Joc 926 PDF

INTERNATIONAL JOURNAL OF CLIMATOLOGY
Int. J. Climatol. 23: 1127–1159 (2003)

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/joc.926
REVIEW
THE ROLE OF THE OCEANS IN CLIMATE
G. R. BIGG,* T. D. JICKELLS, P. S. LISS and T. J. OSBORN
School of Environmental Sciences, University of East Anglia, Norwich, UK
Received 12 July 2002
Revised 5 April 2003
Accepted 5 April 2003
ABSTRACT
The ocean is increasingly seen as a vital component of the climate system. It exchanges with the atmosphere large
quantities of heat, water, gases, particles and momentum. It is an important part of the global redistribution of heat from
tropics to polar regions keeping our planet habitable, particularly equatorward of about 30° . In this article we review
recent work examining the role of the oceans in climate, focusing on research in the Third Assessment Report of the IPCC
and later. We discuss the general nature of oceanic climate variability and the large role played by stochastic variability in
the interaction of the atmosphere and ocean. We consider the growing evidence for biogeochemical interaction of climatic
significance between ocean and atmosphere. Air–sea exchange of several radiatively important gases, in particular CO2 , is
a major mechanism for altering their atmospheric concentrations. Some more reactive gases, such as dimethyl sulphide,
can alter cloud formation and hence albedo. Particulates containing iron and originating over land can alter ocean
primary productivity and hence feedbacks to other biogeochemical exchanges. We show that not only the tropical Pacific
Ocean basin can exhibit coupled ocean–atmosphere interaction, but also the tropical Atlantic and Indian Oceans. Longer
lived interactions in the North Pacific and Southern Ocean (the circumpolar wave) are also reviewed. The role of the
thermohaline circulation in long-term and abrupt climatic change is examined, with the freshwater budget of the ocean
being a key factor for the degree, and longevity, of change. The potential for the Mediterranean outflow to contribute to
abrupt change is raised. We end by examining the probability of thermohaline changes in a future of global warming.
Copyright  2003 Royal Meteorological Society.
KEY WORDS: climate system; air–sea exchange; carbon cycle; sulphur cycle; aerosols; tropical climate; decadal variability;
thermohaline circulation
1. INTRODUCTION
The ocean is an important component of the climate system. It provides the surface temperature boundary
condition for the atmosphere over 70% of the globe. It absorbs over 97% of solar radiation incident on it from
zenith angles less than 50° . It provides 85% of the water vapour in the atmosphere. It exchanges, absorbs
and emits a host of radiatively important gases. It is a major natural source of atmospheric aerosols. Thus,
even a static ocean would significantly influence the climate. However, the ocean is dynamic and its surface
properties will vary on all time scales, allowing great scope for feedbacks between the ocean and atmosphere.
Over the last two decades the importance of the ocean to understanding, and predicting the evolution of,
the climate system has become generally recognized. Now, a climate model needs to possess a coupled ocean
and atmosphere to be taken seriously, because oceanic processes, through the ocean’s thermal and dynamic
inertia, intrinsically contain the long time scales on which climate changes. This development in scientific
understanding of the role of the ocean in climate change can be seen in the various scientific assessment
reports of the Intergovernmental Panel for Climate Change (IPCC). The Third Assessment Report (TAR),
* Correspondence to: G. R. Bigg, Department of Geography, University of Sheffield, Winter Street, Sheffield S10 2TN, UK;
e-mail: g.bigg@uea.ac.uk
Copyright  2003 Royal Meteorological Society

1128 G. R. BIGG ET AL.
published in 2001 (Houghton et al., 2001), through the relevant sections in its chapters, provides a recent
review of the role of the ocean in climate. Here, while introducing the main mechanisms by which the ocean
influences the climate system, we will concentrate on updating and supplementing the IPCC’s TAR.
We focus in this paper on the role of the oceans in natural climate variations, with less consideration
given to their role in modifying the change in climate driven by anthropogenic perturbations to the Earth’s
radiation budget (via greenhouse gas and sulphate aerosol emissions, but also humanity’s influences on ozone
and land-surface characteristics). Only brief references to the ocean’s role in the coming century’s climate
change are given below, as we are not attempting to survey coupled modelling predictions directly relating to
the ocean. One exception is that we review current knowledge about changes in the thermohaline circulation
that may accompany anthropogenic climate change (Section 4.5), because this topic is strongly linked to our
review of its stability and variability. Thus we do not cover the fields of El Niño–southern oscillation (ENSO;
Diaz et al., 2001) and North Atlantic (Marshall et al., 2001) air–sea interactions already well reviewed in
this journal.
After discussing the ocean’s role in sea-level change, and reviewing the ocean’s influence on time scales
within the atmosphere and climate system generally, we proceed to the three major sections of the paper. These
are: (i) biogeochemical links between the ocean and climate; (ii) upper ocean, largely wind-driven, circulation
and interannual to decadal coupling to the atmosphere; and (iii) the thermohaline circulation (THC) and its
role in abrupt change. These topics are normally treated separately, but an important message from this
review is that all three processes are interlinked and the scale and nature of these interactions between the
biogeochemical and purely physical forcing of climate is an area of current research.
1.1. Sea-level change

The heat capacity of the oceans provides a thermal lag on the climate. The heat capacity depends on
the parts of the ocean that are involved: for time scales from seasonal to decadal the ocean mixed layer
is dominant, whereas for millennial time scales the interior ocean is also important and the ‘effective’ heat
capacity is larger, perhaps 50 times that of the mixed layer alone (Wigley and Raper, 1991). This dependence
of effective heat capacity on time scale means that the ocean response (and hence climate lag) is not a
simple exponential approach towards a new equilibrium climate, but instead shows a more rapid response
over the initial few decades of greenhouse gas forcing increase, followed by a more gradual approach to final
equilibrium that may take many centuries (e.g. Senior and Mitchell, 2000).
This oceanic lag on climate change, due to the heat uptake by the oceans, results in the expansion of
the oceans. The thermal expansion component of sea-level rise, at the global scale over the 21st century, is
estimated to be a significant contribution of the total change, compared with the melting of land ice (glaciers
and ice sheets). Under one typical scenario, Church et al. (2001) estimate the thermal expansion over 1990
to 2100 to be in the range 0.1 to 0.4 m, compared with the simulated total sea-level rise of 0.1 to 0.8 m. The
oceans are clearly important, through thermal expansion, for future sea-level changes and the slow approach to
equilibrium discussed above has two important (and related) implications: (i) thermal expansion of the oceans
will continue over many centuries; and (ii) mitigation of climate change through reduction of greenhouse gas
emissions will have less influence on reducing sea-level rise than it will on reducing climate warming. Thus,
21st century climate warming will probably (Church et al., 2001; Raper et al., 2001) result in a commitment
to an equilibrium sea-level rise (due to thermal expansion of the ocean) that is about five times greater than the
rise that is actually realized during this century. This would dominate over the contribution from melting of
glaciers (which may decline as glacier volume decreases), though melting of the Greenland ice sheet (which
is expected to contribute little during the 21st century) may greatly increase if a warmer climate is sustained,
perhaps contributing a much greater 3 to 6 m of sea level rise over the coming millennium (Huybrechts and
De Wolde, 1999).
The ocean’s role in temperature-related sea-level rise is fairly passive, but a dynamical ocean response
is important for regional sea-level changes and may also modify the pattern of climate change. Patterns
of sea-level rise are, however, quite dissimilar between different climate models (Gregory et al., 2001),
indicating that the uncertainty surrounding regional sea-level changes is even greater than at the global
Copyright  2003 Royal Meteorological Society Int. J. Climatol. 23: 1127–1159 (2003)
ROLE OF OCEANS IN CLIMATE 1129
scale. Spatial structures that are semi-robust across models are: (i) above-average Arctic sea-level rise
attributed to an Arctic freshening due to enhanced runoff and precipitation, with a balancing dynamic
response; (ii) a Southern Ocean minimum poleward of 60 ° S, which must require a dynamical explanation,
through upwelling of North Atlantic deep water (NADW), because it is a region of significant oceanic
heat uptake in a warming climate; and (iii) a reduced sea-level rise south of the Gulf Stream and
elevated rise to the north, consistent with a weakening of this surface component of the Atlantic THC
(Section 4.5).
1.2. Time scales and variability

Climate variations can be either a response to external forcing of the climate system or generated
internally within the system. It is useful to define two types of internally generated variability. Passive
variability involves the modulation by ‘slower’ components of the climate system (such as the oceans)
of (typically random) variability in the ‘faster’ components (such as the atmosphere). Active variability is
caused by variations and instabilities of the dynamics of the climate system, and by coupled interaction
between components of the climate system (such as the atmosphere and ocean). Given the random evolution
of passive variability, not relying on particular sequences or patterns of climate variation, this type is
often called stochastic climate variability (e.g. Hasselmann, 1976). A strong dynamical response in the
slower component (the oceans) is not a requirement for passive variability, but it is for active variability;
hence the latter type can be called dynamic climate variability. The ENSO phenomenon is the most
striking example of active climate variability, at least in the relatively short instrumental climate record.
The oceans play a role in all three situations (externally forced, passive/stochastic, and active/dynamic
variations).
The model of stochastic climate variability developed by Hasselmann (1976), in its simplest form, represents
the climate by the heat content of a well-mixed box (representing all or part of the oceans) that is perturbed
by random air–sea heat flux forcing (due to random atmospheric variations) and a dissipative heat flux
that acts to return the temperature of the box toward its unperturbed state. The solution to this system
is a first-order auto-regressive (AR(1)) process, which has a characteristic ‘red’ spectrum of variability.
A red spectrum has greater spectral power (variance per unit frequency band) at low frequencies than at
high frequencies. The oceans, through their thermal inertia, act to damp the random atmospheric variations,
but this damping is most effective at the shorter time scales; at the longer time scales of similar order to
the response time of the ocean box, there is little damping. Thus the oceans convert the ‘white’ spectrum
of atmosphere-only variability (with equal spectral power across all time scales) into a red spectrum of
variability.
A clear demonstration of the importance of the oceans in generating a red spectrum of variability was
give by Collins et al. (2001) using a 1000 year simulation with the HadCM3 coupled climate model. Their
figure 4, reproduced here (Figure 1), shows similar spectral power at intra- and inter-decadal time scales
(implying a white spectrum) over almost the entire land surface, indicated by the unshaded values (between
−0.2 and +0.2). In contrast, the inter-decadal variability is greater than the intra-decadal variability over
almost all the oceans, indicated by the dark shading (>0.2). Although this example highlights the key role
of the oceans, it also implies that the enhanced multi-decadal variability does not extend strongly over
neighbouring land areas (also note the low ratio over the equatorial Pacific, due to the high interannual
variance associated with the simulated ENSO variability exceeding the spectral power at longer time
scales).
Hasselmann (1976) and Mitchell (1976) highlighted the importance of stochastic climate variability at a
time when many scientists were attempting to attribute decadal climate variations to various external forcings
or to strong interactions between atmosphere and oceans. Their insight remains valid today, with the red
spectrum associated with the AR(1) model of passive internally generated noise able to explain much of
the variability of climate on annual to multi-decadal time scales. We give an example applicable to just one
time scale. Figure 2 compares the simulated variance for the Hadley Centre second-generation coupled model
(HadCM2), at a time scale of 20 years, with that expected from AR(1) processes fitted to the model output.
HADCM3-CTL LOG10((20 – 50 YEARS) / (2 – 8 YEARS))

90N
45N
45S
90S
180 90W 0 90E
Figure 1. The ‘redness’ of internally generated 1.5 m air temperature variability in the HadCM3 climate model. The plotted values are
the log10 of the ratio of spectral variance in the 20–50 year band to the spectral variance in the 2–8 year band (isoline interval 0.2,
dark shading (red spectrum) > + 0.2, light shading (blue spectrum) < − 0.2). (Reproduced from Collins et al. (2001) with permission
from Springer-Verlag)
Even with a very large statistical sample (here we use a 1400 year control run), the simulated variance only
exceeds, with statistical significance, that expected from passive variability over parts of the Southern, North
Atlantic and Arctic Oceans. These regions are associated with a small residual climate drift (in the range
0.5 to 1.0 ° C over the 1400 year simulation), and the statistical significance of the enhanced variance to the
southwest of Australia and in the Arctic Ocean is much reduced if the time series are first de-trended. Thus,
only the North Atlantic and part of the Southern Ocean in HadCM2’s control run show variability that cannot
be explained by stochastic processes.
The analysis shown in Figure 2, though, directly applicable only to one climate model and one time scale,
indicates that the role of active variability may be small overall — though where it is significant it certainly
involves ocean–atmosphere interaction at these time scales (Figure 1). Two-way interaction (or coupling),
with the atmosphere responding to ocean variations as well as vice versa, is especially important for generating
modes of enhanced variability and provides the potential for climate predictability on seasonal to decadal
time scales. Given the importance of any seasonal–decadal predictability, the search for true coupling has
been vigorous. The difficulty of separating cause and effect in the real world is hampered by the lack of
accurate records, with good spatio-temporal coverage, of variables (particularly air–sea fluxes) that could be
used (e.g. Cayan, 1992) to distinguish the direction of influence (atmosphere to ocean, or vice versa). Without
these it is not possible to rely solely on empirical studies of observed variability, and model simulations
are required to investigate the existence of coupled fluctuations. In the tropics the coupling is strong and
the internal atmospheric variability is weak and it is possible to identify the ENSO phenomenon clearly
in the Pacific Ocean (see Diaz et al. (2001) for a review). In the other oceans, the signal-to-noise ratio of
coupled variability is weaker and hinders the detection of phenomena involving local air–sea interaction.
Even in the tropics, where the interaction is strong, external influences (ENSO and extra-tropical variability)
can obscure the identification of local modes of variability in the Atlantic (Marshall et al., 2001) and Indian
Oceans (Goddard et al., 2001), whereas in the extra-tropics, the internal atmospheric variability is high and
makes the identification of the influence of the ocean rather difficult. The use of models has not yet clarified
the situation, because the atmospheric response to mid-latitude sea-surface temperature (SST) anomalies is
20-year variance / AR (1) variance

90N
60N
30N
EQ
30S
60S
90S
60E 90E 120E 150E 180 150W 120W 90W 60W 30W 0 30E 60E
Figure 2. Ratio of the variance of 20 year mean temperatures (simulated during a 1400 year control simulation of the HadCM2 climate
model) to that expected from an AR(1) process (fitted to the HadCM2 simulation). Grey shading indicates the areas where the simulated
variance is greater than that expected from an AR(1) process, though only those locations indicated by symbols are significantly greater
(with 95% () or 99% ( ) confidence)
dependent on model and atmospheric state (e.g. Kushnir and Held, 1996; Peng et al., 1997; Rodwell and
Folland, 2002).
2. BIOGEOCHEMICAL INTERACTION
The interactions between the ocean and the atmosphere are major regulators of atmospheric composition,
and hence of climate. It is also clear that human activity will produce changes in the chemistry of the
atmosphere, and hence in the Earth’s climate, over forthcoming decades (Houghton et al., 2001). Here, we
will briefly review the main modes of ocean–atmosphere chemical interaction that affect climate and also
consider some of the palaeo-record evidence for the linkages between atmospheric composition and climate.
We begin with a consideration of the air–sea exchange of the long-lived radiatively important gases, viz.
carbon dioxide (CO2 ), methane (CH4 ) and nitrous oxide (N2 O). Next, the air–sea exchange of more reactive
gases is considered. These influence atmospheric gas-phase chemistry significantly, particularly in relation
to ozone regulation, and yield important amounts of secondary aerosol that influence cloud formation and
albedo. We then consider the role of primary aerosols in the marine atmosphere. Sea spray emitted from the
oceans will be considered first. This has important light-scattering properties in the atmosphere and provides
a reaction site for several important chemical reactions. The oceanic deposition of aerosols produced on land
will then be considered. These can alter biological activity (primary productivity) in the oceans, thereby
changing the air–sea exchange of climatically important gases. Finally, we demonstrate some of the complex
interactions and feedbacks between these exchanges that may afford a measure of climate regulation. Although
water itself is important in climate regulation as a greenhouse gas, it will not be considered here. Since many
modes of air–sea interaction involve gas exchange, the characteristics of that process itself are considered
first before moving on to discuss key chemical species.
2.1. Gas exchange

