4, 79-89, 1991. Bottom Water, in Ocean, Ice and Atmosphere: Interac- Coachman,L., and K. Aagaard, Transportsthrough Bering tions at the Antarctic Continental Margin, Ant. Res. Ser., Strait: Annual and interannual variability, J. Geophys. vol. 75, edited by S.S. Jacobs,and R. Weiss, pp. 151-171, Res., 93, 15,535-15,539, 1988. AGU, Washington, D.C., 1998. Cresswell, G. R., A. Frische, J. Peterson, and D. Quad- Rintoul, S. R., and J. L. Bullister, A late winter hydro- fasel, Circulation in the Timor Sea, J. Geophys.Res., 98, graphic section from Tasmania to Antarctica, Deep Sea 14,379-14,389, 1993. Res., Part I, •6, 1417-1454, 1999. Davis, R. E., Preliminary results from directly measuring Rintoul, S. R., J.-R. Donguy, and D. H. Roemmich, Seasonal middepth circulation in the tropical and South Pacific, J. evolution of upper ocean thermal structure between Tas- Geophys.Res., 103, 24,619-24,639, 1998. mania and Antarctica, Deep Sea Res., Part I, •, 1185- Davis, R. E., D.C. Webb, L. A. Regier, and J. Dufour, The 1202, 1997. autonomousLagrangian circulation explorer (ALACE), Robbins, P. E., and J. M. Toole, The dissolvedsilica budget J. Atmos. Oceanic Technol., 9, 264-285, 1992. as a constraint on the meridional overturning circulation De Ruijter, W. P.M., and D. B. Boudra, The wind-driven of the Indian Ocean, Deep Sea Res., Part I, dd, 879-906, circulation in the South Atlantic-Indian Ocean, I, Numer- 1997. ical experiments in a one-layer model, Deep Sea Res., Part Rosenberg,M., R. Eriksen, and S. Rintoul, Aurora Australis A, 32, 557-574, 1985. marine sciencecruise AU9309/AU9391: Oceanographic Fahrbach, E., G. Rohardt, M. Schroder, and V. Strass, field measurements and analysis, Res. Rep. 2, 103 pp., Transport and structure of the Weddell Gyre, Ann. Geo- Antarct. Coop. Res. Cent., Hobart, Tasmania, Australia, phys., 12, 840-855, 1994. 1995a. Ganachaud, A. S., Large scaleoceanic circulation and fluxes Rosenberg,M., R. Eriksen, S. Bell, N. Bindoff, and S. Rin- of freshwater, heat, nutrients and oxygen, Ph.D. thesis, toul, Aurora Australis marine science cruise AU9407: Mass. Inst. of Tech.- Woods Hole Oceanog.Inst. Joint Oceanographic field measurements and analysis, Res. Program, Cambridge, 1999. Rep. 6, 97 pp., Antarct. Coop. Res. Cent., Hobart, Tas- Georgi, D. T., and J. M. Toole, The Antarctic Circumpolar mania, Australia, 1995b. Current and the oceanicheat and freshwaterbudgets, J. Rosenberg,M., R. Eriksen, S. Bell, and S. Rintoul, Aurora Mar. Res., JO suppl., 183-197, 1982. Australis marine science cruise AU9404: Oceanographic Gordon, A. L., Interocean exchangeof thermocline water, field measurements and analysis, Res. Rep. 8, 53 pp., J. Geophys.Res., 91, 5037-5046, 1986. Antarct. Coop. Res. Cent., Hobart, Tasmania Australia, Gordon, A. L., and E. J. Molinelli, Southern Ocean Atlas, 1996. 200 pp., Columbia Univ. Press, New York, 1982. Rosenberg,M., S. Bray, N. Bindoff, S. Rintoul, N. Johnson, Gouretski, V., and K. Jancke, A new world ocean climatol- S. Bell, and P. Towlet, Aurora Australis marine science ogy: Objective analysis on neutral surfaces, WOCE Rep. cruise AU9501, AU9604, and AU9601: Oceanographic 256/17, World Ocean Circ. Exp. Hydrogr. Prog. Spec. field measurementsand analysis, inter-cruise comparisons Anal. Cent., Hamburg, Germany, 1998. and data quality notes, Res. Rep. 12, 150 pp., Antarct. Heil, P., and I. Allison, The pattern and variability of Coop. Res. Cent., Hobart, Tasmania, Australia, 1997. Antarctic sea-ice drift in the Indian Ocean and west- Schodlock, M.P., M. Tomczak, and N. White, Deep sec- ernPacific sectors, J. Geophys.Res., 10J, 15,789-15,802, tions through the South Australian Basin and acrossthe 1999. Australian-Antarctic Discordance, Geophys. Res. Lett., Heywood, K. J., M.D. Sparrow, J. Brown, and R. R. Dick- 2d, 2785-2788, 1997. son, Frontal structure and Antarctic Bottom Water Flow Sloyan, B. M., and S. R. Rintoul, Circulation, renewal and through the PrincessElizabeth Trough, Antarctica, Deep