Thursday
Apr072011
by 
Bishop Hill
Bishop Hill Climate heroes
Apr 7, 2011 Climate: Sceptics Climate: Surface Eli Kintisch interviews Richard Muller, whose BEST project has been causing something of a stir in recent days. Muller certainly knows how to get attention...
I realized that Watts was doing something that was of importance. The issues he raised needed to be addressed. It made me seriously wonder whether the reported global warming may be biased by poor station quality. Watts is a hero for what he's done. So is [prominent skeptic blogger] Steve McIntyre.
ABC in Australia is also looking at BEST and Anthony W.
Reader Comments (54)
A case of don't do as I do, do as I say.
Muller is a hypocrite.
Those Australians are so backward. This is illustrated by their choice of name for their website - "The Drum". You'd never get away with that in this Country.
http://www.youtube.com/watch?v=NtaegFGgRRQ
A Eli Kintisch YouTube clip about his book, "Hack The Planet" about geoengineering. I do like the analogy stated by Mr. Kintisch, "that there is plenty of drugs out there that work well but doctors don't understand why they work." So much for the billions spent on research and medical trials just to test the potency and efficacy of drugs.
It seems for both Muller and Kintisch that only science that fits an agenda has any worth.
Mac
Give it a rest eh? Let's see what BEST actually delivers.
Have you read Montandon (2011) properly yet BTW?
Especially the key finding that LST trends have been under-estimated due to spatial measurement bias.
Full text pdf here:
http://pielkeclimatesci.files.wordpress.com/2011/02/r-344.pdf
You know in your heart of hearts where this is headed don't you? I mean just squint at any temperature record. They all slope up to the right.
Deriving linear trends for temperature is a completely meaningless excercise.
Derive a linear trend for the complete UAH satellite record (30+ years).
Derive a linear trend for the complete GISS record (130+ years).
Derive a linear trend for the complete HadCrut record (160+ years).
Derive a multitude of trends, you choose.
What does it all mean? It means nothing! You are not seeing the data if you rely on trends.
What has BEST delivered so far? It has made unsubstantiated claims - no supporting evidence - no peer review studies - it has seen Muller piss in Watts pot - it has seen a breach in confidence - it demands better behaviour from others - it has seen Muller give results and interviews to warmists and alarmists.
BEST is a sham.
That, as they say, is not even wrong.
BBD you are not even right, never mind correct.
BEST is a fraud.
"of importance" instead of "important": the USA will be drowned by a rising ocean of needless words.
Mac
Why, then, does Muller say:
Despite your gross misrepresentation of this quote in your first comment, perhaps you can return to it objectively for a moment.
He re-states that BEST data and methodology (code) will be publicly available.
How - by any stretch of the sane imagination - can this possibly be fraud?
So why do you keep attacking BEST with this virulent, unfounded (and libellous) accusation?
Why not wait for the results and the examination of same by Watts, Pielke Snr, Eschenbach, Mosher, Broberg, Hausfather, Lucia, etc?
It's not rational. And nor is this:
Would you like to provide some references to back this statement up? Because I'm interested to see how far you get.
And when you've done the literature search on that, let's get back to Montandon et al.
I'm curious about why you referenced that paper when it clearly shows how LSTs may have been under-estimated.
I thought you believed the opposite. This has me confused.
Or were you flailing around for an argument and accidentally clutched at the wrong straw? Is that it?
BBD -
Perhaps you would like to comment on Doug Keenan's revelation in wsj i.e. that trends are only meaningful once you have proposed and underlying model and validated that model using statistical tests?
If you propose a model with no autocorrelation, you test for it.
If you propose a model with each observation only autocorrelated with the previous one, you test for it. and once it has passed that test, you know how to interpret your discovered trend i.e. is it significant or not.
Is that what Doug was saying?
If you skip the relevant test for autocorrelation, then any trend you claim to have found is meaningless.
matthu
If I understand Keenan aright:
- his analysis does not appear to explain the existence of longer-term trends in temperature data
- he proposes no physical mechanism to account for these trends
- he does not therefore show that the trends present in the various temperature records are invalid
- he does not explain why similar trends are evident in surface temperatures, SSTs and tropospheric temperatures over the 32 year period of satellite observations.
P.S. I think it is actually the residuals that should be tested for autocorrelation - but the basic idea is the same.
