Skip to content

Muller’s Alabama Temperatures Hopelessly Inaccurate

August 8, 2012

By Paul Homewood

It is nearly a year since Professor Muller published his BEST global temperature figures, which made so many headlines at the time. WUWT has neatly summarised some of the fundamental errors Muller made, so I won’t revisit old ground. However the Sunshine Hours blog has highlighted some more problems. I would highly recommending his post on this in full as he has done a lot of work collating the data. You can read it here.

In essence though it seems that there are some huge discrepancies between the BEST figures and the official NCDC ones for parts of the USA. Compare these two graphs.



Figure 1

Annual Temperature


Annual 1910 – 2011 Trend = -0.13 degF / Decade



Figure 2

It is well known and accepted by the NCDC that Alabama, along with the rest of the South East of the US has bucked the trend by getting cooler since their records began in 1895. The trend since 1910 has been a decline of 0.13F/decade as their graph shows.

So why does BEST show a warming trend? According to the table shown below their graph (click on their website for this), the Mean Rate of Change per Century in Alabama since 1910 has been an increase of 0.40C/century (0.72F), so there is a difference of 2.02F/century. Remember as well that NCDC figures are already fully adjusted for TOBS and other factors and that the raw figures show a greater cooling trend than NCDC does. (See footnote).

All this raises the question – do the same discrepancies exist in other parts of the US, and indeed globally? Of course it could be that NCDC’s figures are hopelessly wrong, but perhaps in that case someone should tell them.




The graph below is produced by NOAA, and shows the difference between the old “divisional datasets” that were based on raw temperatures, and the new “ GHCN version” that contains adjustments for TOBS etc. From the 1930’s to the present the new version adds a warming trend of about a degree fahrenheit. It is this new version that the NCDC graph is based on.

So there are now two sets of warming trends introduced:-






  1. August 8, 2012 5:09 pm

    Reblogged this on Tallbloke's Talkshop and commented:
    This issue will not go away. Adjustments to data invite scrutiny.

  2. August 8, 2012 5:09 pm

    Reblogged this on Tallbloke's Talkshop.

  3. Ray permalink
    August 8, 2012 6:37 pm

    Since the above graphs cover different periods and one shows 10 yr/12 month moving averages and the other annual figures, I downloaded the NCDC annual mean temp. from 1895 to 2011 and calculated the BEST annual mean from the monthly data for the same period. Since BEST data for 2011 only goes to November, I based the annual mean on that figure, although that is likely to be too high.
    The NCDC data shows a trend of -0.039c/decade and the BEST data shows +0.066c/decade over the period.
    The NCDC 10 yr average starts above BEST but ends up below it, with a crossover point of 1955.
    To put it another way, all of the NCDC average annual temperatures are higher than BEST up to 1955, and lower than BEST after 1955.
    The rate of gain for BEST over NCDC is about 0.105c/decade.

    • August 8, 2012 8:45 pm

      Thanks Ray.

      Sunshine Hours came to the same sort of conclusion – it’s certainly worth checking out his article if you have not already done so, as I did not want to rehash his numbers.

      I will also be looking at UK figures in the next day or so.

  4. August 8, 2012 7:54 pm

    Great post.

  5. Paul Matthews permalink
    August 9, 2012 8:47 am

    I guess you chose Alabama just as it’s the first alphabetically?

    I Just checked Kentucky – same thing.
    NCDC almost flat, 0.01F/decade
    BEST 20thC warming looks like about 0.1C/decade

Comments are closed.

%d bloggers like this: