Monday
Apr162012
by 
Bishop Hill
Bishop Hill Quote of the day
Apr 16, 2012 Climate: Models One climate modeller we interviewed explained that the climate is a ‘heterogeneous system with many ways of moving energy around from system to system’ which makes the theoretical system being modelled ‘tolerant to the inclusion of bugs'.
From the Pipitone and Easterbrook paper on validating climate model software, currently the subject of a guest post at Judith Curry's.
Reader Comments (129)
BB,
I'm in the process of writing this up anyway, so here is a reply which may be longer than you wanted....
Richard Betts (a research manager at the Met office who posts here from time to time) lead me to the Bern model, via the IPCC report:
The AR4 table he refers to is here. This refers to the Bern model and gives references.
Re-writing the formula given there, they say
"...the decay of a pulse of CO2 with time t is given by
0.22 + 0.26 exp(-t/173) + 0.34 exp(-t/18.5) + 0.19 exp (-t/1.19)" [A]
---------------------------------------------------------------------------------------
Flashback; At the age of about 12 or 13, I was fascinated by audio frequency electronic design.. Practical Wireless magazine published a wonderful artcle explaining how you can analyse even quite complicated RC circuits with tolerable accuracy (max +/- 3dB error) with only simple arithmetic - no need for inversion of matrices with complex coefficients. Analysing and building many dozens of tone controls and other circuits gave me a pretty good physical insight into the behaviour of RC circuits.
Then, as a final year EE undergraduate, I got completely fired up by a course on the analysis and synthesis of passive RC circuits - what physcally can/can't be done by passive RC's, the detailed maths of positive-real complex functions and so on. The end result is that I have an ability to look at an impulse response and instantly see that it could not result from a passive RC circuit.
--------------------------------------------------------------------------------------------------------------
The "box models" for CO2 distribution between deep ocean, atmosphere, biosphere are (if you assume linearity) equivalent to passive RC circuits. (Capacitors representing the reservoirs of CO2 and resistors representing the diffusion paths between reservoirs.) Without going into detail here, it jumped out at me that the formula [A] could not possibly be realised by passive RC circuit.
(They got the formula by simulating the "Bern model" and then fitting arbitrarily chosen exponentials to the response by least squares, not by doing a linear circuit anlysis of the model - which would probably have been quite easy using Spice software...)
As you'll know, radioactive carbon 14C is chemically identical to normal carbon and its compounds have physical properties very close to those of normal carbon. If we can assume linearity of the dynamic systems involved, then the dynamics of 14C O2 will be virtually the same as normal CO2. Linearity means they do not interact (amongst other things). It also means we can look at the reponse to an injected pulse, without being afected by the ongoing dynamic equilibrium interchange between ocean/biosphere/atmosphere.
You'll also know that around 1958 - 63 there were very large atmospheric nuclear explosions and after that essentially none. So there was essentially an impulse of C14 injected into the atmosphere. The half life of C14 is long (>5000yr) so the reduction in atmospheric 14CO2 tells us that it disappeared in pretty well the same way that any CO2 will disappear.
Here is a D14C plot.
This looks like a decaying exponential. Plotted on a log scale, it gives a dead straight line - telling us it really is a simple exponential decay. (Note that this graph plots DeltaC14, not absolute C14 levels, so it does not directly give CO2 dynamics.)
Without at the moment saying which one I think is nearer the reality, the physical measurement of C14 or the Bern model, the fact is they give results for the dynamics of CO2 concentration that are orders of magnitude apart. There is no way they can both be close to reality.
Apr 20, 2012 at 8:20 AM Arthur Dent
Yes, equally a catastrophe in the management of change in organisations as a software catastrophe. Another illustration of how governments have infinite capacity for getting things wrong.
The large projects fail because they ask everyone what they want, and then try to include it.
But no two Hospital/Police/Ambulance regions work in the same way. Or have to same quality of people to describe how they currently work and how they would like to work.
The Consultant Companies in charge want a a big a set of requirements as possible. They are not interested in managing global change. And they often add very little value. And no one actually has the authority to get everyone moving forward together.
Programming is not the failure of these projects, it is the lack of effective process re-engineering, and the correct implementation of any changes.
