The South Atlantic

The central problem of South Atlantic oceanography and paleoceanography is the exchange of heat between South and North Atlantic. More specifically, it is the nature of the North Atlantic heat piracy. Without this heat transfer, northern Europe would have an entirely less clement climate. The necessity to understand the operation of the Atlantic Heat Conveyor provides the rationale for studying both physical and historical oceanography of the South Atlantic. The study of present conditions allows a synoptic view of dynamic interaction of surface- and deep circulation, and associated productivity patterns. The study of past patterns provides clues to the range of possible states and rates of change, and to the long-term stability of the system.

In a recent study (Reid, 1994) maps of the general circulation of the Atlantic Ocean were presented and discussed. The emphasis was placed upon the exchange of waters between the North and South Atlantic as indicated by the patterns of characteristics.The exchange with the waters entering through the Drake Passage involves further discussion of the effect of the Atlantic waters in providing a major defining characteristic of the Circumpolar Water — the layer of warm and saline water. The associated features — the high oxygen and low nutrient concentrations - are also defined well for a long distance down stream in the Antarctic Circumpolar Current. However it is only the vertical maximum in salinity that persists all along the flow around Antarctica and through the Drake Passage into the South Atlantic.This returning salinity maximum is much lower than the salinity maximum from the north that it meets and mixes with in the South Atlantic. Part of their mixture turns back northward into the South Atlantic where it appears as the vertical maximum in salinity seen in the east, but lower in salinity than the maximum seen along the western boundary. The rest of the mixture continues eastward, with its salinity increased well above that of the Drake Passage water and begins another circuit of Antarctica.

We report distributions of the chlorofluorocarbon CFC 11, of CC1₄, and of terrigenic 3He, along three zonal sections across the South Atlantic (WOCE WHP sections A8 – A10, 11.7°S, 19°S, 30°S). The distributions fully reflect the water mass structure of the sections. They reveal a region of comparably much slower water renewal in the range of the Central and Antarctic Intermediate Waters, northeast of the Angola-Benguela Front. For all water masses further down, the distributions demonstrate that renewal is very much slower still, and that it occurs via advective cores adjacent to the respective western boundaries of the basins. The oldest waters are found to be present in a wedge centered in about 3000 m depth and extending from the African slope westward across the Midatlantic Ridge. The tracers are characterized by different input time scales, and these are well apparent in their distributions. The observed correlation between tracers indicates rather steady formation of Upper North Atlantic Deep Water over the past several decades. CC1₄ is powerful in tracing relatively older waters, but in the upper South Atlantic waters our data confirm the decomposition of CC1₄ reported previously.

The sea surface height observations made by the radar altimetry system aboard the TOPEX/POSEIDON satellite, the first satellite dedicated to the study of the global ocean circulation, were used to study the circulation and its variability of the South Atlantic Ocean. Preliminary results from the first 18 month’s worth of data are presented on the mean ocean dynamic topography (1 year average) and its seasonal and mesoscale variabilities. Due to the large uncertainty in the geoid model at small scales, the utility of the dynamic topography is limited to scales larger than 2000 km. Comparisons were made with historical hydrographic observation and contemporaneous simulation by a computer model. Both annual and semiannual harmonics (amplitude and phase) were estimated for the sea level variations at scales larger than 500 km. The strongest seasonal variabilities (both annual and semiannual) were found in the Brazil/Malvinas Confluence and the Agulhas Retroflection, with secondary maxima in the Gulf of Guinea (semiannual) and the western tropical Atlantic (annual). The mesoscale variability revealed by the standard deviation of sea level is similar to previous observations. The effects of eddies on the mean flow, estimated from the eddy Reynolds stress, are important in the Brazil/ Malvinas Confluence and the Agulhas Retroflection.

The data from six zonal sections in the World Ocean Circulation Experiment (WOCE) in the tropical and southern Atlantic are used to describe the distribution of water masses. Due to the high spatial resolution, the structure of temperature, salinity, oxygen, silicate and nitrate displays details related to transport and mixing in this region. Temperature-salinity diagrams are also presented which indicate the effects of branching and recirculation loops in the water mass flow.

Hydrographic data along 11°S occupied in 1983 by the R.V. OCEANUS are used together with various wind climatologies to estimate the annual average transport of heat at this latitude. Some motivation for expecting fairly well-defined estimates at this latitude compared to others comes from the absence of a strong western boundary current. Results include flow in four layers representing the thermocline, Antarctic Intermediate Water, North Atlantic Deep Water, and Antarctic Bottom Water, using zero velocity reference level choices based on property distributions. The annual average heat transport is estimated to be 0.6 ± 0.17 x 1015 W. Previous estimates of the transport at 8–16°S range from 0.2 PW to greater than 1 PW. Interannual variability from the wind field alone leads to interannual heat transport variability of about 0.05 PW. Comparisons with other recent studies at 45–30°S and 11°N are made.

