Bishop Hill Great Evans above
Sep 26, 2015 Climate: Models Jo Nova carries a rather interesting piece today about some work done by her husband David Evans, who thinks he has uncovered a rather major flaw in the mathematics at the core of the basic model of the climate.
The climate models, it turns out, have 95% certainty but are based on partial derivatives of dependent variables with 0% certitude, and that’s a No No. Let me explain: effectively climate models model a hypothetical world where all things freeze in a constant state while one factor doubles. But in the real world, many variables are changing simultaneously and the rules are different.
Partial differentials of dependent variables is a wildcard — it may produce an OK estimate sometimes, but other times it produces nonsense, and ominously, there is effectively no way to test. If the climate models predicted the climate, we’d know they got away with it. They didn’t, but we can’t say if they failed because of a partial derivative. It could have been something else. We just know it’s bad practice.
This sounds plausible to me. What do readers here think?
Reader Comments (199)
Nullius in Verba - You say "Evans would have to show explicitly what PDE he was talking about" because you did not read the post. The partial derivatives to which I am referring are in post 3, which describes the conventional basic climate model, "the basic physics".
Another non-reader.
James G. - You wrote a 630 word comment but obviously did not bother to read the post -- and then claim that it is I who wastes your time. Cheeky!
And since you couldn't be bothered to find out (it's linked from the top of the post you didn't read), I do modelling for a living and am an engineer (electrical, Stanford PhD, six university degrees all essentially in applied maths).
Nic Lewis, this is part of a series. There is a lot more coming. We already know the models fail because of the empirical evidence; this series is explaining the significant problems and uncertainties, and then fixes some of them.
David,
Your post 3 does not show the equations that are solved in a GCM. It shows how one might develop a basic climate model if you already know the feedback response (i.e., how the system responds radiatively to changes in temperature). In a full GCM the feedback response emerges from the model/calculation, it is not an input into the model and the model does not evolve one dependent variable with respect to another dependent variable. As far as I can tell, you're essentially making the same mistake that Christopher Monckton regularly makes. I have a feeling that even people here would find that a rather embarassing association.
"Your post 3 does not show the equations that are solved in a GCM." Well obviously not ...what is your name? There is a hint in the title: "New Science 3: The Conventional Basic Climate Model — In Full". Read post 1 for why this an interesting thing to do.
David,
Well then what are you on about? You're claiming that climate models are wrong because they solve partial derivatives of one dependent variable with respect to another, and yet when asked to show the equations you mean, you acknowledge that these aren't the equations in a GCM. Are you criticising GCMs, or basic box models?
Don't worry, someone will be along in a minute to tell you who I am.
I don't claim the GCMs are wrong because they use partial derivatives on dependent variables.
Perhaps if you read the post (you still haven't?) it would be worth discussing it with you.
@ aTTP
You really haven't read Evans' 3rd post in the series, have you ? Instead of shadow boxing, make direct comment on the actual website
Don't worry, no one cares who you are
Struts his petty face
As Willard at his pace.
===============
David,
I'm referring to the post highlighted here which says
Okay, I see what you're getting at. I was confused by your post 3 where you went on about dG/dT. You're pointing out that you can't - in advance - know what (for example) dw/dx is if w depends on a bunch of other variable (y,z,...). Okay, that's still silly though. Climate models evolve various properties in time and space. At its base are the Navier Stokes equations. The evolution in time and space is done simultaneously in a way that (to within the accuracy of the method) conserve mass, momentum and energy. You then couple that with radiative transfer calculations to determine how these changes influence the energy balance and consequently - the energy/temperature in various parts of the system.
As others have pointed out, there is nothing fundamentally incorrect about this. Many of the methods are very well understood and have been used extensively across the physical sciences and engineering. You're essentially criticising much more than simply climate models. You're criticising modern computational science.
I had, but your earlier posts going on about the feedback responses dependence on T had me rather confused. In some sense I don't think that I could quite bring myself to believe that your critique was simply "they use partial derivatives" because I didn't consider that someone could say something so obviously wrong. Having read those posts I seriously doubt that this is worth discussing with you. You could endeavour to prove me wrong, though.
Sorry, I don't have time to tutor you personally ATTP.
