Downloaded 09/25/21 10:39 PM UTC 1028 JOURNAL of PHYSICAL OCEANOGRAPHY VOLUME 33

MAY 2003 SLOYAN ET AL. 1027

The Paci®c Cold Tongue: A Pathway for Interhemispheric Exchange*

BERNADETTE M. SLOYAN,ϩ GREGORY C. JOHNSON, AND WILLIAM S. KESSLER NOAA/Paci®c Marine Environmental Laboratory, Seattle, Washington

(Manuscript received 29 January 2002, in ®nal form 28 October 2002)

ABSTRACT Mean meridional upper-ocean temperature, salinity, and zonal velocity sections across the Paci®c Ocean between 8ЊS and 8ЊN are combined with other oceanographic and air±sea ¯ux data in an inverse model. The tropical Paci®c Ocean can be divided into three regions with distinct circulation patterns: western (143ЊE± 170ЊW), central (170Њ±125ЊW), and eastern (125ЊW±eastern boundary). In the central and eastern Paci®c the downward limbs of the shallow tropical cells are 15(Ϯ13) ϫ 106 m3 s Ϫ1 in the north and 20(Ϯ11) ϫ 106 m3 sϪ1 in the south. The Paci®c cold tongue in the eastern region results from diapycnal upwelling through all layers of the Equatorial Undercurrent, which preferentially exhausts the lightest (warmer) layers of the Equatorial Undercurrent [10(Ϯ6) ϫ 106 m3 sϪ1] between 125Њ and 95ЊW, allowing the denser (cooler) layers to upwell [9(Ϯ4) ϫ 106 m3 sϪ1] east of 95ЊW and adjacent to the American coast. An interhemispheric exchange of 13(Ϯ13) ϫ 106 m3 s Ϫ1 between the southern and northern Paci®c Ocean forms the Paci®c branch of the Paci®c± Indian interbasin exchange. Southern Hemisphere water enters the tropical Paci®c Ocean via the direct route at the western boundary and via an interior (basin) pathway. However, this water moves irreversibly into the North Paci®c by upwelling in the eastern equatorial Paci®c and air±sea transformation that drives poleward interior transport across 2ЊN.

1. Introduction and Brady (1985) constructed large regional box models and found that upwelling from the EUC leads to the A striking feature of the Paci®c Ocean is the equa- equatorial SST minimum. From the heat budget Wyrtki torial sea surface temperature (SST) minimum that ex- (1981) deduced that the maximum upwelling occurred tends from the coast of the Americas into the central in the upper EUC, while Bryden and Brady (1985) sug- Paci®cÐthe cold tongue (Wyrtki 1981). Knauss (1966) suggested that the salinity, temperature, and nutrient dis- gested that the water upwelling to form the cold tongue tribution in the central and eastern equatorial Paci®c comes equally from meridional and zonal convergence. Ocean was the result of equatorward convergence into One model of the EUC termination in the east has an the Equatorial Undercurrent (EUC) and subsequent up- inertial jet with diapycnal mixing gradually losing vol- welling. Upwelling in the eastern Paci®c Ocean provides ume through upwelling from successively denser layers a route by which subtropical water can in¯uence the (Pedlosky 1988), consistent with the observed eastward tropical SST, either through property anomalies (Gu and cooling of the cold tongue. Observational analysis also Philander 1997) or transport anomalies (Kleeman et al. suggests that the EUC supplies some of the coastal up- 1999). Upwelling also creates the zonal SST gradient welling off South America (Lukas 1986). that is coupled to the Walker circulation. However, the In the present study, inverse methods are used to com- processes that control the intensity and extent of the bine a new mean temperature, salinity, and direct ve- mean cold tongue are not well known. locity climatology (Johnson et al. 2002) with air±sea Many observational and modeling studies have an- ¯ux climatologies and historical conductivity±temper- alyzed the fate of the EUC. Wyrtki (1981) and Bryden ature±depth (CTD) and expendable bathythermography (XBT) data in the eastern Paci®c to estimate a tropical Paci®c circulation that is consistent with the observed * Paci®c Marine Environmental Laboratory Contribution Number data. While the particular focus of the present study is 2440. to determine processes that in¯uence the strength of the ϩ Current af®liation: Department of Physical Oceanography, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts. eastern Paci®c cold tongue, the extension of the model domain across the Paci®c Ocean also enables investi- gation of circulation pathways that transport water from Corresponding author address: Dr. Bernadette Sloyan, Dept. of the South Paci®c to the North Paci®c, feeding the Pa- Physical Oceanography, MS 21, Woods Hole Oceanographic Insti- tution, Woods Hole, MA 02543. ci®c±Indian interbasin exchange. The new climatology, E-mail: [email protected] constructed from 172 contemporaneous meridional

᭧ 2003 American Meteorological Society

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1028 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33

CTD/acoustic Doppler current pro®ler (ADCP) sections TABLE 1. Isopycnal surfaces that de®ne the 11 layers of the inverse taken across the tropical Paci®c, mostly during the model. The 11 layers are grouped into four general water classes, whose average thickness is given. 1990s, greatly improves the quality, quantity and spatial coverage of data available compared to that used in Model ␴␪ Model Thickness earlier volume constraint studies (Wyrtki 1981; Bryden surface (kg mϪ3) layer Water class (m) and Brady 1985). A 25-box inverse model was con- surface structed for the tropical Paci®c that explicitly includes 1 21.3 1 Surface and mixed 30 the diapycnal property transport, meridional eddy heat 21.3 layer 2 21.7 2 ¯ux induced by tropical instability waves (TIW) and 21.7 air±sea transformation driven by horizontal wind-driven 3 22.5 3 mixing and buoyancy ¯uxes. 22.5 This paper is structured as follows. A brief description 4 23.0 4 of the model and data is given in section 2. Initial anal- 23.0 Upper thermocline 120 yses of the individual model boxes showed that these 5 24.0 5 water 24.0 could be grouped into three broad regions. A detailed 6 25.5 6 description of the circulation of each region is given in 25.5 Lower thermocline 70 section 3. Section 4 draws together the circulation of 7 26.1 7 water the three regions and discusses their impact on the Pa- 26.1 ci®c cold tongue. Conclusions are given in section 5. 8 26.3 8 26.3 Thermostad water 238 9 26.5 9 2. Inverse model and data 26.5 The geographical domain of the study extends from 10 26.7 10 26.7 8ЊSto8ЊN and from 143ЊE to the American West Coast. 11 26.9 11 The primary data used are 10 meridional upper-ocean sections of temperature, salinity, and zonal velocity lo- cated at 143ЊE, 156ЊE, 165ЊE, 180Њ, 170ЊW, 155ЊW, 140 W, 125 W, 110 W, and 95 W, thus spanning most Њ Њ Њ Њ with geostrophic u below 250 m. This level is conve- of the tropical Paci®c Ocean. The data and ®tting pro- niently below the core of the EUC. Geostrophic u is cedures used to construct the mean ®eld studied here calculated from the mean temperature and salinity sec- are discussed elsewhere (Johnson et al. 2002). Zonal tions (Johnson et al. 2002) (applying equatorial ␤-plane sections constructed at Ϯ2Њ and Ϯ8Њ latitude further dynamics within 1 of the equator). The vertical shear subdivide the study domain, with additional CTD and Ϯ Њ of u from the ADCP is small between 275 and 375 m. XBT data used to determine mean water property dis- The reference velocities (at each latitude along each tributions and meridional velocity ®elds at Ϯ8Њ between 95ЊW and the American coast. section) are estimated as the mean differences of the This combination of meridional and zonal sections ADCP u and geostrophic u over this depth interval. This divides the tropical Paci®c Ocean into 25 boxes (Fig. procedure reproduces both the strong eastward u as- 3). The inverse model of Sloyan and Rintoul (2000) is sociated with the EUC and subsurface countercurrents developed further and applied to these data to provide [SCCs: northern (NSCC) and southern (SSCC)] and the an estimate of the circulation of the upper tropical Pa- westward u of the South Equatorial Current [SEC: ci®c Ocean that is consistent with the temperature, sa- northern branch SEC(N) between 4ЊNto0Њ and southern linity, velocity, and surface forcing data. The model is branch SEC(S) between 0Њ to 8ЊS] and deeper Equatorial forced with surface buoyancy ¯uxes and wind stress Intermediate Current (EIC; Fig. 1). from the Comprehensive Ocean±Atmosphere Data Set The ADCP provides estimates of both u and the me- (COADS) climatology (da Silva et al. 1994). Eleven ridional (␷) velocity components. However, in the equa- potential density surfaces divide the upper 500 m into torial region the variability of ␷ is large so the ADCP 11 layers (Table 1). Mass (M), temperature (T), and salt data provide only a noisy estimate of the small mean ␷ (S) are conserved in all layers. The model system of (Johnson et al. 2001). Therefore, we estimate ␷ between simultaneous equations is solved by the Gauss±Markov the meridional sections assuming geostrophy (refer- technique (Wunsch 1996). The zonal sections at 8ЊS and enced to 895 m) and Ekman dynamics. Zonal sections 8ЊN and 2ЊS and 2ЊN divide the northern and southern at Ϯ8Њ and Ϯ2Њ latitude are formed from the mean tem- tropical sectors from the equatorial sectors. The many perature and salinity pro®les (Johnson et al. 2002). meridional sections allow exploration of the zonal evo- Mean zonal sections between 95ЊW and the coast of the lution of the circulation. Americas at 8ЊS and 8ЊN are produced from historical While the zonal (u) velocities used in the model are CTD and XBT data using similar mapping techniques directly measured by ADCP to 350±400 m, they are to those applied by Johnson et al. (2002). These sections extended to 500 m by melding the ADCP u estimates extend the model domain from 95ЊE to the coast, and

