Ocean circulation: thermohaline conveyor, gyres, upwelling zones

Intuition Beginner

Ocean water is constantly moving in vast, interconnected patterns. Surface currents are driven by wind and Earth's rotation, forming circular patterns called gyres — clockwise in the Northern Hemisphere, counterclockwise in the Southern. The Gulf Stream is a powerful western boundary current that carries warm water from the Gulf of Mexico toward northern Europe, making Europe much warmer than similar latitudes in Canada.

Below the surface, a much slower but enormous circulation — the thermohaline conveyor — moves water around the globe based on differences in temperature and salinity. Cold, salty water sinks in the North Atlantic, flows south at depth, eventually surfaces in the Indian and Pacific Oceans, then returns at the surface. A full circuit takes roughly 1,000 years.

Upwelling brings deep, nutrient-rich water to the surface along certain coastlines and at the equator. This process fuels some of the most productive fisheries on Earth, including those off Peru and California. Without upwelling, surface waters would be starved of the nutrients that sustain marine food webs.

Visual Beginner

Circulation type	Driving force	Depth	Timescale	Example
Surface gyre	Wind + Coriolis	0-500 m	Months to years	North Atlantic Gyre
Western boundary current	Wind + beta effect	0-1000 m	Weeks to months	Gulf Stream, Kuroshio
Thermohaline conveyor	Density (T, S)	Full depth	~1000 years per circuit	NADW formation
Coastal upwelling	Wind-driven Ekman transport	0-300 m	Days to weeks	Peru, California coasts
Equatorial upwelling	Divergent Ekman transport	0-200 m	Weeks	Eastern equatorial Pacific

Worked example Beginner

London sits at roughly 51.5 degrees north latitude, about the same as Calgary, Canada. Yet London's January average temperature is around 5 degrees Celsius while Calgary's is around -10 degrees Celsius. Why the 15-degree difference?

The Gulf Stream and its extension, the North Atlantic Drift, transport enormous quantities of heat northward. This warm current carries roughly 30 million cubic meters of water per second — over 100 times the combined flow of all the world's rivers. As this warm water reaches the North Atlantic, it releases heat to the atmosphere, which prevailing westerly winds carry across Europe.

The same mechanism explains why Norway's coast remains ice-free despite lying above the Arctic Circle, while similar latitudes in Canada experience months of frozen coastline. The ocean is acting as a vast heat-delivery system, redistributing tropical warmth toward the poles through both surface currents and the thermohaline conveyor.

If the thermohaline circulation weakened — for example, because freshwater from melting ice reduced North Atlantic salinity — this heat transport would diminish, and Europe could cool by several degrees even as the rest of the planet warms.

Check your understanding Beginner

Formal definition Intermediate+

Wind-driven circulation refers to the upper-ocean currents forced by surface wind stress and modulated by the Coriolis effect. The Ekman spiral describes how current direction rotates with depth under steady wind forcing: each successive layer is deflected further from the wind direction, and net (depth-integrated) Ekman transport is 90 degrees to the right of the wind in the Northern Hemisphere.

Subtropical gyres are large-scale circular current systems driven by the curl of the wind stress. Each gyre consists of an equatorward-flowing eastern boundary current (broad, slow, cold), a poleward-flowing western boundary current (narrow, fast, warm), and zonal connecting flows at the equatorward and poleward extremes. The five major subtropical gyres occupy the North and South Atlantic, North and South Pacific, and Indian Ocean basins.

Western boundary current intensification (Stommel 1948) is the phenomenon by which currents on the western side of an ocean basin are much stronger and narrower than those on the eastern side. The beta effect — the variation of the Coriolis parameter $f=2\Omega \sin \varphi$ with latitude — is responsible. To close the vorticity budget of a wind-driven gyre, the return flow on the western side must be narrow and intense.

Thermohaline circulation (also called the meridional overturning circulation, MOC) is the density-driven component of ocean circulation involving the full ocean depth. Water sinks where it becomes sufficiently dense (cold and salty) and rises where it is brought to the surface by wind-driven upwelling or mixing. The primary deep water masses are North Atlantic Deep Water (NADW) and Antarctic Bottom Water (AABW).