The series starts here. If you comment on the posts themselves maybe someone will help you. Without reading and understanding the posts in question, the conversation is a waste of time.
For any onlookers, ATTP's comments have very little to do with the posts or their contents and I invite you to come see for yourself. If ATTP was really sure of his points he'd make them where the posts are made; making assertions at BH without even reading the posts is not the mark of someone who cares about honest conversations.
David,
I read that first post. It's one massive strawman. I'm referring to your later posts where you try to explain in more detail the fundamental issues with climate models, but instead seem to illustrate that even if you're
it doesn't mean that you can't blunder spectacularly.
I would have been extremely surprised if it had been anything else.
My main reason for commenting here was to respond to Andrew's question "what do readers think?". I had little reason to think that commenting on the original posts would be worthwhile. This discussion appears to confirm this.
Could I ask everyone to focus on David Evans' post and avoid general comments about climate models or other aspects of global warming.
Also, please avoid snark and namecalling.
Attp - Ken
It might help all of us if you provided an example that overcomes the consequences of using PDEs.
Something pretty comparable to the examples David gives or will give shortly.
Expressed mathematically, rather than verbally, would also help.
Something that is known to work well despite the concerns of PDEs.
Geoff
Sadly David and Jo have become rather defensive and aggressive in this latest series of posts, no doubt due in part to the very hostile reception given to the previous series on the notch-delay solar idea. Such hostility from them does not bode well for the eventual outcome.
I have to agree with ATTP and JamesG, it is a major strawman to say that confidence in climate predictions is based on a simple analysis model of complex GCM outputs. David will only further shake (and then restore?) confidence in GCMs by dealing with them directly.
David Evans,
Some of your responses here are unhelpful if you believe that you have a valid point to make.
I did read your original post and went back to your previous post to try to dispel my confusion before making any comment. You are right that I was and still am confused, and your exhortations to actually read your posts have done nothing to help me.
It is perfectly possible that I have misunderstood what you intended to say. However I can only interpret what you actually do write, not what you intend to communicate. It seems that I am in good company in my misunderstanding of your intent. I say without heat that, instead of offering personal putdowns, you might just reflect on why it is that a number of literate individuals with skills in applied maths and numerical modeling in particular all find your post unpalatable, despite the fact that most of us, I believe, would support the view that the GCMs are wholly inadequate for the task of predicting climate with any degree of confidence.
If you are going to criticize the GCMs on the basis that there is an error of formulation or solution routine, then you really do need to spell out the formulation actually used and the offending interaction(s). The model which you set out in your “Post 3” is not a formulation used by any of the GCMs, although I accept from my own studies that for some but not all of the GCMs your model does a very fair job of emulating aggregate behavior. If you wish to argue that in fact it is isomorphic with a particular GCM formulation (i.e. it forms an aggregate governing equation for some GCM), then go ahead and prove it by derivation from an actual formulation, and then present an argument that the interactions are not dealt with correctly. I am very happy to read and learn.
As it is, your post leaves the impression that there is a generic problem in numerical analysis that means we cannot solve non-linear systems where there are dependencies amongst the variables of interest. As a generalization, this, as you are no doubt aware, is pure BS and I am willing to accept that this is not what you intended to convey despite what you wrote. However, I am STILL at a loss to understand what you ARE trying to say.
Geoff,
If you consider climate models then the fundamental equations are the Navier Stokes equations which are integrated with respect to time and space. Time and space are totally independent and so there is no major issue with using those equations. If you consider one of the examples David provides a link to in his post number 4, it applies to a situation where the variable you're trying to determine depenp on other quantities that are not fully independent. This doesn't apply to the Navier Stokes equations. There are other quantities that do depend on non-independent variable (evaporation rate and humidity that depend on temperature, for example) but - unless David can actually provide some evidence otherwise - I do not think that these involve solving partial differential equations in the way that David is suggesting. Additionally, that there are inter-dependencies does not mean one cannot solve these coupled equations. As the system evolves, you will explicitly evolve these different quantities and degeneracies will, typically, be broken. It's not that you can't do it, it's more that you can't know how it will evolve in advance, given that these are coupled, non-linear equations.