The role of the South Atlantic in the global climate system, specifically as it concerns the North Atlantic Deep Water driven thermohaline meridional overturning cell, has drawn much attention in recent years. I propose that the configuration of the continents bounding the South Atlantic, specifically the contrast in the southern extreme of South America and Africa relative to the maximum westerlies wind, allow the development of a salty Atlantic Ocean, preconditioning the region for production of North Atlantic Deep Water, whose positive feedbacks further energizes the process. Much of the low salinity upper layer water carried by the South Pacific Current towards the coast of Chile, turns northward into the subtropical South Pacific as the Peru Current. Pacific surface water that does flow through the Drake Passage tends to transverse the South Atlantic, picking up more freshwater enroute drawn from the evaporative regime of the South Atlantic subtropics, into the Indian Ocean. Only a small component of the South Atlantic Current turns northward into the Benguela Current. Warm, saline Indian Ocean thermocline water can slip along the southern rim of Africa into the South Atlantic to further increase Atlantic salinity.

The circulation at mid-latitudes of the South Atlantic Ocean accommodates a substantial, though intermittent, leakage of tropical and subtropical water from the South Indian Ocean past the southern tip of Africa. This transferral of water masses is for the most part, but not exclusively, due to large Agulhas rings that are shed at the Agulhas retroflection. Direct transfer in the form of filaments also occurs. These inter-basin exchanges of waters influence global climate, local ocean dynamics, biogeographic patterns and possibly even local fish recruitment.Agulhas rings are major mesoscale features that average 250 km in diameter and extend to more than 1000 m depth. They drift into the South Atlantic at rates of about 7 km/day (8 cm/s). An estimated 6 to 9 are formed each year with a range of dimensions. Recent studies have estimated the total inter-basin heat flux due to Agulhas rings as being 0.05 PW per annum, and the salt flux as 78 x 1012 kg per annum.A secondary, and minor, exchange between the ocean basins is due to Agulhas filaments. These are detached from the landward edge of the southern Agulhas Current and are then advected into the South Atlantic by the Benguela drift. These features are only about 50 m deep on average, lose most of their heat to the atmosphere, but contribute a salt flux that is estimated to be 29 x 1012 kg per annum, or about 12% of that due to Agulhas rings.Agulhas rings are formed through complex processes. Perturbations in the flow paths of the Agulhas Return Current, and that of the Agulhas Current proper, cause the opposing flows to coalesce and thus to pinch off a ring. In this process quantities of cold, Subantarctic Surface Water are forced into the South Atlantic, further complicating the array of water types already present. The driving forces that bring about perturbations in the Agulhas system are poorly understood. Meanders in the southern Agulhas Current are similar to those observed in other western boundary currents, but are, by comparison, small. A large soliton meander on the current, the Natal Pulse, occurs at irregular intervals, about 8 times per year. Recent analyses suggest that the passage of each Natal Pulse is followed by the spawning of an Agulhas ring. The triggering of a Natal Pulse occurs far upstream in the Natal Bight and may be caused by the impingement of deep-sea eddies on the Agulhas Current, or by fluctuations in the velocity of the current leading to barotropic instabilities.

The Benguela is one of four major eastern boundary current regions of the World Ocean. The oceanography off the west coast of southern Africa is dominated, like the regions off California, Peru and North West Africa, by coastal upwelling; but the Benguela is unique in that it is bounded on both the equatorward and poleward ends by warm water regimes. In this paper we build on review articles which were published during the 1980s by highlighting the main advances in the conceptual understanding of the system since 1985. A large amount of research on this eastern boundary domain has been conducted in recent years by the institutes to which the authors are affiliated. This has given clear definition to four aspects of shelf dynamics, these being poleward flow at depth across the shelf and out into the slope region, the existence of baroclinic shelf-edge jets in the vicinity of the shelf break, barotropic shelf waves and the importance of variation in wind rather than constant strength of wind as a factor controlling upwelling. Short discussions of some, as yet, unpublished findings supplement published works by ourselves and other authors. Of particular importance are the events such as Benguela Ninos in the north and intrusions of Agulhas water in the south. The latter, which can take the form of filaments or rings, influences the oceanography of the region of the “Greater Agulhas Current”, with interactions between Agulhas rings and Benguela shelf waters.

The boundary region between the South Atlantic and Southern Ocean is discussed making use of results from the FRAM model and hydrographic sections. Although, at first, the system appears to be a simple one, with a sub-tropical gyre to the north abutting a zonal jet to the south, the detailed structure is more complex. Part of the complexity is well known and arises from the large scale thermohaline circulation, water being transported north in the surface Ekman layer and at depth, and returning south at intermediate levels. More recent work has shown that, in addition, the sub-tropical gyre and the Antarctic Circumpolar Current combine to form the Deacon Cell which transforms surface stress due to the wind down to mid-depths. Further complications arise from the Agulhas eddies which pass through the regions and the constraints on the circulation due to the bottom topography.

