Click images for more details



Recent posts
Recent comments

A few sites I've stumbled across recently....

Powered by Squarespace
« Happy Christmas to all - Josh 253 | Main | The professionals »

Brian Hoskins, then and now

The question of how fair a representation the Royal Meteorological Society has given of the reliability of climate models has been discussed in a couple of posts now and readers seem to have concluded that if you wade through the obfuscation they are in fact only giving weak support to climate models. I don't think anyone has made the case that they have given a clear and balanced account though. In my opinion, by speaking of “a confidence in the models’ suitability for their application in detection and attribution studies and for quantitative future predictions and projections” the authors strongly imply that such a confidence in fact exists.

I thought it might be interesting to see what the Science-y Grantham Institute at Imperial (as opposed to the eco-warrior bit at LSE) had to say on the subject. The author team behind the submission was headed by Brian Hoskins, who has been quite vocal on the subject of climate models. Famously, he described them as "lousy" and "terrible" in an interview with the Economist in 2010.

Here's what he has to tell politicians on the subject at the end of 2013:

The IPCC AR5 WG1 dedicates one whole chapter of its report (Chapter 9) to the evaluation of climate models.  The summary to this chapter gives a fair assessment of the improvements made by climate modellers since the last Assessment report (AR4) as well as the remaining challenges.  For example, the AR5 WG1 TS suggests that some models may be too responsive to the effects of greenhouse gases.  We do not yet fully and quantitatively understand the slowdown in surface warming (see our answer to Q9 for further comment).

We agree however, with the key conclusion of the authors of Chapter 9 that, “Climate and Earth System models are based on physical principles, and they reproduce many important aspects of observed climate. Both aspects contribute to our confidence in the models’ suitability for their application in detection and attribution studies (Chapter 10) and for quantitative future predictions and projections.”

This does not imply bind acceptance of the results from the current models: continual questioning of the results and improvements in the models is crucial.

The similarity in approach to the Royal Meteorological Society is striking: there is exactly the same ambiguity, allowing the authors insinuate that the scientific community has confidence that the models are suitable for quantitative prediction while giving them plausible deniability in future. However, as you can see, the Grantham statement is at least surrounded by caveats about the unknowns of the climate system, so it's a better summary of the state of play than the weathermen's.

That said, you really don't get a sense of "lousy" and "terrible", do you?

PrintView Printer Friendly Version

Reader Comments (41)

If you admit you don't fully understand the mechanisms, then you don't have a good model. You are in effect admitting flaws. Serious flaws, I'd say. But then folks might come here splitting hairs about whether the flaws are admitted to be serious without addressing what they are doing to fix the models.

Now, I suspect some real work is going on behind closed doors to get the models right. Whether they can overcome their own prejudices and false premises I do not know, but that debate ought to be taking place in public or in the literature, not by implications and forms of wording to supply deniability in some arse-covering exercise.

Dec 22, 2013 at 10:49 AM | Unregistered Commenterrhoda

The summary to this chapter gives a fair assessment of the improvements made by climate modellers since the last Assessment report (AR4) as well as the remaining challenges. For example, the AR5 WG1 TS suggests that some models may be too responsive to the effects of greenhouse gases.

The second sentence does not logically follow from the first as it is not an example of improvements.

Does anyone know what improvements have been made between 2010 and 2013 that have turned lousy and terrible models into models in which it can be claimed we can have "confidence in the models’ suitability for their application in ..... quantitative future predictions and projections"?

Dec 22, 2013 at 10:55 AM | Registered CommenterPhillip Bratby

...I don't think anyone has made the case that they have given a clear and balanced account though....

Of course not! What do you think that these bodies are there for?

They are NOT research bodies supporting the government by feeding them basic data in a digestible form. That is done at a much lower level. At the top end, these organisations are part of the political machine. And the political machine does NOT want surprises, or the truth - it wants support for its policies. Sometimes that support will be unequivocal (when the policies are practical) and sometimes it will be lukewarm (when the policies are insane). But the system will always provide support. That's what it's there for.

Take an example from a completely different part of government, the intelligence services. Remember when the politicians wanted a war with Iraq? The intelligence services duly provided a very similar dossier supporting the idea that Iraq might be a vague threat. Completely untrue, of course, and written in mandarin-speak, so everybody in Whitehall knew that it was untrue - but that was not the issue. The dossier gave the government carte blanche to run their policy and join the war.