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1029

FIG. 1. Salinity (pss-78 shaded) and initial zonal velocity, u (black contours, 10Ϫ2 msϪ1 solid line eastward, dashed line westward) distribution at (a) 180Њ, (b) 155ЊW, and (c) 110ЊW. Also shown are the four potential density (kg mϪ3) surfaces (white contours) that divide the water column into the four general water classes used in this study. resolve eastern boundary currents, such as those off cells (Lu et al. 1998; Johnson 2001). Previous tropical Peru. Paci®c volume budget studies (e.g., Wyrtki 1981; Bryden The construction of zonal sections at Ϯ8Њ and Ϯ2Њ and Brady 1985; Meinen et al. 2001) placed their zonal latitude distinguish the EUC from the SEC and SCCs boundaries at Ϯ5Њ latitude. This latitude was chosen be- and separates the equatorial upwelling and near-equato- cause midlatitude geostrophy and Ekman dynamics be- rial downwelling of the shallow tropical recirculation come unreliable for ␷ closer to the equator. Our study

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1030 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33

uses the techniques of Lagerloef et al. (1999) to estimate The best ®t of ␪s to drifter data was 2.2Њ. This length geostrophic and Ekman ␷ at Ϯ2Њ of the equator. scale is comparable to the ®rst baroclinic mode Rossby The transition between midlatitude and equatorial dy- radius near the equator, 200±250 km (Gill 1982; Chelton namics is problematic. Lagerloef et al. (1999) estimated et al. 1998). Scaling the transition from midlatitude to weighting functions such that the geostrophic velocity equatorial dynamics by the equatorial Rossby radius (Ug) is the weighted sum of equatorial and midlatitude also has some theoretical support. Moore and Philander formulations: (1977) applied a steady wind to an unbounded equatorial

Ugffϭ WU ϩ WU␤␤, (1) ocean that was initially at rest. Their solution showed where U ϭ u ϩ i␷ and the subscripts g, f, and ␤ rep- a transition from zonal acceleration on the equator to a resent the (merged) geostrophic velocity, midlatitude ( f steady Ekman response off the equator that scaled by plane) geostrophy, and equatorial (␤ plane) geostrophy, the equatorial radius of deformation. respectively. Lagerloef et al. (1999) prescribed the Thus, following Lagerloef et al. (1999) the weighted sum of the midlatitude and equatorial geostrophic ␷ at Ϯ2Њ weight functions Wf and W␤ to vary inversely with the meridional structure of the respective error variance of latitude provided an initial estimate of the mean geostrophic across these zonal sections. The same weighting Uf and U␤, such that the minimum variance is given to ␷ the weighted sum of (1). Using surface drifter data, they functions were applied to the Ekman transport at Ϯ2Њ. found that the weight functions can be approximated at Mean ␷ produced by this method was found to be small a given latitude (␪) by a Gaussian function with the in comparison to u, generally Ͻ5cmsϪ1 (Fig. 2). meridional decay scale (␪s) a free parameter, The inverse model conservation equation illustrates the balance among advection, diapycnal transport, and ␪ 2 W ϭ exp Ϫ , W ϭ 1 Ϫ W . (2) surface forcing. For a layer bounded by interface m and ␤ ␪ f ␤ []΂΃s m ϩ 1,

N hmϩ1 ⌬x ␳c (U ϩ b) dz ϩ (Ec) ϩ (wAc) Ϫ (wAc) ϩ (F ϩ F*) Ϫ (F ϩ F*) ϭ 0. (3) ͸ j jj jjmeridional cm cmϩ1 ccmccmϩ1 jϭ1 ͵ []hm

Here ⌬xj is the station spacing at pair j and cj is the ature conservation equations to include the effects of property value per unit mass at this pair. The initial the TIW eddy heat ¯ux. In the upper 60 m between 155Њ estimate of the velocity U is determined from the CTD, and 95ЊW, heat equivalent to a surface heat ¯ux of 100 Ϫ2 ADCP, and XBT data; Ejcj is the meridional Ekman Wm between 2ЊS and 2ЊN was transferred into the property transport at pair j; wcAc is the property dia- equatorial sector from the northern (75%) and southern pycnal transport across isopycnal layer m beneath the (25%) sectors. sea surface, including mechanical mixing in the mixed The solutions determined by inverse methods are layer where isopycnals are nearly vertical; and Fc is the greatly in¯uenced by uncertainties in the initial esti- total transformation driven by horizontal wind mixing mates of the velocity, diapycnal transfer rate and air± and surface buoyancy forcing across an outcropping is- sea ¯uxes (solution variance), and conservation state- opycnal. The subsequent system of simultaneous equa- ments (model variance). A priori layer conservation var- 2 tions is solved for the unknown velocity adjustments bj, iances of (3 Sv) for mass, (2 Sv ϫ layer mean tem- 2 2 the diapycnal property transfer rate wc (with c ϭ M, T, perature) for temperature, and (5 Sv ϫ layer mean salt) S) for the mth interface, and corrections to the wind- for salt are assumed (Sv ϵ 106 m3 sϪ1). An additional driven horizontal mixing and air±sea climatologiesF*c . constraint was that the entire region should balance to In the central and eastern equatorial Paci®c (155Њ± within an a priori variance of (4 Sv)2. No other con- 95ЊW) mixing across the SST fronts bounding the cold straints (e.g., Indonesian through¯ow transport or heat tongue, induced by TIWs, results in a net heat conver- ¯ux) were imposed on the model. gence into the equatorial Paci®c (Hansen and Paul 1984; The errors in absolute u from the ADCP include in- Baturin and Niiler 1997; Kessler et al. 1998). These strumental and navigational error, data processing er- studies suggest that the magnitude of meridional eddy rors, mapping errors, and probably most signi®cantly, heat ¯ux (100±180 W mϪ2) is larger than the air±sea geophysical aliasing errors from sparse temporal sam- heat ¯ux. Hansen and Paul (1984) found that the north- pling. Considering these potential sources of error, we ern TIWs dominate the eddy heat ¯ux term and sug- assume an a priori variance for u of (6 ϫ 10Ϫ2 msϪ1) 2. gested an eddy heat ¯ux ratio of 0.75 to 0.25 between At Ϯ8Њ, ␷ is the sum of geostrophic and Ekman esti- the northern and southern TIWs, respectively. In our mates and we assign a variance of (1 ϫ 10Ϫ2 msϪ1) 2. study we added an additional constraint to the temper- However, the more involved ␷ estimates at Ϯ2Њ (see

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1031

FIG. 2. Salinity (pss-78 shaded) and initial meridional geostrophic velocity ␷ (black contours, 10Ϫ2 msϪ1 solid line northward, dashed line southward) distribution at (a) 2ЊN, and (b) 2ЊS. Also shown are the four potential density (kg m Ϫ3) surfaces (white contours) that divide the water column into the four general water classes used in this study. above) require a higher a priori variance, and we chose tainties of the tropical Paci®c Ocean circulation deter- the value (3 ϫ 10Ϫ2 msϪ1)2. mined by the model. Model solutions run with reason- Previous studies have found signi®cant vertical and able perturbations to the a priori variances are the same, cross-isothermal (ഠ diapycnal) mixing in the tropical within the given error, as those presented here. Paci®c Ocean (e.g., Wyrtki 1981; Bryden and Brady 1985; Johnson et al. 2001; Meinen et al. 2001) and also 3. Regional circulation indicate a strong depth dependence to the vertical velocity. Given the expected depth dependency of dia- The zonal evolution of the tropical Paci®c currents pycnal mixing, we assumed an a priori variance of (2 implied by the inversion (Table 2 and Fig. 3) can be ϫ 10Ϫ5 msϪ1) 2 for the upper nine isopycnal surfaces compared with previous estimates from a variety of Ϫ3 Ϫ6 Ϫ1 2 (␴␪ Յ 26.5 kg m ) and (2 ϫ 10 ms ) for remaining studies, methods and longitudes (Table 3). The esti- Ϫ3 isopycnal layers (26.7 Յ ␴␪ Յ 26.9 kg m ). mated current transports from this study and previous Finally, the errors associated with any air±sea ¯ux studies, at corresponding longitudes, are similar apart climatology are thought to be large (Speer and Tzip- from the North Equatorial Countercurrent (NECC) erman 1992; Barnier et al. 1995; Josey et al. 1999). where the present study de®nition of the NECC as east-

However, error estimates for the climatologies are rarely ward transport between 4Њ and 8ЊN for ␴␪ Յ 25.5 is provided. In this study we assumed an a priori variance more restrictive than that used in previous studies (Gou- which is a factor of (2 ϫ magnitude of wind stress and riou and Toole 1993; Johnson et al. 2002). The study freshwater ¯ux)2 and (50 W mϪ2) 2 for the heat ¯ux. boundary at 8ЊN results in an incomplete sampling of As stated, a priori model and solution variances great- the NECC, particularly in the eastern Paci®c. Also the ly in¯uence the model solution and associated errors. SEC(S) extends to 15Њ±20ЊS (Church and Boland 1983) We have attempted to quantify these variances to the but the present sampling only extends to 8ЊS. Johnson best of our ability and in all cases have tried not to et al. (2002) present a detailed description of the mean underestimate these variances. As a result, the posterior zonal currents and water properties, their seasonal cycle, errors are likely to represent upper bounds on uncer- and ENSO cycle.

