What's wrong with this graph?

A regular here recently offered us this PDF from Willie Soon and Lord Monkton's Science and Public Policy Institute and asks what is wrong with the graph in there.

i-35688596a9dfb47c970ddb0ca104130d-10yr-CO2-temp-thumb-500x339-48295.jpg

As it happens, Michael Tobis has already taken a look at another very similar construction and identified three deceptions (he kindly called "bugs") in what is technically correct data.

  1. different smoothing is used on the two types of data. Temperature is presented in monthly mean whereas CO2 looks to have had the seasonal rise and fall removed. This gives the impression of a steady rise in CO2 in stark contrast to the jitters of temperature.
  2. the choice of vertical scale dramatically exaggerates the change in CO2. In the last hundred years, CO2 has risen 100ppm, temperature about .8 oC yet this graph puts the same .8 oC of vertical temperature scale against only 35 ppm of vertical CO2 scale, a factor of three times inflation.
  3. a very short time frame eliminates 90% of the instrumental record and leaves us with much too little temperature data from which to determine a significant trend. Climate is defined as the 30 yr statistics of weather, you can not see a trend in it from 15 years of data.

Michael kindly provides an improved graph with none of those issues.
i-2254861c805aaebcf1845458da14a4d3-40yr-CO2-temp-thumb-500x378-48299.png

40 years of data is enough to discern the signal from the noise, the vertical scale of the graph is fair to both data sets and the smoothing is monthly for both. The lie SPPI is trying to pedal, that CO2 and temperature do not correlate, has vanished along with the emperor's clothes.

Robert Grumbine has an educational post "Does CO2 over the modern instrumental record correlate with Temperature?" Yes, it does.

More like this

TB's graph also shows the other problem; when I drew my graph that you link to, data for 2009-2010 was unavailable. SPPI is, however, willfully truncating the graph.

The eye tends to emphasize the ends of the graph. The way the graph *looks* is vastly different with the recent uptick rather than the prior downward spike ending the record. In a formal statistical analysis, the difference between the shorter and the longer curves is, of course, small.

The "global warming ended in 1998" meme looks likely to go away this year, so they have to push it for all its worth before the annual record comes out. "Have to", of course, because they are not interested in facts, but only in politics.

Timely post, see here >

Markey panel to address 'deniers' head-on

(05/04/2010)

Paul Voosen, E&E reporter

Rep. Ed Markey (D-Mass.) will seek to further restore the public credibility of climate science this Thursday by hosting several top American researchers in an explanatory hearing that, his office promises, "will address the claims of deniers head-on."

. . .

Also testifying to the panel will be another Briton, Lord Christopher Monckton, a hereditary peer in the House of Lords and prominent critic of the scientific consensus supporting anthropogenic climate change.

Schedule: The hearing is Thursday, May 6, at 9:30 a.m. in 2237 Rayburn.

Witnesses:
_______________

Lisa Graumlich, director, School of Natural Resources and the Environment, University of Arizona, and member of the "Oxburgh Inquiry" panel;

Chris Field, director, Department of Global Ecology, Carnegie Institution of Washington;

James McCarthy, professor of biological oceanography, Harvard University;

James Hurrell, senior scientist, National Center for Atmospheric Research;

Lord Christopher Monckton, chief policy adviser, Science and Public Policy Institute.

There are many, many things wrong with that graph. Let me list a few more:

- The scales are arbitrary. The CO2 Y axis goes from 355 to 390. You could easily change that to (say) 200 to 500, and the graph would look quite different.

- The series have completely different noise. Temperature experiences weather noise, which CO2 does not.

- You should look at a logarithm of the CO2 concentration.

- The effect of a change in the CO2 concentration is not instantaneous. It should be thought of as directly affecting the equilibrium temperature, which is in turn tracked by the observed temperature.

- CO2 is not the only thing that affects climate. There are other greenhouse gases, the Earth's albedo, and solar irradiance.

Thanks for the link to Robert Grumbine's article, which is great. Though I must say, the discussion on correlation and causation in the comments seems disturbingly reminiscent of the days of the tobacco wars. The anthropogenic source of the post-industrial carbon dioxide increase is, I should have thought, hardly among the more controversial findings of climate science!

By Tony Sidaway (not verified) on 04 May 2010 #permalink

Ah yes the old "that graph does not agree with my belief" syndrome "therefore it must be wrong".

Seriously, you can produce any number of graphs and tweak them to make your argument look better.

So sure lets all agree this graph is crap because really the scales can change but the data stays the same.

What does the data tell us, well lets look at the past 100 (1900 to 2000) years and see what it says.

Temps have increased by about 0.5C and most of that increase came in the first 50, CO2 has increased by about 70ppm and interestingly most of that increase came in the second 50 years, and of course as we all know since 2000 CO2 levels have increased to 390 approx since them and the temps have fallen.

So bleat all you like about the scaling, generate all the amateur graphs you like in an attempt to prop up your beliefs but still the facts wont change.

While i am on my soap box lets look at your stupid little theory, according to your theory the temps should have risen a lot more than what they have, do you then say "well maybe something is wrong with theory" no you simply invoke a magical theory that natural variation is "masking" all the heat.

Do you have any evidence to support your original theory of very strong +ve feed backs that will cause all this heat?

Do you have any evidence to support your magicical theory of natural variation is now masking the heat?

Can you identify a model that predicted a masking of said heat?

No you cannot, you do not have one shred of evidence to support any of the original and the magical theory, oh and lets not forget the other magical theory that is invoked when needed to explain the other time the temps dropped when CO2 went up.

And of course who could forget Trenberths triple magical theory to explain all the masking...sorry i mean missing heat in the oceans.

Without any evidence to support your multiple theories one can only conclude that you "believe" in these theories.

Of course there is an alternative theory that states you have over cooked your +ve feed back theory and all we will get from a doubling of CO2 is about 0.5 to 1C of warming.

When i look at any temp V CO2 graph regardless of the scaling i can see merit in the alternative theory can do the same? I mean without invoking unsupported magical theories? Well can you?

As i said you simply believe in GW, now there is nothing wrong in believing in a theory but it is another thing to defend it. How can you defend something based on belief. Ever tried to convince a religious person there is no God?

After watching Skip get his arse kicked by Max on another thread i am sure he will have plenty to say here.

Once again crakar shows how out of touch with reality he is. Just about everyone of his "facts" is wrong. I'm too lazy to point them all out but believe me, he is either very ignorant of climate science or he just likes to tell lies.

By Ian Forrester (not verified) on 04 May 2010 #permalink

"crakar = canards in a box"

Ducks come in boxes now? Well I never, will wonders of the modern age never cease!

Crakar must surely have seen the irony in the phrase "Ever tried to convince a religious person there is no God" since his idea of skepticism appears to exclude those areas of science that disagree with his ideology.

Crakar, have you read the AR4 Working Group 1 report? Perhaps that might help to convince you that the correlation is well justified on scientific grounds. Of course there's a lot more going on besides radiative forcing but the correlation is well established to be regarded as the most likely explanation for the warming since the middle of the last century.

Your alternative hypothesis could be right, of course, but it doesn't seem likely to me that you will be able to communicate that fact by simply deriding the better established (for now) hypothesis.

By Tony Sidaway (not verified) on 04 May 2010 #permalink

Coby,

I agree that the first graph is deceptive, if not meaningless, but your graph is not without its own problems.

I do think the fact that there is a correlation between CO2 and the global average temperature is just that, a fact. But I would like to point out that temperature is going to correlate well with any parameter, physical or not, that has increased linearly since the 1950's. If we examine the percentage of Asian Americans enrolled in institutions of higher education in the US, one would see a linear increase in time. I would suspect that this linear change correlates quite well the increase in global average temperatures, despite the fact that there is no reasonable physical model in which an increase in college enrollment would affect global mean temperatures.

Maybe Joseph can take a look at that one to beef up his argument concerning the meaning of the correlation between CO2 and global average temperature.

More importantly, the second graph is misleading because the two linear increases are put on top of each other. This gives the impression that the correlation is more meaningful than it may or may not be. And the scales don't seem to make sense. What is the y-axis in the second graph? I went to Michael's post and there was no information on the y-axis there either. There needs to be a little more transparency in that respect because it looks as thought the scales have been manipulated to explicitly place the two graphs on top of each other.

