Kininmonth in the Australian

The Australian has published a piece by William Kinimonth arguing that global warming is a natural phenomenon. His argument in his book was that the models used by the IPCC were one dimensional and didn't account for the flow of energy from the tropics to the poles. This is, of course, wildly incorrect as anyone can find out in minutes on the net. So he's dropped that argument, but that means that all he has left is this:

IPCC has made much of the apparent ability of computer models to simulate the climate system; computer models that have been tuned to reproduce the main statistical characteristics of the global climate notwithstanding the uncertainty of representing many of the climate processes. The computer models are claimed to be able to respond correctly as one or more of the boundary conditions are changed but this has not been demonstrated.

Colour me old-fashioned, but if someone wants to argue that the current warming is
natural, I'd like to see their model and data showing this. Complaints that the models currently in use are wrong in same vague unspecified way just don't do much for me.

Tags

More like this

How can mainstream papers keep publishing this rubbish? At Kinninmonths own book launch an invited real climatologist rubbished his claims totally as you pointed out in a previous post.

I challenge the Australian to publish this. Instead Kinninmonth is given status as a "William Kininmonth is a climate consultant who headed the National Climate Centre for more than 10 years. " No mention that this 'centre' has no standing in the international scientific community - to the general public he sounds like an authority.

His article contains:

"Are we in a long-term cooling trend or will the warming of the 20th century continue? Science cannot tell us"

I have offered him a bet at odds of 2:1 in his favour that it will in fact get warmer over the next 20 years. But of course I fully expect him to chicken out like all the rest (some more cases are documented on my blog).

The GISS SI2000 can hindcast accurately too, so he's full of sh*t.I've been following your bet, James. Keep at it!D

Well, don't forget the "What climate change?" army's frequent argument (on Usenet at least): "The IPCC and similar models cannot predict future climate, becuase they were just fitted to recent climate." Apparently, they prefer to use models which wer NOT fitted to recent climate to predict future climate. That explains quite a lot.

By z from Ian Tys… (not verified) on 06 Jun 2005 #permalink

I see we have yet another skeptic criticism of climate models based only on vague generalities about "uncertainties" and claims that the models are "tuned" to give the desired results--once again, with no specific demonstrations of any kind as to where this actually happened in a real climate model or what the impact was.

I never cease to be amazed by the fact that in over 10 years running, with only two exceptions, I cannot point to a single example of any global warming skeptic who attacked the credibility of climate models and even managed to name one correctly, much less demonstrate a specific problem with it. The exceptions were the two recent Douglass, Singer, and Michaels GRL papers on models and upper-air trends (Douglass et al., Geophys. Res. Lett., 2004; 2004b) in which the authors did at least manage to properly name 3 real climate models. But beyond that their "analysis" was so badly botched that the resulting criticisms were useless. For more info, see my critique of skeptic upper-air arguments.

Is it really too much to ask of someone that if they're going to challenge the viability of climate models they at least be able to name one correctly, much less demonstrate a specific, properly researched problem with it? Apparently so....

Is it really too much to ask of someone that if they're going to challenge the viability of climate models they at least be able to name one correctly, much less demonstrate a specific, properly researched problem with it?Sigh...Scott, Scott, Scott.
Doubt-sowing is hard work. Specifics and demonstrating mean that the author has said something...well...specific, which means their mendacity is easier to expose. It takes a lot of work to be so slippery. You should honor their hard work instead of picking nits. The left: always complaining and never offering solutions!Or something.D

Zed, it's a long run from Coronation to Longview.

It is easy to understand why Rupert's Rag would want to publish opinions that denied any need to address anthropogenic factors in climate change. But what is Kininmonth's motivation for his strange fixation on planetary warmings that occurred megayears ago when there is nothing in the records of the last millenium to account for warming except our own activities?

Many critics of global warming, evolution, whatever, fail to understand how science works. They are looking for proof that is a logically conclusive argument of the sort that occurs in logic or mathematics.

Of course they will never find it. No model can "prove" in this sense that global climate will be warmer in future than it is today, just as no model can "prove" that the sun will rise tomorrow. But you'd be a damn fool to assume because of that that it won't.

