# Roy Spencer hides the increase

Possum Comitatus has noticed a very interesting change in Roy Spencer’s presentation of his satelite temperature data.

This is the November version:

And this is the December version:

Spot the difference!

Update: Gavin reminds me that in April Spencer was using a ridiculous degree 4 fit to the data:

If he’d stuck with that, the current graph would look like this:

The polynomial is still decreasing at the end, but the divergence from the data is striking.

1. #1 jakerman
January 16, 2010

cohnite wries:

>*the 3rd poly for instance is useful if you want to demonstrate that there are 2 fluctuations, valleys or hill, in the data.*

Cohnite that be more accurate if if read:

>3rd order polynomials for instance are useful if you want to assume and force *2 fluctuations, valleys or hill, in the data*.

How easliy you flit from [trend analysis](http://scienceblogs.com/deltoid/2009/12/russian_analysis_confirms_20th.php#comment-2180096) to imposing unjustified order of polynomicals to fit your own agenda. As you admit for yourself (but you are in no position to suggest that this is the same for others):

>*People tend to use what ever statistical tool will best manifest whatever point they are seeking to make*

I reiterate, this may be an admission of your practice (and what Lawyers representing an interest do), but its not what truth seeking scientist do. (But it gives us some insight into the processes of your concoction of a “break”).

A truth seeker on the other hand would ask what justifies a third order polynomial as the best representation of UAH temperature?

Even Roy Spencer [seems too embarrassed](http://www.drroyspencer.com/2010/01/is-spencer-hiding-the-increase-we-report-you-decide/) to keep pushing unjustified high order polynomial trends.

Here is a challenge for you cohnite. Provide the formula of your third order poly and use it to forecast temperature anomalies for the next 30 years. Show us the plot you produce.

2. #2 carrot eater
January 16, 2010

So el gordo isn’t saying the cold snap is anything more than regional, nor that it means anything for climate, after all. He’s just pontificating over what effect it might have politically.

Yawn.

3. #3 P. Lewis
January 16, 2010

The CET data go back 350 years.

The Jan average CET is 3.24°C.

The Feb average CET is 3.86°C.

Of the 350 years’ data:

132 Febs were colder than the corresponding Jans, which equates to ~38% Febs colder than the corresponding Jans.

11 (~3%) Febs were the same average temperature as the corresponding Jans.

Which makes 59% of Jan CETs colder than the corresponding Febs.

One doesn’t need a supercomputer. OpenOffice’s Calc or MS Excel will do nicely (or pencil and paper if you must and manually counting), opening/importing the data in CSV format and using =IF(FebJan;1;0) and summing the column totals gives a very quick answer.

4. #4 P. Lewis
January 16, 2010

Hey! What happened to my HTML codes between preview and post?

“=IF(Feb<Jan;1;0), =IF(Feb=Jan;1;0), =IF(Feb>Jan;1;0)”

Ah! The HTML entities transform to the literal characters and then when you post you’re buggered.

5. #5 cohenite
January 16, 2010

Your comment about forcing fluctuations, hills and valleys is not appropriate; any statistical analysis should have a comparative verification of best fit; the use of a Chow test for instance to test for breaks should have its results compared with other methods; this was done at Table 1:

http://arxiv.org/PS_cache/arxiv/pdf/0907/0907.1650v3.pdf

In respect of UAH and polynomials would a 4th order poly, as Spencer has used, or a linear regression best represent the trend in the data; an R2 test should tell you that; why don’t you do one?

6. #6 TrueSceptic
January 16, 2010

103 P.Lewis,

Thanks but I was waiting for el gordo to respond.

You’ll find some interesting figures, e.g., the last monthly average below zero was Feb 86, before that Jan 79, then Jan-Feb 63. I remember all of those 😉

I’ll look tomorrow, but I *think* that the Jan/Feb averages have been closer over the last 50, 30, 20 years.

7. #7 carrot eater
January 16, 2010

Why are we comparing Febs to Jans? You compare Febs to other Febs.

And here we have somebody who actually thinks fitting higher order polynomials through such data is meaningful or appropriate? Wow.

8. #8 P. Lewis
January 16, 2010

Why are we comparing Febs to Jans? You compare Febs to other Febs.

Because the statement from el gordo was:

TS: I don’t have the latest super computer, but the agricultural records of Britain are a better guide as to what is going on. It is traditionally colder in February after a buildup of snow and ice from the previous months.

That’s why.

Of course, el gordo may have been referring to November and the following Feb, or December and the following Feb, or…, but I thought Jan was a good place to start to verify his claim (wrt CET that is).

9. #9 el gordo
January 16, 2010

Traditionally colder in February if you go back over the past 2000 years. This is the crux of the matter, they need to build a model that takes out CO2 and puts in anecdotal evidence from historical observation.

CO2 is a plant food and as a vegan I will defend my food source.

10. #10 jakerman
January 16, 2010

cohnite writes:

That directly contradicts [your claim](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2207636) of a 3rd order polynomial trend being the best representation of UAH temperature.

Your social science chow test has not been justified for this application with cycles overlaying a physically explained trend.
And your claim of a break was smacked down by a paper you [linked to earlier](http://scienceblogs.com/deltoid/2009/12/russian_analysis_confirms_20th.php#comment-2180196).

[Your prediciton](http://landshape.org/enm/wp-content/uploads/2009/07/article-003.png), rather than being justified by the data, instead fits [your admission](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208524) of employing *what ever statistical tool will best manifest whatever point [you] are seeking to make*.

11. #11 jakerman
January 17, 2010

>*In respect of UAH and polynomials would a 4th order poly, as Spencer has used…

When forced to justify himself even Spenser uses a linear trend. As I just stated, it seems even Spenser is [too embarrassed](http://www.drroyspencer.com/2010/01/is-spencer-hiding-the-increase-we-report-you-decide/) to continue with his laughable polynomials. You should have taken that hint. Tim (above) shows how bad Spensor’s 4th order poly compares with reality now.

Cohnite continues:

>*[would] a linear regression best represent the trend in the data; an R2 test should tell you that; why don’t you do one?*

The ratio of CO2 forcing compared to [exogenous](http://tamino.wordpress.com/2009/12/31/exogenous-factors/) and cyclic forcing mean that CO2 only dominates the trend over 20-30 year time intervals. UAH ain’t old enough, but fortunately we other longer records. And you your self have argued that UAH is well correlated with GISS and HadCRUT.

Currently the [30 year running mean](http://www.woodfortrees.org/plot/gistemp/plot/gistemp/mean:360/plot/gistemp/last:360/trend) is well correlated with OLS.

12. #12 Bernard J.
January 17, 2010

According to [cohenite](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208267):

People tend to use what ever statistical tool will best manifest whatever point they are seeking to make; the 3rd poly for instance is useful if you want to demonstrate that there are 2 fluctuations, valleys or hill, in the data.

Let’s take the first part of this sentence first, that “[p]eople tend to use what ever statistical tool will best manifest whatever point they are seeking to make”. This might apply to someone who is attempting post hoc to extract from the data an ‘implication’ that emphasises an a priori-held view, but it does not describe the process of science.

In science one forms an hypothesis, designs an experiment and establishes a priori the analytical/statistical procedures required to process the results, and then conducts the experiment/data collection. Important in this method is that the analytical procedures are selected prior to acquiring the data: this removes experimenter bais, and ensures that the methods used are those most objectively appropriate to the nature of the experiment, and not to the outcome that might be implicitly and subjectively desired by the experimenter.

It’s one of the reasons why there is such a beast as a ‘[null hypothesis](http://en.wikipedia.org/wiki/Null_hypothesis)’, and why scientists employ the many statistical procedures that are ‘[two tailed](http://en.wikipedia.org/wiki/Two-tailed_test)’ – there is often no a priori reason to expect a particular outcome.

You need some serious education about statistics cohenite, and about statistical selection, and also about proper scientific method. Your claim above demonstrates that you are unfamiliar with all three of these essential prerequisites. Your approach is a form of data massaging that no scientist or statistician worth his salt would use.

Now, about the second part of your statement, that “the 3rd poly [sic] for instance is useful if you want to demonstrate that there are 2 fluctuations, valleys or hill [sic], in the data”.

It seems that you are also unfamiliar with various properties of third order polynomials.

A third order polynomial has, at most, one [point of inflection](http://en.wikipedia.org/wiki/Inflection_point). Relevant to your claim is the fact that this point of inflection need not be bounded by a positive and a negative radius: both sides of the curve about a point of inflection may be monotonically increasing (or decreasing). In other words, it is possible for a third order polynomial to have no non-positive (or conversely, no non-negative) second derivatives.

If you do not understand how this might be so, play with [this applet](http://www.univie.ac.at/future.media/moe/galerie/fun1/fun1.html).

What this means is that a third order polynomial need not describe a trajectory characterised by “valleys and hills”. If you don’t believe me, consider the arch-typical cubic, y = x3. “Valleys and hills”? Riiiight…

Of course, your problems with attempting to use curve fitting to datapoints is worse than just this. By using a third order polynomial you are effectively stating that the cubic that you select applies to all of the data in the range that you are considering, which is a very long mathematical bow to draw. Even if you argue that you are merely ‘smoothing’, a third order polynomial is rapidly going to go from being an overfit of limited data, to discarding signal along with noise in larger sets of data.

