DECEMBER 2012 D O W N E S E T A L . 2153

Tracing Southwest Pacific Bottom Water Using Potential Vorticity and Helium-3

STEPHANIE M. DOWNES Program in Atmospheric and Oceanic Sciences, Princeton University, New Jersey, and Research School of Earth Sciences, and ARC Centre of Excellence for Climate System Science, The Australian National University, Acton, Australian Capital Territory, Australia

ROBERT M. KEY Program in Atmospheric and Oceanic Sciences, Princeton University, Princeton, New Jersey

ALEJANDRO H. ORSI Department of Oceanography, Texas A&M University, College Station, Texas

KEVIN G. SPEER Department of Oceanography, The Florida State University, Tallahassee, Florida

JAMES H. SWIFT Scripps Institution of Oceanography, La Jolla, California

(Manuscript received 30 January 2012, in final form 23 July 2012)

ABSTRACT

This study uses potential vorticity and other tracers to identify the pathways of the densest form of Cir- cumpolar Deep Water in the South Pacific, termed ‘‘Southwest Pacific Bottom Water’’ (SPBW), along the 2 28.2 kg m 3 surface. This study focuses on the potential vorticity signals associated with three major dy- namical processes occurring in the vicinity of the Pacific–Antarctic Ridge: 1) the strong flow of the Antarctic Circumpolar Current (ACC), 2) lateral eddy stirring, and 3) heat and stratification changes in bottom waters induced by hydrothermal vents. These processes result in southward and downstream advection of low po- tential vorticity along rising isopycnal surfaces. Using d3He released from the hydrothermal vents, the in- fluence of volcanic activity on the SPBW may be traced across the South Pacific along the path of the ACC to Drake Passage. SPBW also flows within the southern limb of the Ross Gyre, reaching the Antarctic Slope in places and contributes via entrainment to the formation of Antarctic Bottom Water. Finally, it is shown that the magnitude and location of the potential vorticity signals associated with SPBW have endured over at least the last two decades, and that they are unique to the South Pacific sector.

1. Introduction concluded that heat in the buoyant plumes from hy- drothermal vents is effectively diffused unless ‘‘com- The divergence of tectonic plates along midocean peting mechanisms are not overpowering.’’ Stommel’s ridges introduces a gap in the earth’s crust, forming an paper considered the circulation above the East Pacific axial valley, from which geothermal heat is released at Rise, located deep below the South Pacific subtropical a relatively high rate, by both conductive and convective gyre, where little ‘‘overpowering’’ circulation would processes (Wilson 1965; Morgan 1971). Stommel (1982) be expected. Hydrothermal vents are apparently com- mon along midocean ridges, but with a highly inter- mittent distribution; they have recently been identified Corresponding author address: Stephanie Downes, Research School of Earth Sciences, The Australian National University, on parts of the 2600-km-long Pacific–Antarctic Ridge, Canberra ACT 0200, Australia. southwest of the East Pacific Rise (Winckler et al. E-mail: [email protected] 2010). In complete contrast to the East Pacific Rise, the

DOI: 10.1175/JPO-D-12-019.1

Ó 2012 American Meteorological Society Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2154 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

Pacific–Antarctic Ridge lies within the main flow of the slightly converge and flow along the northern flank of Antarctic Circumpolar Current (ACC), presumably an the Pacific–Antarctic Ridge (Gordon et al. 1978; Gille ‘‘overpowering’’ flow. In addition, vigorous eddy activ- 1994). Along the ridge, the ACC flow between its ity and the upwelling of deep waters all coincide with the southern fronts is strong compared to other parts of the multiple hydrothermal sources along this ridge. Southern Ocean where the dominant flow is between the Hydrothermal plumes from the major ocean ridges Subantarctic and Polar Fronts (Rintoul and Sokolov have been traced both near and far afield from the vent 2001). East of the Pacific–Antarctic Ridge, the ACC sources. Veirs et al. (1999) used a stability function along fronts are squeezed through the Udintsev Fracture Zone the Juan de Fuca Ridge to identify which vent sources and the Eltanin Fracture Zone (Gordon et al. 1978; along the ridge are associated with temperature and Patterson and Whitworth III 1990). The ACC intensifies light attenuation signals. They concluded that the sta- as it converges (Gille 1994), is sharply deflected south- bility anomaly was confined to within about 50 km from eastward, and widens as it flows toward Drake Passage the vent source. Hydrothermal signals have also been (Read et al. 1995). The southern edge of the ACC in this detected at large distances from the sources on the ridge sector forms the outer Ross Gyre between 1708E and crests. For example, d3He and manganese have been 1408W (Gordon et al. 1978; Gille 1994; Jacobs et al. 2002; used to trace hydrothermal fluid originating at the East Rickard et al. 2010). Pacific Rise and flowing more than 2000 km west of the Two deep water masses have traditionally been as- ridge (Lupton and Craig 1981; Klinkhammer 1980).1 In sociated with the circulation in the Pacific sector of addition, Johnson and Talley (1997) compared stratifi- the Southern Ocean (Callahan 1972; Patterson and cation measures and temperature-salinity anomalies Whitworth 1990), namely, the oxygen-poor Upper Cir- with d3He to trace various pathways of the hydrothermal cumpolar Deep Water (UCDW) and high-salinity Lower plumes that had originated along the East Pacific Rise. Circumpolar Deep Water (LCDW). These water masses In the South Atlantic Ocean d3He indicates hydrother- are roughly considered to have Indo-Pacific (LCDW) mal plumes also extending thousands of kilometers from and Atlantic (UCDW) origins. Here we are interested in the hydrothermal sources along the southern half of the Lower Circumpolar Deep Water (LCDW), and the Mid-Atlantic Ridge (Ru¨ th et al. 2000). conditions that determine its properties. In terms of Winckler et al. (2010) identified the Pacific–Antarctic density, this layer is defined by the range gn 5 27.98 to 2 Ridge as a major hydrothermal plume source that can be 27.27 kg m 3, where gn is neutral density (Jackett and 2 traced via a d3He plume along the 28.2 kg m 3 neutral McDougall (1997); see also Fig. 3 and Table 1 in Orsi 23 density surface (s2 ’ 1037.12 kg m ) south of the ridge et al. (2002)). Some LCDW recirculates and upwells in along 1508W and across the 678S cruise transect. They the interior of the Weddell Sea, where it also may be concluded that this d3He plume, distinct from the much transformed into Antarctic Surface Water; but part of its stronger d3He source along the East Pacific Rise volume also shoals toward the continental shelf over (Lupton 1998), could be used for tracing the South Pa- a larger zonal extent, where it contributes to the pro- cific abyssal circulation. However, d3He measurements duction of both lighter and denser water masses (Orsi in the South Pacific are sparse, and thus this tracer et al. 1995). LCDW reaching the Pacific–Antarctic cannot solely be used to describe in detail the large- margins participates in the renewal of ventilated deep 2 scale circulation. Here, we expand upon the analysis and bottom waters (gn . 28.15 kg m 3) in the Ross Sea of Winckler et al. (2010) by showing that the d3He sig- (Orsi and Wiederwohl 2009) as well as of lighter (gn , 2 nature they observed coincides with a distinct potential 28.15 kg m 3) pycnocline waters (Wa˚hlin et al. 2010; vorticity signal in the deep South Pacific sector of the Jacobs and Giulivi 2010; Jenkins et al. 2010). Southern Ocean. Throughout the zonal domain of the ACC there is In the South Pacific sector of the Southern Ocean, a well-mixed portion of LCDW, which Orsi et al. (1999) strong interactions occur between the eastward-flowing identified as bottom water within the ACC (ACCbw; ACC and topography (Gnanadesikan and Hallberg density range 28.18 to 28.27), not directly supplied by 2000; Rintoul et al. 2001). The ACC is comprised of new bottom waters sinking in either the northern North fronts or jets (cf. Orsi et al. 1995; Sokolov and Rintoul Atlantic or around Antarctica. Yet, the lateral ventila- 2007). Around 1508E, the ACC strengthens (Che et al. tion of the ACCbw layer has been attributed to multiple 2011), and its path is deflected northward as the jets exports of Modified CDW near the Antarctic Slope Front (ASF) in the Atlantic, Indian, and Pacific sectors of the Southern Ocean (Whitworth et al. 1994; Orsi et al. 1 3 5 3 3 4 2 3 4 d He 100 f[( He/ He)water sample ( He/ He)atmosphere]/ 2001; Orsi and Wiederwohl 2009; Orsi 2010). Here we 3 4 ( He/ He)atmosphereg; Clarke et al. (1969). refer to water in the ACCbw density range in the South

