A reference was made to a paper that has become a classic of a kind on a thread near here recently. Some things get chewed up into bits and buried so deep that you’d think we’d never hear from them again. But they have a habit of showing up from time to time. It concerns an peer-reviewed article Influence of the Southern Oscillation on tropospheric temperature by three Australasian writers, J. D. McLean, C. R. de Freitas and R. M. Carter in the Journal of Geophysical Research (JGR), a refereed journal published by the American Geophysical Union (AGU).
The article purports to look at the relationship between ENSO, or specifically the Southern Oscillation (SOI), and variance in the tropospheric temperature in the tropics. They found that
the results showed that SOI accounted for 81% of the variance in tropospheric temperature anomalies in the tropics.
Carter, we were told, generalises this to global temperature:
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. The available data indicate that future global temperatures will continue to change primarily in response to ENSO cycling, volcanic activity and solar changes.
So the implications are said to be global and that’s it then for all those hucksters masquerading as climate scientists. The games up. They can pack it in and find a real job.
If you can’t get at the original article behind the pay wall the main findings are here.
Tamino at Open Mind was first out of the blocks. He found that it’s old news that ENSO has an effect on global temperature from year to year. But that effect is contained within the event and doesn’t contribute as such to the longer term trend.
Tamino points out that there is a difference between temperature and trend. He reckons they’ve stuffed up with their statistical method:
The fact is that their methodology, the process of estimating derivatives by taking 1-year differences, transforms any trend into a constant and thereby eliminates its impact on all variation and correlation.
Foster et al came to similar conclusions:
This comment first briefly reviews what is understood about the influence of ENSO on global temperatures, then goes on to show that the analysis of MFC09 severely overestimates the correlation between temperature anomalies and the SOI by inflating the power in the 2–6 year time window while filtering out variability on longer and shorter time scales. It is only because of this faulty analysis that they are able to claim such extremely high correlations. The suggestion in their conclusions that ENSO may be a major contributor to recent trends in global temperature is not supported by their analysis or any physical theory presented in that paper, especially as the analysis method itself eliminates the influence of trends on the purported correlations.(Emphasis added)
They also find some trickery (illegitimate this time) in patching the pre-satellite data onto the satellite record.
To me it doesn’t make sense. In the last decade we have been getting La Niñas with higher temperatures than El Niños in earlier decades. El Niños involve an rearrangement of heat within the earth system, not an input of extra heat, so could not result in an upward trend in global temperature across decades. Major volcano eruptions dampen temperature, but the effect washes out after a few years. The sun is on a 11-year cycle and is currently at a minimum when we’ve been having some of the hottest years on the record. And still there has been a strong warming trend in the global surface temperature which is where we live and grow our food.
In addition to the three links already provided here’s a sample of the rest:
Michael Tobis at Only in it for the gold
James Annan at James’ empty blog
Martin Tobis asks whether what they did was incompetence or intentional. He decides the latter. I beg to differ. I’ve met Bob Carter and he’s a nice bloke. He doesn’t have that sort of caper in him.
Greenfyre probably has the most comprehensive treatment including some links demonstrating the previous form of each of the authors.
The further question is how the paper got published in what is supposed to be a peer-reviewed journal. James Annan, in addition to finding one more error, goes into this to some degree.
JGR asks authors to submit the names of five possible reviewers. Although the editors may go elsewhere, the temptation is to be lazy. Also it does seem that JGR has an editorial policy of publishing contrarian climate science. It’s noteworthy, I think, that Morgan and Crystal in their book Poles Apart found that what they called The Sceptics were extremely unsceptical about anti-AGW publications. Just about anything goes as long as it’s anti-AGW.
I want to leave you with some images which I hope will convince you that the paper is an offence to common sense.
This image shows the multivariate index of ENSO.
As explained here ENSO comprises six factors, not just the SOI, which is based on the atmospheric pressure differences between Darwin and Tahiti. In this index the graph is either positive or negative, which is a bit misleading. We are simply not always either in an El Niño or La Niña event. There is ‘normal’ state also. When all the factors line up and reach a certain extremity you have what is called a “major” ENSO event.
If you want just SOI, here’s an annual bar graph up the other way, which is usual:
The main point is that there have been major La Niña events since 1998 and all along the way. Now look at this temperature graph:
Every year since 2000 has been warmer than the 1990s average. Every year in the 1990s has been warmer than the 1980s average. Recent El Niño years are clearly warmer than all earlier El Niño years if you go back a bit. In the main, ENSO involves differences in atmospheric pressure and changes in the heat distribution in the Pacific Ocean, plus such things as changes in wind direction and cloudiness. It’s bleeding obvious that additional heat is coming in from somewhere. It’s hard to see how ENSO could do it.
Then there’s the sun, where the pattern over time looks like this (from NASA GISS’s 2008 summation of temperature trends):
Match that with this image I posted in an earlier post from a piece by James Hansen et al:
The 11-year running mean should eliminate the year to year noise from ENSO and minimize the 10 to 12-year solar cycle. Looks like something else is going on.
I also posted this graph from Tamino where he used statistical analysis techniques to remove entirely the effect of ENSO and the aerosol effect from major volcano eruptions:
Finally, we are told that more than 90% of the heat from global warming ends up in the ocean. Here are two graphs much beloved by Joe Romm at Climate Progress:
The first is oceans to 2000 metres, the second global heat content. To those I’ll add this one from the dreaded IPCC:
I’m aware that the authors have responded to criticisms of their statistical methods (scroll down to “Technical Note from co-authors of study – July 29, 2009″).
Personally I don’t have the skills to arbitrate on the statistical methods, but in view of the above I can’t see how the interannual variance of temperature caused by ENSO could lead to a long term warming trend in surface temperature or a change of that magnitude in the earth’s total heat content. In the long run I think the contrarians, sceptics or whatever you call the anti-AGW mob will be mugged by reality. I’d be absolutely delighted if they are not, but I reckon that’s their lot.
I’ll end with this caution from an interesting post at Rabett Run:
Many non-scientists have no clear idea of how highly rated redundancy is in science. You don’t really believe anything seriously before it has come from several independent sources. And those sources themselves are often internally redundant, like surface temperatures, monthly averaged, correlate over long distances.
An excellent, comprehensive article. Thank you.
The SOI by my intuition is controlled by the Antarctic melt rate. The melt water being non saline floats and absorbs energy from the air as it disburses away form the pole but more importantly driven by the currents up the western side of the South American continent. As this melt water is increasing the SOI should be becoming smoother ie less extreme and more often. The whole system is driven by southward flowing currents on the one hand and by air flows from the upper atmosphere which fall over the Antarctic like a huge blow torch warming the overall ice mass. These air currents are fuelled by tropical heating. So as the tropical weather systems become more energetic due to heat contained by CO2 the upper atmosphere air movements should be increasing. This is how the Earth’s temperature control mechanism works. It is the energy in these air flows that is keeping us cooler than we would expect from global warming intuitively by pushing cold air away from the poles at a greater rate. But of course the whole process is about overall heating rather than local heating. The other invisible is that global warming is most observable in the average night time temperature. It is at night that the CO2 heat trapping is most effective.
There are huge apparent contradictions in the locally observable and the actual situation. It is going to be the Northern Hemisphere that runs out of ice first (most of the CO2 is released in the north and the north has the greatest amounts of methane to be released sooner), and that may happen fairly soon (relatively). Are there any climate scientists out there who have thought about what will happen when there is a global temperature difference between the northern hemisphere and the southern?
BilB, I’m aware that there was a suggestion that changes in the North Atlantic Oscillation (NAO) may drive the El Niños. But I’d have to stick with Wikipedia when it says, “Mechanisms that cause the oscillation remain under study.”
