The subject came up — yet again — whether or not there is a “pause” in global warming recently. Specifically:

Re looking at global warming over the last 15 years. I know it breaks all the “statistical” rules, but just simply eyeballing the graph of (annual average) temperatures shows a distinct change in the trend of global temperatures after 1998. In order to avoid a charge of “cherry picking” I could have said 10 years since 2002 – but to me the break in the trend is pretty clear.

We’ve often dealt with this subject, but since it’s so common, and seems to come from real skeptics as well as fake ones, we’ll address it once again. In depth.

Scientifically the best answer is to look at the other factors which are known to affect temperature and remove an estimate of their influence, better to isolate the trend from the fluctuations. This was done here. Frankly, it’s just plain impossible (really — it’s impossible) that these known factors do not affect global temperature. Their influence over the last decade was examined here. If you haven’t already read that post, it’s a good idea to do so.

But the best scientific answer isn’t always the most persuasive for the nonscientist. So for the moment, let’s ignore all that stuff, not try to account for other (known) factors, just look at the temperature data au naturel.

We’ll examine three global temperature records: GISS (from NASA) because I believe it does the best job estimating the Arctic (which is just about the fastest-warming region on earth); HadCRUT3v because the CRU record was specifically mentioned and unfortunately the new improved HadCRUT4 doesn’t yet go beyond 2010, and UAH because even though it doesn’t start until 1979, it’s a satellite record, and since it’s the work of Christy & Spencer at UAH nobody can even suspect that they’ve “doctored” the record to inflate global warming.

We’ll start with GISS. Here’s the whole thing:

Some things are easy to see and are actually correct — like the overall rise, and the consistent increase since 1975. But to many eyes there seems to be (at least possibly) a levelling off recently. This is more visually apparent if we just plot the post-1975 data:

Of course, there are a lot of ups and downs that nobody takes seriously as a trend reversal. The speedy decline from 1998 to 1999, for instance, covers about 0.4 deg.C, but it’s such a short-term fluctuation that everybody sees it for what it is — a short-term fluctuation. But that whole post-1998 stuff, is that one of those natural fluctuations that can so easily fool the eye, or is it a genuine sign of trend reversal?

Let’s see how the behavior pre-1998 really compares to that post-1998. We’ll use just the data from 1975 to 1998 to estimate the trend, then we’ll extrapolate that trend up to the present. Here’s the result (estimated trend in red, extrapolated trend as a dashed line in blue):

Now let’s use the data from 1998 to now to estimate the more recent trend, and see how it compares to the extrapolation (recent trend estimate also in red):

Interesting! The estimated trend rate post-1998 is less than the estimate for pre-1998, but then we already knew there’ll be random fluctuation. But the mean value of the post-1998 stuff is well above what the pre-1998 trend would have predicted. Even if the trend rate decreased, there’s no basis to say that it actually cooled, not even relative to what was expected from the pre-existing trend. In fact, if I estimate the trend for the entire time span 1975 to now, it has a higher warming rate than either the pre-1998 or post-1998 sections (solid blue line):

Ok — so it sure didn’t cool relative to what was expected. But did the trend rate actually decrease after 1998, or not? I honestly don’t know how to give major visual impact to the answer — I can just crunch the numbers. I did so, computing the trend 1975-now, 1975-1998, and 1998-now, as well as 95% confidence intervals for each. Here they are:

Answer: there is no evidence that the trend rate was any different after 1998 than before.

OK that’s GISS. How about CRU? Here’s the HadCRUT3v data since 1975:

Let’s do the same thing we did with GISS. Here’s the pre-1998 trend (in red) extrapolated to now (blue dashed):

And here’s the estimated post-1998 trend (also red):

The trend rate sure looks different — but looks can be deceiving. But notice that once again, the mean value isn’t lower than was predicted by extrapolating the trend, it’s higher (but not by a statistically significant amount). And once again, the trend over the whole time span is faster than either segment separately (solid blue):

In the mean, it sure didn’t cool off or even slow down. But what about the rates? Here they are:

Once again there’s no statistically significant disagreement. It’s “flirting” with significance, but not as much as it might look — remember there’s uncertainty in all three estimates. And don’t forget this: that CRU is known to underestimate the global trend — especially recently — because it leaves out the fastest-warming region on the planet, the Arctic.