Apr 20, 2012 at 10:21 AM Jiminy Cricket
Programming is not the failure of these projects, it is the lack of effective process re-engineering, and the correct implementation of any changes.
Well. I'd say it's a failure to be willing to start programming before you know what it is you are programming.
But I also know that it's also not easy to go home and say "Hello dear. I just resigned my job on a point of principle. How do you fancy going back to supply teaching?".
Martin A - a facinating analysis of CO2 decay. When / where will you publish?
Aha, MA, an experiment, unplanned as it was. It's not one that will be soon replicated, but fascinating.
I've long thought that the published figures for anthroCO2 remaining in the atmosphere were way too long, simply because they didn't account for negative feedback factors only recruited with higher CO2 levels. But I like your thoughts, too.
================
Maybe in a few weeks or months (but don't hold me to it). I'll probably post it on a website of my own and then provide a link to it from a posting on BH.
It won't be published in a professional journal, for two reasons:
- I doubt that what I am doing is substantially original - a pre-requisite for publication in a proper journal.
- Publication in a proper journal would require proper reference to previous or related work. I don't have access to a university-standard library (nor the time nor inclination) to make a proper study of previous work.
The question of whether fossil fuel CO2 can sigificantly affect the climate depends on two things:
- The effect of atmospheric CO2 concentration on climate. We know that quantifying this depends on the use of GCM models which are intensively discussed and studied, whatever reservations one might have about their use of positive feedback assumptions and inability to be validated.
- It equally depends on the relation between fossil fuel burn rate and increase in atmospheric CO2, which seems to have attracted much less discussion as a question. Clearly, a long residence time would mean that CO2 levels would rise far higher, with greater effect on climate, than a short residence time. The IPCC references the Bern model which predicts long CO2 residence time although many other publications, not referenced by the IPCC, estimate far shorter residence times.
I'm doing this work just for my own interest.
Somehow I have more confidence in things I have worked out myself, from first principles, using simple methods, and data from physical measurements, than I do from published "climate science" papers - even when they give results I would like to believe.
Martin,
Can I urge you to reconsider publication in a peer-reviewd jounrnal? What could be more important than a major revision to the scientific analysis that is set to cost this country £100 billion in CAGW mitigation? As for originality, I am unaware of any similar publications (but that is not to say that they do not exist.
I agree with your comments concerning CO2 concentrations being as much a part of the story as GCMs. Indeed, as the CAGW conjecture relies as much on the CO2 "hockey stick" as the temperature one, I believe that this is an area worthy of more serious study. I, for one, would like to understand why Ernst Beck's sysnthesis of chemical CO2 measurements differs so markedly from the ice core data.
Concerning GCMs (the subject of this thread), I think that their fitness for purpose rests upon a hypothesis, that could be stated:
GCM Hypothesis – Numerical models can provide valuable information on the future of the Earth's climate in the decades to come, particularly in relation to anthropogenic GHG emissions.
The scientific method requires us to attempt to falsify this hypothesis. Here is my attempt, separated into temporal segments, bearing in mind that it has been stated that the same models are used over all timescales:
1. Numerical models are claimed to have increased the accuracy of weather forecasts over a timescale of several days, however, it is just as likely that any improvement in recent decades is the result of modern data collection methods (satellites, weather radar, etc.).
2. Numerical models have been shown to be inaccurate over timescales of a few months. Evidence – the”barbeque summer” argument.
3. Numerical models have been shown to be inaccurate over decadal timeframes. Evidence: empirical data showing a lack of statistically significant warming over the last 10 – 15 years, contradicting model predictions made 10 – 15 years ago (reference – WUWT database). (The counter argument is that “internal variability” (El Nino, La Nina, volcanoes, etc.) mask trend data over a 10- 15 year timeframe).
4. It is claimed that, following a period of internal variability, numerical models will begin to produce valuable trend data, which can be used to analyse the effect of GHG emissions on the climate. There is no evidence to support this claim, indeed it is mathematically counter-intuitive, as as drift and errors (in input data and incorrect or incomplete algorithms) have a cumulative effect in a numerical, time step integration, thereby progressively deviating from reality.
I would welcome your thoughts on this, and those of anybody else.
