UPDATED – see below
Monckton provides these slides for discussion along with commentary related to his recent post on CO2 residence time – Anthony
There is about one molecule of 13C in every 100 molecules of CO2, the great majority being 12C. As CO2 concentration increases, the fraction of 13C in the atmosphere decreases – the alleged smoking gun, fingerprint or signature of anthropogenic emission: for the CO2 added by anthropogenic emissions is leaner in 13C than the atmosphere.
However, anthropogenic CO2 emissions of order 5 Gte yr–1 are two orders of magnitude smaller than natural sources and sinks of order 150 5 Gte yr–1. If some of the natural sources are also leaner in CO2 than the atmosphere, as many are, all bets are off. The decline in atmospheric CO2 may not be of anthropogenic origin after all. In truth, only one component in the CO2 budget is known with any certainty: human emission.
If the natural sources and sinks that represent 96% of the annual CO2 budget change, we do not have the observational capacity to know. However, we do not care, because what is relevant is net emission from all sources and sinks, natural as well as anthropogenic. Net emission is the sum of all sources of CO2 over a given period minus the sum of all CO2 sinks over that period, and is proportional to the growth rate in atmospheric CO2 over the period. The net emission rate controls how quickly global CO2 concentration increases.
CO2 is emitted and absorbed at the surface. In the atmosphere it is inert. It is thus well mixed, but recent observations have shown small variations in concentration, greatest in the unindustrial tropics. Since the variations in CO2 concentration are small, a record from any station will be a good guide to global CO2 concentration. The longest record is from Mauna Loa, dating back to March 1958.
The annual net emission or CO2 increment, a small residual between emissions and absorptions from all sources which averages 1.5 µatm, varies with emission and absorption, sometimes rising >100% against the mean trend, sometimes falling close to zero. Variation in human emission, at only 1 or 2% a year, is thus uncorrelated with changes in net emission, which are independent of it.
Though anthropogenic emissions increase monotonically, natural variations caused by Pinatubo (cooling) and the great el Niño (warming) are visibly stochastic. Annual changes in net CO2 emission (green, above) track surface conditions (blue: temperature and soil moisture together) with a correlation of 0.93 (0.8 for temperature alone), but surface conditions are anti-correlated with δ13C (red: below).
The circulation-dependent naturally-caused component in atmospheric CO2 concentration (blue above), derived solely from temperature and soil moisture changes, coincides with the total CO2 concentration (green). Also, the naturally-caused component in δ13C coincides with observed δ13C (below).
==============================================================
ADDED (the original MS-Word document sent by Monckton was truncated)
==============================================================
The naturally-caused component in CO2 (above: satellite temperature record in blue, CRU surface record in gray), here dependent solely on temperature, tracks not only measured but also ice-proxy concentration, though there is a ~10 µatm discrepancy in the ice-proxy era. In the models, projected temperature change (below: blue) responds near-linearly to CO2 concentration change (green).
In the real world, however, there is a poor correlation between stochastically-varying temperature change (above: blue) and monotonically-increasing CO2 concentration change (green). However, the CO2 concentration response to the time-integral of temperature (below: blue dotted line) very closely tracks the measured changes in CO2 concentration, suggesting the possibility that the former may cause the latter.
Summary
Man’s CO2 emissions are two orders of magnitude less than the natural sources and sinks of CO2. Our emissions are not the main driver of temperature change. It is the other way about.
Professor Salby’s opponents say net annual CO2 growth now at ~2 μatm yr–1 is about half of manmade emissions that should have added 4 μatm yr–1 to the air, so that natural sinks must be outweighing natural sources at present, albeit only by 2 μatm yr–1, or little more than 1% of the 150 μatm yr–1 natural CO2 exchanges in the system.
However, Fourier analysis over all sufficiently data-resolved timescales ≥2 years shows that the large variability in the annual net CO2 emission from all sources is heavily dependent upon the time-integral of absolute global mean surface temperature. CO2 concentration change is largely a consequence, not a cause, of natural temperature change.
The sharp Pinatubo-driven cooling of 1991-2 and the sharp Great-el-Nino-driven warming of 1997-8, just six years later, demonstrate the large temperature-dependence of the highly-variable annual increments in CO2 concentration. This stochastic variability is uncorrelated with the near-monotonic increase in anthropogenic CO2 emissions. Absence of correlation necessarily implies absence of causation.
