The Northern North Atlantic

Humanity today is living in a fast-changing world. Our future depends on our ability to use the planet and its natural resources without destroying them, and to adapt to changes in the global environment in the very near future. This can only be achieved through an intimate understanding of the modem and past environments, and thereby its potential for change, especially with respect to basic rates of change, and to essential properties of the environment.

The ice information discussed below addresses both biological and geological processes in the Greenland Sea. Background information is provided for a better understanding of the effects an ice cover has on the upper water column.

Data on the inorganic carbon system, the distribution of oxygen, nitrate, and phosphate, as well as particle sedimentation and plankton biomass collected from winter 1993 to summer 1996 in the central Greenland Sea show that although this area is a sink for atmospheric carbon throughout the year, relatively little of the carbon fixed by photosynthesis into organic compounds in the surface waters is eventually sequestered in deep waters. Rather, due to intensive biological remineralization of organic matter within the winter mixed layer, the bulk of carbon is retained in the upper few hundred meters of the water column. The sequestration of biogenic carbon is constrained by the depth of the winter mixed layer, in that deep winter mixing effectively increases the depth below which true export can occur. There is potential for increased export with increased rates of deep convection. Likewise, a reduction in heterotrophic recycling in near-surface waters may enhance the effectiveness of the biological pump. However, because of our still limited understanding of the interactions between the biological and solubility pumps in this region, the extent to which export may be enhanced is unclear.

A decade of particle flux measurements providse the basis for a comparison of the eastern and western province s of the Nordic Seas. Ice-related physical and biological seasonality as well as pelagic settings jointly control fluxes in the western Polar Province which receive s southward flowing water of Polar origin. Sediment trap data from this realm highlight a predominantly physical flux control which leads to exports of siliceous particle s within the biological marginal ice zone as a prominent contributor. In the northward flowing waters of the eastern Atlanti c Province, feeding strategies, life histories and the succession ofdominant mesozooplankters (copepods and pteropods) are central in controlling fluxes. Furthermore, more calcareous matter is exported here with a shift in flux seasonality towards summer I autumn. Dominant pelagic processes modeled numerically as to their impact on annual organic carbon exports for both provinces confirm that interannual flux variability is related to changes in the respecti ve control mechanisms.

Annual organic carbon export s are strikingly similar in the Polar and Atlantic Province s (2.4 and 2.9 g m⁻² y^−l at 500 m depth), despite major differences in flux control. The Polar and Atlantic Provinces, however, can be distinguished according to annual fluxes of opal (1.4 and 0.6 g m⁻² y^−l) and carbonate (6.8 and 10.4 g m⁻² y^−l). Interannual variability may blur this in single years. Thus, it is vital to use multi-annual data sets when including particle exports in general biogeochemical province descriptions. Vertical flux profiles (collections from 500 m, 1000 m in both provinces and 300-600 m above the seafloor deviate from the general vertical decline of fluxes due to particle degradation during sinking. At depths> 1000 m secondary fluxes (laterally advectedlresuspended particles) are often juxtaposed to primary (pelagic) fluxes, a pattern which is most prominent in the Atlantic Province.

Spatial variability within the Atlantic Province remains poorly understood, and the same holds true for interannual variability. No proxies are at hand for this province to quantitatively relate fluxes to physical or biological pelagic properties. For the seasonally ice-covered Polar Province a robust relationship exists between particle export and ambient ice-regime (Ramseier et al. this volume; Ramseier et al. 1999). Spatial flux patterns may be differentiated and interannual variability can be analyzed in this manner to impro ve our ability to couple pelagic export patterns with benthic and geochemical sedimentary processes in seasonally ice-covered seas.