Maps of the Antarctic Intermediate Water (AAIW) in the Atlantic, and on a global isopycnal which intersects the AAIW in the south, show the location and properties of the salinity and oxygen extrema associated with the AAIW, and the likely sources of AAIW. These are primarily the surface waters in the southeastern Pacific, which produce the South Pacific AAIW, and surface waters in northern Drake Passage and the Falkland Current loop, which produce the South Atlantic AAIW. This latter source is the primary one for AAIW of the Indian Ocean as well. Winter surface properties and annual-averaged Ekman pumping and S verdrup transport for the southern hemisphere suggest that the formation density of the AAIW is the highest density which can be subducted in the South Pacific. The higher density of AAIW in the South Atlantic may result from more complex processes. The connection between the subtropical gyres of the Atlantic and Indian and between the Indian and Pacific Oceans contributes to modification of AAIW as it spreads tortuously northward around the subtropical gyres. Potential vorticity and AAIW salinity and oxygen illustrate the near barrier between the subtropical and tropical regimes, at about 20°to 25°north and south of the equator. Communication between the regimes is primarily through the western boundary currents.

Direct measurements of magnitude of the northward flow of the Malvinas (Falkland) Current have recently been made with two types of Lagrangian platforms: ALACE floats which cycled between 750-m depth and the sea surface, and 100-m drogued surface drifters. Each data set clearly delineates the path of the Malvinas Current, and the vertical shears inferred from them are commensurate with historical geostrophic shears. Velocities from the surface drifters are used here to adjust geostrophic shears from historical measurements, and the results confirm a large transport of the current, as previously implied by numerical models and a regional inverse calculation. At 42°S, the northward transport of the Malvinas Current in the upper 3000 m appears to be about 70 Sv, several times larger than estimates obtained by adjusting geostrophic shears to assumed levels of no motion. This large barotropie component may have significance in the cross-frontal transfer of intermediate and deep waters from the circumpolar current to the adjacent flow regimes in the South Atlantic, and thus on the inter-basin exchange of water masses.

The Deep Basin Experiment (DBE), a part of the World Ocean Circulation Experiment (WOCE), is presently underway in the Brazil Basin of the South Atlantic. The program objectives and design philosophy are reviewed and early results are presented. Observations from a moored array along the southern boundary and neutrally buoyant float trajectories in the North Atlantic Deep Water and Antarctic Bottom Water are described with emphasis on their relationship to the recent flow schemes offered by Reid (1989). Also discussed are the process of cross isotherm mixing within the intense flow regime of the Vema Channel and observations of long period warming of the bottom water.

The importance of the Deep Western Boundary Current (DWBC) in the Atlantic for the interhemispheric exchange of water masses and of heat is well known, but data to estimate transports and to follow its pathways are sparse, especially in the equatorial Atlantic. New insight into the distribution of water masses in the DWBC, their transports and their variability off Brazil were gained in a contribution to the WOCE (World Ocean Circulation Experiment) program. In these studies, moored current meter measurements were combined with shipboard data from three cruises in the years 1990, 1991 and 1992. Besides tracer (Chlorofluoromethanes CFMs, components F11 and F12) and hydrographic data, direct velocity measurements were carried out using a lowered ADCP attached to the CTD, and the Pegasus profiling system.The estimated transports of deep water in the equatorial Atlantic, net eastward transport of 19–22 Sv at 44°W, and 26.8 ± 7.0 Sv at 35°W, net southward transport of 19.5 ± 5.3 Sv at 5°S, are in the range of previously published estimates farther west and south. The data show significant spatial and temporal variability of the flow field and of the estimated transports as well as variability in the tracer distributions. This can lead to large uncertainties in the interpretation of single cruise observations.The transient tracer distributions (CFMs and tritium) along the DWBC from the northern North Atlantic to 10°S have been used within a box model to estimate the mean spreading velocities of the tracer bearing water masses of the DWBC. Together with assumptions about the mean spatial extent of the DWBC and the vertical velocity structure a mean transport of the DWBC of 9–12 Sv is obtained. These numbers represent the spatially and temporally averaged net DWBC transport, i.e. the deep part of the thermohaline circulation. Thus, the tracer derived estimates are comparable to the estimates from inverse calculations. The different results between the transports obtained from direct observations on one side and from inverse calculations and tracer distributions on the other side are thought to be caused by various recirculation cells along the path of the DWBC. Indications for various recirculation paths of part of the NADW have also been found off Brazil.