Upwelling is the vertical transport of deep water to the surface, occurring along coastlines where winds drive offshore Ekman transport (coastal upwelling), at the equator where Ekman transport diverges (equatorial upwelling), and in the Southern Ocean where westerly winds drive divergent surface flow. Downwelling is the opposite process, where surface water is pushed downward.

Ekman transport and the Ekman spiral

Under steady wind stress on an infinite ocean, the steady-state balance between Coriolis force and friction produces the Ekman spiral. The depth-integrated horizontal transport per unit width is:

$${\mathbf{M}}_E=\frac{\mathbf{\tau }\times \mathbf{k}}{{\rho }_{0}f}$$

where $\mathbf{\tau }$ is the wind stress, $\mathbf{k}$ is the vertical unit vector, ${\rho }_0$ is the reference density, and $f$ is the Coriolis parameter. In the Northern Hemisphere, the transport is 90 degrees to the right of the wind.

The Ekman layer depth, over which the wind-driven current decays, scales as:

$$D_E\approx \sqrt{\frac{2A_v}{|f|}}$$

where $A_v$ is the vertical eddy viscosity. Typical values give $D_E\approx 50$–$200$ m.

Sverdrup balance

The Sverdrup relation connects the vertically integrated meridional transport to the wind stress curl:

$$\beta V=\frac{1}{{\rho }_0}(\nabla \times \mathbf{\tau }{)}_z$$

where $\beta =df/dy$ and $V$ is the meridional transport. Positive wind stress curl drives equatorward Sverdrup transport; negative curl drives poleward transport. This balance holds in the ocean interior where friction is negligible.

Western boundary layer theory

Stommel (1948) showed that adding linear bottom friction to the Sverdrup interior produces a western boundary layer of width:

$${\delta }_S=\frac{r}{\beta }$$

where $r$ is the friction coefficient. Munk (1950) replaced bottom friction with lateral viscosity, giving the Munk layer width:

$${\delta }_M={\left(\frac{A_H}{\beta }\right)}^{1/3}$$

With realistic $A_H$, this yields a boundary current width of about 100 km, consistent with the observed Gulf Stream.

Coastal and equatorial upwelling

When wind blows parallel to a coastline with the coast on its left (Northern Hemisphere), the offshore Ekman transport pushes surface water away from the coast. Mass continuity requires deep water to rise and replace it. The upwelling velocity $w_E$ is related to the wind stress component along the coast:

$$w_E=\frac{{\tau }_{\text{along}}}{{\rho }_{0}fL}$$

where $L$ is the relevant cross-shore length scale. Equatorial upwelling arises because trade winds drive divergent Ekman transport on either side of the equator (to the right north of the equator, to the left south of it), pulling water apart and drawing deep water upward.

The global thermohaline conveyor

Broecker (1991) popularized the "Great Ocean Conveyor" as a schematic of the global thermohaline circulation. NADW forms in the Nordic and Labrador Seas where warm, saline surface water cools and sinks. It flows southward at depth through the Atlantic, joins the Antarctic Circumpolar Current, and branches into the Indian and Pacific basins. After upwelling through mixing and wind-driven processes over centuries, it returns at the surface through the Indonesian Throughflow and around southern Africa. AABW, denser than NADW, forms in the Weddell and Ross Seas and fills the deepest layers of the major basins.

The residence time of deep water — the time a parcel spends at depth before returning to the surface — is approximately 1,000 years, estimated from radiocarbon dating of dissolved inorganic carbon.

Key result: Stommel's two-box model and multiple equilibria of the thermohaline circulation Intermediate+

Henry Stommel's 1961 two-box model demonstrated that the thermohaline circulation can possess multiple stable equilibria — a result with profound implications for abrupt climate change.

