The Data Adjustment Bureau

From Euan Mearns Energy Matters site:

The Data Adjustment Bureau

A few weeks ago I posted on The Vostok Ice Core and the 14,000 Year CO2 Time Lag. Blogger, “And Then There’s Physics” (ATTP) picked up on it and posted a critique. Some useful information came to light in the comments and I said I would respond via a separate post, and so here it is.

In summary, the main evidence presented to refute my findings was a 2001 Nature paper by Cuffey and Vimeux [1]. These authors make many of the same points as I do but set about solving the problem by adjusting the temperature scale in Vostok that is derived from dD in ice.

They try to create a better fit between T and CO2 but fail to do so completely. The adjusted T data still leads CO2 by about 5000 years at the inception. At the Eemian termination and inception the alignment between T and CH4 is EXACT. The temperature adjustment made by Cuffey and Vimeux will destroy that fine relationship and we have to conclude that the Cuffey and Vimeux data adjustment is not valid.

(note that dD is derived from the isotope ratio of  2H/1H where H = hydrogen and 1 and 2 = the atomic mass. 1H is “normal” hydrogen comprising a single proton + electron. 2H is heavy hydrogen comprising a single proton+neuton+electron.)


On reading ATTP’s critique I find that he is to some extent in agreement with what I say in that he accepts that CO2 and CH4 variations are feedbacks to the climate system. Where our views diverge significantly is that he argues that the greenhouse gas (GHG) feedbacks + albedo once they begin to change in a particular direction begin to force the glacial climate system. These are assumptions for which there is no evidence. To the contrary the data I present from Vostok proves it is false to make these assumptions. For starters, as pointed out in my post, at the glacial inception CO2 is high and albedo is low. Following ATTP’s logic, it should continue to get warmer, but it doesn’t. What happens next is that we get plunged into a new glaciation. At the terminations, CO2 is low and albedo is high and The Earth should according to theory get colder and colder. But no. What happens next is that the glaciation ends. This alone proves that CO2, CH4 and albedo are weak forces that do not overcome the strong force that modulates glaciations where the most likely candidate is changing obliquity combined with other orbital factors.

In this post I want to focus on a comment on ATTP by Professor Eric Steig. Here’s what he said:

If Mearns wishes to recycle an old, tired, logically inconsistent idea, the least he could do is update himself with the literature. He’s at least 16 years out of date. For starters, he needs to read Cuffey and Vimeux, 2001 ( and Pedro et al., 2012: Among other things, Cuffey and Vimeux show that the purported lack of correlation between CO2 and temperature at Vostok that Mearns refers (about how CO2 doesn’t fall for thousands of years after temperature does, at the end of the last interglacial) reflects the fact that deuterium/hydrogen ratios are not exactly temperature. Pedro et al. show that the lag is not 800 years. It’s more like 0.

The allegation that I am logically inconsistent has been echoing in my head for a few weeks now. Lets see how Professor Steig’s superior scientific credentials stack up.

The Peer Reviewed Evidence

Let me begin with the Pedro, Rasmussen and Ommen (2012) [2] article:

Tightened constraints on the time-lag between Antarctic temperature and CO2 during the last deglaciation

In my Vostok post I say this in the opening paragraph:

At the termination, CO2 follows dT exactly, but at the inception CO2 does not follow temperature down for 14,218 years. Full glacial conditions came into being without falling CO2 providing any of the climate forcing.

Professor Steig does not seem to have grasped that time lags in Vostok occur on at least two very different time scales, evident from his comment above. It is also not clear that he has grasped the radically different geochemical responses at the glacial inceptions and terminations. At the terminations the time lag of several hundred years is not controversial. And in my post I said this:

Time lags of a few hundred years can be explained by dating errors and calibrating the gas ages to the ice ages (see my earlier post). This has caught the attention of the climate science community and for example Parrenin et al (2013) revised the gas ages in Vostok over the time interval 10,000 to 22,000 years ago that spans the termination that preceded the Holocene interglacial where the established chronology already showed good alignment between CO2 and T (Figure 1). Parrenin et al conveniently forgot to examine the relationship at the inception where the time lag of several thousand years is so large, it is not possible to explain it by a calibration error.

The Pedro et al paper addresses the small time lag at the glacial termination that is not contested by me. Nor is it relevant to my point of focus that is the glacial inception where the time lag is over 14,000 years.

The second paper Steig links to is more relevant and more interesting.

Covariation of carbon dioxide and temperature from the Vostok ice core after deuterium-excess correction (2001) Kurt M. Cuffey & Francoise Vimeux.

Their opening paragraph says this:

As calculated from Vostok ice-core data , the correlation strength of CO2 and dD for the past 150 kyr is r2 = 0.64, and this is limited primarily by the rapid decrease of dD during and immediately following the last interglacial, 100-130kyr before present, BP (Fig. 1). If this dD change really reflects a proportional temperature drop, then more than half of the interglacial-to-glacial temperature change occurred before significant removal of atmospheric CO2, implying a minor contribution of CO2 to the glacial-age climate forcing.

This is the exact point I have made in the two posts I have written on the subject and here we have it confirmed by a publication in Nature.

In summary, what Cuffey and Vimeux have done is to adjust the temperature scale in Vostok in an attempt to make the mismatch between temperature and CO2 go away. In my earlier posts I noted that the mismatch was so large that this could not be fixed by adjusting the time scales and ice age to gas age calibrations, also noting that CH4 and temperature are closely aligned at the inception. Changing the time scale would ruin that excellent alignment.

The result of their data adjustment is illustrated in Figures 1 and 2. Their Figure 1 is analogous to my chart that is reproduced below as Figure 3. Their Figure 2 shows the adjusted temperature curve.

