Wednesday, June 12, 2013

Murry Salby: CO2 is the integral of temperature the past, on short timescales, it has therefore fluctuated rapidly...

Honza [=Jan] U. sent me the following one-hour April 2013 talk by Prof Murry Salby of Australia's Macquarie University:

This astrophysicist and atmospheric scientist has a rather impressive publication record. At the beginning, I was a bit discouraged by Pierre Gosselin's summary that suggested that Salby was making some widespread elementary errors about the direct attribution of CO2 emissions according to their isotopic composition (the extra CO2 we see in the atmosphere generally has a very different composition than the CO2 when we emitted it, because the carbon is being quickly recycled all the time while chemistry doesn't care about the differences between isotopes but it's still true that our additions of CO2 have increased the CO2 concentration).

But I was wrong, Salby isn't doing these particular trivial mistakes and when I ultimately listened to the talk, it looked rather impressive.

He employs various types of statistical models, Fourier transformations, and other things to decode the relationships between CO2 and the temperature. Of course, the temperature is mainly the driver of CO2 and CO2 follows – to say the least, that's the dominant relationship during the glaciation cycles (the time scales from tens to hundreds of thousands of years).

Those things wouldn't be new and I wouldn't listen to another 1-hour talk that just discusses whether CO2 was the cause or the consequence during glaciation cycles. Of course it was the consequence. Whoever still acts as if he were misunderstanding these basic issues is either a hopelessly brainwashed moron or an amazingly dishonest demagogue or both.

But Salby said much more than that.

He argued that the anomalous CO2 concentration may be approximated as the integral of the anomalous temperature, \[

\Delta\,{\rm conc}(CO_2) = \alpha \int \dd t\,\Delta T

\] which sort of explains why it seems to be rising so smoothly (if we ignore the nearly periodic seasonal variations). But Salby has also presented some evidence that the ice record heavily underestimates the fluctuations of the CO2 concentration, especially the high-frequency (short-period) oscillations that occurred a long time ago. If that's true, it's pretty likely that concentrations above 400 ppm may have been rather mundane even before the industrial activity.

Using the Fourier methods, he argues that there is a phase shift of 90 degrees between the temperature and CO2 pretty much at all frequencies. I am not quite seeing how this may be true because at least in the glaciation cycles, i.e. at the 10,000-year time scale, these two quantities are pretty much in sync. How does the phase shift move to 90 degrees for shorter time scales?

And his discussion of the different isotopic composition (C12 vs C13) of the fossil fuels and the present plant life is sophisticated, not the kind of silly caricature I was led to expect. At any rate, Salby concludes that the excess CO2 is caused by the integrated or accumulated positive temperature anomaly in 1920-1940 and 1980-2000 or so and these positive anomalies may be interpreted as noise, not results of any trends.

That sounds nice except that I think it's obvious that the CO2 we have added to the atmosphere has led to some increased CO2 concentrations and the latter increase is comparable to 50% of the former (airborne fraction etc.) – it's not negligible. It doesn't matter that there are 50 times more important contributions to the CO2 atmospheric budget as well. Despite these dominant contributions, a small surplus simply can't become completely invisible.

If you were thinking whether you should listen to that talk, my recommendation is probably Yes. Despite the fact that he is trying to deny some obvious facts – if I understand the discussion about the attribution of an elevated CO2 well and if I am right about its imperfections – he is also saying lots of new things and offering many sophisticated methods that you may want to know about.

At the end, Salby offers some criticisms of the climate models that I only partly agree with. Concerning the agreeable conclusions, he says that the prevailing climate models show CO2 and the temperature essentially as the same thing; in the real world, they're not the same thing at all. These two claims – and their paramount contrast – are self-evidently true.

To mention the propositions I don't quite share, he says that theories can't ever be tested against the past data; tests of predictions of the future are always needed. I disagree with that. It's a historical coincidence whether some data were collected before a theory was written down or after that. A theory is always constructed or chosen according to the data in the past and then it gives predictions for other phenomena as well. Those phenomena may be data to be collected in the future but also additional data that may be collected about the past. Regardless of the timing, such data may be used to strengthen or weaken our confidence in the theory (or rule it out).

See also replies by MeteoLCD (more detailed review than mine!), Anthony Watts' 100+ commenters, The Hockey Schtick, Climate Depot, Tall Bloke, and – from the crazy side of the aisle – John Cook (I agree with some of the criticism) and Deltoid who calls Salby "unhelpful" (for "the cause") LOL. ;-)


  1. Luboš, you write:

    "If that's true, it's pretty likely that concentrations above 400 ppm may have been rather mundane even before the industrial activity."

    - oh yes, surely, see e.g. here: - and there is also the lag of about 5 years... Alexander

  2. "But Salby has also presented some evidence that the ice record heavily underestimates the fluctuations of the CO2 concentration, especially the high-frequency (short-period) oscillations that occurred a long time ago."

