AUGUST 2003
ZHANG ET AL.
1783
Observational Evidence for Flow between the Subtropical and Tropical Atlantic: The Atlantic Subtropical Cells* DONGXIAO ZHANG Joint Institute for the Study of the Atmosphere and Ocean, University of Washington, and NOAA/Pacific Marine and Environmental Laboratory, Seattle, Washington
MICHAEL J. MCPHADEN NOAA/Pacific Marine and Environmental Laboratory, Seattle, Washington
WILLIAM E. JOHNS Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida (Manuscript received 16 January 2002, in final form 6 February 2003) ABSTRACT This study determines the mean pathways and volume transports in the pycnocline and surface layer for water flowing between the subtropical and tropical Atlantic Ocean, using potential vorticity, salinity, geostrophic flow maps on isopycnal surfaces, and surface drifter velocities. In both hemispheres, subducted salinity maximum waters flow into the Tropics in the pycnocline along both interior and western boundary pathways. The North Atlantic ventilating trajectories are confined to densities between about 23.2 and 26.0 s u, and only about 2 Sv (Sv [ 10 6 m 3 s 21 ) of water reaches the Tropics through the interior pathway, whereas the western boundary contributes about 3 Sv to the equatorward thermocline flow. Flow on shallower surfaces of this density range originates from the central Atlantic near 408W between 128 and 168N whereas flow on the deeper surfaces originates from near 208W just off the coast of Africa at higher latitudes. The pathways skirt around the potential vorticity barrier located under the intertropical convergence zone and reach their westernmost location at about 108N. In the South Atlantic, about 10 Sv of thermocline water reaches the equator through the combination of interior (4 Sv) and western boundary (6 Sv) routes in a slightly higher density range than in the North Atlantic. Similar to the North Atlantic, the shallower layers originate in the central part of the basin (along 108–308W at 108–158S) and the deeper layers originate at higher latitudes from the eastern part of the basin. However, the ventilation pathways are spread over a much wider interior window in the Southern Hemisphere than in the Northern Hemisphere that at 68S extends from 108W to the western boundary. The equatorward convergent flows in the thermocline upwell into the surface layer and return to the subtropics through surface poleward divergence. As much as 70% of the tropical Atlantic upwelling into the surface layer is associated with these subtropical circulation cells, with the remainder contributed by the warm return flow of the large-scale thermohaline overturning circulation.
1. Introduction Subtropical cells (STCs) are shallow meridional overturning cells that transport water subducted in the subtropics during the winter season to the Tropics, where it is upwelled to the surface. The upwelled water is modified by air–sea heat exchange and then advected * National Oceanic and Atmospheric Administration Contribution Number NA17RJ1232. Corresponding author address: Dr. Dongxiao Zhang, NOAA/Pacific Marine Environmental Laboratory, 7600 Sand Point Way NE, Seattle, WA 98115. E-mail:
[email protected]
q 2003 American Meteorological Society
back to the subtropics by poleward Ekman flows in the surface layer to complete the STC. The pathways and transports of STCs in the Pacific have been extensively studied both observationally and theoretically (e.g., McCreary and Lu 1994; Liu et al. 1994; Johnson and McPhaden 1999), and have been implicated in changing tropical sea surface temperatures (SST) on decadal timescales through advection of water mass anomalies or changes of circulation (e.g., Gu and Philander 1997; Kleeman et al. 1999; McPhaden and Zhang 2002; Schneider et al. 1999; Hazeleger et al. 2001). Unlike the STCs in the Pacific, fewer studies have been conducted on the STCs in the Atlantic, though several modeling papers (Fratantoni et al. 2000; Inui et al. 2002; Lazar et al. 2002; Malanotte-Rizzoli et al.
1784
JOURNAL OF PHYSICAL OCEANOGRAPHY
2000; Jochum and Malonotte-Rizzoli 2001; Hazeleger et al. 2003) have recently appeared. Malanotte-Rizzoli et al. (2000), for example, based their study on evaluations of the Bernoulli function on isopycnal surfaces and trajectories of floats injected along various subtropical latitudes in a non-eddy-resolving terrain-following coordinate general circulation model (GCM) of the Atlantic. Their analysis suggested that for the South Atlantic STC, nearly all of the equatorward thermocline flow passes through the western boundary, while for the North Atlantic an interior exchange window may exist for surfaces outcropping at latitudes as far north as 228N. The ventilation of the Equatorial Undercurrent (EUC) in the northern and southern subtropics is also suggested in the model simulation of Hazeleger et al. (2003). In contrast, Harper (2000), using a high-resolution global version of the Geophysical Fluid Dynamics Laboratory (GFDL) Modular Ocean Model (MOM) GCM, found that the subducted water reaching the EUC in the Atlantic comes exclusively from the Southern Hemisphere. Particles subducted at the eastern edge of the South Atlantic subtropical gyre near 308S reached the equator both through western boundary and interior pathways. In the Northern Hemisphere, particles subducted at the eastern edge of the northern subtropical gyre were completely entrained into the subtropical gyre recirculation and did not penetrate equatorward. Differences between the model simulations may be attributed to differences in model configurations and the wind products chosen, as well as differences in how the large-scale thermohaline circulation (THC) is simulated in the models. Using idealized numerical experiments, Fratantoni et al. (2000) illustrated that the inclusion of a realistic THC in the Atlantic can reduce the supply of thermocline water to the equator from the North Atlantic and increase the supply from the South Atlantic. Experiments by Inui et al. (2002) using a reduced-gravity GCM also suggest a strong dependence of the northern communication window on the wind stress climatology. These modeling studies are limited by the fact that the target phenomena, namely the shallow subtropical overturning cells, are not well documented from observations. Moreover, the Atlantic is more complicated than the Pacific in terms of tropical–subtropical ocean pathways and transports in and above the thermocline, because this shallow overturning circulation is superimposed on the northward return branch of the thermohaline circulation. While the density range of the Atlantic EUC outcrops in both the Northern and Southern Hemisphere subtropics, most of the water in the EUC seems to be of South Atlantic origin (Metcalf and Stalcup 1967; Wilson et al. 1994). Water mass properties and direct velocity measurements along the western boundary in the South Atlantic (Schott et al. 1995; Schott et al. 1998; Stramma and Schott 1999) have helped to delineate the mean STC in the Southern Hemisphere, where about 20 Sv (Sv [ 10 6 m 3 s 21 ) of the North Brazil Current (NBC) transport above the s u 5