Lest we neglect satellite data, here’s the record from UAH:

We’ll do the same thing again. Here’s the pre-1998 trend (red) extrapolated to the present (dashed blue):

Here’s the post-1998 trend superimposed (also red):

What a difference! The trend rates are about the same, but the mean value is a lot higher. Really — no cooling there. And again, the overall trend is faster than either subsection (solid blue):

As for comparing trend rates, here ya go:

No evidence of a difference.

What’s going on here? All three records share these properties:

– No significant difference in trend rate pre-1998 to post-1998
– Mean value post-1998 at least as high as predicted by extrapolating pre-1998 trend
– Higher rate 1975-now than either subsection separately

Why is that, really? It’s because the separation time, 1998, was chosen because it gives the visual impression of a change, and that’s because of the extreme warmth of 1998. And that’s because of the monster el Nino that year (one of those “known factors”). That makes it worth investigating, but it also means that we should expect lower trend rates both before and after — before, because that interval ends just prior to a high point, and after because that interval starts with a high point. A monster of a high point.

But seriously, can random fluctuations really create a 15-year time span with a negative (albeit not significantly so) trend estimate? I generated some artificial data to mimic the behavior of the CRU data. This is complicated because the data show strong autocorrelation, and it’s not even the simplest kind of autocorrelation (AR(1) noise), it’s a lot closer to ARMA(1,1) noise. So I ran up 100 years of ARMA(1,1) noise with the same autocorrelation as HadCRUT3v data, gave it the same standard deviation as HadCRUT3v data, and added the same trend as HadCRUT3v data. I didn’t do lots and lots of data sets until I got what I wanted, I just did a single run of 100 years. Lo and behold it had a time span with negative trend estimate, not just of 15 years, but 18 years long:

Yes. Random noise really can create 15-year (or longer) stretches with negative (albeit not significantly so) estimated trend rate.

I hope you’ve enjoyed the free meal, but now it’s time to pay for the soup, meaning, endure my little “lecture” (get the Star Trek reference?).

This analysis isn’t simple. Far from it. We compared what would have been expected to what happened. We estimated the uncertainty in estimates. Even that wasn’t simple — we had to compensate for autocorrelation, and not the simplest kind of that. We even briefly contemplated the reasons for the results, both statistical (how does splitting the data at a monster high point affect things?) and physical (what’s the impact of leaving out the fastest-warming region on earth?). But hey, if you really want to get at the truth, if you aspire to deeper understanding than can be had just scratching the surface, that’s what you have to do. It’s called “science.” It works, bitches.

It also takes work.

We didn’t just say “Hey look — it sure looks like the trend changed!” That’s a very natural thing to do. It’s an extremely persuasive argument with non-scientists! A vast amount of scientific experience has shown that it’s also a great way to get the wrong answer.

But for some people, it’s the only approach they’ll accept. That’s not because they couldn’t understand the science if it were properly explained in layman’s terms. It’s because they’re not willing. Do you really think that James Inhofe will invest the effort required to plumb this issue to its depth? Even if he did, do you really think he’d believe it? Or would he refuse to budge from “Hey look!”?

50 responses to “Fifteen

  1. Figure 7 from your paper is illuminating: given the big dip in TSI, we’d expect to see a flattening of the short term trend.

  2. alexandriu doru

    Here i propose a way to disprove the meme”global warming stopped in 1998″
    I put the GISS LOTI trends(via “wood for trees”) in K/decade:
    Start year: 1960 1965 1970 1975

    end year: 1990 .0936 .1290 .1562 .1871
    1995 .0958 .1195 .1320 .1340
    2000 .1160 .1385 .1533 .1625
    2012 .1364 .1521 .1620 .1674

  3. Nebuchadnezzar

    “Yes. Random noise really can create 15-year (or longer) stretches with negative (albeit not significantly so) estimated trend rate.”

    Presumably, if you let it run on long enough, it will generate ‘significant’ negative trends. Significant being, in this case, synonymous with ‘quite unlikely’.

  4. Give them Hell!!! :)

    You keep amazing me. Great analysis. Like trying to explain stuff to toddlers…That’s their level!!

  5. I expect that short term climate variability will be all the rage from those that skepticize when the 2012-2013 ENSO event really kicks in. It is not officially an ENSO yet because that is defined by temperature anomaly in the central Pacific. The ENSO dynamics are working right now though.

  6. It is probably worth noting that HadSST2 has a discontinuity at the start of 1998 presumably due to a switch from ICOADS to NCEP/NRT, which I document here. It may well also impact GISTEMP through the use of the HadISST dataset.
    I’d guess the magnitude of the global impact to be about 0.025C, probably not enough to change your conclusions.

  7. Thanks for the rigorous analysis Tamino.

    With regard to arguments that a layman with little statistical expertise (like me) would understand, I would say that it’s important to mention that the surface and lower troposphere only hold a small proportion of the heat content of the climate system – the vast majority resides in the oceans. Doesn’t this mean that GISS, UAH etc. are inevitably going to be more noisy than the heat content of the whole climate system?

    Also, an El Niño period in GISS/UAH looks like a period of warming but really means that the oceans are losing heat to the atmosphere, so the climate system as a whole is actually cooling, or at least that it’s gaining heat more slowly than it was before. Conversely, a La Niña looks like cooling or a slowdown in warming, but the climate system as a whole is gaining heat faster than in El Niño or neutral years (all other things being equal).

    What’s interesting about the supposed pause or slowdown in global warming of recent years (according to surface/lower troposphere data) is that it doesn’t seem to be evident at all in the ocean heat content data, if Levitus et al 2012 is to be believed. If anything, the oceans to 2km have been gaining heat significantly faster in the last decade than in the previous one. Is it possible that beyond ENSO there are longer timescale processes causing accumulated heat to be directed more into the oceans than into the rest of the climate system in recent years? That would reconcile the different impressions from GISS/UAH vs ocean heat content data.

    • I’ve always wondered about this. La Nina has been dominating El Nino during the first years of the ARGO era. Doesn’t that mean downwelling of heat will be the presentation?

  8. Jim Pettit ("Neapolitan")

    The Denialist Conjecture: “The planet hasn’t warmed in x years” (where x = [current year – 1998]).

    Obviously, we’ll have to live with this line of stellar “reasoning” for a little while longer, until the planet warms enough via CO2 alone or in concert with another El Nino event to top 1998. We’re getting there–though, of course, that will then establish a new denialist baseline: “The planet hasn’t warmed since 2013 (or whenever)”.


    Anyway, thanks for the most excellent deconstruction of this very prevalent “argument”. Bookmarked. Not that I expect it to sway any feeble and hardset Inhofe-ian, Foxed-up minds…

    • climatehawk1

      At least we will get a few years respite, which would be nice.

    • “We’re getting there–though, of course, that will then establish a new denialist baseline: “The planet hasn’t warmed since 2013 (or whenever)”.”

      the difference is next time we’ll be able to point out their past track record of getting it wrong wrt 1998.

      • …….

        best of three?

      • Assuming “best of three” means let’s not decide after one that didn’t go my way – we’ll see who “wins” after the next (either a tie, or you’re now up 2-0)

        …but with the fake skeptics, best out of three will be followed by “three out of five”, and “four out of seven”, on to (n+1) out of (2n+1), where n is the number of failures so far…. Anything to delay or avoid…

  9. Horatio Algeranon

    Something people sometimes forget is that there are actually two aspects to a trend line: the slope (which everyone considers ad nauseam) and the intercept (which is rarely considered, at least on climate blogs).

    If it were really true that there had been “a distinct change in the trend of global temperatures after 1998”, the pre- and post- trend lines should still somehow “match up” in the year in which the supposed ‘change in trend” began.

    In other words, quite unlike the situation shown in the graphs above, the long term and short-term trend lines should show ‘continuity” (no jump) in the year where one trend line ends and the other begins.

    One can “force” the continuity requirement by doing the linear least squares analysis for the recent data (eg, post 1998) with the intercept of the recent (short term) trend line fixed so that the line passes through the last point (for 1998) on the succeeding (long term) trend line. (in other words, so that in 1998, the two trend lines have the same y value)

    If one imposes this requirement on recent short term trends (since 1998, 2000, 2001, 2002, etc), one gets a trend slope for the short-term trend that is much more in keeping with the slope of the preceding long term trend.

    One might argue that if there is a real “step” in temperature in the year where the long trend ends and the short begins, then the assumption about the trend lines “matching” up won’t be valid.

    True enough.

    But these apparent discontinuities between successive trends are very common and it’s simply not plausible that whenever we see them there is a real step.

    When one does linear regression on recent trends without regard to the trend of the preceding period, one is effectively throwing away a lot of valuable information which (absent real steps) “constrains” the subsequent trend (with regard to both intercept and slope). In fact, the last data point of the preceding long term trend contains “years” of information.

    • One can “force” the continuity requirement by doing the linear least squares analysis for the recent data (eg, post 1998) with the intercept of the recent (short term) trend line fixed so that the line passes through the last point (for 1998) on the succeeding (long term) trend line.

      Wouldn’t the best way to constrain for continuity be to find the least-squares best fit for the slope before, the nominal 1998 temperature and the slope afterwards? I.e., let the data before and after 1998 both contribute to determining the intercept point. The maths of that is just beyond my grasp but it seems to me that it ought to be a simple generalisation of normal least squares adding just one more parameter to the solution.

      Generalising to multiple breakpoints seems the obvious next step: 1940, 1975, 1998 or whatever. For n breakpoints the solution would have n+1 slopes and n nominal values.

      • Horatio Algeranon

        The above is not really intended as a method to determine a breakpoint.

        Instead, it’s meant to take the claim that there IS such a breakpoint to its logical conclusion and show what that would actually mean about the short term trend that follows the long term trend ie, about the constraint placed on the short term trend with regard to intercept.

        If one claims that “global warming stopped[or slowed] in 1998”, the presumption is that the temperature started a new “trend” in 1998 or shortly thereafter (either flat or downward)..

        IF that is true, then (barring an actual “step” in the year 1998 –ie, for which temp jumped up and stayed up), logic dictates that the two trend lines still “match up” in 1998.

        It’s hard to believe that NONE of the folks making the claim that “warming stopped (or even slowed)” in a particular year appreciate what that means with regard to the relationship between subsequent (long and short term) trends.


        NONE of these folks (some of them quite competent mathematically) has noticed how much the 1998 coordinate (for example) on the trend line “jumps” when one goes from the long to the short term trend lines? (or indeed, that the amount of this jump can change rather dramatically just by changing the starting year of the short trend slightly?)

        In fact, it is rather curious that the “step” meme seems to have become popular recently in some circles. It’s really the ONLY way that one can avoid the logical requirement of continuity between subsequent trend lines.

        PS The requirement that the short term trend line actually pass through the last point of the long term trend line (ie, that the intercept of the short trend be fixed) could be relaxed a bit by doing a weighted least squares on the recent short period and instead of fixing the intercept, merely include in the analysis the last point on the long-term trend line, (heavily) weighted by the number of points in the long term period.

      • Horatio Algeranon

        After reading your comment again, it appears that the method you describe is the better one, but you’d best ask the resident expert.


  10. “She was right. But at the wrong time.”

  11. First of all I understand the argument over which should be the appropriate temperature series to use. I must admit to being surprised at how much they differ in their short term trends. As I state elsewhere I use the HadCrut3 data simply because I always have done and am familiar with where to get it and how it is formatted and updated. HadCrut4 is not yet a fully developed data set but I am more than happy to use it if it becomes more regularly updated. I also like the fact that HadCrut3 goes all the way back to 1850 – although I wonder how reliable it is prior to the WW1 (the error bars are pretty impressive!).
    Yours is a very interesting analysis, showing technical and mathematical skill that is way beyond me. It is very convincing in its aim of proving that the current (apparent) hiatus in warming could be merely a random fluctuation. The limit to that approach is, of course, that just because something could be random does not mean that it is. Turning points can look random to begin with, apparently similar to many that have gone before, but only with hindsight does their significance become apparent.

    [Response: Of course it *could* have done just about anything — including begun warming at 100 deg.C/yr, last Thursday. But that doesn’t seem very plausible, does it?

    The question is: what is the evidence? The evidence is that there’s no statistically significant sign of a change in the trend. The most likely interpretation, statistically, is that the trend hasn’t changed. As for changes which are not significant but are visually suggestive, those are not just possible — they are *inevitable*. If the “too short term” trend didn’t fluctuate all over its statistically possible range, *then* I’d think something unexpected was happening.

    That the short-term trend estimate should be steady despite the presence of noise is, frankly, impossible.

    And as for the reality of those “other factors” that influence the short term trend, did you read this?

    Of course, statistics can only tell us so much. For further insight we should appeal to the laws of physics.]

    I only have one major question, why do you start your analysis in 1979 when there is data available going back to 1850 (from UEA) and around 1880 for GISS?

    [Response: I begin in 1975 (not 1979) because that’s a “turning point” in the trend. The satellite data analysis begins in 1979 because that’s when the satellite data begin.]

    I would be interested in your views on the following charts. These show a rolling 50 year linear regression applied to the entire HadCrut3 data series followed by the same graph but overlaid with the 30, 20 and 15 year rolling linear regression lines.

    [Response: That’s another post altogether. For the moment I’ll only say that it’s another case in which relying on the “natural” interpretation based on visual impression is a terrific way to get the wrong answer.]

    Since I did this, reading one of the links from a commentator here has made me aware that I should be placing each point at the centre date rather than the last date. This would solve the “offset” problem with the shorter LR sets. I will do a version that way as well at some point.
    The 50 year line looks like it follows a 60 – 63 year cycle (trough to trough) of rising and falling rates of warming with a gently accelerating rise in the rate. In each case the shorter LR line shows a distinct dip in recent years, about where you would suspect the rate of rise to start to decrease if there really is a 63 year (or so) pattern. Unfortunately we only have about 160 years of data. Without at least two peaks and two troughs you cannot definitely infer a cycle, the shortage of data limits our ability to do this so we will just have to wait and see.
    What stands out to my untrained eye is a drop in the 15 year LR line to zero and falling, the first time since the mid 1970s and the sharpest fall since the 1950s. I am aware of all the arguments about trends only being significant for a 30 year period, but a significant shift in the shorter term trends can sometimes herald a change in a longer term trend – so it would be rather short sighted to completely ignore shorter term data.
    Even if the rate of warming does take a 30 year dip, there is still a clear long term acceleration in the rate of warming (the waves in the 50 year LR line are still pointing upward).
    I might see if I can get hold of the GISS data and try the same analysis on that. However, with a demanding day job, wife and three kids time is rather short so I don’t expect to have that done particularly quickly.
    I would stress that I am not saying there is a pattern, just that it looks like there could be one. I have no idea of any mechanism that would drive such a fluctuation, but multi-decadal changes in ocean temperatures have been cited as a possible cause. I am doing the opposite of science – looking at the data without any underlying hypothesis, but there you go – I am no scientist!
    Personally I find your analysis removing the sun, volcanoes and ENSO more convincing than this one (as I am sure you also do!). I have some questions on that, but that is a discussion for another day.

    [Response: There’s a “Climate Data Links” page which will point you to data sets (like NASA GISS data).]

    • Phil Scadden

      TLM. Do you agree with the analysis that ENSO, Volcano and sun do impact the temperature record (visually, statistically, and more importantly, physically, we expect this)? If so, then surely you should account for variations due to those sources before you look for any other unexplained variance. The remaining large-scale character in the record looks to be readily accountable from anthropogenic forcings which is what we expect from purely physical considerations. If you are going to invoke as-yet-undiscovered oceanographic cycles or similar, then you also need to account for why the expected physics of GHG is not affecting the record.

      • TLM. Do you agree with the analysis that ENSO, Volcano and sun do impact the temperature record (visually, statistically, and more importantly, physically, we expect this)? If so, then surely you should account for variations due to those sources before you look for any other unexplained variance.
        Yes, absolutely. I would be interested in seeing an attribution analysis using these three “non-anthropogenic” variables going back as long as we have data – particularly if it can cover the cooling period 1940 to 1975, so we could see what, if any, residual trend this throws up.

        I know there is a view that the cooling from 1940 – 1970 can all be accounted for by aerosols, but due mainly to the lack of solid data on the quantity and effect of aerosols on the temperature record, I am “sceptical” (but you already knew that!) and would need some convincing.

        [Response: Data are sparse but not absent. See this. The cooling influence of aerosols was effectively proved by the accurate predictions (made before observations) of the cooling impact of the Mt. Pinatubo eruption.]