James,


I have offered him a bet at odds of 2:1 in his favour that it will in fact get warmer over the next 20 years.

I'll take your bet, but 2:1 odds don't reflect the message we're getting from the global warming advocates, their claim sounds more like 10:1. So lets split the difference: 5:1. Also, we need to quantify "warmer" so that each side doesn't end up paying based on a mere statistical fluctuation.

So how about this: I'll pay you AUD$10,000 in 2025 if the average global temperature from 2005 to 2025 is at least 2C greater than that from 1985 to 2005 (pick your own definition of "average global temperature" and we can discuss). You pay me $50,000 if the average is not greater by more than 1C. Nobody pays anything from 1C to 2C.

Sound fair? [you no doubt know a lot more than I about expected random fluctuations so feel free to propose alternative temparature ranges or time periods]

Anon, the usual usage of the word "warmer" is a higher temperature, not "at least 2 degrees C" higher. If you won't take a bet of 2:1 on it it being warmer, it seems that you must think that it likely will be warmer.

You want 5:1 odds AND a threshold of 2C in 20 years??? I don't believe that a single climate scientist anywhere in the world has ever suggested that such a rapid temperature rise is remotely possible, let alone odds-on likely.

Funny how it seems to be the sceptics who make the alarmist predictions when asked to put their money where their mouths are.

Tim, what do you mean by "warmer"? Any temperature rise at all? That would not be a bet on human-induced global warming but just a bet on the natural climate variability, which is a different proposition.

I might be happy to take such a bet after appropriate investigation.

James, I deliberately left the question of the temperature range and time period open:

Sound fair? [you no doubt know a lot more than I about expected random fluctuations so feel free to propose alternative temparature ranges or time periods]

Hardly "alarmist" to admit my own ignorance. Just tell me what increase you'd be willing to give me 5:1 odds on.

anon.... if you have no idea about climate, why are you so vocal? Prehaps you might get a good book (no Bill K's is not so good), and go away and learn a little. Climate is actually pretty complicated, which is why some of us get PhD's in the subject...

Your perfectly entitled to your views, but be honest and state they are based on next to no knowledge of the matter.

David

PS anon, 5:1 is rather stacked in your favour. If climate is completely unpredictable and global cooling is around the corner as some geologists claim, why not have even money on 0C change - a cooling I pay double, a warming you pay double. I'm willing to put a couple of $1000's on that.


PS anon, 5:1 is rather stacked in your favour. If climate is completely unpredictable and global cooling is around the corner as some geologists claim, why not have even money on 0C change - a cooling I pay double, a warming you pay double. I'm willing to put a couple of $1000's on that.

If climate is completely unpredictable and global cooling is just around the corner then the global warming advocates are wrong.

I took up James' bet, which is interesting in that it backs global warming. Your bet is not much more interesting than betting on the outcome of a coin flip.

If I know so little about climate change I should be easy to fleece.

Just tell me what increase you'd be willing to give me 5:1 odds on.

Probably something negative. I'm not really that interested in looking at the improbable edges of the distribution, but would much rather arrange bets at close to even odds. While there are people around (like Corbyn, Jaworowski, Lindzen, Bashkirtsev, Mashnich) saying that they expect no warming or even cooling, there should be no need to offer high odds to tempt them into a bet.

Unless, of course, they are lying.

Lindzen won't bet on less than 50:1 odds, and none of the rest will consider any sort of bet at all. You work it out.

James

Ok, what increase at your original 2:1 odds?

I too don't know why they are not interested in betting. If one thinks the current increase is due to random fluctuations (rather than human driven), then historical temperature data combined with a pretty straightforward application of stochastic differential equation theory (option pricing) will tell you whether your odds are good or not.

It may simply be that the historical data shows such high variance that your odds are not good, but to determine that I need to know what increase in the next 20 years you would give 2:1 odds on.

anon.. you don't get it do you. At the moment all the climate risk is worn by those who have something to loose if the climate experts are right, and the "sceptics" are wrong. The sceptics and their backers wear no risk (and your doing the same with wanting a +ve increase for odds stacked in your favour).

