Wednesday
Mar172010
by 
Bishop Hill
Bishop Hill Statistical death-match
Mar 17, 2010 Blogs Climate: Statistics If you have a lot of time on your hands and get excited by tests for unit roots, do have a look at contributions the thread on random walks at Tamino's blog and the longer (much longer!) subsequent thread at Bart Vergheggen's.
VS, for that is his nom-de-blog, reckons that the temperature records are a random walk (or something like one) which means that many standard statistical approaches to measuring trends, including all the ones currently used in climatology, are invalid.
The thread is interesting on another level too, with several days of calm debate at Bart's site rudely interrupted by an invasion of trolls from Tamino's.
Statistics wasn't supposed to be this much fun.
Reader Comments (82)
I am not well read into Steve McIntyre's work, so apologies all around if this has already been done to death.
When I was doing a course in GeoStatistics in the late 90s, It ocurred to me that constructing semi variograms ( http://www.kriging.com/PG1979/Chapter_2/Part1.htm ) for individual climate station's raw data would provide a measure of any trends and periodicity in the data, and give a measure for what period in time data can be extrapolated over, and with what confidence.
Has anyone seen that done?
With respect to the two equillibrium states of the earth's climate : glacial and interglacial, who says that the mechanism is terrestrial, as opposed to orbital or solar variability?
In the long gone days of studying reconstruction of Quaternary environments, the term "Milankovitch cycle" appeared ( http://en.wikipedia.org/wiki/Milankovitch_cycles ).
@Bart Verheggen
"There are more factors influencing climate than only CO2. Aerosol went up a lot in concentration during the 40s-70s (to a large extent due to soaring SO2 emissions)".
Sure and sulfur emissions came soaring down again after the '70s, coinciding with a rapid divergence of ground temperature vs satellite temperature.
BTW Weart is not telling the full story on Arrhenius:
http://www.klimatosoof.nl/klimafiles/images/Untitled-13.png
land vs ocean is even more prominent
http://suprememastertv.com/data/geditor/0903/%5B0%5Dland_and_ocean_temperature_.jpg
Yer Grace
Remarkable! So I shall. Like others I spent a few hours on this and was totally absorbed.
Up with this comment by VS I can put with great agreement and delight -
"WHAT ON EARTH is wrong with this entire clique?
How come NOBODY can discuss NORMALLY with somebody who they disagree with? "
:-)
@Ianl8888, You touched on one of the most important points there, namely READ THE COMMENT BY WHBABCOCK!
From Beenstock & Reingewertz, temperature and co2 are, according to the cointegration tests, unlikely to be related by a simple linear relationship. i.e. as postulated by the great and good, that an increase in co2 leads to a proportional or related increase in temperature.
Instead, the tests suggest that temperature changes in step with the rate of change of co2. So when the rate of increase of co2 is fast, temp increases. If the rate of co2 increase is constant, temp stays constant - no increase.
David Stockwell covers this very well at Niche Modelling.
TG,
Much like saying that in the long run we're all dead. That doesn't imply that we should hurl ourselves into the path of an oncoming train.
Strawman - nobody said we are about to hit a tipping point. The point is that ramping up the greenhouse effect with, according to your view, no real idea of what may result is not exactly reassuring.
Your version of VS's argument is like saying that it is safe to swing a sledgehammer around a house, with no idea which are the supporting walls. All while claiming that a dubious statistical analysis of roof heights around the world means that if you hit something important, then some unknown 'feedback' may cause the roof to levitate above your head.
@Chuckles
No that's the other way round, CO2 absorption speed follows temperature,
http://members.casema.nl/errenwijlens/co2/co2lt_2007.gif
Correlation is not causation.
http://www.youtube.com/watch?v=GPg7v6iUXcw
Philip Bratby,
For example: http://www.agu.org/pubs/crossref/2009/2009JD012105.shtml
Then there is also sea level rise from thermal expansion, the response of much of the biosphere, together with a large amount of ice which seems to have taken a permanent walk and left no forwarding address.