Look at all the justifications for government policy. Look at the HS2 traffic analysis. Remember the support for Identity Cards? Did you even expect a 'clear and balanced account' in these cases? So why are you expecting one over global warming?

Dec 22, 2013 at 11:04 AM | Unregistered CommenterDodgy Geezer

There are known knowns. There are things we know that we know. There are things we know we don't know. But there are also unknown unknowns. There are things we don't know we don't know.

Until they Admit to the above, then this circus will continue

Dec 22, 2013 at 11:08 AM | Registered Commenterandrw100

The second sentence does at least give an example of (arguably) the most important of the "remaining challenges" which might be seen as a small chink in the wall.
As for the rest, I refer you to the answer given by my honourable friend Dodgy Geezer!

Dec 22, 2013 at 11:25 AM | Registered CommenterMike Jackson

A simple law;

Models aren't real.

Dec 22, 2013 at 11:31 AM | Unregistered CommenterRightwinggit


Bish you are too polite. How about, deceit, concealment, fraud. It doesn't matter though its only mega pounds being blown.

Dec 22, 2013 at 11:40 AM | Unregistered CommenterMartyn

"...the recent slowdown in surface warming"

I think he means the halt in surface warming.

Spinning to deceive.

Dec 22, 2013 at 12:00 PM | Registered CommenterMartin A

“Climate and Earth System models are based on physical principles, and they reproduce many important aspects of observed climate.

Before Copernicus astronomers would have said the same thing about Ptolemy's model of the motions of the planets. However it is unfair to compare climate models with the Ptolemaic system. The later could produce fairly accurate predictions of planetary motions for a long time into the future. Climate models have proved pretty useless for predicting the future. However, to give the Devil his due, climate models can be successful at "predicting" the present using historical data. I wonder why that is?

Dec 22, 2013 at 12:13 PM | Unregistered CommenterRoy

There is tooooo much money involved for anyone to take 5 paces forward. I/2 a pace is about as far as anyone is likely to go. Hoskins made a 1/2 forward and a 1/4 backward.

Dec 22, 2013 at 12:27 PM | Unregistered CommenterStephen Richards

'Grantham' he who pays the piper calls the tune, he would not think for a second to pull the funding if 'his department ' failed to meet the task 'he needs ' it to do. He did not get to be very rich but throwing his money away onto idea he did not want . Why anyone thinks the cash he provides for the LSE etc is any different is an good question.

Dec 22, 2013 at 12:28 PM | Unregistered CommenterKnR

The Grantham Institute is just as alarmist. Real science out of the window. No wonder Imperial has sunk down the world University list. It used to be no. 2 to MIT, as no. 1 but even that has slipped. Poor science based on climate alarmist government grants.

Dec 22, 2013 at 12:43 PM | Unregistered CommenterJohn Marshall

"The models are based on physical principles."

I'm sure they are, but maybe the wrong ones. Consider for the sake of argument that our climate last century was mainly due to warming in line with increasing solar activity with further rapid warming due to the temporary cumulative effect of different ocean cycles. Assume that GHG effects are real but negligible because the planet has a range of feedback mechanisms that maintain good thermostatic control.

Now consider that for decades, climate scientists and modellers have concentrated all their resources on the GHG radiative effect. A whole generation of scientists have been taught that CO2 has controlled our climate and was even responsible for the ice ages. Their peers, the establishment, the press and most of the worlds population now believe in CO2 driven climate change.

There is just this irritating problem of the pause....

I don't think they ever possessed the objectivity that is essential in good science. They will find it almost impossible to start thinking the unthinkable. Don't expect an admission that the models are wrong any time soon.

Dec 22, 2013 at 12:53 PM | Unregistered CommenterSchrodinger's Cat

@Schrodinger's Cat: what's more, they have taken on, from Meteorology, a highly-flawed radiative effect analysis, a failure to distinguish between the potential energy flux from an emitter to a sink at absolute zero, which is what they measure, and the net Irradiance, which is what really drives radiative heat transfer.

Dec 22, 2013 at 1:06 PM | Unregistered CommenterMydogsgotnonose

Think of the Councils at Nicaea. The various factions and cults kept at it until they got their story straight as the Emperor directed.