Sometimes, I despair ...
This fixation on 'global temperature' is like taking our eyes off the ball: we do not know what causes this rise!
as BBD said above, the graphs all slope to the right, 'up', I think he meant.
So they do ... but so they have done in the past, and then the slope changed direction downwards.
There hasn't been sufficient research done to explain why, simply because of the ICCP's insistence - with the help of certain individuals in the scientific community! - that it is due to CO2 and to nothing else.
Nobody who posts here can be unaware of the shenanigans which have been going on to keep CO2 as the sole culprit, and the extent to which this has been successfully sold to politicians across the globe to not just rise taxes but to strangle manufacturing industries and to keep their populations poor (energy price!).
Therefore, BEST should be seen as just one more attempt to get something right, in order to support the second pillar of AGW, namely the culpable human act of rising CO2 levels - and forget about that yellow thing in the sky (can't tax that), forget about what is sloshing around the continents (can't tax that either).
I am still appalled that the gate keepers and the MSM are dissing and playing down all those who do proper, observation-based science - be it Anthony Watt's project, be it Courtillot, be it the works by Spencer or Pielke Sr. or Dr Christie, never mind the audits by Steve M.
Being a sceptic does not mean one denies that there was a rise in global temperature (a minute one!), nor that the climate changes. I'm getting sick of the need to say this. it reminds me of political debates, where one has to start with 'I'm not a racist' before one's allowed some mild criticism.
The point is that they are not trends as such, until you have corrected for all of the cyclical data.
For example, if you have a long e.g. 60 year cyclical pattern, then any 20 year period would look like a trend.
Only by examining your residuals after fitting the trend would you discover that your trend does not account for the variability. If you omit this step, then you are effectively explaining your variability as being dependent on some linear indeopendent variable which may not be the case.
Climate scientists prefer not to consider this possibility - so they didn't bother to check whether their residuals were autocorrelated or not. If they ARE autocorrelated, then the statistical tests to be applied for testing significance of the trend are different. Significance may disappear - in which case the measured trend is regarded as being insignificant and the underlying independent variable (CO2 in this case) meaningless as an explanatory variable.
I don't intend to develop this argument any further - I leave that to professional statisticians, as climate scientists may have been better advised to have done as well.
Linear trends can be useful, as long as you're clear about what you're using them for. It's a useful shorthand for describing the movement of data in a particular sense. You have to be very specific however about describing them. (Lucia at the Blackboard is very good at this). Very few climate scientists (none that I've read) I think claim that the temperature trend over time is a linear one, but since we have no other function that (easily) describes the complex data it is useful.
However, in a set of complex data you can 'cherry pick' your trends simply by picking start and end periods, or changing the time scale. As long as you're specific about this, and describe the whole data set or how you selected a subset, then there is no issue. If on the other hand you take a particular view, and choose your parameters to emphasise only that view, and yell 'we're all doomed' as a result, then that is dishonest.
Personally I find the various moving average curves more useful as they describe the dynamic behaviour of the data more effectively. A good way to use short term linear trends is to look at the change of slope with adjacent data sets (e.g. take decadal sets and look at the simple linear trend in each), this will give you the rate of change of trend, which tells you quite a lot!
Ian UK
'The Drum' is an Antipodean slang term that requiresit's inclusion in a supporting phrase to even begin to make sense, such as
"Gimme the drum"
which is loosely translated as "Gimme the goss" or "tell me what's going on."
In that context, it does make sense and is not at all contentious or outrageous. But it's a fairly outdated term, too, which speaks volumes about the ABC's handle on it's own culture.
It's probably worth saying, despite it being obvious, that trends in the data are only showing you what HAS happened, not what is going to happen. That takes theory and hypothesis, and (yes!) modelling. That theory then has to be tested and refined. Studying trends really does not take the argument much further than saying 'this is what happened'. It's the easy bit. [Having said that of course, the data is so ragged that we're not even at the point of agreeing what has happened, so heaven help us in positing what is going to happen].
The BEST project is built on sand.
It's based on the idea that correct statistical analysis of thermometers readings of daily maximum and minimum air temperatures will reveal anything useful about the weather decades in the future.
Getting involved is like talking to a Jehovah's Witness. Hang on - that's the doorbell...
