Southern Ocean mixing, seasonal sea ice, and glacialinterglacial CO2 variation.
Abstract
Several lines of evidence indicate that a change in the rate of stratification and mixing
in the Southern Ocean, either near the surface or at depth (most probably, both), is a
key factor in explaining glacial-to-interglacial atmospheric CO2 change. Here I
discuss how the Southern Ocean is ventilated today, and propose how this may have
been different in glacial time. Deep convection is confined to regions very close to the
continent. The open ocean south of the Polar front is largely ventilated from below,
with rapid mixing apparently driven by interaction of deep currents with topography
driving high rates of mixing well up into the water column. Between the surface and
~1500m depth, the water column is ventilated from above, stabilized by a halocline
that is due in part to sea ice formation and brine rejection, and probably dominated by
the effects of storms. I propose that in glacial time, more copious sea ice formation
towards the Antarctic continent, together with substantial seasonality and melting
further out, resulted in denser bottom water formation and more fresh water near the
surface. The greater stratification at depth caused lower mixing rates there while
greater winter-time sea ice cover reduced mixing towards the surface. The increased
stratification in the glacial deep ocean led to reduced ventilation of the deep ocean as
a whole, allowing the build up there of biologically-transported carbon. This scenario
is consistent with most proxies, including those on the extent of sea ice, the
productivity and nutrient distribution in the Southern Ocean, and the distribution of
13C. I use a box model, similar to that of Toggweiler (1999) to illustrate that, (in
conjunction with other mechanisms known to have influenced atmospheric CO2), this
scenario can reconcile CO2 variation with current proxies. Specific tests are suggested
that would help distinguish this “Southern Ocean seasonal sea ice” mechanism from
others that have been suggested.
Introduction
The possibility that a change in the rate at which the deep sea is ventilated could
lead to changes in atmospheric CO2 was first raised in 1984, when several papers
were published pointing to the possible role of the high-latitude oceans as controllers
of natural CO2 concentrations (Knox and McElroy 1984; Sarmiento and Toggweiler
1984; Siegenthaler and Wenk 1984). The box models on which these these
“Harvardton Bears” papers were based highlighted the dependence of atmospheric
CO2 on a balance between biological productivity and ventilation between the surface
and the deep in the Southern Ocean. Today, water at the surface of the Southern
Ocean (and the North Pacific) contains non-zero mineral macro-nutrients (nitrate and
phosphate), and correspondingly has a higher pCO2 than would be the case if these
nutrients were more fully utilized. Much of the water in the the deep sea is ventilated
from this region, and its “preformed” nutrient and CO2 is set at the surface of the
Southern Ocean. Either increasing biological productivity, or decreasing the exchange
of water between the surface and depth in this region, can cause atmospheric CO2 to
be lower in this kind of model.
1
The results of these papers were initially interpreted in terms of increased
biological export productivity in the polar waters. However, it has become
increasingly clear that proxy evidence does not support the idea of an increased
Southern Ocean productivity in glacial time. There is room for doubt because not all
proxies seem to tell the same story, but a recent “multiproxy” compilation of LGM
export production estimates (Bopp, Kohfeld et al. 2003) suggests a coherent picture:
an increase in productivity in the Subantarctic region, particularly in the Atlantic
sector, in glacial time, but less export production from the region south of the (present
position of) the Polar Front.
An alternative possible explanation for lower glacial CO2 is that the rate of
ventilation of the deep ocean was slower in glacial time. Recently, (Toggweiler 1999)
has revived this idea as an important mechanism for causing lower atmospheric CO2.
In fact, given the constraints put on the problem of atmospheric CO2 change by our
present knowledge, I will argue in this paper that the deep ocean must have been more
slowly ventilated in glacial time. It is however not trivial to understand why this might
have been the case. In view of this, it is important to fully understand how “Southern
component” deep water ventilates the ocean today. Accordingly, the first section of
this paper is entirely concerned with observations of the modern ocean. I then put
forward a suggestion for why, and how, ventilation was different in glacial time.
Some calculations using a box-type model are used to show that this mechanism,
acting in combination with other effects that we know occurred, has the potential to
change atmospheric CO2 by the right amount. This is not the first suggestion for a
mechanism to change ocean ventilation, and I therefore end by comparing it with
other recently proposed mechanisms, suggesting tests that may help to distinguish
between the competing theories.
