1. Introduction
One of the key processes controlling the thermohaline circulation is the interaction between the large-scale circulation and motions on smaller spatial scales. Previous studies of this subject have identified two types of interactions as particularly important: diapycnal vertical mixing, commonly associated with the breaking of internal waves in the ocean, and mixing of heat, salt, and momentum by mesoscale eddies. In particular, it has been demonstrated by a number of studies that vertical mixing plays an important role in determining the intensity and general pattern of the thermohaline circulation as it balances the buoyancy gain caused by upwelling deep water (Munk 1966; Bryan 1987; Marotzke 1997; Gnanadesikan 1999).
Mesoscale eddies affect the mean circulation in an important way by transporting heat and salt, and therefore acting to align the isopycnal surfaces with the eddy fluxes. In particular, horizontal eddy fluxes tend to flatten isopycnals, thus releasing available potential energy. In addition, the advection of isopycnal surfaces by eddies can induce large-scale flows that tend to compensate these transports, as demonstrated by both eddy-resolving numerical simulations (Karsten et al. 2002) and coarse-resolution models (Danabasoglu et al. 1994; Gent et al. 1995). Linear stability theories suggest that the mesoscale eddies should be especially active in the areas where the isopycnals slope steeply, and therefore the level of available potential energy is high (Green 1970; Stone 1972; Visbeck et al. 1997). In particular, Wunsch (1999) finds that the ocean eddies transfer a significant amount of heat in the western boundary current regions of all bounded oceans, and additionally in the Southern Ocean.
The western boundary region in the Atlantic and the regions of the Southern Ocean both play an important role in thermohaline circulation. Although the structure of the flow and density is different in each region, the pressure-driven Atlantic circulation is likely to be greatly affected by eddy-induced changes in the density gradients in both. In the vicinity of the Gulf Stream and the deep western boundary current, the meridional transport of mass is effectively controlled by the meridional pressure gradient (Hughes and Weaver 1994; Rahmstorf 1996; Gnanadesikan 1999), which also affects geostrophic zonal flow. Deep horizontal temperature gradients adjust to the intensity of the eddy activity, decreasing with stronger eddy mixing and subsequently reducing eddy heat fluxes (Bryan 1987).
In the Southern Ocean, the eddy activity is likely to result in larger heat and salt fluxes and thus have a stronger effect on the density distribution than in the mid- and low latitudes for two reasons. First, the horizontal gradients of temperature and salinity are very sharp across fronts in thermal wind balance with the Antarctic Circumpolar Current (ACC). Second, the isopycnal surfaces in the Southern Ocean outcrop to the surface and therefore their movement near the surface is restricted by the surface buoyancy fluxes especially in ocean-only models where these fluxes are specified or otherwise constrained. The eddy-induced transports act to counterbalance strong wind-driven overturning in numerical models (Danabasoglu et al. 1994; Gent et al. 1995). Papers of Marshall et al. (2002) and Karsten et al. (2002) describe and quantify a mechanism in which surface buoyancy fluxes and Ekman pumping, acting to overturn the isopycnals, are balanced by transports of baroclinic eddies in the Antarctic Circumpolar Current. Keeling (2002) even suggests a regime in which the eddy transports can completely compensate the Ekman transport and the Pole-to-Pole Atlantic overturning circulation does not exist. Hallberg and Gnanadesikan (2001) show the importance of the transient and stationary eddies in setting the transport of the Antarctic Circumpolar Current in response to changing wind stress.
The significance of the Southern Ocean eddies is not restricted to the Southern Ocean, because of the role of the Southern Ocean in setting the global density structure, which then effects the global thermohaline circulation. Gnanadesikan (1999) in his elegant theoretical model demonstrates that the Southern Ocean processes and mid- and low-latitude upwelling have comparable influence on the overturning of the North Atlantic Deep Water (NADW). Using scaling arguments, he expresses the pycnocline depth in terms of the southern winds and eddies, and low-latitude vertical diffusion. The Southern Ocean is the origin of the Antarctic Intermediate and Subantarctic Mode Water that returns the outflow of the NADW from the Atlantic and preconditions the high-latitude North Atlantic for forming deep water (Gordon et al. 1992). Weijer et al. (1999) in their two-dimensional model show that the overturning circulation in the Atlantic is much more sensitive to changes in the lateral buoyancy fluxes across 30°S than to the changes in surface buoyancy fluxes over the North Atlantic (see also Weijer et al. 2001). Ocean general circulation models (GCMs) also show significant sensitivity to the amplitude of the winds over the Southern Ocean (Toggweiler and Samuels 1995, 1998; McDermott 1996), as well as to the magnitude of the freshwater flux over the Southern Hemisphere (Wang et al. 1999). Keeling (2002) presents a scheme of the Pole-to-Pole Atlantic conveyor driven from the Southern Ocean by the along-isopycnal penetration of relatively fresh waters into the Atlantic. Saenko et al. (2003) demonstrates that the relationship between densities of the Antarctic Intermediate Water and NADW controls the intensity of the latter cell, and can cause its complete shut down.
The detailed understanding of the dynamics of the eddy transports and their interactions with the large-scale circulation is a challenging problem now, and undoubtedly will be for years to come, because of the complexity of the corresponding dynamics and incomplete data coverage of the World Ocean. A more readily addressable problem is estimating a role of the eddy transports in the dynamical processes that are important for setting global circulation and density structures in the context of numerical GCMs. In this study, we will address this problem by running several extended numerical experiments, a task currently feasible only in coarse resolution GCMs. Although quantitative results will need to be refined when new data and techniques become available, qualitative conclusions will remain valid if backed by sound physical analysis.
Numerical ocean models exhibit significant sensitivity of the global density structure and large-scale circulation to the form and magnitude of eddy transports, as has been demonstrated by several previous studies (e.g., Danabasoglu et al. 1994; Danabasoglu and McWilliams 1995; England 1995; Duffy et al. 1997; Weaver and Eby 1997; England and Holloway 1998; Jiang et al. 1999; England and Rahmstorf 1999; Hirst et al. 2000; Kamenkovich et al. 2000). Eddies in coarse resolution GCMs, however, cannot be explicitly resolved and instead are parameterized as subgrid mixing. As an illustrative example, we discuss here two widely used parameterization schemes, namely, horizontal diffusion and the Gent–McWilliams (GM hereinafter) scheme.
Horizontal diffusion is the oldest parameterization scheme and is still used in some studies—it employs Laplacian mixing in the horizontal direction. Horizontal diffusion results in unrealistically large diapycnal fluxes of heat and salt in regions with sloping isopycnal surfaces. The eddy transports across the western boundary currents can affect meridional heat transport (Böning et al. 1995) and induce basin-scale zonal circulation patterns in numerical models, the so-called Veronis effect (Veronis 1975; Toggweiler et al. 1989; Böning et al. 1995; Lazar et al. 1999). The Veronis effect is widely believed to be unrealistic, and this is considered to be a major disadvantage of using horizontal diffusion in a model.
In this study, we aim to understand the sensitivity of the thermohaline circulation to the form, magnitude and spatial distribution of parameterized eddy transports, and identify the dynamical mechanisms that explain this sensitivity. The objective of this study is twofold. First, we identify dynamical consequences of eddy-induced flattening of isopycnals as simulated by the GM scheme, which we will treat as a more realistic representation of eddy effects than the horizontal diffusion. Second, we explain the significant differences in circulation and density structure in the simulations with the GM schemes and horizontal diffusion. We will compare the relative importance of the eddy transports in the Atlantic and the Southern Ocean and demonstrate that although the strength of midlatitude upwelling is affected locally by the Atlantic eddies, the overall intensity of the NADW overturning is basically governed by eddy activity in the Southern Ocean.