CFM (CCl₃F or freon F-11 and CCl₂F₂ or freon F-12) measurements were made on board the R.V. Atalante as part of a quasi-synoptic survey of the tropical Atlantic Ocean. The work was carried out during the CITHER 1 cruise, part of the French CITHER program (Circulation THERmohaline) between January and March 1993. Two zonal sections at 4°30 S and 7°30N (the A7 and A6 WOCE sections) and two meridional sections at 35°W and 3°50 W were sampled for CFMs between the African and South American continents. The results reported here deal primarily with the North Atlantic Deep Water. The CITHER 1 sections were made just 10 years after the first CFMs snapshot of the tropical Atlantic ocean obtained during the Transient Tracers in the Ocean Program.The detection limit was approximately 0.0025 pmol.kg-1 for both F-11 and F-12. This is sufficient to allow the determination of “apparent” ages and dilution factors for the freon- enriched tongues of the Upper and Lower North Atlantic Deep Water (UNADW centered around 1600 m and LNADW centered around 3800 m).Both zonal sections clearly show the propagation of UNADW into the eastern basin. The eastern meridional section at 3°50W shows the CFM cores extending from 4°S to 3°N with a maximum around 2°S. From the equatorial CFM ratios in the eastern basin, we estimate an eastward velocity close to 2 cm/s.The CFM distributions at the levels of the UNADW and LNADW show a large variability, probably linked to northern and southern deep recirculation gyres. In both sections, CFM enriched cells are clearly the result of reversed currents.The net decrease of CFM in the LNADW between 7°30N and 4°30S is partly the result of the bifurcation of the deep flow north and at the equator. This is confirmed by data taken in November 1992 in the region of the Equatorial Romanche Fracture Zone by the ROMANCHE 2 cruise. The influence of Antarctic Bottom Water is noticeable along the South American continent in the southern section. There is no evidence, through CFM data alone, for a northward flow of this “young” bottom water mass, which is probably blocked by the topography near the sampled areas.

The South Atlantic ocean is widely open to the Indian Ocean and the North Atlantic Ocean, and has a large inflow from the Pacific Ocean through the Drake passage. Strategies of modelling the ocean circulation in this area require to consider inter-ocean exchanges. The paper discusses various numerical approaches of the problem. One consists in modelling the world ocean and to study the South Atlantic as a sub-domain. Exchanges between the various oceans are thus determined by the model, and the circulation obtained for the South Atlantic depends upon the overall model behaviour. It is a consistent way to diagnose the circulation of the water masses in the ocean with coarse resolution models. This approach is also possible at eddy resolving resolution since the global ocean modelling effort undertaken by Semtner and Chervin (1988, 1992), and the simulation of the southern ocean circulation realised by the Fine Resolution Antarctic Model experiment (FRAM, Webb et al. 1991). However, there still are large differences between the various hydrographie or model estimates of the fluxes at the limits of the South Atlantic.Another approach is to limit the modelled domain to the South Atlantic basin, and then have to deal with the complicated problem of open boundaries. In that case, the simulated circulation of the South Atlantic will largely depend upon the fluxes prescribed at the boundaries. This approach requires an a-priori knowledge of incoming mass, heat and salt fluxes, which is generally derived from climatologies of hydrographie data. It appears suited to process studies which characteristic time scales are short compared to the time evolution of the deep circulation, and to sensitivity studies investigating the impact of the fluxes at the boundaries on the estimates of the meridional mass and heat transports.

Mass and heat transports in the South Atlantic as well as exchange flows with the South Pacific and the Indian Ocean are determined by driving a conservative, steady box-model towards the historical temperature (θ) and salinity (s) observations. The optimal model circulation searched for is required (a) to approximately preserve the vertical velocity shear as given by geostrophic calculations and (b) to correctly reproduce the measured distributions of 9 and s Information contained in the θ/s data on baroclinic flows is exploited through constraint (a) and the unknown reference velocities are determined by the model in a way such that the resulting absolute flow velocities produce realistic θ and s fields (constraint (b)). The model is mass, heat and salt conserving and has realistic topography. The adjoint method is applied as an efficient means for calculating cost function gradients needed during the optimization process.Model experiments show that indeed realistic θ and s model distributions can be obtained with flows that are consistent with geostrophy. Moreover, close agreement between measurements and model is obtained for a variety of model velocity fields that differ considerably with respect to strength of the meridional overturning cell and magnitude of meridional heat transports. The maximal acceptable meridional heat transport across 30°S (based on an evaluation of θ/s misfits and deviations from geostrophic shear) amounts to 0.4 PW. Forcing the model to produce larger heat fluxes results in systematic property misfits in the upper layers of the South Atlantic. Contrary to most published heat transport estimates the model also accommodates poleward (southward) heat fluxes of up to -0.5 PW. The best model property fields are obtained for a heat transport across 30°S close to zero. All acceptable model solutions show a dominance of northward flow of Antarctic Intermediate Water (AAIW) over warmer, upper layer waters, and all model temperature and salinity fields.

Long-term particle fluxes have been measured with time-series sediment traps off Cape Blanc (CB), in the southern Guinea Basin (GBS), in the northern one (GBN) and off Namibia (Walvis Ridge, WR). Seasonality was most strongly expressed at the GBS and WR sites. Production half-time (length of time to generate one half of the annual productivity; Berger and Wefer 1990) ranged from 4.6 to 1.8 months suggesting almost constant (e.g at Cape Blanc) to highly-peaked (e.g. at GBS) production systems. Off Cape Blanc, mean annual total flux was 45.3 g m-2 to a water depth of 3204 m, exhibiting substantial variation of 62% (deviation from mean value). Holocene sediment accumulation rates amounted to 17 g m-2 yr-1. At GBN, mean total flux was 36.1 g m-2 at 3939 m of water and the interannual variation reached only 11%. South of the equator at GBS, a total flux of 36.4 g m-2 to a water depth of 3382 m was determined. Bulk sediment accumulation rates in the Gulf of Guinea ranged between 17 and 23 g m-2, but revealed higher values at GBS. At the Walvis Ridge, we obtained a mean total deep-water flux of 31.1 g m-2 with 42% interannual variation; sediment accumulation rates amounted to 15 g m-2. Average annual organic carbon flux to the seafloor was 1.6 (Cape Blanc), 2-2.1 (Guinea Basin) and 3.2 g m-2 (Walvis Ridge); these values are typical for open-ocean-(coastal) upwelling transition systems. They revealed no relationship to literature-derived annual primary production values. Organic carbon preservation was generally poor and estimated as 7.5% (Cape Blanc), 2.9%-3.0% (Guinea Basin) and 1.9% (Walvis Ridge) of the carbon fluxes into the traps, respectively. Carbon accumulation was not related to the deep-water fluxes nor to primary production estimates taken from literature. Mean annual total, lithogenic and biogenic opal fluxes generally increased with depth in the water column (lithogenic fluxes about two-fold), obviously due to the contribution of a substantial fraction of fine-grained, resuspensed material originating at topographic elevations. In contrast, organic carbon fluxes decreased with water depth at all sites following an exponential decline. Surprisingly, the accumulation of refractory lithogenic material in the underlying sediments was only 22–56% of the deep-water fluxes and resembled the subsurface fluxes more closely. We assume that the additional resuspended fraction originated at topographic elevations and was not incorporated yet into the sedimentary record but is being transported in suspension and dispersed in the ocean.

Distribution patterns of potential temperature (8) and of H₄SiO₄ concentrations show that the bottom water in the major part of the Angola Basin originates from the Romanche Fracture Zone. At present the bottom water of the Angola Basin is predominantly lower NADW with a 10–20% admixture of a southern component. South of about 29°S a gradually increasing percentage of the bottom water originates from the Cape Basin, entering through gaps in the Walvis Ridge. Near the deep entrances into the Guinea Basin and the southern tip of the Angola Basin, vertical gradients of temperature and salinity are large. Small changes in the relative production rates of AABW and NADW, assuming that their properties remain the same, could cause a large change in bottom water composition of the Guinea and Angola Basins.The concentrations of bio-SiO₂, in sediments and of H₄SiO₄ in interstitial waters increase in the NE Angola Basin and parts of the Guinea Basin. This reflects enhanced productivity by diatoms, caused by various kinds of upwelling. Upward diffusion of H₄SiO₄ from interstitial waters causes a H₄SiO₄ “excess” of up to 10 μM in deep waters above the Zaire deep sea fan.Bio-SiO₂ in the sediments of the Zaire deep-sea fan contains up to 6% of aluminium, which contracts the amorphous structure and lowers its solubility and dissolution rate, thereby greatly increasing its preservation. The incorporation of Al into bio-SiO₂, mainly occurs in the surface sediment. Gibbsite and possibly kaolinite are considered likely sources of the Al.

The Rio Grande Rise delineates a large-scale topographic barrier in the South Atlantic separating the Brazil Basin to the north from the Argentine Basin in the south. In addition to the Verna Channel the Hunter Channel represents an important conduit for the equatorward flow of Antarctic Bottom Water. Motivated by the lack of reliable topographic charts from the Hunter Channel Region (34°S, 28°W) we have compiled multibeam soundings from a number of different bathymetric surveys taken on board RV Meteor, resulting in a set of three-dimensional images from the region. They are presented in context with remarks on the name “Hunter Channel”, together with selected hydrographic observations and long-term near-bottom current meter records.