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1032 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33 N Њ 7.0 8.0 ; EIC 3 2.3 1.9 0.4 2.4 2.5 Ϯ 2.4 0.4 Ϯ * * * * * * Ϫ 26.3 kg 26.9 kg W P&T Њ * Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ S and 1 Ն Յ Њ

␪ 33.2 ␪ 17.8 All regions W,C, and E

␴ ␴ 0.3 3.8 6.2 5.7 5.3 11.2 16.2 Ϫ Ϫ Յ Ϫ for ; NECC eastward 26.3 kg m 3 Њ Ϫ 2 Յ 1.6 2.8 ␪ Ϯ r and Taft (1987), pre- W ␴ Њ Ϯ Ϯ * * * * * * 2.1 1.8 1.9 1.9 0.6 0.5 2.2 W95 HTM Њ 110 * Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 8.3 of equator; L&F: Lukas and N for 13.7 26.7 kg m Њ Њ 2 6.1 0.8 2.1 6.4 Յ W 15.9 14.3 24.0 Ϯ

␪ Њ Ϫ Ϫ Ϫ ␴ Eastern S and 4 Њ * * * * * * * 0 dbar, respectively; W&K: Wyrtki and ns, eastern Paci®c between 2 and 95 ate current not found at section. Њ Leetmaa 20.2 8.0 2.3 1.1 2.0 2.2 1.6 0.7 1.8 W 110 Њ Âin (1988), SURTROPAC direct measurement * 110 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 6.6 0.1 6.0 5.5 13.3 13.5 30.5 Ϫ W Ϫ Ϫ Њ 9.8 4.3 7.7 29.3 11.0 40.3 30.3 19.2 Ϫ 155 W&K Ϫ Ϫ Ϫ ഠ 2.1 1.8 1.4 0.9 0.6 1.9 2.0 W 125 Њ * Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 17 W the density range extends to 1.0 6.4 4.5 Ϯ Њ 22.0 19.1 25.3 11.4 Ϫ Ϫ Ϫ W Њ and/or eastward transport between the equator and * * * * * * * 3 L&F 155 Ϫ 3.5 25.6 ഠ 1.9 2.1 1.0 0.8 1.2 1.8 2.0 W 140 Ϯ Њ Central * Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ E semiannual between 1984 and 1986, referenced to 1000 m and free-falling direct Њ 32.2 4.0 6.7 7.6 26.4 19.0 12.6 10.4 25.5 kg m Ϫ Ϫ Ϫ Ն

␪ W 5.7 5.1

␴ Њ Ϯ Ϯ for * * * * * * 0.8 1.8 1.9 1.0 1.0 1.5 0.8 W 155 ±150 Њ Њ K&T Њ * 2 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 21.1 35.1 Ϯ 170 180 Ϫ 9.1 3.2 9.0 6.9 , although east of and including 110 24.2 10.9 20.7 3 Ϫ Ϫ Ϫ Ϫ W between 1979 and 1981, referenced to 500 dbar; P&T: Picaut and Tournier (1991), Paci®c Ocean XBT Њ 2.6 1.1 1.2 1.7 1.8 1.1 2.0 0.5 Њ 10.6 26.1 11.1 18.9 41.7 Ϫ 26.3 kg m * Ϫ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Յ

␪ 6.7 7.9 8.3 2.0 ␴ 23.0 13.9 23.6 Ϫ Ϫ E Њ Ϫ Յ 6.5 4.9 * * 17.3 11.4 19.6 37.7 Ϫ G&T . New Guinea Coastal Undercurrent (NGCU) westward transport adjacent to the southern boundary for 24.0 165 Ϫ 3 Ϫ 1.1 1.3 2.0 1.7 0.9 2.2 0.4 E 180 Њ * Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 7.6 4.5 15.3 11.7 20.4 24.8 Ϫ Ϫ 7.6 6.7 7.4 0.3 18.2 13.1 15.1 Ϫ Ϫ ) for the tropical Paci®c currents. The zonal currents are de®ned as SEC westward transport between 8 25.50 kg m 1 Ϫ Ϫ s Յ ) for the tropical Paci®c currents from previous studies at a variety of locations. D&H: Delcroix and He 3 ␪ 1 Western of equator. Asterisks indicate current estimate not given or current not found at particular sections. Ϫ ␴ m of the equator for 22.5 s Њ 7.3 7.0 2.6 13.5 17.5 6 E; DEH: Delcroix et al. (1987), SURTROPAC CTD data at 165 3 E Њ 2 Њ Њ 0.8 1.5 1.4 1.0 1.3 0.6 1.5 E 165 2 m * * * Њ Ϯ 10 Ϯ Ϯ Ϯ Ϯ Ϯ W using XBT data and a mean temperature±salinity relation for the central Paci®c, relative to 450 m and poleward of 6 Ϯ * Њ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ DEH 165 N for ϫ Њ 10 8.8 19.7 14.8 27.4 38.8 2.8 9.6 0.5 2.0 ϫ C; HTM: Hayes et al. (1983), EPOCS data at 110 19.8 11.5 17.1 Њ Ϫ Ϫ Ϫ Ϫ and 150 Њ 4.8 E Њ 1.5 0.8 0.5 1.0 Ϯ E 156 * * * * * * * Њ * * * * Ϯ Ϯ Ϯ Ϯ D&H 165 7.0 Ϫ 8.2 0.8 9.3 14.7 Ϫ Ϫ 2. Transport estimates ( Ϫ 3. Transport estimates ( ABLE Äo mean between 180 T ; EUC eastward transport within . Errors quoted represent the formal error of the inverse method, i.e., the error associated with determining velocity adjustments. Asterisks indic ABLE 3 3 T Ϫ Ϫ Current Current 143 EUC NSCC NECC NGCU SSCC EUC NSCC NECC to a lower temperaturedata of from 13.5 1979 to 1985, poleward of velocity measurements; G&T: Gouriou and Toole (1993), combination of SURTROPAC, US/PRC and TEW spanning 1984 to 1991 for western Paci®c; K&T: Kessle El Nin Firing (1984), from Hawaii-to-Tahiti Shuttle Experiment, direct velocity measurements from pro®ling current mooring and CTD data referenced to 40 Kilonsky (1984), CTD sections of the Hawaii-to-Tahiti Shuttle Experiment (central Paci®c) referenced to 1000 m; Leetmaa (1982), direct observatio deep westward transport lying beneath the EUC;transport SCCs north eastward of transportm approximately poleward 4 of SEC (S) SEC (N) EIC SSCC SEC (S) SEC (N) Total SEC EIC m from free-falling current pro®ler at 165

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1033

FIG. 3. Volume transport (ϫ106 m3 sϪ1) of tropical Paci®c currents. (a) Eastward currents: EUC (black), SCCs (blue), and NECC (pink). (b) Westward currents: SEC(S,N) (black) and EIC (blue). The westward component of the NGCU (green) is also shown. Transports are only given at the boundaries separating the western and central region (170ЊW), central and eastern region (125ЊW), and at extreme western (143Њ and 156ЊE) and eastern (95ЊW) sections. Also shown are the box (25) regions (dashed lines) used in the inverse model. The boundaries are formed from a combination of CTD and ADCP data (Johnson et al. 2002) and historical CTD and XBT data in the eastern Paci®c.

The individual 25 boxes of the tropical Paci®c can steps around the coast of New Guinea with sections at be grouped into three large regions with distinct cir- 8Њ,5Њ, and 3ЊS (Fig. 3). For simplicity these steps are culation patterns: the western region from the 143ЊEto combined and shown schematically as transport at 8ЊS 170ЊW, the central region between 170Њ and 125ЊW, and (Fig. 4 and Table 4). the eastern region from 125ЊW to the coast of the Amer- The outer islands of the Bismarck Archipelago (New icas. The 11 isopycnals describe four general water clas- Ireland and Bougainville Islands) and New Guinea form ses (Table 1). Important features of the tropical Paci®c the western boundary of the equatorial Paci®c Ocean. circulation are described for each region, water class, The SEC impinging on this region is north of its bi- and sector (southern 8Њ±2ЊS, equatorial 2ЊS±2ЊN, and furcation zone at ϳ15ЊS (Church and Boland 1983) and northern 2Њ±8ЊN). Henceforth, we will describe trans- feeds the salty equatorward NGCU and NICU (Tsuchiya ports between these boxes, which combine eastward and et al. 1989; Butt and Lindstrom 1994). Observational westward currents. For example, boxes from 2Њ to 8ЊN and model studies (Pedlosky 1987; Tsuchiya et al. 1989; include both the eastward NECC and part of the west- Pedlosky 1991; Butt and Lindstrom 1994) suggest that ward SEC. the high salinity NGCU and NICU feed the EUC. Tsu- chiya et al. (1989) asked whether the NGCU overshoots a. Western region the equator to then retro¯ect and ¯ow eastward with the The western region circulation is complicated by the EUC, similar to observations (Flagg et al. 1986) of the termination of the SEC and EIC; the development of North Brazil Current in the Atlantic Ocean, and as sug- the EUC, SCCs, and NECC; and connections among gested by theories of equatorial western boundary dy- these currents with the low-latitude western boundary namics (Anderson and Moore 1979). CFC property dis- currents: the New Guinea Coastal Current (NGCC), tributions support such a hypothesis, with a South Pa- New Guinea Coastal Undercurrent (NGCU), New Ire- ci®c CFC minimum extending north to 2ЊN (Fine et al. land Coastal Undercurrent (NICU), and Mindanao Cur- 1994). rent (MC; Tsuchiya et al. 1989; Butt and Lindstrom In our model we ®nd westward transport of the NGCC 1994). An additional complication in the model is that and NGCU at 143ЊE and signi®cant eastward transport the southern boundary of the western Paci®c region at 156ЊE north of 2ЊN (Table 4) with a salinity signature

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1034 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33 similar to that of the NGCU, consistent with the property well 1953; Knauss 1966; Wyrtki and Kilonsky 1984; and theoretical studies mentioned above. In the equa- Lu et al. 1998) is seen. The zonal currents (EUC, SEC, torial sector, at 143ЊE, the eastward transport of the NECC, and SCCs) are fully developed and the ther- Ϫ3 lower thermocline (25.5 Յ ␴␪ Յ 26.3 kg m ) EUC mocline shoals to the east. Equatorial upwelling and and westward component of the NGCU effectively bal- surface poleward transport (Fig. 5 and Table 4) form ance (Fig. 4 and Table 4). the upward and surface limbs of the shallow tropical The inverse solution suggests that the NGCU does, and subtropical cells (Lu et al. 1998). Off-equatorial in fact, overshoot the equator and retro¯ect back to feed downwelling and subsurface meridional equatorward the EUC between 156ЊE and 170ЊW. This plus equa- transport complete the shallow tropical cells. Within the torward ¯ow of upper and lower thermocline water thermostad a complicated interaction among the deep across 2Њ and 3ЊS are major contributors to the increased currents occurs, and Southern Hemisphere water feeds transport of the EUC in the western Paci®c. The re- the extension of the EUC into the thermostad. maining contribution comes from the North Paci®c, Lu et al. (1998) document the indirect observational identi®ed by fresher water within the EUC. and model evidence for the existence of the shallow Tsuchiya (1991) also found northwest thermostad tropical circulation cell. Recent observational analysis transport in the NGCU. He suggested that the thermos- suggests that half of the water upwelling at the equator tad SCCs originate from surface and thermocline water downwells within Ϯ8Њ of the equator (Johnson 2001), of the Tasman Sea, although both southern and northern but does not quantify the subsurface limbs. The present properties are found in the NSCC (Gouriou and Toole study provides further direct observational evidence for 1993; Bingham and Lukas 1995; Johnson and Mc- the shallow tropical cells (Fig. 5 and Table 4). The north- Phaden 1999). Our study shows a complex set of in- ern and southern tropical cells exhibit poleward surface terconnections among the equatorward western bound- transport, diapycnal downwelling (Figs. 6a,c) between ary currents (MC, NGCU, and NICU), the westward Ϯ2Њ and Ϯ8Њ of 8(Ϯ5) and 10(Ϯ9) Sv, respectively, EIC, and the eastward SCCs. equatorward upper thermocline transport across Ϯ2Њ, In this study the SCCs, found within the thermostad and closing equatorial upwelling. The equatorward up- Ϫ3 (26.3 Յ ␴␪ Յ 26.9 kg m ), attain near-constant east- per thermocline transport across Ϯ2Њ effectively bal- ward transport by 165ЊE, although a local maximum is ances the large equatorial diapycnal upwelling of found at 170ЊW (Table 2). Transport of the westward 24(Ϯ4) Sv of upper thermocline water to the surface/ EIC, found beneath the EUC and ¯anked by the SCCs, mixed layer. reaches a maximum between 170ЊW and 180Њ. The The conversion of 24(Ϯ4) Sv of EUC upper ther- SCCs eastward transport, at 170ЊW, is greater than that mocline water to surface/mixed layer (Fig. 6b, right of the westward EIC in the equatorial sector and results panel) results from the combined action of wind-driven in a net eastward thermostad property transport in all horizontal mixing, air±sea heat ¯uxes, and equatorward sectors (Fig. 4 and Table 4). The property transport TIW eddy heat ¯uxes as the EUC shoals along its east- across the southern boundary is poleward between ward path. Vertical mixing associated with the upward 165ЊE and 170ЊW(8ЊS), but farther west between 143Њ slope of the isopycnals (Fig. 6, left panel) does not result and 165ЊE (at 3Њ and 5ЊS) net property transport in the in the transformation of water between the upper ther- thermostad layer of the NGCU/NICU is equatorward. mocline and surface/mixed layer water classes. Between Much of the NGCU ¯ows unmodi®ed through the 155Њ and 125ЊW the assumed TIW eddy heat ¯ux (100 southwest corner of the model domain and leaves at the WmϪ2), distributed in the upper 60 m, adds heat to the western boundary (143ЊW). Eastward and equatorward surface/mixed layer and shoaling upper thermocline. thermostad property transports are found at the zonal The TIW heat ¯ux to the upper thermocline layer results and meridional boundaries of the northern sector. The in an interior diapycnal heat ¯ux of 4 W mϪ2 from the retro¯ection of the NGCU results in equatorward trans- upper thermocline to the surface/mixed layer. The in- port across 2ЊN (165ЊE±170ЊW) that feeds both the inversion correction to the air±sea heat ¯ux increases the creased westward transport of the EIC and the NSCC. initial COADS heat ¯ux from 64 to 81 W mϪ2 (155Њ± The EIC terminates before reaching the western bound- 125ЊW). These heat ¯uxes to the surface/mixed layer ary by poleward transport across 2 S that, together with Њ feed the 204 W mϪ2 net advective (poleward and zonal) the NGCU/NICU, ultimately feeds the SSCC. Equator- ward transport across 8ЊN feeds North Paci®c thermostad water into the NSCC. In the western region the → inversion suggests that isopycnal mixing within the thermostad dominates and diapycnal mixing is not signi®- FIG. 4. Circulation of the western Paci®c region for (a) volume cant. (ϫ106 m3 sϪ1), (b) temperature (PW), and (c) salt (ϫ106 kg sϪ1) transports. Ekman transport is included in the meridional transport estimates. Lateral and zonal transports are solid arrows. Lateral trans- b. Central region ports at 2ЊS and 2ЊN are shown (red arrows) on back face. Diapycnal property transport between layer interfaces for the southern, equator, In the central Paci®c (170Њ±125ЊW), unlike the west- and northern regimes are shown on front face. Water classes are ern region, the classical 3D tropical circulation (Crom- separated by bounding isopycnal surfaces-surface/mixed layer (red),

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1035

FIG.4.(Continued) upper thermocline (green), lower thermocline FIG. 5. As in Fig. 4 but for central Paci®c region. Air±sea trans- (blue), and thermostad (pink). Note that although the boundary of formations are shown at outcropping isopycnal. In the equatorial the southern sector is shown at 8ЊS for simplicity, it is actually found sector, interior diapycnal transports, where signi®cant, are also shown over a range of latitudes in the western region. at the outcropping isopycnal.

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1036 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33 E) Њ heat divergence. These terms balance to within the over- 0.8 0.4 1.2 4.0 0.3 0.2 0.7 1.8

N Ϫ2 Њ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ all formal uncertainty of 23 W m . Equatorial upwelling estimates for the central Paci®c 2.9 1.3 3.2 8.4 1.2 0.1 1.0 1.3 N boundary Њ are often provided in depth coordinates which combine 0.9 0.4 1.7 4.9 0.1 0.3 0.7 1.4 both diapycnal upwelling and the eastward shoaling of N8 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Њ W

Њ isopycnals (e.g., Wyrtki 1981; Poulain 1993; Weisberg r thermocline water; S and 2 Њ 1.8 6.2 3.5 8.1 1.2 1.3 9.9 1.7 95

E, and northern 156 and Qiao 2000; Johnson et al. 2001). In general, these Њ Ϫ Ϫ studies ®nd upwelling from the EUC to the surface/ N and (b) net zonal transport Њ 0.4 0.6 1.9 Eastern mixed layer. The Weisberg and Qiao (2000) and Johnson 4.8 0.9 1.5 Ϯ Ϯ Ϯ S2 Ϯ Ϯ Ϯ * et al. (2001) vertical velocity pro®les showed strongest Њ * l transport of the southern boundary 3.6 0.2 0.8 upwelling across the EUC upper thermocline with de- Southern Equator Northern 6.6 1.3 N, and 8 10.2 Њ Ϫ E, equator 143 Ϫ

