Bishop Hill Speccy on AR5
Sep 20, 2013 Climate: Surface Energy: other The Spectator has a leader article on the Fifth Assessment Report today, and pretty much nails it:
Next week, those who made dire predictions of ruinous climate change face their own inconvenient truth. The summary of the fifth assessment report by the Intergovernmental Panel on Climate Change (IPCC) will be published, showing that global temperatures are refusing to follow the path which was predicted for them by almost all climatic models. Since its first report in 1990, the IPCC has been predicting that global temperatures would be rising at an average of 0.2° Celsius per decade. Now, the IPCC acknowledges that there has been no statistically significant rise at all over the past 16 years.
And the outlook seems to be upbeat too:
As things have worked out, carbon emissions in the rich world have been falling anyway — not due to green taxes but to better technology, like fracking. Global warming is still a monumental challenge, but one that does not necessarily have to be met by taxing the poor off the roads and out of the sky. Sanity is returning to the environmental debate. Let us hope that, before too long, it also returns to British energy policy.
Reader Comments (138)
Dung
The last time CO2 was 400ppm, in the Pliocene, the temperature was 3C warmer than today and the sea level 20M higher. Why are you so sure it wont go the same way this time?
entropic man (2:08 AM) -
While there are considerable uncertainties, Pliocene warming seems to be mainly a consequence of greater oceanic currents, according to GISS: "Estimates based on carbon isotope measurements (Raymo and others, 1992; 1996) indicate that Pliocene atmospheric CO2 levels were, at most, 100 ppm greater than today. Moreover, if we compare Pliocene and modern ocean heat transport distributions (Figure 5) we find that a poleward shift in the peak ocean heat convergence would have been necessary to balance the Pliocene SSTs regardless of the CO2 level. Thus, neither simulation results or data support the conclusion that Pliocene warming was caused entirely by a large increase in atmospheric CO2 content. We cannot rule out, however, that some combination of the altered CO2 and altered ocean heat transport caused the warmer climate of the middle Pliocene."
EM
The graphs in AR4 show roughly 7 interglacials in the current ice age and that the last three were 2degC or more higher than this Holocene period. During those periods CO2 did not get above 280 ppm according to the IPCC report. There are other papers which put the temps much higher eg Watanabe et al., 2003 who said temperatures got up to 6degC higher than today with LOWER CO2. It is not about CO2 EM.
EM
For at least the last 15 years there has been a vested interest in keeping this gravy train on the tracks which, while it doesn't immediately affect the science, ought to make all of us view what science we are being fed by those with the vested interest with a high degree of scepticism.I thought we'd made the point that I don't need to find an alternative to CO2. The warming from the LIA accounts for what you might call the current "ratchet effect" (1970 starts from a higher level than 1910), or so I am led to believe.
Accepting that, it is up to the proponents of AGW to provide the evidence that "this time is different".
"We can't think of anything else so it must be CO2", which is (in simple terms) what we are being told, doesn't quite seem to hack it in my mind. They appear to be very keen on absolving the sun of any responsibility even though it has been unusually active for most of the 20th century. Of course if I'm wrong about the recovery from the LIA (and not being an expert in this field I am taking what I am told on trust here) then the rules of the game may have changed a bit!
But there again if that recovery is finished and the sun has gone quiet again that could also change things in a different direction.
All of which brings us back to the simple fact that we don't know.
But we're betting the farm on the psyentists being right. And I think esmiff explained why in this posting on the Will Hutton thread:
Or do you trust any of those firms not to rip off the general public if they can get away with it? I don't.
PS
And as I wrote yesterday, the debate over the extent to which CO2 is responsible — from 0% to 100% — is increasingly casting doubt on the certainties being espoused by the IPCC. All sides are claiming that the laws of physics support their position. I repeat: we don't know the answers.
A bit late reading this thread but I must say that agree with everything dcardno said (Sep 20, 2013 at 9:20 PM), especially "...given absence of modelled cloud effects in all the GCMs, I don't think we can say that anyone has produced "a good physical model" yet..."
Model apologists never recognise the number of physical processes that the IPCC admit are excluded from their models. Just follow this link…
http://www.ipcc.ch/publications_and_data/ar4/wg1/en/contents.html
…and simply look at the Table 2.11 in Chapter 2.9.1, which lists all possible sources of radiative forcings. Now, compare it to Fig. 2.20 (A) in Chapter 2.9.2 and you’ll see that those listed as ‘Very Low’ Level of Scientific Understanding (LOSU) are not included in any of the models. Then go read Chapter 8.6 (Cf. 8.6.2.3, 8.6.3.2 & 8.6.4) and see what they say about feedback and, especially, clouds.
