Thursday
Jan102013
by 
Bishop Hill
Bishop Hill Spot the difference
Jan 10, 2013 Climate: MetOffice Climate: Models I'm still suffering. Even whisky isn't working. It must be serious.
In the meantime, Paul Homewood has found something interesting about the Met Office's forecasts.
Reader Comments (216)
I too should like to add my thanks to Dr Betts for his contribution here.
The issues of Met Office forecasts is the subject of John Redwood`s blog today. I have posted, inter alia, as follows:
" It would be a considerable public service if you were to prevail upon the Business Secretary to require the Met Office to publish, for the record, a comparison of the successive temperature forecasts they have made over the past 20 years, for what purpose, without any after the event adjustments (as witnessed in the latest hindcasts), and how they compare with the actual reported temperatures over the same period."
I imagine that this should be a relatively simple job to do - presumably it is all on the record, and it would answer the questions many have about the differences between the latest projections (including hindcasts) and earlier projections based on earlier models and the actual observed temperature outcomes.
If you've got the flu, and you've got aches and pains, then Gatorade is very helpful. I didn't mix whiskey with mine, so I can't recommend the mixture. But straight Gatorade has electrolytes which fight dehydration. And dehydration is what causes the aches and pains. The flu suppresses natural thirst, so you dehydrate.
Also wondering if anyone knows if Harribin's 'insider' is likely to be Slingo or someone close to her..... I imagine that Harribin has excellent contacts and can easily obtain such guiding commentary 'off-the-record' whenever a very senior person does not wish to be quoted by name.... I doubt he wastes much time with smaller fry when pursuing this kind of story (with intense public exposure and significance).
I guessed some time back that someone released the data on 24th Dec because all the bosses were already on holiday so didn't know about it.
Looks like this is being confirmed via the Harrabin insider!
HaroldW - like you I think science is wonderful. But no matter how wonderful science is it is simply not possible to predict that which cannot be predicted.
I would like to add that David Attenborough became a believer following a visit to the Met Office, where he delighted in telling everyone about the ability of the MO models to hindcast (now termed previous predictions) past events. Perhaps someone with contacts to the Great Man could direct his attention to this thread.
Bish, whisky is the remedy but don’t dilute it.
Interesting to look back at this climate prediction from the 1970s before the field was corrupted by Hansen, the team and other activist~scientists:
Climate Scientists who were right 30 years ago?
David Porter: very good point about Attenborough. I'd forgotten that his conversion was at the MO. I do remember my deep disappointment shortly afterwards when Simon Jenkins, then of The Times, wrote that if Attenborough now believed then he must, against all his instincts, do the same. But Climategate did for Sir Simon and he now lays into climate doomsters just like any others, just as he used to. So who knows Sir David, to pass on the link? Good question.
I must call attention to Messenger's recall of Julia Slingo's comments to the Feedback show at 4:30 (over on the BH "Rumbling On" thread). This does confirm that she is responding directly to at least some media inquiry and that she sounds "very cross" about the matter .... so it's at least plausible to think that Harrabin may be channeling her or someone close to her. Speculation, yes, but why does that aspect matter? Because we should want to know why it is regarded as such a mistake to have published the (half) decadal forecast on Dec. 24, experimental or not. Why was this act described as so 'innocent'** and 'naive' and 'bloody stupid'??
Why should the public NOT be allowed to know about this research? Whose call is it whether such research is provided to the public? Is it a state secret when there are uncertainties or new projections or model outputs that differ from previous public pronouncements? Is this a scientific organization or a PR organization? etc. So many questions, not enough answers.
==================================================================
** actually the word 'innocent' in context (although it's only a brief tweet, admittedly) seems not to be redundant with 'naive' but more the opposite of 'guilty' as in did someone in the Met Office intend to create this fuss. This will be a revealing period for assessing the amount of political and institutional pressure brought to bear on individual scientists, both explicitly and implicitly. How does one need to speak and behave to be regarded as 'innocent' and not 'guilty' by the @metoffice 'insider' type speaking to Harrabin? Is it Team Science or Team Politics or Team Slingo here....??
Jan 11, 2013 at 6:49 PM | Skiphil
"...it is regarded as such a mistake to have published the (half) decadal forecast on Dec. 24, experimental or not."
It has given the infidels the opportunity to claim they were trying to bury the bad (for them) news.
In the 1970's there was a lot of research on automatic pattern recognition - machine reading of hand-written postcodes and so on.
It was apparent that in practice, these systems performed far worse than the research had predicted.
Slowly it was realised that the problem was that the same data used for training the systems (computing optimised matrices used in the decision processes) was used in testing, to predict their performance.
The MO has embraced the fallacy of "testing on the training data" all along:
James Evans - That looks remarkably like a multi-decadal oscillation??
I suppose a LIA might spoil the plot. I hope I live long enough to at least see which way it is going!!!
Lapogus - That sounds remarkably like a multi-decadal oscillation???
I seem to remember someone in the Climategate emails saying "they will kill us if it just turns out to be just a multi-decadal oscillation".
BTW BTW, I think the Mt Pinatubo reference on the MO graph is a perfect example of how little we understand about the climate system.
Here's an experiment for you:
Find 100 people.
Show them the latest MO decadal forecast graph.
Ask each person to comment on the characteristics of the graph.
Count up how many people say something like "wow, there's a strange dip in the graph around 1991/2."
See if you can find a difference between the number of people counted, and the number zero.
How is the dip in temps in 1991/2 any different from the dip around 1954? Or the dip around 1974? Or the dip around 1984? Or the dip around 1999?
The labelling of Mt Pinatubo seems to be a strange artifact left over from previous versions of the graph. Look at the graphs shown in Paul Homewood's piece, linked to above. The old hindcast that starts in about 1985 shows a large divergence from reality after about 1990. This was blamed on Mt Pinatubo.
If Mt Pinatubo cause the 1991 dip in temps then what caused the indistinguishable dips in 1954, 1974, 1984 and 1999?
Answers on a postcard please.
Martin A:
Isn't this error everywhere in climate models? But on this occasion it's getting attention. High time.
retireddave,
"That looks remarkably like a multi-decadal oscillation??"
The bad news for the MO is that if my model is correct then they have another 18 years to spin the fact that temps aren't going up. Eighteen years. Of spin. Careers will be based on it. Can you imagine?
As Chief Scientist of the Met Office, I'm delighted to have the opportunity to respond to your questions that you, the public, have sent in around the science of climate change (...) but I think there are some clear messages, and these messages are really important, particularly as we go into the really unprecedented negotiations that must go on in Copenhagen in the next two weeks, to address the challenge that we the world face from growing emissions of carbon dioxide in particular. (...)
There's been a lot of questions about carbon dioxide and whether it's generally responsible for climate change, and so the first question would really be, how do we know that carbon dioxide is responsible for the climate change that we have seen and can we prove it?
Well we know that carbon dioxide is a very potent greenhouse gas and (...) by increasing its concentrations we know that it will have a very significant impact on the radiative balance of the planet and lead to changes in temperature. (...)
What carbon dioxide is doing now is enhancing that greenhouse gas effect. So we understand the basic physics of that. It's enhancing that greenhouse effect and leading to an increase in temperature, so we're trapping more energy into the planet because of increasing levels of carbon dioxide. How do we know that it matters? Well you don't have to really think to hard that if you've gone from before we started our industrial activities in the 19th century, carbon dioxide levels were at 280 parts per million by volume. We are now rapidly approaching 390 parts per million which means it's been a 40% increase. Most of that increase has happened in the last 50 years. And if we know that carbon dioxide is a greenhouse gas, it's hard to believe that if you increase it by 40% you're not going to do something to the temperature of the planet.
There is an infantile tone to the public face of much of what passes for climate science over recent decades. The facile alarmism, the obsession with a single factor, the absurd levels of confidence, the name-calling, the scheming, the petulance, the relentless appeals to authority and to consensus, the demands for critics to be prosecuted or demonised, the unwillingness to entertain alternative views, and of course, the bizarre respect for computer models that have proven to be worse than useless. The field is a young one, but it has passed from infancy to influence without ever going through what might be called a growing-up phase. No wonder it has been such easy game for political and financial opportunists to hunt and exploit.
"There is no computer model of world temperature and climate that can explain what has happened without greenhouse-gas induced global warming. None.
Unless we act to curb greenhouse gas emissions, continued warming is not a matter of speculation. It is inevitable. And scientists fear it will accelerate."
Chris Huhne 21 July 2011
OT but this 1959 video shows Bertrand Russel giving advice to CAGW zealots http://www.retronaut.com/2013/01/bill-jones-business-motivation-posters/
Just curious - how does this significant decadal warming to insignificant / possibly no decadal warming, "old model" to "new model", "This is still early days of course, there is still a lot more work to do", etc., etc., etc. fit in with "The science is settled" (peer reviewed!)?
