By Don J. Easterbrook, PhD.
In a paper “Global warming preceded by increasing carbon dioxide concentrations during the last deglaciation”, Shakun et al.(Nature 2012) contend that rising temperature at the end of the last Pleistocene glaciation were preceded by increasing atmospheric CO2. In his usual masterful fashion, Willis Eschenbach has dug deeply into the data used in the paper and shredded the conclusions in it (see http://wattsupwiththat.com/2012/04/06/a-reply-shakun-et-al-dr-munchausen-explains-science-by-proxy/
and http://wattsupwiththat.com/2012/04/07/shakun-redux-master-tricksed-us-i-told-you-he-was-tricksy/#more-60932/. So rather than dwell on the things that Willis has already shown so well, I thought I’d take a look at some of the assumptions and misconceptions that paper is built upon.
When reading a paper like this, I always like to ask myself, what are the basic assumptions that underlie the methodology involved? What contentions are simply stated as fact or generated in a computer model, rather than demonstrated with real, physical evidence? I will confess here that I don’t believe computer models really prove anything. Sure, they can suggest many things and point out areas of interest, but I live the real world and prefer real physical evidence upon which to base important conclusions. That doesn’t mean I discount models out of hand—it simply means that I look for physical evidence to confirm or deny what the models are saying. So I asked myself a series of questions about the basic issues in this paper. Here are some of the questions that I came up with (the answers follow).
1. Can the Antarctic ice cores be dated with sufficient accuracy to establish a firm temperature chronology?
2. Are the 80 temperature proxies used in the paper sufficiently accurate to establish a solid global temperature chronology?
3. Can CO2 in the ice cores be measured with validity and accuracy?
4. Can the difference in the age of the trapped air and the age of the enclosing ice be determined and is it constant with age?
5. Are CO2 measurements from air bubbles valid or do diffusion and the uncertainty in the timing of isolation of air in bubbles render them invalid?
6. Is the data from Antarctic ice cores consistent with data from the Greenland ice cores?
7. Is the temperature chronology of the ice cores and global proxies consistent with the well-dated, global glacial record?
8. Is the so-called ‘see-saw’ of climate changes between hemispheres valid, i.e, are climate changes in the Northern Hemisphere out of phase with those in the Southern Hemisphere?
9. Would correlation between CO2 and temperature necessarily prove that CO2 causes climatic warming?
10. Since CO2 is incapable of causing climatic warming by itself (CO2 makes up only 0.038% of the atmosphere and accounts for only a few percent of the greenhouse gas effect), is there evidence of concomitant increase in water vapor (which causes more than 90% of the greenhouse gas effect)?
11. Is the AMOC the only viable causal mechanism? What about the influence of the Pacific Ocean, which covers about half the Earth’s surface
So, what is the main contention of this paper and what does it imply? The authors claim to have “compelling evidence that rising CO2 caused much of the global warming” at the end of the last ice age, roughly 11,000 to 25,000 years ago. According to the authors, “if you reconstruct temperatures on a global scale – and not just examine Antarctic temperatures – it becomes apparent that the CO2 change slightly preceded much of the global warming, and this means the global greenhouse effect had an important role in driving up global temperatures and bringing the planet out of the last Ice Age.” The crux of their contention is illustrated in their Figure 2.
Shakun et al. Figure 2. The Red line is Antarctic temperature curve based on ice cores; the yellow dots are CO2 measurements from ice cores; the blue line is composite global temperature from 80 proxies.
Willis has sliced and diced the data behind these curves so be sure to read his analyses. I’ll refer to some of his graphs and conclusions but look at the Shakun et al. contentions from a somewhat different angle. Because this is such a marked divergence from the widely held view that CO2 lagged rising temperatures at the end of the last ice age, careful scrutiny must be given to evidence and assumptions upon which this contention is based. Right off the bat, a most surprising conclusion in this paper is that the authors claim that correlation proves cause. Simply showing that CO2 correlates with anything surely doesn’t prove that CO2 was the cause. It’s the same kind of mindset involved with the oft-heard claim that if we have had global warming while CO2 was rising that proves the cause was the rise in CO2. Heck, I had hair before CO2 began to rise, but I don’t blame that on CO2.