Њ creasing but still positive vertical velocity through the S, 2

Њ EUC lower thermocline. Weisberg and Qiao (2000) also 4.7 1.2 1.6

S2 found evidence of downwelling into the thermostad. S, 2 Њ * 0.4 0.4 0.4 2.0 Ϯ Ϯ Ϯ Њ 8 Bryden and Brady (1985) estimated a velocity of 1 ϫ Ϯ Ϯ Ϯ Ϯ 4.5 1.7 1.8 10Ϫ5 msϪ1 across the 23ЊC isotherm between 150Њ and 1.8 1.6 1.2 0.6 Ϫ Ϫ 110ЊW and smaller downwelling into the thermostad. While their transport estimates were made between 5ЊS 1.0 4.7 0.7 0.7 0.4 1.6 0.1 0.5 and 5ЊN, cross-isotherm velocities were calculated as- N Ϯ Ϯ Ϯ Ϯ S. Also in the eastern Paci®c region the 2 Ϯ Ϯ Ϯ Ϯ Њ W Њ Њ suming con®nement to within Ϯ83 km of the equator. 1.4 0.4 1.6 7.0 8.4 0.4 11.6 125

14.0 The estimated mean cross-isothermal velocity of Mei- Ϫ Ϫ Ϫ Ϫ nen et al. (2001) has a similar vertical pro®le to that of S, and 3 Њ 1.0 1.1 4.3 0.7 Bryden and Brady (1985), but with signi®cantly reduced 0.7 0.4 1.9 0.1 Ϯ S, 5 N8 Ϯ Ϯ Ϯ

Њ magnitude because their estimate (also between 5 )

Ϯ Ϯ Њ Ϯ Ϯ Ϯ Њ was not arti®cially con®ned to an equatorial strip. 7.5 1.4 1.1 0.2 1.6 14.5 11. 8 Southern Equator Northern 10.1 Ϫ Ϫ Ϫ Ϫ In the present study the equivalent diapycnal velocity Ϫ Ϫ W), and at the extreme western (southern 165 Њ needed to convert upper thermocline to surface/mixed Central 1.3 3.3 0.9 0.7 layer water (Fig. 6b, left panel) is comparable to that S2 Ϯ Ϯ Ϯ Ϯ Њ

0.6 0.4 0.2 1.6 of Bryden and Brady. However, the inversion shows 0.8 6.4 1.0 Ϯ Ϯ Ϯ Ϯ that the net diapycnal mass transport between the sur- 12.6 Ϫ Ϫ

2.6 3.2 1.2 5.0 face/mixed layer and upper thermocline is dominated ) for the western, cental, and eastern Paci®c regions at 8 1

Ϫ by the effects of wind-driven horizontal mixing across 1.1 1.2 0.9 4.6 s 0.4 0.3 0.5 1.4 3 S2

Ϯ Ϯ Ϯ Ϯ outcropping isopycnals, with the eddy heat ¯uxes in- Њ m Ϯ Ϯ Ϯ Ϯ 8 6 W W. Asterisks indicate water class not found at section. Transport estimates are shown graphically in Figs. 4, 5,

Њ duced by the TIWs adding the necessary additional Њ 3.8 8.0 2.1 3.8 10 7.3 2.0 Ϫ 12.9 11.4 buoyancy required to convert upper thermocline water ϫ 170 Ϫ to surface/mixed layer water (Fig. 5). and 95

Њ In the equatorial sector of the central region the trans- 0.7 0.3 1.4 0.5 4.2 0.6 0.8 1.6 W), central and eastern region (125 Њ Ϯ Ϯ Ϯ Ϯ port of the EUC increases slightly from west to east N Ϯ Ϯ Ϯ Ϯ Њ (Table 2), although signi®cant upwelling of EUC upper 1.0 4.6 5.0 Southern Equator Northern 6.5 1.5 2.3 6.5 10.0 Ϫ Ϫ thermocline water is also seen. The increased transport Ϫ Ϫ Ϫ Ϫ of the EUC is the result of increased transport in the 4.1 0.6 1.4 1.8 lower thermocline and extension of the EUC to the ther- N8 0.7 0.5 0.3 1.4 Ϯ Ϯ Ϯ Ϯ

Њ mostad layer. Increased transport of the lower thermo- Ϯ Ϯ Ϯ Ϯ 2.5 9.6 0.1 cline layer is driven by diapycnal downwelling from the 12.7 Ϫ Ϫ Ϫ 5.6 1.5 7.8

Ϫ EUC upper thermocline and equatorward convergence 10.3 across Ϯ2Њ (Fig. 5, Fig. 6b, and Table 4). Western E 3.8 0.4 1.4 0.8 Њ 1.4 0.6 0.4 0.7 In the southern sector poleward transport of surface/ S2 Ϯ Ϯ Ϯ Ϯ Њ Ϯ Ϯ Ϯ Ϯ mixed layer water across 2ЊS feeds the increased west- ±165 Њ 0.9 0.7 1.8 3.8 3.3 0.6 7.1

15.0 ward property transport from 125Њ to 170ЊW, and to- Ϫ Ϫ Ϫ Ϫ Ϫ 143 Ϫ gether with the air±sea transformation (wind-driven hor- S for simplicity, it actually steps around the coast of New Guinea with sections at 8 Њ 1.4 0.3 0.6 2.4 2.2 1.4 4.5 0.8 izontal mixing, and heat and freshwater ¯uxes) of upper S2 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ

Њ thermocline water to the surface/mixed layer, maintains 8 W) sections for southern, equator, and northern sectors. Transport estimates are given for each water class: S/ML, surface and mixed layer; UTW, uppe

Њ poleward transport across 8ЊS (Fig. 5). The small net 1.3 1.1 2.8 6.2 4.0 3.8 0.7 1.9 Southern Equator Northern Ϫ Ϫ Ϫ Ϫ Ϫ upward diapycnal temperature and salt transports from 4. (a) Total meridional transport (including Ekman) extimates ( the upper thermocline to the surface/mixed layer (Fig.

ABLE 5) result from a slightly larger upward diffusion from T the relatively warm, salty upper thermocline layer that Water class Water class and 7. does not exterd to the America; transport estimates are between 125 LTW, lower thermocline water;is and nominally TW, thermostad given water. at Positive is 8 northward or eastward. Note that in the western Paci®c region while the meridiona and eastern (95 at the boundaries separating the western and central region (170 (a) S/ML UTW LTW TW (b) S/ML UTW LTW TW enters the tropical Paci®c region across 8ЊS. The upward

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1037

FIG. 6. Diapycnal velocity (10Ϫ5 msϪ1, left panel solid line) and transport (106 m3 sϪ1, right panel solid line) for each sector of the central Paci®c: (a) northern, (b) equatorial, and (c) southern. Diapycnal velocities and transports are given for Ϫ3 ␴␪ ϭ 23.0, 24.0, 25.5, 26.3, and 26.9 kg m , shown on common (internal) y axis. Also shown for the equatorial sector is the along isopycnal ``upwelling'' velocity (dashed). Air±sea diapycnal transports (right panel, dot-dashed) are shown at the outcropping isopycnals. Inverse method errors associated with determining adjustments to the diapycnal transport are shaded (right panel). diapycnal salt transport is related to the chimneylike ward thermostad transport across Ϯ8Њ maintains the po- structure of salinity in the central Paci®c (Johnson et tential vorticity and property gradients at the SCC cores. al. 2002). In the northern sector, as in the southern sector, The SCCs diverge poleward as they ¯ow east, and poleward transport in the surface/mixed layer across 8ЊN westward ¯owing deep SEC water wedges between the is maintained by air±sea transformation (wind-driven SCCs and EUC thermostad extension. At 170ЊW the horizontal mixing and buoyancy) of upper thermocline eastward SCCs dominate (Fig. 5). In the southern and water to the surface/mixed layer and net zonal conver- northern sector at 125ЊW the eastward SCCs and west- gence between 125Њ and 170ЊW (Fig. 5). ward deep SEC balance, resulting in small net eastward Johnson and McPhaden (1999) and Rowe et al. (2000) transports. Isopycnal mixing between the NSCC and describe the property homogeneity equatorward of the deep SEC(N) results in a net equatorward transport of SCC cores and the sharp potential vorticity fronts at the thermostad water across 2ЊN. Approximately one-half SCC cores that delineate the equatorward thermostad of the SSCC transported eastward in the equatorial sec- water from the poleward components of the SCCs. In tor, at 170ЊW, moves poleward across 2ЊS between 170Њ the present study, a complex circulation in the ther- and 125ЊW. The SSCC water that remains in the equa- mostad layer is found, with strong horizontal shears torial sector and the equatorward transport across 2ЊN among the deep SEC(N,S), SCCs, EUC, and EIC (see contributes to both the increased westward transport of Fig. 1c). Thermostad property homogeneity is found to the EIC and the extension of the EUC into the ther- result from mixing between the SCCs equatorward of mostad (Fig. 5 and Table 4). Mixing between the pole- their core and the deep SEC(N,S). Deep SEC(S,N) water ward trending SSCC and the SEC(S) reduces the west- replaces SCC water lost to the equator, while equatorward transport of the SEC(S), homogenizes the ther-

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1038 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33

mostad property gradients equatorward of the SSCC core, and allows diapycnal upwelling of thermostad water to the lower thermocline. Equatorward thermostad transport at Ϯ8Њ, into the northern and southern sectors, feeds recently ventilated subpolar mode/intermediate water into the tropical Paci®c.