After following the link to the signal versus noise thread, I am a bit confused about what contributors here consider a 'signal'. Is the 'signal' the fact that both CO2 and global average temperatures show a linear increase? Or that these linear increases correlate well with one another?

If so, I think that's a rather terrible notion of 'signal'. From a signal processing perspective, one would expect to be able to pick out the seasonal modulation frequency in the atmospheric CO2 concentration from the overall global average temperature. It's a well-defined frequency and if CO2 is modeled to have a strong forcing in the overall temperature signal, finding that modulation should not be that difficult. As far as I understand it, however, this has not been done. I don't even see (a primitive signal processing technique, I admit) this modulation in the global average temperature graph Michael has provided. It seems like the modulation is lost in the noise from other climate influences.

Good math, bad math's most recent post on the issue of not knowing what one does not know could be rather informative on the use of words like 'signal' and 'noise' in this thread. Summing it up, when one doesn't fully appreciate the meaning of these ideas in this, or any other, context, one can come to some very questionable conclusions on the implications of scientific information. I think contributors here should think hard about that message when trying to understand what this and many other 'layman's blogs' present on the topic of climate change.

The most important difference between the two graphs is that latter is made with gnuplot whereas the former is made in Excel. That fact alone is enough to establish the relative credibility of the two arguments. ;)

maxwell said:

More importantly, the second graph is misleading because the two linear increases are put on top of each other. This gives the impression that the correlation is more meaningful than it may or may not be. And the scales don't seem to make sense. What is the y-axis in the second graph? I went to Michael's post and there was no information on the y-axis there either. There needs to be a little more transparency in that respect because it looks as thought the scales have been manipulated to explicitly place the two graphs on top of each other.

Have trouble with your eyesight or are you just trying to appear more knowledgeable than you really are?

The information is provided on the graph if you could just read and interpret it.

As for your silly comment about correlation between two completely unrelated parameters (temperature and Asian American enrollment) why do you deniers keep on showing how out of touch with reality you are? There is a physical reason (radiative physics of CO2 molecules) which provides evidence that the correlation between temperature and CO2 concentration is true. Did you know that and are just trying to be a disinformer?

You are pathetic in your clueless attacks on science.

By Ian Forrester (not verified) on 05 May 2010 #permalink

The only thing about Soon's graph that I feel is actually deceptive is the way the time scale is partitioned. Picking your start and stop dates makes a big difference.

The vertical scaling and data smoothing are just ways to influence the reader psychologically. If you take a quick glance at his graph, your gut tells you "no correlation, therefore no causation." Tobis' graph does the same thing by aligning two linear increases and scaling them so they line up visually. In this case, the gut feeling is "We DEFINITELY have correlation, therefore we must have causation."

By CherryBomb (not verified) on 05 May 2010 #permalink

maxwell,

Correlation does not always mean causation, correct. Nor does this post, or the signal vs noise post provide evidence for a causal link between CO2 and temperature. The focus of both posts is very narrow, this one is to demonstrate that the SPPI graph is deceptive and that the CO2 - temperature rise is correlated (a statistical statement that Bob's post carefully validates), the focus of the linked post is the reality of a rising trend in very noisy temperature data.

We agree on that, right? Now if you want evidence that the correlation is not just coincidental, this post is not the place, I can provide other links for that if you like.

As for finding a seasonal CO2 cycle signal in temperature data, I guess that is an interesting enough question but A) I doubt it is discernable as we are talking about detecting fractions of an expected annual rise of .01oC in data that can jump up to 70 times that amount in a single year, so good luck and B) there is no expectation of that tight a coupling. CO2's expected and detected influence is on the multi decadal time scale.

IAN PATRICK FORRESTER!
Go to your room!!
(parent follows child to bedroom, and stands in doorway...ARMS FOLDED)
I see you are still "Mr Ad Hominem".
Do you realise that there is only ONE sentence in your last post that is not attacking/ridiculing Maxwell??
I'd like to quote it if I may....
You said: (ahem!)
"Maxwell said:"
That's it!
That's the only sentence you wrote in that post that is not filled with your poison.
You may remember me Ian. I have called you by your full name before. (I guessed it)
Of course, knowing that you are quite obviously an extremely busy (and highly decorated) Professor of Climate Science, with lectures to get to, and field research trips to attend, I'm VERY much looking forward to your detailed analysis of Maxwell's post. (as quoted)
Please enlighten us all on what real science is??
Please Mister?
I.... think I understand what the term "peer reviewed" means.
I realise that you could never consider a "filthy denier" a peer. (I am one)
I suspect you don't consider too many of the posters on this blog as "peers". (whatever their stance)
But I beg of you, please, Mr Ian Forrester, please forget all that for just a moment and
justify your ad hominem attack on Maxwell with more than just abuse and ridicule. (pretty please?)
Surely you can lower yourself for five minutes?
(It'll only take YOU that long)
Ooooh! I'm all agog at the SCIENCE that is about to be forthcoming!
(there's gonna be graphs everyone!)

Sincerely,
Michael

Post Script... Dear reader/peer,
Am I too, guilty of an ad hominem post?
A thousand apologies.

michael, whoever you are, do you think I give a hoot about anything you think about me or what you post here? I will expose junks science and dishonesty about climate science whenever I come across it.

Get a life.

By Ian Forrester (not verified) on 05 May 2010 #permalink

Howdy Ian.
Actually, I DO think you "give a hoot" and I still await your reasoned, calculated and scientific reply to Maxwell's post.
I simply cannot abide your abusive, evasive and downright RUDE posts.
I suspect you are better than that, and just need more time.
(I myself am both busy and slightly lazy too!)
...
me... agog.... still...

michael, it is not rude to expose dishonesty and just down right falsification of data such as maxwell and other deniers continue to flood the science blogs with.

You should read up on what "ad hominem" actually means. You will find that I do not use "ad hominems" but merely report on the true behaviour of deniers. Too bad that certain people cannot abide the truth about themselves being exposed.

By Ian Forrester (not verified) on 05 May 2010 #permalink

Good grief, Michael.

Max tries to sound clever with correlation-does-not-equal-causation (as if no one thought of that) and you think the respondents are avoiding him.

My God.

Ian,

'There is a physical reason (radiative physics of CO2 molecules) which provides evidence that the correlation between temperature and CO2 concentration is true.'

Correlation has nothing to do with physical reasons or radiative physics. Many models of correlation between physical parameters have been built in the past (most notably, local causation in quantum mechanics) that have been proven wrong. So because we can propose a physical link does not imply that there will be a correlation in reality.

Correlation is a statistical idea that quantifies the amount that two trends mirror each other. Therefore, there is nothing really silly about pointing out that global average temperature will correlate well with any linearly increasing parameter. This is especially important to point when Coby cites a graph that makes it look as though CO2 and global average temps are linked by more than simple correlation.

'I will expose junks science and dishonesty about climate science whenever I come across it.'

So expose what I said as junk. All you've done is call me names without even providing a shred of evidence that what I call out in this post isn't meaningful. It's just...well, I'm not sure that what you've written is really anything.

'You are pathetic in your clueless attacks on science.'

I guess that's why you didn't provide any scientific rebuttal on these 'attacks'? Maybe you should sit down for a few minutes and just take deep breathes when you don't know what to do besides attempt to insult someone's intelligence. It'll help bring down your blood pressure.

Coby,

'...B) there is no expectation of that tight a coupling. CO2's expected and detected influence is on the multi decadal time scale.'

I was under the impression that the vast majority of warming since the turn of the 20th century was due to CO2 forcing. If this is true, one should be able to process the temperature signal to find that seasonal frequency since it is rather monotonic. More importantly, where is the detected CO2 influence multi-decadal temperature signal? In the linear rise of temperature? If so, that's not really a 'detected influence' as much as a correlation. Even you stated that correlation does not equate causation. So how do we detect the influence of CO2?

Again, I think there is some confusion over the notion of 'signal' and how it's getting used here and the post you linked to above.

A serious question to everyone though, why don't we see the seasonal modulation of CO2 concentration increase its peak amplitude as time goes on? If the peak amplitude increased, it would be much easier to detect the CO2 influence in the temperature signal.