BTW, my expert opinion is that the most likely warming over the next 20 years is around 0.5C... but this is irrelevant, what we are dicussing is whether it will warm or not. Science says yes, and the sceptics say no. This is what the bet should be on.

David

David, what is it that I don't "get"?. James proposed a bet. I asked for clarification (and made my own proposed clarification). All I am asking for is what temperature increase James is willing to give me 2:1 odds on. If that increase is only 0 or more, then so be it.

But that seems more like straight gambling than a bet on global warming given the natural (un-human-forced) fluctuation in temperature over 20 year periods (I had a look at the time series back to 1880 last night - it moves a lot; James better hope there are no big volcanic eruptions in the next 20 years :).

But as I pointed out to Tim, it may still be good odds.

In the long term I would expect the market to choose fair odds on warming v cooling (and/or a mid-market value for the expected magnitude of warming). Until such a market is set up, it seems reasonable to try to find some middle ground between the various claims. I am no alarmist and have already bet on warming being below 0.25C per decade over the next 30 years (but unfortunately this is only an internet game with no real money involved). Anyone who will only bet on cooling if given 50:1 odds doesn't actually disagree with the consensus as I understand it. A big volcano would do the job fine, especially when added to a big of natural variability. The IPCC only says the warming is "likely" to be in the range of 0.1-0.2 per decade for the near future.

At the moment, Lindzen and myself have not reached a mutually acceptable wager. He wants 50:1 or more, I have offered 2:1 and said I might go to 3:1 (which as I explain on my web-site, represents equal value to someone who thinks that warming is absolutely certain, and someone who thinks it is 50-50).

I think the most interesting approach to these bets is an options pricing one. If you believe the fluctuations are random then it should be possible to apply options pricing theory in a straightforward manner.

Unfortunately, getting good figures for options pricing depends on having accurate estimates of a stock's volatility. For climate predictions, that means an accurate estimate of the temperature volatility, and over very long time scales. If the data only goes back to 1880 that may be hard to estimate.

The other problem is that the average is highly susceptible to rare volcanic events. So whether the next 20 years is warmer or cooler than the last may just depend on when and how large the eruptions are. Again, not so much a bet on human-forced global warming but on volcanic activity which is far more random.

I have some questions now that I have read a paper or two (this one in particular I found very interesting: http://pubs.giss.nasa.gov/docs/2005/2005_HansenNazarenkoR.pdf
- although it still emphasizes fits by what I consider to be dubious modeling methodology, it talks a lot more about net energy imbalance, which to me is a far more verifiable approach to the global warming question (it seems easier to measure whether there is a net energy imbalance than to predict climate change, and if there is, fairly uncontroversial that it has to show up in temperature increase at some point).

Q1: was the ~3C plunge in T in the 1880s due to Krakatoa?

If the answer to Q1 is yes, then:
Q2: why did it take until ~1940 for the temperature to reach the pre-Krakatoa temp? Did the eruption have that long-term effect?

whoops - looking at the wrong graph - that should be 0.6C plunge, not 3C.

A1 Yes, although it looks like 0.5C to me.

A2 Note from Fig 1a, that there was not just Krakatoa, but also several smaller eruptions including significant ones around 1904 and 1913. The temperature imbalance decays in a roughly exponential manner towards equilibrium, so the last 0.1C or so does takes a long time to fade away.

Q3: The model seems to show considerably less variance than the historical record. The runs in Figure 1B track each other fairly closely, whereas the observations generally oscillate to a much greater extent. There are also periods (eg around 1910) where all the runs are trending in the opposite direction to the observations. Is there an explanation for this?

[As an aside, I don't like the way these plots are done: by using a big thick black line for the model average and thin lines for the runs _and_ observations, it makes the agreement look better than it is. It would be better to plot, say observations in thick red and model average in thick green (or whatever colours), with the individual runs as thin lines. That would make it much easier to eyeball the model average/observation discrepancy.]

