Climate Science Glossary

Term Lookup

Enter a term in the search box to find its definition.


Use the controls in the far right panel to increase or decrease the number of terms automatically displayed (or to completely turn that feature off).

Term Lookup


All IPCC definitions taken from Climate Change 2007: The Physical Science Basis. Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Annex I, Glossary, pp. 941-954. Cambridge University Press.

Home Arguments Software Resources Comments The Consensus Project Translations About Support

Bluesky Facebook LinkedIn Mastodon MeWe

Twitter YouTube RSS Posts RSS Comments Email Subscribe

Climate's changed before
It's the sun
It's not bad
There is no consensus
It's cooling
Models are unreliable
Temp record is unreliable
Animals and plants can adapt
It hasn't warmed since 1998
Antarctica is gaining ice
View All Arguments...

New? Register here
Forgot your password?

Latest Posts


Does CO2 always correlate with temperature (and if not, why not?)

What the science says...

Select a level... Basic Intermediate

Even during a period of long term warming, there are short periods of cooling due to climate variability. Short term cooling over the last few years is largely due to a strong La Nina phase in the Pacific Ocean and a prolonged solar minimum.

Climate Myth...

There's no correlation between CO2 and temperature

"Twentieth century global warming did not start until 1910. By that time CO2 emissions had already risen from the expanded use of coal that had powered the industrial revolution, and emissions only increased slowly from 3.5gigatonnes in 1910 to under 4gigatonnes by the end of the Second World War. 

It was the post war industrialization that caused the rapid rise in global CO2 emissions, but by 1945 when this began, the Earth was already in a cooling phase that started around 1942 and continued until 1975. With 32 years of rapidly increasing global temperatures and only a minor increase in global CO2 emissions, followed by 33 years of slowly cooling global temperatures with rapid increases in global CO2 emissions, it was deceitful for the IPCC to make any claim that CO2 emissions were primarily responsible for observed 20th century global warming."
(Norm Kalmanovitch).

That carbon dioxide causes warming is well established by physics theory and decades of laboratory measurements. This is confirmed by satellite and surface measurements that observe an enhanced greenhouse effect at the wavelengths that carbon dioxide absorb energy. Given the strong causal link between CO2 and warming, what are we to make of periods where CO2 does not correlate with temperature? The most commonly cited example is the recent years since 2002. Over this 7 year period, global temperature has shown little to no trend while CO2 has risen. If CO2 causes warming, shouldn't temperature be rising steadily also?

Figure 1: Annual atmospheric carbon dioxide (NOAA) and annual global temperature anomaly (GISS) from 2002 to 2008.

However, this is a short period as far as climate trends are concerned. To understand recent years in the broader context of long term climate trends, one needs to look at the temperature record over several decades. By comparing carbon dioxide levels to temperature from 1964 to 2008, it becomes apparent that even during a long term warming trend, there are short periods of cooling.

Figure 2: Annual atmospheric carbon dioxide (NOAA) and annual global temperature anomaly (
GISS) from 1964 to 2008.

Internal variability causes dramatic ups and downs in temperature compared to the more gradual long term trend. Consequently, it's possible to select short periods throughout a long term warming period where the warming slows or reverses. For example, the periods 1977 to 1985 and 1981 to 1989 both show little to no warming while CO2 continues to increase. Taken out of context, one might have concluded in 1985 or 1989 that global warming had stopped based on the previous few years data.

Figure 3: Annual atmospheric carbon dioxide (NOAA) and annual global temperature anomaly (GISS) from 1977 to 1985 and 1981 to 1989.

What causes this climate variability? Ocean cycles shuffle heat around the climate by exchanging heat between the ocean and atmosphere. This can have a strong short term effect on global temperature, the most dominant cycle being the El Niño Southern Oscillation. In 2008, the Pacific Ocean was in a strong La Niña phase, leading to unusually cool temperatures throughout the tropical Pacific Ocean. Additionally, the sun was currently in solar minimum, experiencing the lowest solar levels in a century. Solar activity has an 11 year cycle which is estimated to have an effect of around 0.1°C on global temperatures. The combination of solar minimum and La Niña conditions would have a short term cooling effect on global temperatures.

