Well, you all knew that anyway I suppose, but what I didn't realise until just taking a look recently was quite how appallingly bad it is. At http://www.junkscience.com/MSU_Temps/Model_Request.htm they ask "if anyone has managed to recreate say, Earth's global mean temperature track for the period 1880-2000 (we'd accept 1880-1979 or some reasonable facsimile) as GCM output". Errrm, well, yes, you can try reading the TAR. Figure 12.7c: http://www.grida.no/climate/ipcc_tar/wg1/450.htm#fig127 has a nice picture (it even goes back to 1860), for example, and refers you to the appropriate papers. Note that the JS quote fails to say exactly what measure they mean, but later down they say "annual mean temperature track" so presumably they do mean annual mean. But if so, why do they then say "any track that manages to stay within ±1.5 °C..." when, as the TAR pic makes clear, thats far too generous an error margin.
So, I admit, I'm baffled. Have the septics really lied to themselves so often that they have come to believe it? I suppose so.
[Update: if you're mad enough to look at http://www.junkscience.com/MSU_Temps/Warming_Glance.htm you'll find the UAH data plotted, with the excuse that they don't plot RSS because "RSS do not publish Lower Troposphere MSU data". But this is false. The data is available from Rss's ftp site and you'll find it ref'd at a page rather more respectable than JunkSci: [[Image:Satellite_Temperatures.png]]. I've mailed them this, we'll see if they upgrade.]
14 comments:
I wonder, William, how it's possible that whenever you write something like that, it never makes any sense, while the well-known "septics" are usually able to formulate their ideas quite clearly.
The only comprehensible point of your text is that you apparently don't like JunkScience.COM. Could you please try to write the same thing once again, from the scratch?
Funny, it makes sense to me, Lubos.
It makes perfect sence to me. Milloy may be using the personal ignorance ploy, I haven't heard of it, so it must not exist. You have to wonder how much research Milloy has done.
Most people do not follow up by fact checking and few, even the intelligent, have the background information to evaluate wild claims. For example Lumo.
There is little cost to such a policy for junkscience
Hi Lubos, yes I can explain: you have a filter in your mind which rejects certain information you don't like. Even in a case as simple as this. Let me do it in really easy steps:
1. JS asks for people to find GCM results that reproduce the 20C temperature history, and purported to believe that such things are hard to find.
2. Those results have been available for years, in the IPCC TAR, 2001.
3. The TAR is very well known. The figures are not hidden (the one I gave is 12.7, but essentially the same figure is in the summary for policymakers, which is probably the most widely-read document on climate change in the world).
4. Either JS is hopelessly ignorant, or lying, or both.
An additional point is that JS purports to have recieved no answers to its question. This means that either (a) JS readers are also hopelessly ignorant, or that (b) those who aren't (like me) don't consider it worth pointing out the bleedin' obvious to them.
"telling us precisely which model you used, which initialisation files and what forcing values you supplied (so we can indulge that repeatability thing - apparently a bad habit of ours). "
It seems to me that this is deliberately designed to ensure they get no answers. Reasonably knowledgeable people won't want to go rooting for information such as initialisation files. This probably leaves the authors and few other other people who could provide all the information. Any author who did provide the information would be asking to be pestered with silly questions as to how to get the model running. (This isn't easy judging by the delays at climateprediction.)
Is it a bit like expecting Mann to do a huge amount of teaching to help M&M so they can complain about minor issues when he has been careful to avoid drawing too strong a set of conclusions?
crandles
CR - well, sort or. It looks like idiot-cunning to me. Because of course all of that information *is* available via the papers referenced in the TAR.
The silly bit is the "repeatability thing" because of course there's not the slightest hope of those bozos at JS actually patching all that info together, building a GCM, and running it. Trying to pretend that repeatability is a "habit" of theirs is also laughable.
Coming soon: my post on mental filters...
Well, Lumo I have to be boring and sing with the rest of the comments... it probably has something to do with not wonting to understand...
Sorry to be candid, but as far as this debate goes, all of you are dishonest and intellectually limited political activists.
Steve Milloy defines the main problem of GCMs absolutely exactly. It's not any kind of established science because it simply has not reproduced the known temperature record in a way that would confirm any nontrivial prediction (beyond an obvious interpolation of data that is possible even without a correct theory).