Pelagic and ice-associated particle sources have been investigated to determine their contribution to vertical fluxes from upper ocean layers. Process studies were conducted from 1988 to 1997 during various seasons between 72° N and 82° N in the Marginal Ice Zone (MIZ) and open waters of the Greenland Sea. Ice-bound (in-ice and under-ice) particle production begins as early as April, prior to pelagic production, and provides material which may be set free in the course of melting or originate from food-web processes in the under-ice habitat (namely grazing by sympagic amphipods). These particles may be deposited from surface waters, degraded within pelagic food webs or, in the case of autotrophic components, may serve as a seeding population for pelagic production. Findings on transects from the open water into the pack ice stress the overall importance of the MIZ for particle export. The MIZ is characterized by highly variable physical and biological conditions which foster local phytoplankton blooms. Particle exports from the MIZ are very variable but generally high, with prominent autotrophic diatom contributions (up to 60 mg poc m⁻² d⁻¹, 30 mg Opal-Si m⁻² d⁻¹ and 10⁷ diatoms m⁻² d⁻¹). Analyses of algal pigments and their degradation products, combined with microscopical inventories, permit the differentiation of sources of particle export. Freshly produced material from the MIZ can rapidly sediment to great depths, feeding the benthos and affecting sediment geochemistry.

The present study gives an overview of recent research results concerning the distribution of living fossilizable plankton groups (coccolithophores, diatoms, dinoflagellate cysts, radiolarians and planktic foraminifers) in the water column of the Nordic Seas. Information is combined with results of sediment trap investigations on fluxes and the species composition of these fossilizable plankton groups in the Nordic Seas and their relation to oceanographic conditions. In order to validate spatial and temporal occurrence patterns of the plankton assemblages, their fluxes were monitored with sediment traps over several years in three different oceanic regions, characterized by the major surface water masses of the Nordic Seas. In addition, the alteration processes on settling plankton assemblages were investigated at three water depths.

Oceanographic domains in the Nordic Seas can be clearly distinguished on the basis of the export of different plankton assemblages. The Norwegian Sea is dominated by calcareous plankton and the Greenland Sea and the Jan Mayen Current are dominated by siliceous plankton organisms. All fossilizable plankton groups show a corresponding decrease with low interannual variations in the number of species from the temperate Norwegian Sea to the colder Jan Mayen Current and the Greenland Sea. Dissolution is one of the major alteration processes and alters diatom assemblages significantly during sinking and settling. No quantitative signal of siliceous plankton groups is preserved in the deep traps or in sediments. The second major alteration process is lateral advection. This process influences the investigated sites on different scales.

The distribution of calcareous, siliceous and organic-walled planktic microfossils has been investigated in surface sediments from the Nordic Seas, Planktic foraminifers, coccolithophores, radiolarians, diatoms and dinoflagellate cysts generally reflect major oceanographic domains in the distribution of sediment assemblages, The occurrence of fossilizable plankton in the water column and in sediments has been compared to describe processes which alter the composition of assemblages.

The distribution of calcareous microfossil assemblages and calcium carbonate contents in surface sediments is generally related to warm water inflow into the Nordic Seas because dissolution in the water column and sediments does not alter the general composition of assemblages and carbonate fluxes. Most planktic foraminifer and coccolithophore species are adapted to warmer surface waters, and the number of species is low in sediments underlying the colder water masses of the western Nordic Seas. Therefore, the Polar and Arctic Domains cannot be clearly recognized in the distribution of calcareous assemblages.

The composition of siliceous and organic-walled microfossil assemblages is more variable in the Greenland and Iceland Seas than is that of calcareous microfossil assemblages, and permit to separate the Atlantic, Arctic and Polar Domains, Theoretically, the varying environments of the Nordic Seas are better reflected in diatom and radiolarian assemblages than in coccolithophore and planktic foraminifer assemblages, but selective dissolution in the water and sediment columns, particularly in the Greenland Sea, cause a strong distortion of siliceous microfossil assemblages in the sedimentary records.Although opal fluxes and fluxes of siliceous microfossils are much higher in the Greenland Sea than in the Norwegian Sea, this cold-water signal is not preserved in surface sediments.

Therefore, the spatial variability in the sedimentation of carbonate and opal in the water column is only partly reflected in surface sediments. The distribution of biogenic carbonate and opal in sediments suggests that the Nordic Seas are predominately characterized by production related to warm water inflow. Additionally, the low diversit y of calcareous microfossils and the low preservation potential of siliceous microfo ssils limits their applicability in cold-water environments of the Nordic Seas, where organic-walled dinoflagellate cysts may have the best potential to reflect the complex hydrographic conditions in the fossil record.