Our uncertainty about the mechanisms causing glacial-interglacial CO2 change,
even after more than 20 years of research, is sometimes cited as an illustration of our
ignorance of fundamental processes in the Earth system. However, though there are
differences in the detail, it is worth noting that all the viable theories are now
convergent in their most important aspects. Plausible theories for lower glacial
atmospheric CO2 all share a requirement for lower ventilation rates of the deep sea,
obtained by one or more of: higher sea ice cover, near-surface stratification of the
Southern Ocean or a greater deep stratification. They further require more efficient
nutrient utilization leading to lower preformed nutrients in the Southern Ocean in
glacial time (probably aided by greater atmospheric iron supply), all of these changes
serving to partition biologically fixed carbon into the deep sea and away from the
atmosphere.
The global ocean overturning circulation
Figure 1, adapted from (Toggweiler 1994) and (Sloyan and Rintoul 2001), shows a
schematic of the global ocean overturning circulation. The deepest, densest waters of
the world ocean are formed close to the Antarctic continent, by a process of brine
rejection associated with sea ice formation leading to dense shelf waters. These
transfer into the deep ocean by down-slope gravity currents and by convection (the
relative importance of these mechanisms being undefined). They mix with fresh, cold
water as they do so (ref from haine et al) , to form Antarctic Bottom Water (AABW).
North Atlantic Deep Water (NADW), less dense, warmer and saltier than AABW, is
formed by sinking of water in the Northern North Atlantic, and penetrates southward.
Intermediate waters, freshened by net precipitation and sea ice melt, are formed by the
2
northward transport and subduction of Antarctic surface waters by Ekman drift under
the influence of the westerly winds in the Southern Ocean.
There are two contrasting views about the mechanics of the overturning
circulation. The more classical view, articulated by (Munk 1966) and elaborated by
(Munk and Wunsch 1998), is that the steady state circulation is governed by a balance
between vertical advection and downwards turbulent diffusion of buoyancy. In this
“abyssal recipes” picture, the rate-limiting step is how fast buoyancy can mix
downwards. Vertical mixing rates, represented by the turbulent vertical diffusivity κz,
are highly variable with position in the ocean. Measurements show them to be very
low in the thermocline (Ledwell, Watson et al. 1998) and the interior of the oceans
away from rough topography (Toole, Polzin et al. 1994). However, internal waves and
turbulence generated by currents interacting with bottom topography generate
enhanced mixing which can extend up through the water column for kilometres
(Polzin, Toole et al. 1997; Ledwell, Montgomery et al. 2000). These mixing “hot
spots” are generally thought to dominate the overall mixing of the oceans.
The alternative view of the ocean overturning is that it is driven by wind-driven
surface convergence and divergence, especially that associated with the westerlies in
the Southern Ocean (Toggweiler and Samuels 1995; Toggweiler and Samuels 1998).
In this view, the surface Ekman drift associated with the zonal winds near the latitude
of Drake Passage lifts water out of the deeper ocean. This in turn allows a
compensatory sinking of water is elsewhere, i.e in the North Atlantic. This winddriven overturning is not dependent on interior diapycnal fluxes, and can proceed
even as κz in the interior tends to zero.
In reality, ocean ventilation combines elements of both of these modes of
ventilation. Given that rates of diapycnal mixing are agreed to be low in the main
thermocline, wind-driven preocesses will likely dominate vertical transport through
the top kilometre or so of the water column. Some portion the NADW formation is
probably wind-driven by Ekman suction in the Southern Ocean, see for instance
(Webb and Suginohara 2001)). On the other hand, the ventilation of the densest
waters in the world ocean must be largely diapycnally driven. This seems inevitable,
because this water has no surface outrcrop of significant area – it is formed in very
restricted areas indeed. While it is possible to imagine its formation at the surface in
dense but areally restricted plumes due to shelf processes or convection, there is no
known mechanism to return it to the surface where it can be destroyed by air-sea flux
processes in similarly highly restricted regions, so it must be modified in the interior.