This demonstrates the danger of drawing conclusions from one small piece of the puzzle without viewing the broader picture. If one focuses on just the last few years, one might erroneously conclude global warming has stopped. However, by looking at several decades of data, we see a climate that shows strong short term variability. By understanding the mechanisms that cause climate variability, we see that the current cooling is short term variation imposed on the long term warming trend. What about a longer time series? Over the past century, are there any periods of long term cooling and if so, what is the significance?

Figure 4: Green line is carbon dioxide levels from ice cores obtained at Law Dome, East Antarctica (CDIAC). Blue line is carbon dioxide levels measured at Mauna Loa, Hawaii (NOAA). Red line is annual global temperature anomaly (GISS)

Figure 4 compares CO2 to global temperatures over the past century. While CO2 is rising from 1940 to 1970, global temperatures show a cooling trend. This is a 30 year period, longer than can be explained by internal variability from ENSO and solar cycles. If CO2 causes warming, why isn't global temperature rising over this period? To answer this, one needs to recognise that CO2 is not the only driver of climate. There are a number of factors which affect the net energy flow into our climate. Stratospheric aerosols (eg - from volcanic eruptions) reflect sunlight back into space, causing cooling. When solar activity increases, the amount of energy flowing into our climate increases. Figure 5 shows a composite of the various radiative forcings that affect climate.

Figure 5: Separate global climate forcings relative to their 1880 values (

When all the forcings are combined in Figure 6, the net forcing shows good correlation to global temperature. There is still internal variability superimposed on the temperature record due to short term cycles like ENSO. The main discrepancy is a decade centered around 1940. This is thought to be due to a warming bias introduced by US ships measuring engine intake temperature.

Figure 6: Blue line is net radiative forcing (
GISS). Red line is global temperature anomaly (GISS).

So we see that climate isn't controlled by a single factor - there are a number of influences that can change the planet's radiative balance. However, for the last 35 years, the dominant forcing has been CO2.

Intermediate rebuttal written by dana1981

Update July 2015:

Here is a related lecture-video from Denial101x - Making Sense of Climate Science Denial


Last updated on 17 July 2015 by MichaelK. View Archives

Printable Version  |  Offline PDF Version  |  Link to this page

Argument Feedback

Please use this form to let us know about suggested updates to this rebuttal.


1  2  3  4  5  6  7  8  Next

Comments 1 to 25 out of 183:

  1. I think that there are a number of issues here. First is the question of correlation, specifically, the degree of correlation between CO2 and warming comapred to other factors. For instance, my understanding is that water vapor constitutes about 95% of the earth's greenhouse gases. But probably more important is the question of causuality. Just because over the last 35-50 years the earth is warming and at the same time CO2 is increasing, might imply correlation but not causuality. I've read studies that in fact suggest the reverse causuality - that over long geologic periods, CO2 increases naturally follow and probably result from the increases in temperature, not the other way around. Next, even if CO2 increases are both correlated to and causual to temperature increases, to what degree is man-made CO2 an effect (since much of the science of climate change is direct input to political decisions around the world). If water vapor is the dominant (95%) greenhouse gas, and assuming that CO2 is the next highest (which it might not be, taking into account methane), then what percentage of that 5% is man-made? From what I've read, that number appears to be around 3%, which translates to a fraction of a percent for man-made contributions to greenhouse gases. But even that might overstate things from a political policy perspective. I understand that a significant percentage of man-made CO2 emissions are directly related to livestock.
  2. Glen Tamblyn has been posting excellent rebuttals to the Skeptical Science commenter The Inconvenient Skeptic, over on The Inconvenient Skeptic's site page about the Taylor Dome ice core analysis. That is relevant to this Skeptical Science page (There’s no correlation between CO2 and temperature) because The Inconvenient Skeptic's core argument is that there is no correlation between CO2 and temperature. As commentors and moderators have been pointing out to The Inconvenient Skeptic, his detailed comments on that topic do not belong on the Skeptical Science page The Value of Coherence in Science.
  3. Why is it that when I compare the average monthly min/max ratio for 1946 to 2009 for Norfolk Airport Virginia, there is no noticeable change, yet carbon dioxide has virtually doubled? I would expect to see the ratio reducing.
  4. @kdfv #3 Please, share data (for instance 54°F) not adverbs + adjectives (no noticeable, virtually doubled). Your ordered pairs like (1946; 'same same') are not very functional. Links and you transcribing the few significant data will suffice. [Don't forget you have also some things to reply in other posts] @Everybody I would be good to stop replying half cooked comments. For instance, it's easy to choose a couple of years and looking in temperature maps find then a lot of places where temperature have not changed substantially, for instance, as I remember, a third of the Bible Belt for the last 50 or 60 years . But we all know how these sort of 'naive' assertions work, so let the 'innocent' at least to do the work of documenting it.
  5. From here. "NOAA/NCDC Annual Global Temperature Change vs CO2 dataset clearly shows ... rising CO2 had little impact" Utter nonsense. The graph shows annual temperature change, ie T(now) - T(last year). This removes any long term trend and therefore cannot be compared with the long term increase of atmospheric CO2. Any conclusions taken from the website you reference are therefore also utter nonsense. Surely you can do better than a misapplication of basic math and science?
  6. A very devoted scientist at the University of Oslo has put together this graph: Where's the correlation between CO2 and temperature over the last 11000 years? That's pretty long-term...
    Response: [muoncounter] Please restrict image width to 500.
  7. My mistake, I forgot to resize the above image to 450 pixels. Mr. Cook, could you please edit the post to fix the problem?
  8. michaelkourlas @6: Where is this image from? The top graph contains one curve, but is labeled "Air Temperature at the summit of the Greenland ice sheet" on the left and "Approximate global temperature anomaly" on the right. Unless these track exactly (which would seem surprising), something's wrong there. Greenland temperatures wouldn't be useful in comparison. I think the author of the graph should include more recent CO2 values to show that the current spike in temperature is matched by the current spike in CO2. He/She could have used a dotted line similar to the one in the temperature graph. Additionally, I was under the impression that the correlation was only for the latter part of the 20th century (another reason why the author should have included the current CO2 values). Previous to that, I was under the impression that other climate forcings were stronger than CO2; eg, the sun in the early 20th century. As such, it would seem that this graph is a strawman.
  9. michaelkourlas @6: First, as Dr Humlum knows very well, the temperature record of any single site makes a very poor proxy for global temperature. Any such site will always show far more variability than a true global temperature because other locations across the globe will not warm or cool synchronously with the first location. This is easily seen in the following graph of several proxies for temperatures at individual locations during the holocene: As you can see, the individual temperature proxies are all over the place, and the GISP2 record (light blue), which shows four of the five highest peaks, is probably the most variable amongst them. Of course, as you can also see, if you take an average of the individual records there is very little remaining variability, compared to that of GISP2 alone. Indeed, taking a multiple proxy mean shows global temperature variability to be confined to a 1 degree C range, not the 2 degree C range Dr Humlum would have you believe. The multi-proxy mean is not the best reconstruction. This is so because it treats individual site specific reconstructions as of the same value as multi-proxy regional reconstructions; and also because of the eight reconstructions used in this case, only two are from the tropics (33% of the Earth's surface), only two from the Southern Hemisphere extra-tropics, in both cases from Antarctica, and the other four are from the Northern Hemisphere extra-tropics. Without this Northern Hemisphere bias, it is likely the reconstruction would show even less variability. As can be seen from the diagram below, Dr Humlum chose to compound the misleading choice of a very variable single site proxy by using global mean temperatures (which because they are a global average have low variability relative to single sites) to represent the modern era. In this way the false impression is created that modern global temperature temperature change is much less than past episodes, whereas modern global temperature are at levels rarely if ever exceeded in the Holocene. Second, the range of CO2 levels over the Holocene as shown by the Epica core is just 20 ppm. A increase of CO2 from 260 to 280 ppm will result in an increased forcing of just , and a temperature increase of just 0.4 Watts/m^2, and a temperature increase of just 0.32 degrees C. That is significant enough, and may well be why the Holocene has not slid into another glacial, unlike the previous inter-glacials. But it is certainly not enough to swamp the many forms of natural forcing, being of a similar magnitude to changes in solar forcings over the last two thousand years, and smaller than changes to volcanic forcings. So, to summarize, variation in Holocene temperature is much smaller than Dr Humlum purports it to be, and CO2 variation is sufficiently small for non-CO2 forcings to be the primary drivers of temperature in the Holocene. Dr Humlum knows this. I wonder why he does not mention it?
  10. This is a response to a comment on the Skeptic/ Denier thread. J Bob - I don't understand why anyone would use a local/ regional temperature record for such a comparison when there are others with global figures. Such as this graph on this page It also has the great advantage of displaying aerosols (and other forcings) as well as GHGs. Global. Comprehensive. Much better.