In recent years considerable attention has been paid towards gaining a better understanding of the cycling of elements along ocean margins. The recognition of the importance of the carbon cycle for the earth’s climatic system and an evaluation of the role of the ocean in this respect, as well as an estimate of human-induced climatic (and in the near future anthropogenic) change on a global level, can only be obtained by modeling fluxes of energy and elements between coastal region s and the deep ocean as well as between high and low latitudes. This requires better and more detailed insights into exchange and transport processes from the shelf via the continental slope to the deep sea, and a detailed quantification of fluxes between compartments (van Weering et al. 2000).

Sediment transport processes in the northern North Atlantic have been investigated on the basis of various numerical models. A general circulation model has been used to investigate large-scale particle transport, a reduced gravity plume model has been used to investigate particle transport by cascading from the shelves into the deep basins, an ocean slice model has been used to investigate particle exchange processes between a bottom current and the ambient water mass, and a Bottom Boundary Layer model has been used to investigate particle interactions influencing the settling behavior of suspended particles. In this paper, the various processes investigated in these models are described (i) schematically, (ii) on the basis of field data, if available, and (iii) by employing results from numerical simulations. In a first attempt the northern North Atlantic will be divided into separate process defined areas, which can be used in carbon budgeting, for example.

Acoustic mapping and sampling of Holocene and Late Weichselian deglacial sediments combined with oceanographic measurements in the sediment source and accumulation areas on the eastern continental margin of the Norwegian-Greenland Sea provide evidence that episodic transport towards high-accumulation areas is primarily downslope and gravity driven. Triggered by various hydrographic conditions, the runoff repeatedly followed the local gullies and channels of the seafloor.

In contrast, sedimentological and oceanographic data from a topographic sediment trap on the Vøring Plateau at the Norwegian Sea margin indicate upslope sediment transport in the bottom nepheloid layer.

Less comprehensive data from the Greenland Sea margin suggest that the prerequisites to sediment advection in terms of oceanographic structure, availability and sources of sediment differ from those of the eastern margin. The resulting near-bottom sediment transport is less pronounced. Advected sediments reaching the continental rise and the abyssal plain follow an old channel system.

The forcing mechanisms of sediment advection changed dramatically from Late Weichselian to Holocene times due to changes in sea level, sediment availability, and water-mass stratification.

The distribution and structure of zoobenthic communities have been investigated in the northern North Atlantic. The principal goal of these studies is to assess the degree to which benthic community patterns depend on and/ or mediate carbon flux between the pelagic and benthic realms, as well as between seabed, sediment-water interface and benthic boundary layer. A common rationale is that these patterns integrate the impact of environmental factors over longer periods of time and reflect relatively long-lasting or predictably recurrent environmental states, thus providing clues to the relative significance of potential community determinants on a time scale of months to years. Since 1992, several meso-scale field studies have been carried out in three regions at the East Greenland continental margin between 68° N and 81° N at water depths ranging from 40 to 3,700 m. A suite of sampling methods was employed (corers, trawls, seabed imaging) to adequately probe various benthic community fractions, such as foraminifers, poriferans, macrobenthic endofauna, peracarid crustaceans and megabenthic epifauna.A depth zonation in the faunal composition, accompanied by a shift in the predominance of different feeding types and a significant decline in biomass and abundance by as much as two and three orders of magnitude was the most conspicuous general pattern detected. However, in terms of species richness, no common trend for water depth or latitude was perceivable. The general depth zonation of the macrobenthos as well as the spatial concordance of high macrobenthic abundance and biomass with relatively productive hydrographic zones, such as marginal ice zones, polynyas and anti-cyclonic gyres, provide evidence for the importance of water column processes and, hence, for subsequent food availability as major determinants for benthic assemblages and the significance of pelago -benthic coupling in the study area in general. However, for megafaunal species such as echinoderms, community patterns on a 10-1an scale and the dispersion of organisms on a 100-m scale, are best explained by seafloor properties. There is no evidence of direct pelago-benthic coupling, irrespective of water depth. These contrasting findings emphasize that the relative importance of potential community determinants can change with both spatial scale and life traits, e.g. body size, mobility and feeding ecology, of the organisms considered.

