Ocean circulation and climate

The ocean is Earth's largest heat reservoir, absorbing and redistributing solar energy on a planetary scale. Ocean currents transport roughly one petawatt (10¹⁵ watts) of heat from the tropics toward the poles, moderating temperature extremes and shaping the climatic conditions under which terrestrial and marine ecosystems have evolved.² This vast circulatory system operates on two broadly complementary levels: a wind-driven surface circulation that dominates the upper few hundred metres of the water column, and a density-driven deep circulation — often called the thermohaline circulation — that turns over the entire ocean volume on a timescale of roughly one to two thousand years.^{1, 3} Changes in the strength, geometry, or mode of ocean circulation have triggered some of the most dramatic climate shifts in Earth's history, from the glaciation of Antarctica following the opening of the Drake Passage to the abrupt cooling episodes of the last ice age driven by collapses of the Atlantic Meridional Overturning Circulation.

The geological record preserves remarkably detailed evidence of past ocean circulation states through chemical signals locked in the shells of microscopic marine organisms, through the isotopic fingerprints of dissolved elements in seawater, and through the sedimentary architecture of the deep seafloor itself. Decoding these archives has allowed scientists to reconstruct how the ocean has circulated through radically different climate states over the past hundred million years and to assess the vulnerability of the modern circulation system to ongoing anthropogenic warming.^{1, 6}

Surface circulation and wind-driven currents

The surface circulation of the ocean is driven primarily by the frictional stress of prevailing winds acting on the sea surface. The global pattern of surface winds — the trade winds in the tropics, the westerlies in the mid-latitudes, and the polar easterlies at high latitudes — sets up large-scale circular flow patterns known as gyres. In each major ocean basin, a subtropical gyre circulates clockwise in the Northern Hemisphere and counterclockwise in the Southern Hemisphere, a consequence of the Coriolis effect produced by Earth's rotation.² These gyres concentrate warm equatorial water along the western margins of ocean basins, producing intensified boundary currents such as the Gulf Stream in the North Atlantic, the Kuroshio Current in the North Pacific, and the Agulhas Current off southeastern Africa. Western boundary currents are narrow, fast, and warm, transporting enormous quantities of heat poleward, while their eastern counterparts (such as the Canary Current and the California Current) are broad, slow, and cool.²

The Antarctic Circumpolar Current (ACC), the largest and strongest current in the world ocean, is a unique feature of Southern Hemisphere circulation. The ACC flows eastward around Antarctica, driven by the powerful westerly winds that blow unimpeded across the Southern Ocean — the only latitude band on Earth where no continental landmass interrupts the zonal flow. The ACC transports approximately 130 to 185 sverdrups (one sverdrup equals one million cubic metres per second) through the Drake Passage between South America and Antarctica, connecting the Atlantic, Pacific, and Indian ocean basins and thermally isolating the Antarctic continent from warmer waters to the north.^{9, 13} The establishment of the ACC in the Cenozoic was one of the most consequential events in the tectonic history of ocean circulation, as discussed below.

Wind-driven surface currents also influence deeper layers of the ocean through a process called Ekman pumping. In subtropical gyres, the convergence of wind-driven surface flow pushes water downward into the ocean interior, depressing the thermocline and creating a dome of warm water in the gyre centre. In subpolar regions and along certain coastlines, the divergence of surface water draws cold, nutrient-rich water upward from depth, fuelling biological productivity and exchanging gases between the deep ocean and the atmosphere.² This interplay between wind-driven and density-driven circulation makes the ocean a single, interconnected system rather than two independent layers.

