Global Circulation and Water Properties

C H A P T E R

14 Global Circulation and Water Properties In this chapter we summarize the circulation and water properties at a global scale, synthesizing the regional elements from the individual ocean basins (Chapters 9 through 13), and present some evolving views of the global overturning circulation. For courses providing just a limited introduction to the ocean’s circulation and water properties, it might suffice to use materials from Chapter 4 and this chapter, with highlights from the forcing fields in Chapter 5 and introductory materials in the basin Chapters 9 through 13. The surface circulation systems (Section 14.1.1) have been observed for centuries in all of their complexity, and are the best mapped part of the circulation because of ease of access. These circulations impact navigation, pollutant dispersal, the upper ocean’s productive euphotic zone, and continental shelves and coastal zones. As the interface with the atmosphere, the surface layer and circulation are directly involved in ocean-atmosphere feedbacks that affect both the mean states of the ocean and atmosphere and also seasonal to climate scale variability. Just a few hundred meters below the sea surface, some parts of the circulation change dramatically as the wind-driven gyres contract and weaken. At intermediate and abyssal depths (Section 14.1.2), the circulation is

Descriptive Physical Oceanography

dominated by the deep penetration of the most vigorous surface currents, and by circulation associated with large-scale buoyancy forcing and weak diapycnal processes that can change the density of the water internally (Section 14.5). The large-scale circulations include very weak vertical velocities that connect these deeper layers with each other and with the upper ocean, referred to as the overturning circulation (Section 14.2). The overturning circulation includes shallow cells that cycle water within the warmest, lowest density parts of the ocean, which can be important for poleward heat export from the tropics and subtropics. The deeper overturning circulations, connecting intermediate and deep waters to the sea surface, are generally much more global in scope than the wind-driven, upper ocean circulation systems. The grandest scale overturning circulations are those associated with North Atlantic Deep Water (NADW) formation in the northern North Atlantic and Nordic Seas, and with dense water formation in the Southern Ocean. A weaker, smaller scale overturning circulation is associated with North Pacific Intermediate Water (NPIW) formation in the North Pacific. The drivers for the ocean circulation, its variability, and its mixing are the winds and airesea-ice buoyancy fluxes; the tides are an additional source of energy for turbulent

473

Ó 2011. Lynne Talley, George Pickard, William Emery and James Swift. Published by Elsevier Ltd. All rights reserved.

474

14. GLOBAL CIRCULATION AND WATER PROPERTIES

dissipation, which is central to the overturning circulation. In Chapter 5 we presented all of these forcing fields. In Section 14.3 we revisit the ocean’s heat and freshwater transports with an emphasis on their relation to elements of the ocean circulation. Time dependence characterizes fluid flows at all timescales. While this text mostly emphasizes the large-scale, time-averaged circulation, each basin chapter also introduces regions of persistent local eddy variability. Here we summarize the global distribution of eddy variability and associated eddy diffusivity (Section 14.5). A brief overview of climate variability and climate change in the global ocean is provided in Section 14.6, but the main materials are presented in the supplemental material as Chapter S15 on the text Web site http:// booksite.academicpress.com/DPO/; “S” denotes supplemental material. East Greenland Current

Norwegian Atlantic Current

14.1. GLOBAL CIRCULATION 14.1.1. Upper Ocean Circulation Systems The global surface circulation is shown schematically in Figure 14.1. This surface circulation derives most of its characteristic shape and strength from the basin-scale wind forcing in each ocean (Section 5.8, Figures 5.16 and 5.17). The anticyclonic subtropical gyres in each of the five ocean basins are evident, with their poleward western boundary currents: the Gulf Stream and North Atlantic Current (NAC), Kuroshio, Brazil Current, East Australian Current (EAC), and Agulhas Current systems. Each anticyclonic gyre has its eastern boundary current regime: Canary, California, Benguela, Peru-Chile, and Leeuwin Current systems, respectively. The eastern boundary currents

East Kamchatka Current

Bering Strait

Beaufort Gyre

Oyashio Labrador Current North Atlantic Current Canary Current System

Gulf Stream System NEC

Equator

NECC NEUC

North Brazil Current System

40 N

Alaska Gyre

Somali Current System

Subpolar Gyre

Indonesian Throughflow

Kuroshio System

Subtropical Gyre

Benguela Current System

NEUC NECC

NEC Pacific Equatorial/Tropical Current System

Subtropical Gyre

Brazil Current System

California Current System

Subtropical Gyre

Atlantic Equatorial Current System

SEC

North Pacific Current

Equator

SEC SEC

Leeuwin Current Subtropical Gyre

Subtropical Gyre

East Australian Current System

40S

Agulhas Current System

Malvinas Current Weddell Sea Gyre

Antarctic Circumpolar Current System

Peru-Chile Current System

Ross Sea Gyre

Subtropical Gyres Equatorial and Tropical Circulations Intergyre and/or Interbasin Exchanges Polar & Subpolar Current Systems

FIGURE 14.1 Surface circulation schematic. This figure can also be found in the color insert. Modified from Schmitz (1996b).

GLOBAL CIRCULATION

flow equatorward with the exception of the Leeuwin Current, which flows poleward. The higher latitude cyclonic circulations with their equatorward western boundary currents are evident in the Arctic and Nordic Seas, North Atlantic, North Pacific including the marginal seas, and the Weddell and Ross Seas. The respective boundary currents are the East Greenland (EGC) and Labrador Currents, the East Kamchatka Current (EKC) and Oyashio, and the boundary currents of the Weddell and Ross Sea gyres. In the tropics, the quasi-zonal tropical circulation systems are apparent, including equatorial countercurrents, equatorial currents, and lowlatitude western boundary currents. Large-scale tropical cyclonic circulation systems include the zonally elongated North Equatorial Current and Countercurrent “gyres” at 5e10
N in the Pacific and Atlantic, the Angola Dome (South Atlantic), and Costa Rica Dome (North Pacific). While all circulation is time dependent to some extent, tropical circulation variability is particularly strong relative to the mean, with fast responses to changing winds yielding strong seasonal and interannual variability. Of the major western boundary currents, only the Somali Current system in the northwestern Indian Ocean and the circulation in the Bay of Bengal change direction completely (seasonally), responding quickly to the reversing monsoonal winds because of the narrow width of their basins, which reduces the response time to the changing winds. The ocean circulations are connected to each other. The North Pacific is connected to the North Atlantic with a small transport (<1 Sv) through the Bering Strait, through the Arctic, and then southward both west and east of Greenland. The tropical Pacific feeds water into the Indian Oceans through the Indonesian passages with a modest transport (~10 Sv). The three major oceans south of South America (Drake Passage), Africa, and Australia/New Zealand are connected through the Antarctic

475

Circumpolar Current (ACC), with a large transport (>100 Sv). (The Bering Strait and Indonesian Throughflow (ITF) outflows from the Pacific are supplied from the Southern Ocean as well.) There are also many connections with marginal seas that affect water properties within the open ocean. Many of these connections are shown in Figure 14.1; they are mostly discussed in the ocean basin chapters. Surface circulation mapped directly from data has improved dramatically in recent years because of more complete surface drifter data sets and satellite altimetry (see Chapter S16 on the textbook Web site). A drifter-based surface dynamic topography map, which reflects the surface geostrophic circulation, is shown in Figure 14.2a (Maximenko et al., 2009). Globally, the highest dynamic height is in the subtropical North Pacific, which is around 70 cm higher than the highest dynamic topography of the subtropical Atlantic. The lowest dynamic height surrounds Antarctica, south of the ACC. Relatively low dynamic height is found in the subpolar North Atlantic and North Pacific. Surface velocity (Figure 14.2b) highlights include the high velocities and narrow boundary currents, the zonal tropical circulation systems, and the ACC. This total surface velocity field includes both geostrophic and Ekman components. In the geostrophic flow field, depicted by contours of surface dynamic height (Figure 14.2a), complete closed subtropical gyres are missing or distorted in some regions, especially the North Atlantic. But when the total including the Ekman component is considered, the surface circulation appears more gyre-like, and more similar to the 200 m geostrophic flow (Figure 14.3). The streamlines for total surface velocity also indicate regions of convergence, where the streamlines terminate in mid-gyre, and divergence where the streamlines originate in mid-gyre. Convergence and divergence are associated only with the Ekman velocity, since geostrophic flow is non-divergent by definition (Section 7.6).

476

14. GLOBAL CIRCULATION AND WATER PROPERTIES

FIGURE 14.2 (a) Surface velocity streamlines, including both geostrophic and Ekman components; color is the mean speed in cm/sec, and (b) surface dynamic topography (dyn cm), with 10 cm contour intervals. This figure can also be found in the color insert. Source: From Maximenko et al. (2009).

Just below the surface layer, even at 200 dbar, all five wind-driven subtropical gyres are tighter (more localized) than at the surface (Figure 14.3). At this depth there is no Ekman flow, so the total mean velocity is represented by the absolute dynamic topographies in the basin chapters, which the relative dynamic

topography in Figure 14.3 strongly resembles. Compared with the sea surface, the subtropical gyres are shifted toward their strong western boundary currents and extensions, that is, toward the west and the poles. At 1000 dbar relative to 2000 dbar, the anticyclonic gyres retreat even more toward their

GLOBAL CIRCULATION

477

FIGURE 14.3 Steric height (dyn cm) relative to 2000 dbar at (a) 200 dbar and (b) 1000 dbar, using mean temperature and salinity from five years of float profiles (2004e2008). Source: From Roemmich and Gilson (2009).

western boundary current extensions (Figure 14.3b).1 The contrast between maximum dynamic height of the Pacific and Atlantic remains, with the Pacific higher than the Atlantic. The Southern Hemisphere gyres are much more exaggerated at 1000 dbar than in the upper ocean, while the Kuroshio and Gulf Stream gyres are weaker. In Figure 14.3b, the Gulf Stream gyre even appears to have disappeared in this relative velocity calculation, in favor of general 1

northeastward flow into the subpolar gyre, but the absolute geostrophic streamfunction even as deep as 2500 dbar in Chapter 9 (Figure 9.14a) retains a closed anticyclonic Gulf Stream gyre. The subpolar circulations and ACC are more barotropic than the subtropical circulations, with little change in position from the sea surface to the ocean bottom. This marked shift in behavior from the subtropics to the subpolar regions is most likely due to a reduction in

The absolute geostrophic streamfunctions at 1000 dbar in the basin chapters differ somewhat from the relative geostrophic flow at 1000 dbar relative to 2000 dbar in Figure 14.3b, because the 1000 and 2000 dbar circulations are both weak. Therefore the non-zero flow field at 2000 dbar is important to include when computing the 1000 dbar flow relative to 2000 dbar.

