Is ENSO “responsible for recent global warming?” No.

A raging controversy, one almost as hot as the record-breaking heat wave on the North American west coast, has broken out over a recent paper in the Journal of Geophysical Research (Atmospheres), a peer-reviewed periodical published by the American Geophysical Union (AGU).

Authors John McLean, Chris de Freitas and Bob Carter all have a long history of links to climate disinformation groups and associated PR campaigns (many of which lead back to Canadian spinmeister Tom Harris, director of the International Climate Science Coalition).

Now the trio have claimed that their analysis demonstrates that global warming is primarily a result of natural processes with little role for anthropogenic influences such as greenhouse gases.

A barrage of criticism has forced the authors and their champions to backpedal furiously. Along the way, a PDF of the paper has been removed from the New Zealand Climate Science Coalition (NZCSC) website, presumably to rectify an egregious copyright breach. Even worse, the NZCSC parent affiliate, the above-mentioned International Climate Science Coalition (ICSC), was caught substituting its preferred title for the actual one in its link to the paper.

And, of course, a look at the actual data clearly shows that the trio’s touted source of global warming, namely a 1976 “shift” in the El Nino Southern Oscillation (or ENSO), can not possibly account for the recent upward trend in temperatures. All this raises serious questions about JGR’s editorial processes, and leads to the inevitable conclusion that the paper should be withdrawn.

To be sure, much of the paper’s analysis is unremarkable. ENSO has long been recognized by scientists as the main source of short-term ( year-to-year) variability in mean global temperature. But as discussed in relevant posts at RealClimate and Tamino’s Open Mind, that fact has little bearing on the attribution of the warming trend, which is clearly anthropogenic.

Naturally, then, criticism has focused on various unsubstantiated claims by McLean et al, including this concluding sentence:

Finally, this study has shown that natural climate forcing associated with ENSO is a major contributor to variability and perhaps recent trends in global temperature, a relationship that is not included in current global climate models. [Emphasis added].

The authors also claimed that “trends” over the last 14 years were similar for global temperature and ENSO (as measured by the Southern Oscillation Index or SOI), but did not actually present any evidence for the assertion:

The post-1995 period illustrated in Figure 7c shows that the respective trends across the last 14 years will be very similar, albeit from 1997 onward the average global temperature anomaly is consistently slightly higher than the relationship would suggest.

In discussing ENSO and tropical temperature trends, the authors made this unequivocal (but unsupported) claim:

We have shown here that ENSO and the 1976 Great Pacific Climate Shift can account for a large part of the overall warming and the temperature variation in tropical regions.

In two press releases, the trio went even further. The first press release, entitled “Nature, Not Man, Is Responsible For Recent Global Warming”, was apparently produced by the New Zealand Climate Science Coalition and released through Marc Morano‘s Climate Depot website (a project underwritten by the anti-AGW pressure group Committee For a Constructive Tomorrow, better known as CFACT).

The same press release, which featured NZ “corresponding author” Chris de Freitas, was apparently pressed upon New Zealand columnist Paul Gorman by NZCSC “foundation member” Brian Leyland (who also serves as ICSC Secretary). [Hat tip to BigCityLib].

Here, Bob Carter claimed:

The close relationship between ENSO and global temperature, as described in the paper, leaves little room for any warming driven by human carbon dioxide emissions.

And lead author McLean chimed in:

When climate models failed to retrospectively produce the temperatures since 1950 the modellers added some estimated influences of carbon dioxide to make up the shortfall.

A second press release, this one from the Australian Climate Science Coalition (a second ICSC affiliate), was reproduced nearly verbatim at FarmOnline. Naturally, this release focused more on Aussie lead author McLean, who elaborated further on the implications of the study for climate models:

This paper shows that the missing component [of climate models] was the El Niño-Southern Oscillation (ENSO). The difference in output when using these two factors is that one predicts a continuing rise in temperature and the other predicts fluctuations according to the ENSO. The data particularly over the last 10 years indicates that the latter is correct.