Tim:
I read this piece so forgive me for doubting you when I say this, but reaeding it with a clear mind about the issues of global climate change you have not convinced me this article is flawed. Unlike you I don't pretend to know or understand hard scientific evidenc that global warming is man created.
The issue has become so politicised primarily by the hard left that any decent middle of the road person whould scare cannot make a true assessment of the situation. So can you as computer science teacher rally know about a subject that is extremely complex unless you have studied this discipline. Jumping around from one leftist website to another, picking bits here and there does not convince me you have any knowledge in the area other than a simple amateur.

If you were really concerned and wan toknow more shouldn't you enrol in the hard scieces that cover this area?

Just wondering.

DC, your comment is bizarre. I don't get my information about climate science from "leftist web sites", but from scientific web sites. I provide links so you can check for yourself, but you don't seem to be interested in doing that.

DC: "The issue has become so politicised primarily by the hard left that any decent middle of the road person whould scare cannot make a true assessment of the situation."

No, the issue has been politicised by the hard Right - who accuse anyone not toeing the Bush/Howard line of being "a leftist".

Is John McCain a leftist? John Major? Helmut Kohl? George Bush Sr.? The entire Japanese LDP government? The Taiwanese government?

How about the 40+ US Senators who voted for the McCain=Leiberman Climate Stewardship bill?

George Pataki? Arnold Schwartzenegger?

By Ian Gould (not verified) on 09 Jun 2005 #permalink

Ian: of course the issue has been politicized by the left. Just as it has been (more recently) politicized by the right.

This blog is political: you only have to look at the vitriol and scorn dumped on anyone who dares question the "received wisdom" on climate change.

Anon:

That depends on how you define "received wisdom". If you define it as - "current research from peer-reviewed journals", then I tend to agree with you.

Ah yes, the old infallible peer review.

Peer review means review by a group of your peers: ie, people who likely think similarly to you, have similar educational backgrounds, who you may know personally from the conference circuit, who often have similar political views, etc.

Peer review is certainly better than no review at all, but it is certainly not an infallible or completely "independent" process, particularly when reputations are at stake.

Heard of the "Sokal hoax"? "Paradigms"?

How is a statistician criticising the use of statistics in a scientific paper "political"?

By Ian Gould (not verified) on 09 Jun 2005 #permalink

To add to Ian Gould's list, some of the world's biggest businesses have released a statement stating that "We agree that the science is sufficiently compelling to warrant action by both the private and public sector, and ... action must be initiated now." Are they leftist too?

How about this recent article by Michael Howard, the British Conservative party leader? Is he a leftist?

"Peer review means review by a group of your peers: .... who you may know personally from the conference circuit,"...

Peer-review is usually anonymous.

"..who often have similar political views, etc."

And who equally may have totally opposed political views and hate you with a passion.

Which is why, sevewral reviewers are usually asked to comment on a paper in total isolation withotu knowing who the other reviewers are.

By Ian Gould (not verified) on 10 Jun 2005 #permalink

Peer review is a process conducted by humans. It is subject to the same social factors that all human endeavours are subject to - groupthink, formation of communities, paradigm lock, etc.

It is certainly an awful lot better than no review at all, but my point is that one should not worship peer review as guranteeing some kind of absolute truth.

The softer the science the greater the danger. Climatology is pretty hard but not bulletproof by any means.

It is certainly an awful lot better than no review at all, but my point is that one should not worship peer review as guranteeing some kind of absolute truth.

Man, the klaxons went off pretty loud at that statement. Now that my ears have stopped ringing, I see your word worship is the one that got the horns blaring.

D

anon Says:
>June 10th, 2005 at 11:03 am
>Ian: of course the issue has been politicized
>by the left. Just as it has been (more >recently) politicized by the right.

Anon (and others) I am perplexed by suggestions that the science is possibly politicised by/IN the mainstream science community. Can you point to a single significant case where politics has swayed the conclusions of "mainstream" climate change scientists? Surely, there must be some examples up your collective sleves?

The anthropogenic greenhouse affect is an observable fact (see Harries et al. 2001. Increases in greenhouse forcing inferred from the outgoing longwave radiation spectra of the eather in 1970 and 1997, Nature, 410, 355-), behaving exactly as predicted from quantum physics.