478

14. GLOBAL CIRCULATION AND WATER PROPERTIES

stratification, which then allows much deeper penetration of surface signals.

14.1.2. Intermediate and Deep Circulation At intermediate depth (Figure 14.4a), the geostrophic circulation, represented by steric height, retains the western boundary currents,

their recirculations, and the ACC of the upper ocean. It also retains a strongly zonal character in the tropics (where flows are not well resolved in the set of studies used in the figure). Importantly, Deep Western Boundary Currents (DWBCs; Section 7.10) appear by this depth, and there is a transition to the structure of the open-ocean deep flows that are affected by topography, especially the mid-ocean ridges.

FIGURE 14.4 Streamlines for the (a) mid-depth circulation at 2000 dbar and (b) deep circulation at 4000 dbar. (Adjusted steric height, representing the absolute geostrophic flow.) Source: From Reid (1994, 1997, and 2003).

GLOBAL CIRCULATION

Unlike the surface currents, deep currents that are not a deep expression of a surface current have only generic names. They are mostly identified by location. In more detail, each of the five subtropical anticyclonic gyres retains some expression at 2000 dbar, including a western boundary current that separates from the coast and flows eastward, and a very compact anticyclonic circulation on the equatorward, offshore side. The Gulf Stream, Brazil Current, Agulhas, and a deep version of the South Pacific’s anticyclonic gyre (somewhat east of New Zealand at this depth due to topography) are present. The Kuroshio gyre is shifted north of its surface location in Figure 14.4a, but in Chapter 10 we noted that the Kuroshio Extension does extend, weakly, to the ocean bottom at the same location as its surface core (Figure 10.3). The high latitude cyclonic circulations evident at the sea surface are also present in the northern North Atlantic, North Pacific, and south of the ACC in the Weddell and Ross Seas, continuing the near-barotropic character previously noted. Major circulation features that appear at 2000 dbar, but not at the sea surface, include the DWBCs and poleward flows along the eastern boundaries. The DWBCs at 2000 dbar are southward in the Atlantic and mostly northward in the Pacific and Indian Oceans. The Atlantic DWBC carries NADW from the northern North Atlantic southward to about 40
S, where it enters the ACC system. At this depth, unlike at 4000 dbar, there is no DWBC in the subtropical South Pacific, which has a deep-reaching subtropical gyre. However, a northward DWBC does form within the tropical Pacific and can be traced northward along the western boundary to past the northern boundary of Japan, where it encounters a southward DWBC. These flows are more clearly defined at 4000 dbar. In the Indian Ocean, the northward DWBCs are also more easily identified at 4000 dbar (next), but include northward flow along the east coast of Madagascar and a hint of

479

northward flow in mid-basin that follows the Central Indian Ridge. At 2000 dbar, all three oceans export water southward into the Southern Ocean. Part of this southward transport is gathered in broad poleward flows near the eastern boundaries, and is evident in water properties in the South Pacific and South Atlantic (see the basin chapters, 9e13). In the Indian Ocean, poleward flow is evident in water properties west of Australia, but it does not continue southward to the ACC. In the South Atlantic, southward flow of NADW is most vigorous in the DWBC along the western boundary. For a dynamically complete description, we note that the 2000 dbar flows near the eastern boundaries of the North Atlantic and North Pacific are also poleward, indicating that basin-scale cyclonic circulation at this depth is ubiquitous. The circulation at 4000 dbar (Figure14.4b) is greatly affected by topography. Here DWBCs carry deep and bottom water northward from the Antarctic into each of the three oceans. The northward DWBC in the South Atlantic, carrying Antarctic Bottom Water (AABW), shifts eastward at the equator, becoming an eastern boundary current along the mid-Atlantic Ridge (Chapter 9); the DWBC at the continental western boundary in the North Atlantic is southward, carrying the deepest components of NADW. The northward DWBCs in the Indian Ocean follow the ridge systems, east of Madagascar and northward into the Arabian Sea, east of the Southeast Indian Ridge and Central Indian Ridge, and east of Broken Plateau into the western Australia Basin. In the Pacific, the principal northward DWBC is east of New Zealand, flowing through Samoan Passage into the tropics, crossing the equator and then northward along the western boundary and also through the Wake Island Passage and along the Izu-Ogasawara Ridge. In the far northern North Pacific, the DWBC is southward. This is counterintuitive if one assumes (incorrectly) that DWBCs must carry

480

14. GLOBAL CIRCULATION AND WATER PROPERTIES

water away from their sinking sources. There is no surface source of deep water in the northern North Pacific. This DWBC thus best illustrates the dynamics of the deep circulation, which is driven by upwelling from the bottom and deep water layers into the overlying intermediate and shallow layers. According to the Stommel and Arons (1960a,b) solution, upwelling stretches the deep water column thereby creating poleward flow in the deep layers in mid-ocean through potential vorticity conservation (Section 7.10.3). This poleward flow is counterintuitive, since the main flow in the basins is toward the high latitude sources of deep water. In this theoretical framework, the DWBCs are a consequence of closing the mass balance for this upwelling. They are not simply drains of dense water.

14.2. GLOBAL MASS TRANSPORTS AND OVERTURNING CIRCULATION The overturning circulation in each ocean is described in Chapters 9e13. We present here a global picture of the volume/mass overturns. Their role in global heat and freshwater transport is summarized in Section 14.3. The global overturning circulation is complex and three-dimensional, with dominant paths that we attempt to depict in a simplified manner. The student is cautioned that these pathways are not isolated tubes flowing through the ocean. Many transport pathways depicted schematically as narrow “ribbons” are broad and thick flows, covering large regions. There is much circulation and mixing from one “path” to another. Throughout, and especially in the upper ocean, the pathways can be caught up in the much stronger wind-driven circulation. 2

Historically, there has been an emphasis on the “meridional” overturning circulation (MOC), which is important for the latitudinal redistribution of heat, freshwater, and other properties. However, some important elements of the global overturning circulation and these redistributions are not meridional. Inter-basin transports between the Pacific, Indian, and Atlantic Oceans are crucial for the global mean ocean state. Even at the ocean gyre scale, the mean state is maintained by some zonal components, such as between western boundary air-sea heat loss regions and eastern boundary air-sea heat gain regions. Calculation and depiction of the MOC usually includes computing the meridional mass transports across each coast-to-coast, zonal transect in isopycnal (or depth) layers (Section 14.2.1), computing the upwelling and downwelling transports between the layers in closed geographic regions bounded by two transects (Section 14.2.2), and computing the overturning streamfunction to visualize the overturn in two dimensions (Section 14.2.3). Schematics of the overturn (14.2.4) are often constructed to assist interpretation, but are obviously not essential to the calculations. More generally, this methodology applies to any closed region, and could be used for zonal transports across meridional sections, or even transports into an open ocean region enclosed by station data.

14.2.1. Mass Transports in Layers into Closed Regions The overturning circulation is calculated by first defining closed geographical regions within which net mass must be (nearly) conserved.2 For example, a closed region can be defined by two coast-to-coast, top-to-bottom transects (labeled “N” and “S” in Figure 14.5);

The mass balance is not exactly zero because there is a very small exchange of freshwater with the atmosphere. When time dependence is considered, there can also be a time-dependent storage or deficit of mass within some regions; this also is proportionally small when considering large regions.

GLOBAL MASS TRANSPORTS AND OVERTURNING CIRCULATION

North latitude

Western boundary

VN4

Layer 4

W3

VS4

W2

VS3

Layer 2

VS1

e3

rfac

Inte

e2

rfac

Inte

Deep layer

VN2 VS2

Surface layer Intermediate layer

VN3

Layer 3

Layer 1

Eastern boundary

Sea surface

South latitude

W1 Bottom layer

in the mean, the same amount of water must move out of the region through one section as moves in through the other (Section 5.1). For data analysis, the latitudes are those of ocean transects along which the data were collected. For models, any latitudes can be chosen and often many are used, thinking ahead to the overturning streamfunction calculation described in Section 14.2.3. To look at overturn, the closed region is next divided vertically into layers (c.f. i ¼ 1, 2, 3, 4 in Figure 14.5). The boundaries between the layers can be defined in different ways. Exact choices depend on the purposes of the calculation. The usual choices are isopycnals (or neutral density surfaces), constant depths, and sometimes even isotherms (of potential temperature). Isopycnal (isoneutral) surfaces are usually the most informative, because they are directly related to the airesea buoyancy fluxes and diapycnal diffusion that transform waters from one layer to another. The net mass transports in the layers along the southern and northern boundaries of the region

481

FIGURE 14.5 Meridional overturning circulation transport calculation: example for four layers. The mass transports for each layer “i” through the southern and northern boundaries of each layer are VSi and VNi. The vertical transport across each interface is Wi. Arrow directions are those for positive sign; the actual transports can be of any magnitude and sign. The sum of the four transports (two horizontal and two vertical) into a given closed layer must be 0 Sv. The small amount of transport across the sea surface due to evaporation and precipitation is not depicted.

e1

rfac

Inte

VN1

are then calculated (VSi and VNi in Figure 14.5). For hydrographic sections, the transports are usually based on geostrophic velocities from the sections, plus Ekman transports perpendicular to the sections. Calculating the geostrophic velocities from hydrographic station data is not trivial since reference velocities are required (Section 7.6); the overall mass conservation constraint is one of the important inputs for determining the best set of reference velocities. Three different global calculations are superimposed in Figure 14.6a (two based on data and one on a global ocean model) and a fourth is shown in Figure 14.6b; a fifth is represented in Figure 14.9b and c by its overturning streamfunction (Lumpkin & Speer, 2007). The most robust elements of the overturning circulation are common to all of these calculations. For all five analyses in Figures 14.6 and 14.9, mass transports were first computed for a large number of isopycnal layers. These were combined into three or four larger layers representing: the upper ocean above the main pycnocline; a deep

