Bishop Hill Climate magic
Nov 15, 2013 Climate: Surface 
The green side of the climate debate is getting terribly excited about a paper by Kevin Cowtan, a chemist from the university of York, and Robert G. Way, a geographer from the university of Ottawa. They claim to have discovered that the pause is illusory and due to incorrect estimates of temperatures at the poles. With their new whizz-bang method of making up data they claim to have magicked the missing data into existence and, surprise suprise, actually the poles are warming very quickly and the pause doesn't exist.
Judith Curry is having fun with the paper, noting that the methods they use to estimate the missing data are not exactly suited in the particular circumstances of polar temperatures:
- the statistical infilling technique of kriging 'makes no physical sense' when applied across land/ocean/sea ice boundaries
- satellite data is 'not useful' when applied to the polar regions
- data reanalyses are 'not useful' because of temporal inhomogeneities in the datasets that are assimilated.
It will be interesting to see how much traction it gets outside the green ghetto.
Reader Comments (64)
But surely this cannot be correct: Tom Chivers has just told us that
Oh dear - it's all very confusing.You have to wonder what these people are smoking, telekinetic trees, heat that sneeks past multi million sensors they designed and now (but not for the first time eg Stieg) more imagined from thin air data.
Two points immediately spring to mind.
1. Why would a geographer and a chemist want to write a peer reviewed paper on the construction of (et another) global temperature series? Obvisous answer is that is for personal (activist) reasons
2. Their approach makes absolutely no sense from the outset. If you think the satellite data provides a credible estimate of global temperatures, why are you cherry picking some of it to shoehorn into another unrelated series of data? The UAH and RSS data both indicate a pause, how can you logically inisit that taking part of those series and combining them with another series provides a more credible or accurate result??
This is peer reviewed isn't it??
Some idiots actually agreed this was worthy of publication?
Pal-review. All greenie mates together. Simple as that
Green Ghetto - I love it!
This very neatly illustrates the religious nature of the warmists' belief system.
For example, if someone were to actually prove that god did not exist, there are those you would claim that such a proof was in fact evidence for the very existence of god.
These types are completely impervious to rational argument. Their religious fervour is all-consuming.
I remember ages ago Tony Watts having a post up about the exact sane thing happening in Africa where climate scientists were using station data and then extrapolating that data across areas where no data existed...in many cases the areas were separated by thousands of miles!
Secondly, and more problematic, how is it something like this gets through peer review? Do journals now only use "approved" reviewers these days???
Anyway, I'm betting Entropic Mans left trsticle that politicians will be citing this paper (like how they used Lews paper) to support their energy policy follies!
Mailman
I see it now. There has been so much warming that in order for it not to register where the thermometers are it must have gone where the thermometers are not - flown to the poles and dived into the deep dark sea.
http://onlinelibrary.wiley.com/doi/10.1002/qj.2297/abstract
From the abstract:
"Two alternative approaches for reconstructing global temperatures are explored, one based on an optimal interpolation algorithm and the other a hybrid method incorporating additional information from the satellite temperature record."
"Optimal interpolation algorithm " Optimal for who? I think they mean it gives the results they want.
And
" with the hybrid method showing particular skill around the regions where no observations are available."
For me this last sentence sets alarm bells ringing, basically using statistical "hybrid" methods to create data out of thin air. Also how can they quantify or measure "skill" when they cannot compare it to the actual data, which does not exist?
It would be an interesting test to take the temperature records for the whole of the United States (or any other large area with a good temperature record) and apply the same methods as used by the paper. In essence extract a few temperature record series on the West of the USA and a few on the East and see if these "hybrid" and "optimal" methods can be used to accuarately derive the tempertures for the remaining areas in between. Then compare with the historical measured data. Simple.
I have not read the actual paper as it requires subscription, but as an engineer who has to do real life stuff (often involving mathematics) I would classify this paper under the heading of "bo**ocks". A technical term sometimes used in my place of employment.
