Via Jennifer Marohasy, I found a report on three articles in Science Express that put the closing seal on the most significant issue in the debate about the reality of human-caused climate change: the disagreement between climate models and data from satellites and radiosonde balloons. Now as Real Climate observes “the discrepancy has been mostly resolved – in favour of the models.”
There aren’t many scientifically literate sceptics (that is, people open to being persuaded by evidence, but not yet convinced) left on the global warming issue, and this evidence, along with the continued warming being observed at all levels, should convince most of those who remain. There’s a bit more history over the fold.
The big expert on satellite measurements of climate is John R. Christy of the University of Alabama, who has worked with Roy Spencer. His data, which started in 1979, initially appeared to show a cooling trend in the troposphere. Since satellite data seemed free of many of the errors that affect surface measurements, these results were seized on by global warming ‘sceptics�.

However, it was discovered in 1998 that the satellite data had problems of their own, arising from a failure to correct for the gradual decay in their orbits. Correcting for this, and adding more data, the satellite data now showed a slight warming trend, but not as much as the surface data. More data and alternative analysed reduced the discrepancy somewhat, but it remained a problematic issue. Sceptics also pointed to studies of radiosondes carried on weather balloons. These were much patchier, but also failed to match model predictions.
The latest studies correct errors in the procedures used to derive both satellite and radiosonde time trends, largely eliminating the discrepancy. The most conservative estimates of tropospheric warming, those of Christy and Spencer now show warming of 0.12 C per decade, compared to a surface trend of 0.19 C per decade. Other estimates show tropospheric warming equal to or greater than that at the surface, as predicted by the models.
In view of the predictive success of the models, in which climate change is driven by human-caused accumulation of greenhouse gases, and the repeated failure of empirical challenges, it’s hard to see how anyone without strong preconceptions on the topic can reject the conclusion that, in all probability, we are experiencing human-caused climate change. Most of the models predict that the warming is likely to accelerate in the future.
John Quiggin 
We seem to have strayed from the topic of weather balloon readings but some big issues have emerged. I agree that there may be a psychological barrier to alternative fuels based on coal, shale or crops such as trees. When an extraordinary amount of visible effort is needed to keep each car on the road people will feel uncomfortable. On the other hand petroleum is like drawing blood; you get an energy rich liquid through pinpricks on the Earth’s surface, not huge quarries or hectares of forest. However if you add the requirement of greenhouse neutrality then biofuels win over oil shale and coal . Since even that is distasteful to some the final option seems to be using less fuel period.
Ender,
We’ve also had this argument before – at Club Troppo, I believe. I think you also made the same points last time:
Insurance companies think global warming is a problem – yes, but their motives aren’t exactly pure: they get to charge more for bearing risk if the perception of risk increases.
Decision-making under stress – I’m bored of this one: “we have a rapidly closing window to reduce the effects of AGW. After this time has passed it really does not matter a pair of fetid dingo’s kidneys what we do as the die will be cast and we will have to live with the effects whatever they are.” The problems here are that:
1) We don’t know whether observed GW is anthropogenic or not;
2) We don’t know what the “window” for decsion-making is; and
3) We don’t really know what the effects of AGW will be, if it occurs.
Faced with that magnitude of uncertainty, I propose the logical course of action is more research until we can answer these questions.
I’m still stuck on Fyodor’s initial assertion: “That the models fit historical data is trivial and expectedâ€?. The subsequent discussion has not clarified what he actually meant by that. People have talked about “data-fitting”, “prediction” and “model limitations” in very non-specific terms. And references to “skeptical-cum-solipsistic philosophy” and “wishy-washy brand of do-nothing “libertarianâ€? policy preferences” don’t advance the discussion in this particular area. (Geat fun, though. Keep it up.)
Now, I’ve been aware of global climate modelling for some time (I work in smaller-scale ocean modelling myself). I know something about the structure of the climate models and the processes they represent. Over the last decade or two I have heard conference talks about simulating “present” climate (though that’s a moving target), pre-industrial climate, palaeoclimate (eg Last Glacial Maximum) and climate sensitivity. During this time there has clearly been a process of iterative development (simulate, validate, improve) that you might, at a stretch, call data fitting. (But Fyodor this has not been trivial. That was the word that first attracted my attention. I would call it heroic.)
In recent times there have been several exercises in which a model, resulting from the above iterative development, has been run for the 20th century, more or less, subject to time series of various “forcings”. These time series have, as far as I know, been derived *a priori*. They’re subject ot uncertainty, of course, but they’re best estimates from available knowledge. The model has been run several times on the same forcings with randomly different initial conditions, generating an ensemble of different “predictions”. (Yes, I know they’re not actual predictions of the future.)
Two things have surprised me about the results of these exercises: 1) for a an integral quantity like global mean surface temperature, the scatter between the members of the enseble has been rather small; 2) the mean of the ensemble has tracked the observed global mean surface temperature very well. But don’t take my word for it, follow the links in this RealClimate post:
http://www.realclimate.org/index.php?p=100
The people who do this stuff are aware that their forcing time series are uncertain and try various combinations and perturbations. They conclude that some of the temporal variations are due to the natural forcings (volcanoes, solar) and some due to the anthropogenic ones (tropospheric aerosols, greenhouse gases) but you need both to get a good match.
Now I can see that dishonest modellers could pervert this process, by choosing the forcings to get the results they want. But they appear to be open about what they have done and I am not aware of anyone having proposed forcing time series that are substantially different *and* plausible. (Eg you could propose that solar forcing has risen in the second half of the 20th century, but as I understand it there is no a priori reason to believe this has happened and recent measurements strongly indicate that it hasn’t happened.)
So Fyodor, the models (iteratively develpped or fitted as you will) have been driven by a priori forcing time series and have produced surprisingly good hindcasts of the global mean temperature over the last century or so. What problem do you have with that?