Pressuming you believe in conservation of energy, you then have no choice in accepting that we are inputting an extra few W/M^2 of energy into the lower troposphere because of this human induced process.
Conventional science (which has no published alternative) tells us that this will warm the globe.. and geez guess what is happening... the globe is warming.

I am yet to meet a sceptic who has a robust and published quantitative explanation for the warming we have seen, or a mechanism to explain where the energy from the enhanced greenhouse affect is going. As for answering why the energy going into warming equals the input from the anthropogenic greenhouse affect... this hasn't registered on the sceptics radar, as even their contorted physical view of the world can't jump through this loop.

David

David,

I think we should ask the skeptics to put up or shut up. Since they claim the clmate models used by mainstream scientists are defective then by all means let's see them produce an alternive model which has an equally good back-casting capacity and which also spontaneously generates micro-phenomena such as hurricanes.

Science is about testing the predictive capacity of competing theories. The state of the art in climate science theories are global computer models of the climate.

Until the skeptics can come up with their own model, they don't even have a horse in the race.

I'll point out here that climate modelling probably isn't that expensive when compared to the amount of money the fossil fuel industry and the US and Australian governments have put into astro-turf and economic modelling to prove the cost of abatement is too high.

I'd be very sdurprised if offers to generate such models haven't been received by these bodies. The fact that no results of such studies have ever been published (so far as I know) suggests that it's been pretty damn hard to come up with a credible, internally consistent climate model which doesn;t show anthorpogenic warming.

By Ian Gould (not verified) on 13 Jun 2005 #permalink

Ian - "I think we should ask the skeptics to put up or shut up"

The trouble with this is this is exactly what they do - they shut up - at least for the current discussion when asked for specifics or data or research.

However when a new discussion starts up they are in there swinging again with the same old refuted arguments.

David,


Pressuming you believe in conservation of energy, you then have no choice in accepting that we are inputting an extra few W/M^2 of energy into the lower troposphere because of this human induced process.

The results are nowhere near as clear-cut. Eg, the latest from GISS

Table 1 shows the forcings from all sources. Total forcings come out to 1.80 +- 0.85 W/m^2, with a net imbalance of 0.85 W/m^2.

Not exactly an "extra few W/M^2".

And take a look at the results in Figure 1. In the text the model is described as a "good fit" to the observations, but I don't get that from Figure 1B at all. The observations are far more variable than the model and often trend in the opposite direction to the model simulation (eg around 1910). This discrepancy is disguised by the way the results are plotted - thick line for the model average and thin line for the observations. Makes the agreement look better than it is.

My question is: to what extent could I predict the historical temperature just as well with a simple polynomial function of the forcing variables (Figure 1A)? If a low-order model does it, then to what extent are the climate models simply statistical fits as opposed to modeling real physics?

I have downloaded the source for this model and I am going to run my own simulations. The code is surprisingly short (about 50,000 non-comment lines of code), but in fortran so I am currently converting to a more modern language to make it easier to understand. My day job keeps me from spending much time, but it should be interesting.

From my reading so far I think the whole global warming debate is not well-served by these kind of predictive climate models. The question of net-energy imbalance can be answered without needing to model the temperature. And if we can establish conclusively net energy influx, it would seem to be straightforward conservation of energy to conclude that the temperature must go up and/or the ice must melt at some point, regardless of what the models predict.

I believe someone has asked for a septic model, anon, not an example of dissembling from some Googler septic searching for a crumb somewhere, using big ol' words.

I'm a little blinded by the polysyllaby...wait...wait...OK.

Anon, do let us know when you have, Galileo-like, collapsed the worshipful pillars of climate science. Or we can hear the crash and attribute it to you, either way.

Audit away, M&M-like, brave anon!

D

Dano,

How is my post dissembling? I am referring to actual data. Do you dispute my interpretation of the models? Or my figures (taken from the climate papers themselves)? If so, why?

Sarcasm is the lowest form of wit, although I think you could lay claim to something even lower.