So let’s look at each of questions posed above.
- How accurate is the dating of Antarctic ice cores? How can you date ice that has nothing in it that can be directly dated? The Shakun et al. paper states that they use the methodology of Lemieux-Dudon et al. (2010), which involves construction of a model using estimates of snow accumulation rates, temperature, firn densification rates, and ice flow rates, all of which vary from glacier to glacier and from glaciation to interglaciation (thus introducing large potential errors). The modeling data is then modified by matching with tephra horizons, sulfate spikes, δ18O, firn densification model results, and orbital tuning. All of the assumptions built into the modeling are cumulative, resulting in large possible age errors. As Lemieux-Dudon point out “One special feature of glaciological models is a large model error due to unresolved physics and errors on the forcing fields, clearly affecting the quality of the inferred dating scenarios.” What this means of course is that the age determinations of the Antarctic cores are, at best, educated guesses with large uncertainties. Because chronology is so critical to the Shakun et al. contention, the ages of the Antarctic cores shown in their Figure 2 cannot be considered accurate.
- Are the 80 temperature proxies used in the paper sufficiently accurate to establish a solid global temperature chronology? Willis Eschenbach has made a detailed analysis of the data used to construct the global temperature curve in Figure 2 of Shakun et al.(see this in his web posting) He plotted individual curves for each of the 80 temperature proxies used to create Figure 2 in the Shakun et al. paper. What he found was large variability in the data, which led him to conclude that “The variety in the shapes of these graphs is quite surprising Yes, they’re all vaguely alike, but that’s about all. The main curiosity about these, other than the wide variety of amounts of warming, is the different timing of the warming.” When he ploted all the individual proxies all together (see below), the scatter is readily apparent, leading him to conclude: “It’s clear that there is warming since the last ice age.” “But if you want to make the claim that CO2 precedes the warming? I fear that this set of proxies is perfectly useless for that. How on earth could you claim anything about the timing of the warming from this group of proxies? It’s all over the map.”
===============================================================
Dr. Easterbrook requested this correction below saying:
As one of your readers pointed out, Willis used ‘Year’ for his time scale (meaning years BC, rather than years BP). I didn’t notice this (geologists always use years BP for events older than a few thousand years), so there really isn’t a discrepancy between the Shakun global curve and Willis’s data points. That graph and the text with it should be replaced with the attached file “dje response to Nature paper x.doc”) or it can just be removed from the posted version. That also means that the YD shown vertical time lines in the previous graph needs to be moved over 2000 yrs so we might as well just Willis’s graph (see attached file).
Sorry for the glitch–my fault–I should have caught it, but the thought never occurred to me that Willis would use 25,000 years BC. It doesn’t change any other of the other material.
Large scatter of individual data points on Willis’s plot from the 80 proxies used in the construction of the Shakun et al. temperature curve. I’ve added lines to show the age of Younger Dryas interval, which doesn’t correspond to the dip in the Shakun et al. temperature data.
Just for fun, I superimposed the curves on Shakun et al. figure 2 over Willis’s data point plot (see below). Because the global temperature curve (the blue curve) was presumably derived from the data in Willis’s plot, it should fit well with it. Interestingly, it doesn’t. I’ve shown with a blue arrow the dip in temperature that corresponds to the Younger Dryas and a black arrow pointing to what should be the same dip in temperature on the plot of individual data points. Other arrows point to similar differences for the end of the Younger Dryas. Now you would think that since the Shakun et al. blue curve was constructed from the individual data points shown on the graph, the two should surely be compatible! I’ve also shown on the graph the well-established age of the Younger Dryas—note that the Shakun et al. global temperature data points show a dip in temperature (presumably the Younger Dryas) that is considerably younger. Makes you wonder!
==============================================================
3. Can CO2 in the ice cores be measured with validity and accuracy?
4. Can the difference in the age of the trapped air and the age of the enclosing ice be determined and is it constant with age?
5. Are CO2 measurements from air bubbles valid or do diffusion and the uncertainty in the timing of isolation of air in bubbles render them invalid?