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC DECEMBER 2012 D O W N E S E T A L . 2155

3 TABLE 1. Cruises of analyzed hydrographic (including He) data. Note: the CLIVAR S4P line also includes western (1728E), central (1708W), and eastern (1508W) Ross Sea sections that extend south of 678S to about 728S. Also, the P06 cruise had three legs and the chief scientists for each leg are listed. The cruise chief scientists and the Principal Investigator (PI) for 3He are also included.

Line Latitude Longitude Ship Year Months Chief Scientist(s) 3He PI Pacific WOCE P06 328S 1538E–718W Knorr 1992 May–Jul H.Bryden/M.McCartney/J.Toole W.Jenkins WOCE P14 678–408S 1708E Discoverer 1996 Jan–Feb J.Bullister/G.Johnson — WOCE P15 678S–08 1708W Discoverer 1996 Feb–Mar R.Feely/M.Roberts — WOCE P16 628–218S 150.58W Knorr 1992 Oct–Nov J.Reid W.Jenkins WOCE P17 698–548S 1358W Knorr 1992 Dec–Jan J.Swift P.Schlosser WOCE P17N 348–578N 1458W Thomas Thompson 1993 May–Jun D.Musgrave J. Lupton WOCE P19 698–548S888W Knorr 1992 Dec–Jan J.Swift P.Schlosser WOCE P18 678S–238N 1038W Discoverer 1994 Feb–Apr B.Taft/G.Johnson W.Jenkins CLIVAR P16 718–168S 1508W Revelle 2005 Jan–Feb B.Sloyan/J.Swift P.Schlosser CLIVAR P18 698–278S 1038W Ronald H. Brown 2008 Jan–Feb G.Johnson/A.Orsi [Leg 2] P.Schlosser Southern WOCE S4P 678S 1628E–718W Akademik Ioffe 1992 Feb–Apr M.Koshlyakov/J.Richman P.Schlosser WOCE S03 668–448S 1408E Aurora Australis 1994 Dec–Feb S.Rintoul P.Schlosser CLIVAR S4P 678S 1688E–738W Nathaniel B. Palmer 2011 Feb–Apr J.Swift/A.Orsi P.Schlosser Atlantic WOCE A08 118S488W–138E Meteor 1994 Mar–Jun T.Mu¨ ller/W.Zenk W.Roether WOCE A09 198S378W–98E Meteor 1991 Feb–Mar G.Siedler A. Putzka WOCE A10 308S478W–158E Meteor 1992 Dec–Jan R.Onken W.Roether WOCE A21 638–558S688W Meteor 1990 Jan–Mar W.Roether W.Roether CLIVAR A095 238S428W–138E James Cook 2009 Mar–Apr B.King — CLIVAR A13.5 548S–58N08 Ronald H. Brown 2010 Mar–Apr J.Bullister/R.Key P.Schlosser Indian WOCE I02 88S398–1058E Knorr 1995 Dec–Jan G. Johnson/ B. Warren W.Jenkins WOCE I03 208S498–1138E Knorr 1995 Apr–Jun W.D.Nowlin Jnr. W.Jenkins WOCE I08S 658–308S908E Knorr 1994 Dec–Jan M.McCartney/T.Whitworth III P.Schlosser WOCE I09S 658–348S 1158E Knorr 1994 Dec–Jan M.McCartney/T.Whitworth III P.Schlosser CLIVAR I06S 688–338S308E Revelle 2008 Feb–Mar K.Speer/T.Dittmar W. Jenkins

Pacific Ocean as ‘‘Southwest Pacific Bottom Water’’ (ACC), 2) lateral eddy stirring, and 3) heat and strati- (from herein, SPBW), based on the potential vorticity fication changes in bottom waters induced by hydro- and d3He signals it acquires at a middepth source lo- thermal vents. We trace the circulation of a water mass we cated far from the Antarctic continental slope. The term Southwest Pacific Bottom Water over roughly two name ‘‘ACCbw’’ refers to a water type restricted to the decades, using results from the World Ocean Circulation ocean floor in the ACC region and then exported equa- Experiment (WOCE; 1991–2000) and Climate Variability torward (see temperature and salinity distribution in plates and Predictability (CLIVAR; 2000 onward). 220–221 in Orsi and Whitworth 2005). Traces of SPBW The outline of the remainder of the manuscript is as signals are found downstream, both to the east at the bottom follows. We begin with a description of the cruise data of southern Drake Passage, and to the west at deep levels of and define potential vorticity used here as a tracer. We the southern Ross Gyre. The interior vertical mixing of new then describe the pathways of distinct potential vorticity bottom waters from the Ross Sea needed for the regional signals in the Southwest Pacific bottom water density AABW to continue equatorward over the ridges is there- class from the Pacific–Antarctic Ridge to the Ross Sea fore partly reflected in the SPBW characteristics. and Drake Passage. We assess the influence of other The vertical layering of water masses along the path of major ridges and finally summarize and discuss our results. the ACC and the formation of AABW in the Ross Sea have been extensively documented; however, our un- derstanding of the interaction between these major el- 2. Data and methods ements of the deep stratification remains incomplete. a. WOCE and CLIVAR cruise data We focus our study on the potential vorticity and other tracer signals associated with three major dynamical pro- This study uses hydrographic data from multiple cruises cesses occurring in the vicinity of the Pacific–Antarctic (Table 1; Fig. 1). All of the data and final cruise reports are Ridge: 1) the flow of the Antarctic Circumpolar Current available from the CLIVAR and Carbon Hydrographic

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2156 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

FIG. 1. Map of cruise sections used in our analysis. Red sections denote those collected during the WOCE era (1990 to 1996), and blue dashed sections after 2005 (during the CLIVAR pe- riod). The P16, P18, and S4P lines were sampled during both the WOCE and CLIVAR periods. Magenta dots and numbers indicate GEOSECS cruise stations, and these are referred to later in the paper in Fig. 7.