Though correlation between anthropogenic emissions and annual variability in net emissions from all sources is poor, there is a close and inferentially causative correlation between variable surface conditions (chiefly temperature, with a small contribution from soil moisture) and variability in net annual CO2 emission.
Given the substantial variability of net emission and of surface temperature, the small fraction of total annual CO2 exchanges represented by that net emission, and the demonstration that on all relevant timescales the time-integral of temperature change determines CO2 concentration change to a high correlation, a continuing stasis or even a naturally-occurring fall in global mean surface temperature may yet cause net emission to be replaced by net uptake, so that CO2 concentration could cease to increase and might even decline notwithstanding our continuing emissions.
Natural temperature change and variability in soil moisture, not anthropogenic emission, is the chief driver of changes in CO2 concentration. These changes may act as a feedback contributing some warming but are not its principal cause.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
William Astley says:
November 25, 2013 at 1:17 am
You are ignoring the fact as pointed out by Salby that the recent CO2 changes in the atmosphere correlate with the integral of ocean temperature rather than change in anthropogenic CO2 emissions.
William, I am not the least interested in the origin of CH4, because there is no correlation between ocean temperature (integrated or not) and recent emissions of CH4 as can be seen as levels in the atmosphere. Neither are the current levels comparable with historical levels. By coincidence (?) sharply increasing when humans started to use it. Thus in my opinion, humans are the cause of the increase.
eric:”Thus the A-CO2 theory is coherent with the evidence that you and Bart present.”
Sorry, I must have missed a trick. Which part of your “A-CO2” account shows 8ppmn/K/year relationship between CO2 and SST ?
reminder:
http://climategrog.wordpress.com/?attachment_id=233
http://climategrog.wordpress.com/?attachment_id=223
I’d say this drives the stake thru. Thanks for the time integral.
“Absence of correlation necessarily implies absence of causation.” This statement is not true. Independence implies zero correlation. Zero correlation does not imply independence.
Mr. Gossage is incorrect in stating that absence of correlation does not necessarily imply absence of causation. In logic, while it is true that correlation does not necessarily imply causation, it is also true that absence of correlation necessarily implies absence of causation.
Here is an example. It is known by experiment that adding greenhouse gases to the atmosphere ought to cause some warming, and, conversely, that warming the atmosphere ought to add greenhouse gases to the atmosphere. If for a sufficiently long period CO2 concentration rises and global temperature rises too, then it is possible (though not certain) that one is causing the other to change, and it is not possible to deduce merely from the correlation which (if either) of the two datasets was causative of change in the other.
On the other hand, if for a sufficiently long period the concentration of CO2 continues to rise but global temperature does not, then the absence of correlation necessarily implies that the CO2 increase has not caused the temperature to fail to rise. Something else must be causing that, and it must be sufficient to offset any warming that the CO2 on its own might have caused.
Likewise, if a dataset increases monotonically, as CO2 does, and another dataset, say global temperature, varies stochastically, then we know that the stochasticity of the temperature dataset cannot have been caused by the monotonicity of the CO2 dataset.
Entire textbooks have been written on the causal laws. The principle that absence of correlation necessarily implies absence of causation is well established, and can be demonstrated by various formal methods, including propositional calculus.
Ferdinand Engelbeen says:
November 25, 2013 at 3:37 am
Dear Lord Monckton,
Some addition:
Here a graph of human emissions/yr, the measured increase in the atmosphere/yr and the calculated increase/yr by applying a factor to human emissions (“the airborne fraction”) and the increase/yr as function of the excess pressure above equilibrium:
http://www.ferdinand-engelbeen.be/klimaat/klim_img/dco2_em4.jpg
The function assumes:
– 1850 equilibrium CO2 level 290 ppmv
– CO2 increase by T (medium/long term): 8 ppmv/K
– 0.6 K linear T increase 1960-2010
– removal factor 2.15/110 (ppmv/ppmv)
As you can see, the trend can be fitted quite nicely with a single linear decay rate which is directly excess pressure related and hardly influenced by temperature.
The variability around the trend is caused by short term reactions of oceans and plants on temperature and humidity changes, as Salby said, but also by several before him.