On tracing the path of this water away from the Antarctic, it becomes clear (as
described further below) that it is so modified, its properties being altered in the deep
sea, well away from the surface, implying large interior diapycnal fluxes. Therefore, if
ventilation of the deepest ocean was slower in glacial than interglacial time, as
Toggweiler (1999) has suggested, this is likely to be a largely a question of changing
diapycnal mixing rates. This presents a theoretical challenge. Why should deep ocean
mixing be slower in glacial time?
Recently, (Munk and Wunsch 1998) have pointed out that the energy required to
drive the abyssal circulation is of the same order as the total energy available from
wind and tidally generated currents in the deep sea. Plausibly therefore, the overall
control on the rate at which the deep MOC turns is one of energy limitation. The
power required to maintain the diabatic abyssal circulation is readily calculated. From
a scale analysis, the rate at which potential energy must be added to raise bottom
water being formed with a volume flux of Q is W ~ Qgh , where  is the density
difference between the initial and final densities of the water, g the acceleration due to
3
gravity and h is the scale height of the density change. If the available power supply
is constant, then, other things being equal, Q might be expected to be inversely related
to , with the total buoyancy flux Q remaining constant. Similarly, scale analysis
of the steady-state advective-diffusive balance (e.g. that employed by (Munk 1966))
gives us the relation Q ~ A z / h between the overturning rate, the area A over which
upwelling occurs, the vertical mixing rate and the scale height. Substituting for Q in
the first relation we obtain W ~ A z g , showing that if the power is indeed
constant, then vertical mixing rates should be inversely proportional to density
difference. Recent papers by Nilsson and colleagues (Nilsson, Brostrom et al. 2003;
Nilsson, Brostrom et al. 2004) have explored in greater detail the implications of an
energy-limited overturning for the modern day circulation with conceptual and
numerical models. Here we are content with noting that there is some reason to
suggest that mixing and overturning rates may decrease as density anomaly increases.
Suppose then, that in glacial time the density contrast between the deepest waters and
those overlying them was greater than it is today: this might help explain why the
deep sea was apparently less well ventilated at LGM.
What do we know about stratification of the deep ocean in glacial time? Figure 2,
from the work of (Adkins, McIntyre et al. 2002), shows estimates of the temperature
and salinity in the deep ocean at LGM from measurements on pore waters, compared
with present day conditions. Contours of potential density referenced to 4000m are
also shown. The difference between the stratification of the glacial and modern ocean
is striking. Salinities are greater everywhere, which is to be expected because the
greater volume of fresh water bound up in icecaps in glacial time resulted in a saltier
ocean. However, whereas the modern abyssal ocean is largely stratified by
temperature differences, the density differences in the deep glacial ocean are mostly
due to salinity. Temperatures are close to freezing at all the sites in the glacial ocean,
but the deep Southern Ocean shows as much saltier than the other locations. The
density anomaly between the Southern Ocean site and the others is about three times
larger than it is today, and a crude application of the above (much over-simplified)
reasoning might then suggest that the mixing between this Southern component and
the overlying waters was three times slower at LGM.
Mixing and ventilation of the modern-day Southern Ocean,
In recent years a good deal of new information has become available relating to rates
of diapycnal mixing in the Southern Ocean. In the near-surface there have been
several direct measurements using tracer releases (Law, Abraham et al. 2003),
Goldson, 2004). In the region of the Scotia Sea, studies making use of new techniques
using lowered acoustic doppler current profilers have suggested very high mixing
rates in the region of the Scotia Sea (Heywood, Garabato et al. 2002; Garabato, Polzin
et al. 2004). Estimates using CFCs in the broad plume of bottom water originating
from the Weddell Sea and flowing towards the Indian Ocean have indicated high rates
of mixing between this and the overlying water (Haine, Watson et al. 1998).. These
estimates are drawn together in Table I. There is a considerable range of values,
particularly in the deep ocean. However, a consistent picture emerges: in the upper
water column, mixing across and below the summer-time seasonal pycnocline is ~10-5
m2s-1, though with evidence of rates that are several times higher during passage of
storms (Goldson, 2004). In the deep Southern Ocean however rates are ~10-3 m2 s-1,
ranging higher still immediately over rough topography. Results of inverse models
constrained with hydrographic data can also be used to infer diapycnal fluxes, but in
4
the Southern Ocean, these broad averages are difficult to separate into fluxes due to
air-sea interaction and those due to interior mixing (Ganachaud and Wunsch 2000;
Sloyan and Rintoul 2001).