@maxwell: Just trying to get my head round what you're saying. It appears that your statements boil down to:

1) Correlation does not imply causation (i.e. CO2 correlates with temp but that may be an artefact).

2) Because we can propose a physical link does not imply that there will be a correlation in reality (i.e. the physics that state that CO2 will have a +ve effect on temp does not imply they will correlate)

Is this more or less correct?

To parse the two statements: Just because CO2 should affect temp (according to physics) it doesn't mean that the correlation shown between the two is anything other than coincidence.

Does that sound right?

A serious question to everyone though, why don't we see the seasonal modulation of CO2 concentration increase its peak amplitude as time goes on?

Ok, I'll be the bonehead that bites.

Why would we expect that? If by amplitude you mean the increase from the individual concentration lows then I don't see why that should change. Is there a reason it should?

If you mean the changing *absolute* concentration seen at each peak, then it looks they are going steadily up to me.

If the peak amplitude increased, it would be much easier to detect the CO2 influence in the temperature signal.

Again, I don't understand why you're saying this, either.

But I would like to point out that temperature is going to correlate well with any parameter, physical or not, that has increased linearly since the 1950's

That's called a spurious correlation. The correlation between temperatures and CO2 is simply not spurious. There are ways to tell.

@ Joseph:

It may help to elucidate on how to tell that the correlation is not spurious - "there are ways" is a little woolly.

"A serious question to everyone though, why don't we see the seasonal modulation of CO2 concentration increase its peak amplitude as time goes on? If the peak amplitude increased, it would be much easier to detect the CO2 influence in the temperature signal." - Maxwell

Like Skip, I'm waiting to hear your reasoning on this.

By Dappledwater (not verified) on 05 May 2010 #permalink

It may help to elucidate on how to tell that the correlation is not spurious - "there are ways" is a little woolly.

What I've suggested is to detrend the series and then compare them. This is fairly easy to do. You can do it with a spreadsheet. The last post I've written on this is here.

There are more sophisticated/complicated analyses you could use, which I'm honestly not too familiar with, e.g. a Granger causality analysis.

I try for the poor man's criteria: Is there "support for" the causal hypothesis if the three key elements of causation (time order, correlation, ruling out of obvious lurking variables) are established.

I'll leave scientific "certainty" to gods, philosophers, and drunks--and not necessarily in that order. I assume I'm not the first, but freely adopt the identity of the last two in concert.

Chris S,

'2) Because we can propose a physical link does not imply that there will be a correlation in reality (i.e. the physics that state that CO2 will have a +ve effect on temp does not imply they will correlate)"

That's correct. I was not making this point in the particular case of CO2 forced warming, but rather a general statement in reference to Ian's assertion that because we have a physical model, a correlation should be expected. I think this reasoning is essentially flawed.

Joseph,

correlating a rise in temperature to non-physical parameters is definitely spurious. I was trying to point this fact out to illustrate that having correlation is not as meaningful as the second graph of this post seems to make it seem.

Statistical tools (I've never heard of the Granger causality analysis) can be very useful for understanding data, but it seems to me that a well designed experiment will always beat them in terms of providing a clear physical understanding of the problem at hand, whatever it may be.

Skippy and DW,

my reasoning on the issue of increased modulation depth is muddy to say the least. I can't really imagine a model for why or how this would happen. That the depth of modulation would increase is really just hope in the face of a very challenging signal processing problem.

I was asking the contributors here to see if anyone had heard anything about this. It's interesting to me.

Cheers.

maxwell said:

a general statement in reference to Ian's assertion that because we have a physical model, a correlation should be expected. I think this reasoning is essentially flawed.

Well, I think that your reasoning is essentially flawed. We do not have a model (though the effects can be modeled) we have a system obeying physical laws. Why do you deniers insist on confusing the concept of physical laws and models? Why do you try and mislead everyone with everything you say?

You are a pathetic lot, you deniers. You should go and read how science, as opposed to politicking, is actually done, you may learn some thing.

By Ian Forrester (not verified) on 05 May 2010 #permalink

I remember looking at a school book (if i find source will supply) where they showed a graph of CO2 and Temp together. The graph was drawn in a way that would show CO2 lagged temp if reading from left to right. Unfortunately the time scale on the bottom read from right to left therefore giving the young impressionable minds the idea that temp lags CO2. Now this could have been a extremely huge mistake or it could have been fraud.

Coby to label the SPPI graph as a fraud is a huge over statement. To lable the graph as incorrect is also an inaccurate statement as the data, that is CO2 and temp measurements correspond to the years displayed etc are correct, yes? Complain all you like about the scaling but the intent of the graph was to provoke some into using logic and common sense rather than clinging to magical unsupported theories.

I see the following few posts were once again full of name calling but not much substance, i see Ian has come up with a new way to avoid having to address the magical theory issue by claiming he is too lazy to do so.

Skip,

Here is my laymans response to your laymans questions post 31.

1, Time order, i assume here you mean time lag? If so and using Coby's graph it would appear as though CO2 is driving the temps at an almost instantaneous response (no time lag) of course if you look at the two in very small windows of time you are bound to get a correlation of sorts, pure luck will produce this. However if you look over longer time frames it is easy (for some) to see this time order shows the opposite. In fact at times there seems to be no correlation at all.

2, As suggested above if any correlation could be derived over longer time frames it would suggest CO2 responds to temp not drives it, also where there is no correlation suggests CO2 is not a primary driver and other factors play a larger role.

3, Lurking variables, this is probably the most important aspect as what are the variables that lurk? What is the cause of the divergence of the past decade? The current theory is

A, Natural variation

Now superseded by

B, Equipment "out of cal" or
C, Data processing or
D, Heat in a place we cannot look.

But this is where it all falls down because we cannot explain nor predict such things. Therefore if we cannot explain why at times (such as the last decade)there is a lack of correlation then how do we know when there is a correlation it is nothing but coincedence.

maxwell,

I was under the impression that the vast majority of warming since the turn of the 20th century was due to CO2 forcing.

"vast majority" is an overstatement, but CO2 is definately the largest factor in an almost entirely anthropogenic warming trend.

http://www.ipcc.ch/publications_and_data/ar4/wg1/en/faq-2-1-figure-2.ht…

If this is true, one should be able to process the temperature signal to find that seasonal frequency since it is rather monotonic.

I don't see how this follows at all. We can detect a 100 year influence, therefore we should be able to detect a 6 month influence? Why? 6 months of increasing CO2 should cause ~.007oC rise in temperature on average. The noise is 100x stronger than that signal.

More importantly, where is the detected CO2 influence multi-decadal temperature signal? In the linear rise of temperature? If so, that's not really a 'detected influence' as much as a correlation. Even you stated that correlation does not equate causation. So how do we detect the influence of CO2?

You might try the IPCC AR4 chapter on detection and attribution: http://www.ipcc.ch/publications_and_data/ar4/wg1/en/ch9.html

maxwell:

"That's correct. I was not making this point in the particular case of CO2 forced warming, but rather a general statement in reference to Ian's assertion that because we have a physical model, a correlation should be expected. I think this reasoning is essentially flawed."

I think we can agree on this (though I'd like to see you produce some evidence for the claim). However it seems that you are using this reasoning to dismiss the correlation that is shown.

I'll repeat the parsed statement that I gave above:

"Just because CO2 should affect temp (according to physics) it doesn't mean that the correlation shown between the two is anything other than coincidence."

Is this a fair representation of your argument?

@ Crakar:

You need to read Darrell Huff's classic "How to Lie With Statistics".

"the intent of the graph was to provoke some into using logic and common sense"... Quite the opposite I fear.

Maxwell 13:
You're correct that any two generally increasing curves will correlate well. Though, given how cold the 70s were, it's surprising to hear you describe the last 50 years as steadily increasing temperatures. Still, the general trend certainly was up.

As my post was aimed at the very elementary question, one which I see fairly routinely gotten wrong, of whether temperature and CO2 correlate (they do, quite well) I didn't pursue more subtle questions. It has been done, both in an elementary way on a blog and with heavy duty methods in the professional literature.