There are much more appropriate ways of ‘smoothing’ (which is not the same as ‘curve fitting’, such as a cubic fit is), and these have been covered by [Tamino](http://tamino.wordpress.com/2009/05/11/dangerous-curves/), and even by [Tim Lambert](http://scienceblogs.com/deltoid/2009/01/sixth-degree_polynomial_fits_j.php).

The point also remains that third order polynomials fall down in ‘describing’ the terminal data points in a dataset, as well as they might describe the datapoints in the middle of the dataset, where one it attempting to fit a point of inflection somewhere in the range of the dataset. And of course third order polynomials are very poor at extrapolation beyond dataset ranges, taking into account that one should not (or at least, rarely and with careful qualification) use simple curve fits as extrapolation tools.

If your argument was that you selected a third order ploynomial* in order to capture hills and valleys, then your level of analytical sophistical is asymptotic to zero, and indeed may actually reach it in a number of steps that themselves approach zero.

And for what it’s worth, when third order polynomials do have both negative and positive second derivatives, they demonstrate one ‘valley’ and two hills (where one assumes that the ‘incline’ coming out of the side of the ‘valley’ away from the point of inflection is the slope of a hill: technically it isn’t, as a ‘hill’ should not stretch to infinity).

The bottom line is this, a third order polynomial might have interesting mathematical uses, but they have no place in describing the trend of data such as the global temperature record of the last century. Anyone who employs them is simply demonstrating their statistical ignorance and naïvety.

(* the typo seemed to suit the context so well that I just had to leave it in…)

13. #13 Dale Husband
January 17, 2010

Spencer is attempting some damage control after he was busted.

http://www.drroyspencer.com/2010/01/is-spencer-hiding-the-increase-we-report-you-decide/

January 16th, 2010 by Roy W. Spencer, Ph. D.
One of the great things about the internet is people can post anything they want, no matter how stupid, and lots of people who are incapable of critical thought will simply accept it.

I’m getting emails from people who have read blog postings accusing me of “hiding the increase” in global temperatures when I posted our most recent (Dec. 2009) global temperature update. In addition to the usual monthly temperature anomalies on the graph, for many months I have also been plotting a smoothed version, with a running 13 month average. The purpose of such smoothing is to better reveal longer-term variations, which is how “global warming” is manifested.

But on the latest update, I switched from 13 months to a running 25 month average instead. It is this last change which has led to accusations that I am hiding the increase in global temperatures. Well, here’s a plot with both running averages in addition to the monthly data. I’ll let you decide whether I have been hiding anything:

Note how the new 25-month smoother minimizes the warm 1998 temperature spike, which is the main reason why I switched to the longer averaging time. If anything, this ‘hides the decline’ since 1998…something I feared I would be accused of for sure after I posted the December update.

But just the opposite has happened, with accusations I have hidden the increase. Go figure.

Needless to say, I don’t believe him. I think he got caught and is making excuses for what he did after the fact. And note that he allows no comments on his blog entry.

14. #14 Bernard J.
January 17, 2010

Commenting about [cohenite’s use of a third order polynomial](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2207636) to describe the last several decades of global temperature, [jakerman](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208524) noted in a paraphrasing of cohenite:

3rd order polynomials for instance are useful if you want to assume and force 2 fluctuations, valleys or hill, in the data.

[Emphasis mine]

to which [cohenite responded](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208577):

Your comment about forcing fluctuations, hills and valleys is not appropriate; any statistical analysis should have a comparative verification of best fit…

I agree with jakerman’s comment cohenite. Your fitting of a third order polynomial forces the data to take your preconceived notion of the shape of the trajectory that you imagine the data describe: it does not identify accurately what is happening underneath the noise in the system.

As to a “comparative verification of best fit”, and:

In respect of UAH and polynomials would a 4th order poly, as Spencer has used, or a linear regression best represent the trend in the data; an R2 test should tell you that; why don’t you do one?

do you understand why simply obtaining an r-squared value close to 1 is irrelevant, if the curve you select is not appropriate to the phenomenon/a that you are attempting to describe, and especially if you are breaking the ‘law’ of parsimonious description?! By your logic you should be selecting a higher order polynomial that you did, because you’d eventually be able to get an r-squared value greater than 0.99.

I’ll give you an enormous clue – it has something to do with selecting curves that miss real movement in the data and, especially in the case of your fourth-order comment, with curves that are fitted to noise in the data.

One can fit arbitrarily-selected equations to any dataset such that the r-squared value is greater than would be obtained from an equation that might actually reflect a genuine mathematical descriptor of the underlying phenomenon/a: this does not mean that such mined equations are suitable. It does mean that you might simply be finding an equation that incorporates irrelevant noise into your description of underlying trend, and it does not mean that your nifty r-squared = 0.999 equation is more appropriate than a valid smooting technique that nevertheless gives a greater sum of residuals.

Seriously, tell us why r-squared is the be-all and end-all in this matter, and why other matters are irrelevant. If you are not sure, perhaps David Stockwell or your wife might provide the answer for you.

Oh, and:

I’ve [asked you previously](http://scienceblogs.com/deltoid/2009/07/ahh_mclean_youve_done_it_again.php#comment-1819185) without response: I would be most interested to know what exactly what statistical, or indeed any other type, of contribution you specifically made to the ‘structural break’ piece.

15. #15 Stephen Gloor (Ender)
January 17, 2010

Tim – something to add to your long running corrections that will never be made. Do you think Bolty will repost Dr Spencer’s new graph with the upward linear trend on his blog?

I think pigs would be looking at the homeland security watch list when accepting passengers before this would happen.

16. #16 cohenite
January 17, 2010

BJ; as usual you don’t read what I write or don’t write; i did not say R2 was the “be-all and end-all” in any matter; it merely is one means of verification of fit; for instance the R2 of the linear regression of UAH monthly temps is 0.2874; obviously something is lacking, what would you suggest? Jackerman has suggested 30 month running mean which is somewhat larger than Spencer’s.

As for this;

Have a look at what is being described; the 3rd order poly of the differences is almost dead straight; don’t you think that constitutes some sort of validity of using the 3rd polys on the main data? As for the law of parsimony; what could be more spare than a Chow test for breaks? And you guys seem to think that its only me that harps on about the break in temperature; I have posted significant references before [and they are not ‘smacked down’ by the Wu paper]; how about opening your mind.

17. #17 jakerman
January 17, 2010

cohnite writes:

>*As for this;
Have a look at what is being described; the 3rd order poly of the differences is almost dead straight; don’t you think that constitutes some sort of validity of using the 3rd polys on the main data?*

It’s like I say, ‘why use a linear trend when a 3rd order poly can get you same result’, hey cohnite?

And the double bonus is you can claim a cooling trend, then when you get called on the sophistry, just say its “*almost dead straight*”.

18. #18 jakerman
January 17, 2010

cohnite writes:

>*you guys seem to think that its only me that harps on about the break in temperature; I have posted significant references before [and they are not ‘smacked down’ by the Wu paper]*

Your inapplicable social science chow test inferred “break” used in this application is smacked down by [Wu’s trend analysis](http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1986583/figure/F3/).

And if you doubt Wu’s work rejects your projection, how well do you think [your projection](http://landshape.org/enm/wp-content/uploads/2009/07/article-003.png) fits with Wu’s trend analysis? Please reconcile the two.

19. #19 jakerman
January 17, 2010

cohnite writes:

>*use[s] what ever statistical tool will best manifest whatever point [he is] seeking to make*

Then cohnite projected his own practice onto others.

20. #20 cohenite
January 17, 2010

from the Wu et al paper;

“The linear trend is no doubt the poorest one, the intrinsically determined overall adaptive trend is a major improvement over the linear fitting, and the multidecadal trend catches essentially the meaningful variability and change associated with the annual GSTA, showing even greater improvement”

The Chow test is the best way of depicting the multidecadal trend.

21. #21 TrueSceptic
January 17, 2010

108 P.Lewis,

Realistically, it can only be Jan or Feb in the N Hemisphere. Does anyone know of anywhere where it’s consistently some other month? You might get Dec some years, but not too often.

22. #22 TrueSceptic
January 17, 2010

109 el gordo,

We have instrumental records going back to 1659. Whatever happened before that is based on various proxies and therefore less accurate. In any case, who cares about February 1,000 years ago? We’re talking about this February, about which you made a claim. Again, wanna bet?

We are talking about the temperature record itself here. The causes of changes in that are a different issue.

What do you imagine animals are fed, the ones we rear for meat? Do you think that meat-eaters don’t care about the source of their food as much as you?

23. #23 Bernard J.
January 17, 2010

CO2 is a plant food and as a vegan I will defend my food source.

When was the last time that the biosphere suffered from an ecologically functional lack of photosynthetic activity due to insufficient CO2 in the atmosphere? What was (were) the consequence(s)?

24. #24 TrueSceptic
January 17, 2010

106 me,

OK, I’ve just done that. Here are the CET Jan & Feb averages over recent, shorter periods:-

From, Jan, Feb

1950, 4.04, 4.08

1980, 4.47, 4.48

1990, 4.86, 5.03

2000, 5.02, 5.03

2009 is the end year in all cases.

So my recent impression was correct, and el gordo is still wrong.

25. #25 Dan R
January 17, 2010

Did anyone else catch this from Dr. Spencer recently?

“Their science thus enters the realm of faith. Of course, there is always an element of faith in scientific inquiry. Unfortunately, in the arena of climate research the level of faith is unusually high, and I get the impression most researchers are not even aware of its existence.” – Jan 12th 2010.

This is the same Dr. Spencer who states:

“In relation to the basic claims of Christianity, do what I did! Read the Bible. Judge it for itself. Put it to the test. [I am confident that you too will find the Bible not only to be in agreement with proven facts of science](http://theevolutioncrisis.org.uk/testimony2.php), but also to be the book which will lead you to a personal faith in God the creator of all things.”

And:

“[Indeed, I was convinced of the intelligent design arguments based upon the science alone](http://www.tcsdaily.com/article.aspx?id=080805I).”

26. #26 TrueSceptic
January 17, 2010

124 Dan,

Yes, Spencer is delusional. But we knew that.

27. #27 Stu
January 17, 2010

He’s not delusional because he’s a Christian, he’s delusional because he’s accusing ‘climate science’ of mixing science and faith, something he has done a great deal of himself, yet apparently he doesn’t see the irony.

I’m a Christian who is heavily interested in Science (well I am a scientist, I have a master’s in meteorology), and I suffer a certain level of cognitive dissonance. However, science has by no means shaken my faith in God, if anything it strengthens it. I do think I regard the Bible more critically than most Christians.

The realm of rational science seems to be mostly atheistic, which is why I’m sometimes put off when discussions of climate science lead to accusations that (fundamentalist, if you want to use that word) Christianity and climate science denial go hand in hand. At least Ian Plimer has bucked that trend.

Anyway, my long and rambling point, even if it is a blunt one, is that the occasional snide comments about faith, mainly Christianity, can be quite hurtful to someone like me who loves both God and science.

Even denial depot does it sometimes, and I LOVE denial depot!

28. #28 TrueSceptic
January 17, 2010

126 Stu,

Young Earth Creationism is delusional. Most Christians (or Jews) are not YEC.

29. #29 janama
January 17, 2010

An Urgent Call to Action:
Scientists and Evangelicals Unite to Protect Creation
January 17, 2007
National Press Club, Washington, D.C.

“We believe that the protection of life on Earth is a profound moral imperative. It addresses without
discrimination the interests of all humanity as well as the value of the non-human world. It requires a new moral
awakening to a compelling demand, clearly articulated in Scripture and supported by science, that we must
steward the natural world in order to preserve for ourselves and future generations a beautiful, rich, and
healthful environment. For many of us, this is a religious obligation, rooted in our sense of gratitude for Creation
and reverence for its Creator.”

Signed by
James E. Hansen Ph.D.
Director
NASA Goddard Institute for Space Studies
Columbia University Earth Institute

perhaps you should check WUWT where Dr Spencer has just replied to your, and others, pathetic accusations.

30. #30 dhogaza
January 17, 2010

Janama, we already know of Spencer’s lame excuse for arbitrarily changing the rolling average length.

We think he’s fibbing.

31. #31 el gordo
January 17, 2010

Richard Dawkins seems to think ‘intelligent design’ is not so far fetched.

BEN STEIN: What do you think is the possibility that Intelligent Design might turn out to be the answer to some issues in genetics or in evolution.

DAWKINS: Well, it could come about in the following way. It could be that at some earlier time, somewhere in the universe, a civilization evolved, probably by some kind of Darwinian means, probably to a very high level of technology, and designed a form of life that they seeded onto perhaps this planet. Now, um, now that is a possibility, and an intriguing possibility. And I suppose it’s possible that you might find evidence for that if you look at the details of biochemistry, molecular biology, you might find a signature of some sort of designer.

32. #32 cohenite
January 17, 2010

el gordo; Dawkins is taking the piss; the clue is by “Darwinian means”.

33. #33 tresmal
January 17, 2010

Re:El Gordo’s post above. Finally something I can comment on. First, that interview segment comes from the breathtakingly dishonest pseudodocumentary Expelled. Second, the interviews were obtained deceitfully. Then the interviews were edited to present the scientists in the worst possible light and to distort what they were trying to say. Third Dawkins does not give any credence to alien intelligences being involved in our evolution, he was merely proposing a hypothetical in order to answer the question: How would we detect Intelligent Design?
I don’t know if this is representative of the quality of El Gordo’s sources, but if it is, it confirms the opinion of him that so many people here have of him.

34. #34 MapleLeaf
January 17, 2010

This from Kirtman (J. Climate 1997),

“Typically, the time between warm events is around 3 to 5 yr; however, there is also considerable modulation of the ENSO cycle on decadal timescales (Wang 1995; Zhang et al. 1997).”

And this from An and Wang (J. Climate 2000)

“In the late 1970s, the ENSO cycle exhibited frequency change. The oscillation period increased from 2–4 yr (high frequency) during 1962–75 to 4–6 yr (low frequency) during 1980–93.”

Again, Spencer has no physical basis for changing to a 25-month avg.. He is trying to weasel, using ENSO as an excuse for changing the averaging period, but his averaging period is not consistent with that typically associated with ENSO (which has quite a variable period, so using a fixed avg. for the purpose cited by Spencer does not make sense).

35. #35 TrueSceptic
January 17, 2010

132 tresmal,

The source was typical. The reasoning was no better.

36. #36 el gordo
January 17, 2010

Could be right, cohers. Wouldn’t be the first time I’ve been caught out.
I have observed ufos and admit to bias, so its only natural that I saw parody as truth.

37. #37 Bernard J
January 17, 2010

BJ; as usual you don’t read what I write or don’t write; i did not say R2 was the “be-all and end-all” in any matter

(Snigger…)

Cohenite.

As usual you don’t read what I write or don’t write – I did not say that you said that “R2 was the “be-all and end-all”. I simply asked you to explain why you thought that it was, and why other matters were irrelevant.

Then there is this comment from you:

As for this;

Have a look at what is being described; the 3rd order poly of the differences is almost dead straight; don’t you think that constitutes some sort of validity of using the 3rd polys on the main data?

Oh, I had a look at the data alright, cohenite – I had a good look. In fact, I took the [raw global UAH data from 1979 to date](http://www.ncdc.noaa.gov/oa/climate/research/uahncdc.lt), and derived polynomials from orders 1 through to 6.

What I obtained was very interesting…

The graph itself [is here](http://i47.tinypic.com/nc0b2r.jpg). As a consequence of the fact that I have only a borrowed computer at present, I could only construct the graph in Excel, and thus I had no way of differentiating the different polynomial fits using the ‘add trendline’ option. Frustratingly, I could not manually reconstruct anything higher than a first order polynomial using the coefficients given by the trendline function, because Excel rounded some of them to too few decimal places. Using the coefficients given, the resultant lines were simply rubbish, and thus there was no straight-forward way to add my own stippling to the trendlines in order to identify them.

So, for those who are interested, the polynomial orders of the fits on the graph are, from top to bottom, and on the left hand side and the right hand side termini respectively:

1. 3 (LHS): 2 (RHS)
2. 6 (LHS): 1 (RHS)
3. 4, 5 (LHS): 3 (RHS)
4. 2 (LHS): 6 (RHS)
5. 1 (LHS): 4, 5 (RHS)

As might be gathered from the above listing, and from the graph, the 4th order and the 5th order polynomial fits were almost identical in trajectory.

Take some time to carefully look at the various fits, cohenite, and tell us why the third order polynomial is your fit of choice. Please feel free to elaborate on the concept of parsimony in the giving of your explanation, to the extent that any reasonable statistician or scientist would expect.

And whilst you’re thinking about it, you might like to consider the particularly interesting R-squared values for each of the polynomial fits, listed below in the same order as the polynomials themselves:

1. 0.2903
2. 0.2998
3. 0.3426
4. 0.3466
5. 0.3467
6. 0.3475

Whilst you’re contemplating the answer to [the question about your contribution to the ‘structural break’ paper](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208934), you might also think about how you would go about seriously justifying – in the manner of a methodology in a peer-reviewed paper, your choice of a third order polynomial over other smoothing techniques.

38. #38 Bernard J.
January 17, 2010

BJ; as usual you don’t read what I write or don’t write; i did not say R2 was the “be-all and end-all” in any matter

(Snigger…)

Cohenite.

As usual you don’t read what I write or don’t write – I did not say that you said that “R2 was the “be-all and end-all”. I simply asked you to explain why you thought that it was, and why other matters were irrelevant.

Then there is this comment from you:

As for this;

Have a look at what is being described; the 3rd order poly of the differences is almost dead straight; don’t you think that constitutes some sort of validity of using the 3rd polys on the main data?

Oh, I had a look at the data alright, cohenite – I had a good look. In fact, I took the [raw global UAH data from 1979 to date](http://www.ncdc.noaa.gov/oa/climate/research/uahncdc.lt), and derived polynomials from orders 1 through to 6.

What I obtained was very interesting…

The graph itself [is here](http://i47.tinypic.com/nc0b2r.jpg). As a consequence of the fact that I have only a borrowed computer at present, I could only construct the graph in Excel, and thus I had no way of differentiating the different polynomial fits using the ‘add trendline’ option. Frustratingly, I could not manually reconstruct anything higher than a first order polynomial using the coefficients given by the trendline function, because Excel rounded some of them to too few decimal places. Using the coefficients given, the resultant lines were simply rubbish, and thus there was no straight-forward way to add my own stippling to the trendlines in order to identify them.