    Finally! I never found the smoothness of the paleo records too credible. There must be something in the ice cores that lowers the peaks and raises the troughs. For example, wouldn't a year with a "bumper crop" of volcanic eruptions and consequent cooling make for a rather steep dip in the temperature curve? Yet we don't see such abrupt fluctuations in the graphs.

  3. Regarding the 90 degree phase shift, this is part of what CommieBob says on WUWT:

    "All the data is simply explained using a first order low pass filter analogue. Over periods shorter than one time constant, such filters integrate. Over long periods, the inputs and outputs are directly proportional. So simple. There is nothing about the waveforms that suggests that it is worthwhile to go for a more complicated relationship. As always, one should apply Occam’s Razor."

  4. This is something that has puzzled me for a couple of years. If Arctic summer insolation is in phase with the ice melt rate and ice melt rate is a proxy for temperature, then why are the global temperature proxies 90 degrees phase shifted? The only reason I could think of is that equatorial temps are more inphase with ice volume or energy in the climate system. If that were the case then CO2 would also be more in phase with global temps as well.

  5. Eugene, the point is that the CO2 graph above is utter BS - it is not global conentration, but local.

  6. Fair and insightful answer. May I add one more silly detailed question? How does one calculate the time constant or at least that the time constant is longer than decades but shorter than thousands of years?

  7. I read somewhere (sorry, I forget where) yesterday that Salby doesn't claim that man's contribution to the CO2 increase is negligible; he says it may be as high as 20% of the increase.

  8. Analyses of ice cores indicate that CO2 lags temperature by about 200 to 800 years, so they are not in near synchronicity. Furthermore, temperature is falling when CO2 is maximal and rising when CO2 is minimal. The current rising trend of CO2 may be ocean out gassing plus human additions.

    I don't understand what Salby is doing.

  9. On a 100K yr glacial cycle, wouldn't it be that CO2 and temperature *are* in near synchronicity? The values would be moving in opposite directions for only about 1K yrs over the cycle.

    If he was using the Arctic summertime ice melt rate or the rate of change of δ18O as his temperature proxy, then they would be phase shifted as prescribed. But, IIRC the temperature reconstructions are determined by ice volume and δ18O concentration, so only a small phase shift (200-800yr) is generally accepted. So I don't understand what Salby is doing either.

  10. What is the diffusion coefficient of CO2 in ice?

  11. This is a very good point. I have often been puzzled that the cores show CO2 fluctuations. Shouldn't they diffuse out? The delta O2 doesn't both me as much as it is fixed into the ice water itself.

  12. He specifically says that, given our limited understanding of the nature of carbon dioxide sources and sinks, it is not currently possible to determine what the human contribution to net carbon dioxide emission actually is.

  13. Alex, it's from the Mauna Loa observatory. They have been measuring atmospheric CO2 content for some fifty years now. True, this content oscillates annually, with the Northern Hemisphere moving in the opposite direction to the SH. But in this Wolfram Alpha clip which I saved for you, you can choose a "deseasonalized" representation of the data.

  14. I tried to contact Salby at the email address given on his Maquarie U home page

    but it failed permanently with "The email account that you tried to reach is disabled." I wonder what that's about.

  15. "I am not quite seeing how this may be true because at least in the glaciation cycles, i.e. at the 10,000-year time scale, these two quantities are pretty much in sync. How does the phase shift move to 90 degrees for shorter time scales?"
    The 90 degree phase shift was discussed and shown to hold using modern CO2 and instrument data. This cannot be done with ice core data because the ice core data is both magnitude buffered and time shifted. But if you had the real data from that time period, the 90 degree relationship would still hold.

  16. I'd think Canada would welcome some warming.

  17. The diffusion speed is proportiotional to the Air-Ocean concentration gradiënt (Fick Law), modulated with a temperature signal, and is on annual average into the ocean, therefore the ocean is a net sink and not a source. The CO2 gradiënt is also from land to sea and from north to south, more evidence that the CO2 source is on land and not in the ocean.

  18. Dear Jason,
    this dramatic text

    shockingly clarifies the outrageous reasons why your e-mail failed to be delivered.


  19. WowDiscussionWithLubosMotlsOct 26, 2013, 11:02:00 PM

    Do you say Lubos that CO² is rising because of temperature and not the opposite?

  20. First of all, this is a review of opinions of Murry Sulby, not mine. Second, in the most important contexts where a correlation between CO2 and the temperature is actually observable, like in the glaciation cycles, yes, the temperature changes are the drivers and the CO2 concentration changes as a consequence.

    No, in the present era, the CO2 isn't changing because of any changes of the temperatures - after all, the temperatures weren't really changing for 20 years or so. CO2 is increasing in the atmosphere because we're burning carbon-based fuels. It doesn't have any negative consequences.