Dec 22, 2013 at 1:16 PM | Unregistered Commenterhunter

"...quantitative future predictions and projections." Oh, I see, never mind the quality, feel the width.

Dec 22, 2013 at 1:35 PM | Unregistered CommenterSnotrocket

I swear that the evidence that I shall give, shall be the truth, the whole truth and nothing but the truth, except when it comes to climate and climate models, so help me Politicians.

Dec 22, 2013 at 2:19 PM | Unregistered CommenterMartyn

Roy, the Ptolemaic planetary model is my favourite example of a model that worked well for the wrong reasons. The great hydrologist Vit Klemes noted that the Ptolemaic planetary model gave the right results despite a profound misconception while the Copernican model worked well because it was based on a scientifically correct principle. Klemes went on to note that:

"Had computers been available at the time so that a more refined fitting of the Ptolemaic model could easily be implemented and adding another epicycle would mean just going once more through a DO- loop, this model could still be with us today and space exploration would still be only a fantasy."

Dec 22, 2013 at 2:56 PM | Unregistered Commenterpotentilla

Until that classically simple summary of the state of knowledge of any subject by Donald Rumsfeld I always had problems when I said "i don't know what I don't know". Quick often it was a conversation stooper. Now all one needs do is quote that and say come back when you understand the problem (with the theory) and we can have a proper discussion.

My feeling is that the unknown unknowns are forgotten/ignored in too many fields and the mis-application of Sherlock Holmes' "How often have I said to you that when you have eliminated the impossible, whatever remains, however improbable, must be the truth?"

Dec 22, 2013 at 3:16 PM | Unregistered CommenterSandyS

This re-positioning of his views by Brian Hoskins reminds me of another leading player who seems to have beefed up their perception of what is possible with "models" in very short order back in 2009.

Dec 22, 2013 at 3:26 PM | Registered Commenterretireddave

Anyone else finding BH is loading very slowly (or not at all) lately?

Dec 22, 2013 at 4:13 PM | Unregistered CommenterTC

TC - yes I have noticed that a couple of times recently. Seems OK at the moment.

Dec 22, 2013 at 4:21 PM | Registered Commenterretireddave

Yesterday morning. I nearly gave up and shut down the system twice to see if it was a problem my end. OK today, though.

Dec 22, 2013 at 4:44 PM | Registered CommenterMike Jackson

Seems to be a problem with Chrome. Firefox seems to be working fine though. Unfortunately I'm a fan of Chrome (normally).

Dec 22, 2013 at 4:51 PM | Unregistered CommenterTC

I'm beginning to wonder whether Google is doing a Reddit? Just jesting .. I think.

Dec 22, 2013 at 4:53 PM | Unregistered CommenterTC

I've had a few problems with firefox today and yesterday, but it's OK at the moment.

Dec 22, 2013 at 4:59 PM | Registered CommenterPhillip Bratby

Here's a way to scare people. Tell them that things are going to be so different in the future that all their past experience is no longer reliable as guide to their future.

That's quite unsettling. What are we to do?, they might well ask. You've got their attention, now move in for the kill.

Tell them about your computer models that warn of big rises in temperature. Tell them of your chums who have found all sorts of ways in which a rise of temperature is really really bad for you (and for lots of other people as well). Remind them that their fond memories of warmer weather in the past are not to be trusted, nor are they to take seriously talk of medieval or Roman warm periods, nor in general are they to take anything at face value that suggests warmer has been better anywhere. That's all in the past after all. No, they must rely on your computer models. Why is that?, they might ask. Well, one reason is that basic, simple theory suggests not-very-scary rises in global mean temperature per doubling of ambient CO2, and we have quite a while to wait for a doubling at current rates of growth. 'Steer the client away from that', is the guidance in the sales-manual for WarmAlarm Inc's products and services. Tell them it is too simple. Tell them the system is so complex that only computer models will do.

Do not whatever you do get involved in comparing model outputs to real data, and especially do not get into the business of comparing forecasts with observations, especially forecasts which are beyond what the punters might well have come up with themselves using 'past experience' and.or simpler models of the kind that would fit in a decent programmable calculator. You can try the 'we don't make predictions, only projections' line - that's good enough for folks who are already quite far down the road of buying your products and services - they want them for all sorts of reasons of their own after all.