So, for those who are interested, the polynomial orders of the fits on the graph are, from top to bottom, and on the left hand side and the right hand side termini respectively:

1. 3 (LHS): 2 (RHS)
2. 6 (LHS): 1 (RHS)
3. 4, 5 (LHS): 3 (RHS)
4. 2 (LHS): 6 (RHS)
5. 1 (LHS): 4, 5 (RHS)

As might be gathered from the above listing, and from the graph, the 4th order and the 5th order polynomial fits were almost identical in trajectory.

Take some time to carefully look at the various fits, cohenite, and tell us why the third order polynomial is your fit of choice. Please feel free to elaborate on the concept of parsimony in the giving of your explanation, to the extent that any reasonable statistician or scientist would expect.

And whilst you’re thinking about it, you might like to consider the particularly interesting R-squared values for each of the polynomial fits, listed below in the same order as the polynomials themselves:

1. 0.2903
2. 0.2998
3. 0.3426
4. 0.3466
5. 0.3467
6. 0.3475

Whilst you’re contemplating the answer to [the question about your contribution to the ‘structural break’ paper](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208934), you might also think about how you would go about seriously justifying – in the manner of a methodology in a peer-reviewed paper, your choice of a third order polynomial over other [smoothing techniques](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208861).

39. #39 jakerman
January 17, 2010

Cohnite provides a master class in sophistry and ducking from reality:

>*”The linear trend is no doubt the poorest one, the intrinsically determined overall adaptive trend is a major improvement over the linear fitting, and the multidecadal trend catches essentially the meaningful variability and change associated with the annual GSTA, showing even greater improvement”*

Wu eta al were saying that the linear trend is the worst fit of the the three over 150 years of increaing atmosphereic CO2 (not the last 30 years of rapid warming).

Your chow test has not been justified in this application it is at complete odds with Wu’s analysis which shows the trend is expalined with increasing warming trend with an overlay of short term cycles.

cohnite writes
>*The Chow test is the best way of depicting the multidecadal trend.*

You have not supported this statement, you’ve merely asserted it. Wu’s trend analysis slaps this bunk down.

Wu’s work shows the Chow test to be unjustified, amoung the useless ways of depicting the multidecadal trend.

40. #40 Connor
January 17, 2010

Nice work, Tim, you’ve caused Spencer some umbrage by the look of it 😀

http://www.drroyspencer.com/2010/01/…rt-you-decide/
Quote:
“One of the great things about the internet is people can post anything they want, no matter how stupid, and lots of people who are incapable of critical thought will simply accept it.”

“I’m getting emails from people who have read blog postings accusing me of “hiding the increase” in global temperatures when I posted our most recent (Dec. 2009) global temperature update. In addition to the usual monthly temperature anomalies on the graph, for many months I have also been plotting a smoothed version, with a running 13 month average. The purpose of such smoothing is to better reveal longer-term variations, which is how “global warming” is manifested.”

41. #41 cohenite
January 17, 2010

Well BJ, I’m used to you verballing me but now you’re verballing yourself; this is what you said:

“Seriously, tell us why r-squared is the be-all and end-all in this matter” [comment 114]

Now, you’re splitting hairs by saying you did not say I said this; splendid LAYWERING BJ!

Anyway, you’ve done polys 1-6 of the UAH data but if you have a look at the graph again you will see that poly3 is used not just on UAH data but also SH ocean and Global data; I explain why a 3rd poly was used at comment 93; again I point out that the 3rd poly of the DIFFERENCES is almost dead straight with the land/ocean difference dropping in the last 5 years; my point was that this tends to kill off the pipeline concept because, well, you figure it out BJ.

Now, I think you are being mischievious asking about my contribution to the Break paper; but being obliging I’ll try and knock up a polynomial graph depicting my input.

I don’t have any “hot spot nonsense” so I won’t bother trying to understand that any further.

Try and lift your game BJ.

42. #42 cohenite
January 17, 2010

jakerman; the break paper looks at the Australian and global temperature data from 1910 to the present; its not as long as the Wu data but it is considerably longer than 30 years; Wu finds the multidecadal trend [MT] the most reliable; so does the break paper which justifies the MT on the nominated break dates [see Table 1].

43. #43 Grendel
January 17, 2010

If Dr Spencer keeps extending his period (say maybe as far as a 100-month rolling average) he’ll quickly find that detracts from any arguement he might make that the world is cooling.

For entertainment purposes I recommend taking the data and running a 100-month rolling average!

44. #44 thefordprefect
January 17, 2010

Going back to hiding the incline I have asked what has happened to the 1km temperature that used to be on the AMSU site This showed a rapid rise but then the CHLT data was discontinued in Nov. 2009:
http://img69.imageshack.us/img69/2104/amsutemptrends.png
This data was then replaced with sea surface temperature

What I find most strange is that so many sceptics consider AMSU data the most accurate – “all the others have been adjusted”. Satellite temperatures must be some of the most tweaked temperatures around – Temperature is not being measured at all it is a microwave proxy, The height of measurement is not a spot height – it is the combined “temperature” from a range of heights.

45. #45 jakerman
January 17, 2010

cohnite,

>Your chow test has not been justified in this application it is at complete odds with Wu’s analysis which shows the trend is expalined with increasing warming trend with an overlay of short term cycles.

>*if you doubt Wu’s work rejects your projection, how well do you think your projection fits with Wu’s trend analysis? Please reconcile the two.*

Your running continued running away from this reconciliation demonstrates a disingenous approach. But you have more or less addmited to this disingenous appraoch, as [I previously observed](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208524).

46. #46 jakerman
January 17, 2010

James Hansen [shows](http://www.columbia.edu/~jeh1/mailings/2010/20100115_Temperature2009.pdf) what GISS temp would be if it covered the same area as HadCRUT.

47. #47 Bernard J.
January 17, 2010

Curiouser and curiouser…

Suspecting that you would [take offence](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210074) to the fact that I used the global [UAH data](http://www.ncdc.noaa.gov/oa/climate/research/uahncdc.lt) rather than the ocean data subtracted (for whatever reason you might have had) from the global data, as [you did](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208992), I repeated my [previous exercise](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210009).

[This graph](http://i47.tinypic.com/2ir3sj4.jpg) is somewhat clearer, because I swapped the gammy borrowed mouse with my own, so that this time I had a functioning right button. Anyway, I coloured the fitted polynomial trendlines thus, the listed order reflecting the order of the polynomials themselves:

1. orange
2. red
3. black
4. green
5. aqua
6. cyan

The green fourth order polynomial line is difficult to discern because it is overlaid by the aqua fifth order polynomial.

The R-squared values are:

1. 0.067
2. 0.0737
3. 0.078
4. 0.0837
5. 0.0839
6. 0.0867

You will note that I chose as a reference the deadest, straightest of the polynomial lines on [your graph](http://i50.tinypic.com/2nia4xd.jpg). You will note too that your legend refers only to “globe, land, ocean, and ‘diff'” – there is no mention, as [you imputed](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210074) of the SH (southern hemisphere) in your legend. There’s also no mention of the southern hemisphere in your post[ at #93](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208267): WTF are you bringing it into the discussion for?!

So, my [questions](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210009) stand:

Take some time to carefully look at the various fits, cohenite, and tell us why the third order polynomial is your fit of choice. Please feel free to elaborate on the concept of parsimony in the giving of your explanation, to the extent that any reasonable statistician or scientist would expect.

And whilst you’re thinking about it, you might like to consider the particularly interesting R-squared values for each of the polynomial fits…you might also think about how you would go about seriously justifying – in the manner of a methodology in a peer-reviewed paper, your choice of a third order polynomial over other smoothing techniques.

In addition, you might comment on the flatness of all of the polynomial fittings, no matter their order/degree, and you might like to explain, given [your comment](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208992) that

… the 3rd order poly of the differences is almost dead straight; don’t you think that constitutes some sort of validity of using the 3rd polys on the main data?

why ‘straightness’ is a criterion for selecting a third order polynomial fit (as opposed to any of the other orders, which are similarly ‘straight’), and why the ‘straightness’ of the “global minus ocean” data has any bearing on the simple global data.

For extra credit, I am eager to know why you think that [your](http://i50.tinypic.com/2nia4xd.jpg) ‘global – ocean’ data trend down, and [mine](http://i47.tinypic.com/2ir3sj4.jpg) trend up.

I don’t have any “hot spot nonsense” so I won’t bother trying to understand that any further.

I see that you are in denial (yet again) and are now running away from your hot spot garbage: I will take it then that you acknowledge that my arguments are correct, and that your misinterpretation/misrepresentation of the physics of hot spots is exactly that – misinterpretation and/or misrepresentation.

Now, you’re splitting hairs by saying you did not say I said this; splendid LAYWERING BJ!

get over yourself cohenite. I will leave it to others here to decide if I am correct in what I said, and if you are attempting to blur the truth (in typical lawyerly fashion), probably because your nads are being strewn all over the thread.