All the discussion here here about hindcast, backcast, frontcast assabout cast is totally irrelevant. Discussion is not purposeful, not valuable it's useless.
As Dr Richard said , "the point is" and the point is that these model cannot predict anything, nothing, de nada, aucune chose, rien. Discussing climate models is navel contemplation in the extreme.
Yes they are great toys but really, predictive? About as predictive as Mystic Meg. Hindcast or whatever cast, it is only very approximate and will rapidly diverge after a couple of weeks.
Let's just wait and see what 2017 brings (lower temps I fancy) but in the meantime keep one eye on the thimble and the other on the pee. (deliberate)
Jan 11, 2013 at 6:49 PM Registered Commenter Skiphil
Re the Slingo reaction.
Are we to infer that the release of the new forecast was unauthorised by 'management' ?
And by further inference, that 'management' wanted the experimental forecast suppressed from public eyes?
The field is a young one, but it has passed from infancy to influence without ever going through what might be called a growing-up phase. No wonder it has been such easy game for political and financial opportunists to hunt and exploit.
Jan 11, 2013 at 8:06 PM | John Shade
Well said John. You are always the sensible comment in amongst the .....
I would call it adolescent science. All about feet stamping, sulking and demanding money to buy some more toys.
I wrote my last post before seeing this one of yours or I would not have bothered.
Just curious - how does this significant decadal warming to insignificant / possibly no decadal warming, "old model" to "new model", "This is still early days of course, there is still a lot more work to do", etc., etc., etc. fit in with "The science is settled" (peer reviewed!)?
Jan 11, 2013 at 9:10 PM | BruceA
New model, more to do, please send the money to MO, Exeter, England. See easy.
@Stephen Richards
While no reasonable person would believe anyone can foretell the future for decades ahead, it always amazes me how readily they accept that a silicone chip contraption somehow can. If we don’t know how to predict the future, we can’t program a computer to predict the future. Forget computers, forget science, forget math, it’s actually as simple as that.
http://thepointman.wordpress.com/2013/01/11/the-shape-of-things-to-come-snailbats-halsays-scarems-lewpapers-and-dickpols/
Pointman
Rationalisation?
Jan 11, 2013 at 9:56 AM | steveta
You're welcome!
Correct, the exact reasons for the different are still being looked at - since it's a completely new model, there's a lot to do here. It was indeed limited to 5 years because of the much greater computational expense of HadGEM3 compared to HadCM3.
Richard Betts
"computational expense"
Please elaborate, is this strictly monetary or does it hog too much time stopping other operations?
I take it is your own in house hardware?
The story has reached the antipodes (via the all seeing eye of the GWPF), and for the first time we see the two forecast graphs superimposed clearly.
http://www.thegwpf.org/climate-revisions-validate-sceptics/
Jan 11, 2013 at 1:16 PM | Jeremy Poynton
In my opinion, it's not so much about whether there is an optimal global mean temperature, but more about the actual change in this (and more importantly, changes in regional climates, especially rainfall, and changes in sea level). If the world was already, say, 4 degrees C warmer than it is now, and it had been at that state for centuries or millennia, then human society would probably have simply grown up adapted to those conditions. We wouldn't have cities in many of the coastal locations where they are at the moment, because some of those locations would be under water. However, we wouldn't think that was a problem because it had always been that way. However, since we do already have cities in low-lying coastal areas which would be flooded in a warmer world, a move to such a warmer world presents us with a problem - either we try to stop the warming happening, or we live with the consequences and spend money on protect or relocating vulnerable cities.
Changes in global mean temperature are merely one indicator of a changing climate. As you say, there is nothing inherently special about any particular global mean temperature (provided it stays within the bounds suitable for life, i.e.: not literally freezing or boiling!) It's just that changes in the mean indicate that changes that do have an impact will probably follow - like sea level change for example. That's my view, anyway.
Hi Green Sand
The latter - taking up computing resources needed for other things (like the daily weather forecast). Yes, it's in-house supercomputers.
H/t Alex Cull for link to the Julia Slingo comments to the BBC:
https://sites.google.com/site/mytranscriptbox/home/20130111_fb
I look forward to seeing how her regret that 'appropriate messaging' is needed will be explained.
Richard Betts,
Many thanks for your contributions, if you get chance, after solving all the other 200 or so issues raised on this thread, could you please throw some light on the following?