        If you are going to invoke as-yet-undiscovered oceanographic cycles or similar, then you also need to account for why the expected physics of GHG is not affecting the record.
        Agreed. I wish I had the skills and education to do that. By the way, are there not already “discovered” oceanographic cycles such as AMO, PDO and ENSO? Has there been any research that looks at their long term, historic, effect on the global temperature record?

      • TLM: are there not already “discovered” oceanographic cycles such as AMO, PDO and ENSO? Has there been any research that looks at their long term, historic, effect on the global temperature record?

        BPL: Plenty. Granger causality runs from surface temperature to the AMO, not the other way around. ENSO has no effect on the trend, just annual variations. The PDO isn’t really an oscillation, despite its name.

  12. First of all, thanks again for the long reply. I am aware you don’t have to do this – I just hope that it is as interesting and helpful to others as it is to me.
    [Response: I begin in 1975 (not 1979) because that’s a “turning point” in the trend. The satellite data analysis begins in 1979 because that’s when the satellite data begin.]
    Do we know why the turning point is there?
    Have you (or anybody else) run a similar analysis between the 1940 turning point and the 1975 turning point?
    Or the 1910 turning point and the 1940 turning point?
    Is it possible to estimate how much of the cooling 1940-1975 was due to the sun, Volcanoes and ENSO?
    Do we have the data to do this?

    I am not expecting all of this from you but pointers to some papers or web sites elsewhere would be fine.

    I am quite prepared to be convinced that these apparent visual patterns in the record are either illusory or can be explained by known factors. I have not read the IPCC reports cover to cover so I would not be surprised if you told me that it is in there somewhere already!

    • TLM: Is it possible to estimate how much of the cooling 1940-1975 was due to the sun, Volcanoes and ENSO?

      BPL: Yes. Here’s an example.

      Check out the analysis of variance at the bottom of the page.

    • Phil Scadden

      TLM – You can see here for more discussion whether it is PDO. Invoking ocean cycles to explain a trend implies that there must be a long term transfer of heat from the ocean to the atmosphere to satisfy conservation of energy. Instead we see the ocean heat content steadily increasing. The challenge though was really about what could negate the known physics of GHGs.

      I would say it is imprudent to be looking for far-fetched explanations for the surface temperature record when there is a perfectly obvious explanation in AGW which is consistent with known physics.

  13. One point that has interested me about the various intellectual stances that have been labelled “skeptical” is: what is the default position? For example, one common version of skepticism seems specific to claims of supernatural activity. The default position in that case is that there is as yet no convincing evidence of supernatural activity, and therefore any new observations, even prior to analysis, will be assumed to have scientifically describable causes.

    With global warming skeptics, the default seems to me to be less grounded in scientific history. They seem to treat the idea of global warming the same way a skeptic of the supernatural treats a ghost story, despite the fact that the vast majority of the scientists studying the field tell us otherwise. So this form of skepticism isn’t a preference for science over “woo-woo,” it seems instead to be an assumption that science *is* the “woo-woo.”

    People who find themselves ostensibly rational and scientific-minded, but “skeptical” of “AGW,” then end up having to do some weird contortions to stay on the side of, for example, James Inhofe, versus the large majority of climate scientists.

    For any James Randi-type skeptic, doesn’t the default have to be that the scientists are more likely to be right than the political think-tank when the two are at odds?

  14. Tamino notes that “Do you really think that James Inhofe will invest the effort required to plumb this issue to its depth? Even if he did, do you really think he’d believe it? Or would he refuse to budge from “Hey look!”?”

    Too true. That would also seem to apply to Dr. Roger Pielke Snr. and others who should know better, but are not willing to invest the effort to undertake a proper analysis. For them, it is much easier to guess estimate by eyeballing trends and let their bias condition the answer (that is, confirmation bias).

  15. John Nielsen-Gammon has an interesting take on how to look at the resent warming trend, or trends – he plots trends for El Nino type years, La Nina type years, and Neutral years –
    He has a followup where he tries varying assumptions here

    • Re John Nielsen-Gammon, now that is a very clever, simple and convincing explanation! Separating the El Nino , La Nina and neutral years into separate series with their own trend lines makes perfect sense and does not need any general appeal to random variation. The climate may be complex, chaotic even, but it is not “flip a coin” random.
      Thanks for that, it has gone straight on my “favourites” list.
      Thanks to Tamino for the information on aerosols and BPL regarding the 1940-75 period. I feel that some of the gaps in my knowledge are being filled now and I am getting a better understanding of how the variables influence the trends.
      I hope others have found this exercise as useful as me and not an annoying distraction. There are lots of us out here genuinely looking for explanations who find that getting slapped down as a troll when they dare to ask questions dispiriting. This stuff is far from obvious and the information poorly organised on the web generally – so a polite response that treats the poster as an intelligent human being and points them in the right direction will more likely get them on your side.