Paul_K probably says it best
From now reading the full post and the follow-up rather than the short passage we were asked to comment on, I don't get the impression that Dr Evans knows much about programming this type of numerical model. It isn't just a once-through calculation; internal variables are iterated until they converge so all dependencies are properly accounted for as far as possible. All of the underlying maths and physics are approximated by discretisation techniques via a mixture of crude and elegant techniques. So it's not any kind of a discovery he makes and there is no need for such high-handed dismissal of the modelers as not 'serious maths guys'. It is not the underlying maths that contains the main errors but the fact that it is an intractable problem in the first place due to the missing physics. The reason all feedbacks are related to temperature btw (or pressure in coupled models) is because otherwise the model is unsolvable.
This is only a simulation though - if it correctly simulates the real world to a reasonable standard then however you got there is legitimised. Of course if the model fails to match reality spatially or temporally then usually it has zero legitimacy or credibility and that is what the alarmists are trying to ignore by pretending that models and data have equal legitimacy. That false notion is the place to focus most criticism!
And btw if you add your comment at Jo Nova and are insufficiently ingratiating to her husbands genius then you will be verbally abused - so physician heal thyself! I will be interested in his final post with the new model. The so-called pde 'error' is something for a report appendix.
JamesG,
Just for the record, you are perhaps describing what should be rather than what is. You are describing a fully implicit solution routine. The released GCM codes that I examined a couple of years ago used a semi-implicit scheme with a concomitant likelihood of large-but-inestimable error propagation. There may be a fully implicit model out there somewhere, but, if so, I don't know where it is.
Perhaps one of the knowledgeable people on here could bear with me for an instant and answer the question I've posed many times over the years, are there any testable intermediate results which we could use to determine the validity of the models. It really is no good to me if the ONLY THING I EVER SEE of model results is this stupid ensemble graph stretching out years in the future. Nobody should accept that, not even a Texas housewife. Tell me what results you get that we can actually compare against observations. Not the daft global temperature anomaly, that is not really an observation at all, merely an average of a lot of real observations not compatible in any real way with the model ensemble graph as we see it. The whole caboodle is designed to deceive. Where is a checkable result? If the result does not check out, doesn't that make the temperature result, no matter how it fits, invalid?
Errors magnify. Errors put through imperfect algorithms will magnify the errors dramatically.
Redbone,
Your own description of how heat is transferred in Earth's system underscores that calling CO2 an insulator is not such a bad turn of phrase.
CO2 acts to slow down heat loss in the Earth system.
That is close enough to insulation.
Setting requirements that it behaves the same as a literal blanket is as hard headed as those who reject calling GHG's GHG's because the Earth is not actually a greenhouse.
Changing the lapse rate is part of the mechanism by which CO2 acts to slow cooling.
Hunter, does it change the lapse rate? Some explanations offered to me right here on the Hill claimed that the TOA height was changed and the lapse rate remained the same giving higher surface temps. But of course it was all averaged out far too early in the workings and was more notio0nal than demonstrable.
And it slows down heat loss. Now I've always grasped that bit, but for how long? I don't see why it is not on the order of seconds rather than centuries. Heat will not be denied. Entropy ditto. If it can't find a way out by one channel it will use another. Kinda like dependent variables?
This is getting very tiresome. That the good Bishop should bring the work of Mr Evans to the notice of those who are unaware of Jo Nova’s blog (hopefully, very few) has to be to his credit. For those who wish to use this site to dissemble the work of Mr Evans, without, it soon becomes obvious, actually reading any of his work is getting very, very irritating – and (for a change, though not a particularly refreshing change) the culprits are not just the usual suspects. However, it could give us an indication of how much we can truly believe the proclamations of others who have demonstrated that they are not willing to become familiar with the subject under discussion. It hurts that many of them have managed to present believable arguments in the past; now I have to question how much I can trust them and their depths of research.
For those who think that Mr Evans has made errors, or is wholly and utterly wrong, take yourself over to Jo Nova and discuss it with him there, rather than force him to trawl the many sites around so that he may answer your criticisms (though that you could be hoping that your comment will not be spotted so never be addressed by him, thus giving you the dubious claim that you raised a point he could not answer so kudos to you, is another possibility that cannot be dismissed). Use this site merely to discuss the progress that is being made with the argument, rather than attacking the proponent from behind a barricade of distance and (hopeful) anonymity.