See e.g. the lecture of Pieter Tans from sheet 11 on:
http://esrl.noaa.gov/gmd/co2conference/pdfs/tans.pdf
Monckton of Brenchley says:
November 25, 2013 at 8:25 am
Mr. Gossage is incorrect in stating that absence of correlation does not necessarily imply absence of causation.
The monotonic increase of sealevel caused by melting ice, waming oceans, tectonic movements etc. doesn’t correlate at all with the tide gauge measurements, while waves, tides and storm surges do correlate.
Does that imply that
– the first series of variables are not the cause of the (very small) trend?
– the second series of variables are the cause of the trend?
thanks for clearly stating the assumptions. I’m not sure I agree with them but at least we are clear where we’re starting from. That’s good.
However, CoB and myself were asking about your flux rate derivation of 51 year time constant, could you explain the maths behind your calculation. ie why the ration of those two quantities gives the time const .
eric1skeptic says:
November 25, 2013 at 3:09 am
You are just not getting it. You cannot get a sensitivity in ppmv/K when you have one in ppmv/K/unit-of-time. You do not get a finite change for a finite change in temperature.
“The latter depends on specific conditions at the equator where CO2 is generally released and at the poles where it is generally absorbed.”
Yes. That is why I have been advising Ferdinand that it is futile to try to match things too perfectly here using these bulk global averaged measurements. What we really need is a globally weighted average temperature to bring out more directly those portions of the globe which are most affected.
Nevertheless, the SNR is so strong, that we already see the emerging phenomenon in the global averages to a remarkable degree of fidelity.
Dear Lord Monckton,
From “Introduction the Theory of Statistics” by Mood, Graybill, and Boes. 3rd Ed. Section 4.5:
Corollary to Theorem 9: “If X and Y are independent variables then Cov[X,Y] = 0”
Remark: “The converse of the above corollary is not always true; that is Cov[X,Y] = 0 does not always imply that X and Y are independent, as the following example shows.”
While there may be many cases for which the two statements are true, that is not always so.
Regards
Brett
Bart: “What we really need is a globally weighted average temperature to bring out more directly those portions of the globe which are most affected.”
It would be useful to try to establish which regions are key in this short term reaction. While I agree in principal with your comment, a quick scan around the major basin averages (eg N. Atlantic etc.) did not reveal any region that correlated better (or as well as) global.
That maybe due to the wide latitude range of these subsets which I had extracted for other purposes.
KNMI climate explorer makes it very easy to pull out “rectangular” long/lat subsets, so looking into what regions dominate this relationship may be interesting.
Mr. Engelbeen and Mr. Gossage question my assertion that, while correlation does not necessarily imply causation (with which they both agree), absence of correlation necessarily implies absence of causation (with which they both disagree).
Mr. Engelbeen takes the following instance:
“The monotonic increase of sealevel caused by melting ice, waming oceans, tectonic movements etc. doesn’t correlate at all with the tide gauge measurements, while waves, tides and storm surges do correlate.
“Does that imply that
– “the first series of variables are not the cause of the (very small) trend?
– “the second series of variables are the cause of the trend?”
His first statement really says tide gauges are not accurate as methods of measuring sea-level rise, which he believes to exist and to be monotonic. However, sea-level rise is not monotonic, as even the largely fictional graph of the University of Colorado shows. As for whether there has been sea-level rise recently, during the entire eight-year period of the ENVISAT satellite, sea level was shown as rising at a rate equivalent to 3 cm/century – well within the measurement uncertainties. Also, Peltier et al. (2009) reported that analysis of the gravitational anomalies measured by the GRACE satellites showed there was no sea level rise over the period of study, and that sea level might actually have fallen. His example fails on factual grounds. If the tide-gauges are not showing sea-level rise to be monotonic, they are correct. If they are showing sea-level fluctuations as being caused – at least in the short term – by waves, tides, and storm surges, they are again correct.
Mr. Gossage quotes from a textbook of statistics, “Introduction the Theory of Statistics” by Mood, Graybill, and Boes. 3rd Ed. Section 4.5:
“Corollary to Theorem 9: “If X and Y are independent variables then Cov[X,Y] = 0.
“Remark: “The converse of the above corollary is not always true; that is Cov[X,Y] = 0 does not always imply that X and Y are independent, as the following example shows.”
Mr. Gossage has things the wrong way about. The statement that the covariance of independent variables is aye nil is the statisticians’ way of stating that absence of correlation necessarily implies absence of causation.