Data Office (CCHDO; http://cchdo.ucsd.edu/). We have in the abyssal ocean stratification with the use of hy- also analyzed 1973/74 hydrographic data from the drographic data that is high resolution in the vertical Geochemical Ocean Sections Study (GEOSECS), also direction. Potential vorticity (Q) quantifies the ratio of available from CCHDO. None of these data were col- the combined relative vorticity (z) and planetary vor- lected during winter; however, our research focuses on ticity (or Coriolis parameter; f ; negative in the Southern deep and abyssal waters so we do not expect any sea- Hemisphere) to the thickness of the water column, and sonal bias. The main variables of interest are pressure, is given by potential temperature, and salinity, measured using (z 1 f ) ›r conductivity–temperature–depth (CTD) instruments with Q 52 , (1) an accuracy of at least 2 m, 0.0058C, and 0.005, respectively. r ›z CTD results are routinely reported at 1-dbar resolution, but we subsampled that data every 10 dbar. Horizontal where r is the mean density, and the local vertical station spacing for WOCE and CLIVAR is routinely change in density over that in depth is given by ›r/›z. We assume that the relative vorticity is negligible for 55 km. Potential vorticity, a measure of stratification and n a focus for our study, is calculated using neutral density. large-scale flows, and we use neutral density g . Neutral density is based on temperature and salinity and Hence, spatial location of the cast, and is calculated using software f ›gn Q 52 . (2) detailed in Jackett and McDougall (1997). The tracers, gn ›z d3He, silicate and chlorofluorocarbons were sampled using Niskin-type bottles and a Rosette and typically have ac- We assume a positive sign for potential vorticity through- curacies of 1.5%, 2%, and 1%, respectively. Most of these out our text, figures, and equations. A small potential cruises collected 36 bottle samples at each station; however, vorticity corresponds to a small change in density with for these tracers only silicate was measured in all bottles. respect to depth, and hence weak stratification (and a All of the WOCE and CLIVAR cruise data have been well mixed fluid). subjected to multiple levels of quality control and are be- In a layered framework, the potential vorticity equa- lieved to be of very high quality by modern standards. The tion can be written as GEOSECS data were high quality for the time (1970s), but are not quite as accurate or precise as the newer data. by $h w+ 2 w+ h f 1 u 5 T B 1 k=2 . (3) b. Potential vorticity f h h f h Potential vorticity is a measure of stratification in the The first term represents the meridional advection of water column, and we are able to detect subtle changes planetary vorticity. The second term (h is layer thickness)

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC DECEMBER 2012 D O W N E S E T A L . 2157

212 21 21 23 FIG. 2. WOCE P16 potential vorticity (color, 310 m s ) overlaid with 28.2 kg m neutral density surface (yellow) and (a) d3He (black contours; %; asterisks denote sampling locations for d3He), (b) vertical temperature 2 2 gradient (black contours; 8C dbar 1), and (c) vertical salinity gradient (black contours; CTD dbar 1). The gradients are the vertical difference in temperature or salinity with respect to pressure. We draw the reader’s attention to the southward-flowing upwelling Lower Circumpolar Deep Water, represented by the salinity maxima [.34.73; red 2 2 2 2 dashed contours in (c)]. (d) CLIVAR P16 potential vorticity (color, 310 12 m 1 s 1) overlaid with 28.2 kg m 3 2 neutral density surface (yellow) and vertical temperature gradient (black contours; 8C dbar 1). is vortex stretching because of the isopycnal component 3. Results of the velocity (conservative stretching). The third term a. Identification of potential vorticity signals along represents the diapycnal (nonconservative) stretching the Pacific–Antarctic Ridge because of diapycnal fluxes at the top (T) and bottom (B) of the layer, and the fourth term is the isopycnal Winckler et al. (2010) illustrated the existence of diffusion of potential vorticity (where mixing k is as- a unique d3He plume south of the Pacific–Antarctic sumed to be constant). Ridge along the WOCE P16 section, as well as along the Potential vorticity is not conserved near the plume 678S WOCE S4P section. We found that this distinct source. Heat injected from the earth’s mantle and flow d3He plume signal of over 10% at the ridge crest around over the bottom can generate diapycnal fluxes w+ that 598S coincides with a potential vorticity minima signal modify potential vorticity. Frictional torques experi- (Fig. 2a). The minimum in the potential vorticity of less 2 2 2 enced by the abyssal ocean will alter the local density than 5 3 10 12 m 1 s 1 rises above the ridge along the 2 field and break the conservation constraint for potential 28.2 kg m 3 surface (see yellow contour). Directly be- vorticity as well. Large-scale lateral eddy stirring k does low the potential vorticity minima, we find a maximum 2 2 2 not directly modify density since the stirring is essen- in the potential vorticity of 1.1 3 10 11 m 1 s 1 that is tially along isopycnals, but it does smooth out potential evident south of 608S. We note that the potential vor- vorticity gradients on these isopycnals, spreading tracers ticity of the plume is not conserved near the ridge crest, away from sources. These nonconservative influences where it is influenced by geothermal heat, eddies, and may be a factor in setting the potential vorticity signals bottom friction. South of 618S and downstream of the of SPBW once it moves away from the direct influence ridge, the plume retains a potential vorticity close to 9 3 2 2 2 of the plume source. 10 12 m 1 s 1. The density range of SPBW encompasses

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2158 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

3 FIG. 3. WOCE P16 potential temperature (8C) vs d He (%) for stations south of 558S, be- tween 800 and 3500 m. The potential temperatures associated with the potential vorticity (Q) 2 2 2 2 2 2 minima and maxima signals (9 3 10 12 m 1 s 1 and 1.1 3 10 11 m 1 s 1, respectively) are denoted by the red vertical lines.