Milloy's statement is obviously true to anyone who knows at least a little bit about the climate. Milloy also defines very clear, fair, and completely comprehensible rules of the game for anyone who wants to question his (and my) assertion.
If you think that there exists a GCM that reproduces the temperature record, he would love to hear from you. Just send him the model plus the initialization files and input data.
None of you has anything that has even a remote chance to satisfy this rudimentary requirement. You are just a gang of political activists who refer to other political activists (and maybe even believe them) just like in the fairy-tales about the Emperor's new cloths.
Even with the models based on this belief - models that you actually don't have on your hard disk and most likely you can't even have - the agreement is very poor (see the third graph). It's just much less poor than an interpolation of a graph with 3 generic parameters (while the actual ones use many more to make it work). If someone thinks that this vague inaccurate similarity can serve as an argument in favor of that model's predictions, then he has unrealistic ideas about probability theory.
Garbage in, garbage out.
Lubos, you're ranting. But its entertaining so I've left it up.
But you write too many words. Lets simplify: of my steps 1-5 in the wors-of-few-syllables version, which is the *first* step that you consider to be false?
I read the "fine print":
any track that manages to stay within ±1.5 °C (i.e. 2 to 3 times the estimated change since 1880 or roughly twice the error margin for the estimated global mean 1961-1990) of a recognised historical global mean temperature reconstruction (GHCN, NCDC...) will be deemed "accurate" provided it is reasonably consistent (i.e. not using the entire 3 °C range starting low and ending high but a similar track perhaps a degree or two high or low).
Note that approximating a 30-year average at some point in the 20th Century is meaningless unless you can demonstrate a reasonable facsimile of the whole century's annual mean temperature track (we've managed to generate "correct" mean temperatures with truly bizarre temperature tracks while probing some models' coded sensitivity to various forcings).
Figure 12.7c manages to remain roughly within 0.5K of the temperature anomaly from 1860 to 2000.
I don't know if it is the greatest of models, seeming sometimes to go in a direction opposite to the observations, but perhaps that much of climate is irreducibly stochastic. In any case, it meets the conditions, and I too am not sure what Lubos Motl is upset about.
Anyway, I should thank him, 'cause his rants led me to this blog and to Imbrie & Imbrie.
"Steve Milloy defines the main problem of GCMs absolutely exactly. It's not any kind of established science because it simply has not reproduced the known temperature record in a way that would confirm any nontrivial prediction (beyond an obvious interpolation of data that is possible even without a correct theory)."
Irrespective of the success of GCMs at reproducing the temperature record they do qualify as science according to definitions produced from scientific philosophy. It qualifies because it integrates current scientific theory into a computational model. This isn't a process which is scientifically unsound per se.
The global warming hypothesis certainly has more going for it in terms of its philosophy when compared to other theories such as string theory - that was for the arrogant Lumo. This is because string theory can't be empirically proven yet where as global warming at least has a chance at achieving this and therefore being classed as science.
Seriously now, I hope Lumo has success in his field because it could answer important questions.
To disprove the hypothesis of global warming you will have to come up with some real empirical evidence to falsify it. That is the only legitimately scientific way to poke holes in the theory.
Rubbish. Mann's hockey stick model was shown to be garbage by pure mathematical reasoning. Not a scintilla of real-world experimental evidence was necessary. You could plug a numeric hash of the names in the telephone book, or white noise from quasars, into the model and get the same result.
And your tone is coming dangerously close to forgetting where the burden of proof lies. It lies with the alarmists who blame every tornado and frost on anthropogenic climate change. So far they have not convinced me of anything other than the dynamic nature of the normal atmosphere and climate.
You've either misread or misunderstood what I meant here, so I'll make it obvious.
When I said "to disprove the hypothesis of global warming" I was referring to the hypothesis of anthropogenic global warming not the MBH Hockey Stick paper.
"And your tone is coming dangerousl....."
Err, the current empirical evidence does support AGW in case you hadn't kept up to date with the current literature. You don't seem to be familiar with the scientific method: observe, theorise and hypothesise, observe, observe more and if there continues to be a lack of evidence to the contrary put more trust in the theory, but beware of new observations. Which was my point, the scientific method directs us to find evidence to disprove the theory, the world of science isn't like a court of law. The owness is on scientists to find evidence that breaks the current theory.
And finally, why, if the MBH methodology has been discredited, do other groups manage to reproduce similar results on different datasets with different methodologies?
Post a Comment