What Instrumental Record did Mann Use?

There is a new paper involving the omnipresent Michael Mann that is going to get lots of attention.  The paper isn’t even out yet, but already I am very intrigued because the data used in the charts is clearly not the data that it says it is.  The graph in the paper shows significantly more warming that the data it claims to be.  For a peer-reviewed document that should be the gold standard, this one is already seriously lacking at the first cursory review.

At first glance I saw something wrong with the top chart in this figure.

The Inconvenient Skeptic

Kemp, Mann 2011

They claim to be using the HADCrutv3 data for the red instrument portion of the record, but whatever the data is, it is NOT the HADCrutv3 data.

Here is the temperature section of the graphic by itself.

The Inconvenient Skeptic

Kemp, Mann 2011 - Temperature Portion Only.

What it shows is that the global temperature is nearly 1.0 °C at the modern day.  The data it references has little resemblance to the chart above.

To make sure that I was using the latest data, I went and got the latest data for the HADCrutv3.

Here is what it looks like.

The Inconvenient Skeptic

Annual HADCrutv3 Temperature Anomaly

Notice right away that the actual HADCrutv3 data never exceeds 0.5 °C.  The graphic in Mann nearly reaches 1.0 °C.  A difference of at least 0.4-0.5 °C.  The Mann paper is showing double the warming for the HADCrutv3 data.

Part of the problem is that the HADCrutv3 data starts above the 0.0 °C anomaly.  Something that the actual data does not do.  So I started playing around to get a graphic that is comparable to what the Mann paper has.  This seems to be a common problem that Mann papers have.

The best fit from what I can tell is a 0.5 °C offset on a 20 year moving average.  Here is what that looks like if I compress the x-axis.

The Inconvenient Skeptic

(Red) best fit for the Mann temperature data is a 20 year moving average with a +0.5 °C offset.

This is not as smooth as the Mann graphic, but the inflection points are as close as I could fit with only the graphic to go by.  The offset is clearly part of the data manipulation that Mann used to splice the instrumental data to the reconstruction.

Here is a closer view of the Mann temperature chart.

The Inconvenient Skeptic

Kemp, Mann 2011 - Temperature Portion Only. 0.4 and 0.8 °C lines added for clarity.

It is clear that the Mann paper uses an offset of ~0.5 °C from the actual HADCrutv3 data.  If this is a mistake it is of enormous proportions to have an offset like that to splice data.  I am sure others will find the Mann data and take it apart like it usually is, but this mistake was so obvious that it needed immediate attention.



I am not good with graphic manipulation, but I did get a transparent overlay of the offset and the Mann version of the HADCrutv3 chart that was used.

The Inconvenient Skeptic

Not easy to see, but the (Light Green) is the 0.5 °C offset of the 20 year average. A pretty good match to what is used in the Mann paper.

Posted in Bad Science and Fear and Misinformation and Skeptic by inconvenientskeptic on June 20th, 2011 at 6:29 pm.


This post has 10 comments

  1. I don’t see a problem. It is obvious that Mann is using the HADCrutv3 data with a 0.5 degree C offset, so your entire ‘argument’ falls apart into nothing.

    What you HAVE convinced me of, though, is that you are in no way interested in the facts, merely in sliming anybody who is not an AGW-denier, the facts be damned.

  2. inconvenientskeptic Jun 20th 2011

    The only indication is that it is relative to reconstructed data. That an offset of half a degree is tucked into that is not trivial.

    A more robust method would be to make the reconstruction relative to modern measurement. Since one method is known to be direct measurements of temperature and the other is less accurate reconstructed data.

    The method is misleading at best. That I am unwilling to accept misleading information is what I do. I am sorry if you do not approve.

  3. You know Bill, if you’re adding 0.5 degrees to data supposedly showing global warming, isn’t it Mann who isn’t interested in the facts? What this post has convinced me of is that Mann is either sloppy, desperately looking for what he wants to find, or outright lying. When the degree of warming is one of the major issues at hand (along with billions of dollars of research funding and possible regulations that could impact everyone) it seems to me that pointing out that Mann apparently added in a half a degree to exaggerate his viewpoint is worth discussion.

  4. It’s worse that we thought.

    Mann used upside down proxies. Again.


  5. inconvenientskeptic Jun 21st 2011

    So it would appear that Mann used series of data that could not be calibrated, but then re-calibrated the modern measurements to the series that could not be calibrated. Doing so required an offset of +0.5°C in the modern data, but that value is really meaningless because the tree-ring data he used was gibberish.

    Seems like a robust method to me…….

    Meanwhile.. What I wanted to focus on this week is is the zonal temperature trends. I guess that will have to wait.

  6. I encourage the zonal temperature trends research. I always love breaking down data to see the pattern.

    What I’d love to see is the 5×5 HADCRUT grid visualized to see where the “warming” is hiding.

    Lubos took a look at RSS data since Jan 2001:

    -0.40 °C / century: globally
    -1.16 °C / century: tropics
    +0.22 °C / century: North extratropics
    -0.19 °C / century: South extratropics
    +3.83 °C / century: Arctic
    -1.27 °C / century: Antarctica
    -4.84 °C / century: contin. USA
    -0.23 °C / century: North Hemisphere
    -0.58 °C / century: South Hemisphere


    Steve Goddard has an itneresting look at how localize sea level increase is too:


  7. LtCusper Jun 22nd 2011

    Different anomaly base periods are indicated. Left to the reader not to be confused I guess. Worth discussing the issue.

    The Kemp et. al. paper Fig. 2(A): “Data since AD 1850 (red) are HADCrutv3 instrumental temperatures. Values are relative to a preindustrial average for AD 1400–1800.”

    The HADCrutv3 data site linked: “…temperature anomaly (from the base period 1961-90).”

  8. inconvenientskeptic Jun 22nd 2011


    As I said in the previous comment, it is not a robust method to calibrate “good” data to less precise reconstructed data. (ok, technically I was being sarcastic).

    The appearance that the Kemp chart gives is HIGHLY misleading. That is why it is important to stick to a more consistent reference. Switching references defeats the purposed of using anomaly data in the first place.

  9. HADCRUTv3 = variance adjusted version of HadCRUT3

    I don’t think they should be using a variance adjusted data set for this.

  10. Travis Jul 11th 2011

    Thank you, Skeptic, for your hard work.

    I frequent this site and recommend it to others who have doubts about the role of mankind in climate change.

    I’ve found your methodology and willingness to engage your opponents with facts (and civility) to be refreshing.

    Keep up the good work. I, for one, will be buying your book.

Web Design & Dev by

Mazal Simantov Digital Creativity