Does Antarctica initiate glaciations?

Attempts to explain the last million years of glacial cycles have concentrated mainly on the growth of Northern Ice sheets. Traditionally the summer insolation at 65N is thought to control the growth and extent of ice sheets, with interglacials coinciding with summer melt-back. A new paper ‘Hemispheric sea ice distribution sets the glacial tempo‘ argues instead that Antarctic sea-ice sets  the timing of glaciations.

The Southern Hemisphere has a higher capacity to grow sea ice than the Northern Hemisphere, where continents block growth. New research shows that the expansion of Southern Hemisphere sea ice during certain periods in Earth’s orbital cycles can control the pace of the planet’s ice ages. Credit: Jung-Eun Lee / Brown University

Variations in the earth’s orbit (Milankovitz cycles) change the intensity of polar summer insolation and consequent summer melt back. Obliquity affects both poles equally, whereas precession of the equinoxes  is modulated by the earth’s eccentricity. The extent of sea ice in the Arctic is limited by the closeness of northern continents, whereas Antarctica is not. Lee et al. argue that at minimum summer insolation over Antarctica, amplified by high eccentricity,  leads to a rapid growth in annual sea ice coverage in the southern hemisphere. This reduces the earth’s net albedo enough to initiate a new ice age leading consequently also to the growth in northern ice sheets.  This asymmetry in the response of sea ice to precession between North and South poles is enough, they argue, to kick start a new ice age.

Currently perihelion coincides with the Antarctic summer insolation at a maximum. Despite this an asymmetry in sea ice cover averaged between 1980-2015 is evident.

Sea ice extent (million km2) NH SH
Max 13.27 16.39
Min 5.40 2.57
Difference 7.87 13.82

The authors use a climate model to calculate that June perihelion leads to sufficient increase in summer antarctic sea ice to reduce the net annual insolation for the earth, triggering a new ice age thus reducing albedo.

drop in net shortwave energy flux at the top of the atmosphere for the clear (dotted) and all-sky (solid) conditions from the June perihelion and to the December perihelion calculated by climate model GFDL CM2.1.  They compare different CO2 levels and ‘All sky’ is the realistic situation in the Antarctic.
Credit: Jung-Eun Lee / Brown University

The paper does not discuss how ice ages end, but proposes that glaciations can only start when the precession term is amplified by maximum eccentricity  leading to the 100,000y cycles.  Until 1 million years ago obliquity alone drove the glacial cycle until the earth cooled below a threshold. This is a nice idea to explain the 100k problem but does it work? Here are the calculated summer maxima of insolation calculated from the work of Laskar et al.

Maximum and total solar insolation calculated at the poles during last 600,000 years. The Arctic and Antarctic summer insolation are modulated by eccentricity and out of phase by half a precession cycle. Both the total annual insolation and the N-S asymmetry clearly show the underlying effect of the 41,000 obliquity signal.

One problem is that eccentricity itself follows a 400,000 years super cycle so that some summer minima in the Antarctic (dashed line) nearer minimum eccentricity  are actually larger than others at maxima eccentricity. Obliquity affects both poles equally which is why it alone was sufficient to drive glacial cycles fro the previous 3 million years.

Now let’s see if  minima in  Antarctic insolation match up to the onset of glaciations.

There is indeed some correlation but it is not fully convincing. Basically the problem is that there is only a 12000y gap between maximum northern and maximum southern polar summer insolation. So why should only 1 in 6 southern minima initiate a new glacial cycle and likewise 1 in 6 northern minima herald a new interglacial? Perhaps only those minima coincident with minimum obliquity count. This leads to a slightly better correlation with cooling. If we now just look at the last glacial cycle and compare the Greenland (NP) ice data with that from Antarctica, there is further insight.

Comparison of Antarctic temperature and dust measurements(top) with Greenland data (bottom). The dashed curve is the Summer insolation at the South Pole while the Blue curve is that at the North Pole. Interglacials are triggered by a rise in NP insolation.

It is remarkable how similar the overall Arctic/Antarctic  temperature dependence is. Whatever initiates a glaciation and terminates it is a global phenomena without significant time delay. This favours a global change in albedo and probably also connected heat flow from Antarctica to the Arctic and vice versa through the AMOC. Is the fall in temperature 12000 years ago in the Antarctic caused by by the fall in summer insolation? If so then the next glacial cycle will surely begin within 5000y as Antarctic summer insolation is currently in decline.

The new proposal in this paper is that sea ice growth is unhindered in the Antarctic Ocean but limited in extent in the Arctic. This produces an imbalance in global albedo whenever Antarctic summer insolation reaches an (eccentricity enhanced)  minimum. However despite this nice idea, I am dubious that this paper is a breakthrough in understanding glacial cycles. It still seems to need the Arctic to end ice ages even if the Antarctic starts them. Yet something else is needed as well to short circuit those intermediate insolation minima which apparently have little effect until the every 100 ky eccentricity peaks. This is the same problem as relying on 65N insolation to pace glaciations.

About Clive Best

PhD High Energy Physics Worked at CERN, Rutherford Lab, JET, JRC, OSVision
This entry was posted in Climate Change, climate science, Ice Ages and tagged , . Bookmark the permalink.

5 Responses to Does Antarctica initiate glaciations?

  1. Euan Mearns says:

    Clive, as you know d_Insolation resulting from orbital cycles is almost zero on a global scale. d_Insolation at 65?N at summer solstice is a desperate attempt to try and reconcile Milankovitch with observed glacial cycles. But even there, the d_Insolation is feeble in explaining the drastic planetary consequences of glacial cycles.

    Last Fall I had an article published in The Alpine Journal (thats right rock climbing) (I’m away to send you a copy). We concluded that Milankovitch could be a red herring.

    I believe it is the pattern of ocean and atmosphere circulation that controls the glaciations and they are in turn modulated by The Sun, in particular variations in spectrum that correlate with d_magnetism.

    We don’t understand what causes the glacial cycles and therefore have zero chance of understanding climate change today. d_magnet somehow has a second order correlation with orbit. CO2 simply follows the resulting dT, but at inceptions lags it by several thousand years.

    • Clive Best says:

      Hi Euan,

      We certainly don’t understand glacial cycles over the last 800,000 years. However we do know that obliquity was the driver for the preceding 3 million years. They follow the obliquity cycle almost like clockwork. This makes sense as high obliquity brings more sunshine to both poles simultaneously while low obliquity reduces solar radiation.

      So yes the mystery of the longer glaciations as the earth cooled has not been solved. Enhanced heat circulation to polar regions is one possibility that causes glacial collapse. Another possibility is that CO2 levels fall so low that northern forests die out, turn to dust and increase the ice surface albedo. Under very cold conditions the poles are like deserts with very little precipitation. Now we have the above proposal and various hysteresis type scenarios of glacial collapse.

      I don’t think Milankovitch is a red herring, but I do think it now needs a helping hand from an as yet unknown source.


  2. Rod says:

    I have read some of your articles with interest. It is good to read someone who takes a view from the physics rather than writing a simulation which does not seem to work very well. Only recently I found an article that the temperature on Mt Kenya was 5 deg C higher than any model predicted.

    When I read physics many years ago there was only one perihelion of Earth and that occurred in January (around the 4th depending on leap days). Maybe you mean solstices which do occur in June and December but these are very different in terms of physical and geometry to perihelion.

  3. Ed Bo says:

    Rod: There is only one perihelion per year. As the post says, this perihelion now occurs during the Antarctic summer (January). But due to the precession (wobble) of the earth’s spin, that event varies slowly through the year, with a cycle of about 23,000 years.

    The graph titled “June Perihelion Minus December Perihelion…” is comparing years in which the (single) perihelion occurred in June with (different) years in which it occurred in December.

  4. Javier says:


    We are always back to square one on the glacial cycle.

    There is no 100 Kyr glacial cycle. It is a mathematical artifact. Temperatures on the planet follow obliquity. They respond much less to precession or eccentricity.

    Interglacials, when correctly defined and accounting for a few cycle irregularities clearly follow the obliquity cycle.

    If you correct for the early (long) MIS 11 interglacial, and the late (short) MIS 7 interglacial and place their start at the date they should have started, the following interglacials get also placed at the right distance, and EVERY interglacial is positioned according to the obliquity cycle. Try to do that trick with the 100 kyr cycle where lots of interglacials are badly positioned.

    Blue curve obliquity. Red curve eccentricity normalized.

    And there is a rule that it is never broken. When obliquity falls interglacials end. There is a variable delay due to thermal inertia of average 6500 years. It is the same that the coldest part of the winter comes with a delay respect to the winter solstice.

    So falling obliquity has always ended interglacials. We can only discuss how it does it. No reason to think that this time will be different, unless one has faith. On God or on CO2, there is choice.

Leave a Reply