Guest essay by Jeff Patterson
In a recent post on RealClimate, the author examines the statistical significance of the “The Pause” using a technique recently in vogue called changepoint analysis (CPA). The basic idea is to subdivide a time series into intervals and determine if a statistically significant change in the regression slope can be detected at the interval breakpoints. For a given number of break points, all potential break point positions are tried and the best fit is recorded. The number of break points is increased by one and the analysis is re-run. This continues until no significant reduction in the residual is obtained. The breakpoints at which a significant change is detected are called “changepoints”.
Since CPA is designed to answer the question, “has something changed” (we use it where I work to monitor the defect rate of electronic assemblies as a process control metric), one can forgive the naive application to global temperature undertaken in the aforementioned post. The author’s basic thesis is that since a CPA analysis detects no significant recent change in the slope of the GISS dataset there is no pause. Unfortunately, the analysis is of no value because, as is commonly known, the CPA cannot be used on auto-regressive time series. This can be easily demonstrated. Here’s a random sample of an ARIMA[3,1,1] process (This is not to infer the climate can be modeled as an ARIMA process. CPA fails for any integrative process, a class which in all likelihood the climate falls within.)
Figure 1 Simulated climate data from an ARIMA process
If we run this random data through R using the standard changepoint package we get:
Figure 2 – Changepoint analysis using R
The CPA algorithm detected three significant “changepoints” in a process known to have none.
So while I place no value in the analysis, ironically I actually agree with the author’s contention that all this talk of a pause is gibberish. The fact of the matter is that there has been no statistically significant increase in the rate of warming over the entire observable temperature record. Here is yet another way to demonstrate this unassailable fact.
Start with the Cowtan and Way modified Hacdrut4 global temperature series:
Figure 3 – HadCRUT4v2
Subdivide the series into 640-month intervals, where each interval is offset 2 months from the previous interval (638 month overlap). Plot the least-squares, best-fit slope (in °C/decade) for each interval.
Figure 4 – Sliding window regression
Add the best-fit linear regression to the above.
Figure 5 – Sliding window slope regression with best-fit line
Over the 32 year period from 1963 to 1985 the rate of warming increased from .01 in °C/decade to .15 °C/decade, not significantly different from the -.03 to .1 change that occurred from 1893 to 1930.
As we decrease the interval length, the data gets noisier but we can get a better idea of the recent behavior. The conclusion remains the same.
Figure 6 – 640 and 320-month interval slope regression
Conclusion
One benefit of the recent discussions on the so called “pause” in global warming is a healthy re-focusing on the empirical data and on the failure of climate models to accurately reflect climate dynamics. Yet to speak of a pause infers that the rapid warming that occurred at the end of the last century reflects the true, post-industrial trend. As the analysis above shows, there is no empirical evidence to support the notion that that period was particularly unusual, much less that it was due to anthropogenic effects.
In short it is in my view incorrect to term the nearly 20 year slowing in the rate of warming as a pause. Rather it is the natural (and perhaps cyclical) variation around a warming trend that has remained constant at ~.008 °C/decade2 since the late 1800s. There is no empirical evidence from the temperature record that mankind has had any effect one way or the other.
Hi Jeff Why does this remind me of S.P.C.s
“CPA fails for any integrative process, a class which in all likelihood the climate falls within.”
These seems to be a basic assumption in your analysis. Please elaborate. Thanks.
Please see my response to Brandon Shollenberger below.
Infer/imply. Get ’em right. Please.
I can’t wait to read the correction on RealClimate.
” . . .warming trend that has remained constant at ~.008 °C/decade . . . ”
Shouldn’t that be per YEAR ? OR 0.08 °C/decade . . ?
. .OK. sorry, I see it is ‘an INCREASE in the rate of warming . . not just ‘the rate of warming. I was thinking of the warming of 0.8°C over about a century, when I made the previous comment.
Whether it is due to humans or not, at least we all can agree that it’s been warming and there is no pause.
Two words: statistical significance.
“it’s been warming”
Since when? We can agree that it has been warming since the last glacial maximum. Since the medieval warm period, not really. Since 1880, yes; since 1997, no. Since the PETM, alarming cooling indeed.
Wrong. Dead wrong.
Today’s “pause” of 18 – 2 months has lasted nearly as long the original “Warming period” of 1976 – 1998.
But wait, how could there have been a warming period of 1976-1998? Using the methodology in Ross McKitrick’s paper which was promoted here as demonstrating a 19 year “pause” in HadCRUT4, there was a pause in warming from 1983-1996 (also in HadCRUT4).
Does that mean there was a 10+ year “pause” in the middle of “the original ‘Warming period’ of 1976 – 1998”?
I suppose it would help if I included a link to the post discussing the calculations I refer to. Oops!
Please look at figure 5 again RACookPE1978
You make my point: Figures 4, 5, and 6 are “slopes” of the original temperature plot. As such, they show the changes in temperature with time. (Its acceleration periods, its steady periods, and its decreasing periods of how fast temperatures are changing with time.)
Thus, when the slope of such a curve is zero, the change in temperature increase over time is zero, which shows that the temperature itself is constantly rising or falling at the same rate (degrees per decade for example). When the slope of Figures 4, 5, and 6 is highest, the temperature curve has “bent” (or inflected) which indicates a change from a warming to a cooling period, or from a cooling period to a warming period.
Broadly speaking, a sharp heating period from 1910 to a broad maximum in 1945, and then a flat or slowly cooling period from 1945 through 1976, then another period from 1976 up towards today’s broad flat peak between 1998 – 2015.
Then? Most likely a slow cooling period of 33-35 years, or a flat period of 33-35 years.
But thereafter?
Are we now at the peak of the Modern Warming Period? Or do we wait until 2060-2070 for that peak?
lol. Brandon can’t see the woods or the trees.
It is NOT a pause: it is a STOP. Won’t know if it was a pause, until it is not a stop any longer.
It has been cooling since the Holocene Climatic Maximum 6,000 years ago. Each of warming periods that followed – Minoan, Roman, Medieval, and the one we are in now – was not as warm as its predecessor. Current warming is just a natural rebound from the coldest period of the past 10,000 years, the Little Ice Age (1450-1850 AD). Historical climate “ignorati” don’t recognize that climate change did not begin with Al Gore’s birth.
+1.
An excellent, concise ‘big picture’ summary.
Bill 2 December 7, 2014 at 7:56 pm
Whether it is due to humans or not, at least we all can agree that it’s been warming and there is no pause.
Careful with that straw you are clutching Bill – it is increasingly fragile.
Sorry, Professor Wegman’s caveat applies here: climate scientists should work with statisticians to determine the validity of their techniques.
Most climate variables present as times series that are non-stationary as are most economic variables. We should not be surprised then that econometricians have developed statistical techniques appropriate to time series data. Granger and Engle received the Nobel Prize for their cointegration approach.
http://www.kva.se/globalassets/priser/nobel/2003/sciback_ek_en_03.pdf
Based on this econometric technique called polynomial cointegration analysis an Israeli group concluded,
“We have shown that anthropogenic forcings do not polynomially cointegrate with global temperature and solar irradiance. Therefore, data for 1880–2007 do not support the anthropogenic interpretation of global warming during this period.”
Beenstock, Reingewertz, and Paldor, Polynomial cointegration tests of anthropogenic impact on global warming, Earth Syst. Dynam. Discuss., 3, 561–596, 2012
URL: http://www.earth-syst-dynam-discuss.net/3/561/2012/esdd-3-561-2012.html
There were critiques of the methodology and the group slightly altered their conclusion to say that they had not disproved the AGW hypothesis.
In my opinion their study shows that, based on the data, they could not dismiss the null hypothesis.
No, Information Theorists, not Statisticians.
Some people seem to forget or choose to ignore that there was a significant El Nino event in the 1997 /1998 period and that a similar magnitude event hasn’t happened since.
Which was cancelled out by the large La Ninas in 1999/2000.
Comment on Beenstock here:
Comment on “Polynomial cointegration tests of anthropogenic impact on global warming” by Beenstock et al. (2012) – Some fallacies in econometric modelling of climate change
D. F. Hendry and F. Pretis
http://www.earth-syst-dynam-discuss.net/4/219/2013/esdd-4-219-2013.html
A reply by Beenstock et.al http://www.earth-syst-dynam-discuss.net/4/C118/2013/esdd-4-C118-2013-supplement.pdf
.Evidently analyzing time trends is not child’s play.
“The second misunderstanding is about greenhouse gas theory. Our clearly
stated motivation was not to refute this well-established theory
(see sections 1 and 4).
Below we reiterate why the time series data that we and others use are
inappropriate for these purposes.”
I think it is time to admit that when trying to analyze climate data without establishing what the data is(statistically) is unsupportable as science.
From another anonymous reviewer:
” For the length of time series we have it is hard to come to very strong conclusions about the nature of the series using simple tests and, therefore, the possible relations between them.”