The blog approach, mentioned by a commentor to my post, is to first detrend both the CO2 and temperature series and look at the correlation of the deviations. Still quite simple, and addresses your concern about the correlation being merely a matter of two generally increasing series. The correlation is still quite high. See
http://residualanalysis.blogspot.com/2009/12/statistical-proof-of-anthropogenic.html for the full discussion of his results. This doesn't address specifically the seasonal cycle of CO2, which requires more advanced methods.

To get to the heavy duty methods, see David J. Thomson's Dependence of global temperatures on atmospheric CO2 and solar irradiance, PNAS, 94, 8370-8377, 1997. Of course also follow up the citations therein, especially to his 1995 paper in Science. And read the papers which cite to those two. As a nutshell version for blog purposes here, I'll retype the abstract of his 1997 paper:

"ABSTRACT Changes in global average temperatures and of the seasonal cycle are strongly coupled to the concentration of atmospheric CO2. I estimate transfer functions from changes in atmospheric CO2 and from changes in solar irradiance to hemispheric temperatures that have been corrected for the effects of precession. They show that changes from CO2 over the last century are about three times larger than those from changes in solar irradiance. The increase in global average temperature during the last century is at least 20 times the SD of the residual temperature series left when the effects of CO2 and changes in solar irradiance are subtracted."

Statistical tools (I've never heard of the Granger causality analysis) can be very useful for understanding data, but it seems to me that a well designed experiment will always beat them in terms of providing a clear physical understanding of the problem at hand, whatever it may be.

@maxwell: But you do realize that there are well-designed experiments (and very precise data, actually) showing that CO2 and other greenhouse gases absorb electromagnetic radiation at certain wavelengths but not others, right? This is known since at least the mid 1800s.

Check out the HITRAN database, or spectracalc.com, or the MODTRAN tool.

Chris S,

'I think we can agree on this (though I'd like to see you produce some evidence for the claim). However it seems that you are using this reasoning to dismiss the correlation that is shown.'

I'm not sure which 'claim' you are referring in this comment. Could you please be more specific? That will help me produce some evidence for it.

Also, if you read my very first comment, I stated in no uncertain terms that I see the correlation between CO2 and global mean temperatures as fact. So I don't know how one can come to the conclusion I am trying to dismiss this correlation. If it seems I am saying that, I apologize for the confusion. I'm not trying to dismiss the correlation.

'"Just because CO2 should affect temp (according to physics) it doesn't mean that the correlation shown between the two is anything other than coincidence."

Is this a fair representation of your argument? '

No, it's not a fair representation. I think there is definitely an influence of CO2 (or any other greenhouse gas) on the climate, global mean temperature in this case. The question remains, however, how much of an influence?

Ian's point considered the idea that since we had a model for how the emitted terrestrial radiation interacts with greenhouse gases, we should expect a correlation. I merely pointed out that a physical model is not sufficient for a correlation to appear in reality, and tried to use non-local quantum effects as evidence of this fact.

When researchers first tried to measure these non-local effects as predicted by Bell's theorem, they all believed that Einstein's local theory of relativity would prevail as correct. They, however, were incorrect and Bell's theorem is now considered the more appropriate model for how reality behaves.

That's the larger point I am making. Correlation in measured data reflects something about reality, while a physical model may or may not reflect something about reality. One must fit the model to data.

In the case of greenhouse gas (CO2 specifically) forced warming, I think that the data fits the model to a degree that allows to see that there is an influence. Again, the question remains as to what amount of the historical temperature trend is due to increased CO2 concentrations.

My problems with the second graph don't have to do with whether or not there is correlation, but more to do with the fact that the relative scales aren't present and whoever made the graph did so in a way that makes it seems as though there is a one-to-one relationship between CO2 concentration and global mean temperature, which is not supported by data. Such a quantifiable relationship is still unknown and will likely remain so into the near to mid-future (25 years I say to stick to something).

Ian,

it seems you're having trouble discriminating between the idea of a 'climate model', which you've been told is closer to a computer simulation, and a 'physical model' which uses laws to predict how a particular system will behave and is used to test against experiments. The idea that sunlight comes to the earth, causes heating, the earth emits radiation and greenhouse gases interact with that emitted radiation is a model. No bones about it. It's not 'law'.

There are equations like those of James Clerk Maxwell, Lambert, Stefan and Boltzmann which are implemented into the physical model and, as you point out, we can run computer simulations of the physical model, but it's a model nonetheless.

You may want to consider the idea that what you know about the world isn't all there is to know. That maybe, just maybe, you don't know everything or that your way of seeing things isn't the only 'correct' way of interpreting what's happening around you. If you're just going to close yourself off from increasing your understanding of the world, then you're really missing out.

Coby,

'I don't see how this follows at all. We can detect a 100 year influence, therefore we should be able to detect a 6 month influence? Why? 6 months of increasing CO2 should cause ~.007oC rise in temperature on average. The noise is 100x stronger than that signal.'

Again, I think there is an issue concerning the use of the word 'signal'.

I agree that simply looking that the linear increase over some small time period is not likely to be helpful. One needs a large N to properly do this. Looking at the zero frequency component in the temperature series due to CO2 will be excessively noisy, as it is in many experiments.

I am saying that if process the temperature signal and look for contributions that are evolving at the same frequency as the seasonal changes in CO2 concentrations, there may be something. I'm not aware of anyone trying this though I haven't looked that hard.

If CO2 is the 'largest' contribution to the change in temperature we've measured, then there should be at least as strong modulation at the frequency of seasonal changes in CO2 as there is in the measured CO2 signal itself. That's a rather easy conclusion to make. If other contributions to the temperature signal are swamping the CO2 contribution (noise in this context), then it's hard to make the argument that CO2 is really that significant of a contributor to the overall signal, in my opinion.

I have studied Chap. 9 of the AR4, as well as other sources on the topic, and have come away with one important conclusion. Since we didn't monitor the climate before CO2 concentrations increased, it's very hard to find its contribution to the overall signal. That is, there was no 'control' measurement made of the climate system that allows us to test for the causal link between global mean temperature and CO2 concentration. Researchers, Hergel I think was the lead author on the section, are doing what they can, but it doesn't seem as though they have 'detected' any influence to this point.

Joseph,

'But you do realize that there are well-designed experiments (and very precise data, actually) showing that CO2 and other greenhouse gases absorb electromagnetic radiation at certain wavelengths but not others, right? This is known since at least the mid 1800s.'

Not only do I realize this fact, I do these types of measurements on a daily basis as well as constructing models that help us predict where we will and will not see changes in the interactions of molecules with electromagnetic fields.

My experiments, and those of the past, are not that informative, however, when trying to parse out the causal effects of atmospheric CO2 on the climate. In the vast majority of the cases, the light used to do such experiments is highly controlled versus the unpredictable nature of sunlight and solar ejections which causes a variety of problems ranging from low signal to noise, as has been pointed out already, to anomalous changes in the energy budget of the earth.

More than that, all atmospheric dynamics, every single one, can be neglected when a spectroscopic experiment is carried out in the lab. There is no Coriolis force, no Hadley cell, no jet stream and no biosphere (therefore, no people).

I do think that these experiments inform us enough to know that there is a relationship between the concentration of greenhouse gases and the amount of emitted terrestrial radiation that passing back to outer space, which does in turn affect the energy budget of the earth. When it comes to detecting a CO2 contribution in the global mean temperature signal, however, the point group and normal modes of CO2 or other greenhouse gases are not going to be very useful. That was the context in which I discussing experiments versus statistical 'tricks' to find causal relationships.

I think you knew that though. You're a smart guy.

NIST Chemistry Webbook is also a great resource for spectroscopic and other molecular data.

Cheers

maxwell, just want to make sure you dont miss the references Robert Grumbine provided in comment #38, was which stuck in moderation while the thread moved past. If I read it right, the signal you are asking about can be discerned.

If CO2 is the 'largest' contribution to the change in temperature we've measured, then there should be at least as strong modulation at the frequency of seasonal changes in CO2 as there is in the measured CO2 signal itself. That's a rather easy conclusion to make.

I still don't understand why you think this, Max.

If other contributions to the temperature signal are swamping the CO2 contribution (noise in this context), then it's hard to make the argument that CO2 is really that significant of a contributor to the overall signal, in my opinion.