    [DB] Thank you for setting a great example!

  11. J Bob:
    "You asked why I used local temps instead of global? The reason is that there where no “global” temperatures prior to 1850, if you would call it that. Even in 1850, almost all were in the Northern Hemisphere. But there were quite a number of individual station records starting in 1659, so one has a longer time span to compare one variable (temperature) to another (CO2), hence the use of Ave14. You will notice I included the HadCRUT NH as a check on Ave14, and the patterns do seem to track, especially when both are compared to CO2. Both Ave14 & HadCRUT showed strong variations, while CO2 showed none. If I added a HadCRUT global plot, it would appear to change nothing. Your plot, you referenced by Meehl, uses models, and again show little compassion of GHG to temperature, except they both go up over time. I don’t see the up and down motion that would give stronger correlations. This process is my idea of a skeptic."
    1) First, any sensible comparison between CO2 and temperatures should compare CO2 forcing with temperature as CO2 and its forcing are not linearly correlated. 2) Although CO2 forcing is considered to be a factor prior to 1950, it is not considered to be the dominating factor by anybody prior to that time for short term (< millennial) time scales (see figure 5 above). Therefore we do not expect a strong short term correlation between CO2 forcing and temperature prior to 1950. 2a) Because of that, temperature fluctuations prior to 1950 are not direct evidence against the greenhouse effect. They are interesting evidence of the scale of natural variability, but using a regional temperature record is inappropriate for determining that variability for reasons given in 9 above. 3) Once all GHG forcings, anthropogenic aerosols and volcanic aerosols are all taken in to account, the decadal correlation between forcings and temperature is remarkable (see figure 6 above). The correlation of annual variability is poor because annual change in CO2 forcing is small, and because annual variation is dominated by the El Nino Southern Oscillation. 4) If you insist on doing a simple comparison of CO2 to temperatures, you should do so using Southern Hemisphere temperatures which, because the NH and SH air masses mix poorly, was largely isolated from the aerosol burden which is so significant, and so significantly distorting in the NH. Differences between the Arctic and the Antarctic means that the Arctic amplifies, while the Antarctic suppresses the warming, but the SH still provides a better natural laboratory if you are not going to do a full work up of all factors:
  12. J Bob "...referenced by Meehl, uses models, and again show little compassion of GHG to temperature, except they both go up over time. I don’t see the up and down motion that would give stronger correlations. This process is my idea of a skeptic. Your thoughts?" My thought currently is "Oh bother!" Once upon a time I had a lovely little display showing changes in nighttime temperatures in early 20th century. Can't find it (I'm sure some very thoughtful person will eventually direct me to the right place.) 'Twould be handy for one point I might make. Without it, I'll stay mum on that one. As for thinking about your "idea of a skeptic", I'm not so sure. The up and down squiggles you're looking for could just be the 'noise' from various natural variation or they could be those familiar roller-coaster dips we see after major volcanic eruptions. My first response as a skeptic would be - why do the people who know more than me about this stuff not consider it the same way I've been thinking about it? Read a bit, or a lot, more. Then decide whether I understand it well enough to see what they're getting at - or decide I've spent enough time on it and just accept it until someone comes up with the brighter, shinier, clearer version.
  13. Tom Curtis @ 11, your 4 point argument has a few problems. 1-Nothing was stated about linear correlation, or linkage between CO2 & global temperature being linear. And if you have ever worked on non-linear systems, most do respond to input changes, such as aircraft stall, or even ferrous transformer cores (except in full saturation). 2-So why was CO2 not considered a dominant factor prior to 1950? Seems to be a giant leap of faith, like in 1950, CO2 woke up. Although it’s a convenient point, since global temperature was bottoming out from the 1942-1955 ~0.45 deg. C (HadCRUT3gl) dip. 3-Even the Krakatau eruption in 1883 didn’t seem to phase the global temperature, with all the material dumped into the atmosphere, nor did it seem to effect the Lawdome CO2 data. Nor did CO2 change during in the 1880-1910 (HadCRUT3gl) 0.5 deg. C dip, or the longer 1775-1860 (Ave14). 4-I looked at NH data, since that is where most of the long term temperature data sets reside. As far as gases mixing, I assume the gas diffusion law still works between the NH & SH, so CO2 distribution should be pretty uniform globally. The one area we might agree on would be the increase in particulates in the atmosphere, such as the Asian Brown Cloud. Here is a higher resolution graph.