Chris,
Your first paragraph is confusing, but your I agree with your concluding point, i.e. it looks like we have resolved the surface-satellite discrepancy.
Re: your second paragraph, no, I don’t have the data to hand (and can’t be bothered to look for it) and, given we do not have accurate temperature readings earlier than the 20th century (if that), the actual magnitude of temperature changes can only be known approximately. We do know, however, the direction of temperature changes, which varied, and we do know the changes were significant enough to be noticed by historical sources, and to be evident in tree rings, ice core samples etc. It’s also worth pointing out that in just the past fifity years we have seen increases and decreases in temperature. The Earth’s surface temperature clearly changes without human interference, and we don’t know why. Despite this, a lot of people (let’s call it a “consensus”) are certain that recent warming is attributable to human action.
Re: your third paragraph, reread JQ’s initial comments. The title for this thread is “Climate change modellers vindicated”, not “Surface data vindicated”, so I’m not constructing a strawman here.
Re: your concluding paragraph, your assertion that the Earth is hotter now than it’s been in the past 1,000 years is contentious, to say the least. I agree with your sentiment, though: the data on recent warming looks solid, so we really should spend more time asking why the Earth has warmed. Unless you’ve got all the answers already, of course, in which case I’m all ears.
Plagiarised Nabsy on shorter Strocchi: Fyodor is really pissing me off!
I hope it didn’t take you a whole day to come up with that exceedingly concise and pithy contribution to the subject of this thread.
Mark Hadfield,
By the statement, “That the models fit historical data is trivial and expected.� I meant that it is not surprising that climate models developed to simulate climate change tend to match historical data, because their output is usually tested for a posteriori predictive power against historical trends.
The risk in the model building and testing process is that the model builder works through so many iterations to “fit” the model to the historical data, that they lose the rigour of the science used in determining the key variables and the structure of the model, thereby producing a model that works really well at predicting the past, but may be lousy at predicting the future.
I hope that clarified my position for you. Can you now clarify something for me? Can you elaborate on the following statement, as I don’t understand what you mean.
Hi again, Fyodor
How do you do those verical-line quote thingies? I am new to blog discussions.
Quoting your paragraph:
***
By the statement, “That the models fit historical data is trivial and expected.� I meant that it is not surprising that climate models developed to simulate climate change tend to match historical data, because their output is usually tested for a posteriori predictive power against historical trends.
***
Aha! Now we’re getting somewhere. It is my understanding that the models have not been tuned to fit historical trends. They have been tuned (iteratively developed) to represent various aspects of equilibrium climate accurately *then* run with historical trends in forcing *then* the trends they produce in global mean surface temperature have been validated against the instrumential record. But I have a lot of reading to do about this.
Now my paragraph that you don’t understand
***
In recent times there have been several exercises in which a model, resulting from the above iterative development, has been run for the 20th century, more or less, subject to time series of various “forcings�. These time series have, as far as I know, been derived a priori.
***
I’m talking about the GISS hindcast for 1880-2000 described in these links
http://www.realclimate.org/index.php?p=100
http://www.realclimate.org/index.php?p=148
A few years ago I read about something similar by the UK Hadley Centre people.
Howdy to you too, Mark,
There’s bound to be someone around who can explain this more elegantly, but the quote thingies are done by opening an arrow quote, typing “blockquote”, closing with an arrow quote, pasting in the text you want to quote, opening another arrow-quote, forward slash, “blockquote”, then closing arrow quote.
Christ this is a clumsy explanation, but it should look like a pointier version of this:
(blockquote)text(/blockquote)
In case you’re wondering, I don’t know how to type in the arrow-quote without actually launching its function. Hope that helps.
Thanks for the links – I’ll have a read.
Mark,
I’ve read the links you’ve provided, and I think you’ve confused the meaning of a priori in your original statement. Through many iterations the models the links refer to have been used to, as you say, hindcast (a posteriori) the historical experience, but the track record on actual (i.e. truly a priori) prediction remains extremely limited. If we take JQ’s “start” point of 1997, models have only been forecasting (as opposed to hindcasting) for around eight years. As I said previously, this is not a long time frame for testing the predictive validity of these models.
carry on…
wordpress sucks.
How about some hotlinks to the basic functions, near the comments box?
Now that’s the clearest thing you’ve said.
Fyodor
Yeah, you’re probably right about “a priori”. Strictly speaking it means “beforehand”. My point was that the forcing time series were *not* fudged to get the results the modellers wanted. At least I don’t think they were…
I think you’re setting the bar of “predictive validity” too high. Natural scientists don’t usually have the luxury of running well-controlled experiments. Mankind’s current greenhouse-gas experiment is an interesting one, but even it is imperfect. Luckily, mankind seems intent on applying a really big greenhouse forcing so the signal will probably be visible above the confounding factors.
Let’s say we had a perfect model. Our forecasts of (say) global mean surface temperature over the next 30 years would still be limited by: uncertainty in projecting greenhouse concentrations; uncertainty about the other forcings; intrinsic variability (but the models suggest that it’s not all that largeif you average over the whole earth for a decade or so).
Let’s say in 30 years time we do have very good climate models and very good time series for the forcing factors. Let’s say we run a hindcast and find it matches the observations very well. You will still be able to say that it’s only a hindcast and agreement is only to be expected (but please don’t call it “trivial”).
I don’t think I can contribute much more to this discussion without doing more background reading…
Cheers.
Fyodor:
“The problems here are that:
1) We don’t know whether observed GW is anthropogenic or not;”
Well yes we do however you are in the minority of people that still think that it is not. In the opinion of most scientists in the world Global Warming is anthropogenic.
“2) We don’t know what the “windowâ€? for decsion-making is; and”
Again yes we do. If the rate on CO2 was drastically reduced now then the warming could be held to less than 2 degrees. This has been confirmed in a range of scenerio modelling. If you want proof just wait. After this time the CO2 already in the atmosphere will raise the temperature from 5 to 10 degrees and no decrease in CO2 will stop it. The research is quite clear and we have enough information.