The sediment water interface represents a physical boundary between the water column and the sediment, which thus separates two environments with processes of significantly different intensities and spatial and temporal dimensions. Furthermore, the sediment water interface (SWI) does not only separate the present day environment from the paleoenvironmental record, but is also a layer where pelagic signal s are strongly modified before their final documentation over geological time scales. Thus, the SWI is one of the key areas of the global carbon cycle, where carbon is by burial removed from, or returned by remineralization to the atmosphere ocean system. Therefore, the quantification of fluxes reaching the seafloor and the process oriented understanding of degradation and remineralization pathways are important for a dynamic comprehension of marine geochemical cycles, the coupling between present dayand paleo-processes and the suitability of proxies applied for paleo-reconstructions.

Benthic communities below the euphotic zone are strictly dependent on food supply through the water column, with only few exceptions at hot vents and cold seeps. The magnitude of this supply of organic carbon is mainly determined by temporal and spatial patterns of the export flux of pelagic primary production from the euphotic zone. In contrast to the northern Pacific, where a sequence of ecosystems exists along a latitudinal gradient, in the Greenland-Iceland-Norwegian Seas (GIN Seas) north ofIceland two very different pelagic environments are present: export regimes are separated meridionally and exposed to the same external forcing of solar radiation. At the eastern boundary of the GIN Seas the influence of the Gulf Stream results in an environment similar to that found at lower latitudes and is well documented for biological processes in the water column (v. Bodungen et al. 1995) and at the seafloor (Graf et al. 1995). Seasonal ice-cover together with hydrographic and nutrient conditions define the continental margin of East Greenland as a typical polar environment.

Surface sediments from the high-latitude southern and northern Atlantic Ocean have been investigated with respect to the benthic cycle of particulate organic carbon (POC) and biogenic silica (BSi). Special emphasis has been given to the magnitude of benthic fluxes, their spatial distribution, their relation to particle-trap studies, and their ability to trace ice-related features.

In both polar regions enhanced fluxes of POC to the seafloor were observed for polynyas, reflecting the enhanced primary production of recurrent ice-free surface-water formation in these regions. Very similar fluxes were observed for organic carbon in the Weddell Sea and the northern North Atlantic. Within the northern North Atlantic regional differences in POC fluxes were investigated using an empirically derived relationship between benthic POC flux, primary production, and water depth. The calculated spatial budget suggests that 2% of the primary production is exported below 1000 m water depth and that 1.2% reaches the seafloor. These values are quite similar to temperate regions within the world oceans.

Sediments of the Norwegian-Greenland Sea were analysed for concentration, and enzymatic degradation potential of organic material, and additional environmental parameters related to microbial enzymatic activities. The pool of extracellular enzymes in the sediments has to be characterized as tolerant against variations of temperature and pressure. Instead, the availability of decomposable organic carbon was the main factor regulating enzymatic activities. Concentration and decomposition of organic matter varied strongly dependent upon location. Sediments from the Jan Mayen Fracture Zone colonized by epibenthic foraminifera revealed low organic carbon (C) and nitrogen (N) concentrations, low C/N ratios, and high enzymatic activities as compared to the other sediments investigated (Barents Sea, Kolbeinsey Ridge, East Greenland Shelf, Lofoten Basin, Vøring-Plateau).The occurrence of macrofauna and their biogenic structures greatly enhanced enzymatic activities. The existence of epibenthic foraminifera coincided with steep gradients of enzymatic activities in the uppermost sediment surface layers. The metabolic status of individual foraminifera could be detected by their enzymatic activities. One active foraminifer could already account for the hydrolytic activity measured in one cm⁴ of surface sediment. In sediments across the continental slope of the Barents Sea, general patterns of the decomposition and preservation of the deposited organic matter strongly dependent on the nutrient supply and the benthic colonization could be demonstrated. Sediments located on the slope with very high sedimentation rates showed a good preservation of the deposited organic material with C/N ratios close to ratios of surface sediments. Sediments located at the base of the slope with a very low supply with organic material revealed a preferential relative enrichment of N over C (low C/N ratios) of the residual organic material. Sediments located on the shelf with intermediate sedimentation rates were characterized by an incomplete oxidation of the deposited organic matter thus leading to the relative enrichment of Cover N (high C/N ratios). These observations demonstrate that the original signature of the sedimented organic material can be overprinted by biological activities. The mode of overprinting (relative enrichment of either C or N) is dependent upon the sedimentation rates of organic matter and the benthic colonization.

