No Warming for Fifteen Years?

The recent cooling shown in most global temperature datasets except GISS has been a spot of contention for the AGW believers. Global cooling has been an embarassing topic for those who have made careers of predicting worse than we thought warming every few weeks. This is especially contentious point for scientists who’s careers depend on warming.

A comment sent to me by email pointed to this post by Dr. Eric Steig on another blog.

21. At 9:42pm on 09 Nov 2009, Eric Steig wrote:Paul,

It is well known this “warming has stopped” idea is a complete red herring, as we have patiently explained here:

and here:

Your article merely promotes further confusion and misinformation by adding the mark of credibility that “BBC meterologist” provides. The preceding comment by “timjenvey” is a typical example characteristic of the misuse of the data that is the result. (Why does he choose October only, for example?)

Now Dr. Steig is mostly correct in my opinion, and I don’t disagree with him on this. However, the argument should have been warming did pause – a little, it simply doesn’t prove anything about CO2 driving climate. But the pause has caused some red faces WRT computer models – see Lucia(the Blackboard) and Chad (treesfortheforest). I am a believer in some level of global warming but the arguments used in the links above cut both ways.

Here is a quote from Real Climate:

The mean of all the 8 year trends is close to the long term trend (0.19ºC/decade), but the standard deviation is almost as large (0.17ºC/decade), implying that a trend would have to be either >0.5ºC/decade or much more negative (< -0.2ºC/decade) for it to obviously fall outside the distribution. Thus comparing short trends has very little power to distinguish between alternate expectations.

So, it should be clear that short term comparisons are misguided, but the reasons why, and what should be done instead, are worth exploring.

The point Real Climate makes is that the trends are inside the statistical significance limits of the trend so the argument that global warming has stopped is false. Tamino actually took it even further with this post about Bjorn Lomborg where Bjorn made the true statement below.

Temperatures in this decade have not been worse than expected; in fact, they have not even been increasing. They have actually decreased by between 0.01 and 0.1C per decade.

Tamino makes the same claim as RC but somehow misses the point that temps actually are NOT increasing while using some math to show that the lack of increase doesn’t disprove the long term trend. His point reads that statistical short term noise has an effect on the trend signal so the trend isn’t real. Actually he’s wrong and what it means is that of course the trend is real but that the short term trend means an increased probability of pausing of long term trends but doesn’t prove them false. As above, the great thing about this approach Real Climate and Tamino have taken is that it cuts both ways. If the negative trend isn’t significant and doesn’t count, that also applies to the positive trend. — so let’s have a little fun with the team.


The plot works like this – The thick black line is the trend from x year to present in degrees C per decade. The Red line is the lower confidence limit and the blue is the upper – calculated with corrections for AR1 autocorrelation Quenouille style as used in Santer et al. If the black line stays between the confidence limits, it represents a statistically insignificant trend. The code for the calculation is shown here:

### Get trend
fm=lm(window(dat, st, en)~I(time(window(dat, st, en))))

### Initialize variables
N=length(window(dat, st, en))

### Calculate sum of squared errors

### Calculate OLS standard errors

### Calculate two-tailed 95% CI

### Get lag-1 ACF
resids=residuals(lm(window(dat, st, en)~seq(1:N)))
r1=acf(resids, lag.max=1, plot=FALSE)$acf[[2]]

### Calculate CIs with Quenouille (Santer) adjustment for autocorrelation

So in Figure 1 black line crosses below the upper significance level (blue line) in 1995 and even touches the lower limit (red line) in 2001. So using our climate scientists arguments applied to HadCRUT show there has been no significant warming in the past 15 years (since 1995). It just clears the line in 1994 so 16 years is barely significant.

Below is UAH:


No significant warming since 1993 – 17 YEARS!


RSSNo significant warming since 1993 17 years – no significant warming!!

and GISS.

GISSOnly 11 years. In my opinion GISS has become an embarassment with it’s corrections but people don’t know it yet. I hope to live long enough to see the mess corrected (go surfacestations) but recently a prominent GISS scientist publicly promised future warming would continue and would cancel this inconvenient downward trend. The GISS trend is completely out of whack in recent years, so it looks like GISS is coming through on their promise.

What we do know from this is that despite massive increases in industry and constant worse than we thought bombardment from the press despite the need to spend trillions on socializing industry, three of four measurements show no significant global warming for the last 15 years and came very close to clearing the 17 year mark.

The title could have been: Real Climate and Tamino Claim No Global Warming for Past Fifteen Years!!

I wonder if Real Climate and Tamino will want to change their opinions?

37 thoughts on “No Warming for Fifteen Years?

  1. My impression is that RC and Tamino’s method of computing the uncertainty intervals due to “weather noise” generally makes then larger. That would result in no statistically significant warming for a longer periods.

  2. Jeff,

    Interesting post. I don’t think RC and Tamino will agree with you (even though you are of course correct). They already “know” what the truth is…. in a religious sort of way.

