There is quite a bit of buzz surrounding a talk and pending paper from Prof. Murry Salby the Chair of Climate, of Macquarie University. Aussie Jo Nova has excellent commentary, as has Andrew Bolt in his blog. I’m sure others will weigh in soon.
In a nutshell, the issue is rather simple, yet powerful. Salby is arguing that atmospheric CO2 increase that we observe is a product of temperature increase, and not the other way around, meaning it is a product of natural variation. This goes back to the 800 year lead/lag issue related to the paleo temperature and CO2 graphs Al Gore presented in his movie an An Inconvenient Truth, Jo Nova writes:
Over the last two years he has been looking at C12 and C13 ratios and CO2 levels around the world, and has come to the conclusion that man-made emissions have only a small effect on global CO2 levels. It’s not just that man-made emissions don’t control the climate, they don’t even control global CO2 levels.
Salby is no climatic lightweight, which makes this all the more powerful. He has a strong list of publications here. The abstract for his talk is here and also reprinted below.
PROFESSOR MURRY SALBY
Chair of Climate, Macquarie University
Atmospheric Science, Climate Change and Carbon – Some Facts
Carbon dioxide is emitted by human activities as well as a host of natural processes. The satellite record, in concert with instrumental observations, is now long enough to have collected a population of climate perturbations, wherein the Earth-atmosphere system was disturbed from equilibrium. Introduced naturally, those perturbations reveal that net global emission of CO2 (combined from all sources, human and natural) is controlled by properties of the general circulation – properties internal to the climate system that regulate emission from natural sources. The strong dependence on internal properties indicates that emission of CO2 from natural sources, which accounts for 96 per cent of its overall emission, plays a major role in observed changes of CO2. Independent of human emission, this contribution to atmospheric carbon dioxide is only marginally predictable and not controllable.
Professor Murry Salby holds the Climate Chair at Macquarie University and has had a lengthy career as a world-recognised researcher and academic in the field of Atmospheric Physics. He has held positions at leading research institutions, including the US National Center for Atmospheric Research, Princeton University, and the University of Colorado, with invited professorships at universities in Europe and Asia. At Macquarie University, Professor Salby uses satellite data and supercomputing to explore issues surrounding changes of global climate and climate variability over Australia. Professor Salby is the author of Fundamentals of Atmospheric Physics, and Physics of the Atmosphere and Climate due out in 2011. Professor Salby’s latest research makes a timely and highly-relevant contribution to the current discourse on climate.
Salby’s talk was given in June at the International Union of Geodesy and Geophysic meeting in Melbourne Australia. He indicates that a journal paper is in press, with an expectation of publication a few months out. He also hints that some of the results will be in his book Physics of the Atmosphere and Climate which is supposed to be available Sept 30th.
The podcast for his talk“Global Emission of Carbon Dioxide: The Contribution from Natural Sources” is here (MP3 audio format). The podcast length is an hour, split between his formal presentation ~ 30 minutes, and Q&A for the remaining time.
Andrew Bolt says in his Herald Sun blog:
Salby’s argument is that the usual evidence given for the rise in CO2 being man-made is mistaken. It’s usually taken to be the fact that as carbon dioxide concentrations in the atmosphere increase, the 1 per cent of CO2 that’s the heavier carbon isotope ratio c13 declines in proportion. Plants, which produced our coal and oil, prefer the lighter c12 isotope. Hence, it must be our gasses that caused this relative decline.
But that conclusion holds true only if there are no other sources of c12 increases which are not human caused. Salby says there are – the huge increases in carbon dioxide concentrations caused by such things as spells of warming and El Ninos, which cause concentration levels to increase independently of human emissions. He suggests that its warmth which tends to produce more CO2, rather than vice versa – which, incidentally is the story of the past recoveries from ice ages.
Dr. Judith Curry has some strong words of support, and of caution:
I just finished listening to Murry Salby’s podcast on Climate Change and Carbon. Wow.
If Salby’s analysis holds up, this could revolutionize AGW science. Salby and I were both at the University of Colorado-Boulder in the 1990′s, but I don’t know him well personally. He is the author of a popular introductory graduate text Fundamentals of Atmospheric Physics. He is an excellent lecturer and teacher, which comes across in his podcast. He has the reputation of a thorough and careful researcher. While all this is frustratingly preliminary without publication, slides, etc., it is sufficiently important that we should start talking about these issues. I’ll close with this text from Bolt’s article:
He said he had an “involuntary gag reflex” whenever someone said the “science was settled”.