I am satisfied with the level of my game: if I were you, I certainly would not be. But heck, why rely upon our own subjective opinions – if you dare, ask the other readers here what they think of our exchanges.

If anyone besides you sees deficiency in anything that I have said, I am happy to address it, or to ask for input if I can’t.

I don’t see you doing anything of the sort, and I certainly don’t see you taking any of an ever-growing number of criticisms of your poor science, and actually giving a valid justification for them, let alone attending to the highlighted deficiencies.

But then, this is what it’s all about, really, isn’t it? You’re not actually interested in real science; you’re simply trying to find pseudoscientific crud that presents as being a credible disproof of AGW, and you’re trying to dupe as many as are prepared to hear your message of inaction.

It must sting awfully that you are being called out about it.

Of course, if I am wrong and you seriously believe that you are playing at ‘real science’, then I invite you yet again to make your points at Open Mind, or at RealClimate, and see what the professionals in the field think about your so-called analyses.

48. #48 jakerman
January 17, 2010

cohnite,

Since you have refered to it here, can you again provide a link to your Chow test structurl break paper here as well?

49. #49 Bernard J.
January 17, 2010

Stcokwell’s ‘structural break’ piece, with Anthony Cox’s name upon it apparently for decoration (although cohenite promises that he’ll attempt to “knock up a polynomial graph depicting [his] input”), is [here](http://arxiv.org/PS_cache/arxiv/pdf/0907/0907.1650v3.pdf).

50. #50 Bernard J.
January 18, 2010

Cohenite.

This interests me so much that I think that the question bears repeating:

For extra credit, I am eager to know why you think that your
‘global–ocean’ [Poly.(Diff G-O)] data trend down, and mine
trend up.

After all, your ‘global’ points are trending up at a faster rate than your ‘ocean’ points, which as I would expect. Why then should your “[almost dead straight](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208992)” ‘Poly.(Diff G-O)’ fitted trend decrease over the time period 1979-2009, and what does such a trend inmply about the relative heat capacity of oceans, compared to that of the globe overall?

51. #51 MapleLeaf
January 18, 2010

Why are we still entertaining Cohenhite on this thread? S/he has effectively hi-jacked the thread and it detracting from discussion of Spencer’s manipulation of the MSU data. Correct me if I am wrong, but his/her posts have nothing to do with Spencer.

52. #52 Dhogaza
January 18, 2010

MapleLeaf, that appears to be the denialist MO no matter what subject of a thread, or in what venue. Hijack it back to irrelevancies.

That’s actually the denialist MO in the real world, too. Ignore rising temps, melting glaciers, diminished arctic sea ice, rising sea levels, and concentrate on stuff like the CRU e-mail theft, Al Gore being “fat”, etc.

53. #53 jakerman
January 18, 2010

Bernard, thanks for the link to cohnite’s paper.

How robust is the Stockwell-cox prediction? Well, its an artifact of HadCRUT, which has poor global coverage [compared to GISS](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210151).

But [here are](http://www.woodfortrees.org/plot/gistemp/plot/gistemp/from:1910/to:1978/trend/plot/gistemp/from:1978/to:1997/trend/plot/gistemp/from:1997/trend) the same periods in the GISS.

Further more Stockwell makes the baseless predction that ‘structural breaks’ will not reoccur before 2050 -despite the evidence that the temperature of recent years (2008 in particular) being dominated by cycles (solar miniuma and La Nina) which will or have turned to warming phase.

54. #54 MapleLeaf
January 18, 2010

Dhogaza, I suspect that your observations are correct, again.

On a somewhat relevant note to this thread. I think I suggested earlier that January 2010 would be the 2nd warmest in the satellite record, not sure what is going on but global temps in the mid and lower trop have been increasing rapidly for over a week now, and even if they do start declining, it still looks like Jan 2010 will be the warmest on the satellite record. The global SATs and trop temps typically lag the peak of the ENSO by about 6 months, so this rapid increase in January lower and mid trop temps is somewhat perplexing/intriguing as the ENSO seems to have peaked in December.

Any guesses what is causing the marked warming in the AMSU data?

55. #55 jakerman
January 18, 2010

And [here is](http://www.woodfortrees.org/plot/uah/plot/uah/from:1978/to:1997/trend/plot/uah/from:1997/trend) the ‘structural break’ period using UAH temperature. No drop in warming slope at all.

Sorry cohnite, back to the drawing board for you and David. At least one of you acknowledged that lack of fractional diferencing parameter for HadCRU3GL global were so hight (0.50) that “*results must be interpreted more cautiously*”.

56. #56 jakerman
January 18, 2010

Mapleleaf writes:

>*not sure what is going on but global temps in the mid and lower trop have been increasing rapidly for over a week now, and even if they do start declining, it still looks like Jan 2010 will be the warmest on the satellite record.*

Maple who cares what data shows the globe is warming, its means [nothing unless Prague is warming](http://wattsupwiththat.com/2010/01/15/uah-satellite-data-has-record-warmest-day-for-january/).

57. #57 cohenite
January 18, 2010

Sorry if you got left behind MapleLeaf but the relevance is Roy Spencer coped some flack for using higher order polys to depict the temperature ‘history’; I’m just exploring whether polys do have a role to play in presenting trends.

BJ, you say this:

“After all, your ‘global’ points are trending up at a faster rate than your ‘ocean’ points, which as I would expect. Why then should your “almost dead straight” ‘Poly.(Diff G-O)’ fitted trend decrease over the time period 1979-2009, and what does such a trend inmply about the relative heat capacity of oceans, compared to that of the globe overall?”

I don’t know why your G-O polys go up; obviously this shows that ocean temps are increasing more than global and would confirm heat being stored in the oceans; the graph I presented shows the opposite which as I said would indicate that there is no heat being stored in the ocean; I checked WFT for a quick comparison from 1998;

The graph of course from 1979 shows consistent upward trends in all indices. The OHC data generally shows downward trends from about 2003 [the changeover date] so I’ll have a think about that.

jakerman; you use the UAH data to show that there is no break; but your graph result is described in the paper at page 8;

“This is known as a change point model, characterized by abrupt changes in the mean value of the underlying dynamical system, rather than a smoothly increasing or
decreasing trend. The confidence in 1978 as a break-date is further strengthened by the results for global temperatures since 1910, indicating the series could be described as gradually increasing to 1978 (0:05 0:015C per decade), with a steeper trend thereafter (0:15 0:04C per decade).”

The point you miss is described in the following paragraph:

“The Chow test since 1978 finds another significant break-date in 1997, delineating an increasing trend up to 1997 (0:13 0:02C per decade) and non-significant trend thereafter (0:02 0:05C per decade). Contrary to
claims in Easterling & Wehner (2009) that the 10 year trend since 1998 is arbitrary, structural change methods indicate that 1997 was a statistically defensible beginning of a new, and apparently stable regime.”

58. #58 jakerman
January 18, 2010

cohnite, you miss every point I made [here](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210434)and [here](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210148)

59. #59 jakerman
January 18, 2010

>*Neither of the paragraphs you cite support your projection nor support you against the critiques I made.

60. #60 Bernard J.
January 18, 2010

I don’t know why your G-O polys go up; obviously this shows that ocean temps are increasing more than global and would confirm heat being stored in the oceans

Amd with this comment you have snookered yourself cohenite.

Think about it carefully. If global temperatures are increasing at a greater rate than are ocean temperatures, then the progressive results of the subtraction of ‘ocean’ from ‘global’ will also be greater.

Another way to look at it is that the difference between ‘global’ and ‘ocean’ is increasing with time – thus the trend of the subtraction resulting in this ongoing increase will also be increasing. It is certainly not correct to say that because the trend of the subtraction of ‘ocean’ from ‘global’ is increasing over time, then “obviously this shows that ocean temps are increasing more than global” over time.

As to the matter of “heat being stored in the oceans”, you are forgetting (or you simply do not understand) that water has a vastly greater heat capacity than the atmosphere, and can ‘store’ relatively much greater quantities of heat without reflecting the same increase in temperature than would the atmosphere, if that heat were instead ‘stored’ there.

I’m not nearly close to finishing with you yet cohenite/Anthony Cox, but I believe that a pause in procedings is necessary in order that you might contemplate addressing the ever-increasing tally of your howlers, distortions, and other deliberate and/or ignorant mishandlings of science.

Have a careful read of the many posts by myself and others on this thread and on previous ones, and ask yourself where you have actually ever made a substantive point in reply… especially one that in any way diminishes the import of the weight of evidence and of the informed opinion of the world’s best professional climatologists.

61. #61 jakerman
January 18, 2010

Severn reasons why the structural break is inappropriate for making future projections:

1) These structural breaks have no predictive capacity;

2) The prediction of Stockwell and yours is based on the unsupportable assumption ‘structural breaks’ will not reoccur before 2050 -despite the evidence that the temperature of recent years (2008 in particular) being dominated by cycles (solar miniuma and La Nina) which will, or have already begun turn to warming phase, such forcing and cycles are governed by physical mechanisms;

3) The break for HadCRUT is dodgy based on its own tests (fractional diferencing parameter = 0.5);

4) You cite your paper saying:

>*The Chow test since 1978 finds another significant break-date in 1997, delineating an increasing trend up to 1997 (0:13 0:02C per decade) and non-significant trend thereafter (0:02 0:05C per decade)*

Howerver the non-significant trend your refer to after 1997 is an artifact our your [sample being too short](http://tamino.wordpress.com/2009/12/15/how-long/).