Because these questions are all inter-related let’s consider them together. The validity of measurement of CO2 from bubbles in ice cores has been challenged in a number of studies. There are several basic problems: (1) air becomes trapped in ice during the conversion of snow to firn to ice. Air in the snow/firn phase remains in contact with surface air until it turns to ice and seals off air bubbles from further mixing with surface air. The depth at which sealing occurs varies considerably, depending on the rate of firn densification, and may extend to more than 100 meters and take a thousand years or more. This means that the age of air in a bubble is not the same as the age of the inclosing ice. Snow densification rates vary considerably between temperate and polar glaciers and between glacial and interglacial climates, making it difficult to measure and date adequately. In any case, rates are not likely to be constant. (2) a second problem results from possible diffusion along the walls of an air bubble, which can upset the CO2 concentration in the bubble. These and other problems mean that measurement of CO2 in ice cores is not straight forward—measurement of CO2 concentrations in ice bubbles and determination of the age of the air are likely to be quite variable. General trends are apparent in CO2 ice core measurements, but variability in CO2 concentrations and age remains problematic.
At this point, answering the remaining questions is quite obviously going to take some time, so they will be considered in Part 2, coming soon.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Heck, I had hair before CO2 began to rise, but I don’t blame that on CO2.
I too had hair before CO2 began rising – but can I now blame my lack of it to rising CO2 levels – ie to be inversely proportional to CO2 concentrations?
Some decades ago, my geophysical colleagues developed a model of how a concealed discrete body of magnetic rock would disturb the normal flux pattern of the Earth’s magnetic field. This started in the days when calculations were done on a mechanical machine with a handle that you pulled to perform a preset calculation. The hard work was completed in the years before the PC sat on so many desks.
By marking out a surveyed grid (with altitudes) and using a sensitive magnetometer, the natural field was mapped, usually at an area where airborne survey showed a discrete anomaly. Then, hypothetical bodies were fitted, usually with an ellipsoidal shape, but with length and direction of X,Y and Z axes variable, as well as the angles between them. There was usualy a need to drill for a sample to get an idea of remanent magnetism so that work could be done on the induced magnetism perturbations. After modelling several hundred discrete bodies near Tennant Creek, Northern Territory, the geophysicists became uneeringly good at predicting the position and disposition of the body, whose top could be 300 meters below land surface.Thus, about 5 new gold/copper/bismuth mines were found, some of them quite valuable, but invisible to other surface exploration methods. The originators were mainly Mr L.A. Richardson and C.B. Kirkpatrick.
This is mentioned in the context of the sentences above, “I will confess here that I don’t believe computer models really prove anything. Sure, they can suggest many things and point out areas of interest, but I live the real world and prefer real physical evidence upon which to base important conclusions.”
My words here are give hope that models can go fairly close to a ‘proof’ if perturbing variables are understood well enough to be eliminated or compenstated.
As an aside, one can philosophise that a reason for success was the need for each geophysicist to work with the data manually, over and over, to get a personal ‘feel’ for it. There are dangers in the closed, black box approach that seems to get more emphasis these days than getting hands dirty from actual data.
It has been shown recently that CO2 diffuses through ice (Scripps Oceanographic Institute). It follows Fick’s Second Law of Diffusion. Very slowly, but we are talking kiloyears here. It diffuses from areas of high concentration to low concentration. SO high concentration bubbles will diffuse toward the outside, to be in equilibrium ONLY when the concentration in the bubble is equal to concentration outside. Well-known physics. Nothing can be really trapped! It is properly said, then, that the situation is metastable. Even mylar balloons lose helium in a short while, and mylar is very impermeable, relatively speaking.
Every gas diffuses through every solid. Even isotopes become concentrated in the heavier isotope, even though the difference is an atomic weight or two.
Diffusion is a Physical Law.
ALSO, when the ice core is removed from “down under”, where there exists high pressure at the bottom of the ice column, the pressure is removed, and the ice bubbles, retaining the pressure from below, diffuse faster toward lower pressure at the surface. Diffusion then becomes directly proportionally increased in accordance with the pressure gradient. This is additive to the diffusion gradient.
