Click
Here
GEOPHYSICAL RESEARCH LETTERS, VOL. 35, L09704, doi:10.1029/2008GL033236, 2008
for
Full
Article
North Atlantic Deep Water collapse triggered by a
Southern Ocean meltwater pulse in a glacial climate state
J. Trevena,1 W. P. Sijp,1 and M. H. England1
Received 11 January 2008; revised 13 March 2008; accepted 25 March 2008; published 7 May 2008.
[1] It is generally accepted that surface freshwater
anomalies in the Southern Ocean drive increases in North
Atlantic Deep Water (NADW) formation via a bipolar
density see-saw. We find that a Southern Ocean freshwater
pulse of comparable magnitude to meltwater pulse 1A,
shuts down, instead of strengthens, NADW in a glacial
climate simulation. Unlike the modern-day simulation, the
glacial experiment is associated with a more fragile North
Atlantic thermohaline circulation, whereby freshwater
anomalies that propagate into the North Atlantic are able
to dominate the bipolar density see-saw. Meltwater pulses
over the North Atlantic and subsequent NADW shut down
are often invoked to explain cold ‘Heinrich Events’
appearing in the paleoclimate record. Our results suggest
that triggers for NADW collapse may also originate from
the southern hemisphere in glacial epochs. Once NADW
collapses, North Pacific Deep Water develops, consistent
with a North Pacific/North Atlantic see-saw triggered from
the Southern Ocean. Citation: Trevena, J., W. P. Sijp, and
M. H. England (2008), North Atlantic Deep Water collapse
triggered by a Southern Ocean meltwater pulse in a glacial
climate state, Geophys. Res. Lett., 35, L09704, doi:10.1029/
2008GL033236.
1. Introduction
[2] Recent studies have shown the sensitivity of North
Atlantic Deep Water (NADW) formation to perturbations of
heat or freshwater (FW) over the Southern Ocean. For
example. Knorr and Lohmann [2003] showed that NADW
could be triggered from the ‘off’ to ‘on’ state by forcing a
glacial to non-glacial transition in sea surface temperature
(SST) over the Southern Ocean. Weaver et al. [2003] found
that a NADW ‘off’ to ‘on’ transition can be triggered if an
amount of FW roughly equivalent to meltwater pulse 1a
(MWP1a; occurring 14.6 ka before present (BP) and raising
global sea levels by 20 m) is applied to the Southern
Ocean. These results support the density see-saw hypothesis
first put forward by Bryan [1986], in which a density
decrease in one hemisphere favours deep overturning in the
opposite hemisphere. In the case of heating or freshening
over the Southern Ocean, the density of the Southern Ocean
decreases, favouring an increase in NADW formation.
[3] Studies continue to suggest that variations in NADW
are involved in past rapid climate change [e.g., McManus et
al., 2004]. The results of Weaver et al. [2003] and Knorr
1
Climate Change Research Centre, University of New South Wales,
Sydney, New South Wales, Australia.
Copyright 2008 by the American Geophysical Union.
0094-8276/08/2008GL033236$05.00
and Lohmann [2003] suggest that surface buoyancy
changes in the Southern Ocean could have contributed to
the last Northern Hemisphere deglaciation - via increased
northward oceanic heat transport when NADW is invigorated. This effect will be greatest in models when there is a
large difference between glacial and interglacial NADW
overturning. For example, Montenegro et al.’s [2007]
present-day NADW overturning is 21.6 Sv and their glacial
NADW overturning is 13.7 Sv, so a transition from glacial to
interglacial overturning requires an increase of just 7.9 Sv only about a third of the 21.6 Sv increase that would occur if
the initial glacial overturning was zero (as in work by
Weaver et al. [2003] and Knorr and Lohmann [2003]).
Moreover, there is much evidence to suggest that NADW
was only weakened rather than collapsed at the time of the
glacial period for which the most data are available - the
Last Glacial Maximum (LGM), 21 ka BP [e.g., LynchStieglitz et al., 2007].
[4] Compared to the present-day climate, glacial NADW
could have been more vulnerable to FW perturbations due
to weakened North Atlantic (NA) overturning. Furthermore,
glacial climates had cooler ocean temperatures over all
regions, not just over the NA from weakened or collapsed
NADW. In models this could tip the bipolar density see-saw
less in favour of NADW than if a cooling over just the NA
is considered. To test the sensitivity of a glacial climate
characterised by weak NADW formation to Southern Ocean
FW input, we will apply FW pulses to the Southern Ocean
in both modern-day and glacial climate simulations.
