Monday
Jul212014
by 
Bishop Hill
Bishop Hill New paper from the Netherlands on C02 emissions
Jul 21, 2014 Climate: carbon budget 'A paper published today in Atmospheric Chemistry and Physics finds that only about 3.75% [15 ppm] of the CO2 in the lower atmosphere is man-made from the burning of fossil fuels, and thus, the vast remainder of the 400 ppm atmospheric CO2 is from land-use changes and natural sources such as ocean outgassing and plant respiration.'
Read all about it at Climate Depot and H/T to Swiss Bob in the BBC and Lawson comments.
Reader Comments (61)
Bart: If your f(T) term has the form k(T-T0), you may not find much room for human emissions to make much of a difference*. However, if you f(T) term depends only on recent temperature change (say in the last month or several months, then you will probably find plenty of room for human emissions to drive rising CO2 levels, but still see temperature dependence in the d(CO2) vs T plot. Consider again the data for the 97/98 El Nino and the following La Nina, the biggest temperature swing in the record. During the 6/97-6/98 El Nino, CO2 rose an usually large seasonally-adjusted 3.34 ppm. However, during the 6/98-6/99 La Nina - one of the largest temperature drops on record, CO2 STILL rose 1.41 ppm. So, you want to model an anthropogenic increase of 4 ppm, an anthropogenically enhanced natural uptake of 2 ppm (driven by CO2 near 400 ppm rather than 280 ppm) and temperature dependent emission due to the temperature spike in 97/98 of an extra 1.3 ppm and a temperature dependent "emission" of -0.6 ppm (uptake of 0.6 ppm) due to the temperature fall in 98/99.
One possible reservoir for the CO2 released or taken up with temperature change is the mixed layer of the ocean. The surface of the ocean responds to seasonal changes in SWR and turbulent mixing produces seasonal changes down to about 100 m. The solubility of CO2 in water decreases with rising temperature. As turbulent mixing transfers heat through the mixed layer, more CO2 could be released from this layer (if the water is warming) or taken up by this layer (if the water is cooling). Temperature change occurs within months, so the CO2 change could happen as quickly.
*Of course, any sensible mechanism for explaining changes in atmospheric CO2 must include human emission.
Bart wrote: "I expect then we will be treated to the climate establishment scratching their heads, and wondering why sink activity seems to have picked up, as more and more of the human emissions appear to be getting sequestered."
The climate establishment knows that the rate of accumulation of CO2 in the atmosphere has been less than expected from total human emissions. However, it is not simply that "human emissions" appear to be missing. Other than their lack of C14, human emissions can't be distinguished. It is simply that, as atmospheric levels of CO2 have risen, the rate of some uptake processes has also risen. This enhanced uptake provides the illusion that 50% of human emissions have "disappeared" into sinks. CO2 changes in response to all emissions (natural, anthropogenic, or stimulated by recent temperature rise) and all uptake (natural, natural but enhanced by current high levels of CO2, or stimulated by recent temperature fall).
"If your f(T) term has the form k(T-T0), you may not find much room for human emissions to make much of a difference*. However,..."
There is no "if". There is no "however". The empirical data show that the f(T) term is undeniably k(T-T0). The only arbitrary constants which could be distributed among the two possible inputs are k and T0.
If you do not adjust k, there is no room for anthropogenic inputs, because anthropogenic inputs are not constant, but have a distinct trend, and cannot be soaked up in the constant T0. The amount by which you can adjust k is limited, because you must match the variability. Ergo, there is little room for anthropogenic inputs.
" It is simply that, as atmospheric levels of CO2 have risen, the rate of some uptake processes has also risen."
Occam's Razor - there is no need for such exotic behavior. There is no need to conjecture that the Earth is somehow behaving precisely in a manner to make it appear that CO2 is being driven entirely by temperature related pumping. The simplest explanation is that it is, indeed, being driven almost entirely by temperature dependent pumping, and anthropogenic inputs are rapidly sequestered.
Seems to me Nullius in Verba has it right, though I find it easier to visualise if I DON'T think in terms of pocket-money and gobstoppers. The proportion of anthropogenic CO2 in the atmosphere is no indicator of whether or not it is responsible for the rise.