The textbook’s “Remark” to the effect that absence of causation does not necessarily imply absence of correlation is correct. For instance, if events A and B did not cause one another, they may both have been caused by a third event C, in which event they may or may not be correlated with one another.
Bart says:
November 25, 2013 at 9:09 am
You cannot get a sensitivity in ppmv/K when you have one in ppmv/K/unit-of-time. You do not get a finite change for a finite change in temperature.
That is exactly where you, Salby and others go wrong. The atmosphere gets to a new equilibrium with the oceans and vegetation when the temperature changes with a finite amount. There is a finite change in CO2 for a finite change in temperature. That is what Henry’s law dictates for ocean waters and what happens for the uptake/release from vegetation.
Greg Goodman says:
November 25, 2013 at 10:47 am
“It would be useful to try to establish which regions are key in this short term reaction.”
This could be a monumental study requiring a great many man-hours. The proper weighting might even be time varying with the hemispheric seasons, or particularly heavily weighted in relatively small regions, possibly due to undersea geologic activity. I suppose I would focus on the tropical and the polar oceans, as regions of upwelling and downwelling waters, first.
I might suggest latitude weighting bands. You could actually do a fit of a few weighting coefficients to the dCO2/dt. Granted, this might produce only an apparent agreement over the existing data sets which might not hold up over time. Then again, it might.
I don’t intend to attempt that analysis myself, but if you’ve a mind to, by all means go for it. You have alluded to Salby’s “surface conditions”. Perhaps he has started down a path in this direction, and no doubt has a better comprehension of what might be appropriate than I. Someday, hopefully, he will find just what he is looking for to make an incontrovertible case, and will publish it. I suspect that is the reason for his reticence so far – he probably wants to have an airtight case before confronting the firestorm which is sure to erupt.
Ferdinand Engelbeen says:
November 25, 2013 at 11:22 am
“The atmosphere gets to a new equilibrium with the oceans and vegetation when the temperature changes with a finite amount.”
Not if new CO2 is being introduced to the surface system via, e.g., the temperature dependent outgassing of CO2 from CO2 rich upwelling waters. You know quite well this is my argument.
Here is that argument as a reminder. Something like this must be going on, because the data clearly show a ppmv/K/unit-of-time dependence.
Monckton of Brenchley says:
November 25, 2013 at 11:10 am
Dear Lord Monckton,
Back to our favorite subject:
There is no correlation between the rate of change of CO2 in the atmosphere and human emissions.
There is a strong correlation between the rate of change of CO2 in the atmosphere and changes in temperature.
Does that imply that human emissions are not the cause of the increase in the atmosphere (the trend)?
Does that imply that the nice correlation with the temperature variability also proves that temperature is the cause of the increase in the atmosphere (the trend)?
Bart says:
November 25, 2013 at 11:34 am
Here is that argument as a reminder. Something like this must be going on, because the data clearly show a ppmv/K/unit-of-time dependence.
I had quite a lot of comments there. But as usual…
The data are a mix of a trend (caused by humans) and a variability (caused by temperature variability). There is not the slightest indication in the data that the trend is caused by temperature
There is no increase in emissions from the oceans visible in the residence time or the 13C/12C ratio trend or the 14C/12C bomb spike trend…
What kind of a fanciful world you all must live in, you can see a prehistoric graph of air samples over 800,000 years old, through multiple interglacial cycles (some of which had temperatures slightly above our current temperature at their peak) and their CO2 concentrations never went above 300 ppmv.
You can look at this 800,000 year record and notice that our current CO2 concentration is almost 133% greater than the highest level recorded in at least 800,000 years, and that the increase has occurred since 1880, in only about 135 years.
And you can, somehow, convince yourselves that it has nothing to do with human emissions???
My favorite quote:
“the demonstration that on all relevant timescales the time-integral of temperature change determines CO2 concentration change to a high correlation, a continuing stasis or even a naturally-occurring fall in global mean surface temperature may yet cause net emission to be replaced by net uptake, so that CO2 concentration could cease to increase and might even decline notwithstanding our continuing emissions.”
. . .amazing
baffling and amazing
you all live in a very magical world.
Greg Goodman says:
November 25, 2013 at 10:47 am
It would be useful to try to establish which regions are key in this short term reaction. While I agree in principal with your comment, a quick scan around the major basin averages (eg N. Atlantic etc.) did not reveal any region that correlated better (or as well as) global.