The South Atlantic is tightly coupled to the North Atlantic climate amplifying system. At present, enormous amounts of heat are delivered across the equator to the north, with surface and subsurface waters. The return flow occurs at depth, within the coldwater sphere. During the last glacial the Atlantic Heat Conveyor was much less efficient, that is, the North Atlantic heat piracy is a positive feedback on climate change. This positive feedback is an important ingredient in the orbitally driven climate cycles. The current (that is, late Quaternary) conditions in the South Atlantic are the result of a long evolution of climate and geographic boundary conditions, which started with the opening of the basin at the end of the Jurassic and in the early Cretaceous, by continental breakup and seafloor spreading. Todays margins contain the ancient deposits of a narrow trough with restricted access, including evaporites. Warm-ocean sediments accumulated during the Cretaceous, including organic-rich deposits indicative of poorly oxygenated deep waters. Sinking of the sea floor from cooling of the lithosphere, and ridges produced as hot spot tracks (from Tristan da Cunha on the Mid-Atlantic Ridge) determined the main features of the bathymetry. The leitmotifs of Cenozoic evolution are general cooling (from mountain building and associated regression, and from reduction of atmospheric CO₂), the closure of the world-encircling tropical Tethys Ocean and opening of passages in the south, linking ocean basins through a circumpolar Cold Ring. Overall regression and associated polar deepwater production forced new patterns of biogenous deposition which resulted in a large-scale global drop of the Carbonate Compensation Depth (CCD) about 40 million years ago. At the same time, the isotopic ratio in the element strontium in seawater (as captured by calcareous fossils) started a long trend toward more radiogenic values, indicating increased supply of continental material. The major reorganization in deepsea sediments (the Auversian Facies Shift) in the late Eocene is also expressed as the onset of deposition of rather pure pelagic carbonates, with opaline and organic-rich sediments being increasingly restricted to ocean margins. Continued cooling eventually led to large-scale deepwater formation in high latitudes, which is expressed in the first great cooling step in the deep sea, at the end of the Eocene. The second great cooling step saw the buildup of ice on Antarctica, roughly 15 million years ago, presumably after considerable reduction of atmospheric CO₂. The third great cooling step consists of ice buildup around the North Atlantic, a step that moved the system into modern climate dynamics. Concerning the third step, it is commonly surmised that the closing of the Panama Straits was responsible for its timing (about 3 million years ago). We propose (Panama Hypothesis) exactly the reverse: in fact, the emergence of the Isthmus greatly favored North Atlantic heat piracy, so that northern glaciations were delayed by several million years. After initial onset of northern glaciations (7 to 6 million years ago) it took another 3 million years of mountain building and CO₂ reduction to attain sustained glaciations (3 million years ago). This period of delay is the well-know warm period of the early Pliocene. The story of the onset of northern glaciations is further complicated by the fact that cooling first enhances NADW production, before the onset of northern glaciations, and then obstructs it, presumably by reduction of evaporation and by sea ice formation. The identification of the switch point of NADW production, from negative to positive feedback, is vital for the understanding of the ocean’s role in climatic change in the late Neogene.

A series of model experiments to investigate the data requirements of an inverse model of the Atlantic is described. Results of a first attempt to reconstruct the circulation system of the Atlantic for the last glacial maximum (about 20,000 B. P.) are shown. The amount of data used is quite small compared with an application to the present day Atlantic, for which the model was originally developed. Although the results reflect the difficult data situation, some features of the flow field of the glacial Atlantic are recovered. With some extensions the inverse modelling approach seems to be very promising.

There has been a major contradiction between benthic foraminiferal Cd/Ca and δ13C data concerning the labile nutrient chemistry of the Southern Ocean during the Last Glacial Maximum (LGM). Cd data indicates that LGM South Atlantic nutrient concentrations were as low as they are today, indicative of a persistent influx of nutrient-depleted North Atlantic Deep Water (NADW). δ13C data indicates that LGM South Atlantic nutrient concentrations were much higher than at present (even higher than anywhere else in the ocean at that time), and these data have been interpreted as signifying the complete shutdown of the export of NADW into the global ocean. This paper examines both true geochemical differences and various confounding foraminiferal artifacts for both tracers. While many different processes and artifacts affect both tracers in the margin, we conclude the discrepancy is mainly due to the „Mackensen Effect“of low foraminiferal δ13C as a result of high carbon flux to the sediments, and that LGM Atlantic Sector Southern Ocean nutrient concentrations remained similar to the levels encountered today.

Micropaleontological and oxygen isotope analyses of planktonic foraminifera from North Atlantic Ocean and Southern Ocean sediment cores have been used to reconstruct temperature and salinity of surface waters during the last glacial maximum. Whereas the Norwegian-Greenland Sea and the high latitudes of the North Atlantic experienced a large negative anomaly, the Southern Ocean maintained salinity values similar or slightly higher than the modern ones around Antarctica, thus favouring winter convection and deep water formation in the Southern Hemisphere. These data have been used to force the zonally averaged, three-basin ocean model of Louvain-La-Neuve. The model reproduces the main trends of the geochemically constrained glacial Atlantic circulation and suggests that the glacial production of Antarctic Bottom Water was slightly higher than the modern one, whereas that of North Atlantic Deep Water was reduced by about 40%.

The central problem of late Quaternary circulation in the South Atlantic is its role in transfer of heat to the North Atlantic, as this modifies amplitude, and perhaps phase, of glacial- interglacial fluctuations. Here we attempt to define the problem and establish ways to attack it. We identify several crucial elements in the dynamics of heat export: (1) warm-water pile-up (and lack thereof) in the western equatorial Atlantic, (2) general spin-up (or spin-down) of central gyre, tied to SE trades, (3) opening and closing of Cape Valve (Agulhas retroflection), (4) deepwater E-W asymmetry. Means for reconstruction are biogeography, stable isotopes, and productivity proxies. Main results concern overall glacial-interglacial contrast (less pile-up, more spin-up, Cape Valve closed, less NADW during glacial time), dominance of precessional signal in tropics, phase shifts in precessional response. To generate working hypotheses about the dynamics of surface water circulation in the South Atlantic we employ Croll’s paradigm that glacial - interglacial fluctuations are analogous to seasonal fluctuations. Our general picture for the last 300 kyrs is that, as concerns the South Atlantic, intensity of surface water (heat) transport depends on the strength of the SE trades. From various lines of evidence it appears that stronger SE trades appeared during glacials and cold substages during interglacials, analogous to conditions in southern winter (August).

