In their seminal paper on the Vostok Ice Core, Petit et al (1999)  note that CO2 lags temperature during the onset of glaciations by several thousand years but offer no explanation. They also observe that CH4 and CO2 are not perfectly aligned with each other but offer no explanation. The significance of these observations are therefore ignored. At the onset of glaciations temperature drops to glacial values before CO2 begins to fall suggesting that CO2 has little influence on temperature modulation at these times.
As discussed at the end of this post, consideration of the geochemical cycles of CO2 and CH4 in ice, permafrost, terrestrial and oceanic biospheres and in deep ocean water during freeze – thaw glacial cycles suggests that it is inevitable that CO2 and CH4 are going to correlate with temperature in a general way. This correlation shows that CO2 and CH4 are controlled by temperature and so provides no evidence for CO2 or CH4 amplifying temperature signals that are linked to orbital cycles.
Figure 1 The location of Antarctica, Vostok and other ice core locations.
The Russian Vostok Antarctic base lies 1300 km from the S pole, close to the centre of the Antarctica continent at an elevation of 3488 m. It currently receives 2.6 mm precipitation per year. Average temperature is -55˚C and the record low is -89.2˚C which is below the freezing point of CO2. Vostok is one of the most hostile places on Earth.
There is a history of drilling various ice cores at Vostok. The main ice core, the subject of this post, was drilled in 1995. The Vostok ice core is 3310 m long and represents 422,766 years of snow accumulation. One year is therefore represented by only 7.8 mm of ice. Vostok is a cold, cold desert and the very slow ice accumulation rate introduces significant uncertainties to the data.
In addition to ice cores, Vostok is famous for the sub-glacial lake that lies beneath that has been mapped as one of the largest lakes in the world covering 14,000 sq kms. It is clearly a lot warmer under the ice than on its surface.
Figure 2 Vostok scenery
Data: Temperature, CO2 and CH4
In comparing the temperature, CO2 and CH4 signals in the Vostok ice core, it is important to understand that the temperature signal is carried by hydrogen : deuterium isotope abundance in the water that makes the ice whilst the CO2 and CH4 signals are carried by air bubbles trapped in the ice. The air bubbles trapped by ice are always deemed to be younger than the ice owing to the time lag between snow falling and it being compacted to form ice. In Vostok, the time lag between snow falling and ice trapping air varies between 2000 and 6500 years. There is therefore a substantial correction applied to bring the gas ages in alignment with the ice ages and the accuracy of this needs to be born in mind in making interpretations. Vostok data can be downloaded here.
Note that in all my charts time is passing from right to left with the “present day” to the left. The present day (year zero) is deemed to be 1995, the year that the cores were drilled. The GT4 time scale of Petit et al is used .
The methane concentrations in gas bubbles and temperature variations in Vostok are incredibly well aligned, especially at the terminations and return to glaciation when temperature variations are at their greatest. (Figure 3).
Figure 3 Methane and temperature variations. Note how methane and temperature are particularly strongly aligned at the terminations and during subsequent decline back to glacial conditions.
This shows that the ice age to gas age calibration is good. But does it show that methane variations of ±200 ppbV (parts per billion) are amplifying the orbital control of glaciations?
The fit of CO2 to temperature is actually not nearly so tight as for CH4. There is a persistent tendency for CO2 to lag temperature throughout and this time lag is most pronounced at the onset of each glacial cycle “where CO2 lags temperature by several thousand years”  (Figure 4).
Figure 4 CO2 and temperature appear well-correlated in a gross sense but there are some significant deviations. At the terminations, the alignment is as good as observed for methane. But upon descent into the following glaciation there is a time lag between CO2 and temperature of several thousand years. Petit et al  make the observation but fail to offer an explanation and to take the significance into account preferring to make instead unsupportable claims about CO2 and CH4 amplifying orbital forcing.