“Anyone who thinks the science of this complex thing is settled is in Fantasia.”
Dr Roy Spencer has suspected something similar, See Atmospheric CO2 Increases: Could the Ocean, Rather Than Mankind, Be the Reason? plus part 2 Spencer Part2: More CO2 Peculiarities – The C13/C12 Isotope Ratio both guest posts at WUWT in 2008. Both of these are well worth your time to re-read as a primer for what will surely be a (ahem) hotly contested issue.
I’m pretty sure Australian bloggers John Cook at Skeptical Science and Tim Lambert at Deltoid are having conniption fits right about now. And, I’m betting that soon, the usual smears of “denier” will be applied to Dr. Salby by those two clowns, followed by the other usual suspects.
Smears of denial and catcalls aside, if it holds up, it may be the Emily Litella moment for climate science and CO2 – “Never mind…”
Actually, Richard, I was not aware of that model. I have, however, looked over simpler models which apparently are based on it. The basic components are all the same: atmospheric, oceanic, and terrestrial reservoirs, anthropogenic and non-anthropogenic inputs and sequestration outputs.
The problem I have is not with the basic structure (though I do have qualms about the assumption of non-time varying parameters), but with the lack of uniqueness between model assumptions and agreement of the model output with major observables. When you have more parameters to fit than you have data to fit them, you have to rely on subjective criteria to constrain the problem so that it can be solved. The modelers have constrained it such that the bandwidth of the resulting CO2 regulation is very low. But, a step back to view the forest reminds us that low bandwidth systems are very bad at maintaining an equilibrium within tight bounds.
I believe it is very likely that the modelers have assumed essentially deterministic behavior, and so have not given thought to the effects of variability and “noise” in the inputs and sink capacities. It is very easy to observe that the measurements do not pick up much if any influence from apparent variations in the recorded emissions – one can do it just eye-balling the data. That could be merely a phantom due to corrupted emissions reporting. However, when I couple that with the tight regulation of CO2 levels as indicated by the ice cores, I naturally reach a conclusion that A) the ice core measurements also are corrupted or B) the regulation of CO2 is, in fact, high enough bandwidth that small input variations from nominal conditions are effectively attenuated below the noise.
As I stated at Dr. Curry’s site of CO2 levels:
John,
BZZZZT. Wrong. John, as soon as you said this: “This rise cannot be temperature driven because there has been no change in temperature. It is driven by human emissions.”, I could tell you haven’t been able to grasp what has been said from any of my longer posts, or you are purposely distorting it to make believe it can’t be true. The only thing I can recommend is that you read them again. The math and units are quite straightforward. Read the one about 1850 being 0.7°C cooler, and try to imagine a temperature increase of 0.7°C occurring over 160 years. Every year (or month), apply the rate of your choice, in ppm/°C/[time] and calculate the effect. Do the same for the next period. Add the effects together. You will see an exponential function emerge. Excel is good for this sort of thing.
Bonus points: Assume an equilibrium effect, (a rate) such that the new total accumulated ppm for the period in consideration is used to calculate an offsetting factor that serves to reduce the effect of the above function compared to using the full delta T, the distance from month(n) to original temperature at 1850 (-0.7°C in our case). At what rate will the compensating function result in ppm at year 2011 of 390? What are the factors you used? If you can answer this, I’ll be confident you actually get it.
Michael D Smith says:
August 9, 2011 at 7:34 pm
Michael, I can only think we are talking at cross purposes as when we look at basic rates of change and sensitivity you seem to be in a different place to me (and everyone else).
Let’s start from the very basics. Sensitivity (in our context) is estimated (or determined) from the relationship between the change in the dependant variable (x) over the independent variable (y). For example, if you believe that increasing CO2 concentrations are solely responsible for the increase in temperature since 1850, then the sensitivity can be estimated as follows:
Delta_F = 1.7 w/m2 (forcing due to CO2 since 1850)
Delta_T = 0.7 deg C (temp rise since 1850)
Sensitivity due to CO2 = 0.7/1.7 = ~0.41 deg C per w/m2 (note the units).