VOLUME 33
26.8 kg m 23 surface is known to retroflect into the EUC. However, little is known observationally about interior ocean pathways and transports in either hemisphere, or about the western boundary pathway in the North Atlantic. The focus of this manuscript will be on describing the mean pathways and volume transports in the pycnocline and surface layer for water flowing between the subtropical and tropical Atlantic Ocean. In the following, we first briefly introduce the procedures we use to derive a high resolution climatology for the hydrography of Atlantic in section 2. Section 3 presents geostrophic velocities and hydrographic properties on isopycnal surfaces in the pycnocline, which are used to characterize the STC pathways. In section 4, we quantify the strength of Atlantic STCs by deriving their transports of interior and western boundary branches. In section 5, poleward return flows of the STC are estimated from the surface drifter data and Ekman divergence. Section 6 quantitatively links equatorward convergence in the pycnocline to the surface divergence through the upwelling in the tropics. The last section provides a summary of results. 2. Data and data processing a. Hydrography climatology The primary dataset for this study is a combination of the World Ocean Database (Conkright et al. 1999) from the National Oceanographic Data Center (NODC) and new hydrographic data collected during the World Ocean Circulation Experiment (WOCE) in the 1990s available from the WOCE Hydrographic Program Office. Quality control of this merged dataset follows the procedures suggested by Curry (1996) and Lozier et al. (1995), including the elimination of duplicate data. A total of 86 131 casts with both temperature and salinity measurements reach a depth of at least 1200 m, which is the reference level we use for our geostrophic velocity estimates in the interior ocean (see section 4). The number of casts available for defining water mass properties at shallower levels is considerably greater (e.g., 166 941 casts reach at least 300 m; Fig. 1a). By decade, the number of available cast ranges between about 7000 and 30 000 reaching 1200 m and between 13 000 and 53 000 reaching 300 m, with maximum sampling taking place in the 1970s and 1980s (Fig. 1a). The seasonal distribution of data points is reasonably uniform in the tropical Atlantic, especially in the North Equatorial Countercurrent region, where seasonal variability is most significant. The number of available casts reaching 1200 m for the four seasons (December–February, March–May, June–August, and September–November) is 2308, 3676, 3636, and 3168, respectively, within 108 of the equator. Flow in the Atlantic STCs is concentrated on isopycnal surfaces that outcrop and are ventilated in the
AUGUST 2003
ZHANG ET AL.
FIG. 1. (a) Number of hydrographic cast extending deeper than 300 and 1200 m, respectively, in the Atlantic between 408S and 508N. (b) Center of 0.258 lat 3 0.258 lon bins, with T–S profiles extending deeper than 1200 m for the period 1950–2000.
subtropics. Temperature (T), salinity (S), geostrophic streamlines, and potential vorticity (N 2 f /g, where N is the buoyancy frequency, f is Coriolis parameter, and g is gravitational acceleration) are calculated on these isopycnal surfaces using the all the available hydrographic data from 1950–2000. Calculations are performed using the HydroBase analysis package (Lozier et al. 1995; Curry 1996), which implements isopycnal averaging to grid individual profiles along density surfaces. Averaging and gridding on isopycnal surfaces has the advantage in that it avoids the creation of unrealistic water properties that can occur when averaging on depth or pressure levels in strong frontal and boundary regions (Lozier et al. 1995). The scattered station profiles are first averaged into bins on a 0.258 latitude 3 0.258 longitude grid. Figure 1b shows the locations of all bins for which mean T–S profiles extend to 1200 m or deeper. The resulting bin-averaged profiles are then objectively mapped onto a 0.58 3 0.58 grid using the objective analysis of Mariano and Brown (1992), with zonal and meridional decorrelation scales set at 58 longitude 3 28 latitude. Since the property (T, S, and their derived properties, such as potential vorticity) profiles are transformed to isopycnal coordinate individually before av-
1785
FIG. 2. Drogued surface drifter deployment locations and contoured drifter days per degree of longitude and latitude.
eraging, the high-frequency vertical movements of isopycnal surfaces are taken into account when calculating geostrophic velocity and other properties on these surfaces. The mean transport determined by integrating the resulting isopycnal velocities and depths are called Lagrangian mean transport in Hazeleger et al. (2003), who further shows the critical importance of the Lagrangian mean in correctly representing the overturning circulations in their model results. b. Wind products Various wind products are used to calculate the Ekman divergence in the surface layer. Ensemble mean Ekman flow will be used in our analysis. These wind climatologies are calculated from 1) the Comprehensive Ocean–Atmosphere Data Set (COADS) for 1961–93 (da Silva et al. 1994), 2) Servain winds for 1961–99 (Servain et al. 1985), 3) the National Centers for Environmental Prediction (NCEP) reanalysis for 1958–2000 (Kalnay et al. 1996), 4) European Centre for MediumRange Weather Forecasts (ECMWF) wind analysis for 1979–2000 (see online at www.ecmwf.int), and 5) ERS1 and ERS-2 satellite winds (ERS) for 1991–2000 (see online at www.ifremer.fr/cersat). Winds were converted to stress using a drag coefficient based on a standard algorithm (Large and Pond 1981) for Servain, NCEP,
1786
JOURNAL OF PHYSICAL OCEANOGRAPHY
VOLUME 33
FIG. 3. Geostrophic velocities (cm s 21 ) and streamlines (referenced to 1200 m) on three isopycnal surfaces in the upper ocean. Winter outcrop lines (Feb and Mar in the Northern Hemisphere; Aug and Sep in the Southern Hemisphere) are shown.
and ECMWF; COADS and ERS provide stress products directly. c. Surface drifter data The positions and velocities of drogued surface drifters available in the Atlantic Ocean between 408S and 508N were obtained from Global Drifter Program (Han-
sen and Poulain 1996). The positions were averaged to daily values and the associated daily drifter velocities were then objectively mapped onto a 0.58 3 0.58 grid using decorrelation scales of 28 latitude 3 58 longitude. The deployment locations and number of drifter days are shown in Fig. 2. The uneven data coverage is not only influenced by the drifter deployment but also by the ocean currents. The number of drifter days tend to
AUGUST 2003
ZHANG ET AL.