2. Model
[5] This study employs the University of Victoria Intermediate Complexity Climate (UVic) model [Weaver et al.,
2001], which comprises an ocean general circulation model
coupled to an energy-moisture balance model of the atmosphere, and a dynamic-thermodynamic model for sea-ice.
The atmospheric model includes advection and diffusion of
moisture and the sea-ice model includes prediction of the
ice thickness and ice area fraction in each grid cell. The
climate model is linked to prescribed land-ice that is static in
time, but can be set to values between the present-day and
21 ka BP. Full details of the model set up are available in
work by Trevena et al. [2008], though in this study we use
the recently incorporated tidal mixing scheme of Jayne and
St. Laurent [2001]. From both the modern-day and glacial
simulations, we apply FW to the region south of 63S,
equivalent to a 25 m global sea level rise over 500 years
(experiments CNTRLFW and GLACIALFW, respectively).
Experiments with smaller FW pulses, equivalent to global
sea level rises of 20 m and 10 m, produced similar results to
those described below.
L09704
1 of 6
L09704
TREVENA ET AL.: NADW COLLAPSE FROM SOUTHERN OCEAN
L09704
for an overview). Maximum SST cooling of over 5.0°C is
seen in the Northern Hemisphere north of 45°N, consistent
with PMIP2 simulations [Braconnot et al., 2007].
[8] Figure 2 shows the overturning streamfunction in
CNTRL and GLACIAL at steady state. NADW formation
weakens to 14.3 Sv in experiment GLACIAL (a net reduction of 23%), consistent with the intermediate NADW
strength during the LGM proposed by Lynch-Stieglitz
et al. [2007]. The formation areas also shift slightly
southward (Figure 2d). NPIW in GLACIAL strengthens
slightly to 2.1 Sv. AABW sinking decreases from 3.6 Sv to
2.7 Sv, whilst the abyssal cell barely changes, exhibiting an
equilibrium value of 11.6 Sv. The response of CNTRL and
GLACIAL to meltwater anomalies is discussed below in
Section 3.
3. Results
Figure 1. Simulated sea surface temperature (SST) difference from CNTRL for the GLACIAL simulation. Values are
based on a 1-year average. Contour interval is 0.5°C.
2.1. Modern Day Simulation
[6] The modern-day CNTRL experiment is a 3000-year
run forced with pre-industrial insolation and CO2 (280 ppm),
and pre-industrial land-ice. The equilibrium control state
generates approximately 17.6 Sv of NADW (Figure 2b), with
14.5 Sv exported across 32°S, close to the 17.9 ± 3.1 Sv
proposed by Lumpkin and Speer [2007]. North Pacific
Intermediate Water (NPIW) is weak (1.3 Sv) and shallow
(<500 m; Figure 2a), being the result of localised convection
of mode waters. Formation of AABW is characterised by a
cell of strength 3.6 Sv (not shown), corresponding to
overturning adjacent to Antarctica. Another measure of the
model’s AABW strength, which depends less on zonal
aliasing, is the net northward flow of bottom water in the
abyssal cell centred near 40°S. This cell has strength 11.7 Sv
in CNTRL.
2.2. Glacial Simulation
[7] For the GLACIAL experiment, land-ice is prescribed
to values estimated for 21 ka BP (based on the ICE-4G reconstruction of Peltier [1994]). Orbital configurations are
also set to 21 ka BP, and CO2 is set to 200 ppm and the
model is run for 3000 years to equilibrium (the actual net
CO2-equivalent is likely even less during the LGM). The
resulting glacial climate state has a global mean SST and
SAT cooling relative to CNTRL of 2.0°C and 3.1°C
respectively. Global SST differences between experiments
GLACIAL and CNTRL are shown in Figure 1. While our
GLACIAL run is not intended as an exact paleoclimatic
reconstruction, but moreover as a generic glacial climate
state, it is of interest to compare the simulation to LGM
observations and models where possible. The SST field of
the glacial simulation reproduces the large scale estimated
features of the LGM ocean. Cooling over the tropics varies
from >2°C near the west coasts of Africa and South
America, to between 1.5°C and 2.0°C over the central
Indian Ocean. These values are colder than CLIMAP, but
consistent with recent proxy studies (see Pinot et al. [1999]
[9] During the first 100 years of the GLACIALFW experiment, NADW increases from 14.3 Sv to 17.4 Sv (Figure 3a).