"Issues with anomaly assimilation in DePreSys:"
"Case study of the North Atlantic rapid warming in the mid 1990s"
http://www.met.reading.ac.uk/~swr06jir/presentations/JIR_Utrecht_poster2.pdf
Are these the issues that HadGEM3 is designed to resolve?
Edit:- "designed to resolve" is wrong but you will get my drift, I hope?
Invest in a turkey its a shame to only half cook it.
Richard Betts:
"Jan 11, 2013 at 9:56 AM | steveta
'I think Richard Betts has fully explained the misunderstanding regarding the white lines - it appears to be simply a misreading of a couple of words, no malice or jiggery-pokery, but perhaps sloppy language, from the Met Office. Thanks are due to Richard for taking the time to clear it up.'
You're welcome!"
I think we need a new term. Cherry feeder.
Jan 11, 2013 at 12:34 PM | ThinkingScientist
You're quite correct. The model is purely deterministic, so re-running the model does indeed give exactly the same output every time, provided that the initialisation data and parameters are exactly the same down to the bit level. Even slight differences result in a different outcome - the famous "butterfly effect". This is a very important fact when testing updates to the model - if changes have been made that are not intended to affect the scientific calculations but merely for some operational reason (like making the code more efficient) then we do "bit-comparison" tests - the new version must give exactly the same result as the old version, down to the bit level, and if it doesn't then there has clearly been an unintended change. Of course this doesn't work with updates to the science, as this is by definition expected to affect the outcome, but nevertheless it's very useful for checking non-science updates.
However, if there is a long-term external forcing (solar or GHG changes) then the simulated long-term (multi-decadal) trend in response to this is roughly the same irrespective of the initial conditions. The outcome is not the same in detail, but is similar in terms of long-term averages and trends.
Jan 11, 2013 at 1:08 PM | Geckko
The red shadings show the spread of results from simulations with different initial conditions.
Pharos,
"The story has reached the antipodes (via the all seeing eye of the GWPF), and for the first time we see the two forecast graphs superimposed clearly."
Thanks. You can see there that the two forecasts only differ by a fifth as measured from a completely arbitrary baseline. Personally, I like the number 75/123. There must be a baseline that gives that as the fractional difference. What's your favourite fraction?
Skiphil
I think the "naive and bloody stupid" remark by Harrabin's "insider" (not me, in case you were wondering!) refers to the lack of awareness by the scientists involved that this would be a result of extremely high interest to the media and general public, and that hence they should tell the press office about it before putting it on the web.
There was definitely nothing malicious or underhand here. The decadal forecasts are produced every December as one of the deliverables in our climate research programme. The deliverables are always due at the end of the month, and this one was due at the end of December, and I imagine that my colleagues were going to be away over the Christmas holiday period so delivered it before they went away - the last working day being Christmas Eve.
The guys who work on the decadal forecast are definitely honest, objective scientists who are working hard in pushing the boundaries of science by trying to make forecasts of natural variability over the next few years. Personally I think they are quite brave to do this in public, as it's all very new and (as we saw) doesn't always work - but the new model is great improvement. They could quite easily do this all behind closed doors so as to never get criticised, but that wouldn't be good for science. In my view they are to be applauded for being so open and putting their results out on the web as soon as they are produced! Since they don't actually work on anthropogenic climate change (since that is only relevant over longer periods) it probably never occurred to them that a change in the five-year forecast would be misunderstood as having implications for long-term anthropogenic warming. Their only mistake was to forget that not everyone understands that 5 years is not particularly relevant to the long-term warming, and hence not realise that a forecast of reduced 5-year warming would be of great interest. Had they realised that, they could have told the press office so that an announcement could be made, in order to avoid any misconceptions about deliberately "burying" this. But I guess you live and learn!
Thanks for all the comments above. As I feared, rather too many to respond to them all individually, but I hope I've addressed most of the key ones. Sorry if I've missed any important ones, but it's time I called it a night now. Also, thanks for staying polite!
'night all!
Cheers
Richard
James Evans
Looks more like the difference between what you were told you would get for your annuity, compared to what you will actually get.
"...either we try to stop the warming happening,..." Jan 11, 2013 at 10:58 PM | Richard Betts
Please Richard, how do you propose we do that, (if it were ever possible)?
Richard Betts: 'night all!
Perchance to dream that the decadal forecast guys have a handle on climate sensitivity.