Cumbrian Lad
HADCRUT annual means:
http://woodfortrees.org/plot/hadcrut3vgl/mean:12
Jack Hughes
I thought the idea behind BEST was a reanalysis of past temperature data in order to determine whether systematic error or bias was present in other key reconstructions, eg GISTEMP, HADCRUT etc?
Also, in your view, is it scientific fraud? As Mac has now claimed so often that I long ago lost count?
"It's based on the idea that correct statistical analysis of thermometers readings of daily maximum and minimum air temperatures will reveal anything useful about the weather decades in the future."
It doesn't even say much about the current weather. Take the temperature every minute for 24 hours and then integrate the area under the curve. Do this every day and graph those results. Now you have some data that means something.
BBD, that is a nice example of a graph that is not well labelled. Now I'm quite a regular user of Wood for Trees, it is a really good tool, but if you take one of its graphs out of the blue it's not quite clear what it's showing. Your short descriptor could imply it's a graph of global mean temperatures, and it's not clear from the graph linked to that it is in fact mean anomalies. (This information is there in the drop downs, but you have to know what it means and where to look for it). A better example of how it should be done is at http://www.metoffice.gov.uk/hadobs/hadcet/ where it's clear what is being plotted (I know it's a different data set to the one you reference, it's the labelling I'm calling attention to not the data).
Clarity in all things!
Cumbrian Lad
I said it was HADCRUT annual means.
HADCRUT is a gridded anomaly dataset - from your obvious familiarity with the temperature records I assumed that you knew this, which indeed appears to be the case.
There was absolutely no intent to mislead.
I didn't provide the graph for you to complain about the supposed lack of labelling. I put it up because you mentioned the dangers of over-reliance on linear trends.
What do you think this curve tells us about GATA over the last century, and the last 50 years?
Sam Hall
BEST is a climatological reconstruction.
Climate is not weather.
Cumbrian Lad
The linked graph shows the 5 year running mean for the full HADCRUT3 (variance adjusted global mean) temperature series:
http://woodfortrees.org/plot/hadcrut3vgl/mean:60
How would you characterise the change in GATA over the last century and the last fifty years?
Cumbrian lad, " It's probably worth saying, despite it being obvious, that trends in the data are only showing you what HAS happened, not what is going to happen."
True in an ideal world.
But, we Aussies are not altogether stupid, having achieved one of the best standards of living on the globe. Being not stupid means checking assertions and above all, checking data. When you mine down to the very fundamental temperature data, you find to your dismay that the version on which YOU depend has some very dubious adjustments. Therefore, a prudent person would not say anything about trends, before confirming that they are based on accurate figures. They are not.
Upon which trend do you rely in this graph? Each line was presented on the Net or by BOM in official products that were simply downloaded at the time and presented here. No tricks at this end. The tricks are with the original data, which I have studied for some years for Darwin.
http://www.geoffstuff.com/spaghetti%20Darwin.jpg
It's amazing how propaganda can still suck in the populace.
Geoff Sherrington
I wonder if you mean me, rather than Cumbrian Lad?
Either way, I wonder what your views are on the trends in tropospheric temperature derived from two independent analyses of the satellite data by RSS and UAH.
I need hardly remind everyone that Roy Spencer is responsible for the UAH analysis:
http://woodfortrees.org/plot/rss/plot/rss/trend/plot/uah/plot/uah/trend
I'm still trying to figure out why so many people (from all sides of the debate) are attacking Dr Muller. I would agree that it would have better for his team to have completed their analysis, and published a complete paper, with data and code, before the Congressional testimony. But the timing was not his choice.
To question the conclusions of the temperature record, given the documented poor location of many of the sensors, is only common sense. And the homogenization algorithms also are surprising in their effects. So I for one am very grateful that an independent team with significant statistical expertise, are reviewing them. It may well be that they come to the conclusion that even the poorly-sited thermometers can (in large numbers, anyway) produce a relatively decent estimate of trend. It might not be the idea way to measure, but one can't go back in time and re-position the sensors. It has to be a good thing to reduce the "error bars."
Let's let them complete their analysis, and then look at the methodology and not the results. In fact, I'd like them to publish a paper with only the methodology first, in the hope that the criticism will not be based on the "bottom-line" answer -- but that seems unlikely, and in any case, it wouldn't be more than a day before someone runs the algorithm and publishes numbers anyway.