The presence of gas hydrates in oceanic sediments along the Norwegian Continental Margin is documented in high-frequency near-vertical and wide-angle seismic reflection data. The base of the hydrate stability zone (HSZ) is detected in reflection seismic sections by the occurrence of a strong bottom simulating reflector (BSR). Below the BSR, a low-velocity layer is interpreted as a gas-bearing zone. Modeling of the HSZ as a function of temperature and pressure shows a distinct thinning of the HSZ at the Norwegian Margin from the Last Glacial Maximum (LGM) to the present time. This highly dynamic HSZ system can release large amounts of methane from subseafloor hydrate reservoirs into the oceanosphere. Based on these modeling results, considerable amounts of carbon have left the submarine hydrate reservoirs of the Norwegian Continental Margin area since the LGM. A reasonable estimation for the released methane total volume (Standard Temperature and Pressure (STP)) lies between 3 and 13.10¹³m³ and for carbon mass between 16 and 67 GT.

Study of natural climate change is important for many reasons. These studies explain processes that drive climate change, interactions between the atmosphere, biosphere, oceans and glaciers.Afirm knowledge of climate change of the past (paleoclimate) is necessary in order to better model future change.

Five plankton groups, including diatoms, radiolarians, coccolithophores, foraminifers, and dinoflagellate cysts, were synoptically analyzed in six sediment cores and two sediment traps from the Norwegian-Greenland Sea and the North Atlantic in order to provide more detailed insights into the paleoclimatic and paleoceanographic evolution and the development of plankton assemblages of the northern North Atlantic during the last 15, 000 years. Based on Q-mode factor analyses, cold, warm, transitional, and relict assemblages were calculated for each of the plankton groups. Data from the different plankton groups complement one another, although they are not always consistent. However, the multiple plankton-group data set is able to bridge intervals in which single groups lack preservation or the ability to react to changes. Synoptically interpreted, the results provide a detailed picture of the response of plankton assemblages to environmental changes during the time period investigated, which includes the Bølling/ Allerød interstadial, the Younger Dryas cold spell, Termination I_B, and, in all likelihood, also the “8, 200 Event”, and the Hypsithermal (approximately 8-4 ¹⁴C ky BP).

The advantages and limitations of various proxy indicators of physical boundary conditions, such as sea-surface temperature (SST) and salinity (SSS), circulation patterns, deep-water ventilation, nutrient cycles and the response of the biota (productivity and carbon-flux), have been investigated in the Greenland-Iceland-Norwegian (GIN) Seas. Low sea-surface temperatures, the presence of sea ice, strong seasonality and vigorous thermohaline circulation in this area cause severe difficulties for the application of several classical temperature, salinity, nutrient and flux proxies. Quantitative estimates are provided for temperature, salinity and the seasonal duration of seaice cover, although margins of error are still larger than in the North Atlantic south of Iceland. However, proxies based on benthic foraminifera, reflecting biotic processes at the benthic boundary layer (i. e. assemblage composition and isotopic signals of benthic foraminifera), exhibit potential as indicators of seasonal carbon flux and deep-water ventilation in the Greenland-Iceland-Norwegian Seas.

