1. Introduction
Recent studies about global warming or multidecadal variability in the climate system indicate a poleward shift or long-term oscillation of the westerly wind over the Southern Ocean (e.g., Cai et al. 2003). The responses in the ocean to changes in wind stress over the Southern Ocean were investigated in previous studies (e.g., Oke and England 2004). In the latitude band containing the Drake Passage, meridional geostrophic flow zonally integrated above the sill depth must be zero. The northward Ekman drift driven by westerly wind stress over the Southern Ocean must be compensated by the southward flow primarily below the sill depth as geostrophically balanced flow. These constraints make the winds over the Southern Ocean play important roles in the global ocean circulation, which contributes to the interhemispheric heat transport maintaining the current climate. In their numerical experiments, Toggweiler and Samuels (1995) found that the increased westerly wind stresses over the Southern Ocean enhance the remote northern sinking meridional overturning in the North Atlantic (NA cell), and they named the phenomenon the Drake Passage effect. Here, the term “Drake Passage effect” is loosely defined as the correlation between the Southern Ocean westerly winds and the NA-cell strength.
Toggweiler and Samuels (1995) proposed a mechanical explanation of the Drake Passage effect; that is, the northward Ekman transport at the latitude of Cape Horn controls the North Atlantic Deep Water (NADW) production rate in the constraints of the zonally connected channel. They related the NADW outflow into the Southern Ocean with the Ekman pumping in the circumpolar belt using their experiment results. In the context of the mechanical Drake Passage effect, Nof (2000) demonstrated that the meridional overturning transport is given simply by the wind stresses in those cases without explicit diabatic forcing. In contrast, Rahmstorf and England (1997) argued that thermohaline forcing controls the NA cell, and the robustness of the wind effect reported by Toggweiler and Samuels (1995) is due to the use of a thermal-restoring boundary condition without temperature feedback to the atmosphere.
Tsujino and Suginohara (1999), using closed basin model experiments, determined that the wind forcing imposed on one hemisphere enhances the meridional overturning circulation sinking at the end of the other hemisphere. They noticed that the enhanced surface heating in the cyclonic wind-driven gyre in the former hemisphere is balanced with enhanced surface cooling in the deep-water formation region in the latter hemisphere. Remote wind-driven overturning in a closed basin was also examined by Klinger et al. (2004). Klinger et al. (2003) demonstrated that winds outside the circumpolar channel significantly influence the NA cell. They tried to combine the anomalous Ekman pumping with the anomalous cross-equatorial flow followed by the anomalous northward heat transport, while Nof and Boer (2004) called attention to the necessity of distinguishing the interhemispheric flow from the northward Ekman drift. Using a realistic ocean model, Hasumi and Suginohara (1999) demonstrated that the increase in buoyancy loss in the North Atlantic is balanced against the increase in buoyancy gain in the Southern Ocean when the wind stresses were changed over the Southern Ocean.
Some of these studies have demonstrated that the deep-water formation rate in the Northern Hemisphere can be increased by the wind forcing over the Southern Ocean, through change in the thermal state of the ocean rather than mechanical increase in the deep southward flow at the latitude of the Drake Passage. However, most of these previous studies argued the Drake Passage effect based on results from numerical experiments using thermal-restoring surface boundary conditions, which complicated the estimation of the role of buoyancy fluxes in the Drake Passage effect. Although Rahmstorf and England (1997) used an ocean model with a simple atmospheric feedback model, they did not confirm explicit buoyancy fluxes.
Although the present study uses density-restoring boundary conditions in an idealized ocean model, the relations between the NA cell and the buoyancy fluxes are examined. Furthermore, the capacity of wind stresses over the Southern Ocean to modify the density field in the remote North Atlantic and the NA cell is discussed, considering their temporal evolution. The processes through which the Southern Ocean wind affects the remote overturning were not clear in previous studies, and this report clarifies the problem. The validity of the mechanical explanation of the Drake Passage effect is examined, the probable thermodynamic processes are defined, and an attempt is made to identify the dominant factor in the effect.
The next section describes the idealized model configurations, boundary conditions and experiment methods. Section 3 presents the results. Section 3a investigates the response of the NA cell in a single-basin model forced by the Atlantic reference density profiles to the change in the Southern Ocean wind stress and the response to that change by deep outflow into the Southern Ocean in a twin-basin model. Section 3b attempts to discern how the increased wind stress over the Southern Ocean can affect the remote NA-cell strength. To deal with the limitation of these idealized experiments, some results from realistic experiments are presented in section 4. Sections 5 and 6 are devoted to discussion and summary.
2. Numerical experiments with idealized configurations
A primitive equation model is used that employs three prognostic variables: potential density, zonal, and meridional velocities. The computational loads are reduced using only potential density as an active tracer, not temperature and salinity. In situ density is obtained as a function of potential density and pressure. The hydrostatic, Boussinesq, and rigid-lid approximations are assumed. The complete convective adjustment is applied for statically unstable conditions. The weighted upcurrent scheme (horizontal, upcurrent 0.7 plus centered 0.3; vertical, upcurrent 0.2 plus centered 0.8) is used to calculate the advection terms of density. The generalized Arakawa scheme (Ishizaki and Motoi 1999) is used to calculate the momentum advection terms.
Figure 1 illustrates the topography of the model ocean. Idealized single-basin and twin-basin models with a periodically connected passage are used. The west and east basins in the twin-basin configuration are assumed to be a simplified realization of the Atlantic and the Pacific Oceans. The twin-basin configuration is necessary to produce the interbasin deep circulation that is a large component of the abyssal circulation. The model ocean ranges from 60°S to 60°N in the meridional direction. The single-basin model is 60° wide in the zonal direction, and a passage periodically connects the basin at the latitude band from 55° to 45°S. The depth of the ocean is 5125 m, except for the passage region that is 2875 m deep. The twin-basin experiment with longitudinal width 120° is separated into two basins by a meridional wall with 2° width to the north of 35°S. The depth of the flat bottom and the periodic passage are the same as in the single-basin model.
In the single-basin experiments, the model ocean is initially at rest with vertically stable and horizontally homogeneous densities ranging from 26 at the top to 27.15 at the bottom. The model was run until a quasi-steady state was achieved, with the Tpac1 through Tatlslw forcings and W0 through W3 wind stresses. These solutions are designated Tpac1W0, TatlslwW3, etc. The Bryan (1984) accelerating method is used in these integrations. Thus, there are solutions for 20 cases: 5(Tpac1, Tpac2, Tmid, Tatl, and Tatlslw) × 4(W0–W3). The primary focus is on the Tatl results.
In the twin-basin experiments, the west and the east basin each have a different reference density profile. In the (Tatl+Tpac1) experiment, Tatl forcing is applied to the west basin and Tpac1 forcing to the east basin. The Tatl forcing to the west and the Tpac2 forcing to the east are referred to as (Tatl+Tpac2). These buoyancy forcings are combined with four wind forcings: W0, W1, W2, and W3. Each integration is done until a quasi-steady state is achieved. Thus, there are eight cases for the twin-basin experiments: 2[(Tatl+Tpac1) and (Tatl+Tpac2)] × 4[W0–W3]. The most realistic of these experiments is (Tatl+Tpac2)W3.
3. Results
a. NA cell and the buoyancy fluxes
1) The single-basin experiments restored to the Atlantic-type densities
To estimate the NA-cell strength, the meridional overturning streamfunction is calculated from the zonally integrated meridional mass transport. The streamfunctions in TatlW0 and TatlW2 are given in Fig. 3. The NA cell in TatlW2 is stronger than that in TatlW0 by about 4 Sv (Sv ≡ 106 m3 s−1). Most of the meridional Ekman transport (hereinafter ET) driven by the wind stress at 45°S must be compensated by the net southward transport below the sill depth of the passage because of the constraint of the circumpolar channel between 45° and 55°S. It is qualitatively consistent with the Toggweiler and Samuels (1995) theory, that the increase in the NA cell is nearly equal to ET in W2 (4.6 Sv).