Thermohaline circulation and the global conveyor belt

Beneath the wind-driven surface layer, the deep ocean is filled by cold, dense water masses formed at high latitudes where surface waters become sufficiently dense to sink to great depth. This density is governed by two properties: temperature and salinity — hence the term thermohaline circulation. Cold water is denser than warm water, and saline water is denser than fresh water; where both conditions are met simultaneously, surface water can become dense enough to descend thousands of metres and spread laterally along the ocean floor, driving a slow but volumetrically enormous overturning circulation.^{1, 2}

In a simplified but influential conceptualization introduced by Wallace Broecker in 1991, the thermohaline circulation can be visualised as a global conveyor belt: warm, salty surface water flows northward through the Atlantic, loses heat to the atmosphere at high latitudes, becomes dense, and sinks to form deep water. This deep water flows southward through the Atlantic, joins the circumpolar flow around Antarctica, and is distributed into the deep Indian and Pacific oceans, where it gradually warms and upwells, completing the circuit.³ Although this metaphor oversimplifies the true three-dimensional complexity of ocean circulation, it captures the essential feature that the Atlantic is a net exporter of deep water and a net importer of warm surface water — a configuration that gives the North Atlantic an anomalously mild climate for its latitude.

The two principal deep water masses that drive the lower limb of this overturning are North Atlantic Deep Water (NADW) and Antarctic Bottom Water (AABW). NADW forms in two main regions: the Nordic Seas (the Greenland, Iceland, and Norwegian seas), where cold, dense overflow water spills southward across the Greenland-Scotland Ridge through the Denmark Strait and the Iceland-Scotland gap; and the Labrador Sea between Canada and Greenland, where intense winter cooling and wind forcing drive open-ocean convection to depths exceeding 2,000 metres.² AABW forms around the margins of Antarctica, principally in the Weddell Sea and the Ross Sea, where brine rejection during sea-ice formation in coastal polynyas creates extremely cold and dense water that sinks along the continental shelf and fills the abyssal basins of the global ocean below approximately 4,000 metres.²⁰ Together, NADW and AABW ventilate the deep ocean, supplying it with dissolved oxygen and drawing down carbon dioxide from the atmosphere on timescales of centuries to millennia.

The Atlantic Meridional Overturning Circulation

The Atlantic Meridional Overturning Circulation (AMOC) is the organised system of northward surface flow and southward deep return flow in the Atlantic basin that constitutes the single most important component of the global thermohaline circulation for Earth's climate. The AMOC transports approximately 17 sverdrups of warm water northward in its upper limb and returns a comparable volume as NADW in its deep limb, delivering roughly 1.3 petawatts of heat to the North Atlantic and its bordering continents.^{2, 5} This northward heat transport is responsible for making western Europe several degrees warmer than equivalent latitudes in North America and is a primary driver of the interhemispheric temperature asymmetry that characterises Earth's modern climate.

The driving forces of the AMOC have been debated for decades. Two principal mechanisms have been identified: deep-water formation driven by surface buoyancy loss at high northern latitudes, and wind-driven upwelling in the Southern Ocean. In the traditional view, the AMOC is driven from the north — surface cooling and evaporation in the Nordic and Labrador seas create the dense water whose sinking pulls warm surface water northward. An alternative view, supported by theoretical arguments and modelling studies, emphasises the role of Southern Ocean winds in drawing deep water upward along steeply tilted density surfaces, effectively pulling the deep limb of the AMOC from below.² In reality, both mechanisms contribute, and the AMOC is best understood as a system maintained by the interplay of northern sinking, southern upwelling, and internal mixing processes that convert deep water back into lighter water within the ocean interior.^{2, 5}

A crucial property of the AMOC is its potential for bistability — the existence of two stable states under the same external forcing conditions. In a landmark 1961 paper, Henry Stommel demonstrated with a simple two-box model that thermohaline circulation driven by opposing temperature and salinity gradients can exhibit two equilibrium states: a vigorous "on" mode resembling the modern AMOC, and a weak or collapsed "off" mode in which deep-water formation shuts down.⁴ The physical mechanism underlying this bistability is the salt-advection feedback: a vigorous AMOC imports salt into the North Atlantic (because evaporation exceeds precipitation in the subtropics), which reinforces the density of surface waters and sustains deep-water formation. If the circulation weakens, the salt supply diminishes, surface waters freshen, and the weakening is amplified in a positive feedback loop that can drive the system into a collapsed state.^{4, 5} This theoretical prediction has been confirmed in numerous climate models and is central to understanding the abrupt climate changes recorded in the geological record.²⁴

Proxies for past ocean circulation

Reconstructing the circulation of ancient oceans requires indirect methods, because direct instrumental measurements of ocean currents extend back only a few decades. Paleoceanographers rely on several classes of geochemical and sedimentological proxies, each of which records a different aspect of past ocean conditions.