Unless its effects are *long term* and the noise is just short term impacts. I'm struggling to see why this is such a stumbling block, Max.

Reno weather provides a perfect analogy. The "signal" caused by the Northern Hemisphere's gradual tilt toward the sun is causing a gradual warming. It will eventually hit 100+ in August. But in the mean time we have *wild* fluctuations in daily temperature. It was 29F when I rode in this morning, wrapped in winter body armor, whereas just yesterday I was in a t-shirt and shorts! Its the chaotic weather systems that come off the pacific and into the Sierras. But it doesn't change the fact that the overall trend until about September will be warming.

The tilting of the Earth on its axis over the course of months is like CO2 over the course of decades. The daily weather fluctuations are like the ENSO cycles, etc.

Max, I am sorry, but by your logic a flat earth advocate could say, in response to my argument about the "long" term effect of axial tilting of a round planet, "Well, if the the signal is so strong then why don't you see a stronger short term correlation between tilting and temperature? Seems like there's a lot of other stuff going on here that you aren't accounting for."

Argh.

maxwell is a typical denier throwing confusion about everything all over the place. He uses big words to try and convince people that he knows everything there is to know about radiation physics, climate science and in fact all there is to know about science.

This is complete nonsense. The fact that CO2 causes an increase in temperature is not a "model" it is known to be true from theoretical physics and it has been shown to be true by careful measurements, you know, the stuff that scientists do all the time.

Get over it maxwell, CO2 causes warming of the earth, increasing levels of CO2 concentration cause increasing temperature rise. Anthropogenic emissions of CO2 are increasing and causing the measured increase in CO2 in the atmosphere. All of these are carefully measured and are not models.

Get over it, AGW is real, is causing problems now and will cause increasingly more serious problems in the near term future. You are just using your "science" as a reason to convince politicians to do nothing. That is why you are a denier.

By Ian Forrester (not verified) on 06 May 2010 #permalink

While maxwell take some time to read the links I gave, I'll add a few comments here as it seems he hasn't been able to explain clearly enough for most people what he wants. In saying that, I'm assuming that I actually do understand what he wants, which may be untrue. Still, what I understand him to mean is something interesting.

Start with the facts that CO2 is a greenhouse gas and it has risen by about 110 ppm since 1880. Given those two, we expect there to be a warming. A warming is indeed observed, so you also find a good correlation between the two (hence my post that Coby linked to).

It's also true, though, that CO2 varies during the course of a single year, by 10-15 ppm. If we expect the 110 ppm to cause a temperature change in the earth, we'd expect the 10-15 also to cause a temperature change in the course of a year. We'd expect something rather smaller, of course, which makes detecting its effect more challenging. But, since that 10-15 ppm cycle was occurring before anthropogenic sources were important, it is already reflected in the climatological observations. (This cycle is due to growth and decay of plants in the northern hemisphere through the year -- much more land in the north, so much more effect globally following the norther hemisphere cycle.)

Since that 10-15 ppm is already in the climatology, the signs of anthropogenic influence on CO2, and thence to temperature, would come from the changes in the magnitude of the seasonal cycle and its timing. If I recall correctly, the amplitude change is 1-2 ppm, and the timing of the maximum/minimum CO2 levels has changed by only a few days.

So in looking for the effects of anthropogenic changes on the CO2 in the seasonal cycle of temperature, you're looking for something about 1% the size of the total CO2-induced climate change of the last 130 years, and a phase shift (the timing of the seasonal maximum/minimum) of less than 1% of the period. I expected intuitively that this would be undetectable without a lot more than 100 years of data. But then (about the time they came out) I read Thomson's papers. And he showed how it is that the effects are indeed detectable. And they're in the direction expected by simple consideration of the fact that CO2 is a greenhouse gas.

That does help.

Thanks R.

Indeed, a sweet and simple explanation of the magnitude of the effect. Like you, I'm surprised those 1% effects can be teased out of the data and I'm impressed Thomson did it. My knowledge of stats is fairly basic, so I couldn't really follow the time-series stat discussion therein, unfortunately.

When it comes to detecting a CO2 contribution in the global mean temperature signal, however, the point group and normal modes of CO2 or other greenhouse gases are not going to be very useful. That was the context in which I discussing experiments versus statistical 'tricks' to find causal relationships.

@maxwell: Are you asserting that the normal mode of a CO2 molecule in 'nature' is different to what it is in the lab, so its infrared absorption properties would also be different? Do you have evidence of this if so?

Are you also saying that the point group of CO2 is different in nature than in the lab?

"The tilting of the Earth on its axis over the course of months is like CO2 over the course of decades. The daily weather fluctuations are like the ENSO cycles, etc" - Skip

Errr Skip, might be the way you phrased it, but that's not right, Axial tilt variation occurs over long geologic timeframes, but the seasonal variation is due to the tilt relative to Earth's plane of orbit around the sun:

http://www.youtube.com/watch?v=DuiQvPLWziQ

By Dappledwater (not verified) on 06 May 2010 #permalink

oh. right.

I'd make a good AGW denier . . .

Ian Forrester @43 - Correct, but I, likewise, found for once he posed an interesting question.

Rob Grumbine @38 - Thanks for the link to the paper, I'll have a read.

By Dappledwater (not verified) on 06 May 2010 #permalink

Obama man,

You got the wrong website by mistake, suggest you go to whatreallyhappened.com

Back to reality,

Firstly to Chris, yes i agree stats can and are used to make ones case look a lot better than it really is. The original intent of the graph was to highlight a rising CO2 whilst temps have stagnated. Coby then used the graph to suggest a denier (me) was trying to pull the wool over everyones eyes. To be honest the scaling issue never crossed my mind as my intent was to merely remind all that temps are still not playing the game.

Skip,

Thats OK i understood what you meant, you do raise a good point in that if the temps/weather vary in one day or from one to the next then short term observations may be useless.

Max,

Are raising two points here? From a standpoint of CO2 being the major climate driver then we should see (a) seasonal changes in temp as we do in CO2 and (b) a higher temp as CO2 increases.

I am not sure about (a) i tend to agree with Skip and others on short term stuff (seasonal) but with (b) i do agree which was the whole point of the graph as i said above.

After reading all above posts (and all were good) is it fair to say that we are indeed measuring/seeing the warming from increased CO2? however can we see any warming from the theoretical +ve feed backs?

A follow up question to Joseph's (47), i read a study (cant remember which) where they tried to do GHG experiments and yes lab conditions are not the same as real world. Having said that i seem to recall CO2 to be a stable molecule so maybe not much would change however H20 was a different story as the molecule would react differently under different conditions which of course made it for less predictable in the real world, hopefully someone out there can shed a lot more light (pun intended) on the subject than i.

@maxwell

"I'm not sure which 'claim' you are referring in this comment. Could you please be more specific? That will help me produce some evidence for it."

It's OK, you go on to give an example in Bell's theorem.

Also, if you read my very first comment, I stated in no uncertain terms that I see the correlation between CO2 and global mean temperatures as fact. So I don't know how one can come to the conclusion I am trying to dismiss this correlation. If it seems I am saying that, I apologize for the confusion. I'm not trying to dismiss the correlation.

It's primarily the statement "But I would like to point out that temperature is going to correlate well with any parameter, physical or not, that has increased linearly since the 1950's." And others that are similar

"The question remains, however, how much of an influence [of CO2]?

That's the larger point I am making. Correlation in measured data reflects something about reality, while a physical model may or may not reflect something about reality. One must fit the model to data.

In the case of greenhouse gas (CO2 specifically) forced warming, I think that the data fits the model to a degree that allows to see that there is an influence. Again, the question remains as to what amount of the historical temperature trend is due to increased CO2 concentrations."

A question that, in the broad sense remains to be answered. The IPCC has a range of sensitivity estimates, some say the range is set too high, others too low.

"My problems with the second graph don't have to do with whether or not there is correlation, but more to do with the fact that the relative scales aren't present and whoever made the graph did so in a way that makes it seems as though there is a one-to-one relationship between CO2 concentration and global mean temperature, which is not supported by data. Such a quantifiable relationship is still unknown and will likely remain so into the near to mid-future (25 years I say to stick to something)."