Figure 1 The variation in dD (T) and CO2 throughout the last glaciation as recorded by the original Petit et al paper [3]. Compare with Figure 3 and you will see that my rendering of the data is identical to that shown here. The offending data is that centred on 110,000 years where T falls significantly unaided by falling CO2. Cuffey and Vimeux view this as a serious problem and set about fixing it by adjusting the dD / dT data. At no point do they consider the possibility that the obseravtions are correct and the global warming theory driven by dCO2 could be wrong.

Figure 2 Analogous to Figure 1 but this time showing T adjusted for what is considered spurious dD behaviour. The main adjustment actually occurs across the peak of the Eemian interglacial. The correction has the effect of extending the duration of the Eemian to about 20,000 years.

Figure 3 Temperature and CO2 comparison in the Vostok ice core [1] from my original post that shows the exact same features as discussed by Cuffey and Vimeux for the interval 160,000 years to present.

It is not easy to explain the science behind what Cuffey and Vimeux have done. The starting point is the fact that deuterium (D) measured in ice is not directly temperature but is instead a proxy for temperature. To convert dD to a temperature requires a series of assumptions about the behaviour of D in the atmosphere where the water vapour was originally incorporated as clouds and during transport to the South Pole and eventual deposition as snow. Suffice to say that converting dD to dT is somewhat a black art.

Cuffey and Vimeux hypothesise that “excess D” is the problem in Vostok and apply a non-linear correction to the data that greatly improves the match of dCO2 to dT (Figure 2). The justification for doing so is derived from an anomalous relationship between dD and d18O. This could be valid but here we hit a problem. I have not been able to find any d18O data for Vostok ice and Cuffey and Vimeux do not present any either.

Information about source-region climate variations is thus essential for quantitative interpretations of ice-core dD, and, following previous work (6,11,12) we obtain this information from non-standard covariation of dD and d18O, defined as the deuterium excess (d = dD – 8d18O; we will refer to this as `the excess’). Theory predicts (11-13) that the excess of polar precipitation varies in response to source-region temperature and relative humidity (see Methods), and that source-region temperature signals can be separated from other in ̄influences using marine and ice-core isotopic data (70. Calibrated versions of this theory successfully explain major features of the global distribution of excess in precipitation (14), including polar snow (11,12).

Its seems from the methods section that Cuffey and Vimeux have used a d18O model for the evaporative source area to correct the data and at this point I become immediately sceptical about what has been done. This does not appear to be a data-driven but rather a model-driven correction.

For the past 150 kyr we estimate Ddm directly from radiometrically dated sea-level history (23-27) and calibrate this to the Last Glacial Maximum ocean composition inferred from pore water (9) (see Supplementary Information). For the older record (150±350 kyr BP), sea-level data of this quality are not available, and we instead use the foram d18O history (28), which is less desirable because it includes temperature effects. The GT4 timescale is used for the Vostok data (1).

Fortunately, it is not necessary to have a detailed critique of the methodology employed to discount this work since two major obstacles remain.

First, the opaque methodology improves the match of dT and dCO2 throughout much of the core but at the key interval at the inception the mismatch remains. And we now have a large bounce in temperature at about 150,000 years in an interval where CO2 does not change at all. In their concluding sentence, Cuffey and Vimeux say this:

A further result relevant to carbon-cycle studies is that much of the very long delay (10-15 kyr) and disproportionate magnitude of the CO2 decrease relative to the dD decrease during the last glacial inception is an artefact. The actual delay between CO2 decrease and temperature change was probably not more than 5 kyr: such a delay is much more easily explained in terms of the time required for ocean circulation response and deep-water ventilation. None-the-less, the correspondence of CO2 and temperature during the glacial inception was not as strong as during the last deglacial transition (22).

This large mismatch remains a showstopper. During the last glacial inception, temperatures fell and rose significantly unaided by CO2. Let us remind ourselves what Professor Steig had to say:

Pedro et al. show that the lag is not 800 years. It’s more like 0.

Maybe, but Cuffey and Vimeux conclude that the lag remains a stonking 5000 years.

Second, as mentioned on numerous occasions, CO2 and CH4 are closely aligned throughout much of the Vostok ice core but at the inceptions they are not aligned. This can be explained by different biogeochemical response times of CH4 and CO2 in response to cooling.

Figure 4 Comparison of T and CH4 in the Vostok ice core. Note how tight this covariance is across the Eemian interglacial providing strong evidence that the T curve does not need to be adjusted.

It is important to look closely at the relationship between CH4 and T at the last inception (that is the descending T curve below and to the right of the word Eemian). The profiles match EXACTLY as they do at each of the terminations and inceptions (note at the inception: CH4 and T are exactly aligned; CH4 and CO2 are not aligned, therefore CO2 and T are not aligned). For me this is wonderful geochemical data that shows CH4 is more responsive to temperature change than CO2. Terrestrial processes dominate CH4 while CO2 is heavily influenced by oceanic processes that introduces very large time lags at the inceptions. The Cuffey and Vimeux correction to T will utterly destroy this fine relationship between T and CH4 from which I conclude the adjustments are invalid. I rest my case.


I am interested to collect more examples of climate data being adjusted to make it match the theory, if commenters can help me out. I’m aware of the temperature adjustments in Iceland and the near ubiquitous adjustments to global land temperature-time series brought about by homogenising data. I’m sure there must be more examples.


[1] Kurt M. Cuffey & Francoise Vimeux (2001) Covariation of carbon dioxide and temperature from the Vostok ice core after deuterium-excess correction; Nature 412

[2] J. B. Pedro, S. O. Rasmussen and T. D. van Ommen (2012) Tightened constraints on the time-lag between Antarctic temperature and CO2 during the last deglaciation; Clim. Past, 8, 1213–1221

[3] 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 |

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s