Cast off the shoes! Follow the Gourd!
I think we are making a terrible mistake poking fun at these climate scientists all the time. What would we do if they all took umbrage and went on strike ?
Where would we be then, eh?
I'm sure Steve Mc did a post on kriging some time ago. Perhaps he will have a look at it. Mind you, sooner or later he's going to throw up his hands and say "enough of this idiocy, I've got better things to do". That would be completely understandable, but it would be a sad day for those who still believe that honesty, accuracy and accountability form the basis of science.
@gekko "a geographer and a chemist" so I'm sure that Steve Jones will ensure that this paper is not mentioned on the BBC by enforcing the "they are not climate scientists rule"
..oh silly me that rule only applies to people making skeptical arguments, every Greenpeace activist, solar panel salesman and their dogs are allowed to express opinions in favour of the "catastrophy is certain " line.
EternalOptimist at 10:43 AM
Climate science strike!
What a wonderful thought, we can but dream. However I do think that these climate so called scientists seem to take great pleasure in how persecuted and put upon they are. Perhaps in their minds they see it as confirmation of their righteousness and taking the true path.
Once again a paper is released by peer review that is conceptually wrong. That's okay - peer review is not a badge of absolute merit, a concept many of us here are familiar with.
However yes many AGW proponents would claim that just by passing peer review this paper must be correct. And furthermore that you need to publish a rebuttal to disprove it.
This actually shows ignorance of the scientific method. A paper can be factually wrong - as in the data is incorrect but the approach and method are sound. An example is that there was a hidden bias signal that is only found through more experiment.
Then of course a paper can have a methodology issue - as in the assumptions and calculations are incorrect.
Lastly and most importantly a paper can be conceptually wrong. As in creating data and then using the data as proof of something.
For this category you don't need rebuttal. You just need a simple sentence: "Wrong...by inspection". Maths has been doing this for years.
So this paper as pointed out by others here is wrong by inspection. The basic concept does not justify the outcomes.
Robert Way seems to be on the staff at SkS: http://www.skepticalscience.com/team.php
Nutticelli has fallen for it of course (as I just wrote on twitter, Leo DiCaprio's phrenology lesson in Django Unchained reminded me of Skeptical Science for a very good reason) but...does anybody know if any of the big guns has endorsed this paper?
Hang on a minute... So "the extra heat" is finding its way into the ocean, slipping quietly past all the land and satellite based monitoring - OK. But, the BBC et al are now banging on about "ocean acidification" caused by increases in dissolved CO2. I'll admit that I last did chemistry at "A" Level - and it was so long ago that we had to write the answers in Latin, but ISTR that the solubility of gases decreases with increasing water temperature... So WTF? It can't be both. Maybe ZDP or Entropic, or someone else with far greater scientific knowledge than I, can explain it to me.
I would recommend Lucia's for a more rational review of the paper and what it is doing
http://rankexploits.com/musings/2013/cotwin-way-trends-compared-to-ar5-model-trends/
This paper seems to be another in a long line that illustrate just how poor and unscientific some branches of science have become. Many scientists and academics no longer sit in postions of high intellectual and ethical esteem but are more like a sixth former with some dodgy science project to cobble together.
The reviewers of this paper should have got alarmed after a basic sniff test and chucked it straight back to the authors with a note saying "when you have graduated/grown up/sobered up try again" or "stop taking the piss." However as it is climate science this does not happen.
I understand that much of climate science is not real science, but one of these guys is a chemist for goodness sake and must have to do something real now and then. Don't they have some basic decency, integrity or common sense? Do they not realise how this makes them look? Or perhaps climate science has infected more of the academic world then we realise? Fools who cannot see what they are.
I am sure there are others here who realised all this long ago and can express it much better than I can.
Ah, I see.
It's no longer Global Warming.
It's Places-where-there-don't-happen-to-be-any-thermometers-or-even-people Warming.