Dano,

Anon's offer strikes me as a reasonable first effort at what I asked for. Although while it would provide for macro-level back-castign and forecasting i don't see how it could be tested against other models for its ability to generate phenomena such as hurricanes.

Certainly it sounds like a better effort that msot of the so-called skeptics offer.

By Ian Gould (not verified) on 13 Jun 2005 #permalink

If anyone's interested in what large and well-informed players do when asked to put up big bets with their own money on the subject of global warming, try asking for a quote on long-dated buildings insurance in the Ruhr floodplain.

Hmmm. Last night my interpretation of Anon's post seemed reasonable. In the light of day, however, it no longer seems accurate nor reasonable.

Apologies to all.

D

Since no one denies that there is a random (chaotic) component in climate, why should differences in short term forcasts from the data (which itself has noise) be disqualifying?

By Eli Rabett (not verified) on 14 Jun 2005 #permalink

I am not saying that the differences between the model and data should be disqualifying - as you rightly point out, some disagreement is to be expected.

But in view of the limited dynamic range of the data (-0.5C to +0.5C from 1880 to 2003, Figure 1B), the deviations between the model and observation seem considerably greater (at least to my uneducated eye) than that which could reasonably be attributed to random variation.

Examples: from 1884 to 1886 the observations drop by 0.5C but the models don't track that nearly so well. For the 10 years around 1910 the models and observations are trending in the opposite direction . Same for just prior to 1920. Generally, the observation variance appears much greater than the model variance. Also, if you terminate the series at 1980, there seems to be no statistically significant warming trend at all (all the meaty action is post 1980).

All this variation may be attributable to observation errors, but if that is the case, fitting models to the data is made more questionable not less.

The models are very complex with many tunable parameters. The time series seems to have few significant (in the statistical sense) features: sharp declines due to volcanos and the apparent ramp from 1980 (although without a longer series to calibrate against even the latter is not clearly significant). If the rest of the features in the data are "noise", what does Figure 1B really tell us about the models? (my guess: not a whole lot).

Anon,

You repeatedly fail to distinguish between forced predictable temperature changes on decadal through centenial time scales, and naturally generated unpredictable variations on time scales of a couple of years. The argument you present is a straw man, and has the logic of denying the predictability of the seasons on the basis of our failure to accurately predict the weather at 7 days.

The IPCC (See figure 4 in the TAR) makes this issue very clear by using ensembles of historical climate runs to comapre against observations; that is they deliberately smooth out the impossible to predict year on year wiggles, to focus on the possible to predict decadal and centenial time scale changes. Global warming only emerges above the noise of weather on long times scales.
Any one who compares single model runs with a single set of observations is either pushing an agenda or simply has no idea; it is junk science.

You clearly have never worked with climate models, otherwise you would realise that their basic formulation is not tunable. It is based on conservation of mass, energy, momentum, and moisture. The governing equations are called the primitive equations because they are so basic...
Their is no option to tune any of these laws. Parameterisations are introduced as a by product to allow more detailed representation of small scale features; these again are based on observational physics, not some whimsical construction of the modellers mind. You can strip a model back to its bear bones largely or wholly stripping out parameterisations - it is called an energy balance model - and the result you get is pretty much the same as what you get when you add the works (see the IPCC TAR) - ie a general circulation model. The reason why we use 4 dimensional general circulation models (3 space and 1 time dimension) is because they allow us to describe the spatial and temporal behaviour of the atmosphere, and predict more complex variables such as rainfall, wind, etc.

One question you need to answer is if the models are so tunable, why do they all pretty much say the same thing? Sure, the warming at a doubling of CO2 varies between about 2 and 4C, but the up-shot is that it is large, and will take our climate beyond anything seen for many 1000s (if not 1000000s) of years.

As for your comment about radiative forcing, whether this is a couple or a few W/M^2 is irrelevant. We are adding to the greenhouse gas forcing at a rate of around 0.5 W/M^2 which means the target is rapidly changing. The current greenhouse gas forcing is just short of 3W/M^2.