1787
FIG. 3. (Continued )
be small in swift current areas such as the equatorial and western boundary regions. Also, surface divergence areas tend to prevent drifters from entering or staying, but convergence areas tend to keep the drifters. The choice of decorrelation scales for the objective mapping takes into account both the characteristics of tropical ocean circulation and the drifter data coverage. 3. Pycnocline flows To show isopycnal flows in detail, we plot the geostrophic streamlines as defined in Zhang and Hogg (1992) and velocity vectors along three isopycnal surfaces (Fig. 3), representing three levels in the upper pycnocline. The reference level is chosen to be 1200 m. According to direct current measurements of Stramma and Schott (1999) this level is close to the depth of minimum current velocities and is the boundary between the northward flowing Antarctic Intermediate Water (AAIW) and southward flowing North Atlantic Deep Water (NADW). Calculations using different reference levels ranging from 1000 to 1400 m are tested and show little difference in pycnocline flows. Velocity calculations stop within 28 of the equator, where geostrophy becomes a tenuous constraint and cross-streamline flow is more likely to occur. To better understand the characteristics of flow patterns off the equator and the linkage between the tropical and subtropical Atlantic, planetary potential vorticity (PV) and salinity maps on 25.4 s u are shown in Figs. 4a and 4b. Streamlines on these three surfaces (Figs. 3a–c) illustrate the interior communication windows in both the
North and South Atlantic. Also shown are the wintertime outcrop lines of these isopycnal surfaces. The bottom of the mixed layer is defined as the depth at which density is 0.125 kg m 23 greater than at the sea surface. In the Northern Hemisphere, flow reaching the equator in the interior on the shallowest density surface (24.4 kg m 23 ) originates primarily in the central Atlantic between 458–358W and in the latitude band 128–168N (Fig. 3a). Flow reaching the equator in the interior on the 25.4 kg m 23 surface originates between 188 and 228N, 328 and 228W (Fig. 3b), while flow reaching the equator on deeper surfaces originates from near 208W just off the African coast (Fig. 3c). Flow across 108N is concentrated in the westernmost part of the basin near 408– 508W and is in the equatorward direction. In the South Atlantic, flow reaching the equator in the interior on shallower surfaces originates in the central part of the basin (along 108–308W between 108–158S), while flow reaching the equator on deeper surfaces originates at higher latitudes in the eastern part of the basin, similar to the North Atlantic. On the two shallower surfaces, the ventilation pathways from the South Atlantic are spread over a wide interior window, which at 68S extends from 108W to the western boundary. At densities of 25.4 kg m 23 and deeper, a large part of the interior flow converges on the western boundary south of Cape San Roque (;68S) and presumably turns northward in the North Brazil Undercurrent (NBUC) toward the equator. Some of this interior flow could also bifurcate southward in the Brazil Current to join the South Atlantic subtropical gyre. Interior pathways are more circuitous in the Northern
1788
JOURNAL OF PHYSICAL OCEANOGRAPHY
VOLUME 33
FIG. 4. (a) Absolute value of planetary potential vorticity (10 212 m 21 s 21 ) and (b) salinity on the isopycnal surface s u 5 25.4 kg m 23 . Outcrop lines (Feb and Mar in the Northern Hemisphere; Aug and Sep in the Southern Hemisphere) are shown by blue contours.
Hemisphere because of the presence of a PV ridge extending from the eastern boundary near 158N to the western part of the basin near 108N (Fig. 4a). This high PV ridge underlies the intertropical convergence zone (ITCZ) where surface wind stress curl pumps the pycnocline up toward the surface and vertically compresses density surfaces. It tends to block the flow of waters
between the subtropical and equatorial Atlantic (McCreary and Lu 1994). In contrast, the PV field in the Southern Hemisphere is more uniform, allowing for a more direct interior pathway between the subtropics and the equator. The salinity field of Fig. 4b is characterized by lowsalinity tropical waters sandwiched between the high
AUGUST 2003
ZHANG ET AL.
1789
FIG. 5. Trajectories of subducted waters on 25.4 s u . The time interval between the adjacent triangles is 1 yr. Also shown are the two sections south of Cape San Roque, through which the subducted waters in the pycnocline flow into the western boundary to form the western boundary route of the Southern Hemisphere STC.
salinities of Northern Subtropical Underwater (NSUW) and Southern Subtropical Underwater (SSUW) (Lambert and Sturges 1977). These high-salinity waters are also referred to as ‘‘Salinity Maximum Waters’’ (Tomczak and Godfrey 1994) and are formed by subduction in the transition zone between the Tropics and subtropics where E 2 P (evaporation minus precipitation) reaches a maximum value. A salinity front, roughly along the axis of North Equatorial Current (NEC), originates near Cape Verde and extends to the South American shelf near 58N, separating the most saline subtropical water from the low-salinity tropical water. The southward and southeastward bending of the 36.2 salinity contour between 78 and 108N, 408 and 508W is consistent with the advection of northern subtropical water into the Tropics by the pycnocline flow. In the Southern Hemisphere, a similar front extends from the northern rim of the South Atlantic subtropical gyre to Cape San Roque (near 68S, 358W), roughly along the axis of South Equatorial Current (SEC). The high salinity of southern subtropical origin is seen to be advected across the equator along the North Brazil shelf via the NBC and NBUC. As shown in Fig. 3, there are two regions of strong current in the ocean interior. In the western basin close to the North Brazil shelf, the vigorous southeastward flow in the pycnocline is due to the convergence of Northern and Southern Hemisphere water carried by the NBC retroflection and the Guiana Coastal Undercurrent (Schott et al. 1998). This southeastward flow joins the
EUC west of 358W. The second region is between 48 and 88N, where the eastward flow of the North Equatorial Countercurrent/North Equatorial Undercurrent (NECC/NEUC) system extends across the entire basin on all three surfaces. The flow is deflected equatorward in both the interior and at the eastern boundary after colliding on the African coast near 108E. On the deeper density surfaces (s u 5 25.4 and 26.0 kg m 23 ) the northern part of NECC/NEUC turns northward around the eastern flank of the Guinea Dome. An interesting and well-known feature of the salinity field (Fig. 4b) is the high-salinity tongue extending eastward along the equator associated with the EUC. On its way, the high-salinity tongue draws in waters from both north and south of the equator as shown by Fig. 3. A similar subsurface convergence toward the equator was observed in direct current measurements at 1.58N, 288W and 1.58S, 288W by Weisberg et al. (1979). An indication of a weak high-salinity tongue is also seen along ;58N, associated with the NECC/NEUC, which receives water from both the NEC and the NBC retroflection, as well as the northern branch of the South Equatorial Current (nSEC) (Schott et al. 1998; Stramma and Schott 1999; Bourles et al. 1999). Lagrangian trajectories emanating from the subduction zones computed from the seasonal geostrophic velocity fields on the s u 5 25.4 kg m 23 surface are shown in Fig. 5. Although seasonal variations of the tropical currents are strong, especially the NECC in the western