Blast furnace simulation uses similar modelling techniques and similar physics to climate models.
At the bottom of the link :-
"Transferable expertise:
Mathematical modelling of complex technical systems, numerical solution of partial differential equations."
There would seem to be no problems using partial differential equations in this example of heat and material flow simulation.
Any attempt to discredit climate models solely because they use partial differential equations is, on its own, insufficient.David Evans would need to show specifically that, though valid elsewhere, partial differential equations are not appropriate to the specific case of CMIP5.
Perhaps slightly off topic.
When you compare CMIP5 output updated for recent forcings with the actual temperature record, you get the second graph here.
I find it difficult to understand how the close correspondence between the predicted temperatures using updated forcings (black line) and the temperature records can be described as wrong.
If you think that is the case, EM, why not go over to Jo Nova’s site and discuss your observations with Mr Evans? (Which is exactly what I mooted in the post directly above yours, but I do not expect you to take any note of that.)
Or are you one of those who are happy to criticise something you have not read, as seems to be common with others on this site? Perhaps the chore of clicking on a link and reading is just too much, when it is so much easier to see who the author is, then allow personal prejudices and imagination to fill in any blanks. The other option is that there is an attempt to impress others on this site with the argument against Mr Evans without actually engaging in argument with Mr Evans.
This is one of the very, very rare situations in which I disagree with you Penfold. I am not happy that many people have tried to tell me that I am in some way at fault for not reading the original (and subsequent) posts on Jo Nova made by Mr Evans. I am commenting on the post by his grace and I am perfectly entitled to do just that :)
First episode of the new Danger Mouse series was on C BBC tonight!
Sorry you think that way, Dung (and no need to look so beetle-browed at me like that!). I will comment on what has been said in links, but I try not to argue with the linked author by proxy – should I have any points I would like to raise with the original author, I try to make them on their site, though that may not always be feasible. I will hoik out quotes from them to verify my impressions gained, but, should it come to actually discussing the nitty-gritty, I would rather do that with the person concerned. You could look on my preferred methods as a case of, “Hey! Come and see what I’ve found!” then return to the site and engage – one classic (for me) was when Mr Alder and I took on the mite of Richard Telford about ocean “acidification”.
Dang! Missed it! (Or would you prefer – “Dang, Dung, deng done gorn miss dit”?)
OK, I will venture a start here. I seems to me that Evans is talking about feedbacks that depend on other things than temperature. If these are ignored, then that's a potential inaccuracy. That is an issue that requires rather deep expertise. However, the stuff about partial derivatives is I think a needless formalism that just confuses the issues. The problem here would be inadequate models of the feedbacks which neglect (possibly) important terms. So that's a valid point I think, but hardly one that requires a lengthy series of posts. :-)
May I humbly suggest that comments be closed on this, as its actually quite unhelpful.
By all means point to Jo / David's site and suggest THAT COMMENTS BE LEFT THERE!!!
Commenting here is largely a waste of precious time and keystrokes, and about as effective as peeing in the ocean and expecting to measure a rise.
Please - comment where it will do most good.
Paul_K finally came to the actual post at JoNova and left this comment. Here is the reply, since he hasn't bothered to reproduce it here or answer there:
Paul, your comment is based on misinformation by commenters on BH, who introduced a straw man.
“If you are going to criticize the GCMs on the basis that there is an error of formulation or solution routine,…”. I didn’t.
How much plainer can I make it? In post 1 I explained why the conventional basic climate model is of significance, in post 3 I analyzed the basic model and showed all its equations in the full case (post 2 was the simplified case), and then post 4 is about misuse of the partial derivatives by the basic model. Post 4′s overview begins “The basic model relies heavily on partial derivatives.”
Note the absence of statements about GCMs, the Navier Stokes equation, and grids. The GCM details were injected into the conversation at BH by commenters at BH; they did not originate here. Perhaps you were being misled. Hence my exhortations at BH to read the posts :)
Later in this series I will point out that the GCMs do indeed share some of the same architectural shortcomings as the basic model, but as it happens, use of partial derivatives is not one of them. (I think I have also made the architectural shortcomings point about GCMs in comments already.) If you are interested, please stick around.