Given these facts, I’m constantly amazed at the faith people put in current models and the degree to which they will argue that it has to be CO2 because nothing else matches the data, even when that match is demonstrably poor. I also continue to be disappointed at the way people like Dr Betts never want to discuss this issue and always ‘disappear’ from the conversation whenever I raise it.
EM I am afraid I have come to this thread a little late as I have been away overt he weekend, but lets take just one point at a time.
You have claimed at the top of this thread that you need 20 years of data to show no trend. This is not the first time you have claimed this. What is your source for this? What statistical argument or reference leads to that statement? What is the technical basis for that statement?
And just so we can remind ourselves about basis statistics, the detection of a trend is not dependent on the number of samples in isolation, it is dependent on the signal:noise versus the strength of the trend being detected.
So what's your source for that claim that we need 20 years of data? Your statement contradicts at least two other sources I know of, so you need to provide a reference. When we have resovled that point, I'll move onto other mistakes in your subsequent posts.
Hahhahahhahhha!
You're a classic case of a terrible court jester entropic. Classically the jester is supposed to be humorous with wit and irony; instead you just open that upper vent of yours and make up things.
Now why don't you post the statistical analysis backing up your bs? That not warming length of time goalposts have been moved back since before Trennyberth uttered his famous 'travesty' phrase. Don't know what I'm writing about? heh heh
Sep 20, 2013 at 1:03 PM | Unregistered Commenterentropic man
I see 20 years is now the magic number. Funny how it's always a few years longer than the current pause.
The smell of desperation gets ever stronger.
EM has been challenged on his 20 years claim several times but never responded. If he expects to be taken seriously he needs to give a valid reference. Otherwise its just BS, repeated until the meme takes hold, EM is playing games here like RealClimate do, with very long, involved posts with lots of references but closed discussion lines. Its a play, trying to control the story. The level of detail and the sticking to a particular story line is also looking very...unusual.
ThinkingScientist.
Patience, I have a life and am not surgically attached to my computer.
This is Statistics 101.
To show that a trend in time series data is significant you need to demonstrate that the difference between means is larger than the range of variation expected if there was no trend. This takes time.
I said this in a reply to a comment from ZedsDeadBed, but its's been censored, so I'll try again by example.
Consider a temperature record such as GISS or Hadcrut4 with a linear long term warming trend of 0.1C/decade. That series also shows stochastic variation of +/- 0.1C due to weather, volcanoes etc. The stochastic variation is independant of the long term trend. The actual data has a slightly greater warming rate since 1970, but this is close enough for a simplified example.
Over short periods the change due to the long term trend is smaller than the short term variation, and cannot be distinguished from it. Over a sufficiently long period the difference exceeds that expected from short term variation,
The temperatures at the end of the series need to be enough higher that the top of the range of variation at the start is below the bottom of the range of variation at the end. If the two variation ranges do not overlap, you can be confident that the difference is due to the trend and not just stochastic variation.
In my example the minimum difference between means necessary to show the trend would be 0.2C. At 0.1C/decade it would take two decades for this to occur.
If the rate of change were larger, the trend would become significant after a shorter period. If the slope were less, the time period would be longer.
You might like to try the exrcise I suggested to Radical Rodent earlier in this thread, which illustrates the principle.
A fuller explaination would express these concepts in terms of variance, standard deviation and 95% confidence limits.
You might also remember two interviews with Professor Jones, which touched on this subject.
http://news.bbc.co.uk/1/hi/sci/tech/8511670.stm
You said yourself- "And just so we can remind ourselves about basis statistics, the detection of a trend is not dependent on the number of samples in isolation, it is dependent on the signal:noise versus the strength of the trend being detected."
Perhaps you haven't spotted that this applies to temperature series. The noise is the stochastic variation and the signal is the long term warming trend. The strength of the signal is the rate of change.
Moderator
Sorry, my computer hung up and I've duplicated the last post. Please edit.
ThinkingScientist, good spot :-)
EM: Your reply is empty of references and is merely your, rather patronising, opinion. I don't need a lesson in statistics 101, I need to hear from you the technical basis for arguing that you need "20 years" of data. What is your criteria for testing for statistical significance and what are your assumptions about sample depedence? Whose statistical model are you following in this? You appear to be saying its "your opinion". And then have a rather niave statement about stochastic variation and the height of the points at one end of the graph being different to the other end. You are going to have to do a lot better than that if you want to be taken seriously here.
Your assertion that the length of time is the only criteria is simply not true: even a simple Student T test for spurious correlation in OLS requires the number of samples and the R^2, not just the number of samples, to establish significance.
EM, either stop arm waving and pretending you know this stuff or give us a hard reference to an established statistical method, with a justification for selecting it.
EM: Just for the record, if you take HadCRUT4 annual data for the period 1999 - 2012 (just 14 years) and run a Student T test** you will find there is no significant warming (R=0.43; Nsamp = 14; 95% CI, probability of spurious correlation 12.5%).
I selected that period because it avoids the spike in 1998 and avoids 2013 which is incomplete.
If you take the same period and add a linear trend with a slope of just 0.027 degC/decade (note: per decade) to that then you get a statistically significant correlation using a Student T test (R=0.54; Nsamp=14; 95% CI; probabillity of spurious correlation 4.8%).
So, on the assumption of independent samples, a real trend of 0.2 degC/decade will stand out like the testicles on my dog and be highly statistically significant over a period as short as 14 years, based on a Student T test. The probability of spurious correlation would be effectively zero (my calculation returns 0.0% - there's a roundoff there somewhere, but its very small). Even a linear trend of just 0.1 degC/decade added would give a statistically significant correlation using a Student T test, with the probability of spurious correlation being just 0.3%.
What's your calculation of statistical significance based on? More arm waving?
**Note: other statistical tests are available and may be more appropriate
Thinking Scientist
Independant 1-tailed t-test. If you are using GISS data, each year's average is based on multiple daily measurements from some 3500 stations. Sufficient?. Call it a minimum of 35,000 measurements for a 5-year average.
Make your mind up. Two posts ago you complained that I produce "very long, involved posts with lots of references"
Your last post complained bcause I did not produce a "very long, involved post with lots of references".
How about a side bet? £100 to a charity of our choice. If the 5-year moving average centred on the year 2018 is significantly higher, to 95% confidence, than the 5-year moving average centred on 1998 I win.
Calculation to be made using GISS land-ocean temperature data, using an independant 1-tailed t-test. Since my statistics is rustier than yours, I'll let you make the calculation, subject to independant check.
GISS give the current 95% confidence limits for the 5-year average as +/-0.04C, so the confidence limits for the 1998 and 2010 averages are just overlapping.
For those who dont know what we're talking about, try here.
http://www.gla.ac.uk/sums/users/jdbmcdonald/PrePost_TTest/chooset1.html
Thinking scientist.
Note that nobody, especially the professionals, are claiming significant warming for the most recent decade. You've just created a straw man and then proved it cannot exist.
That's a handy stats package you've got there, pity I no longer have access to such kit.
Thinking scientist
You win regarding the 20 years requirement. Since I regard this pause as temporary, the bet's still on offer if you want it.
Gentlemen
Statistics is very useful for analysing past temperature records but it has no usefulness in predicting anything or explaining anything. The greenhouse effect is a physical process which either happens or does not happen and it is then either large or small. The IPCC stated that it was the largest driver of global temperature, overwhelming natural drivers. The amount of human produced CO2 is rising year on year and the absence of warming in one year let alone 15 consecutive years, shows that the physical process is not happening in the way it was predicted to happen.
EM I note that you have not responded to my points about interglacial temperatures and in fact you have only responded to TS out of half a dozen posters who put questions/points to you.
EM says "You win regarding the 20 years requirement"
Thank you for conceding the point formally, although I should point out for other readers that I do not necessarily agree that a Student T test is the most appropriate test, partly because it assumes independent samples. However, I should also point out that the statement by EM about 35,000 samples is erroneous, they are averaged into single year values. I was running Student T tests on just 14 samples (the number of years) which is much more pessimistic and still shows a very low trend to be detectable.
Regarding your comment "That's a handy stats package you've got there, pity I no longer have access to such kit", whilst I do have access to very substantial software packages, I did the calculations I showed here in Excel. Excel has a built-in function for Student T tests called TDIST, so I am sure you can do the calculations yourself if you have a home computer with Excel installed. If not, R can be downloaded for free.
Anyway, I hope we will hear no more from you about how you need 20 years for validation. I recall Trenberth stated you need 15 years and Phil Jones 17 years, although neither give the basis for their statements and I don't think anyone has formally recommended the appropriate statistical method to use on this type of data (Steve McIntyre, any suggestions? Its an interesting question what method should be used to calculate significance of trends on the temperature annual series)
Regardng the bet, I am aware of it and am interested to note how you have proposed it here. The only bet I ever place is on the Grand National, so sorry not interested.