Martin D. Interesting posts. Related to this, at http://www.john-daly.com/dietze/cmodcalD.htm there is a long mail thread discussing the subject ("The CO2 excess lifetime and C14 puzzle"). There is a useful description about half way down (search for "Answers to questions, finally"). Maybe this has a bearing on your calculations (or maybe not :-)
Previous post should have been addressed to Martin A. Sorry.
BB,
Thank you. I remember looking at that site some time back. The thing that astounded me (from what I recollect) were some of the elementary misconceptions of dynamics and of physical science of some of the scientist who replied to Peter Dietze. Including one who thought that the chemistry of C14 is different from the chemistry of C12.
Climatologists do not take kindly to the idea that electrical circuits can have the same dynamic behaviour as climate systems. The fact that the differential equations are the same passes them by. I posted some simple questions on Realclimate some time back, but the resulting hostility had to be seen to be believed.
_________________________________________________________________
Roger,
Thank you. Let's take things step by step. If, after BH readers have seen what I'll have written and had the chance to point out any fallacies, we can see if it makes sense to take things further.
But don't kid yourself. The Great Delusion is not going to come to an end because someone has refuted some paper referenced by the IPCC. It has taken root and it has a life of its own, independent of any science.
On GCM models...
I was amazed when I first read that the Met Office use essentially the same model for all timescales. I still find it hard to beleive.
To anyone who has simulated physical systems of any sort and had to validate the simulations, the idea that you can use a model to predict changes in climate is just laughable. The system is so complex, with so many aspects that are poorly understood, and with so many approximations and linearisations needed that whole thing should have been laughed out of court years ago.
And yet there are apparently sincere, well qualified people taking the effort seriously. So far as I can see, they are simply hypnotised by groupthink.
Here is what the Met Office itself says about the validation of their models:
To me a statment like that is so ludicrous, I'd be embarrassed to publish it.
It is mendacious at various levels. It implies that, except for assumptions about greenhouse gas emissions and perhaps a few other things, their models could predict the future precisely.
It repeats the fallacy that the ability of a model to reproduce the data used to construct it is a validation of its ability to predict the future. A completely incorrect model can reproduce the past with zero ability to predict the future. And, if the physical system is chaotic, even a precisely correct physical model may have no ability to predict the long term future.
Does it? As someone (Dawkins?) said, if I state my hypothesis that there is a yellow teapot in orbit around the sun, it does not oblige someone else to falsify my ludicrous claim.
To me, the Met Office's claim to be able to predict the future climate is on the same level of probability of being a valid claim as the orbiting teapot. So to go back to the original subject of this thread, a few bugs here and there in the code of a GCM does not really affect the value of what comes out of it.
Thanks Martin,
I agree with everything that you say, except for the need to falsify the GCM hypothesis. While it may be as barking mad as Dawkins' orbital teapot, the teapot is not going to cost us £100 billion!
Good luck with your paper.
Teapots
In this case not the fundamentalist atheist but a previous atheist Betrand Russel
http://en.wikipedia.org/wiki/Russell's_teapot
Martin A. The reasons you gave for not publishing in a journal are, if you will excuse me, rather serious. Please don't take this as an insult, as it is not meant as such.
In doubting that your work is original and failing to reference previous work, you effectively admit that you haven't read the existing literature and that the work is the product of your intellect and experience alone. But you are attempting to overturn 'accepted science' (whatever you might think of that). It is equivalent to claiming that you are uniquely capable of deducing the dynamics of the carbon cycle; that you know all there is to know without reference to earlier work. That is quite some claim, and while it might be true, the probability is low. As a result, the work will not be taken seriously.
BB, I agree with what you said, but is that not the point of peer review - to weed out the papers that are not original or are just plain incorrect?
AD: Thank you for the Russell teapot background. I think Dawkins may have recycled the argument.
RL, BB: Thank you for your comments.
As I said before I'm doing this work just for my own interest. If anyone else finds it of interest, that's great. But if it happens that "As a result, the work will not be taken seriously", I won't give a toss.
Note that I am not "effectively admitting" that I have not read the existing literature. I have stated it explicitly.
I don't have the resources (library access, time) to do it. Nor do I have the motivation, since I am doing this work for my own interest, rather than to add to the scientific literature and I have other things that keep me fully occupied. I'd estimate six months to a year of full-time effort to read the entire relevant literature.