Similarly, if you believe temperature is mainly or wholly responsible for the rise in CO2 Concentrations (ppm) since 1850 you can again calculate the sensitivity, i.e.
Delta_CO2 = 110 ppm (incease in CO2 since 1850)
Delta_T = 0.7 deg C (temp rise since 1850)
Sensitivity of CO2/Temp = 110/0.7 = ~157 ppm/deg C (again note the units)
It’s as simple and as basic as it gets. We have an increase in temperature of 0.7 deg which it’s claimed causes a 110 ppm rise in CO2 concentrations. That ‘s an increase of ~157 ppm for every degree C rise.
You then say
The math and units are quite straightforward. Read the one about 1850 being 0.7°C cooler, and try to imagine a temperature increase of 0.7°C occurring over 160 years.
I just have.
Every year (or month), apply the rate of your choice, in ppm/°C/[time] and calculate the effect.
I chose not to use a time period of one month or one year I used a time period of 160 years. Let’s got through your units, i.e. ppm/°C/[time] to illustrate
ppm = 110
°C = 0.7 deg/160 years or 0.0044 deg/year if you prefer
time = 160 years
Sensitivity = 110/0.0044/160 = ~157 ppm/ degC
Let’s just check the units – we have
ppm / degC/yr / yr = ppm * yr/degC / yr
= ppm/degC ( since the yr units cancel out)
Yep – all seems fine.
Do the same for the next period. Add the effects together. You will see an exponential function emerge.
Whooaa – hold on a minute, Michael, why would YOU expect to see an exponential function emerge. You are claiming that ppm is driven by temperature. Temperature hasn’t risen exponentially – has it? As it happens ppm has risen exponentially but that’s because of the growth in human emissions.
Why don’t you go through the first few iterations (i.e. first few years) of the exercise you suggest and post the results. I can then see what you’re doing. As it is I’m having to second guess what you’re thinking. It has just occurred to me what you might be doing but I’ll hold fire til you respond.
Excel is good for this sort of thing
Thank you for that, Michael. I first used Excel around 20 years ago when the university where I was employed migrated from Lotus 123.
Michael D Smith says:
August 9, 2011 at 7:34 pm
“You will see an exponential function emerge. “
Michael – correct me if I am wrong, but from light reading of your posts, I am under the impression that what you are arguing is essentially that there is heavy low pass filtering going on and the yearly variations are heavily attenuated, whereas the steady inputs at lower frequency are passed through at full strength.
That is, if the time constant were 160 years, the annual variations should be attenuated by
alpha = 1/sqrt( 1 + (2*pi*160)^2) = 0.001
whereas a ramp would be passed through with just a little phase lag. Thus, in this case, using the sensitivity to annual temperature variations, say, to gauge the sensitivity to steady warming could give you a value underestimated by a factor of 1000.
Bart says:
August 9, 2011 at 2:41 pm
The point here is that I have ripped your mass balance argument to shreds. Your claim was that the mass balance proved that the rise was wholly anthropogenic. That claim is revealed as sloppy reasoning and utterly false.
We can repeat the whole discussion here, but for those interested, have a look at Judith Curry’s blog:
http://judithcurry.com/2011/08/04/carbon-cycle-questions/
Bart and Michael Smith:
Bart,
re your post at August 9, 2011 at 7:10 pm.
Yes. Our agreement is in your statement saying;
“However, when I couple that with the tight regulation of CO2 levels as indicated by the ice cores, I naturally reach a conclusion that A) the ice core measurements also are corrupted or B) the regulation of CO2 is, in fact, high enough bandwidth that small input variations from nominal conditions are effectively attenuated below the noise.”
All the reservoir models (i.e. Berne, Engelbeen, etc.) assume there are no natural “small input variations from nominal conditions” (or output variations) so the anthropogenic emission acts as an extraneous effect and not a variation. Indeed, they clearly state this assumption because they use accountancy principles to “prove” the anthropogenic emission is accumulating in the air.
Michael Smith:
I am amazed and in admiration of your patience but I see it is getting exhausted by John Finn (as mine did).
Something one says once may be overlooked, it is probably a deliberate oversight if is still ignored when repeated , and when explained yet again but ignored again there is no point in continuing the ‘discussion’. Indeed, I reached the conclusion that John Finn was deliberately trying to disrupt rational discussion.