Reaction, of course, was swift. RealClimate called the paper “atrocious”, while Tamino at Open Mind demonstrated that the “differencing” analysis technique employed completely removed any trends whatsoever.

Back in New Zealand, National Institute of Water and Atmospheric Research principal climate scientist James Renwick told Paul Gorman:

They have used their favourite versions of the radiosonde data and don’t discuss the possible issues with some of the radiosonde and satellite data. They try and confuse the year-to-year variation and they have deliberately taken out the trend in their analysis.

Renwick also said the conclusions were unsupported by the paper’s analysis.

[The concluding statements] strike me as being questionable at best, not based on anything that’s been shown. It is a real surprise it got through the peer review.

Picking up on the same theme, James Annan focused on what appeared to be a grievous lapse in editorial process at AGU journals. Commenters everywhere wondered how such a travesty could have been published. Annan’s answer seemed the most likely – that the AGU’s system of letting authors nominate reviewers was an “open invitation to game the system.”

It’s also worth noting that Chris de Freitas has been involved in editorial shenanigans before. Half the editorial team of the journal Climate Research, including chief editor Hans Van Storch, resigned in protest over the journal’s’ refusal to repudiate an anti-AGW piece that was accepted by de Freitas when he was an editor there. And one of the co-authors of that paper, Willie Soon, just happened to have been a reviewer of the de Freitas article published by the Bulletin of Canadian Petroleum Geology in 2002. [See PDF at p. 27].

Since he is the brother of the present CSPG Editor [Tim de Freitas], the paper was handled by one of our most capable Associate Editors, Dale Leckie, and review by Willie Soon, an Astrophysicist at the Harvard-Smithsonian Center for Astrophysics and Sonja Boehmer-Christiansen, the scientific editor of Energy and Environment and a climate scientist at the University of Hull, UK. These individuals are recognized globally for their contributions to climate research. They also recommended publication of the paper with minor revision. However, their names were not included in the acknowledgements at the end of the paper, which was an oversight.

Ah, yes, Sonja Boehmer-Christiansen, the “scientific” editor of the contrarians’ favourite periodical, Energy and Environment, is “recognized globally”, all right – mainly for her decision to publish McKitrick and McIntyre without even bothering with pesky peer review, not to mention the truly atrocious George-Ernst Beck paper on historical CO2 concentrations. Now that’s how to “game the system”!

Anyway, in the face of the unrelenting attacks, a third press release has now been issued, with control reverting to the NZCSC. (Interestingly, this is the only one of the three that is actually available in its original form).

In contrast with the claim of the first NZCSC release, entitled “Nature, Not Man, Is Responsible For Recent Global Warming”, the latest headline backpedals somewhat: “Compelling New Research Shows Nature Rules Climate”. The news item announcing the latest press release has the rather illogical title “Three Australasian Scientists Find AGW Link To Southern Oscillation”.

More hilarity awaits in the news release itself, as the authors actually claim that the whole exercise of “differencing”, with its attendant detrending, was only done to find the correct lag between the Southern Oscillation Index (SOI) and mean global temperature (MGT).

Regarding the various previous claims about “trends”, they have this to say:

The results in Figure 7 clearly show that the SOI related variability in MGT is the major contribution to any trends that might exist, although the McLean et al study did not look for this. The key conclusion of the paper, therefore, is that MGT is determined in most part by atmospheric processes related to the Southern Oscillation.

That’s it? That’s the “compelling” analysis that shows that a shift in ENSO “is responsible for recent global warming”?

While we absorb fully the fact that key findings of this paper are pretty much admitted by the authors themselves to be empty hand waving, here are a couple of other interesting footnotes.