Benthic foraminiferal oxygen isotopes (δ¹⁸O) are the workhorse proxy of paleoceanography. The oxygen isotopic composition of calcium carbonate shells precipitated by bottom-dwelling foraminifera reflects both the temperature of the ambient bottom water and the δ¹⁸O of the seawater itself, which varies with the global volume of ice stored on land. By stacking δ¹⁸O records from dozens of globally distributed deep-sea sediment cores, Lisiecki and Raymo (2005) constructed the LR04 reference curve, a continuous record of deep-ocean temperature and ice volume changes extending back 5.3 million years through the Pliocene and Pleistocene.⁶ This stack reveals the rhythmic glacial-interglacial oscillations driven by orbital forcing and provides a chronological framework for correlating events across ocean basins.

Benthic foraminiferal carbon isotopes (δ¹³C) serve as a tracer of deep-water mass geometry and ventilation. The δ¹³C of dissolved inorganic carbon in seawater decreases as organic matter is remineralised at depth, so "young" deep water recently formed at the surface carries a high δ¹³C signature, while "old" deep water that has accumulated remineralised carbon is isotopically depleted. In the modern ocean, NADW has a characteristically high δ¹³C (approximately +1.0 per mille) and AABW has a lower value (approximately +0.3 to +0.5 per mille), enabling paleoceanographers to map the extent of these water masses through time by analysing the δ¹³C of fossil benthic foraminifera from different depths and latitudes.⁷ During glacial periods, maps of benthic δ¹³C reveal a dramatically reorganised Atlantic, with a shallower and weaker NADW cell overlying an expanded tongue of southern-sourced deep water — evidence that the AMOC was substantially reduced during ice ages.^{1, 7}

Neodymium isotopes (ε_Nd) provide an independent tracer of water mass provenance. The neodymium isotopic composition of seawater reflects the age and composition of the continental rocks from which dissolved neodymium was derived through weathering and river input. Because different ocean basins are bordered by rocks of different ages, their deep waters carry distinct ε_Nd signatures: North Atlantic deep water has a characteristically unradiogenic value (approximately −13), reflecting the ancient cratonic rocks of Canada and Greenland, while Pacific deep water is more radiogenic (approximately −4), reflecting younger volcanic terranes. These contrasts allow ε_Nd to be used as a quasi-conservative water mass tracer that is largely independent of biological processes, complementing the carbon isotope proxy.⁸

Sediment drift deposits (contourites) provide physical evidence of past bottom-current activity. Contourites are sedimentary bodies deposited or reworked by persistent bottom currents flowing along bathymetric contours, and they form large elongated mounds and channel systems on the continental slopes and abyssal plains of all ocean basins. Their grain size, accumulation rate, and internal stratigraphy record changes in the vigour and direction of deep-water flow over millions of years. Major contourite systems such as the Eirik Drift south of Greenland, the Gardar Drift in the Iceland Basin, and the Feni Drift west of the British Isles have been drilled by the International Ocean Discovery Program and have yielded high-resolution records of NADW variability through the Neogene and Quaternary.²¹

Principal proxies for reconstructing past ocean circulation^{6, 7, 8, 21}

Gateway events and circulation reorganisation

The geometry of ocean basins is set by plate tectonics, and the opening or closing of oceanic gateways between continents has repeatedly reorganised global ocean circulation, with profound consequences for climate. Three gateway events of the Cenozoic era have received particular attention from paleoceanographers.