‘3) We don’t really know what the effects of AGW will be, if it occurs.”
No we don’t – however I have never seen you write that you might be wrong. I have. You do not know that the effects of GW whether it is antropogenic or not will be.
“Faced with that magnitude of uncertainty, I propose the logical course of action is more research until we can answer these questions.”
We will only answer these questions when the effects are irreversible. We have enough information to act now. I never said the insurance industry has pure motives – it is affecting their bottom line. They make money from calculating and assessing risk. If you cannot listen to most of the world’s scientists telling you there is a problem then at least listen to the industry that is the best at assessing risk. They are telling us that there is a non-zero risk that the climate will change.
Ender,
1) Opinion is not science.
2) Reread my previous comments about the limitations of scenario modelling.
3) You’ve never seen me write that I might be wrong? I have, but for your benefit let me state it again, categorically: I might be wrong.
Conjecture.
Again, opinion is not knowledge. How can you be certain that you have enough information when you don’t know how the Earth’s weather works?
I’m always listening, and I do follow the progress of the scientists. Do I believe certain scientists’ forecasts for global temperature in 2055? No, I don’t, for many obvious reasons. Do I question the motives of the insurance industry in modelling catastrophe risk? Absolutely. Do I think their views are as tainted as the astroturf outfits JQ regularly chastises? Absolutely. WTF does “…non-zero risk that the climate will change.” mean? History alone tells us that climate will change. That’s not in dispute.
Some climate sceptics are prepared to put their money where their mouths are.
http://www.guardian.co.uk/uk_news/story/0,3604,1552092,00.html
I’m sorry if Fyodor is confused by my explanation of how he was misleading but that’s often the way it is when someone is misleading i.e. showing how someone is misleading is often not a trivial process.
So Fyodor can’t be bothered to look for data to support his bare-faced assertion that “The Earth has warmed and cooled by ‘many’ multiples of this amount several times over just the last milleniumâ€?. It’s all very well waffling on about how past temperatures have produced a record but until you can come up with a backed-up number you’ve made nothing more than an assertion. It’s just my opinion but I don’t think many people are interested in bare-faced assertions by nobodies like Fyodor.
When I said “the original issue of this thread”, the intended meaning was the issue that motivated this thread (the resolution of the temperature record discrepancy). I don’t see any problem with pointing out a profound consequence of this, even if it’s not the same consequence that Pr Q wrote about.
I didn’t actually assert that the Earth is hotter now than it’s been in the past 1,000 years. I implied that I thought it was but more importantly I had previously written that a deduction from the surface instrument and satellite records since 1980 and McKitrick’s reconstruction pre-1980 showed this to be true. So what’s Fyodor’s problem with saying the earth’s surface is the warmest it’s been in a thousand years? Does he wan’t to just say we’re 99% certain the earth’s surface is the warmest in a thousand years? i.e. does he have a problem in this context with the way language is often used. I guess usually when people are 99% certain of something they don’t usually bother saying “99% certain”. Or does he think all the climate reconstructions are complete rubbish?
James Annan has made the easiest $10,000 in the history of scientific work. I wonder if those guys are looking for further people to take their money off them.
That Lindzen wants odds of 50-1 shows just how much confidence he has in his own claims. Effectively he is saying “there is only a 2% chance that I am right, but the world should risk global catastrophe on that basis”. Actually this is not quite fair, if there was some slight warming Lindzen could still reasonably claim to be right, but would lose his money. However, this could be easily settled with a bet of warming of more than a certain amount. I think Lindzen wouldn’t take that either without favourable odds – in other words he knows he is lying, and putting the world in danger in the process.
Getting back to Ffyodor, let me see if I have summarised the position correctly:
Thousands of the world’s finest minds have spent decades trying to model the global climate. They have succeeded in very accurately modelling what has been happening through the period 1900-1997, even though when they started there were divergent results for some of this period. That is when they started the landbased and satellite readings were very different. They created the models, and these matched the land based readings closely, but no matter how hard anyone tried they could not be made to match the satellite readigs. The satellite analysis has since been shown to be wrong, vindicating the models.
Warming since 1997 has been very much in line with what the models predicted, vindicating the models again.
The models indicate that if we keep pumping CO2 into the atmosphere at current rates within about 25 years we will have a global disaster that will dwarf the second world war, and a few decades after that the survival of humanity starts to come into question.
But its still not time to do anything. After all, of the time to predicted disaster available in 1997 we’ve only used a quarter testing the models. Maybe we should use half before we do anything, leaving us with only half the time to actually fix things.
Fyodor, if you actually lived your life this way you would be dead long ago, having failed to get out of the way of a car because you couldn’t be sure it was actually there.
Fyodor Says: August 19th, 2005 at 8:18 am
I am not that angry at Fyodor. I take some time out to take Fyodor’s comments to pieces as a public service and private pleasure. I am sorry that he wastes other peoples time. I notice that Pr Q has given up on trying to argue with him, so someone has to do it.
His intellectual tactics are borderline troll masquerading as conceited big mouth intellectual. His combination of arrogance and ignorance, an incoherent jargon with incendiary bluster, has a superficial plausiblity that just begs to be put down.
It is a pity to see others mesmerized into watching a fool dig himself in deeper and alarming to see him entice them into the bog. Although he himself is apparently incorrigible, hanging him out to dry will encourage the others. Plus it is fun, since it is always enjoyable to highlight a big mouth putting his foot in it.
Fyodor seems to have started this wild intellectual goose chase by confusing two aspects of mathematical modelling, roughly corresponding to the distinction between historical construction and theoretical application. :
How do we know that the measurement data is a representative set of possible values?
Does the model describe well the properties of the system between the measurement data (interpolation)?