The model divides the ocean into a low-latitude box (warm, saline due to excess evaporation) and a high-latitude box (cold, fresher due to excess precipitation and runoff). A hydraulic flow connects the boxes, whose strength $q$ is proportional to the density difference:

$$q=k(\alpha \Delta T-\beta \Delta S)$$

where $\alpha$ and $\beta$ are the thermal and haline expansion coefficients, $\Delta T$ and $\Delta S$ are the temperature and salinity differences between the boxes, and $k$ is a hydraulic constant. The salinity difference is maintained by a surface freshwater flux $F$ and modified by the advective transport $q$:

$$\frac{d(\Delta S)}{dt}=|q|\Delta S+2F$$

The steady-state solutions reveal three possible equilibria: a thermally driven mode (large $q$, dominated by $\alpha \Delta T$), a haline-driven mode (reversed $q$, dominated by $\beta \Delta S$), and an unstable saddle point. For certain parameter ranges, both stable modes coexist — the system exhibits hysteresis.

The physical interpretation is significant. The modern Atlantic thermohaline circulation is in the thermally driven mode: warm surface water flows northward, cools, and sinks. If freshwater forcing (from melting ice sheets or increased precipitation) were to reduce $\Delta S$ sufficiently, the circulation could abruptly switch to the haline mode or shut down entirely. Paleoclimate evidence suggests such a shutdown occurred during the Younger Dryas cold period (approximately 12,900–11,700 years ago), when a massive freshwater pulse from melting ice sheets appears to have disrupted NADW formation, cooling the North Atlantic region by several degrees within decades.

This hysteresis behaviour means that the thermohaline circulation is not smoothly reversible. Once a threshold is crossed and the circulation collapses, restoring the original freshwater forcing may not restore the original circulation state — the system remains on the alternate branch of the hysteresis loop until the forcing is reversed well past the original transition point.

Exercises Intermediate+

Advanced results Master

Geostrophic currents and dynamic height

Away from boundaries and the Ekman layer, the large-scale ocean circulation is approximately in geostrophic balance:

$$f\mathbf{k}\times {\mathbf{v}}_g=-\frac{1}{{\rho }_0}{\nabla }_{h}p$$

The geostrophic flow is parallel to lines of constant pressure (isobars). In practice, oceanographers compute the geostrophic velocity from the density field using the dynamic height (or geopotential anomaly). The thermal wind relation gives the vertical shear of the geostrophic velocity:

$$\frac{\partial {\mathbf{v}}_g}{\partial z}=-\frac{g}{{\rho }_{0}f}\mathbf{k}\times {\nabla }_h\rho$$

This allows computation of the velocity field relative to a reference level (a "level of known motion"), a method called the dynamic method developed by Sandstrom and Helland-Hansen around 1900.

Sverdrup transport and the island rule

The Sverdrup balance can be integrated across a basin to yield the total meridional Sverdrup transport:

$${\Psi }_S(x)=\frac{1}{\beta {\rho }_0}{\int }_{x_e}^x(\nabla \times \mathbf{\tau }{)}_{z}d{x}^{\prime }$$

where $x_e$ is the eastern boundary. Godfrey's island rule extends this to compute the net transport around an island or through a passage, including the Indonesian Throughflow, from the wind stress curl integrated over the area bounded by the relevant coastline.

Stommel-Arons model of abyssal circulation

Stommel and Arons (1960) developed an analytical model of the deep ocean circulation driven by a uniform upwelling through the abyssal ocean and point sources of deep water formation. The model predicts that deep western boundary currents carry water equatorward from the formation regions, and the interior flow is poleward — the opposite of the surface Sverdrup circulation. This prediction was confirmed by the discovery of a deep western boundary current along the east coast of North America (the "deep western boundary current" observed by Swallow and Worthington in 1957).

Rossby and Kelvin waves in the ocean

Rossby waves (planetary waves) propagate westward in the ocean with phase speed:

$$c_{Rx}=-\frac{\beta g^{\prime }H}{f^2}$$

where $g^{\prime }$ is the reduced gravity and $H$ is the layer thickness. They are the primary mechanism by which the ocean adjusts to changes in wind forcing on timescales of months to years.

Kelvin waves propagate eastward along the equator (and poleward along boundaries) with speed $c=\sqrt{g^{\prime }H}$. Equatorial Kelvin waves play a central role in ENSO dynamics: a westerly wind burst in the western Pacific excites a downwelling Kelvin wave that propagates to the eastern Pacific, depressing the thermocline and initiating El Nino warming.