  • The link to the NZCSC version of the paper no longer works. Most likely, this is related to the fact that the NZCSC appeared to have been in egregious violation of the authors’ copyright agreement with the AGU. That agreement permits alternate electronic publication only of unformatted versions of published articles, and these must be hosted only at the authors’ personal websites.
  • Even worse, in its original news item on the paper, the International Climate Science Coalition had actually substituted the title of the first press release for for the actual title in its link to the paper:

ICSC July 26 short 2That’s right – according to the ICSC, the paper’s title was “Nature, not Man, is responsible for global warming.” Stop the presses!

It’s worth noting that all three authors are “advisors” to the ICSC, a sort of umbrella group that brings together contrarians from Canada, the United Kingdom, Australia and New Zealand. The ICSC also has a special relationship with the New Zealand and Australian “coalitions” – all three websites are even hosted at the same service provider in Arizona.

But for now, we’ll leave the authors and their dizzying array of affiliations with various “science” lobby groups aside, as that subject deserves a post (at least) of its own.

Turning (at last) to the data, we now examine comparisons of the Southern Oscillation Index (SOI) and various measures of mean global temperature. As explained by the Australian Bureau of Meteorology, the SOI “is calculated from the monthly or seasonal fluctuations in the air pressure difference between Tahiti and Darwin.” Negative values generally indicate El Nino conditions associated with warming episodes; for that reason SOI and temperature are plotted on opposing axes. (For those interested, an ASCII version of monthly SOI values is available here).

A couple of additional notes: McLean et al use a scaling of about 4.3 SOI for each deg. C of tempertaure. Here we have used a slightly different scale, namely 5 SOI for each deg. C. Also we use a six-month shift in SOI. As we are mainly interested in comparing trends over decades, however, these small differences in scaling and shifting are of negligible import.

First we show annual SOI averages plotted against the two standard lower tropospheric temperature series (UAH and RSS).


This chart makes clear a fact obscured in the multi-panel monthly data presentation in McLean et al.: The SOI slight downward trend is manifestly opposite to the warming lower troposphere (LT) temperature trends. It is difficult to understand, then, how ENSO could possibly be driving global warming.

Nevertheless, we’ll examine various ways of correcting for volcanic cooling, which is apparent in 1982-4 (El Chichon) and 1991-3 (Pinatubo).

First we’ll consider the recent period 1996-2008, which is free of major volcanic influence. This is the period in which McLean et al claimed to diiscern “similar” trends for SOI and mean global tempertaure. Recall also their related claim in the latest press release that their results “clearly showed that SOI related variability in MGT is the major contribution to any trends that might exist.” [Emphasis added].

The monthly series is obscured by wide variation. Nevertheless, over this period SOI is very slightly downward (indicating cooling influence), while both UAH and RSS have warming trends above 0.1 deg C per decade.

SOI UAH RSS 1996 month

The annual plot is clearer, and shows a virtually flat SOI against rising LT temperatures:


Finally, we’ll take a look at decade-over-decade changes in SOI and global average temperature (this time, we’ll include surface global tempertaure as well, with series from HadCRU and NASA GISTEMP). The first chart shows a clear upward trend in all temperature series, while SOI rises slightly in the 1990s, but falls back in the 2000s.

SOI Decade

Even if we remove the years of significant volcanic cooling, there is still no evidence of a dominant ENSO influence on temperature trends. The most that could be said is that variations in SOI may have had a slight dampening effect on warming in the tropospheric data sets, but even thatsmall effect is not evident in the decadal surface record.

SOI Decade NV

To sum up:

  • The McLean, de Freitas and Carter paper presented unsubstantiated conclusions that are contradicted by a cursory analysis of the very data presented.
  • There is widespread agreement among climate scientists that this paper should not have passed review and should not have been accepted for publication.
  • The authors actively participated in a deceptive public relations campaign that trumpeted and exaggerated the paper’s claims, a campaign that even substituted a press release headline for the true title of the paper.
  • The authors permitted an egregious breach of copyright in the dissemination of the paper in its published form at the New Zealand Climate Science Coalition website.

