One of the most memorable moments in Al Gore’s movie “An Inconvenient Truth” occurs as Gore discusses the ice-core records for CO2 and temperature over the previous 7 “ice-ages”. After graphically showing the strong correlation between CO2 and temperature in these records, Gore is forced to mount a cherry-picker in order to physically point to the height on the graph that CO2 is predicted to reach over the next 50 years. The implication is clear: if CO2 is predicted to get so high, and given that temperature follows CO2, just imagine how high the temperature will be!

In several places I have seen the scientific basis for this argument questioned. Drawing on researchers such as Fischer et al. (1999), “skeptics” have argued that the ice-core records actually show that temperature changes cause CO2 changes not the other way around. Fischer et al. (1990) observed that temperatures in ice-core records lag CO2 on the order of 400 to 1000 years. CO2 clearly cannot be causing temperature changes, it is argued, if it is following temperature changes; the ice-core records imply that CO2 is a product of temperature change, not its cause.

A succinct and representative example of this kind of argument can be found in the “The Great Global Warming Swindle” (26s + 1m 30s below).

This disagreement raises a number of questions. Did Fischer et al. really find that CO2 increased *after* temperature? Has Fischer et al.’s result been refuted? If CO2 didn’t initiate interglacials, then what did? Why did CO2 increase with temperature? And finally, if CO2 followed temperature changes, doesn’t that prove that CO2 doesn’t cause temperature changes?

Did Fischer really find that CO2 increased *after* temperature?


You can download the full Fischer et al. (1999) paper to read here, and it confirms the first half of hwat “The Great Global Warming Swindle” movie says, that CO2 increases come after temperature increases. Fischer et al. write that:

… during the penultimate warm period, CO2 concentrations reach their maximum 400 +/- 200 years later than Antarctic temperatures

and that:

The time lag of the rise in CO2 concentrations with respect to temperature change is on the order of 400 to 1000 years during all three glacial-interglacial transitions … The good agreement of the delta-age model with the measured value for the present supports the idea that at least the lag at the beginning of the warm periods is real.

Has the Fischer result been refuted?

Definitely not.

The Fischer et al. (1999) result has not only been strengthened by recent results, but it itself confirmed a long history of predictions and observations of a lag between CO2 and temperature.

This literature starts over 20 years ago, when Lorius et al.’s (1990) hypothesised that natural temperature perturbations caused changes to oceanic processes which in turn released CO2 into the atmosphere. This theme continued in the literature, when the lag was even acknowledged by the IPCC’s Assessment Report in 2007:

High-resolution ice core records of temperature proxies and CO2 during deglaciation indicates that antarctic temperature starts to rise several hundred years before CO2 … (Chapter 6 p 444)

The IPCC cites studies like Caillon et al. (2003) who “confirms that CO2 is not the forcing that initially drives the climatic system during a deglaciation”, and that observation continues to be upheld by more recent studies, (e.g. Clark (2009)).

In addition, Lorius et al.’s basic hypothesis - that natural temperature perturbations cause ocean processes to release CO2 into the atmosphere - continues to be the most widely-accepted theory of the glacial succession today (see Sigman & Boyle (2000) for a recent review).

So if CO2 didn’t initiate interglacials, then what did?

The short answer seems to be: Milankovitch cycles.

The theory is a fairly mathematical (Wikipedia has an easy-to-read introduction) but the basic idea is that the Earth’s orbital geometry is not perfect. For example, the Earth’s orbit around the sun is not perfectly circular but rather slightly elliptical, the Earth wobbles on its axis as it spins around the Earth, and the tilt of the Earth’s axis changes as well. These various wobbles and deviations from a perfect model happen at regular intervals, and they also enhance each other or partially cancel each other out depending upon their relative timing, leading to a rather complicated but regular and overall pattern. This overall pattern is known as the Milankovitch cycles.

The significance of the Milankovitch cycles to climate is that the wobbles cause changes in the amount and distribution of solar radition hitting the Earth. This is referred to as the orbital forcing.

