Monday 24 April 2017

"Hiatus": Signal and Variability

Stefan Rahmstorf, Grant Foster and Niamh Cahill just summarized the statistical evidence for the mirage people call the "pause" of global warming in their new article: "Global temperature evolution: recent trends and some pitfalls."

The Open Access paper is clearly written; any natural scientist should be able to follow the arguments. The most important part may be a clear explanation of the statistical fallacies that lead some people to falsely claim there was such a thing as a "hiatus" or "slowdown".

Suppose that Einstein had stood up and said: I have worked very hard and I have discovered that Newton got everything right and I have nothing to add. Would anyone ever know who Einstein was? ... The idea that we would not want to be Einstein, if we could overturn global warming ... how exiting would that be? Of the tenth of thousands of scientists there is not one who has the ego to do that? It's absurd, it is absolutely unequivocally absurd! We are people.

I have studied the "hiatus" problem hard (1, 2, 3, 4), read this new paper and I have nothing to add. Unfortunately.

Well, okay, maybe one thing. Just because a trend change is not statistically significant, does not mean you cannot study why it changed. It only means that you are likely looking at noise and thus likely will not find a reason. But if you think there may be a great reward in the result that can make high-risk research worthwhile. Looking at how small the trend differences are and knowing how uncertain short-term trends are, I am not going to do it, but anyone else is welcome.

That there was no decline in the long-term trends also does not mean that it is not interesting to study the noise around this trend. The biggest group in the World Climate Research Program studies Climate variability. That by itself shows how important it is.

This blog is called Variable Variability. I love variability. It is an intrinsic property of complex systems and its behaviour over temporal and spatial averaging scales can tell us a lot about the climate system. It also has large impacts. Droughts and floods fuelled by El Nino are just one example. It is a pity most people just want to average this away.

One man's noise may be another man's music

Now that we take the climate system into unknown territories predictions of the seasonal, annual and decadal variability have become even more important to plan ahead and protect communities. Historian Sam White suggests that the problem of the little ice age in Europe was not the cold winters, but the unpredictability of the weather. Better predictions will help a lot in coping with climate change and already produce useful results for the tropics.

Variability lovers of the world, let's stand up for the importance of our work and not try to faithlessly justify it with middle of the road research on overstudied averages.

Related reading

Science Media Centre asked three scientists for a reaction to the study: expert reaction to climate hiatus statistics

Cranberry picking short-term temperature trends

Statistically significant trends - Short-term temperature trend are more uncertain than you probably think

How can the pause be both ‘false’ and caused by something?

Atmospheric warming hiatus: The peculiar debate about the 2% of the 2%


Rahmstorf, Stefan, Grant Foster and Niamh Cahill, 2017: Global temperature evolution: recent trends and some pitfalls. Environmental Research Letters, 12, No. 5,


  1. A couple of points about the paper

    I like very much the fact that the authors emphasize that a change in the rise of GMST and the issue of comparison with model projections are separate issues and it seems to me that distinction is not always made. (I read a paper once a few years back - 2013/4 ? - that used the term "hiatus" to describe the divergence of the observational record from model projections which just seemed sloppy.)

    On a wider point, one can look at the GMST rise either as an indicator of the underlying warming trend - in which case it is valid to talk of ENSO, volcanic events and variations in TSI as "noise" - or as a thing in itself to which all these factors contribute. I think a lot of confusion (perhaps intentional, from some people) about the "hiatus" has been generated because it is not clear how they are "using" GMST. This paper explains very nicely that a "hiatus" in the trend in GMST does not necessarily imply a hiatus in the trend of Global Warming

  2. Agree, those are two fully different questions.

    Calling the differences between model and observations a "hiatus" is somewhat weird. That presupposes a reason while studying whether there is a discrepancy and if there is, what the reason is: the model itself, the model input, the observations and its data processing or the comparison methodology.

    Climate modellers making such comparisons often do not seem to be aware that model ensemble spread is not uncertainty. When it comes to short-term variability, the uncertainty is about twice as large. In ensemble NWP this difference is a main part of the work.

  3. "That there was no decline in the long-term trends also does not mean that it is not interesting to study the noise around this trend. "

    THIS. And I would add that it doesn't mean that there aren't explanations for that noise around the trend.
    I think this is a crucial part of understanding how the hiatus can be both false and real, as you once argued.

  4. Nice post, as always.

    Similar to the related fad in EBM-derived sensitivity, we should not forget how very easy to produce such papers are.

    Erratum: I think you may want "mediocre" rather than "middle of the road." While the literal meanings are similar, the latter connotes mainstream and the former connotes deficiency.

  5. Thanks.

    "Mediocre" is a bit to negative for me. There is a reason it is the dominant line of thinking, it is valuable. I just feel it is done too much.

  6. Hmm, was assigning significance to the "pause" ever the dominant line of thinking? I guess the IPCC may have made it that in one sense, but they got rather a lot of pushback on that. My impression was that it was more defensive political move than science. Was it dominant in your field?

  7. Hard to say, I do not know of any "hiatus" consensus studies. :-) Would also be hard to do because proponents of the idea keep on making new definitions.

    The homogenisation community are people who are good at data analysis and statistics; I do not remember any of them claiming it was real. There seems to be quite a few on twitter from the South of England, though.


Comments are welcome, but comments without arguments may be deleted. Please try to remain on topic. (See also moderation page.)

I read every comment before publishing it. Spam comments are useless.

This comment box can be stretched for more space.