482

14. GLOBAL CIRCULATION AND WATER PROPERTIES

(b)

80˚W

40˚W

0˚

80˚N

40˚E

80˚E

2.5

160˚W

Surface Intermediate Deep Bottom

40˚N

20˚N

160˚E

120˚W

Mass transports (Sv)

60˚N

0.7

15.3 6.8 7.0

120˚E

80˚N

1.0

40˚N

0.6 1.8

19.0

60˚N

1.0

2.8

2.4

20˚N

1.0

0.4

10.0

0˚

0˚

3.9 10.1

20˚S

1.1 40˚S

7.7

17.6 3.9

5.2

7.6

5.9 20˚S

6.2

3.6

10.3

10.7

13.6

5.6

2.5 0.5

3.1 60˚S

40˚S

60˚S 80˚S

80˚S 80˚W

40˚W

0˚

40˚E

80˚E

120˚E

160˚E

160˚W

120˚W

FIGURE 14.6 Net transports (Sv) in isopycnal layers across closed hydrographic sections (1 Sv ¼ 1  106 m3/sec). (a) Three calculations from different sources are superimposed, each using three isopycnal layers (see header). Circles between sections indicate upwelling (arrow head) and downwelling (arrow tail) into and out of the layer defined by the circle color. This figure can also be found in the color insert. Source: From Maltrud and McClean (2005), combining results from their POP model run, Ganachaud and Wunsch (2000), and Schmitz (1995). (b) Fourth calculation based on velocities from Reid (1994, 1997, 2003), with ribbons indicating flow direction and oveturn locations schematically. Source: From Talley (2008).

483

GLOBAL MASS TRANSPORTS AND OVERTURNING CIRCULATION

layer that includes North Atlantic, Pacific, and Indian Deep Waters; and a bottom layer that is mainly dense Antarctic water (Lower Circumpolar Deep Water, LCDW or AABW). For the upper ocean layer, robust results are (a) net northward mass transport through the entire Atlantic (also including intermediate water in Figure 14.6b), (b) southward transport out of the Indian Ocean, (c) northward transport into the Pacific, (d) westward transport from the Pacific into the Indian Ocean through the Indonesian passages (ITF), and (e) northward transport out of the Pacific into the Atlantic through the Arctic (Bering Strait). Along these pathways, there is also weak upwelling into the warm water path from deeper layers within the Pacific and Indian Oceans. Deep water in Figures 14.6 and 14.9 is transported southward through the length of the Atlantic and southward out of the Pacific. These are the NADW and Pacific Deep Water (PDW), respectively. Deep transport in the Indian Ocean in Figure 14.6a is small and northward. When the deep Indian layer is subdivided, as in Figure 14.6b, the thinner layers have nearly balancing northward and southward transports of about 6 Sv; these are NADW moving northward and Indian Deep Water (IDW) moving southward at a slightly lower density. Bottom water moves northward from the Antarctic into all three oceans. Figure 14.6b shows the northward penetration of bottom water into the subtropical North Atlantic as well, while the thicker layer used in Figure 14.6a subsumes this Antarctic water in the bottom part of the southward-moving NADW. The layer transports differ from one section to another within each map. Therefore there is convergence or divergence between the sections. This results in upwelling and downwelling, as described next. The weak overturn of the North Pacific is also depicted in Figure 14.6b. This cell transports approximately 2 Sv of warm water northward, and slightly denser NPIW southward.

14.2.2. Upwelling and Downwelling Returning to the method (Figure 14.5), we next calculate the vertical (diapycnal) transport through the interfaces between layers within the closed regions. The transports and velocities for each layer i are MTi ¼ VNi  VSi þ Wi  Wi1 ¼ 0

(14.1a)

Wi ¼ VNi þ VSi þ Wi1

(14.1b)

wi ¼ Wi =Ai

(14.1c)

in which the vertical transport through the bottom (“W0”) is zero. Ai is the area of each interface and wi is the average vertical velocity through the interface. The upwelling or transport Wi , in units of Sverdrups, across the top interface of each layer is calculated from the divergence of the horizontal transports in the closed region plus the upwelling transport across the bottom interface (Eq. 14.1b). We start with the bottom layer, which has no flow through its bottom boundary, labeled i ¼ 1 in Figure 14.5. The total transport MT1 into the closed region must be 0 by continuity (Eq. 14.1a). This yields the upwelling or downwelling transport W1 across the upper interface of the bottom layer since there is no transport through the ocean bottom. The average upwelling velocity w1, in m/sec, across this interface is this transport divided by the surface area A1 of that interface. Next move upward through each of the layers and find the sum of the transports through the side boundaries and through the bottom interface (VNi, VSi, and Wi1). This yields the net transport Wi through the upper interface of this box. Continue this for all layers up to the surface. Because the overall velocity calculation should have been carried out with mass conservation (including Ekman transport in the uppermost layer), there should be no net upwelling or downwelling across the sea surface, which is the upper interface of the topmost box.

484

14. GLOBAL CIRCULATION AND WATER PROPERTIES

For example, if there is a net flow of 2 Sv northward into the southern side of the bottom box and a net flow of 1 Sv northward out of the northern side of the box, then there must be a net loss of 1 Sv within the closed bottom region. Therefore, 1 Sv must upwell across its top interface. The average upwelling velocity, with an interface surface area of, say, 1013m2, would be 105cm/sec. Now, switching back to results for the actual global ocean (Figure 14.6), we find: (a) the net lateral (meridional) transport requires net downwelling from the surface to the deep water in the northern North Atlantic, (b) there is also downwelling from the upper ocean and deep water to bottom water in the Antarctic, and (c) there is diapycnal upwelling of bottom water in all three oceans in the low latitude region between about 30
S and 24
N (the locations of the zonal hydrographic sections). While most of this upwelling is into just the overlying deep water layer, some reaches the upper ocean in the Indian and Pacific Oceans.

The “downwelling” process in the North Atlantic is localized deep convection in the Greenland and Labrador Seas followed by entrainment of surrounding waters; this is the production of NADW (Chapter 9). The “downwelling” in the Antarctic is localized brine rejection combined with entrainment that increases its volume tenfold, mainly along continental shelves and near ice shelves; this is the production of AABW (Chapter 13). “Upwelling” in low latitudes is associated with eddy diffusion, driven by deep turbulence, which has large geographical heterogeneity (see Sections 7.3.2 and 14.5; Figure 14.7). In the Indian and Pacific Oceans, this upwelling produces the Indian Deep Waters (IDW) and the PDW from upwelled AABW. In the Atlantic Ocean, the upwelled AABW joins the NADW. In greater detail, in the Indian Ocean there is upwelling from the bottom to the deep water, from the deep water to the intermediate layer, and even a small amount of upwelling to the thermocline layer. The South Pacific has similar

FIGURE 14.7 Modeled upwelling across the isopycnal 27.625 kg/m3, which represents upwelling from the NADW layer. This figure can also be found in the color insert. Source: From Kuhlbrodt et al. (2007); adapted from Do¨o¨s and Coward (1997).

GLOBAL MASS TRANSPORTS AND OVERTURNING CIRCULATION

processes, with net inflow in the bottom layer and outflow in all layers above it, hence with net upwelling into each of the layers, although the quantities and vertical distribution differ from the Indian Ocean. In the Southern Ocean, using a finer division of layers than in Figure 14.6, there is also diapycnal upwelling from the deep water to the upper ocean. The upwelling source waters are largely the IDW and PDW that enter the Southern Ocean at a slightly lower density and shallower depth than the NADW. All three northern source deep waters (NADW, PDW, and IDW) physically upwell here to near the sea surface, mostly adiabatically along the steeply sloped isopycnals. The actual Southern Ocean diapycnal “upwelling” can occur mostly near the sea surface where airesea buoyancy flux can directly transform the upwelled water. The airesea buoyancy flux map of Figure 5.15 shows the requisite (small) net heating and net precipitation along the ACC that create lighter surface waters. Part of the adiabatically upwelled water also experiences cooling and brine rejection, hence buoyancy loss, and sinks to make the deep and bottom waters in the Antarctic. The actual location of diapycnal upwelling (buoyancy gain) is likely very complex. Observations and budgets are as yet relatively sparse so we do not have a detailed picture from observations. Much more detail is available from general circulation models than from data, and suggest very localized processes. For the isopycnal layer associated with NADW, much of the diapycnal upwelling occurs in the Southern Ocean south of the ACC, but there is also enhancement along the equator and in other regions associated with the circulation and with complex topography (Figure 14.7).

14.2.3. Meridional Overturning Streamfunction A final quantitative step in depicting the overturning circulation is to compute the meridional

485

overturning transport streamfunction for each ocean and for the globe. The overturning streamfunction is one of the basic diagnostic outputs for ocean models used to study climate, as in the Coupled Model Intercomparison Project. In Chapter 7, we introduced the concept of a streamfunction for geostrophic flow in the horizontal plane (Eq. 7.23f): the velocity is parallel to the streamfunction and its magnitude is equal to the derivative of the streamfunction in the direction perpendicular to the flow. Therefore, the geostrophic streamfunction is the horizontal integral of the geostrophic velocity field. The overturning transport streamfunction is conceptually similar. It is calculated and plotted in a vertical plane, with a single horizontal direction. For the MOC, this horizontal direction is north-south. At any given latitude, the overturning streamfunction J is the vertical integral of the mass transport, summed from the bottom of the ocean (bottom layer) to the surface: Ji ¼

N P i¼1

JðzÞ ¼ JðrÞ ¼

Vi

R z R xeast 0

xwest

R r R xeast o

xwest

vðx0 ; z0 Þdx0 dz0

(14.2a,b,c)

vðx0 ; r0 Þdx0 dr0

The transport streamfunction J has units of transport (Sv). The discrete sum form (14.2a) is the calculation that is actually carried out in N layers that can be defined in depth or density (or any other pseudo-vertical coordinate). Upper case Vi is the transport through the section in each layer, that is, the integral of velocities in that layer, however the layer is defined. For the more mathematical reader/ student, two integral forms are given in Eq. (14.2b,c), to make explicit the difference between integrating in depth or in density. Lower case v is the velocity (in m/sec) perpendicular to the transect, which proceeds from one coast, at xeast, to the other coast, at xwest.