However, the strength of the NA cell does not respond simply to ET alone. An example is provided in Fig. 4. Although ET in TatlW3 is equal to that in TatlW1 (2.3 Sv), the NA cell in the former is larger than in the latter by 2 Sv (Fig. 4, bottom). These results suggest that the wind stresses outside of the channel can affect the NA cell, as demonstrated by Tsujino and Suginohara (1999) and Klinger et al. (2003). Next, the results are compared with respect to the changes in the thermodynamic structure. The enhancement of the wind driven gyres is displayed in Fig. 5 (left). The difference in potential density and meridional velocity between TatlW3 and TatlW1 on the zonal section at 30°S is presented in Fig. 5 (right). The anomalous northward flow is restricted to the near surface layers (top right panel), most of which are denser in TatlW3 than in TatlW1 (bottom right) because of the enhanced upwelling in the anomalous cyclonic gyre (subpolar gyre) seen between the wind stress maximum latitude (37.5°S) and the northern end of the channel latitude (45°S) in TatlW3. Since the reference density σ* is identical, the restoring surface boundary condition gives more buoyancy to the Southern Ocean in TatlW3 than in TatlW1 (Fig. 5, center). In TatlW3, the pycnocline north of the wind stress maximum is lowered by the anomalous Ekman downwelling, and the anomalous buoyancy gain in the increased northward Ekman transport region keeps the subsurface layers less dense there (Fig. 5, bottom right). On the other hand, anomalous warm current is restricted to the narrow western boundary region (top right), so there is only a slight loss in surface buoyancy, but more heat is transported from the lower latitudes to the wind maximum latitudes.
The differences in zonal mean potential density between TatlW3 and TatlW1 are given in Fig. 6 (top panel). An anomalous subpolar gyre is spun up only in TatlW3, and that region is denser than in TatlW1 because of the pumping up of the denser subsurface water. As the surface water becomes denser, more buoyancy is given in the manner mentioned above. The anomalous buoyancy gain and the enhanced anticyclonic gyre seen north of the westerly wind stress maximum (subtropical gyre) maintain the deeper pycnocline in that region. The warm region spreads to the northern part of the basin and increases its buoyancy loss. Thus, the wind stress changes outside of the channel can affect not only the buoyancy flux in the Southern Ocean but also the density field that sets the NA cell in the Northern Hemisphere.
The ET effect on the NA cell can be explained in a similar way by comparing TatlW1 with TatlW0 and TatlW2 with TatlW1 (Fig. 6, middle and bottom panels). Anomalous Ekman upwelling at the circumpolar channel brings denser water to the surface. The denser surface layer receives more buoyancy from the atmosphere. An anomalous subtropical gyre and the warm western boundary current are spun up to the north of the channel and lower the pycnocline there. Each panel in Fig. 6 demonstrates that the subsurface layers are less dense to the north of the channel, when ET is large. The difference in zonal mean density between TatlW0 and TatlW1 (Fig. 6) indicates that less dense water spreads into the Northern Hemisphere and increases the buoyancy loss there, as in the top panel. The changes in the density distribution in the Northern Hemisphere correspond to the NA-cell enhancement, as discussed later. It should be pointed out that any noticeable difference in zonal mean density cannot be seen between TatlW2 and TatlW1.
The distributions of the buoyancy fluxes in TatlW0 and TatlW2 are displayed in Fig. 7. Cooling in the Northern Hemisphere occurs in the warm western boundary current region and high latitudes (shaded). The zonally integrated buoyancy loss and gain are given in Fig. 8. The adapted wind stresses in TatlW2 increase the buoyancy gain in the Southern Ocean, which is balanced against the increased buoyancy loss in the Northern Hemisphere, as reported by Hasumi and Suginohara (1999).
The top panel of Fig. 9 depicts the relationship between the NA-cell strength and the sum of buoyancy loss in the Northern Hemisphere. The increase in the latter corresponds to the increase in the sum of buoyancy gain in the Southern Ocean. Since the changes in wind stress are restricted to the south of 22.5°S and the surface waters in the low latitudes are too light to mix with the Southern Ocean waters, the buoyancy fluxes in the low latitudes are not affected by the wind changes (Fig. 8) and do not participate in the global heat budget changes. The reason why the increased buoyancy gain in the Southern Ocean affects the remote NA-cell strength is discussed in section 3b.
The NA-cell strength is proportional to the buoyancy flux, exhibiting the correlation between the meridional overturning mass transport and the meridional heat transport in the basin. It should be noted here that the NA-cell strength is set by the global buoyancy forcing, while the change in ET does not sufficiently explain the NA-cell enhancement (Fig. 9, bottom).
2) The surface density-restoring rate and the Drake Passage effect
The response of the NA cell in Tatlslw cases (closed circles in Fig. 9, bottom) to various wind stresses is weaker than in the Tatl cases (triangles). These results correspond to the claim asserted by Rahmstorf and England (1997): the robust Drake Passage effect in Toggweiler and Samuels (1995) is due to the use of a thermal-restoring boundary condition. Under a slow density-restoring boundary condition (Tatlslw), the increase in the Southern Ocean wind does not cause much buoyancy gain in the Southern Ocean, leading to weaker NA cells than in Tatl. The Drake Passage effect under a realistic surface boundary condition is explored in section 4.
3) The twin-basin experiments and the deep outflow into the Southern Ocean
Figure 9 includes the results from the twin-basin experiments, where the NA-cell strength and the sum of buoyancy loss are calculated in the west basin. The buoyancy flux change in the northern east basin is so small that the increase in the buoyancy loss in the northern west basin is balanced against the increase in the buoyancy gain in the Southern Ocean in the steady state.
The NA-cell strength in the twin-basin model is larger than that in the single-basin model under an identical wind stress (bottom panel of Fig. 9). This enlargement is explained by the doubled area of the Southern Ocean, where the buoyancies are poured from the atmosphere. Although the NA-cell enhancement might be explained by the doubled ET based on the mechanical Drake Passage effect, the relation between the NA cell and ET is loose (Fig. 9, bottom), while the top panel shows tight correlation between NA-cell strength and buoyancy fluxes. In both single-basin and twin-basin models, the increased westerly winds increase the buoyancy gain in the Southern Ocean, change the density distribution in the northern (west) basin, and enhance the NA cell, increasing the buoyancy loss there.
In the twin-basin configuration, part of the deep water produced in the west basin upwells in the east basin, where the upward heat advection and downward heat diffusion balance. Toggweiler and Samuels (1995) related the magnitude of the NADW outflow into the Southern Ocean to ET. The following discusses the relationship among the deep outflow into the Southern Ocean, the northern upwelling circulation in the east basin [Indo-Pacific (IP) cell], and ET.