1790
JOURNAL OF PHYSICAL OCEANOGRAPHY
basin, waters subducted from a limited area between 228 and 328W can reach the equator from the Northern Hemisphere through a corridor between 408 and 508W across 108N (as suggested by the mean velocity field in Fig. 3). The subducted waters that flow into the western boundary near 108N can go either northwestward into the Caribbean or southeastward to join the NBC retroflection. The pathways of these waters are not resolved in our analysis since our geostrophic calculation stops near the continental shelves shallower than 1200 m. In the Southern Hemisphere, however, subducted waters can get to the equator via interior pathways or in the western boundary layer after joining the NBUC south of Cape San Roque (Stramma and Schott 1999). The time for subducted waters on 25.4 s u surface to reach within 28 of the equator is 5–6 years in the Northern Hemisphere and 3–4 years in the Southern Hemisphere. These times are consistent with Inui et al.’s (2002) numerical simulation forced by the COADS wind climatology, which generates STCs with similar exchange windows in both hemispheres. 4. Meridional transport in the pycnocline To quantify the strength of STC interior pathways, we calculate the equatorward transport in the pycnocline across 108N and 68S. The 108N zonal section is chosen because it lies along the potential vorticity ridge, across which a portion of the NEC turns southward into the NECC in the western basin. The reason for choosing 68S is because it is at the tip of Cape San Roque, and SEC water that reaches the South American coast south of this cape flows into the western boundary current. These latitudes are also choke points for equatorward interior STC transports in a diagnostic study of wind climatologies by Huang and Wang (2001). The transport integration along 68S and 108N excludes recirculations associated with the tropical cell (Lu et al. 1998; Hazeleger et al. 2003), which is a shallow overturning circulation formed by a part of the equatorial upwelling and off-equatorial downwelling confined within 58 latitude of the equator. To determine the total meridional transport we must add an Ekman contribution to the geostrophic flow. Consistent with observational studies on Ekman layer transports and mixed layer depth (e.g., Davis et al. 1981; Price et al. 1987; Chereskin and Roemmich 1991; Wijffels et al. 1994), we assume an Ekman transport distribution with 3/4 of the transport in the mixed layer and 1/4 of the transport extending into the thermocline to a distance equal to one-half of the mixed layer depth. As noted previously, we define the mixed layer depth as the depth where the density reaches a value 0.125 kg m 23 greater than the surface density, which is consistent with several recent studies (e.g., Qiu and Huang 1995). The Ekman transports are calculated from the ensemble mean wind climatology. Figures 6a and 6b illustrate the total volume transport
VOLUME 33
(Ekman 1 geostrophic) distribution across 68S and 108N in isopycnal layers, excluding the mixed layer. The interior equatorward transport in the pycnocline is between 23.2 and 26.0 s u at 108N and between 23.6 and 26.2 s u at 68S. In the remainder of this paper we will refer to this layer of equatorward flow as the ‘‘pycnocline layer.’’ The poleward transport in the overlying layers is due to the stronger poleward Ekman transport than the equatorward geostrophic transport, and we will henceforth refer to this as the ‘‘surface layer.’’ At 108N, the poleward transport below the pycnocline layer, between about 26.0 and 27.4 s u , corresponds to the interior pathway of the THC in the South Atlantic Central Water (SACW) layers (Arhan et al. 1998; Stramma and Schott 1999; Blanke et al. 1999). At 68S, the southward transport between 26.4 and 26.8 s u is the result of the strong interior recirculation of SACW (recirculating southward from the NBUC to SEUC) which is also seen in the Blanke et al. (1999) simulation and shown in Stramma and Schott’s (1999) circulation schematic. Note that our transport integration across the ocean interior at 68S stops at 338W and excludes the northward NBUC in the western boundary layer. The southward interior flow in the layers between 26.4 and 26.8 s u clearly separates the northward pycnocline flow of the STC and the northward flow of AAIW below 27.0 s u (Fig. 6b). The 26.2 s u surface is also the deepest on which water can be subducted along the path of the SEC and Benguela Current off the coast of South Africa. Figures 6c and 6d show the cumulative transport across 108N and 68S in the layers with net equatorward interior flow in the pycnocline (Figs. 6a,b), integrated westward from the eastern boundary. At 108N, the strongest equatorward flow occurs near 408W and between 518 and 558W. These two regions are separated by a small cyclonic recirculation cell (Fig. 3) located between 448 and 518W. At 68S, the cumulative equatorward transport increases steadily from 158W toward the western boundary. Excluding the recirculation in the western boundary between 558 and 628W (Figs. 3 and 6c), the net transport across 108N between the eastern boundary and 558W, defined as the Northern Hemisphere STC interior transport, is 2 6 0.7 Sv. The net transport across 68S between the eastern boundary and 338W (28 longitude off the Cape San Roque) is defined as the southern STC interior transport and is estimated to be 4 6 0.5 Sv. To derive error bars, we smoothed the cumulative transport curves in Figs. 6c and 6d using spline fitting and define the errors as the largest deviations across the basin between the smoothed and the original cumulative transport curves from the eastern boundary. The logic behind this method is that these deviations can be caused by the aliasing of incompletely resolved high-frequency and small-spatial-scale oceanic processes, including seasonal variability and mesoscale eddies in shipboard data. We also performed Monte Carlo simulations to estimate uncertainties by calculating geostrophic transports using
AUGUST 2003
ZHANG ET AL.
1791
FIG. 6. Net meridional volume transport (Sv) excluding the mixed layer in isopycnal layers from the African coast to (a) 608W across 108N and to (b) 348W across 68S; Meridional volume transport (Sv) in the pycnocline layers with net equatorward transport (23.4–26.0 s u at 108N and 23.6–26.4 s u at 68S) zonally accumulated from the African coast westward along (c) 108N and (d) 68S.