The paper is organized as follows. Following a description of the numerical model, we will present a brief summary of differences between two control simulations with the horizontal diffusion and the GM scheme in section 2. The results of numerical experiments aimed at understanding the role of eddy transports in explaining these differences are analyzed in section 3. In section 4, we interpret the numerical results using modified scaling of Gnanadesikan (1999). The summary and discussion are presented in section 5.
2. Model description and control runs
The model employed for this study is based on the Geophysical Fluid Dynamics Laboratory (GFDL) Modular Ocean Model (MOM3) code (Pacanowski and Griffies 1999). The horizontal resolution is 4° in longitude and 3° in latitude. The model has 25 levels in the vertical direction with resolution increasing from 17 m at the surface layer to 510 m at the bottom. The model is global, and the bathymetry is derived from the Scripps topography. The maximum depth is 5 km; the minimum depth is 50 m. No-slip conditions for velocity are applied at all lateral walls; linear bottom drag is applied at the bottom. Boundary conditions for tracers are insulating at the solid lateral walls and bottom. Either the horizontal diffusion or the GM scheme is used (see Table 1). Vertical diffusivity varies from 0.25 × 10−4 m2 s−1 at the surface to 1.0 × 10−4 m2 s−1 at the bottom. This profile reflects the increase of the vertical mixing in the direction from thermocline to the deep ocean (Bryan and Lewis 1979), and the intensification of mixing by rough bottom topography (Polzin et al. 1997). While it is surely possible that the vertical mixing and eddy transports interact, we will separate their effects by keeping the vertical mixing fixed throughout all the experiments, reserving consideration of their interaction to future work.
To differentiate between water masses of Northern Atlantic and Southern Ocean origins, we employ a technique of “dye tracers” (Cox 1989). We use three dyes, each having a source at the surface in one region and a sink at the surface over the rest of the domain. The restoring boundary conditions with a restoring time scale of 50 days over 50 m are used following Goodman (2001) to represent these sources and sinks. The first “dye” (“Southern Ocean tracer” hereinafter) is restored to 1 at the surface over the Southern Ocean (south of 30°S), while it is restored to 0 in the rest of the surface domain. The third dye (“North Atlantic tracer”) has a source in the surface of the North Atlantic (north of 50°N). The second dye has a source in the remainder of the domain. As Cox (1989) pointed out, these tracers in steady state possess a very useful property, namely, that at each grid point the sum of their concentration is unity. It is thus possible to determine what proportion of any grid box is occupied by water of the Southern Ocean or North Atlantic origin.
We then identify a water parcel as that of the Southern Ocean origin, if the concentration of Southern Ocean tracer is the highest among all other dyes. This dyeing is used for identifications of intermediate water and Antarctic Bottom Water (AABW) inflows as water of Southern Ocean origin at middepth and near the bottom, respectively. The definition of the NADW is, however, more ambiguous because of strong mixing between the NADW and the rest of the water in the Atlantic. For example, substantial vertical mixing between the southward flowing deep water leaving the Atlantic and northward flowing intermediate water manifests itself in the too shallow a maximum in Southern Ocean tracer. We, therefore, choose to identify a water parcel as of the NADW origin if the concentration of the Southern Ocean tracer is not the highest among all colors and the parcel lies below 750 m in the Atlantic.
a. Control simulations: Summary of the differences
Two control simulations, one with horizontal diffusion (experiment HDC), another with the GM scheme (experiment GMC), were performed (Table 1). The differences between the two cases are similar to those reported in the previous studies (Danabasoglu and McWilliams 1995; Duffy et al. 1997; Weaver and Eby 1997; Hirst and McDougall 1998; Jiang et al. 1999; England and Rahmstorf 1999; Kamenkovich et al. 2000, 2002; Hirst et al. 2000) and are only briefly summarized here. The reader is referred to the aforementioned studies for more detailed description of these effects.
Both the zonally averaged meridional velocities and the thickness of the NADW overturning cell are larger in HDC than in GMC (Figs. 1a,b). Most of the subsurface Atlantic Ocean in the low and midlatitudes is significantly denser in the GMC (Fig. 2), with the largest difference in the deep ocean (see, e.g., Duffy et al. 1997). Because of the Veronis effect (Böning et al. 1995), almost 7 Sv (1 Sv ≡ 106 m3 s−1) upwells in the Northern Hemisphere Atlantic in HDC (Fig. 1b; Table 2) and a little more than 60% of the NADW formed in the Atlantic leaves the basin. By contrast, only 2.5 Sv upwells in the North Hemisphere in GMC, and 10.5 Sv of the Atlantic origin leaves the basin. The pycnocline (Table 2) is also significantly deeper1 in HDC, consistent with stronger NADW overturning (Gnanadesikan 1999).
The Southern Ocean isopycnals tend to be more horizontal in GMC than in HDC (Fig. 3); this difference manifests itself in smaller meridional gradients in the subsurface Southern Ocean and weaker geostrophic currents in the ACC. The meridional circulation in the Southern Ocean driven by northward Ekman flows acts to overturn isopycnals and eddies in the Southern Ocean work to counterbalance this tendency (Gent et al. 1995; Karsten et al. 2002). This effect of eddies is apparently stronger in the GM case because of the tendency of the GM velocities to flatten the isopycnals. In contrast, the isopycnal surfaces are oriented nearly vertical in HDC; HDC is also characterized by vigorous convection in the Southern Ocean (Danabasoglu and McWilliams 1995). The AABW (Figs. 1 and 4) is noticeably denser in GMC than in HDC (Fig. 2), which is explained by reduced magnitudes of cross-isopycnal mixing in the Southern Ocean in the former case. Stronger mixing acts to erode the density of this water mass in HDC (England 1993).
3. Relative role of the eddy transports in the Southern Ocean and Northern Hemisphere
Our goal in this section is to identify mechanisms controlling the sensitivity of the global density and flow structure to parameterization of horizontal eddy transports (section 2), to evaluate the relative roles of these mechanisms in the Southern Ocean and Northern Hemisphere Atlantic, and thus to explain the difference between GMC and HDC.
We note that an explanation of the reduction in the NADW overturning in GMC based on the dependence of the overturning on vertical diffusivity has to be dismissed beforehand. First, the vertical diffusion is the same in both experiments with some modest differences due to the different depth of the NADW upwelling and the nonconstant profile of vertical diffusivity. Second, the zonal-mean Eulerian velocities change very little in the North Atlantic with the addition of the eddy-induced transport velocities u* and υ* [(1)]. Therefore, the contribution of the eddy-induced transports of heat into the total vertical advective heat flux at midlatitudes is relatively small, and any compensation of diffusive fluxes by the GM heat fluxes cannot be significant. Alternative explanations for the reduction in overturning intensity are given below.
Since the thermohaline circulation is basically driven by density gradients, the reduction of the NADW formation in GMC relative to HDC must be tied to differences in the density structure of the two simulations. The difference, however, is very small near the surface, where variations in density structure are constrained by the surface boundary conditions. In the convective region north of 55°N, the difference is also small at all depths since the entire water column is strongly ventilated from the surface at the same constant specified density. Because of this nearly fixed density at the surface and at all depths at high latitudes, any change in the subsurface mid- and low-latitude density will lead to changes in the depth of the pycnocline and in the meridional pressure gradient. For example, increases in low- and midlatitude density in GMC would shoal the pycnocline and decrease the pressure gradient, resulting in a weaker meridional flow (Hughes and Weaver 1994; Rahmstorf 1996; Gnanadesikan 1999).