The southward transport of the deep water originating from the northern west basin is estimated on the σ−ϕ plane. The streamfunctions for cases (Tatl+Tpac2)W1 and (Tatl+Tpac2)W3 are displayed in Fig. 10, separated into three parts: the Southern Ocean, the west basin, and the east basin. In (Tatl+Tpac2)W1, deep water formed in the west basin outflows below the σ0 = 27.7 surface, across the 35°S section. Its southward transport (SF35) is 5.6 Sv. Below the σ0 = 27.7 surface, the east basin receives 5.2 Sv of the water that upwells in it. The residual 0.4 Sv of water must upwell in the Southern Ocean and return into the west basin above the σ0 = 27.7 surface. In (Tatl+Tpac2)W3, although ET is the same as that in (Tatl+Tpac2)W1, deep southward transport below the σ = 27.65 surface from the west basin is 6.8 Sv, and the east basin receives 4.8 Sv of the water below the σ = 27.65 surface, with 2.0 Sv of upwelling in the Southern Ocean. The quantities are estimated in the same way as above for the eight cases with the twin-basin configuration and listed in Table 1. Note that the upwelling in the east basin is larger in (Tatl+Tpac1) cases than in (Tatl+Tpac2) cases by about 0.5 Sv with an identical wind forcing. As the vertical density gradient is larger in the former than in the latter, the diffusively driven IP cell is also larger in the former.
Figure 10 and Table 1 also demonstrate that the IP cell makes a large branch of the deep outflow from the west basin and is not affected by the wind stress changes in the Southern Ocean. This means that the change in deep outflowing transport SF35 depends on the change in upwelling in the Southern Ocean. The upwelling estimated on the σ–ϕ plane corresponds to the buoyancy gain in the Southern Ocean, which is balanced with buoyancy loss in the northern west basin. The top panel in Fig. 11 compares the buoyancy loss in the northern west basin with SF35. The bottom panel compares SF35 to ET. It should be noted that the changes in SF35 and upwelling in the Southern Ocean cannot be explained consistently by the change in ET.
b. The temporal evolution of the density field and the overturning
This section presents the processes through which the buoyancy anomalies excited by the wind stresses in the Southern Ocean affect the remote NA cell.
1) Zonal mean density
The following discusses the temporal evolutions of the zonal mean density field and the NA cell in the west basin after W3 forcing was adapted to the (Tatl+Tpac2)W0 solution. Figure 12 illustrates how they change with time. After 2 yr of integration, subsurface water to the north of the westerly wind maximum becomes less dense and that to the south of the maximum becomes denser. The anomalous overturning seems to weaken the original NA cell in the vicinity of 20°S. Eight years later (in the tenth year), the region of less dense water (by 0.05 kg m−3) intrudes into the northern midlatitudes in the west basin. The anomalous NA cell then starts growing. After 100 yr of integration, the whole ocean seems to be warmed up, except for the southernmost part and the surface just to the north of the circumpolar channel. The developed NA cell can also be seen. The temporal evolution of the meridional section panels in Fig. 12 indicates that the buoyancy anomalies excited by the wind stresses in the Southern Ocean spread into the northern part of the west basin and strengthen the NA cell.
2) Shear transport streamfunction as a visualizing tool for vertically integrated density anomaly
As wind stress is not imposed on the Northern Hemisphere in these experiments, vertically integrated horizontal transport must be zero in the lowest order. Closed ψST contours around a relative high center then correspond to a clockwise circulation in the upper layers and an anticlockwise circulation in the lower layers. The high center also represents the less dense anomalies near those areas. A low center indicates a reversed pattern of circulation and density. Water cannot flow across a wall, so the contour lines of ψST running into lateral boundaries represent the flow to the boundaries in the upper (lower) layers and the flow from the boundaries in the lower (upper) layers. Therefore, these contour lines indicate the downwelling (upwelling) or ageostrophic flows there.
Figure 13 depicts ψST in the northern west basin in the final states of (Tatl+Tpac2) × W0, W1, and W3. The features seen in the single-basin experiments (not shown) are qualitatively the same as in Fig. 13. The narrowness of the downwelling region (Marotzke and Scott 1999) near the northeast corner is visualized by the concentration of ψST contour lines. As ψST represents the vertically integrated buoyancy anomalies, Fig. 13 demonstrates that the increase in the Southern Ocean wind makes the northern west basin more buoyant, enhances NA cell, and increases the buoyancy losses there.
3) The density anomaly propagation on plan-view map
The shear transport streamfunction introduced above is used to illustrate the temporal evolution of vertically integrated buoyancy anomalies and geostrophic overturning flow. The transient experiment as analyzed in section 3b(1) is also used here. The ψST deviations from the initial state (Tatl+Tpac2)W0 in the northern part of the basins are shown on the time series of plan-view maps (Fig. 14). The W3 forcing is adapted abruptly, so the path of the buoyancy anomalies excited by the wind can be detected. This transient experiment is similar to an experiment by McDermott (1996), in which the westerly wind stress over the Southern Ocean is strengthened and weakened. McDermott (1996) reported the initial response of ocean to wind changes in a two-basin model.
The following processes, not seen in Fig. 14, are deduced to occur in the Southern Hemisphere and near the equator (e.g., Kawase 1987; McDermott 1996). The buoyant anomalies at the western boundary region excited by the increased buoyancy gain in the Southern Ocean propagate northward as Kelvin waves, and then eastward along the equator. At the eastern boundary, the anomalies separate into three branches propagating southward, westward, and northward. The southward branch passes around the Cape of Good Hope in the current model and enters the east basin. The west branch travels back along the equator as reflected Rossby waves.
Figure 14 demonstrates that the north branch anomalies propagate northward along the eastern boundary in the west basin. These baroclinic Kelvin waves cannot enter the northernmost region of the basin where the water column is hardly stratified (McDermott 1996). The anomalies thus lock at the northeastern corner of the west basin and subsequently propagate westward as baroclinic Rossby waves (see 5th-, 10th-, 20th-, and 40th-year states in the left panels of Fig. 14). The buoyant anomalies corresponding to high ψST accumulate in the western part and intensify the meridional overturning current at the western boundary. The density field is significantly modified in the west basin (60th-, 120th-, 200th-, and 300th-year states in right panels of Fig. 14). On the other hand, in the northeastern basin, the anomalies propagate along the equator, the eastern and northern boundaries without obstacles like the vertically homogenized region in the west basin. Furthermore, anomalies at the eastern boundary propagate westward as Rossby waves. The northwestern part of the east basin is left behind, and the relatively low ψST region forms there (see 20th-year panel). However, as the buoyant anomalies continuously pour into the east basin, the relative low is buried and fades out, and only the anomalous zonal overturning remains (Fig. 14; 300th year).
The high-ψST anomaly in the west basin corresponds to the enhanced upper northward flow and lower southward flow at the western boundary, indicating intensification of the NA cell. The low ψST anomaly seen for a time in the east basin corresponds to the weakening of the IP cell. However, in the final state when the relative low ψST region does not exist, the IP cell in the east basin is almost as strong as in the initial state, despite the increased wind stresses over the Southern Ocean (Table 1). In contrast, the NA cell responds robustly to the changes in the buoyancy fluxes caused by the adapted westerly wind stresses over the Southern Ocean (Fig. 9).
4. Experiments with realistic configuration
The idealized configuration in the experiments presented in the previous sections has the following problems. 1) The idealized topography and wind stress distribution. In particular, the zonal length scale and the latitudinal location of the Drake Passage are far different from the real situation. 2) Use of potential density as a prognostic variable. In reality, seawater density is a function of temperature, salinity, and pressure, which should be prognostic variables in a realistic model. 3) The density-restoring surface boundary condition. It is less realistic than a flux boundary condition. 4) Use of horizontal diffusion scheme. It is less appropriate for parameterizing subgrid-scale eddies than GM90 in a coarse resolution model.
