On the Fundamental Preference for Atlantic Overturning as Demonstrated by an Idealised Reverse World Experiment

Abstract

The thermohaline circulation (THC) is an important component of the climate system which transports large amounts of heat northwards in the Atlantic sector. It is still unclear why there is no counterpart in the Pacific, and whether this absence is coincidental or fundamental. We investigated the relative importance of various geographical asymmetries in forcing an Atlantic preference in a coupled climate model using a novel method in which the direction of rotation of the Earth was reversed. We found that the Atlantic THC collapsed in the reverse rotation experiment, and a THC started up in the Pacific, and that the response was forced by changes in the zonal and meridional steric gradients resulting from changed freshwater flux forcing. In particular, migration of the precipitation maximum from the Maritime Continent region in the reverse experiment allows a build-up of salinity to develop in the subtropical Pacific, which on advection north-east leads to the development of a meridional steric gradient and facilitates deep convection. Meanwhile, in the Atlantic, eastwards water vapour export into Africa increases massively, and the subtropical Atlantic gets more saline, but the North Atlantic and Mediterranean both freshen substantially, and this eventually over-rides the low latitude salinity increase, and collapses the Atlantic THC. The climate system appears to have a fundamental preference for Atlantic overturning over Pacific overturning, driven by the interaction of geographical asymmetry of ocean basins and their associated freshwater catchments with the net freshwater forcing. The export of water vapour across Panama in the current climate favours Atlantic overturning, but is not the sole determinant, and it is clear that the literature focus on Atlantic to Pacific freshwater export is too simplistic.

Full text

Page 1 of 22
On the Fundamental Preference for Atlantic Overturning as Demonstrated by an Idealised
Reverse World Experiment
Robert B. Thorpe (UK Met Office), FitzRoy Road, Sowton, Exeter, Devon, UK
Met Office Internal Report, November 2005
Abstract
The thermohaline circulation (THC) is an important component of the climate system which transports
large amounts of heat northwards in the Atlantic sector. It is still unclear why there is no counterpart
in the Pacific, and whether this absence is coincidental or fundamental. We investigated the relative
importance of various geographical asymmetries in forcing an Atlantic preference in a coupled climate
model using a novel method in which the direction of rotation of the Earth was reversed. We found
that the Atlantic THC collapsed in the reverse rotation experiment, and a THC started up in the Pacific,
and that the response was forced by changes in the zonal and meridional steric gradients resulting
from changed freshwater flux forcing. In particular, migration of the precipitation maximum from the
Maritime Continent region in the reverse experiment allows a build-up of salinity to develop in the
subtropical Pacific, which on advection north-east leads to the development of a meridional steric
gradient and facilitates deep convection. Meanwhile, in the Atlantic, eastwards water vapour export
into Africa increases massively, and the subtropical Atlantic gets more saline, but the North Atlantic
and Mediterranean both freshen substantially, and this eventually over-rides the low latitude salinity
increase, and collapses the Atlantic THC. The climate system appears to have a fundamental
preference for Atlantic overturning over Pacific overturning, driven by the interaction of geographical
asymmetry of ocean basins and their associated freshwater catchments with the net freshwater
forcing. The export of water vapour across Panama in the current climate favours Atlantic overturning,
but is not the sole determinant, and it is clear that the literature focus on Atlantic to Pacific freshwater
export is too simplistic.
Introduction
The thermohaline circulation (THC) is an important component of the climate system, both in terms
of the large amount of heat it transports to high northern latitudes (Broecker, 1991), and in terms of
its notable asymmetry in both latitude and longitude, with a marked preference for overturning in the
northern hemisphere (northern sinking as termed by Marotzke and Willebrand (1991)) which is
concentrated in the Atlantic basin. Warren (1983) drew attention to this Atlantic focus which results
in large differences in the magnitude and nature of meridional heat transport in the Atlantic and
Pacific. In the Pacific basin, 75% of the meridional heat transport is accounted for by the shallow gyre
transport, whereas in the Atlantic, a similar percentage is associated with the formation of deep water
and the related meridional overturning (Talley, 1999). In the current climate, the meridional
overturning is responsible for transporting large amounts of heat northwards to high latitudes in the
Atlantic sector, and the temperate climate of the North Atlantic sector depends at least partly upon
the heat supplied by the THC. For example Bergen, Norway (60 degrees N, 5 degrees E, 44m above
sea level) has a mean January temperature of 1 degree C, and a mean July temperature of 15 degrees
C, whilst the corresponding figures for Juneau, Alaska (58 degrees N, 134 degrees W, 8m above sea
level) are -3 and 14 degrees C.