c. Eastern region In the eastern region the EUC continues to shoal to 110ЊW (e.g., Johnson et al. 2001). Farther east it shifts southward and terminates (Lukas 1986). Poleward surface transport, off-equator downwelling and subsurface equatorward transport identify the shallow tropical cells (Fig. 7 and Table 4). Equatorial upwelling that exhausts the lightest layers of the EUC, exposing an increasingly denser, colder EUC to the east, results in the eastward cooling of the cold tongue. Lower thermocline and thermostad water that upwell to form the cold tongue are exported into the Northern Hemisphere. As the SCCs approach the eastern boundary, they diverge poleward (Johnson and McPhaden 1999; Kessler 2002) and northern and southern branches of the SEC develop with deep extensions to the thermostad layer. Interior equatorward transport of lower thermocline and thermostad water imports Southern Hemisphere mode water into the tropical region. In the model the northern and southern boundaries at Ϯ8Њ are extended from 95ЊW to the coast of the Amer- icas (Fig. 3). For simplicity the meridional transports across these sections are combined with the meridional transport across Ϯ8Њ between 125Њ and 95ЊW and shown schematically between these longitudes (Fig. 7). While zonal transports at 95ЊW are shown, the diapycnal property transports for the region between 95ЊW and the Americas (from 8ЊSto8ЊN) are folded into the equatorial sector (Fig. 7 and Fig. 8b). The shallow tropical overturning cell is found in the northern and southern sectorsÐpoleward surface transport (surface/mixed layer in the northern region and upper thermocline in the southern sector), downwelling [northern 7(Ϯ12) Sv, and southern 10(Ϯ7) Sv], and equatorward subsurface transport (Fig. 7, Table 4, Figs. 8a,c). In the southern sector the downward limb of the Ϫ3 shallow tropical cell is found across ␴␪ ϭ 24.0 kg m , which is within the upper thermocline layer (and thus not shown in Fig. 7, see Fig. 8c). However, the subsurface equatorward convergence does not completely balance the surface poleward divergence. Instead, upwelling from all layers of the EUC feeds the poleward transport (Fig. 8b).

←

estimates between 125Њ and 95ЊWatϮ8Њ. Sections at Ϯ2Њ are between FIG. 7. As in Fig. 4, but for eastern Paci®c region. Air±sea trans- 125Њ and 95ЊW. Whereas zonal transports at 95ЊW are shown at the formation are shown at the outcropping isopycnal. At the eastern eastern boundary, including those in the equatorial sector, the dia- boundary historical CTD and XBT data close the Ϯ8Њ sections to the pycnal property transports for the region between 95ЊW and the Amer- American coast; those transports are combined with transport icas are folded into the equatorial region.

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1039

FIG. 8. As in Fig. 6, but for eastern Paci®c sectors. Diapycnal velocity and transport for the region between 95ЊW and the American coast are folded into the equatorial region.

Observational (Lukas 1986) and model (Pedlosky COADS heat ¯ux of 83 W mϪ2, essentially unchanged 1987, 1988) studies describe the termination of the by the inversion and the 24 W mϪ2 diapycnal heat ¯ux, EUC. Lukas (1986) suggests that the EUC terminates feed the 184 W mϪ2 net advective (poleward) heat di- via upwelling east of the Galapagos Islands between the vergence. These terms balance to within the overall for- equator and 5ЊS. Pedlosky (1988) suggests that the EUC, mal uncertainty of 44 W mϪ2. modeled as an inertial jet in which cross-isopycnal mix- In the southern sector poleward upper thermocline ing is included, is exhausted before reaching the eastern transport at 2ЊS, westward transport at 95ЊW and equa- boundary by gradual upwelling from successively dens- torward transport at 8ЊS, which mark the beginnings of er EUC layers. The present study suggests that both the SEC in the model domain, feed the increased west- mechanisms are active in the termination of the EUC. ward transport of the SEC(S) from 95Њ to 125ЊW as well In the equatorial sector at 125ЊW, the EUC brings as air±sea transformation of upper thermocline water to upper and lower thermocline and thermostad water into the surface/mixed layer (Fig. 7 and Table 4). Interior the region that, together with diapycnal upwelling of diapycnal downwelling (mass and temperature) from the mass from all layers of the EUC (Fig. 7 and Fig. 8b), upper thermocline to the lower thermocline layer is seen, feeds the net poleward transport at 2ЊS and air±sea trans- while diapycnal upwelling of salt results in a salt chim- formation of 13(Ϯ4) Sv of EUC upper thermocline wa- ney similar to, although weaker than, that seen in the ter to the surface/mixed layer. The upwelling halves the central region (Fig. 7). transport of the EUC between 125Њ and 95ЊW and the Johnson and McPhaden (1999) suggest that the pole- lightest layers of the EUC are exhausted, preferentially ward spreading SCCs form the eastward limb of elon- exposing an increasingly denser, colder EUC to the east. gated cyclonic gyres, while Rowe et al. (2000) also ®nd Much of the diapycnal upwelling (70%) of EUC lower anticyclonic gyres. In the present study we ®nd pole- thermocline water occurs in the coastal region between ward transport across 8ЊN supporting a cyclonic recir- 95ЊW and the American coast. Between 125Њ and 95ЊW culation, while deep westward transport in the SEC(N,S) the TIW eddy heat ¯ux (100 W mϪ2) and direct air±sea in the eastern Paci®c support the existence of the an-

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1040 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33 ticyclonic gyres (Table 2). However, we ®nd that the salinity maximum subtropical mode water increases the southern cyclonic and anticyclonic gyres are not com- transport of the shallow overturning lower limb. This pletely closed recirculations. source of Southern Hemisphere water forms part of a Thermostad water in the southern sector is transported net interhemispheric exchange. equatorward at 8Њ and 2ЊS (Fig. 7 and Table 4), prin- Eastward shoaling of the thermocline coupled with cipally between 125Њ and 110ЊW. In the equatorial sector diapycnal upwelling from the EUC thermostad through the Southern Hemisphere input and eastward transport the upper thermocline layers, in the eastern region, re- of EUC thermostad water at 125ЊW feed the poleward sults in the equatorial sea surface temperature minimum, ¯ow at 2ЊN and the diapycnal upwelling of EUC ther- the (Paci®c) cold tongue. The spatial structure of the mostad water to EUC lower thermocline water. Further cold tongueÐeastward cooling of sea surface temper- poleward transport of thermostad water across 8ЊN oc- atureÐis determined by diapycnal upwelling that pref- curs between 125ЊW and the Americas where the NSCC erentially exhausts the lightest layers of the EUC along turns north under the Costa Rica Dome (Johnson and its eastward transit of the eastern region, allowing deep- McPhaden 1999; Kessler 2002). er, colder, EUC lower thermocline and thermostad water Equatorward thermostad water transport across the to upwell east of 95ЊW and adjacent to the coast of the southern boundary feeds Southern Ocean mode/inter- Americas. Thus the EUC terminates both through in- mediate water into the tropical Paci®c. This water fol- terior upwelling (Pedlosky 1988) of 10(Ϯ6) Sv of upper lows a convoluted circulation path and eventually sup- thermocline water and upwelling adjacent to the Amer- plies equatorward thermostad transport at 2ЊS. Once icas (Lukas 1986) of 9(Ϯ4) Sv of lower thermocline within the equatorial region this water may directly up- and thermostad water. The buoyancy (heat) input needed well into the lower thermocline feeding the equatorial to drive the upwelling from the EUC layers is 0.10 PW upwelling or circulate within the northern equatorial and 0.13 PW, respectively. Air±sea transformation of anticyclonic cell. 13(Ϯ4) Sv of the upwelling EUC water to the surface/ mixed layer and poleward transport across 2Њ and 8ЊN between 95ЊW and the American coast allow an inter- 4. Circulation of the tropical Paci®c Ocean and hemispheric transport to occur in the eastern Paci®c. impacts on the cold tongue From this study, differences in the strength and depth The inverse model shows that the circulation of the of equatorial upwelling between the central and eastern tropical Paci®c can naturally be separated into three region are notable (Fig. 6b and Fig. 8b). (Recall that in regions. West of 170ЊW the SEC feeds the Southern the eastern region diapycnal velocity and transports be- Hemisphere low-latitude western boundary currents tween 95ЊW and the American coast and 8ЊN and 8ЊS (NGCU, NICU). These equatorward boundary currents are folded into the equatorial upwelling. This may result are essential elements in the development of the EUC in an underestimate of the equatorial diapycnal velocity, and, together with the EIC, in the development of the but not transport, in the eastern Paci®c.) In the central SCCs. We ®nd that the NGCU crosses the equator in region diapycnal upwelling is found above the core of Ϫ3 the far west and retro¯ects to the equator across 2ЊN the EUC (␴␪ Յ 24.0 kg m , ϳ75 m) and weak dia- between 143Њ and 165ЊE. This, together with direct input pycnal downwelling below the EUC core (␴␪ Ն 25.5 from the NICU (2ЊS) results in the eastward salinity kg mϪ3). Air±sea transformation across outcropping is- Ϫ3 increase of the EUC (Gouriou and Toole 1993). Net opycnals (␴␪ Յ 23.0 kg m ), principally through wind- equatorward transport across the southern boundary re- driven horizontal mixing, is signi®cantly larger than the sults from these large western boundary transports west interior diapycnal transport. However, the heat ¯ux re- of 165ЊE, partly compensated by poleward (Ekman) quired to support the air±sea transformation is nearly transport east of 165ЊE. This is the signature of the SEC equally derived from the inversion corrected COADS shifting poleward as it approaches the western boundary heat ¯ux of 81 W mϪ2 between 155Њ and 125ЊW and where mixing in the Solomon and Bismark Seas partly the imposed 100 W mϪ2 TIW eddy heat ¯ux. In the erodes the high salinity SEC signature. These waters eastern region diapycnal upwelling extends from the form the western boundary currents that eventually feed thermostad to the surface/mixed layer (␴␪ Յ 26.3 kg the EUC. mϪ3, ϳ175 m). In this region interior diapycnal mixing In the central and eastern regions the downward limbs dominates and air±sea transformations play a minor of the shallow tropical overturning cell are 15(Ϯ13) Sv role, apart from in the northern sector. in the north and 20(Ϯ11) Sv in the south. The tropical Bryden and Brady (1985), in the central Paci®c, found cells are closed by subsurface equatorward transport, similar cross-isotherm (ഠdiapycnal) upwelling veloci- equatorial upwelling, and poleward surface transport ties above the core of the EUC and smaller upwelling across Ϯ2Њ. The lower limbs of the shallow tropical cells below the core. The differences in diapycnal velocity feed relatively high and low salinity southern and north- estimates below the EUC core between the current study ern subtropical water, respectively, into the equatorial and the Bryden and Brady (1985) study is likely a result region that reinforces the sharp salinity front at the equa- of the previous study estimating the upwelling rate over tor. In the southern sector the equatorward transport of an area that combined the central and eastern Paci®c