I don't think I am "attempting to overturn accepted science" - I think there are plenty of refereed publications that indicate a short atmospheric residence time of injected CO2. I imagine that any analysis I do is likely to agree with the most, if not all of those published papers. However, the IPCC chose to disregard those many peer-reviewed papers and refer to the Bern model, whose predictions seem to be vastly different from the rest of the literature.
"It is equivalent to claiming that you are uniquely capable of deducing the dynamics of the carbon cycle".
I disagree that I am doing the equivalent of making such a claim. (And note that my interest is limited to empirical observation of the atmospheric residence time characteristics of injected CO2, not understanding the dynamics of the carbon cycle with its various reservoirs.)
However, if pressed, I might concede that there are probably zero "climate scientists" who have anywhere near my depth of understanding and my level of accomplishment and publication record in research into the mathematical theory and my record of innovation in the engineering practice of signal processing and system dynamics.
Thanks again for your comments, which I'll keep in mind.
Good stuff Martin. It's the best answer to BB's base attempts at provocation.
Roger Longstaff: yes you are right. But I imagine that thousands of papers are submitted for publication and only a minority go out to peer review. Most will be rejected before that stage by the publishers. Remember that reviewers are not paid, so publishers cannot afford to test reviewers' patience with papers that do not follow accepted conventions (accepted across all sciences, not just climate).
Martin, as I don't know what the 'A' stands for, clearly I have/had no idea of your publication and research record. As I said, my comments were in good faith.
Thanks BB. The beauty of Martin's work is that it gives a half life (chemical, not nuclear) for CO2 residency that can be used in a differential equation (about 12 years - is that right Martin?). I will cetainly use that in any further work that I might do (as an amateur) in this area. It is simple, elegant, and does not require the torture of data - real science, as far as I am concerned.
BB: Thank you for your comments - I appreciated all of them.
My own experience both as author of papers and as reviewer is as follows. Nearly all papers submitted to a journal will be sent to review - an editor will only reject papers without review if there is a problem such as not being in the subject area of the journal, being of excessive length or otherwise not following the journal's guidelines. Of those, I'd hazard a guess that maybe at least half finish up being published - a paper will normally only be recommended for rejection by its reviewers if not original, erroneous, trivial or not in subject area covered by the journal. (I'm talking about maths/physics/engineering - as we have seen, things seem different with some climate science journals.)
Because of the "publish or perish" aspect of academic life, anyone whose papers are repeatedly rejected will probably migrate to a different career path in a shortish time span.
Roger - yes, it seems that a 1st order differential equation applies. That's the right order of magnitude for the CO2 half-life, but wait until I've plugged in the latest data for a specific figure. I have just received C14 data up to 2011.
BTW - I think we are all amateurs - it's a term with a long and respectable tradition.
the theoretical system being modelled (is) ‘tolerant to the inclusion of bugs'
Translation: it gives the answer we want even if there are mistakes in it.
Martin, a question (hopelessly O/T, sorry bish):
If we subtract the "bomb pulse" from TOTAL C14 in the atmosphere, can we quantify the ratio of anthropogenic CO2 to "natural" CO2 as a function of time? (assuming that anthropogenic CO2 is from fossil fuels, where the C14 has decayed).
Martin A, I'm not sure I understand your model completely but based on your discussion of 14CO2 pulses I wonder if you might be missing something. Basically, there must be two timescales at work. Atmospheric CO2 can exchange with CO2 in the upper ocean. From what I've read, that occurs fairly quickly (half-life of 10 years or so) but it is also reversible. And there's a lot of CO2 in the upper ocean - much more than in the atmosphere. If that was the only thing that happened, and you suddenly added a load of 14CO2 to the atmosphere, then the concentration of 14CO2 would drop off with a half life of 10 years, but the total amount of CO2 in the atmosphere wouldn't change.
In fact, there are other processes that remove CO2 even from the reservoirs that can exchange with the atmosphere. Those are the ones that would drag CO2 concentration down if anthropogenic emissions dropped. Their speed is not so well known. Consensus-hewing IPCC sorts obviously think they are very slow.
