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)
Good spot Paul.
Has Hansen been transferred ?
Heads should roll at the Met Office.
you must be reaaly ill. we'd spotted that a few days ago in the comments ;-) ;-)
recalculated model 'output' is not the same as a 'forecast'
unless they've redifined the word 'forecast' or 'prediction/projection.
"I'm still suffering. Even whisky isn't working. It must be serious."
If you can type a cogent posting, the dosage is obviously inadequate.
Get well soon.
Most of us consider a forecast to be something that is published before the event. We need to ask the MO when these previous forecast were published (and whether there were any other forecasts published at the same time).
It seems when the MO decide to build a new model they probably start an iterative process of tweaking the inputs and running the model against historic data in order to measure how well they can track historical temperatures. They use the results of this process to fine-tune their inputs and repeat the whole process. So far so good ...
After several hundred attempts they have fine-tuned their inputs sufficiently to be well-pleased with themselves. But then they confuse everyone - including themselves no doubt (I'm being charitable here) - by referring to the output of this exercise as their previous forecast!
No doubt Richard Best will ask someone to issue an apology...
It sounds as though they ran their model again with tweaks to match the known reality in order to look less stupid.
@ matthu
"We need to ask the MO when these previous forecast were published....."
We need to ask the MO to show examples of their published previous 'forecast(s)'.
If you read the text in the decadal forecast it explains that the new one is a "retospective forecast":
"During 2012 our decadal prediction system was upgraded to use the latest version of our coupled climate model. The forecasts and retrospective forecasts shown here have been updated to reflect this change."
In other words they did a new one to match the most recent historical data.
It could be that the model run they're showing now was one they had prepared earlier but had thrown aside at the time and have now dug out again. Either way the whole forecast/projection/prediction capability of climate science is a patent mess.
It seems ever more obvious there is nothing useful going on – just pick and mix as and when politics or reputation polishing demands.
I'm guessing that as this is a new model and once again starts with its initialisation, it then provides a different scenario as it has changed something(s) in its parameters. Indeed, I'd be surprised if it didn't hindcast as well as it did, because I think it's meant to.
There would be two good aspects of this (if I'm anywhere near the mark - doubtful, frankly) - the recent flatlining is now almost produced by the model, and as a result, it's now nowhere near as scary looking. I wonder what was changed in the forcings, oceans, solar etc to achieve this?
Climate science has become corrupt with political agendas, lobbyists, misinformation, fake results and dodgy presentation. The initial headlines form the public impression then the MSM lose interest.
Publishing retrospectively adjusted model predictions may have limited scientific interest but I doubt if that was the main motivation for the Met Office when they published their press release.
It could be that the model run they're showing now was one they had prepared earlier but had thrown aside at the time and have now dug out again. Either way the whole forecast/projection/prediction capability of climate science is a patent mess
I don't think so. Their model (they only have one I think) has too much CO² warming in it. We know this from past 'projections' and Paul Hudsons note (9/10 projections are too high). I think they tweeked the inputs to the required answer. As our more able colleagues here penetrate the mire of this output I'm sure other peculiarities will be observed.
Sadly Richard Betts is unlikely to be back for a while but never mind.
Is Dr Betts seeing a therapist ? The cognitive dissonance must be extreme by now ... in fact, it may be worse than we thought.
[BH adds: Please try not to antagonise other commenters]
Bishop, I sympathise...but I think I understand the problem..."Even WHISKY" isn't working.
As I see it, you can't spell WhiskEy properly north of the border.
So it's not the REAL stuff and therefore it has no effect.
Try Bushmills instead!
It comes from the oldest distillery in the world (as I understand it) so it must be the original recipe.
Everything else is merely a copy.
PW
P.s. anyway, whatever you are imbibing, I wish you a speedy recovery.
The white lines show hindcasts, ie: model simulations started from older initial conditions and then run onwards, and compared with the observations to see how well the model does. The point here is that the hindcasts with the new model (HadGEM3) compare better with the observations than the old model (HadCM3) and so this gives us more confidence in the new model.
These decadal forecasts use "initialised forecasting" techniques, ie: the models are started at the observed state for the current time - as distinct from the long-term climate projections that start back in pre-industrial times, run through the 20th Century and then on into the 21st Century, meaning that they can't be expected to capture the exact year-by-year variations that the initialised forecasts are attempting to capture. Because the initialised forecasts are started off at, say, the right place in an ENSO cycle, they potentially can capture the natural variations arising from ENSO and other modes of variability. This is still early days of course, there is still a lot more work to do, but you can see from the 2012 figure that the hindcasts show the model agreeing with the observations reasonably well (and better than the HadCM3 hindcasts as shown in the 2011 figure).
The first time that these initialised forecasting techniques were used for decadal forecasting was this paper published in 2007. So this was the first time there was actually a proper forecast looking forward in time - anything before then is a hindcast. This is the case for all versions of the decadal forecast that you might find.
Dear Bishop,
Forget the whisky. Brew yourself a drink of honey, lemon, ginger and garlic in roughly equal parts with a glassful of hot water after boiling. This will strengthen your immune system which will rescue you quickly/
The MO claim that the new forecast (plus hindcast) is the result of a "new model" and that they published it as soon as "the computer" had finished running the new model.
Just what exactly can be new? They do not know anything new about how the climate works, there is no new scientific theory even so what is new?
As others have already said; just the tweaks are new to make it look better.
Who exactly is being fooled by this, ah yes sorry it is of course the politicians.
Richard you are indeed a brave man.
However good your new model is you should not have published hindcasts as if you always predicted the new temperature graph, that is totally dishonest. The old predictions just disappeared?
New Model? No CO2-AGW?
AlecM
Get up to date on Rhoda's GHE demo discussion mate ^.^
Richard Betts
Thanks Richard, please, please get the descriptions changed. We have been round this too many times. It must be possible to light upon a simple description that does not make people immediately think that they are looking at forecasts.
Thanks Richard, so I was not too far away, surprisingly. Apart from resetting the initialisation, does the new model continue to use the same levels of the various natural and manmade forcings, or have they been tweaked to produce a better fit?
Bish
I do not like Whiskey but Bushmills is something else :) Please get wells soon ^.^
As a Scot, I don't approve of Bushmills, but I'll buy a bottle to celebrate when the MO management are sacked.
Richard,
"The point here is that the hindcasts with the new model (HadGEM3) compare better with the observations than the old model (HadCM3) and so this gives us more confidence in the new model."
Is that really the point here? I know it's your point, but I would say there are more important points. Like not calling hindcasts "previous predictions", while simultaneously deleting your real previous predictions.
You mentioned before that the mislabeling of hindcasts was to avoid using jargon. Maybe it's better to use jargon, if the alternative is to mislead.
"These decadal forecasts use 'initialised forecasting' techniques [.....................] can capture the natural variations arising from ENSO and other modes of variability."
In other words, short term predictions are easier than long term ones. This isn't a sciency concept. Try guessing where Fulham will be in the league in a couple of weeks time. Then try guessing where they'll be in a few years time. It's not a concept that requires a lot of fancy words.
Dung, I for one looked at Paul Homewood's article and immediately assumed that the new model was not trying to 'hide' the previous predictions, but instead showed how it was better at predicting the past than HADcm3, resulting in the downward adjustment of future temps. I think it's a good thing, because the result has been all the hoohah which has brought the plateauing temperatures into the much wider MSM.
I don't think castigating the MetO for running a more realistic model is ideal. By all means, the behaviour of the senior politicos there has been an entirely different matter, as you'd expect, but the scientific method (if it's not working, adjust the model to better reflect reality) is at least poking its head above the parapet.
Hi Dung, the original predictions are published in the literature for all to see, in the paper I linked to above.
Bish, I hope it's single malt you're consuming, or it won't work. I always take some zinc tablets too for any cold / flu / sore throat stuff. Beefs up the immune system and works wonders. Get well soon.
Sorry about that.
Hope his lordship gets well soon.
Last night 10/1/13 I awoke and switched wireless on.
Have a listen to BBC 5 live i player from 0215 to 0225'ish.
Some listener was browbeaten by some Aussie warmist about the lastest Met Office embarrasment
Listen and cry/laugh.
Dung & Schrodinger's Cat on Bushmills...I will meet you 2/3rd of the way and share the bottle with you both by appointment. (and with my wife's permission!)
PW
The narrative on the graph clearly states
Previous predictions starting from June 1960, 1965, ..., 2005 are shown as white curves,
No mention of hindcasts. If indeed they are hindcasts, the description should be changed.
More importantly, a separate graph should be issued that compares the original predictions with actuals.
Without this, the graph is highly misleading. (Or is that the intention?)
Richard Betts -
Two quick questions, if I may.
1) Do you know why the recent MO update using the HadGEM3 is only for a five-year period rather than ten?
2) HadGEM3 is not listed on KNMI Explorer among the CMIP5 models. Do you know what it projects for those scenarios, or how it compares with (say) HadGEM2?
2013. If the documents liberated via Donna have serious flaws and get shredded by free-thinking subject-matter experts. If the AGU auto-dissolves in a pool of shame. If the Met Office Climate Initiative continues to make a complete ARse of itself. If the internet survives a bit longer. If the membership of the Royal Society gathers feathers, and warms up some tar for their next AGM. If the Labour Party and its print wing in the Guardian and broadcasting wing in the BBC spot some less vexatious bandwagon to ride upon for their ambitions than climate change. If the word 'sustainable', and all the silly baggage it carries, becomes everywhere reviled and banished to some dusty darkness. If the odious IPCC simply fades away with a whimper. If the Bish finds the right malt (note, no 'e' to be seen). Why then, 2013 could be a right corker of a year.
Hi HaroldW
1) Because HadGEM3 is much more computationally expensive to run than HadCM3, being higher resolution and having much more detailed equations, so the hindcasts to check the model's performance were only done for 5 years as opposed to 10 years. Notice that the red bands (the hindcasts) are only 5 years in length in the 2012 figure, but 10 years in the 2011 figure.
2) HadGEM3 is still under ongoing development so couldn't be included in CMIP5 - models had to be "frozen" (ie: no more changes) in order to included in the comparison. So, it's not been used for those scenarios yet
@Richard Betts
I genuinely appreciate your inputs here - I read your comments with care and note that your politeness and general acuity greatly ameliorate my irritation at the "we do not debate in public" sidestep. Thank you for your efforts here
But (had to be one), the phrase "retrospective forecast" almost caused me to fall off my chair. It is a perfect example of a complete oxymoron. Do the BOM actually mean "hindcast" here, or is this too direct ?
Green Sand: thanks, I'll feed that back!
Stun: the forcings are unchanged, but the forcings don't make much difference on 5-year timescales anyway, see Figure 4 in this classic paper by Ed Hawkins.
I thought bad news got buried, not good news. Or does bad/good depend on degrees of malice?
@Richard Betts
"The white lines show hindcasts, ie: model simulations started from older initial conditions and then run onwards, and compared with the observations to see how well the model does. The point here is that the hindcasts with the new model (HadGEM3) compare better with the observations than the old model (HadCM3) and so this gives us more confidence in the new model."
Could this be rewritten as?
We started from scratch, back in 1950, fiddled with the input parameters, ran the model a few dozen times and went with the one that gave us the best fit for the known record, then ran it on for 5 years. We hope that we might just have fiddled things so that it makes a half good attempt at a forecast this time."
Or am I being too cynical?
ianl8888
Yes, "retrospective forecast" means hindcast here.
Don, if only it were that easy :-)
Richard Betts (11:03 PM) -
Thanks for the reply. Has the new model been run over a long enough time-span to establish a likely TCR or ECS?
ssat, nothing was being "buried" here! The decadal forecast is updated every December, and goes on our science pages as it's ongoing research and not intended to be a forecast for public use (it's not yet been shown to be useful to anyone, although we hope it will be when we've developed the technique further). It's not particularly relevant to global warming as it's about near-term natural variability rather than the long-term trend.
Richard,
"The point here is that the hindcasts with the new model (HadGEM3) compare better with the observations than the old model (HadCM3) and so this gives us more confidence in the new model."
P.S.
It looks to my eyes that the old hindcasts were for 10 year periods, while the new ones are for 5 year periods. Is that correct?
If you get to re-set the initial conditions every 5 years instead of every 10 years, then does the improvement really reflect an advance in understanding?
HaroldW, great questions again - but no, not to my knowledge. (I should emphasise that I am talking about my colleagues' work here, so there might be other ongoing stuff that I'm not aware of yet). As I say, it's still under development anyway, and the experimental decadal forecast is part of that development. It will be used for climate change work in due course.
James Evans
Correct, see my response to HaroldW above.
It's not a matter of "resetting" the model every few years and comparing the long-term trend - we compare the model against observations for each individual hindcast period. To check the old vs. new, just look at the first 5 years of the 10-year forecasts in the 2011 figure.
Anyway, goodnight folks, thanks for the discussion!
Cheers
Richard
I think someone found a line of code where the fudge factor should have been 3 not 4.
Richard Betts
"Green Sand: thanks, I'll feed that back!"
Richard, sorry, I appreciate your efforts, I know you are busy, but that is the third time in the last 12 months you have told me, on this particular subject, that you will "feed it back".
Either the message is not getting through or it is being willfully ignored. These are simple, easy ways to improve communication and therefore promote trust. It the MO cannot or will not take such obvious steps it can only further degrade its failing reputation.
However I note the clarification in:-
http://www.metoffice.gov.uk/research/news/decadal-forecasting
and if it is a product of your previous endeavors, thank you! Now please lets get a grip and make it clear in the actual forecasts.
Time to start gaining the attention of those concerned!
Richard,
Thanks again for the reply.
While I applaud the reserve of describing the new model as "not yet been shown to be useful to anyone," I can't help but wonder if such a phrase might also apply to the prior model which appears to be fairly consistently high in its temperature predictions at least. In the words of one of my favorite characters, "You might very well think that; I couldn't possibly comment."
Just made my first visit to Paul Homewood's website.
I noticed the sub-header on his homepage:
A discussion of that statement would be a significant post in its own right.
Applied, it would devastate the IPCC: 1) assessment processes wrt lack of any meaningful openness and transparency, 2) biasing of processes for personnel selection, 3) cherry picking of research uncritical of alarming AGW by CO2 from fossil fuels.
Well done on your website Paul Homewood.
John