486

14. GLOBAL CIRCULATION AND WATER PROPERTIES

To calculate the overturning streamfunction J (14.2a), it is preferable to subdivide the water column at each latitude into a large number of layers, many more than the 3 or 4 depicted in Figures 14.5 and 14.6. Again, the optimal layers are isopycnal or isoneutral, rather than defined in depth, although most published overturning streamfunctions are computed with depth layers. The overturning streamfunction is calculated at each available latitude (very few for hydrographic data; many for an ocean model). The transport streamfunctions for all latitudes can then be contoured as a function of latitude and vertical coordinate (Figures 14.8 and 14.9). If the layers are defined by isopycnals or neutral density surfaces, the streamfunction can be projected back to depth coordinates by choosing the average depth of the isopycnals at each latitude. The depiction of the overturn in each separate ocean from a global ocean model (Figure 14.8) is representative of most published calculations, although actual numerical values of the overturn (in Sv) differ. These individual ocean overturns were described in Chapters 9e13. 1. The Atlantic has an NADW cell with sinking in the north and an AABW cell with inflow from the south and upwelling into the NADW layer. 2. The combined Pacific/Indian overturn includes inflow of bottom waters (mostly AABW) that upwell into the deep water and thermocline layers, mostly in the Southern Hemisphere and tropics. 3. Meridional overturns in the Northern Hemispheres of the Pacific and Indian are weak. The 2 Sv overturn of NPIW is apparent, as is the weak, deeper overturn of Red Sea Water (RSW) in the Indian Ocean. The global overturning streamfunction is constructed by summing the layer transports for all oceans at each latitude (Figure 14.9). The major features, found in all recent calculations, are (a) shallow subtropical-tropical overturning cells

with sinking at about 30
latitude and rising in the tropics; (b) the large deep cell due to NADW with sinking in the north, occupying most of Northern Hemisphere water column and extending southward to about 35
S; (c) the deep cell centered in the Southern Hemisphere at 3000 m, with northward bottom flow (AABW) and southward deep flow, mainly as PDW and IDW; and (d) a top-to-bottom overturn in the south next to Antarctica that forms AABW. These are all principally connected to diapycnal processes, with downwelling and upwelling limbs. When the globally averaged overturning streamfunction is calculated in depth layers (Figure 14.9a,c) rather than isopycnal layers (Figure 14.9b), the Southern Ocean also includes a strong surface-intensified overturning cell with sinking around 50
S and upwelling between 35 and 50
S. This is called the “Deacon cell.” It mostly disappears when the overturning is calculated in isopycnal layers, that is, it is not associated with a large amount of diapycnal flux. The Deacon cell is mostly due to (adiabatic) flow along isopycnals that change depth and latitude: in the Southern Ocean there is a large component of northward flow that returns southward at the same density but at greater depth (Do¨o¨s & Webb, 1994; Kuhlbrodt et al., 2007). There is also well-documented diapycnal upwelling in the Southern Ocean (Chapter 13). The part of the Deacon cell that is mostly due to depth-averaging is the downwelling limb between 50 and 40
S. The “diabatic” Deacon cell (Speer, Rintoul, & Sloyan, 2000), which involves diapycnal transport, includes northward Ekman transport in the surface layer across the ACC, fed by upwelling from deep waters that mostly rise to the surface adiabatically (along isopycnals) as they move southward across the ACC. Buoyancy gain at the sea surface in the ACC vicinity then allows these waters to move northward, decreasing in density. They are mostly deposited into the Subantarctic Mode Water (SAMW) layer just

GLOBAL MASS TRANSPORTS AND OVERTURNING CIRCULATION

487

FIGURE 14.8 Meridional overturning streamfunction (Sv) from a high resolution general circulation model for the (a) Atlantic, (b) Pacific plus Indian, and (c) Indian north of the ITF. The Southern Ocean is not included. Source: From Maltrud and McClean (2005).

488

14. GLOBAL CIRCULATION AND WATER PROPERTIES

north of the ACC and then move into the Southern Hemisphere gyre circulations to the north.

14.2.4. Overturning Circulation Schematics Here we use schematics to summarize the elements of the global overturn, based on the preceding transport, upwelling, and streamfunction calculations. All such schematics are incomplete since they cannot represent the complexities of the large-scale circulation or eddying and time-dependent paths. Therefore they should always be interpreted cautiously. Richardson (2008) presented a good history of such overturning schematics from the earliest in the nineteenth century to the present.

The widely known popularized depiction of the global NADW cell, often referred to as the “great ocean conveyor,” is shown in Figure 14.10 (after Broecker, 1987, 1991, which were based on Gordon, 1986). Although this diagram has deep deficiencies in terms of representing the actual global overturn, it is nevertheless useful for public education: it is simple and it is global. It nicely illustrates the Atlantic-Pacific/Indian asymmetry in deep water formation, with sinking somewhere near the northern North Atlantic but not in the other two oceans. This particular simplification captures only a part of the global overturn because the important multiple roles of the Southern Ocean were intentionally excluded; other overly simplified descriptions do not include the essential roles of the Indian and Pacific Oceans.

FIGURE 14.9 Global meridional overturning streamfunction (Sv) for (a) a global coupled climate model with high resolution in latitude. Source: After Kuhlbrodt et al. (2007). (b, c) For hydrographic section data at several latitudes, plotted as a function of neutral density and pressure; contour intervals are 2 Sv. The white contours are typical winter mixed layer densities; gray contours indicate bathymetric features (ocean ridge crests). Source: After Lumpkin and Speer (2007).

489

GLOBAL MASS TRANSPORTS AND OVERTURNING CIRCULATION

(b) 24.7

6

26.99

level γ n

27.6 27.88

-14

28

-2

-10 0

28.06 28.11

18

12

28.2

(c)

80S

62S

32S

24N

12

1000

60N

80N

-2

-10 -14

2000

p (dbar)

48N

0

20 10

3000

0

20 10

4000

0

−10 −20

5000 6000 80S

62S

32S

24N

48N

60N

80N

Latitude

FIGURE 14.9 (Continued).

The next three simplified schematics illustrate the elements that we consider essential for a comprehensive teaching presentation of the global overturn. A number of global overturn schematics capture most of the aspects; in particular we note Gordon (1991), Schmitz (1995), Lumpkin and Speer (2007), and Kuhlbrodt et al. (2007). The global overturning can be divided into two major, connected global cells, one with dense water formation around the North Atlantic and the other with dense water formation around Antarctica. These are the NADW 3

and AABW cells, respectively. These two cells are interconnected, especially in the Southern Ocean, complicating any simple representation of the overturn. A third, weak overturning cell is found in the North Pacific, forming a small amount of intermediate water (NPIW); it is mostly unconnected to the NADW/AABW cells, but is included because it contrasts the weakness of dense water formation in this high-latitude region with that in the high-latitude regions of the Atlantic and Southern Ocean. Essential features for the global NADW cell are as follows.3 The NADW cell begins with

We ignore along-isopycnal exchange of deep waters between oceans. We also have to ignore the finer steps of the large-scale upwelling process, which could better be likened to hundreds of steps on different staircases in a building of many floors, rather than a single leap from one very thick layer to another.

490

14. GLOBAL CIRCULATION AND WATER PROPERTIES

FIGURE 14.10 Simplified global NADW cell, which retains sinking only somewhere adjacent to the northern North Atlantic and upwelling only in the Indian and Pacific Oceans. See text for usefulness of, and also issues with, this popularization of the global circulation, which does not include any Southern Ocean processes. Source: After Broecker (1987).

warm water entering the Atlantic from the Indian, via the Agulhas, and from the Pacific, via Drake Passage. This upper ocean water moves northward through the entire length of the Atlantic (becoming first lighter and then denser), and then sinks at several sites in the northern North Atlantic (Nordic Seas, Labrador Sea, and Mediterranean Sea). These denser waters flow south and exit the Atlantic as NADW. Bottom water (AABW) also enters the Atlantic from the south. It upwells into the bottom of the NADW layer in a diffusive process. The warm Indian Ocean source water for NADW includes water from (1) the Pacific via the ITF, (2) the southeastern Indian Ocean south of Australia (sourced from the Southern Ocean and also somewhat from the Pacific), and (3) upwelling from the underlying deep water layers within the Indian Ocean. The ITF waters in the

Pacific Ocean originate from upwelling from deep and intermediate waters within the Pacific (South Pacific and tropics) and from the upper ocean in the southeastern Pacific. The upper ocean source waters from Drake Passage arise in the southeastern Pacific (SAMW and some Antarctic Intermediate Water). Now following the NADW as it leaves the Atlantic, part enters the Indian Ocean directly around the southern tip of Africa, contributing to the IDW. Most enters the ACC, where it upwells. Here it becomes the source for deep water formation around Antarctica. This is the main connection of the NADW and AABW cells. The AABW cell begins with this NADW upwelling to near the sea surface around Antarctica, where it is subjected to brine rejection in polynyas (Chapter 13). The densest waters thus formed sink; the part that escapes

HEAT AND FRESHWATER TRANSPORTS AND OCEAN CIRCULATION

northward across topography and into the main ocean basins is referred to as AABW (although the densest bottom waters are confined to the Southern Ocean). This AABW moves northward at the bottoms of the Atlantic, Indian, and Pacific Oceans. In all three oceans, AABW upwells into the local deep water, that is, into the NADW, IDW, and PDW. Because there are no volumetrically important surface sources of dense water in the northern Indian and Pacific Oceans, this upwelled AABW is the primary volumetric source of the IDW and PDW, whereas AABW is only a minor component of NADW. The IDW and PDW (which can be traced by low oxygen because they are composed of old, upwelled waters) flow southward into the Southern Ocean above the NADW layer because they are less dense than NADW. Here, like NADW, they upwell to the sea surface. However, they upwell farther north than NADW because they are less dense. The upwelled IDW/PDW in the Antarctic feeds two cells: (1) northward flux of surface water across the ACC that joins the upper ocean circulation, accomplished initially by Ekman transport; and (2) the dense AABW formation, which then recycles this mass back through the deep water routes, along with the NADW. The first of these is a major source of the upper ocean waters that then feed northward to the NADW formation region, again connecting the AABW and NADW cells. The vertical pathways connecting NADW, AABW, and also, importantly, IDW and PDW, are illustrated in Figure 14.11c, which is a collapsed, two-dimensional version of Figure 14.11a and b. If we tried to sketch the NADW and AABW cells directly from a global meridional overturning streamfunction (Figure 14.9a), they would appear to be completely separate. This is incorrect as the global average is missing the important basin-specific roles of the Indian and Pacific upwelling and diffusive formation of IDW and PDW, which are

491

high-volume water masses with large associated meridional and upwelling transports.