Over the same period in CNTRLFW, the response is similar
in magnitude, with NADW increasing from 17.6 Sv to
20.5 Sv. This increase is due to the density see-saw effect
[Saenko et al., 2003], whereby freshening the Antarctic
Intermediate Water (AAIW) formation regions in the
Southern Ocean leads to a decrease in local density, an
increase in the density contrast between the Southern
Ocean and the NA, and thereby increased overturning in
the NA.
[10] After this initial period, the glacial climate responds
very differently. In GLACIALFW, NADW steadily decreases,
attaining a value of zero by 1000 years, where it remains
stable. The Atlantic meridional streamfunction at 3000 years
(Figure 2f) confirms the lack of NADW formation in the
final state of this experiment. The initial reduction in
NADW formation in response to the FW perturbation arises
from fresher NA surface conditions due to the penetration of
this salinity anomaly into the Atlantic. The resulting reduction in salt transport into the subpolar gyre leads to further
freshening of this net precipitation zone, thus reinforcing the
reduction in NADW formation. Factors contributing to the
continued suppression of NADW formation after its initial
collapse include the appearance of an AAIW reverse cell
[Sijp and England, 2006] and the appearance of North
Pacific Deep Water (NPDW) [Saenko et al., 2004]. By
contrast in CNTRLFW, NADW initially decreases to 14.4 Sv
at around 540 years, but it then increases back to just over
20 Sv by year 1000, before decreasing smoothly to 18.1 Sv
by year 2500 (Figure 3a).
[11] In both experiments, the NADW decrease after the
initial increase results from the Southern Ocean FW anomaly being transported northwards of the Antarctic Circumpolar Current (ACC) via Ekman transport [Trevena et al.,
2008], where it then intrudes into the NA via the Benguela
Current and the Gulf Stream. Figures 4a and 4b show a
similar negative sea surface salinity (SSS) anomaly in the
Atlantic at year 100 for both CNTRLFW and GLACIALFW.
Yet despite this similarity, there is a very different response
in the NA. In the modern-day climate the extra FW in the
surface layers of the NA does not induce a collapse of
NADW; yet in the glacial climate, NADW shuts down
(Figure 3a).
2 of 6
L09704
TREVENA ET AL.: NADW COLLAPSE FROM SOUTHERN OCEAN
L09704
Figure 2. Meridional overturning stream functions for the (left) Indo-Pacific and (right) Atlantic for (top) steady-state
CNTRL and (middle) GLACIAL; and for (bottom) year 3000 GLACIALFW. Values are based on a 1-year average. Contour
interval is 2 Sv. (1 Sv = 1 106 m3s 1).
[12] Compared to CNTRL, the GLACIAL run has a
thicker annual mean sea-ice coverage over the NA, and
sea-ice also extends further to the south (figure not shown).
This insulates the ocean from the atmosphere, reducing airsea heat fluxes that promote NA overturning [Lohmann and
Gerdes, 1998]. The overturning thus weakens and shifts
southward. Reduced overturning results in weaker northward salt transport into the NA, with surface waters 0.7 psu
fresher compared to CNTRL (Figure 3b). The effect on
density of this fresher water is offset by the fact that the
GLACIAL waters of the NA are 5°C colder. Thus, despite
very different temperature-salinity (T-S) signatures, the
water density over NADW formation regions is very similar
between GLACIAL and CNTRL (Figure 3b). By contrast,
over AAIW formation regions, surface density is 0.3 kg/m3
greater in GLACIAL than CNTRL due to 2°C cooler
surface waters (Figure 3b). The bipolar density see-saw is
thus situated much closer to its tipping point in GLACIAL,
and is more sensitive to the FW anomaly that propagates
into the NA after a few hundred years. In addition, the
increased dependency of density on salinity in the colder
GLACIAL simulation (due to the non-linearity of the
Equation of State), means that the FW anomaly induces a
greater increase in surface buoyancy (by 0.1 kg/m3), and
thus also surface stratification, in the colder glacial case.
[13] When NADW is suppressed in GLACIALFW, it
causes more high salinity water to accumulate in the North
Pacific (NP), where it increases overturning; consistent with
the NP/NA see-saw proposed by Saenko et al. [2004].
Strengthened NP overturning then increases northward salt
transport into the Pacific basin, increasing the salinity over
the NP, and further re-enforcing the initial change that
produced it. In particular, NPIW deepens and increases to
12.2 Sv by year 1800 in GLACIALFW, whereupon it remains
stable, thus becoming NPDW (Figures 2e and 3). In the
GLACIAL simulation, the finely balanced bipolar density
see-saw between the Southern Ocean and the NA (Figure 3b)
leads to a larger response of NADW to the FW anomaly
3 of 6
L09704
TREVENA ET AL.: NADW COLLAPSE FROM SOUTHERN OCEAN
L09704
Figure 3. (a) Overturning time-series for North Atlantic (blue) and North Pacific (red) in the GLACIALFW (solid lines) and
CNTRLFW (dashed lines) experiments. (b) Temperature-Salinity (T-S) plot showing surface density of NADW and AAIW
formation regions in the CNTRL state (squares) and GLACIAL state (circles).
originating in the Southern Ocean compared to CNTRL.