subsets of the dataset near the eastern and western limits of the basin. These calculations resulted in essentially the same error bars. Choosing alternative reference depths (for example, 1000 and 1400 m) leads to relatively small differences in pycnocline transport values (0.3–0.6 Sv) in the two hemispheres. Unlike in the interior Atlantic, we are unable to estimate the pycnocline transport in the western boundary region due to hydrographic data limitations and uncertainties in the reference level. At 108N, however, six ADCP surveys have suggested a southeastward western boundary current in the pycnocline with a mean transport of about 3.3 6 1.0 Sv (Wilson et al. 1994; Bourles et al. 1999; Schott et al. 1998). This southeastward boundary current is only evident between 100 and 300 m and is referred to as the Guiana Coastal Undercurrent (GCU). Mayer et al. (1998) linked the GCU to the western boundary current of the wind driven tropical gyre. Whether an average from six sections is representative of an annual mean transport is questionable, due to high eddy activity in this region. In the Southern Hemisphere, the STC pathways are more uniform and direct than their Northern Hemisphere counterparts. At 68S, the communication window is much wider, extending from 108W to the western boundary. It
is evident that streamlines in the SEC system flow into the western boundary (Figs. 3, 5), where it bifurcates into the NBC/NBUC and Brazil Current. Assuming a bifurcation point at 128S (Mayer et al. 1998; Mayer and Weisberg 1993), we can derive the Southern Hemisphere western boundary STC transport by integrating pycnocline flows into the boundary above 26.2 su south of Cape San Roque, through a meridional section along 338W between 68 and 128S and through a zonal section 128S between 368 and 338W (Fig. 5). The inferred western boundary STC transport is 6.0 6 1.2 Sv in the South Atlantic at 68S. The error estimate takes into account aliasing of high-frequency short-space-scale variability and the uncertainty of the SEC bifurcation (which was tested using alternative bifurcation latitudes between 108 to 148S in the calculation of northward transport in the western boundary pycnocline). Our estimate is consistent with a 5 Sv northward flow between 23.4 to 26.2 su in the NBUC, as estimated from Schott et al.’s (1998) mean ADCP section at 58S (their Fig. 15). 5. Surface layer transports We can compute Ekman divergence using either surface wind data or surface velocities measured by surface
1792
JOURNAL OF PHYSICAL OCEANOGRAPHY
VOLUME 33
FIG. 7. Mean surface layer velocity field (cm s 21 ) derived from surface drifter data using the objective mapping method of Mariano and Brown (1992). Velocities with relatively large mapping error are plotted out as light vectors in the data void areas.
drifters to examine the surface flow in the surface limb of the STC. Figure 7 shows the gridded mean velocities derived from the objective mapping of drifter data drogued at 15 m. Swift currents are visible in the Tropics with maximum mean speeds of more than 30 cm s 21 . These strong currents include the eastward NECC along about 58N, the westward SEC along the equator, and the northwestward NBC along the north Brazil coast. Meridional poleward Ekman divergence in the interior ocean is also evident, particularly near the equator between 358W and 58E. The divergence field calculated from our gridded velocity field is shown in Fig. 8 (smoothed by a Hanning filter with a window of 38 in latitude and 128 in longitude). Positive divergence implies upwelling and negative divergence (i.e., convergence) implies downwelling. Besides well-defined regions of equatorial upwelling and coastal upwelling off Africa, upwelling also occurs within the NBC and its retroflection and in the NECC, consistent with a recent modeling study of Blanke et al. (1999). Upwelling is largest near the equator and shifted slightly south along 18S in the eastern basin. Downwelling zones along 28N and 48S correspond to the poleward limits of the tropical cell. Another strong downwelling zone can be seen along about 88N between the NEC and NECC, just south of the upwelling region that extends westward from North Africa at 108–128N. To understand how the upwelled water is expelled out of the Tropics, we compute the poleward transports (V i )
across 108N and 68S using the surface drifter velocity: V i 5 # L y Dmix dx. Here, y is the drifter velocity, Dmix is the depth of momentum mixed layer, and L is the width of the basin at each latitude. Based on the section-averaged velocity structure from cross-basin shipboard ADCP surveys in the tropical ocean (Chereskin and Roemmich 1991; Wijffels et al. 1994; Garzoli and Molinari 2001), the momentum mixed layer is taken to be 40 m in the tropical Atlantic. The errors are estimated from piecewise averages of the surface current measurements in longitudinal bands, 618 latitude. Specifically, the interior sections along 108N and 68S are divided into 9 and 10 segments, respectively, of 58 longitude width, which is the decorrelation scale we used in the objective mapping. The western boundary current region is defined as from 578W to the western boundary at 108N and from 338W to the coast of Brazil at 68S. The standard errors of velocity in these boxes are multiplied by the width of segment and momentum mixed layer depth to produce the standard errors of meridional transport, along 108N and 68S every 58 longitude. The accumulation of these transport errors (represented by square root of the sum of squared standard errors) across the basin along 108N and 68S are defined as the total error for the interior surface current transport at the two sections. As in Fratantoni (2001), the number of independent measurements, that is, the degrees of freedom for the standard error calculation, in a given box was determined by the number of different drifters and in-
AUGUST 2003
ZHANG ET AL.
1793
FIG. 8. Mean surface divergence (10 27 s 21) calculated from the velocity field in Fig. 7. Solid contours correspond to upwelling (positive divergence); dashed contours to downwelling (convergence). Hatched areas outline the regions where the divergence calculation has large uncertainties.
dependent samples per drifter (in terms of the number of Lagrangian integral timescales taken to be 10 days) in the box. The surface transport through 108N derived from the drifters is 11.0 6 1.5 Sv northward, with 4.0 6 0.8 Sv going north in the vicinity of the western boundary between the coast and 578W, and 7.0 6 1.3 Sv flowing northward in the interior. A total of 9.8 6 1.4 Sv flows southward across 68S, with a surface transport in the western boundary of 0.5 6 0.4 Sv northward and an interior transport of 10.3 6 1.4 Sv southward. The increase of northward western boundary transport in the surface layer from 68S to 108N is consistent with Schott et al. (1998), who showed the subsurface intensified NBUC becomes a surface intensified NBC north of the equator. The 4 Sv of surface western boundary transport across 108N is consistent with the model results of Fratantoni et al. (2000) and is related in part to transport by North Brazil Current Rings and coastal currents that form a northward continuation of the surface NBC (Johns et al. 1998). In fact, many of the surface drifters show looping trajectories near the western boundary that are indicative of North Brazil Current rings. The total poleward surface transport, that is, the surface divergence, across the two zonal sections 108N and 68S is 20.8 6 2.1 Sv. Another way to estimate the poleward surface transport in the interior is to calculate the Ekman transport from surface winds and the surface geostrophic transport
from hydrographic data, and add them together. The surface layer geostrophic transport estimated from the hydrographic data in the interior ocean is 0.5 6 0.1 Sv southward across 108N and 1.8 6 0.1 Sv northward across 68S. The ensemble mean Ekman transports calculated from various wind products, across 108N and 68S in the interior (excluding the boundary region west of 578W at 108N and west of 338W at 68S) are 8.1 6 0.4 Sv northward and 11.9 6 0.6 Sv southward, respectively. The total surface transport in the interior ocean is therefore poleward across both sections, 7.6 Sv northward at 108N and 10.1 Sv southward at 68S. These transports are consistent with the respective values of 7.0 6 1.3 Sv and 10.3 6 1.4 Sv derived from the surface drifter data. 6. Discussion In section 4 the net equatorward convergence of pycnocline transport across 108N and 68S associated with the STC is estimated to be 15 Sv, with 5 Sv (2 interior and ;3 western boundary) from the subduction in the Northern Hemisphere and 10 Sv (4 Sv interior and 6 Sv western boundary) from the Southern Hemisphere. This is 6 Sv smaller than the 21 Sv poleward surface divergence. According to Roemmich (1983), 6 Sv of the northward THC return flow is converted from SACW to surface water through upwelling in the tropical Atlantic.