Conclusion: Unless one instantaneously tests the bubbles, AND goes back in time to when the bubbles were deposited, one can never know the original concentration.
Of course, if one could go back in time, why not just test the air? Measure temperature while we are at it?
I think the last two figures are incorrect. Seems that the Shakun diagram is in time before present while Eschenbach’s plot is in calender year.
Well said otherwise.
The difference between the timing of the YD and temperature proxies’ drop makes me nervous; this isn’t just a bookkeeping error confusing BP and BC dating, is it? Don’t laugh; engineers can even confuse km and mi. when setting Mars Probe rocket decel timings ….
maybe the co2 came out of the cold polar waters when it warmed rather than from the warmer equatorial ones that were already depleted?
once a soda is flat, it can’t fizz any more.
The proxies themselves often take time to respond and adjust to environmental changes, and these responses vary greatly between different proxies and between different locations, therefore it is not valid to use globally averaged proxies to determine timing relationships beyond a certain error margin.
A basic example would be sea level, it takes time to melt ice at a certain temperature, so therefore any measurement of palaeo-sea level would be offset from the period of warmth with which it is related, as the ice that would melt from such warmth would take a certain period of time to then effect the sea level (and then effect other proxies such as corals, and so on). You can also get negative and flatline responses once certain point is reached (e.g with tree rings and warmth), as well as feedback effects which can also increase the level of response. These sort of things vary greatly between proxies, and therefore any strong conclusions from averaged analyses of past proxies must be treated with utmost caution; in particalar conclusions regarding timing relationships and magnitude.
Academics know this, its just that they sometimes forget, to get a paper or to try and make a point.
But the same goes for the whole AGW thing, my main problem with it is that I think they have got the general response rates of the earth wrong, for much the same reasons as illustrated above. The earth will warm much more slowly than their models suggest (think hundreds to thousands of years, not decades).
Geoff Sherrington says:
April 8, 2012 at 8:21 pm
…
As an aside, one can philosophise that a reason for success was the need for each geophysicist to work with the data manually, over and over, to get a personal ‘feel’ for it. There are dangers in the closed, black box approach that seems to get more emphasis these days than getting hands dirty from actual data.
More importantly your colleagues ground-truthed the model by boring at the anomally locations.
Don: On your last points about dating the carbon dioxide and ice there are two papers that cast doubt on the use of isotopic oxygen and carbon for this purpose. The first examines the use of 18O:16O ratio to date the ice cores. This suggests that as 18O is heavier than 16O less is evaporated from the oceans and so less ends up in the snow that makes up the ice cores.
http://chiefio.wordpress.com/2012/03/02/doubting-oxygen-isotopes/
The second compares the determined ice core carbon dioxide values from 13C:12C ratio with contemporary values from plant stomata.
http://wattsupwiththat.com/2012/03/03/why-william-d-nordhaus-is-wrong-about-global-warming-skeptics-being-wrong/#more-58200
Re Geoff Sherrington says:
April 8, 2012 at 8:21 pm
The mechanical calculator you describe might be the Curta, called the coffee grinder. Here is one of many links:
http://www.vintagecalculators.com/html/the_amazing_curta.html
I wonder what the effects of algae and fungus growing at the surface of the ice has on CO2 distribution? In the desert, fungus pumps nutrients towards the surface of the soil.
Don Easterbrook asks “Can the difference in the age of the trapped air and the age of the enclosing ice be determined and is it constant with age?” He notes that “the age of air in a bubble is not the same as the age of the inclosing ice.”
The study by Caillon et al. (2003) says “One way to circumvent this difficulty is to use records of atmospheric CO2 content and temperature contained only in the trapped gases.”
This graph http://www.friendsofscience.org/assets/documents/FOS%20Essay/CaillonTermIII%203.jpg
from the paper Caillon et al. (2003) finds that the CO2 increase lagged Antarctic deglacial warming by 800 years. The authors measured the isotopic composition of argon40 and CO2 concentration in air bubbles in the Vostok core during the end of the third most recent ice age (Termination III), 240,000 years before the present. The graph shows CO2 and Argon40 isotope, with the time scale shifted by 800 years to present the best fit overlay. Both the CO2 and the Argon40 are gases extracted from the same air bubble in the ice core. The Argon40 is not in the surrounding ice. There can be no time lag between these two measurements. Deuterium in the ice is often used to measure the ice temperature. The Argon40 has a R2 correlation factor with deuterium of 0.85, indicating that Argon40 is an excellent temperature proxy of the trapped air. The temperature is not recorded when the snow falls, but both temperature and CO2 are recorded at some later time when there is no longer exchange with the atmosphere.
