Antarctic temperature plots through the history of the debate:
NASA 2004:
NASA 2007:
Steig et al 2009:
O’Donnell et al 2010:
Some historical perspective:
“If your result needs a statistician then you should design a better experiment”
Ernest Rutherford quotes Baron Rutherford of Nelson. New Zealander born British Chemist who laid the groundwork for the development of nuclear physics by investigating radioactivity. Nobel Prize in 1908. 1871-1937
Personally, besides all the original obtuse PCA statistical sophistry, the buried weather stations, and other issues, now I simply think the issue boils down to: it’s weather, not climate.
See here as to why.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Ernest Rutherford said a lot of stupid things. That was one of them.
Another was: “Anyone who expects a source of power from the transformation of the atom is talking moonshine.”
Agreed (scroll up two pictures)
Juice says “Ernest Rutherford said a lot of stupid things. That was one of them.”
Lord Rutherford, the man who split the atom probably did say stupid things. Who doesn’t? Albert Einstein once said “God does not play dice”, an later had to apologise to Heisenberg.
But Rutherford’s comment “If your result needs a statistician then you should design a better experiment” is not stupid, but profound. Juice, if you want to disagree with the arguments presented here, then address the arguments directly. Don’t waste our time with your ad hominem attacks on a Nobel prize-winner who won that prize in the days when it really meant something.
A better comment would be “If your result needs a statistician and you don’t have a real one, then you should hire somebody to do it right.”
The thing is… it’s easy to learn enough about almost anything to be able to find problems in someone’s experiment – but it’s really hard to be good at something without some solid experience in the field. Most climate scientists might be great at their own subspecialty, but have little to no experience in the things that are currently destroying their research results. Like statistics, or computer programming, or solar astronomy.
Or, as one climate scientist once angrily told me, “insolation is a constant.” What?
Juice says:
February 13, 2011 at 10:53 am
Okay, we know you don’t like Rutherford. Do you have anything of substance to contribute? Do you have the courage to tell us the rest of what you think? Lay it out for us. Be brave.
John Johnston says:
February 13, 2011 at 11:22 am
Juice says “Ernest Rutherford said a lot of stupid things. That was one of them.”
Physicist Rutherford is clearly making the point that a good physical experiment would be even better one if what one concludes from it does not depend on the finer details of a statistical analysis of the results. A very reasonable observation I would have thought.
The use, or rather abuse, of statistics is at the heart of the current problem with science.
The hunt for a signal amidst noise makes sense when looking into phenomena where the signal has intelligence (i.e. radio signals). Cleaning up the noise has a clear objective since you already know what the signal is supposed to be.
When looking into natural phenomena all you can do is find patterns within the noise and then assume these have meaning. I am sure that many of the resolved patterns do have meaning, but too often we see that what is assumed greatly colors the interpretation.
Beyond the fundamental problem of the meaning of a resolved signal lies the problem of cherry picking of parts of the noise to “help” in tuning a signal.
I understand the motive in cleaning up input to remove random noise and bad data, but too often this filtering is done to make a case for reasons not is the data, but in the researcher’s bias. And to make matters worse, when this is pointed out, the researchers either can’t understand the criticism, or do and don’t care.
It is very unfortunate for science that this kind of statistical abuse has been allowed to prosper. It has permeated to the top of academia and most scientific organizations.
I doubt that changing the current paradigm will occur in my lifetime, but there are indications that there is movement toward a recognition of the problem.
If you are going to use ‘statistics’, how good is your data sample? Do you have all of the highs and lows of temperature variations over the last one billion, three billion or 6.5 billion years? If not, how good is your sample, then? What, one millionth of the total data set represents the whole set?.. Crawl back into your cave and turn out the lights. Worthless opinion is like anal sphinctors, eveyone has one.
#1 and #4 say it all, so what happeed at NASA after 2004?
In the referenced WUWT thread “Mike” said:
“Both papers show that most of Antarctica is warming.”
Can one even say that? I would each paper has a statistical model that purports to show warming across the entire Antarctica continent based upon sparse data from around the edge of the continent. The real issue is do the models reflect reality in any way? From our experience of the US we would say that that a handful of stations around the edge of the continental US and a statistical model could not in any way model the entire temperature map of the US.
I think its just numbers with no mechanism for validation.
I may be naive about such things but I am amazed that anyone would ignore the one bit of Antarctica which sticks out, free of sea ice, into the Pacific Ocean during the decade that the PDO peaked, and ending with a big Modoki El Nino. If you stick your arm into a hot oven, your arm will get hot.
Can some explain why Steig is still insisting, along with A. Revkin, that O’Donnell is planning to admit he was wrong on the major points? This is Steig’s Feb 9th post at RC which remains up:
“Ryan O’Donnell made a series of serious of allegations against me at ClimateAudit, in the context of our friendly dispute about whether his new paper in the Journal of Climate supports or ‘refutes’ my own results, published in Nature.
To his credit, Ryan has offered to retract these allegations, now that he is a little better acquainted with the facts.”
I keep harping away at this, and I apologize for that. I guess I’m either crazy or naive. Or both. I mean, how can he leave this up knowing that as of Feb 11, it’s been demonstrated via O’Donnells actual response that his apology is confined to one tangential issue as below..
“Subsequent to my post on Feb 7, 2010 here, Eric Steig informed me by email that he had not seen our Response to his Third Review, as I had previously assumed. I apologize for my misunderstanding on this point, which was, however, incidental to the major concerns expressed in my post. A more detailed response on matters raised in Steig’s most recent RC post and other issues will be forthcoming.”
Evidently, as deep as my cynicism goes, it’s not deep enough. I apologize for hammering away at this, and I don’t have time to read through all the comments to see if it’s been discussed. But such brazen gamesmanship, when it can so easily be show to be such, surpasses understanding. Or my understanding anyway..
Delingpole has a good summary of the Steig affair. He concentrates on the abuse of peer review, not on the statistics.
http://blogs.telegraph.co.uk/news/jamesdelingpole/100075232/realclimategate-hits-the-final-nail-in-the-coffin-of-peer-review/
I would suggest using the following image: http://science.nationalgeographic.com/wallpaper/science/photos/clouds/antarctic-clouds/
Antarctica is very large, very cloudy and very cold. There are few historical temperature records and those are sparse, gappy and noisy. Temperature reconstructions, like those shown above, are speculative, although they may be interesting.
DirkH writes,
Delingpole has a good summary of the Steig affair. He concentrates on the abuse of peer review, not on the statistics.
Delingpole has a strong political spin, sure enough, but shows that he understands nothing about peer review. Steig’s reviews were those of someone who thought O’Donnell’s paper had flaws but also merit; he helped make it good enough to publish.
“Personally, besides all the original obtuse PCA statistical sophistry, the buried weather stations, and other issues, now I simply think the issue boils down to: it’s weather, not climate.” Or to be blunt you dont understand the discussion.
Did you even pass high school math?
REPLY:Did you even read the link? Attack the messenger with juvenile prose, ah, spoken like a true anonymous coward. Troll on brother, but do it somewhere else. – Anthony
Phil writes,
Antarctica is very large, very cloudy and very cold. There are few historical temperature records and those are sparse, gappy and noisy. Temperature reconstructions, like those shown above, are speculative, although they may be interesting.
Sparse data is not the same as no data, or pure speculation. Steig’s and O’Donnell’s papers are about how to use the data we have as effectively as possible.
Gneiss says:
“Steig’s reviews were those of someone who thought O’Donnell’s paper had flaws but also merit; he helped make it good enough to publish.”
I think Steig was pushed into a corner. He had no choice but to give in. O’Donnell’s paper was published depite Steig, not because of him.
See John Brignell’s view of refereeing. It’s only a few paragraphs.
Phil says:
February 13, 2011 at 2:35 pm
And cold. Very, very cold.
Smokey writes,
I think Steig was pushed into a corner. He had no choice but to give in. O’Donnell’s paper was published depite Steig, not because of him.
Steig was not pushed into a corner or forced to give in. Reviewers have no veto power, and they don’t have to agree. Disagreement is in fact very common, and editors are the ones who make the call.
For what looks like the clearest account of how things actually went down, I recommend Neilsen-Gammon. It’s quite different from the emotional versions on this site, and contradicts O’Donnell’s accusation that Steig suggested ridge regression. But it’s not all the same as Steig’s recollection either.
http://blogs.chron.com/climateabyss/
Gneiss says: “Steig’s reviews were those of someone who thought O’Donnell’s paper had flaws but also merit; he helped make it good enough to publish.”
It was already good enough to publish, according to Reviewers B and C. Reviewer A was the only one who required a major rewrite. After all was said and done, Steig 2009 was still toast.
Gneiss says: “Sparse data is not the same as no data, or pure speculation.”
Untrue in this case. In Antarctica, there’s so little coherent data as to make zero difference between sparse and none. Remember the Trenberth epigram re Steig 2009? “It is hard to make data where none exist.”