Based on the 13C/12C ratio changes, the 1-3 year variability in CO2 comes mainly from the tropical forests: with an El Niño, rain patterns change and temperature increases. That gives that several parts of the Amazon get drier but also that soil bacteria are working harder.
But it will not be easy to find regional precipitation levels for the Amazon forest…
Ferdinand Engelbeen says:
November 25, 2013 at 11:53 am
First, do you then acknowledge that your comment at November 25, 2013 at 11:22 am is not at all as categorical as you allege? That if CO2 rich waters are upwelling as I hypothesize, then that would lead to just such a temperature dependency as we observe?
Monckton of Brenchley says:
November 25, 2013 at 11:10 am
Mr. Engelbeen and Mr. Gossage question my assertion that, while correlation does not necessarily imply causation (with which they both agree), absence of correlation necessarily implies absence of causation (with which they both disagree).
Yea its the difference in “if/then” and “if and only if”. There really is no arguing beyond that…sorry mathematics is a cruel mistress.
Greg Goodman says:
November 25, 2013 at 9:02 am
However, CoB and myself were asking about your flux rate derivation of 51 year time constant, could you explain the maths behind your calculation. ie why the ration of those two quantities gives the time const .
As usual, I was preparing a (too) long reaction, but by being hasty completely destroyed it with one keystroke…
Thus here in short (?). based on Paper 1 of Pettersson:
If the equilibrium constant is less than 0.05, more than 95% of the emission will be removed from the reservoir and the process may be considered as practically irreversible. Under such conditions, the relaxation time will differ less than 5% from the turnover time and be practically equal to it.
and
The turnover time (β) is normally defined as the amount of compound being present in the reservoir divided by the flux rate at which the compound is removed:
β = Amount/Flux
Warning: the “turnover time” in this case is NOT the mass of CO2 going in and out within a year caused by temperature, it is the mass of CO2 which is removed per year as result of the difference in pressure between current atmosphere and atmosphere at equilibrium. For clarity we still will use “relaxation time”.
Contrary to the Bern model, the deep oceans and vegetation can remove 99% of all extra CO2 above equilibrium, thus reaching near equilibrium with only one relaxation time (*).
In fact there is one extra factor: the ocean surface layer, which can absorb about 10% (19% according to the Bern model?) of the change in the atmosphere with a relaxation time of 1.2 years. That is part of the overall relaxation time, thus doesn’t play a visible role now, but when the emissions should stop, the fast equilibrium reactions of the surface layer will reverse and release CO2 into the atmosphere as levels are falling. That will increase the overall relaxation time, but that is not the case for now.
(*) That needs some clarification:
The deep oceans receive their increased levels of CO2 from a few cold places near the poles. These have a pCO2 down to 250 μatm, while the atmosphere is at 400 μatm (~ppmv). That is what pushes CO2 into the deep oceans. To saturate the oceans in the current circumstances would need to increase the pCO2 of the oceans up to 400 μatm, without further increase in the atmosphere.
Near the poles, the uptake thus is not a problem, not even in the far future. The Revelle (buffer) factor doesn’t play a role in deep waters.
The up to current human emissions all can be absorbed and still give only 1% increase in deep oceans and atmosphere at equilibrium…
I hope that is clear now?
Ferdi: ” the 1-3 year variability in CO2 comes mainly from the tropical forests: with an El Niño, rain patterns change and temperature increases. ”
Interesting. There are aspects of the phase relationship that I have not reconciled yet, so I’m open to ideas. Once again if you could point to evidence of what you say rather than expecting us to take it on faith it would be good.
“That is exactly where you, Salby and others go wrong. The atmosphere gets to a new equilibrium with the oceans and vegetation when the temperature changes with a finite amount. There is a finite change in CO2 for a finite change in temperature. That is what Henry’s law dictates for ocean waters and what happens for the uptake/release from vegetation.”
You have stated that the time constant is 51 years . That suggests about 250 years to get close to equilibrium, assuming no continued changes in other parameters, which of course there are.
It would lead to a 50y lag in response to a steady increase in driver. It may be interesting to see how well that plays out in relation to changes in both temp and atm CO2.
BTW if you could explain how you derive the 51y lag mathematically, it would be fairly essential to adopting the idea.