Page 2 of 22
Evidence from various paleoclimate indicators suggests that rapid climate changes have occurred in
the geologically recent past in the Atlantic sector (Broecker et al., 1985; Severinghaus and Brook,
1999). Indeed the relative stability of the Holocene climate seems to have been atypical in the context
of the much longer period covered by ice core records (Petit et al., 1999). Conversely, the Pacific sector
appears to have been much less variable on centennial and millennial timescales, and this may be
related to the possible absence of deepwater formation at any time during the Quaternary (Keigwin,
1987). Hence the vulnerability of the Atlantic region to abrupt climate change in the future appears
higher than that of the Pacific, and tied up with the asymmetric character of the present-day THC
(Duplessy et al., 1992; Jouzel et al., 1995).
Warren (1983) noted that there was a striking difference between the near surface salinities in the
North Atlantic and North Pacific, with North Pacific water being sufficiently fresh that even when
cooled to freezing, it is only able to sink to about 150m (Reid, 1973), whereas the more saline water
in the North Atlantic is able to sink to great depth in the Nordic (Meinke et al., 1997), Irminger (Bacon
et al., 2003), and Labrador Seas (Krahmann et al., 2003).
Warren (1983) and Emile-Geay (2003) suggested that the North Atlantic was much more saline than
the North Pacific because of higher evaporation rates in the former basin, and greater isolation of the
North Pacific from lower latitude water masses caused by changes in the wind stress forcing in the
two basins. However, as Warren (1983) conceded, neither of these factors is really independent of the
existence of the thermohaline circulation in the Atlantic (Warren, 1983; Hughes and Weaver, 1994).
The higher evaporation in the North Atlantic is at least partly a consequence of the warmer water in
the Atlantic, and this is associated with the heat transported by the meridional overturning, whilst the
winds stress forcings are also materially affected by the presence of the overturning. It therefore
remains unclear whether the absence of deep water formation in the North Pacific is coincidental
(being a function of the previous history of the system) as suggested by Stommel and Arons (1960), or
fundamental.
There are obvious geographical differences between the Atlantic and Pacific basins that could lead to
a fundamental behavioural asymmetry. Firstly, the Atlantic extends much further north, and has a
deeper connection with the Arctic. Secondly, there is a net transport of freshwater between the
Atlantic and Pacific. The narrow and relatively low-lying Isthmus of Panama allows the export of water
vapour from Atlantic to Pacific via the North-East Trades (Weyl, 1968; Zaucker et al., 1994; Latif et al.,
2000), but the Pacific to Atlantic freshwater transport is blocked by the Rockies and Andes. Thirdly,
the Pacific has a much greater longitudinal extent than the Atlantic, and Schmitt et al. (1989) have
suggested that this causes the latter to be more susceptible to the incursions of cold, dry continental
air that would encourage evaporation and heat loss. Fourthly, salty outflow from the Mediterranean
can contribute to the high salinity of near surface water in the North Atlantic (Reid, 1979), whereas
no counterpart to this exists in the Pacific. And fifthly, Reid (1961) has suggested that the greater
southward extent of South America relative to South Africa may act to impede the transport of
freshwater from Pacific to Atlantic via the Antarctic Circumpolar Current (ACC).
In the limit of identical geometry and surface forcing, there would be no preference for thermohaline
overturning in either basin (or if zonally symmetric either hemisphere), and any differences would be
due to the prior history of the system. The extent to which the various asymmetries contribute to a
preference for overturning in the Atlantic has been explored in a range of modelling studies. Marotzke
and Willebrand (1991) used an idealised ocean general circulation model (OGCM) with identical
“Atlantic” and “Pacific” basins, connected by a southern circumpolar channel, and forced by idealised
freshwater flux profiles with peak evaporation at the equator. This favoured deep water formation in
the northern part of both basins, a preference apparently forced by north-south geographical