Richard,
"Personally I think they are quite brave to do this in public, as it's all very new and (as we saw) doesn't always work - but the new model is great improvement..........."
That's great. However, it reminds me of that moment from Seinfeld - "anyone can just take a reservation":
http://www.youtube.com/watch?v=A7uvttu8ct0
Anyone can just *make* a prediction. The point is to *hold* to the prediction. To stand by the prediction. To measure yourself against the prediction. Not much point if all you do is wait five minutes and then just make another prediction.
Meanwhile, the public and the politicians seem to think that the Met Office understands climate change. And I'm paying for that right now. I'm not paying for it metaphorically. I'm literally having to pay for it. And I'm damn poor.
TJan 11, 2013 at 11:41 PM Richard Betts
Good time to call it a day, I have no doubt that Skiphil will reply but in the meanwhile:-
I do not understand the "December deliverable". The 2005/7 to 2011 Decadal Forecasts ran from Sept to August each year, on a rolling monthly basis, so why now this year, does it become a December deliverable?
I try very hard to understand what the MO is trying to do, but with your present day outfit it is as my old man used to say "like platting fog". Communication and clarification has become a total anathema in virtually anything it produces. It displays all the symptoms of a rudderless ship.
Whilst it may be seen as a simple internal communication problem to you, I can assure you that on the outside it is considered to be a total shambles by both sides of the indivisible divide.
There is a problem Richard and it is not, repeat not, that others cannot understand your message, it is simply that in its present format your message is incomprehensible.
'night all!
Cheers
Green Sand
Hi, Richard Betts, thank you as always for your comments and feedback here. I guess the 'naive and bloody stupid' phrase seemed rather strong so I thought it hinted at something more than neglecting to tell the press office or write up a press release, as in 'why did they allow this to be public at all'?
Anyway, I'm sure few are still awake on that side of the pond so I will sign off, thanks to all, and glad to know the kraken is stirring (be well, Andrew, and all)......
Richard it is relevant to climate chnage..
ie it would apear to falsify Vicky Pope highly publicized pre Copenhagen predictions for the next 5 yrs (from them)
ie instead of those yrs being warmer, we now have acknowledgement of a longer slowdown.. I'm sure you are aware of the implications of a possible 20 yr pause to many model 'projections.. and to the politics.
Hi Richard,
Thanks for answering that question for me, much appreciated. Any possibility you could address the other 5 labelled (a) - (e) that are further up the thread?
Thanks in advance!
remeber highly public warming warning for next few yrs (Dec 200() the issue is that the Public is not aware that these were on thge warm side, and temps have 'paused----
http://www.bbc.co.uk/blogs/paulhudson/2009/12/global-temperature-predictions.shtml
The Majority - More warming:
1) Met Office Hadley Centre. At least half of the years between 2010 and 2015 will be hotter than the previous hottest year on record (set in 1998).
A more recent press release issued during the Copenhagen climate conference said next year is 'very likely' to be the hottest year on record.
This is part of a new forecast for 2009-2019 which shows a range of possible outcomes for each year but the "central estimate" for 2010 shows a figure of about 0.55C above the average - clearly in excess of 1998. The temperatures then steadily climb to a high of about 0.7C above the average by 2015.
2) NASA: Their most recent predictions (early 2009) have been used in the absence of anything new. But effectively it's the same as the Met office, in that a new global record is expected next year.
3) Lean and Rind: research funded by NASA: Global surface temperatures to increase by 0.15C +- 0.03C from 2009 to 2013 inc.
Richard
'then human society would probably have simply grown up adapted to those conditions. We wouldn't have cities in many of the coastal locations where they are at the moment, because some of those locations would be under water. '
Surely this is what mankind has always done, there are complex remains of civilisation off the coast of many locations around the world. Why does this generation now have the belief that man can control nature on such a scale?
I agree that to study and predict climate optimises our opportunity to adapt but to infer that changes in climate can be attributed to and therefore modified by man is approaching a God complex. Locally we can affect the climate with cities and land change, but globally with around 80% of the globe unpopulated!
Mmmm I wouldn’t be too impressed by the “information” gleaned from the MO insider since it would only seem to fit the MO’s MO of exculpatory hindsight projection ;)
I don't know if this has been mentioned but another curious thing I notice about the new model projection is that it looks far more amazingly accurate starting from further back in time than the previous one, but I notice it appears to be made up of different start points joined together. The caption says
Could Richard Betts answer if this is the case and how these start point were decided upon and if the ellipsis represents many other joins in between?