1794
JOURNAL OF PHYSICAL OCEANOGRAPHY
VOLUME 33
FIG. 9. Schematic of the pathways and volume transports (Sv) of equatorward pycnocline flow. The western boundary transport in the South Atlantic includes 6 Sv of warm return flow of the THC. The dark shaded areas show upwelling regions in the tropical Atlantic. The light shaded areas mark the region where subducted salinity maximum waters can potentially reach the equator. The STC pathways that are not well determined in this study are displayed as dashed lines. Latitudes 108N and 68S are marked by solid lines and the equator is shown by the long dashed line.
After complicated recirculation through zonal tropical currents, this water is then expelled northward from the tropics by poleward Ekman flow in the ocean interior to join the westward NEC (Schmitz and McCartney 1993). This 6 Sv of upwelling associated with the THC occurs through the 26.2 s u surface (Roemmich 1983), which defines the bottom of the STC subsurface limb and the top of the SACW. These waters can be traced back into the western Indian Ocean (Sprintall and Tomczak 1993; Tomczak and Godfrey 1994), and are brought into the Atlantic by the Agulhas Current and Agulhas Current eddies and carried northward by the Benguela Current and SEC to join the NBC/NBUC south of 108S (Stramma and Schott 1999). The upwelling of these waters from intermediate depths is in addition to upwelling associated with converging pycnocline transports in the STCs. A schematic of the inferred transports and pathways of pycnocline water that converges on the equator and upwells into the surface layer is given in Fig. 9. Interior flow in the Northern Hemisphere STC takes a zigzag pathway towards the equator, sweeping around the potential vorticity barrier under the ITCZ. Pycnocline waters of the Southern Hemisphere STC flow more directly towards the equator. Also included in the western boundary transport at 68S is another 6 Sv related to the THC return flow, which upwells through the base of pycnocline and into the surface layer together with the pycnocline convergence from STCs. The 3-Sv transport
of the Northern STC in the western boundary is based on limited number (6) of ADCP sections summarised in Bourles et al. (1999). However, with the other branches of STC determined by our geostrophic calculation and the 6 Sv related to THC return flow from Roemmich (1983), 3 Sv is exactly the transport needed to have 21 Sv subsurface convergence to balance the divergence in the surface layer of tropical Atlantic, suggesting it is a reasonable estimate. The areas where the subducted water could potentially reach equatorial latitudes are indicated by light shading in Fig. 9. They are determined by the intersections of the streamlines of pycnocline flow that enter the equatorial communication windows with the winter outcrops of these density surfaces (e.g., as shown in Fig. 3). These areas are coincident with regions of high subduction rate (25–50 m yr 21 ) estimated in Qiu and Huang (1995) for the Northern Hemisphere based on Levitus climatology (Levitus 1982) and in Lazar et al. (2002) in their numerical simulation for the Southern Hemisphere. Upwelling regions, indicated in Fig. 9 by dark shading, are based on total transport divergence in the surface layer from our drifter analysis. They are, however, also generally coincident with the major convergence areas in the pycnocline (Fig. 3) and agree well with numerical modeling results (Blanke et al. 1999). A depiction of the zonally integrated mass balance in the upper tropical Atlantic based on our results is given in Fig. 10. The dark pathways are the Northern
AUGUST 2003
ZHANG ET AL.
FIG. 10. Schematic of the zonally integrated circulation (Sv) in the upper ocean of tropical Atlantic. Note the upwelling is not concentrated on the equator, but is spread out in the Tropics with more upwelling to the north of the equator (see text for details). The dark routes represent flows in the Northern and Southern Hemisphere STCs. The light routes represent the warm return flow of the THC that upwells into the surface layer.
and Southern Hemisphere STCs, which are self-balanced with the same amount of water diverging poleward in the surface layers as is flowing equatorward in the thermocline. The light pathway corresponds to that part of the THC warm return flow that enters the Tropics from the South Atlantic, upwells through the thermocline into the surface layer, and flows into the Northern Hemisphere. This flow balances part of the cold southward flowing Deep Western Boundary Current in the deep ocean. Of the 21 Sv upwelling, 15 Sv is associated with the STCs with 5 Sv from the Northern Hemisphere and 10 Sv from the Southern Hemisphere, through both the interior and western boundary pathways. The remaining 6 Sv is associated with the THC. 7. Summary The focus of this study has been on describing the pathways of the STC and quantifying the volume fluxes associated with these pathways between the subtropics and the Tropics. To better resolve flows in the pycnocline and the complex equatorial current system, a highresolution climatology of hydrographic measurements was developed in the Atlantic between 408S and 508N, based on all the available temperature and salinity measurements in the past 50 years. We have also made use of surface drifter data and surface wind stress products to infer poleward surface transports and divergences. Salinity and pycnocline flow maps suggest that the EUC is fed by salinity maximum waters originating in the subtropical gyres. The pathways of the subducted water in the Southern Hemisphere to the EUC are relatively direct and include a western boundary route through the NBUC/NBC and NBC retroflection (e.g., Schott et al. 1988), and an interior route across 68S between the western boundary and 108W (Figs. 3 and 9). Serpentine pathways towards the equator in the Northern Hemisphere are suggested by potential vorticity, salinity, the geostrophic current fields, and the trajectories of subducted waters forced by seasonal cycle of pycnocline flows. The high-salinity water of NSUW
1795