To prove whether the results in the previous section are independent of these idealizations and simplifications, additional experiments were conducted with 1) realistic topography, 2) temperature and salinity as prognostic variables, 3) heat and moisture fluxes calculated by bulk formulas, and 4) isopycnal thickness diffusion scheme (GM90). The main purpose is to show that the buoyancy flux anomalies excited by the wind stress anomalies in the Southern Ocean can explain significant changes in the NA cell more consistently than the northward Ekman transport at the Cape Horn latitude (ET in this section).
a. Experimental design
A brief description of the global ocean model with realistic topographies (Fig. 15) is presented in the appendix. The surface fluxes (heat, moisture, and wind stress) are calculated through bulk formulas using sea surface temperature and prescribed atmospheric conditions. Artificial salinity-restoring is not used.
Temperature and salinity distribution in the initial state of the model are given by the Polar Science Center Hydrographic Climatology (PHC; Steele et al. 2001). The numerical model is integrated with prescribed atmospheric boundary conditions (the pilot phase of Ocean Model Intercomparison Project; Röske 2001) for 400 yr. This experiment will be referred to as EX1. Two experiments are branched out from the 200th-year state in EX1 and integrated for another 200 yr. In one of the two branches, westerly wind stresses are added at the latitudes from 65° to 50°S throughout the year (Fig. 16). In the other, westerly wind stresses are added to the north of Cape Horn latitude (55°S in the realistic model). The former will be called EX2 and the latter EX3, and their respective wind stresses correspond to W2 and W3 in the idealized experiments. These three experiments do not reach equilibrium in a few hundred years of integration and exhibit decreasing trend of changes in the NA cell. However, the NA-cell strength in EX2 and EX3 differs sufficiently from EX1 for meaningful argument.
b. Results from realistic experiments
The differences of zonal mean density and meridional overturning between EX2 and EX1 after 200 yr of integration after the branch point are depicted in Fig. 17 (upper panels). The density differences are smaller than 0.1 kg m−3 almost everywhere in the North Atlantic (left). The differences between EX3 and EX1 are displayed in Fig. 18. The zonal mean density differences are much larger than EX2–EX1. These pictures correspond to Figs. 6 and 12, discussed in the previous section. The meridional overturning streamfunction maximum in the North Atlantic is 13.8 Sv in EX1, 16.9 Sv in EX2, and 26.9 Sv in EX3. The NA cell in EX2 is larger than that in EX1 by only 3.1 Sv, while that in EX3 is twice as strong as that in EX1. The NA-cell differences can be compared with those presented in Fig. 12.
The North Atlantic heat loss in EX2 is only a little larger than in EX1 (Fig. 17, lower-left panel). The heat loss in EX3 is larger than in EX1 by >200 W m−2 at the maximum. These pictures should be compared with Figs. 7 and 13 from section 3. The enhanced NA cell induces the anomalous cooling in the northwestern North Atlantic. The IP-cell strength is estimated at 30°S, and the streamfunction minimum (anticlockwise; negative) is at 3500-m depth in each of EX1 (−13.3 Sv), EX2 (−10.8 Sv) and EX3 (−12.7 Sv). The IP cell in EX2 is weaker than that in EX1 by 2.5 Sv (Fig. 17, lower-right panel). The IP-cell difference between EX3 and EX1 is only 0.6 Sv.
In the following discussion, the Southern Ocean region will be referred to as SH, and NA will indicate the North Atlantic region (Fig. 15). Figure 19 displays the temporal evolution of the heat fluxes in SH (Fig. 19a) and in NA (Fig. 19c), the NA-cell strength (Fig. 19b), the IP-cell strength (Fig. 19d), and ET estimated by the meridional overturning streamfunction maximum in the vicinity of the Cape Horn latitude (Fig. 19e). The IP-cell strengths are presented in absolute values. When the wind stresses are added, the heat loss in the SH region rapidly decreases (anomalous heat gain; Fig. 19a). The SH mean heat fluxes are larger than in EX1 by 0.6 W m−2 in EX2 and 4.5 W m−2 in EX3, and the differences remain almost constant for 200 yr. On the other hand, the heat loss in the NA region increases with the slow enhancement of the NA cell (Figs. 19b,c), compensating the unbalanced global heat budget. After 200 yr of integration, NA heat losses are larger than EX1 by 1.6 W m−2 in EX2 and 10.9 W m−2 in EX3, and the differences continue to expand. Since the heat content differences between EX3 and EX1 are much larger than those between EX2 and EX1 (indicated by the zonal mean density differences seen in top left panels in Figs. 17 and 18), the NA heat flux in EX3 is much larger than in EX2. The changes in SH heat flux in response to the increased westerly wind stresses in the Southern Ocean (Fig. 19a) can explain the change in the NA cell (Fig. 19a) that is tightly correlated to the change in NA heat flux (Fig. 19c).
It should be noted that the NA-cell change in EX3 is four times as large as in EX2 (Fig. 19b), while ET in EX3 is approximately identical to EX1 and smaller than EX2 by about 16 Sv (Fig. 19e). The increase in ET does not explain the NA-cell (nor deep outflow) enhancement successfully, as seen in the results from idealized experiments (Figs. 9 and 11; Table 1). These results suggest that the essential factor in the Drake Passage effect is the change in the heat fluxes in the Southern Ocean that modifies the density distribution in the global ocean.
For several decades, the IP cell is decayed after the wind stress increased, but it tends to recover the strength in further integration (Fig. 19d). This IP-cell response is qualitatively the same as in the idealized experiments [section 3b(3)], where the overturnings in the east basin are almost independent of ET (Table 1). Thus, IP cells in EX2 and EX3 are expected to become nearly equal to that in EX1 after thousands of years of integration, and then the wind-induced changes in the meridional overturning are concentrated to the changes in NA cells.
In EX2 and EX3, the westerly wind stresses are added at a certain latitudinal range without increasing the scalar wind speeds that increase the heat flux, evaporation, and wind stress through bulk formulas there. Other experiments were conducted where the sea surface scalar wind speed was increased consistently with the increased wind vector. The results are qualitatively the same as those described here.
5. Discussion
The major conclusion from this study that differs from Toggweiler and Samuels (1995) is that the increase in ET does not explain the NA-cell enhancement sufficiently when the Southern Ocean winds increase. The deep outflow from the North Atlantic does not correspond to ET (Table 1). It is striking that the NA-cell enhancement in EX3–EX1 (13 Sv at the 200th year) becomes several times as large as that in EX2–EX1 (3 Sv), while ET in EX2 is larger than in EX3 by 16 Sv (section 4). These results qualitatively differ from the conclusion by Klinger et al. (2003), where the channel winds produce 3 times as much NA-cell enhancement as that from the winds outside the channel. The difference in topographies and surface boundary conditions between their study and this one may be the reason.
This paper has demonstrated that the Drake Passage effect should be understood through the global buoyancy flux changes. Increased westerly winds induce the increase in buoyancy gain and excite the buoyancy anomalies in the Southern Ocean. These anomalies propagate into the North Atlantic as described by Kawase (1987) and McDermott (1996). The buoyancy anomalies especially affect the density distribution, intensifying the zonal density gradient in the North Atlantic, and enhance the NA cell, increasing its buoyancy loss. In a steady state, the increase in buoyancy loss in the North Atlantic is balanced against the increase in buoyancy gain in the Southern Ocean, as shown by Hasumi and Suginohara (1999). Thus, the buoyancy flux anomalies induced by the increased westerly wind stress in the Southern Ocean successfully predict the final NA-cell enhancement.
Based on this understanding of the Drake Passage effect, the probable NA-cell response to the poleward shift of the westerly winds over the Southern Ocean accompanying global warming (e.g., Cai et al. 2003) can be considered. The buoyancy flux changes in the Southern Ocean and in the North Atlantic dominantly affect the NA-cell enhancement in the context of the Drake Passage effect. The buoyancy gain in the Southern Ocean might increase as the result of the poleward shift of the westerly wind jet and temperature front in the atmosphere. However, an increase in the buoyancy loss in the North Atlantic cannot be predicted, since the surface air warming in high latitudes might weaken the deep-water formation.