However, there are some other points in your posts in this thread that I may have time to take issue with....
EM says "Note that nobody, especially the professionals, are claiming significant warming for the most recent decade. You've just created a straw man and then proved it cannot exist."
I have done no such thing. Please show me where I have done this. I have stated that if you add a trend to the current data set, of the slope that you proposed, it would be statistically signficant with just 14 years of data, not 20 years as you had claimed. I am showing, with a trivial worked example, that your claim is not ture, and you have accepted that (which is genuinely gracious of you).
dung
Contrary to the rumours that I'm a front name for a group of lobbyists, there's only me, and there are other demands on my time.
One at a time, please.
You deny that CO2 changes can cause temperature changes and any attempt to discuss the matter involving radiative physics invites censorship. Thus any further discussion of the feedback hypothesis that "temperature changes drive CO2 changes and the CO2 changes then drive further temperature changes" is likely to be sterile.
I've given a represetative paper on the mechanisms linking temperature and CO2 changes during glacial/interglacial cycles.
https://edit.ethz.ch/umweltphysik/education/biogeochem_cycles/reading_list/broecker_paloce_98.pdf
Thinking Scientist
Peace? We've probably sqeexed enough juice out of this for one thread.
loves the casual use of "deny" in that post....true expertise!
Entropic Man
No worries. Thanks for conceding the point about 20 years openly and graciously. Fair play.
ThinkingScientist
Thank you. I'm not inflexible. Show me clearly that I'm wrong and I'll agree.
"Science is a way of trying not to fool yourself. The first principle is that you must not fool yourself, and you are the easiest person to fool.".
Richard Feynnman
HaroldW was right. In the absence of specific calculations my old rule of thumb is to expect significant difference when the 95% confidence limits of two means no longer overlap. That now seems too conservative.
Entropic Man
Sep 23, 2013 at 5:25 PM | Unregistered Commenterentropic man
You set up 3 straw men, introduced a red herring and used the word deny in a very short post, congratulations!
You seem to have plenty of time and energy to write many posts criticising others but less when it comes to defending your own posts.
I can see that you enjoyed the discussion you had with TS but you have criticised many people (including me) not just TS and it is only right that you respond.
I did not "deny that CO2 changes can cause temperature changes" and I did not once mention radiative physics.
The link you gave me was for a paper written in 1998 (even before Al Gore's film was made) and it talks about temperature and CO2 rising together at the termination of a glacial period/start of an interglacial. At that time it was not possible to zero in on short time periods in the ice cores. Surely you realise that today it is known that that CO2 lagged temperature by a minimum of 800 years and as much as 2000 years at the start and end of interglacials?
I would be most grateful if you would respond to all the points I have raised, they were made in response to all the criticisms that you made.
Dung: The publication of the Vostok ice core, with its 800 year lag the wrong way (for CAGW) was published in 1999. So you have a good point.
I think that EM should spend less time arm waving and posting irrelevent references and focus on one point at time. Less scattegun, more focussed. I think EM needs to pause and remember that the general level of knowledger at BH is both diverse and comprises many specialists in a variety of relevent areas to a climate discussion. Some of the posts by EM are pretty patronising and technically weak. Maybe its the school teacher. We are not O level students.
Regarding your point about lags, clearly the Vostok ice core shows a lag of 800 years and the current best temperature data shows a CO2 lag too. If CO2 lags at all time scales, how does it drive temperature?
Dung
The most recent work suggests that the increases in temperature and CO2 at the start of the Holocene were synchronous. There was no lag.
http://www.sciencemag.org/content/339/6123/1060.full
Dung, ThinkingScientist
Also remember Shakun et al's paper suggesting that the orbital changes at the start of the current interglacial may have triggered an increase in atmospheric CO2 which either went along with the temperature increase, or may have run ahead of it.
http://www.nature.com/nature/journal/v484/n7392/full/nature10915.html
EM: That argument makes no rational sense because:
(a) CO2 always lags temperature, whether temperature is increasing or decreasing
(b) If CO2 is such a powerful GHG, how do you get back to an ice age? You would need an even bigger negative forcing agent
EM
I am willing to believe that you are not a troll, however you are exhibiting troll like behaviour; you do not respond to all the points put to you and you are happy to throw out references to papers and let others do the work. You have now given me two papers to read (that prove your point), I debunked the first one and now I debunk the second one ^.^
The second paper is utter tosh but then it was published in Nature so no surprise there. It purports to show that the layers of "firn" laid down each year (before eventually becoming ice) hold CO2 that is not the same age as the ice itself. As the "firn" forms and more layers form on top, pressure and density increase and some air is squeezed out to the surface while the rest forms bubbles. The implication in the paper is that over time these bubbles move upwards until they are finally locked in when ice forms. This is an unproven theory and is not repeated or confirmed anywhere else, it directly contradicts the British Antarctic Expedition report for instance.