2 the low potential vorticity signal along the 28.2 kg m 3 Ridge, between the mean temperature 0.48 and 0.68C surface. (Fig. 3). The WOCE P16 section at 1508W lies within 30 km Hydrothermal plumes generated close to, but up- downstream of potential plume source regions identi- stream, of the WOCE P16 section undergo several fied by Winckler et al. (2010) (see their Fig. 3). We infer modifications. Geothermal heat released into the water that the WOCE P16 section is close to a plume source, column at the Pacific–Antarctic Ridge would de- though we cannot determine the exact distance with the stabilize the water column, thus reducing the potential presently available measurements. We calculate the tem- vorticity. However, potential vorticity can also be 2 perature anomaly in density space along the 28.2 kg m 3 influenced by the enhanced flow over a rough bottom surface, defined as the difference between the background and mixing induced in the canyons on the flank of the temperature away from the ridge at 568S (taken to be ridge (Thurnherr et al. 2005). 0.68C) and the temperature at the ridge crest at 59.58S. To determine if additional mixing sources are at play, We find that the Pacific-Antarctic Ridge crest has a we can estimate the scale (H) over which the plume ef- 0.078C temperature anomaly, which compares well with fects might occur (without knowledge of the exact depths the 0.048C anomaly found at the Juan de Fuca Ridge of the sources we are limited to scaling estimates). We (Cannon and Pashinski 1997). use the equation from Speer and Helfrich (1995), H ’ 23 1/4 The localized minimum in the vertical temperature 3.76(FoN ) , where 3.76 is a coefficient based on gradient (i.e., the local change in temperature with depth) laboratory and in situ observations. The local buoyancy along WOCE P16 clearly coincides with the potential frequency squared, N2 52gQ/f is based on the Coriolis vorticity minima (Fig. 2b); however, the relatively shal- parameter (f), the acceleration due to gravity (g 5 2 lower minimum of the vertical salinity gradient is un- 9.81 m s 2), and the potential vorticity (Q) estimated in 2 related to the potential vorticity signals around the Eq. (2), and has units of s 2. The source buoyancy flux 23 22 21 4 23 28.2 kg m surface (black contours in Fig. 2c). We also (Fo) is typically between 10 and 10 m s (Speer observed a strong correlation between the vertical tem- and Helfrich 1995). Taking the local background po- perature gradient and the potential vorticity signal for the tential vorticity of the vent source along the WOCE P16 212 21 21 CLIVAR P16 section (Fig. 2d). The temperature gradi- section to be 5 3 10 m s and the range of Fo ent is weaker in the modern data, possibly because of the values, we predict a vertical scale between 300 and 530 m. small offset in location and large time difference between Given that the plume appears to have vertical scales 2 occupations. Along the WOCE P16 line, there is a clear substantially larger, rising to the 28.2 kg m 3 surface that d3He maxima within the vicinity of the Pacific-Antarctic flows ;1 km above the ridge crest (Fig. 2), it is likely

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC DECEMBER 2012 D O W N E S E T A L . 2159

212 21 21 3 23 FIG. 4. Potential vorticity (color, 310 m s ) overlaid with d He (thin black contours; %) and 28.2 kg m neutral density surface (yellow) for WOCE Pacific diagonal P16 line [between (1358W, 628S) and (1478W, 588S)] , the meridional lines P17, P18, P19, and the zonal line S4P. Asterisks denote sampling locations for d3He. Note that only one d3He sample was available for the P17 cruise, hence we have omitted d3He contours and asterisks from this 2 2 2 2 2 2 section. The 10 3 10 12 m 1 s 1 (11 3 10 12 m 1 s 1 for P19) thick black contour in each panel highlights the SPBW 2 potential vorticity minima (along the 28.2 kg m 3 surface) and maxima. (bottom right) (Drake Passage, i.e. WOCE 2 A21) potential vorticity (color) overlaid with d3He (solid black contours) and the 28.0 kg m 3 (upper dashed con- 2 tour) and 28.2 kg m 3 (lower dashed contour) density surfaces. that other mixing mechanisms do occur at the ridge crest Rona 1989). The plume (along with bottom mixed layer and on its flank. However, distinguishing these mixing contributions) attains eventually a minimum value in 2 2 2 mechanisms is beyond the scope of this study. potential vorticity of about 9 3 10 12 m 1 s 1 and is 2 As it rises, the plume transports warmer waters into advected along the 28.2 kg m 3 surface both poleward the surrounding fluid and entrains cooler ambient wa- into the Ross Gyre and downstream along the ACC ters until it reaches a height where the background southern fronts. We find that the layer affected by the density is similar to its own (Joyce et al. 1986; Speer and plume flows deeper than the salty LCDW (Fig. 2c).

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2160 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

FIG. 5. Schematic of the Pacific Ocean bathymetry (blue hues; m) overlaid with the fronts that encapsulate the ACC (from north to south): the SAF, PF, SACCF, and the SBDY. GEOSECS stations are indicated by magenta dots with stations shown adjacent in white boxes with magenta text. The WOCE P14, P15, P16 (P16A), diagonal P16 line (P16D), P17 (P17E), P18, P19, and S4P lines are denoted in red. The CLIVAR P16, P18 and S4P cruises are shown by blue dashed lines. The CLIVAR S4P cruise is divided into the main cruise section along 678S (S4P), and the western, central, and eastern Ross Sea sections (RW, RC, and RE, respectively). Potential plume source locations are taken from Winckler et al. (2010). The bold part of the cruise lines denotes regions where we observe SPBW.

There is in addition a dynamical effect of the mixing is the forcing mode, and H is the previously calculated and plume source near the bottom. Once the plume height the plume rises before equilibrating with the reaches its maximum vertical spreading height, it is background buoyancy. The first three modes and an transported westward via a diffusive phase speed or average plume rise height of 450 m give a westward 2 2 eastward via advection (Joyce and Speer 1987). The 28.2 phase speed of less than 2 3 10 3 cm s 1. neutral density surface lies within the coldest portion of We estimate the eastward advection along the WOCE LCDW, and hence within the deep meridional over- P16 section using the thermal wind equation: 3 turning cell (Speer et al. 2000), so when the d He tracer ð is injected from the vent, it spreads laterally in the ab- g ›r u 5 dz, (5) sence of ambient currents. We estimate the phase speed r0f ›y of the long baroclinic Rossby waves (Joyce and Speer 1987) using the formulation: where the r is the ocean density, with a mean of r0 5 2 1028 kg m 3. The meridional density gradient is in- bga C 5 u , (4) tegrated over depth (z), giving a mean eastward flow m 2 2 z 2 f m of at least 1 cm s 1, 500 m above the ridge crest. The eastward advection, due to the ACC in this case, is one where the beta plane approximation b 5 ›f/›y ’ to three orders of magnitude greater than the diffusive 2 2 2 10 11 m 1 s 1 is the latitudinal variation in the Coriolis phase speed, implying that the plume is entirely passive 2 2 2 parameter, f ’21.25 3 10 4 s 1. The thermal expan- and swept downstream along the 28.2 kg m 3 surface. 2 2 sion coefficient, a ’ 3 3 10 4 8C 1, and g is acceleration Other studies have quantified a larger zonal velocity due to gravity. The background vertical temperature within the ACC (e.g., Renault et al. 2011; Zhang et al. 24 21 gradient uz is estimated as 5 3 10 8Cm from Fig. 3. 2012), further emphasizing our conclusion that advec- The vertical wavenumber m 5 np/H, where n 5 1, 2, ... tion via the ACC dominates the plume’s pathway.