Page 3 of 22
asymmetry. Hughes and Weaver (1994) extended this study by exploring the effects of some of the
real-world longitudinal asymmetries through widening the Pacific basin, extending the Atlantic further
north, and adding a sill at the southern connection between the basins (representing the Drake
Passage). These changes had the effect of strongly biasing the model in favour of North Atlantic
overturning (though North Pacific overturning could still happen if the water mass structure in the
Southern Ocean was strongly modified), and suggests that the basin asymmetries are collectively
significant, though Hughes and Weaver (1994) were not able to determine the extent to which each
factor might contribute individually.
Here we report the results of an experiment that addresses the issue in a novel light, using a coupled
climate model (Jones et al., 2005) in which the direction of rotation of the Earth is abruptly reversed.
The rationale for this is that some of the geographical features are invariant under rotation (the
relative width and extent of the basins), whereas one would expect the impact of the Drake Passage
(due to reversal of the ACC), relative southwards extent of South America and South Africa, and water
vapour export across Panama to be reversed, with the impact on Mediterranean outflow being unclear
a priori. The experiment also allows for an investigation of the extent to which the previous history of
the system (implied by initialisation from Levitus and Boyer (1994) temperature, and Levitus et al.
(1994) salinity) determines its evolution under radically altered boundary conditions.
Experimental Design
We use a low-resolution version of HadCM3, based on the FAMOUS ocean (Jones, 2003) coupled to a
5o x 7o atmosphere, and tuned to be as similar as possible to HadCM3, using the methodology of Jones
et al. (2005). The version we use here has the identifier ADLYA, and differs from the 4th phase control
of Jones et al. (2005) – identifier ADTAA – having RHcirt of 0.636 rather than 0.714, and with slightly
different large scale precipitation parameters. Henceforth, in this paper, FAMOUS refers to variant
ADLYA. The FAMOUS model is used because the long timescales involved in adjustment of the deep
ocean to the imposition of the reverse rotation mean that HadCM3 is too slow in the context of
currently available computing resource. The climate of variant ADTAA is described in detail elsewhere
(Jones et al., 2005). Here we show that the near surface temperature, meridional heat transports, and
freshwater budget for variant ADLYA to give the reader an idea of the skill of this model. Though the
model skill is less than that of HadCM3, it exceeds that of the unfluxadjusted HadCM2, and is much
faster than the latter, so it represents a good compromise between speed and skill.
Figure 1 shows the near surface temperatures for a) FAMOUS and b) HadCM3. FAMOUS captures the
general patterns to first order, and the differences at low latitudes are generally small, except over
the Sahara. However, most of the area south of 60oS and north of 50oN is far too cold in FAMOUS,
especially in the Atlantic sector, where differences in excess of 10oC occur quite widely. This may be
the result of differences in the heat transport between the two models (Figure 2). Though Northern
Hemisphere heat transports are very similar to HadCM3, the ocean heat transports are consistently
lower in FAMOUS, especially in the Atlantic sector at mid to high latitudes. This is associated with a
somewhat weaker Atlantic meridional circulation at these latitudes (Figure 3) which is a consequence
of the shallower topography resulting from reduced ocean resolution in FAMOUS.
Several studies have suggested that the Atlantic freshwater budget, and the latitudinal distribution of
this budget may be important determinants of the sensitivity of the thermohaline circulation. Figure
4 shows the Atlantic freshwater budget (precipitation + run off – evaporation) for FAMOUS and
HadCM3. There is greater net precipitation at high northern latitudes in FAMOUS (on the face of it this