14.3. HEAT AND FRESHWATER TRANSPORTS AND OCEAN CIRCULATION The global circulation redistributes heat and freshwater within and between the ocean basins. In Chapter 5, the heat and freshwater budgets, airesea fluxes, and meridional transports were described. Here we briefly describe the components of the circulation that redistribute heat and freshwater. Globally averaged, heat is transported meridionally by the ocean from the tropics to higher latitudes. The largest heat gains are in the tropics, with heat gain also in upwelling regions such as the eastern boundary currents. Individually, the Pacific and Indian Oceans move heat poleward. The Atlantic Ocean transports heat northward throughout its length to balance the combined Gulf Stream and Nordic Seas heat loss regions. The meridional heat transports are mostly associated with the upper ocean circulations, which are wind driven. The shallow tropical cells carry heat from the tropics to the subtropics. The subtropical gyres then carry the heat toward the enhanced heat-loss regions of their western boundary currents. The somewhat cooled water returns southward, subducted into the upper part of the subtropical gyres. This leads to a net poleward heat transport in all five anticyclonic subtropical gyres (Figure S14.2 seen on the textbook Web site). The cyclonic subpolar gyres of the North Pacific and North Atlantic also transport heat poleward, with warmer surface inflow in the east cooling to form the colder, denser waters in the northern and western parts of both gyres (NPIW and Labrador Sea Water/NADW). In the subtropical North and South Pacific and Indian Oceans, this upper ocean gyre

492

14. GLOBAL CIRCULATION AND WATER PROPERTIES

process accounts for almost all of the net poleward heat transport. However, in the North Atlantic, the Gulf Stream gyre accounts for only part of the northward heat transport (about 0.4 PW of 1.2 PW total in the calculation in Talley, 2003). In the South Atlantic, the net heat transport is northward (~0.4 PW), toward the equator, even though the upper ocean gyre carries heat southward (~0.1 PW). The formation of NADW, associated with heat loss in the northern North Atlantic and Nordic Seas, accounts for the remaining northward heat transport, due to northward volume transport of warm upper ocean water and southward

return of the new, cold NADW (Figure 5.12 and Figures S5.9 and S14.3 from the textbook Web site). Freshwater is transported by the oceans from regions of net precipitation and runoff to regions of net evaporation. The tropical cells export freshwater poleward from the rainy Intertropical Convergence Zone toward the net evaporation centers (Figure 5.4a). On the poleward side of the evaporation centers, the subtropical gyres transport freshwater equatorward (salty water poleward in the western boundary currents, and freshened subducted water toward the evaporation centers).

FIGURE 14.11 Global overturning circulation schematics. (a) The NADW and AABW global cells and the NPIW cell. (b) Overturn from a Southern Ocean perspective. Source: After Gordon (1991), Schmitz (1996b), and Lumpkin and Speer (2007). (c) Two-dimensional schematic of the interconnected NADW, IDW, PDW, and AABW cells. The schematics do not accurately depict locations of sinking or the broad geographic scale of upwelling. Colors: surface water (purple), intermediate and Southern Ocean mode water (red), PDW/IDW/UCDW (orange), NADW (green), AABW (blue). See Figure S14.1 on the textbook Web site for a complete set of diagrams. This figure can also be found in the color insert. Source: From Talley (2011).

493

HEAT AND FRESHWATER TRANSPORTS AND OCEAN CIRCULATION

(c)

Southern Ocean wind-driven upwelling & surface buoyancy flux

Low, mid-latitude upper ocean waters SAMW, AAIW Pacific-Indian upwelling & diffusion PDW/IDW

UCDW

Antarctica

LCDW AABW formation (brine rejection)

NA

DW

AABW

FIGURE 14.11

(Continued).

PDW/IDW formation (diffusion)

NADW formation (convection)

494

14. GLOBAL CIRCULATION AND WATER PROPERTIES

Of the deeper overturning cells, only the NADW and NPIW overturns carry a significant amount of freshwater equatorward. Both of these cells consist of saltier poleward flow of surface water that is freshened and joined by higher latitude fresh water (Nordic Seas, Arctic and Bering Strait input for the NADW), with southward flow of fresher water. The other three major deep water overturns in the global circulation d formation of AABW, IDW, and PDW d have little impact on either heat or freshwater transport. In the case of IDW and PDW, this is because they form by upwelling of AABW and NADW, so alteration of their properties is due to diapycnal diffusion, which is a slow, weak means of change compared with direct airesea fluxes at surface outcrops. In the case of AABW, even though there is direct atmospheric forcing, the heat and freshwater transports are small because the source water is already cold, so it can be cooled only slightly more and can only be freshened by a limited amount and still remain dense enough to sink.4 Heat and freshwater are also transported in the global overturning circulation by the ITF, moving 10 to 15 Sv from the Pacific to the Indian Ocean, and by flow through the Bering Strait, moving less than 1 Sv of low salinity water (32.5 psu) from the Pacific to the Atlantic. The ITF loop exports heat and freshwater from the Pacific because the ITF is warmer and fresher than the compensating inflow into the Pacific from the Southern Ocean. In the Indian, the ITF imports heat and freshwater input because the ITF is warmer and fresher than the Agulhas outflow that drains the ITF water. Bering Strait exports freshwater from the Pacific to the Atlantic, because the flow through the strait, at 32.5 psu, is fresher than the 4

volumetrically compensating inflow from the Southern Ocean.

14.4. GLOBAL PROPERTY DISTRIBUTIONS We return here to the global perspective of water properties introduced in Chapter 4. We first describe the global pattern of sea level height since it is partly related to the temperature/salinity distribution. We then focus on global summaries of the water masses that were introduced in Chapters 9e13, here relating the property structures to the global circulation and overturning.

14.4.1. Sea Level The ocean’s mean surface height distribution (relative to the global mean surface height) can be inferred from the global dynamic topography of Figure 14.2a (Maximenko et al., 2009). Actual surface height (relative to the geoid) is close to the dynamic height divided by g ¼ 981 cm s2 (Eq. 7.28). The dynamic topography also yields the surface geostrophic circulation. Using the global map, we compare the corresponding large-scale features in each ocean. For instance, the surface height difference from west to east across the North Pacific subtropical gyre is about 70 cm. In contrast, the west-to-east drop across the North Atlantic subtropical gyre is about 40 cm. There is a similar contrast between the South Pacific and South Atlantic subtropical gyres of about 70 cm versus 40 cm difference. This means that there is more equatorward volume transport in the Pacific subtropical gyres than in the Atlantic gyres.

Although the upwelled Antarctic surface waters incorporate a large amount of freshwater in the Antarctic, the newly forming AABW can only be freshened a limited amount and still retain a density that is high enough to allow sinking. The remaining freshwater stays in the upper ocean and is exported in the Southern Ocean’s upper ocean overturns, contributing to Antarctic Intermediate Water (Talley, 2008).

495

GLOBAL PROPERTY DISTRIBUTIONS

The simplest reason is that the Sverdrup transport in the Pacific is proportionally higher than in the Atlantic because the Pacific is that much wider and the winds, and hence Ekman pumping, are similar. Looking at the global scale, the very low surface height in the Southern Ocean contrasts with the rest of the world ocean. The high gradient between the low Southern Ocean pressure and high pressure just to its north marks the eastward geostrophic flow of the ACC, which is principally wind-driven. Separate from, and somewhat masked by these wind-driven gyre differences, a remarkable global feature is the overall higher surface height in the Pacific compared with the Atlantic. This is associated with the relatively lower density of the Pacific compared with the Atlantic, which is associated with the lower mean salinity of the Pacific.

60˚W

80˚N

0˚

60˚E

14.4.2. Water Mass Distributions Water masses in the upper ocean, at intermediate depth (below the pycnocline), in the deep ocean (2000e4000 m), and near the bottom are presented here mostly using schematics; maps and sections were shown in Chapters 9-13. Only a subset of the water masses introduced in previous chapters are included, but these are representative of most of the processes that determine the property structures. The upper ocean water masses are represented here by mode waters, reviewed in more detail in Hanawa and Talley (2001; Figure 14.12). (Unrepresented by this schematic are the upper ocean water masses associated with subduction d the Central Water and Subtropical Underwater of the main pycnocline of each ocean basin.) All mode waters are associated with strong fronts, most of which are

180˚

120˚E

120˚W

1

60˚ 26.9–27.75 26.2

40˚ 7

26.5–26.8

26.5

25.2

24 25.4 24–25.4

20˚ –

0˚

25.5

20˚ 0

25.5 26.2–26.7

26.2–26.3

26.0 26.0

40˚ 26.85 27.1

26.95

60˚

80˚S

FIGURE 14.12 Mode Water distributions, with typical potential densities and schematic subtropical gyre, and ACC circulations. Source: After Hanawa and Talley (2001). Medium grays are STMWs in each subtropical gyre. Light grays are eastern STMWs in each subtropical gyre. Dark grays are SPMW (North Atlantic), Central Mode Water (North Pacific), and SAMW (Southern Ocean).

