OK thanks Dung but I was not put out in any way by BYJ's comment. I had intended to add that our IQ, computed by BYJ's undisclosed formula, might well finish up needing negative or even imaginary numbers to compute but that would not bother me. But my 14 minutes was up.

OK, Missy, I think I see what you are saying. But I don't think the peak at 10 months in his cross correlation function can be due to the effect you described.

Salby's graph of the cross correlation function between CO2 and temp (at 09:05 - I have the benefit of having captured his viewgraphs as .gif files) - does not seem to confirm what you have said.

There is no doubt that the CO2 record has a sinusoidal (approximately) ripple at 1 cycle per year. I think you have assumed that the temperature record will also have such a ripple (which would be hidden in the noise, as it is not visible in the temp record he displays).

You have said that the correlation at +10 months is because CO2 peaks around 10 months after the peak temperature.

I think there probably is no periodic component in the temperature record so there will be nothing to correlate with the periodic component in the CO2 record.

I think that this is because he's using *global* temperatures, so summer temperatures in the North are balanced by winter temperatures in the South, with essentially no cyclic summer/winter fluctuation in the global record.

If you cross correlate two periodic signals (of the same frequency), you'll get a cross correlation function that is periodic also at the same frequency, in this case, one cycle per year. But Salby's cross correlation function does not appear to be periodic at one cycle per year (nor at any other frequency). If your explanation were correct, the cross correlation function should have positive peaks at +10, +22, +34, ... and at -2, -10, -22,... months.

The instrumental record is only 50 years long so I'd say his assertion that the CO2/temperature relation in the instrumental record is the same as in the proxy record is plausible but is not incontrovertible, unless I have missed a point.

Well I assumed your formula was IQ = 100/Np but I wasn't sure.

Comments noted.

Well Martin, I don't suppose I can shake your faith in Salby whatever I say. But just as a final shot, I'll add a few thoughts.

I would not use the term "faith", which implies a belief that will not be changed by the presentation of alternative explanations that explain things better, or by contradictions between theory and observation.

What he has presented so far seems to me to provide a better explanations of some things than mainstream climate science does and he seems to have avoided using concepts that have no physical existence, yet which form the foundations of much of mainstream climate science. Not to overlook that his work looks as if it should be capable of making predictions capable of subsequently being compared with observations.

I am doubtless showing my lack of deep understanding of correlation (which you can all gloat over), but it seems to me that if, as I think we agreed, the correlation is acting upon the wiggles (ignoring the linear trend) then _all_ it can do is show a relation between the wiggles - the seasonal variations in CO2 and temp. What is wanted in contrast is a relation between the trend in CO2 and the trend in temp. It stretches my understanding too far for me to say whether Salby can instead have correlated the _trends_. As you can guess, I find it hard to imagine that would give a meaningful 'lag' between the two.

I'd agree that correlation is only useful for showing relations between records with lots of wiggles (ie degrees of freedom, duration × bandwidth product or whatever you want to call it). If you have long trends that extends over most of the the length of your record, then I don't think that correlation calculations can tell you anything much about the relation between them.

On the other hand, if short term fluctuations in temperature have high correlation with short term fluctuations in CO2, which implies a physical effect linking the two, then it's hard for me to see why the same effect should not also link long term fluctuations in the two things. (In other fields, you can have effects where only rapid fluctuations have an effect but there are clear physical reasons for this.)

You are clearly correct that there should be no ripple in the temperature - it is a global temperature after all. But that may be asking too much of the global temperature series (whichever one was used). Most of the land mass and the measurements are in the northern hemisphere so it might be logical that a bias towards the NH exists in the data. Whether a possibly small ripple is consistent with a correlation of 0.5, I don't know. The lack of peaks at other months is indeed a challenge to my suggestions, one that I cannot counter.

No, it's not obvious that the Southern hemisphere temperature should be the precise complement of the Northern hemisphere temperature. A Fourier spectrum analysis of the global average temperature record over a long enough period should reveal the presence of a discrete frequency component at 1 cycle per year, if there is one. Dunno if such an analysis has ever been done.

I wish you, Martin, all the best - I'll leave the rest of you to slap each others backs now as my presence is clearly unwanted.

Yes, sorry about that. All the best to you too Missy, and thanks for bringing up points which provoked the need to think.

A comment by Bart at the discussion of David Coe's paper refers to Salby's work and the (net CO2 emission/temperature) relationship, with some possible explanations.

http://bishophill.squarespace.com/blog/2013/8/1/a-new-look-at-the-carbon-dioxide-budget-part-2.html?lastPage=true#comment20240508

In a subsequent comment, Bart gave the answer to a question posed by Missy: (in my words)

"OK, so the wiggles in the record of CO2 emissions may be correlated with temperature, but this says nothing about trends so Salby is incorrect"

[trends equate to fluctuations with low frequency content; wiggles have more high frequency content.]

Bart said:

"It is also unphysical to assume that temperatures can drive CO2 variations while something else drives the long term behavior without any phase distortion appearing at the crossover frequency where they blend."

Bart continued...

"It also runs afoul of Occam's Razor, in that the temperature accounts for both the long and the short term without needing any additional input. As Laplace would say, we have no need of the hypothesis of an additional significant driver."

Aug 2, 2013 at 6:40 PM Bart

Correct URL for Bart's comment

"A comment by Bart at the discussion of David Coe's paper refers to Salby's work and the (net CO2 emission/temperature) relationship, with some possible explanations."

http://bishophill.squarespace.com/blog/2013/8/1/a-new-look-at-the-carbon-dioxide-budget-part-2.html#comment20240508

From Jo Nova on WUWT and on her own blog:

Murry Salby responds to critics

Murry Salby responds to the attacks on his record

Blockbuster: Planetary temperature controls CO2 levels — not humans