On the other hand, however large it is, that's no indication that the rise isn't due to something else entirely. Anthropogenic could be 10%, and uptake could have expanded handily to take care of it. Ocean outgassing, on the other hand might have increased 20%, completely blowing the system. Unlikely, I admit, but our information on these things is so vague.
The informed sceptic opinion these days seems to be that the CO2 increase really is definitely anthropogenic. I find that a little disturbing, because the reason given is usually "What else could it be?" - to my mind a typically warmist mode of argument.
I'd rather say "I don't know," - and then go on to say what I do know. That's how science is supposed to progress; a little bit of certainty at a time.
Bart wrote: There is no "if". There is no "however". The empirical data show that the f(T) term is undeniably k(T-T0). The only arbitrary constants which could be distributed among the two possible inputs are k and T0.
In science, an experiment, especially a single experiment, doesn't prove a theory! You can say that k(T-T0) is CONSISTENT with the data, but anything further begins to resemble religion conviction rather than science. Hypothesis like yours may become theories if they survive other tests and if all other hypotheses are found to be incompatible with the data. You haven't actually tried to see if a conventional hypothesis (such as I outlined) is also consistent with the data. Normally, people challenging your hypothesis would be expected to show that an alternative is possible, but they have already been published. And those alternatives also take into account the ice core record and the fact that any sensible theory of atmospheric CO2 must include human emissions. These failing make your proposal extremely unlikely and unworthy of discussion - except for the fact that your derivative plot demands a larger role for temperature dependence than I expected.
You appear to be proposing that human CO2 emissions are being rapidly sequestered by some unique mechanism. Occam's Razor says that the natural processes operating before anthropogenic emissions determine the fate of all CO2, whether it comes from burning fossil fuel or natural emissions. Le Chatelier's Principle explains why natural uptake (but not natural emission) has increased as atmospheric CO2 has risen. Enhanced uptake isn't an exotic conjecture needed to explain this particular data set. A high-school chemistry test question on equilibrium and Le Chatelier's Principle might ask a student to predict what will happen in this situation.
"And those alternatives also take into account the ice core record and the fact that any sensible theory of atmospheric CO2 must include human emissions."
But, they do not match the short term variations in the derivative. They are simply top level matching of low order polynomial behavior, a matching which is not at all unlikely to appear merely by chance. As you say, correlation is not causation. Low frequency correlation of that sort is not-causation squared. And, that low frequency correlation, as I have explained and shown, is presently diverging. The rate of change of atmospheric CO2 has stalled along with temperatures, while emissions continue accelerating.
" You can say that k(T-T0) is CONSISTENT with the data, but anything further begins to resemble religion conviction rather than science."
You do not appear to be getting the point. The atmospheric concentration rate of change matches k(T-T0). If you want to add something in, you have to choose values for k and T0 which allow it to be added in. If you change k, you cannot change it much, because you will not match the variation. You cannot vary T0 to allow anthropogenic inputs in because they have a trend, and T0 is effectively a constant over the modern era.
"You appear to be proposing that human CO2 emissions are being rapidly sequestered by some unique mechanism."
Why unique?
"Enhanced uptake isn't an exotic conjecture needed to explain this particular data set."
It is if you presume anthropogenic inputs are the reason for the rise, because the rate of change of atmospheric CO2 has settled out in the last decade plus, while anthropogenic emissions continue accelerating.
I'm not going to argue anymore. It's all been said upthread. This is a slam dunk. It is very ordinary behavior for a feedback system. If you do not see it, keep watching. You will.
Bart: I got the point: The hypothesis that dCO2 = k(T-T0) is CONSISTENT with observations at Mauna Loa. Unfortunately, that hypothesis isn't consistent with earlier observations.
You say that an 0.5 degC increase in temperature in the last 60 years has increased atmospheric CO2 by 80 ppm. However, when temperature fell at least this much during the LIA, CO2 fell only 10 ppm. An 80 ppm change in CO2 is what was seen during the last ice age when the temperature change was 5 degC, not 0.5 degC. Based on the change during the LIA, 0.8 degC of warming since 1900 has increased CO2 by roughly 10-15 ppm, man has emitted enough CO2 to raise CO2 by 225 ppm and enhanced uptake driven by higher levels of CO2 has removed about 120 ppm.