The opening of the Tasman Gateway between Australia and Antarctica, which deepened to allow significant throughflow between approximately 35.5 and 30 million years ago during the late Eocene and early Oligocene, was one of two tectonic prerequisites for the establishment of a circum-Antarctic current. As Australia drifted northward from Antarctica, the progressively widening and deepening seaway allowed first shallow and then deep-water exchange between the South Indian and South Pacific oceans.¹⁰ Neodymium isotope records from fossil fish teeth recovered during Ocean Drilling Program Leg 189 indicate that the deep Tasman Gateway opened at approximately 33 million years ago, aligning with the dramatic shift toward cooler global temperatures at the Eocene-Oligocene transition.^{10, 13}

The opening of the Drake Passage between South America and Antarctica completed the circumpolar pathway and permitted the full development of the Antarctic Circumpolar Current. The timing of this event remains debated, with estimates ranging from the late Eocene (approximately 41 million years ago) to the early Oligocene (approximately 30 million years ago), depending on whether shallow or deep throughflow is considered.¹¹ Kennett (1977) proposed that the thermal isolation of Antarctica by the ACC was the primary cause of Cenozoic Antarctic glaciation, arguing that the circumpolar current blocked the poleward transport of warm subtropical waters and allowed the Antarctic continent to cool to the point where a permanent ice sheet could form.⁹ More recent work has suggested that declining atmospheric CO₂ concentrations played a co-equal or even dominant role in triggering Antarctic glaciation, but the tectonic opening of the Southern Ocean gateways remains an essential part of the narrative.^{11, 13} Scher et al. (2015) used neodymium isotope records to demonstrate that a vigorous ACC developed approximately 30 million years ago, when the Tasman Gateway had drifted into the latitude band of the strong westerly winds, providing the dynamical forcing necessary to drive a powerful circumpolar flow.¹³

The closure of the Central American Seaway and the rise of the Isthmus of Panama in the late Pliocene, approximately 4.6 to 2.6 million years ago, represents the most recent major gateway event. Before the isthmus formed, a broad seaway connected the tropical Atlantic and Pacific oceans, allowing the exchange of surface water between the two basins. As the isthmus gradually emerged through volcanic and tectonic activity, the seaway constricted and the Atlantic became increasingly isolated from the Pacific. Haug and Tiedemann (1998) demonstrated, using carbon isotope and carbonate records from Ocean Drilling Program cores in the Caribbean, that the progressive closure of the seaway intensified the Gulf Stream, increased the salinity of Atlantic surface water by blocking the inflow of fresher Pacific water, and strengthened NADW formation.¹² Paradoxically, this intensification of the AMOC may have contributed to the onset of Northern Hemisphere glaciation by delivering more moisture to high northern latitudes, where increased snowfall could accumulate into continental ice sheets.^{12, 22}

Ocean anoxic events

During the warm Cretaceous period (approximately 145 to 66 million years ago), when global temperatures were far higher than today and no permanent polar ice sheets existed, the ocean experienced episodes of widespread oxygen depletion known as oceanic anoxic events (OAEs). These events are recorded in the sedimentary record as distinctive layers of organic-rich black shale deposited on the seafloor under conditions where dissolved oxygen had been so severely depleted that aerobic organisms could not survive.¹⁴

Two OAEs stand out in the Cretaceous record. OAE 1a (the Selli Event, approximately 120 million years ago, in the early Aptian) and OAE 2 (the Bonarelli Event, approximately 94 million years ago, at the Cenomanian-Turonian boundary) each lasted on the order of several hundred thousand to one million years and left globally recognisable organic carbon enrichments in marine sediments. Both events are associated with massive volcanism from large igneous provinces, which injected CO₂ into the atmosphere, warmed the climate, and accelerated the hydrological cycle. Enhanced continental weathering delivered excess nutrients (particularly phosphorus) to the oceans, stimulating primary productivity and increasing the biological oxygen demand at depth. When oxygen consumption by the decomposition of sinking organic matter exceeded the supply of dissolved oxygen from ocean circulation, large volumes of the deep ocean became anoxic.^{14, 19}