And also, the light shed on the poor siting by Mr Watts has already caused NOAA to improve some sites, as Watts has reported -- I wonder if there's actually a progamme afoot to make a general improvement to the siting quality.
HaroldW
Hear hear!
Wish you'd showed up earlier on this thread.
Good post Harold.
Your way - publish method, defend method, adjust method, get agreement that method is correct, then publish the results it gives - is a good way forward.
I'd also expect it to show a general warming, since there is probably enough good data in the existing flawed studies to support this, even if the location, the rate, and the significance of any global average remain unconvincing.
The trouble is that when such a study is published, ecofascists will try to argue it proves their entire case and so let's switch the lights out right now.
It leaves untested the questions of what's caused it, what it will be in the future, what could be done about it, and whether it's worth it.
The latter stuf is why I am a sceptic. Because there are no convincing answers to any of these, ecofascists have adopted a strategy of claiming that anyone who questions the solution must also dispute the temperature record. As a matter of fact I do, but not if it's done properly and it is incidental, not central, to the important questions.
I tend to agree with HaroldW. Let's let Muller get on with things - he was dealt a difficult hand by being asked to testify so soon after getting started (an offer he presumably could not refuse). Though I could believe anything in climate science, I would have been surprised had his reconstruction been very different from Hadcrut gisstemp etc. But on the correct vertical scale all those curves are pretty flat.
And the curve tells us nothing about why it has warmed, nor whether it will warm further. For that, we need a BEST GCM. With fully open book physics and uncertainties. That would be fun.
Being asked to testify, one can always refuse.
I don't trust Muller after I learned about his carbon-related nonsense.
He is a CAGWist, any questions about that?
Alexander K (11:14 am) got to comment on my post because it was removed, I assume for lack of relevance (or possibly suspected racist overtones). It was a bit tongue in cheek, I admit, following a recent scandal when an otherwise upstanding citizen was pilloried for referring to "jungle drums" in the context of spreading rumours.
OOPS! Should have said BEFORE, not BECAUSE in the first line. I'll get mi coat...
Shub
Perhaps Muller should have refused to testify (although I do not know if he could then have been censured or compelled - others may be able to clarify this). But he did. It doesn't make him the devil. He was asked about preliminary results by his own government and he answered its questions.
He is a CAGWist, any questions about that?
I don't care if he's the King of Old Siam so long as he makes BESTs code and data publicly available.
I don't care if he's the King of Old Siam so long as he makes BEST's code and data publicly available.
We are living in an interglacial period, the Holocene..
Petit et al(1999) pointed out that of the last five interglacials, four of them were on average about 2 degrees C warmer than the present.
And Sime et al(2009) "maximum interglacial temperatures over the past 340,000 years were between 6.0°C and 10.0°C above present-day values".
We are talking about comparing a data base of highly sophisticated recorders over a short time frame..of about 120 ? years.
And some are screaming..this is "unprecedented" !!
Too state that..you must be able to prove that previous interglacial warming periods..warmed faster etc than now..otherwise..you have no rational argument..
What temperature proxy, has the ability to show "precise" measurements of temperatures in time periods of 120 years..or smaller increments.?
Regarding the drum/abc science reporting..
Awful is not even a good description..the "journalists" never have the time to research anything..
Just cut and paste and march in tune..
The reaction of some climate skeptics to the preliminary results of the BEST team has been near hysterical. Years of debate and bickering with the CAGW dogmasphere have left them feeling bitter, and they have copied the attitude of 'if you are not one of us, then you are one of them'. I don't think these skeptics are much interested in the transparency the Berkeley team seek to introduce to science in general and climate science in particular. No, sir, they are seeking the correct result for their side.
Willis Eschenbach, Anthony Watts and Roger Pielke Sr have all lessened their credibility in my estimation with the way they reacted to Richard Muller's testimony to the Congress. They ought to have seen it for what it is; a preliminary finding on a tiny fraction of the data that agrees generally with the trends identified by others. Instead, these 'skeptics' run to their blog and cry out "like woman who's just learned her husband was cheating on her" (Yes, Muller, thanks for that, too).
BBD - Muller was asked to testify.
I don't see any reason why he should not have testified.
But he can be criticised for the way he testified and for what he said.