  14. J Bob @13: 1) There has been an increase of CO2 from 285 to 390 ppm since 1850, or 37%. Over the same period the forcing from CO2 has increased from approximately 30 W/m^2 to 31.7 W/m^2, or 5.7%. Running a direct comparison between CO2 and temperature therefore significantly distorts expectations of the effect of CO2 forcing. Your insistence on making a comparison known to be misleading rather than one known to be accurate is not a problem with my argument. Rather, it is a problem with your credibility. 2) Comparison of estimated forcings for all well mixed Green House Gases (second column) compared to the direct and indirect aerosol effect (second last and last columns respectively) show that until about 1950 they are always close to balancing each other out. Occasions when they are not quite in balance either coincide with exceptionally cool periods when the aerosol forcing is stronger (1910), or rapid warming when the GHG forcing is stronger (late 30's). After 1950, and particularly after the mid-1970's, GHG forcings begin to exceed aerosol forcings on a consistent basis. 3) Talking about just CO2 forcings is, once again, a futile exercise. It is like trying to predict how long it takes for a feather to fall from the Tower of Pisa using Newton's laws of motion and gravity alone. Including the effect of anthropogenic aerosols results in a good match to the various temperature fluctuations. It would be an even better match if we have been underestimating their effect as recently suggested by Hansen 4) Long lived components of the atmosphere like GHGs are well mixed between SH and NH. Short lived components like aerosols are not. Consequently the SH has a similar GHG forcing to the NH, but a much reduced aerosol forcing. It therefore provides a natural experiment which better quantifies the effect of CO2 forcings. That is straight forward enough. That you prefer a temperature record where we expect a maximum contaminating effect from aerosols (Europe) and at the same time refuse to factor in aerosol forcings into your calculation is very telling.
  15. J Bob, you do realise that particulates are included in aerosols?
  16. scaddenp @ 15, yes, from 1959. Principles of Heat Transfer by Frank Kreith, p.211, section on Radiation from Gases, Vapors & Flames. Tom Curtis @ 14, you quote a lot of estimated data based on math models. It would be better if you used actual data. I'm not thrilled about proxy data from ice cores, but it gives a sense of assurance if corresponds to actual measured data. I don't think I'd bet the farm on estimated model data.Those charts I presented above, can be traced to actual data sets, as well as some of the CO2 data. Somehow I feel more comfortable with some actual test data. I've seen what can happen when models go wrong, due to limited evaluation.
  17. J Bob @16, Translated: You want to treat the experiment as containing a single variable when you know they contain multiple variables. Right, got it.
  18. In item 6 above, charts of CO2 and Temperature are displayed for the last 11,000 years. Despite imperfections that have been pointed out for those charts, it is reasonable to ask the question "Where is the correlation between CO2 and Temperature ?" The same question applies for similar charts drawn for 1850 to 2011. From 1910 to 1940 temperature increased at a rate similar to the rate of temperature increase between 1975 and 2000. The rate of change of CO2 concentration was lower between 1910 and 1940 than it was between 1975 and 2000. From 1940 to 1975 temperature decreased while CO2 increased. From 2000 to 2011 CO2 has continued to increase but the rate of temperature increase has declined. For the 12 months of any year CO2 varies by much less than the un-correlated, larger variation in temperature. The data shows no correlation between Temperature and CO2. Charts of the measured data show that temperature is affected by other factors more than it is affected by CO2.
  19. Tom Curtis, I don't know how many systems you had to identify, with multiple inputs, output (MIMO) & parameters, when many of the inner workings were not well defined, and non-linear. One of the first things you do, is vary a single input, and then note the output response. You continue to do this with all the inputs, in order to gain an understanding of what is going on within. As you gain experience, one starts to look at multiple input changes, and subsequent responses. This is not an experiment, this is real life.
