RealClimate logo

What does the lag of CO2 behind temperature in ice cores tell us about global warming?

Filed under: — group @ 3 December 2004 - (Français)

This is an issue that is often misunderstood in the public sphere and media, so it is worth spending some time to explain it and clarify it. At least three careful ice core studies have shown that CO2 starts to rise about 800 years (600-1000 years) after Antarctic temperature during glacial terminations. These terminations are pronounced warming periods that mark the ends of the ice ages that happen every 100,000 years or so.

Does this prove that CO2 doesn’t cause global warming? The answer is no.

The reason has to do with the fact that the warmings take about 5000 years to be complete. The lag is only 800 years. All that the lag shows is that CO2 did not cause the first 800 years of warming, out of the 5000 year trend. The other 4200 years of warming could in fact have been caused by CO2, as far as we can tell from this ice core data.

The 4200 years of warming make up about 5/6 of the total warming. So CO2 could have caused the last 5/6 of the warming, but could not have caused the first 1/6 of the warming.

It comes as no surprise that other factors besides CO2 affect climate. Changes in the amount of summer sunshine, due to changes in the Earth’s orbit around the sun that happen every 21,000 years, have long been known to affect the comings and goings of ice ages. Atlantic ocean circulation slowdowns are thought to warm Antarctica, also.

From studying all the available data (not just ice cores), the probable sequence of events at a termination goes something like this. Some (currently unknown) process causes Antarctica and the surrounding ocean to warm. This process also causes CO2 to start rising, about 800 years later. Then CO2 further warms the whole planet, because of its heat-trapping properties. This leads to even further CO2 release. So CO2 during ice ages should be thought of as a “feedback”, much like the feedback that results from putting a microphone too near to a loudspeaker.

In other words, CO2 does not initiate the warmings, but acts as an amplifier once they are underway. From model estimates, CO2 (along with other greenhouse gases CH4 and N2O) causes about half of the full glacial-to-interglacial warming.

So, in summary, the lag of CO2 behind temperature doesn’t tell us much about global warming. [But it may give us a very interesting clue about why CO2 rises at the ends of ice ages. The 800-year lag is about the amount of time required to flush out the deep ocean through natural ocean currents. So CO2 might be stored in the deep ocean during ice ages, and then get released when the climate warms.]

To read more about CO2 and ice cores, see Caillon et al., 2003, Science magazine

Guest Contributor: Jeff Severinghaus
Professor of Geosciences
Scripps Institution of Oceanography
University of California, San Diego.

Update May 2007: We have a fuller exposition of this on a more recent post.

4 Responses to “What does the lag of CO2 behind temperature in ice cores tell us about global warming?”

  1. 1
    alex says:

    Although I don’t have data to support the following claims, and can’t recall specific papers on the topic (undoubtedly my following thoughts are some sort of compilation of papers I have read), a reasonable guess for the lag has to be based on the biological response of phytoplankton to changing environmental conditions. Offhand, it seems reasonable that as the Milankovitch (orbital) forcings initiate the onset of the glacial termination there may be some sort of decrease in the zonal sea surface temperature gradient which would in turn lead to less zonal winds. Less zonal winds would seem to reduce vertical mixing in the surface ocean and induce a subsequent stratification of surface waters. The surface waters may be additionally stratified by an increase in melt water remaining at the surface (reducing the salinity of surface water, consequently increasing water column stability). These combination of these factors do not bode well for carbon fixers such as phytoplankton who are dependent on intense vertical mixing as a source of upwelled nutrients. Less carbon fixation would result in more CO2 remaining in the atmosphere as the glacial termination proceeds, acting as a positive feedback to the initial forcing. Of course, this sort of large scale response would take a long time to occur, and may in fact not happen until well into the glacial termination. Just some thoughts. Perhaps someone with more study in the field could present the conventional wisdom on this topic.

  2. 2
    Ferdinand Engelbeen says:

    May I disagree with this article?

    The correlation between CO2 and temperature in the pre-industrial 420,000 years, according to the Vostok ice core is surprisingly linear (some 8 ppmv for 1°C, see: ) and includes shorter (~800 years) for glacial-interglacial transitions and longer (up to thousands of years) lags for CO2 vs. temperature for interglacial-glacial transitions. See e.g. the previous interglacial at:

    While the fast glacial-interglacial transitions may hide which leads what and to what extent, the much slower (depending of the length of the interglacial) interglacial-glacial transitions make it clear. When the temperature decreases (7°C), CO2 levels remain high. When CO2 levels fall (some 50 ppmv), there is no measurable effect on temperature at all. This contradicts the possibility that CO2 plays an important role in the onset of both glaciations and deglaciations.

    Response The above statement is incorrect. The correlation being discussed is between CO2 and deuterium/hydrogen isotope ratios in snow (as archived in an ice core), an imperfect measure of temperature. It has been shown that when the deuterium/hydrogen ratios are corrected for the influence of temperature changes at the ocean surface (from which the water that fell as snow originally evaporated), the correlation becomes even more remarkable, and the apparent drop in temperature thousands of years before the drop in CO2 disappears. See Cuffey and Vimeux, Science, 2002.

    Further, even if we assume that the average of current climate models is right, the temperature increase from a CO2 doubling is around 3°C, or ~1°C for a 100 ppmv rise. Which is 1/10th of the >10°C rise seen in the last deglaciation.

    Moreover, the Taylor Dome ice core, also reveales a lag of 1200 +/- 700 years between CO2 and temperature on shortes time scales (D-O events). See:

    Rapid temperature swings like the end of the Younger Dryas (probably less than a few decades) are followed by CO2 changes.

    Even the past 1,000 years, showed a lag of ~50 years of CO2 vs. temperature for the Law Dome ice core, but the temperature data disappeared from the Internet…

    And since the industrial revolution, sea surface temperature changes, like El Niño, induce peaks in the CO2 increase rate, some 6 months after the onset of the event…

    Thus all together, all historic data point to a lag of CO2 after temperature changes, without much influence of CO2 on temperature when that happens…

  3. 3
    David Holland says:

    Wow! Are you really saying that we have no idea what starts to warm up our world from an ice age but know with near certainty what has caused the warming of the last three decades?

    From my now somewhat distant scientific education I recall that it takes some 80 times more heat to turn the ice to water than to raise its temperature by a single centigrade degree. With sea levels 125m or so lower a significant proportion of the planets water must have been in the form of ice. The â??unknown processâ?? you refer to would have had to supply far more extra heat than the CO2 feedback, which was able to take over some 800 years later.

  4. 4

    Re: Climate Change Deni