Table I: Estimates of diapycnal mixing rates in the Southern Ocean
method
Depth
Location
Value (10-4m2s-1)
(Law, Abraham et
al. 2003)
SF6 tracer
release
50-100m
61S,
0.11 0.2
Goldson. (2004)
SF6 tracer
release
50-100m
(Garabato, Polzin
et al. 2004)
LADCP shear
+ hydrography
0-200m
Scotia Sea, Drake
Passage
0.1 – 0.3
(Haine, Watson et
al. 1998)
CFC budget
> 3500 m
Abyssal S. Ocean
-
52
(Heywood,
Garabato et al.
2002)
LADCP and
basin budget
> 3500m
Scotia Sea
39 
(Garabato, Polzin
et al. 2004)
LADCP shear
+ hydrography
> 2000m
Scotia Sea
5 – 100
East Drake
Passage
10 – 1000
Atlantic and
Indian sector,
-
~10
Source
Upper water
column
?
Deep water
This paper
CFC budget
> 2000m
As shown, for example, by (Haine, Watson et al. 1998) chlorofluorocarbon
concentrations can be useful for illuminating the processes of ventilation from the
surface to the interior of the Southern Ocean. Figure 3 shows a compilation of CFC11 concentrations on five WOCE sections from Drake passage through the Atlantic
and Indian Ocean sectors.. The concentration scale has been adjusted so that
concentrations below 1 pmol kg-1 are highlighted. Deep waters at Drake Passage show
almost undetectable CFCs. Concentrations are also low north of about 50S.
However, the deep outflow of recently ventilated bottom water from the Weddell Sea
is clearly seen as a core at about 60S in both the Atlantic Ocean sections. The section
through the Eastern Scotia Sea shows relatively high concentrations throughout the
water column, due perhaps to mixing of this deep outflow up through much of the
water. Likewise, all the other sections show enhanced concentrations in the deepest
waters south of 60S, and easily measurable concentrations right up through the water
5
column. The more easterly sections also show evidence for extensive “local” deep
water formation close to the Antarctic continent, not associated with the Weddell or
Ross Sea outflows.
Figure 4. shows the means of the CFC concentrations from the Atlantic sections,
averaged north of 50S and south of 50S. The profiles south of 50S reinforce the
impression of a CFC sourced at the bottom that mixes up through the water column to
depths ~1000m. If this is the origin of the mid-depth CFC, a simple scale analysis can
be used to estimate the order of magnitude of the vertical mixing that must be
responsible, as follows: CFCs have a time scale for rising in the atmosphere of order
20 years, as shown in the inset to Fig. 4. The profiles in Fig. 4 have length scales of
order 1000m or more. From these scales, a mixing rate of order z ~ (length scale)2/(2
x time scale) ~ 10-3 m2s-1 follows. This is broadly compatible with the deep mixing
rates obtained by other workers, as summarized in Table 1.
It might be argued that the mid-water CFC content comes not from up-mixing from
the bottom source, but is injected directly onto the intermediate density surfaces by
surface-driven buoyancy changes, for example convection or dense plume formation
associated with sea ice formation. The structure of the CFC sections does not provide
much support for this hypothesis however. All the sections except the one south of
Australia show a distinct minimum in concentrations south of 60S and at 1 to 2 km
depth, suggesting that the CFC further to the north has not primarily got there by
advection from the south. Perhaps then, all the injection occurs by convective
processes east of the Antarctic Peninsula and in the Scotia Sea region, with the newlyformed water subsequently advected eastward in the ACC. This route for the
formation of mid-water CFCs is probably important, however the model of (Haine,
Watson et al. 1998) considered only data from east of the prime meridian, and still
found high mixing rates.