modification of Antarctic mode and intermediate water, Sea Res., Part I, 46, 1181-1200, 1999. J. Phys. Oceanogr., in press, 2000a. Jackett, D. R., and T. J. McDougall, A neutral density vari- Sloyan, B. M., and S. R. Rintoul, The Southern Ocean able for the World's Oceans, J. Phys. Oceanogr.,27, 237- limb of the global deep overturning circulation, J. Phys. 263, 1997. Oceanogr., in press, 2000b. Jacobs,S.S., On the nature and significanceof the Antarctic Sokolov, S., and S. R. Rintoul, Some remarks on interpo- Slope Front, Mar. Chem., 35, 9-24, 1991. lation of nonstationary oceanographicfields, J. Atmos. Kalnay, E., and et al., The NCEP/NCAR 40-year reanalysis Oceanic Technol., 16, 1434-1449, 1999. project, Bull. Am. Meteorol. Soc., 77, 347-471, 1996. Sokolov, S., and S. R. Rintoul, Circulation and water masses Macdonald,A.M., The global oceancirculation: A hydro- along WOCE section Pl1: Papua New Guinea to Tasma- graphic estimate and regional analysis, Prog. Oceanog., nia, J. Mar. Res., 58, 223-268, 2000. 41,281-382, 1998. Tchernia, P., and P. F. Jeannin, Quelques aspects de la cir- Meyers, G., R. J. Bailey, and A. P. Worby, Geostrophic culation oceanique Antarctique reveles par l'observation transport of Indonesianthroughflow, Deep Sea Res., Part de la derive d'icebergs, Rep. 3, 92 pp., Cent. Natl. d'Etud. I, 4{2, 1163-1174, 1995. Spatiales, Mus. Nat. Hist. Nat., Paris, 1983. Nowlin, W. D. J., and M. Clifford, The kinematic and ther- Toole, J. M., and B. Warren, A hydrographicsection across mohaline zonation of the Antarctic Circumpolar Current the subtropical South Indian Ocean, Deep Sea Res., Part at Drake Passage,J. Mar. Res., d0 suppl., 481-507, 1982. I, do, 1973-2019, 1993. Orsi, A. H., T. W. Whitworth III, and W. D. Nowlin Jr., White, W. B., and N.J. Cherry, Influence of the Antarctic On the meridional extent and fronts of the Antarctic Cir- Circumpolar Wave upon New Zealand temperature and cumpolar Current, Deep Sea Res., Part I, d2, 641-673, precipitation during autumn-winter, J. Clim., 12, 960- 1995. 976, 1998. Peterson,R. G., On the transport of the Antarctic Circum- White, W. B., and R. Peterson, An Antarctic Circumpolar 2832 RINTOUL AND SOKOLOV: TRANSPORT VARIABILITY OF THE ACC

Wave in surfacepressure, wind, temperature and sea ice Wyrtki, K., The flow of water into the deep sea basins of the extent, Nature, oø80,699-702, 1996. western South Pacific Ocean, Aust. J. Mar. Freshwater Whirworth, T., III, Zonation and geostrophicflow of the Res., 12, 1-16, 1961. Antarctic CircumpolarCurrent at Drake Passage,Deep Wyrtki, K., The subsurface water masses in the western Sea Res., Part A, œ7,497-507, 1980. South Pacific Ocean, Aust. J. Mar. Freshwater Res., 13, Whirworth, T., III, Monitoring the net transport of the 18-47, 1962. Antarctic CircumpolarCurrent at Drake Passage,J. Yaremchuk, M. I., N. L. Bindoff, J. SchrSter, D. Nechaev, Phys. Oceano#r.,lo ø, 2045-2057, 1983. and S. R. Rintoul, On the zonal and meridional circula- Whirworth, T., III, and R. G. Peterson,The volumetrans- tion and ocean transports between Tasmania and Antarc- port of the Antarctic Circumpolar Current from three- tica, J. Geophys. Res., this issue. year bottom pressuremeasurements, J. Phys. Oceanogr., 15, 810-816, 1985. Whitworth, T., III, W. D. Nowlin Jr., and S. J. Worley, The net transport of the Antarctic CircumpolarCurrent S. R. Rintoul and S. Sokolov, CSIRO Marine Re- throughDrake Passage,J. Phys. Oceanogr.,12, 960-971, search and Antarctic Cooperative Research Centre, 1982. GPO Box 1538, Hobart, Tasmania 7001, Australia. Whitworth, T., III, A. H. Orsi, S.-J. Kim, W. D. Nowlin ([email protected]) Jr., and R. A. Locarnini,Water massesand mixingnear the Antarctic SlopeFront, in Ocean,Ice and Atmosphere: Interactions at the Antarctic ContinentalMargin, Ant. Res. Set., vol. 75, edited by S.S. Jacobs,and R. Weiss, (ReceivedJuly 15, 1999; revisedApril 19, 2000; pp. 1-27, AGU, Washington, D.C., 1998. acceptedJune 12, 2000.)