The atmospheric temperature distribution cannot be independent of the winds. The westerly wind over the Southern Ocean should be correlated with the meridional atmospheric temperature gradient there, which directly affects the buoyancy flux on the sea surface. This limits the adaptability of the results in this paper to the behavior of the real ocean. Understanding the relation between the NA cell and the Southern Ocean winds in the real climate system in more detail requires a study that includes a physically consistent distribution of atmospheric temperature, humidity, wind stress, precipitation, river runoff, and the feedbacks from the other components in the climate system.
It is noteworthy that the IP cell is decayed for a while following the westerly wind stress adaptation. In the final state, however, the IP-cell strength returns to nearly the same as it was in the initial state. Similar responses are seen in the realistic configurations. It is useful to know how the IP cell responds when the Southern Ocean winds are shifted and/or intensified. In a few decades after the wind intensification, the change in the IP cell is larger than in the NA cell. Therefore, a faster change in the signal of the Southern Ocean wind may be detected in the Indo-Pacific Ocean, but a hundred years must pass before the Drake Passage effect will be observed.
The thrust of this study is not the effect of the eddy-induced transport on the NA cell (e.g., Kamenkovich and Sarachik 2004). However, it was at least confirmed that the major results from the simplified experiments with a horizontal diffusion scheme are also robust in the experiments with GM90. Since one of the major roles of ocean eddies is to flatten isopycnal surfaces, the NA cell in an ocean model decays with large GM90 coefficients. It would be beneficial to use an adequate eddy parameterization in a coarse-resolution model and/or to develop an eddy-resolving model to simulate the real ocean.
6. Summary
Idealized experiments with various forcings and experiments with realistic configurations were conducted to clarify the processes in the Drake Passage effect. The following discusses the major processes dominant in the Drake Passage effect and presents conclusions gathered from this paper.
Increased westerly wind stresses over the Southern Ocean excite buoyancy anomalies there. These wind stresses bring dense subsurface water up to the surface in the anomalous subpolar gyre or the channel latitudes south of the westerly wind stress maximum and transport it to lower latitudes. Under the surface density-restoring boundary condition, denser sea surface water induces anomalous buoyancy gains in the Southern Ocean. Simultaneously, the enhanced southward western boundary current transports more buoyant water originating in lower latitudes to the westerly wind stress maximum latitude (Fig. 5). Thus, the subsurface water in the anomalous subtropical gyre spun up north of the wind maximum becomes more buoyant than that in the weaker wind case (Fig. 6). A density-restoring time scale affects the buoyancy flux responses to the winds in this phase.
These anomalies propagate into the Northern Hemisphere as baroclinic waves, modify the density distribution, and simultaneously enhance the NA cell. The vertically homogenized region at high latitude supports the NA-cell enhancement. The buoyancy losses in the North Atlantic correlate to the NA-cell strength.
The NA cell can continue to grow until the cooling (buoyancy loss) in the North Atlantic catches up with the warming (buoyancy gain) in the Southern Ocean. Therefore, the buoyancy flux change in the Southern Ocean is the essential factor in the Drake Passage effect. Since the winds outside the channel latitudes significantly affect the buoyancy flux, the increase in ET is not a good measure of the NA-cell enhancement.
Acknowledgments
We thank Dr. Hiroyuki Tsujino, the members of the Oceanographic Division of Meteorological Research Institute, and Dr. Ryo Furue for pleasant discussions and helpful comments. Two anonymous reviewers are acknowledged for constructive criticism that helped to improve the manuscript. This work is funded by the Meteorological Research Institute.
REFERENCES
Borowski, D., R. Gerdes, and D. Olbers, 2002: Thermohaline and wind forcing of a circumpolar channel with blocked geostrophic contours. J. Phys. Oceanogr., 32 , 2520–2540.
Bryan, K., 1984: Accelerating the convergence to equilibrium of ocean–climate models. J. Phys. Oceanogr., 14 , 666–673.
Cai, W., P. H. Whetton, and D. J. Karoly, 2003: The response of the Antarctic Oscillation to increasing and stabilized atmospheric CO2. J. Climate, 16 , 1525–1538.
Cox, M. D., 1987: Isopycnal diffusion in a z-coordinate ocean model. Ocean Modell., 74 , 1–5.
Gent, P. R., and J. C. McWilliams, 1990: Isopycnal mixing in ocean circulation models. J. Phys. Oceanogr., 20 , 150–155.
Hasumi, H., and N. Suginohara, 1999: Atlantic deep circulation controlled by heating in the Southern Ocean. Geophys. Res. Lett., 26 , 1873–1876.
Hellerman, S., and M. Rosenstein, 1983: Normal monthly wind stress over the world ocean with error estimates. J. Phys. Oceanogr., 13 , 1093–1104.
Hunke, E. C., and J. K. Dukowicz, 2002: The elastic–viscous–plastic sea ice dynamics model in general orthogonal curvilinear coordinates on a sphere-incorporation of metric terms. Mon. Wea. Rev., 130 , 1848–1865.
Ishikawa, I., H. Tsujino, M. Hirabara, H. Nakano, T. Yasuda, and H. Ishizaki, 2005: Meteorological Research Institute Community Ocean Model (MRI.COM) Manual (in Japanese). Tech. Rep. 47, Technical Reports of the Meteorological Research Institute, 189 pp.
Ishizaki, H., and T. Motoi, 1999: Reevaluation of the Takano–Oonishi scheme for momentum advection on bottom relief in ocean models. J. Atmos. Oceanic Technol., 16 , 1994–2010.
Kamenkovich, I., and E. Sarachik, 2004: Mechanisms controlling the sensitivity of the Atlantic thermohaline circulation to the parameterization of eddy transports in ocean GCMs. J. Phys. Oceanogr., 34 , 249–266.
Kara, A., P. Rochford, and H. Hurlburt, 2000: Efficient and accurate bulk parameterizations of air–sea fluxes for use in general circulation models. J. Atmos. Oceanic Technol., 17 , 1421–1438.
Kawase, M., 1987: Establishment of deep ocean circulation driven by deep-water production. J. Phys. Oceanogr., 17 , 2294–2317.
Klinger, B. A., S. Drijfhout, J. Marotzke, and J. R. Scott, 2003: Sensitivity of basinwide meridional overturning to diapycnal diffusion and remote wind forcing in an idealized Atlantic–Southern Ocean geometry. J. Phys. Oceanogr., 33 , 249–266.
Klinger, B. A., S. Drijfhout, J. Marotzke, and J. R. Scott, 2004: Remote wind-driven overturning in the absence of the Drake Passage effect. J. Phys. Oceanogr., 34 , 1036–1049.
Leonard, B. P., 1979: A stable and accurate convective modeling procedure based upon quadratic upstream interpolation. J. Comput. Methods Appl. Mech. Eng., 19 , 59–98.
Leonard, B. P., M. K. MacVean, and A. P. Lock, 1993: Positivity-preserving numerical schemes for multidimensional advection. NASA Tech. Memo. 106055, ICOMP-93-05, 62 pp.
Marotzke, J., and J. R. Scott, 1999: Convective mixing and the thermohaline circulation. J. Phys. Oceanogr., 29 , 2962–2970.
McDermott, D. A., 1996: The regulation of northern overturning by Southern Hemisphere winds. J. Phys. Oceanogr., 26 , 1234–1255.
Mellor, G. L., and T. Yamada, 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. Space Phys., 20 , 851–875.
Mellor, G. L., and L. Kantha, 1989: An ice–ocean coupled model. J. Geophys. Res., 94 , 10937–10954.
Nof, D., 2000: Does the wind control the import and export of the South Atlantic? J. Phys. Oceanogr., 30 , 2650–2667.
Nof, D., and A. M. Boer, 2004: From the Southern Ocean to the North Atlantic in the Ekman layer? Bull. Amer. Meteor. Soc., 85 , 79–87.
Oke, P., and M. England, 2004: Oceanic response to changes in the latitude of the Southern Hemisphere subpolar westerly winds. J. Climate, 27 , 1040–1054.
Rahmstorf, S., and M. England, 1997: Influence of Southern Hemisphere winds on North Atlantic Deep Water flow. J. Phys. Oceanogr., 27 , 2040–2054.
Röske, F., 2001: An atlas of surface fluxes based on the ECMWF Re-Analysis—A climatological dataset to force global ocean general circulation models. Tech. Rep. 323, Max-Planck-Institut für Meteorologie, 31 pp.
Steele, M., R. Morley, and W. Ermold, 2001: PHC: A global ocean hydrography with a high-quality Arctic Ocean. J. Climate, 14 , 2079–2087.
Toggweiler, J. R., and B. Samuels, 1995: Effect of Drake Passage on the global thermohaline circulation. Deep-Sea Res., 42 , 477–500.
Tsujino, H., and N. Suginohara, 1999: Thermohaline circulation enhanced by wind forcing. J. Phys. Oceanogr., 29 , 1506–1516.
Tsujino, H., H. Hasumi, and N. Suginohara, 2000: Deep Pacific circulation controlled by vertical diffusivity at the lower thermocline depth. J. Phys. Oceanogr., 30 , 2853–2865.
APPENDIX
Description of the Model with Realistic Configuration
The following briefly describes the model (MRI. COM; Ishikawa et al. 2005) used in section 4 of this paper. The horizontal grid interval is 2° (zonal) × 1° (meridional) in generalized orthogonal coordinates (Fig. 15). The model ocean consists of 48 vertical levels, with thickness increasing from 4 m at the surface to 250 m below 2250 m. The momentum advection scheme is the same used in the idealized experiments. The Quadratic Upstream Interpolation for Convective Kinematics with Estimated Streaming Terms (QUICKEST; Leonard 1979) and the Uniformly Third Order Polynomial Interpolation Algorithm (UTOPIA; Leonard et al. 1993) are used as the vertical and horizontal tracer advection schemes. An isopycnal diffusion scheme (Cox 1987) and GM90 are used with coefficient 500 m2 s−1 (above 200-m depth, the GM90 coefficient monotonically decreases to zero at the surface). The vertical diffusivity and viscosity are set by the turbulence closure level 2.5 scheme (Mellor and Yamada 1982). The background vertical diffusivities follow those reported by Tsujino et al. (2000). The seawater densities are calculated by an accurate equation of state (Ishikawa et al. 2005). Thermodynamic (based on Mellor and Kantha 1989) and dynamic (Hunke and Dukowicz 2002) sea ice components are implemented in the model.
Latent and sensible heat fluxes and evaporation are calculated through bulk formulas using surface air temperature, dewpoint, and wind speed used in the pilot phase of the Ocean Model Intercomparison Project (P-OMIP; Röske 2001). Wind stresses are also calculated from surface air temperature and wind velocity through the bulk formulas based on Kara et al. (2000). Insolation and long-wave radiation also participate in the heat fluxes. Prescribed precipitation and river runoff participate in the freshwater fluxes. Artificial salinity-restoring is not used.
Topographies of model ocean. (left) Plan view of single-basin model with periodically connected passage ranging from 55° to 45°S. (center) Lateral view of the basin. It is common to single-basin and twin-basin models. Flat bottom is 5125 m deep, while the circumpolar channel is 2875 m deep. (right) Plan view of twin-basin ocean with the same passage as the single-basin ocean. A wall extending from the northern boundary to 35°S separates the ocean at the midlongitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Topographies of model ocean. (left) Plan view of single-basin model with periodically connected passage ranging from 55° to 45°S. (center) Lateral view of the basin. It is common to single-basin and twin-basin models. Flat bottom is 5125 m deep, while the circumpolar channel is 2875 m deep. (right) Plan view of twin-basin ocean with the same passage as the single-basin ocean. A wall extending from the northern boundary to 35°S separates the ocean at the midlongitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Topographies of model ocean. (left) Plan view of single-basin model with periodically connected passage ranging from 55° to 45°S. (center) Lateral view of the basin. It is common to single-basin and twin-basin models. Flat bottom is 5125 m deep, while the circumpolar channel is 2875 m deep. (right) Plan view of twin-basin ocean with the same passage as the single-basin ocean. A wall extending from the northern boundary to 35°S separates the ocean at the midlongitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Profiles of forcings. (left) Surface reference density σ* vs latitude. Damping rate of restoring is proportional to (20 days)−1 for Tpac1, Tpac2, Tmid, and Tatl and (80 days)−1 for Tatlslw (profile of Tatlslw is identical to that of Tatl). (right) Westerly wind stress vs latitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Profiles of forcings. (left) Surface reference density σ* vs latitude. Damping rate of restoring is proportional to (20 days)−1 for Tpac1, Tpac2, Tmid, and Tatl and (80 days)−1 for Tatlslw (profile of Tatlslw is identical to that of Tatl). (right) Westerly wind stress vs latitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Profiles of forcings. (left) Surface reference density σ* vs latitude. Damping rate of restoring is proportional to (20 days)−1 for Tpac1, Tpac2, Tmid, and Tatl and (80 days)−1 for Tatlslw (profile of Tatlslw is identical to that of Tatl). (right) Westerly wind stress vs latitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction in (top) TatlW0 and (bottom) TatlW2; contour interval is 1 Sv. The solid lines denote positive values, and the broken ones denote negative values. The region with positive values corresponds to the clockwise circulation around the region. The thick dotted line traces the circumpolar passage. The northern sinking circulation cell (NA) can be seen in the two panels, and the northward Ekman transport at 45°S (ET) can be seen in the bottom panel (see text).
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction in (top) TatlW0 and (bottom) TatlW2; contour interval is 1 Sv. The solid lines denote positive values, and the broken ones denote negative values. The region with positive values corresponds to the clockwise circulation around the region. The thick dotted line traces the circumpolar passage. The northern sinking circulation cell (NA) can be seen in the two panels, and the northward Ekman transport at 45°S (ET) can be seen in the bottom panel (see text).
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction in (top) TatlW0 and (bottom) TatlW2; contour interval is 1 Sv. The solid lines denote positive values, and the broken ones denote negative values. The region with positive values corresponds to the clockwise circulation around the region. The thick dotted line traces the circumpolar passage. The northern sinking circulation cell (NA) can be seen in the two panels, and the northward Ekman transport at 45°S (ET) can be seen in the bottom panel (see text).