What more is needed to prod the AGU and the Journal of Geophysical Research to do the right thing? The paper should be withdrawn, and the editor responsible disciplined. Now.


18 responses to “Is ENSO “responsible for recent global warming?” No.

  1. Pingback: Mother nature’s sons — Hot Topic

  2. Andrew Hurley

    Great summary…I’ve been watching the various bits of this story unfold but here they all are in one easy to follow collation.

    I skimmed Lucia’s post at The Blackboard and was wondering if your chart of declining SOI and rising temperature trends is the graphical representation of what she was talking about with the 4.2C/century temperature trend discovered by McLean et al.

    [DC: I think Lucia may be referring to the constant term in one of the “differential” correlations. AFAIK, that’s a separate issue, but I haven’t looked into it in any detail.]

  3. Note that McLean et al used UAH mid troposphere data, not LT as they claimed in the paper, and UAH MT shows only 0.04 K/decade.

    They call it ‘global atmospheric temperatures’, ‘lower atmospheric temperatures’, and most often ‘mean global temperature’ (!) in the press release.

    [DC: I realize that McLean et al say they use “lower troposphere”, but pointed to the mid-troposphere MT data set. I’ve now checked figure 7 carefully against the UAH data set and can verify that the LT data set was in fact used. So the reference to the UAH MT data set URL was a mistake. (Another mistake not caught by the editor was the conflicting labeling in Figure 7 – two of the panels refer to “SOI derivative”, while the third refers to “SOI”.)

    The paper itself takes a gratuitous and unsupported swipe at the surface temperature records. So McLean et al are definitely buying into the contrarian meme that the satellite data sets are somehow more reliable. And of course they prefer UAH to RSS.]

  4. I was amused to follow the link above to “ponder the Maunder”, where the former wunderkind of the skeptics. Kristen Byrnes, working I think for a school project, neatly anticipated the thesis of McLean et al:

    I will demonstrate that a negative trend in the El Nino Southern Oscillation (more and stronger La Ninas) from 1945 to 1975 and a positive trend in the ENSO from 1975 to present (more and stronger El Ninos) correlates better with global temperature changes than greenhouse theory. Thus, ENSO is probably the largest contributor to global warming in the past 30 years.

    [DC: That’s definitely an amusing footnote. Still, I may have to disable the related posts “feature” – somehow it only finds contrarian posts (except when I’ve done a multi-part series)! I guess there are just a whole lot more of those.

    By the way, I appreciate all your contributions, of course, but especially those on the contrarian blogs. Your single handed deflation of the vicious attacks on Rahmstorf at CA deserves some kind of award. Thanks for that.]

  5. Jim Galasyn

    Nice, keep up the good work!

  6. Deep–
    The negative trend in SOI and the increase in the estimate of the remaining trend based on the constant in MdF&C’s article are interlaced.

    If you go through the algebra for a linear regression, you’ll see that if one fits:

    dT=a dSOI + m

    and discovers “a” is positive (as in the MdF&C) paper and the average value of ‘dSOI’ is negative for your sample set, then the trend ‘m’ in this fit will be higher than you would have obtained by fitting a trend to

    T=m time + b
    using the undifferenced data, and not including the SOI.

    You’ll find that the difference in the two trends is approximately equal to the product (a * average(dSOI)) (There will be a bit of slop, but the agreement will be pretty close.)

    So, in some ways, we are discussing the same thing. I thought a shorter discussing only one metric would be easier for people to grasp, so I didn’t discuss both ways to look at it and to see that the issue is not complicated.

    Still it’s good that it gets discussed both ways because some people will better appreciate the fact that the product (a dSOI) means the change in SOI would cause us to expect a drop in temperature rather than explaining away the warming. Other people will better appreciate that the value of ‘m’ increased.