It's a woodfortrees graph with CO2 offset by -340. I agree that, to some extent, this forms a false impression but the first graph perform the same trick (overlaying the two lines in the same way) and the second graph illustrates the (to me greater) falsehood that is the truncation of the time series. Regarding this quantifiable relationship, the trouble with waiting (say) 25 years to firm up the science before acting will mean, should the higher range of sensitivity estimates apply, that the task will be a huge degree harder than if we act now. Of course if the sensitivity turns out to be at the low end of the estimates then the task will still be much larger in 25 years than it is already.

crakar:
"Coby then used the graph to suggest a denier (me) was trying to pull the wool over everyones eyes."

Make no mistake here, crakar, I think of you as a victim, not the perptrator.

Dr. Grumbine,

thanks for the link and paper. It's looks like a very interesting analysis. It'll take me some time to come to grips with the math.

I do find it interesting that given the import of this paper it was not cited in chapter 9 of the AR4. I would imagine that a paper that finds solar and CO2 forcings accounting for almost 90% of the observed variation in the global mean temperature would appear in the chapter on the 'detected' CO2 signal of the most comprehensive report on climate change put together by the scientific community.

We'll see what I find on it.

Thank you though. I want you to know that after recommending Washington and Parkinson, I now have it on my night stand. Thanks for that too.

Joseph,

'Are you asserting that the normal mode of a CO2 molecule in 'nature' is different to what it is in the lab, so its infrared absorption properties would also be different?'

Umm, no. I am asserting that even though the normal modes of CO2 are the same in the atmosphere as they are in the lab, as well as any other parameter that we can determine in the lab, knowing that information doesn't really help find the CO2 signal in the historical temperature time series.

It was that context in which I was discussing experiments versus statistical methods.

Information on the normal modes or point group of CO2 is important in formulating a simple physical model, which even Arrenhius could do (without quantum mechanics or group theory) over 100 years ago. This information, however, did not make it any easier for him to quantify the influence of CO2 in climate.

Chris S,

'It's primarily the statement "But I would like to point out that temperature is going to correlate well with any parameter, physical or not, that has increased linearly since the 1950's." And others that are similar.'

I don't understand what about this statement invalidates the notion that I think the correlation between temperature and CO2 concentrations is fact. I was pointing out that correlation is meaningful to a certain point.

'Regarding this quantifiable relationship, the trouble with waiting (say) 25 years to firm up the science before acting will mean, should the higher range of sensitivity estimates apply, that the task will be a huge degree harder than if we act now.'

How are we quantifying the 'degree' of difficulty in dealing with human emitted CO2? How is implementing changes to the way we produce energy, transport ourselves and produce goods to maintain our standard of living going to be harder in the future than it already is?

This is the argument I don't really understand. You raise concerns over my pointing out specifics about correlations and then basically speculate on what will or will not be easier or harder in future based on totally qualitative measures.

As I told skippy, when it comes to political action, it becomes a matter of opinion. In my opinion, we don't need more action than more research and the same slower progress toward more efficient consumption of resources. You might disagree, but that's simply your opinion. I think it's hard to make a definitive argument for or against 'action' because it just becomes a matter of opinion at some point.

Cheers.

Just out of interest.

When i look at this graph i dont see a lot of correlation, could it be that either our measurements that far back are not very accurate or is it that there was simply was no correlation or i cant read a graph :-))

Look under "Similarities with our Present World" heading

http://www.geocraft.com/WVFossils/Carboniferous_climate.html

Maxwell @55

The IPCC reports focus on what has been learned since the last report, rather than on a complete survey of everything ever learned. For all that many papers are cited, not all of them are. Not even close. And not even all the papers that I would want to see cited if I were writing a given chapter.

To take an easy example, the variations in the earth's orbit are discussed. The oldest paper on the topic that gets cited is Berger, 1978. While that's now old, the idea (as you can get an idea from the title of Berger, 1988) comes from Milankovitch, 1941. And if you were more interested in the topic, you'd also read Milankovitch's 1930 and 192x works, and Croll and Adhemar from the 1800s, none of which are cited, but all of which are important in understanding the history of the idea of the influence of orbital variations on climate. Scientists tend not to be historians; I have some leanings that way is all.

I also was not trying to give you busy work in suggesting that you read the papers that cite Thomson's 1997 and 1995. It could be that later work has found an error in what he did. I doubt it, but it could have happened.

I'll mention too that over at my blog I hang out a post from time to time specifically for questions to be asked. They're always open. The most recent is February's 'Dare to ask a question!' I don't always have answers, and sometimes it takes a while to get one when I do. But usually I or a reader can shed some light.

Dr. Grumbine,

'It could be that later work has found an error in what he did.'

From what I have read, it seems that researchers now lean more strongly on sulfate aerosols as a component in the historical time series of temperature. In Thomson's analysis he finds that he can get almost 90% of variation in temperature from just solar and CO2 concentration changes. I think that if there is an error, it is likely in that physical model.

Also, when GISS put together their overview of climate model simulation results in 2002, internal variability was considered as a 'source' for some of the warming. I haven't thoroughly read that monster of a paper, so I'm not sure how much weight they gave to such internal factors causing warming.

'To take an easy example, the variations in the earth's orbit are discussed. The oldest paper on the topic that gets cited is Berger, 1978. While that's now old, the idea (as you can get an idea from the title of Berger, 1988) comes from Milankovitch, 1941. And if you were more interested in the topic, you'd also read Milankovitch's 1930 and 192x works, and Croll and Adhemar from the 1800s, none of which are cited, but all of which are important in understanding the history of the idea of the influence of orbital variations on climate. Scientists tend not to be historians'

I think this is not really pertinent to the notion that Thomson's paper is 'old' and therefore not worth citing.

Milankovitch's (and I would imagine older work as well) is now part of the standard canon of climate science. I'm sure the important points and contributions that older work made are now in text books that the current generation of climate scientists used in training as students. In many case, I'm sure those text books get cited in place of older work.

For the same reason, when I write a paper on molecular physics, I'm not going to cite Niels Bohr for the idea of quantized electronic orbitals. If one doesn't know he contributed that, then they likely won't get much out of the paper in the first place.

Thomson's paper, from just 13 years ago, was in a high profile journal for a conference called 'Carbon Dioxide and Climate Change'. Not only that, when checking the references, I found a 'Thompson' who was cited three times with papers dating back to 1998. That's just as old as Thomson's paper.

Again, I have not read the paper thoroughly so I'm not sure what the issue is. I do know that using measures like times cited, publication merit and institution can be meaningful.
In the case of climate science, I don't see why we wouldn't want to see if the most high profile scientific organization had used a particular publication (especially from as revered a journal as PNAS) to come to their conclusions.

Thanks for your comments.

Cheers.

maxwell and crackar, please review this linked video featuring Dr. Richard Alley entitled "The Biggest Control Knob: Carbon Dioxide in Earth's Climate History"

http://www.agu.org/meetings/fm09/lectures/lecture_videos/A23A.shtml

It will correct several misconceptions the pair of you seem to have.

By the way maxwell your continued spouting of such spurious term as "point group" and "normal modes" really are showing that you are trying to blind us with pseudo-science when you say (yes I know you later denied it) that CO2 in a lab reacts differently from CO2 in the atmosphere. You get full marks for obfuscation but a zero on your chemistry test since you know (or should know) that they are meaningless in the context which you are trying to use them i.e. effects of IR on atmospheric temperature versus lab measurements. That is obfuscation at its worst since naive people will actually think that you know what you are talking about.

By Ian Forrester (not verified) on 07 May 2010 #permalink

maxwell says:"If so, I think that's a rather terrible notion of 'signal'. From a signal processing perspective, one would expect to be able to pick out the seasonal modulation frequency in the atmospheric CO2 concentration from the overall global average temperature. It's a well-defined frequency and if CO2 is modeled to have a strong forcing in the overall temperature signal, finding that modulation should not be that difficult. As far as I understand it, however, this has not been done."

The anomaly plots removes seasonal variation in temperature, so one most emphatically should not expect to correlate that with seasonal CO2 variation. You could do it, just add the mean monthly temperatures back to the data.

By t_p_hamilton (not verified) on 07 May 2010 #permalink

Umm, no. I am asserting that even though the normal modes of CO2 are the same in the atmosphere as they are in the lab, as well as any other parameter that we can determine in the lab, knowing that information doesn't really help find the CO2 signal in the historical temperature time series.