Richard
I am amazed and in admiration of your patience but I see it is getting exhausted by John Finn (as mine did).
Richard
You don’t seem to be able to post a comment without mentioning my name at least once. This seems a bit surprising since you appear to think I have nothing worthwhile to add to the discussion. Do you have a bit of a crush on me, perhaps?
Anyway, as it happens, I’m interested in Michael’s line of thinking. I think he’s got something screwed (quite badly) but I’m not sure where. He comes across as fairly numerate so I’m interested in finding out if there is a key factor in his analysis to which he has a ‘blind’ spot.
‘Back of the envelope’ calculations suggest he’s miles out.
Now I fully understand that you’re out of you’re depth with this stuff. That’s ok we can’t all be qualified in the relevant disciplines but in might be best if you just stepped back from this issue, Richard, as you’re not really helping.
Bart says:
August 10, 2011 at 1:48 am
I’m not sure Michael is trying to making any of arguments you suggest. It appears to me that he’s got a basic misunderstanding on how to interpret the rates of change. This is from one of his posts
Post 1979 is, say, 29 years (my data ends in 2008). So the required rate is 50/0.4/29 or 4.3ppm/°C/year. (good answer, wrong method) Now, the other one… 100ppm/0.7°C/108 years = 1.32ppm/°C/year
The numbers refer to (a) ~50 ppm and ~0.4 temp rise between 1979 and 2008 (29 years) and (b) ~100 ppm and ~0.7 temp rise between 1860 and 2008 (108 years).
Now in a way he’s not wrong with his calculations. That is, if temp is the sole driver, a ~0.4 deg rise over 29 years does produce an average rate of 4.3 ppm/°C for EVERY year of the rise but that’s a fairly meaningless statistic. He seems to think this is the sensitivity or response rate – it’s clearly not.
The 0.4 and rise covered 29 years so the sensitivity becomes 4.3 x 29 = ~125ppm/degC
Richard S Courtney says:
August 10, 2011 at 2:22 am
All the reservoir models (i.e. Berne, Engelbeen, etc.) assume there are no natural “small input variations from nominal conditions” (or output variations) so the anthropogenic emission acts as an extraneous effect and not a variation. Indeed, they clearly state this assumption because they use accountancy principles to “prove” the anthropogenic emission is accumulating in the air.
Who said that? Of course there are small natural input variations, but indeed they are small on short term. The influence is visible as a short-term variability of about 4 ppmv/degr around the trend. The very long term variability is visible in ice cores in century to millenia scale: 8 ppmv/degr.C.
Thus the third reason why there is little variation in ice cores is that there was little natural variability on decadal to century scale, not more than what is seen as natural variability on sub decadal or on century or larger scale…
The problem with the ice cores is that there is no way to actually confirm the model for how CO2 is trapped and held over many thousands of years. I suspect that, if we could actually run a closed loop experiment in the lab, we would find surprising results which differed significantly with the expectations. Because, we always do when we go to the lab…
John Finn says:
August 10, 2011 at 6:11 am
“I’m not sure Michael is trying to making any of arguments you suggest.”
If not, then consider me making them. Where does it lead?
All right, since nobody took me up on that challenge, let’s look at the data Michael has provided.
In his chart here, I see a temperature component amplitude at one point of about 0.3 degC with a CO2 derivative amplitude of about 0.1 ppmv/month, or 1.2 ppmv/year, and a period of about 2 years. So, frequency in radians is w = 2*pi/2 = 3.14 rad/year.
Suppose the temperature variation at this frequency is described by dT = A*sin(w*t), and the resulting CO2 derivative variation is described by CO2dot = B*w/sqrt(1+(tau*w)^2) where tau is the time constant of the transfer function. The sensitivity at zero frequency is given by B/A in ppmv/degC. We get
B/A = (1.2/0.3)*sqrt(1+(tau*w)^2) /w
With tau = 160 years, we get 640 ppmv/degC.
Clearly, that’s too much. So, let’s try tau = 30 years. We get 120 ppmv/degC.
So, if we have a time constant of maybe 30 years, there is enough sensitivity for the temperature to be driving most of the CO2 increase.
Check my math, and make comments as desired.
