Discussion > Statistical Analysis of Global Temperature (so-called) Data is Inherently Nonsense
So what? To say analyses of temperature data (e.g. those that try to infer future movements) are guesses is to state the obvious. But not all guesses are alike. Just being a guess doesn't make something worthless unless you equate the word "guess" with, say, "random". A guess is an estimate based on prior information. If there is no prior information, the guess will indeed be meaningless. The more prior information there is, the better the guess is likely to be. Did you ever assume anything different?
Raff
Daffodils
Brought up on the South Downs, I knew that towns and villages at the bottom, had Daffodils in flower a few weeks earlier. The height difference is about 500 feet, and temperature can be 1-2C different.
I would love to know whether any of the big stately homes, Kew Gardens etc maintain garden diaries going back 100 years to the first world war, 200 years back to Waterloo, or even further back, recording first snow drop, crocus, daffodil, tulip. I fully appreciate that spring bulbs are NOT a record of winter temperature, but can any patterns produce something more realistic than a Hockey Stick? Link with the Central England Temperature (CET) records?
Obviously there are other note worthy annual events to a gardener such as tree blossoms, roses, vegetables and fruit coming into season.
Or has anyone ever done it, found no correlation with the Hockey Stick, and binned it?
golf charlie
Dec 28, 2015 at 5:59 PM | Unregistered Commentergolf charlie
and Martin, there is absolutely tons of proxy data that show past climate. Obviously it is open to misuse by poor interpretation but even with that loads of it contradicts the current global warming narrative very clearly. For example just look at historical glacier extent and see that doesn't seem to fit CO2 as a driver at all.
Rob Burton
Rob Burton, thank you for that. Records of glaciers tend to go back as far as the era of photography, and I would certainly agree that they are a better guide to gradual changes, up and down in average temperatures ( whatever they are) over 5-10+ years. Photos simply dated to a year don't necessarily help, if there is a big difference between the end of the spring thaw, and the autumnal freeze up, for example in the Alps, when in the summer, there is almost no snow at all.
Hence my curiousity about hand written gardeners diaries, which may go back to pre-Victorian times?
Gilbert White (try google) was one of the first naturalists, and he wrote stuff down in the late 1700s, based on his observations in Selbourne, Hampshire, in/on the South Downs between Alton and Petersfield. Next time I am near his former home, I might just buy one of his books!
golf charlie
For example just look at historical glacier extent and see that doesn't seem to fit CO2 as a driver at all.I don't know what you are referring to but maybe you imagine that people think CO2 is the only influence on climate...
Raff
I don't know what you are referring to but maybe you imagine that people think CO2 is the only influence on climate...
Thanks. So you are a denier after all?
clipe
It makes it more interesting when you add a couple more lines. Which way will the next trend go? And, does it really mean anything, at all?
Radical Rodent
So what? To say analyses of temperature data (e.g. those that try to infer future movements) are guesses is to state the obvious. But not all guesses are alike. Just being a guess doesn't make something worthless unless you equate the word "guess" with, say, "random". A guess is an estimate based on prior information. If there is no prior information, the guess will indeed be meaningless. The more prior information there is, the better the guess is likely to be. Did you ever assume anything different?
Dec 28, 2015 at 1:57 PM | Raff
To say analyses of temperature data (e.g. those that try to infer future movements) are guesses is to state the obvious.
I agree, but not for what you'd regard as the right reasons. They are guesses because there is no valid stochastic model available. Not because a "guess is an estimate" (which it is not).
"A guess is an estimate based on prior information." If you think that then you use the word "guess" differently from the rest of the English speaking world.
The Met Office does not say not say “Here is our guess for tomorrow's weather”. There are many books on prediction and estimation theory. But none on “guessing theory”. The pilot asks the navigator for an estimate of the remaining flight time, not a guess.
A guess is a surmise reached on the basis of little or no information other than the range of what is possible.
An estimate is something produced by processing available relevant information. It may be done mentally, or by using approximate formulas, or by using specific algorithms (eg least squares linear prediction theory).
(Note that a prediction is estimate of a future state. Likewise a forecast.)
Martin A
And, does it really mean anything, at all?
Dec 29, 2015 at 12:09 PM Radical Rodent
No. Nothing at all. Other than "from here to here, the mean rate of change turned out to be xxx".
They are just lines. Without an understanding of the underlying mechanisms your guess is as good as EM's, or mine, or the Met Orifice's.
Martin A
Golf Charlie
If my memory serves me correctly I believe that the Japanese may have many hundreds of years worth of records of first flowering of Cherry Blossoms in various parts of the country. Cherry blossom is an important part of Japanese culture; one of there specially designed kamikaze planes was called the Cherry Blossom. A reckon Google might help if it's something you want to follow up.
SandyS
Raff
"A guess is an estimate based on prior information "
Feynman famously pointed out that you then calculate the consequences of the guess and compare them with the
Results of experiments or observation. If these do not agree with the calculated consequences then the guess is wrong.
In this case the calculated consequences of the guess i.e. the rate of increase of temperature guessed from the temperature series are the output of climate models - none of which confirm the
guess. Thus the guess ( in this case that temperatures would rise at a certain rate in lockstep with atmospheric CO2
concentration is wrong. QED
Spectator
SandyS, I am not on a personal mission, but Japanese Cherry Blossom flowering, sounds like a rather more reliable treemometer, than chopping down the trees and measuring the rings! It also sounds like a non invasive technique that should appeal to genuine environmentalists.
I suppose some pseudoGreens have looked into it, and discovered results that did not match with their consensus view of history.
Any viewers here interested to find out?
After Fukishima, Japan suffered a Green backlash. As Japan doesn't really have any fossil fuels, common sense has prevailed, and pragmatism has returned. Japan is not in a position to fight a war with the Green Blob.
golf charlie
Martin A: I know. I was being rhetorical.
Radical Rodent
My dictionary says of the verb guess:
estimate or conclude (something) without sufficient information to be sure of being correctThat does seem to fit the case of analyzing temperature data. You say that a guess is a
surmise reached on the basis of little or no information other than the range of what is possiblebut having little or no information is different from having insufficient information. And it doesn't fit the example of climate and temperature where there really is a lot of information available. You may say that we can't make useful predictions because we don't have a perfect model, but again, so what? We'll never have perfect information, however long we wait, so we must make the best of what we have.
Raff
Spectator, can you show me where climate scientists has said, as you suppose, that "temperatures would rise at a certain rate in lockstep with atmospheric CO2 concentration". That is a strawman used exclusively by those who claim to be "skeptical". Don't bother searching academic articles. You wont find it.
Raff
Raff fancies himself a barrister it seems in the Court of Climate Orthodoxy.
Despite your protestations of being a reasonable person, and open to arguments, It seems your whole approach is to suppress rather than encourage. Control rather than facilitate. than I do hope you are not a teacher or a lecturer.
I am serious, if you could time travel yourself you would fit in so well at a Stalin or Rákosi show trial. Perhaps the Spanish inquisition would suit you better?
I can only speak for myself, but it is tiring, and people shouldn't give you any authority by engaging with you.
This isn't off topic, as this incompetent unimaginative engineer will, when he has the time over the weekend, add a comment to this discussion. But I am not a statistician so you can prepare your case in advance.
Media Hoar
Raff, where did climate science predict that temperatures would stop rising, or pause? It was certainly something of concern to Phil Jones in the Climategate EMails. Or is that old hat/failed science aswell?
golf charlie
Raff
"Spectator, can you show me where climate scientists has said, as you suppose, that "temperatures would rise at a certain rate in lockstep with atmospheric CO2 concentration". That is a strawman used exclusively by those who claim to be "skeptical". Don't bother searching academic articles. You wont find it."
Very interesting comment - are you saying that climate models are not based on increasing CO2 in the atmosphere causing higher water vapour content leading to atmospheric warming - or that such models are so based but this basis has no justification in climate science?
Spectator
Spectator, I think the figures are based on optimistic assumptions about climate scientist projected lifetime earnings.
golf charlie
"Sigh"
You sceptic ignoramuses might like to look up "stochastic".
The temperature record is stochastic in the short term, year on year, due to weather and ENSO.
On a longer scale, decades to centuries, it responds to forcings such as changing CO2.
The result is as you see , short term random variation around a long term trend due to CO2
The idea that temperatures should rise in lockstep with CO2 is not from the scientists at all. It is another simplistic denialist straw man.
Entropic man
EM, "really big deep sigh" are you sure you are upto simplistic rehashes of Schmidt?
golf charlie
Spectator, what do you understand by the word "lockstep" when you said "temperatures would rise at a certain rate in lockstep with atmospheric CO2 concentration"? If you mean the temperature is determined directly and immediately by CO2, why would a climate model ever be necessary if anyone thought that? We could just take the change in CO2 apply a sensitivity value and that would give us expected temperature change.
Raff
I think the climate obsessed are taking data analysis for the science of climate to where the eugenics supporters got phrenology in regards to identifying those worthy of enlightenment.
hunter
Sigh"
You sceptic ignoramuses might like to look up "stochastic".
The temperature record is stochastic in the short term, year on year, due to weather and ENSO.
On a longer scale, decades to centuries, it responds to forcings such as changing CO2.
The result is as you see , short term random variation around a long term trend due to CO2
The idea that temperatures should rise in lockstep with CO2 is not from the scientists at all. It is another simplistic denialist straw man.
Dec 30, 2015 at 6:33 PM Entropic man
Hello EM - I see you have found another word to add to your vocabulary. Looked it up did you?
"The temperature record is stochastic in the short term, year on year, due to weather and ENSO."
A bit of a tautology there that does not actually explain anything.
English Translation: "The short term temperature record varies randomly because of short term random variations."
"On a longer scale, decades to centuries, it responds to forcings such as changing CO2."
English Translation: "The longer term temperature record varies in unpredictable ways because of factors that are neither understood nor predictable."
The idea that temperatures should rise in lockstep with CO2 is not from the scientists at all. It is another simplistic denialist straw man.
Oh really? I seem to remember seeing a plot of the average of the GCM temperature predictions used by the IPCC, overlaid with atmospheric CO2 graph. The term 'lockstep' was used to point out the fact that the two curves were essentially identical.
Could you please help me understand where the GCM temperature predictions used by the IPCC came from if not from "scientists"?
Martin A
This thread is to continue discussion (if the discussants wish to) that was distracting BYIJ's "Predictions for 2016" thread.
Here is a graph of global temperature (so-called) data where EM fitted a line (blue, sloping) and I fitted a line (green, flat) graph
My view is that almost any sort of statistical analysis of global temperature (so-called) data is pretty much the equivalent of saying "here is my guess" - or "looking at that cloud over there, I can see a dragon, which confirms my dragon hypothesis".
I say this because to draw conclusions from statistical analysis of a time series (eg stating confidence intervals for something or other) you need to have a valid model for the process that generates the time series. And that is something you don't have in the case of global temperature (so-called) data.
There are, in principle, two ways to obtain a model for the stochastic process that generates a time series:
1. Empirical Analysis
Observe the time series for long enough and, from the observed values, construct a model that can generate a time series with similar statistical properties. To use this method, it is necessary that the time series is stationary ie its statistical properites don't change with time.
This method can't be used with global temperature (so-called) data because [A] only very short records are available [B] the process is pretty obviously nonstationary.
2. From Detailed Knowledge of the Physical Mechanism
From detailed knowledge of the physical processes involved, (eg from the differential equations describing it) you can in principle construct a model for the stochastic process that generates the time series.
This method can't be used for global temperature (so-called) data because a correct detailed understanding of the physical mechanism is lacking.
Bottom Line:
As I said above, if your model is just a guess, your statistical analysis is just a guess.