This response is large enough to then tip the NP/NA seesaw, now triggered from the Southern Ocean instead of the
NA or NP as in previous studies.
[14] The initial increase of NADW in GLACIALFW causes
a regional SAT warming of 0.5°C over the NA (Figure 4c).
The Southern Ocean cools during this time, in agreement
with the results of Weaver et al. [2003]. Once NADW
decreases to zero however, regional SAT cools by over
3.5°C in the NA, similar to the difference between a NADW
‘on’ and ‘off’ state in the modern-day climate [Weaver et al.,
2003]. In contrast, the differences in global SAT in the
CNTRL climate are minimal (not shown), since the experiment returns to close to its initial state by 3000 years
(Figure 3a). The increase of NPDW formation in GLACIALFW
causes SAT over the NP to increase by over 1.5°C by
1400 years, where it remains steady (Figure 4d).
4. Summary and Conclusions
[ 15 ] While the simulations presented here are not
attempts at paleoclimate reconstructions, the FW pulse we
apply to the Southern Ocean is comparable in magnitude to
MWP1a. Although this pulse is unlikely to have originated
solely in the Southern Hemisphere [Peltier, 2005], it almost
certainly comprised a significant contribution from the
Antarctic ice-sheet [e.g., Basset et al., 2005]. Further experiments with a reduced FW pulse amounting to an equivalent
global sea level rise of 10 m and 20 m, instead of 25 m, also
resulted in a shut down of NADW in the GLACIALFW
simulation. We have therefore shown that a plausible
Antarctic meltwater discharge may cause cessation of
NADW formation in glacial climate states due to the
advection of the resulting FW anomaly into the NA.
Meltwater pulses in the Northern Hemisphere, and subsequent NADW shut down, are often invoked to explain cold
Heinrich events, or the Younger Dryas of 13 – 11.5 ka BP.
Our results suggest that cold periods in the NA could also
be triggered by meltwater events originating in the Southern
Hemisphere.
[16] After the collapse of NADW, the spin up of NPDW
supports recent proxy studies that advocate a possible
convection cell in the northwest Pacific during deglaciation
[Ohkouchi et al., 1994; Minoshima et al., 2007]. The latter
argue that such an increase in North Pacific ventilation was
temporary; though in our simulation the NADW ‘off’ state
and concurrent NPDW ‘on’ state remain stable. However,
we fix land-ice and orbital parameters, and the model
precludes a free response in atmospheric circulation to the
perturbations applied (Arzel et al. [2008] find that important
feedbacks to the wind stress can occur in models when this
feature is enabled). Reconstructions suggest the western
Laurentide ice sheet adjacent to the Pacific mostly disappeared between 15 ka and 12 ka BP [Peltier, 1994]; it is
possible that the resultant FW injected into the NP could
have caused NPDW to collapse. Under this scenario,
NADW would then re-initiate to modern-day values due
to the NP/NA see-saw effect.
[17] Since ice-sheets amplify climate signals, models
with interactive land-ice are required to more fully explain
the mechanisms leading to glacial-interglacial transitions.
However, models without this feature can provide insight
into the way ocean circulation responses change in a cold
4 of 6
L09704
TREVENA ET AL.: NADW COLLAPSE FROM SOUTHERN OCEAN
L09704
Figure 4. (top) Changes in sea surface salinity (SSS) for (a) GLACIALFW and (b) CNTRLFW after 100 years. (bottom)
Surface air temperature (SAT) differences for the GLACIALFW experiment at (c) 100 years and (d) 3000 years. All values
are based on a 1-year average. Contour intervals are 0.25 psu and 0.5°C respectively.
climate compared to a warm climate. Using an intermediate
complexity model with prescribed land-ice, we have shown
that an Antarctic freshwater pulse can shut down, instead of
strengthen, NADW in a glacial climate simulation.
[18] Acknowledgments. This research was supported by the
Australian Research Council and Australia’s Antarctic Science Program.