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1041

FIG. 9. Net volume transport (ϫ106 m3 sϪ1) across the northern, western, and southern boundaries. The Ϫ3 mean ␴␪ of meridional transport across the southern boundary is ␴␪ ϭ 24.0 kg m at the western boundary, Ϫ3 Ϫ3 ␴␪ ϭ 25.5 kg m for the central region, and ␴␪ ϭ 26.7 kg m for the eastern region. Across the northern Ϫ3 boundary (8ЊN) the mean ␴␪ density of the equatorward and poleward transport is ␴␪ ϭ 26.3 kg m for Ϫ3 Ϫ3 the western region, ␴␪ ϭ 22.5 kg m for the central region, and ␴␪ ϭ 25.5 kg m for the eastern region. regions. The reader is also reminded that while the pre- in the Paci®c cold tongue and provides a mechanism by vious study transport estimates were made between Ϯ5Њ which Southern Hemisphere mode/intermediate water of the equator, the cross-isothermal velocities were cal- can in¯uence the equatorial Paci®c sea surface temper- culated assuming con®nement to a narrow equatorial ature. The properties of these waters, including their strip. potential vorticity, are reset by surface forcing in the Net equatorward transport of 13(Ϯ13) Sv of Southern equatorial region and thus can be transported poleward Hemisphere water (Fig. 9) forms the Paci®c branch of across 2ЊN in the basin interior. The upwelling in the the Paci®c±Indian interbasin exchange (Macdonald eastern Paci®c is one route by which Southern Ocean 1998; Gordon et al. 1999; Sloyan and Rintoul 2001). mode/intermediate water, participating in the Indian± At the southern boundary of the central region, the trans- Paci®c interbasin exchange, irreversibly crosses the port occurs in the upper thermocline layers (salinity equator. maximum), while at that of the eastern region it occurs in the thermostad layer. In the southern sector of the 5. Conclusions western region, poleward and westward transport between 165ЊE and 170ЊW, principally surface/mixed layer Mean meridional temperature, salinity, and zonal ve- and upper thermocline water, feeds the equatorward locity sections (Johnson et al. 2002) and historical CTD low-latitude western boundary currents. The Southern and XBT data were combined with COADS surface Hemisphere water participates in the tropical Paci®c climatologies by inverse methods to estimate the three- Ocean circulation. The result, in the northern sector, is dimensional circulation of the tropical Paci®c Ocean. enhanced eastward transport in all layers in the western The inversion showed that the circulation of the tropical region and net northward transport in the surface/mixed Paci®c Ocean can be divided into three regions: western layer in the eastern region. (143ЊEto170ЊW), central (170ЊW±125ЊW), and eastern Input of southern origin upper thermocline water in (125ЊW±eastern boundary). In the western region mix- the central Paci®c de®nes the interior (basin) pathway ing between terminating westward currents (SEC and by which South Paci®c subtropical water enters the trop- EIC) and low-latitude western boundary currents de- ical Paci®c region (Johnson and McPhaden 1999). The termines the properties of the eastward EUC, SCCs, and upper thermocline water increases the westward SEC(S) NECC. These interactions occur within similar density transport and strengthens the lower limb of the Southern horizons and diapycnal property transports are not sig- Hemisphere shallow tropical overturning cell. Thus, ni®cant. equatorward transport in the southern sector of the cen- The shallow tropical cells [15(Ϯ13) Sv in the northern tral region can feed both direct equatorward interior sector and 20(Ϯ11) Sv in the southern sector] are seen transport and indirect transport through the equatorward in the central and eastern regions. In the central region low-latitude western boundary. Thermostad water, equatorial upwelling of 24(Ϯ4) Sv from above the core whose origin is mode/intermediate water of the Southern of the EUC de®nes the upper limb of the combined Ocean, enters the tropical Paci®c in the eastern region shallow tropical and subtropical cells. Decreased pole- and at the western boundary, circulates in and mixes ward transport between Ϯ2Њ and Ϯ8Њ and diapycnal with the SCCs, EIC and deep SEC(N,S). Equatorial up- downwelling from the surface/mixed layer to the upper welling of thermostad water in the eastern Paci®c results thermocline form the poleward and downward limb of

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC 1042 JOURNAL OF PHYSICAL OCEANOGRAPHY VOLUME 33 the shallow tropical cell. In the central region the trop- Oceanic and Atmospheric Research. Timely comments ical cells are closed by subsurface equatorward transport from Peter Niiler on the effects of TIW on the equatorial of upper thermocline water that nearly balances the heat budget were useful. Comments from two anony- equatorial upwelling. In spite of this upwelling, EUC mous reviewers and Eric Firing improved the manu- transport increases due to equatorward input of lower script. thermocline and thermostad water, part of the subtropical cells. Similar to this study, modeling studies (e.g., Lu and McCreary 1995; Blanke and Raynaud 1997) REFERENCES found that the shallow tropical overturning cells do not reduce the net transport of the EUC. Anderson, D. L. T., and D. W. Moore, 1979: Cross-equatorial inertial jets with special relevance to very remote forcing of the Somali The large errors on the meridional and diapycnal Current. Deep-Sea Res., 26, 1±26. transports re¯ect the upper bounds on uncertainties of Barnier, B., L. Siefridt, and P. Marchesiello, 1995: Thermal forcing these components of the tropical Paci®c circulation. The for a global ocean circulation model using a three year clima- ADCP velocities result in zonal current transports that tology of ECMWF analysis. J. Mar. Sys., 6, 363±380. Baturin, N. G., and P. P. Niiler, 1997: Effects of instability waves in are well determined. Improving the initial estimate of the mixed layer of the equatorial Paci®c. J. Geophys. Res., 102, meridional transport via direct velocity and shear ob- 27 771±27 793. servations will greatly reduce the uncertainties of the Bingham, F. M., and R. Lukas, 1995: The distribution of intermediate meridional and diapycnal mass and property transports. water in the western equatorial Paci®c during January±February The Paci®c cold tongue in the eastern region results 1986. Deep-Sea Res., 42, 1545±1573. Blanke, B., and S. Raynaud, 1997: Kinematics of the Paci®c equa- from isopycnal shoaling and from diapycnal upwelling torial undercurrent: An Eulerian and Lagrangian approach from through all layers of the EUC. The spatial structure of GCM results. J. Phys. Oceanogr., 27, 1038±1053. the cold tongueÐsea surface temperature cooling in an Bryden, H. L., and E. C. Brady, 1985: Diagnostic model of the three- eastward directionÐis due to the preferential diapycnal dimensional circulation in the upper equatorial Paci®c Ocean. J. Phys. Oceanogr., 15, 1255±1273. upwelling of lightest (warmer) EUC upper thermocline Butt, J., and E. Lindstrom, 1994: Currents off the east coast of New near 125ЊW. Upwelling of 10(Ϯ6) Sv to the east ex- Ireland, Papua New Guinea, and their relevance to the regional hausts the lightest EUC water, exposing cooler, denser undercurrents in the western equatorial Paci®c Ocean. J. Geo- EUC water as it approaches the Americas. Adjacent to phys. Res., 99, 12 503±12 514. the coast of the Americas 9(Ϯ4) Sv of EUC lower ther- Chelton, D. B., R. A. DeSzoeke, M. G. Schlax, K. El Naggar, and N. Siwertz, 1998: Geographical variability of the ®rst baroclinic mocline and thermostad water upwells, resulting in the rossby radius of deformation. J. Phys. Oceanogr., 28, 433±460. coolest sea surface temperature. Church, J. A., and F. M. Boland, 1983: A permanent undercurrent A net 13(Ϯ13) Sv transport from the southern to the adjacent to the Great Barrier Reef. J. Phys. Oceanogr., 13, 1747± northern Paci®c Ocean is implied by the inversion. The 1749. Cromwell, T., 1953: Circulation in a meridional plane in the central net equatorward transport across the southern boundary equatorial Paci®c. J. Mar. Res., 12, 196±213. is achieved at the western boundary via western bound- da Silva, A. M., C. C. Young, and S. Levitus, 1994: Algorithms and ary currents (NGCU and NICU), via interior transport Procedures. Vol. 1, Atlas of Surface Marine Data 1994, NOAA of upper thermocline salinity maximum water in the Atlas NESDIS, 83 pp. central region and via equatorward transport of ther- Delcroix, T., and C. HeÂin, 1988: Observations of the equatorial intermediate current in the western Paci®c Ocean (165ЊE). J. Phys. mostad water in the eastern region. Southern Hemi- Oceanogr., 18, 363±366. sphere water that enters the tropical Paci®c Ocean ÐÐ, G. Eldin, and C. HeÂin, 1987: Upper ocean water masses and moves into the North Paci®c by upwelling in the eastern transports in the western tropical Paci®c (165ЊE). J. Phys. Ocean- equatorial Paci®c and air±sea transformation which ogr., 17, 2248±2262. Fine, R. A., R. Lukas, F. M. Bingham, M. J. Warner, and R. H. drives the poleward transport across 2Њ and 8ЊN between Gammon, 1994: The western equatorial Paci®c: A water mass 95ЊW and the Americas. In particular, Southern Ocean crossroads. J. Geophys. Res., 99, 25 063±25 080. thermostad (mode/intermediate) water that enters the Flagg, C. N., R. L. Gordon, and S. McDowell, 1986: Hydrographic tropical Paci®c participates in the circulation of and and current observations on the continental slope and shelf of mixing between the SCCs, EIC, and deep SEC(N,S), the western equatorial Atlantic. J. Phys. Oceanogr., 16, 1412± 1429. eventually feeding the dense thermostad water that up- Gill, A. E., 1982: Atmosphere±Ocean Dynamics. Academic Press, wells in the eastern Paci®c. This suggests that ther- 662 pp. mostad water upwelling in the eastern Paci®c is one Gordon, A. L., R. D. Susanto, and A. F®eld, 1999: Through¯ow route by which the Paci®c±Indian interbasin exchange within Makassar Strait. Geophys. Res. Lett., 26, 3325±3328. Gouriou, Y., and T. Toole, 1993: Mean circulation of the upper layers is achieved. of the western equatorial Paci®c Ocean. J. Geophys. Res., 98, 22 495±22 520. Acknowledgments. This work was performed while Gu, D., and S. G. H. Philander, 1997: Interdecadal climate ¯uctuations BMS held a National Research Council Research As- that depend on exchange between the tropics and extratropics. sociateship Award at NOAA's Paci®c Marine Environ- Science, 275, 805±807. Hansen, D. V., and C. A. Paul, 1984: Genesis and effects of long mental Laboratory (PMEL). Support was also provided waves in the equatorial Paci®c. J. Geophys. Res., 89, 10 431± through the NOAA Of®ce of Global Programs (PAN 10 440. American studies program) and the NOAA Of®ce of Hayes, S. P., J. M. Toole, and L. J. Mangum, 1983: Water-mass and

