International Journal of Inactivism (now supplanted by Decoding SwiftHack)


Surface stations redux

Filed under: Anthony Watts,surface stations — stepanovich @ 20:15

cite as: F. Bi. 2008. Surface stations redux. Intl. J. Inact., 1:122–125

Hmm, so Anthony Watts and John Goetz are still blowing smoke over the US surface temperature station records. As before, their overall claim is that many surface stations are “not-so-well maintained or well sited”, causing their temperature measurements to be biased towards a warming trend. And as of writing Watts, is still showing Orland, CA, and Marysville, CA, as examples of ‘good’ and ‘bad’ stations respectively — the ‘good’ station shows cooling, while the ‘bad’ station shows warming, or so the claim goes.

Now, if this were true, then if we take the differences between successive temperature measurements from the two stations, the differences should form an upward trend with time — that is, they should become bigger and bigger as time goes by. To see if this is true, I wrote a little C++ program to calculate the differences and perform linear regression to find the least squares trend.1 Here’s the result:

Uh-oh. Far from showing an upward trend, the differences actually go downwards. If the ‘badly sited’ surface stations are causing a bias, the bias will seem to be in the opposite direction of what inactivists such as Watts are saying!

Of course, flesh wounds such as these are no obstacle to the inactivist — if the data don’t listen to you, you can just throw away some of the data, again and again, until they do listen. So a fun diversion to consider is this: can we throw away data from the beginning of the temperature records in such a way that we end up with an upward trend of the differences? To answer this question, I computed

  • the gradient of the linear least squares trend from Mar 1883 to the present;
  • the gradient of the linear least squares trend from Apr 1883 to the present;
  • the gradient of the linear least squares trend from May 1883 to the present; …

(This is also in the C++ program, by the way.) Here’s the graph I obtained:

Well… clearly, for a majority of starting points, the trend in the differences is downwards. However, if we pick a starting point between Feb 1909 and Oct 1926, we end up with a trend that’s upwards. Therefore, if you want to ‘prove’ that the ‘bad siting’ of the Marysville station causes its temperature measurements to show a warming bias, be sure to find a flimsy excuse to cherry-pick a starting point within this range. ☻


  1. The program takes three arguments — the file names for the two temperature records, and a label — and outputs a gnuplot file which can then be rendered in, say, PostScript. By the way, note that implementing linear least squares regression is pretty straightforward — it takes less than 15 lines of code, and that’s in C++!

Update 2008-08-13: John Goetz complained here that I’m mischaracterizing his blog entry as showing a warming bias in the ‘bad’ stations, while it doesn’t try to show any bias. Well, I’ll grant that; I’ll also wait to see if Goetz decides to inform Watts (on the same blog) that the ‘bad’ stations do not in fact give a warming bias, because Goetz will try to correct him won’t he…

Another update 2008-08-13: Watts asks,

What specific datasets did you use?

Fair question. I’m using data sets that I wrote about before — that is, temperature records after homogeneity adjustments — and I’m uploading them here.



  1. (John Goetz: It’s clear you didn’t read this post before complaining that I wasn’t reading your post. That’s uncool, don’t you know?)

    Comment by frankbi — 2008/08/13 @ 03:07 | Reply

  2. I love math, it never lies, and proves so often who does.

    Comment by The Chemist — 2008/08/13 @ 22:53 | Reply

  3. The Chemist:
    As a friend of mine (in the social sciences) puts it, math is a nice, hard discipline. And like most nice, hard things, it makes a satisfying THWACK! when you smack the ignorant upside the head with it.

    Comment by Brian D — 2008/08/13 @ 23:39 | Reply

  4. Brian D, The Chemist:

    Ha. 🙂 Well, Watts’s latest excuse for dismissing the above analysis is that I’m using homogeneity-adjusted data, which is Wrong because the data aren’t Raw enough or something. Now, if the homogeneized data are what the climatologists are using to build their global warming theory on — which I presume they are — then I don’t see any reason not to use them.

    And then there’s the good old ‘even if…’ backpedalling schtick.

    Comment by frankbi — 2008/08/14 @ 06:24 | Reply

RSS feed for comments on this post. TrackBack URI

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

Create a free website or blog at

%d bloggers like this: