Monday
Aug052013
by 
Bishop Hill
Bishop Hill Akasofu's model
Aug 5, 2013 Climate: Models In recent months there has been a lot of attention devoted to the failure of the GCMs to predict the pause/standstill/hiatus in temperature rise seen for the last decade or more. Mike Kelly points me to this recent paper in the open access journal Climate. It's by the prominent sceptic Syun-Ichi Akasofu, whose naive model of global temperature change gets attention every few years.
The model essentially superimposes a multidecadal oscillation on an upward linear trend representing the recovery from the little ice age, and it gets attention because it performs so well. In the latest paper, the model is compared to observations for the period to 2012.
;)
Good eh?
Reader Comments (81)
And AMO is a misnomer - it's a global multidecadal oscillation (GMO).
http://www.woodfortrees.org/plot/hadcrut4gl/from:1880/detrend:0.762/plot/hadcrut4gl/trend/detrend:0.762/plot/esrl-amo/plot/esrl-amo/trend
Well it's not that dissimilar to the Swanson/Tsonis model which IS based on physics. We all know there is a cyclical fluctuation short term, so why not a cyclical fluctuation long term as well? Recovery from the little ice age (LIA) is as good an explanation as any because nobody has any valid theory about why the LIA happened other than somehow by the suns cycles. Swanson/Tsonis did not specify if the underlying linear rise was from manmade CO2 or a natural recovery from LIA. They seemed to impute that the idea that it may be manmade is enough for panic. Well to keep getting funded they would say that wouldn't they? But a little warming is beneficial according to everything that has ever been written about historical warm periods and even according to the IPCC; who came up with the 2 degree number (based on no physics whatsoever) as the cut-off between beneficial/benign and potentially harmful.
Aug 6, 2013 at 1:18 AM Richard Betts
They are very sour, these purely empirical models.
Well, if it wasn’t for the poisoning of climate studies by such as the IPCC leadership and uncounted numbers of agitators and sundry opportunists, we would have an innocent and delightful topic for discussion here – the use of very simple models, and how and why they might be more successful than the monster GCMs at least over the short timescales we have for checking them all out. Alas the innocence is lost now that the issue has been so dramatically politicised by people treating climate models with far more respect than I think they deserve.
Simple models have been used recently only for global mean temperatures, as far as I know, but there seems to be no reason why they could not be tried on cloud cover, precipitation, etc etc wherever half-decent data exists. It would also be interesting to see how they might handle regional and single-station records.
I have now learned of more such models thanks to this thread. New ones for me are:
(1) Page:
He notes “In the last few months there have been numerous discussions on the WUWT site and amongst establishment scientists questioning the validity of climate models as a source of useful predictions about future temperature trends. Notably, the UK Met Office has reported on "The Recent Pause in Global Warming" for which they have no good explanation. The fact is that, as will be discussed later, their models are incorrectly structured and the modelling approach is inherently useless for making predictions. A much better approach is to recognise and project forward quasi-cyclic quasi-repetitive patterns in the temperature, oceanic system and solar driver data as was done in the 30 year forecast reviewed here.”
Simple modeling maybe, but big conclusions: “I concluded ,as might any person of reasonable common sense and average intelligence given these simple observations that solar activity and our orbital relations to the sun were the main climate drivers. More specific temperature drivers were the number of hours of sunshine, the amount of cloud cover, the humidity and the height of the sun in the sky at midday and at Midsummer . It seemed that the present day was likely not much or very little outside the range of climate variability for the last 2000 years and that no government action or policy was required or would be useful with regard to postulated anthropogenic CO2 driven climate change.”
Thanks: Aug 5, 2013 at 4:50 PM | Norman Page
(2)Ziskin and Shaviv: As you would expect from this source, this is sharp and penetrating stuff and looks to this humble observer as well worth a lot of further study – which is to say I for one would need a lot more time and effort to have any chance of following it! The authors note right at the end “Furthermore, we show in the present work that a simple energy balance model can shed significant light on the understanding and quantification of the climate system, and in particular, that such models can improve our understanding of the solar–climate link”
Thanks: Aug 5, 2013 at 8:23 PM | Noblesse Oblige
(3) Evans: This one looks like a time-series method fitted to recent data and projected ahead.
Thanks: Aug 5, 2013 at 8:31 PM | James Evans
(4) DocMartyn:
Where he finds and subtracts cycles, looks at residuals, and makes projections. In a comment on that post, Scarfetta claims that he did all that earlier and notes “The 60-year cycle, which is quite evident from 1850 to 2013, is just one of the several oscillations that characterize the climate system and in my papers I am talking about several other oscillations in addition to the 60-year cycle.