Martin - my question @ 8.47am was answered when I found the "Suess effect", which considers exactly this problem. Reading around it a bit, most of the discussion regards C13, which may have ambiguous sources and sinks. I would think that C14 analysis - quantified by the bomb data - would give a clearer result.
Apr 22, 2012 at 9:42 AM Jeremy Harvey
Jeremy, Thank you for the question/comment. I think it is far from impossible I am suffering from misconceptions, so I welcome critical comments.
I'm writing this stuff up in the hope that I when I post it, then any misconceptions and errors will be pointed out by people looking at it critically. Plus pointing out any implicit assumptions I may be making without being aware of it, or explicit assumptions that may be unjustified.
Here, in rough outline, is what I am assuming about the point you have made:
The overall system of CO2 reservoirs (surface ocean, deep ocean, middle deep ocean, ice, biosphere, lithosphere (or whatver it's called - where CO2 gets absorbed in making limestone etc) is immensely complex and incapable of being modelled in detail. But let's assume we can simplify the whole thing to nothing more than:
atmosphere <=CO2=> shallow ocean
with CO2 continually being exchanged between them. If the whole thing is in equilibrium (eg no fossil fuel CO2 being injected into the atmosphere, climate is constant), the rate of diffusion from atmosphere to ocean will be constant and equal to the rate of diffusion from ocean to atmosphere.
Here is the essential and critical assumption: The system is linear.
(I think I am making numerous other assumptions but they can be discussed later.)
Another assumption (not essential but simplifies things) is that the capacity of the ocean to absorb CO2 is infinite. This means that the concentration of CO2 in the ocean remains constant, irrespective of how much has been absorbed/released by the ocean recently. It also means that an injected dollop into the atmosphere eventually will reduce to zero. That is to say, the atmospheric CO2 concentration will eventually reduce to what it was before the dollop injection.
These assumptions mean that the dynamic exchange between the atmosphere and ocean continues at its constant rate of exchange, unaffected by the sudden injection of a dollop of 'labelled' (14C O2) carbon dioxide. Because C14 is radioactive and was injected by bomb testing in a big dollop (roughly doubling the atmospheric CO2) C14 measurements tell us what happens to an injected dollop.
Because of linearity, there is no interaction between the 14C O2 absorption by the ocean and the ongoing dynamic equilibrium of non-radioactive CO2,
There would equally be no interaction between the ongoing equilibrium and an injected dollop of non-radioactive CO2, though the response to the latter has no way of being directly measured.
Linearity also means that the time to for an injected dollop decay (eg reduce to 50%, 1/e, or any other fraction you wish) is independent of the magnitude of the dollop. Because of linearity, if we know the impulse response - the response to an injected dollop - we can then readily compute the response to any time profile of CO1 injection.
This misses out detailed arguments and was written quickly but I hope it makes sense.
Please let me know if this:
- seems to make sense
- misses the point
- is incomprehensible
- seems wrong
Thank you for your help with this.
SkS has an article on this: http://www.skepticalscience.com/co2-residence-time.htm
I think they confused. They seem to be confusing the ongoing equilibrium exchange with what happens to an additional injected quantity of CO2.
So far as I can see, the (statistical) atmospheric lifetime of a CO2 molecule is the same as the lifetime of a substantial injected quantity. Using an appropriate definition of lifetime of course.
Martin A, what you wrote was mostly comprehensible, but I do think it is wrong in one important respect. You treat 14CO2 as being a different species than 12CO2, with a different reservoir. This is not right. To a first approximation, they can be treated as exactly the same thing, and it is just a book-keeping exercise to work out which one is where. The exchange processes for CO2 can be treated linearly (first-order kinetics), though.
It helps to look at this diagrammatic representation of the carbon cycle. It is credited to the IPCC so it must be right ( ;-) ) - no, I really do mean that it must be about right for the present purposes. There's a similar, but more complicated plot on Wikipedia.
If instead of focusing only on the atmosphere and the ocean, as you and I both did above, and instead lump the upper ocean and the biosphere together, you have basically three important places where there is CO2: the atmosphere (A), the biosphere and upper ocean (B), and the deep oceans (D). A contains about 660 billion tons of "Carbon" (I guess that is carbon-equivalent of CO2 - so all masses henceforth should be multiplied by 44/12, the relative mass of CO2 and carbon, but that does not change anything important so I have not done it), B contains much more - about 3,000 billion tons, and D contains an effectively infinite amount.