You found this post unpalatable because it was misconstrued at BH. It takes a fair bit of imagination or mischievousness to read my posts and think they mean that “there is a generic problem in numerical analysis that means we cannot solve non-linear systems where there are dependencies amongst the variables of interest.” It wasn’t you that made this leap initially at BH, but it would have helped if you’d checked by reading the posts.
To summarize what I am saying: the partial derivatives used by the basic model do not, technically, exist, and they are not empirically verifiable — so they are a poor basis for a model. We will use this clue in later posts to construct a better basic model.
Does that make sense?
Not really. The basic climate model you can write as
N(t) = C d(Delta T)/dt = Delta F(t) - lambda Delta T,
where N(t) is the system heat uptake rate, C is the heat capacity, Delta T is the change in temperature, Delta F(t) is the change in forcing, and lambda is the feedback response. It seems to me that you are referring to this kind of model.
Well, the term lambda certainly has units of W/m^2/K, but that doesn't mean that these simple models are using partial differential equations. Typically, either lambda is assumed to be constant so that one can estimate how much warming will take place for that feedback response, or lambda is determined by trying to fit to observations (surface temperatures). I think there are some examples where people have inserted non-linearities into these basic models, but normally the feedback response is assumed to depend linearly on temperature.
So, it seems as though you're savaging a strawman. If you can construct a better basic model, please do so, but that doesn't mean that your criticism of the current basic models is justified.
Oops, the link to Paul-K's comment did not come out in the previous comment -- his comment is here.
For any onlookers, ATTP is a PR guy who still has not read the posts and is hoping you won't bother, and there is no hope of an honest conversation. If you are curious, the series starts here and the project home is here.
David,
Firstly, I'm not a PR guy. Secondly, I have. Thirdly, that you would both describe me incorrectly and infer a motive to me that you cannot know to be true, might suggest that you're the one who is reluctant to engage honestly. Where are the partial differential equations in your basic climate model? That you can write the feedback response as
f = [G(T + h, ....) - G(T)]/h
does not mean that the basic model is explicitly solving partial differential equations. This is not complicated.
Mr Evans: I can understand your frustrations, but it probably would be a good idea, both for the argument and for your health, to ignore most of what is said on here, and only engage with those who can be bothered to go to your site and actually read what you are saying, and discuss it directly with you (as, no doubt, the likes of Manic BC and Rereke are doing – it's I while since I've had a look). Leave this site for those who want to pick over the argument in general, or for those who wish to pontificate at great length about something which they have no knowledge of, in the odd belief that it makes them look good.
When someone has put as much time and effort into an enterprise as Mr Evans has quite clearly done,then it would only be polite to read his work and withhold comment until such time as conclusions are reached.
It would further be even more polite to comment on his site ,instead of sniggering like school children behind his back here.
Entirely understandable that BH should point out the work and site ,but lets play in Mr Evans' garden hay!
very good initiative
navier stokes is very good to design that next revolutionary venturi pipe.
mayybe, very maybe it is helpful to design a new windmill or turbine blade..
NOT for predicting climate.
all REAL modellers of the past twist and turn in their graves at so much imbecillic number crunching
Too many people are trying to force the issue here. This is the Bishop Hill blog and people are entitled to say what they like within reason and certainly commenting on the many and varied topics raised by his grace is lauded and not frowned upon. Andrew made it perfectly clear that the action started on Jo Nova and therefore we were all aware that we could read more over on that blog. However his grace ended his opening post by saying the following:
"This sounds plausible to me. What do readers here think?"
Following that request it seems perfectly reasonable that a BH regular might go to JN, read the posts there and come back to BH and make a comment however they equally well might simply respond to the original post here. Should there be anyone who objects to that process then I care not one jot ^.^
ATTP,
Do you believe that there is a net forcing arrived at by summing all the individual forcings? Where might I find the justification for that in plain language?
Do you believe the Heat Capacity C is a constant? What's the mass of the object you have in mind?
Does not your formula contain many components which are in reality inter-dependent and/or unquantifiable?
(And no, I am not taking this to Jonova to expose my ignorance..)
rhoda,
Do you believe that there is a net forcing arrived at by summing all the individual forcings? Where might I find the justification for that in plain language?