Be afraid. Be very afraid
This is sokal-on-stilts.
It's a false flag operation to discredit 'the cause'. Like all good parody it looks real until you examine it closely.
The heat isn't hiding in the deep ocean remember Al Gore told us it's millions of degrees a few miles into the crust. It's obvious all that heat is going down through all the air water and crust directly into Gore's furnace that's why we can't detect it. I expect Brian Cox will be along in a minute to explain that the heat engages in quantum teleportation straight down, there might even be a paper in the works!
That CO2 it’s tricky magical stuff, it bends the known laws of physics, chemistry and geology. You couldn’t make it up, all the stuff it does, could you?
Yet the only warming data we had in the first place was in the Arctic and Siberia - plus of course the guesstimated warming for the oceans that magically disappeared when more accurate data came onstream. And everyone knows the Antarctic is cooling: The previous attempt by Mann/Steig to smear it all with peninsular data were largely ignored. So I trust this will meet the same fate. It just increases the tabloidization of science journals.
I'm still waiting for anyone to do any data adjustment that produces some cooling - just to balance things up a bit. Funny every temperature sensor being discovered to run cold: Not such a common phenomenon in any other field of science or engineering!
@ Black Dog, 1204 & 1210 abobe.
Don't you know not to feed the gobshite troll?
It just encourages it.
(Gobshite...Loud-mouthed person who talks a lot, but nothing with any value - as in shite coming out of their gob).
PM Walsh
The emphasis on a zero (or minimal) slope has always been misplaced, although perhaps it's useful from a rhetorical point of view. More important is whether the models' projections, on which mitigation's value is predicated, are accurate. In this regard, the key takeaway from Judith Curry’s column is this: “The bottom line remains Ed Hawkins’ figure that compares climate model simulations for regions where the surface observations exist. This is the appropriate way to compare climate models to surface observations, and the outstanding issue is that the climate models and observations disagree.”
My initial comments based on a scan through read of the paper:
1. For those unfamiliar with the jargon, for a linear weighted estimator based on minimum variance, kriging is optimal.
2. Spatial resolution of temperature measurements, satellite measurements. How has the temperature data been upscaled to the grid resolution (variance correction) for HadCRUT or has only point estimation been used?
3. Comments on benefits of kriging:
Yes, kriging is an exact interpolator but (a) the measurements should only be matched exactly if there is no measurement error (if not – use kriging with measurement error) and (b) temperature measures are point values, we are estimating the average over a region (block kriging of the original points onto the cell). Therefore what are the support corrections for (a) HadCRUT (b) the kriging estimator used and (c) the satellite data?True, but this is only a desirable property (a) when interpolating not extrapolating and (b) if the variable under consideration is stationary within the coordinate domain.
Yes, this is an advantage
How can a cell contain station position information? At what scale are the HadCRUT measurements ie are they point estimates per cell or are they upscaled to the grid scale (with the corresponding variance adjustment)?
So why was kriging with measurement error not used then?
4. The hybrid calculation – what is the correlation coefficient between the two measures? Are they at the same support (measurement) scale? Why was collocated co-kriging or kriging with external drift not used?
5. The usual way to show cross-validation is by cross-plot eg predicted vs actual, error as a function of actual etc. Why are these displays not shown instead of a table which can hide many problems?
6. What was the neighbourhood search strategy used for the kriging?
7. Was this computed on a sphere? If on a regular grid of 5 x 5 deg , then it is unequal area and the kriging estimator is incorrect as used as the cell size is not taken into account – this is critical for quantities that are dimensionless for example. Was the kriging 2D (not spherical) or 2D (spherical) or 3D (either XY/Lat/Long) in either flat grid or spherical plus temporal change?
8. Experimental variogram not shown, what is the model fit to the data? The range of the model is not stated, nor can we see the quality of the fit of the variogram model to the experimental variogram. If the range of the variogram corresponds to the distance you want to extrapolate ie the arctic, then the spreading of values will depend on (a) the variogram range chosen and (b) the local neighbourhood search chosen. Or is this a unique neighbourhood?