It is therefore no surprise that CO2 and CH4 show significant differences (Figure 5) with CO2 lagging CH4 in a fashion similar to the lag between CO2 and temperature.
Figure 5 CO2 lags methane in a manner similar to the lag between CO2 and temperature. This time lag requires an explanation rooted in the geochemical environments that are both emitting and sequestering these gases. Petit et al  devote surprisingly little space to explaining the physical processes behind the CO2 and methane variations at all.
Petit et al  appear to have been more eager to emphasise the similarities than to report the important differences…
The overall correlation between our CO2 and CH4 records and the Antarctic isotopic temperature is remarkable (r2 1⁄4 0:71 and 0.73 for CO2 and CH4, respectively). This high correlation indicates that CO2 and CH4 may have contributed to the glacial–interglacial changes over this entire period by amplifying the orbital forcing along with albedo, and possibly other changes.
In fact the high correlation is best explained by CO2 and CH4 both responding to temperature change as opposed to “causing it” and there is zero evidence from this data that amplification of orbital forcing has taken place, which is not to say that it has not happened.
Figure 6 provides an expanded view of the last glaciation where it can be seen quite clearly that there is a time lag of about 8,000 years between temperature falling and CO2 being pumped down. The temperature fell to glacial conditions (-6˚C) with CO2 at interglacial values (265 ppmV). Methane fell immediately with temperature but CO2 did not. This suggets that CO2 has little control over the main structure of the glacial cycle that is controlled by orbital forcing. There are similar time lags at the beginning of each glacial cycle (Figure 4). This is clearly an important and reproducible geological process or sequential combination of processes.
Figure 6 Detail of the last 150,000 years showing how CO2 lags temperature by about 8,000 years following the Eemian inter-glacial. Full glacial conditions were established with inter-glacial CO2 concentrations.
The cyclicity of the CO2 and methane needs to be interpreted in terms of flux, sources and sinks. When the concentration rises this shows that the rate of production exceeds the rate of removal and vice versa. Envisaging glacial cycles there are a multitude of processes that one can imagine influencing both CO2 and CH4 flux. For example, sea level rise and fall flooding or draining land, vegetation growth and decay, changes to soils, ice sheets and permafrost melting, changes in ocean bio-productivity, changes in ocean circulation, in particular thermohaline circulation.
CH4 and CO2 rise together with temperature at the terminations and it is tempting to suggest that the source for these two gases is the same. This is likely to be only partly true. The most prominent source for the CH4 is likely to be melting permafrost around and beneath melting northern hemisphere ice sheets. This will also release some CO2. The ice itself also contains small amounts of both gases. The most likely source for most of the CO2 is considered to be the oceans where warming seawater can hold less CO2. It is straight forward to explain the concordant rise of CH4 and CO2 with temperature at a time of rapid warming and ice sheet melting. When the warming halts so does the rise of CO2 and CH4, but then, with greenhouse gases at a maximum things turn colder. This alone suggests that greenhouse gases play a minor role in modulating glacial temperature and climate.
So why do CH4 and CO2 not follow each other down during cooling? There is not actually a sink for CH4. It is destroyed rather in the atmosphere by reaction with sunlight and oxygen to form CO2. The residence time is rather short, about 10 years. And so once added to the atmosphere it is quickly destroyed by conversion to CO2. The rapid warming that marks the beginning of an interglacial is normally followed in short order by rapid cooling. One can imagine the permafrost gradually freezing again, resulting in a reduction of the methane flux, the rate of destruction overtakes the rate of release and the concentration falls.
The large time lag for CO2 is not so easy to explain. At the termination and during the warming phase one has to imagine poleward migration and growth of forests. I can only guess that the mass of the terrestrial biosphere increases. I don’t know what may happen to the mass of the ocean biosphere which is often more productive in cold water? I can also speculate that thermohaline circulation is established or amplified enabling the partial degassing of the deep, carbon rich ocean. It is difficult to fit these pieces together in a quantitative way but suffice to say that warming leads to an increase in atmospheric CO2. So why does cooling not draw CO2 down again immediately?