Given that they're all defined as the change in TOA flux in the absence of any kind of restoring influence, I can't see why you wouldn't be able to simply add them. Energy is a scalar, so if one forcing produces a change in energy flux, you can simply add that to the change due to another forcing. Here's the GISS forcing site. It doesn't justify it, but it certainly does it.
Do you believe the Heat Capacity C is a constant? What's the mass of the object you have in mind?
The model that I presented was a very simple one-box model, where C is taken to be the heat capacity of the land/atmosphere and the upper ocean (normally about the upper 100m). In that context it is indeed constant.
If you want to include the deep ocean, you can add an extra box and couple the two systems through a simple diffusion type term. You can make it even more complex if you want to include latitudonal variations, and you can add extra boxes if you want more stratification. It was just intended as an illustration of a basic model.
It is just a simple model. The assumption is that if there is an external forcing, the system will gain energy, the temperature will rise, and the change in temperature will produce feebacks that further influence the temperatures. In this model, however, the feedbacks are typically assumed to be known, and to be constant. You can probe how it depends on the strength of the feedbacks and, by comparing to observations, may be able to estimate what the feedback response is. However, these basic models do not allow you to determine the feedback response in a more fundamental way. This is different to GCMs in which the feedback response emerges from the model/calculation, rather than being assumed to be known in advance.
I point at this:
> The GCM details were injected into the conversation at BH by commenters at BH; they did not originate here.
then at this:
> Later in this series I will point out that the GCMs do indeed share some of the same architectural shortcomings as the basic model [...]
In other words, Dr^n David Evans will not address PaulK's argument because it's a strawman, but will shortly talk about what PaulK said. To say something related to what PaulK said.
Gold. Always believe in your soul. You got the power to know.
It seems to me that David Evans had intended his world changing discovery (?) to be revealed on a slow release basis and is miffed that some BH readers got there first and spoiled it ^.^
Rhoda
",,,,,are there any testable intermediate results which we could use to determine the validity of the models...."
If it was me then I'd ensure that the model was spatially correct in steady state before using it for anything else. This at least gives you some confidence prior to doing a time-dependent analysis. There was a post on this blog earlier this year which showed that the Giss model was least spatially correct and the model that was most spatially correct had almost no positive feedback. There was a similar result s few years ago where of all the models used in the IPCC roundup only one could match reality (a Russian model) that had no positive feedback. Alas positive feedback is what turns mild, beneficial warming into thermageddon so it is absolutely necessary if you want to influence policy. Note that the IPCC range actually included a model with a 1.1K rise; ie equivalent to zero-feedback - so they are covered in any event.
However folk in the climate community are cunning in that they usually find one model somewhere that shows one thing correctly hence arguing with them about model validity is frustrating: You argue the models are bad at something, meaning most models, and they find the one model that does that thing ok but everything else badly; if you talk about another quantity then they find a different model; you say you can't just switch models like that and they say 'that's why we do an ensemble'. Sheer frustration! You may also remember the farrago where actual Phd mathematicians James Annan and Gavin Schmidt were arguing that a wider model range of errors could just clip the range of observations and hence were not as bad as we skeptics were saying - especially if they also adjusted the observations. This resulted in a paper by Santer which could not even get a good result without truncating the obs. The upshot of this line argumentation is that the worse the models are - the better they are. How such dreck gets published is beyond me! But worse is that some people are easily persuaded because they want to believe it.
By the way, the AR4 report showed a sample model spatial output of the Pacific versus reality and it was a very good match. What they didn't say in AR5 was that this model was from CSIRO (I recognised it), meant for cyclone prediction and was thus very heavily seeded with real world values in order to force it to simulate well. All the authors of AR5 wrote was that this was an example of what they were aiming at and it should not be misconstrued as actual model output. Well of course the alarmists didn't read that bit, didn't know the origin of the plot and hence took it exactly how the authors intended - that the models were just great at simulation.
ATTP,
Thanks. I'd say that formula is naive, but I understand that it represents acknowledged over-simplification in order to explain concepts rather than claiming to be the start point of a model.
I can't buy a two-box ocean. It seems to me that wherever you put the boundary between boxes the heat capacity of the disputed portion overwhelms the mere atmosphere. In mass at least thirty feet of water is the whole atmosphere. For a GCM to work you'd need to model the entire hydrosphere, bio components and all. It can be tried, but it can't be done.