  20. J Bob, you have to use proxy data because you dont have any other reliable set. Your situation is 19 is ideal but that's not the real world. All variables are varying at the same time so you need multivariate tools. tobyth2 - see CO2 is not the only driver of climate. You only expect the correlation when CO2 is the primary forcing. The better question to ask is, since climate theory predicts temperature change to be a function of all forcings in operation, how well is temperature (on climatic scales of 30 year average) predicted by those forcing. Answer. very well indeed.
  21. 19, J. Bob,
    I don't know how many systems you had to identify, with multiple inputs, output (MIMO) & parameters, when many of the inner workings were not well defined, and non-linear.
    You seem to have experience and understanding with complex systems. You should of course, then, understand that the climate is a complex system, and that teasing out an exact, easily recognized CO2 to temperature correlation would be a fool's errand, correct? You would not expect to have done this with the systems with which you have experience, correct?
    One of the first things you do, is vary a single input, and then note the output response. You continue to do this with all the inputs, in order to gain an understanding...
    Yes, but the only place where this can be done is in models (which is another subject). In real life there are some very clever ways to do this, by for instance studying the short term responses to volcano eruptions, or paleoclimate data, or other clever ideas. But overall, this simply isn't possible. This is a different kind of problem than you are used to, and requires a different sort of thinking.
    This is not an experiment, this is real life.
    Precisely. And for this reason, meeting your insistence on a direct correlation between CO2 and temperature is simply unrealistic. That doesn't mean that CO2 can't be doing what it does, it only means that your demand for impossible proof itself proves and disproves nothing. Unless I'm misunderstanding your request...
  22. J Bob @19, I am quite aware that the procedure for identifying the effects of particular independent variable in a multiple variable system is to determine how the dependent variable changes as you change one independent variable while holding all others constant. You in contrast are comparing one independent variable to the dependent variable while simply ignoring the other independent variables. Not only that, you insist on making your comparison where an unmonitored (by you) independent variable is known to have fluctuated by a great deal. The quality of your method is well parodied by tonyth2 (whether he intended it as parody or not) when he says:
    "For the 12 months of any year CO2 varies by much less than the un-correlated, larger variation in temperature. The data shows no correlation between Temperature and CO2."
    Frankly I'm surprised he didn't also mention the daily variation in temperatures. Just imagine, it was 5 degrees C at 4:00 am this morning, and now its 25 degrees C and the CO2 content of the atmosphere has not changed at all: therefore CO2 is irrelevant to the Earth's temperature. Is that the sort of argument you like? It sure looks like it from this position.
  23. scaddenp @20, you say "All variables are varying at the same time so you need multivariate tools". True, I had a feeling I should have kept my MATLAB & System Identification Toolbox. Used it several times. Have you ever used it?
  24. No. That is not a toolbox that I have heard of. I seldom need that kind of statistical analysis and would look to R instead of MATLAB for the tools if the statistics Toolbox wouldnt cover it. (That said, it does have multivariate tools but I have never used them). However, that's the approach of the Benestad paper.
  25. scaddenp @ 24, MATLAB is an analysis software package that leans toward matrix analysis decomposition. It is used extensively in industry. It has a number of "toolboxes", that range from Statistics, Process control systems, Image processing, finance, etc. It's not cheap, but it does do a pretty good job. One of the better features is the "linking" of different "toolboxes" as needed, as well as the Simulink simulation package. MATLAB

1  2  3  4  5  6  7  8  Next

Post a Comment

Political, off-topic or ad hominem comments will be deleted. Comments Policy...

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.

Link to this page

The Consensus Project Website


(free to republish)

© Copyright 2024 John Cook
Home | Translations | About Us | Privacy | Contact Us