Way back in 1976, building on the then-decades-old knowledge of the Milankovitch cycles, Hays et al. produced strong statistical modelling evidence linking the Milankovitch cycles to the succession of glaciations, and concluded that the Milankovitch cycles were the fundamental cause of the glacial transitions. The theory that Milankovitch cycles initiate glacial transitions can be traced in its development and confirmation through the literature - including the papers we’ve read above (Losius et al., 1990, Fischer et al. 1999, and Clark 2009) - and remains the theory with the strongest evidence, albeit incompletely understood. It is so well-accepted that the IPCC even acknowledged it as the strongest theory in its most recent report (see FAQ 6.1):

there is strong evidence that [the ice ages] are linked to regular variations in the Earth’s orbit around the Sun, the so-called Milankovitch cycles … These cycles change the amount of solar radiation received at each latitude in each season … and they can be calculated with astronomical precision … Climate model simulations confirm that an Ice Age can indeed be started in this way…

Aside: How do oceanic processes release CO2 into the atmosphere?

This doesn’t seem to be entirely clear.

In the IPCC’s words (Box 6.2 p 446) “The exact mechanisms behind the interplay between temperature and absorption and release of CO2 by the ocean is still being studied …”. The rest of that page summarises various theories and their pros and cons. I also found a review by Sigman & Boyle (2000, Nature) which is easy-to-follow and provides a summary of the state of recent theory. They favour a theory that involves changes in the upwelling of nutrients and productivity in the southern ocean, which is also acknowledged as having “a strong argument” for its importance by the IPCC (p 446).

If CO2 followed temperature changes, doesn’t that prove that CO2 doesn’t cause temperature changes?

So in summary, Milankovitch cycles change the distribution of solar energy hitting the earth, this causes temperature increases, and in turn the temperature increase causes a release of CO2 due largely to oceanic processes. This explains why CO2 increases follow temperature increases, and even the IPCC admits that all of this is true, so that’s the end of the story, right?

Well, not quite.

Let’s continue to read what Fischer et al. (1990) had to say about the lag in the very study that denialists are quoting:

Atmospheric CO2 concentrations show a similar increase for all three [glacial] terminations, connected to a climate-driven net transfer of carbon from the ocean to the atmosphere … Variations in atmospheric CO2 concentrations accompanying glacial-interglacial transitions have been attributed to climate-induced changes in the global carbon cycle (5, 6), but they also amplify climate variations by the accompanying greenhouse effect … [emphasis mine]

Fischer et al. (1990) is basically saying that, although the deglaciation was initiated by orbital forcing, the carbon (CO2 and CH4) released by the oceans into the atmosphere continued and amplified that initial warming via the greenhouse effect.

So why would they say that? We already have strong evidence that the temperature changes were initiated by orbital forcing, so why should we invoke CO2 and an additional explanation for temperature change as well?

The reason is that there is a problem with orbital forcing theory, and it is a “problem” that had been known from the very start; the magnitude of temperature changes caused by orbital forcing alone is not enough to account for the total warming and cooling over glacial transitions. It also doesn’t account for it’s specific shape (fast warming slow cooling), the dominance of the relatively weak 100,000 year orbital pattern, nor the similar timing and amplitude of climate change in both the Northern and Southern hemispheres.

The seminal Hays, Imbrie & Shackleton (1976) paper (that first one that really convinced the scientists of the role of orbital forcing) alluded to as much when they modelled the orbital forcing on climate “without identifying or evaluating the mechanisms through which climate is modified by changes in the global pattern of incoming radiation” and suggested that a full explanation including an explanation of the 100,000 year cycle’s dominance “probably requires an assumption of nonlinearity”.

As the coauthor Imbrie explained in a later paper (Imbrie & Imbrie 1980):

Croll (1) pointed out that orbitally driven variations in the annual energy influx … are on the order of … 0.1 percent. Like others who have considered the matter since … he concluded that the direct influence of these variations was probably too small to be detectable…

They continued saying that “early models [citing papers from 1969 - 1971] suggested that the climatic response to orbital changes was too small to account for the succession of Pleistocene ice ages”.

At that time, their view was that a new generation of models [1976 - 1979] relying upon a nonlinear response of Northern Hemisphere ice sheets might solve the problem, as alluded earlier in their Hays, Imbrie & Shackleton (1976) work.