Phil C
Let me add that the post supports the view of climate scientists of a moderation of the rate of global warming in the naughties, a moderation that seems to be coming to an end.
Figure 6 shows that the rate of global warming is accelerating on the average – so what is the fuss about?
Which average? Which dataset? Which period ?
You are missing the point of the article the rate of warming is not accelerating.
If a period without SUVs and mega coal plants had the same rate of warming as a period with SUVs and mega coal plants, we know the warming is not due to SUVs and mega coal plants. The fuss is about the urban legend disguised as science that warming is primarily anthropogenic.
“Figure 6 shows that the rate of global warming is accelerating on the average”
Figure 6 shows that the average rate of global warming (the slope of the dotted line) is constant.
” – so what is the fuss about?”
My point exactly.
I’ve been looking at the met office data for England at http://www.metoffice.gov.uk/public/weather/climate-historic/#?tab=climateHistoric.
I’ve noticed that all the stations I have looked at show about a 1°C max and min annual temperature step up from 1988. I have searched for a reason for this in sun spots, ocean indexes and volcanoes without sucess. Then Chernobyl suddenly popped into my mind. I googled the date of Chernobyl and it was 1988.
England and Wales suffered a lot of fallout from Chernobyl. A coincidence?
It is there in the CET as well.
just ran a graph on the CET data (min values) 1980 onwards ,data shows a darn good cooling from start 1988 to mid 89 followed by an uptick, so I see a step `back up` (even some of the months values show a negative slope from 1980 to end 2013)
regards
That’s strange jono. I constructed the Yearly CET data from the daily CET. It shows a warming since 1988.
I downloaded the CET Yearly data and it shows a cooling. Strange that my CET Yearly agrees with the individual stations?
Chernobyl happened in April 1986
By what mechanism would the meltdown effect a temperature series?
Thanks I must get a new pair of reading glasses. The CET max does show it starting 1986.
I don’t know why nuclear fallout would affect temperature Kevin, it’s just something I noticed.
I remember well being at the Expo 86 worlds fare at Vancouver in 1986 and visiting the USSR pavilion and their embarrassment about Chernobyl. So I knew right away that the 1988 date is obviously wrong.
http://bobbea.com/expo-86/ussr.html
There is a pause because the oscillation with a 60 year period was ignored.
The acceleration in the warming starts before human emissions were knee high to a grasshopper and with the 1940s blip put back where it belongs, there isn’t even that.
Why is this post using 640 and 320 month windows? No explanation is given for these parameters, nor is the effect of the parameter choice even considered. It seems completely random and arbitrary, and it makes the rest of the calculations suspect.
More troubling, why does this post say:
When this is not remotely true? There are many different approaches to changepoint analyses. Some can’t account for autocorrelation. Some can. This post claims to demonstrate CPA can’t do it:
But that’s only using a single approach to CPA, and it is done without considering the parameters which go into the approach. The R changepoint package being used has a parameter which controls the sensitivity of the algorithm to changes. You could likely find values for it which would result in no breakpoints. Even if that weren’t true, there are ways of testing the significance of the breakpoints you find. You can’t argue a methodology is bad because it finds a number of breakpoints if the same methodology would tell you those breakpoints are insignificant.
And that’s when using only the simplest of changepoint analyses. As the figure shows, each of those changepoints accompanies a flat segment. There’s no reason to believe the temperature record is made up of flat segments. The RealClimate post doesn’t use flag segments either. Instead, it uses slopes. I’m not sure how appropriate it is to fit straight lines to the temperature record to try to detect changepoints, but it is far, far better than fitting flag segments.
It also makes it trivially easy to account for the autocorrelation this post says can’t be accounted for. If you want to account for autocorrelation when fitting lines to the temperature record, all you have to do is add a parameter to your regression to account for whatever form of autocorrelation you believe the data has. Problem solved.
Pretty much nothing this post says about changepoint analysis is right. I’m not sure about the other things it says. I haven’t been presented any justification for them, so even if they are right, I have no reason to believe it.
“Why is this post using 640 and 320 month windows? ”
The window length was chosen to demonstrate that the temperature “trend” is constantly evolving about a central mean.
“No explanation is given for these parameters, nor is the effect of the parameter choice even considered.”
Halving the interval shows the (non) effect of the parameter choice. The central mean is the same in both cases (see fig. 6)
” Some can’t account for autocorrelation. Some can. This post claims to demonstrate CPA can’t do it:”
The issue is not autocorrelation. Processes which contain constructive feedback with memory when forced with random fluctuations exhibit rapid transitions toward the extremes which appear to the CPA algorithms be a change in slope. This is a well known limitation of CPA. My request to the author to explain how this issue was addressed went unanswered. The onus is on the person making a claim of proof to show that the model used was appropriate for the task at hand.