For targets showing more resistance, you may decide to give up on them. They may be a tiny potential market, and business is pretty good with the early adopters and softer touches anyway. To reduce the damage they might do by bad-mouthing your arguments, try conceding a lot of ground. Yes, the models are useless for predictions. Sure they fail to agree with observations just about everywhere you look. Yes they can't do clouds very well, nor ocean currents, nor biological processes, nor precipitation, nor Europe, nor the south poles,. nor hurricanes, and so on and on. Let them see that you can feel their anxieties. Assure them that with a bit more money, and a bit more time, the models will get better and better. Flood them with complexity. Get them to think that computers are the only way out of it. They may not be much good, but they're the best we've got. That kind of thing.

What you must avoid, at all costs, is any hint that we'd all be better off if these models had never been invented. That they have been a substantial distraction from good theory and experiment in the real world. Why bother with that messy area, when you can happily discuss virtual worlds without end? Climate projections which have local authorities planning for heatwaves, droughts, overheating classrooms, grapes growing on the moorlands, and sea levels rising at unprecedented rates, not to mention 'more extremes of every kind', can lead to money being wasted. What have they been spending on that they would not have been spending on if they had not received the scary guidance? What other opportunities were lost as a result? What if 'past experience' was in fact quite a useful guide to the future? Mustn't go there.

Dec 22, 2013 at 6:00 PM | Registered CommenterJohn Shade

Same as Philip Bratby, if it persists (OK currently) I'll try Chrome, Opera and Safari and post the results. Being a paranoid sort of person I regularly use a different browser and search engine just to keep GCHQ on their toes.

Dec 22, 2013 at 6:17 PM | Unregistered CommenterSandyS

Here's a useful link to help decide if a blog failing to load or reported as unavailable is just a problem at your end:

Now, get back on topic and read my previous comment for starters! :)

Dec 22, 2013 at 6:24 PM | Registered CommenterJohn Shade

...Being a paranoid sort of person I regularly use a different browser and search engine just to keep GCHQ on their toes....

Not a problem. We watch everything at the packet level. The only additional work that you cause is that we note that you are using different browsers and search engines occasionally, but this is automatically logged, so requires no human intervention.

However, using different browsers is a suspicious marker, so you are now on our files as moderately suspicious. This is useful to us, because we need a lot of slightly suspicious people to justify our budget proposals of several billions to expand our remit. If we could not claim that we were saving the world from suspicious people, we would have had to close down operations when our last 'enemy', the Soviet Bloc, stopped being an enemy in the early 1990s.

So thank you for keeping us in work.


GCHQ and NSA staff

Dec 22, 2013 at 6:38 PM | Unregistered CommenterDodgy Geezer

John Shade
" What have they been spending on that they would not have been spending on if they had not received the scary guidance? "

Well until December 2010 what they weren't spending on, when they should have been, was salt and snow clearing equipment.

The Met Office guidance to local authorities, utilities and many companies, was warming forever and drought forever - until it was cooling and the floods arrived.

I would guess that some these organisations understand the limitation of the advice they get now.

The idea that a forecast of climate change advice can be made for any period in the future makes snake oil selling a viable line of work.

Dec 22, 2013 at 7:43 PM | Registered Commenterretireddave

This type of a re-positioning and flipping between different meanings of concepts is quite common in the climate debate. Give it a few years and we sceptics will comfortably be back to being the idiots again.

Dec 22, 2013 at 9:27 PM | Registered Commentershub

""The author team behind the submission was headed by Brian Hoskins, who has been quite vocal on the subject of climate models. Famously, he described them as "lousy" and "terrible" in an interview with the Economist in 2010.""

To be fair what he actually said was:

"Interviewer: Tell us the war story. How bad were the climate models when you started out?
Brian Hoskins: Ah, they were pretty lousy, and they're still pretty lousy, really. They were terrible. They had difficulty representing the general sort of high pressure over the poles, and then the general westerly winds in middle latitudes, say. They were approximately right, but -

Interviewer: - and this was how far back?

Brian Hoskins: You go back to the - well, I'm afraid I go back to the '70s, and that's when - yeah.

So the "terrible" refers to the 1970's and the "pretty lousy" refers to 2010 one presumes. He may have changed his mind since then.

The damning part of the current opinion to me is:
"" For example, the AR5 WG1 TS suggests that some models may be too responsive to the effects of greenhouse gases.""
If they have not got the response to greenhouse gases right and that is the only thing that policy makers are currently interested in then they are as useless as a piece of seaweed hung on a nail. That admission is worth framing for posterity.