Ice-age cooling of the central equatorial Atlantic Ocean reflects both equatorial upwelling and advection of cool water off the southern-hemisphere eastern boundary, but the largest contribution appears to be advection. This conclusion is based on planktonic foraminiferal assemblages that vary over the last ~300 ky in the tropical Atlantic and Pacific Oceans. Core-top maps of these assemblages reveal relationships to 1) the tropical-subtropical warm pool, 2) equatorial and coastal upwelling, and 3) eastern boundary current advection. The faunal indices are not sensitive to selective dissolution, as they do not correlate with measures of preservation such as calcite fragmentation or water depth. The sequence of changes going into a glacial interval is 1) initial cooling associated with equatorial upwelling driven by trade winds, followed by 2) amplified cooling by intensified meridional winds along the SE Atlantic margin which advected cool Benguela Current water to the central equatorial Atlantic. The Gulf of Guinea maintained its present cyclonic gyre circulation in glacial time, bounded by a front similar to the present Angola-Benguela Front. We prefer a model in which thermal gradients in the southern hemisphere drive changes in the South Equatorial Currents.

In order to reconstruct Late Quaternary variations of surface oceanography in the east- equatorial South Atlantic, time series of sea-surface temperatures (SST) and paleoproductivity were established from cores recovered in the Guinea and Angola Basins, and at the Walvis Ridge. These records, based on sedimentary alkenone and organic carbon concentrations, reveal that during the last 350,000 years surface circulation and productivity changes in the east-equatorial South Atlantic were highly sensitive to climate forcing at 23- and 100-kyr periodicities. Covarying SST and paleoproductivity changes at the equator and at the Walvis Ridge appear to be driven by variations in zonal trade-wind intensity, which forces intensification or reduction of coastal and equatorial upwelling, as well as enhanced Benguela cold water advection from the South. Phase relationships of precessional variations in the paleoproductivity and SST records from the distinct sites were evaluated with respect to boreal summer insolation over Africa, movements of southern ocean thermal fronts, and changes in global ice volume. The 23-kyr phasing implies a sensitivity of eastern South Atlantic surface water advection and upwelling to West African monsoon intensity and to changes in the position of the subtropical high pressure cell over the South Atlantic, both phenomena which modulate zonal strength of southeasterly trades. SST and productivity changes north of 20°S lack significant variance at the 41-kyr periodicity; and at the Walvis Ridge and the equator lead changes in ice volume. This may indicate that obliquity-driven climate change, characteristic for northern high latitudes, e.g fluctuations in continental ice masses, did not substantially influence subtropical and tropical surface circulation in the South Atlantic. At the 23-kyr cycle SST and productivity changes in the eastern Angola Basin lag those in the equatorial Atlantic and at the Walvis Ridge by about 3500 years. This lag is explained by variations in cross-equatorial surface water transport and west-east countercurrent return flow modifying precessional variations of SST and productivity in the eastern Angola Basin relative to those in the mid South Atlantic area under the central field of zonal trade winds. Sea level-related shifts of upwelling cells in phase with global climate change may be also recorded in SST and productivity variability along the continental margin off Southwest Africa. They may account for the delay of the paleoceanogreaphic signal from continental margin sites with respect to that from the pelagic sites at the equator and the Walvis Ridge.

Planktic foraminifera data from three cores of the Angola-ZaYre margin are used to reconstruct palaeopositions of the Angola-Benguela Front (ABF) between the warm Angola Current and cold Benguela Current for the last 180,000 years. Strong northward shifts occurred in stages 4 and 3.3–3.1, but not in stages 6 and 2. The Benguela Current did not penetrate into the Gulf of Guinea. The southernmost positions, not far from the present one, were occupied in stages 5.5 and 1, but also in stage 6.3. The record of the shifts contains significant variance in the 23 ky-1 orbital frequency band and there are indications for a strong 100-ky-1 frequency component. There is also much variance at 15 ky-1, the sum frequency of the 23-ky-1 and the absent 41-ky-1 cycles. This cyclicity is a real feature in the records.The Benguela Current system performs a combination of two types of precessional (23 ky-1) movements: shifts of the ABF to the south and north precede swings between more zonal and more meridional directions respectively of the Benguela Oceanic Current by 6.6 ky. The extreme positions of the ABF may be described by the 100-ky-1 eccentricity component which is also documented in cores from Walvis Ridge and Cape Basin. Both the 23-ky and 100-ky shifts of the ABF are in phase with the meridional movements of the Subtropical Convergence zone (STC) in the southern Indian Ocean, and with advection variations in the equatorial Atlantic. The movements of the STC are probably driven by meridional displacements of the belt of westerly winds over the southern hemisphere. These caused displacements of the circumpolar fronts, the SE trades and the Benguela Current system, and made the cool advection at the equator fluctuate.The northernmost positions of the ABF coincide with strong sea-surface temperature minima in the Arabian Sea and the equatorial Pacific Ocean. These minima are probably the result of increased advection of relatively cool surface water from the south which is also caused by the northward eccentricity-driven movement of the westerlies and the associated oceanic polar fronts.