    The black line represents the “best estimate” of the true trend, even if the trend is not significant at 95% confidence. So for all four temperature metrics (RSS, UAH, Hadley, GISS) the best estimate for the trend is negative starting in 2001, and for all but GISS, negative since 2000. Calculating the limits for 90% confidence might be worthwhile, since it would demonstrate that at least some of the negative trends starting after 2001 are “real” at 90% confidence. (I would try this, but I do not have nor know R.) When the actual temperatures are compared to the IPCC projections of “about +0.2 C per decade”, actual trend already deviates from the IPCC projections at 95% confidence staring in about 2001… the first year they made the “0.2 per decade” projection!

    The ongoing El Nino may change the negative post-2001 trends of course, but if El Nino dies off soon, then negative trends of longer than 10 years will continue to make the IPCC projections based on GCMs look bad. At some point most people will say to the IPCC “These models can’t be accurate”.

  3. This episode reinforces a long-held impression of mine. What I am sure the guys at RC haven’t yet realized (who’s Tamino anyway), is that by making AGW as water-tight as possible they are turning it into something as vaporous as irrelevant.

    If anything we read, measure, hear, experience, compute is invariably further evidence for AGW, then obviously AGW is an all-encompassing, all-singing, all-dancing idea that explains absolutely nothing at all.

    I don’t think we’re there yet. But close.

  4. Excellent. This really cuts to the heart of the matter. Sure, they are right, you can’t look at the last 7 or 10 years and claim it’s been cooling – but alternatively you can’t turn around and claim it’s been warming either – the confidence intervals cut both ways.

    Also, if things continue at about the same temperature levels we have now, it doesn’t look like it will be long before there is a statistically significant negative trend.

  5. Dear Jeff, I came across this quotation just the other day, from that great observer of human nature, your fellow countryman, Mr. Mark Twain.

    Mark Twain – “There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact”

    I think this is bang on and for me, sums up the current nature of the climate debate and the science. In my opinion, we need a lot more observational based science on the climate and in particular, we require at least an additional 20 year observational baseline of satellite reading and the new ocean sensor array. We are now beginning to collect data sets that will lead to much better earth models – we just need to give it some time.


  6. According to Tamino, we would need to see 15 or 16 years of no new records to be statistically significant that warming has stopped, if you treat temp as trend + noise.

    This is OK, but the idea that warming doesn’t stop until you see a statistically significant change is ludicrous. So in 2014 December 31, warming hasn’t stopped, and you should reach no conclusions about the warming for the 21st century, but when those new numbers come out for 2015, then the warming has stopped?

    Other variations on this theme is that it is still one of the warmest X years on record. So therefore warming hasn’t stopped unless you lower temperatures all the way down to the average, or perhaps a bit below to get the running mean back down.

  7. Jeff, in Ross and my original submission on Santer, we made almost precisely the same ironic point – that the argument used to assert that there was no statistically significant difference between models and observations cut two ways and showed no statistically significant difference from no trend.

    One of the reviewers demanded that this be removed. He called it a “side show” and said that is had no place in “any published submission” and that it “simply distracts from what could be scientifically useful.”

  8. With respect to GIStemp, if you haven’t looked at E.E. Smith’s blog lately, he appears to have uncovered a .6 C bias for the U.S. temperatures, an artifact of dropping of thermometers in the last couple of years. See Jeff’s link to ChiefIO blog, above.

  9. #7, So it’s fair to test for warming as long as you don’t notice the lack of significant trend. Sounds like a fair review.

    #9 Al, I’ve read it a few times but EM doesn’t do enough graphing and you really have to read carefully to get the message. I’d like to replicate his work someday just to get a feel for what they’re doing. You learn so much more doing it than reading it.

  10. Has Grant made a post yet to say that you have no credibility, so you don’t merit his attention yet? If so, it’s his admission that he has no rebuttal – and his 4 loyal readers should “move on.” If not, he’s probably trying to devise a two box model to refute your/his assertion.

    Life is grand. 😀

  11. If someone is bored one day, why not perform the same analysis on stock market prices and demonstrate the complete absence of credit crunch. Then challenge Mr Tamino (note the respect) to purchase some shares I have at the 30 year trend price.

  12. Looking at your code, you’re testing 95% CI two tailed, which is really testing for no significant warming or cooling at alpha = .025 each. If you change your Ho to “there is no significant cooling at 95% CI”, you’ll reject it on all 4 data sets (eyeballing it). Mike S.

  13. Thanks MikeN. It says “only if the record lasts beyond 2012 do we have statistically significant evidence of any change in the global warming pattern“.

    Finally, something to work about!

  14. I followed the Tamino link and enjoyed the continued attempts to reduce climate to an algebraic formula.

    The truth is that the temperature data from the individual stations are often highly flawed. If you glue together a variable number of already highly flawed stations, you will end up with an extremely highly flawed ‘global’temperature that can’t be used to prove anything.