References
Arzel, O., M. H. England, and W. P. Sijp (2008), Wind-stress feedback
stabilisation of North Atlantic Deep Water formation during glacial times,
J. Clim, in press.
Basset, S. E., G. A. Milne, J. X. Mitrovica, and P. U. Clark (2005), Ice sheet
and solid earth influences on far field sea-level histories, Science, 309,
925 – 928.
Braconnot, P., et al. (2007), Results of PMIP2 coupled simulations of the
mid-Holocene and Last Glacial Maximum—part 2: Feedbacks with emphasis on the location of the ITCZ and mid- and high latitudes heat
budget, Clim. Past, 3, 279 – 296.
Bryan, F. (1986), High-latitude salinity effects and interhemispheric thermohaline circulations, Nature, 323, 301 – 304.
Jayne, S. R., and L. C. St. Laurent (2001), Parameterising tidal dissipation
over rough topography, Geophys. Res. Lett., 28, 811 – 814.
Knorr, G., and G. Lohmann (2003), Southern Ocean origin for the resumption of Atlantic thermohaline circulation during deglaciation, Nature,
424, 532 – 536.
Lohmann, G., and R. Gerdes (1998), Sea ice effects on the sensitivity of the
thermohaline circulation, J. Clim., 11, 2789 – 2803.
Lumpkin, R., and K. Speer (2007), Global ocean meridional overturning,
J. Phys. Oceanogr., 37, 2250 – 2262.
Lynch-Stieglitz, J., J. F. Adkins, W. B. Curry, and T. Dokken (2007),
Atlantic meridional overturning circulation during the Last Glacial Maximum, Science, 316, 66 – 69, doi:10.1126/science.1137127.
McManus, J. F., R. Francois, J. M. Gherardi, L. D. Keigwin, and S. BrownLeger (2004), Collapse and rapid resumption of Atlantic meridional circulation linked to deglacial climate changes, Nature, 428, 834 – 837.
Minoshima, K., et al. (2007), Deep water ventilation in the northwestern
North Pacific during the last deglaciation and the early Holocene (15 – 5
cal. kyr b.p.) based on AMS 14C dating, Nucl. Instrum. Methods Phys.
Res., Sect. B, 259, 448 – 452.
Montenegro, A., M. Eby, A. J. Weaver, and S. R. Jayne (2007), Response of
a climate model to tidal mixing parameterization under present day and
Last Glacial Maximum conditions, Ocean Modell., 19, 125 – 137,
doi:10.1016/j.ocemod.2007.06.009.
Ohkouchi, N., H. Kawahata, M. Murayanam, and M. Okada (1994), Was
deep water formed in the North Pacific during the late Quaternary?
Cadmium evidence from the northwest Pacific, Earth Planet. Sci. Lett.,
124, 185 – 194.
5 of 6
L09704
TREVENA ET AL.: NADW COLLAPSE FROM SOUTHERN OCEAN
Peltier, W. R. (1994), Ice age paleotopography, Science, 265, 195 – 201.
Peltier, W. R. (2005), On the origins of meltwater pulse 1a, Quat. Sci. Rev.,
24, 1655 – 1671.
Pinot, S., et al. (1999), Tropical paleoclimates at the Last Glacial Maximum: Comparison of paleoclimate modeling intercomparison project
(PMIP) simulations and paleodata, Clim. Dyn., 15, 857 – 874.
Saenko, O. A., A. J. Weaver, and J. M. Gregory (2003), On the link
between the two modes of the ocean thermohaline circulation and the
formation of global-scale water masses, J. Clim., 16, 2797 – 2801.
Saenko, O. A., A. Schmittner, and A. J. Weaver (2004), The AtlanticPacific seesaw, J. Clim., 17, 2033 – 2038.
Sijp, W. P., and M. H. England (2006), Sensitivity of the Atlantic thermohaline circulation and its stability to basin scale variations in vertical mixing, J. Clim., 19, 5467 – 5481.
L09704
Trevena, J., W. P. Sijp, and M. H. England (2008), Stability of Antarctic
Bottom Water formation to freshwater fluxes and implications for global
climate, J. Clim, in press.
Weaver, A. J., et al. (2001), The UVic Earth system climate model: Model
description, climatology, and applications to past, present and future
climates, Atmos. Ocean, 39, 361 – 428.
Weaver, A. J., O. A. Saenko, P. U. Clark, and J. X. Mitrovica (2003),
Meltwater pulse 1a from Antarctica as a trigger of the Bolling-Allerod
warm interval, Science, 299, 1709 – 1713.
M. H. England, W. P. Sijp, and J. Trevena, Climate Change Research
Centre, University of New South Wales, Sydney, NSW 2052, Australia.
([email protected])
6 of 6