“CO2 derivative variation is described by CO2dot = B*w/sqrt(1+(tau*w)^2)*sin(w*t)…”
Bart says:
August 10, 2011 at 9:10 am
The problem with the ice cores is that there is no way to actually confirm the model for how CO2 is trapped and held over many thousands of years.
Actually it was done: some Japanese group measured the migration of CO2 in “Vostok” conditions, against ambient pressure. There was some modest migration, but not very relevant for the real ice core, as there is little pressure difference between ice at 2,000 m and 1,999 m depth.
A theoretical calculation, based on CO2 levels near remelted layers in the Siple Dome ice core did show some migration, which gives a broadening of the resolution of 10% at mid-depth and 100% at full depth.
More important is that there is no flattening of the ratio between CO2 levels during an interglacial and a glacial period over each 100,000 years back in time. If there was even the slightest migration, that difference should fade over each period back in time.
Thus based on this evidence, an overlap of 20 years with direct measurements and similar levels found in ice cores of complete different places (coastal: high accumulation, higher temperatures, inland: low accumulation and temperatures) for the same period, I suppose that the data obtained from ice cores are pretty accurate.
Note the formula can be simplified for large tau to approximately
So = CO2dot*tau/dT
where So is the sensitivity at dc (zero frequency). Again, CO2dot is the amplitude of the derivative of the cyclical portion of CO2 concentration at a given “high” frequency, and dT is the amplitude of the temperature variation at the same frequency with which the CO2 variation is correlated.
Ferdinand Engelbeen says:
August 10, 2011 at 2:10 pm
20 whole years, huh? Theoretical calculations? Uh-huh…
Did the “Japanese group” travel back in time to measure the ambient CO2, then confirm the measurement when they got back? No?
Color me unconvinced.
“Actually it was done: some Japanese group measured the migration of CO2 in “Vostok” conditions, against ambient pressure. There was some modest migration, but not very relevant for the real ice core, as there is little pressure difference between ice at 2,000 m and 1,999 m depth.”
I think it takes about 150 to 200 years for the snowfall precursor to ice layers in the Arctic and Antarctic to become compressed and sealed enough for migration to cease.
That would account for all that we observe namely:
i) A hockey stick type rise in CO2 content in bubbles of air in the ice over the past 150 years or so.
ii) The inconsistency observed pre and post about 1850.
iii) the apparent lack of variability for many millennia into the past.
Note that where water is concerned CO2 flows from warmth to cold and maximum cold is in the air above the ice. That is how CO2 can still migrate upward out of the ice when the atmospheric CO2 content is higher. The suggestion that CO2 at 200ppm in air bubbles held within the ice cannot migrate to CO2 at 400pm in the air above is apparently false.
But nonetheless after 150 to 200 years the air and CO2 held in the ice becomes sealed off from interaction with the atmosphere.
Bart says:
August 10, 2011 at 3:04 pm
Bart,
1. Etheridge (1996) measured CO2 directly in firn from not yet closed air bubbles and in closed bubles in ice, via the normal way (ice crushing under vacuum, CO2 measured over a cold trap). At closing depth ice and firn CO2 were the same. Thus there is no (measurable) change in CO2 level by the closing process.
2. 20 years overlap show that the highest resolution ice cores track the atmospheric composition, once the ice is completely sealed. Thus once sealed, there is no migration over a short period (contrary to the idea of some persons who expect that CO2 is migrating/pressed out of the ice to the surface, while the pressure increases).
3. Any migration further down the ice would be faster for ice cores with higher temperatures, but ice cores with large differences in average temperature show the same CO2 levels (+/- 5 ppmv) for the same periods.
4. Any migration flattens the data over a larger period. The highest CO2 levels are 10-20 kyr long over the interglacials, the lowest levels 100 kyr long over the glacials, thus any migration would lower mainly the highest values. But the CO2 values track the temperature values with a constant ratio over the full 420 kyr Vostok record, recently confirmed by the 800 kyr Dome C record.
I am convinced, mainly by point 4.
I know you are convinced, Ferdinand. I am not. In industry, before any complex component is released to the buyer, it must undergo a battery of tests, including most importantly the end-to-end tests. In these, the entire integrated product is tested, with real-world inputs which must reproduce expected behavior. Even though all the subcomponents have passed their tests. Even though there is no reason at all to have any doubt that the product will work as planned.