It is all just about energy flows, Brian, and the key sources of bulk energy both positive and negative are plain to see. Of course there are to be many hidden inter-relationships but the main forces are well known. The one that routinely never gets mentioned or appreciated is the upper atmosphere air flow volumes and energy content. We tend to focus on the surface effects as these air masses rise or fall. What climate science is all about is proving and quantifying what is regarded as being plainly obvious. And what a difficult task that is.
By the way the real proof is in the energy content details that you have regularly presented.
That’s a really nice post by Tamino. It does seem like the paper contains a basic statistical error. The authors’ note to critics just seems to suggest that they added the trend back in at the end of their model, not hard to do with a time series.
It appears that they’ve just shown that a large proportion of the variation around the trend is due to the ENSO…
Well I’m disappointed, I clicked through looking for info about dung beetles.
My mum’s farm has a very healthy population of dung beetles, released by CSIRO in a test a few years ago, a very much self-sustaining population now.
Yep – Tamino’s point is valid. The authors committed a simple logical error so that the second quoted paragraph is a pure non sequitur.
In fact the error is so obvious that if I was the Journal’s editor I’d be savaging the referees for not spotting it immediately (not boasting or anything, but I’d spotted it before I even read below the fold).
Yeah Brian, catchy heading got me too. Anticipation had me reading through, dreading the prospect that our imported (African) dung beetles were yet another casualty of AGW.
“El Niños involve an rearrangement of heat within the earth system, not an input of extra heat, so could not result in an upward trend in global temperature across decades.”
I urge you read more widely and think a little harder.
“If you can’t get at the original article behind the pay wall the main findings are here.”
That main findings link adds a new layer of obfuscation by splitting McLean’s GTTA/ENSO graph (figure 1 in the excellent Skeptical Science rebuttal) into three, at 1980 and 1995. The effect is to obscure the otherwise clear warming trend.
To demonstrate it, I’ve done the same with surface temps in my comment at Skeptical Science. A “trick” to “hide the incline”?
EarlH,
What Brian has said is more or less correct. The cause of El Ninos is entirely to do with the redistribution of energy, the part that you could pick issue with is about whether you take into account the energy entering the atmosphere that is then redistibuted under the atmospheric consequences of the El N surface temperature. But that energy entering the atmosphere is relatively constant baring cloud cover considerations.
Great post, as per usual.
Agree with patrickg @13.
Thanks Brian, another great post!
Shame about the lack of dung beetles, though.
Sorry about the title. I put it there as a working title and started the post with reference to a pile of pooh deposited on the other thread. Which kind of represents how I think of it.
But in these posts you can dig yourself a hole and drop yourself in it, plus I like to be a bit less inflammatory. So I toned down the first para, but forgot about the title.
I’m fairly confident that the story I was attempting to give in the visuals will turn out to be right, but I’m not a full bottle on ENSO. I’ll come back to this later tonight.
Meanwhile those of you who are mathematically literate have confirmed what Tamino said. Having read his posts over a period of time I have a great deal of confidence in him. Also the authors in Foster et al represent a critical mass of impressive expertise.
Jonathon @ 11 thanks especially for the link to Skeptical Science. A good post and an amazing list of articles at the end.
Brian, I can’t see how the North Atlantic can have enough effect to defeat the effect of the mass of cold water coming off the Antarctic with relation to the El Nino. The higher probability is that Antarctic flows could interfere with the conveyor current flowing around the bottom of Africa and in that way appearing to form a link of North Atlantic events and the El Nino. The article you linked to from my scan read does not say absolutely that the Norh Atlantic drives El Nino, it simply points to a correlation.
But the most telling indicator is the ocean energy content steadily rising. You can only try to hide that by tilting the graph paper sideways 45 degrees to make the trend seem horizontal.
BilB, I didn’t follow everything you said @ 2, but let me have a go, while real climatologists will probably shake their heads and wonder. I think McLean & co are saying that the Hadley circulation transports the extra heat generated near the tropics in an El Nino to higher latitudes. This in itself is a stretch. Anyway it all seems to depend on atmospheric and oceanic circulation and heat transport.