An obvious thought is that this is linked to thermal inertia of the oceans. That the land and atmosphere had cooled with the oceans lagging a few thousand years behind. A simple way to check this was to compare Vostok CO2 against the ocean temperature record as recorded by the d18O signatures of globally distributed benthic foraminifera  (Figure 7). There is a similar time lag in the oceans between temperature (d18O) and CO2 (Figure 7) so the thermal inertia idea does not work.
Figure 7 There is a similar time lag between CO2 from Vostok and the temperature record of benthic foraminifera in the N Atlantic  showing that the slow pump down of CO2 has nothing to do with the thermal inertia of the oceans.
So what may actually be going on? A few months ago Roger and I had a series of posts on Earth’s carbon cycle. We never really got to the bottom of it but in the process learned a lot and turned up much interesting data. I made three interim conclusions 1) deep ocean water contains much more carbon than the surface, and because of this 2) the much publicised oceanic CO2 solubility pump cannot exist and 3) most CO2 is removed from the atmosphere by photosynthesis – trees on land and phytoplankton in the oceans . This may help us to understand the CO2 time lag. The deep oceans contain vast amounts of carbon, the product of rotting plankton at depth, and when the oceans warm or overturn, this C can be released to the atmosphere, quickly. But the return trip is not so simple since this depends on photosynthetic rates. In short, it seems that the oceans can exhale CO2 much more easily than it can be inhaled again.
On land, the re-creation of northern hemisphere ice sheets will kill high latitude forests and cause global migration of climatic belt boundaries towards the equator. Killing forests reduces the size of the terrestrial CO2 pump whilst simultaneously adding a source of CO2 – rotting wood. This will tend to offset the oceanic biosphere’s ability to pump CO2 down during the cooling phase.
- Over four glacial cycles CO2, CH4 and temperature display cyclical co-variation. This has been used by the climate science community as evidence for amplification of orbital forcing via greenhouse gas feedbacks.
- I am not the first to observe that CO2 lags temperature in Vostok  and indeed Petit et al  make the observation that at the onset of glaciation CO2 lags temperature by several thousand years. But they fail to discuss this and the fairly profound implications it has.
- Temperature and CH4 are extremely tightly correlated with no time lags. Thus, while CO2 and CH4 are correlated with temperature in a general sense, in detail their response to global geochemical cycles are different. Again Petit et al  make the observation but fail to discuss it.
- At the onset of the last glaciation the time lag was 8,000 years and the world was cast into the depths of an ice age with CO2 variance evidently contributing little to the large fall in temperature.
- The only conclusion possible from Vostok is that variations in CO2 and CH4 are both caused by global temperature change and freeze thaw cycles at high latitudes. These natural geochemical cycles makes it inevitable that CO2 and CH4 will correlate with temperature. It is therefore totally invalid to use this relationship as evidence for CO2 forcing of climate, especially since during the onset of glaciations, there is no correlation at all.
 J. R. Petit*, J. Jouzel†, D. Raynaud*, N. I. Barkov‡, J.-M. Barnola*, I. Basile*, M. Bender§, J. Chappellaz*, M. Davisk, G. Delaygue†, M. Delmotte*, V. M. Kotlyakov¶, M. Legrand*, V. Y. Lipenkov‡, C. Lorius*, L. Pe ́ pin*, C. Ritz*, E. Saltzmank & M. Stievenard† (1999) Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica. NATURE | VOL 399 | 3 JUNE 1999 |
 Jo Nova: The 800 year lag – graphed
 Lisiecki & Raymo (2005) A Pliocene-Pleistocene stack of 57 globally distributed benthic D18O records. PALEOCEANOGRAPHY, VOL. 20, PA1003, doi:10.1029/2004PA001071
 Energy Matters: The Carbon Cycle: a geologist’s view