Does the model describe well events outside the measurement data (extrapolation)?
A common approach is to split the measured data into two parts; training data and verification data. The training data is used to train the model, that is, to estimate the model parameters (see above). The verification data is used to evaluate model performance. Assuming that the training data and verification data are not the same, we can assume that if the model describes the verification data well, then the model describes the real system well.
Fyodor has spent the better part of this thread arguing that the data discrepancies between satellite and surface reports is improperly resolved because the rotten modellers have only fitted (meaning retrodictively interpolated) the data to their models. Now he says that Global Warming (GW) models are ok since the data discrepancies have been properly resolved in favour of the GW model which is in accord (meaning predictively extrapolates) with data provisions and revisions! So all is forgiven to those “reckless” and “naive” modellers.
This means that Fyodor implicitly concedes that models show that GW is occurring. I take it that he has back-flipped from his initial super-skepticism about modelling, in general, and the Anthropogenic GW model, in particular? One will wait in vain for a self-correction on this point.
Instead of treating this discrepancy-resolution, as any fair and reasonable person would, as a predictive triumph for the AGW model he now withdraws to a falls back position – the AGW models fits key parts of space but not enough periods of time. It is, allegedly, too early to tell if the trend is cyclical or secular.
Fyodor Says: August 19th, 2005 at 9:57 am
Who died and made Fydor the Pope of indefinitely extending the duration of time series? A model that consistently predicts a year-on-year trend for more than a few years looks a pretty good bet to me. What are the odds that random noise or accidents will all tip the trend in the same direction?
Fyodor insists that, deep down, he is really just a critical onlooker who cares only for the advancement of knowledge. Well they all say that. Will he set some benchmarks or litmus tests for the validation of his hyper-skeptical theory of Global Warming? How many years of increasing temperature, and how high will the temperature have to go, before he admits he is wrong? I would be inclined to say that Hell would Freeze over before we saw that happening, if it were not for the inapt natue of the metaphor.
Fyodor also suggest that GM might not be anthropic in source, so maybe it is geologic, oceanic, cosmic, what ever, and it is only a coincidence that the temperature gets hotter now that 6 bill humans are all burning just about anything combustible. Or perhaps little devils are secretly cranking up everyones thermometer when scientists are not looking.
By this time we are all so exhausted trying to follow the Byzantine path of Fyodor’s bizarre self-justification we might just be fooled into believing a little diabolical intervention.
Chris, I wasn’t confused by your explanation of how I was misleading, but the point of the paragraph. You start off by quoting my comment on the historical variation in temperature, then leap to a non-sequitur on satellite data, with no relevance to my quote. It’s not my fault you can’t structure a paragraph.
Re: your second paragraph, it’s true I can’t be bothered to track down the actual data, but here is a sample of charts from several different reconstructions, several of which support my earlier assertion. As I said before, however, these are “reconstructions” of data sets, not accurate observations.
Re: your third paragraph, I was under the impression that JQ determined the original issue of this thread. I’ll stand corrected if you can prove you’re the evil genius manipulating the good Prof.
Re: your fourth paragraph, yes, I do have problem with bald statements like “the earth’s surface is the warmest it’s been in a thousand years”, because we simply don’t know. And, yes, you did say it. The quote from comment #31395 is:
The reconstructions above suggest you’re right, but you didn’t qualify the statement with the caveats surrounding the integrity of the underlying reconstructed “data”. I apologise for being pedantic, but you’re the one fixated on the data.
Stephen L,
I’ll address your post in parts, if I may.
They haven’t actually modelled what happened from 1900 to 1997, as they only started modelling in the early 1990s. I hate to go over this again, but, given they already had the data from, say, 1900 to 1997, making the models fit the historical data was not a stretch. In fact, it’s expected. My point is that it’s much harder to create a model that can reliably predict future temperature and, as the models have only been around for, say, eight years, we don’t have much evidence to validate or invalidate them. A simple univariate trend model could have “forecast” the results we’ve seen, but that’s not evidence of predictive power. Correlation is not causation.
IMO, that the surface and satellite observation differed was not a problem with the models per se, but with the hypothesis that global warming was actually taking place. I think this is a secondary issue to whether models are reliably predictive or not [the models simply reflect the data and structure that go into them: garbage in, gargage out], but JQ notes that several GW sceptics seized upon this as evidence of weakness in the models. I’m not particularly hung up on this issue, as I think JQ has shown that the discrepancy is fast being resolved in favour of the surface data, and that global warming is taking place.
Eight years worth of model results for a complex and volatile system like the Earth’s weather is not conclusive about the models’ predictive power. As I state above, a simple trend model could have produced a similar result, but been totally worthless as a forecasting tool.
The models may suggest that, but if the models are unreliable, so are their forecasts. Back to square one.
Again, nobody has any accurate forecast of what will happen in the future, or when. This talk of deadlines is totally unscientific and redolent of hysteria rather than rational decision-making.
Very droll. Let me stretch the analogy further: I usually look in both directions, and listen, to make sure no car is coming before I cross the road. You’re suggesting we only need to look in one direction to make that decision.
Shorter Strocchi: just to prove I’m not really, really angry at Fyodor, I’m going to have a stab at discussing the thread topic, and prove comprehensively that I am, in fact, a buffoon.
Love those snappy one-liners, Auntie Jack. It’s good to see you’ve cut back on the aimless verbosity. However, please read my comments before responding – it’ll be less embarassing for you.
Fyodor
“1) Opinion is not science.” – Sorry in their scientific opinion which is based on their studies of the atmosphere – this OK?
“2) Reread my previous comments about the limitations of scenario modelling.” The models give a range of scenerios and are reliable enough to assess the risk.
“3) You’ve never seen me write that I might be wrong? I have, but for your benefit let me state it again, categorically: I might be wrong.”