We agree that there is temperature dependence, BUT the temperature dependent emission of CO2 can follow some other function of temperature BESIDES k(T-T0). I suggested that this month's temperature-dependent CO2 emission from the mixed layer could depend on SST warming over this month. dCO2 = k*dT. In complete terms: dCO2 = k*dT + 0.5*aCO2. (Roughly half of the anthropogenic CO2 is consumed by enhanced uptake since 1960.)
Google "George Tenet" and "slam dunk".
"The hypothesis that dCO2 = k(T-T0) is CONSISTENT with observations at Mauna Loa."
Hence, to put anything else in, you have to subtract part of that out. But, there is no part you can substantially subtract out, and substitute with anthropogenic emissions, which will allow as good a fit.
The fit is across both low frequency (the trend), and the variation. Fitting those two components requires only one value of k. The odds of that occurring by pure happenstance are essentially nil.
"Unfortunately, that hypothesis isn't consistent with earlier observations."
A) Those observations are unverifiable, hence matching them superficially with some model could be worse than ignoring them. Rather than provide information, they could be providing mis-information.
B) There is no requirement that the dynamics be time invariant. In the modern era, we can say what they are using the best, most certain, most precise data available. And, since the lion's share of the rise occurred in the modern era, that is all we need.
"You say that an 0.5 degC increase in temperature in the last 60 years has increased atmospheric CO2 by 80 ppm."
No! I do not say that at all. I say that there is evidently a temperature dependent pumping action into the atmosphere going on. I described earlier in this thread how such activity could occur.
"... man has emitted enough CO2 to raise CO2 by 225 ppm and enhanced uptake driven by higher levels of CO2 has removed about 120 ppm."
Which proves nothing. Feedback systems typically attentuate disturbances. That's the very reason we use feedback loops in hi-tech devices. The Earth's CO2 feedback loop is closed by sink activity. If the sinks are powerful (and by all indications, they are), they can easily adjust to effectively remove essentially all human inputs. Substantial changes then require a shift in the natural equilibrium dynamics. Any such natural shift will have temperature dependence. This is precisely what we observe. Emissions, on the other hand, are not temperature dependent.
" In complete terms: dCO2 = k*dT + 0.5*aCO2."
This model does not fit 1958-present. There is a trend in dT. If you choose k to match the variability in dT to that in dCO2, you wil also match the trend in dCO2 quite closely. Putting in 0.5*aCO2 on top of that will produce essentially double the observed trend. Given the amount you can adjust k to maintain a reasonably good fit with the variability, the amount of aCO2 you can get by with adding is substantially less than 0.5.
Given that you can get a darn good match with 0*aCO2, Occam's razor recommends you should just dispense with it entirely.
"There is a trend in dT"
Note: Your nomenclature is imprecise, and I have tried to adapt to it. I assumed by "dCO2", you really meant the derivative of CO2 with respect to time, and by dT, you meant the difference between current temperature and an appropriate baseline.
If, however, you meant to remove the trend in temperature by differentiating it, then your model would have a 90 deg phase shift for the variational components, and that also does not fit the data.
Bart: My nomenclature was indeed imprecise. I have been using dT to be deltaT, the change in T over a month or perhaps a quarter. Ideally we would use SST for T because my assumption is that CO2 is coming out of the mixed layer of the ocean. I'm not sure how long it takes heat from the surface of the mixed layer to penetrate and release CO2 to the atmosphere, but the mixed layer responds to seasonal changes, so the lag in the response can't be particularly long. dCO2 is deltaCO2 over the same period as deltaT.
"dCO2 is deltaCO2 over the same period as deltaT."
Then, the phase does not match. There is a 90 degree lag between temperature and atmospheric CO2. That indicates an integral relationship, such as the dCO2/dt = k*(T-To) I have proffered.
It is a mathematical equivalence relationship for natural, minimum phase systems. You cannot change the phase and amplitude of the response arbitrarily: 90 deg lag = integral.
If you are interested in the details, the equivalence relationship arises from an obscure bit of mathematics known as the Bode Gain-Phase Theorem.