flows westward in the NEC, skirting the region of high potential vorticity under the ITCZ, then flows southward and southeastward into the NECC. Part of these waters, along with South Atlantic STC waters retroflected from the NBC, may upwell along the path of the NECC in the upwelling regions under the ITCZ and near the African coast. Some of these waters may also become entrained into the westward nSEC and then be drawn into the EUC, consistent with numerical modeling results (Malanotte-Rizzoli et al. 2000; Inui et al. 2002). Geostrophic transport calculations using all the available hydrographic data suggest that the pycnocline flow in the interior Northern Hemisphere STC is 2.0 6 0.7 Sv southward across the high potential vorticity ridge along 108N. Previous studies have suggested a southward thermocline flow in the western boundary near 108N (Wilson et al. 1994; Bourles et al. 1999) of 3 Sv on average, which forms the Northern Hemisphere STC western boundary route. Flow in the Southern Hemisphere STC consists of 4.0 6 0.5 Sv in the interior and 6 Sv in the western boundary flowing northward across 68S in the pycnocline. The equatorward pycnocline flows of the Northern and Southern Hemisphere STCs upwell into the surface layer and eventually depart the tropical Atlantic via poleward divergence. The equatorward converging pycnocline transports, together with the 6 Sv of upwelling in the tropical Atlantic associated with the warm return flow of the THC according to Roemmich (1983), yield a total of 21 Sv of upwelling into the surface layer between 68S and 108N (Figs. 9 and 10). This upwelling rate is in excellent agreement with the surface divergence (21 6 2.1 Sv) derived from the surface velocity field of available drifter data. The export of upwelled water from the tropical Atlantic surface layer is accomplished through 11 Sv northward flow (4 Sv in the western boundary and 7 Sv in the interior) across 108N and about 10 Sv net southward flow across 68S (0.5 Sv northward in the surface layer of western boundary and 10.3 Sv southward in the interior). Besides the well-identified upwelling regions in the tropical Atlantic (i.e., the equatorial upwelling and upwelling near the African coasts), the divergence of surface drifter velocities also suggests upwelling within the NBC, the NBC retroflection, and the NECC, which are also simulated by the numerical model of Blanke et al. (1999). Upwelling in the tropical Atlantic has major effects on the upper ocean heat budget and on the interhemispheric transport of heat by the ocean. In addition to its importance in the mean heat budget of the tropical Atlantic, the spatial and temporal variations in the upwelling may have implications for tropical Atlantic decadal variability (TAV)—the so-called dipole mode or cross-equator SST gradient mode (Nobre and Shukla 1996). The results of our study suggest that as much as 70%, or 15 Sv, of the upwelling into the surface layer is associated with the STCs. The timescales for waters subducted in the subtropics to reach
1796
JOURNAL OF PHYSICAL OCEANOGRAPHY
the upwelling regions at lower latitudes are about 3–6 yr, indicating a potential role for the STCs in modulating the surface layer heat balance on these timescales. Acknowledgments. We thank Greg Johnson of NOAA/PMEL for suggestions on the processing of surface drifter data. The first two authors acknowledge support of NOAA’s Office of Global Program’s Atlantic CLIVAR Program. The third author acknowledges support from the National Science Foundation under grants OCE-9730322 and OCE-9811531. This publication is partially supported by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA Cooperate Agreement NA17RJ1232. REFERENCES Arhan, M., H. Mercier, B. Bourles, and Y. Gouriou, 1998: Hydrographic sections across the Atlantic at 7830N and 4830S. DeepSea Res., 45, 829–872. Blanke, B., M. Arhan, G. Madec, and S. Roche, 1999: Warm water paths in the equatorial Atlantic as diagnosed with a general circulation model. J. Phys. Oceanogr., 29, 2753–2768. Bourles, B., R. L. Molinari, E. Johns, W. D. Wilson, and K. D. Leaman, 1999: Upper layer currents in the western tropical North Atlantic (1989–1991). J. Geophys. Res., 104, 1361–1375. Chereskin, T. K., and D. Roemmich, 1991: A comparison of measured and wind-derived Ekman transport at 118N in the Atlantic Ocean. J. Phys. Oceanogr., 21, 869–878. Conkright, M. E., and Coauthors, 1999: World Ocean Database 1998 Version 2.0. National Oceanographic Data Center Internal Rep. 14, 118 pp. Curry, R. G., 1996: Hydrobase—A database of hydrographic stations and tools for climatological analysis. Woods Hole Oceanographic Institution Tech. Rep. WHOI-96-01, 44 pp. da Silva, A. M., C. C. Young, and S. Levitus, 1994: Algorithms and Procedures. Vol. 1, Atlas of Surface Marine Data, NOAA Atlas NESDIS 6, U.S. Department of Commerce, 83 pp. Davis, R. E., R. deSzoeke, and P. P. Niiler, 1981: Variability in the upper ocean during MILE. Part II: Modeling the mixed layer response. Deep-Sea Res., 28A, 1453–1475. Fratantoni, D. M., 2001: North Atlantic surface circulation during the 1990’s observed with satellite-tracked drifters. J. Geophys. Res., 106, 22 067–22 093. ——, W. E. Johns, T. L. Townsend, and H. E. Hurlburt, 2000: Low latitude circulation and mass transport pathways in a model of the tropical Atlantic Ocean. J. Phys. Oceanogr., 30, 1944–1966. Garzoli, S. L., and R. L. Molinari, 2001: Ageostrophic transport in the upper layers of the tropical Atlantic Ocean. Geophys. Res. Lett., 28, 4619–4622. Gu, D.-F., and S. G. H. Philander, 1997: Interdecadal climate fluctuations that depend on exchanges between the tropics and extratropics. Science, 275, 805–807. Hansen, D. V., and P. M. Poulain, 1996: Quality control and interpolations of WOCE/TOGA drifter data. J. Atmos. Oceanic Technol., 13, 900–909. Harper, S., 2000: Thermocline ventilation and pathways of tropical– subtropical water mass exchange. Tellus, 52A, 330–345. Hazeleger, W., M. Visbeck, M. Cane, A. Karspeck, and N. Naik, 2001: Decadal upper ocean variability in the tropical Pacific. J. Geophys. Res, 106, 8971–8988. ——, P. Vries, and Y. Friocourt, 2003: Sources of the Equatorial Undercurrent in the Atlantic in a high-resolution ocean model. J. Phys. Oceanogr., 33, 677–693. Huang, R. X., and Q. Wang, 2001: Interior communication from the subtropical to the tropical oceans. J. Phys. Oceanogr., 31, 3538– 3550.