Actually, what this strongly suggests is that the satellite data (ie real observations) add nothing to the estimation? What is the correlation between the satellite data and the real data? Zero?Why the 1200 km cutoff, where does this come in? What range has been used for the variogram model? What is the neighbourhood search range, or is a unique neighbourhood used?
With the post above, it would be nice if the authors might comment here. I will not post at Judith Curry's because there is just way too much clutter.
These are my initial thoughts, some of the notes I would have made back to the authors had I been a peer reviewer for this paper.
Scottie
"were to actually prove that god did not exist"
As Douglas Adams noted...
"I refuse to prove that I exist," says God, "for proof denies faith, and without faith I am nothing."
"But," says Man, "the Babel fish is a dead giveaway, isn't it? It could not have evolved by chance. It proves that You exist, and so therefore, by Your own arguments, You don't. QED"
"Oh dear," says God, "I hadn't thought of that," and promptly vanishes in a puff of logic.
I don't see how can "kriging", with respect to temperature, can work across a phase change of water, i.e. where there is both water and ice present. It appears to require a continuous function, and a physical phase change is, by definition, not continuous.
Disappointing comments here. All of you heard that (a) the paper supports the AGW supporters, (b) someone here says it's bad - and all lined up to criticize a paper you didn't read and don't understand (there are a couple of exceptions)? Sigh. I'm mostly a skeptic, but I can be disappointed in skeptics too. See both the threads from Curry and Lucia, where the authors reply to the criticisms and explain some of the details, AND where Curry and Lucia and Mosher and others come back with more discussion. It sounds like a reasonable, serious attempt. And according to all accounts, it doesn't disprove the Pause at all, regardless of what nonsense others are saying about it. - Oh, then it's okay then.
Michael Hart: "I don't see how can "kriging", with respect to temperature, can work across a phase change of water, i.e. where there is both water and ice present. It appears to require a continuous function, and a physical phase change is, by definition, not continuous."
I don't see this as a problem myself as I would expect the temperature to be a continuous property and especially its monthly average. However, the properties are almost certainly not stationary in this context (eg variance of data on land I would expect to be larger than for the sea) and this might lead one to regard the problem as multivariate (eg land and sea data), or use some other form of kriging which can deal with the data in different categories.
pogo: "banging on about "ocean acidification" caused by increases in dissolved CO2...the solubility of gases decreases with increasing water temperature..."
You're correct that the solubility of CO2 (in particular) in water decreases with increasing water temperature. Recalling Henry's law, the concentration of dissolved CO2 is proportional to the pCO2 in the atmosphere. It is the coefficient which decreases with temperature.
So there are two effects operating: first, the increase in atmospheric concentration of CO2, which tends to increase the amount of dissolved CO2; and second, the decrease of the Henry coefficient, which tends to decrease the amount of dissolved CO2. The first effect, which ceteris paribus would result in approximately a 50% increase, dominates the second.
thinking scientist -
I would recommend posing these questions at Lucia's site.
It is generally agreed that the global surface-to-lower troposhere amplification factor (as derived from the laspe rate enhancement) is, or should be, approximately 1.2. So, if the satellites have now suddenly become super-accurate and reliable, let's go the whole hog and use them to estimate the global surface temperature trend. By ratioing the current RSS decadal trend of .127C down by 1.2, the new and improved GISS' decadal trend (since 1979) would come in around .106C instead of its feverish .159C. Better than we thought! This method has the additional benefit of saving a lot of money, since we no longer require either the GISS and HADCRUTx temperature series, nor the associated expensive and sophisticated efforts of the TOBS adjusters and the UHI non-adjusters.