496

14. GLOBAL CIRCULATION AND WATER PROPERTIES

well-known strong currents, such as the Gulf Stream, Subantarctic Front, and so forth. These fronts have strongly sloping isopycnals that favor lower stratification and hence deeper mixing on the warm side of the front. Subtropical Mode Waters (STMWs) are associated with each subtropical western boundary current. STMWs fill a large portion of the western subtropical gyres. Each STMW has a temperature of around 16 to 19
C; the ubiquity of this temperature is due to the similarity of western boundary current separation latitudes and the surface temperature distribution in each subtropical gyre (Figure 4.1). However, the potential densities of the STMWs differ greatly because of the salinity differences between the oceans. The North Atlantic is the saltiest, so it has the densest STMW, and so forth. STMW formation mechanisms include deep winter mixed layer outcrops close to the strong fronts, preconditioned by the isopycnal slopes associated with the fronts, and cross-frontal transports driven by wind or the dynamics of the front. Each STMW is subducted into the interior of its subtropical gyre and becomes isolated from the sea surface within a few hundred kilometers of the front. Eastern STMWs (lighter grays Figure 14.12) in each ocean basin are less dense than STMWs in most oceans, and are the least dense class of mode waters shown in the map. They are found where the fronts that define the gyres swing southward, except in the North Atlantic, where the Azores Current is the relevant front, and head directly for the Strait of Gibraltar. Subpolar Mode Water (SPMW) in the North Atlantic is associated with the northeastward flow of the NAC and the cyclonic Irminger and Labrador Sea circulations. Within the NAC separation region, SPMW functions like an STMW, and subducts southward into the subtropical North Atlantic circulation. SPMW in the northeastern North Atlantic is associated with the northeastward branches of the NAC that enter the Norwegian Sea. SPMW in the

northwestern North Atlantic ultimately becomes the new Labrador Sea Water (LSW) that sinks and spreads away from the Labrador Sea (McCartney & Talley, 1982). Formation of SPMW is like that of STMW: deep winter mixed layers on the warm side of the strong fronts. Central Mode Water in the North Pacific is associated with the eastward flow of the North Pacific Current’s subarctic front, which is more or less a continuation of the Oyashio, and lies north of the Kuroshio front. Again, this front favors deep mixed layers on its warm side. SAMW is the series of mode waters along the northern side of the Subantarctic Front (McCartney, 1977, 1982). Like STMWs, these are associated with deep winter mixed layers within several hundred kilometers or less of the strong front. The SAMWs subduct northward into the subtropical gyres where they become an important part of the pycnocline. The best-developed (thickest, lowest stratification) SAMWs are found in the eastern Indian Ocean and across the Pacific where winter mixed layers are thickest. After subducting northward into the pycnocline, these SAMWs carry tracers associated with large surface ventilation (high oxygen, chlorofluorocarbons; CFCs) far into the Indian and South Pacific. The major intermediate waters of the global ocean, each identified by a vertical salinity extremum, are shown in Figure 14.13. The greens and blues are low salinity intermediate waters and the oranges are high salinity intermediate waters. Each of these intermediate waters forms predominantly in a localized region (locations marked on the map) and then is advected by the circulation. Each is associated with the global overturning circulation, in that formation involves a conversion of surface waters to densities that reach to intermediate depths, below the pycnocline. The overall reach of each intermediate water is greater than indicated by the location of its vertical extremum, which is simply an imperfect marker of the spread of water from a surface source. For instance, most of the

497

GLOBAL PROPERTY DISTRIBUTIONS

80˚N

60˚W

0˚

60˚E

120˚E

60˚

Labrador Sea Water 27.8 σθ

40˚

20˚

Mediterranean Water 28.0 σθ

180˚

120˚W

North Pacific Intermediate Water 27.0 σθ

Red Sea Water 27.7 σθ

0˚

20˚

40˚

Antarctic Intermediate Water 27.1 σθ

60˚

80˚S

FIGURE 14.13 Low- and high-salinity intermediate waters. AAIW (dark green), NPIW (light green), LSW (dark blue), MW (orange in Atlantic), RSW (orange in Indian). Light blue in Pacific: overlap of AAIW and NPIW. Light blue in Indian: overlap of AAIW and RSW. Cross-hatching: mixing sites that are particularly significant for the water mass. Red dots indicate the primary formation site of each water mass; fainter dots mark the straits connecting the Mediterranean and Red Seas to the open ocean. The approximate potential density of formation is listed. This figure can also be found in the color insert. Source: After Talley (2008).

Okhotsk Sea water that provides the NPIW salinity minimum in the subtropical North Pacific resides in the subpolar gyre; however, it is not a vertical salinity extremum, and is therefore not as easily identified. The three major low salinity intermediate waters d LSW, NPIW, and AAIW d result from relatively fresh, cold, dense water at subpolar latitudes that sinks beneath the warmer, saltier, lighter subtropical waters. The formation mechanism differs for each of these water masses: LSW, deep convection and sinking in the Labrador Sea; NPIW, brine rejection and sinking in the Okhotsk Sea followed by strong mixing in the Kuril Island passages; AAIW, deep mixed layers and underlying fresh subantarctic water sinking in the Drake Passage region and subducted smoothly northward into the Pacific and subducted with large mixing northward into the Atlantic/Indian.

The temperatures of these three low salinity intermediate waters are similar: 3 to 5
C. Their densities differ greatly because the overall salinity of their respective oceans differs. NPIW forms in the relatively fresh North Pacific and is the freshest and least dense of the intermediate waters, while LSW forms in the salty North Atlantic and is the saltiest and most dense. The high salinity intermediate waters, Mediterranean Water (MW) and RSW, result from high evaporation and winter cooling in the Mediterranean/mid-east region. The resulting deep convection is strongly localized within the Mediterranean and Red Seas. The dense waters flow out over sills through narrow straits to join their respective ocean circulations, both sinking to intermediate depth and entraining large amounts of ambient ocean water as they equilibrate at depth in the open ocean. Once equilibrated,

498

14. GLOBAL CIRCULATION AND WATER PROPERTIES

they are much warmer than the low salinity intermediate waters, of the order 12 to 15
C, but they are dense because of their high salinity. NADW and AABW (or LCDW) are the two large-scale dense/bottom water masses that are always included in describing the global overturning circulation. Both are formed at the sea surface by buoyancy loss due to airesea fluxes. The overturn associated with NADW is estimated at 15 to 20 Sv. The overturning estimates associated with AABW range from 12 to more than 25 Sv. IDW and PDW are not represented in the maps included here. Both are formed by upwelling of bottom water (AABW) and admixture of NADW within their respective oceans. Downward diffusion from above modifies their properties relative to their deep source waters.

(a)

They have little or no surface source, and therefore the water mass decompositions shown in Figures 14.14 and 14.15 focus on NADW and AABW. The sources of NADW and AABW are represented by the map showing the location of a deep isopycnal in Figure 14.14a. At this density, these two sources are separated by topography south of Nova Scotia and Newfoundland (which thus under-represents the global reach of NADW as described in the following paragraphs). The dense source of Nordic Seas Overflow water is deep convection east of Greenland. The multiple, distributed sources of AABW are brine rejection due to sea ice formation in polynyas within the Weddell and Ross Seas and at several locations along the Antarctic coast.

Location of the isopycnal σ4 = 45.92 80˚N

60˚W

0˚

60˚E

120˚E

180˚

120˚W

Greenland Sea Deep Water 28.1 σθ

40˚

80˚N

80˚S

60˚

Nordic Seas Overflow Water 27.9 σθ 46.1 σ4

20˚

0˚

20˚

40˚

Antarctic Bottom Water: 27.7-27.9 σθ

60˚

45.9-46.2 σ4

Adélie Land Bottom Water 80˚S

Weddell Sea Bottom Water

Ross Sea Bottom Water

FIGURE 14.14 Deep and bottom waters. (a) Distribution of waters that are denser than s4 ¼ 45.92 kg/m3. This is approximately the shallowest isopycnal along which the Nordic Seas dense waters are physically separated from the Antarctic’s dense waters. At lower densities, both sources are active, but the waters are intermingled. Large dots indicate the primary formation site of each water mass; fainter dots mark the straits connecting the Nordic Seas to the open ocean. The approximate potential density of formation is listed. Source: After Talley (1999). (b) Potential temperature (
C), and (c) salinity at the ocean bottom, for depths greater than 3500 m. Source: After Mantyla and Reid (1983).

GLOBAL PROPERTY DISTRIBUTIONS

499

FIGURE 14.14 (Continued).

The global maps of bottom potential temperature and salinity in Figure 14.14b,c show the contrasting warm, saline Nordic Seas and cold, fresh AABW properties. The NADW occupies most of the bottom of the North Atlantic and

eastern South Atlantic, as also seen in the water mass decomposition considered next (Figure 14.19). The colder AABW occupies the Southern Ocean, the western South Atlantic, and dominates in the Indian and Pacific Oceans.

500

14. GLOBAL CIRCULATION AND WATER PROPERTIES

FIGURE 14.15 Fractions of NADW and AABW. (a) Fraction of NADW on the isoneutral surface gN ¼ 28.06 kg/m3 (s4 ~ 45.84 kg/m3, at a depth of 2500e3000 m north of the ACC; G. Johnson, personal communication, 2009). (b) Fraction of AABW in the bottom water (with the remaining fraction being mostly NADW). Source: From Johnson (2008). Both maps are from an OMP analysis using as inputs the properties of NADW at a location just south of Greenland, downstream from the Nordic Seas Overflows, and of AABW in the Weddell Sea. The complete figures are reproduced on the textbook Web site as Figures S14.4 and S14.5.

