Is Stratospheric Cooling Proof of CO2-Induced Global Warming?

[Gokul43201]

I think that what Andre is pointing at is that it cannot be a simple, direct positive feedback and that's it.

Vanesch: I can't even say I understand how that is done. If the signal was cleaner, and we were looking at the correct time scales, then I might understand at least some of that. But, in what we have, the RMS noise level seems to be at least comparable to changes in the signal during the estimated lag or lead times. I just don't understand why there shouldn't be that level of "conflict in trend" (no matter what the transfer function is) in such a noisy signal. The size of this "conflict" seems to be what I would expect from this noise level.

But that's also just some naive intuition on my part. Is this wrong?

 

[vanesch]

Vanesch: I can't even say I understand how that is done. If the signal was cleaner, and we were looking at the correct time scales, then I might understand at least some of that. But, in what we have, the RMS noise level seems to be at least comparable to changes in the signal during the estimated lag or lead times. I just don't understand why there shouldn't be that level of "conflict in trend" (no matter what the transfer function is) in such a noisy signal. The size of this "conflict" seems to be what I would expect from this noise level.

But that's also just some naive intuition on my part. Is this wrong?

I see what you mean. I also just do some eyeballing! The main RMS noise level seems to be rather high-frequency, and I would guess that if one does some linear regression on the 3 parts (the "horizontal" part before the rise, the "rising slope", and then the part in the "conflicting region", that one would find a slight decrease. To do this seriously would require to put up a model and to do a statistical test on the compatibility between the model and the data, and more importantly to have a reliable model of the noise on these data (white noise, red noise, ... ). Now, you said at a certain point that in "saturation" of course the positive feedback is gone now, that's true. However, it cannot be "gone now", and work again at full power for the next rise (unless the model has changed in between). So we have to assume that the positive feedback is working fully in the region between 14 000 and 12000 years, in which case a negative slope is somehow problematic.

Of course, the question is of where the burden of proof resides. If Andre says "I've explicitly demonstrated that no positive feedback mechanism is possible" then that is a strong claim (a kind of no-go theorem), which will need much more proof than just this plot, I agree. It would imply the rejection of the whole family of dynamical models containing a positive feedback, and furthermore an explicit specification of that exact family.

However, it if it is "I emit some doubts about these data to be compatible with a given feedback mechanism as is verbally described on < fill in your preferred AGW blog > ", then I'd say, this is just a request of "show me that famous model of yours that reproduces these data" and which *demonstrates* the existence of that famous positive feedback.

 

[Gokul43201]

Of course, the question is of where the burden of proof resides. If Andre says "I've explicitly demonstrated that no positive feedback mechanism is possible" then that is a strong claim (a kind of no-go theorem), which will need much more proof than just this plot, I agree. It would imply the rejection of the whole family of dynamical models containing a positive feedback, and furthermore an explicit specification of that exact family.

However, it if it is "I emit some doubts about these data to be compatible with a given feedback mechanism as is verbally described on < fill in your preferred AGW blog > ", then I'd say, this is just a request of "show me that famous model of yours that reproduces these data" and which *demonstrates* the existence of that famous positive feedback.

Here's what has been said (in the other thread that he linked):

So I used this high resolution graph of Antarctica's EPICA Dome C ice cores during the last glacial temination between 20,000 and 10,000 years ago, to demonstrate that there is no positive feedback because the typical behavior of positive feedback is not seen.

(emphasis mine)

Maybe it would be a good idea to acknowledge the limitations of that demonstration?

 

[vanesch]

Maybe it would be a good idea to acknowledge the limitations of that demonstration?

It's what I did, no ?

Now, maybe we can make this more explicit, and try to write a kind of toy model that could eventually be refuted by these data, just for the fun of it, like (guessing right here on the spot):

Two state variables, T (temperature) and C (CO2 concentration) ;

dT/dt = A C + B + f(t)

C = U T(t-t0) + V + g(t)

where T(t-t0) indicates a delay, and f and g are external driver functions.

A, B, U and V being the system constants to be estimated. We take it to be a linear model (we could introduce some non-linearities, but then extrapolating this model to other regions of operation is even more tricky).

If it were possible to demonstrate that there exist no 4 numbers A, B, U, and V which make statistically compatible predictions with the data, then we have falsified this toy class of models.

This is a trivial toy model of course. We could start here. It corresponds to the bare bones implementation of the verbal statement about the feedback in its most trivial and simple form.

edit: note that this toy model reduces to something much simpler:
dT/dt = alpha T(t-t0) + beta + gamma(t)

with alpha and beta constants, and gamma(t) a driving function.

 

[Skyhunter]

Vanesch,

Concerning the lower stratospheric cooling, this is usually seen as a proof for CO₂-induced global warming. The simple explanation is that as more IR radiation is withhelt by the lower troposphere, less of it reaches the stratosphere so this one cools.

That is not quite correct. The dominant CO₂ absorption band, 15µm is saturated in the lower atmospheric levels meaning that fewer photons that can be absorbed by CO₂ reach the stratosphere. Carbon dioxide having a zero dipole tends to spontaneously radiate. Hence stratospheric CO₂ radiates away more energy than it absorbs.

http://www.atmosphere.mpg.de/enid/2__Ozone/-_Cooling_nd.html"