Micky H Corbett wrote:
"That's okay - peer review is not a badge of absolute merit, a concept many of us here are familiar with. However yes many AGW proponents would claim that just by passing peer review this paper must be correct"
Can anybody give some examples of this, preferably as strong as possible? This is not a snarky question, I'm interested in examples of this, particularly in climate science. To be clear, I'm interested in examples where peer review is claimed as explicitly as possible as something like a 'badge of absolute merit' rather than a first pass or a plausible result.
If there are examples of AGW proponents claiming that in relation to this specific paper, all the better.
HaroldW, "Recalling Henry's law, the concentration of dissolved CO2 is proportional to the pCO2 in the atmosphere."
Yes, at equilibrium. But neither natural sinks, nor natural sources, are at equilibrium. They are not even at a steady state.
Robin Guenier,
"But surely this cannot be correct: Tom Chivers has just told us that
'the latest research shows that most of the extra heat has been ploughed into the oceans (hence the erroneous claims, based on surface temperatures, of a “pause” in warming).'
Oh dear - it's all very confusing."
Why is that confusing? Lewandowsky explained it to us a while ago:
"While consistency is a hallmark of science, conspiracy theorists often subscribe to contradictory beliefs at the same time – for example, that MI6 killed Princess Diana, and that she also faked her own death."
http://climateaudit.org/2013/11/07/more-false-claims-from-lewandowsky/
Climate Scientists have found the missing heat hiding where it can't be detected at the bottom of the ocean. And at the same time they have found it hiding just out of sight at the North Pole. To find it twice is pretty impressive, I think. That's how good they are at finding these things.
JK
I think you are splitting hairs but to answer your question:
To be exact, I used AGW proponents as it encompasses media as well as any proclamations by scientists studying in the field or even just generalists.
One immediate example was Brian Cox's Science Brittanica, the one where he looked at peer review. The editor of Nature was seemingly arguing that peer review IS a badge of merit. That a paper, though not perfect, has had a good going through.
Another example is that Michael Mann seems to think that his papers, especially MBH98 and related ones, are somehow above reproach. He gets very testy when they are critiqued. But he doesn't explictly say they are a "badge or merit" though but he does appear to believe they are correct.
And also there was the very Parlimentary Meeting that our Grace attended in which various bodies appeared to use the peer review literature as a reference for the work being good and that it was correct. No mention that you still have to analyse what a paper has to say and make up your own mind. Which is actually what our Grace argued.
I don't know if any scientists have blatantly said "peer review means a badge of merit" or "this means it is correct" but the inference is there in climate science more than other branches I've been involved in. The IPCC appear to use this line of reasoning when announcing the scope of their reports.
And also peer review is simply that it passes a minimum review. The purpose of publishing is to seed further work and study in the academic forum. The reviewers themselves are never going to catch everything. They might not even catch an obvious flaw in logic or approach if they are working in the same field, as you can lose perspective of things sometimes and be prone to confirmation bias.
Nov 15, 2013 at 11:41 AM | Unregistered Commenter Black Dog
"Do they not realise how this makes them look?"
Probably, but I'll bet they don't care how they look to you and me, but more likely how they look to somebody they think is important. And those important people are probably not impressed by objectiveness.
JK, I doubt you will find many specific examples of such claims about peer review. The way it generally works (not just in climate science) is the larger, more extravagant claims are made in the talks, speeches, press releases, TV interviews, and presentations to investment capitalists, not in the peer-reviewed literature.
However, even the BBC recently recognised some of the dichotomies that crop up, with their report of the recent Nobel Prize in Economics being awarded to individuals with, reportedly, completely contradictory theories. Appropriately enough the title was "Are markets 'efficient' or irrational?" Doubtless all the competing contradictory theories were peer-reviewed.
PM Walsh at 2013 at 1:45 PM
Hi
Yes I know, you are totally correct about the troll. I was off work today and coming here was better than cleaning the house or stuff like that, whilst my wife was at work. It is just so hard to resist engaging someone who may well read the Guardian, but when I come away from it all I feel the need for a shower and a good scrub.