The geochemistry of OAEs provides evidence that ocean circulation in the Cretaceous differed fundamentally from the modern pattern. In the absence of polar ice caps and with reduced equator-to-pole temperature gradients, the thermohaline circulation was likely weaker and driven by warm, saline subtropical surface waters sinking at mid-latitudes rather than by cold polar waters. This so-called halothermal circulation may have been less efficient at ventilating the deep ocean, predisposing it to oxygen depletion during episodes of enhanced nutrient input.¹⁴ Positive feedback loops amplified and prolonged the anoxic conditions: under anoxic bottom waters, phosphorus that would normally be bound in sediments was released back into the water column, further stimulating productivity and carbon burial in a self-reinforcing cycle that could sustain OAE conditions for hundreds of thousands of years.¹⁹

Abrupt circulation changes in the ice ages

The Pleistocene glacial-interglacial cycles of the past 2.6 million years provide the most detailed record of abrupt ocean circulation changes and their climatic consequences. During each glacial period, the AMOC weakened substantially from its modern strength, reducing the northward heat transport that warms the North Atlantic region. Benthic δ¹³C records show that glacial NADW was shallower and less extensive than its modern counterpart, with nutrient-depleted southern-sourced water filling much of the deep Atlantic below approximately 2,500 metres.¹ Superimposed on this baseline weakening were two classes of even more dramatic, short-lived disruption: Heinrich events and Dansgaard-Oeschger oscillations.

Heinrich events are episodes during which massive armadas of icebergs were discharged from the Laurentide ice sheet (and to a lesser extent from European ice sheets) into the North Atlantic, depositing distinctive layers of ice-rafted debris (IRD) across the ocean floor. Six major Heinrich events have been identified in the last glacial period, occurring at roughly 7,000 to 10,000 year intervals between approximately 60,000 and 16,000 years ago.¹⁵ The freshwater released by the melting of these vast iceberg fleets reduced the density of North Atlantic surface waters, suppressing or shutting down deep-water formation and severely weakening or collapsing the AMOC. Proxy evidence indicates that Heinrich events were associated with cooling of 5 to 10 degrees Celsius in the North Atlantic region, widespread aridity in the Northern Hemisphere tropics, and a compensating warming in the Southern Hemisphere — a pattern consistent with a bipolar seesaw response to a reduction in northward oceanic heat transport.^{15, 23}

The Younger Dryas (approximately 12,900 to 11,700 years ago) is the best-studied example of an abrupt cooling event linked to AMOC disruption. As the great Northern Hemisphere ice sheets were melting at the end of the last glacial period, a sudden return to near-glacial conditions interrupted the warming trend across much of the North Atlantic region and persisted for approximately 1,200 years. The prevailing explanation invokes a massive release of glacial meltwater into the North Atlantic, either through the St. Lawrence River system or via a northward re-routing of proglacial drainage, which freshened the surface ocean and shut down NADW formation.²³ McManus et al. (2004) used the sedimentary ²³¹Pa/²³⁰Th ratio — a kinematic proxy for the rate of overturning circulation — to demonstrate that the AMOC was sharply reduced during the Younger Dryas and recovered rapidly at its termination, coinciding with the resumption of warming.¹⁶ The speed and severity of the Younger Dryas event, which cooled Greenland by approximately 10 degrees Celsius within decades, illustrate the capacity of the AMOC to act as a switch in the climate system, amplifying and transmitting regional forcing into hemispheric and global climate responses.^{1, 16}

Estimated AMOC strength during key climate events of the last 20,000 years^{1, 16}

Ocean heat transport and climate regulation

The ocean's role in climate extends far beyond the redistribution of heat.

By absorbing approximately 90 percent of the excess heat trapped by anthropogenic greenhouse gases and roughly 30 percent of the CO₂ emitted by human activities, the ocean acts as an enormous buffer that has moderated the pace and magnitude of surface warming over the industrial era.⁵ This buffering capacity is intimately linked to the overturning circulation: the sinking of dense water at high latitudes carries heat and dissolved carbon into the deep ocean interior, sequestering them from the atmosphere for centuries to millennia. The efficiency of this sequestration depends on the rate of overturning — a faster AMOC draws more heat and carbon into the deep ocean, while a weaker AMOC reduces the ocean's capacity to absorb and store these quantities.^{2, 5}