EDDY VARIABILITY AND DIFFUSIVITY

The effect of vertical mixing is apparent in these global maps. The western Indian Ocean has higher bottom salinity than the entering AABW, which fills the bottommost layer. This is due to downward mixing from the overlying higher salinity RSW. In the Pacific Ocean, the bottom salinity distribution also indicates downward mixing from overlying waters: in the southwestern Pacific, a deep vertical salinity maximum remains from the NADW influence in the circumpolar deep waters, and this creates higher salinity at the bottom. Farther north in the Pacific, the bottom waters are fresher and warmer, which is partially due to the elimination of the densest bottom waters through upwelling (diapycnal mixing) and downward mixing of overlying fresher water. Full explanation requires detailed consideration of properties on deep isopycnal surfaces. The global reach of NADW is demonstrated using the fraction of NADW (compared with AABW) on an isopycnal surface that typifies the high salinity core in the ACC (Figure 14.15 and Figure S14.4 seen on the textbook Web site). The maps were computed by G. Johnson (personal communication, 2009) using his (Johnson, 2008) application of optimum multiparameter analysis (OMP) (Section 6.7.3) to global water masses. Reid and Lynn (1971) were the first to show and describe the global pattern of salinity on this NADW isopycnal. Salinity on nearly the same isopycnal surface in the Southern Ocean is shown in Figure 13.17. Water with a large fraction of NADW reaches southward from its source in the northern North Atlantic, down along the western side of the South Atlantic, and then eastward at 20e30
S. There is a transition to a lower NADW fraction (i.e., salinity) between 30 to 40
S, which indicates the onset of much more AABW. A tongue of higher NADW fraction (i.e., salinity) spreads eastward south of Africa and then in patches along the core of the ACC eastward into the Pacific Ocean. The pattern of somewhat higher fractions (salinity) then extends northward

501

into the Pacific Ocean in the DWBC just east of New Zealand. Along the entire path, the NADW fraction (higher salinity) decreases downstream as the AABW fraction increases. The global reach of AABW is also demonstrated with OMP analysis applied at the ocean bottom (Figure 14.15b and S14.5 from the textbook Web site). The pattern is similar to that on the deep isopycnal (Figure 14.15a). AABW dominates the world ocean’s bottom water, but with a respectable NADW fraction of about 0.3. AABW is mostly blocked from crossing the mid-Atlantic and Walvis Ridges in the South Atlantic, so NADW dominates the eastern South Atlantic as well as the North Atlantic. In both the deep and bottom water maps, AABW covers significantly more of the ocean than NADW. Johnson (2008) estimated that two-thirds of the deep and bottom water arises from AABW and one-third from NADW. If the overturning rates for NADW and AABW of 19 and 28 Sv from Figure 14.5 are used together with Johnson’s (2008) calculations of the volumes of NADW and AABW, a residence time of about 500 years for both water masses is obtained. If, however, the overturning rates for the two water masses are more like 17 Sv each, as summarized in Johnson (2008), the residence times differ, with NADW around 500 years and AABW around 870 years. Uncertainties in these values are large. As both water masses are in similar deep/bottom water environments, differences in residence times would rely on differences in how they are affected by geographically heterogeneous diapycnal diffusion.

14.5. EDDY VARIABILITY AND DIFFUSIVITY This introductory textbook is mostly written from the point of view of a mean circulation and simple departures from the mean, including some seasonal and climate variability, and

502

14. GLOBAL CIRCULATION AND WATER PROPERTIES

energetic, recurring time-dependent features such as Gulf Stream or Agulhas rings. However, all regions of the ocean have some level of eddy variability, which is defined as the departure of instantaneous velocities or sea surface/isopycnal heights from the mean. Variability can range from random noise, to wavelike disturbances, to closed, coherent features. Eddy variability is responsible for stirring in the nearly horizontal (along-isopycnal) direction, and is thus critically important to along-isopycnal eddy diffusivity (Sections 7.2.4 and 7.3.2). Warm- and cold-core rings generated by the meandering of major currents are large, closed features that one might typically associate with “eddies.” On the other hand, eddy variability in the central parts of the gyres may look more like spectral noise. Some kinds of eddies extend from the sea surface to the bottom, while others are concentrated in the surface layer, and others can be embedded entirely within subsurface layers. Horizontal, eddy length scales are kilometers to thousands of kilometers. Timescales are typically weeks to months, but can sometimes be many years for coherent vortices such as Meddies (Chapter 9). This is considered to be the ocean’s mesoscale. These are the length and timescales associated with planetary waves such as Rossby and Kelvin waves. The most important length scale for this variability is the Rossby deformation radius, which depends on latitude and vertical stratification (Section 7.7.4, Figure S7.30 on the textbook Web site). A new category of shorter submesoscale variability, mostly associated with the surface layer, is now being vigorously analyzed through theory, modeling, and observations. Because this layer is so shallow, the horizontal spatial scales are small, on the order of kilometers. (This can be thought of as related to an internal Rossby deformation radius using a vertical length scale of about 100 m rather than 1000 m.) Internal waves and tides (Sections 7.5.1, 8.4, and 8.6) have even shorter timescales. This high frequency variability is critically important

for ocean turbulence and hence diapycnal diffusivity and mixing (Sections 7.3 and 7.4). We therefore present some recent global results for near-inertial and tidal variability.

14.5.1. Eddy Energy and Lateral Eddy Diffusivity Distributions The basic physics concepts of kinetic and potential energy were described in terms relevant to the ocean in Section 7.7.5. Eddy kinetic energy (EKE) is calculated using departures of the instantaneous (synoptic) velocity from the mean velocity, regardless of how the mean is defined (leaving some ambiguity that should be carefully described in any given study). EKE maps are almost always based on lateral currents and not on the vertical velocities, which are much smaller (but important for the diapycnal eddy diffusivity described in the following section). EKE for surface currents was first derived from ship drift observations, but is now much more easily constructed from surface drifter velocities and from surface velocities derived from altimetric surface heights. EKE maps for deeper levels are calculated from Lagrangian float observations; moored current meter arrays are also used to calculated eddy energy locally. Eddy potential energy is calculated using departures of instantaneous sea surface height and isopycnal heights from their mean values; currently satellite altimetry data are valuable for this, and in situ Argo profiling float data set will also be valuable after many more years of data are collected. Surface EKE (Figure 14.16 and Figure S14.6 seen on the textbook Web site) is mostly related to the mean current speeds (Figure 14.2b, Section 14.1). Eddy energy has several sources, including current instabilities (Section 7.7.5). Mean flows with strong velocity shear in both the horizontal and vertical tend to be the most unstable. Horizontal shear generates barotropic instabilities that draw energy from the shear. Vertical shear in geostrophic flow is associated

EDDY VARIABILITY AND DIFFUSIVITY

503

FIGURE 14.16 Eddy kinetic energy (cm2 s2) from surface drifters. Source: From NOAA AOML PHOD (2009). A complementary figure based on satellite altimetry (from Ducet, Le Traon, & Reverdin, 2000) is reproduced in Figure S14.6c on the textbook Web site. This figure can also be found in the color insert.

with sloped isopycnals; the eddy energy is generated through release of the potential energy of the sloping isopycnals through baroclinic instability. Overall, satellite altimetry analysis has reinforced the importance of flow instabilities, and especially of baroclinic instability, in all regions (Stammer, 1998). On the other hand, mid-ocean eddies away from strong currents can be generated by mechanisms other than flow instability, that is, through direct wind forcing. An easily visible example in Figure 14.16 is the high EKE band just west of Central America; these are eddies in the Gulf of Tehuantepec, forced by very strong winds through the adjacent mountain passes (Figures 5.16 and 10.21). Although it is mostly related to the currents, the EKE distribution differs in important ways from the speed distribution (Figure 14.2b and Figure S14.6 seen on the textbook Web site). The strongest eddies, such as Agulhas rings, propagate away from the mean flow that created them, accounting for broader EKE maxima compared with the speed (mean kinetic

energy) maxima. The most striking large-scale difference of EKE from the mean speed distribution is the high EKE in bands around 20 to 30
latitude, especially in the Pacific and Indian Oceans, but also with a signature in the Atlantic. These regions have low mean surface velocities, yet the enhanced EKE stands out starkly in the global mean. These regions have shallow eastward surface flow with underlying westward flow (the Subtropical Countercurrents in the Pacific, Eastern Gyral Current in the Indian, and the Azores Current and Subtropical Countercurrents in both hemispheres in the Atlantic). This vertical shear is associated with tilted isopycnals and enhanced baroclinic instability (e.g., Palastanga, van Leeuwen, Schouten, & deRuijter, 2007; Qiu, Scott, & Chen, 2008). Horizontal eddy diffusivity can be calculated from the eddy variability measured by Lagrangian drifters. The highest values of surface eddy diffusivity might exceed 2  104 m2/sec (2  108 cm2/sec) (Figure 14.17a and Figure S14.7 on the textbook Web site). The pattern of surface diffusivity corresponds roughly to the

504

14. GLOBAL CIRCULATION AND WATER PROPERTIES

FIGURE 14.17 (a) Horizontal eddy diffusivity (m2/sec) at the sea surface (color) with mean velocity vectors, based on surface drifter observations. Source: From Zhurbas and Oh (2004). (b) Eddy diffusivity ellipses at 900 m based on subsurface float velocities. Colors indicate different scales (see figure headers). Source: From Davis (2005). The Atlantic surface map and Indian 900 m map from the same sources are reproduced in Chapter S14 (Figures S14.7 and S14.8) on the textbook Web site. Both Figures 14.7a and 14.7b can also be found in the color insert.

EDDY VARIABILITY AND DIFFUSIVITY

EKE pattern. The eddy diffusivity is not exactly proportional to EKE because timescales for diffusion depend on location and on the underlying processes that create the variability (Lumpkin, Treguier, & Speer, 2002; Shuckburgh, Jones, Marshall, & Hill, 2009). The map in Figure 14.17a shows a scalar diffusivity, calculated using a modified version of Davis’ (1991) method; the full horizontal diffusivity is a tensor, hence can have different values in the zonal and meridional directions, since velocity variations can be in any direction relative to the mean velocity. Below the sea surface, out of reach of satellites and surface drifters, eddy statistics are more difficult to compile.5 Horizontal eddy diffusivities based on subsurface floats at 900 m (Figure 14.17b) have maximum values of about 0.8  104 m2/sec, which are robustly less than those at the sea surface despite differences in computation methods, including the use of ellipses that show the directional difference in diffusivity. As at the sea surface, the 900 m eddy diffusivity is high mostly where currents are strong. It is also mostly isotropic (the ellipses are “round”) except in the tropics, where it is highly directional, with much larger values in the east-west direction, matching the strongly zonal direction of the tropical currents (Davis, 2005).

14.5.2. Observed Scales, Speeds, and Coherence of Eddy Variability This is a very brief introduction to the large amount of work describing the ocean’s eddy variability. A simple time-space display (Hovmo¨ller diagram) of sea-surface height (SSH) anomalies at mid-latitude in each ocean 5

505

(Figure 14.18) reveals generally westward propagation of the dominant features, which is typical behavior for Rossby waves (Section 7.7.3). Similar patterns are found at almost all latitudes, except near the equator where eastward-propagating Kelvin waves are also found, and in strong eastward flows such as the ACC that advect the variability to the east (a Doppler shift). Phase speeds calculated from SSH plots like those of Figure 14.18 yield robustly westward propagation, close to the speeds of firstmode baroclinic Rossby waves (Figure 14.19; Chelton & Schlax, 1996; Stammer, 1997). However, the difference from simple Rossby wave speeds is important: the observed speeds are almost twice as fast at mid-latitudes. The non-Rossbywavelike behavior of the variability is likely due to nonlinear interactions with other modes (e.g., Wunsch, 2009), and to the prevalence of coherent eddies that propagate westward (Figure 14.21). Such coherent eddies are nonlinear by definition. Frequency and wavenumber spectra (Section 6.5.3) provide statistical information about variability observed from satellites, current meters, and so forth. The directional wavenumber spectrum from satellite altimetry in Figure 14.20b again shows that most energy propagates westward (solid curve) rather than eastward (dashed). In the frequency spectrum, the annual cycle is the most energetic signal (peak indicated by dashed vertical line in the left panel), because this is the strongest external forcing frequency, associated with seasonal changes. Other than this peak, the frequency spectrum is relatively smooth. The spectra in Figure 14.20 are nearly flat at lower frequencies and wavenumbers (longer

The best EKE estimates are made at long-term current meter moorings, which tend to be deployed in dynamically interesting regions such as the Gulf Stream, with little sampling elsewhere. Subsurface floats provide information at their target depths. Acoustically tracked floats provide the best statistics since their locations are observed nearly continuously, but they are not global. Profiling floats that are tracked when they surface, approximately every 10 days, can provide global statistics, although global maps are not yet available.

506

14. GLOBAL CIRCULATION AND WATER PROPERTIES

FIGURE

14.18

Surface-height anomalies at 24 degrees latitude in each ocean, from a satellite altimeter. This figure can also be found in the color insert. Source: From Fu and Chelton (2001).

EDDY VARIABILITY AND DIFFUSIVITY

FIGURE 14.19 (a) Westward phase speeds (cm/sec) in the Pacific Ocean, calculated from the visually mostdominant SSH anomalies from satellite altimetry. The underlying curves are the fastest first-mode baroclinic Rossby waves speeds at each latitude. (b) The ratio of observed and theoretical phase speeds, showing that the observed phase speeds are generally faster than theorized. Source: From Chelton and Schlax (1996).

time and space scales), and steeply sloped at higher frequencies and wavenumbers. (The flat parts are called “white” because white noise includes all frequencies with roughly the same energy; the sloped parts are called “red” because they slope up to higher energy at lower frequency.) The spectra transition from flat to sloped at a relatively well-defined frequency

507

and wavenumber; this can be called a “cutoff” frequency or wavenumber. The cutoff marks a shift in the physical processes that dominate the white versus the red parts of the spectra. The spectral slopes are consistent with generation of energy through baroclinic instability (Section 7.7.5) rather than external changes in forcing (mainly wind; Stammer, 1997). One explanation for the lack of detailed correspondence between observations of westward propagation and the obvious initial explanation of Rossby waves is that a great deal of energy is actually contained in coherent eddies, which differ from Rossby wave behavior in many ways, but retain the westward propagation of Rossby waves. In each ocean basin chapter (9e13), we described some of the major eddy (ring) formation and propagation locations. These phenomena often have names (“Halmahera eddy,” “Gulf Stream rings,” “Agulhas rings,” “Queen Charlotte Eddy,” “Brazil Current Rings,” etc.) because they occur so frequently in a given location and so greatly dominate variability and often transport of properties near those locations. There have been indications of more widespread coherent eddies in Lagrangian data sets (e.g. Shoosmith, Richardson, Bower, & Rossby, 2005). Coherent vortices (eddies) have now been shown, from satellite altimetric data, to exist in large regions of the oceans (Figure 14.21; Chelton, Schlax, Samelson, & de Szoeke, 2007). These maps of vortices resemble, to some extent, the EKE maps of Figure 14.16. The major bands of eddies are in the western boundary currents/ extensions, in large bands at mid-latitudes where flow is eastward (subtropical countercurrents, Azores Current), and in the ACC. The eddies mostly propagate westward; in the eastward flow of the ACC, they are advected eastward. These eddies are strongest (highest SSH) in the western boundary current extensions, but their populations are highest in the ACC and mid-latitude bands (subtropical

508

14. GLOBAL CIRCULATION AND WATER PROPERTIES

FIGURE 14.20 (a) Frequency and (b) wavenumber spectra of SSH in the eastern subtropical North Pacific, using 15 years of satellite altimetry observations. The dashed line in (a) is the annual frequency. In the wavenumber panel, solid is westward propagating, and dashed is eastward propagating energy. Source: From Wunsch (2009).

countercurrents). The eddy diameters are generally larger than the Rossby deformation radius (Figure S7.30 seen on the textbook Web site), exceeding 200 km in the regions most dominated by eddies, and dropping off to about 100 km at higher latitudes (Chelton et al., 2007).

14.5.3. Diapycnal Diffusion and Near-Inertial Motion In the ocean’s overturning circulation, density is altered along the circulation path (Section 7.10). Surface water becomes dense enough to sink to great depth mostly in well-defined small regions. The dense waters eventually return to lower density, due to downward diapycnal eddy diffusion of buoyancy (Figure S7.40 on the textbook Web site). In the ocean’s interior, this is a weak and slow process, associated with turbulence generated by internal wave breaking (Section 7.3.2). Based on the observed ocean stratification and simple models, the globally averaged diapycnal eddy diffusivity is

about 104 m2/sec (Munk, 1966; see Sections 7.3.2 and 7.10.2). Within the main pycnocline the diapycnal eddy diffusivity is much smaller, of the order 105 m2/sec (e.g., Gregg, 1987; Ledwell, Watson, & Law, 1993). On the other hand, diapycnal diffusivity as observed near the ocean bottom, and up into the water column over regions of very rough topography, is higher than the Munk value (Figure 7.2). If isopycnals are raised up to near the sea surface through mechanical forcing (e.g., Ekman suction due to wind stress curl) or due to sloping associated with strongly sheared geostrophic currents, then much higher diffusivities are available to waters on those isopycnals because of the much higher levels of near-surface turbulence due to wind and airesea flux forcing. This uplift occurs in shallow cells in the tropics and along the eastern boundaries, where upwelling is strong, and also in the Southern Ocean and subpolar North Pacific, where open-ocean isopycnals rise up to the sea surface.

EDDY VARIABILITY AND DIFFUSIVITY

509

FIGURE 14.21 Tracks of coherent cyclonic and anticyclonic eddies with lifetimes of more than 4 weeks, based on altimetric SSH, color coded by a “nonlinearity parameter,” which is the ratio of velocity within the eddy compared with the eddy propagation speed. White areas indicate no eddies or trajectories within 10 degrees latitude of the equator. This figure can also be found in the color insert. Source: From Chelton et al. (2007).

510

14. GLOBAL CIRCULATION AND WATER PROPERTIES

Wind-forced near-inertial motion (Figure 7.4) in the surface layer is expected to result in higher diapycnal diffusivities there; the geographical distribution of this motion should be an indication of geographic variations in surface layer diffusivity. The near-inertial motion has been mapped globally from the drifter data set (Figure 14.22a). Mean speeds of the near-inertial motions are 10 cm/sec, ranging up to much higher than 20 cm/sec beneath the atmosphere’s mid-latitude storm tracks and in the eastern tropical Pacific and western tropical Atlantic. Observed inertial current radii are 10e30 km (Chaigneau, Pizarro, & Rojas, 2008). Energy spectra from surface drifter velocities show the prevalence of inertial energy at all latitudes (Figure 14.22b). This is important because it demonstrates that the wind energy is indeed concentrating in the inertial band, and therefore energy for the surface layer’s turbulent mixing is largely near-inertial. Thus the map of inertial energy in Figure 14.22a reflects the spatially varying capability of the surface ocean to mix. The inertial frequency depends on latitude, going from 0 at the equator to almost 2 cycles per day at 70
latitude (solid curve in the figure). There is also energy at low frequencies at all latitudes, which is largely due to geostrophic motion, and energy at the semidiurnal period (2 cycles per day; vertical yellow bands in the figure) mainly due to tides.

14.6. CLIMATE AND THE GLOBAL OCEAN In present usage, climate variability refers to natural climate variability and climate change refers to anthropogenically forced variations in climate. The latter is also referred to as “global change.” We include climate in an oceanography text not necessarily because of oceanatmosphere feedbacks, which might be weak in all but the tropical modes, but because

FIGURE 14.22 Near-inertial motion. (a) Average inertial current speeds (cm/sec), based on surface drifters. Source: From Chaigneau et al. (2008). (b) Rotary power spectra in 2.5 degree latitude bins in the Pacific Ocean. The solid curve is the inertial frequency at each latitude; the dashed curve is twice the inertial frequency. Negative frequencies rotate counterclockwise and positive frequencies rotate clockwise. Source: From Elipot and Lumpkin (2008). Both Figures 14.22a and 14.22b can also be found in the color insert.

climate variability and change affect ocean variability in properties and circulation. All of the remaining text, figures, and tables relating to climate variability have been moved to Chapter S15 (Climate Variability and the

CLIMATE AND THE GLOBAL OCEAN

Oceans) on the supplemental Web site for the textbook, which also includes climate variability materials from each of the basin chapters. In the supplement, we present figures and a table of the modes of the interannual, decadal, and longer

511

term climate variability that appear to have the most imprint on the ocean. We also discuss changes in ocean properties (temperature, salinity, oxygen) and to some extent circulation, as they relate to climate variability and climate change.