I have always loved the term gobshite, having grown up in a wee country where it is often used. Such a wonderful word, it is the sort of thing Mr Heaney may well have used.
JK,
There was a demo where the "climate concerned" were "armed only with peer-reviewed science"
link to pic
The words "peer review" have been bandied about so much that a lot of people really do see it as some kind of stamp of approval. An example from the world of advertising...
source
JK "Can anybody give some examples of this, preferably as strong as possible? This is not a snarky question, I'm interested in examples of this, particularly in climate science. To be clear, I'm interested in examples where peer review is claimed as explicitly as possible as something like a 'badge of absolute merit' rather than a first pass or a plausible result."
I recently complained to the BBC about their use of Marcott's temperature graph as being misleading, pointing out that following comments by the authors the Met Office removed the paper and discussion from their web site but the BBC uses its "hockey stick" as a poster child (see http://www.bbc.co.uk/news/science-environment-24204323)
The BBC news online response to the complaint was:
That's it, the entirety of the response to 6 highly relevent criticisms pointing out factual errors. But its peer reviewed, so it must be true. Its just like if you read it in The Sun, it must be true.
It will be interesting to see how much traction it gets outside the green ghetto.
Well if it gets into the AGW dogma , it will 'have to be defended unto to death ' no matter how rubbish it is .
Remember this is 'religion not science we are dealing with.
To me, this paper has to be arrant nonsense: cobbling temperature extrapolations together from areas where no data-gathering devices are sited appears to rank right up there with Lewandowsky's latest exploits, where he has used populations which are null sets. Complicated nonsense is still nonsense, no matter how cleverly written.
I suspect many scientists in the Green camp have become so accustomed to having their reviewers wave nonsense through that those scientists have lost touch with reality.
Alexander K: Its not arrant nonsense. There is nothing intrinsically wrong with kriging to obtain the estimates, and showing the uncertainty. There is nothing wrong with using a related measured (eg the satellite data) and using it as a constraint. What I don't understand is why they don't use standard methods and use standard results. Eg cross validation plots and variogram displays. No sign of these in the paper, yet they would appear in any paper in an oil industry paper, soil science paper or mathematical geology paper where these techniques are used regularly. Why have they commented that there are measurement errors and then not used kriging with measurement errors? Or they use a related variable but have not attempted a comparison wth co-kriging, co-located co-kriging or kriging with external drift. These are standard methods, well understood and highly appropriate for the problem under consideration.
Also, as far as I can tell so far, they are kriging on a degree based regular grid, but not accounting for the unequal area associated with the cells on the grid projection. So simple to correct for (I may be wrong, perhaps the authors can correct me on this point). Why write all this stuff in Python, from scratch, when the kriging estimation could be done in a well known industry standard package, such as Isatis? Why don't they report the neighbourhood search, state the variogram model used, show the match to experimental data? There is much to praise in the approach , but so much to criticise in the execution too.
thinkingscientist-
"7. Was this computed on a sphere? If on a regular grid of 5 x 5 deg , then it is unequal area and the kriging estimator is incorrect as used as the cell size is not taken into account – this is critical for quantities that are dimensionless for example."
Interesting comment. It would help explain what I estimated to be the satellite trend needed in the Arctic.
Based on what the RSS trend is for 60N to 82N (WUWT posted it today), and the fact that the Arctic is less than 4% of the global surface area, the 0.33C/decade RSS trend measured there (UAH is similar) is five times too low to generate the hybrid trend claimed in the paper. The RSS Antarctic trend is actually slightly negative, so it makes things worse.
Chris Y: there is a further problem if this computed on an XY grid (albeit in degreees lat/long) - are they taking account of the fact that the left and right edges of the grid are actually adjacent to each other on a sphere? It is not at all clear from the paper whether this is taken into account. Along with all the other points I mention.
Thinkingscientist....your questions might get answers over at Lucia's Blackboard blog, where the atmosphere is calmer and ultra-scientific.