@maxwell: I see what you mean now, but it would be wrong. Knowing the absorption properties of CO2 as observed under controlled conditions is quite helpful. If you only had the instrumental temperature record to go by, and (say) a hypothesized algorithmic relationship between CO2 and equilibrium temperature, you can still come up with a model, but it won't be as accurate. That's especially true when you're considering multiple greenhouse gases, because of the curse of dimensionality.

Let me try to illustrate with an analogy (although I couldn't think of any really obvious ones.) Suppose you wanted to model sea level rise as a function of temperature. You can do it by just looking and statistically analyzing instrumental data, but it would be much better if you can start out with some fundamental knowledge about ice melt and thermal expansion derived in the lab under controlled conditions, wouldn't it?

t_p_,

the second graph presented in this post does have the seasonal variation of CO2 concentrations. It was this graph I was discussing.

Ian,

'You get full marks for obfuscation but a zero on your chemistry test since you know (or should know) that they are meaningless in the context which you are trying to use them i.e. effects of IR on atmospheric temperature versus lab measurements.'

That comment should be directed at Joseph, not me. He said,

'@maxwell: But you do realize that there are well-designed experiments (and very precise data, actually) showing that CO2 and other greenhouse gases absorb electromagnetic radiation at certain wavelengths but not others, right? This is known since at least the mid 1800s.'

bringing in the idea that the lab experiments were of importance in this discussion. I merely pointed out that knowing particular molecular parameters like a point group or normal modes do not inform us on how to find the CO2 contribution in the historical temperature time series. In was in the context of the CO2 contribution to the historical temperature time series that I discussing experiments versus statistical manipulation. He took that comment out of context, as you have done, yet again.

Furthermore,

'By the way maxwell your continued spouting of such spurious term as "point group" and "normal modes"...'

the point group and normal modes of CO2 are not spurious in the larger context of the greenhouse effect because it is the point group and normal modes (well, a combination of them) that informs us that CO2 absorbs IR light and is a greenhouse gas. Those might be ideas that some others here find interesting or want to look further into. They are interesting to me at least.

Cheers.

maxwell said:"t_p_,

the second graph presented in this post does have the seasonal variation of CO2 concentrations. It was this graph I was discussing."

Note that I said temperature anomaly does not include seasonal variations. This is critical to your point "From a signal processing perspective, one would expect to be able to pick out the seasonal modulation frequency in the atmospheric CO2 concentration from the overall global average temperature."

In other words, one graph has seasonal variation, the other has it removed. There is absolutely no reason to expect to see the seasonal variation in the temperature anomaly graph. It is trivial to add it back in.

By t_p_hamilton (not verified) on 07 May 2010 #permalink

the point group and normal modes of CO2 are not spurious in the larger context of the greenhouse effect because it is the point group and normal modes (well, a combination of them) that informs us that CO2 absorbs IR light and is a greenhouse gas.

It is spurious in the sense that there are databases that already tell us what we want to know, i.e. the absorbance of a gas at a given wavelength, temperature & pressure.

Anyone who has studied chemistry is aware of IR absorption by various molecules. maxwell's "modes" refer to the different vibrational states which are excited by the specific wave lengths. My hazy memory of this suggests that there are four modes with CO2, however, I seem to remember that only three are IR active (asymmetric stretch, vertical bend and horizontal bend; the symmetric stretch is not IR active).

This is of course immaterial to temperature increase when LWR is emitted from the surface. When the energy is taken in by the IR active molecules it doesn't matter which mode is involved. The molecules could be wagging their wings or even giving deniers the finger, what is important is the quantity of energy retained and re-emitted.

maxwell is doing a great job of muddying the waters with all his grandiose words.

The bottom line is CO2 is IR active, thus increasing concentrations result in more energy being retained in the atmosphere and thus leading to rising global temperature and climate change.

Do not get confused by maxwell's obfuscation, it looks like he may be succeeding, don't let him.

By Ian Forrester (not verified) on 07 May 2010 #permalink

The bottom line is CO2 is IR active, --Ian

Exactly.

Ian,

'maxwell is doing a great job of muddying the waters with all his grandiose words.'

I don't understand what the hell you guys are arguing about. It's like you needed something to argue about and manufactured this.

Joseph's original comment to me, in the context of my discussing climate experiments versus statistical manipulations was,

'But you do realize that there are well-designed experiments (and very precise data, actually) showing that CO2 and other greenhouse gases absorb electromagnetic radiation at certain wavelengths but not others, right? This is known since at least the mid 1800s.'

My response to him in comment #40 was,

'Not only do I realize this fact, I do these types of measurements on a daily basis as well as constructing models that help us predict where we will and will not see changes in the interactions of molecules with electromagnetic fields.

My experiments, and those of the past, are not that informative, however, when trying to parse out the causal effects of atmospheric CO2 on the climate.'

Then you said in response to notion of normal modes in particular or molecular physics in general,

'This is of course immaterial to temperature increase when LWR is emitted from the surface. When the energy is taken in by the IR active molecules it doesn't matter which mode is involved.'

which was my point the whole time! The information determined in the lab doesn't really inform our ability to do climate science other than to tell us CO2 is a greenhouse gas. I appreciate you working so hard to make my argument for me. It's good to have such partnerships in knowledge.

You guys have a nice weekend.

maxwell, stop obfuscating. You are just sowing confusion so that your morals can be spread to others. You are a delayer or as Frank calls them, inactivist. You deny that there is a need to do something about AGW.

Muddying the waters by use of grandiose words and terms may work with some people but it won't work where contributors have a reasonable knowledge of science.

All you are doing is using delaying tactics. You continually turn people's arguments around and try and confuse them. You are just playing with words, not discussing science. You are not a lawyer or politician are you? That is how they behave.

By Ian Forrester (not verified) on 07 May 2010 #permalink

IAN PATRICK FORESTER GO TO YOUR ROOM!

(It makes no sense in context but I've decided I want to do that from time time.)

"maxwell is a typical denier throwing confusion about everything all over the place"

I disagree, Ian. Maxwell's questioning comes from a place of some knowledge and much curiosity, it seems to me. With so many know-it-alls in these here debates, a genuinely skeptical, inquisitive approach is something to be cherished. I've been disappointed to see genuine enquirers alienated by people leaping to conclusions, probably edgy from the frustrations of dealing with true denialists. But I suspect in this case Maxwell isn't the skittish type.

Barry, I beg to differ but you are entitled to your opinion.

By Ian Forrester (not verified) on 07 May 2010 #permalink

Thank you for a very useful website.

You asked: what is wrong with this graph? I would respond: many things.

1) The time-series is just too short given between-year variability. Hence Phil Jone's absolutely correct but widely mis-interpreted comment about what happens when trying to calculate statistically signifant trends using noisy time-series data. You just can't do it.

2) The time-series plotted use different smoothing methods based on monthly (temperature) versus smoothed (seasonally detrended) CO2 data. That is inappropriate if one intends to use a common X-axis time scale to overlay two series. Either justify the choice of smoothing methods or just don't do it at all (e.g., see Krebs 1989).

3) The choice of left-hand and right-hand Y-axes is arbitrary. Yes, the plots look nice and they fill up the vertical space in such a way as to portray a convincing correlation to untrained eyes. In your words they "are fair to both data sets" and represents what any decent graphics designer would do. I strongly disagree with both you and Michael in this regard. You miss the essential point. One could choose any Y-axis scaling BUT that would not change the essential statistical correlation...the two variables are indeed highly correlated regardless of Y axis scaling.

4) WoodForTrees.org is a marvelous interactive resource and a useful visualization tool. But seeing more reporting of actual test results might improve the scientific literacy of viewers to this (and other) sites. Many people use Excel, which contains a minimal but useful array of statistical functions that allow calculation of simple Pearson correlations. That and a freely available set of statistical tables (google is your friend) would allow visitors to download raw data, set up a dataset, run a test, and evaluate to their own satisfaction what the results suggest.