First of all I recall reading that the heat content of the entire atmosphere is the same as that contained in the upper 3 metres of the ocean. The ocean is on average 3.6km deep, and a fair bit of that approaches 0C, so there you have an enormous heat sink. I can’t readily find the average sea surface temperature, but it can’t be far behind the average land-ocean temperature. So if we are going to have remixing without a net heat input, then you’d expect cooling rather than heating of the average surface temperature.
Here’s your latitudinal circulation patterns from this article.
Here’s your ocean surface currents from this article.
The other main factors seem to be upwelling and the Walker circulation which changes configuration during an El Nino.
There’s a good explanation of the Walker circulation during an El Nino here.
As that reference says, the winds normally blow from east to west across the tropical Pacific to give a temperature 8-10C higher on the western side. During an El Nino, the winds fail and the water heats more in situ in the central and eastern Pacific. Also, some upwelling of cool water on the eastern edge is normal and this lessens or ceases during an El Nino.
The upwelling factor in addition to the surface current sweeping up from the Antarctic seems a critical factor, which is much enhanced during a La Nina. This is well described in this Skeptical Science article. During a La Nina the winds are stronger, pushing the warm surface water further west, which is displaced in greater part by upwelling of cool deep water. Makes for good fishing off S America.
So I’m still struggling to see how a periodic El Nino, whose affects seem to dissipate quickly, can transport heat to the polar regions, which are warming more than twice as fast as the tropics, and sustain a longer term surface temperature rise, including the 70% of the planet covered by oceans.
It’s easier to see how the La Nina leaves a significant imprint on upper ocean heat content (see Figure 4 at Skeptical Science). In fact, both do, according to Figure 4, with the La Nina having the greater impact in the tropics. But they are only blips on the upward trend which seems to continue strongly in neutral ENSO conditions. That’s what McLean et al should try to explain.
As to the influence of the North Atlantic, I haven’t seen an explanation of a mechanism yet. I suspect that the driver of ENSO is the configuration of the atmospheric pressure (SOI and Walker), which affects the winds. So what drives that? But I confess that the interaction between these major systems is new territory for me, and I think not well-understood by the professionals.
Great post. I’m a skeptic but this is the sort of thing that helps things along.
No mention of denialism and the alleged vast conspiracy that pays evil think tanks to go and attack ‘the science’.
Careful Brian, that reference to the vast and apparently unlimited heat sink capacity of the oceans will invite all manner of mischief! Rather like the old idea of dumping nuclear waste into the intertectonic plate fault lines.
One question I have is about the models predicting increasing rates of temperature change.
While lots of discussion focuses on the existence or otherwise of the presence of climate change driven by increasing temperatures no-one seems to be engaged with the the previously often mentioned “tipping” points.
Has this idea been lost in the mess of the CRU email imbroglio?
Monckton was mentioning that the rate of change was negligible according to one set of calculations but where is the current thinking on this issue expecting us to go?
(Completely off thread are you considering any posts about BSE affected countries and the importation of beef from such places?)
MTF,
The discussion has nothing what ever to do with the reality. The ice is still melting…faster now, the methane is still gushing out of the peat bogs of the arctic north…faster now, The tropical weather systems continue to develop…stronger now, the ocean heat is building…steadily, tropical zone earthquakes are continuing…slightly more frequently and more severe, weather continues to change…inconveniently, and insurance rates continue to rise…unaffordably. Fires rage, floods deluge, snow blankets, oil depleats, economies crisis, people die. But the television still works,….so settle back and relax. Don’t worry, man.
actually Brian, I wonder if the original paper contains an obvious sign of the dodge that Tamino mentions. The intercept for any time series model with annual differences should be (as Tamino observes) the magnitude of the incline. I bet when they analysed the data, the authors saw that incline statistic, but they haven’t given any evidence of it in their publicly available summary. It should be in the paper, but I wonder if they have buried or not printed the coefficients which don’t pertain to the correlation term. This is pretty standard time-series analysis stuff.