In the rest of this section, we discuss possible ways in which eddy transports can affect the meridional density structure and the meridional mass transport in the North Atlantic. Namely, we consider eddy effects in 1) the low and midlatitudes of the Northern Hemisphere Atlantic, 2) the Southern Ocean at intermediate depths, and 3) Antarctic Bottom Water.
a. Northern Hemisphere Atlantic
The meridional flow at low and midlatitudes in the Northern Hemisphere Atlantic can be affected locally by the eddy-driven horizontal exchanges between the deep western boundary current and lighter interior in two major ways, as indicated in the introduction. First, the eddy transports can change the horizontal density distribution, thus affecting geostrophic currents. Second, strong upwelling due to the unrealistic Veronis effect might also draw more water from the North Atlantic, thus directly influencing the NADW production.
In this section, we isolate the effect of eddy transports in the low and midlatitudes of the Northern Hemisphere Atlantic, where practically all of the upwelling takes place (Fig. 1). We conduct an experiment, GMNH, in which both mixing schemes are present but with the mixing coefficients made spatially varying. In particular, the isopycnal diffusion and the GM coefficients are set to 103 m2 s−1 in the Northern Hemisphere of the Atlantic but zero elsewhere; whereas the horizontal diffusion coefficient is 0 in the Northern Hemisphere of the Atlantic and is set to 103 m2 s−1 in the rest of the domain. This experiment is effectively HDC but with GM replacing HD in the North Atlantic only. We expect experiment GMNH to reduce the upwelling relative to HDC, so that relative effect of Northern Hemisphere upwelling can be assessed. In addition, we expect isopycnal surfaces in the Northern Hemisphere to be more flat in GMNH than in HDC because of the GM eddy transport's effects on density.
The maximum zonally averaged meridional overturning in the Northern Hemisphere is very close to that in HDC and is equal to 19.25 Sv (Fig. 5a and Table 2). The upwelling of the NADW in the Northern Hemisphere is, however, greatly reduced in comparison with HDC and reaches only 2.25 Sv (Table 2), an amount close to that in GMC. Weak upwelling in the Atlantic combined with vigorous NADW formation result in a strong outflow of the deep water of the Atlantic origin; 15.5 Sv leaves the Atlantic basin. The northward inflow of water of the Southern Ocean origin at middepth is also very strong and amounts to 15.75 Sv; the small difference between inflow and outflow of the NADW (Table 2) is explained by the mixing of the NADW with the AABW. Vigorous overturning of NADW and active volume exchanges between the Southern Ocean and the South Atlantic in the upper 2500 m can be contrasted to the markedly weakened circulation of the AABW in the Atlantic. Only 0.5 Sv of the northward flowing bottom water of Southern Ocean origin crosses 30°S and this bottom water does not reach the equator at all, thus overturning entirely in the Southern Hemisphere.
The density in this experiment is somewhat larger everywhere than in HDC, but noticeably smaller than in GMC; see Figs. 6a and 6b and the Northern Hemisphere values in Fig. 7. The difference in density between GMNH and HDC is very small in the upper 1000 m of the Atlantic, despite the fact that the entire Northern Hemisphere Atlantic is governed by the GM mixing and consequently the upwelling is very weak. The pycnocline depth is 100 m (9.5%) shallower than in HDC (Table 2). The meridional pressure gradient in GMNH is therefore somewhat weaker, which can support a weaker meridional mass transport according to Gnanadesikan (1999).
The western boundary current in the upper 1000 m in GMNH are indeed slower than in HDC (Fig. 8e). The zonally integrated transport is nevertheless as large in GMNH as in HDC. This is explained by a more unidirectional western boundary current. Similarly to GMC, the isotachs in GMNH (Fig. 8c) are more horizontal in the western boundary region than in HDC and, unlike HDC, the horizontal recirculation just below 1000 m is weak (Figs. 8b,c). The local GM mixing in GMNH is more effective in flattening isopycnals in the Northern Hemisphere, which reduces the curvature of the isopycnals and, consequently, the shear in meridional velocity.
The total mass transport by the NADW cell also depends on its vertical extent (Gnanadesikan 1999; Kamenkovich and Goodman 2000). In particular, the overturning cell in the North Atlantic in GMNH penetrates deeper than in HDC (Figs. 1b and 5a), which is consistent with the weaker AABW inflow. The AABW in GMNH is very similar in density to that in HDC, since the horizontal eddy transports in the Southern Ocean are the same; the density at depth in the Northern Hemisphere is, in contrast, higher than that in HDC (Fig. 6b). The resulting deep meridional density contrast between the AABW and deep waters in the Northern Hemisphere nearly reverses, which prevents the meridional flow of the bottom water from penetrating far into the Atlantic (Kamenkovich and Goodman 2000). Weakening of the AABW flow leaves more space for the circulation of NADW, thus acting to enhance the transport of the latter.
We therefore conclude from this experiment that although horizontal eddy transports in the Northern Hemisphere control the upwelling rates and influence the horizontal structure of the currents, the NADW formation rate is mostly affected by the eddy dynamics in the Southern Hemisphere. The southward NADW outflow from the Atlantic is determined by the difference between the NADW formation rate and the total transport of the recirculating upwelling. The density structure in the Northern Hemisphere is mostly responsive to changes in the eddy transports south of the equator, but is also influenced by the local eddy fluxes. The vertical extent of the overturning cell is affected by the density contrast between the NADW and the AABW. The next step is to directly estimate the role of the Southern Ocean eddy transports in changing density and flow structure in the model.
b. Southern Ocean
In the Southern Ocean, the more horizontal orientation of the isopycnals in GMC leads to isopycnals from the subsurface Southern Ocean extending farther northward (Figs. 2 and 3), thus making the more northern part of the Southern Ocean denser. As a result, the inflow of the Southern Ocean water at intermediate depths (around 700 m) is denser in GMC, even though the magnitude of the inflow is similar between the two cases (Table 2). In particular, the difference in the density of 0.2 kg m−3 between HDC and GMC is characteristic for most of the Atlantic at intermediate depths south of 50°N and can be traced to the Southern Ocean being both colder and saltier in GMC at the intermediate depth. Increased low-latitude density results in shoaling of the pycnocline and in decreasing of the meridional pressure contrast with high latitudes. We next demonstrate the importance of this effect for the NADW circulation.
We conduct the experiment GMSO, in which the isopycnal diffusion and the GM coefficients are set to 103 m2 s−1 in the Southern Ocean (south of 30°S) but 0 elsewhere, whereas horizontal diffusion coefficient is set to 0 in the Southern Ocean and to 103 m2 s−1 in the rest of the domain (Table 1). This is effectively the HDC experiment but with horizontal diffusion replaced by the GM scheme in the Southern Ocean only. Thus, in this experiment, the eddy transports in the Southern Ocean are determined by the GM scheme whereas the eddy mixing in the rest of the global domain is represented by horizontal diffusion.