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC DECEMBER 2012 D O W N E S E T A L . 2161

212 21 21 23 FIG. 6. Potential vorticity (color and black contours, 310 m s ) overlaid with 28.2 kg m neutral density surface (yellow) for Pacific lines P16, P18, and S4P. (top) Cruises during the WOCE period are shown; (bottom) CLIVAR cruise lines are shown. b. Tracing SPBW downstream of the Front (SBDY), as shown by the bold parts of the cruise Pacific–Antarctic Ridge lines in Fig. 5. Along the WOCE S4P section, the SPBW potential vorticity and d3He signals are present We trace SPBW potential vorticity signals along the across the entire section, as noted in the Winckler et al. path of the ACC in the WOCE cruises south and east of (2010) study. This indicates that SPBW flows poleward the Pacific–Antarctic Ridge (Fig. 4). We find that po- of the Pacific–Antarctic Ridge, in the Ross Gyre (dis- tential vorticity signals along the WOCE P16 diagonal cussed further in section 3d) and downstream along the section sampled between (1358W, 628S) and (1478W, ACC. 588S), as well as the southern parts of the meridional East of the S4P section, we have no evidence that the sections (P16, P17, P18, P19; Figs. 2 and 4), coincide with potential vorticity or d3He signals associated with SPBW the d3He tracer (;10%). The stronger d3He signal survive the intense mixing encountered in Drake Pas- (;14%) in the WOCE P18 and P19 sections are asso- sage (Fig. 4; bottom right panel). The stronger d3He 2 ciated with the East Pacific Rise at a lower density signal (;17%) on the 28.0 kg m 3 surface represents 2 (;28.0 kg m 3; Fig. 4). SPBW is carried along the ACC the diluted signal from the East Pacific Rise hydro- between the Polar Front (PF) and Southern Boundary thermal vents that is advected into the ACC from the

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2162 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

FIG. 7. GEOSECS Pacific stations (Legs 7, 8, and 9) south of 408S for pressure greater than 1000 dbar. Shown is d3He (%) in neutral 2 density space (gn;kgm 3). The symbols for each station are shown in the legend, and station locations are shown in Fig. 1. Stations in the vicinity of the Pacific–Antarctic Ridge are: 282 (purple squares), 286 (green circles), and 287 (orange triangles).

north in the Southeast Pacific (Lupton 1998; Well et al. 2003; Naveira Garabato et al. 2007). The observations presented in our results thus far show strong evidence 3 of SPBW potential vorticity and d He signals that are FIG. 8. Potential vorticity (color and black contours, 2 2 2 2 advected along the path of the ACC east of the Pacific- 310 12 m 1 s 1) overlaid with 28.2 kg m 3 neutral density sur- Antarctic Ridge. In the next section we will demon- face (yellow) for WOCE Pacific lines P14 and P15. Unfortunately, 3 strate that the SPBW we trace in this study originates d He was not sampled along these lines, and thus we only show potential vorticity. along the Pacific-Antarctic Ridge, and not upstream of the ridge. Remarkably, across the WOCE and CLIVAR periods c. Signals upstream of the Pacific–Antarctic Ridge the magnitude and location of the potential vorticity minima and maxima in the SPBW density range remain To examine the uniqueness of the SPBW potential 2 unchanged (Fig. 6), and the 28.2 kg m 3 surface remains vorticity and d3He signals found in the Pacific basin, we almost fixed in depth for the P16, P18 and S4P sections. analyze cruise data directly upstream of the WOCE P16 We can also trace the d3He signal pre-WOCE using section (WOCE P14 and P15; Fig. 8) and three Indian Pacific data collected during the GEOSECS (Fig. 7). Ocean lines upstream of the Pacific–Antarctic Ridge, There are two d3He signals present south of 408S. The namely I08S, I09SS, and S03 (Fig. 9 and Table 1). In Fig. 8 2 first is the strong d3He signal of approximately 20% the 28.2 kg m 3 surface does not intersect the Pacific– 2 near 28.0 kg m 3 that originates along East Pacific Antarctic Ridge (;648S), but rather flows at least 500 m Rise (station 322; 438S, 1308W), and dissipates by the above the ridge crest. We observe a maxima in the po- 2 2 2 time it reaches stations 282 and 290. The second d3He tential vorticity (1.1 3 10 11 m 1 s 1) south of the ridge 2 maxima of ;10% occurs near 28.2 kg m 3 for Sta- crest along both the P14 (;64.58S) and P15 (;668S) 2 tions 286 and 287 (678Sto568S, 1708Eto1708W), that sections. The maxima is below the 28.2 kg m 3 surface lie around the Pacific–Antarctic Ridge. This second for the P14 section (similar to the maxima illustrated in 2 d3He signal agrees in magnitude and density with the Fig. 4), however, the maxima and 28.2 kg m 3 surface WOCE P16 (1508W) observations and other Pacific coincide along the P15 line. The potential vorticity min- sections. The GEOSECS, WOCE and CLIVAR d3He ima for the P14 section is difficult to observe in Fig. 8, data presented here provide evidence of hydrother- however, it is clearly highlighted in Fig. 9a; the P15 mal vent activity along the Pacific-Antarctic Ridge near section minima is less well defined as data unavailable at 2 2 the 28.2 kg m 3 surface with d3He of ;10% since the densities greater than ;28.21 kg m 3. In addition, the early 1970s. GEOSECS station 282, located north of the crest near

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC DECEMBER 2012 D O W N E S E T A L . 2163