Variations in benthic foraminiferal δ13C from a suite of cores located in the equatorial western Atlantic document that the production rate of North Atlantic Deep Water has varied considerably during the last 400,000 years. The cores are located on the eastern slope of Ceara Rise and monitor the chemical variation in deep waters over the depth range of 3200 to 4300 m. Peak production apparently occurred during interglacial stages with isotopic stage 7.3 (216ka) exhibiting the greatest bathymetric extent of NADW. Minima in NAD W bathymetric range occurred during glacial maxima, when significant northward penetration of deep water from the Southern Ocean occupied large portions of the western Atlantic. Minimum production of NADW and maximum penetration of southern-source deep water occurred during glacial stages 4, 8 and 10. During these periods, there is no evidence for NADW at depths below 3200 m.Climate change and deep water production are linked but the records are complicated because maxima in NADW production do not necessarily occur during the most extreme interglaeials. Also there is a water depth-related difference in the phase relationships between ice volume and deep water production. In the deepest sites, changes in deep water production (as denoted by benthic foraminiferal Δδ13C) appear to lead changes in ice volume (as denoted by benthic foraminiferal δ18O) for both the eccentricity (100 kyr) and obliquity (41 kyr) periods of orbital geometry, providing partial confirmation of the Imbrie et al. (1992) model of a deep water-climate link. But the shallower cores do not lead ice volume at the 100 kyr period. Furthermore, NADW production lags ice volume in the precessional period at all water depths, unlike the records used in the Imbrie et al. (1992) model. Higher frequency, sub-orbital oscillations in benthic foraminiferal 8I3C imply that NADW production is occurring at frequencies like those seen in ice core records. These oscillations in NADW production appear to be restricted to intervals when ice volume was greater than today’s.

Carbonate dissolution data (sand contents) and δ13C records of the epibenthic foraminifer Cibicides wuellerstorfi from 12 gravity cores are used to reconstruct the history of deep water circulation in the South Atlantic for the last 360,000 years. The cores were selected from depth-sections in four basins (Brasil-, Guinea-, Angola- and Cape Basins) in water depths between 2900 m and 4600 m. The depth-transect approach allows removal of mean global shifts as well as local productivity effects from the paleo-property records and extraction of variations which are due to changes in deep water chemistry and/or circulation in the South Atlantic. As a result of the reduction of NADW during the last glacial maximum the Southern Component Water was higher in the water column and extended farther north than it does today. This glacial water mass can be divided into an upper part (USCW) with δ13C values between 0.2%o and 0.7%o and a lower part (LSCW) characterized by values of -0.2%o to 0.2%o. The boundary, marked also by the calcite lysocline, was at 3800 m water depth near the equator and rose slightly toward the Southern Ocean. The asymmetry observed in bottom water circulation today (LCDW in western basins and in the Cape Basin, NADW in eastern basins below 4000 m) was not present. From comparison to a deep western Pacific core (ODP 806B; Bickert et al. 1993) there is evidence that the nutrient-enriched but oxygen-depleted LSCW resembles the glacial Pacific Deep Water. This is also true for the older glacial stages 4, 6, 8 and 10.

Downcore clay mineral fluctuations in Late Quaternary sediment cores from the southeastern South Atlantic and adjoining Southern Ocean are of low amplitude. North of the Antarctic Circumpolar Current/Weddell Gyre boundary, small-scale variations, particularly of clay mineral ratios, essentially monitor cyclic changes of deep water advection in response to climatic oscillations.Kaolinite and chlorite are of most reliable palaeoceanographic significance. Chlorite originates from southern high-latitudes (Antarctic Peninsula, Patagonia) and is transported and distributed by northward advection of southern-source deep water. Northern-source deep water carries pedogenic kaolinite from the tropical regions to the Southern Ocean. The opposite advection of kaolinite and chlorite by both deep water masses is displayed by a latitudinal zonation of kaolinite/ chlorite ratios with higher values to the north. Downcore kaolinite/chlorite variations exhibit raised ratios during warm climatic stages compared to glacial periods. Cross spectral analyses reveal that kaolinite/chlorite ratios are coherent and in phase with global ice volume in the 41-kyr and 100-kyr periods of obliquity and eccentricity.By relating kaolinite/chlorite ratios in modern deep-sea sediments to the present-day deep water distribution, a model of past deep water advection may be inferred from downcore variations of kaolinite/chlorite ratios. Thus, northern-source water reaches not farther south than 40°S during glacial maxima. During interglacial optima northern-source deep water is injected into the Southern Ocean to latitudes around 55°S.