The new equilibrium state is reached with the meridional circulation shown in Fig. 5b. The first thing to notice is that the maximum value of the meridional overturning is remarkably close to that in the GMC run (around 12.5 Sv in both). Although the large horizontal eddy transports associated with horizontal diffusion act in the entire Atlantic basin north of 30°S, they seem to have little impact on the total amount of the NADW formed in the Northern Hemisphere. By contrast, the mid- and low-latitude upwelling of the NADW in the Northern Hemisphere of the Atlantic basin is controlled by local horizontal eddy transports (the Veronis effect). Approximately 5.75 Sv of deep water upwells in the area (Table 2) only 1 Sv less than in the HDC. The upwelling in GMC, in contrast, is significantly smaller and amounts to only 2.5 Sv in total.
This strong “short circuiting” of the overturning in the Northern Hemisphere combined with a modest amount of NADW produced at high latitudes results in a very weak southward outflow of the NADW in the South Atlantic. Only 6.5 Sv of the water of the Atlantic origin enters the Southern Ocean at 30°S, significantly less than the outflows in either the HDC or GMC experiments. The combination of strong upwelling, weak NADW formation and weak outflow of NADW from the Atlantic makes the circulation pattern in this experiment less realistic than in either the HDC or the GMC experiments. As in the previous experiment, we observe that the outflow is more sensitive to the magnitude of the upwelling than is the NADW formation rate.
The pycnocline depth (725 m) is much closer to the value in GMC (650 m) than that in HDC (1050 m). The two effects of the eddy transports on the density structure in the upper part of the ocean compete in GMSO. The density of the northern part of the Southern Ocean in GMSO has a similar structure to that in GMC, since the Southern Ocean is governed by the same GM eddy dynamics. The Southern Ocean isopycnals are more horizontal than in either HDC or GMNH. The resulting density of the intermediate water entering the South Atlantic is higher than in HDC and is very close to that in GMC. The inflow of denser water makes subsurface low-latitude waters denser than in HDC, lifts the pycnocline, and ultimately decreases the meridional pressure contrast with high latitudes. On the other hand, the large horizontal eddy transports at mid- and low latitudes in GMSO and HDC lead to similar values of upwelling and cross-isopycnal fluxes of heat and salt, which acts to bring the Northern Hemisphere density in GMSO closer to that in HDC.
Local recirculation of water leads to similar zonally integrated transports between GMC and GMSO despite swifter currents in the former case. Larger meridional pressure gradients in GMSO result in somewhat stronger meridional transport of the western boundary currents (Gnanadesikan 1999) as compared with that in GMC (Fig. 8e). However, significant horizontal recirculation is evident just below 1000 m in experiments HDC and GMSO with a noticeable southward flow at the same depth as the lower portion of the northward flowing western boundary current. Horizontal diffusion permits larger curvature of the isopycnal surfaces and horizontal velocity shear in the Northern Hemisphere is greater than in GMC or GMNH. Strong upwelling in GMSO also contributes to stronger zonal flow and larger zonal shear in the meridional velocity.
The vertical extent of the NADW cell in GMSO is smaller than in either GMC or HDC cases (Figs. 1a,b and Fig. 5b). The total thickness of the NADW cell is affected by the density of the deep ocean and the vertical extent of AABW, and AABW therefore can influence the overturning of NADW (England 1993; Gnanadesikan 1999; Kamenkovich and Goodman 2000). AABW is subject to a limited amount of mixing by eddy transports in GMSO and GMC, and is as dense as it enters the Atlantic at 30°S in GMSO as in the GMC (Fig. 6c). The deep ocean in GMSO, as a result, has a very similar density to that in GMC. In contrast, the water of the upper Atlantic in GMSO is lighter than that in the GMC (Fig. 6c), and therefore AABW can restrict NADW to even shallower depths than in GMC, reducing the strength of the NADW cell and increasing the transport of the AABW cell (England 1993; Kamenkovich and Goodman 2000). This vertical contraction of the NADW cell may mask any strengthening of the cell by other factors. It is therefore difficult to estimate a role of the midlatitude eddy transports in altering the NADW circulation before the analysis of the relative role of AABW in the following section 3c.
The results so far suggest that the main effect of the GM eddy transports in the Southern Ocean is to enhance stratification in the Southern Ocean and increase density of the inflowing intermediate water into the Atlantic which subsequently reduces the formation of NADW. To confirm this conclusion, we carry out an additional experiment, SOdense, in which the HDC experiment is repeated but with the Southern Ocean density increased. We artificially raise buoyancy loss over the surface of the Southern Ocean by decreasing restoring temperature in Tobs in (2) by 1°C and increasing restoring salinity Sobs in (3) by 0.2 psu everywhere between 66° and 30°S. The density in the upper 1000 m of the South Hemisphere Atlantic is now higher than in HDC (Fig. 9a) and is in fact very close to the corresponding values in GMSO (Fig. 6d). In contrast, the density and transport of the inflowing AABW hardly differ from that in HDC. The resulting maximum NADW overturning is 15.25 Sv (Fig. 9b and Table 2). Higher density of the Southern Ocean in this experiment is clearly the only reason for the NADW circulation being weaker than in experiment HDC, since the horizontal mixing is the same.
We have thus observed that the parameterized eddy transports in the Southern Ocean strongly influence meridional density structure and consequently control the NADW formation rate in the Atlantic. A denser inflow of the intermediate water shoals the low-latitude pycnocline and reduces the meridional density contrast between low latitudes and the northern North Atlantic. The Northern Hemisphere eddy transports due to horizontal diffusion have two counteracting effects on the NADW formation: They decrease density at low latitudes, but they also lead to a larger horizontal shear in velocity and stronger recirculation in the horizontal plane. As in section 3a, AABW controls vertical penetration of the NADW cell, which affects the transport of the latter.
c. Role of AABW
This section aims at explicitly estimating the role of AABW in our study by conducting experiments in which the inflow of AABW into the Atlantic is absent. Although the magnitude of the inflow of AABW to the North Atlantic is similar between GMC and HDC (Table 2), AABW in GMC extends to shallower depth and recirculates differently in the two cases. Denser bottom water in GMC acts to constrain NADW from reaching larger depths, and thus affects the vertical extent of the NADW cell and its meridional transport (Gnanadesikan 1999; Kamenkovich and Goodman 2000). AABW is very sensitive to mixing in the Southern Ocean (England 1993) and the deep Atlantic (Kamenkovich and Goodman 2000), and can therefore be an important factor in explaining the sensitivity of NADW to the Southern Ocean eddy transports.
We initially attempted to stop the AABW formation by decreasing surface density near the Antarctic coast and thereby stopping the convection that leads to the formation of the AABW. This, however, failed to entirely cease the AABW propagation into the Atlantic. Although the AABW formation was easily shut off in the HD case, the formation of this water mass is remarkably robust in the GM simulations under any reasonable perturbation in the surface density. Large alterations of the surface density led to significant changes in the meridional density structure and in the significant reduction in the transport of the ACC. To avoid making Southern Ocean density and flow structure too unrealistic, we therefore chose to topographically block the AABW penetration into the Atlantic. We put a zonal ridge at 28.5°S in the Atlantic, which is 2240 m deep and stretches across the entire Atlantic. Note that the depth of the ridge corresponds to the depth of the NADW outflow in HDC, which is the deepest between the two control experiments. In addition, the wintertime enhancement of the surface salinity near the Antarctic coast which was added to improve the AABW formation and sinking (see section 2) was also removed.