Papers may be downloaded from my web-site where at the bottom I am updating my forecast every month an demonstrating that far outperforms any climate model used by the IPCC for accuracy.” Scarfetta (scroll down to the 6th figure, unfortunately they are not numbered, to see his latest comparison).”
Thanks: Aug 5, 2013 at 11:48 PM | DocMartyn
Finally, here are some simple methods from the 1920 to 1940 era that I think deserve some attention . They are in the book ‘Physics of the Air’ by W.J.Humphreys first published in 1920 but with many editions since.
He describes a very simple computation (model) to estimate the impact on global temperature of doubling or halving ambient CO2 levels. In each case, he finds the effect would be ‘no more than about 1.3 C’ (page 622, of my 1940 edition). This fits in with the broad-brush conclusion drawn earlier in the book than the carbon dioxide theory of ice-ages is not a very convincing one. Referring to several experimental studies in laboratories with columns of CO2, he writes
So, science was ‘unsettled’ back then as well!
Thanks: Aug 6, 2013 at 1:18 AM | Richard Betts (welcome back! - your comment mentioned Callendar and the McIntyre post, and that reminded me of Humphreys)
@ topper
the "standstill" in global temperature change will soon turn into a marked fall in temperature which will extend until about 2030....politicians around the world who are presently convinced by "warmist" will...recognize that they have been sold a pup. It should be the end of the global warming industry.
I'm not so sure. A major consideration in the way climate psyence data is organised is the anticipated length of a psyentist's career in the field.
Your typical aspiring climate psyentist wanting to go to UEA gets 3 Bs in Geography, Biology and Politics A Levels aged 18. He takes a year off backpacking, and goes to UEA aged 19, graduating with his 2:1 in Climate Psyence aged 22. He takes another year off, then gets a research job making stuff up for the grey literature relied on by IPCC assessment reports. He gets his MSC aged about 27 and his PhD aged about 30.
His PhD will contain the laughable nonsense that sets him up for a career in the public sector. Once the pension is banked, it can all fall apart without consequences. Ergo, whatever he puts in it needs to survive for about 30 years, but nobody following behind him need defend it further.
The most significant cycle in climate psyence is thus this ~30-year personal career cycle. It is a useful period in other ways too. There are only about 150 years of data, so by splitting it into five cycles, climate psyentists can kid themselves there's adequate data to analyse trends (if there were 50 years of data, they would certainly argue that there are 10-year cycles instead with a 50-year supercycle, or make something else up).
Thus, sometime after these morally incompetent ethical midgets retire, we can expect to see the first pieces debunking their work making it through pal review.
These pieces will be published long after a shivering world has noticed that CAGW hasn't been happening, of course, but science will thereby retain its reputation for taking lots of time and money to state and describe the bleedin' obvious.
Meanwhile, the viciousness we see from the likes of Mann et al is because they are in the position of Guido the pimp in Risky Business. Joel has noticed how easy and lucrative it can be to become an amateur pimp, but is memorably warned off by Guido: "In times of economic uncertainty, never ever f*** with another man's livelihood." McIntyre et al did just that by debunking the Hockey Stick right away rather than politely waiting 30 years.
The circular reasoning of cagw trumps a mild extrapolation of observations? 97% of people with a dog in the fight believe that!
"The Livingston and Penn Solar data indicate that a faster drop to the Maunder Minimum Little Ice Age temperatures might even be on the horizon.If either of these actually occur there would be a much more rapid and economically disruptive cooling than that forecast above which may turn out to be a best case scenario"
Aug 5, 2013 at 4:50 PM | Norman Page"
Interestingly enough Leif Svalgaard is also "speculating" that we may be moving towards a Maunder Minimum in a comment on WUWT "The sun is about to have a flipping magnetic field reversal"
"I have reasons to believe that we may have another Maunder-type minimum, but that is speculation, of course, hence the may. You can see some of my reasons for this here: http://www.leif.org/research/SSN/Svalgaard12.pdf" He is being somewhat circumspect, but I believe that Leif Svalgaard would not have gone on record with such a statement unless he believed it was a distinct possibility. Not sure how the GCM followers would cope with that if such a slump in temperatures materialised.
So, like the man said, the peak-to-peak is measured over 30 years (unless you have just invented a new meaning for the term "peak-to-peak").
Martin A --
Entropic Man used "peak-to-peak" to compare the crest of one cycle to that of the following cycle, possibly unaware of the engineering term "peak-to-peak amplitude" describing the range of an oscillation (for which EM would presumably use "peak-to-trough").
Lord B is right.
Everyone jumps to criticise the simple model, yet it gives a much better data fit than the hugely expensive models that require £30m computers. So what does it mean?
The Akasofu model fits some data but may have no connection with reality, so it will probably not have predictive skills. The expensive climate models do not match the data at all and therefore do not match reality and have no predictive skills.
@ thinkingscientist
Akasufo's model is simply empirical and it is therefore easy to poke holes by saying it has no physical significance, but that doesn't mean his trend + periodic behaviour is wrong, just unexplained.
Yep - analogously, people were pretty good at predicting low and high tides based on the moon cycle long before the theory of gravity came along to explain it. It wasn't necessary to know why it worked to know that it worked.
AIUI you can get achieve pretty accurate weather forecasting by assuming that that tomorrow will be like yesterday. Not a single Met Office salary and final salary pension required, either.
I believe that Lord Beaverbrook gets the closest to the nub of the questions about models. Models are useful in trying to "understand" what has happened in the past but they have absolutely no reliability when it comes to predicting the future. One day that will happen but it is a long way down the research road from here.
I apologise to Richard Betts but the fault lies with the MO. Many many years ago I saw a representative of the MO on a local news program, he was entering a room containing the then "super advanced computers" being used for modelling. As the man entered the room he turned to the camera and warned us not to base to much on the results of the models, after all he said, they only predict what they are programmed to predict. How refreshing is that and how different from today?
The whole problem is the importance placed on the models by the MO which is then used by politicians.
What is needed is someone like my guy from the MO to tell the government that the same is true today, how about you Richard?
@Richard Betts.
A couple of points.
Firstly, it is the GCMs that overfit ("flux adjustments" anyone?) . Exactly how many paramters is Akasufo using?? More or less than your average GCM?
Secondly, It doesn't matter a jot if there is "no physics" in the model. Any model that presents itself as a representation of the real world physics need to outperform this model in order for it not to be cylindrically filed. Something climate modellers are loathe to do given the time an money sunk into their failing concoctions.
Thirdly - and I am surprise a climate scientist didn't pick up on this one. If a naive model is outperforming complex highly fitted model with low degrees of freedom, isn't there an extremely attractive research opportunity trying to work out if the naive model is indeed picking up an unknown factor and if so what the hell is it?
Instead we have so-called scientists looking at a model that considerably outperofrms their physical theory and simply waving their hands dismissively while wittering "but there is no physics in it". No sh!t Sherlock - but it has given some strong indication that the physics that is in your model is wrong or at the very least dangerously incomplete.
Am I missing something?
Isn't the physics related to the Sun, the Maunder Minimum and the increase in solar activity since then?
Swiss got me to thinking,
If the sun is involved in what the temperature of the earth is, wouldn't correct models of the sun need to be incorporated into projective earth climate models?
Andrew
If the GCM's are based on physics, how come their output is all over the shop? Either only one is right or they are all wrong.
At Least no physics was harmed in the construction of the Akasofu model.
It wasn't my intention to criticize this model per se, what I have doubts about is ready acceptance of this one because it says something one might like. That other set of models, the GCMs, have problems of their own, they don't reproduce anything particularly well and they (their proponents) don't even chuck out the bad ones, presumably for political reasons. Nevertheless, the only hope of ever getting this right is some sort of GCM. The ones we have now are only good as stepping stones getting us to a decent stab at a model. But that is not going to happen as long as they are politically constrained.
I'll repeat my null. Nothing much is happening. Nothing new or special is happening. Whatever does happen is best dealt with by adaptation. And everyone who says they are certain about anything in climate is a liar or daft.
Oh, and when the ice age returns, the Met will be the last to know or acknowledge it.
Perhaps the GCMs could be improved by removing some of the variables. First get rid of the positive water vapour feedback, then remove the GHG effect. Yep, that does it.....
Wow. I'm listed as a model:
John Shade:
"(3) Evans: This one looks like a time-series method fitted to recent data and projected ahead."
I'm not a rich man. In fact I'm not even a poor man. I work for the kind of money that most of you wouldn't even get out of bed for.
But I'm prepared to say this:
I bet my model (based on cut-and-paste technology) out-performs the Met Office model. I'll bet 1,000 pounds that my model does better than the MO model over the next 10 years. Any takers?
I've tried similar bets over the last couple of years, and strangely I've had no interest.
If you're up against spineless twats, there's not much more you can do.
@Geckko "are loathe"
Fraid not, "loth" = reluctant, "loathe" = abominate. Choose your words more carefully Geckko or you might be misunderstood.
Simon -
Here in the States, at least, "loath" is more common than "loth". Which unfortunately makes it easier to mistake it for "loathe".
"The sun is about to have a flipping magnetic field reversal"
Aug 6, 2013 at 12:58 PM | John Peter
Rlax. It does that every 11 years.
Martin A
We may well be using the term "peak-to-peak" in different senses. I am using it in a colloquial sense to refer to the period from one temperature maximum to the next, about 60 years.
Would you be using it in the sense that a peak is a maximum distance from the datum in either direction? In that case a maximum peak and the subsequent minimum peak would be 30 years apart.
"Instead we have so-called scientists looking at a model that considerably outperofrms their physical theory and simply waving their hands dismissively while wittering "but there is no physics in it". No sh!t Sherlock - but it has given some strong indication that the physics that is in your model is wrong or at the very least dangerously incomplete."
Aug 6, 2013 at 4:00 PM | Unregistered CommenterGeckko
We cannot let them get away with the claim that there is physics in their model. More specifically, their suggestion that their model contains a complete physical theory is nothing but a suggestion. They have never shown us the physical hypotheses which describe the behavior of clouds or of water vapor in an environment of rising CO2. An incomplete physical theory cannot be a good physical theory. A physical theory that is incomplete on just the crucial points, clouds and water vapor, is useless at present and holds no hope for the future.
When they say that they have physics in their model, they mean two things. The first is that they have created some elaborate differential equations which are expected to someday simulate the behavior of clouds and water vapor even without a physical theory of the same. In other words, construct and revise enough models until the incremental process leads to perfection. That is a fool's errand.
The second is the detestable presumption that they have identified all natural processes that can contribute to climate change. They believe that they know what all the causes are, what their effects are, and they need only enlighten us to the fact that CO2 causes all other climate changes. They assert this knowing full well that they have no account of natural variability and have no idea what the earth is capable of on its own.
When it comes to planning your trip, which would you prefer: a bus timetable based on observations of past buses, or one starting in the engine compartment with the laws of thermodynamics? The latter would quickly become very large and intricate, but it would remain utterly out of its depth despite looking very sophisticated.
Would you be using it in the sense that a peak is a maximum distance from the datum in either direction? In that case a maximum peak and the subsequent minimum peak would be 30 years apart.
Aug 6, 2013 at 11:54 PM Entropic Man
Yes, that's the conventional use of the phrase and I assumed that was what you meant. Another commenter evidently assumed that you were using the phrase in the conventional sense too.
Martin A
I'm a biologist talking to two electonics engineers. In my convention a cycle is a complete positive and negative oscillation. Consider the wavelength of light as an analogy. Your convention describes what I would regard as a half wavelength.
Bart
The 130 years of the instrument record is a composite of three signals. These are the 11 year solar cycle, the 60 year cycle and the long term trend.
The solar cycle is not relevant to Akasofu's paper.
Our processing of the data separated the two remaining signals . Each graph displayed one of them.
Your graph detrended the data, which removed the long term signal and left the 60 year cycles. Not surprisingly, the two iterations of the cycle looked the same.
http://www.woodfortrees.org/plot/hadcrut4gl/from:1900/detrend:0.75
My graph generated linear regressions for each of the two wavelengths. This removed the 60 year cycles and left the long term accelerating trend.
http://www.woodfortrees.org/plot/hadcrut4gl/from:1880/to:1940/trend/plot/hadcrut4gl/from:1940/to:2000/trend
We are both right.
Your convention describes what I would regard as a half wavelength.
Aug 7, 2013 at 1:53 PM Entropic Man
EM - no we are at cross purposes.
The term 'peak-to-peak' is normally understood as the difference between the most positive value and the most negative value of a periodic waveform. So it's (in the case of a voltage signal) a voltage, not a time.
In the case of a sinusoid, the positive and negative peaks are obviously a half-cycle apart.
Martin A
Got it.I'll avoid using "peak-to-peak" in this context.
What would you suggest as an appropriate term for the interval between two positive peaks in the 60 year temperature cycle we are discussing?
"The period"? [a standard term for the duration of one cycle of a periodic function.]
"the time between adjacent maxima"? [longer but unlikely to be misunderstood and applies even if not periodic]