Gas exchange rates between the atmosphere and the oceans depend on the concentration difference between
the two reservoirs and on the exchange rates (Liss and Merlivat, 1986; Liss et al., 1997; Nightingale et al.,
2000). In the case of some reactive gases emitted from the ocean and rapidly degraded in the atmosphere,
atmospheric concentrations are effectively zero. In other cases, such as CO2 , the system is close to equilibrium,
with concentrations in both the atmosphere and surface ocean important in controlling fluxes. In the case of
gases such as the chlorofluorocarbons (CFCs) and their replacements that are only emitted to the atmosphere,
the overall direction of the flux is from the atmosphere to the oceans at present. In the future, emissions from
the ocean may become significant as the atmospheric emissions are phased out. For some gases produced in
the oceans, such as the important sulphur-containing gas dimethyl sulphide (DMS), the surface waters are
supersaturated and the overall direction of flux is from the ocean to the atmosphere. Rates of gas exchange
depend on physical factors at the interface, such as wind and wave effects, temperature and possibly the
presence of surfactants. Parameterizations of the effect of wind on gas exchange rates are available, but
these still have rather large uncertainties and contribute significantly to overall uncertainties on gas exchange
80
70
60
50
k600 (cm hr-1)
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20
Wind speed (m s-1)
Figure 3. Tracer data plotted against wind speed at 10 m. Solid circles represent data from four tracer experiments in the North Sea. The
solid squares are the Georges Bank dual tracer data of Asher and Wanninkhof (1998). The solid triangle is an estimate from the Florida
Shelf (Wanninkhof et al., 1997). The thin solid line represents the Liss and Merlivat (1986) relation, the short dashed line represents the
relation of Wanninkhof (1992), and the longer dashed line represents the relation of Smethie et al. (1985). The thick dark line represents
a quadratic deconvolved fit to the North Sea data. (Diagram courtesy of P. Nightingale)
rates (Figure 3). Recent evidence suggests that there may be an upper limit to the exchange coefficient
at hurricane-force wind speeds, as the sea surface becomes covered with a layer of foam (Powell et al.,
2003).
2.2. Carbon dioxide

The dominant greenhouse gas, apart from water vapour, is CO2 . There is a vast natural cycle in which
carbon is cycled between the land surface, oceans, atmosphere and lithosphere. It is now well established
that CO2 concentrations have changed naturally over glacial–interglacial cycles and that they have increased
over the last 100 years or so as a result of human activity (see figure 3.2 of Houghton et al. (2001)). It is
also clear that the changes in CO2 concentrations are closely linked to and influence climate itself. Increases
in CO2 concentration arising from human activities represent a significant perturbation of this cycle (see
figure 3.1a and b of Houghton et al. (2001)). Much of the CO2 emitted by human activity does not stay in the
atmosphere, but rather has been stored (at least temporarily) in the oceans or on the land surface (vegetation
and soils). It has been estimated, for example, that about a quarter of emitted fossil fuel CO2 is now stored
in the oceans (Feely et al., 2001). The exchange of CO2 between the oceans and the atmosphere has both a
physical and a biological component; though the two are linked, however, since biological processes respond
to physical forcing (Watson and Liss, 1998; Feely et al., 2001; Zeebe and Wolf-Gladrow, 2001; Watson and
Orr, in press).
2.2.1. Physical exchange. CO2 is more soluble in seawater at lower temperatures. Ocean circulation carries
cold surface water with a high CO2 content into the deep ocean. This water eventually mixes or upwells to
the surface again, warms and in the process loses some of its CO2 to the atmosphere. This simple solubility-
driven system is complicated by the chemistry of seawater, which contains bicarbonate, and carbonate ions,
which react with CO2 via the reaction
−
CO2 + H2 O + CO3 2− −
−
−−
−
− 2HCO3 (1)
The net effect is to increase the solubility of CO2 in seawater considerably over what it would be in the
absence of such reactions. It is clear from basic chemical principles that increasing CO2 concentrations will
drive this reaction to the right, and thereby increase the dissolution of CO2 , and in this sense the ocean acts as
a large sink for anthropogenic CO2 . However, as this reaction proceeds it tends to lower the pH of seawater
slightly, and this will act to shift the equilibrium for the reaction converting CO3 2− into HCO3 − and thereby
reduce the capacity for further CO2 uptake. In the future, increasing water temperatures will also tend to
reduce the capacity of the oceans to take up CO2 .
Current estimates (Takahashi et al., 1997) suggest that the major oceanographic sinks for CO2 are in the
North Atlantic, particularly the Norwegian Sea, and in the Southern Ocean. However, the limited nature of
regular measurements of CO2 concentrations throughout most of the world oceans, coupled with uncertainties
over the exchange rate, means that there is currently an uncertainty of about a factor of two in calculations
of the air–sea exchange of CO2 , and particularly that of anthropogenic CO2 .
2.2.2. Biological exchange. In addition to this physical sink for CO2 in the oceans, CO2 is also taken up
during primary production in the surface of the oceans and converted into plant biomass. This is analogous
to the growth of plants on land, although in the oceans the process is conducted by microscopic algae
(phytoplankton) that have a very short life cycle compared with terrestrial plants. Overall, the rates of CO2
uptake by plants on land and in the oceans are similar, but the turnover rates in the surface ocean are very
much faster. Most of the CO2 taken up by phytoplankton is rapidly returned to the water and the atmosphere
when the phytoplankton die or are eaten. However, some is lost to the deep sea and sediments as sinking
particles, where it can remain for long periods of time relative to the rate of recent increase in CO2 from
human activity. This loss to the deep sea depends on the ecology of the phytoplankton involved. In general, the
phytoplankton community in high-latitude regions is capable of exporting a significant proportion of carbon
to deep water. By contrast, the phytoplankton in tropical waters are primarily sustained by rapid and intense
recycling within the surface layers with rather little loss to deep waters. Although these generalities are widely
accepted, the detailed factors regulating primary productivity in the ocean, and the way these will respond
to global change, are too poorly known to allow them to be included in global climate models. Since ocean
primary production is not limited by CO2 supply, however, most models assume that primary productivity
and associated net CO2 uptake has not changed due to anthropogenic perturbations. This assumption may not
be valid, as discussed below, since human activity (or natural long-term geochemical cycles) can alter total
primary productivity and also change phytoplankton species composition.
A further ecological issue centres on phytoplankton that manufacture a calcium carbonate (CaCO3 ) skeleton,
particularly the very abundant Emiliania huxleyi. In making their skeleta these algae will release CO2 according
to the reaction
Ca2+ + 2HCO3 − −
−
−−
−
− CaCO3 + CO2 + H2 O (2)
Thus, growth of calcareous algae will involve net release of CO2 in formation of their carbonate skeletons.
The factors that promote growth of E. huxleyi, or other calcareous algae, in preference to other algal groups
that make silica or organic skeletons, are not understood. Hence, it is not currently possible to predict changes
in the abundance of these species in the future.
The IPCC reviewed several models that predict future trends in CO2 uptake by the oceans (Houghton et al.,
2001). In all cases the physical processes of CO2 uptake were well represented, but biological processes and
the effects of pH changes were not so well represented. All models predict the oceans will continue as a net
sink for anthropogenic CO2 throughout the 21st century. However, the limitations of these models, particularly
in terms of their representation of biological processes, means that such predictions are necessarily uncertain.
The atmospheric record of CO2 increase shows significant interannual variability. Often, but not always,
the largest atmospheric increases correlate with El Niño events. However, even if the correlation were perfect,
it is still difficult to separate the roles of land and ocean uptake in their impacts on the observed atmospheric
increases.
We now consider some of the more important gases that are produced in the oceans and which have
important effects on climate once they are in the atmosphere.
2.3. Methane and nitrous oxide

The oceans are also a significant source of other important long-lived greenhouse gases, particularly CH4
and N2 O (Nevison et al., 1995; Bates et al., 1996; Lelieveld et al., 1998; Seitzinger and Kroeze, 1998; Upstill-
Goddard et al., 2000). Both gases are produced biologically within the oceans. Higher productivity coastal and
upwelling systems are the primary source. Atmospheric concentrations of both CH4 and N2 O have increased
in recent times, probably as a result of human activity (see figures 4.1 and 4.2 of Houghton et al. (2001)).
The detailed biological mechanisms for the production of these gases in the ocean are not particularly well
understood, and hence it is not possible to predict likely changes into the future, or indeed adequately to
explain changes seen in the past. Production of both gases takes place associated with chemically reducing
(low oxygen) conditions in sediments or the water column, though the mechanism of production of each is
different. It is possible that increasing inputs of nutrients, such as nitrate, into coastal zones have increased
productivity, thereby increasing the extent of reducing environments and leading to increased emissions of
these compounds (Naqvi et al., 2000). Increasing human populations in the future are likely to intensify such
pollution pressures on coastal zones. Such increases in productivity may also increase CO2 uptake by coastal
waters and will also change the pattern of emissions of sulphur gases, discussed below (Jickells, 1998).
There are also substantial stores of methane hydrates in the oceans; destabilization of these could result in
significant emissions to the atmosphere (Gornitz and Fung, 1994). This scenario is more probable for coastal
methane hydrates. In a recent striking demonstration of this, a trawler dredged up 100 kg of methane hydrate
while fishing off Vancouver Island, Canada (Spence and Chapman, 2001).
2.4. Sulphur gases

One of the most important of the short-lived gases emitted from the oceans is DMS (Liss and Galloway,
1993; Andreae and Crutzen, 1997; Watson and Liss, 1998). This gas is produced by phytoplankton, probably
as part of their regulation of internal cell ionic strength, but the amounts released by different species vary
markedly. Furthermore, it is now clear that there is intense cycling of DMS and its precursor compounds in
the surface ocean, processes that can substantially influence DMS fluxes to the atmosphere. DMS emissions
are quantitatively large in the context of the natural sulphur cycle, and in the remote atmosphere, marine
emissions presently dominate the natural cycle. Anthropogenic sulphur emissions (mainly as SO2 from
combustion sources) are now comparable in size to natural emissions, though their geographic distribution is
quite different (Penner et al., 2001). Anthropogenic emissions are centred on industrialized regions of North
America, Europe and Asia; the DMS emissions come from all ocean areas, although they are particularly
strong from some regions where certain algae that are major DMS emitters dominate (Kettle et al., 1999).
The northeastern Atlantic Ocean is one such region of strong emissions, since there are regular blooms of
coccolithophores in this area.
DMS breaks down rapidly in the atmosphere on a time scale of a day or so, depending on atmospheric
chemical conditions (Smet et al., 1998), to yield primarily sulphuric acid. This sulphuric acid then forms
aerosols in its own right, or by reacting with ammonia (NH3 ) gas (which is also emitted from the oceans)
to form ammonium sulphate salts. In the remote atmosphere, sulphuric acid and ammonium sulphate salts
dominate the fine mode aerosol and can influence production of cloud condensation nuclei (CCN). On this
basis, it has been proposed that DMS emissions could form part of a climate regulation system in which
phytoplankton emissions of DMS alter CCN abundance and thereby influence cloud albedo, which in turn
influences phytoplankton growth to regulate climate (Figure 4), the so-called ‘CLAW’ hypothesis (Charlson
et al., 1987).
Research (Ayers et al., 1991; Boers et al., 1994) has demonstrated that increased emissions of DMS can
influence CCN abundance, though even this linkage is complex. It is also now clear that much of the sulphuric
acid formed from DMS oxidation does not form new aerosol particles; rather, it is oxidized on, or rapidly
adheres to, other particles, such as sea salt and dust. However, despite this, DMS emissions can influence CCN
formation by promoting the growth of smaller particles and making them more effective CCN (Andreae et al.,
1999). We also now know from ice-core records (Saltzman et al., 1997) that sulphate aerosol concentrations
Cloud formation Increased Increased cloud

concentration albedo
of cloud droplets
Cloud
condensation
nuclei
Loss of solar
radiation
Particle
formation
2− Lower Less
SO4
surface near-surface
temperature solar flux
Oxidation
DMS gas
Atmosphere
Ocean
DMS Biological
(in solution) production
of DMS (±?)
Figure 4. The CLAW hypothesis illustrating potential climate regulatory feedbacks associated with marine sulphate emissions
have changed in phase with climate cycles over glacial–interglacial time scales. The linkages between cloud
albedo, radiation levels, phytoplankton growth and DMS emissions, which complete this proposed cycle, are
currently not established. These interactions cannot, therefore, be adequately represented in global climate
models. However, preliminary representations do suggest that the kind of climate feedbacks discussed can
change the global temperature by 1 ° C (Turner et al., 1996). As discussed below, it is likely that the real
climate feedback linkages between the atmosphere and the oceans are, in fact, more complex than suggested
in the CLAW hypothesis, though the role of sulphate aerosol is clearly important in climate control.
Another climatically important sulphur gas emitted from the oceans is carbonyl sulphide (COS), which is
produced primarily in the coastal zone from the photolysis of dissolved organic material. Production rates for
this gas are lower than for DMS, but it has a rather long atmospheric lifetime (4 years) and hence exerts an
important influence on stratospheric aerosol, since it too is oxidized to sulphuric acid (Andreae and Crutzen,
1997; Uher and Andreae, 1997).
2.5. Halogen and non-methane hydrocarbons

It has been demonstrated over recent years that the oceans are a source of a wide variety of other gases to
the atmosphere. Two important groups in terms of climate regulation are non-methane hydrocarbons NMHCs;
(Broadgate et al., 1997) and halocarbons (Carpenter et al., 1999). The former react in the atmosphere with
oxidizing species, such as OH radicals, and thereby impact the oxidizing capacity of the atmosphere (and
hence the removal of methane from the atmosphere — see Osborn and Wigley (1994) for an estimate of this
indirect climatic influence) and, in turn, could influence ozone cycling. Although concentrations are lower,
their reactivity is in many cases higher than that of methane. Halocarbons include a range of low molecular
weight gases containing halogens, such as chlorine, bromine and iodine; examples include methyl iodide and
bromoform. All of these gases react in the atmosphere, releasing halogen atoms that become involved in
catalytic ozone-destruction cycles. Depending on the reactivity of the gases, this may impact tropospheric or
stratospheric ozone. In either case, since ozone is a greenhouse gas, these emissions will influence climate,
particularly in remote areas where anthropogenic impacts are least. They also influence the oxidizing capacity
of the atmosphere, and hence the methane and NMHC concentrations. Marine halocarbon emissions may also
impact aerosol formation (O’Dowd et al., 2002).
The biological mechanisms controlling the production of these gases are poorly understood and global fluxes
are uncertain. However, it is clear that the production and cycling processes for these gases are different from
one another and different from those of the sulphur gases, suggesting that incorporating the cycling of these
gases into global models using a simple relation to biological productivity is going to be very challenging
(Baker et al., 2000).
2.6. Aerosols
2.6.1. Sea salt aerosol production. Sea salt is emitted from the oceans to the atmosphere by processes of
bubble bursting and wave breaking. The production of sea salt is usually described as an exponential function
of wind speed, though the exact relationship is not particularly well characterized (O’Dowd et al., 1997; Rae
et al., 2000). Total production is estimated at 5900 Tg year−1 (1 Tg = 1012 g). The material produced may
be modified from bulk seawater composition, since bubbles will escape via the sea surface microlayer, a
region known to be enriched in organic matter. It is now known that many of the aerosol particles in the
atmosphere contain organic matter, but its source and impact on aerosol properties are major uncertainties
that will challenge atmospheric chemists for many years to come (Jacobson et al., 2000).
The sea salt formed represents a major source of aerosols over the ocean, and hence an important source
of light scattering and CCN. The lifetime of a sea-salt aerosol is a matter of days, depending on its size,
i.e. much less than the mixing time of the atmosphere. Sea-salt production rates will vary widely across
the oceans, depending on wind speed and whitecap formation, so the effects on light scattering and CCN
formation will vary markedly across the oceans. In addition, it is now clear that the sea-salt aerosol acts as
a very important site for a variety of chemical reactions, including DMS oxidation (see above) and reactions
with acids (see below). The latter can release halogens from the aerosol, and thereby have an important effect
on ozone cycling
2.6.2. Aerosol deposition to the oceans. Primary productivity in the ocean is limited by a variety of factors
in different places, particularly light and the supply of key nutrients, including nitrogen and iron (Falkowski
et al., 1998). It is estimated that the atmospheric supply of nitrogen to the oceans is comparable to that from
rivers; the main external source of iron to the oceans is from atmospheric dust deposition. Hence, atmospheric
deposition can affect ocean productivity (Jickells, 2002). Most primary productivity in the oceans is driven
by the large-scale internal cycling of nutrients, hence the assumption noted earlier, that primary production
and biological uptake of CO2 by the oceans has not changed as a result of human activity. As noted above,
this is certainly not true for the coastal zone, because of anthropogenic inputs. It may also not be true for the
open ocean if the changes in external nutrient supply are large enough to change ocean productivity.
The atmospheric nitrogen input to the oceans is estimated at 200–275 Tg year−1 (Jickells, 2002). This has
increased substantially as a result of human activity and has large spatial gradients decreasing away from
major source regions in developed countries. This pattern of distribution is expected to change substantially in
the future, with increased industrial activity in Asia in particular (Galloway and Cowling, 2002). Atmospheric
nitrogen inputs to the ocean are of sufficient magnitude that it may enhance the uptake of CO2 and emissions
of biogenic trace gases such as DMS by the oceans (Jickells, 2002). It has also been argued that short-term
pulses of high nitrate and ammonium deposition, which can occur in coastal systems, may be of sufficient
magnitude to perturb the ecosystem. Such perturbations may change the species composition and hence the
emission of trace gases. Coastal zones are particularly vulnerable in this regard because they are close to
large-scale NOx and NH3 emission sources. In addition, atmospheric reactions taking place when polluted
continental air meets marine air, particularly involving gas-phase nitric acid and sea-salt aerosol, can act to
increase deposition rates and thereby focus atmospheric deposition into coastal areas (Jickells, 1998). In a
corollary of this process, Rosenfeldt et al. (2002) have argued that sea-spray emissions from coastal waters
can act to overwhelm the suppression of cloud and precipitation formation arising from fine aerosol particle
emissions associated with combustion sources. Sea spray thereby promotes precipitation and acts to clean the
atmosphere.
It has recently become clear that the availability of iron limits primary productivity in several marine areas,
particularly the Southern Ocean (Falkowski et al., 1998). The main external source of iron to the oceans is
from atmospheric dust, which is primarily derived from desert regions (Figure 5). Total global dust fluxes
Sep., Oct., Nov
EAOT
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
Husar Stowe and Prospero,1996
Figure 5. Satellite figure showing global aerosol distribution. Note strong signals associated with dust transport from North African and
Asian deserts
Figure 6. (a) Chlorophyll measurements during the SOIREE experiment in the Southern Ocean showing dramatic increases in chlorophyll
in the iron-fertilized area (open circles) compared with the unfertilized area (dark circles) (Boyd et al., 2000); (b) DMS measurements
during the SOIREE experiment in the Southern Ocean showing dramatic increases in DMS in the iron-fertilized area (open circles)
compared with the unfertilized area (dark circles). (Courtesy of S.M. Turner)
are estimated to be 400–1000 Tg year−1 , though this varies substantially from year to year depending on
aridity and climatic changes in transport pathways (Jickells and Spokes, 2001). The Southern Ocean is most
remote from desert regions. Hence, the phytoplankton living there are most vulnerable to such iron stress. It
has been demonstrated (Boyd et al., 2000) that addition of iron increases primary productivity, CO2 uptake
by the surface waters and DMS emissions (Figure 6). It is clear that, over glacial–interglacial cycles (see
Section 4), the inputs of dust to the oceans have varied dramatically, associated with both changes in aridity
and in wind strength. It has been argued that increased dust input to the Southern Ocean during the last
glacial maximum might have been sufficient to increase productivity enough to contribute to the decrease in
CO2 seen in ice-core records, though this conclusion is controversial (Watson, 2001). Human activity may
have already significantly modified dust fluxes and may further alter them in the future (Tegen et al., 1996;
Ridgwell et al., 2002).
Iron may also influence primary productivity in other ocean areas by another mechanism. Most phyto-
plankton groups cannot directly utilize the abundant supply of dissolved N2 gas. The biological fixation
of N2 requires large amounts of energy and specialized physiological adaptation. The enzymes involved
also require rather large amounts of iron. It has been suggested, therefore, that in tropical ocean areas
where high light and temperature levels favour nitrogen-fixing phytoplankton, iron supply may influence
nitrogen fixation rates. It is then suggested that increased nitrogen fixation rates would relieve nitrogen
stress throughout the community, thereby increasing productivity (Falkowski et al., 1998; Berman-Frank
et al., 2001). This increase would again favour specific algal groups, and hence would selectively impact
the emissions of the trace gases considered above. The recognition of the key role of iron in atmospheric
and oceanic biogeochemical cycles is recent. It has focused a lot of research into the cycling of iron in
the oceans and the effect of iron availability on phytoplankton populations (Turner and Hunter, 2001).
This research is still in its infancy, but it is clear that iron stress affects different phytoplankton groups
to varying degrees, and hence may alter trace gas emission rates selectively. One key issue revealed by
this research is the question of the solubility of iron from atmospheric dust (Jickells and Spokes, 2001).
This is believed to be very low (probably a few percent) and controlled by various factors including aerosol
pH. Since pH depends on factors including DMS emission and subsequent conversion to sulphuric acid,
there is the potential for linkages to the CLAW hypothesis discussed earlier. Iron solubility also depends on
photochemical processes in the atmosphere and surface water and on the complexation of iron by organic
ligands. Both these factors offer the opportunity for complex linkages and feedbacks between the various
aspects of ocean–atmosphere chemistry discussed here and changes associated with global change in the
future.
2.7. Interlinking cycle
A key conclusion from this brief review is that atmosphere–ocean biogeochemical interactions are
demonstrably important in regulating atmospheric chemistry, including radiatively active compounds, and
hence climate. It is clear that these cycles are interlinked and so may allow feedbacks for the regula-
tion of climate cycles. Some of these are illustrated in Figure 7, though it is important to note that not
all the potential interactions described here are included, to make the diagram easier to follow. There
may also be important interactions we have yet to identify. Although the complexity of the possible
linkages presents a daunting challenge to biogeochemists, the complexity also offers the possibility of
rather stable climate regulation by these feedbacks. It is clear that our knowledge of these interactions
is currently inadequate to provide quantitative inputs to global climate models. The new IGBP SOLAS
(Surface Ocean — Lower Atmosphere Study) project aims specifically to improve our understanding of
these interactions. Further information can be found on the SOLAS Website at http://www.uea.ac.uk/env/
solas.
Ozone
DUST
FeIIIs FeIId+OH
Cl SO2+OH H2SO4
Sulphate Aerosol
Ammonium Sulphate Radiation NH4+
Sea Budget
Salt Fe
SO2 NH3
DMS
NH3/NH4+
Nitrate
Algae
DMS Degradation/Loss
Figure 7. Combined feedback diagram for some chemical interactions associated with air–sea biogeochemical exchange
3. ATMOSPHERE–OCEAN INTERACTION INVOLVING THE WIND-DRIVEN CIRCULATION
3.1. Tropical air–sea coupling in the Atlantic and Indian Oceans

El Niño is the most well-known form of air–sea interaction within the tropics, and is covered by Diaz et al.
(2001). Until recently, it was supposed that climate variability in the Pacific controlled much of the variability
in other tropical ocean basins (e.g. Latif and Barnett, 1995), and, as seen earlier, stochastic variability is also
important at longer time scales. Recently, however, coupled processes internal to both the Atlantic and Indian
Oceans regions have been identified. The Pacific is clearly important for both sub-annual (Saravanan and
Chang, 2000; Camberlin et al., 2001) and decadal (Latif, 2001) variability of the tropical Atlantic. Here, our
focus is on coupled interactions mainly confined to the Atlantic. Within the last decade of the 20th century
the seemingly strong link between the Indian monsoon (and hence Indian Ocean) and El Niño has apparently
broken down (Saji et al., 1999). Here, we will also examine processes within the Indian region linked to this
change of behaviour of the climate system.
The tropical Atlantic is now recognized as having two separate modes of internal variability: a dipole mode,
with SST and sea-level pressure (SLP) anomalies of opposite sign across the equator, and an equatorial mode,
with SST anomalies centred on the eastern equatorial Atlantic (Figure 8; see also Handoh et al. (submitted)).
The latter (Latif and Grotzner, 2000) is analogous to the El Niño mode in the Pacific. There is sometimes
thought to be a correlation between the two modes, but this is not robust (Murtugudde et al., 2001) and
is probably a reflection of the geometry of the Atlantic basin, with the bulge of West Africa potentially
constraining equatorial anomalies to extend south of the equator but less readily north.
The degree of coupling between air and sea in the dipole mode is hotly debated. At the time of preparation
of the IPCC TAR it was believed that the SST and SLP anomalies of the dipole mode led to the wind
field’s enhancement of evaporation in the region of the cold SST anomaly and suppression of evaporation
above the warm anomaly. This would reinforce the original pattern, and so lead to its persistence, even to
decadal time scales (Carton et al., 1996; Chang et al., 1997). Atmospheric general circulation model (GCM)
results, forced with a dipole SST field, reinforced this hypothesis (Okumura et al., 2001) and suggested a link
to the North Atlantic oscillation (NAO). A new positive feedback mechanism was identified in that study,
involving evaporative enhancement of low-level sub-tropical stratiform clouds, with the consequent reflective
loss of solar radiation and so of further cooling. However, both observational and modelling studies allowing
coupling between the ocean and atmosphere suggest that the dipole mode is not as strongly coupled as once
thought. Saravanan and Chang (2000), Servain et al. (2000) and Chang et al. (2001) highlight the importance
of oceanic processes in the coupling, and the latter suggest that oceanic advection acts as a negative feedback
to the atmospherically induced positive feedback mechanism discussed above. The atmospheric modelling
study of Sutton et al. (2000) also casts doubt on coupling being responsible for the dipole mode. Several
authors have suggested that air–sea coupling is too weak to sustain an oscillation in the dipole mode, and that
stochastic forcing, perhaps linked to the NAO, is required to force the Atlantic basin into this mode (Enfield
et al., 1999; Chang et al., 2001; Li, 2001).
The equatorial mode has received less attention, but appears a better candidate for exhibiting distinct and
persistent air–sea coupling. In a relatively simple coupled model of the tropical Atlantic, Zebiak (1993)
showed the potential for an oscillatory mode, with an equatorial maximum, to be set up and to persist for a
number of cycles before decaying. This mode was entirely independent of the Pacific. It had an annual period,
being linked to perturbations of the annual cycle. The coupling between atmosphere and ocean was evident
through coordinated westward movement of anomalies. This can be explained as follows: oceanic heating
(cooling) leads to enhancement (suppression) of atmospheric convection, resulting in enhanced (reduced)
atmospheric inflow from the west counteracting (enhancing) the prevailing easterly winds and decreasing
(increasing) evaporation. This causes the warm (cold) SST anomaly to move westwards, with a speed similar
to, but less than, the prevailing upper-ocean current near the equator. Once the anomaly reaches the western
Atlantic and moves into the atmospheric convective cell over South America the anomaly in the (trailing)
wind field can induce a Kelvin wave response in the central-west equatorial Atlantic Ocean that is opposite
in effect to the just ended coupled anomaly. The eastward-travelling Kelvin wave generates an oppositely
Figure 8. (a) Dipole mode (empirical orthogonal function (EOF) mode 2: explained variance of 20%) of tropical Atlantic SST anomalies.
A 12 month Butterworth low-pass filter (order 12) was applied to the GISST monthly SST data (1948–99), prior to EOF analysis. The
seasonal cycle is removed. Contour interval is 0.1 ° C. Zero contours are omitted. (b) Same as (a), but for monopole mode (EOF mode
1: explained variance of 28%). (Figure courtesy of I. Handoh)
signed anomaly in the upper eastern Atlantic Ocean a few months later (Figure 9). The Sutton et al. (2000)
modelling study supports the coupled nature of this equatorial mode.
This mode has been found in observations (Servain et al., 2000; Handoh and Bigg, 2000; Handoh et al.,
submitted), with the most recent oscillation between a warm and cold phase being in 1996 to 1997 (Handoh
and Bigg, 2000). This particular occurrence shows why a correlation between the two modes is sometimes
found, as there was a strong South Atlantic anomaly within the atmospheric and oceanic fields feeding into
the equatorial signal, but practically no effect north of the latitude of the Gulf of Guinea (Handoh and Bigg,
2000). There was thus not a real mixing of modes in this event, but a conventional EOF analysis gives
significant strength to both modes. Handoh and Bigg’s (2000) study found some evidence that the Atlantic
has exhibited longer oscillations, as suggested by Zebiak’s (1993) modelling results, e.g. in the late 1960s,
but in general the influence of other effects, most notably from the Pacific, prevents this.
Large-scale modes of ocean–atmosphere interaction are also found in the tropical Indian Ocean. During
the 1990s, the well-established link between El Niño and the Indian monsoon failed. In 1994, the southeast
tropical Indian Ocean experienced unusually cold SSTs and southeasterly winds (Behera et al., 1999). Through
Warm Phase
warming cooling
Upwelling Downwelling Upwelling

discouraged Rossby wave encouraged
Cold Phase
cooling warming
Upwelling Upwelling Upwelling

encouraged Rossby wave discouraged
Figure 9. Schematic of the Atlantic coupled mode and its feedbacks, showing the warm phase leading to a succeeding cold phase
a combination of enhanced upwelling along the eastern shore of the Indian Ocean with evaporative cooling
offshore, atmosphere convection was suppressed in the region. Anomalous moisture transports associated
with these anomalies caused enhanced convection further west and north, over the central Indian Ocean,
India and East Asia. A positive feedback maintained this state for some months, allowing a good monsoon
despite 1994 being an El Niño year. In 1997, another period of enhanced upwelling and cooling in the
eastern, equatorial, Indian Ocean gave rise to similar convective anomalies over East Africa, with westward
Rossby wave propagation in the ocean maintaining the western basin ocean surface warming and enhanced
atmospheric convection through the year (Webster et al., 1999). There is evidence that this pattern is an
important Indian Ocean coupled anomaly, independent of the state of the southern oscillation (Saji et al.,
1999).
3.2. Variability involving wind-driven circulation in the extra-tropics

