Is the historical temperature record reliable?

The hockey stick*.

A recent paper  (published in the March 2011 issue of the Annals of Applied Statistics) by Blakeley McShane of Northwestern University and Abraham Wyner of the University of Pennsylvania has raised doubts about the reliability of the historical temperature record. The paper re-opens the controversy over the “hockey stick graph”, a graph constructed mainly from tree ring “proxy” data by US climate scientist Michael Mann. The graph shows temperatures remaining fairly steady and then rising sharply since the industrial revolution. But the two US statisticians  conclude in their paper that the evidence for a ”long-handled” hockey stick (where the shaft of the hockey stick extends to the year 1000 AD) is lacking in the data. The “fundamental problem“, according to the researchers, is that “there is a limited amount of proxy data which dates back to 1000 AD; what is available is weakly predictive of global annual temperature“.


Smoothed reconstructions of large-scale (Northern Hemisphere mean or Global mean) surface temperature variations from six different research teams. Source: NRC (2006).


McShane and Wyner state in their paper  that historical temperatures reconstructed from proxies such as tree rings and ice coresdo not predict temperature significantly better than random series generated independently of temperature” and that climate scientists “have greatly underestimated the uncertainty of proxy based reconstructions and hence have been overconfident in their models”. They believe that the long flat handle of the hockey stick is more a feature of the data processing and “less a reflection of our knowledge of the truth”. Regression of high dimensional time series is indeed a very complex problem with many traps and in this case in may be that the number of truly independent observations is just too small for the accurate reconstruction of the historical temperature record.

p.s. Non-specialists are encouraged to read the introduction, the conclusion as well as the second section (controversies) of the paper. A series of discussion papers related to this study can be found on the following links:  1, 23, 4, 5


CRU Northern Hemisphere annual mean land temperature is given by the thin black line and a smoothed version is given by the thick black line. The forecast produced by applying the Lasso to the proxies is given by the thin red line and a smoothed version is given by the thick red line. The in-sample mean is given by the horizontal blue line. The forecast produced by ARMA modeling is given by the thin green line and a smoothed version is given by the thick green line. The Lasso and ARMA models and the mean are fit on 1850-1968 AD and forecast on 1969-1998 AD.


Backcasts to 1000 AD from the various models considered in this section are plotted in grey. CRU Northern Hemisphere annual mean land temperature is given by the thin black line with a smoothed version given by the thick black line. Three forecasts are featured: regression on one proxy principal component (red), regression on ten proxy principal components (green), and the two stage model featuring one local temperature principal component and ten proxy principal components (blue).



Backcast from the Bayesian Model of section 5 of the paper. CRU Northern Hemisphere annual mean land temperature is given by the thin black line and a smoothed version is given by the thick black line. The forecast is given by the thin red line and a smoothed version is given by the thick red line. The model is fit on 1850-1998 AD and backcasts 998-1849 AD. The cyan region indicates uncertainty due to \epsilon_t , the green region indicates uncertainty due to \beta, and the gray region indicates total uncertainty.



* The hockey stick: A multiproxy reconstruction of Northern Hemisphere surface temperature variations over the past millennium (blue), along with 50-year average (black), a measure of the statistical uncertainty associated with the reconstruction (gray), and instrumental surface temperature data for the last 150 years (red), based on the work by Mann et al. (1999).

Tagged , ,

6 thoughts on “Is the historical temperature record reliable?

  1. thinks says:

    I feel that in the scale of geological time, the last 200 years, and the last 1,000 years are segments too short to represent a greater picture, or to point to a trend attributable to specific events, as in a cause-and-effect theory. In fact, we know that the average temperature and climate of the planet has been unusually steady since the last ice age. Taking into considerations periods of dry deserts and bitter colds that have been documented using various techniques and sources and representing billions of years, including the theory of “Snowball Earth”, it seems that predicting the climate’s immediate course and attributing a cause, is closer to the realm of crystal balls than verifiable science. That, plus the fact that thermometer data is far more accurate, and there is far more and detailed data, than the sources for before the Industrial revolution.

    Although all models show an upward trend for the last two hundred years, we should see graphs of data reaching back for thousands of years to establish whether a cause-and-effect theory attributable to human interference (through industrial activity) could be supported, or whether this fluctuation ha srepeated itself before, as it surely has. Also, we should consider the so-called little ice age that occurred in Europe after the so-called middle-ages warm up. The trend of the last 200 years is much more likely to be an “aftershock” of those fluctuations.

    • epanechnikov says:

      I think I tend to agree with your view. As mentioned in the paper “the number of truly independent observations is just too small for the accurate reconstruction of the historical temperature record”. This however does not mean that we should not be very conscious…

      • thinks says:

        Agreed, about being self-conscious. However, if we were born an evolved of the earth, like everything else in the biosphere, is not each action we take something that the Earth does to itself, rendering our assumption of interference plain arrogance? Mom may want her temp to go up a bit and she may like a bit more CO2… There was much more of that earlier before all that green stuff produced all the O. Such an accommodation will also get rid of that pesky organism that was used to create it, therefore no harm done? On the other hand, if that organism were to self consciously ignore mommie’s evolutionary random plans and look after its own little self, would that not be constitute a break from nature? Just musing! 🙂

      • epanechnikov says:

        You are right to say that we are a part of the universe and our actions are, by definition, something that the universe does to itself. However it is also true that the accumulation of knowledge has made us capable to fundamentally change the way our (close) universe functions. But as you nicely implied, our effects until now have probably been weaker than “mommie’s evolutionary random plans”…

        “would that not be constitute a break from nature?…”

        Good question but I think that all organisms act in a selfish matter. The difference is that we can affect the universe more than any other living organism.

      • thinks says:

        So, consider this: We, among other things are the self-consciousness of the universe, and believe, and are probably right, that organisms like us (of which type there are many, many in the universe, surely) do have more of an effect than lower organisms. We may, within our own anthropomorphic understanding of ourselves, use the knowledge, the self consciousness of the universe, and debate. Human forces and influences come into play, “save the Earth”, “pass the buck” etc. Whatever ends-up happening it is still what the Earth/Universe does to itself. We only think we have a choice, when in fact we are part of the random process of a greater picture. I’m horrified I said that because it sounds fatalistic and that is the last thing I will ever consciously support.

      • epanechnikov says:

        The way I see it is that knowledge creates uncertainty. And that self-consciousness is just a mechanism which controls the size of extra randomness we spread into the universe. Is self-consciousness a perfect control mechanism? That is something, I dare to say, which we won’t know until it is too late (or hopefully never).

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 )

Google+ photo

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

Connecting to %s