The resulting circulation exhibits no AABW transport into the Atlantic in either HD or GM simulations (expts HDnB and GMnB; see Figs. 10a,b and Table 2). The NADW cell in either case almost reaches the bottom in the Northern Hemisphere. The total transport of the meridional overturning and of the outflow of NADW in both GMnB and HDnB increases as compared with the control simulations GMC and HDC because more space is available for the circulation in the absence of AABW. In contrast, the amount of the Northern Hemisphere upwelling hardly changes at all in either case. The deep ocean becomes significantly less dense in the absence of AABW. Because AABW is denser in GMC than in HDC, the absence of AABW in GMnB reduces the density difference between GMnB and HDnB (cf. Fig. 11a and Fig. 2d). The reduction is, however, modest, which suggests that factors other than AABW are as important in making the deep water in GMC denser than in HDC. First, the reduction in the NADW formation cools the deep ocean (Manabe and Stouffer 1988). Second, the flattening effect that mesoscale eddies have on isopycnals raises isopycnal surfaces in the North Atlantic thus increasing density of the deep ocean. Third, the stronger ventilation of the deep ocean, through mixing along the isopycnal surfaces that outcrop in the high latitudes of the North Atlantic, cools the deep water.
We now repeat the experiments GMNH and GMSO without AABW (experiments GMNHnB and GMSOnB). The main results from the experiments with AABW (GMNH and GMSO of sections 3a and 3b) are reproduced here with only minor differences. The NADW overturning is similar between the experiments with the same eddy parameterizations in the Southern Ocean (GMnB and GMSOnB, HDnB and GMNHnB), whereas the Northern Hemisphere upwelling is similar between experiments with the same schemes in the Northern Hemisphere (GMnB and GMNHnB, HDnB and GMSOnB). A peculiar feature of the circulation in GMSOnB is a meridional overturning cell in the deep ocean (Fig. 10c). The concentration of the Southern Ocean tracer in this cell does not exceed 0.15; the streamlines of zonal-mean currents in Fig. 10d are closed within the Atlantic. The overturning cell is therefore a local recirculation most likely driven by the deep Veronis effect of the large horizontal diapycnal mixing in the Atlantic north of the Southern Ocean.
The density differences between GMNHnB and HDnB, and between GMSOnB and GMnB, are presented in Figs. 11b and 11c. The density differences in the upper 2000 m between all four cases without AABW are qualitatively similar to the differences between experiments with AABW (cf. Figs. 11 and 6). The pycnocline shoals when GM eddy transports are present in the Southern Ocean, whereas the Northern Hemisphere cross-isopycnal eddy fluxes of the horizontal diffusion tend to deepen the pycnocline. It is noteworthy that the range for the pycnocline depths in the experiments without AABW is narrower than in the experiments with AABW. The lack of AABW permits NADW to reach the bottom, which apparently limits the sensitivity of the pycnocline depth to changed physics in our model and complicates calculation of D using (4).2 The introduction of the GM mixing to the Northern Hemisphere in GMNHnB changes the upper-layer density only slightly in comparison with HDnB (Fig. 11b). In contrast, the water in the deeper layers of the Northern Hemisphere becomes denser, because of the increase in the density of NADW. The density in GMSOnB is intermediate in values in HDnB and GMnB for all depths including bottom layers (Fig. 11c).
We conclude that the role of AABW in explaining the sensitivity of the NADW formation to eddy transports in the Southern Ocean is smaller than the role of the intermediate water. Some differences between experiments with AABW and those without AABW are, however, noticeable. In particular, the density of the upper 1000 m of the Atlantic is less affected by the form of parameterized eddy mixing in the Northern Hemisphere in the experiments without AABW. The absence of AABW makes the density of the water below the main pycnocline more sensitive to the Northern Hemisphere eddy mixing.
4. Scaling for the role of the Southern Ocean and Northern Hemisphere eddy transports
In the previous sections, we have described how the flow and density structure respond to changes in the form and magnitude of parameterized eddy transports. We have concluded that Southern Ocean eddies strongly affect thermohaline circulation, and that the intermediate water plays a major role in linking the Southern Ocean and the North Atlantic. In this section, we will use the scaling laws for the thermohaline circulation originally derived by Gnanadesikan (1999) to quantitatively interpret the results. In particular, our goal is to understand why the midlatitude diapycnal fluxes, while controlling the mid- and low-latitude upwelling, have a limited role in affecting the depth of the low- and midlatitude pycnocline and the intensity of NADW formation.
Equation (10) suggests that a relative role of the Veronis effect in causing changes in the pycnocline depth can be measured by a nondimensional parameter ε = (kH/AI)(LyLn/LxLm). This ratio is 0.13 in our study, and thus the importance of the Veronis effect for the depth of the pycnocline and NADW formation rate is small compare to the role of the Southern Ocean eddies. It is noteworthy, that the significance of the Veronis effect decreases with increasing width of the Southern Ocean in a model, which emphasizes the importance of having a realistic geometry in a numerical study.
Our next step is to compare solutions of (10) with our numerical results. We take the nondimensional constant C to be 0.35. The resulting values for Tn, Tu, and D are presented in Table 3. These numbers can be compared with the four experiments of sections 2–3: GMC (AI = 1000, kH = 0), HDC (AI = 0, kH = 1000), GMSO (AI = 1000, kH = 1000), and GMNH (AI = 0, kH = 0).3 A large difference between NADW formation for cases with AI equal to 0 and those with non-0 AI is evident. In agreement with Gnanadesikan (1999), the results confirm the strong sensitivity of the pycnocline depth and the NADW formation to the intensity of the eddy transports in the Southern Ocean. The scaling also supports the conclusions of section 3 that the cross-isopycnal horizontal mixing in the Northern Hemisphere that causes the Veronis effect controls upwelling, but has a modest effect on the pycnocline depth and the NADW overturning.
5. Discussion and summary
This study describes an important mechanism that allows eddy-controlled changes in the density structure in the Southern Ocean to affect the meridional circulation in the Atlantic. We find that eddy transports in the Southern Ocean can considerably modify the density of the intermediate water entering the Atlantic from the south. The intermediate water influences the low-latitude stratification in the upper ocean, which changes the meridional pressure difference and therefore the intensity of the meridional overturning in the Atlantic basin. This mechanism explains, in particular, significant differences in the flow and density structure between numerical simulations with the Gent–McWilliams mixing parameterization and those with horizontal diffusion. To separate the roles of the eddy transfer in the Southern Ocean and in the Northern Hemisphere in maintaining thermohaline circulation, we conducted a set of numerical experiments, in which the horizontal diffusion is replaced by the GM mixing in either the Southern Ocean or in the Northern Hemisphere. In these experiments, we varied only the horizontal eddy fluxes, fixing all other parameters. For example, we kept the vertical diffusion fixed in order to separate effects of the mesoscale eddies and that of vertical mixing by the internal waves. We did not carry out a complete sensitivity study to a wide range of diffusion parameters, concentrating instead on qualitative effects of eddy transports.
The meridional flattening effect of the GM mixing tends to shoal isopycnals that outcrop at high latitudes in the Southern Ocean, which leads to the denser inflow of the waters at intermediate depths. The Southern Ocean isopycnals outcrop at the surface while their surface movement is restricted by surface buoyancy fluxes. Any meridional displacement of subsurface isopycnals therefore changes vertical stratification to a much larger degree than in the subsurface layers in mid- and low latitudes. For example, a northward displacement of Southern Ocean isopycnal surfaces raises subsurface isopycnals, which increases the density at intermediate depths where the water masses entering the Atlantic basin from the south originate. The consequent increase in the low- and midlatitude density leads to the shallower low-latitude pycnocline, smaller meridional pressure gradient, and weaker meridional flow (Gnanadesikan 1999). To confirm our conclusions, we conducted a horizontal diffusion experiment, in which increase in the density of the intermediate water is achieved by changing surface boundary conditions. The resulting NADW overturning is substantially weakened in comparison with the control run.
