Home > Deconstruction > Archibald 2008 Redux

Archibald 2008 Redux

2012 May 20

In this presentation, I will demonstrate that the Sun drives climate, and use that demonstrated relationship to predict the Earth’s climate to 2030. It is a prediction that differs from most in the public domain. It is a prediction of imminent cooling.

Solar Cycle 24: Implications for the United States, Archibald, 2008

I contacted Archibald earlier this year as I attempted to reproduce a couple of the graphs in his 2008 Solar Cooling presentation. He responded promptly but wasn’t able to provide me any help telling me that the computer he used for that presentation had crashed some time before and he had no alternate sources. Nor did he recall the methods used.

I began with Archibald’s chart entitled ‘A US Rural Data Set.’ I note that 1) he used GISTEMP data and 2) listed the five stations he used by name and 3) that he smoothed the annual average

I know that the source of GISTEMP US data is USHCN, so I turned there for the station data. Strangely, all the USHCN data begins in 1895, but the first time tick on the chart is for 1893.

Archibald has chosen five widely distributed, rural stations as representative of the continental United States. It was easy to locate the Station IDs by name in the USHCN station inventory. The stations are mapped below.

093754 31.9881 -81.9522 61.0 GA GLENNVILLE 3NW
094170 33.2842 -83.4681 74.7 GA HAWKINSVILLE
098535 32.6875 -84.5197 195.1 GA TALBOTTON
161411 32.5133 -92.3478 54.9 LA CALHOUN RSCH STN
314055 35.0536 -83.1892 1170.4 NC HIGHLANDS

Archibald mentions that he smoothed annual averages, but does not recall the exact method. Examining several different smooths, the best fit I could find was a 3 year moving average, which fits well in recent decades, but diverges somewhat in middle decades while maintaining a similar variation throughout. I cut the station data at the year 2003, the last year shown on the original chart. This fit is shown below.

Having found a reasonable fit to the station data, I turn next to the “Projected Temperature Profile to 2030.” The pre-2003 rural US data set has been further smoothed and, strangely, cooled by about 0.5C. This may be due to some sort of averaging with the solar smooth shown earlier in the presentation. Nevertheless, I found a lowess smooth (f=0.04) of the 3 year moving average creates a reasonable likeness of the 2030 projection chart. I have added the post-2003 data points and the continued lowess in red.

This is not a perfect reconstruction by any means. The data sets are not the same. The range appears to be different. The visual matching may be impaired by poor scaling in the background images. Nevertheless, it should be adequate for qualitative comparisons of future observations to the original temperature projection.

Combining the rural US data set we saw earlier and the projected temperature response to the length of Solar Cycle 23, this graph shows the expected decline to 2030.

The temperature decline will be as steep as that of the 1970s cooling scare, but will go on for longer.

Solar Cycle 24: Implications for the United States, Archibald, 2008

code is here

  1. Ned
    2012 May 22 at 6:14 am

    Ooh, can I play this game, too?

    I went to GISTEMP and looked for five rural stations that were slightly better distributed than Archibald’s five stations:

    Station_Name Lat Lon Stn_ID Pop Years
    Albia 3 Nne 41.1 -92.8 425001301120 < 10,000 1901 – 2012
    Zion Np 37.2 -113 425004297170 < 10,000 1904 – 2010
    South Lincoln 44.1 -73 425004376120 < 10,000 1895 – 2012
    Moro 45.5 -120.7 425003557340 < 10,000 1917 – 2012
    Oconto 4 W 44.9 -88 425004762080 < 10,000 1895 – 2012

    Here's a map of those stations:

    link to map

    I downloaded the monthly data from GISTEMP, converted them to anomalies (in degrees C), and calculated the mean of the five stations for each month. Then I calculated three-year moving averages:

    link to graph of 36-month means

    I then compared this to your version of Archibald’s graph, including the prediction out to 2030:

    link to comparison

    It looks a bit different, doesn’t it?

    For fun, I also ran a 36-month LOESS smoothing function on the monthly data. The result was quite striking:

    link to 36 month LOESS version

    I suspect Mr Archibald will not be rushing to embrace these results.

  2. 2012 May 22 at 10:08 pm

    Nope, not those results. For that matter,I’m not sure he even embraces his own results. Based on his comment about this post at WUWT, he seems to be trying to put distance between himself and his previous work.

  3. 2012 May 30 at 6:39 am

    Wondering if the mismatch in the charts is do to nonlinear changes to the image while transferring between documents such as excel -> powerpoint -> pdf

  4. 2012 May 30 at 8:15 am

    I got a ping from http://scienceblogs.com/stoat/2012/05/25/global-cooling-and-spooks-again/

    So… the mystery to me is, why does anyone take Archibald seriously enough to bother reproducing his stuff?

  5. 2012 May 30 at 8:33 am

    Archibald is occasionally taken seriously in other quarters.
    Sometimes a picture/chart/graph is worth a 1000 words.

  6. Ned
    2012 May 31 at 5:49 am

    Yikes. I only read TOD occasionally, so I guess I missed that.

    From your comment there, I’m guessing that you, or others, had pointed out that Archibald is not a particularly reliable source of analysis … but the comments were removed.

  7. Ned
    2012 May 31 at 6:02 am

    I see that Archibald has a new post at WUWT, on Peak Saudi Oil. The locals over there don’t seem to like it very much. If he had added a section at the end advocating coal-to-liquids, and/or promoting Keystone XL, it probably would have gone over better.

  1. 2012 May 28 at 1:24 pm
  2. 2012 May 30 at 10:41 pm
Comments are closed.