      [Response: You’re quite right.

      But we do get genuine trolls too.]

      • TLM – Genuine queries such as yours are always more than welcome.

        I have to agree about the John Nielsen-Gammon blog – it’s clear and convincing. Foster and Rahmstorf 2011 was quite convincing to me, but I do have a background in multiple linear regression; the John Nielsen-Gammon post doesn’t need that level of mathematic background to show what’s happening.

      • Alex the Seal

        Good article but the first sentence is poorly worded. “global temperatures have not gone up very much” should read “surface temperatures …”. as his charts do not include oceanic heat.

      • Alex the Seal

        Scratch that – looks like they do??

  16. TLM asks “By the way, are there not already “discovered” oceanographic cycles such as AMO, PDO and ENSO? Has there been any research that looks at their long term, historic, effect on the global temperature record?”

    I’ve just done such an analysis. You can see it at:

    • Do be warned that the analysis linked to by Ron Manley, while it does look at AMO, PDO & ENSO, takes no account of ENSO (& PDO although that is more understandable) . Also it admits to entirely ignoring climate feedbacks, it mentions in passing delay between forcing & temperature changes but takes zero account of it, ignores strat.H20 & trop.O3 forcing as well as all anthropogenic negative forcing. It then proceeds to demonstrate how the remaining set of numbers (LLGHG forcing, TSI & AMO) used with zero regard to any physical climate processes can be then curve-fitted onto the HadCRUT3 temperature record and and the equations then used to predict future global temperatures. The large variance between this prediction and IPCC projections, according to the analysis, is because the IPCC is simply wrong & Ron Manley right. If that is not proof enough, Ron Manley’s prediction is also less scary and (at 6,000 words) his analysis is not quite as lengthy.

      • Al Roger. A lot of what you said is true and in the posting I mention most of your points as being potentially problematically. The regression model (though I generally try to avoid using the word ‘model’ for regression analysis) does though explain 88.7% of the variance in annual temperatures even without the detailed science.

        The key point, as you mentioned and as I recognised, is the apparent absence of water vapour feedback. If all the temperature increase from 1850 to the present was due to the increase in GHGs this implies a sensitivity of 1.1 °C per doubling of GHGs – the accepted sensitivity in the absence of water vapour feedback. So, either there has been no water vapour feedback from any of the periods of warming during 160 years, or part of the warming has in fact been due to water vapour feedback in which case the sensitivity is less than generally accepted. A couple of months ago I saw a reference to a paper which explained why water vapour feedback has been delayed. Can anyone point me to it?

        [Response: I’d say the key point is the lack of appropriate treatment of different thermal inertia of multiple components of the earth system. See this.]

      • Ron Manley
        You say at the top of you analysis:- “What I describe is a model which seems to suggest that … Whilst I can’t really believe it myself, I can’t find anything wrong with it. Maybe you can.”
        Well, yes I can. All you have demonstrated is that with a data set with increasingly rising values (it could be world population), plus a noisy data set (shipping lost in the Bermuda triangle is probably noisy enough) and yet another set with a 60-70 year oscillation (variation in Earth’s rotation or the change in solar system’s centre of mass give a good periodicity. But AMO gives period and noise!), you can reproduce a graph that near enough fits the HadCRUT3 record. Add a few more variables and you can make it trumpet Dixie.
        Why should it not be possible to do this? There are no constraints. H2O feedbacks, GHG forcing? The model doesn’t need them so, hell, they probably have no effect on world temperature. Predictions contradicting IPCC projections? Great!
        But you cannot simply ignore physical reality like this. And you cannot ignore what your model represents. Consider your own deliberations over which AMO series to use – why should AMO be 60% more powerful (while all the other factors diminish in power, aerosols by a whopping 70%!) just because the ESRL set is used instead of NCAR? That is plain bonkers!
        Why in your model should TSI have only 65% the power of GHGf? Why does you model show Pinatuba depressing temperature by only 0.05 deg C? And for heaven’s sake, why does it not account for the thermal inertia of the planet?
        You say you can’t find anything wrong with your model. I say I cannot find anything right with it!