5) In science one generally chooses not to throw away data if they might sometimes later become useful. Choosing a start date of 1979 is fine if you only wish to analyse satellite data. Choosing 1956 might be fine if you only wanted to compare the first quantitative ice measurement data (from the USS Nautilus voyage in 1958) with more modern methods. Choosing HadCrut (from 1880 versus NOAA/GISS data from 1900) also makes a difference depending on what you wish to do. But for your purposes, the Mauna Laua CO2 extend back to 1958, and that should have been the start date for these graphs.

6) That "correlation does not equal causation" hardly constitutes a "new idea"; it was discussed some time ago by scientists who were demonstrably smarter than me and who therefore preferred arguments based on "the preponderance of evidence" (e.g., Popper 1968).

"What is wrong with this graph?" is not as useful a question as "what other mechanism might explain recent observed observations on the strength of known physics, chemistry, and biological principles"? Enquiring minds would like to know...

<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
Krebs, C.J. 1989. Ecological methodology. HarperCollins (New York, NY). 654 pp.

Popper, K.R. 1968. The Logic of Scientific Discovery. Second edition. Hutchinson (London). 127 pp.

By wanderers2 (not verified) on 07 May 2010 #permalink

Gotta agree with barry on maxwell. Maxwell is clearly capable of learning the relevant science.

By t_p_hamilton (not verified) on 08 May 2010 #permalink

crakar, would you please look at the graph here and let me know if you think it has any problems. The data it is based on is accurate.

Coby,

I followed your link in post #74, and I nearly fell off my chair laughing!! Thanks for that!

At the start of the link, there are a couple of follow-on links, one of which is too the interestingly titled "The Climate Scum" website. Here is a direct link to a hilarious post.
http://theclimatescum.blogspot.com/2010/05/correlation-what-correlation…

If you don't wish to follow the link and have your brain fried by the sheer idiocy of what deniers think, here is a quote for you too all laugh (or cry) at:

"....This simple graph disproves two of the most egregarious lies of the IPCC cabal: (1) it is getting warmer and (2) the warming is due to CO2. There is clearly no correlation between temperature and CO2! The increase in CO2 is instead caused by the medieval warming period: as we know from the ice cores, CO2 lags 800 years behind temperature.....Using the current CO2 levels as a proxy for past temperatures, we can also conclude that it was about 100 degrees warmer 800 years ago, a period of great prosperity when they built the great cathedrals in Europe, Richard Lionheart (to the right) defeated the islamofascists, Chinese fleets sailed around the Arctic ocean and the Inca and Aztec civilizations were at their peaks. If 100 degrees Kevin was so good for humans 800 years ago, why should we worry about a couple of tens of degrees during the 21st centuries? Only the bedwetting ninnies and nincompoops of the IPCC watermelon (green outside, red inside) fruit sallad could object, because they hate humankind!..."

So - current warming is caused by the MWP. It was 100 degrees warmer 800 years ago (that's 100 degrees KEVIN - LMFAO!!!), and we shouldn't worry about a few tens of degrees of warming!!! And islamic terrorists were a problem even then apparently.

But of course, as moronic as this sounds, it gets worse. How about this for a quote from "DenialDepot", one of the other links on the page:
http://denialdepot.blogspot.com/2009/04/co2-levels-may-have-been-over-2…

"....The scientific consensus is that in 1200AD the climate was excruciatingly hot. Plagues and disease swamped the world and people were dying of heat exhaustion. The situation got so bad that the Vikings moved to a large snow covered island north of Iceland to escape the heat. The ice on this island subsequently melted and so the Vikings called it Greenland. Greenland soon became the wine and farming capital of the world....Didn't last though. Eventually the climate shifted back again and the Vikings were wiped out by ice...."

Of course, as they explain, the reason it was so hot in 1200AD was because CO2 levels were 5 times (2,000ppm) what they are today. I'm not sure where the evidence for this comes from, but there you have it folks!!! Proof positive that it is all a fraud - just ask your local village idiot.

And where do you think i get all my info from Mandas? The villiage idiot knows all like the base hairdresser remember?

By the way if you are going to pursue a career in comedy you will need to expand your repertoire. He are some ideas you could mull over.

GW is causing sheep in Scotland to shrink.

If you are going to make a factual documentary about AGW how do you get your message across without actually revealing any facts? You simply throw in a cartoon and some snippets from a Hollywood disaster movie, the only hard part is keeping a straight face when you accept your award.

Selling carbon trading to the world is easy but once again the only hard part is keeping a straight face when you collect the cheque(s).

How do you sell a book about increasing occurrences of cyclones/hurricanes when they have actually been decreasing? You simply put a photoshopped image of about 10 cyclones all over the face of Earth on the front cover.

Do you think Mann is a big fan of ice hockey or do you think he prefers the more girly version? I think the latter.

How do you get a photo of a poley bear floating on an ice cube where none exist? Photo shop of course.

In light of the recent Long Cao study do we now chop down all the trees?

Supplementary, do you think it would be better if we sprayed the Amazon with Agent orange rather than employ lumber jacks as the aerosols produced by the turbine engines will add to the cooling effect?

And finally all artifacts that are discovered under retreating glaciers located in areas inhospitable today which show human habitation no more than 1,000 to 2,000 years ago are placed there under clandestine operations and are all part of Lord Monktons global network of climate change denying conspirators. A network whose only concern is to ensure the destruction of the human race by any means possible.

Barry and t_p_,

thanks for the vote of confidence. It seems that most of the time, when someone questions the conclusions put forth by Coby here (or others elsewhere), this questioning automatically elicits a defensive response from like-minded contributors.

This is a very interesting scientific issue for me and I have put quite some time into trying to understand it more thoroughly.

One question to t_p_, on the issue of seasonal variations in the temperature (or anomaly), if this quantity is supposed to reflect a global mean, why doesn't the averaging over the surface of the earth get rid of the seasonal variation? That is, why does this variation have to be 'taken out'? Shouldn't the averaging over the whole earth get rid of this variation since, on average, there are no seasons?

Thanks again for your comments.

Cheers.

Dr. Inferno and Baron von Monckhofen are so good at Poe, that they apparently fooled mandas @75.

Shouldn't the averaging over the whole earth get rid of this variation since, on average, there are no seasons?

That's an interesting question. I'll go off on a tangent here, but I've looked at the seasonality of the equator in GHCN v2 (using my open source tool - GHCN Processor.) There's clear temperature seasonality at the equator, with a high in March and April, and a low in July and August. This pattern is sustained for several degrees north of the equator -- which is pretty much contrary to what you'd expect for the NH.

It's not a huge difference between seasons - maybe one degree Celsius.

My guess is that it has to do with the Earth's current eccentricity.

maxwell, the seasonal cycle in global average temperatures is mostly due to the much greater proportion of land in the NH. The oceans have much more thermal inertia so the greater area of ocean in the south causes a smaller seasonal cycle than in the north. The land area difference is also why CO2 has a seasonal cycle even though the N and S cycles are opposing.

There is a similar difference for a similar reason in the N and S anamoly, see here for example.

Interesting about the equatorial pattern though...perhaps is has more to do with rainfall and cloudiness.

maxwell asked:"One question to t_p_, on the issue of seasonal variations in the temperature (or anomaly), if this quantity is supposed to reflect a global mean, why doesn't the averaging over the surface of the earth get rid of the seasonal variation? That is, why does this variation have to be 'taken out'? Shouldn't the averaging over the whole earth get rid of this variation since, on average, there are no seasons?"

I did not know until recently that the global mean was not the same for each month. However, it is a fact, and we know what that means for a beautiful sounding idea (seasons average out).

Have you looked at an advanced undergraduate text, such as Houghton's Physics of Atmospheres? That is the one I have perused - won't pretend I have worked the problems, but then again I am not claiming skepticism about its accuracy.

By t_p_hamilton (not verified) on 10 May 2010 #permalink

I pointed this out on Grumbine's website too.

A much more appropriate variable to correlate with is e.g. net forcings. And even then you need to invoke a simple atmospheric-ocean model like the 2-box models used by Tamino, Lucia, Nick Stokes and others. In this model, there was a 1-year delay + an e.g. 30 year delay. But you still need to use total forcings not just the CO2 portion.

There is no expectation of a direct linear correlational relationship between CO2 and GMT. If you play this game long enough, you'll end up getting burned, because data are model are guaranteed to eventually diverge.