“As the figure shows, each of those changepoints accompanies a flat segment. There’s no reason to believe the temperature record is made up of flat segments.”
You misinterpreted the plot. CPA is based on the cumulative sum of the data. The flat regions on the R plot show the regions of constant slope in the data.
“Why is this post using 640 and 320 month windows?No explanation is given for these parameters ”
The initial interval (>50 years) was chosen because that’s at the upper limit of the (ever increasing) time-span climatologists claim is required for detection of a change in trend (and non coincidentally model falsification) . The 640 month plot shows the slope is oscillating about a central mean which has remained constant from the pre-industrial period to the present.
” nor is the effect of the parameter choice even considered.”
Halving the interval shows the (non) effect of the parameter choice. The central mean (dashed line of figure 6) is the same for both cases.
“There are many different approaches to changepoint analyses. Some can’t account for autocorrelation. Some can.”
Autocorrelation isn’t the issue. Processes which contain constructive feedback and memory exhibit rapid transitions to the extremes when forced with random fluctuations which cause false positive detection in the CP algorithm. This is a well known limitation of CPA. My request to the author for an explanation on how this issue was addressed went unanswered. The onus is on the person making the claim that the analysis tool is suitable to the underlying process.
” As the figure shows, each of those changepoints accompanies a flat segment. ”
You misinterpreted the plot. The flat lines in the R plots show the regions of constant slope and do not imply the slope is zero over the region.
Jeff Paterson, this explanation:
Provides no justification for your specific parameter choices. You could have just as easily picked any of a hundred other window lengths. You either need to show this specific choice was justified or show the choice of window doesn’t affect your results.
Nonsense. More often than not, scaling something by a factor of two will cause the results to exhibit largely similar patterns. That means getting a similar pattern when scaling by two is non-informative.
Except you said it was by claiming CPA cannot be done on a “auto-regressive time series.” Auto-regressive time series will necessarily exhibit autocorrelation. You cannot hand-wave away comments about autocorrelation as not mattering while simultanesouly claiming autoregression does matter.
I’d be more bothered by you not getting an answer to your question if you didn’t just argue about burden of proof while dismissing an entire type of CPA with nothing more than a handwaved claim that they can’t be done. Or, you know, if you hadn’t selectively chosen parts of my comment to address, ignoring things for no apparent reason (like the significance testing which can be done for breakpoints). Or if you didn’t constantly resort to hand-waving, like when you say:
Because that’s actually exactly what they imply. R’s basic changepoint package uses cpt.mean to find a change in mean values of a series. A change in mean values necessarily requires each segment be assigned a slope of zero.
I believe it is you, Jeff Patterson, who has misinterpreted the R plot. Those flat lines very much appear to show what Brandon Shollenberger thinks they show. If the aim were to divide the plot into segments of constant, non-zero slope, then surely the breakpoints would come at the ends of periods of strong up and down trends, not right in the middle of them!
Nigel Harris, it’s easy enough to verify I am right about the R function. The function is cpt.mean. Anyone can look at its documentation and see what it does. Alternatively, one could just recognize an abbreviation for “changepoint mean” likely calculates changepoints for mean values, not changepoints in trend values.
And really, if the function was calculating changepoints in trends, why would it it show every segment as having 0 slope? Why would it calculate trend changepoints then refuse to show any trends and instead show unrelated flat lines?
This is one of those things where I’m not sure how much more obvious it can get >.<
I should have mentioned that the choice of the process model used to demonstrate the false positive CPA issue was informed by this paper from the MET office (http://www.metoffice.gov.uk/media/pdf/2/3/Statistical_Models_Climate_Change_May_2013.pdf) which, while not endorsing it as a valid representation of the climate system, found it most likely of the models tested to produce samples which mimic the climate process. For the purpose at hand, all that was required is a process which produced “climate-like” sample ensembles. To estimate the ARIMA parameters, I used Mathematica’s process estimation function on the Hadcrut4v2 dataset.
I guess you can say it has been warming if you pick a starting point when it was cooler. In 1955 you could have said it was cooling if you picked your staring point as 1930.
In this case the starting point picked was 1880 and temperatures do appear higher in 2000 than in 1880. But unfortunately for climate “it’s just physics” propaganda, the warming cycles do not coincide with CO2 concentrations in the atmosphere and as the article points out the rate of temperature increases starting in 1880 and 1960 are the same.