David

>

Yes but given the uncertainties in measuring the physical variables that are plugged into those equations and the different valeus given in different sources in the scientific literature it is probablly possible to choose the values most likely to give the outcome you want.

I have no reason to assuem this is happening but my experience with econometric models suggests that that is how a biased modeller would shape the results.

By Ian Gould (not verified) on 14 Jun 2005 #permalink

Ian's description is what I mean when I talk about "tuning". The climateprediction.net experiments make the dependency of the model outcomes on the choice of the (uncertain) physical constants quite explicit (it is this dependency that they are investigating).

Unconscious tuning is very hard to avoid and even detect without independent validation of the models. This phenomenon has been observed in experiments as uncontroversial as measuring physical constants (eg electron charge). Each successive measurement fits within the error bars of the previous one, but very often the precise later values lie completely outside the range of uncertainty of the early experiments, which may have been performed decades earlier.

The most sensible explanation of this phenomenon is that no experiment is truly objective, and every experimentalist has sufficient uncertainty about their own methods that they are reluctant to contradict the last group's measurements. So where there is wiggle room in the results (and there is always wiggle room), the bias is towards making your new results closer to the previous results.

None of this invalidates the climate modeling, it just suggests with complex models and little data you need to be extremely careful.


Any one who compares single model runs with a single set of observations is either pushing an agenda or simply has no idea; it is junk science.

But I am not talking about individual runs. I am talking about an ensemble average of 5 runs, which was deemed sufficient by the climate researchers themselves. Go look at the GISS paper I referred to.

And I am not talking about small unpredictable deviations based on a couple of years of data - I am talking about extreme short-term deviations (1884 - 1886) or smaller sustained long-term deviations (around 1910). I understand perfectly well that small, short-term deviations are not likely to be significant.

Krakatoa exploded August 27 1883. Without looking at all at what you're on about, anon, I'm fairly confident that fact is likely to be no less relevant than the (to me) strange belief that two year anomalies in observational climate data ought to be reproduced in climate models. Sorry if I'm just misunderstanding you.

The observed temp shot back up to the pre-Krakatoa level in 1885, but none of the models come close to that.

Anyway, I think this focus on individual parts of the series is a little beside my original point, which is that well-fitting climate models should not be necessary to answer what I consider to be the fundamental question of global warming: is there a net energy imbalance and can we control it by controling CO2 concentrations?

Sensitivity of the global temperature to CO2 concentration may turn out to be a positive thing. After all, it provides us with a mechanism to smooth out the fluctuations in temperature (ice ages) that have hitherto been completely beyond our control.

Imagine: It's the year 4050 and an ice age is coming due to oscillations in the earth's orbit, or because a close shave with a comet knocked the earth off-track. So the IPCC (or its successor) recommends that instead of fusion we start burning coal for power (actually, we'd probably just burn a huge pile of it since there's no point building a power station just to burn the coal)

Well-fitting models give us confidence that we have captured most of the important processes and their predictions are likely to be reasonably good. No-one would prefer a model which simulated the last 150 years badly over one which did a good job!

Certainly, parameters can be adjusted, but for the most part, models are tuned to look reasonable at the present day, and then the ~150 yer simulation is itself a validation. We don't have enough computer power to run for 150 years, tweak a parameter, run for another 150 years etc etc. (I can't promise the latter never happens.)

A greater question about models is: do they let us make policy decisions?

If we demand perfection before we make a decision, nothing will get done. If we demand that GCMs do a really good job before we accept them, great. We should also demand that other models do a really good job too. Economic models, say. They don't achieve a great deal of accuracy, but they are used for policy decisions all the time.

If we use this accuracy logic, then we should demand that economic models achieve the same degree of accuracy that we demand of GCMs, else we don't use them.

D

This debate is really about "models vs principles".

I am not suggesting we shouldn't try to model climate. That's a decent a scientific endeavour as any. But it is not clear to me that a really accurate model is a necessary prerequisite for policy decisions (it is sufficient of course).

A simple analogy. Say we discovered a comet on route to hit the earth. We don't need to model the impact process to the tenth decimal place, tracking every consequence to know that it is going to make a huge mess. All we need to know is "how big" and maybe "how steep" so we can determine the size of the impact.