He could have testified that his work was in the very early stages and that the necessary checks and balances had not yet been completed and that any conclusions that could be drawn from the work to date could seriously mislead the Congress.
However, he has said what he has said to Congress.
He has also promised that when complete, the work and the operating programs that generate it will be available for all to examine, criticise and improve (the last my addition to this arguement).
Now I am sceptic of BEST, but let us approach it with an open, but a careful and enquiring mind.
Not looking for the smallest fault to destroy the lot, nor swallowing it either without question, hook, line and sinker.
AusieDan
We appear to be in agreement, more or less.
Have you read this thread?
The question Dan, is, why is Muller not approaching the 'worlds anomaly' with a cautious, enquiring and open mind himself?
Therefore he does not deserve any such courtesy (from 'us', that is). Now, of course, he deserves a basic minimum courtesy, but that's about it. In this aspect, it does appear quite puzzling that Watts and Morano 'attacked' him, but amongst these, Watts' case is the weakest. He is haggling with Muller on procedural grounds (did not ask permission before reaching conclusions about data etc) and he has committed the same scientific strategic blunder that Muller has. I quote:
Muller did the same thing with AGW. "I don't trust the guys who are doing reconstructions (CRU) but it will be really bad if the globe is warming, which it is".
The "global anomaly" is a meaningless metric, a smokescreen in effect. It is perhaps a crude, and yet abstruse thermodynamic notion; how is it a climatic quantity?. The climate science community is absolutely hesitant and shifty when comes to answering the question: "Will the global temperature change cause the climate to change, or does 'climate change' cause the global temperature to change?". Sometimes they want to have it this way, and sometimes the other way. The US surface record is 2% of the world's surface record. How much time can one spend over it?
Mosher, Nick Barnes, Zeke, Chad, Tom, Dick and Harry are all doing temperature reconstructions. Big deal. Muller wants to another one. Big deal. Good publicity for calculating a useless metric though.
Shub
I think you are mistaken and that this is a poor 'sceptical' argument which detracts from the credibility of scientific scepticism about the potential effects of AGW on future climate. But who cares what I think.
Dr Roy Spencer thinks you are mistaken too. Does that get your attention?
Read the rest here:
http://www.drroyspencer.com/2010/05/in-defense-of-the-globally-averaged-temperature/
There's been an astonishing amount of bollocks on this thread re Muller and BEST. I'm still shaking my head in dismay.
Muller says:
"R.M.: [Their goal is] to generate long continuous methods. … If there was a change, [like] a station moved, they would adjust the data to try to eliminate that. [But] it makes me very uncomfortable when you adjust the data. … [So] we just cut the data at that point [and create two shorter records]. It means we wind up in our analysis with [not very many] continuous records."
We must keep our eye on the pea. Muller is doing two things:
1. He is altering data and attempting to hide the fact. When a station is moved and that move causes some noticeable change in the temperatures reported, Muller is selecting some point and snipping the record at that point. Then he is treating the two snipped measurement series as two stations. The logic of this is bizarre. He starts with readings A through D. There is some noticeable change between A and D, at point B, that is apparently caused by a station move. He snips at point B and creates two records, A through B and C through D. The two series AB and CD are treated as two stations. If they prove to have the same trend then it is concluded that the station move had no effect. You do see that the pea is cleverly hidden in his hand, right? What about the trend of AD? He cleverly did away with that! What if it differs from AB and CD? Doesn't matter because the pea is in his hand. This surely deserves the name "novel statistical technique." We must hear from McIntyre and McKitrick on this.
2. His brilliant new research does nothing but extract trends from data. It does not attempt to address the quality of the raw data. He makes the simplifying assumption that all complaints about stations that were moved, stations that are poorly sited, and stations that have come under the influence of UHI are complaints to the effect that these stations gave reports that are artificially high. He took care of that with his novel statistical technique; the pea is hidden in the hand. However, that was never the complaint. Rather, the complaint was that the condition of the stations clearly showed that their management was incompetent and all records from them should be rejected. Poor siting and station moves of the kind recorded by Watts are matters of gross negligence on the part of management. The other matter is UHI which is systemic. But Muller hasn't even addressed the question.