About the recent behaviour(pause)(fig.6) is the conclusion valid that the increase in cooling equals the actual anthropogenic contribution? That would mean that the anthropogenic contribution until now is no more than appr. 0,4 K and this number includes effects not related to CO2. This is half of the IPCC number and spans a 150 year. Nothing to worry about, at least in the short term.
No, the conclusion is that the rate of warming (or cooling) is a constantly evolving parameter of the climate system which oscillates about a constant trend line that shows no anthropogenic contribution. If we are going to attribute the increase in the rate of warming late last century to CO2, we can’t account for the nearly equivalent increase that took place prior to industrialization, or the equal but opposite decrease in the rate from 1930 to 1960, or the decrease that has occurred since the turn of this century in the face of ever increasing CO2 concentrations.
More and more micro-splitting hairs on the “pause” — it’s ridiculous.
It’s really simple — global temps haven’t changed significantly in 15 yrs or whatever.
Its even simpler. Santer et. al said that a temperature trend (e.g. No statistically significant warming) outside the CMIP5 warming envelope would have to last for 17 years to falsify those models. McKitrick showed that (depending on dataset) there has been no statistically significant warming for 16, 19, or 26 years. The models are falsified by Santer’s own warmunist criterion –Independent of whether or how CPA is used. Easiest is to just plot the temperature data versus the CMIP5 archive and use the finely calibrated Mark 1 eyeball instrument to detect the discrepancy and its duration.
Of course not. Temperature records don’t come with labels that indicate how much if any of the observed changes are affected by mankind. But physics tells us that a significant increase in greenhouse gases will have an impact on surface temperatures.
NO. “One simplified Global Warming THEORY using simplified physics over long-time periods using computer simulation of greatly simplified approximations of incredibly complicated physical reactions tells the politicians paying for those computer simulations that there MIGHT BE a measureable impact on surface temperatures.”
Actual measurements during periods when CO2 has been steady, rising, and falling show there has been no detectable man-caused global warming (or cooling) over many long periods of time. Your theory using “simplified physics” is incomplete.
Unfortunately, the physics don’t tell us what the sensitivity of the system is to greenhouse gases and therefore cannot inform us as to their impact relative to the (also unknown) natural variation. Hence the reliance on models which have thus far been shown to be quite dubious.
Physics tells us that a glass jar of CO2 will absorb more long wave radiation than a glass jar of O2. Because of that, it is not unreasonable to think that CO2 has a positive affect on climate temp. From observation it appears that if it does have an affect, it is below the noise of the system and undetectable with current data. That could change with more/better data or a higher concentration of CO2. Panic if it seems fitting.
NO! Physics only tells us that ‘absent feedbacks’ increases in greenhouse gasses ‘should’ cause an increase in temperature. Without a thorough understand of the system and the physics involved you don’t know if the increase in gasses will cause an increase, decrease or have no effect on temperature.
The elephant in the room is still, and always has been, the feedbacks. Without a thorough understanding of those one cannot tell whether any applied changes to the system will have any effect at all. The assumptions used for the feedbacks by the warmists (i.e. that they are all positive) are totally unworkable in any control system, manmade or natural.
It is the central fallacy of the climate obsessed that they have *the* interpretation of the physics regarding Co2 and climate. It permits them to ignroe reality and to dismiss pesky data to continue their apocalyptic mantra…and calls for ever more money.
This analysis seems to me to SUPPORT the warmists: projection of the trends indicates that the next 20+ years will not be a cooling period, but just a slow warming. The warmists say that natural variation is an add-on to CO2 warming; the skeptics say that CO2 is an add-on to natural variation. The question is dominance. A dominant, continuing natural warming trend is NOT in the skeptic’s position story, as there is no science for it; it would also be indistinguishable from CO2-induced warming.
For CO2 to be the subordinate add-on, we need a period of actual cooling, not just lessened warming. I admit that I am in the future-cooling camp. This work makes me wonder why.
The long term warming trend is the same it has been since well over a century ago, well before rising CO2 could have established it. Only after removing that warming trend can you consider the effect of CO2. When you do that, there isn’t much warming left which could be attributed to rising CO2. Just because we do not know the cause of the long term warming trend is no reason to shove it into the CO2 slot.
Thanks for your thoughts, Bart.
Let’s be clear: I believe in natural variation or factors being dominant, CO2 being subordinate. But a natural trend that is positive will be indistinguishable from CO2 forced, as I mentioned. A “pause” in the temperature record could be easily interpreted as a result of a NEGATIVE natural variation on top of a CO2 forced rise, allowing an internally coherent explanation of strong CO2 forcing.