This is perhaps why some people prefer to build all these kinds of models as linear regressions with correlated residuals – you can’t bury anything in the detail then, and you always estimate the trend. I wonder what the full content of that paper is, and whether it obscures some of this information…
MTF, I’d like to do something on BSE, but I don’t think I have time.
The “tipping points” are always around the place. On climate sensitivity, my understanding could be a bit faulty, but here goes.
The usual approach tries to work it out from observations of short-term feedbacks (Charney, up to 30 years) based on observations, necessarily short, using models to some degree. You get outcomes like those shown on Figure 2 on this post or the 55ppm line on Figure 1. This boils down to 3C plus or minus 1.5C, with greater uncertainty on the upper side, hence the “fat tail”. I’ve never seen it said as such, but I suspect the fat tail is there because of uncertainty about tipping points. Other than that the major uncertainties (plus or minus 1.5C) relate to the net effects of clouds and aerosols.
Roger Jones has said he doesn’t like this linear approach to expressing uncertainty, I think because it tends to quantify the uncertainty of non-linear effects, which are essentially unquantifiable. I think that’s what he’s saying.
Hansen’s recent work on this is important. First, he has looked at the issue in the paleorecord, especially in the early Cenozoic (65 to 35mya) when there were no major ice sheets, but also coming out of the last ice age. He puts climate sensitivity down as 3C plus or minus 0.5C when you take longer term responses such as albedo on ice and ice sheet response out of play. But he reckons if you add in those factors, plus vegetation response (more stuff grows further north, reducing albedo, without forming a decent carbon sink), as you must in the situation we are in, climate sensitivity is doubled.
Please note that this approach is based on observations, albeit indirectly through proxies.
There has been one recent detailed look by a bunch of researchers at one event a few million years ago, which worked out climate sensitivity at 4.5C. Obviously we need more studies like that.
When you think about it this could explain why scientists are continually surprised by how fast things are occurring, because future scenarios based on models have only been based on short term feedbacks.
Secondly, he reckons that we are in tipping point territory already for such things as the drying out of the Amazon and methane emissions. On the latter, he reckons the methane clathrates continually build up and are then released at some point over spans of millions of years. The last big one was the PETM 55mya, when he reckons most likely the methane went off as a feedback caused by a 2-3C temperature rise which happened through GHG, mostly CO2, forcing. He reckons it takes a few million years only to reload the methane gun, which is well and truly loaded now. We’ve escaped because of the long term cooling in the last 50 million years.
On the positive side, the methane thing last happened over around 10,000 years. It’s not going to blow up in your face in terms of human time-scales, but will make mitigation a whole lot harder to mitigate if it really starts in earnest.
Third, Hansen reckons the threshold for an ice sheet free world is 450ppm, plus or minus 100. That’s why he chose 350ppm as the first target we should aim at. To quote him, he says “it would be exceedingly foolish and dangerous to allow carbon dioxide to approach 450 ppm.”
So these days he’s not at all keen on 2C which he had earlier subscribed to but abandoned sooner than most. Science generally and politics hasn’t caught up with him yet.
So the science isn’t settled, but it’s probably heading in an even more alarming direction.
I understand that model builders are now trying to build in long term feedbacks, but the difficulty is that they can’t do this very easily on the basis of observations from the instrumental record. We have no experience of ice sheets disintegrating and reforming.
sg @ 20, confession time. I haven’t seen the paper. I could get it, but it would take me some trouble, and given my lack of statistical skill I’d take more notice of what the likes of Tamino say rather than anything I could nut out for myself.
The James Annan link says they had the help of a statistician, so there is a question as to how skilled the authors are at statistics, how competent their help was and what he thinks about what they ended up publishing and more particularly the conclusions the authars, especially Carter, have drawn which may not have support in the paper.
I’d really be interested if someone with the requisite skills would comment along these lines.