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction in (top) TatlW1, (middle) TatlW3, and (bottom) difference between them; contour interval is 1 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction in (top) TatlW1, (middle) TatlW3, and (bottom) difference between them; contour interval is 1 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction in (top) TatlW1, (middle) TatlW3, and (bottom) difference between them; contour interval is 1 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences between TatlW3 and TatlW1. (left) Horizontal streamfunction; contour interval is 5 Sv. The region with positive values corresponds to an anomalous clockwise circulation. Contour line of zero is not drawn. (center) Anomalous buoyancy flux at the surface; contour interval is 1 × 10−10 kg m−1 s−1. Shading denotes negative values (anomalous buoyancy loss). (top right) Anomalous meridional velocity in 30°S section; contour interval is 1 cm s−1. Shading denotes southward anomaly. (bottom right) Difference in potential density; contour interval is 0.05 kg m−3. Shading denotes the region where potential density is denser in TatlW3 than in TatlW1.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences between TatlW3 and TatlW1. (left) Horizontal streamfunction; contour interval is 5 Sv. The region with positive values corresponds to an anomalous clockwise circulation. Contour line of zero is not drawn. (center) Anomalous buoyancy flux at the surface; contour interval is 1 × 10−10 kg m−1 s−1. Shading denotes negative values (anomalous buoyancy loss). (top right) Anomalous meridional velocity in 30°S section; contour interval is 1 cm s−1. Shading denotes southward anomaly. (bottom right) Difference in potential density; contour interval is 0.05 kg m−3. Shading denotes the region where potential density is denser in TatlW3 than in TatlW1.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences between TatlW3 and TatlW1. (left) Horizontal streamfunction; contour interval is 5 Sv. The region with positive values corresponds to an anomalous clockwise circulation. Contour line of zero is not drawn. (center) Anomalous buoyancy flux at the surface; contour interval is 1 × 10−10 kg m−1 s−1. Shading denotes negative values (anomalous buoyancy loss). (top right) Anomalous meridional velocity in 30°S section; contour interval is 1 cm s−1. Shading denotes southward anomaly. (bottom right) Difference in potential density; contour interval is 0.05 kg m−3. Shading denotes the region where potential density is denser in TatlW3 than in TatlW1.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences in zonal mean density between (top) TatlW3 and TatlW1, (middle) TatlW1 and TatlW0, and (bottom) TatlW2 and TatlW1; contour interval is 0.05 kg m−3. Shading denotes the region where water in the former case is denser than in the latter.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences in zonal mean density between (top) TatlW3 and TatlW1, (middle) TatlW1 and TatlW0, and (bottom) TatlW2 and TatlW1; contour interval is 0.05 kg m−3. Shading denotes the region where water in the former case is denser than in the latter.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences in zonal mean density between (top) TatlW3 and TatlW1, (middle) TatlW1 and TatlW0, and (bottom) TatlW2 and TatlW1; contour interval is 0.05 kg m−3. Shading denotes the region where water in the former case is denser than in the latter.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Buoyancy flux at the surface in (left) TatlW0 and (right) TatlW2; contour interval is 1 × 10−10 kg m−2 s−1. Shading denotes negative values (buoyancy loss).
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Buoyancy flux at the surface in (left) TatlW0 and (right) TatlW2; contour interval is 1 × 10−10 kg m−2 s−1. Shading denotes negative values (buoyancy loss).
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Buoyancy flux at the surface in (left) TatlW0 and (right) TatlW2; contour interval is 1 × 10−10 kg m−2 s−1. Shading denotes negative values (buoyancy loss).
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Zonally integrated buoyancy gain (reduction of the density) and buoyancy loss at the surface in (left) TatlW0 and (right) TatlW2.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Zonally integrated buoyancy gain (reduction of the density) and buoyancy loss at the surface in (left) TatlW0 and (right) TatlW2.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Zonally integrated buoyancy gain (reduction of the density) and buoyancy loss at the surface in (left) TatlW0 and (right) TatlW2.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Sum of buoyancy loss in the Northern Hemisphere vs the NA-cell strength. Open circles denote single-basin experiments Tpac2 and Tmid. Triangles denote Tatl and the closed circles Tatlslw experiments. Rectangles and diamonds denote twin-basin experiments (Tatl+Tpac1) and (Tatl+Tpac2). The NA-cell strength and the buoyancy loss are integrated in the west basin for twin-basin experiments. The solid line denotes the linear regression for single-basin experiments, and the broken line that for twin-basin experiments. (bottom) Northward Ekman transport at 45°S (ET) vs the strength of NA cell. The symbols are the same as at top.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Sum of buoyancy loss in the Northern Hemisphere vs the NA-cell strength. Open circles denote single-basin experiments Tpac2 and Tmid. Triangles denote Tatl and the closed circles Tatlslw experiments. Rectangles and diamonds denote twin-basin experiments (Tatl+Tpac1) and (Tatl+Tpac2). The NA-cell strength and the buoyancy loss are integrated in the west basin for twin-basin experiments. The solid line denotes the linear regression for single-basin experiments, and the broken line that for twin-basin experiments. (bottom) Northward Ekman transport at 45°S (ET) vs the strength of NA cell. The symbols are the same as at top.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Sum of buoyancy loss in the Northern Hemisphere vs the NA-cell strength. Open circles denote single-basin experiments Tpac2 and Tmid. Triangles denote Tatl and the closed circles Tatlslw experiments. Rectangles and diamonds denote twin-basin experiments (Tatl+Tpac1) and (Tatl+Tpac2). The NA-cell strength and the buoyancy loss are integrated in the west basin for twin-basin experiments. The solid line denotes the linear regression for single-basin experiments, and the broken line that for twin-basin experiments. (bottom) Northward Ekman transport at 45°S (ET) vs the strength of NA cell. The symbols are the same as at top.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction on the σ–ϕ plane. (top) (Tatl+Tpac2) W1, (bottom) (Tatl+Tpac2) W3, (left) Southern Ocean, (center) west basin, and (right) east basin; contour interval is 1 Sv. (left) Shading denotes the outcropping of the region at each latitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction on the σ–ϕ plane. (top) (Tatl+Tpac2) W1, (bottom) (Tatl+Tpac2) W3, (left) Southern Ocean, (center) west basin, and (right) east basin; contour interval is 1 Sv. (left) Shading denotes the outcropping of the region at each latitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional overturning streamfunction on the σ–ϕ plane. (top) (Tatl+Tpac2) W1, (bottom) (Tatl+Tpac2) W3, (left) Southern Ocean, (center) west basin, and (right) east basin; contour interval is 1 Sv. (left) Shading denotes the outcropping of the region at each latitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Sum of buoyancy loss in the Northern Hemisphere against the southward transport of deep water across the 35°S section estimated on σ–ϕ planes (SF35). (bottom) Northward Ekman transport at 45°S against the outflow of deep water (SF35). The symbols are the same as in Fig. 9.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Sum of buoyancy loss in the Northern Hemisphere against the southward transport of deep water across the 35°S section estimated on σ–ϕ planes (SF35). (bottom) Northward Ekman transport at 45°S against the outflow of deep water (SF35). The symbols are the same as in Fig. 9.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Sum of buoyancy loss in the Northern Hemisphere against the southward transport of deep water across the 35°S section estimated on σ–ϕ planes (SF35). (bottom) Northward Ekman transport at 45°S against the outflow of deep water (SF35). The symbols are the same as in Fig. 9.