The work may well be right. But I don’t like that it shows a change in magnitude and not one of direction for the next few decades. For CAGW to be knocked off the regulatory scene, we need a temperature decline of at least the 1940 -1965 period. Preferably deeper, so that the radiative forcing of CO2 will have to be acknowledged as signficantly less than natural variation. The models that would be run with a reduced CO2 forcing would have to compensaate for the 1850 – 2010 period with increased natural variation factors. The benefits of CO2 management would fall away. Threats of anniliation would survive, but the efforts would be mitigation, not legislation.
The argument from ignorance IS the underpinning argument of CAGW. We shouldn’t jump on CO2 just because the theory fits. No: but few people will acknowledge that their analytical work produces, mostly, internally coherent reasons consistent with observation. Not “the” reasons, but reasons that in a hindsight, curve-fitting manner, acceptable.
Climatologists and other religious people, plus engineers by profession mistake “an” answer for “the” answer regularly. I call it the Unique Solution Syndrome: all problems have either one correct answer or one “best” answer, according to this system of belief. Once “the” solution has been determined by consensus, all other possible solutions BY DEFINITION are either wrong or, at a minimum, inferior. CO2 does it all, God is an old, white European with a bushy beard, and the Tacoma Narrows bridge is a perfect design. None of these conclusions or positions can be changed until striking observations are make that are impossible, not just difficult, to ignore.
We need a period of distinct cooling to end the CAGW disaster, not just a minimal heating period.
Doug Proctor says:
The warmists say that natural variation is an add-on to CO2 warming; the skeptics say that CO2 is an add-on to natural variation.
There’s an easy way to show who’s right: a verifiable, testable, quantified measurement of AGW, showing the specific fraction of human-caused global warming, out of total warming. A percentage would do. Is it 89%? Is it 2.6%? Is it 50.0%? No one knows, because no one can measure AGW.
Since there are no such measurements, everything is only an opinion; a conjecture. I see one of two possible reasons for the absence of any measurements of AGW, since just about every physical process can be measured:
1) There is no AGW. It doesn’t exist. Since I reject that, I think it is…
2) AGW is so minuscule that it’s below the background noise. That’s why AGW isn’t measureable. There are no instruments currently sensitive enough.
Of course, that completely destroys the climate alarmists’ case: if AGW is that insignificant, we should promptly end all public spending on what is a non-problem, and concentrate on important things.
I would add a question to Doug’s comment: does he believe that the putative warming effect has so precisely balanced global cooling, that temperatures have remained completely flat for the best part of two decades? Amazing if true. Because the recent warming is large enough to measure, but AGW isn’t.
Hey, DB,
Read my reply to Bart, above.
I think the amount of AGW warming is indeed miniscule. I think the dominant, natural factors are, well, dominant, so that what we are seeing is the variations of natural factors at work.
A couple of years ago I worked out my own version of temperature changes in the Central UK, posted on Tallbloke’s Talkshop. (I am not well versed in physics, and so had to work out for myself some of the principles; the result of which is the article is awkward and overly long). I convinced myself that there was a CO2 signal but it was of no importance. What counted was global energy changes from AMO-PDO shifts, plus a determinable change in cloud cover.
What I found – not interpreted, but graphed – was a cycle of increased “bright” sunshine hours and maximum daytime temperatures that was time-dependent in magnitude. The time dependency tied to AMO-PDO cycles. The cycles showed the warming up to the ’40s and then the cooling to the ’60s, and then another, stronger cycle that ended – according to the graph – in 2010.
Many say curve-fitting or correlations can’t be used to determine causitive reasons, and they are right … while demonizing potential insights into hidden causitive factors. I’m looking to 2015 as a watershed year. I want the IPCC, Al Gore and David Suzuki to be relegated to the dustbins of history for all the harm, alarm and self-serving egocentricism they caused and represent. But I think a period of real cooling is what it will take to do this. Anything less remains a spin of the CAGW narrative.
Doug Proctor,
Thanks for your reply. In reading my comment, it didn’t really look like I intended. Sorry if I came across so abruptly, or however it looked to you.
Anyway, I don’t know much about solar physics, but I do know something abouty the effect of CO2. This graph shows why there is no warming resulting from the current rise in CO2. That beneficial trace gas could go up another 25% – 30%, and we still couldn’t measure the resulting global warming.
CO2 is only a 3rd order effect, which is swamped by higher order effects. But “carbon” can be taxed, because almost every industrial process emits CO2 — and that can be measured.