Parameterized eddy transports and upwelling at low and midlatitudes in the Atlantic also have a noticeable effect on the density and flow structure. Unrealistically large cross-isopycnal fluxes of heat caused by horizontal diffusion induce strong upwelling, which short-circuits a portion of the total volume of NADW formed in the high latitudes—the well-documented Veronis effect. Scaling arguments suggest that this upwelling is at least as strong as the upwelling induced by vertical diffusion. The Veronis effect act to deepen the pycnocline and enhance the NADW formation, although the effect is considerably smaller than that of the intermediate water, according to both the numerical results and theoretical scaling. The upwelling caused by horizontal mixing has little direct influence on the amount of the NADW formation and therefore indirectly affects the outflow of NADW from the Atlantic. The magnitude of the exchange of deep water between Southern Ocean and the Atlantic is not directly determined by the eddy transports in the Southern Ocean, but rather appears to be merely a residual between the NADW formation and the Atlantic upwelling. The parameterized eddy transports also affect horizontal flow structure in the low and midlatitudes in the Northern Hemisphere. Isopycnals flattened by the GM eddy transports correspond to the slower geostrophic currents, but at the same time to the more unidirectional meridional flow. In contrast, steeper isopycnals correspond to swifter currents, but can also lead to more significant horizontal recirculation.
The parameterized eddy transports of heat and salt strongly affect the Antarctic Bottom Water (AABW), whose properties in turn influence the vertical penetration of the NADW cell. For example, increased density contrast between the AABW and NADW cells in the run with the GM scheme in the Southern Ocean acts to shoal the latter cell and decrease its transport. The situation is reversed in the run with the GM scheme in the North Hemisphere, in which NADW is the deepest among all cases. The role of AABW in this study is smaller than that of the intermediate water, however, as demonstrated by our experiments with the blocked AABW propagation into the Atlantic. In the absence of the AABW flow, the effect of the low and midlatitudes on the NADW formation becomes more visible, but the difference with the experiments with AABW is not substantial.
The dependence of the meridional circulation and density structure in the Atlantic on the density of the inflowing intermediate water illustrates an important dynamical role of the Southern Ocean and offers a new interpretation of some previous results. The sensitivity of the meridional overturning on the magnitude of the Southern winds (Toggweiler and Samuels 1995, 1998; McDermot 1996), and on the magnitude of the buoyancy fluxes over the Southern Ocean (Wang et al. 1999) may both be explained by modifications in the density structure in the Southern Ocean and resulting changes in the density of the intermediate water. Thus, the steeper orientation of isopycnals by stronger winds or larger surface buoyancy fluxes decreases the stratification and therefore tends to decrease the density of the intermediate water and strengthen the overturning in the Atlantic, according to the mechanism described in this study. Böning et al. (1995) use an Atlantic-only model with density at the southern boundary prescribed to observations throughout the entire oceanic depth, and report almost no difference in the maximum meridional overturning between their horizontal diffusion and GM simulations despite much larger North Hemisphere upwelling in their horizontal diffusion case. The specification of density with depth at the southern boundary of the Atlantic in Böning et al. (1995) effectively eliminates the effect of the eddy-controlled Southern Ocean densities demonstrated here and therefore removes the apparent contradiction with our results.
This study emphasizes the necessity of a realistic representation of the formation and propagation of the Circumpolar Mode Water and Antarctic Intermediate Water. An increased resolution, inclusion of the sea-ice processes, and a representation of the topographic overflows, all should improve the simulation of AABW. Variations in the surface density in the Southern Ocean were severely restricted in this study for the simplicity of the analysis, whereas they too can be greatly affected by the eddy fluxes and small-scale air–sea exchanges, all of which needs to be studied in the context of a fully coupled model.
Acknowledgments
We thank Anand Gnanadesikan for his suggestions on the experimental strategy and on the interpretation of results. Comments from Mathew England, Peter Gent, and anonymous reviewers helped to improve this article and are gratefully acknowledged. This work was supported by the National Science Foundation through the Office of Polar Programs under grant 0126208, and by a grant from the NOAA Office of Global Programs to the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) Center for Science in the Earth System. This publication is funded by JISAO under NOAA Cooperative Agreement No. NA17RJ11232.
REFERENCES
Böning, C. W., W. R. Holland, F. O. Bryan, G. Danabasoglu, and J. C. McWilliams, 1995: An overlooked problem in model simulations of thermohaline circulation and heat transport in the Atlantic Ocean. J. Climate, 8 , 515–523.
Bryan, F., 1987: Parameter sensitivity of primitive equation ocean general circulation models. J. Phys. Oceanogr, 17 , 970–985.
Bryan, F., and L. J. Lewis, 1979: A water mass model of the world ocean. J. Geophys. Res, 84 , 2503–2517.
Cox, M. D., 1989: An idealized model of the World Ocean. Part I: The global-scale water masses. J. Phys. Oceanogr, 19 , 1730–1752.
Danabasoglu, G., and J. C. McWilliams, 1995: Sensitivity of the global thermohaline circulation to parameterizations of mesoscale tracer transports. J. Climate, 8 , 2967–2986.
Danabasoglu, G., J. C. McWilliams, and P. R. Gent, 1994: The role of mesoscale tracer transports in the global ocean circulation. Science, 264 , 1123–1126.
Duffy, P. B., K. Caldiera, J. Selvaggi, and M. I. Hoffert, 1997: Effects of subgrid scale mixing parameterizations on simulated distributions of natural C14, temperature, and salinity in a three-dimensional ocean general circulation model. J. Phys. Oceanogr, 27 , 498–523.
England, M. H., 1993: Representing the global-scale water masses in ocean general circulation models. J. Phys. Oceanogr, 23 , 1523–1551.
England, M. H., 1995: Using chlorofluorocarbons to assess ocean climate models. Geophys. Res. Lett, 22 , 3051–3054.
England, M. H., and G. Holloway, 1998: Simulations of CFC content and water mass age in the deep North Atlantic. J. Geophys. Res, 103 , 15885–15901.
England, M. H., and S. Rahmstorf, 1999: Sensitivity of ventilation rates and radiocarbon uptake to subgrid mixing in ocean models. J. Phys. Oceanogr, 29 , 2802–2827.
Gent, P. R., and J. C. McWilliams, 1990: Isopycnal mixing in ocean circulation models. J. Phys. Oceanogr, 20 , 150–155.
Gent, P. R., J. Willebrand, T. J. McDougall, and J. C. McWilliams, 1995: Parameterizing eddy-induced tracer transports in ocean circulation models. J. Phys. Oceanogr, 25 , 463–474.
Gnanadesikan, A., 1999: A simple predictive model for the structure of the oceanic pycnocline. Science, 283 , 2077–2079.
Gnanadesikan, A., R. D. Slater, and B. L. Samuels, 2003: Sensitivity of water mass transformation and heat transport to subgridscale mixing in coarse-resolution ocean models. Geophys. Res. Lett.,30, 1967, doi:10.1029/2003GL018036.