Another word about checkable intermediate results. Do they exist? Are they used internally by modelling groups? With what result?
Sorry I accidentally wrote AR5 rather than AR4 a couple of times.
The reason some of us skeptics are a little upset by the rubbishing of pdes with dependencies by Dr Evans is because we are defending our own models which use similar techniques. But then we always validate our programs anyway (and not just by hindcasting which is wholly insufficient) - so we know these techniques work ok much of the time. No doubt climate simulation is a worthy effort and surely Dr Evans can contribute. My main complaint is that they may be useful tools for some thing but they are not even remotely adequate for policy. And the oft-used argument that they 'are all we have' is not true.
Rhoda: you talk about forcings, a term that does seem to have arisen around climate “science”, and one that I had never heard before I started reading about climate “science”. While you might even be as ignorant as I am (seeing your response to aTTP, above, don’t bet on that), and use the term as it seems to be one that others understand, could you not question what it is that forcings are supposed to be, erm, forcing? Surely, a “forcing” is merely a condition of the environment upon which certain effects depend; for example, is the glass in a window a “forcing”, preventing air from leaving or entering the room? Or is it a factor of the window, the presence of which influences the conditions within the room?
Mind you, the term, “forcings” does sound so utterly false, it is arguably quite fitting to be used in the whole climate “science” farrago.
One thing I do not understand is why there is the belief that such a vast, chaotic system that is the Earth’s climate can rendered down to a few trivial equations, such that we can claim to exercise some control over it. To put it simply – we do not understand the system; we do not know all the variables that could exist; we do not know all the variables that DO exist; and we do not know how each and every variable interacts with the other variables; in other words, we really do not understand what is going on. All we can realistically do is observe patterns from which we can make tentative short-term predictions (a.k.a. weather reports), or note the changing of the seasons. Each and every one of the many models (over 100, now, I believe) has been shown to be wrong, yet many still believe that they are right (and all, curiously, seem to lead to catastrophe. Odd, that). My understanding is that Mr Evans is attempting to produce a way of conclusively showing that the models are not (and, possibly, cannot be) correct, with an alternative offered. Whether he will succeed or not is a moot point, but he has the courage to post his research, so far, for feed-back from others to question, verify or challenge, in order for him to progress further.
rhoda,
The upper ocean is well-mixed. In other words, if you add energy in the upper parts, it mixes throughout that layer quickly (a few years). The regions below this are, however, not well-mixed (in other words, water from the upper ocean does not mix into the lower ocean very quickly). Hence energy transport from the upper ocean to the lower ocean is quite slow and can typically be represented as diffusive. That's not to say that there isn't any mixing, but simply that the upper ocean attains energy equilibrium with the atmosphere very quickly, while the deeper ocean takes a very long time.
That's kind of the point. The C in the equation I presented is the heat capacity of the atmosphere and upper ocean, since these two regions reach energy equilibrium quite quickly and can be - approximately - treated as a single system. You can introduce a small lag time which does improve the model slightly.
I don't really know what you mean by this.
Radical Rodent: I don't really buy the concept of forcing but I let that go for the sake of discussion. You can't get anywhere in understanding the other person's position if you dispute every premise.
Intermediate results: What does each model say about heat in the ocean, for example. Not the global average temp in twenty years but the heat in the ocean, or the rate of heat transfer into the ocean, this year or next. One of the shocking things put out by the climate folks in the plethora of explanations for the 'pause' was heat in the oceans. I said then, and repeat, that if you didn't see that in the models the models are no good. If it's an unmodelled phenomenon, then go away and do some more models. Now heat in the ocean is not the only thing which could be used to see how the models are performing. There's tropospherical temps, there's LWIR sat measurements, there's albedo, there's all sorts of things that MUST exist as variables in the model. Drag them out and let's see them. Then we'll throw away the models which got them wrong, won't we? No, we'll say 'I don't really know what you mean by this.'.
Harry Dale Huffman wrote:
"Everyone has had nearly 5 years to learn this (actually, nearly 25 years, since the relevant Venus data was obtained, from which ANYONE could have, and should have, done the Venus/Earth comparison I did--so I maintain there has not been a competent climate scientist in that length of time)."
Huffman's "comparison" ignored the albedo of each planet. It is rubbish.