5) The difference in slope either side of the 1997 break disappears when using either [GISS](http://www.woodfortrees.org/plot/gistemp/plot/gistemp/from:1910/to:1978/trend/plot/gistemp/from:1978/to:1997/trend/plot/gistemp/from:1997/trend) or UAH, so its the opposite of robust;

6) GISS Temp has [better global coverage](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210151) than HadCRUT;

7) Your dodgy projection is smacked down by the trend analysis done they Wu (discussed is earlier links). The best fit is an increasing warming trend with an overlay of short term cycles.

62. #62 Connor
January 18, 2010

I’m pretty clueless when it comes to statistics, can anyone show me what a 10-year moving average would look like using Wood For trees?

http://www.woodfortrees.org/plot/uah/

63. #63 Ezzthetic
January 18, 2010

From Spencer’s defence of Intelligent Design:

the nucleus of every human cell is a digitally coded database containing more information than Wikipedia …

Surely we could solve this whole Intelligent Design question by examining the Human Genome, and checking if there’s a “citation needed” flag against it.

64. #64 jakerman
January 18, 2010

Connor, WFT use months, so 10-year moving average is [mean:120 months](http://www.woodfortrees.org/plot/uah/plot/uah/mean:120).

65. #65 TrueSceptic
January 18, 2010

137, 138 Bernard,

The graph itself is here. As a consequence of the fact that I have only a borrowed computer at present, I could only construct the graph in Excel, and thus I had no way of differentiating the different polynomial fits using the ‘add trendline’ option.

I don’t understand. Although the Add Trendline function does not itself have the options you would expect, if you right-click > Format Trendline, afterwards, you can change style, colour, and weight for each one.

66. #66 TrueSceptic
January 18, 2010

144 thefordprefect,

Tamino has compared UAH and RSS at least once. Other than the known errors in UAH, corrected some years ago, there is [something weird going on](http://tamino.wordpress.com/2008/10/21/rss-and-uah/)

67. #67 TrueSceptic
January 18, 2010

165 me,

Ignore that Bernard, but you could’ve told us why. 😉

Umm, in Windoze is there another way of doing a right-click?

68. #68 P. Lewis
January 18, 2010

Shift + F10 works in Word, so I presume it would in Excel.

69. #69 Bernard J.
January 18, 2010

TS.

Sorry about my lack of explanation the first time ’round. I was tired, incredulous at the laywer’s denseness, and frustrated with the computer and its mouse. I realised later that there is a drop-down way of doing the same thing, but as I never usually use Excel to graph anyway, I wasn’t inclined to stuff around at the time.

The original graph in glorious polynomial colour is [here](http://i47.tinypic.com/117bj3q.jpg), with the same colour-coding [as used at #147](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210203).

I think that it shows quite clearly that cohenite is trying to “hide the increase”, a la Spencer’s ‘trick’ back in April 09…

70. #70 Bernard J.
January 18, 2010

Just to add: left-click on the trend line, and drop-down ‘Format’ is the other way that I realised afterward.

Either way, Excel would not be my graphing package of choice!

71. #71 David Duff
January 18, 2010

Well, I couldn’t see what he was “hiding” given that unlike some climate scientists he publsihed both graphs. Anyway, now he has published both layouts on one graph and it shows about as clearly as a boil on the end of your nose an increase in temperatures over the last 30 years. If you call that “hiding” what do you call what Mann and Jones ‘et al’ have been doing?

72. #72 MapleLeaf
January 18, 2010

Janet, re #156. Thanks, for spoiling my day. Just kidding, read your link to WUWT. I love the following:

“However, there is another, more important consequence of these numbers. And it is the following: the global mean temperature is irrelevant for you and for everyone else, too.”

If they could harness the energy for all that spinning, they will have solved our global energy problems. The article is filled with “cute’ smiley and wiking faces– looks like something written on twitter.

73. #73 Dr Who
January 18, 2010

Latest satellite data show a huge spike in temperatures – http://discover.itsc.uah.edu/amsutemps/amsutemps.html . January 2010 could well be the warmest January on record.

74. #74 cohenite
January 19, 2010

Yes, you are correct BJ; G-O with an increasing divergence [G is decreasing at a lessor rate than O] should produce an upward trend; my mistake, it means of course that the O temperature is declining [not increasing] and there is no pipeline which was my point at 141. I won’t bore you with the usual NODC graphs, or the usual papers from Levitus, Ishll & Kimoto, Dominques, Dipuccio, Wijffels, Willis, Loehle, all of which show a post 2003 decline in OHC, but none explain the ARGO transition spike; instead I’ll refer you to a new paper by Douglass and Knox which studied OHC to 750 metres and found no evidence for a pipeline effect but evidence for breaks;

http://www.sciencedaily.com/releases/2009/08/090814103237.htm

So, BJ, your 3rd poly graph was better than mine and truly shows a decline in OHC [at the very least relative to G and L] while Global and Land temperatures have increased over the same period; due perhaps to UHI?

75. #75 jakerman
January 19, 2010

cohnite writes:

>*it means of course that the [Ocean] temperature is declining [not increasing] and there is no pipeline which was my point at 141.*

It means nothing of the sort, you’ve just been corrected and still didn’t take the care required to interpret the data.

76. #76 cohenite
January 19, 2010
77. #77 jakerman
January 19, 2010

Keep doing your part, we make a pretty good team in exposing the cherry picking, misrepresentions, and fact filtering employed by denialists.

We’ll keep point out your errors in as readable way as we can, in order that fair minded readers can judge what the denialist game is.

78. #78 TrueSceptic
January 19, 2010
79. #79 Bernard J.
January 19, 2010

It doesn’t rain cohenite, but it pours…

G-O with an increasing divergence (G is decreasing at a lessor [sic] rate than O) should produce an upward trend…

“G[lobal temperature]” is increasing at a greater rate than is “O[cean temperature]”.

… my mistake, it means of course that the O[cean] temperature is declining not increasing…

Um, no, it means nothing of the sort.

Think carefully about what [I said](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210584):

If global temperatures are increasing at a greater rate than are ocean temperatures, then the progressive results of the subtraction of ‘ocean’ from ‘global’ will also be greater.

Figured it out? Mmmm?

According to the UAH data, and given the period 1979 to 2009, ocean temperatures are increasing, but overall global temperatures are increasing at a greater rate. It looks [like this](http://i47.tinypic.com/jhwmxd.jpg). I don’t think that the graph needs much explaining, except to say that it uses the same [UAH data](http://www.ncdc.noaa.gov/oa/climate/research/uahncdc.lt) as my [previous graphs](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210203), and that it shows that global temperatures are rising, and ocean temperatures are rising at a slower rate, and that the subtraction (whatever it might signify in your mind) of ocean temperature from global temperature thus describes a trajectory that, by simple mathematical consequence, is also rising.

It appears that you are struggling with the concepts underlying the ‘subtraction’ of trends. I could explain it using calculus, and specifically first derivatives, but I fear that I might lose you, or simply end up being [laywered](http://scienceblogs.com/deltoid/2009/12/russian_analysis_confirms_20th.php#comment-2158923) for weeks to come, so I will try to communicate the idea to you with simple graphs.

Scenario 1:
If global temperature and ocean temperature are both increasing at the same rate, then [their trajectories will be parallel, and the subtraction of ocean temperature from global temperature will describe a ‘horizontal’ path](http://i47.tinypic.com/r9n0o7.jpg).

Scenario 2 (reflecting the ‘real’ world…):
If global temperature and ocean temperature are both increasing, but global temperature is increasing at a greater rate than is ocean temperature, then [their trajectories will both be positive in slope, and the subtraction of ocean temperature from global temperature will describe an increasing path somewhere, in value of slope, between ‘horizontal’ and that of global temperature](http://i50.tinypic.com/25i2mfc.jpg), depending on the relative differences between the global and the ocean temperature trends.

Scenario 3:
If global temperature and ocean temperature are both increasing, but ocean temperature is increasing at a greater rate than is ocean temperature, then [their trajectories will both be positive in slope, and the subtraction of ocean temperature from global temperature will describe a path that is decreasing over time](http://i47.tinypic.com/wl6oud.jpg).

For [your statement that ocean temperatures are decreasing](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2212855) to actually be true, you would need a different dataset than the [UAH one]( http://i47.tinypic.com/jhwmxd.jpg) that you yourself referred to.

For your statement, in the post linked in the previous paragraph, that the subtraction of ocean temperature from global temperature describes a negatively-sloped trend to hold, ocean temperature would have to rise at a greater rate than does global temperature, which again would require a different dataset to the one that you used. This statement also happens to contradict [your insistence that the oceans are cooling](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2212855).

As to your reference in that last link, to [Douglass et al](http://www.sciencedaily.com/releases/2009/08/090814103237.htm), it is impossible to say exactly what their work implies without actually being able to read the (currently unpublished) paper, but I will make several points.