The “carbon” scare is a government-originated hoax. It is perfect for massive government growth. But it would be bad for everything else.
The flaw in nearly all flawed ‘regression analysis’ studies is made before a single calculation has taken place. The most greivous error is in the selection of an inappropriate mathematical model. If the physics of the process is not reflected in the equations of the model as used for calculation, only self-delusion can result. I have personal experience in this from several exercises: describing the performance of a flame ionization detector in a gas chromatograph under regimes where the stoichiometry transitions from complete to incomplete combustion, describing the flow rate of gases through a micron-sized orifice as the pressure varies from low (laminar flow) to high (turbulent flow), and describing the viscosity of mixtures of disparate gases as a finction of composition. In all cases, using a theoretically well-founded mathematical model as a data template for the regression analysis gave me an excellent handle on the relevant parameters and the observational errors.
In the case of ‘fitting’ temperatures to a mathematical model, there are only four components to the model that must be considered: (1) a constant, (2) a monotonic secular change (ever increasing or ever decreasing, but not necessarily constantly so – best modeled as a consistently positive or negative rate of change to the temperature, (3) periodic variations in temperature describable with a Fourier analysis, and (4) non-periodic (chaotic) variations that cannot be accounted for with any combination of the above factors. If this last is significant, then any effort to predict the future data is doomed to failure.
Yep. So many of them use such glaringly inappropriate models. The conclusions of such studies are almost uniformly junk.
Well so we have some exciting new mathematical games we can play with a given data set of EXACTLY KNOWN NUMBERS.
I emphasize this point, because the NUMBERS in a data set ARE in fact exactly known numbers.
This make no claim as to whether those numbers are actually exact values of any physical variable or quantity; only the numbers in the data set are exact. They may not even refer to anything real.
So fine. Have fun with your new mathematical prestidigitation.
Just remember, your results tell us EXACTLY NOTHING about ANYTHING ELSE except that particular set of numbers in your data set.
Statistical mathematics is strictly non-predictive of anything outside the data set.
Rather it is the natural (and perhaps cyclical) variation around a warming trend that has remained constant at ~.008 °C/decade2 since the late 1800s. There is no empirical evidence from the temperature record that mankind has had any effect one way or the other.
Was there a change in the late 1800s?
The Annals of Applied Statistics
2014, Vol. 8, No. 3, 1372–1394
DOI: 10.1214/14-AOAS753
© Institute of Mathematical Statistics, 2014
CHANGE POINTS AND TEMPORAL DEPENDENCE IN
RECONSTRUCTIONS OF ANNUAL TEMPERATURE:
DID EUROPE EXPERIENCE A LITTLE ICE AGE?
BY MORGAN KELLY AND CORMAC Ó GRÁDA
University College Dublin
Some of us think so 🙂 http://wattsupwiththat.com/2013/10/04/the-great-climate-shift-of-1878/
Rather it is the natural (and perhaps cyclical) variation around a warming trend that has remained constant at ~.008 °C/decade2 since the late 1800s.
Which is of course a section of the positive phase of the ~1,000 year cycle that is responsible for the Minoan, Roman and Medieval Warm periods and the concomitant cold periods such as the Dark ages.
The late 20th century warming 1970-2000 that has so excited the AGW fantasists is the positive phase of the ~60 year cycle that correlates quite well with the North Atlantic Oscillation, and is statistically practically identical with the warming period 1910-1940, followed by the 30 year cooling period 1940-1970 – both of which periods are a considerable embarrassment to the Warmies, who claim that the advent of detectable anthropogenic influence on the Earth’s temperature was 1940.
So there has been no detectable anthropogenic influence whatsoever on the Earth’s climate – not by greenhouse gases nor by sundry aerosols – in the measured temperature record, and the billions – trillions? – of dollars, pounds or whatever that has been wasted on this utterly non-existent threat will shortly be recognised as the biggest “scientific” mistake in all of human history.
And you can take that to the bank.
I’m confused. The figures seem to show the conclusion the post is disputing. Fig4. shows the sliding slope (deg/time) of a ~50 year period average as a function of time. We see it agrees with Fig3: The climate was cooling (negative values in Fig 4) in the 1800s and warming since 1900 (positive values.) If you used a shorter trend period you might find more warming and cooling periods but the overall picture would be the same. Around 1960 the slope almost gets down to zero because of the flat or cooling trend from 1940-1970 in Fig 3, but is still positive on a 50 year average.