ENSO dynamics: Bjerknes feedback and the delayed oscillator

The Bjerknes feedback is the positive feedback between sea surface temperature, atmospheric convection, and trade wind strength in the equatorial Pacific. Weaker trade winds allow warm water to spread eastward, reducing the SST gradient. The weaker SST gradient further weakens the trade winds. This feedback can amplify a small initial perturbation into a full El Nino event.

The delayed oscillator model ( Suarez and Schopf 1988, Battisti and Hirst 1989) adds the essential negative feedback: the westward-propagating Rossby wave reflected as an equatorial Kelvin wave at the western boundary. The reflected wave arrives at the eastern Pacific with a delay determined by basin-crossing times, acting to terminate the El Nino and initiate La Nina. The interplay of the fast positive Bjerknes feedback and the delayed negative feedback produces self-sustained oscillations on interannual timescales.

Decadal oscillations: PDO, AMO, NAO

The Pacific Decadal Oscillation (PDO) is the leading mode of monthly SST variability in the North Pacific (poleward of 20 degrees N). Its positive phase features a warm eastern Pacific and cool central Pacific; the negative phase reverses this pattern. Regime shifts, such as the 1976-1977 transition, reorganize North Pacific ecosystems and affect salmon runs.

The Atlantic Multidecadal Oscillation (AMO) operates on 60-80 year timescales and affects Atlantic hurricane activity, Sahel rainfall, and European climate. The North Atlantic Oscillation (NAO) represents the fluctuation in the strength of the Icelandic Low and Azores High, governing the vigour of mid-latitude westerlies and the storm track position over the North Atlantic.

These oscillations arise from interactions between ocean heat content, atmospheric circulation, and — in some cases — the thermohaline circulation itself. Distinguishing natural oscillatory variability from anthropogenic trends is a central challenge in climate detection and attribution.

Turbulence and mixing: diapycnal diffusion and internal waves

The Munk (1966) abyssal recipe showed that maintaining the observed ocean stratification requires a globally averaged diapycnal (cross-density) diffusivity of approximately $10^{-4}$ m${}^2$/s, three orders of magnitude larger than the molecular diffusivity of heat. This discrepancy implicates turbulent mixing as the dominant mechanism for vertical exchange in the ocean.

The primary energy source for deep ocean mixing is the breaking of internal waves generated by tidal flow over rough topography. The internal tide, produced when the surface tide oscillates over mid-ocean ridges and seamounts, radiates energy into the ocean interior where it cascades to smaller scales and dissipates, mixing the surrounding water. Tracer release experiments have confirmed enhanced diapycnal diffusivity near rough topography, reaching $10^{-2}$ to $10^{-1}$ m${}^2$/s locally.

Ocean general circulation models and observational networks

Ocean general circulation models (OGCMs) solve the primitive equations (Boussinesq, hydrostatic Navier-Stokes on a rotating sphere) on global grids with horizontal resolution now reaching 1/12 degree or finer. Parameterizations of sub-grid-scale mixing (vertical diffusivity, mesoscale eddy fluxes via Gent-McWilliams or other schemes) remain a major source of uncertainty.

The Argo float network, operational since the early 2000s, provides global coverage of temperature and salinity profiles in the upper 2,000 m through a fleet of roughly 4,000 free-drifting profiling floats. Satellite altimetry (TOPEX/Poseidon, Jason series, Sentinel-6) measures sea surface height to centimetre accuracy, enabling global mapping of geostrophic currents, mesoscale eddies, and sea level rise. Together, Argo and altimetry have transformed ocean observing from sparse ship-based measurements to near-real-time global coverage.

AMOC weakening observations

The Atlantic Meridional Overturning Circulation (AMOC) has been directly monitored by the RAPID/MOCHA array at 26.5 degrees N since 2004. Observations show AMOC transport of roughly 17 Sv (1 Sv = $10^6$ m${}^3$/s) with significant variability. Proxy reconstructions (sea surface temperature patterns, sediment records) suggest the AMOC has weakened by approximately 15 percent since the mid-20th century, consistent with climate model predictions under increasing greenhouse gas concentrations. Whether this weakening represents a fluctuation or the beginning of a sustained decline is an active research question with major implications for European climate and global heat redistribution.