The effect of this rapid mixing on the properties of newly formed AABW are
substantial. Figure 5 shows potential temperatures and densities at 3500m depth in the
Southern Ocean. The tongue of cold, dense water, moving out from the Weddell Sea
region is particularly clear, but as it transits eastward it warms, and such cold waters
are not found north of the ACC. According to the measurements of Haine et al., the
Weddell sea water has a transit time to the Crozet-Kerguelen region of only ~40
years, but this is long enough for its temperature to be substantially modified by
mixing with the warmer water overlying and surrounding it. Thus the AABW which
forms the bottom water for most of the world’s oceans is substantially less dense than
the water which first sinks to the bottom in the Southern Ocean.
The upper water column in the Southern ocean is today largely stratified by a
halocline. The surface water is relatively fresh, and the Ekman flux drives this water
to the north until it is subducted under the warmer and saltier waters north of the
Subantarctic front to form Antarctic intermediate water. There are two potential
sources for the freshening that converts salty upwelling circumpolar water into fresh
AAIW. One is meteoric water -- net precipitation minus evaporation, and the other is
fresh water formed from ice melt, for which the associated salt has been removed
from the mixed layer by brine rejection processes.
It is of interest to calculate the order of magnitude of each of these processes today.
Observationally-derived records such as COADS are unreliable in the sparsely
monitored Southern Ocean, so we use the long-term averages of NCEP re-analysis
data to examine the precipitation minus evaporation (P-E) balance. Figure 6 shows
NCEP annual, zonally-summed net freshwater, accumulated northward from 85S.
6
The net P-E flux between 85S and 60S is 0.32 Sv, while between 85S and 50S it
is 0.68 Sv.
To find the amount of fresh water produced by fractionation of sea water in the
formation of sea ice, we need to estimate the volume of sea ice formed annually. The
area covered seasonally by sea ice is about 1.5 x 107 km2 {Zwally, 2002 #200}.
Relatively few observations of ice thickness have been made. We need to be careful
also, to ensure that we count only ice formed from freezing of sea water, and not the
additional accumulation of snow on top of this. The observations of {Worby, 1996
#205} in the Bellingshausen and Amundsen seas suggest an average ~0.5m for this
ice thickness. The work of {Harms, 2001 #207} suggests total thcknesses of ~0.3m
along the Greenwich meridian, but thicker ice (one to two meters) in the Weddell Sea.
If we assume that on average the ice is 0.5 m thick, using the area above we derive a
figure of 0.24 Sv for the annual average rate of fresh water production due to
Southern Ocean seasonal sea ice formation. Thus we estimate that today, ice
formation contributes more than 40% of the net fresh water budget of the surface
ocean south of 60S, and about 25% of the budget south of 50S. Due to the sparsity
of information on the thickness of the ice, there is large uncertainty on these figures,
but it is clear that the contribution from sea ice is important. In glacial time, when the
hydrological cycle was weaker (there was less evaporation because sea surface
temperatures were lower) but the area covered seasonally by sea ice was about twice
as great, ice would likely have been the dominant contributor to the fresh water
budget south of the Polar front.
Role of: Temperature, salinity, growth of terrestrial biosphere,
Proxy evidence: salty deep water near the freezing point, low Antarctic productivity
Del13C
The need for a ventilation-related mechanism
The need for slower ventilation of the deep ocean. – a puzzle
Energy-limited mixing of the deep ocean ventilation.
How can deep ocean in glacial
References
Adkins, J. F., K. McIntyre, et al. (2002). "The salinity, temperature, and delta O-18 of
the glacial deep ocean." Science 298(5599): 1769-1773.
Bopp, L., K. E. Kohfeld, et al. (2003). "Dust impact on marine biota and atmospheric
CO2 during glacial periods." Paleoceanography 18(2): art. no.-1046.
Ganachaud, A. and C. Wunsch (2000). "Improved estimates of global ocean
circulation, heat transport and mixing from hydrographic data." Nature
408(6811): 453-457.
Garabato, A. C. N., K. L. Polzin, et al. (2004). "Widespread intense turbulent mixing
in the Southern Ocean." Science 303(5655): 210-213.
Haine, T. W. N., A. J. Watson, et al. (1998). "The flow of Antarctic bottom water to
the southwest Indian Ocean estimated using CFCs." Journal of Geophysical
Research-Oceans 103(C12): 27637-27653.
7
Heywood, K. J., A. C. N. Garabato, et al. (2002). "High mixing rates in the abyssal
Southern Ocean." Nature 415(6875): 1011-1014.