The potential exists (Latif and Barnett, 1994) for active, coupled variability in the extra-tropics to be
generated by air–sea interactions related to those that operate in the tropical Pacific to produce ENSO
variability (e.g. Battisti and Hirst, 1989). Positive local feedbacks between SST and atmospheric forcing
could sustain perturbations (counteracting dissipative processes), while the atmospheric response might also
drive basin-scale ocean circulation changes that, after a delay determined by the reaction time of the ocean,
provide a negative feedback capable of removing or reversing the SST anomaly. There are two key differences
in the extra-tropics. First, the time scale of any oscillation, being governed by the response time of the ocean
circulation, will be longer (inter-decadal rather than interannual) because Rossby wave propagation across
the North Pacific or North Atlantic sub-tropics takes 10 to 20 years. Second, extra-tropical SST anomalies
have a weaker influence on the overlying atmosphere. Thus, we may, at best, observe damped oscillations
at a multi-decadal time scale that are forced by stochastic (chaotic) atmospheric variability rather than being
self-sustaining.
The ECHAM/HOPE coupled ocean–atmosphere GCM studied by Latif and Barnett (1994, 1996) has a
stronger air–sea coupling in mid latitudes than some other GCMs, strong enough to generate self-sustaining
oscillations in the absence of stochastic forcing (Xu et al., 1998). When the North Pacific sub-tropical gyre is
anomalously strong, the poleward advective heat transport of the western boundary current is enhanced, leading
to a positive SST anomaly in the Kuroshio extension region of the North Pacific. The ECHAM/HOPE model
responds with a Pacific–North American (PNA) atmospheric circulation pattern, associated with weakened
westerlies over the sub-tropical gyre. These lower the heat loss to the atmosphere and also reduce the
entrainment of cooler sub-surface water into the mixed layer; the combined effect is a positive feedback
that sustains the SST anomaly (Latif and Barnett, 1996). The weakened wind stress, however, leads to a
slowing of the sub-tropical gyre and ultimately to reversal of the SST anomaly due to reduced advective
heat transport by the gyre. The ECHAM/HOPE coupled model also exhibits similar variability in the North
Atlantic (Grotzner et al., 1998), though interactions with the heat transport of the THC complicate the picture
there (e.g. Delworth et al., 1993; Timmermann et al., 1998).
Variability in some other coupled models, with weaker air–sea coupling in mid latitudes, is more readily
explained as a response to stochastic atmospheric forcing — albeit with a structure determined by similar
governing processes to those described above. The identification of coupled gyre–atmosphere modes in the
real Atlantic and Pacific basins is limited by the relative shortness of appropriate observations, though there
is some evidence of multi-decadal Pacific (Latif and Barnett, 1994) or decadal Atlantic (Deser and Blackmon,
1993) variability with similar atmosphere and SST patterns to those simulated. Pierce (2001), however, finds
greater similarity between observations and the stochastic-type simulations.
Different mechanisms may operate in the Southern Hemisphere because the Southern Ocean is zonally
unbounded and thus the gyre-type system is not present in the same form. At higher latitudes, a coupled
ocean–ice–atmosphere mode, termed the Antarctic circumpolar wave (ACW), has been tentatively observed
(White and Peterson, 1996) and also simulated (Christoph et al., 1998). Anomalies in the coupled system
appear to propagate eastwards around the Southern Ocean with a period of 4 to 5 years, taking 8 to
10 years to complete a circuit, due to the wavenumber 2 structure of the anomalies. SST anomalies may
be simply advecting eastwards with the Antarctic Circumpolar Current, but may also generate atmospheric
SLP anomalies that reinforce the SST anomalies by modifying Ekman heat transport. Positive SST anomalies
are then associated with an in-phase retreat of the Antarctic sea-ice edge. The mode appears to be excited by
ENSO variability, via atmospheric teleconnections to the South Pacific. Indeed, it has been argued (Yuan and
Martinson, 2001) that this excitation is the dominant process, generating a standing wave mode that is more
important than the propagating part of the system.
3.3. Glacial to interglacial atmospheric circulation change and impact on the ocean
During glacial periods the atmospheric circulation, particularly over the North Atlantic, was very different
from today. The physical effect of the Laurentide and Eurasian Ice Sheets, both several kilometres high,
blocking atmospheric flow on either side of the northern Atlantic was to deflect and split the main westerly
circulation of the mid latitudes. The westerlies in the Pacific sector were split into two arms, travelling
south and north of the Laurentide Ice Sheet; over the Atlantic the recombined flow was deflected south of
the Eurasian Ice Sheet (Kageyama et al., 1999). This southerly deflection caused the oceanic polar front
and North Atlantic Drift to be further south, at latitudes around northern Iberia, and the North Equatorial
Current also to move equatorward (Vink et al., 2001). This southward orientation of the sub-tropical gyre
and atmospheric wind fields were mutually supportive: oceanic heat transport, and hence storm generation,
occurred further south, leaving much of western Europe without the atmospheric or oceanic heat sources of
today.
Nevertheless, there was some ocean heat transport into the glacial northern Atlantic. It has been known
for a decade that the glacial Norwegian Sea was, at least seasonally, ice free (Veum et al., 1992), although
during peak glaciation the supply of warm Atlantic water may have been of sub-surface origin (Bauch
et al., 2001). Both observations (Weinelt et al., 1996; Kuijpers et al., 1998) and ocean models (Wadley
et al., 2002) suggest that deep-ocean convection, and hence the endpoint of upper-ocean heat flow, occurred
in the sub-polar Atlantic south of Greenland and Iceland. Radiochemical data (Yu et al., 1996) suggest a
similar overturning strength to today. Atmospheric GCMs of the glacial climate forced by CLIMAP SSTs
simulate very cold atmospheric temperatures in the northern Atlantic, which would certainly encourage
deep convection. However, a recent coupled model simulation has produced warmer glacial North Atlantic
atmospheric temperatures and SSTs (Hewitt et al., 2001), which is more consistent with the observational
evidence for significant heat transport north of the polar front.
Some regions of the global atmosphere and ocean system experienced little difference from today during
the last glacial period, including much of the Southern Ocean sector (Matsumoto et al., 2001). This was once
thought to be true of the tropics, but widespread observational and modelling evidence have suggested that the
glacial tropical oceans were 1–2 ° C cooler than today (e.g. Lee et al., 2001; Kitoh et al., 2001). This cooling
may be due to the decrease in moisture in the free troposphere that accompanied global cooling during the last
glacial period (Seager et al., 2000). Less moisture, and less condensation in deep tropical cloud layers, would
have permitted more terrestrial radiation to escape into space. Cooler tropical temperatures, in turn, have been
modelled to produce a strong teleconnection impact on the mid-latitude circulation (Yin and Battisti, 2001),
with significant warming over the Laurentide Ice Sheet, but cooling over the Eurasian Ice Sheet. A region
of the tropics showing warming rather than cooling is the monsoon-dominated area of the Indian Ocean.
The cooler Asian continent led to a weaker summer monsoon, and hence reduced upwelling by around 60%
(Overpeck et al., 1996) in the northwestern Indian Ocean.
4. THE THERMOHALINE CIRCULATION
4.1. Deep-water formation and the conveyor belt theory

The THC of the global ocean is the major mechanism by which the ocean contributes its share of the
poleward heat transport required to counteract the Earth’s surface radiational imbalance, particularly in the
Atlantic (Stocker et al., 2001). Significant changes in THC will therefore affect the global radiation budget
through altering the atmosphere’s surface boundary conditions, and so lead to major climatic change. The
classical view of this circulation is that deep convection during the winter of either polar hemisphere produces
dense, cold, saline water that then flows equatorward and slowly upwells through the global ocean over time.
This upwelled water then returns in warm upper-ocean currents to the convection sites to replenish the cycle,
with a time scale of order 1000 years. Figure 10 shows a schematic illustration of this hypothesis. This general
picture is still current, but all components of it are undergoing significant revision (Houghton et al., 2001).
This global circulation is also an important control on nutrient circulation, and hence on biogeochemical
controls on climate.
The convection arm of the THC is driven by two mechanisms. One is buoyancy loss at the sea surface in
the open ocean caused by wintertime cooling. The surface layer becomes denser than the layer(s) below and
Figure 10. Schematic of the global THC. The broken lines represent the major components of the surface circulation. The solid lines
show the deep water emanating from regions denoted by open circles. Upwelling, particularly concentrated in the Southern Ocean and
over rough topography, closes the circuit. (Figure 1.15 of Bigg (1996), reproduced with permission of Cambridge University Press)
mixing (especially in late winter) can occur to significant depths, generally in localized plumes (see Marshall
and Schott (1999) for a review), although recent evidence suggests that sub-mesoscale eddies may be an
important mechanism in producing deeper convection (Gascard et al., 2002). The depth to which convective
mixing occurs depends on the relative densities in the water column and the integrated surface heat and
freshwater fluxes over the winter. There are a number of regions in the major ocean basins of the global
ocean where convection in the order of 1000 m depth occurs on a regular basis, e.g. the Southern Ocean, the
northern Atlantic and the northern Pacific. However, within the main deep-connected waters of the global
ocean it is only in the Labrador Sea that winter convection extends to sufficient depths to contribute to deep-
water formation. Nevertheless, this only makes up a portion of the deep water deriving from the Northern
Hemisphere. The remainder comes from the Greenland Sea, where open-ocean convection creates dense
water that mixes with inflow through the Fram Strait from the Arctic (Mauritzen, 1996) and then water in
the Norwegian Sea to spill over the Greenland–Iceland–Scotland (GIS) Ridge and sink into the global ocean
basin. This, in conjunction with the deep water from the Labrador Sea, makes up the majority of NADW.
There appears to be an oscillation between the contribution of these two sources to NADW, depending on
the decadal state of the NAO (Dickson et al., 1996), the Greenland Sea being favoured more in the NAO’s
negative phase. The accompanying variation in the strength of the THC may well be within the error bounds
on transport estimates of 15–20 Sv (Schmitz, 1995), but recent heat transport estimates from further south in
the Atlantic suggest greater sensitivity (Koltermann et al., 1999).
Deep convection also occurs in the Bering Sea (Warner and Roden, 1995) and the northwestern Mediter-
ranean. However, the sills separating these basins from the global ocean are sufficiently shallow to prevent
the deepest, densest waters mixing with the main global deep water. About 1 Sv of relatively dense, warm,
salty water does enter the Atlantic through the Strait of Gibraltar. However, after mixing with waters in the
Gulf of Cadiz this is sufficiently light to spread out at depths of 1000–1500 m. We will discuss the fate of
the Mediterranean outflow in more detail in Section 4.4.
The second convection mechanism contributing to deep-water formation is through densification aided by
the release of salt during sea-ice formation or sub-ice shelf accretion (Makinson and Nicholls, 1999), most
typically on a continental shelf. Dense plumes are formed that slide down the continental slope. This process
occurs around the Antarctic coast (Orsi et al., 1999), and particularly in the Weddell Sea and under the
Ronne–Filchner Ice Shelf (Makinson and Nicholls, 1999), and in similar conditions in the Ross Sea, to form
the majority of Antarctic Bottom Water. The latter spreads out into every major ocean basin except the Arctic.
The warm and salty Norwegian Coastal Current also undergoes transformation through this process in the
Barents Sea, but provides a warm intermediate water layer that enters the Arctic at a depth of a few hundred
metres. It has been proposed that this water recirculates into the Greenland Sea at depth through the Fram
Strait to form a major component of the overflow water that exits into the Atlantic across the GIS Ridge
(Mauritzen, 1996). The actual source of overflow water is more likely to be from a combination of Nordic
seas convection and Arctic sources (Turrell et al., 1999).
It has long been known from radiocarbon measurements that even in the North Pacific, as far from deep-
water formation sites as it is possible to get, the age of the deep water is of the order of 1000 years. There
is thus a mechanism to upwell the deep water back to the surface that allows for complete renewal of the
deep waters on a millennial time scale. Formerly, it was thought that this upwelling occurred slowly over
the entire ocean, although observed open-ocean mixing rates were too small by an order of magnitude to
explain this (Ledwell et al., 1993). However, recent measurements of vertical mixing in the deep ocean
have shown that internal tidal mixing over the rough bathymetry of mid-ocean ridges has mixing rates
reaching 1–3 × 10−4 m2 s−1 (Egbert and Ray, 2000; Ledwell et al., 2000). In addition, there is about 30 Sv
of wind-driven upwelling in the Southern Ocean.
The return path of water to the deep convection zones has been formulated as in Figure 10. The resupply
of the convection zones around Antarctica readily occurs from the Antarctic Circumpolar Current through the
Weddell and Ross gyre circulations. However, the supply of water to the northern Atlantic involves a much
longer path. How much this represents the real trajectories of water parcels is a moot point. However, ocean
heat flow is consistent with transport north throughout the Atlantic. Some of this is derived from transport
in eddies spun-off the Agulhas Current into the Atlantic as it retroflects into the southern Indian Ocean
(McDonagh et al., 1999) — the so-called warm water path — but there is also transport of water through
Drake Passage that is transferred from the Falkland Confluence into the South Atlantic — the cold water path.
During the last decade there has been debate over the relative magnitudes of these two northward routes,
but modelling studies suggest about a quarter of the Antarctic Circumpolar Current’s strength is supported
by the THC (Gent et al., 2001) and that the warm and cold paths each contribute approximately 50% to
the northward transport in the southern Atlantic (Wadley and Bigg, 2002). The Mediterranean outflow also
contributes heat, salt and mass to northward transport in the Atlantic, but it seems most likely that, in the
current climate at least, this water affects the polar regions through its mixing with the overlying water rather
than as a direct water mass entering the Nordic seas (McCartney and Mauritzen, 2001).
4.2. Freshwater fluxes and state change

The THC is largely driven by thermal contrasts, with convection dominated by the presence of extreme
winter cooling. Nevertheless, deep convection occurs in cold polar regions, where salinity, due to the non-
linearities in the equation of state, has a strong influence on the density of seawater. Freshwater fluxes in these
regions are therefore vital for determining the location and regional amount of deep-water formation. The
freshwater flux to these regions is not purely through the precipitation–evaporation balance, but also through
the transport of relatively fresh sea ice, icebergs, and water masses with significant contributions from such
sources. The precipitation–evaporation balance itself is still poorly enough known for basin-wide estimates
to not always be of certain sign (Stocker et al., 2001). In addition, there is evidence that atmospheric models
poorly represent these fluxes, both in present-day (Lambert and Boer, 2001) and glacial simulations (Wadley
et al., 2002).
The transport of freshwater through ice advection can be very significant. Although there are insufficient
icebergs issuing into the northern Atlantic in concentration to have a significant impact on the freshwater
balance, this is not so for the Southern Ocean. A large proportion of the icebergs issuing from Antarctica are
confined to the Coastal Current for long times and undergo much of their melting here. Iceberg trajectory
modelling suggests that this freshwater flux is of the same magnitude as the precipitation–evaporation balance
along much of the continental shelf (Gladstone et al., 2001). As this is a zone of potential deep-water
formation, variations in the iceberg melt could have a significant impact on the southern arm of the THC.
In the northern Atlantic, sea-ice transport in the East Greenland Current and freshwater flux through the
Canadian Archipelago are important for the freshwater balance. The flux of freshwater through both these
routes is greater than from iceberg melt and run-off from Greenland, although the total is still approximately
an order of magnitude smaller than the precipitation–evaporation flux to the northwest Atlantic. However,
the convective activity in the Labrador Sea is controlled by this freshwater flux, meaning that the North
Atlantic overturning is 5% (Goosse et al., 1999) to 20% (Wadley and Bigg, 2002) less vigorous than it would
be in the absence of these fluxes. Sea ice may even have a positive feedback role in the THC (Schiller
et al., 1997). More sea-ice production occurring following a lessening of the THC and local cooling results in
export of more freshwater towards regions where convection may still be occurring. This may lead to further
suppression of convection, and yet more sea-ice formation, although with a negative feedback of salinification
of the underlying water.
The coupled climate system generally appears to be stable with respect to small perturbations in the
THC. Anomalous freshwater input, while reducing seawater density, also tends to lead to cooling, and hence
increased density. In Section 4.3 we will see that the THC can be pushed too far through freshwater fluxes, and
change to a new state. However, small perturbations have their amplitude reduced by the thermal negative
feedback. Compensatory heat transport within the atmosphere also leads to stabilization. Ganopolski and
Rahmstorf (2001) suggest that convection wholly south of the GIS Ridge is the most stable state for North
Atlantic deep-water formation, but the present-day situation, with some convection north of the GIS Ridge,
is quasi-stable (Lohmann and Gerdes, 1998; Ganopolski and Rahmstorf, 2001). Weijer et al. (2001) suggest
that the presence of heat transport through the Agulhas Current into the Atlantic contributes to this stability.
With this heat transport removed, the North Atlantic is much more sensitive to perturbations in the freshwater
transport from the Pacific through the Bering Strait.
Variation in freshwater flux has also been linked to the onset of Northern Hemisphere glaciation cycles
around 2–3 million years ago (Driscoll and Haug, 1998). Modelling work suggests that closure of the
Panamanian isthmus 3 million years ago would have increased the THC in the North Atlantic through
strengthening of the Gulf Stream (Maier-Reimer et al., 1990)) and the trade winds carrying moisture across
the Isthmus of Panama, making the Atlantic saltier and so more prone to convection. More heat transport
to the northern Atlantic is hypothesized as causing increased precipitation over Eurasia through enhanced
storminess. The additional run-off to the Arctic freshened the surface waters and allowed sea-ice formation
to occur more readily. The ice–albedo feedback could then cool the near-Arctic, allowing glaciation to occur
during astronomically favourable times.
The impact of the net loss of freshwater from the Atlantic via atmospheric transport across the Isthmus
of Panama is an important factor in making the North Atlantic the saltiest sub-polar ocean. This water is
replaced by salt-carrying waters from the Indian Ocean and southwest Atlantic. Convection is therefore more
likely in the North Atlantic because of the water’s enhanced salinity, and hence greater density when the
upper ocean is cooled in winter. It is correspondingly more difficult to stop convection from freshwater inputs
because of this enhanced salinity.
4.3. Thermohaline catastrophes

Ocean modelling work during the 1990s suggested dramatic effects resulting from variations in the
freshwater flux to the ocean (e.g. Marotzke and Willebrand, 1991; Power and Kleeman, 1993; Mikolajewicz
and Maier-Reimer, 1994; Bigg et al., 1998). More recent coupled model results (Rahmstorf, 1995; Ganopolski
and Rahmstorf, 2001; Rind et al., 2001a,b) have suggested that there are bifurcation points in the dependence
of THC overturning strength on freshwater forcing, so that the THC could exist in more than one stable, or
quasi-stable, state. Thus, returning to a previous state may not be merely a matter of reversing the freshwater
forcing, as once the ocean circulation has entered a particular state it may be resistant to change (Figure 11).
Although caution is always warranted when considering surprising findings from coarse climate and ocean
models, there is evidence for such major reorganizations of the ocean circulation during the last glacial period,
accompanying short-lived but large-scale input of icebergs to the northern Atlantic, so-called Heinrich events
Figure 11. Response of the North Atlantic overturning circulation to slowly changing freshwater forcing in high latitudes. Arrows
indicate the direction in which the freshwater forcing is changing and point ‘f’ shows a state very different to today’s, but with identical
freshwater forcing. (After Rahmstorf (1995), with permission from Nature Publishing Group, www.nature.com/nature)
(van Kreveld et al., 2000). This is also thought to have occurred more recently, during deglaciation and
the Younger Dryas (Stocker and Marchal, 2000; Bauch et al., 2001; Fagel et al., 2001). These thermohaline
catastrophes, whether modelled or in the observational record, have major consequences for climate. Both
observational (Renssen et al., 2001) and modelling (Manabe and Stouffer, 2000; Ganopolski and Rahmstorf,
2001) evidence suggests that the transitions can occur within a few decades, although some modelling
work does suggest a longer response time (Rind et al., 2001b). If the density of the northern Atlantic is
lessened to the extent that deep-water production is shut off, then the THC rearranges, with much reduced
northward transport of heat and salt in the North Atlantic. Regional atmospheric cooling of 5–10 ° C around
the northern Atlantic has been forecast to be produced in such a situation (Rahmstorf, 1995), consistent
with temperature changes during the Younger Dryas cold episode. The Atlantic westerlies adjust with a
more southerly storm track. Some modelling suggests a degree of compensation within the THC, with the
Southern Ocean production increasing. This has been seen in both ocean-only models (Bigg and Wadley, 2001)
and coupled models (Rind et al., 2001b). In the latter, the cooling of the Southern Ocean accompanying a
decrease in NADW formation leads to more sea ice and so greater salinification of the Antarctic shelf
waters.
The impact of adding sufficient freshwater to convection regions to shut off local deep-water formation,
thus causing large-scale adjustment of the THC and regional, if not global, climate, is clear. However, it is less
obvious how internal mechanisms could cause a switch in the reverse direction, from weak to strong North
Atlantic overturning. Ganopolski and Rahmstorf (2001) found that deep-water formation south of Iceland was
a stable climate mode, with overturning north of the GIS Ridge being a marginally unstable state. This is
consistent with the tendency for adjustment between Labrador Sea and Greenland Sea convection modulated
by the NAO (Dickson et al., 1996) and the evidence for the usual state of the glacial THC to be of a similar
strength to today (Yu et al., 1996; Matsumoto and Lynch-Stieglitz, 1999), but based south of the GIS Ridge
(Weinelt et al., 1996). Thus, switches in the formation region of NADW are related to modulation of local
surface density, with possible feedbacks into the North Atlantic atmospheric circulation. However, the reason
for a cycle or reversal of a more extreme switch of THC state is less clear.
Observational evidence suggests that the climate recovers from THC collapse, and recovers rapidly. At the
Younger Dryas, but also following the cold phases of the cyclic Dansgaard–Oeschger events during the last
glacial period, it is becoming clearer that relatively warm climates, and substantial NADW formation, was
resumed over a few decades to a century (van Kreveld et al., 2000). This has been ascribed to the injection of
salt into the North Atlantic through the extension of seasonal sea-ice formation well into mid latitudes. Slow
accumulation of salinity could eventually lead to a sudden onset of convection as the surface waters reached
critical density. Resumption of convection would then pull warmer water from the sub-tropics, stabilizing the
convection process and leading to sudden atmospheric warming. There is, as yet, no firm evidence to support
this mechanism.
Ocean-only models, however, almost invariably remain in a collapsed state, even if the surface forcing
producing this is removed (Bigg et al., 1998). The one exception is discussed in Section 4.4, as in that
case the presence of an oceanic feedback with the Mediterranean is crucial. Coupled models are more able
to reverse a thermohaline shutdown. Manabe and Stouffer (2000) saw their THC return to an active North
Atlantic state through a monotonic increase in North Atlantic overturning once the imposed freshwater forcing
leading to collapse was removed. Here, the additional freshwater was slowly mixed throughout the ocean,
allowing the surface salinity in the northern Atlantic to recover gradually, aided by associated adjustment
of the atmospheric circulation. A relatively gentle resumption of NADW formation was found on the long
time scale. Nevertheless, at the onset and removal of the freshwater anomaly, multi-decadal oscillations in
climate and THC behaviour were found. Ganopolski and Rahmstorf’s (2001) coupled model also showed
temporary cessation of NADW formation during a freshwater catastrophe. However, the model developed at
the Goddard Institute for Space Science has a response to this forcing that is permanent (Rind et al., 2001b).
A definitive modelling solution to explain THC catastrophes and their consequences is still awaited, although
the real ocean recovers in about 1000 years from such perturbations, e.g. leading to the end of the Younger
Dryas.
4.4. The potential impact of the Mediterranean outflow on climate
Historically, the outflow of warm salty water from the Mediterranean Sea, spreading out across the Atlantic
at depths around 1000–1500 m, has been thought to have a potential impact on deep-water production in the
Nordic seas (Reid, 1979). It was believed that a remnant of this water passed across the GIS Ridge through
the Faroe–Shetland Channel and preconditioned the subsurface water for convection by reducing its density.
However, recent observational evidence (Hill and Mitchelson-Jacob, 1993; McCartney and Mauritzen, 2001)
suggests that no traces of Mediterranean outflow water (MOW) survive so far north. Thus, vertical mixing of
MOW in the main Atlantic may reduce intermediate water density, but the impact of MOW is remote rather
than local to the Nordic seas.
Nevertheless, the impact of the MOW could be greater at other times in the past or future, or if the strength
of the outflow increased. It is thought that production of MOW practically ceases during sapropels, periodic
times of suppressed deep-water formation in the Mediterranean (Rohling, 1994). It has been hypothesized
(Johnson, 1997) that a significant enhancement of MOW would allow this water mass to make a major
contribution to the intermediate waters of the North Atlantic. Upon mixing of this warmer, salty intermediate
water to the surface of the northwest Atlantic and Baffin Bay during winter the atmosphere above it would
be warmed, leading to increased storminess. More winter storms in this area lead to increased snowfall, and
the possibility for sufficient snowfall to accumulate for increased glaciation in the islands off northeastern
Canada. A coupled modelling study (Rahmstorf, 1998) suggested that the impact of a realistic increase in
MOW would lead to increased warming in the atmosphere above the Atlantic, but not of sufficient magnitude
for the mechanism outlined above.
The impact of the Mediterranean on the North Atlantic circulation, and hence the climate, thus appears
small. However, almost no climate models can resolve the Mediterranean adequately, and particularly the
exchange through the Strait of Gibraltar. Even if a few model ocean grid points are used to represent the
Mediterranean then the exchange processes are normally parameterized very crudely, if allowed to vary at
all. Thorpe and Bigg (2000) tackled the question of the future impact of the Mediterranean on the North
Atlantic in reverse, by using a fine-resolution ocean model of the Mediterranean basin, forced by output from
an enhanced greenhouse transient run of the Hadley Centre’s climate model (HadCM2). This suggested that
the MOW would warm and become more salty over the next century, with the former change dominating
to make the density of the outflow lessen. The MOW level in the Atlantic was thus raised by a few tens of
metres by 2100.
The curvilinear grid ocean model of Wadley and Bigg (2000) is one of the few global models that can
be run for millennial time scales that can also attempt to incorporate the Mediterranean within the formal
grid structure. The grid has spatially varying resolution, through the displacement of the North Pole to
Greenland, allowing a finer resolution over the northern Atlantic. This model does tend to overestimate the
impact of the MOW in the deep North Atlantic, through producing generally denser outflow than reality.
However, one of their climate runs in particular (Bigg and Wadley, 2001) shows the potential impact of
the MOW. Figure 12 shows the behaviour of the North Atlantic overturning circulation, and the strength
of the Mediterranean outflow, for several thousand years during an ocean-only run started from a glacial
ocean circulation. Freshwater was added to the northern Atlantic, equivalent to a 1 mm day−1 increase in
the freshwater flux. The North Atlantic THC quickly collapsed, but after several hundred years the surficial
freshwater had been advected to other basins or mixed deeper, and the salinity of the upper North Atlantic
increased again. Figure 12 shows that a critical point for sparking vigorous convection in the Mediterranean
was passed around year 700, with a large, if temporary, rise in MOW production. The MOW mixing into
the North Atlantic further raises upper-ocean salinity, causing more Mediterranean convection bursts as the
salinity of the Mediterranean inflow is raised, and eventually deep convection starts again in the North Atlantic.
Note that, because the basic atmospheric forcing is glacial, North Atlantic convection occurs south of the
GIS Ridge, enabling the impact of the MOW to be felt directly in convection zones. As this run does not
have coupling to the atmosphere, eventually the excess salt is removed from the Atlantic and the overturning
is shut down again. Nevertheless, this simulation shows the potential for MOW to influence the global THC
directly, and hence climate.
Figure 12. Variation of the annual average strength of the meridional overturning streamfunction of an ocean model run of glacial
conditions: (a) for the Atlantic and (b) the Mediterranean outflow. (Figure 7 of Bigg and Wadley (2001), reproduced with permission
from John Wiley and Sons, Ltd)
4.5. Changes to the thermohaline circulation due to anthropogenic change

When forced with increasing concentrations of greenhouse gases, coupled ocean–atmosphere GCMs exhibit
a range of responses for the Atlantic meridional overturning circulation, from a virtually complete cessation
(Manabe and Stouffer, 1999; Bi et al., 2001) to no change (Latif et al., 2000; Gent, 2001; Sun and Bleck,
2001). Most models, however, simulate a gradual weakening of the THC (see the compilation of Cubasch
et al. (2001), reproduced here as Figure 13). There are two reasons for this wide range of responses, and hence
for the large uncertainty in the future strength of the THC. The first is that some of the key processes involved
Figure 13. Simulated changes of the Atlantic THC under a scenario of increasing greenhouse gas and sulphate aerosol forcing produced
by nine different coupled ocean–atmosphere models. The circulation strength (Sv) is shown as anomalies from each simulation’s
1961–90 mean. (Redrawn from Cubasch et al. (2001) with permission from the IPCC  2001)
in maintaining and modifying the THC are poorly observed and poorly represented in climate models, such
as convection, diapycnal mixing and overflow currents. We will not consider these in detail here. The second
reason is that there are a number of competing effects, related mainly to changes in air–sea fluxes. When
the resultant change is equivalent to the difference between two larger terms, and is subsequently amplified
by the feedback processes discussed earlier, then the sign and magnitude of the resultant is quite sensitive
to relatively small changes in the magnitude of the larger terms. Thus, relatively minor differences in the
simulated response of the air–sea flux fields to climate change can cause much larger differences in the THC
response.
The key local (to the sinking region) and remote effects, and associated feedbacks, are shown in Figure 14.
The first effect (1) is ubiquitous across all models: a heating of the oceans leading to a reduction in density of
the surface water in the sinking region and a slowing of the THC. This then initiates the feedback processes
discussed earlier: reduced (advective and convective) transport of heat and salt to the surface of the sinking
region, the former leading to a relative cooling and a negative feedback, and the freshening associated with the
latter represents a positive feedback. The reduced transport of heat to the northern North Atlantic also results
in a positive feedback by enhancing the air–sea heat flux into the region (see figure 7 of Thorpe et al. (2001)).
This ‘positive’ feedback, although it can never overcome the negative feedback associated with reduced heat
transport, is important for sustaining a reduced THC on longer time scales, because it induces enhanced
column-integrated heating of the northern North Atlantic relative to other oceans and thereby weakens the
meridional gradient of steric height that drives the THC (Thorpe et al., 2001).
Figure 14. Schematic illustrating the local, remote and feedback influences on the THC under greenhouse warming. Q and F represent
heat and freshwater, T and S represent surface temperature and salinity in the northern North Atlantic, ‘flux’ represents atmosphere–ocean
fluxes in the northern North Atlantic, ‘transport’ represents advective and convective transports to the surface of the northern North
Atlantic, and ↑ and ↓ represent increases and decreases respectively
The second effect (2) in Figure 14 is that the warmer climate is (in most models) associated with an
anomalous freshwater input into the northern North Atlantic Ocean. Annual precipitation increases north of
30–40 ° N over the Atlantic (figure 9.11 of Cubasch et al. (2001)) because the intensified hydrological cycle
drives an enhanced atmospheric meridional transport of freshwater. Evaporation is also enhanced as a result
of warmer SSTs, but in most models this does not match the increased precipitation. Indeed, if a model’s
THC slows markedly and inhibits North Atlantic warming sufficiently, then increased atmospheric stability
can prevent any evaporation increase (Russell and Rind, 1999), i.e. remote effect (3) in Figure 14. In at
least one model (Gent, 2001), however, a THC maintained at full strength allows strong warming of North
Atlantic SSTs, which, together with increased surface wind speeds, drives an increase in evaporation greater
than the simulated precipitation increase (and hence the second effect (2) in Figure 14 is reversed and the
THC strength is maintained). The freshwater input from the melting of mountain glaciers (and possibly from
increasing calving around the Greenland ice cap) is not included in any of the current GCMs, and hence the
simulated change in freshwater flux is slightly biased to too little freshening of the ocean (this effect may be
quite minor, but deserves investigation).
The balance between local effects (1) and (2) depends upon details of the simulated air–sea flux responses
and their interaction with thermohaline-driven heat and salt transports. Carefully designed experiments with
GCMs have gone some way towards isolating the relative influence of the heat and freshwater effects, with
model-dependent results. Dixon et al. (1999) demonstrate that both effects contribute to the THC weakening
in the Geophysical Fluid Dynamics Laboratory model, but with freshwater flux changes taking the dominant
role, whereas in the Hamburg ECHAM3/LSG model it is the effect of surface warming that is dominant
(Mikolajewicz and Voss, 2000). The latter result is in agreement with some intermediate complexity models,
e.g. that of Rahmstorf and Ganopolski (1999), who found a dominant heat flux effect during the first decades
of their multi-century simulations.
Recent modelling studies have highlighted the role of remote forcing of the North Atlantic surface salinity.
Gent (2001) notes that the decreasing freshwater input from the Arctic (in the form of sea ice that melts
in the northwest Atlantic) contributes to an increase in salinity that prevents a weakening of the THC in
the National Center for Atmospheric Research’s CSM model (remote effect (5) in Figure 14). A number of
recent simulations find that the atmospheric export of freshwater from the tropical Atlantic basin to the Pacific
basin (which is important for driving the present-day global THC) may be enhanced under warming scenarios
(remote effect (4) in Figure 14). The ECHAM3/LSG model (Mikolajewicz and Voss, 2000) simulates such an
enhancement, but the resulting salinification of the Atlantic is insufficient to reverse the high-latitude (local)
effects discussed above. In ECHAM4/OPYC (Latif et al., 2000), however, the freshwater export enhancement
is three times stronger (than in ECHAM3/LSG), and is sufficient to prevent any weakening of the THC. Latif
et al. (2000) speculate that the fine meridional resolution (0.5° ) in the tropics of their ocean GCM (OPYC) is
critical in obtaining this response: the enhanced resolution improves its representation of the equatorial internal
ocean wave dynamics involved in the ENSO phenomenon, and may enable the model to simulate a warming
pattern, in response to greenhouse forcing, that resembles the El Niño pattern (in the eastern Pacific). Enhanced
Atlantic freshwater export is observed during present-day El Niños (Schmittner et al., 2000), though this does
not exclude export changes driven by other mechanisms. In HadCM3, with an intermediate meridional ocean
resolution (1.25° ), Thorpe et al. (2001) also find an enhanced export of freshwater from the tropical Atlantic
basin that enhances the ocean advection of salt to the northern North Atlantic Ocean (more than offsetting
the reduced ocean velocity associated with the weakened THC in this model). In HadCM3, however, this
enhanced freshwater export is not associated with an El Niño-like warming pattern in the Pacific Ocean
(Thorpe et al., 2001), and it is not strong enough to prevent a weakening of the THC (though it moderates
the weakening).
The possible existence of bifurcation points (discussed in Section 4.3) in the response of the Atlantic THC
to changes in surface forcings could influence both the short-term behaviour (i.e. an abrupt change if the
positive feedbacks indicated in Figure 14 are sufficiently strong to pass a bifurcation point) and the multi-
century behaviour (whether a modified THC strength is maintained or recovers after greenhouse forcing is
stabilized). There are no indications in Figure 13 that any of the GCM simulations have entered a different
mode during the 21st century, because there are no abrupt changes and none of them shows a complete
cessation of the meridional overturning streamfunction in the Atlantic. Some models however, have shown,
that this outcome may be a function of the both the magnitude and the rate of greenhouse forcing increase
(e.g. Stocker and Schmittner, 1997; Stouffer and Manabe, 1999).
When greenhouse forcing is stabilized, some models exhibit a recovery of the THC back to near its initial
strength (e.g. Bi et al., 2001; Raper et al., 2001: figure 9) and others simulate a reduced THC strength that
is sustained — and therefore represents a new steady state. A new steady state does not necessarily imply
the existence or passing of a bifurcation point: the new state will be under different boundary conditions
and may simply be a response to these new conditions. In the HadCM3 model, for example, Thorpe et al.
(2001) diagnose that the THC stabilizes with a strength 20% below the initial state and attribute this to a
changed tropical Atlantic freshwater budget (linked to the enhanced export of freshwater discussed above)
balancing density changes in the sinking region. In other models, however, the THC may follow either the
recovery pathway or the sustained new steady-state pathway, depending on the details of the greenhouse
forcing applied (e.g. Manabe and Stouffer, 1994; Stocker and Schmittner, 1997; Stouffer and Manabe, 1999).
This indicates the existence of bifurcation points and that the forcing can determine whether or not they are
passed.
The impact on surface climate of a change in the Atlantic THC strength will be combined with the
continuing atmospheric signal due to greenhouse-gas-induced warming. Given that a reduction in THC would
greatly reduce the oceanic component of the northward heat transport in the Atlantic, it would clearly reduce
temperatures in the North Atlantic Ocean and surrounding land areas (perhaps especially those downstream
in the mid-latitude band of prevailing westerly winds). In these regions, a reduced THC and greenhouse
warming would have opposite effects (in terms of temperature, at least), and thus the net effect depends
on the relative strengths of the two signals. A dramatic cooling of the region would only occur if a large
reduction in THC strength occurred before significant greenhouse warming had taken place, and therefore
it is an abrupt change in the THC during the first half of the 21st century that would be of most concern.
Figure 15. Changes in mean 1.5 m air temperature (‘.degC) simulated by HadCM3 following a forced, complete collapse of the Atlantic
THC. Statistically insignificant changes are white, other changes are shaded grey, with warming (mainly in the Southern Hemisphere)
indicated by the ‘+’ marking. Data were provided courtesy of Michael Vellinga (Hadley Centre, UK) from the simulation reported in
Vellinga and Wood (2002)
The GCM simulations shown in Figure 13 show a more gradual weakening of the THC (or no change in
some cases), and simulate reduced warming (rather than dramatic cooling) in the North Atlantic region as a
whole, with localized cooling in only some simulations. Model simulations (e.g. Vellinga and Wood, 2002;
Figure 15) are required to obtain quantitative estimates of the response to a THC collapse, especially for
variables other than temperature, to take into account biogeochemical changes and especially when combined
with the signal of greenhouse warming (the combination might be non-linear).
Figure 15 gives one such example (using data from the simulation reported by Vellinga and Wood (2002)),
showing the surface air temperature response following a forced cessation of HadCM3’s Atlantic meridional
overturning circulation. Cooling of more then 5 ° C is restricted to oceanic regions of the northern North
Atlantic Ocean, with most land areas showing less then 2 ° C of cooling. A weaker, but widespread, warming
is simulated over the Southern Hemisphere. The actual climate change following a partial collapse of the
Atlantic THC due to anthropogenic climate forcing would be some fraction of the temperature change given
in Figure 15 combined (possibly non-linearly) with the anthropogenic climate-change signal.
ACKNOWLEDGEMENTS
TJO was supported by UK NERC under the COAPEC thematic programme (NER/T/S/2000/00327). We
would like to thank Itsuki Handoh for supplying Figure 9 and a reviewer for helpful comments on improving
the manuscript.
REFERENCES
Andreae MO, Crutzen PJ. 1997. Atmospheric aerosols: biogeochemical sources and their role in atmospheric chemistry. Science 276:
1052–1058.
Andreae MO, Elbert W, Cai Y, Andreae TW, Gras J. 1999. Non-seasalt sulfate, methanesulfonate and nitrate aerosol concentrations
and size distributions at Cape Grim Tasmania. Journal of Geophysical Research 104: 21 695–21 706.
Asher WL, Wanninkhof R. 1998. The effect of bubble-mediated gas transfer on purposeful dual-gaseous tracer experiments. Journal of
Geophysical Research 103: 10 555–10 560.
Ayers GP, Ivey JP, Gillett RW. 1991. Coherence between seasonal cycles of dimethyl sulphide, methanesulphanate and sulphate in
marine air. Nature 349: 404–406.
Baker AR, Turner SM, Broadgate WJ, Thompson A, McFiggans GB, Vesperinin O, Nightingale PD, Liss PS, Jickells TD. 2000.
Distribution and sea–air flux of biogenic trace gases in the eastern Atlantic Ocean. Global Biogeochemical Cycles 14: 871–886.
Bates TS, Kelly KC, Johnson JE, Gammon RH. 1996. A reevaluation of the open ocean source of methane to the atmosphere. Journal
of Geophysical Research 101: 6953–6961.
Battisti DS, Hirst AC. 1989. Interannual variability in the tropical atmosphere/ocean system: influence of basin state, ocean geometry
and non-linearity. Journal of the Atmospheric Sciences 46: 1687–1712.
Bauch HA, Erlenkeuser H, Spielhagen RF, Struck U, Matthiessen J, Thiede J, Heinemeier J. 2001. A multiproxy reconstruction of the
evolution of deep and surface waters in the subarctic Nordic seas over the last 30 000 years. Quaternary Science Reviews 20: 659–678.
Behera SK, Krishnan R, Yamagata T. 1999. Unusual ocean–atmosphere conditions in the tropical Indian Ocean during 1994.
Geophysical Research Letters 26: 3001–3004.
Berman-Frank I, Cullen JT, Shaked Y, Sherrell RM, Falkowski PG. 2001. Iron availability, cellular iron quotas and nitrogen fixation
in Trichodesmium. Limnology and Oceanography 45: 1249–1260.
Bi DH, Budd WF, Hirst AC, Wu XR. 2001. Collapse and re-organisation of the Southern Ocean overturning under global warming in
a coupled model. Geophysical Research Letters 28: 3927–3930.
Bigg GR. 1996. The Oceans and Climate. Cambridge University Press: Cambridge.
Bigg GR, Wadley MR. 2001. Millennial changes in the oceans: an ocean modeller’s viewpoint. Journal of Quaternary Science 16:
309–319.
Bigg GR, Wadley MR, Stevens DP, Johnson JA. 1998. Simulations of two last glacial maximum ocean states. Paleoceanography 13:
340–351.
Boers R, Ayers GP, Gras GC. 1994. Coherence between seasonal variation in satellite-derived cloud optical depth and boundary layer
CCN at a mid-latitude Southern Hemisphere station. Tellus B 46: 123–131.
Boyd PW, Watson AJ, Law CS, Abraham ER, Trull T, Murdoch R, et al. 2000. A mesoscale phytoplankton bloom in the polar Southern
Ocean stimulated by iron fertilisation. Nature 407: 695–723.
Broadgate WJ, Liss PS, Penkett SA. 1997. Seasonal emissions of isoprene and other reactive hydrocarbons from the ocean. Geophysical
Research Letters 24: 2675–2678.
Camberlin P, Janicot S, Poccard I. 2001. Seasonality and atmospheric dynamics of the teleconnection between African rainfall and
tropical sea-surface temperature: Atlantic vs. ENSO. International Journal of Climatology 21: 973–1005.
Carpenter LJ, Sturges WT, Penkett SA, Liss PS, Alicke B, Hebestreit K, Platt U. 1999. Short-lived alkyl iodides and bromides at Mace
Head Ireland: links to biogenic sources and halogen oxide production. Journal of Geophysical Research 104: 1679–1689.
Carton JA, Cao XH, Giese BS, daSilva AM. 1996. Decadal and interannual SST variability in the tropical Atlantic Ocean. Journal of
Physical Oceanography 26: 1165–1175.
Cayan DR. 1992. Latent and sensible heat-flux anomalies over the northern oceans — the connection to monthly atmospheric circulation.
Journal of Climate 5: 354–369.
Chang P, Ji L, Li H. 1997. A decadal climate variation in the tropical Atlantic Ocean from thermodynamic air–sea interactions. Nature
385: 516–518.
Chang P, Saravanan R, Ji L, Hegerl GC. 2000. The effect of local sea surface temperatures on atmospheric circulation over the tropical
Atlantic sector. Journal of Climate 13: 2195–2216.
Charlson RJ, Lovelock JE, Andrea MO, Warren SG. 1987. Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate.
Nature 326: 655–661.
Christoph M, Barnett TP, Roeckner E. 1998. The Antarctic Circumpolar Wave in a coupled ocean–atmosphere GCM. Journal of Climate
11: 1659–1672.
Church JA, Gregory JM, Huybrechts P, Kuhn M, Lambeck K, Nhuan MT, Qin D, Woodworth PL. 2001. Changes in sea level. In
Climate Change 2001: The Scientific Basis, Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, Maskell K,
Johnson CA (eds). Cambridge University Press: Cambridge; 639–693.
Collins M, Tett SFB, Cooper C. 2001. The internal climate variability of HadCM3, a version of the Hadley Centre coupled model
without flux adjustments. Climate Dynamics 17: 61–81.
Cubasch U, Meehl GA, Boer GJ, Stouffer RJ, Dix M, Noda A, Senior CA, Raper S, Yap KS. 2001. Projections of future climate change.
In Climate Change 2001: The Scientific Basis, Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, Maskell K,
Delworth TL, Manabe S, Stouffer RJ. 1993. Interdecadal variations of the thermohaline circulation in a coupled ocean–atmosphere
model. Journal of Climate 6: 1993–2011.
Deser C, Blackmon ML. 1993. Surface climate variations over the North Atlantic Ocean during winter: 1900–1989. Journal of Climate
6: 1743–1753.
Diaz HF, Hoerling MP, Eischeid JK. 2001. ENSO variability, teleconnections and climate change. International Journal of Climatology
21: 1845–1862.
Dickson RR, Lazier J, Meincke J, Rhines P, Swift J. 1996. Long-term coordinated changes in the convective activity of the North
Atlantic. Progress in Oceanography 38: 241–295.
Dixon KW, Delworth TL, Spelman MJ, Stouffer RJ. 1999. The influence of transient surface fluxes on North Atlantic overturning in a
coupled GCM climate change experiment. Geophysical Research Letters 26: 2749–2752.
Driscoll NW, Haug GH. 1998. A short circuit in thermohaline circulation: a cause for Northern Hemisphere glaciation? Science 282:
436–438.
Egbert GD, Ray RD. 2000. Significant dissipation of tidal energy in the deep ocean inferred from satellite altimeter data. Nature 405:
775–778.
Enfield DB, Mestas-Nunez AM, Cid-Serrano L. 1999. How ubiquitous is the dipole relationship in the tropical Atlantic sea surface
temperatures? Journal of Geophysical Research 104: 7841–7848.
Fagel N, Robert C, Preda M, Thorez J. 2001. Smectite composition as a tracer of deep circulation: the case of the northern North
Atlantic. Marine Geology 172: 309–330.
Falkowski PG, Barber RT, Smetacek V. 1998. Biogeochemical controls and feedbacks on ocean primary production. Science 281:
200–206.
Feely RA, Sabine CL, Takahashi T, Wanninkhof R. 2001. Uptake and storage of carbon dioxide in the ocean. The global CO2 survey.
Oceanography 14: 18–32.
Galloway JN, Cowling EB. 2002. Reactive nitrogen and the world: 200 years of change. Ambio 31: 64–71.
Ganopolski A, Rahmstorf S. 2001. Rapid changes of glacial climate simulated in a coupled climate model. Nature 409: 153–158.
Gascard JC, Watson AJ, Messias MJ, Olsson KA, Johannessen T, Simonson K. 2002. Long-lived vortices as a mode of deep ventilation
in the Greenland Sea. Nature 416: 525–527.
Gent PR. 2001. Will the North Atlantic Ocean thermohaline circulation weaken during the 21st century? Geophysical Research Letters
28: 1023–1026.
Gent PR, Large WG, Bryan FO. 2001. What sets the mean transport through Drake Passage? Journal of Geophysical Research 106:
2693–2712.
Gladstone R, Bigg GR, Nicholls KW. 2001. Icebergs and fresh water fluxes in the Southern Ocean. Journal of Geophysical Research
106: 19 903–19 915.
Goddard L, Mason SJ, Zebiak SE, Ropelewski CF, Basher R, Cane MA. 2001. Current approaches to seasonal-to-interannual climate
predictions. International Journal of Climatology 21: 1111–1152.
Goosse H, Deleersnijder E, Fichefet T, England MH. 1999. Sensitivity of a global coupled ocean–sea ice model to the parameterization
of vertical mixing. Journal of Geophysical Research 104: 13 681–13 695.
Gornitz V, Fung I. 1994. Potential distribution of methane hydrates in the world oceans. Global Biogeochemical Cycles 8: 335–347.
Gregory JM, Church JA, Boer GJ, Dixon KW, Flato GM, Jackett DR, Lowe JA, O’Farrell SP, Roeckner E, Russell GL, Stouffer RJ,
Winter M. 2001. Comparison of results from several AOGCMs for global and regional sea-level change 1900–2100. Climate Dynamics
18: 225–240.
Grotzner A, Latif M, Barnett TP. 1998. A decadal climate cycle in the North Atlantic Ocean as simulated by the ECHO coupled GCM.
Handoh IC, Bigg GR. 2000. A self-sustaining climate mode in the tropical Atlantic, 1995–1997: observations and modelling. Quarterly
Journal of the Royal Meteorological Society 126: 807–821.
Handoh IC, Matthews AJ, Bigg GR, Stevens DP. Submitted for publication. Dominant coupled ocean–atmosphere interannual modes
in the tropical Atlantic: part 1, the northern tropical Atlantic mode. Journal of Climate.
Hasselmann K. 1976. Stochastic climate models, I: theory. Tellus 28: 473–485.
Hewitt CD, Broccoli AJ, Mitchell JF, Stouffer RJ. 2001. A coupled model study of the last glacial maximum: was part of the North
Atlantic relatively warm? Geophysical Research Letters 28: 1571–1574.
Hill AE, Mitchelson-Jacob EG. 1993. Observations of a poleward-flowing saline core on the continental slope west of Scotland. Deep-Sea
Research Part I: Oceanographic Research Papers 40: 1521–1527.
Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, Maskell K, Johnson CA (eds). 2001. Climate Change 2001:
The Scientific Basis. Cambridge University Press: Cambridge.
Huybrechts P, De Wolde J. 1999. The dynamic response of the Greenland and Antarctic ice sheets to multiple-century climatic warming.
Jacobson MC, Hansson HC, Noone KJ, Charlson RJ. 2000. Organic atmospheric aerosols: review and state of the science. Reviews of
Geophysics 38: 267–294.
Jickells T. 1998. Nutrient biogeochemistry of the coastal zone. Science 281: 217–222.
Jickells T. 2002. Emissions from the oceans to the atmosphere, deposition from the atmosphere to the oceans and the interactions
between them. In Challenges of a Changing Earth, Proceedings of the Global Change Open Science Conference, The Netherlands,
10–13 July 2001, Steffen W, Jager J, Carson D, Bradshaw C (eds). Springer-Verlag: Berlin; 92–96.
Jickells TD, Spokes LJ. 2001. Atmospheric iron inputs to the oceans. In The Biogeochemistry of Iron in Seawater, Turner DR, Hunter KA
(eds). Wiley: Chichester; 85–121.
Johnson RG. 1997. Ice age initiation by an ocean–atmospheric circulation change in the Labrador Sea. Earth and Planetary Science
Letters 148: 367–379.
Kageyama M, Valdes PJ, Ramstein G, Hewitt C, Wyputta U. 1999. Northern Hemisphere storm tracks in present day and last glacial
maximum climate simulations: a comparison of the European PMIP models. Journal of Climate 12: 742–760.
Kettle AJ, Andreae MO, Amouroux D, Andreae TW, Bates TS, Berresheim H, Bingemer H, Boniforti R, Curran MAJ, DiTullio GR,
Helas G, Jones GB, Keller MD, Kiene RP, Leck C, Levasseur M, Malin G, Maspero M, Matrai P, McTaggart AR, Mihalopoulos N,
Nguyen BC, Novo A, Putaud JP, Rapsomanikis S, Roberts G, Schebeske G, Sharma S, Simo R, Staubes R, Turner S, Uher G. 1999.
A global database of sea surface dimethylsulfide (DMS) measurements and a procedure to predict sea surface DMS as a function of
latitude, longitude, and month. Global Biogeochemical Cycles 13: 399–444.
Kitoh A, Murakami S, Koide H. 2001. A simulation of the last glacial maximum with a coupled atmosphere–ocean model. Geophysical
Koltermann KP, Sokov AV, Tereschenkov VP, Dobroliubov SA, Lorbacher K, Sy A. 1999. Decadal changes in the thermohaline
circulation of the North Atlantic. Deep-Sea Research Part II: Topical Studies in Oceanography 46: 109–138.
Kuijpers A, Troelstra SR, Wisse M, Nielsen SH, van Weering TCE. 1998. Norwegian Sea overflow variability and NE Atlantic surface
hydrography during the past 150 000 years. Marine Geology 152: 75–99.
Kushnir Y, Held I. 1996. Equilibrium atmospheric response to North Atlantic SST anomalies. Journal of Climate 9: 1208–1220.
Lambert SJ, Boer GJ. 2001. CMIP1 evaluation and intercomparison of coupled climate models. Climate Dynamics 17: 83–106.
Latif M. 2001. Tropical Pacific/Atlantic Ocean interactions at multi-decadal timescales. Geophysical Research Letters 28: 539–542.
Latif M, Barnett TP. 1994. Causes of decadal climate variability over the North Pacific and North America. Science 266: 634–637.
Latif M, Barnett TP. 1995. Interactions of the tropical oceans. Journal of Climate 8: 952–964.
Latif M, Barnett TP. 1996. Decadal variability over the North Pacific and North America: dynamics and predictability. Journal of
Climate 9: 2407–2423.
Latif M, Grotzner A. 2000. The equatorial Atlantic oscillation and its response to ENSO. Climate Dynamics 16: 213–218.
Latif M, Roeckner E, Mikolajewicz U, Voss R. 2000. Tropical stabilization of the thermohaline circulation in a greenhouse warming
simulation. Journal of Climate 13: 1809–1813.
Ledwell JR, Watson AJ, Law CS. 1993. Evidence for slow mixing across the pycnocline from an open-ocean tracer-release experiment.
Nature 364: 701–703.
Ledwell JR, Montgomery ET, Polzin KL, St Laurent LC, Schmitt RW, Toole JM. 2000. Evidence for enhanced mixing over rough
topography in the abyssal ocean. Nature 403: 179–182.
Lee KE, Slowey NC, Herbert TD. 2001. Glacial sea surface temperatures in the sub-tropical North Pacific: a comparison of U-k37,
delta 18O, and foraminiferal assemblage temperature estimates. Paleoceanography 16: 268–279.
Lelieveld J, Crutzen PJ, Dentener FJ. 1998. Changing concentration, lifetime and climate forcing of atmospheric methane. Tellus B 50:
128–150.
Li ZX. 2001. Thermodynamic air–sea interactions and tropical Atlantic SST dipole pattern. Physics and Chemistry of the Earth, Part B:
Hydrology, Oceans and Atmosphere 26: 155–157.
Liss PS, Galloway JN. 1993. Air–sea exchange of sulphur and nitrogen and their interaction in the marine atmosphere. In Interactions
of C, N, P and S Biogeochemical Cycles and Global Change, Wollast R, MacKenzie FT, Chou L (eds). Springer: Berlin; 249–281.
Liss PS, Merlivat L. 1986. Air–sea exchange rates: introduction and synthesis. In The Role of Air–Sea Exchange in Geochemical
Cycling, But-Menard P (ed.). Redell: Norwell, MA, 112–127.
Liss PS, Watson AJ, Brock EJ, Jahne B, Asher WE, Frew NM, Hasse L, Korenowski GM, Merlivat L, Phillips LF, Schluessel P,
Woolf DK. 1997. Report Group 1 — physical processes in the microlayer and the air–sea exchange of trace gases. In The Sea
Surface and Global Change, Liss PS, Duce RA (eds). Cambridge University Press: Cambridge; 1–33.
Lohmann G, Gerdes R. 1998. Sea ice effects on the sensitivity of the thermohaline circulation. Journal of Climate 11: 2789–2803.
Maier-Reimer E, Mikolajewicz U, Crowley T. 1990. Ocean general circulation model sensitivity experiment with an open central
American isthmus. Paleoceanography 5: 349–366.
Makinson K, Nicholls KW. 1999. Modeling tidal currents beneath Filchner–Ronne Ice Shelf and on the adjacent continental shelf: their
effect on mixing and transport Journal of Geophysical Research 104: 13 449–13 465.
Manabe S, Stouffer RJ. 1994. Multiple century response of a coupled ocean-atmosphere model to an increase of atmospheric carbon
dioxide. Journal of Climate 7: 5–23.
Manabe S, Stouffer RJ. 1999. The role of thermohaline circulation in climate. Tellus 51: 91–109.
Manabe S, Stouffer RJ. 2000. Study of abrupt climate change by a coupled ocean–atmosphere model. Quaternary Science Reviews 19:
285–299.
Marotzke J, Willebrand J. 1991. Multiple equilibria of the global thermohaline circulation. Journal of Physical Oceanography 21:
1372–1385.
Marshall J, Schott F. 1999. Open-ocean convection: observations, theory, and models. Reviews of Geophysics 37: 1–64.
Marshall J, Kushnir Y, Battisti D, Chang P, Czaja A, Dickson R, Hurrell J, McCartney M, Saravanan R, Visbeck M. 2001. North
Atlantic climate variability: phenomena, impacts and mechanisms. International Journal of Climatology 21: 1863–1898.
Matsumoto K, Lynch-Stieglitz J. 1999. Similar glacial and Holocene deep water circulation inferred from southeast Pacific benthic
foraminiferal carbon isotope composition. Paleoceanography 14: 149–163.
Matsumoto K, Lynch-Stieglitz J, Anderson RF. 2001. Similar glacial and Holocene Southern Ocean hydrography. Paleoceanography
16: 445–454.
Mauritzen C. 1996. Production of dense overflow waters feeding the North Atlantic across the Greenland–Scotland Ridge. 1. Evidence
for a revised circulation scheme. Deep-Sea Research Part I. Oceanographic Research Papers 43: 769–806.
McCartney MS, Mauritzen C. 2001. On the origin of the warm inflow to the Nordic seas. Progress in Oceanography 51: 125–214.
McDonagh EL, Heywood KJ, Meredith MP. 1999. On the structure, paths, and fluxes associated with Agulhas rings. Journal of
Geophysical Research 104: 21 007–21 020.
Mikolajewicz U, Maier-Reimer E. 1994. Mixed boundary conditions in ocean general circulation models and their influence on the
stability of the model’s conveyor belt. Journal of Geophysical Research 99: 22 633–22 644.
Mikolajewicz U, Voss R. 2000. The role of the individual air–sea flux components in CO2 -induced changes of the ocean’s circulation
and climate. Climate Dynamics 16: 627–642.
Mitchell Jr. JM 1976. Overview of climatic variability and its causal mechanisms. Quaternary Research 6: 481–493.
Murtugudde RG, Ballabrera-Poy J, Beauchamp J, Busalacchi AJ. 2001. Relationship between zonal and meridional modes in the tropical
Atlantic. Geophysical Research Letters 28: 4463–4466.
Naqvi SWA, Jayakumar DA, Narekar P, Naik H, Sarma VVSS, D’Souza W, Joseph S, George MD. 2000. Increased marine production
of N2 O due to intensifying anoxia on the Indian continental shelf. Nature 408: 346–349.
Nevison CD, Weiss RF, Erickson III DJ. 1995. Global oceanic emissions of nitrous oxide. Journal of Geophysical Research 100:
15 809–15 820.
Nightingale PD, Malin G, Law CS, Watson AJ, Liss PS, Liddicoat MI, Boutin J, Upstill-Goddard RC. 2000. In situ evaluation of
air–sea gas exchange parameterizations using novel conservative and volatile tracers. Global Biogeochemical Cycles 14: 373–387.
O’Dowd CD, Lowe JA, Smith MH, Davison B, Hewitt N, Harrison RM. 1997. Biogenic sulphur emissions and inferred non-sea-
salt–sulphate cloud condensation nuclei in and around Antarctica. Journal of Geophysical Research 102: 12 839–12 854.
O’Dowd CD, Jimenez JL, Bahreini R, Hagan RC, Seinfeld JH, Hämeri K, Pirjola L, Kulmala M, Jennings SG, Hoffmann T. 2002.
Marine aerosol formation from biogenic iodine emissions. Nature 417: 632–636.
Okumura Y, Xie SP, Numaguti A, Tanimoto Y. 2001. Tropical Atlantic air–sea interaction and its influence on the NAO. Geophysical
Orsi AH, Johnson GC, Bullister JL. 1999. Circulation, mixing, and production of Antarctic bottom water. Progress in Oceanography
43: 55–109.
Osborn TJ, Wigley TML. 1994. A simple model for estimating methane concentration and lifetime variations. Climate Dynamics 9:
181–193.
Overpeck J, Anderson D, Trumbore S, Prell W. 1996. The southwest Indian Monsoon over the last 18 000 years. Climate Dynamics
12: 213–225.
Peng S, Robinson WA, Hoerling MP. 1997. The modeled atmospheric response to middle-latitude SST anomalies and its dependence
on background circulation states. Journal of Climate 10: 971–987.
Penner JE, Andreae M, Annegarn H, Barrie L, Feichter J, Hegg D, Jayaraman A, Leaitch R, Murphy D, Nganga J, Pitari G. 2001.
Aerosols, their direct and indirect effects. In Climate Change 2001: The Scientific Basis. Houghton JT, Ding Y, Griggs DJ, Noguer M,
van der Linden PJ, Dai X, Maskell K, Johnson CA (eds). Cambridge University Press: Cambridge; 289–348.
Pierce DW. 2001. Distinguishing coupled ocean–atmosphere interactions from background noise in the North Pacific. Progress in
Powell MD, Vickery PJ, Reinhold TA. 2003. Reduced drag coefficient for high wind speeds in tropical cyclones. Nature 422: 279–283.
Power SB, Kleeman R. 1993. Multiple equilibria in a global ocean general circulation model. Journal of Physical Oceanography 23:
1670–1681.
Rae F, Dingenen RV, Vignati E, Wilson J, Putaud J-P, Seinfeld JH, Admans P. 2000. Formation and cycling of aerosols in the global
troposphere. Atmospheric Environment 34: 4215–4240.
Rahmstorf S. 1995. Bifurcations of the Atlantic thermohaline circulation in response to changes in the hydrological cycle. Nature 378:
145–149.
Rahmstorf S. 1998. Influence of Mediterranean outflow on climate. EOS Transactions of the American Geophysical Union 79: 281–282.
Rahmstorf S, Ganopolski A. 1999. Long-term global warming scenarios computed with an efficient coupled climate model. Climatic
Change 43: 353–367.
Raper SCB, Gregory JM, Osborn TJ. 2001. Use of an upwelling-diffusion energy balance climate model to simulate and diagnose
A/OGCM results. Climate Dynamics 17: 601–613.
Reid JL. 1979. On the contribution of the Mediterranean Sea outflow on climate. Deep-Sea Research 26: 1199–1223.
Renssen H, Isarin RFB, Vandenberghe J. 2001. Rapid climatic warming at the end of the last glacial: new perspectives. Global and
Planetary Change 30: 155–165.
Ridgwell AJ, Maslin MA, Watson AJ. 2002. Reduced effectiveness of terrestrial carbon sequestration due to an antagonistic response
of ocean productivity. Geophysical Research Letters 29: 10/1029/2001GLO14304.
Rind D, deMonocal P, Russell G, Sheth S, Collins D, Schmidt G, Teller J. 2001a. Effects of glacial meltwater in the GISS coupled
atmosphere–ocean model — 1. North Atlantic deep water response. Journal of Geophysical Research 106: 27 335–27 353.
Rind D, Russell G, Schmidt G, Sheth S, Collins D, deMonocal P, Teller J. 2001b. Effects of glacial meltwater in the GISS coupled
atmosphere–ocean model — 2. A bipolar seesaw in Atlantic deep water production. Journal of Geophysical Research 106:
27 355–27 365.
Rodwell MJ, Folland CK. 2002. Atlantic air–sea interaction and seasonal predictability. Quarterly Journal of Royal Meteorological
Society 128: 1413–1443.
Rohling EJ. 1994. Review and new aspects concerning the formation of eastern Mediterranean sapropels. Marine Geology 122: 1–28.
Rosenfeldt D, Lahav R, Khain A, Pinsky M. 2002. The role of sea spray in cleansing air pollution over the ocean via cloud processing.
Science 297: 1667–1670.
Russell GL, Rind D. 1999. Response to CO2 transient increase in the GISS coupled model: regional coolings in a warming climate.
Saji NH, Goswami BN, Vinayachandran PN, Yamagata T. 1999. A dipole mode in the tropical Indian Ocean. Nature 401: 360–363.
Saltzman ES, Whung PY, Mayewski PA. 1997. Methanesulfonate in the Greenland Ice Sheet Project 2 Ice Core. Journal of Geophysical
Research 102: 26 649–26 657.
Saravanan R, Chang P. 2000. Interaction between tropical Atlantic variability and El Niño–southern oscillation. Journal of Climate 13:
2177–2194.
Schiller A, Mikolajewicz U, Voss R. 1997. The stability of the North Atlantic thermohaline circulation in a coupled ocean–atmosphere
general circulation model. Climate Dynamics 13: 325–347.
Schmittner A, Appenzeller C, Stocker T. 2000. Enhanced Atlantic freshwater export during El Nino. Geophysical Research Letters 27:
1163–1166.
Schmitz WJ. 1995. On the inter-basin thermohaline circulation. Reviews of Geophysics 33: 151–173.
Seager R, Clement AC, Cane MA. 2000. Glacial cooling in the tropics: exploring the roles of tropospheric water vapour, surface wind
speed, and boundary layer processes. Journal of the Atmospheric Sciences 57: 2144–2157.
Seitzinger SP, Kroeze C. 1998. Global distribution of nitrous oxide production and N inputs in freshwater and coastal marine ecosystems.
Global Biogeochemical Cycles 12: 93–113.
Senior CA, Mitchell JFB. 2000. The time-dependence of climate sensitivity. Geophysical Research Letters 27: 2685–2688.
Servain J, Wainer I, Ayina HL, Roquet H. 2000. The relationship between the simulated climatic variability modes of the tropical
Atlantic. International Journal of Climatology 20: 939–953.
Smet E, Lens P, van Langenhove H. 1998. Treatment of waste gases contaminated with odorous sulphur compounds. Critical Reviews
in Environmental Science Technology 28: 89–117.
Smethie WM, Takahashi T, Chipman DW. 1985. Gas exchange and CO2 flux in the tropical Atlantic Ocean determined from Rn-222
and pCO2 measurements. Journal of Geophysical Research 90: 7005–7022.
Spence GD, Chapman NR. 2001. Fishing trawler nets massive “catch” of methane hydrates. EOS Transactions of the American
Geophysical Union 82(50): 621–627.
Stocker TF, Marchal O. 2000. Abrupt climate change in the computer: is it real? Proceedings of the National Academy of Sciences of
the United States of America 97: 1362–1365.
Stocker TF, Schmittner A. 1997. Influence of CO2 emission rates on the stability of the thermohaline circulation. Nature 388: 862–865.
Stocker TF, Clarke GRC, Le Treut H, Lindzen RS, Meleshko VP, Mugara RR, et al. 2001. Physical climate processes and feedbacks.
In Climate Change 2001: The Scientific Basis, Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, Maskell K,
Stouffer RJ, Manabe S. 1999. Response of a coupled ocean–atmosphere model to increasing atmospheric carbon dioxide: sensitivity to
the rate of increase. Journal of Climate 12: 2224–2237.
Sun S, Bleck R. 2001. Atlantic thermohaline circulation and its response to increasing CO2 in a coupled atmosphere–ocean model.
Geophysical Research Letters 28: 4223–4226.
Sutton RT, Jewson SP, Rowell DP. 2000. The elements of climate variability in the tropical Atlantic region. Journal of Climate 13:
3262–3284.
Takahashi T, Feely RA, Weiss RF, Wanninkhof RH, Chipman DW, Sutherland SC, Takahashi TT. 1997. Global air–sea flux of CO2 :
an estimate based on measurements of sea–air pCO(2) difference. Proceedings of the National Academy of Sciences of the United
States of America 94: 8292–8299.
Tegen I, Lacis AA, Fung I. 1996. The influence on climate forcing of mineral aerosols from disturbed soils. Nature 380: 419–422.
Thorpe RB, Bigg GR. 2000. Modelling the sensitivity of Mediterranean outflow to anthropogenically-forced climate change. Climate
Dynamics 16: 355–368.
Thorpe RB, Gregory JM, Johns TC, Wood RA, Mitchell JFB. 2001. Mechanisms determining the Atlantic thermohaline circulation
response to greenhouse gas forcing in a non-flux-adjusted coupled climate model. Journal of Climate 14: 3102–3116.
Timmermann A, Latif M, Voss R, Grotzner A. 1998. Northern hemispheric interdecadal variability: a coupled air–sea model. Journal
of Climate 11: 1906–1931.
Turner DR, Hunter KA (eds). 2001. The Biogeochemistry of Iron in Seawater. Wiley: Chichester.
Turner SM, Nightingale PD, Spokes LJ, Liddicoat MI, Liss PS. 1996. Increased dimethyl sulphide concentrations in seawater from
in situ iron enrichment. Nature 383: 513–517.
Turrell WR, Slesser G, Adams RD, Payne R, Gillibrand PA. 1999. Decadal variability in the composition of Faroe Shetland Channel
bottom water. Deep-Sea Research Part I: Oceanographic Research Papers 46: 1–25.
Uher G, Andreae MO. 1997. Photochemical production of carbonyl sulphide in the North Sea water: a process study. Limnology and
Upstill-Goddard RC, Barnes J, Frost T, Punshon S, Owens NP. 2000. Methane in the southern North Sea: low-salinity inputs, estuarine
removal and atmospheric flux. Global Biogeochemical Cycles 14: 1205–1217.
Van Kreveld S, Sarnthein M, Erlenkauser H, Grootes P, Jung S, Nadeau MJ, Pflaumann U, Voelker A. 2000. Potential links between
surging ice sheets, circulation changes and the Dansgaard–Oeschger cycles in the Irminger Sea, 60–18 kyr. Paleoceanography 15:
425–442.
Vellinga M, Wood R. 2002. Global climatic impacts of a collapse of the Atlantic thermohaline circulation. Climatic Change 54: 251–267.
Veum TE, Jansen E, Arnold M, Beyer I, Duplessey JC. 1992. Water mass exchange between the North Atlantic and the Norwegian Sea
during the last 28 000 years. Nature 356: 783–785.
Vink A, Ruhlermann C, Zonneveld KAF, Mulitza S, Huls M, Willems H. 2001. Shifts in the position of the North Equatorial Current
and rapid productivity changes in the western tropical Atlantic during the last glacial. Paleoceanography 16: 479–490.
Wadley MR, Bigg GR. 2000. Implementation of variable time stepping in an ocean general circulation model. Ocean Modelling 1:
71–80.
Wadley MR, Bigg GR. 2002. Impact of flow through the Canadian Archipelago on the North Atlantic and Arctic thermohaline circulation:
an ocean modelling study. Quarterly Journal of the Royal Meteorological Society 128: 2187–2203.
Wadley MR, Bigg GR, Rohling EJ, Payne AJ. 2002. On modelling present day and last glacial maximum oceanic δ 18 O distributions.
Global and Planetary Change 32: 89–109.
Wanninkhof R. 1992. Relationship between wind speed and gas exchange over the ocean. Journal of Geophysical Research 97:
7373–7382.
Wanninkhof R, Hitchcock G, Wiseman WJ, Vargo G, Ortner PB, Asher W, Ho DT, Schlosser P, Dickson ML, Masserini R, Fanning K,
Zhang JZ. 1997. Gas exchange, dispersion, and biological productivity on the west Florida shelf: results from a Lagrangian tracer
study. Geophysical Research Letters 24: 1767–1770.
Warner MJ, Roden GI. 1995. Chlorofluorocarbon evidence for recent ventilation of the deep Bering Sea. Nature 373: 409–412.
Watson AJ. 2001. Iron limitation in the oceans. In The Biogeochemistry of Iron in Seawater, Turner DR, Hunter KA (eds). Wiley:
Chichester; 9–39.
Watson AJ, Liss PS. 1998. Marine biological controls on climate via the carbon and sulphur geochemical cycles. Philosophical
Transactions of the Royal Society of London Series B 353: 41–51.
Watson AJ, Orr JC. In press. Carbon dioxide fluxes in the global ocean. In Ocean Biogeochemistry: A JGOFS Synthesis, Fasham M,
Field J, Platt T, Zeitzschel B (eds). Springer-Verlag: Berlin.
Webster PJ, Moore AM, Loschnigg JP, Leben RR. 1999. Coupled ocean–atmosphere dynamics in the Indian Ocean during 1997–98.
Nature 401: 356–360.
Weijer W, De Ruijter WP, Dijkstra HA. 2001. Stability of the Atlantic overturning circulation: competition between Bering Strait
freshwater flux and Agulhas heat and salt sources. Journal of Physical Oceanography 31: 2385–2402.
Weinelt M, Sarnthein M, Pflaumann U, Schulz H, Jung S, Erlenkeuser H. 1996. Ice-free Nordic seas during the last glacial maximum?
Potential sites of deepwater formation. Palaeoclimates 1: 283–309.
White WB, Peterson RG. 1996. An Antarctic circumpolar wave in surface pressure, wind, temperature and sea-ice extent. Nature 380:
699–702.
Wigley TML, Raper SCP. 1991. Internally generated natural variability of global-mean temperatures. In Greenhouse-Gas-Induced
Climatic Change: A Critical Appraisal of Simulations and Observations, Schlesinger ME (ed.). Elsevier: Amsterdam; 471–482.
Xu W, Barnett TP, Latif M. 1998. Decadal variability in the North Pacific as simulated by a hybrid coupled model. Journal of Climate
11: 297–312.
Yin JH, Battisti DS. 2001. The importance of tropical sea surface temperature patterns in simulations of last glacial maximum climate.
Yu EF, Francois R, Bacon MP. 1996. Similar rates of modern and last-glacial ocean thermohaline circulation inferred from radiochemical
data. Nature 379: 689–694.
Yuan XJ, Martinson DG. 2001. The Antarctic dipole and its predictability. Geophysical Research Letters 28: 3609–3612.
Zebiak SE. 1993. Air–sea interaction in the equatorial Atlantic region. Journal of Climate 6: 1567–1586.
Zeebe RE, Wolf-Gladrow D. 2001. CO2 in Seawater: Equilibrium, Kinetics and Isotopes. Elsevier Oceanographic Series, Elsevier:
Amsterdam.