However none of the above matters one jot since the temperature signal (Oxygen isotopes) and the CO2 signal are sharing the same bubble!
dung
I havent had time!
Briefly
Thinking scientist
Normally CO2 is not a driver. The default long term pattern for climate is that insolation drives temperature and then CO2, water vapour and albedo changes amplify the temperature trend. Warming and cooling take place because the changes are slow enough for the system to remain close to equilibrium, so a small decrease in insolation can gradually decrease CO2 and water vapour and increase ice albedo.
Offhand I can think of three examples in which CO2 acts as the driver of temperature.
1) At the end of a Snowball Earth episode CO2 from vulcanism accumulates until the greenhouse effect becomes strong enough to start the planet warming.
2) Large scale vulcanism such as the Siberian Traps releases large amounts of particates, aerosols and CO2. The initial effect is cooling from the particulates and aerosols. When they settle the CO2 remains and triggers warming.
3) An industrial civilization burns fossil fuels, rapidly increasing CO2 and triggering warming.
dung
Intersting climate papers are cited by other scientists and used as the basis for further research. They are "debunked" by spin-sceptic propoganda sites. Funny, that.
" However none of the above matters one jot since the temperature signal (Oxygen isotopes) and the CO2 signal are sharing the same bubble!"
You're a little confused here. CO2 is trapped in gas bubbles. Oxygen isotopes are in water molecules. Thus the bubbles migrate upwards in the firn to later layers, while the ice ,and the oxygen isotopes therein, tend to stay at the original level.
You can see the movement of the bubbles yourself. Put a bottle of water in the freezer overnight, then watch it thaw.
Perhaps it is not me who is confused:
Take a look at this link, particularly the sections "Oxygen isotope dynamics" and "Application of oxygen-isotope dynamics to Paleoclimate".
Both refer repatedly to oxygen isotopes in water.
http://academic.emporia.edu/aberjame/student/tinsley1/webpage1.html
EM
Obviously I agree with what you said about the information in the link you gave, it talks of isotopes in water not atmosphere. My understanding of what is happening here is that scientists investigate various aspects of past climate. Some researchers melt ice cores to find out about the past temperature of the oceans, others are interested in the atmosphere because the CAGW scare is about surface temperature and atmospheric CO2 not ocean temperature and any CO2 dissolved in the oceans.
The process described in the paper I linked grinds the ice at sub zero temperatures so that the bubbles of air are released but not contaminated by the contents of the water/ice. The result is a clean picture of what is going on in the atmosphere at a time determined by the Oxygen isotopes within the air.
dung
You're right.
Lots of isotopes in the air. Three from carbon, two from oxygen, two(?) from nitrogen(old age is setting in). Peplow's technique lets you sample isotopes of carbon, oxygen etc in the trapped air.
Levels of carbon 14 give calibration for carbon dating and a proxy for cosmic ray intensity. Carbon 13 helps with judging the proportion in the atmosphere of fossil fuel carbon dioxide (which is low in carbon 13).
Except in water vapour, I dont remember that oxygen 18 levels in air vary much, though I may be wrong.
My own understanding of O18 as a temperature proxy is that the heavier water molecules with oxygen 18 instead of oxygen 16 are less able to evaporate, so are less frequent in snowfall and ice cores than in the ocean they came from. Oxygen 18 water evaporates more easily from warmer water, so a greater proportion of oxygen 18 water in an ice core indicates higher sea surface temperatures.
There's a lot more complexity to all the isotope studies used in climate science, so I imagine we've both missed the nuances!
This has been a very useful thread. I've certainly learned a lot. Sorry other things have got in the way.
I'm organising Halloween at work; plus two model railway exhibitions, an archery competition and a trip to England.
EM
You are a gentleman sir ^.^ and if you are in the Midlands and need any help or a place to rest or refresh, give me a bell on 01827 717136 ^.^
I read about the fact that O16 evaporates more easily but I do not remember reading that O18 never evaporates. Just a few O18 molecules are necessary in order to date the CO2 I think. If it were possible to do, it would make sense to try and analyse both the air trapped in the bubbles AND the surrounding ice. This would show whether or not the movement of bubbles after lock in was an issue or a red herring.