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC MAY 2003 SLOYAN ET AL. 1043

transport variability at 110Њ in the equatorial Paci®c. J. Phys. Macdonald, A. M., 1998: The global ocean circulation: A hydro- Oceanogr., 13, 153±168. graphic estimate and regional analysis. Progress in Oceanog- Johnson, G. C., 2001: The Paci®c Ocean subtropical cell surface limb. raphy, Vol. 41, Pergamon, 281±382. Geophys. Res. Lett., 28, 1771±1774. Meinen, C. S., M. J. McPhaden, and G. C. Johnson, 2001: Vertical ÐÐ, and M. J. McPhaden, 1999: Interior pycnocline ¯ow from the velocities and transports in the equatorial Paci®c during 1993± subtropical to the equatorial Paci®c Ocean. J. Phys. Oceanogr., 99. J. Phys. Oceanogr., 31, 3230±3248. 29, 3073±3089. Moore, D. W., and S. G. H. Philander, 1977: Modeling the equatorial ÐÐ, ÐÐ, and E. Firing, 2001: Equatorial Paci®c Ocean horizontal oceanic circulation. The Sea, E. D. Goldberg, Ed., Marine Mod- velocity, divergence, and upwelling. J. Phys. Oceanogr., 31, eling, Vol. 6, Wiley and Sons, 319±361. 839±849. Pedlosky, J., 1987: An inertial theory of the equatorial undercurrent. ÐÐ, B. M. Sloyan, W. S. Kessler, and K. E. McTaggart, 2002: Direct J. Phys. Oceanogr., 17, 1978±1985. measurements of upper ocean currents and water properties ÐÐ, 1988: Entrainment and the termination of the equatorial un- across the tropical Paci®c during the 1990s. Progress in Ocean- dercurrent. J. Phys. Oceanogr., 18, 880±886. ography, Vol. 52, Pergamon, 31±61. ÐÐ, 1991: The link between the western boundary currents and Josey, S. A., E. C. Kent, and P. K. Taylor, 1999: New insights into equatorial undercurrents. J. Phys. Oceanogr., 21, 1553±1558. the ocean heat budget closure problem from analysis of the SOC Picaut, J., and R. Tournier, 1991: Monitoring the 1979±1985 equa- air±sea ¯ux climatology. J. Climate, 12, 2856±2880. torial Paci®c current transports with expendable bathythermo- Kessler, W. S., 2002: Mean three-dimensional circulation in the north- graph data. J. Geophys. Res., 96, 3263±3277. east tropical Paci®c. J. Phys. Oceanogr., 32, 2457±2471. Poulain, P.-M., 1993: Estimates of horizontal divergences and vertical ÐÐ, and B. A. Taft, 1987: Dynamic heights and zonal geostrophic velocity in the equatorial Paci®c. J. Phys. Oceanogr., 23, 601± transports in the central tropical Paci®c during 1979±84. J. Phys. 607. Oceanogr., 17, 97±122. Rowe, G., E. Firing, and G. Johnson, 2000: Paci®c equatorial subsurface countercurrent velocity, transport, and potential vorticity. ÐÐ, L. M. Rothstein, and D. Chen, 1998: The annual cycle of SST J. Phys. Oceanogr., 30, 1172±1187. in the eastern tropical Paci®c, as diagnosed in an ocean GCM. Sloyan, B. M., and S. R. Rintoul, 2000: Estimates of area-averaged J. Climate, 11, 777±799. diapycnal ¯uxes from basin-scale budgets. J. Phys. Oceanogr., Kleeman, R., J. P. McCreary Jr., and B. A. Klinger, 1999: A mech- 30, 2320±2341. anism for generating ENSO decadal variability. Geophys. Res. ÐÐ, and ÐÐ, 2001: Circulation, renewal and modi®cation of Ant- Lett., 26, 1743±1746. arctic mode and intermediate water. J. Phys. Oceanogr., 31, Knauss, J. A., 1966: Further measurements and observations on the 1005±1030. Cromwell Current. J. Mar. Res., 24, 205±240. Speer, K., and E. Tziperman, 1992: Rates of water mass formation Lagerloef, G. S. E., G. T. Mitchum, R. G. Lukas, and P. P. Niiler, in the North Atlantic Ocean. J. Phys. Oceanogr., 22, 93±104. 1999: Tropical Paci®c near-surface currents estimated from al- Tsuchiya, M., 1991: Flow path of Antarctic Intermediate Water in timeter, wind, and drifter data. J. Geophys. Res., 104, 23 313± the western equatorial South Paci®c Ocean. Deep-Sea Res., 38, 23 326. s273±s279. Leetmaa, A., 1982: Observations of near-equatorial ¯ows in the east- ÐÐ, R. Lukas, R. A. Fine, E. Firing, and E. Lindstrom, 1989: Source ern Paci®c. J. Mar. Res., 40 (Suppl.), 357±370. waters of the Paci®c equatorial undercurrent. Progress in Ocean- Lu, P., and J. P. McCreary Jr., 1995: In¯uence of the ITCZ on the ography, Vol. 23, Pergamon, 101±147. ¯ow of thermocline water from the subtropical to equatorial Weisberg, R. H., and L. Qiao, 2000: Equatorial upwelling in the Paci®c Ocean. J. Phys. Oceanogr., 25, 3076±3088. central Paci®c estimated from moored velocity pro®lers. J. Phys. ÐÐ, ÐÐ, and B. A. Klinger, 1998: Meridional circulation cells and Oceanogr., 30, 105±124. the source waters of the Paci®c equatorial undercurrent. J. Phys. Wunsch, C., 1996: The Ocean Circulation Inverse Problem. Cam- Oceanogr., 28, 62±84. bridge University Press, 442 pp. Lukas, R., 1986: The termination of the equatorial undercurrent in Wyrtki, K., 1981: An estimate of equatorial upwelling in the Paci®c. the eastern Paci®c. Progress in Oceanography, Vol. 16, Per- J. Phys. Oceanogr., 11, 1205±1214. gamon, 63±90. ÐÐ, and B. Kilonsky, 1984: Mean water and current structure during ÐÐ, and E. Firing, 1984: The geostrophic balance of the Paci®c the Hawaii-to-Tahiti Shuttle experiment. J. Phys. Oceanogr., 14, equatorial undercurrent. Deep-Sea Res., 31, 61±66. 242±254.

Unauthenticated | Downloaded 09/25/21 10:39 PM UTC