upstream sections that is primarily associated with in- teraction of the density surfaces and the regional to- pography. We find no clear deflections in d3He for the 2 sections upstream of P16 (Fig. 9b), including at 28.2 kg m 3 where we observe decreases in potential vorticity (Fig. 9a). In contrast, the Pacific sections (P16, P18, P19 and S4P) 2 show a clear elevation in d3He (to ;10%) at 28.2 kg m 3, followedbyasharpdeclineind3He at higher densities. The lower Indian d3He signature (;8.5% near 2 28.05 kg m 3) is derived from the Central Indian Ridge (see Fig. 11) and other tropical sources. d. The Ross Gyre and Ross Sea Recent high-resolution model- and data-based studies have indicated large eddy mixing and topographic roughness on the southern flank of the Pacific-Antarctic Ridge where the ACC and Ross Gyre interact (Thompson 2008; Lu and Speer 2010; Nikurashin and Ferrari 2011). Eddy mixing provides a means for the SPBW potential vorticity signals to be transported poleward from the Pacific–Antarctic Ridge, and across and between the ACC fronts. Recent data from the CLIVAR S4P cruise (Fig. 10) includes both repeated stations of the WOCE S4P zonal line along 678S, and three meridional sec- tions in the Ross Gyre that are essentially poleward continuations of the P14, P15, and P16 lines (see Figs. 1, 23 FIG. 9. (a) Neutral density (kg m ) vs potential vorticity 3 2 2 2 5, and 8). d HedataarenotyetavailablefortheCLI- (310 11 m 1 s 1) for the region restricted by density range 27.9 , n 23 VAR S4P cruise, but we can trace the SPBW potential g , 28.4 kg m and between the Polar Front and Southern 23 Boundary Front (see also Fig. 5). Shown are the WOCE Indian vorticity minima and maxima near the 28.2 kg m lines I08S, I09S, S03, and the WOCE Pacific lines P16, P18, P19, surface along the Eastern and Central Ross Sea sec- and S4P. Note that P16 (the most discussed cruise in this study) is tions (Fig. 10). denoted by a bold black curve. The vertical dotted line denotes the In the Central Ross Sea section, the SPBW potential 23 28.2 kg m surface. Note that I08S, I09S and S03 are located vorticity signal is evident until it reaches the continental upstream of the Pacific–Antarctic Ridge and are indicated by small- 2 8 dashed curves. (b) As in (a), but with neutral density (kg m 3)vs shelf around 74 S, where deep waters are modified and d3He (%). P14 (red) and P15 (dark blue) are included in panel (a) then mix with shelf waters to form AABW (e.g., Orsi in bold large dashed curves, but d3He data was not available on et al. 1999, 2002). The interannual variability in the prop- these two cruises so they are omitted from panel (b). erties of SPBW and Circumpolar Deep Waters and Ross Gyre strength are principal drivers in AABW pro- duction rates (Assmann and Timmermann 2005). From the WOCE P14 section (Fig. 5), indicates a d3He max- the Central and Eastern Ross Sea sections, we conclude 2 ima near 28.2 kg m 3 (Fig. 7). The P14 (and to a lesser that the SPBW originating at the Pacific-Antarctic Ridge extent, P15) minima in potential vorticity near the 28.2 flows poleward via the Ross Gyre. The modification of 2 kg m 3 surface combined with the GEOSECS station SPBW north of the continental shelf, and the formation 282 d3He data imply hydrothermal plume sources of AABW on the shelf dilutes the potential vorticity ;900 km farther upstream than the region proposed in signal (Western Ross Sea section; Fig. 10). Winckler et al. (2010) (see yellow bars in Fig. 5 for their potential plume sources). 2 4. Discussion and conclusions Near 28.2 kg m 3, we also observe a strong deflection in the P16 potential vorticity curve, and less so for P18, We use a combined PV and d3He tracer analysis to P19 and S4P at a slightly lighter density (Fig. 9a). The identify the mechanisms modifying the properties and upstream I08S, I09S, and S03 curves show no distinct circulation of Southwest Pacific Bottom Water (SPBW) deflection near this density. However, there is a poten- in the South Pacific and show the SPBW potential vor- 2 tial vorticity minima evident at ;28.05 kg m 3 for the ticity signature and large-scale pathways change minimally

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2164 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

212 21 21 23 FIG. 10. Potential vorticity (color, 310 m s ) overlaid with the 28.2 kg m neutral density surface (yellow contour) for the CLIVAR line S4P. (top left) The zonal 678S part of the line (referred to as ‘CLIVAR S4P’ in the text) with the longitude of the zonal Ross Sea sections denoted by red vertical lines. Also see Fig. 5 for section locations. The remaining panels show potential vorticity and neutral density surfaces along the Western, Central and Eastern 2 2 2 Ross Sea sections (RW, RC and RE, respectively). The 10 3 10 12 m 1 s 1 black contour in each panel highlights the SPBW potential vorticity minima and maxima. over the past two decades. This provides a physical in- Use of high resolution potential vorticity and helium data terpretation of the pathway of hydrothermal plumes have provided a means of separating these deep water originating from the Pacific–Antarctic Ridge, as de- masses in the Pacific Ocean. scribed by Winckler et al. (2010). We also investigated the potential vorticity and d3He Hydrothermal activity can be identified using several along several other cruise lines around the global ocean tracers, such as manganese, silicate, and germanium (e.g., and found no ridges where the two tracers were co- Klinkhammer 1980; Mortlock et al. 1993). We analyzed incident (Fig. 11). The WOCE I03 line intersects the silicate data available on South Pacific WOCE cruises; Central Indian Ridge at ;658E. Helium from the ridge however, high silica also originates from other regions flows northeastward and in a cyclonic path around the surrounding the Pacific–Antarctic Ridge, in the North Ninety-East Ridge (e.g., Srinivasan et al. 2009; Pacific, and Ross Gyre. Hence, the silicate found near Drijfhout and Naveira Garabato 2008). We also ob- the ridge vent sites is thus not necessarily indicative of serve the 15% d3He signal at 88SontheI02cruiseline plumes. Cholorfluorocarbons (CFCs) and oxygen have (figure not shown). Along the Juan de Fuca Ridge (408– been repeatedly used to trace deep and bottom waters in 458N; WOCE P17N), a d3He signal of 24% can be ob- the Southern Ocean (Orsi et al. 1999, 2002). However, served; however, the spread of the plumes originating along LCDW and SPBW are both old, well mixed water masses this ridge are focused in their source region to within a de- with a low CFC signal, and are thus difficult to separate gree or so of longitude (Veirs et al. 1999; Cannon and Pa- using CFCs. In the recent CLIVAR S4P Ross Sea sec- shinski 1997). In the South Pacific (WOCE P06; Fig. 11), tions, we identified distinct nutrient signals (oxygen, ni- hydrothermal plumes originating along the East Pacific trate, and phosphate) that were evident but irrelevant to Rise flow westward and counterclockwise over the ridge, the SPBW potential vorticity signals (figure not shown). then southeastward toward Drake Passage (Lupton 1998;

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC DECEMBER 2012 D O W N E S E T A L . 2165

FIG. 11. Four cruise lines that intersect prominent ridges with hydrothermal activity. Potential vorticity (color, 2 2 2 310 12 m 1 s 1) overlaid with d3He (black contours; %) for WOCE lines I03 (208S), P17N (1458W), P06 (328S), and A10 (308S). Black dots denote sampling locations for d3He.

Well et al. 2003; Naveira Garabato et al. 2007; Bianchi et al. local mixing (Thurnherr et al. 2005). These examples 2010). illustrate the dependence of these two tracer signals We observe the high d3He signal (.24%) on both on background fields as well as local processes. sides of the East Pacific Rise at 328S(WOCEP06; After analyzing several cruise lines near other major Fig. 11). Johnson and Talley (1997) suggested that ocean ridges, we infer that the SPBW potential vorticity stratification and d3He signals associated with hy- coinciding with d3He signals are likely unique to the drothermal plumes downstream of the East Pacific Pacific–Antarctic Ridge. Several physical processes co- Rise are likely to coincide in the absence of vertical incide along the Pacific–Antarctic Ridge to form the mixing. Along the Mid-Atlantic Ridge a distinct, but potential vorticity and d3He signals and then transport weaker, d3He signal of ;6% has been observed along, them across the South Pacific (see Fig. 12): and equatorward of, the WOCE A10 section (Fig. 11; Well et al. 2001; Ru¨ th et al. 2000). Primordial helium d Hydrothermal vents along most of the ridge inject (defined by anomalies relative to the background) significant heat into the water column, destabilizing plumes emanating from the Mid-Atlantic Ridge at 308 the stratification. and 118S spread across the western Atlantic basin d The efficiency of eddy mixing is large and persistent (Ru¨ th et al. 2000). In contrast, potential vorticity along the entire Pacific Antarctic Ridge for the in- signals created above the ridge have been related to termediate and deep ocean (Lu and Speer 2010).