Joc 926 PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Joc 926 PDF

Hochgeladen von

Copyright:

Verfügbare Formate

INTERNATIONAL JOURNAL OF CLIMATOLOGY

Int. J. Climatol. 23: 1127–1159 (2003)

Copyright  2003 Royal Meteorological Society

1.1. Sea-level change

1.2. Time scales and variability

HADCM3-CTL LOG10((20 – 50 YEARS) / (2 – 8 YEARS))

20-year variance / AR (1) variance

2.1. Gas exchange

2.2. Carbon dioxide

2.3. Methane and nitrous oxide

2.4. Sulphur gases

Cloud formation Increased Increased cloud

2.5. Halogen and non-methane hydrocarbons

Sep., Oct., Nov

2.7. Interlinking cycle

3. ATMOSPHERE–OCEAN INTERACTION INVOLVING THE WIND-DRIVEN CIRCULATION

3.1. Tropical air–sea coupling in the Atlantic and Indian Oceans

Upwelling Downwelling Upwelling

Upwelling Upwelling Upwelling

3.2. Variability involving wind-driven circulation in the extra-tropics

4. THE THERMOHALINE CIRCULATION

4.1. Deep-water formation and the conveyor belt theory

4.2. Freshwater fluxes and state change

4.3. Thermohaline catastrophes

4.4. The potential impact of the Mediterranean outflow on climate

4.5. Changes to the thermohaline circulation due to anthropogenic change

Das könnte Ihnen auch gefallen