It is not clear to me that climate models are not to some extent modeling the exact details when all we care about is "how big and how steep".

BTW, economic models are often not intended to be predictive, but more an embodiment of our understanding of economic processes, to help policy makers understand the impact of the levers at their disposal. And they seem to work ok at that.

Of course an accurate model is not sufficient for policy decisions! That would still leave us arguing about what priorities and principles we should adopt. Unfortunately, all too often, people aren't prepared to be honest about their real priorities and principles, which is why rather than debating those issues, they have to battle over the science intead in order to bolster the claim that the science supports them.

"A simple analogy. Say we discovered a comet on route to hit the earth. We don't need to model the impact process to the tenth decimal place,"

I would not mind modelling it so the minimum kick necessary to deflect it past the earth could be calculated. Also with celestial mechanics there is always a small degree of doubt due to numerical solutions of n body problems - a model would be very useful.

But it is not clear to me that a really accurate model is a necessary prerequisite for policy decisions (it is sufficient of course).

Fair enough. Why the energy expenditure pooh-poohing GCM performance, then?

D

Anon is floating a red herring out there. Prediction of the effect of raising greenhouse gas concentrations has been about the same on all scales for about 100 years, from the simplest Arhennius model to the 1D models to the current 3D models. We have good and sufficient answers to how big and how steep. A lot of the debate now is about where and what would be the effect of taking actions. (Tip o the hat to Ender)

By Eli Rabett (not verified) on 16 Jun 2005 #permalink

Red herring may be the term, Eli, but I'm not fully convinced that's what anon is doing.

I don't yet have a catchy term for anon's type of argumentation here - you know the one: look at a little detail, extrapolate it up to the entire system, then declare the entire system problematic. My phrase is: ants finding a crumb and declaring it a picnic, but I can't get it down to a shorter pithy phrase.

Best,

D


Prediction of the effect of raising greenhouse gas concentrations has been about the same on all scales for about 100 years, from the simplest Arhennius model to the 1D models to the current 3D models.

Really? The discussion in the literature is not nearly so confident. The error bars are large (sometimes up to a factor of 2). Eg: GISS


Why the energy expenditure pooh-poohing GCM performance, then?

"Pooh-poohing" is a little pejorative. I don't think the performance is as good as is claimed; I think there is a potential overfitting problem; and currently policy decisions do appear to be based on the performance of GCMs.

Good enough for you Dano?


I don't yet have a catchy term for anon's type of argumentation here - you know the one: look at a little detail, extrapolate it up to the entire system, then declare the entire system problematic. My phrase is: ants finding a crumb and declaring it a picnic, but I can't get it down to a shorter pithy phrase.

Dano, how am I looking at one little detail? My posts address very general issues. You may not be able to find the right language to describe my arguments, but it is easy to describe yours, although quite unprintable.

Go back and read the historical literature on climate sensitivity, about 1-5 C since Arrhenius on 2x CO2. It jiggles a bit but not much and the range does not change. I have read the literature.

The greenhouse effect accounts for about 20 C of warming on the surface of the earth and what you are referring to is the uncertainty in the effect given a perturbation.

By Eli Rabett (not verified) on 16 Jun 2005 #permalink

Eli,


Prediction of the effect of raising greenhouse gas concentrations has been about the same on all scales for about 100 years.

and


Go back and read the historical literature on climate sensitivity, about 1-5 C since Arrhenius on 2x CO2.

are not mutually consistent statements where I come from.

Economic models do a terrible job at prediction. And we make policy decisions from their output every day.

I think, anon, you should expend your energy taking apart economic models instead, as we have been making decisions with them for far longer than with GCMs.

Jus' sayin' that's the biggest bang for your buck, if'n you're really interested in how we make policy.

HTH,

D

Anon, are you still around and interested in betting? It's not clear to me if you were actually interested in betting or just wanted Annan to state his odds without participating in a bet.

If you do want to bet, let us know so we can talk about 2:1 odds in favor of global warming skeptics, and 2:1 odds in favor of the mainstream position.