So now take the logical step and analyse the consequences of you being wrong. You being wrong means is that the global climate will change and people like you will have prevented action to reduce CO2. We will not be prepared and potentially billions of people could have their lives affected. Pretty bad consequences for being wrong!!!
“We will only answer these questions when the effects are irreversible.
Conjecture.”
Not conjecture at all. The CO2 we are putting into the atmoshere will stay around for 50 to 100 years even if we ceased CO2 production now. So the buildup if it is left long enough will not be reversible at least in our lifetime.
“Again, opinion is not knowledge. How can you be certain that you have enough information when you don’t know how the Earth’s weather works?”
We have enough information to assess the risk.
“ises? Absolutely. WTF does â€?…non-zero risk that the climate will change.â€? mean? History alone tells us that climate will change. That’s not in dispute.”
History might tell us that the climate changes but not exactly when. History also tells us that sometimes it changes around the times when CO2 rises in the atmosphere. Without anthopogenic CO2 and human actions to reduce carbon sinks the climate might not be changing right now. You cannot say that there is a 100% chance that the climate will NOT change in the next 20 or 40 years due to AGW. While you cannot say this then there is a risk. This risk should be mitigated by action to reduce one of the main drivers of climate change – high levels of atmospheric CO2.
Despite the easy dismissal of the argument about the relationship between the sun and climate change it is very hard to feel comfortable with the declarations about the projections for the warming of the climate and who and what is responsible when the explanations are constantly changing. No doubt someone will explain with gusto why reports such as the following are irrelevant but it is difficult to believe the doomsayers when the only elements which seems to remain unchanged in the debate are political.
“Typically, warming of the climate leads to increased melting rates of sea ice cover and increased precipitation rates. However, in the Southern Ocean, with increased precipitation rates and deeper snow, the additional load of snow becomes so heavy that it pushes the Antarctic sea ice below sea level. This results in even more and even thicker sea ice when the snow refreezes as more ice. Therefore, the paper indicates that some climate processes, like warmer air temperatures increasing the amount of sea ice, may go against what we would normally believe would occur.�
http://www.universetoday.com/am/publish/sea_ice_may_increase_in_antarctic.html?1682005
Scientists face difficult challenges in predicting and understanding how much our climate is changing. When it comes to gases that trap heat in our atmosphere, called greenhouse gases (GHGs), scientists typically look at how much of the gases exist in the atmosphere.
However, Drew Shindell, a climatologist at NASA’s Goddard Institute for Space Studies, New York, NY, believes we need to look at the GHGs when they are emitted at Earth’s surface, instead of looking at the GHGs themselves after they have been mixed into the atmosphere. “The gas molecules undergo chemical changes and once they do, looking at them after they’ve mixed and changed in the atmosphere doesn’t give an accurate picture of their effect,” Shindell said. “For example, the amount of methane in the atmosphere is affected by pollutants that change methane’s chemistry, and it doesn’t reflect the effects of methane on other greenhouse gases,” said Shindell, “so it’s not directly related to emissions, which are what we set policies for…..
After isolating each greenhouse gas and calculating the impact of each emission on our climate with a computer model, Shindell and his colleagues found some striking differences in how much these gases contribute overall to climate change……
According to new calculations, the impacts of methane on climate warming may be double the standard amount attributed to the gas. The new interpretations reveal methane emissions may account for a third of the climate warming from well-mixed greenhouse gases between the 1750s and today. The IPCC report, which calculates methane’s affects once it exists in the atmosphere, states that methane increases in our atmosphere account for only about one sixth of the total effect of well-mixed greenhouse gases on warming
http://www.physorg.com/news5258.html
What makes it particularly suss for me is that projections are shouted as predictions and those predictions are confidently, though always with negative outcomes, extrapolated over decades.
Ros – Reading the article that you posted it is very interesting that CH4 could be more of a forcing than we thought. This is both good and bad. Good because we can reduce CH4 reasonably easily and maybe keep some of our fossil fuels that we need so badly.
Really really bad because of this: http://www.newscientist.com/article.ns?id=mg18725124.500
“THE world’s largest frozen peat bog is melting. An area stretching for a million square kilometres across the permafrost of western Siberia is turning into a mass of shallow lakes as the ground melts, according to Russian researchers just back from the region.
The sudden melting of a bog the size of France and Germany combined could unleash billions of tonnes of methane, a potent greenhouse gas, into the atmosphere.”
So of this is true I guess we are pretty well screwed.
This really goes to what Fyodor has been saying that we should not stop researching climate change. However it does not invalidate what I am saying that we have enough information now to assess that the risk of future climate change is sufficient to take action on reducing greenhouse gases.
Fyodor
One last try
I hate to point out you haven’t gone over it yet, beyond making vague statements about “data fitting”. Please explain *how* you think the modellers made the models fit the historical (20th century trend) data. By fudging the forcing time series? By fiddling with a hidden knob in the back of the computer? Please tell.
Mark Hadfield Says: August 20th, 2005 at 2:31 pm
Well said Mark.
Fyodor assumes because some models have the GIGO form it follows that all models are suspect. That is, what Stove calls, the epistemology of the “silent scream”, where humanity is locked inside its intellectual structures, with no way of reliably transacting knowledge to or from the outside world. A recipe for “directionless quibble” from the one who knows best.
He has failed to grasp the elementary distintion between empirical data fitting (interpolation) and theoretical data predicting (extrapolation). The AGW model is of course robust for historical data, and now tracks new data. In fact, the AGW model has now received independent corroboration from the satellite data, plus a steady time serial verification from the surface data.
This would be enough to satisfy anyone but a paid hack of the coal industry or a malicious intellectual mischief maker. No prizes for guessing which category Fyodor fits into.
Ender throw in enormous peat bogs in Indonesia that could dry out and burn releasing huge amounts of CO2.