Dec 22, 2013 at 10:29 PM | Unregistered CommenterIvor Ward

"The models are based on physical principles." Are there two sets of physical principles, one for climate prediction and one for weather forecast? Is that why we have a reliable 100-year climate prediction, but not a reliable 100-hour weather forecast?

Dec 23, 2013 at 12:15 AM | Unregistered CommenterCurious George

So Hoskins does not 'fully or quantitatively understand the slowdown in surface warming.' Again, no mention of any possible fault in the model maker's assumption of climate sensitivity. At least there was no arm waving, and talk about deep oceans.

Dec 23, 2013 at 10:18 AM | Unregistered CommenterPeter Stroud

I've always assumes that the 'Based on Physical Principles' comment is of a similar status to drama being described as 'Based on Historical Events' - true, but only for a certain (and variable) value of true. After all, are Braveheart or U571 historically accurate?

Certainly there is some solid physics in the climate models - absorption spectra for CO2 and other GHGs, equations of fluid flow (although noting of course that these are chaotic and so do not resolve to a unique solution), and probably a few other things that are reasonably well constrained. The problem is that as well as these 'known knowns', the models include quite a few 'known unknowns' auch as aerosol effects and quite a few things that cannot be modelled accurately but have to be parameteriased (e.g. cloud formation and storm development, presumably convective heat flow). And then of course you have the possibility of 'unknown unknowns'...

Dec 23, 2013 at 10:25 AM | Unregistered CommenterIan Blanchard

We all know the models are worthless. It is obvious that negative feedbacks dominate or the world would have passed a 'tipping point' at some time in its history, after all CO2 concentration has been much higher before. Anyone with a slightly technical background should be asking for very clear proof of the positive feedbacks employed in the models. It is also obvious that a little (natural) warming & increased CO2 is beneficial.

The reality is that global warming / climate change / extreme weather is just too lucrative, from scietivists obtaining grants, prestige & perks (conferences) to governments taxing us for our own good. The left latch on because anything that increases taxes and grows government is good, the eco-terrorists protestors want to be disruptive but they also attract the stupid who want to 'return to nature' (but with iphones & skiing holidays). Many businesses are well aboard the bandwagon for a share of the subsidies funded from extra tax. The only casualties are the taxpayers.

There's a great deal of money in it, few of the troughers will go quietly.

Dec 23, 2013 at 10:40 AM | Unregistered Commenterjaffa

I agree.

Dec 23, 2013 at 12:46 PM | Unregistered CommenterSchrodinger's Cat

Suppose I had 1% confidence in the models before. After the latest work, I have 2% confidence.

Would it be accurate to say that the work has boosted my confidence, or that it has contributed to my confidence in the models’ suitability...? Strictly speaking, yes.

But if I announced "the work has boosted my confidence in the models" to someone who didn't know the background, would that person be totally misled? Clearly, yes.

Dec 23, 2013 at 1:08 PM | Unregistered CommenterHK

I respectfully direct any of you who have the time in the frenzied run up to Christmas, to read Nic Lewis's formidable presentation to the Energy and Climate Committee's discussion on the IPCC's Fifth Assessment (see Bish's feature: December 18th)..
In particular, see Nic's damning comments about the use of 'subjective priors' when running climate models.
Basically, this means that you, in your opinion, can put in whatever starting data you want. This contrasts with 'objective priors' which imply unbiased industry standards.

Never forget, folks, that the IPCC was set up on the following premise: 'We think that man-made carbon dioxide is dangerously warming the planet . Prove it...'

Dec 23, 2013 at 1:30 PM | Unregistered CommenterSherlock1

Robert S McNamara (US Secretary of State for Defence during the Vietnam War) told his air force chiefs after his discovery that they were using the number of buildings destroyed by bombs as a critical success factor.

Measure what is important; don't make important what you can measure

I fear that the "modelling community" may have fallen into the same trap. Not content with modelling a process that is far from fully understood, using parameters to disguise their ignorance, they have then "bet the farm" on it all being based around CO2 - which is (relatively) easy to measure - and which seems to have declining impact as more empirical studies are done.

Dec 24, 2013 at 9:25 AM | Registered Commenterpogo

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>