    If you measure from the valley of a temperature cycle like Hansen and Jones did, you really shouldn’t be too surprised when temperature rises as you go up towards the next summit.


  15. Doesn’t it follow that I can state that a global cooling started in 1998 with a trend of -1.7C per century and you can’t disprove this unless you show me a warmer point than 1998 in the last 16 years?

  16. Fairly meaningless article. It’s apparent that you did a little work on your code. But it the process is lacking in explanation which means I have to go through the exercise of decoding it before anything meaningful can be drawn from the conclusions.

    The general idea is good. I’ll be sure to go through the exercise to see what the real results will be when the general idea is properly implemented. I’ve been looking for something fun to do.

    My initial impression is GIGO, garbage in, garbage out. That you’ve simply created a piece of code that makes pretty plots but is statistically meaningless.

    If your really think your results are valid, you should get some support. It won’t be hard to do if your results really are valid. There are plenty of people that would just love to find something solid to poke a hole in AWG with. You might just get famous.

    On the other hand, there may be another reason that Real Climate and Tamino would ignore your calculations. And it isn’t some conspiracy to hide the truth. It’s simply that your procedure is useless.

    I just get a sense that we won’t be seeing the headline, “Jeff Id proves Global Warming is a hoax” soon.

  17. Er a question(s) from a non-statisician.

    The process that drives temeperature change is poorly understood if, of course, one ignores the warmists dependance on CO2 + unproven positive feedback.

    Standard deviation is derived from a “normal” distribution. Why should the temperature trend observe this artifact of stats theory?

    Hadley and GISS use statistical torture to derive a GAT temp which itself is something of an illusory concept, based upon a falling number of stations, jiggery pokery to get sea temps, make and loose the original records, dom not publish their adjustment methodology and then claim accuracy to a tenth of a degree when the original measurements were probably no better than +/- 0.5C or worse.

    Do a mind game. Imagine we had an accurate thermometer record going back 2,000 years and this showed the periodic warming and cooling, e.g RWP, Dark Ages cooling, MWP, LIA. The trend that would be derived would of course depend where the measurement started from but one that say started in 950 AD would show a warming trend that would crash into the LIA 300 years later. So if the good citizens of the 12th century listened to Steig, they would have destroyed their economy and then suffered even more as the LIA took hold.



  18. #22, I agree it’s a meaningless article but why have RC and Tamino done multiple posts to demonstrate it and say cooling isn’t real.

    That you’ve simply created a piece of code that makes pretty plots but is statistically meaningless.

    It is equally as meaningful as the stats that claim cooling over the last decade haven’t happened.

    I just get a sense that we won’t be seeing the headline, “Jeff Id proves Global Warming is a hoax” soon.

    How would I disprove something I believe in?

  19. #22 Ya know John, I did expect someone would take advantage of my absence to make unreasonable statements. Lucia has verified at least part of this post, it looks like SteveMcIntyre has produced similar results. The methods are used in climatology. If you want all the code I’ll send it to you in a turnkey form. There is enough above to figure out where it’s wrong though. When you do, let me know because then you can write a letter to Ben Santer and explain to him how his whole ‘climate models are good’ paper is garbage.

  20. I’m getting the impression here that the climate/weather is essentially white noise and that if you have enough different methods of calculating trends then you can choose whatever trend you want. Am I right or wrong.

  21. #30 – it’s a red noise (autocorrelation) but trend and significance have uncertainty based on assumptions. I’ve been digging into this for a while now and nothing is completely clear. Beware of those who claim to know the answers.

  22. The only problem with this kind of analysis, assuming that the technical details are statistically correct, is that is selects a preferred subset of the data. I’ve done this kind of analysis before and is doesn’t have sufficient meaning when put into the context of all the data. The question I had to ask myself was, why this set? Why not choose the last ten years? Why not the last 7? Where do you draw the line to determine that the set you have chosen is preferential of the set that has one year less or one year more in the data? And, the problem is that, as less and less data is used, the level of statistical significance decreases. At some point, it just isn’t enough data.

    I was looking at this back in 2009. I haven’t “decoded” the code, translated into my own program so I truly understand what I’m looking at. I believe, thought, that this basically is what the analysis is attempting to do. But, if I am reading it right, the results are simply saying that 15 years is simply not enough data to determine a trend with significance, not that there is no trend.

    The Earth is a petri dish and we are a fungus in it. At some point, the fungus simple fills the petri dish. The question isn’t “if”, it is “when” and if not now, why not now? And if not now, isn’t it a good thing to be working on?

    The bigger issue isn’t one of significance. There are things that are statistically significant that just aren’t important. Importance is expectation times consequences. More difficult yet is when it is expectation times unclear downside consequence.

    At some point, when standing in the middle of a train trestles, with the train bearing down on you, it behooves you to toss the calculator and just run like hell.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s