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2166 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

212 21 21 3 FIG. 12. WOCE P16 potential vorticity (color, 310 m s ) overlaid with d He (black contours; %) as shown in Fig. 2a. Asterisks denote sampling locations for d3He. White dashed arrows denote UCDW and LCDW that flow southward from the North Pacific and upwell in the ACC region, bound by the SAF to the north, and the SBDY to the south. The SACCf and PF lie within the ACC. The low potential vorticity signature of modified bottom waters (orange dashed arrows) emphasizes that which is induced by the hydrothermal plume source at the Pacific–Antarctic Ridge. Lateral eddy mixing along the ridge (white 2 spiral) and along the 28.2 kg m 3 surface, the ACC eastward flow, and the Ross Gyre circulation transport the low potential vorticity signature/high d3He signal (i.e., SPBW) downstream.

d The ACC overlies the vents along the Pacific–Antarctic been performed extensively over ridges such as the Mid- Ridge. This sweeps them downstream, between the Atlantic Ridge, Juan de Fuca Ridge, and East Pacific Polar Front and Southern Boundary. Strong westerlies Rise (e.g., Lupton 1998; Ru¨ th et al. 2000, and references driving the ACC induce a steep gradient in the density therein), but would be valuable in this region, since de- surfaces across the Pacific–Antarctic Ridge. Bottom composing the diapycnal and along isopycnal eddy mix- waters originating in the Ross Sea flow equatorward to ing, deep water upwelling, hydrothermal activity, and the ridge, carrying a low potential vorticity signal, and ACC flow requires better information about the abyssal mix with lighter LCDW. The multiple hydrothermal interactions with the midocean ridge. sources along the Pacific–Antarctic Ridge (Winckler et al. 2010) provide abundant opportunity for deep Acknowledgments. The authors wish to thank waters to interact with the hydrothermal plume within 2 M. Nikurashin, J. Sarmiento, and two anonymous re- the ACC along the 28.2 kg m 3 surface. Thus, different viewers for useful comments and discussion. We greatly layers within the ACC can have vastly different origins. appreciate the efforts of those who collected and pro- cessed samples on the cruises mentioned in this paper. Several questions remain unanswered. From the re- SD was supported by the Office of Science (BER), U.S. sults presented here and in Winckler et al. (2010), it is Department of Energy, Grant DE-FG02-07ER64467 evident that more detailed surveying of the Pacific– and NOAA Grant NA07OAR4310096. RK was sup- Antarctic Ridge and surroundings are required, to de- ported by National Ocean and Atmospheric Adminis- termine exact vent locations, source strengths and local tration, Grant 08OAR4320752. KS acknowledges sup- spreading rates. Such high-resolution surveys have port from NSF OCE-0622670 and NSF OCE-0822075.

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC DECEMBER 2012 D O W N E S E T A L . 2167