And the oceans are warming. As for El Nino, this is not the first nor even the largest El Nino in recent memory. The oceans still accumulated heat after 1998 so why won't they this time?
ENSO just shuffles heat from one place to another, and meanwhile more heat arrives. Arguing that this means cooling is like arguing that you lose money by moving it from one pocket to another. Or indeed from one interest bearing account with regular deposits to another. There is still more coming in than going out.
@Hans,
That's a most interesting graph Hans, but I think you are talking about physics, and they are talking about statistical tests?
Have a look at David Stockwells posts on co-integration, it describes it extremely well.
Breaking Story: Whilst everyone is beating themselves around their heads with statistics is seems real world data from the scientific past is beginning to assert itself once more.
http://hidethedecline.eu/pages/posts/decline-temperature-decline-1940-78-the-cold-data-war-170.php
Could the 1930s and 1940s have been warmer than the past couple of decades?
.... and who adusted the temperature data from the 1950s, 60s and 70s to make this period warmer than it was?
It would seem not only was the MWP a pesky nuisance, so too was the warm 1930s and 40s and the much colder 1950s, 60s and 70s.
Not sure if this has been posted already but it is a link (from someone called Josh) on the thread in question...
Science News on statistics - Odd are its wrong
And is a good read
Josh -- you had me until "the math rooted in the same principles that guarantee profits for Las Vegas casinos" 2nd par. Getting in early for April Foll were you?
It's possible that Marco in that thread is in fact Tamino.
I too spent some time reading the discussion and found it enlightening. I also started skipping any post of dhogaza's around half way through and it was then faster to read and stayed closer to the main points.
Hans Erren,
SO2 emissions soared up between the 1940s and 1940s but (globally) didn't soar down afterwards. See eg fig 1 here: http://www.pnl.gov/main/publications/external/technical_reports/PNNL-14537.pdf
Moreover, aerosols cause cooling, so their decrease after they have increased unmasks the greenhouse warming. The warming is *not* attributable to *cooling* aerosols.
Ah the battle of the data
http://members.casema.nl/errenwijlens/co2/stern2005fig3.gif
ref: David I. Stern: Global sulfur emissions from 1850 to 2000, in: Chemosphere, Vol. 58, 2005, S. 163–175, doi:10.1016/j.chemosphere.2004.08.022 online data: http://www.sterndavidi.com/datasite.html
Sulfur emissions in the Netherlands peaked in 1966 and are currently at 1890 level.
According to David stern the sulfur peak occured in the 70's which is also in agreement with german "waldsterben" treeringdata of Wilson and Elling
http://www.eeo.ed.ac.uk/homes/rwilson6/Publications/WilsonandElling2004.pdf
On a hot summerday your airconditioning breaks down and it slowly gets warmer in your house. Where's the heat coming from, from the airconditioning or from the warmth outside?
Wrong analogy, on a hot summerday open the shutters and it rapidly gets warmer in your house. Where's the heat coming from, from the shutters or from the sun outside?
Nice to see Bart here, I enjoyed the thread on his site (mostly) and we need more...what?..here. (Trying not to insult or demean anyone here...) People who are happy to argue an anti-sceptical view? People engaged in climate science?
Or, just people like Bart!
@Bart Verheggen
Thank you for hosting the thread - extremely interesting on a large number of levels
I recommend that you re-read the whbabcock post (March 17). The accusation that the B&R paper '"transgresses" known physics is a straw man
Hans Erren,
Well do you want the real answer or the 'sceptic' answer ("it isn't getting warmer in the house! it's UHI! the house has been this warm before so it couldn't have been caused by you opening the shutters! look, I have your emails!").