And, you know what? That is where you find most of the problems. Talk to someone who has endured the ordeal, and you will learn. Experience it yourself, and it will rock your world.
Frankly, I would love to believe that the ice core data is perfect. It would pretty much clinch the argument for strong feedback if it were. Without that well-behaved data, I cannot make that claim, though it would also rip the rug out from under those who claim current CO2 levels are “unprecedented”.
Bart says:
August 10, 2011 at 12:05 pm
With tau = 160 years, we get 640 ppmv/degC.
Clearly, that’s too much. So, let’s try tau = 30 years. We get 120 ppmv/degC.
What’s the justification for tau being 30 years?
Bart says:
August 10, 2011 at 2:24 pm
Note the formula can be simplified for large tau to approximately
So = CO2dot*tau/dT
Right – that figures. As tau increases (tau*w)^2 will also increase (faster), hence as tau -> infinity sqrt(1+(tau*w)^2) -> tau*w. I can’t quite see the relevance though. For anyone who is remotely interested in this drivel, Bart is effectively saying
If X is large then sqrt(1 + X^2) ~ = X i.e. the value of 1 becomes less significant.
Ok, Bart, you’ve shown us how clever you are and I agree that you have a ‘model’ that gives the required sensitivity IF the time constant can be justified. So are you saying tau=30 years is a valid figure?
Bart says:
August 10, 2011 at 6:04 pm
I know you are convinced, Ferdinand. I am not. In industry…..
In industry (in the UK at least) we have certain words for people who, when losing the argument, resort to ‘intellectual intimidation’ in an attempt to ‘wrong foot’ their opponents. I say “argument” but there really isn’t one.
Humans emit ~8GtC into the atmosphere every year. An amount equivalent to approximately half of that is added to the atmosphere every year. Can you think of any reason at all why atmospheric CO2 levels should continue to rise if human emissions suddenly stopped.
Further to my earlier post: in his post on August 10, 2011 at 12:05 pm. Bart says:
All right, since nobody took me up on that challenge, let’s look at the data Michael has provided.
In his chart here, I see a temperature component amplitude at one point of about 0.3 degC with a CO2 derivative amplitude of about 0.1 ppmv/month, or 1.2 ppmv/year, and a period of about 2 years…
I still think you have a problem with dT. In the following year dT 0. Remember the temperature plot uses actual temp anomalies – NOT derivatives. You’ve effectively got a positive signal followed by a negative signal (although steadily positive over time) – but CO2dot remains positive throughout.
Not sure my last post makes sense. It looks like my use of angle brackets to denote “less than” and “greater than” might have screwed the formatting.
During deglaciation the two varied simultaneously,* but during times of cooling the CO2 changed after the temperature change, by up to 1000 years. This order of events is not what one would expect from the enhanced greenhouse effect.
http://www-das.uwyo.edu/~geerts/cwx/notes/chap01/icecore.html
* All studies I’ve seen show Carbon dioxide levels always follow both temperature rises and falls. You’re not dealing with this. The time spans show that Carbon dioxide levels are irrelevant to driving these natural temperature changes. You’re not dealing with this.
Instead what you propose, that CO2 drives temperature changes, is magic. c800 years in advance of a change, Carbon Dioxide magically decides that it will drive these changes.
John Finn says:
August 11, 2011 at 1:36 am
“What’s the justification for tau being 30 years? “
I don’t know. Why would I need a justification to show the conditions under which an hypothesis is viable?
“Ok, Bart, you’ve shown us how clever you are…”
I was just trying to make it easier to see that the reliance on tau was effectively linear. Sorry if that offends you.
“…we have certain words for people who, when losing the argument…”
And, we have a celebrity exemplar of losers who insist they are “winning”.
“Can you think of any reason at all why atmospheric CO2 levels should continue to rise if human emissions suddenly stopped[?]”
Deep ocean upwelling? Sustained loss of sea plankton? I don’t know. But, I don’t need to know in order to detect patterns.
“I still think you have a problem with dT. In the following year dT 0.”
Do the words “amplitude” and “component” not carry the same meanings in Britain that they do in the States? If so, it appears you are not very well versed in signal processing terminology and methods.
“Not sure my last post makes sense.”
You got that right.