Goodman, P., 1998: The role of North Atlantic deep water formation in an OGCM's ventilation and thermohaline circulation. J. Phys. Oceanogr, 28 , 1759–1785.
Goodman, P., 2001: Thermohaline adjustment and advection in an OGCM. J. Phys. Oceanogr, 31 , 1477–1497.
Gordon, A. L., R. F. Weiss, W. M. Smethie Jr., and M. J. Warner, 1992: Thermocline and intermediate water communication between the South Atlantic and Indian Oceans. J. Geophys. Res, 97 , 7223–7240.
Green, J. S., 1970: Transfer properties of the large-scale eddies and the general circulation of the atmosphere. Quart. J. Roy. Meteor. Soc, 96 , 157–185.
Hallberg, R., and A. Gnanadesikan, 2001: An exploration of the role of transient eddies in determining the transport of a zonally reentrant current. J. Phys. Oceanogr, 31 , 3312–3330.
Hirst, A. C., and T. J. McDougall, 1998: Meridional overturning and Dianeutral transport in a z-coordinate model including eddy-induced advection. J. Phys. Oceanogr, 28 , 1205–1223.
Hirst, A. C., S. P. O'Farrell, and H. B. Gordon, 2000: Comparison of a coupled ocean–atmosphere model with and without oceanic eddy-induced advection. Part I: Ocean spinup and control integrations. J. Climate, 13 , 139–163.
Hughes, T. M., and A. J. Weaver, 1994: Multiple equilibria of an asymmetric two-basin ocean model. J. Phys. Oceanogr, 24 , 619–637.
Jiang, S., P. H. Stone, and P. Malanotte-Rizzoli, 1999: An assesment of the GFDL ocean model with coarse resolution. Part I: Annual-mean climatology. J. Geophys. Res, 104 , 25623–25646.
Kamenkovich, I., and P. Goodman, 2000: The dependence of the AABW formation on vertical diffusivity. Geophys. Res. Lett, 27 , 3739–3742.
Kamenkovich, I., J. Marotzke, and P. H. Stone, 2000: Factors affecting meridional heat transport in ocean general circulation models. J. Phys. Oceanogr, 30 , 175–194.
Kamenkovich, I., A. Sokolov, and P. H. Stone, 2002: An efficient climate model with a 3D ocean and statistical dynamical atmosphere. Climate Dyn, 19 , 585–598.
Karsten, R., H. Jones, and J. Marshall, 2002: The role of eddy transfer in setting the stratification and transport of a circumpolar current. J. Phys. Oceanogr, 32 , 39–54.
Keeling, R. F., 2002: On the freshwater forcing of thermohaline circulation in the limit of low diapycnal mixing. J. Geophys. Res.,107, 3077, doi:10.1029/2000JC000685.
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.
Lazar, A., G. Madec, and P. Delecluse, 1999: The deep interior downwelling, the Veronis effect, and mesoscale tracer transport parametrizations in an OGCM. J. Phys. Oceanogr, 29 , 2945–2961.
Levitus, S., and T. P. Boyer, 1994: Temperature. Vol. 4, World Ocean Atlas 1994, NOAA Atlas NESDIS 4, 117 pp.
Manabe, S., and R. J. Stouffer, 1988: Two stable equilibria of a coupled ocean–atmosphere model. J. Climate, 7 , 5–23.
Marotzke, J., 1997: Boundary mixing and the dynamics of three-dimensional thermohaline circulations. J. Phys. Oceanogr, 27 , 1713–1728.
Marshall, J. H., H. Jones, R. Karsten, and R. Wardle, 2002: Can eddies set ocean stratification? J. Phys. Oceanogr, 32 , 26–38.
McDermott, D. A., 1996: The regulation of northern overturning by Southern Hemisphere winds. J. Phys. Oceanogr, 26 , 1234–1255.
Munk, W., 1966: Abyssal recipes. Deep-Sea Res, 13 , 707–730.
Pacanowski, R. C., and S. M. Griffies, 1999: MOM 3.0 manual. Documentation user's guide and reference manual, NOAA/Geophysical Fluid Dynamics Laboratory, Princeton, NJ, 668 pp.
Polzin, K. L., J. M. Toole, J. R. Ledwell, and R. W. Schmitt, 1997: Spatial variability of turbulent mixing in the abyssal ocean. Science, 276 , 93–96.
Rahmstorf, S., 1996: On the freshwater forcing and transport of the Atlantic thermohaline circulation. Climate Dyn, 12 , 799–811.
Redi, M. H., 1982: Oceanic isopycnal mixing by coordinate rotation. J. Phys. Oceanogr, 12 , 1154–1158.
Saenko, O. A., A. J. Weaver, and J. M. Gregory, 2003: On the link between the two modes of the ocean thermohaline circulation and the formation of global-scale water masses. J. Climate, 16 , 2797–2801.
Stone, P., 1972: A simplified radiative-dynamical model for the static stability of rotating atmospheres. J. Atmos. Sci, 29 , 405–418.
Toggweiler, J. R., and B. Samuels, 1995: Effect of Drake Passage on the global thermohaline circulation. Deep-Sea Res, 42 , 477–500.
Toggweiler, J. R., and B. Samuels, 1998: On the ocean's large-scale circulation near the limit of no vertical mixing. J. Phys. Oceanogr, 28 , 1832–1852.
Toggweiler, J. R., K. Dixon, and K. Bryan, 1989: Simulations of radiocarbon in a coarse-resolution World Ocean model. Part 1: Steady state prebomb distributions. J. Geophys. Res, 94 , 8217–8242.
Veronis, G., 1975: The role of models in tracer studies. Numerical Models of the Ocean Circulation, National Acadamy of Science, 133–146.
Visbeck, M., J. Marshall, T. Haines, and M. Spall, 1997: One the specification of eddy transfer coefficients in coarse-resolution ocean circulation models. J. Phys. Oceanogr, 27 , 381–402.
Wang, X., P. H. Stone, and J. Marotzke, 1999: Global thermohaline circulation. Part I: Sensitivity to atmospheric moisture transport. J. Climate, 12 , 71–82.
Weaver, A. J., and M. Ebby, 1997: On the numerical implementation of advection schemes for use in conjunction with various mixing parametrizations in the GFDL ocean model. J. Phys. Oceanogr, 27 , 369–377.
Weijer, W., W. P. M. de Ruijter, H. A. Dijkstra, and P. J. van Leeuwen, 1999: Impact of interbasin exchange on the Atlantic overturning circulation. J. Phys. Oceanogr, 29 , 2266–2284.
Weijer, W., W. P. M. de Ruijter, and H. A. Dijkstra, 2001: Stability of the Atlantic overturning circulation: Competition between Bering Strait freshwater flux and Agulhas heat and salt sources. J. Phys. Oceanogr, 31 , 2385–2402.
Wiebe, E. C., and A. J. Weaver, 1999: On the sensitivity of global warming experiments to the parametrization of sub-grid scale ocean mixing. Climate Dyn, 15 , 875–893.
Wunsch, C., 1999: Where do ocean eddy fluxes matter? J. Geophys. Res, 104 , 13235–13249.