REFERENCES Mortlock, R. A., P. N. Froelich, R. A. Feely, G. J. Massoth, D. A. Butterfield, and J. E. Lupton, 1993: Silica and germanium in Assmann, K. A., and R. Timmermann, 2005: Variability of dense Pacific Ocean hydrothermal vents and plumes. Earth Planet. water formation in the Ross Sea. Ocean Dyn., 55, 68–87. Sci. Lett., 119, 365–378. Bianchi,D.,J.L.Sarmiento,A.Gnanadesikan,R.M.Key, Naveira Garabato, A. C., D. P. Stevens, A. J. Watson, and P. Schlosser, and R. Newton, 2010: Low helium flux from the W. Roether, 2007: Short-circuiting of the overturning circulation mantle inferred from simulations of oceanic helium isotope in the Antarctic Circumpolar Current. Nature, 447, 194–197. data. Earth Planet. Sci. Lett., 297, 379–386. Nikurashin, M., and R. Ferrari, 2011: Global energy conversion rate Callahan, J. E., 1972: The structure and circulation of deep water in from geostrophic flows into internal lee waves in the deep ocean. the Antarctic. Deep-Sea Res., 19, 563–575. Geophys. Res. Lett., 38, L08610, doi:10.1029/2011GL046576. Cannon, G. A., and D. J. Pashinski, 1997: Variations in mean Orsi, A. H., 2010: Recycling bottom waters. Nat. Geosci., 3, 307– currents affecting hydrothermal plumes on the Juan de Fuca 309. Ridge. J. Geophys. Res., 102 (C11), 24 965–24 976. ——, and T. Whitworth III, 2005: Southern Ocean. Vol. 1, Hy- Che, S., Z. Linlin, and Y. Xiaomei, 2011: Stream-coordinate drographic Atlas of the World Ocean Circulation Experiment structure of oceanic jets based on merged altimeter data. Chin. (WOCE), International WOCE Project Office, 223 pp. J. Oceanol. Limnol., 29, 1–9. ——, and C. L. Wiederwohl, 2009: A recount of Ross Sea waters. 3 Clarke, W. B., M. A. Beg, and H. Craig, 1969: Excess he in the sea: Deep-Sea Res. II, 56, 778–795. Evidence for terrestrial primordial helium. Earth Planet. Sci. ——, T. Whitworth III, and W. D. Nowlin, 1995: On the meridional Lett., 6, 213–220. extent and fronts of the Antarctic Circumpolar Current. Deep- Drijfhout, S. S., and A. C. Naveira Garabato, 2008: The zonal di- Sea Res. I, 42, 641–673. mension of the Indian Ocean meridional overturning circu- ——, G. Johnson, and J. Bullister, 1999: Circulation, mixing, and lation. J. Phys. Oceanogr., 38, 359–379. production of Antarctic Bottom Water. Prog. Oceanogr., 43, Gille, S. T., 1994: Mean sea surface height of the Antarctic Cir- 55–109. cumpolar Current from Geosat data: Method and application. ——, S. S. Jacobs, A. L. Gordon, and M. Visbeck, 2001: Cooling J. Geophys. Res., 99, 18 255–18 273. and ventilating the abyssal ocean. Geophys. Res. Lett., 28, Gnanadesikan, A., and R. W. Hallberg, 2000: On the relationship 2923–2926. of the circumpolar current to Southern Hemisphere winds in ——, W. M. Smethie Jr., and J. L. Bullister, 2002: On the total input coarse-resolution ocean models. J. Phys. Oceanogr., 30, 2013– of Antarctic waters to the deep ocean: A preliminary estimate 2034. from chlorofluorocarbon measurements. J. Geophys. Res., Gordon, A. L., E. Molinelli, and T. Baker, 1978: Large-scale rel- 107, C83122, doi:10.1029/2001JC000976. ative dynamic topography of the Southern Ocean. J. Geophys. Patterson, S. L., and T. Whitworth III, 1990: Antarctic sector of the Res., 83, 3023–3032. Pacific. Physical Oceanography, G. P. Glasby, Ed., Elsevier Jackett, D. R., and T. J. McDougall, 1997: A neutral density vari- Oceanography Series, Vol. 51, Elsevier, 55–93. able for the world’s oceans. J. Phys. Oceanogr., 27, 237–263. Read, J. F., R. T. Pollard, and A. I. Morrison, 1995: On the Jacobs, S. S., and C. F. Giulivi, 2010: Large multidecadal salinity southerly extent of the Antarctic Circumpolar Current in the trends near the Pacific–Antarctic continental margin. J. Cli- southeast Pacific. Deep-Sea Res. II, 42, 933–954. mate, 23, 4508–4524. Renault,A.,C.Provost,N.Senne´chael, N. Barre´, and A. Kartavtseff, ——, ——, and P. A. Mele, 2002: Freshening of the Ross Sea during 2011: Two full-depth velocity sections in the Drake Passage in the late 20th century. Science, 297, 386–389. 2006–transport estimates. Deep-Sea Res. II, 58, 2572–2591. Jenkins, A., P. Dutrieux, S. S. Jacobs, S. D. McPhail, J. R. Perrett, Rickard, G. J., M. J. Roberts, M. J. M. Williams, A. Dunn, and A. T. Webb, and D. White, 2010: Observations beneath Pine M. H. Smith, 2010: Mean circulation and hydrography in the Island Glacier in West Antarctica and implications for its re- Ross Sea sector, Southern Ocean: Representation in numeri- treat. Nat. Geosci., 3, 468–472. cal models. Antarct. Sci., 22, 533–558. Johnson, G. C., and L. D. Talley, 1997: Deep tracer and dynamical Rintoul, S. R., and S. Sokolov, 2001: Baroclinic transport variability plumes in the tropical Pacific Ocean. J. Geophys. Res., 102, of the Antarctic Circumpolar Current south of Australia. 24 953–24 964. J. Geophys. Res., 106 (C2), 2815–2832. Joyce, T. M., and K. G. Speer, 1987: Modeling the large-scale in- ——, C. W. Hughes, and D. Olbers, 2001: Ocean circulation and fluence of geothermal sources on abyssal flow. J. Geophys. climate. The Antarctic Circumpolar System, Academic Press, Res., 92, 2843–2850. 292–294. ——, B. A. Warren, and L. D. Talley, 1986: The geothermal heating Ru¨ th, C., R. Well, and W. Roether, 2000: Primordial 3He in South of the abyssal subarctic Pacific Ocean. Deep-Sea Res., 33, Atlantic deep waters from sources on the mid-Atlantic ridge. 1003–1015. Deep-Sea Res. I, 47, 1059–1075. Klinkhammer, G. P., 1980: Observations of the distribution of Sokolov, S., and S. R. Rintoul, 2007: Multiple jets of the Antarctic manganese over the East Pacific rise. Chem. Geol., 29, 211– Circumpolar Current south of Australia. J. Phys. Oceanogr., 226. 37, 1394–1412. Lu, J., and K. Speer, 2010: Topography, jets, and eddy mixing in the Speer, K. G., and P. A. Rona, 1989: A model of an Atlantic and Southern Ocean. J. Mar. Res., 68, 479–502. Pacific hydrothermal plume. J. Phys. Oceanogr., 94, 6213– Lupton, J., 1998: Hydrothermal helium plumes in the Pacific 6220. Ocean. J. Geophys. Res., 103, 15 853–15 868. ——, and K. R. Helfrich, 1995: Hydrothermal plumes: A review of ——, and H. Craig, 1981: A major Helium-3 source at 158S on the flow and fluxes. Physical Oceanography, Geological Society, East Pacific Rise. Science, 214, 13–18. 373–385. Morgan, W. J., 1971: Convection plumes in the lower mantle. Na- ——, S. R. Rintoul, and B. Sloyan, 2000: The diabatic Deacon Cell. ture, 230, 42–43. J. Phys. Oceanogr., 30, 3212–3222.

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC 2168 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 42

Srinivasan, A., Z. Garraffo, and M. Iskandarani, 2009: Abyssal Well, R., J. Lupton, and W. Roether, 2001: Crustal helium in deep circulation in the Indian Ocean from a 1/128 resolution global Pacific waters. J. Geophys. Res., 106, 14 165–14 177. hindcast. Deep-Sea Res. I, 56, 1907–1926. ——, W. Roether, and D. P. Stevens, 2003: An additional deep- Stommel, H., 1982: Is the South Pacific helium-3 plume dynami- water mass in Drake Passage as revealed by 3He data. Deep- cally active? Earth Planet. Sci. Lett., 61, 63–67. Sea Res. I, 50, 1079–1098. Thompson, A. F., 2008: The atmospheric ocean: eddies and jets in Whitworth, T., III, W. D. Nowlin Jr., A. H. Orsi, R. A. Locanini, the Antarctic Circumpolar Current. Philos. Trans. Roy. Soc. and S. G. Smith, 1994: Weddell sea shelf water in the Brans- London, A366, 4529–4541. field Strait and Weddell Scotia Confluence. Deep-Sea Res. I, Thurnherr, A. M., L. C. St. Laurent, K. G. Speer, J. M. Toole, and 41, 629–641. J. R. Ledwell, 2005: Mixing associated with sills in a canyon Wilson, J. T., 1965: Evidence from ocean islands suggesting not the midocean ridge flank. J. Phys. Oceanogr., 35, 1370– movement in the Earth. Philos. Trans. Roy. Soc. London, 1381. A258, 145–167. Veirs, S. R., R. E. McDuff, M. D. Lilley, and J. R. Delaney, 1999: Winckler, G., R. Newton, P. Schlosser, and T. J. Crone, 2010: Locating hydrothermal vents by detecting buoyant, advected Mantle helium reveals Southern Ocean hydrothermal venting. plumes. J. Geophys. Res., 104, 29 239–29 247. Geophys. Res. Lett., 31, L05601, doi:10.1029/2009GL042093. Wa˚hlin, A. K., X. Yuan, G. Bjo¨ rk, and C. Nohr, 2010: Inflow of Zhang, L. L., C. Sun, and D. X. Hu, 2012: Interannual variability of warm Circumpolar Deep Water in the central Amundsen the Antarctic Circumpolar Current strength based on merged Shelf. J. Phys. Oceanogr., 40, 1427–1434. altimeter data. Chin. Sci. Bull., 57, 2015–2021.

Unauthenticated | Downloaded 09/30/21 01:40 PM UTC