Knox, F. and M. B. McElroy (1984). "Changes In Atmospheric Co2 - Influence Of
the Marine Biota At High- Latitude." Journal Of Geophysical ResearchAtmospheres 89: 4629-4637.
Law, C. S., E. R. Abraham, et al. (2003). "Vertical eddy diffusion and nutrient supply
to the surface mixed layer of the Antarctic Circumpolar Current." Journal of
Geophysical Research-Oceans 108(C8): art. no.-3272.
Ledwell, J. R., E. T. Montgomery, et al. (2000). "Evidence for enhanced mixing over
rough topography in the abyssal ocean." Nature 403(6766): 179-182.
Ledwell, J. R., A. J. Watson, et al. (1998). "Mixing of a tracer in the pycnocline."
Journal of Geophysical Research-Oceans 103(C10): 21499-21529.
Munk, W. and C. Wunsch (1998). "Abyssal recipes II: energetics of tidal and wind
mixing." Deep-Sea Research Part I-Oceanographic Research Papers 45(12):
1977-2010.
Munk, W. H. (1966). "Abyssal recipies." Deep Sea Research 13: 707-730.
Nilsson, J., G. Brostrom, et al. (2003). "The thermohaline circulation and vertical
mixing: Does weaker density stratification give stronger overturning?" Journal
of Physical Oceanography 33(12): 2781-2795.
Nilsson, J., G. Brostrom, et al. (2004). "On the spontaneous transition to asymmetric
thermohaline circulation." Tellus Series a-Dynamic Meteorology and
Oceanography 56(1): 68-78.
Polzin, K. L., J. M. Toole, et al. (1997). "Spatial variability of turbulent mixing in the
abyssal ocean." Science 276(5309): 93-96.
Sarmiento, J. L. and J. R. Toggweiler (1984). "A New Model For the Role Of the
Oceans In Determining Atmospheric Pco2." Nature 308: 621-624.
Siegenthaler, U. and T. Wenk (1984). "Rapid Atmospheric Co2 Variations and Ocean
Circulation." Nature 308: 624-626.
Sloyan, B. M. and S. R. Rintoul (2001). "Circulation, renewal, and modification of
Antarctic mode and intermediate water." Journal of Physical Oceanography
31(4): 1005-1030.
Toggweiler, J. R. (1994). "The Oceans Overturning Circulation." Physics Today 47:
45-50.
Toggweiler, J. R. (1999). "Variation of atmospheric CO2 by ventilation of the ocean's
deepest water." Paleoceanography 14(5): 571-588.
Toggweiler, J. R. and B. Samuels (1995). "Effect Of Drake Passage On the Global
Thermohaline Circulation." Deep-Sea Research Part I-Oceanographic
Research Papers 42: 477-500.
Toggweiler, J. R. and B. Samuels (1998). "On the ocean's large-scale circulation near
the limit of no vertical mixing." Journal of Physical Oceanography 28(9):
1832-1852.
Toole, J. M., K. L. Polzin, et al. (1994). "Estimates of Diapycnal Mixing in the
Abyssal Ocean." Science 264(5162): 1120-1123.
Webb, D. J. and N. Suginohara (2001). "Oceanography - Vertical mixing in the
ocean." Nature 409(6816): 37-37.
8
Figure 1: Potential temperature-salinities of deep waters at the LGM derived from porewater
measurements at four ODP sites by Adkins et al., 2002. The modern values at these sites are
also shown. Contours are of 4. The open arrow shows the modern mean ocean salinity, while
the filled arrow is the average salinity adjusted for a 125m drop in eustatic sea level, similar
to that at the LGM. The inset shows the locations of the four sets of measurements, with the
depths in metres written by each symbol.
9
Cumulative, area integrated net freshwater (Sv from 85°S)
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
80
70
60
50
40
30
20
10
Latitude (° S)
Figure . Net precipitation – evaporation for the southern hemisphere, integrated zonally and
between 85S and the latitude on the ordinate. Values at 50S and 60S are highlighted by the
dotted lines. The calculation used the time-average of the NCEP re-analyses results from
1968-1996 (available from the NOAA-CIRES Climate Diagnostics Center,
www.cdc.noaa.gov).
10