Biomarkers are organic mole cules synthesized by specific source organisms (e.g. phytoplankton, terrigenous plants) which survive depo sition to sediments in a recognizable form. They can thus be considered to be chemical fossils. In some cases, the source organism is very well constrained (e.g. bacteriochlorophyll d, from strictly anaerobic photosynthetic bacteria), although sometimes the source is more general (e.g. chlorophyll a derived from photosynthetic organ isms). The utility of these organic components as paleoproxies largely depends on their resilience to early degradation processes during sedimentation, which must be relatively high so that the component is found in the sedimentary column. For instance, (1993) argued that some of the more diagnostic components are sometimes the least stable, giving as an example the high lability of carotenoids (e.g. Repeta and Gagosian 1987) which have some of the higher biologic al specificity of most other lipid classes. Other factors, such as specificity of source s (broad vs. restricted), will constrain the type of application. For instance, n-alkanes distributions in marine sediments are often related to terrigenous higher plant material s, and their long-range transport by wind (e.g. Simoneit et al. 1977). Thus, these components have a global distribution, and the interpretation of the n-alkane data from a global vs. local source standpoint can be ambiguous, as their source and associated windvector may be very difficult to determine.

Centennial- to millennial-scale changes in global climate over the last 60 ky were first documented in ice cores from Greenland, with ice sheets around the North Atlantic and its thermohaline circulation (THC) as prime candidates for a potential trigger mechanism. To reach a new quality in understanding the origin and causal links behind these changes, two strategies were intimately tied together in this synthesis, high-resolution 3-D ocean modeling and paleoceanographic reconstructions. Here, five time series with a time resolution of several decades and various time slices of surface and deep-water paleoceanography were established from hundreds of deep-sea cores for the purpose of monitoring rapid changes across the North Atlantic and testing or initiating model results. Three fundamental modes were found to operate Atlantic THC. Today, mode I shows intensive formation of North Atlantic Deep Water (NADW) and strong heat and moisture fluxes to the continents adjacent to the North Atlantic. Peak glacial mode II leads to a reduction in NADW formation by 30-50%, in line with a clear drop in heat flux to Europe. The glacial Nordic Seas, however, remain ice-free during summer and little influenced by meltwater, in contrast to the sea west ofIreland, where iceberg meltwater blocks an eastbound flow into the Norwegian Sea and induces a cold longshore current from Faeroe to the Pyrenees. The subsequent Heinrich 1 (HI) meltwater mode III leads to an entire stop in NADW and intermediate-water production as well as a reversed pattern of THC, stopping any heat advection from the central and South Atlantic to the north. In contrast to earlier views, the Younger Dryas, possibly induced by Siberian meltwater, began with mode I and ended with mode III, continuing into the Preboreal. Modeling the impact of modes I to III on the global carbon budget, we find that the atmosphere has lost 34-54 ppmv CO₂ from interglacial to glacial times, but has gained 23-62 ppmv CO₂ at the end of HI within a few decades, equivalent to 33-90% of modem, man-made CO₂ release. The robust 1500-y Dansgaard- Oeschger (D-O) cycles and their multiples of as much as 7200 years, the Heinrich event cycles, are tied to periodical changes between THC modes I/II and II/III. In the Irminger Sea rapid D-O coolings are in phase with initial meltwater injections from glaciers on East Greenland, here suggesting an internal trigger process in accordance with binge-purge models. Ice rafting from East Greenland and Iceland occurs only 240-280 y later, probably inducing a slight sea-level rise and, in tum, Heinrich ice rafting from the Laurentian ice sheet during H1, H2, H4, H5. At H1 a major surge from the Barents shelf has lagged initial cooling by 1500 y and entails the most prominent and extended reversal in Atlantic THC over the last 60 ky (probably also at the end of glacial stage 4, at H6). Meltwater stratification in the Inninger Sea reaches its maximum only 640 y after initial meltwater injection and induces, via seasonal sea-ice formation, brine-water injections down to 4 km water depth, signals leading the classic D-O jump to maximum warmth by only 125 y. It may be inferred from this short-phase lag that brine water-controlled deep-water formation probably entrains warm water from further south, thereby forming the key trigger mechanism for the final tum-on of the Atlantic THC mode II roughly within a decade (or mode I, in case of favorable Milankovitch forcing).