After this performance by Muller, one can only conclude that he is a shill for the Warmista. His reasoning is downright childish. More important, like all Warmista climate scientists, he has no instinct for the explanation or for criticism. His only instinct is to get the numbers out. Technicians could do just as well. We wanted scientists whose instincts are for explanations, physical hypotheses, criticism and candid discussion. Muller appearance before the US Congress and this interview are shameful.
Roy Spencer is quoted:
"The globally averaged surface temperature is directly connected to the globally averaged tropospheric temperature through convective overturning of the atmosphere. This is about 80% of the mass of the atmosphere. You cannot warm or cool the surface temperature without most of the atmosphere following suit."
Then why does Pielke, Sr., claim that we should use ocean temperatures only? If the "global average temperature" is such a clear concept, would someone please tell Pielke?
Theo Goodwin -
1. When a station moves, or there is a change in the type of thermometers used, or the method of recording changes, it is quite plausible that the measurements before the change are not commensurate with those afterward. Computing a trend over two (or more!) disparate types of measurement can only mix apples and oranges. The current method (well, the most widely used I believe) is to make a standard adjustment - e.g. if a move alters the elevation of the station, the temperatures are adjusted by a fixed number of K per km of elevation change. E.g. Peterson used 5.3 K/km. Presumably this is meant to be an average lapse rate, but at a given location, the local average lapse rate might be significantly different. One could make similar observations about other effects such as equipment changes. Breaking a station's record into two smaller ones allows the software to determine a best-fit for the offset due to station move. I think that it is generally a superior procedure to let the data speak for itself, and allow the algorithm to determine what effect the change had, rather than applying a uniform adjustment. And certainly better than ignoring the changes! One has to realize that station moves can be over many miles, from urban to rural, etc; the "before" and "after" sites may be no more related to each other than to the neighbouring station in the next city. [As an aside, the caveat I would offer to letting the algorithm determine adjustments, is that in unusual circumstances it might select a physically unlikely adjustment. So I would hope that differences between these broken-apart records would be examined, and unusual adjustment values examined by hand, comparing them to e.g. the "standard" adjustments. I never trust any computer algorithm too much! -- I've written too much software myself.]
2. I understand that BEST have undertaken to examine the data sets for biases such as UHI, and they have developed (or are developing) algorithms to assess these. Watts spoke highly of the promise of these methods, and I have no reason to doubt that the enterprise is being followed without a prejudgment of the results.
HaroldW
Thanks for stepping up. You are doing what Muller should have done before going public with results. You are showing interest in matters that Muller has not addressed at all. You might prove to be the scientist that Muller seems not to be. However, I do note that you did not address my criticisms. Why am I not surprised? However sophisticated your methods, you too are changing the data. I hope your comment starts some candid discussion among people on this blog, Best, the people at WUWT, and just everyone. I do not believe that it will. I believe that BEST is yet another manifestation of our ruling elite. I predict that BEST will speak down to us. So far, I am correct in this prediction. However, I will be overjoyed to find that I am wrong.
At this time, my concern is with Muller. I want Muller to address my criticisms. Also, I want him to explain exactly what he means when he says, or permits others to suggest, that he was in some way compelled to address Congress. Did Congress take his wife hostage? You see, sir, Muller's suggestion that he had to testify and that he had to say what he did are nothing but fodder for guffaws, at least here in the US. When the guffaws die down, everyone volunteers "Childish!" Maybe there are cultural differences between us, but the particulars of your defense of Muller mean that I would not trust you to mow my lawn.
Theo Goodwin
Please read the whole Spencer article before commenting on it further.
Pielke Snr (and others) point to OHC as a better measure of energy imbalance (radiative imbalance at TOA) within the climate system because it can store so very much energy compared to the atmosphere.
The problem is that our estimate for OHC may be broken. Please see the 'It's the ocean, stupid!' thread below this one.
Nor does it 'mean' - or even remotely imply - that attempts to calculate and monitor the global average temperature anomaly are inherently flawed, unworkable or dishonest.
This is a disgraceful tone to take with a commenter who has taken time to try and address questions you raise on this thread in obvious good faith. You owe HaroldW an apology.
BBD and HaroldW
I apologize. Brits are ever more civil than Yanks. However, on the matter of Muller, from the bottom of my heart (a Yank thing, I know) Muller strikes me as behaving in exactly the same way as Brittany Spears. He is doing nothing but drawing attention to himself for the purpose of drawing more attention to himself.