However it didn’t turn out that way. Even when the nonlinear response of Northern ice was included, it was not enough to account for the magnitude of the warming, nor the other attributes of the record including the hemispheric synchronisation. By 1990, Lorius et al. observed that while “the main features of the Vostok temperature record … support the role of orbital forcing in Pleistocene glacial-interglacial cycles”, as Hays et al. (1976) showed, “The orbital forcing is, however, relatively weak when considered on an annual globally averaged basis”, and despite Imbrie & Imbrie’s (1980) expectation for an answer from new models

The amplification of this forcing, the observed dominant 100-kyr cycle and the synchronized termination of the main glaciations and their similar amplitudes in the Northern and Southern hemispheres cannot easily be explained despite developments including the nonlinear response of ice sheets to orbital forcing

But all wasn’t lost, because Lorius et al. (1990) had a new piece of information from the ice-cores - that the CO2 levels tracked glacial/interglacial temperature changes - which provided an explanation for the gap in orbital forcing theory. They continue:

The discovery of significant changes in climate forcing linked with the composition of the atmosphere has led to the idea that changes in the CO2 and CH4 content have played a significant part in the glacial-interglacial climate changes by amplifying, together with the growth and decay of the Northern Hemisphere ice sheets, the relatively weak orbital forcing and by constituting a link between the Northern and Southern Hemisphere climates. The observed changes in CO2 and CH4 imply modification in their sources and sinks that probably involve very different processes such as ocean circulation and marine production for CO2 …

And so this is exactly what Fischer et al. (1999) were referring to when they said that CO2 increases from temperature-induced changes in the global carbon cycle “also amplify climate variations by the accompanying greenhouse effect

If chickens have been observed to hatch from eggs, doesn’t that prove that chickens don’t lay eggs?

Lorius et al.’s (1990) basic idea - an initial orbitally forced temperature change subsequently amplified by the release of carbon from the oceans - remains the dominant theory for glacial transitions. The most recent IPCC report states (FAQ 6.1):

Although it is not [the glacial transition’s] primary cause, atmospheric carbon dioxide (CO2) also plays an important role in the ice ages. Antarctic ice core data show that CO2 concentration is low in the cold glacial times (~190 ppm), and high in the warm interglacials (~280 ppm); atmospheric CO2 follows temperature changes in Antarctica with a lag of some hundreds of years. Because the climate changes at the beginning and end of ice ages take several thousand years, most of these changes are affected by a positive CO2 feedback; that is, a small initial cooling due to the Milankovitch cycles is subsequently amplified as the CO2 concentration falls. Model simulations of ice age climate (see discussion in Section 6.4.1) yield realistic results only if the role of CO2 is accounted for [emphasis mine]

That final sentence there is very important. Not only did CO2 increase the temperatures during glacial terminations, but we cannot explain the full magnitude of the temperature changes without CO2.

The science behind CO2 and the greenhouse effect was already well-established by fields outside the study of ice-ages, and does not need the ice-core records to prove its validity (ask me later if you want to know more). However the realisation that it was the missing piece in the glacial transitions story is powerful confirmation of the theory and a solid empirical demonstration of CO2’s influence on global climate.

Further reading:

A page-turner of an account of the history of the study of glacial/interglacial transitions and how they shed light on the role of CO2 and the greenhouse effect:

A review by Sigman & Boyle (2000) which provides an easy-to-follow summary of the state of recent theory regarding the mechanisms behind CO2 response to temperature:

Jeff Severinghaus, Professor of Geosciences from Scripps Institution of Oceanography, answers a reader’s question about what the lag means:

An article from the popular science press basically laying out the same argument as presented here:


Caillon, N., Severinghaus, J.P., Jouzel, J., Barnola, J.-M., Kang, J. and Lipenkov, V.Y. (2003) “Timing of atmospheric CO2 and Antarctic temperature changes across Termination III.” Science 299: 1728-1731.

Clark, P.U. (2009) “The Last Glacial Maximum”, Science 325:710-714

Fischer, H., Wahlen, M., Smith, J., Mastroianni, D. and Deck, B. (1999) “Ice core records of atmospheric CO2 around the last three glacial terminations”, Science 283(5408):1712

Hays, J.D., Imbrie, J., Shackleton, N.J. (1976). “Variations in the Earth’s Orbit: Pacemaker of the Ice Ages”. Science 194 (4270): 1121-1132

Imbrie, J. and Imbrie, J.Z. (1980) “Modeling the climatic response to orbital variations”, Science 207:943-953

Lorius, C., Jouzel, J., Raynaud, D., Hansen, J. and Le Treut, H. (1990) “The ice-core record: climate sensitivity and future greenhouse warming”, Nature 347:139-145

Sigman, D.M. and Boyle, E.A. (2000) “Glacial/interglacial variations in atmospheric carbon dioxide”, Nature 407:859-869