The conclusion is that temperature changes cause the CO2 changes. Whatever caused the temperature change must be many times more important as a temperature driver than the change in CO2 because for 800 years, temperatures are declining while CO2 is increasing. A link to the Caillon paper and further discussion is at:
http://www.friendsofscience.org/assets/documents/FOS%20Essay/Climate_Change_Science.html#Lead
In a comment response to one of his articles, Willis indicated that he used a BC/AD timeline vs the BP timeline used by Shakun. Did you account for that or does that ~2000y difference explain your YD offset. Terry
An avalanche search dog can sniff human scent through several feet of snow. If scent (gases related to bacteria on epithelials) can percolate through the snow in minutes, I’d expect CO2 would move through snow also. As snow is compressed, air is expelled to the surface, leaving only a relatively small example of the air remaining between the ice crystals. But that remaining air should represent a very similar relative composition of the various gases. However, if CO2 diffuses through ice (as mentioned previously), precise dating and quantifying the relative concentration at any given time would be very fuzzy indeed.
All this is akin to measuring mass differences of micrograms with a spring scale and using the average of several thousand measurements out to four decimal places.
dead on deadwood!
A couple of points. The proof that C02 causes warming is not from corelation. Its from experiemental evidence and basic engineering.
The question is how much warming.
The paper argues this.
1. The INITIAL cause of the warming was a change in orbital parameters. THEN
2. Excess warmth in NH. Then
3. ICE melting and a fresh water pulse THEN
4. A warming in the SH, leading to
5. MORE c02 outgassed from the ocean.
C02 both leads ( enhances) and lags ( as a feedback) the temperature rise.
It doesnt really change our view of things but adds in a few more details about the sequence.
No study of paleo climate change change the physics of c02 causing warming. It does. how MUCH is the key question.
To that end. The IPCC argues that the mean number for climate sensitivity is 3C . Whats that mean. That mean if we have 3.7Watts of addditional forcing ( from the sun or C02 ) we will see 3C of temperature rise. And further, doubling C02 from 280 to 560 will cause 3.7W of forcing.
Is sensitivity 3C as Hansen and the IPCC claim?
This study by Shakun implies a LOWER number actually. It implies 2.5C
Yes. Shakun adds another win for Lukewarmers.
When I drew attention to the CO2 and temperatures issues around the YD in Willis’s thread, I noted that CO2 appears to flatten before the YD start and rise before the end of the YD.
Given that some kind of catastrophic event caused the YD, its strange CO2 ‘anticipated’ the YD.
The Antarctic Cold Reversal is considered to have started 1k years before the start of the YD and ended 1k years before the end of the YD.
Its odd that the NH and SH both had cold periods out of phase with each other and certainly suggests there are age calibration issues, probably in the SH/Antarctic CO2 and temperature proxies.
Don J. Easterbrook
“Simply showing that CO2 correlates with anything surely doesn’t prove that CO2 was the cause.”
Because CO2 is a greenhouse gas and has been shown as the principal control knob governing the climate of the for the last half billion years.. (http://tinyurl.com/6ppedy6) (http://droyer.web.wesleyan.edu/PhanCO2%28GCA%29.pdf) It is not a bold claim to attribute this trail of temperature change to the rise in CO2 as the driver of the change.
“I had hair before CO2 began to rise, but I don’t blame that on CO2.”
CO2 is a greenhouse gas..
“What he found was large variability in the data, which led him to conclude that”, “The variety in the shapes of these graphs is quite surprising Yes, they’re all vaguely alike, but that’s about all. The main curiosity about these, other than the wide variety of amounts of warming, is the different timing of the warming.”[…] “It’s clear that there is warming since the last ice age.” “But if you want to make the claim that CO2 precedes the warming? I fear that this set of proxies is perfectly useless for that. How on earth could you claim anything about the timing of the warming from this group of proxies? It’s all over the map.”