Sediment cores from the Nordic seas covering the past five climatic cycles have been investigated to elucidate the climate-induced relation between the pelagic and benthic realm from studies of foraminifera. A comparison of the total number of benthic and planktic foraminiferal tests (specimens per gram sediment) reveals corresponding fluctuations over the entire time range investigated. Highest abundances are normally observed during peak interglacial periods, whereas glacial periods are marked by generally reduced numbers offoraminiferal tests. Despite an overall similarity, on a spatial basis, the relative proportion of planktic and benthic foraminiferal abundance seems to have varied between each interglaciation. Moreover, distinct differences in species composition characterize some interglacial periods and short time intervals. Because these compositions have had no modem analogue at any time during the present interglacial (Holocene), it is suggested that they result from oceanographic conditions other than those that prevail in the Nordic seas today.

A high-resolution study of the past 25 ka reveals that benthic and planktic foraminifer increased in number after the end of the last glaciation, implying that changes in postglacial water masses had a direct impact on sea-surface and -bottom bioproductivity. This direct linkage between pelagic and benthic faunal productivity persists, although with some notable variations, throughout the Holocene. Furthermore, based on a high correlation coefficient between thermophile surfacewater species and the most dominant benthic suspension feeder, a strong pelagic-benthic coupling gives evidence of a continuous vertical connection of surface and bottom habitats in the Nordic seas during this time.

On the basis of the oxygen isotopic proxy for atmospheric temperature which has been measured in the GRIP and GISP2 deep ice-cores from Summit, Greenland, it is well known that climate in the North Atlantic sector of the planet was highly variable throughout the period of time conventionally referred to as Oxygen Isotope Stage (OIS) 3. This variability consists of millennium time-scale oscillations clustered in packets in which each successive oscillation has a somewhat smaller amplitude than the last. The time scale of individual packets is approximately 10 kyr. A hydrodynamic theory of the basic Dansgaard-Oeschger oscillation has been developed which comprises the millennium time-scale variability based upon a reduced model of the thermohaline circulation (THC). In the present paper the most recent results delivered by this hydrodynamic model are compared with those provided by a much simpler dynamical systems model based upon the assumption that the low-frequency time dependence of the THC may be described by interactions between the time evolving uniform salinities ascribed to each of three interacting boxes. The simple model is shown to agree remarkably well with the predictions of its more complicated hydrodynamic counterpart insofar as its qualitative behavior is concerned.

In recent years, a great wealth of new glacial sea-surface temperatures and salinities have been reconstructed from sediment core data for the intermediate to high latitudes of the North Atlantic. In the present study, the most recent sea-surface temperature and salinity data from the North Atlantic Ocean have been compiled to develop a physically consistent three-dimensional oceanographic scenario of circulation at the Last Glacial Maximum. For a hierarchy of numerical experiments, two general ocean circulation models, driven by traditional restoring boundary conditions, have been used. For the global experiments, a descendant of the Hamburg Large-Scale Geostrophic (LSG) ocean model has been employed, and regional experiments have been carried out with the Geophysical Fluid Dynamics Laboratory (GFDL) Modular Ocean Model (MOM). Both models are linked via the output of the global model, which is used for three-dimensional temperature and salinity restoring along the artificial boundaries of the regional model. This coupling of two models with coarse and fine resolutions offers a unique opportunity to exploit both basin-wide sediment core data coverage as well as the high spatial resolution provided by the cores from high northern latitudes.

This concluding chapter presents a critical appraisal of the research carried out within the framework of SFB 313. We will try to outline the scientific perspectives and problems as well as our approach to their solution. Moreover, we shall describe the scientific highlights of the results and define areas where we followed questionable assumptions and therefore failed to achieve what we had hoped for. A complete documentation of the various activities of SFB 313 (including an overview of research cruises, etc.) can be found in Schafer and Schroder-Ritzrau (1999).