Hans - or you could have the alarmist "answer":
"I've already decided on the outcome of any investigation into this and the answer is what I say it is and I'll back it up with my poorly executed and inappropriate statistics and I'll show how little breadth and depth of understanding I have by refusing to tackle and confront obvious shortcomings and inconsistencies in my argument and I'll get all my colleagues to rally round to drown out any points of view we don't like and I'll respond ad tedium to valid points on any blogs I can with armwaving and shouting and I'll keep it up until you all go away and and and and and and ....."
where "and" tends to infinity. Or something like that.
Your Grace has been noticed by the frequenters of Barts climate space. Regrettably he and the ever opinionated Marco do not seem to share the local enthusiasm for your Graces' literary efforts.
I suspect that Marco is, as ever, deeply confused in his comment further down. Montford, Monckton, what's in a name?
http://ourchangingclimate.wordpress.com/2010/03/19/what-do-we-know-about-climate-change-peter-sinclair-climate-crocks/#comment-2006
I stumbled on this site while looking for something else, and even though I don't understand what is meant by random walk, stochastic, deterministic, or unit root, I read the whole thing all the way through. VS, you have a fan club. I am now going to find a mathematician who will tell me what those words mean.
This climate business is better than any book I've ever come across, and it's real time drama and global.
Chuckles,
Eh? Where is the mention of Monckton?
If I am understanding this correctly, an important part of the dissension is about the nature of the temperature dataset. B&R's paper makes claims as to the nature of that set, that it contains a "unit root", and it would seem from the analyses presented on the various blogs that this is correct. The prime "anti" argument appears to be that although the data looks like that, it isn't "really" like that so the analyses based on the appropriate statistics for such datasets is invalid. Thus the apparent independence of CO2 concentration and temperature is an artifact of the statistical treatment. Is this a valid argument ?
If data has all the characteristics necessary to define the presence of a unit root, in what way is the fact that theoretically it shouldn't in some ways (being bounded for example) sufficient to alter the appropriate statistical treatment ? It is interesting that David Stockwell, using different methods altogether, arrives at the same conclusion as B&R.
I also suspect that much of the special pleading about the "true" nature of the temperature data is based on the utterly unfounded belief (wish ?) from those espousing the view that they know how the real dynamical atmosphere works to a sufficient level of detail. Going with that appears to be a belief that the existing models adequate and accurately reflect that understanding.
Certainly however (to paraphrase commenter VS), it is not a good look for a theory when the ONLY data available to evaluate your theory gives results that invalidate that theory when correctly treated. Of course climate science is full of examples used to support the various theories with "proofs" based on the improper and misleading use of statistics.
Thanks to BV for this post. Who would have thought that the existence or not of a unit root in a time series process could have produced such a span of apoplectic fits amongst Tamino's hangers-on.
Thanks also to VS for his exceptionally well argued input. Read it all, but at the very least read his classic put-down of the execrable Josh Halpern (aka Eli Rabett). Who submerged and remained at snorkelling depth for a long period afterwards.
Sorry Frank, didn't see your post there.
Just my terrible phrasing, I was trying to say that the good bishop is hardly given to polemic in his writings, and that Marco was perhaps confusing him with Monckton.
This was an excellent thread - mind you it took me several days to get through it. I have to agree once I started skipping the posters with empty messages and vitriol (pretty much at the same time VS did) then I started moving faster. Alex also contributed greatly. I would also agree with ianl and chuckles that the best summary was whbabcock. I believe VS also enjoyed and agreed with his summary although no other commenter besides Bart's referred to it.
Other stuff - (a bunch of) hand waving by dhogoza concerning B&R trying to overturn modern physics, Willis E. rebutted and dhog did not answer but certainly kept hand waving.
Also VS had some good questions for Eduardo who showed up.
Once more I thoroughly enjoyed it - wish we could have more reasonable discussions like it.
I can't read your website in IE 6.9, I juts ifgured I minght let you knoiw?
great vdeso
I amk havinjg a helluva itke reading your ewbsitfe in Opera 1.3, I jjust thought I would telkl you about it?
great videos