    • RM, I’ve run a Sims partial-F test for Granger causality on a hundred plus years of AMO and dT data. The direction runs unequivocally form temperature to the AMO, not the other way.

  17. from, not form… it’s early…

    • In my posting I do recognise that the use of the AMO raises a number of issues. First of all I used annual values which minimises the potential problem of lags. For the four parameter version of the model the F value with AMO used with no lag is 332.9 and the explained variance is 88.7%. With the AMO shifted by one year the equivalent values are 215.9 and 82.5%. In other words with annual data and when the other parameters are included the affect of the AMO is virtually coincident. I also recognise that it is unlikely that the AMO itself is the driver; it is better to consider it as an index of an underlying pseudo-oscillation probably related to ocean currents.

      • “pseudo-oscillation”

        Apropos, for a pseudo-analysis.

      • A free down-load of my latest non-award-winning-composition to the first person to correctly name the person who coined the term “AMO”…

        Hint: it sure wasn’t Anthony Watts!

      • Jim Pettit ("Neapolitan")

        Oh, pick me! Pick me! ;-)

        Michael Mann is the culprit; he wrote about it in his excellent book The Hockey Stick and Climate Wars:

        “The multidecadal oscillation I’d helped discover would nonetheless become a cause célèbre among climate change contrarians. It would even get a name: the “Atlantic multidecadal oscillation” (AMO)—a moniker I coined off the cuff in a phone interview with science writer Dick Kerr. The AMO appeared to be real, and at least partly responsible for certain phenomena, such as the acceleration of recent warming in parts of the Arctic, that some had attributed to anthropogenic climate change. Other phenomena that have been blamed on the AMO, such as the increase in Atlantic hurricane activity in recent decades, arguably have nothing to do with it all. That hasn’t stopped climate change contrarians, however, from dragging out the AMO as a favorite catch-all explanation for just about any observed climate trend. At times I have felt like I helped create a monster.”

        I think Watts coined the phrase, “There’s a sucker born every minute,” didn’t he? Oh, wait, that was P.T. Barnum. Okay, well, same thing…

      • Horatio Algeranon

        name the person who coined the term “AMO”…

        Hint: it sure wasn’t Anthony Watts!

        That may be, but Watts does have an indisputable claim to “LAMO”

  18. Kevin Mc K. The first written used of AMO was by Richard Kerr in 2000 in an article in Science called “A North Atlantic Climate Pacemaker for the Centuries”. supports this attribution in their glossary ( In his recent book Michael Mann claimed that he suggested the phrase to Kerr in telephone conversation.

    Can I claim the prize?

    • Jim had the answer I was thinking of–and what a delicious irony it is, in view of all the misusers of AMO over the last decade or so!–but you certainly added valuable detail. Call it a tie.

      Gentlemen, write me off-line for an email to send you the music, and let me know if you prefer audio (mp3–be aware that this is a computer rendition, as the piece I have in mind is still awaiting its premiere) or music notation (PDF.)

      • He’s been hinting in his responses at RC that he wants to do an RC article on the AMO. Wish he would get it done.

  19. Thanks to those made comments. I’ll be modifying the posting to take account of them.

    I have recognised all along that the use of the AMO as a parameter is problematic. Teleconnections with it are well established but how these teleconnections are transmitted and what the index actually represents is not clear. I have always been aware that the AMO includes what, in terms of accuracy, might be called ‘beneficial noise’. That the regression equation accurately represents the 1998 peak temperatures shows that the AMO has an ‘imprint’ of ENSO. Using a smoothed AMO results in a larger coefficient for Aerosol forcing (-0.78 v -0.28) so the measured AMO also carries an ‘imprint’ of volcanic effects.

    That said, the fact the regression equation, despite its defects and despite the fact that its parameters do not detect any sign of water vapour feedback, gives a better fit than the AOGCMs to the 1910 to 1945 period, and its 1980 forecast up to 2011 was more accurate than AOGCMs run around 2004, gives pause for thought.