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(left) Zonal mean potential density in the west basin for (top to bottom) (Tatl+Tpac2) W0 solution and for changes after W3 forcing is adapted in the initial state; contour interval is 0.05 kg m−3. Shading indicates that the region is denser than the initial. The states after 2, 10, and 100 yr of integration are displayed. (right) Meridional overturning streamfunction in the west basin for (top to bottom) initial state and for changes in the streamfunction after the wind stress is switched on; contour interval is 1 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(left) Zonal mean potential density in the west basin for (top to bottom) (Tatl+Tpac2) W0 solution and for changes after W3 forcing is adapted in the initial state; contour interval is 0.05 kg m−3. Shading indicates that the region is denser than the initial. The states after 2, 10, and 100 yr of integration are displayed. (right) Meridional overturning streamfunction in the west basin for (top to bottom) initial state and for changes in the streamfunction after the wind stress is switched on; contour interval is 1 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(left) Zonal mean potential density in the west basin for (top to bottom) (Tatl+Tpac2) W0 solution and for changes after W3 forcing is adapted in the initial state; contour interval is 0.05 kg m−3. Shading indicates that the region is denser than the initial. The states after 2, 10, and 100 yr of integration are displayed. (right) Meridional overturning streamfunction in the west basin for (top to bottom) initial state and for changes in the streamfunction after the wind stress is switched on; contour interval is 1 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Shear transport streamfunction (ψST; contour interval is 4 Sv) and (bottom) buoyancy gain (contour interval is 1 × 10−10 kg m−2 s−1) in the west basin of the Northern Hemisphere for (left) (Tatl+Tpac2) W0, (center) (Tatl+Tpac2) W1, and (right) (Tatl+Tpac2) W3.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Shear transport streamfunction (ψST; contour interval is 4 Sv) and (bottom) buoyancy gain (contour interval is 1 × 10−10 kg m−2 s−1) in the west basin of the Northern Hemisphere for (left) (Tatl+Tpac2) W0, (center) (Tatl+Tpac2) W1, and (right) (Tatl+Tpac2) W3.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
(top) Shear transport streamfunction (ψST; contour interval is 4 Sv) and (bottom) buoyancy gain (contour interval is 1 × 10−10 kg m−2 s−1) in the west basin of the Northern Hemisphere for (left) (Tatl+Tpac2) W0, (center) (Tatl+Tpac2) W1, and (right) (Tatl+Tpac2) W3.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Anomaly of the shear transport streamfunction (ψST) in the twin-basin experiment (Tatl+Tpac2) after W3 forcing is adapted in the final state of the (Tatl+Tpac2) W0 experiment. (left) From (top) fifth year to (bottom) 40th year; contour interval is 0.2 Sv. (right) From (top) 60th year to (bottom) 300th year; contour interval is 1.0 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Anomaly of the shear transport streamfunction (ψST) in the twin-basin experiment (Tatl+Tpac2) after W3 forcing is adapted in the final state of the (Tatl+Tpac2) W0 experiment. (left) From (top) fifth year to (bottom) 40th year; contour interval is 0.2 Sv. (right) From (top) 60th year to (bottom) 300th year; contour interval is 1.0 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Anomaly of the shear transport streamfunction (ψST) in the twin-basin experiment (Tatl+Tpac2) after W3 forcing is adapted in the final state of the (Tatl+Tpac2) W0 experiment. (left) From (top) fifth year to (bottom) 40th year; contour interval is 0.2 Sv. (right) From (top) 60th year to (bottom) 300th year; contour interval is 1.0 Sv.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Topography of the model on the general orthogonal coordinates used in section 4. Grayscale denotes the water depth (m). Thin lines represent geographical longitudes and latitudes drawn at intervals of 30°. Thick rectangles define the NA and SH regions.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Topography of the model on the general orthogonal coordinates used in section 4. Grayscale denotes the water depth (m). Thin lines represent geographical longitudes and latitudes drawn at intervals of 30°. Thick rectangles define the NA and SH regions.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Topography of the model on the general orthogonal coordinates used in section 4. Grayscale denotes the water depth (m). Thin lines represent geographical longitudes and latitudes drawn at intervals of 30°. Thick rectangles define the NA and SH regions.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Zonal mean westerly wind stresses. Thin black solid line represents the standard westerly wind stresses in P-OMIP configuration. Thick black dashed line indicates the stresses added in EX2. Thick black solid line shows the stresses added in EX3. Thick gray solid lines depict the northern and southern end latitudes of the Drake Passage in the model with realistic topography.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Zonal mean westerly wind stresses. Thin black solid line represents the standard westerly wind stresses in P-OMIP configuration. Thick black dashed line indicates the stresses added in EX2. Thick black solid line shows the stresses added in EX3. Thick gray solid lines depict the northern and southern end latitudes of the Drake Passage in the model with realistic topography.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Zonal mean westerly wind stresses. Thin black solid line represents the standard westerly wind stresses in P-OMIP configuration. Thick black dashed line indicates the stresses added in EX2. Thick black solid line shows the stresses added in EX3. Thick gray solid lines depict the northern and southern end latitudes of the Drake Passage in the model with realistic topography.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences between EX2 and EX1 after 200 yr of integration since branching out. (top left) Zonal mean density (σ2) in the North Atlantic; contour interval is 0.05 kg m−3. The dark shadings denote positive values (denser in EX2 than in EX1), and the light shadings denote negative values. (right) Differences in (top) NA cell and (bottom) IP cell; contour interval is 1 Sv. The streamfunction is not defined in the solid black areas. (lower left) Heat flux (into the sea) differences in the North Atlantic; contour interval 50 W m−2. Dark shading corresponds to negative values (more heat loss in EX2 than in EX1) and the light shading positive values. The contours over land denote geographical latitude and longitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences between EX2 and EX1 after 200 yr of integration since branching out. (top left) Zonal mean density (σ2) in the North Atlantic; contour interval is 0.05 kg m−3. The dark shadings denote positive values (denser in EX2 than in EX1), and the light shadings denote negative values. (right) Differences in (top) NA cell and (bottom) IP cell; contour interval is 1 Sv. The streamfunction is not defined in the solid black areas. (lower left) Heat flux (into the sea) differences in the North Atlantic; contour interval 50 W m−2. Dark shading corresponds to negative values (more heat loss in EX2 than in EX1) and the light shading positive values. The contours over land denote geographical latitude and longitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Differences between EX2 and EX1 after 200 yr of integration since branching out. (top left) Zonal mean density (σ2) in the North Atlantic; contour interval is 0.05 kg m−3. The dark shadings denote positive values (denser in EX2 than in EX1), and the light shadings denote negative values. (right) Differences in (top) NA cell and (bottom) IP cell; contour interval is 1 Sv. The streamfunction is not defined in the solid black areas. (lower left) Heat flux (into the sea) differences in the North Atlantic; contour interval 50 W m−2. Dark shading corresponds to negative values (more heat loss in EX2 than in EX1) and the light shading positive values. The contours over land denote geographical latitude and longitude.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Same as Fig. 17, except the differences are between EX3 and EX1.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Same as Fig. 17, except the differences are between EX3 and EX1.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Same as Fig. 17, except the differences are between EX3 and EX1.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Temporal evolution of (a) SH mean heat flux, (b) NA-cell strength, (c) NA mean heat flux, (d) IP-cell strength, and (e) northward Ekman transport at the Cape Horn latitude estimated by the meridional overturning streamfunction. Thin solid lines denote EX1, thick dashed lines are used for EX2, and thick solid lines represent EX3.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Temporal evolution of (a) SH mean heat flux, (b) NA-cell strength, (c) NA mean heat flux, (d) IP-cell strength, and (e) northward Ekman transport at the Cape Horn latitude estimated by the meridional overturning streamfunction. Thin solid lines denote EX1, thick dashed lines are used for EX2, and thick solid lines represent EX3.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Temporal evolution of (a) SH mean heat flux, (b) NA-cell strength, (c) NA mean heat flux, (d) IP-cell strength, and (e) northward Ekman transport at the Cape Horn latitude estimated by the meridional overturning streamfunction. Thin solid lines denote EX1, thick dashed lines are used for EX2, and thick solid lines represent EX3.
Citation: Journal of Physical Oceanography 37, 8; 10.1175/JPO3112.1
Meridional transports of the deep water originated in the west basin (SF35; Sv). Upwelling in the east basin and the Southern Ocean are estimated in the σ–ϕ plane (see text). The northward Ekman transport (ET) at 45°S (Sv) is also shown for each case.