First, your implication that changes in ocean currents, and thus in relative amounts of heat exchange with the atmosphere, represent a “break” mechanism, does not challenge the fundamental tenets of AGW. If you think that it does so challenge it, can you explain exactly where in the deep oceans all of this heat has been stored, and how it managed to be stored for so long in a climate that hasn’t, for millennia, been warmer than today – the mediæval warm period, whatever its magnitude, notwithstanding (unless you believe that it was strong enough, and persisted long enough, to sequester the heat that has been released over the last century). Exactly how does heat stay at the cold bottoms of oceans for whatever unspecified period would be required in order to now pump the energy required to drive the last century’s warming?

Given that the thermohaline conveyor turns over about once every thousand years, if I recall correctly, storing a century (or more) worth of heat seems to be a neat trick… If there is an underlying millennial (or whatever) cycle of heat sequestration and release such as your claims might be said impute, why is it that the planet’s plants and animals show no adaptation to it? If they had experienced deep-ocean driven releases of heat in the past, there would not be the profound phenological shiftings that are presently being observed, and that will inevitably result in extinctions far above the background, non-climate human impacts notwithstanding.

Second, if you propose that the warming over the last century has been caused by shifts in deep ocean currents, do you understand exactly how much energy has been stored in these mysterious deep ocean repositories prior to the Industrial Revolution, or how much extra geothermal heat (see point 3 below) is required to provide this deep ocean energy; and how do you account for its release at the same time in history as the Industrial Revolution commenced? Just what is causing this putative heat to be released now?

Third, in light of Douglass’ et al claim that the source is geothermal, how then do you dismiss the science of CO2 (and other greenhouse gas) radiation physics?

Fourth, if the extra heat is geothermal in origin, why is there no obvious terrestrial reflection of whatever phenomenon is leading to its release? And if contemporary warming were due to marine geothermal sources only, how is such a phenomenon being missed by the decades of global oceanographic study? To warm the planet at the observed rate, there must be something profound indeed occurring underneath the planet’s oceans.

I could continue for paragraphs yet, but I’ll allow to you chew on that for starters, and perhaps let others have a bash at you too.

So, BJ, your 3rd poly graph* was better than mine and truly shows a decline in OHC [at the very least relative to G and L] while Global and Land temperatures have increased over the same period; due perhaps to UHI?

No, cohenite, my graphs, derived from the UAH data, do not show a decline in ocean heat content, where ocean heat content is reflected by ocean temperature. Everything I have shown you is consistent with reputable temperature data sources, and with the fact of global warming. You persist in your attempts to twist the data in order to show something that you want it to show, but which has no basis in fact or in physics. And [laywering](http://scienceblogs.com/deltoid/2009/12/russian_analysis_confirms_20th.php#comment-2158923) away as you are wont to do, this time with mention of the discredited urban heat island excuse, does not gain you any credibility.

Finally, if, by your [Wood For Trees link](http://www.woodfortrees.org/plot/hadcrut3vgl/from:1994/to:2010/scale:0.1/mean:10/plot/hadsst2gl/from:1994/to:2010/scale:0.1/mean:10/plot/uah/from:1994/to:2010/scale:0.1/mean:10) you are imputing that downward ‘noise’ fluctations represent a more general decline in heat content, I would simply advise you to refresh your understanding of the sort of phenomena that drive the overall amount of energy stored in the planet’s oceans. [This is one place to start](http://www.skepticalscience.com/What-causes-short-term-changes-in-ocean-heat.html): there are many more. It is entirely consistent with an underlying “pipeline” effect that ocean surface (0 to 700m) heat content might demonstrate short-term decline, whilst still representing a huge source of thermal inertia for longer term atmospheric warming.

Of course I don’t expect you to cede the point, or indeed any point that fundamentally challenges your ideology, but at least I have put it out there that you are demonstrably pushing another barrowload of [boy-cow poo](http://scienceblogs.com/deltoid/2009/12/russian_analysis_confirms_20th.php#comment-2192651).

* Cohenite, I am sure that any polygraph of mine would be better than yours.

80. #80 Bernard J.
January 19, 2010

Doh!

Scenario 3:
If global temperature and ocean temperature are both increasing, but ocean temperature is increasing at a greater rate than is ocean global temperature…

81. #81 luminous beauty
January 19, 2010

Cohenite,

Both you and the Science Daily writer do not understand the physics of [‘heating in the pipeline’](http://pielkeclimatesci.wordpress.com/2010/01/04/guest-weblog-by-leonard-ornstein-on-ocean-heat-content/).

That SSTs have declined in the past few years is balanced by the fact that upper ocean heat content has remained [steady](http://www.nodc.noaa.gov/OC5/3M_HEAT_CONTENT/).

82. #82 cohenite
January 19, 2010

Truesceptic; I wanted to show the lack of time lapse between the 3 indices; ergo no pipeline [or at least a bit of the puzzle showing no pipeline since only SST is used]; for temperature trends this may be better;

BJ, you may be a better grapher than me but you have no sense of irony at all; and you’ve verballed me again; I said:

” all of which show a post 2003 decline in OHC, but none explain the ARGO transition spike”

Luminous thinks OHC is “steady”; not since 2003; the significance of 2003 is explained here:

http://landshape.org/enm/possible-error-in-ohc/#more-3180

Now as a matter of interest what is the resident think tank’s opinion about 2003?

83. #83 Grendel
January 19, 2010

As interesting as this discussion has become I am still waiting for Cohenite to respond to Bernard J’s analysis in point 112 of why fitting method to data after the data has been collected is an unreasonable approach. He danced around the point but didn’t address it substantially – nor did Cohenite adequately respond to the critique of the use of third order polynomials as his only response (seemed to be – and I paraphrase what I interpreted it as here but am willing to be corrected) “because it looks right”.

This is rather like watching a tennis match where Bernard is using a Wilson Pro and Cohenite selected a used toothpick.

84. #84 jakerman
January 19, 2010

cohnite writes:

>*Luminous thinks OHC is “steady”; not since 2003*

Once again cohnite takes refuge in the noise to hide from the signal.

6 years of data (from 2003) is too short to decern a reliable trend from the noise. This is emphasised by [conflicting results](http://www.skepticalscience.com/Does-ocean-cooling-disprove-global-warming.html) of the qeustion of warming/cooling from 2003.

85. #85 jakerman
January 19, 2010

Grendel,

I think cohnite was hoping to distract from [those questions](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2214571).

Thanks for re-raising them.

86. #86 cohenite
January 19, 2010

Grendel; I don’t disagree with that point which is why I referred to the Wu et al paper with its adaptive analysis. As for statistical analysis generally I find this helpful;

http://landshape.org/enm/ten-commandments-of-statistics/

But I stress my statistical experience is at a graduate level only; even so I find BJ’s hyperbole tends to mask the worth of much of what he is saying; in respect of the 3rd poly I only introduced that to look at the pipeline effect and explained that at comment 93; BJ says a cubic poly only has one inflection; that doesn’t contradict what I said which was 2 fluctuations or valleys or hills;

http://en.wikipedia.org/wiki/File:Polynomialdeg3.svg

Poorly expressed but nonetheless still essentially correct even if you take a pedantic approach. I don’t use toothpicks; dental floss is more sanitary.

87. #87 Grendel
January 19, 2010

“As for statistical analysis generally I find this helpful”

What since landshape posted the link on 12 January?

No issue with the commandments themselves – fine guidelines, however posting them up does not mean they are being used (other than by Peter MacIntyre and his students).

I’d be much happier with the data you were presenting (or attempting to present) if you had a coherent approach to the analysis and I think those 10 commandments are a very good place to start. I too enjoy playing with various analytical techniques to make the lines wiggle on a graph – that doesn’t make it meaningful however.

When playing tennis, dental floss is even less effective than a toothpick.

88. #88 cohenite
January 20, 2010

“When playing tennis, dental floss is even less effective than a toothpick”

No doubt Grendel, but much more practical and, as I say, sanitary if one is sitting down to some of the splendid dishes featured at your website

89. #89 Grendel
January 20, 2010

At last – something on which we can agree. Floss is better than toothpicks for dental hygiene purposes.

90. #90 Bernard J.
January 20, 2010

Yes, you are correct BJ; G-O with an increasing divergence (G is decreasing at a lessor rate than O) [sic] should produce an upward trend; my mistake, it means of course that the O temperature is declining (not increasing) [sic] and there is no pipeline which was my point at 141.

is based upon the trends of the [graph](http://i50.tinypic.com/2nia4xd.jpg) (spanning 1979-2009) that you produced; this was also your point at [#141](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210074) – just as your original discussion at [#93](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2208267) about trends, and your statement that “there is no heat being stored in the ocean” [at #157](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2210494), arose from the same graph… the one that you originally linked to at [#77](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2207636).

Your “post 2003 decline in OHC” came late in the piece, after you were being called on your many other errors; and in this context it was pointed out to you by [myself](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2213233) and [jakerman](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2213081) that comment on ocean temperature/heat content trends are ill-advised in the context of such short time-frames. It also contradicts the work of [Levitus et al](ftp://ftp.nodc.noaa.gov/pub/data.nodc/woa/PUBLICATIONS/grlheat08.pdf).

As to your fixation-of-the-moment with 2003 ocean heat content, I am happy to follow this diversion if you first make your points about ocean heat content at [this thread on RealClimate](http://www.realclimate.org/index.php/archives/2008/06/ocean-heat-content-revisions/). In the mean time, how about you stick to the points that have already been made on this thread?