Page 4 of 22
is a surprise, as the storm tracks are weaker in the coarser resolution atmosphere, but it is due to
reduced evaporation from the cooler ocean in FAMOUS), and the equatorial precipitation maximum
is broader and displaced to the north.
Despite these differences in surface climatology and heat transports, the sensitivity of the model’s
thermohaline circulation to anthropogenic forcing and freshwater hosing has been tested and found
to be similar to HadCM3.
The Timescale of Adjustment to the Reverse Rotation
The various parts of the climate system react to the imposition of the reverse rotation on different
timescales. The atmospheric circulation largely adjusts to the new regime within the first month, as
do the zonal SST gradients, and the deep boundary currents migrate from the west to the east of the
ocean basins within the first decade. As a result of the initial atmospheric circulation and associated
SST and convection changes, the Atlantic overturning strength at 55oN declines by about 50% to 5 Sv
in the first century or so (Figure 5). This is followed by several centuries of much slower decline, in
parallel with the control simulation. The Atlantic THC undergoes a brief reversal between years 430
and 450, then recommences for about 50 years before reversing permanently after about 500 years.
There is effectively no Pacific overturning at 55oN in either the control or reverse rotation, until around
year 300, when high latitude Pacific overturning commences in the reverse experiment, accompanied
by high latitude warming and the onset of deep convection in the Pacific sector as the accumulated
forcing from the altered surface boundary conditions triggers a large-scale oceanic re-organisation.
The initial impact of the reverse rotation on near surface temperatures, and its modification by the
subsequent oceanic re-organisation can be seen in Figure 6, which compares the average near surface
temperature change for the first century, and for years 500-600, by which time the Atlantic THC has
collapsed, and a strong circulation has established itself in the Pacific. As can be seen from Figure 6a,
the initial adjustment is dramatic, especially in Western Europe, which cools by around 10oC, and in
Eastern North America, Siberia, Japan, and China, which warm by similar amounts. All these changes
would be expected on the basis of shifts in the prevailing wind patterns, but more surprising is the
warming of the Canadian Pacific coast and Alaskan Panhandle, when simple consideration of the likely
circulation change would suggest that this region should cool strongly. Thus the response to reverse
circulation is not as simple as swapping east and west coasts, or turning Victoria into Vladivostok.
Figure 6b shows the impact of collapsing the Atlantic THC and starting one in the Pacific on the near
surface temperature. The whole North Pacific region warms strongly as a result of the commencement
of deep convection, and the rest of the Northern Hemisphere also tends to warm slightly, though this
is offset over North America, and in the North Atlantic, where the collapse of the Atlantic THC leads
to a local cooling. The Southern Hemisphere cools generally, with the cooling being pronounced in the
circum-Antarctic region. Thus the Pacific THC appears to be less efficient at transferring heat from
Southern to Northern Hemispheres than that of the Atlantic.
Thermohaline Circulation (THC) Response and Depth-Integrated Steric Gradient
As a consequence of the changed boundary conditions brought about by the reversal in direction of
rotation, the thermohaline circulation moves from Atlantic to Pacific basins. How does this come
about? Assuming geostrophy and a linear relationship between the zonal and meridional steric
gradients implies that there ought to be a linear relationship between the meridional pressure

Page 5 of 22
gradient and the strength of the overturning (Park, 1999), as found by Hughes and Weaver (1994),
Rahmstorf (1996), and by Thorpe et al. (2001) in HadCM3, and Thorpe (2005) in HadCM2. We
investigated whether a change in density gradients was responsible for the change in overturning seen
in this experiment. Figure 7 shows the time evolution of maximum overturning north of 40oN,
maximum surface air temperature at 50-65oN, maximum depth of the mixed layer, and depth
integrated steric height gradient (from 3000m to the surface) for the Atlantic and Pacific basins.
It is clear that all four quantities in both Atlantic and Pacific basins evolve together, with the changes
in ocean circulation being associated with a decrease in steric gradient, vigour of mixing, and heat loss
to the atmosphere in the Atlantic sector, with the reverse happening in the Pacific. An initial change
in the depth-integrated steric height gradient appears to lead the other changes in the Pacific.
Figure 8 shows the relationship between the meridional overturning strength, and the zonal and
meridional depth integrated steric gradients in the Atlantic and Pacific basins. Following the initial
adjustment phase lasting around a century (triangles in Figure 8), where the Atlantic THC weakens
with the zonal density gradient, but without a corresponding change in the meridional density
gradient, results are consistent with the concept of geostrophic balance and inter-relation of zonal
and meridional pressure gradients in which the quantities are linearly related. The clear implication is
that the “collapse” of the Atlantic THC is driven by the disappearance of favourable pressure gradients
in the Atlantic basin, consistent with the first theoretical framework of Park (1999) and as found by
Hughes and Weaver (1994) and Thorpe et al. (2001).
Although the relationship between the overturning and the meridional steric gradient is less obviously
linear for the Pacific, there is still a clear link between the two, whilst the zonal and meridional
gradients remain tightly coupled, suggesting that the same theoretical framework broadly applies to
both the spin up in the Pacific and the collapse in the Atlantic.
Impact of Changes in Surface Flux Forcing on the Steric Gradient
The elimination of the Atlantic steric gradients and development of those in the Pacific result from
changes in the distribution of salinity; the thermal contributions from a warmer North Pacific and
colder North Atlantic attempt to operate in the opposite direction, but is over-ridden. The salinity
changes are caused by changes in the transport of freshwater by the reverse-world atmospheric
circulation. Atmospheric transports of freshwater, and the associated atmospheric freshwater
divergence (i.e. evaporation minus precipitation) are shown for the control and reverse rotation case
(Figure 9).
The control pattern (Figure 9a) shows net evaporation in the subtropics, especially towards the
eastern side of the oceans (except in the Indian Ocean), and net precipitation over tropical land,
particularly over the Maritime Continent (Indonesia) and north-east of Australia. The reverse rotation
case (Figure 9b) shows the same general pattern of net precipitation in the tropics and at mid-
latitudes, and net evaporation in the subtropics, but the longitudinal distribution is different as a result
of the global reversal of typical wind directions. The net evaporation maxima are now generally shifted
to the west of the ocean basins (including in the Indian Ocean), and the primary precipitation
maximum has been shifted eastwards, so that it now occurs in the east Pacific around 120oW. The
precipitation maxima over South America and Africa are pushed west, the former being weakened
and the latter enhanced so that it becomes comparable in strength with that in the Pacific.

Page 6 of 22
These changes in the global freshwater budget act on the ocean catchment areas to radically affect
the ocean freshwater budgets, and hence the global thermohaline circulation. Figure 10 shows the
global freshwater budget, partitioned into Atlantic/Arctic, Pacific, Indian, Mediterranean, Hudson Bay,
and Black Sea/Aral Sea catchment areas. The latter drains internally in HadCM3 and FAMOUS, but the
Hudson Bay and Mediterranean catchments are connected to the Atlantic. Somewhat surprisingly, the
Pacific actually exports freshwater to the Atlantic in the control, with the import across Central
America more than being offset by freshwater export in the mid-latitudes. With the direction of most
of the transports reversing as expected in the reverse world experiment, the Atlantic budget actually
becomes more saline, though gains from the Pacific are largely offset by exports to the Indian Ocean,
with an additional amount being lost to the Hudson Bay region. The Indian Ocean budget becomes
broadly neutral, whilst the Hudson Bay area is subject to strong freshening. On their own, these
changes would tend to stabilise the Atlantic THC, but of course the Atlantic cannot be viewed in
isolation from the Mediterranean and Hudson Bay areas, as all three regions are directly connected.
Taking all three catchments together, net freshening increases from 0.12Sv to 0.4Sv, with most of it
concentrated in the North Atlantic, near the sites of deep water formation in the control climate. The
marked impact of the fresher Mediterranean Outflow Water (MOW) can be seen in Figure 11, as it
spreads out across the North Atlantic, freshening the intermediate waters, and removing the pre-
conditioning referred to by Reid (1979). Over time, this contributes to the collapse of the Atlantic THC
when deep water formation and the required steric gradient can no longer be sustained. The picture
in the Pacific is more straightforward. The Pacific exports freshwater to the Atlantic (the import across
Panama so prominent in the literature (Latif et al., 2000; Thorpe et al., 2001) notwithstanding), but
gains even more from the Indian basin, and is a large net recipient of freshwater in the control. This
situation changes in the reverse rotation case, with imports from the Atlantic now more than offset
by the very large export to the Indian Ocean, so the Pacific becomes evaporative to the tune of more
than 0.6 Sv (a change of over 1 Sv relative to the control), resulting in a significant build-up of salinity,
which on advection north leads to the establishment of a THC.
Discussion and Conclusion
The absence of deep water formation in the Pacific is an interesting question which has been
addressed by many authors (e.g. Warren, 1983; Hughes and Weaver, 1994; Emile-Geay et al., 2003),
though it is still not entirely clear whether this is fundamental to the climate system, and if so what
properties of the real world may force this asymmetry. The presence of an open southern connection,
and the associated wind forcing appear to enforce a preference for northern sinking, and it is
reasonable to suppose that geographical asymmetry likewise enforces a preference for Atlantic
overturning at the expense of the Pacific, but how it might do this has remained unclear.
Five asymmetries have been hypothesised as being important, namely a greater northwards
latitudinal extent, lesser longitudinal extent, Atlantic to Pacific freshwater transport across Panama,
the greater southern extension of South America than South Africa, and the contribution of salty MOW
to preconditioning the Atlantic for deep convection. We investigated the relative importance of these
asymmetries in a coupled climate model using a novel method in which the direction of rotation of
the Earth was reversed. We found that the Atlantic THC collapsed in the reverse rotation experiment
and a THC started in the Pacific, Examination of the meridional and zonal depth-integrated steric
gradients in both basins showed that the two were tightly coupled, consistent with the first theoretical
picture of Park (1999), and suggested that the THC response was being governed by these steric
gradients. Hence the Atlantic THC collapses and a Pacific THC develops because changes in the density

Page 7 of 22
structure of the ocean under reverse rotation eliminates the Atlantic steric gradients, and allow them
to develop in the Pacific.
The re-organisation of the THC in the experiment demonstrates that the effect of the initial conditions
can be over-ridden by changing the boundary conditions and strongly suggests that the preference for
an Atlantic THC is fundamental to the climate system, and not simply a product of recent history. The
experiment also suggests that those boundary conditions that are invariant under the rotation
transformation (open Bering Strait, greater northward extent and lesser longitudinal extent of the
Atlantic) are most probably not responsible for this preference. Because the changes in density
structure appear to be forced primarily by the atmospheric changes, with the ocean responding to
this forcing (Figure 12), that tends to suggest that the asymmetries connected with the ACC, the extent
of South America, and the directionality of Drake Passage flow, are less likely to be causally linked to
the Atlantic overturning, although this remains to be demonstrated conclusively through further
experimentation.
The prime cause of the ocean re-organisation therefore appears to be the surface freshwater flux
forcing. Migration of the precipitation maximum from the Maritime Continent region (Estoque 1982;
Neale and Slingo, 2003), and export of water vapour from Pacific to Atlantic in the reverse rotation
experiment allows a build-up of salinity to develop in the subtropical Pacific, which on advection north
leads to the development of a meridional steric gradient and facilitates deep convection. Meanwhile
in the Atlantic, although the import of freshwater across Panama is compensated for by export across
Africa, the freshening of the Mediterranean eventually feeds freshwater back into the North Atlantic
via less saline MOW, whilst there is also a strong injection of freshwater from the Hudson Bay region.
These changes lead to a reduction in the zonal and meridional steric gradient and deep convection,
and eventually a spin-down of the THC.
Assuming that all the relevant feedbacks are represented in a qualitatively reasonable manner, we
conclude from the experiment that the climate system has a fundamental preference for Atlantic
overturning over Pacific overturning, driven by the interaction of geographic asymmetry of ocean
catchments and net freshwater flux forcing. The export of water vapour across Panama has to be seen
in this context; it favours Atlantic overturning but is not the sole or even prime determinant. The
Maritime Continent precipitation maximum in the west Pacific and the salt supplied to the Atlantic via
the Mediterranean are also vital factors. The relative importance of these factors and the possible
impact of orography remains to be quantified, but it is clear that the literature focus on Atlantic to
Pacific freshwater export is too simplistic.
Acknowledgements
This work was supported by the UK Department for Environment, Food, and Rural Affairs under
contract PECD 7/12/37. Jose Rodriguez developed the freshwater diagnostics used in the analysis, and
the manuscript benefitted from discussions with Tim Johns, Jonathan Gregory, and Chris Jones, who
also led on the development of parameterisations used in the FAMOUS model framework.
References
Bacon, S., W.J. Gould, and Y.L. Jia. Open-ocean convection in the Irminger Sea. Geophysical
Research Letters, 30, Article No. 1246, 2003

Page 8 of 22
Broecker, W.S. The great ocean conveyor, Oceanography 4, 79-89, 1991
Broecker, W.S., D.M. Peteet, and D. Rind. Does the ocean-atmosphere system have more than one
stable mode of operation? Nature, 315, 21-26, 1985
Duplessy, J., M. Labeyrie, M. Arnold, M. Paterne, J. Duprat, and T.V. Weering. North Atlantic sea
surface salinity and abrupt climate changes, Nature, 358, 485-488, 1992.
Emile-Geay, J., M.A. Cane, N. Naik, R. Seager, A.C. Clement, and A. van Geen. Warren revisited:
atmospheric freshwater fluxes and “why is no deep water formed in the North Pacific?”. Journal of
Geophysical Research, 108, 10.1029/2001JC001058, 2003.
Estoque, M.A. Large-scale pressure disturbances over the Indonesian Maritime Continent. Journal of
the Meteorological Society of Japan. 60, 1143-1152, 1982
Hughes, T.M.C. and A.J. Weaver. Multiple equilibria of an asymmetric two-basin model. Journal of
Physical Oceanography, 24, 619-637, 1994
Jones, C.D. A fast ocean GCM without flux adjustments. Journal of Atmospheric and Oceanic
Technology, 20, 1857-1868, 2003
Jones, C.D., J.M. Gregory, R.B. Thorpe, P.M. Cox, J.M. Murphy, D.M.H. Sexton, and P. Valdes.
Systematic optimisation and climate simulation of FAMOUS, a fast version of HadCM3. Climate
Dynamics, 25, 189-204, 2005
Jouzel, J., R. Vaikmae, J. Petit, M. Martin, Y. Duclos, M. Stievenard, C. Lorius, M. Toots, M. Melieres,
L. Burckle, M. Barkov, and V. Koylatkov. The two-step shape and timing of the last deglaciation in
Antarctica. Climate Dynamics, 11, 151-161, 1995
Keigwin, L.D. North Pacific deep water formation during the last glaciation. Nature, 330, 362-364,
1987
Krahmann, G., M. Visbeck, W. Smethie, E.A. D’Asaro, P.B. Rhines, R.A. Clarke, J. Lazier, R.E. Davis, P.P.
Niiler, P.S. Guest, J. Meinke, G.W.K. Moore, R.S. Pickart, and W.B. Owens. The Labrador Sea Deep
Convection Experiment data collection. Geochemistry Geophysics Geosystems, 4, Article No. 1091,
2003
Latif, M., E. Roeckner, U. Mikolajewicz and R. Voss. Tropical stabilisation of the thermohaline
circulation in a greenhouse warming simulation. Journal of Climate, 13, 1809-1813, 2000
Levitus, S. and T.P. Boyer. World Ocean Atlas 1994, Volume 4: Temperature. NOAA/NESDIS E/OC21.
US Department of Commerce, Washington DC, 99pp, 1994
Marotzke, J. and J. Willebrand. Multiple equilibria of the global thermohaline circulation, Journal of
Physical Oceanography, 21, 1372-1385, 1991
Meinke, J., B. Rudels, and H.J. Friedrich. The Arctic Ocean Nordic Seas thermohaline system. ICES
Journal of Marine Science, 54, 283-299, 1997
Neale, R., and J. Slingo. The maritime continent and its role in the global climate: a GCM study.
Journal of Climate, 16, 834-848, 2003
Park, Y.-G. The stability of the thermohaline circulation in a two-box model. Journal of Physical
Oceanography, 29, 3101-3110, 1999

Page 9 of 22
Petit, J., J. Jouzel, D. Raynaud, N. Barkov, J.-M. Barnola, I. Basile, M. Bender, J. Chappellaz, M. Davis,
G. Delaygue, M. Delmotte, V. Kotlyakov, M. Legrand, V. Lipenkov, C. Lorius, L. Pepin, C. Ritz, E.
Saltzmann, and M. Stievenard. Climate and atmospheric history of the past 420,000 years from the
Vostok ice core, Antarctica. Nature, 359, 429-436, 1999
Rahmstorf, S. On the freshwater forcing and transport of the Atlantic thermohaline circulation,
Climate Dynamics, 12, 799-811, 1996
Reid, J.L. On the temperature, salinity, and density differences between the Atlantic and Pacific
oceans in the upper kilometre. Deep Sea Research, 7, 265-275, 1961
Reid, J.L. Northwest Pacific Ocean waters in winter. Technical Report 2, John Hopkins Oceanographic
Studies, 85pp, 1973
Reid, J.L. On the contribution of Mediterranean Sea outflow to the Norwegian-Greenland Sea. Deep
Sea Research, 26, 1199-1223, 1979
Schmitt, R.W., P.S. Bogden, and C.E. Dorman. Evaporation minus precipitation and density fluxes for
the North Atlantic. Journal of Physical Oceanography, 19, 1208-1221, 1989
Severinghaus, J., and E. Brook. Abrupt climate change at the end of the last glacial period inferred
from trapped air in polar ice. Science, 286, 930-934, 1999
Stommel, H. and A.B. Arons. On the abyssal circulation of the world ocean – II an idealised model of
the circulation pattern and amplitude in oceanic basins. Deep Sea Research, 6, 217-233, 1960
Talley, L.D. Some aspects of ocean heat transport by the shallow, intermediate, and deep
circulations, in Mechanisms of Global Climate Change at Millennial Timescales, edited by P.U. Clark,
R.S. Webb, and L.D. Keigwin, volume 112 of Geophysical Monographs Series, p1-22, AGU,
Washington DC, 1999
Thorpe, R.B. The impact of changes in atmospheric and land surface physics on the thermohaline
circulation response to anthropogenic forcing in HadCM2 and HadCM3. Climate Dynamics, 24, 449-
456, 2005
Thorpe, R.B. J.M. Gregory, T.C. Johns, R.A. Wood, and J.F.B. Mitchell. 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, 2001
Warren, B.A. Why is no deep water formed in the North Pacific? Journal of Marine Research, 41,
327-347, 1983
Weyl, P.K. The role of the oceans in climate change: a theory of the ice ages. Meteorological
Monographs, 8, 37-62, 1968
Zaucker, F., T.F. Stocker, and W.S. Broecker. Atmospheric freshwater fluxes and their effect on the
global thermohaline circulation. Journal of Geophysical Research, 99, 12443-12457, 1994

Page 10 of 22
Figure 1: Comparison of near surface (1.5m) atmospheric temperatures for a) FAMOUS, b) HadCM3,
and c) FAMOUS-HadCM3. Note the difference in scale, with annual mean plots a) and b) ranging
from -50oC to +30oC, and the difference ranging from -12 to +12 oC.

Page 11 of 22
Figure 2: Comparison of a) global and b) Atlantic FAMOUS and HadCM3 atmosphere and ocean heat
transports.

Page 12 of 22
Figure 3: Comparison of the average Atlantic meridional overturning streamfunction for years 260-
500 for a) FAMOUS, b) HadCM3, and c) FAMOUS-HadCM3. The circulation is similar to first order in
the two models, but is shallower and weaker in FAMOUS, particularly at mid to high latitudes as a
consequence of the more constricting topography.

Page 13 of 22
Figure 4: Comparison of the zonal Atlantic freshwater budget for a) FAMOUS and b) HadCM3.

Page 14 of 22
Figure 5: Time evolution of the overturning strength at 55oN in the Atlantic and Pacific basins in the
reverse rotation experiment.

Page 15 of 22
Figure 6: Average near surface (1.5m) temperature change for a) the first 100 years, and b) year 500-
600 minus year 0-100, showing the impact of THC shutdown in the Atlantic and spin-up in the
Pacific.

Page 16 of 22
Figure 7: Time evolution of a) maximum overturning strength, b) maximum 50-65oN near surface
temperature, c) maximum mixed layer depth, and d) north-south depth integrated steric gradient in
the Atlantic (top) and Pacific (bottom) basins.

Page 17 of 22
Figure 8: The relationship between a) maximum overturning strength and the depth-integrated
meridional steric gradient between 65oN and 55oS, b) the zonal and meridional steric gradients in the
Atlantic sector (to 55oS) and c) and d) the equivalent of a) and b) for the Pacific. The points marked
by triangles show the behaviour of the system in the first 100 years, when it is reacting to the
sudden imposition of reverse rotation.

Page 18 of 22
Figure 9: Net freshwater divergence for a) control, and b) reverse rotation. Units are 1.0e-5 Kg m-2 s-
1.

Page 19 of 22
Figure 10: Changes in net catchment freshwater budget for the control and reverse rotation during
the first 100 years.

Page 20 of 22
Figure 11: Changes in ocean salinity at 2100m in the Atlantic after 300 years. The signal of the
fresher Mediterranean Outflow Water (MOW) and its spread across the North Atlantic can clearly be
seen.

Page 21 of 22
Figure 12: The relative contribution of freshwater flux forcing from the atmosphere, and ocean
transport to changes in oceanic salinity. Values above 0.5 indicate that the atmospheric transport is
more important than the oceanic transport, and above 1 indicates that the atmospheric transport
effectively determines the response. Values below 0.5 indicate that the ocean transport is more
important, and below 0 indicate that the ocean transport effectively determines the response. The

Page 22 of 22
freshening of the Mediterranean and the Pacific is caused by the atmospheric forcing, the freshening
of the Atlantic is driven primarily by the ocean via Mediterranean Outflow Water.