Connections Master

Connections to atmospheric circulation

Ocean surface currents and atmospheric circulation form a coupled system. Wind stress drives the upper-ocean Ekman transport and surface gyres, while sea surface temperature patterns feed back on the atmosphere by modulating convection, pressure gradients, and precipitation. The Gulf Stream's warm surface waters heat the overlying atmosphere, intensifying the mid-latitude westerlies and contributing to the storm track over the North Atlantic. The ENSO cycle exemplifies tight ocean-atmosphere coupling on interannual timescales (Unit 27.04.02).

Connections to marine ecosystems

Upwelling zones are the engines of marine biological productivity. The Peruvian upwelling system supports one of the world's largest fisheries (anchoveta). When El Nino suppresses upwelling, fish populations collapse, with cascading effects on seabirds, marine mammals, and the fishing industry. The thermohaline circulation also influences ecosystems by regulating nutrient supply to the surface ocean over long timescales.

Connections to climate change

The thermohaline circulation is a major component of Earth's climate system, transporting roughly 1.3 PW of heat poleward in the North Atlantic. Model projections under high-emission scenarios indicate AMOC weakening of 20-45 percent by 2100. A full collapse, though considered low probability, would cool the North Atlantic region by several degrees, shift tropical rain belts southward, and accelerate sea level rise along the northeast North American coast. The hysteresis identified by Stommel's model means that once a collapse occurs, it may be irreversible on human timescales.

Connections to plate tectonics and Earth history

The configuration of ocean basins — controlled by plate tectonics — determines the geometry of ocean circulation. The opening of the Drake Passage (30-40 Ma) enabled the Antarctic Circumpolar Current, thermally isolating Antarctica and contributing to glaciation. The closing of the Isthmus of Panama (3 Ma) redirected Atlantic circulation, intensifying the Gulf Stream and possibly triggering Northern Hemisphere glaciation. Past ocean circulation states are recorded in sediment proxies (carbon isotope ratios, Cd/Ca ratios, ${\delta }^{13}$C) that reveal changes in deep water formation and ventilation.

Connections to the carbon cycle

The solubility pump and the biological pump together determine the ocean's capacity to absorb atmospheric CO2. The thermohaline circulation transports carbon-rich deep water to the surface on millennial timescales, while downwelling carries surface carbon into the deep ocean. Changes in overturning rate therefore affect the ocean's carbon uptake. A weaker AMOC would reduce the ocean's ability to absorb CO2 from the atmosphere, amplifying warming through a positive feedback loop.

Connections to sea level and coastal hazards

Western boundary currents influence regional sea level. The Gulf Stream maintains a sea surface slope of roughly 1 m across its width through geostrophic balance. If the Gulf Stream weakens, the sea level on the US east coast could rise by 10-30 cm independently of global sea level rise. Changes in wind-driven gyre circulation also affect coastal sea level through Sverdrup dynamics and Ekman pumping.

Historical and philosophical context Master

Ekman and the wind-driven spiral (1905)

Vagn Walfrid Ekman, a Swedish oceanographer, developed his theory of wind-driven ocean currents in response to Fridtjof Nansen's observation during the Fram expedition (1893-1896) that Arctic ice drifted 20-40 degrees to the right of the wind. Ekman's 1905 paper showed that the Coriolis effect combined with vertical friction produces a spiral of current direction with depth and a net transport 90 degrees to the right of the wind. This result remains one of the cornerstones of physical oceanography.

Stommel and western intensification (1948)

Henry Stommel, working at the Woods Hole Oceanographic Institution, published his seminal 1948 paper demonstrating that the beta effect causes western boundary current intensification. His simple analytical model — a rectangular basin with wind forcing and linear friction — captured the essential physics of why the Gulf Stream is narrow and fast while the return flow on the eastern side is broad and slow. This work established the theoretical framework for understanding wind-driven gyres and inspired decades of research on western boundary layers.

Munk and the viscous boundary layer (1950)

Walter Munk extended Stommel's model by replacing bottom friction with lateral (eddy) viscosity, producing the Munk layer. His 1950 paper also provided observational confirmation by comparing predicted boundary current widths with measurements of the Gulf Stream and Kuroshio. Munk later made foundational contributions to understanding ocean mixing, proposing in 1966 that the observed abyssal stratification requires a diapycnal diffusivity of $\sim 10^{-4}$ m${}^2$/s — the "abyssal recipe."

Broecker and the Great Ocean Conveyor (1991)

Wallace Broecker, a geochemist at Columbia University's Lamont-Doherty Earth Observatory, synthesized decades of oceanographic research into the "Great Ocean Conveyor" concept in his 1991 Oceanography paper. While an oversimplification of the actual circulation, this schematic captured the public imagination and highlighted the potential for abrupt climate change through thermohaline circulation collapse. Broecker's earlier work on radiocarbon dating of ocean water masses provided the observational basis for estimating deep water residence times.

Stommel's two-box model (1961)

Stommel's 1961 paper in Tellus demonstrated that a simple two-box model of the thermohaline circulation possesses multiple stable equilibria. This was the first theoretical indication that the ocean circulation could undergo abrupt transitions between qualitatively different states. The model's implications were not fully appreciated until the 1980s, when paleoclimate data from ice cores revealed rapid climate shifts during the last glacial period, and numerical ocean models confirmed the possibility of thermohaline circulation collapse under freshwater forcing.

The observational revolution: Argo and satellite altimetry

The launch of the Argo float network in the early 2000s and the continuous operation of satellite altimeters since 1992 transformed physical oceanography from a data-sparse to a data-rich science. Before Argo, the deep ocean was sampled by infrequent research cruises separated by vast gaps in space and time. Argo's 4,000 profiling floats provide global coverage of temperature and salinity to 2,000 m depth on a continuous basis, enabling direct monitoring of ocean heat content, salinity changes, and — through the RAPID array — the strength of the AMOC itself.

Philosophical dimensions: multiple equilibria and irreversibility

The existence of multiple equilibria in the thermohaline circulation raises a fundamental point about the Earth system. Climate need not respond smoothly to gradual forcing. Small changes in freshwater input can trigger large, abrupt, and potentially irreversible shifts in the ocean circulation and hence in regional climate. This nonlinearity challenges the intuition that small causes produce small effects, and it underscores the difficulty of predicting climate thresholds — the boundaries between stable states — before they are crossed.

Bibliography Master

1. Tarbuck, F. K. & Lutgens, E. J. (2018). Earth Science (15th ed.). Pearson. Ch. 10: Ocean circulation.

2. Talley, L. D., Pickard, G. L., Emery, W. J. & Swift, J. H. (2011). Descriptive Physical Oceanography: An Introduction (6th ed.). Academic Press. Ch. 1-5: Thermohaline and wind-driven circulation.

3. Stommel, H. (1948). "The westward intensification of wind-driven ocean currents." Transactions of the American Geophysical Union, 29, 202-206.

4. Stommel, H. (1961). "Thermohaline convection with two stable regimes of flow." Tellus, 13, 224-230.

5. Munk, W. H. (1950). "On the wind-driven ocean circulation." Journal of Meteorology, 7, 79-93.

6. Ekman, V. W. (1905). "On the influence of the Earth's rotation on ocean currents." Arkiv for Matematik, Astronomi och Fysik, 2, 1-52.

7. Broecker, W. S. (1991). "The Great Ocean Conveyor." Oceanography, 4, 79-89.

8. Sverdrup, H. U. (1947). "Wind-driven currents in a baroclinic ocean; with application to the equatorial currents of the eastern Pacific." Proceedings of the National Academy of Sciences, 33, 318-326.

9. Stommel, H. & Arons, A. B. (1960). "On the abyssal circulation of the world ocean — I. Stationary planetary flow patterns on a sphere." Deep Sea Research, 6, 140-154.

10. Munk, W. H. (1966). "Abyssal recipes." Deep Sea Research and Oceanographic Abstracts, 13, 707-730.

11. Wunsch, C. (1996). The Ocean Circulation Inverse Problem. Cambridge University Press.