The geological record demonstrates that changes in ocean heat transport have driven some of the most dramatic climate transitions in Earth's history. The thermal isolation of Antarctica by the ACC following the opening of the Drake Passage and the Tasman Gateway allowed the Antarctic ice sheet to grow, fundamentally altering Earth's albedo and energy balance.⁹ The intensification of the Gulf Stream following the closure of the Isthmus of Panama delivered warm, moist air to northern high latitudes, providing the moisture source for the continental ice sheets of the Pleistocene.¹² The abrupt cooling events of the last glacial period were driven not by changes in the total energy budget of the planet but by the redistribution of existing oceanic heat — when the AMOC weakened, heat that would normally have been transported to the North Atlantic remained in the Southern Hemisphere, producing the characteristic bipolar seesaw pattern observed in ice-core and sediment records.^{1, 23}

The sensitivity of global climate to the configuration and strength of ocean circulation underscores a fundamental principle of Earth system science: the climate is not governed by radiative forcing alone but by the dynamic coupling of the atmosphere, ocean, cryosphere, and solid Earth. Tectonic changes operate on timescales of millions of years and set the boundary conditions — the geometry of ocean basins, the positions of continents, and the existence of oceanic gateways — within which faster processes such as orbital forcing, ice-sheet dynamics, and greenhouse gas feedbacks drive the climate variability recorded in the geological record.^{1, 9}

Modern AMOC stability and future outlook

The geological evidence that the AMOC has weakened or collapsed repeatedly in the past, each time producing abrupt and severe climate impacts, has focused intense scientific attention on the vulnerability of the modern AMOC to anthropogenic climate change. Under continued greenhouse gas emissions, climate models project that the North Atlantic will warm and freshen as a result of increased precipitation, enhanced river runoff, and accelerated melting of the Greenland ice sheet. Both effects reduce the density of surface waters in the deep-water formation regions, weakening the AMOC through the same salt-advection feedback mechanism identified by Stommel's box model.^{4, 5}

The Coupled Model Intercomparison Project (CMIP) has consistently shown that the AMOC weakens under warming scenarios, with projections ranging from a 10 to 50 percent reduction in AMOC strength by 2100 under high-emission pathways. However, most coupled climate models do not simulate a complete AMOC collapse within the twenty-first century, in part because many models may underestimate key feedbacks such as Greenland ice-sheet melt.⁵ Hofmann and Rahmstorf (2009) demonstrated with intermediate-complexity models that the AMOC exhibits a hysteresis response to freshwater perturbations: once a critical threshold of freshwater input is crossed, the circulation collapses abruptly and does not recover even if the perturbation is removed, because the system has entered the alternative stable state predicted by theory.²⁴

Observational evidence has added urgency to these concerns. Ditlevsen and Ditlevsen (2023) applied statistical early-warning indicators to AMOC proxy time series and estimated that a tipping-point transition could occur as early as the mid-twenty-first century under current emission trajectories, though the confidence intervals on such estimates remain wide.¹⁷ Van Westen et al. (2024) used a state-of-the-art global climate model to identify a physics-based early warning signal — the freshwater transport by the AMOC at the southern boundary of the Atlantic — and found that this indicator suggests the present-day AMOC is on a trajectory toward tipping, although the timing of any potential collapse remains deeply uncertain.¹⁸ The consequences of a full AMOC collapse would be severe and far-reaching: rapid cooling of the North Atlantic and western Europe by several degrees, disruption of tropical monsoon systems, sea-level rise of up to one metre along the eastern seaboard of the Americas, and reduced ocean carbon uptake that would accelerate atmospheric CO₂ accumulation.^{5, 17}

The paleoceanographic record is unequivocal that the AMOC is capable of rapid and dramatic state changes. The challenge for modern climate science is to determine how close the present system is to a critical threshold, and whether ongoing changes in North Atlantic freshwater balance are approaching the conditions that triggered past collapses. While substantial uncertainty remains in both the timing and the probability of an AMOC tipping event, the geological precedent for abrupt circulation change — and the catastrophic climate impacts that accompanied it — provides a compelling argument for treating AMOC stability as one of the most consequential risks in the Earth system under continued anthropogenic warming.^{1, 5, 18}