VOLUME 33
Inui, T., A. Lazar, P. Malanotte-Rizzoli, and A. Busalacchi, 2002: Wind stress effects on subsurface pathways from the subtropical to tropical Atlantic. J. Phys. Oceanogr., 32, 2257–2276. Jochum, M., and P. Malanotte-Rizzoli, 2001: Influence of the meridional overturning circulation on tropical–subtropical pathways. J. Phys. Oceanogr., 31, 1313–1323. Johns, W. E., T. N. Lee, R. Beardsley, J. Candela, and B. Castro, 1998: Annual cycle and variability of the North Brazil Current. J. Phys. Oceanogr., 28, 103–128. Johnson, G. C., and M. J. McPhaden, 1999: Interior pycnocline flow from the subtropical to the equatorial Pacific Ocean. J. Phys. Oceanogr., 29, 3073–3089. Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471. Kleeman, R., J. P. McCreary, and B. A. Kinger, 1999: A mechanism for generating ENSO decadal variability. Geophys. Res. Lett., 26, 1743–1746. Lambert, R. B., and W. Sturges, 1977: A thermohaline staircase and vertical mixing in the thermocline. Deep-Sea Res., 24, 211–222. Large, W. G., and S. Pond, 1981: Open ocean momentum flux measurements in moderate to strong winds. J. Phys. Oceanogr., 11, 324–336. Lazar, A., T. Inui, P. Malanotte-Rizzoli, A. J. Busalacchi, L. Wang, and R. Murtugudde, 2002: Seasonality of the ventilation of the tropical Atlantic thermocline in an ocean general circulation model. J. Geophys. Res., 107, 3104, doi:10.1029/2000JC000667. Levitus, S., 1982: Climatological Atlas of the World Ocean. NOAA Prof. Paper 13, 173 pp. and 17 microfiche. Liu, Z., S. G. H. Philander, and R. C. Pakanowski, 1994: A GCM study of tropical–subtropical upper-ocean water exchange. J. Phys. Oceanogr., 24, 2606–2623. Lozier, S. M., W. B. Owens, and R. G. Curry, 1995: The climatology of the North Atlantic. Progress in Oceanography, Vol. 36, Pergamon, 1–44. Lu, P., J. P. McCreary, and B. A. Klinger, 1998: Meridional circulation cells and the source waters of the Pacific Equatorial Undercurrent. J. Phys. Oceanogr., 28, 62–84. Malanotte-Rizzoli, P., K. Hedstrom, H. Arango, and D. Haidvogel, 2000: Water mass pathways between the subtropical and tropical ocean in a climatological simulation of the North Atlantic ocean circulation. Dyn. Atmos. Oceans., 32, 331–371. Mariano, A. J., and O. B. Brown, 1992: Efficient objective analysis of dynamical heterogeneous and nonstationary fields via the parameter matrix. Deep-Sea Res., 39, 1255–1271. Mayer, D. A., and R. H. Weisberg, 1993: A description of COADS surface meteorological fields and the implied Sverdrup transports for the Atlantic Ocean from 308S to 608N. J. Phys. Oceanogr., 23, 2201–2221. ——, R. L. Molinari, and J. F. Festa, 1998: The mean and annual cycle of upper layer temperature fields in relation to Sverdrup dynamics within the gyres of the Atlantic Ocean. J. Geophys. Res., 103, 18 545–18 566. McCreary, J. P., and P. Lu, 1994: On the interaction between the subtropical and equatorial ocean circulation: Subtropical cell. J. Phys. Oceanogr., 24, 466–497. McPhaden, M. J., and D. Zhang, 2002: Slowdown of the meridional overturning circulation in the upper Pacific Ocean. Nature, 415, 603–608. Metcalf, W. G., and M. Stalcup, 1967: Origins of the Atlantic Equatorial Undercurrent. J. Geophys. Res., 72, 4959–4975. Nobre, P., and J. Shukla, 1996: Variations of sea surface temperature, wind stress, and rainfall over the tropical Atlantic and South America. J. Climate, 9, 2464–2479. Price, J. F., R. A. Weller, and R. R. Schudlich, 1987: Wind-driven ocean currents and Ekman transport. Science, 238, 1534–1538. Qiu, B., and R. X. Huang, 1995: Ventilation of the North Atlantic and North Pacific: Subduction versus obduction. J. Phys. Oceanogr., 25, 2374–2390. Roemmich, D., 1983: The balance of geostrophic and Ekman trans-
AUGUST 2003
ZHANG ET AL.
port in the tropical Atlantic Ocean. J. Phys. Oceanogr., 13, 1534–1539. Schmitz, W. J., and M. S. McCartney, 1993: On the North Atlantic circulation. Rev. Geophys., 31, 29–49. Schneider, N., A. J. Miller, A. Alexander, and C. Deser, 1999: Subduction of decadal North Pacific temperature anomalies: Observations and dynamics. J. Phys. Oceanogr., 29, 1056–1070. Schott, F. A., L. Stramma, and J. Fischer, 1995: The warm water flow into the western tropical Atlantic boundary regime, spring 1994. J. Geophys. Res., 100, 24 745–24 760. ——, J. Fischer, and L. Stramma, 1998: Transports and pathways of the upper-layer circulation in the western tropical Atlantic. J. Phys. Oceanogr., 28, 1904–1928. Servain, J., J. Picaut, and A. J. Busalacchi, 1985: Interannual and seasonal variability of the tropical Atlantic Ocean depicted by sixteen years of sea surface temperature and wind stress. Coupled Ocean–Atmosphere Models, J. C. J. Nihoul, Ed., Elsevier, 211–237.
1797
Sprintal, J., and M. Tomczak, 1993: On the formation of Central Water in the southern hemisphere. Deep-Sea Res., 40, 827–848. Stramma, L., and F. A. Schott, 1999: The mean flow field of the tropical Atlantic Ocean. Deep-Sea Res., 46, 279–303. Tomczak, M., and J. S. Godfrey, 1994: Regional Oceanography: An Introduction. Pergamon Press, 422 pp. Weisberg, R. H., L. Miller, A. Horigan, and J. A. Knauss, 1979: Velocity observations in the equatorial thermocline during GATE. Deep-Sea Res., 26, 217–248. Wijffels, S., F. Firing, and H. Bryden, 1994: Direct observations of the Ekman balance at 108N in the Pacific. J. Phys. Oceanogr., 24, 1666–1679. Wilson, W. D., E. Johns, and R. L. Molinari, 1994: Upper layer circulation in the western tropical North Atlantic ocean during August, 1989. J. Geophys. Res., 99, 22 513–22 523. Zhang, H., and N. G. Hogg, 1992: Circulation and water mass balance in the Brazil Basin. J. Mar. Res., 50, 385–420.