Meridional overturning streamfunction (Sv) in the Atlantic for two control simulations, (a) GMC and (b) HDC. Eddy-induced transport velocities are not added in (a). The Southern Ocean is not shown. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning streamfunction (Sv) in the Atlantic for two control simulations, (a) GMC and (b) HDC. Eddy-induced transport velocities are not added in (a). The Southern Ocean is not shown. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning streamfunction (Sv) in the Atlantic for two control simulations, (a) GMC and (b) HDC. Eddy-induced transport velocities are not added in (a). The Southern Ocean is not shown. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic: (a) GMC, (b) HDC, (c) based on Levitus temperature and salinity, and (d) difference between values in GMC and HDC. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic: (a) GMC, (b) HDC, (c) based on Levitus temperature and salinity, and (d) difference between values in GMC and HDC. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic: (a) GMC, (b) HDC, (c) based on Levitus temperature and salinity, and (d) difference between values in GMC and HDC. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) in the upper 1000 m of the Southern Ocean zonally averaged around the globe in GMC (solid) and HDC (dashed)
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) in the upper 1000 m of the Southern Ocean zonally averaged around the globe in GMC (solid) and HDC (dashed)
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) in the upper 1000 m of the Southern Ocean zonally averaged around the globe in GMC (solid) and HDC (dashed)
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Concentration of the dye tracers (see text) zonally averaged in the Atlantic in two control simulations: (a), (b) GMC and (c), (d) HDC. (a), (c) Southern Ocean tracer; (b), (d) North Atlantic tracer
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Concentration of the dye tracers (see text) zonally averaged in the Atlantic in two control simulations: (a), (b) GMC and (c), (d) HDC. (a), (c) Southern Ocean tracer; (b), (d) North Atlantic tracer
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Concentration of the dye tracers (see text) zonally averaged in the Atlantic in two control simulations: (a), (b) GMC and (c), (d) HDC. (a), (c) Southern Ocean tracer; (b), (d) North Atlantic tracer
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning streamfunction (Sv) in the Atlantic for (a) GMNH and (b) GMSO. Eddy-induced transport velocities are not added. The Southern Ocean is not shown. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning streamfunction (Sv) in the Atlantic for (a) GMNH and (b) GMSO. Eddy-induced transport velocities are not added. The Southern Ocean is not shown. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning streamfunction (Sv) in the Atlantic for (a) GMNH and (b) GMSO. Eddy-induced transport velocities are not added. The Southern Ocean is not shown. The upper 1000 m are enlarged
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between (a) GMC and GMNH, (b) GMNH and HDC, (c) GMC and GMSO, and (d) GMSO and HDC
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between (a) GMC and GMNH, (b) GMNH and HDC, (c) GMC and GMSO, and (d) GMSO and HDC
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between (a) GMC and GMNH, (b) GMNH and HDC, (c) GMC and GMSO, and (d) GMSO and HDC
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) area-averaged in the Northern Hemisphere of the Atlantic for four cases: GMC (solid with stars), HDC (dashed with stars), GMSO (solid with circles), and GMNH (dashed with circles). Only values below 250 m are shown for better presentation
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) area-averaged in the Northern Hemisphere of the Atlantic for four cases: GMC (solid with stars), HDC (dashed with stars), GMSO (solid with circles), and GMNH (dashed with circles). Only values below 250 m are shown for better presentation
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Potential density (referenced to 1000 m; kg m−3) area-averaged in the Northern Hemisphere of the Atlantic for four cases: GMC (solid with stars), HDC (dashed with stars), GMSO (solid with circles), and GMNH (dashed with circles). Only values below 250 m are shown for better presentation
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional velocity (m s−1) at 45°N for (a) GMC, (b) HDC, (c) GMNH, and (d) GMSO. Upper 3500 m are shown only. (e) Meridional velocity zonally averaged within 20° east of the solid wall at 45°N in the Atlantic for GMC (solid with stars), HDC (dashed with stars), GMSO (solid with circles), and GMNH (dashed with circles). Velocities in the uppermost layer are not shown
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional velocity (m s−1) at 45°N for (a) GMC, (b) HDC, (c) GMNH, and (d) GMSO. Upper 3500 m are shown only. (e) Meridional velocity zonally averaged within 20° east of the solid wall at 45°N in the Atlantic for GMC (solid with stars), HDC (dashed with stars), GMSO (solid with circles), and GMNH (dashed with circles). Velocities in the uppermost layer are not shown
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional velocity (m s−1) at 45°N for (a) GMC, (b) HDC, (c) GMNH, and (d) GMSO. Upper 3500 m are shown only. (e) Meridional velocity zonally averaged within 20° east of the solid wall at 45°N in the Atlantic for GMC (solid with stars), HDC (dashed with stars), GMSO (solid with circles), and GMNH (dashed with circles). Velocities in the uppermost layer are not shown
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
(a) Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between SOdense and HDC. (b) Meridional overturning in the Atlantic for SOdense. Southern Ocean is not shown in (b)
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
(a) Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between SOdense and HDC. (b) Meridional overturning in the Atlantic for SOdense. Southern Ocean is not shown in (b)
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
(a) Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between SOdense and HDC. (b) Meridional overturning in the Atlantic for SOdense. Southern Ocean is not shown in (b)
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning (Sv) in the Atlantic for four experiments without the AABW (note a zonal ridge at 30°S): (a) GMnB, (b) HDnB, (c) GMNHnB, and (d) GMSOnB
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning (Sv) in the Atlantic for four experiments without the AABW (note a zonal ridge at 30°S): (a) GMnB, (b) HDnB, (c) GMNHnB, and (d) GMSOnB
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Meridional overturning (Sv) in the Atlantic for four experiments without the AABW (note a zonal ridge at 30°S): (a) GMnB, (b) HDnB, (c) GMNHnB, and (d) GMSOnB
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between (a) GMnB and HDnB, (b) GMNHnB and HDnB, and (c) GMnB and GMSOnB
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between (a) GMnB and HDnB, (b) GMNHnB and HDnB, and (c) GMnB and GMSOnB
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Difference in the potential density (referenced to 1000 m; kg m−3) zonally averaged in the Atlantic between (a) GMnB and HDnB, (b) GMNHnB and HDnB, and (c) GMnB and GMSOnB
Citation: Journal of Physical Oceanography 34, 7; 10.1175/1520-0485(2004)034<1628:MCTSOT>2.0.CO;2
Horizontal mixing in the experiments
Volume transports (Sv) and pycnocline depth D (m); see text for the definitions. Note that, in the runs with AABW, the amount of the inflow of the Southern Ocean water is larger than the outflow of the NADW water, since part of the outflow is of the Southern Ocean origin. Since NADW and AABW mix and can change their composition, the NADW maximum overturning is not equal to the sum of the NH upwelling and NADW outflow
Solutions of (10) for four choices of mixing coefficients. Results can be compared with the following experiments of section 3: GMC (AI = 1000, kH = 0), HDC (AI = 0, kH = 1000), GMSO (AI = 1000, kH = 1000), and GMNH (AI = 0, kH = 0)
Joint Institute for the Study of Atmosphere and Ocean Contribution Number 948.
The definition of the pycnocline depth (4) is based on the assumption that the ocean is much deeper than D. This definition, therefore, works well when the NADW cell rests on top of AABW and does not “feel” the bottom. In the absence of AABW, however, we find that D defined by (4) is too sensitive to a choice of the reference level H, and the results become less trustworthy.
Note that the experiments with different surface density (SOdense) and basin geometry (four runs without AABW of section 3c) require a different value of constant C. In addition, comparison of the pycnocline depths D for the experiments without AABW with solutions of (10) is not straightforward because of apparent difficulties in using (4) for these cases (see section 3c).
