clearclimatecode.org

Posted by William M. Connolley on September 15, 2008
(4)
More »

clearclimatecode.org is a project of Nick Barnes, who occasionally comments here. Its an attempt to re-write the byzantine GISS historical record processing software into something that won’t disgrace the 21st century, namely Python (boo hiss, but mt will be happy).

So the bugfixes are nice and clearly described, and are things like “STEP 0 of the GISTEMP code is in FORTRAN. This makes it unclear to the public.” :-)

The aim is to re-implement the GISS algorithms, but in clearer code and in a more portable way, ideally producing identical results except for bug-fixes.

They have found one real bug so far: “GISTEMP STEP0 discards the final digit of every USHCN datum”. Before anyone gets overexcited, GISS already know, have fixed it, and it makes no noticeable difference to the overall result.

Someone else, I think, was doing an open data averaging project, but I forget where. [Update: www.opentemp.org by John Van Vliet (thanks IH). Not sure if that is still active -W]

(4)
More »

Comments

  1. #1 mankoff
    2008/09/15

    bad link to ccc.

    [Thanks. Fixed now, I hope -W]

  2. #2 Nick Barnes
    2008/09/15

    Thank you for this link.
    clearclimatecode.org will one day have a mailing list, some online data visualisation, and maybe a blog, but I have a day job and have spent the last couple of weeks getting CCC ready for our presentation at PyconUK yesterday. So for now and for a while to come there’s just a big tarball to download, and the PNGs of our Pycon presentation. Email comments (and especially offers of help) to staff@clearclimatecode.org.

  3. #3 SomeBeans
    2008/09/16

    Isn’t this what John Van Vliet did with his OpenTemp application? (http://www.opentemp.org/main/) Worth a look anyway…

  4. #4 JohnV
    2008/09/16

    My little OpenTemp program was built to do an independent analysis of the USA48 temperatures. I put it together in response to the SurfaceStations survey and suggestions by some that GISTEMP was wildly inaccurate.

    I was able to show that the GISTEMP historical trends for the USA48 very closely matched a temperature history derived from only the “best” stations (as determined by SurfaceStations). Good times.

    I haven’t touched the code for about 10 months, but it’s still available if anyone’s interested. I’d like to resurrect and extend it but real life has been getting in the way.