I beginning to think we are actually going to have change on the extreme end of the scale. I heard somewhere that the IPPC actually have been being very conservative looking for a consensus and have been ignoring the evidence for a more extreme temperature change.
Get ready folks its going to be a hell of a ride.
>I’m beginning to think we are actually going to have change on the extreme end of the scale. I heard somewhere that the IPPC actually have been being very conservative looking for a consensus and have been ignoring the evidence for a more extreme temperature change.
I forget the details but apparently the IPCC consensus forecasts are simply the arithmetic mean of the different models, when you start adjusting for the different confidence intervals and the differing quality of the models (based on back-casting capacity and the like), the likely warming figures shift up drastically.
Of course, any such statistical manipulation is immediately suspect.
Fyodor,
On the one hand you have a computer model – input a affects output b according to a given mathematical formula and outputs c,d,e and e according to other formula; ditto for several thousand more inputs and outputs. Plug in numerical values for the inputs and you can run the model forward or back in time.
On the other hand, you have empirical data for the actual past sdtate of the climate.
Explain ot me, exactly, why you think its a trivial exercise to make the output of the first exercise match the empirical date, remembering that the model also has to demonstrate predictive value over the short term.
I know a bit about computerised econometric modelling – mainly from the perspective of having been a government employee commissioning such studies and looking at the results – back casting is not the simple process you seem to think and demonstrating the capacity to back-cast is normally regarding as an important demonstration of the robustness of a model.
Ian:
That’s interesting: in the atmospheric science community it’s called “hindcasting”.
And then we have “nowcasting”, which actually means short-term forecasting, based on simple extrapolations of the current state.
But I digress…
test
Mark, “data fitting” has nothing to do with fudging data, or fiddling with knobs (except, perhaps, figuratively). Data fitting, as I stated here, is working and reworking, through many iterations, a model until its output, based on independent data input, matches the dependent variable being modelled. In the case of global warming, an example would be modelling global mean temperature as a function of many variables, such as solar radiation, volcanic activity oceanic feedback, cloud cover etc. Now, how is this model constructed? Do we set the parameters of the model based upon known scientific principles and relationships, or do we calibrate the model so that it matches the recent historical record? This may seem like a trivial difference, but it’s very important in retaining the scientific integrity of the model. My concern is that many models may be so manipulated to fit the data (i.e. so that they look accurate in hindcasting), rather than based upon sound knowledge of the underlying science (i.e. which would make them perform well in forecasting). As Ian Gould points out, the ability to “predict“ historical observations a posteriori is a good test of a model, but if such testing is used upfront to drive the model calibration, you risk losing the integrity of the model – all it becomes is a “fitâ€? to the past, not a useful model for predicting the future.
Well done, Jack: you’re learning. You may now address me as Dr. Pavlov.
Your latest effort shows yet again, that your willingness to over-reach your limited intellect and understanding is unmatched. You’ve done nothing more than regurgitate what others have said, less elegantly. It is, however, a huge improvement in focus and brevity, and I suppose I should throw you a bone by correcting your mistakes.
For your homework, please look up the meanings of “interpolation� and “extrapolation�. There’s no single “AGW model� and of course the models are historically robust – I said that point was trivial, and you’re simply repeating me. The recent work on satellite data doesn’t “corroborate� any models; it confirms the surface data.
Full marks for “time serial verification�, however. That’s a really impressive Stroppylogism. Honestly, you’re downright embarrassing when you make this stuff up.
Ian,
It is a relatively trivial exercise to make a mathematical model “fit” empirical data. It is done by selecting independent variables and coefficients for those variables that produce the best fit to the historical data. As you point out, this does not necessarily produce a reliably predictive model, and this is the point that I want to stress. Producing a model that can reliably predict future outcomes is much more difficult than producing a model that can “back-” or “hindcast”.
I am also familiar with models of many kinds, and it’s not my intention to trivialise the work of the modellers. Rather, I want instead to emphasise how hard it is for them to produce really good models for such a complex system, and how sceptical we should be when using the output of models we know are likely to be flawed, given: a) the risks inherent to modelling; and b) our limited understanding of the forces affecting global temperature.
You may be interested in one technique they use in weather forecasting these days. They make a number of projections based on slight variations of the initial parameters. Typically they get groups of similar results, but not one single group, and maybe a few outliers. They then estimate the probable outcome based on the groups. This rests on having the computer power to do more than one projection before overtaken by events. Presumably something similar could be done with climate modelling; I don’t know if it has been done, though.
PML presumably they do. I recall several years ago the National Climate Centre saying that 7 out of 12 models predicted an El Nino.
But good forecasters also look out the window before issuing a forecast. There is a lot to observe at presesnt. Huge frozen peat bogs in Western Siberia melting, bits falling off Greenland and the Antarctic, loss of sea ice in the Arctic, northern Canada turning mushy, glaciers in rapid retreat just about everywhere, loss of snow on the Australian Alps consistently over 50 years, an increase of average temperatures in Australia of 1 degree, markedly drier autumns and winters, and finally, cricket matches being completed with not too much rain in England. I’m sure I’ve missed about 150 important signals manifestly clear to all who watch the weather.
An old bloke with relatives in the Fassifern Valley (around Boonah, SW of Brissie) told me the other day that the farmers are confused and don’t know what to plant anymore when they do get rain.
Not conclusive, mind you, but it’s getting to look awfully suspicious.
Just because a decent climate model doesn’t come along until 1997 doesn’t mean we can’t go back to say 1975 and use only measurements that were made up to that time to test if the model gives accurate forecasts from that time onwards. Those forecasts from pre-1975 measurements can be tested against post-1975 measurements. Such tests would be part of the normal testing of the climate model.
Fyodor doesn’t give the impression that he thinks anyone might have done such tests. I hope he doesn’t assert that noone has done such tests because that would be a pretty brave assertion to me.