I find BJ’s hyperbole tends to mask the worth of much of what he is saying

What of my comments on scientific matters do you consider hyperbolic?

BJ says a cubic poly only has one inflection; that doesn’t contradict what I said which was 2 fluctuations or valleys or hills

Relevant to your claim is the fact that this point of inflection need not be bounded by a positive and a negative radius: both sides of the curve about a point of inflection may be monotonically increasing (or decreasing). In other words, it is possible for a third order polynomial to have no non-positive (or conversely, no non-negative) second derivatives.

If you do not understand how this might be so, play with [this applet*](http://www.univie.ac.at/future.media/moe/galerie/fun1/fun1.html).

What this means is that a third order polynomial need not describe a trajectory characterised by “valleys and hills”. If you don’t believe me, consider the arch-typical cubic, y = x3. “Valleys and hills”? Riiiight…

*”This applet” refers to the fourth applet on the page linked. I suggest that you play with it, because it makes a mockery of your claim about third order polynomials, because such need not show the “valley/hill” pattern with which you are so enamoured. Worse, valleys and hills might be fitted, within a studied range, using fourth order polynomials or higher, so what makes your selection of a third order polynomial so justified?

You’re full of laywerism cohenite, but I have yet to see in any of your posts a single substantive answer to the challenges to your dismal grasp of science.

91. #91 TrueSceptic
January 20, 2010

182 cohenite,

Huh? The Scale function just rescales the Y-axis values. You would use this to compare different datasets such as CO2 ppm and temp (or you could use Normalise). If you compare [your WFT](http://www.woodfortrees.org/plot/hadcrut3vgl/from:1994/to:2010/scale:0.1/mean:10/plot/hadsst2gl/from:1994/to:2010/scale:0.1/mean:10/plot/uah/from:1994/to:2010/scale:0.1/mean:10) with [mine](http://www.woodfortrees.org/plot/hadcrut3vgl/from:1994/to:2010/mean:10/plot/hadsst2gl/from:1994/to:2010/mean:10/plot/uah/from:1994/to:2010/mean:10/plot/hadcrut3vgl/from:1994/to:2010/trend/plot/hadsst2gl/from:1994/to:2010/trend/plot/uah/from:1994/to:2010/trend) you can see that the only difference apart from my added trend lines is that your temps are x0.01.

92. #92 TrueSceptic
January 20, 2010

184 jakerman,

93. #93 Bernard J.
January 20, 2010

I think that cohenite’s point about the comparison of the three indices was to imply that there is no obvious thermal flywheel effect, such as one might observe, for example, with soil temperatures over an annual seasonal cycle.

I’ve been sitting on commenting about it thus far, just to see where he had intended to take the idea, but now’s as good a time as any to indicate that his point misses the mark: however closely, or otherwise, the ocean and the atmospheric temperatures are tied, does not inform about the oceans’ capacity to affect the manner in which the planet continues to warm, given the radiative physics of a CO2/other GHG-enriched atmosphere, or given the physics of the movement, to thermal equilibrium, of the planet’s larger bodies of water.

Cohenite is on a roll with his confusion about the science of anthropogenic global warming, and how it is performed and interpreted. Still, for him it doesn’t matter: all he needs to do as [secretary](http://scienceblogs.com/deltoid/2009/05/plimer_and_arctic_warming.php#comment-1651965
) of the [as-yet unregistered](http://www.aec.gov.au/Parties_and_Representatives/party_registration/index.htm) Australian Climate Sceptics Denialists Party is to sow confusion about the science of anthropogenic global warming, and how it is performed and interpreted.

For cohenite, it’s never been about understanding the truth – indeed, spending to much time focussing on that part of the issue is just downright inconvenient.

94. #94 TrueSceptic
January 20, 2010

193 Bernard,

I was referring to his pointless Scale Factor, which can have nothing to do with showing any time lag.

95. #95 Bernard J.
January 20, 2010

I wanted to show the lack of time lapse between the 3 indices; ergo no pipeline…

His scaling is indeed a pointless exercise. In fact, it is worse than pointless, because his ordinate scale indicates anomalies ten times smaller than the real ones – a rather nasty piece of sleight-of-hand. Anyone not familiar with what the scaling on WFT does could leave a reading of that graph with a very different understanding of temperatures than they otherwise should.

Cohenite is either even [more confused](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2215316) than I had previously thought, or he is a greater master of mendacious misrepresentation of scientific fact than even one as cynical as I had given him credit for.

Whichever of the alternatives (and they are not mutually exclusive) is the case, it only serves to emphasis my comment in my last post that:

Still, for him it doesn’t matter: all he needs to do as secretary of the as-yet unregistered Australian Climate Sceptics Denialists Party is to sow confusion about the science of anthropogenic global warming, and how it is performed and interpreted.

Cohenite’s behaviour in all of this recalls a similar comment from [Michael Tobis](http://initforthegold.blogspot.com/2010/01/luciagate.html) (hat-tip to Former Skeptic for [drawing attention to Tobis’ thread](http://scienceblogs.com/deltoid/2010/01/open_thread_38.php#comment-2215480)):

Indeed, the non-sensible side usually makes some effort at sounding sensible (though apparently in the Palin era this is less necessary), precisely to make the choice more difficult.

Michael expands upon this observation, and it’s worth reproducing his subsequent paragraphs even though most of the subject matter has been raised on Deltoid before:

Let me say again, I think the number of ill-intentioned people among the opposition is probably quite small, and presuming ill intent on the part of any specific opponent is a bad idea.

On the other hand, to look at the situation, and see how many accusations are flying about, suffices to prove that there are bad guys at work, though it doesn’t suffice to prove which side they prefer and to what degree of unanimity. When accusations of gross malfeasance exist, either the accuser or the accused or perhaps both pretty much have to be in the wrong, i.e., unethical. Judgment must be applied to find out which is which, but the presence of a malign influence is already pretty much guaranteed.

I think, though, that Lucia has missed the point quite thoroughly.

The point isn’t that there are some actual bad guys among the various fools and egotists who plague us, though I have no doubt that there are some. The point is that scientific opposition and legal/political opposition are based on very different views of ethical behavior. When a scientific mind encounters a scientific mind on the opposite side of a question, if all is going well the exchange of ideas can be beneficial for both parties, because whatever opposing goals the parties may have are (at least ideally) subordinate to their shared goal of discovering the truth. This ideal may usually be honored in the breach, of course, but both scientists operate in a larger context where discovery of the truth (at least ideally) is the institutional goal of all participants.

That these ideals are under stress is not something I would contest, but let’s agree at least that this is the ideal. Wiener refers to nature as the opponent of the honest scientist as a “devil” in the Augustinean sense; an evil of incompleteness, not of competitive agency.

In legal and political circumstances, the opposition really is not just seeking the triumph of their ideas, they are seeking the triumph of their ideas irrespective of truth. That is, winning the battle is what counts; the actual validity of their position is secondary and the prospect of shifting sides under the weight of evidence is nil.

There follow several more paragraphs which, for the appearance of brevity I will omit. However, although it has been said many times on Deltoid by Jeff Harvey, myself, and many others, Michael’s last sentence is apposite and deserves inclusion here too:

Please remember, as Deech56 points out in the comments, that in the courtroom of science-related fact, Nature is the judge, and I don’t mean the journal.

All that is important for cohenite and his colleagues though is that they make a buck or several, and/or maintain their ways of life for as long as possible, before Nature passes sentence…

96. #96 jakerman
January 20, 2010

>*I think that cohenite’s point about the comparison of the three indices was to imply that there is no obvious thermal flywheel effect*

Beside cohnite (and RP snr) [misrepresenting the pipleline warming](http://scienceblogs.com/deltoid/2010/01/roy_spencer_hides_the_increase.php#comment-2213398), if cohnite is interested in termal intertia, then he missed the most relevent index. When it comes to thermal inertia and stroage, Sea Surface Temperature is not an adequate proxy for Ocean Heat Content.

97. #97 Bernard J.
January 29, 2010

We seek him here, we seek him there,
Those Frenchies seek him everywhere!
Is he in heaven? Is he in hell?
Where is that damn, elusive Pimp…

…er… bugger the verse.

Cohenite, have you given up defending the indefensible?

98. #98 carrot eater
February 2, 2010

Looks like Spencer is still using higher-order polynomial fits.

http://wattsupwiththat.com/2010/02/01/spencer-natural-variability-unexplained-in-ipcc-models/

Would somebody please, please teach the man how to smooth data before he hurts himself?

The rest of the post is silly as well. He makes a statistical ‘model’ of unforced variability with probably 6 fitted parameters (and even then, the fit isn’t all that great), adds it to the mean of the ensemble of GCMs, sees the result is running hotter than the most recent observations, and then decides this is evidence the GCMs have an overly high sensitivity. Right. And he finishes it off by failing at logic, or at least, understanding how paleoclimate observations constrain the sensitivity.

99. #99 Chris O'Neill
February 3, 2010

Looks like Spencer is still using..

Maybe he wants to get in a pre-emptive strike before announcing the warmest January on record.

100. #100 joni
October 7, 2010

I don’t know if you’ve seen, but Dr Spencer is “surprised” now (Sept 2010) that the temperatures are not falling.