The exchange A <--> B occurs fairly quickly, because as you can see, there's about 200 billion tons go each way each year (120 into/out of plants, 90 into/out of the upper ocean). To a first approximation, there's little exchange with D (none shown on plot).
So, if you suddenly add a slug of 14CO2 to the atmosphere (say, 1,000 tons). What will happen, if we neglect the exchange with D?
The first year, the atmosphere A will contains 1,000 tons of 14CO2. But during that year, about 1/3 (200/660) of ALL the CO2 in the atmosphere goes into B. And a similar amount of CO2 comes out of B into A. As a result, A now contains ca. 660 tons of 14CO2, and B contains about 340 (assuming there was none there to start with, which I suggest we do, to make things more simple). The total amount of CO2 (12CO2 and 14CO2 - and indeed 13CO2) in A and B has not changed. The next year, again 1/3 of the CO2 in A goes into B and gets replaced by CO2 from B. Here, we need to be careful because there's now some 14CO2 in B, but roughly speaking, another 220 tons of 14CO2 go into B, but a small amount does come out also, so the drop is a bit smaller than 220 - say 200, leaving you with 460 tons. Then the next year, another 1/3 goes.
If you write the corresponding differential equation and integrate, you can get analytical expressions for the amount of 14CO2 in A and B. These will show a decay with a half-life of about 2 years. We've made some approximations, though, and I think if you have a more realistic model, you end up with a half-life of about 10 years as I mentioned previously.
One important thing to note here is that the total amount of CO2 in the atmosphere HAS NOT CHANGED. The only way that can happen is if you further have CO2 moving from B to D - as a first approximation, we assumed that does not happen. In fact, the Wikipedia page shows that exchanges between B and D do occur, and I guess that if the total amount of CO2 in A + B goes up, then what happens is you get a net transfer of CO2 from B to D. But this is fairly slow. It is this sort of transfer which would ultimately make the amount of CO2 in A go down if human-linked emissions stopped. You can get some idea of how fast it would happen from knowing things such as that year on year, roughly one half of anthropogenic CO2 emissions do not end up in the atmosphere, so must instead be in B or D. But I think the speed of exchanges to D is not perfectly known.
My view is that the IPCC picture whereby the drop in CO2 concentration in the atmosphere upon stopping emissions is slower than the known 10-year carbon cycle between A and B is certainly correct. Your 14CO2 pulse example can only tell you about the A <--> B exchange. I'm less sure that the picture whereby it is VERY slow is right - I think the part of the carbon cycle involving deep oceans is not perfectly understood, and I assume that within the wiggle-room of plausible interpretations, IPCC-types assume the "worst case" scenario of very slow exchange rather than just slow.
I hope that this, in turn, makes some kind of sense!
... I saw your 8:28 AM post only after clicking 'Create' on mine. Basically, I'm agreeing with SkS (!) - I think they are right in qualitative terms, when they write:
As you'd expect from SkS, they go on to claim that the transfer into 'D' is very very slow, and suggest that anyone who doesn't believe that is an evil denier :-), but I think that is less well-known.
Jeremy,
"Martin A, what you wrote was mostly comprehensible, but I do think it is wrong in one important respect. You treat 14CO2 as being a different species than 12CO2, with a different reservoir. This is not right. To a first approximation, they can be treated as exactly the same thing, and it is just a book-keeping exercise to work out which one is where. The exchange processes for CO2 can be treated linearly (first-order kinetics), though."
I agree.
Both with what you say I am doing in treating 14CO2 as having a different reservoir and with what you say about 14CO2 and 12CO2 being the same thing (to a close approximation). Obviously, to convince you, I need to do something different than just repeat my previous words. (Linearity, which means that the superposition therem applies, is the key to my argument.)
Let me carefully re-read what you (and SkS) have said and then work out how to present my argument in a way that I hope you will find convincing before getting back to you.
If you would care to send an email to spamfree1 [funny email sign] v8pilot.com, we could continue offline. Otherwise, I'll post my reply here.