Saying that models have made forecasts from 1997 to now is thus missing the big picture. It may well be that the warming in 8 years is significantly less than any short term warming in the last 1000 years (although saying “many times” less is stretching it) but that’s missing the real point which is that warming in the last 30 years (about 0.6 degrees C) is more than any short term change in the last 1000 years, according to peer-reviewed estimates. Missing the real point is being misleading. I’d be very surprised if there was no model that confirmed this warming using measurements from before 1975.
I may well be that JQ determined the original issue of the thread. Of course it may well be that the conclusions about the satellite data came into JQ’s head without him reading about it. Then again, maybe fyodor just has a problem with semantics.
I asked fyodor if he wanted to just say we’re 99% certain the earth’s surface is the warmest in a thousand years but he ducked the question.
No, I don’t and yes, it would be.
Re: the “big picture” in your third paragraph, the temperature data you’re talking about over the past 1,000 years is reconstructed. We simply don’t know – with accuracy – what the temperature record was like in the centuries before the 20th.
Re: your last question, I apologise for not answering it – I wasn’t sure which of the four in that last paragraph wasn’t rhetorical. The answer is no, we’re not 99% certain. Determining such a level of statistical confidence for volatile data that is not observable is difficult, to say the least.
Another part of the big picture is that the models can be tested for more than just 8 years of forecasting ability. Saying that testing for just 8 years of forecasts is an insufficient test is missing the point. The point is that the models forecasting abilities can be tested for forecast periods of 30 years or more. What forecast period is a sufficient test?
Yes we don’t know what the temperature record was, long before the 20th Century, and this means that it was absolutely pointless trying to find out what temperatures probably were because they could have been absolutely anything. All the research on paleoclimate has been a complete waste of money. Seriously though any scientific research worth its salt makes probability range estimates of its results and indeed this has been done as shown in this Bureau of Meteorology jpeg. Such estimates can allow us to make a probability estimate of whether it is now the warmest in 1000 years. BTW I thought some of my earlier questions were reiterative and none were rhetorical.
Chris,
It’s not a true test if you already know what the results should be and can cheat by changing the model so that the output matches the “expected” result.
I don’t really understand why you keep harping on about current temperatures being the hottest in the last 1,000 years. We know from palaeoclimatic research [See? It wasn’t entirely pointless.] that temperature varies greatly over time, so if it is true what does it prove?
Fyodor Says: August 23rd, 2005 at 9:24 am
This passage is diagnostic of Fyodor’s methodological confusion and intellectual delusion. He is trapped, poor fool, inside the iron cage of his own solipsistic epistemology. Worse still, instead of trying to escape he has set about hardening the bars.
I love the sneering “scare” quotation marks around the word “corroboration” – a perfect example of what Stove calls neutralizing success-words.
And likewise for Fyodor to heap faint praise on “corroboration”. Thus the madness in Fyodor’s methodology encourages us to believe that empirical corroboration has not only not-occurred in this particular instance, it is perhaps a bad thing in general! Thus the enquiring mind, intoxicated with Fyodor’s epistemological poison, continues to scream in solipsistic silence.
Fyodor does not appear to understand that independent data corroboration is what counts, in positivist methodology, as model confirmation. This is because positivists (the epistemology of most scientists) are not that fussy about the the explanatory elaboration of models – so long as their construction does not violate fundamental principles.
They are “unscrupulous opportunists” (Einstein) who just want the black box (model) to crank out predictions (extrapolations) that track the data (dependent variable). In some cases, the more unrealistic and abstract the model the better, so long as the data is tracked accurately (Friedmans notorious F-twist).
Fyodor believes that the surface data, which describes the behaviour of the dependent variable, is somehow contaminated by association with the, as yet not fully worked out, AGW model. The surface data is one measure of the (AGW-predicted) dependent variable which is global warming in general. Other measures could be ice cores, oceanic metrics etc.
If the satellite data is in independent accord with the surface data then it does not matter how the AGW model is constructed – it is robust over time and accross space. The AGW is now geographicly, as well as historcally, robust. It has survived critical tests extrapolated accross a number of dimensions, not just interpolation along a single dimension.
The satellite data is, in short, independent confirmation not an independent variable. It is not self-generated as an independent variable in the model, it is independently collected thats all.
Fyodor, who suffers from a self-inflicted dose of the “silent screaming” epistemology that has spread like wildfire throughout the slums of academe over the past generation, simply cannot get it into his head that independent data is just that – independent not model-cooked. It is hard to believe that someone who purports to lecture experts in the field could dedicated so much brainpower to justify such a stupid pun.
It is important for commenters to note how much ground that Fyodor, the slippery customer, has conceded in this little aside. This is clear from the spate of “-casting” jargon that has just spread like a rash on this thread, all thanks to Fyodor’s intellectual mischief-making.
He started off questioning the very idea of robust models for complex phenomenon (call this “non-casting”). Then he said that the models are only robust for the 1997+ time period of dedicated observations (“nowcasting”) because rotten modellers only interpolated pre-cooked data to fit their dodgy model. Now he says that, well of course, the models will fit data going back
It is important for commenters to note how much ground that Fyodor, the slippery customer, has conceded in this little aside. This is clear from the spate of “-casting” jargon that has just spread like a rash on this thread, all thanks to Fyodor’s intellectual mischief-making.
He started off questioning the very idea of robust models for complex phenomenon (call this “non-casting”). Then he said that the models are only robust for the 1997+ time period of dedicated observations (“nowcasting”) because rotten modellers only interpolated pre-cooked data to fit their dodgy model. Now he says that, well of course, the models will fit data going back
It is important for commenters to note how much ground that Fyodor, the slippery customer, has conceded in this little aside. His retreat has been covered by an outbreak of �-casting� jargon that has spread like a rash accross this thread since he decided to make pronouncements on the subject.
He started off questioning the very idea of robust models for complex phenomenon (call this “non-castingâ€?). Then he said that the models are only robust for the the 1997-plus time period of dedicated observations (“nowcastingâ€?) because rotten modellers only interpolated pre-cooked data to fit their dodgy model. Now he says that, well of course, the models will fit data going back before 1997 (“hindcasting”). It is only a matter of time before he concedes on “forecasting” (sorry, but Fyodor’s kind of intellectual baboonery encourages bad puns).
Normally someone who backtracked this fast would be laughed out of court as an impostor. But give credit to Fyodor, the bluster that powers his unparalleled combination of arrogance and ignorance comes in handy in when covering hasty backtracks.
It has been a relief for me that others have done most of the heavy lifting in disentangling fact, fantasy and fallacy from Fyodor’s plausible crankery. (Commenters have no idea how hard it is to provethat water typically runs down hill to a man who does not really believe in the concept of probable proof.)
But I must also confess to feeling a certain shame in enjoying the spectacle of knowledgeable persons making mincemeat of Fyodor’s arguments, until nothing of his initial position is left but a tattered remnant. The scene exerts a certain gruesome fascination, like being a ghoulish on-looker at a car-wreck amongst. However, there is not much redeeming intellectual merit, so for shame we must avert our gaze.
Shorter Stroppy: Fyodor is still pissing me off, AND he’s had the last word!
What, no Dr. Pavlov? And after all I’ve done for you, you go and throw yourself off the wagon with a reversion to incoherent spray. You get a big Fyodor “F” for “Fail”, Auntie Jack.
Also, please take care you don’t burn yourself on that Stove. The way you go on about him suggests the book pages might be stuck together.
Fyodor Says: August 29th, 2005 at 12:30 pm
No Fyodor, this-world reality has the last word. Fydodor has a big mouth which he is always sticking a foot in. But even Fydor’s yawning orifice cannot accomodate the whole world of facts.
Question: What kind of fool thinks Australian political parties are not more cultural conservative now than a decade ago, that there was no significant Islamic sectarian aspects to East Timor civil strife, that Muslim jihadists did not show at least as much proactive aggression as Christian Crusaders during the Holy Wars and that Darwinism does not apply to humans at sub-special classifications?
Answer: the Fyodor-ian kind of fool, thats who.
This is the kind of foolish and mischief making attitude exhibited by the sort of person who denies the reality of anthropogenic global warming, and probably the civilising effect of gun control, simply for the self-indulgent pleasure of deological disputation or maybe because they are just plain dumb. And to give his insane, or insincere, epistemological ideas a bit of publicity.
Oh how Fyodor would have loved the medieval period when his peculiar blend of ideological rigidity and specious hair splitting was accorded a much higher status than the kind of blogspatting that he adores indulging in these days.
It is clear that Fyodor’s solipsistic mind is impervious to rational argument and incorrgible to empirical evidence. However, through the miracle of technological inventions and financial institutions, we have an alternative way of resolving the dispute that does not involve resorting to the cerebral domain of the mind. I am talking about a having a little wager, the sort of acquisitive activitiy that would appeal to the livid reptilian lobes of Fyodor’s mind, at least.
Fyodor is, after all, a true believer in the perfectability of financial markets. So he would have to be impressed that his brother AGW skeptics are prepared to put their money where their mouths are, instead of their feet, like Fyodor.
So, if he is prepared to stump up some cash, I am prepared to match him dollar for dollar on a bet that AGW is real and will continue to increase in magnitude. I think that the general conditions of the Mashnich/Bashkirtsev-Annan wager look fair enough to me. Pr Q to hold the bets and deciding fair value odds.
The winnings, I suggest, should be directed to a reputable charity, as directed by the winner. The loser to eat crow in a media to be specified by Pr Q – perhaps this blog who knows?
No bet that involves long term predictions of complex phenomenon is a sure thing. However, in any argument with Fyodor one is entitled to resort to a supplementary epistemological principle in order to deal with what we might call the “Fyodorising” problem (To Fyodorise : verb. To destroy facts in a ‘Fyodorian’ fashion, to ‘pillage’ the truth, poor or no fact checking, wild accusations, conspiracy theories etc.. – with apologies to Auberon Waugh).
This Fyodorising thinker is a sort of reverse compass to the true position. I am pretty confident, given Fyodor’s previous abysmal track record on intellectual disputes, that the world will not be at great pains to seperate
“Fyodor the foolish” from his money.
Shorter Stroppy: No, I have the last word!
Jackpot!
So I take it that Fyodor is chickening out on my offer of a bet on the prospects of AGM? What a wimp! Although, given his form, its probably the smartest move Fyodor has made all day.
Absolutely, Jack. I have no idea what the temperature’s going to be in 10 years. That’s the point of being a sceptic: I don’t believe.
Why should I bet on it? I’ll leave that kind of activity to the credulous and financially incompetent. You, for example.
Fyodor Says: August 29th, 2005 at 6:15 pm
No. The point of a skeptic is to only believe something for which there is reasonably hard evidence. Not to be a “directionless quibbler” or someone engaged in promulgating a solipsistic epistemology.
Fyodor’s statement implies that the world’s long-term meterological modellers are mostly woolly minded suckers. Coming from a world-class scientst this would be considered a deplorable lapse in academic manners. Coming from a loud-mouthed lightweight like Fyodor it is just pathetic.
As for financial savviness, I know at least one v-e-r-y scientificly minded multi-millionaire, with a great track record in the futures market, who wants a piece of that action, on the AGW side. And he sure aint Fyodor. So I guess that means Fyodor is all talk and no action, which we already knew anyway.
I’m sorry, Jack. I didn’t mean to imply that the world’s long-term meterological modellers are mostly woolly minded suckers. I thought it was clear that you are the woolly-minded sucker. Hope that clears up any confusion.
I’m very curious about your “friend”, however. Are you sure he’s not a teensy bit like your other imaginary friends?