    [DC: I have to admit I didn’t look at your analysis, although I saw a reference to it at Tamino’s, so my comment was no doubt hasty. To the extent that the constant term in the correlation of derivatives reflects the trend left over after incorporating the “explanatory” ENSO, yes, I see that this represents a higher temperature trend than the case without ENSO. That’s basically equivalent to what I’ve shown: “cooling” SOI (which confusingly is actually a positive trend) against a strong warming trend in global temperature.

    The details are complicated, because the two analyses are not exactly the same; I used six-month lag and annual or monthly averages, rather than rolling 12-month averages. I’ve noticed that even between monthly and annual series, the trends in SOI can be different. But your essential point remains.

    Also SOI and T are negatively correlated, are they not? I guess I’ll have to dig into the McLean et al correlation of derivatives when I have time.

    Anyway, to me it’s a lot simpler and clearer to compare SOI and temperature trends. If McLean et al wanted to make claims about the attribution of global warming, that’s what they should have done in the first place.]

  7. Deep Climate

    Following up on comments above from Lucia and Andrew Hurley:

    I considered the best fit for the two following regressions (excuse the informality):

    1) T = -0.005 SOI + 0.136t – 0.147

    2) T = 0.126t – 0.120

    T is the UAH LT monthly anomaly, SOI is Southern Oscillation Index (7 months earlier) and t is time measured in decades.

    Period was 1980-present.

    So, from this point of view, after accounting for ENSO, the trend rises from ~0.13 C/decade to ~0.14 C/decade, an increase of 0.01 C/decade.

    From skimming Lucia’s analysis, it appears she derives, so to speak, a much bigger difference from the McLean et al analysis – a rise from 0.12 deg./decade to 0.42 C/decade after accounting for ENSO.

    (By the way, the slight difference in non-ENSO trend comes from the later starting date in the analysis here).

    The principle is the same, though – the opposing (cooling) trend in SOI raises ENSO-accounted trend in LT.

  8. Deep
    For the mathematical point about the relationship between the trend, and the average level of SOI and your numerical comparisons doesn’t quite work for reasons that are a bit obscure. The size of the apparent difference depends on whether you are comparing cases with the volcanic periods taken out to the raw data or not.

    What I’m mean to say is if you fit
    dT= m dSOI + b


    dT = c

    For the same exact set of (dT, dSOI) pairs, you’ll find that

    (c-b) = m * average(dSOI).

    If ‘c’ is the trend that is not explained by dSOI, and ‘b’ is the trend explained by dSOI, the difference is the product of the regression coefficient and the average of dSOI.

    This is always true and can be shown with the actual equations underlying linear regression. So, that relation is very tight.

    The fits to the time series are related to the fits for the difference in the time series and you can show similar relations for the un-SOI corrected trend and the SOI corrected trend.

    So, qualitatively, what you are seeing is related to what I was discussing.

    But yes, our numbers are not coming out exactly the same because we aren’t doing exactly the same things.

    Quantitatively, since we have small amounts of data noise will introduce some “slop” into the comparison of the effect of SOI on the remaning trend computed based on the time series (T, SOI) as opposed to the difference in the time series (dT, dSOI).

    (Whether or not someone dropped out a few years will also affect the numerical values we find.)

    Anyway, to me it’s a lot simpler and clearer to compare SOI and temperature trends. If McLean et al wanted to make claims about the attribution of global warming, that’s what they should have done in the first place.

    It might be. I think it’s good to have it explained both ways.

    I focused on the constant not because it’s simper or clearer but because MdF&C actually provide numbers for the constant but never mention what they found for the average of the SOI or the derivative of the SOI. So, basically, my argument is based on their own numbers in their own paper, using their own method, you would need to say “warming, worse than we though!”

    You point out that the argument falls apart based on the trend in SOI having the wrong sign. Yep. That too.

    This paper … has…. problems. . .