Gneiss says: “Steig’s and O’Donnell’s papers are about how to use the data we have as effectively as possible.”
Absolutely untrue. O’Donnell’s paper shows that Steig’s statistical methods, such as they were, were seriously flawed. O’Donnell was not trying to create a definitive (or more “effective”) reconstruction with the data. He says: “…The only claim we make is – given the data and regression method used by S09 – that the answer is different when the method is properly employed. Period.”
The most telling graphic is in the O’Donnell (2010) paper. It shows the temperature trends on the peninsula with trends that are not statistically significant “grayed” out. What this shows is that there is pretty much no significant anything except warming on the peninsula.
Gneiss says: “Delingpole…shows that he understands nothing about peer review. ”
Another untruth. You evidently didn’t follow the link, Gneiss:
http://blogs.telegraph.co.uk/news/jamesdelingpole/100075232/realclimategate-hits-the-final-nail-in-the-coffin-of-peer-review/
Delingpole obviously understands all the critical points of peer review well enough, just as I don’t have to study horse anatomy to know a dead horse when I see one.
An educated guess may be educated, but it is still a guess. For background see my previous comment:
First, Schneider & Steig 2004 state (pg 5):
However, this ~0.5 K reference may be conservative as the actual quote from Comiso 2000 is:
However, Town 2007 has different estimates for the South Pole (which may be taken as indicative of the East Antarctica plateau – the largest area of Antarctica) (pg 556):
Second, please look at slide 27 here. This refers to figure 5 of Comiso & Stock 2001. They show that the mean monthly cloud fraction anomalies from 1982 to 1999 are decreasing by
Third, did Steig 2009 in-fill the cloudy pixels using the clear-sky pixels by taking into account the cold bias in the clear-sky pixels? They don’t mention that they did in the text and, since they didn’t disclose their satellite data, it isn’t clear what they did. Furthermore, Town 2007 seems to say that clouds significantly warm the surface in Antarctica. If the cloud fraction is decreasing, then wouldn’t the “warm” fraction would also be decreasing? Combined, wouldn’t failing to take into account one or both of these create a warm bias in the trend?
Fourth, if you try to take these apparently non-trivial issues into account when doing the hind-cast of the matrix for the pre-satellite era, what assumption do you use for the trend in the cloud fraction? Given the decreasing trend in the satellite era, would it be reasonable to cast that same trend backwards for the pre-satellite era? Would it be reasonable to assume the pre-satellite era cloud fraction was constant? If so, on what basis? The surface data in Antarctica seems to show an early warming, followed by a recent cooling. That is, the surface data trend would seem to show an inflection point. Wouldn’t it be reasonable to then assume that the cloud fraction trend for the last 50 years also had an inflection point? If so, when? If not, why not? The AVHRR-derived matrix is back-cast by anchoring it to the surface station records of the pre-satellite era. To what could you anchor a back-cast of the cloud fraction trend? Is there anything or wouldn’t it be mostly speculative?
I submit all of the above as support for stating that the images in this post are speculative (not purely speculative, but still speculative), but without claiming perfection. My logic may be faulty, so I have tried to explain it in case I have misunderstood something.
References:
(Comiso 2000) Comiso, J. C. Variability and trends in Antarctic surface temperatures from in situ and satellite infrared measurements. J. Clim. 13, 1674–1696 (2000).
(Schneider and Steig 2004) Schneider, D. P., Steig, E. J. & Comiso, J. Recent climate variability in Antarctica from satellite-derived temperature data. J. Clim. 17, 1569–1583 (2004).
(Comiso and Stock 2001) Comiso, J. C and L. V. Stock, 2001: Studies of Antarctic cloud cover variability from 1982 through 1999. Proc. of the Int. Geosci. and Remote Sensing Symposium, vol. 4, 1782-1785.
(Town 2007) Town, M. S., Walden, V. P., Warren, S. G., Cloud Cover over the South Pole from Visual Observations, Satellite Retrievals, and Surface-Based Infrared Radiation Measurements, J. Clim. 20, 544-559 (2007)
P.S. Steig 2009 claims that there is a net long-term warming trend in Antarctica. Comiso and Stock 2001 show a long-term (1982-1999) steady decreasing trend in cloud fraction. Given the warming of the surface due to cloud cover explained in Town 2007, this should mean a long-term cooling trend in Antarctica. Which is it: warming or cooling? And which can be more accurately measured by satellite: Cloud Fraction or Surface Temperatures? I submit this as more support for these images as speculation, although not pure speculation.