In Fig5 we see the linear fit to the slope as a function of time, i.e. the average slope of the slope, the average second derivative. The linear fit has a positive slope, indicating that on average the warming is accelerating. If it were constant then the linear fit would be a horizontal line (above the x-axis for warming, at zero for no change.) This is consistent with the ‘instantaneous’ slope which is more positive near the present than in the past. There are sizable oscillations around the linear trend, but the trend is there. I don’t see how you can argue based on these figures that the climate scientists are wrong. You could try and show that the linear best fit isn’t statistically significant somehow, but that is not shown here as far as I can tell.
Yes, if the results presented here show anything, its that the rate of warming accelerated in the 20th century!
So the conclusion should be that it is “incorrect to term the nearly 20 year slowing in the rate of warming as a pause. Rather it is the natural (and perhaps cyclical) variation around a warming trend that has accelerated at a constant rate of ~.008 °C/decade2 since the late 1800s.” (what else has accelerated since the late 1800’s I wonder…)
I don’t know of anyone who has seriously suggested that AGW started at the turn of the 20th century but lets pretend that’s the case and that the trend will persist unabated. If you extrapolate the parabola implied by the ~.008 °C/decade2 you get to warming <2 degsC by 2100. Since that is below the IPCC target, can we declare victory and go home?
The increase in the rate of warming has been constant and at the same value since long before GHGs could have possibly had an impact.
Your figures don’t show that. They show that the 50 year trend has considerable variation. If you model that with a straight line fit you assume a constant 2nd derivative. From your graphs we can’t tell what the 3rd derivative might be and I suspect we really would be pushing into areas of too little data/noise to measure a constant 3rd derivative. You could bound it though.
You say GHG couldn’t possibly have an impact but I don’t see how you can assume that. Industrialization was well under way by the late 1800s. You can crudely fit emissions with a parabola too. Obviously the early ones are too small to have a measurable impact under other variations by themselves, but fitting emissions and temps with a parabola doesn’t disprove AGW.
Of course, the proper way to do this isn’t to arbitrarily fit different measurements with trends, but to understand the physics that links them, model it, and see how consistent the models are with reproducing observations. The only thing you get from these graphs is: temperatures are increasing and the warming is accelerating on average so far, since the late 1800s.
” … ironically I actually agree with the author’s contention that all this talk of a pause is gibberish. The fact of the matter is that there has been no statistically significant increase in the rate of warming over the entire observable temperature record. ”
Quite right .
It’s hard to get across what a near noise level molehill this whole politically useful hysteria is all about , a change of just 0.3% in our estimated mean temperature which is only about 3% warmer than the about 279K ( varying +- 2.3K from peri- to aphelion ) of a gray ball in our orbit .
tadchem December 8, 2014 at 11:35 am wrote :
“The flaw in nearly all flawed ‘regression analysis’ studies is made before a single calculation has taken place. The most greivous error is in the selection of an inappropriate mathematical model. If the physics of the process is not reflected in the equations of the model as used for calculation, only self-delusion can result.”
Quite right .
You can do all the stats on all the observational data you want , and you still will not understand the planet’s temperature until you understand the physics I come at this as an APL programmer for whom to understand a quantitative relationship is to implement it so I can explore its parameter space
But even at this most basic level “climate science” parts paths with the classic quantitative analytical method of physics . Because it has to , because its nakedness , if not outright fraud , is quickly apparent .
But if the ~60-year cycle held (cool circa 1900, warm in mid-1930s, cooling to ~1978 when alarmist types were flapping about doom from global cooling, warming to ~1998) we should be halfway down the cooling part of a cycle.
But…. there may well be several influences of different periodicity, that will reinforce or counteract depending on where each is in its cycle at a particular time (i.e. now).
Of course your figures don’t cover recent years.
The “peak” you mentioned is a broad span of ten years – not a “point” at 1945 (up from 1910, down to 1976) but a slow sinewave-like rise from a low broad valley in 1910 to a broad curve centered about 1945. Then a slight down curve towards another valley in 1976, then a broad rise up towards the Modern Warming Period flat-topped curve between 1998 and 2010-2012. Yes, if you think of a thirty five year half-wave, we’d be halfway along (1998->2014), but that’s not the curve we follow.
And, if one assumes the Modern Warming Period is a long 200 year near-plateau between the 1000 year low point of 1650 and the Modern Ice Age of ~2500, then today’s broad MWP will probably stay near the 2000-2010 level for quite some time.
HotWhopper wrote a disparaging critique of this post. I have posted a response here (http://alturl.com/5qwnv) which also addresses similar comments above.
Jeff Patterson