    [DC: Can’t argue with that. ]

  9. Pingback: Meet Alan Gibbs, builder of amphibious HumVees and “climate science” coalitions « Deep Climate

  10. It is interesting to note the scramble for information on the characteristics of the authors of this paper before considering the science behind it or the author’s qualifications. It is also noteworthy, that most scientists who read a paper, will generally read it several times over a period of several weeks, whether or not they agree with the findings and no matter how familiar or otherwise they might be with the topic before putting pen to paper. We also note that in previous posts, the opponents of McLean et al. have been quick to demand a peer reviewed paper rather than a comment on the internet. In this case, the appearance of a peer reviewed paper questioning the wisdom of blaming all of global warming on carbon dioxide brings out the best of insults extending right up to the publishers of JGR!

    Let’s just wait and see what appears next month or the monhs following in JGR.

    [DC: I think you’ll see a response in JGR sooner rather than later, if comments at Open Mind are anything to go by.

    Many of the authors’ most absurd claims were made in press releases disseminated through contrarian blogs. I can see no reason why they should not be answered in the real science blogs like Open Mind and RealClimate]

  11. Having read the comments made on the so called “Open Mind” by Tamino and others, none appears to give one any confidence that the McLean et al. paper will be contradicted scientifically any time soon. The suggestion for example that the ENSO conditions can act to change the Global Temperature on an annual or short term basis does not provide any reason that a particular sustained condition of that phenomenon can not produce a particulat temperature distribution over the globe which controls the effective temperature, as it does.

    Most if not all all of the comments seem to concentrate on trying to diminish the status of the authors and JGR reviewers, which immediately leaves the distinct impression that the critics lack the confidence and obviously the ability to discredit the science by properly formulated counter arguments. This has the effect of reducing what could be a stimulating debate on the science, to a simplistic, and dare I say at times quite childish, attempt to make unsubstantiated claims about the aims of the authors.

    On a personal note, I am at present enjoying a discussion with a person who takes the completely opposite view of the Climate Change debate from mine and we both agree that we are really enjoying the process with out any insults or politics. This person a retired professor of physics sought me out for this purpose. We disagree but respect absolutely each others views and spend our time trying to show scientificaslly where the other is wrong. We concede points to each other and argue others to the death, just as scientifi debate used to be. It is very refreshing. I compare this with the continuing unscientific material which is used in vain attempots to degrade the Mclean paper.

    John Nicol

    [DC: The authors have repeatedly claimed both in the paper, and in three press releases, that their analysis demonstrates that a shift in ENSO is “responsible for recent global warming” and accounts for “any temperature trends that might exist.” Those assertions are clearly false, as are a number of others made along the way.

    Please reread this comment by McLean et al and Tamino’s response.

    Tamino says: Expect a comment on your paper to appear soon in JGR. I can hardly wait to see how you’ll respond there.

    I have no reason to doubt him. ]

  12. Pingback: Climate change Bytes and Blogs VI « Greenfyre’s

  13. After wading through all your invective, it seems to me your point is simply that McClean et al overstate their case. Although I haven’t read their paper, I think you are probably right.

    It looks to me like climate is chaotic in the same way that weather is chaotic. Which is part of the reason it seems likely the warmists have overstated their own case.

    My opinion only.

    [DC: McLean et al do not merely “overstate” their case. They have no case. The analysis here and elsewhere shows that SOI has had a slight cooling effect over the last thirty years, although it can explain much of the interannual variability in temperature.]

  14. Pingback: Revisión científica, pero no es El Niño « Cambio climático

  15. Pingback: In the beginning: Friends of Science, Talisman Energy and the de Freitas brothers « Deep Climate

  16. I like how John Nicol avoids the fact that the substance of this paper, such as it is, has already been dealt with. Quick! Change the subject! The denialists blew it again!

  17. Pingback: McKitrick gets it wrong on IPCC « Deep Climate

  18. Pingback: Mann exonerated by PSU inquiry: “No substance to the allegation” « Deep Climate