Willis is simply negating the fact the Northern Hemisphere warmed slower than the Southern Hemisphere, when you read the paper and not take Willis word on the subject, you can clearly see that there are many different warming start dates for the proxy’s http://i44.tinypic.com/34gncox.jpg the paper is getting a “global” temperature record. Is it not obvious that around the world during a major climatic shift with large areas of ice receding and forest growing over thousands of years, you are going to get a mix of proxy temperature records!! Each proxy record is a peer reviewed study which is referenced in the paper. If you have a problem with one of them, by all means read!!
“Comparison of the Shakun et al. global temperature curve with the data from which it was constructed. The blue arrows point to the Younger Dryas dip in temperatures and the black arrows point to the ending of the Younger Dryas. The two should match, but don’t.”
This point amazed me, the paper is a global record of temperatures, and the Younger Dryas dip was only really seen in the Northern Hemisphere. It is all explained in the paper and I have already shown the graph which explains it, http://i44.tinypic.com/34gncox.jpg What is even more bizarre is how you show this on a graph that is not averaged and by the look of it are only considering the yellow dots on the graph??
http://www.ncdc.noaa.gov/paleo/abrupt/data4.html
As others have noted there is an error in position of YD is the second graph here. Similarly in the overlay , please correct this. Labeling an axis “year” is insufficient if you do not specify whether it’s AD or BP, though the range of the data makes it obvious.
The YD provides an important check on the timings which, despite the error here, still don’t line up. Though it is a lot nearer than it appears here this would seem to be a key factor in determining any lag between the two. Perhaps posting the Shakun graph itself would be helpful (fair usage) since the paper is pay-walled.
Steven Mosher says:
April 8, 2012 at 10:35 pm
“Yes. Shakun adds another win for Lukewarmers.”
If your maths is right, you make a good point..
I don’t understand what is interesting about this study. Carbon dating at the 15,000 year level isn’t accurate enough for what they are trying to do. The accuracy decreases by 2 to the power of the number of half lives (5,730 years). Given that 1/2 a half life is +/- half a century the error at 3 half-lives is on the order of time (a millennium) that you are trying to measure. Also ice core dating errors and the proxy dating errors are independent and affected by different factors.
They have dated parts of the same animal (from frozen tundra) as 15,000 and 21,000 years old. We don’t know which end of the tree they are dating.
They would have a dating accuracy of better than 1% to make this study useful.
This doesn’t even touch the issues that the proxies for temperature would have to be reasonably accurate and respond to temperature and not moisture and other variables.
The ice cores at least have the advantage that the CO2 and temperature signals are related to the same relative layers even though they may be skewed or modulated.
PS the offset shown here looks like >=2500 years so there will be something like a 500y offset when dates are corrected. This is just about what Shakun needed to revert the previous work showing CO2 lagged temps.
As ever totally unrealistic uncertainty estimates enable drawing a conclusion that is no way justified by the data.
Mashing 80 very different proxies from all around the globe and showing the minimalistic error bars is complete fiction. This is from the same school of post-impressionist science as Mann’s hockey stick. I’m not impressed.
Distant past is almost imponderable.
More recent past is controversial, the hockey stick and like.
One thing is almost a certainty
– CO2 atmospheric content can’t affect changes in the Earth’s magnetic field
– global temperature change can’t affect changes in the Earth’s magnetic field
regardless of which one came first
Not so certain is that the sun’s magnetic oscillations can’t affect changes in the Earth’s magnetic field.
– If the changes in the Earth’s magnetic field for the last 350 years plotted against hockey stick result: no correlation
– If the changes in the Earth’s magnetic field for the last 350 years plotted against the Loehle’s temperature ‘non tree ring’ reconstructions result: good correlation.
http://www.vukcevic.talktalk.net/LLa.htm
It is to the ones who earn leaving from the CO2 hypothesis, to make sure that their profession is safeguarded by maintaining the ‘science is settled’ status.