RealClimate

Comments

RSS feed for comments on this post.

  1. Thank you. This is the kind of informative “look inside” climate science article that RealClimate does very well, perhaps uniquely well.

    Comment by SecularAnimist — 24 Apr 2014 @ 3:20 PM

  2. That is the good thing about standards, there are always so many to choose from.

    Comment by Dikran Marsupial — 24 Apr 2014 @ 3:33 PM

  3. One class of tools I’ve found helpful to deal with this are the units packages. These are available in all the popular data analysis languages (MATLAB, R, Python), and many even support uncertainty. For example, a precip variable can be created with the code

    precip = nc.load[“precip”] * uu.mm/uu.day

    And then later easily convert with precip.to(uu.m/uu.year)

    Alternatively

    book_length = (20. * ureg.centimeter).plus_minus(2.)

    I use the Pint library for Python http://pint.readthedocs.org/en/latest/

    Comment by Mankoff — 24 Apr 2014 @ 5:09 PM

  4. ” … using Fortran for data analysis is no longer very efficient.”

    fortran isn’t quite dead yet. I am biased by my training, but for some recent analysis i was doing, fortran still beat R,Matlab,Octave, Mathematica and the like by orders of magnitude in running time. And if i am running on a beowulf, fortran is my weapon of choice every time for numerics (distributed text processing is another matter …). The most important thing is that the numeric libraries for fortran are very old, very well debugged, and their failure modes are well known. It is now possible to import most of these libraries into environments such as R, so we sorta can have the best of both worlds in some situations.

    but, as i said, i am biased.

    and the flip side is that i can develop and debug a program in the latter group of environments an order of magnitude faster, which is more important in a great many cases.

    sidd

    Comment by sidd — 24 Apr 2014 @ 8:39 PM

  5. So technically there are different climate calenders?

    Comment by Sanele Ncube 14295068 — 25 Apr 2014 @ 5:25 AM

  6. “… temperature: Kelvin, Fahrenheit, or Centigrade”

    What, not Celsius?

    Of coures, Centigrade was very like Celsius but it had a slightly different definition (in terms of the freezing and boiling points of water rather than, via Kelivn, absolute zero and the triple point of water) but mostly it’s just an old name for the scale which isn’t much used these days except in countries which are particularly cantankerous in their use of units (the US and, to a lesser extent, the UK and a few other English speaking countries) – a large part of the problem in the first place.

    Comment by Ed Davies — 25 Apr 2014 @ 5:48 AM

  7. One of the strengths of IDL has been its adaptability in IO along with its back compatibility with FORTRAN formatting statements. The licensing fee is a drawback though.

    Comment by Chris Dudley — 25 Apr 2014 @ 8:32 AM

  8. Fortran: Any language that gets widespread use lasts … a long time, see:
    Languages, Levels, Libraries, and Longevity in ACM Queue.

    In the 1970s, at Bell Labs we went through nightmares trying to get all the various operations support systems (400) to use similar coding for data.

    Comment by John Mashey — 25 Apr 2014 @ 7:09 PM

  9. Maybe you should also talk to those palaeo people Rasmus. Have a look here: http://gergs.net/2014/03/earth-temperature/. Just 8 sources, but a total of 4 different base dates, 3 different reference intervals, 3 different (and not always obvious) spatial coverages … and, of course, 8 completely different data formats.

    One big plea. If you’re going to publish a csv of you data, by all means use fixed column widths, but please please pad them with a tab character instead of spaces. Makes data import much simpler in many environment; not everyone is writing READ(3,8H(10F5.2)) … does that still work?

    Comment by GlenFergus — 27 Apr 2014 @ 11:06 PM

  10. Is the meta-data discussed, also considered an extra layer for a common modeling infrastructure, or a standalone framework? And how relevant is the Earth System Modeling Framework, or more recently known as ESPC? Thanks.

    Comment by prokaryotes — 30 Apr 2014 @ 11:30 AM

  11. A semantic web organized via an ontology is a laudable goal. This removes much of the ambiguity on search terms and on definitions and when tied into a logic engine, you can start to do interesting things. I used the JPL SWEET ontology on my semantic web server: http://entroplet.com/

    Like all good things, the more effort you put into the organization, the more payback you can get. It does take a lot of discipline, which is why these things are still not as popular as free-form search and organization such as Google.

    Comment by WebHubTelescope — 30 Apr 2014 @ 5:56 PM

  12. two classics in this vein

    1
    http://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/

    2
    also, you probably know the 1st hubble scope had a multimillion dollar piece of glass, the primary mirror, that had spherical aberration

    well, while they were building this piece of glass, there was an old time telescope maker – the kind of guy who would grind glass by hand
    And he built, out of cardboard and such, a test jig, and he told PerkinElmer, you multi milldollar mirror has 1o of spherical aberration
    So PE build a multimillion dollar test rig which said the mirror was fine…

    Comment by ezra abrams — 2 May 2014 @ 4:52 PM

Sorry, the comment form is closed at this time.

Close this window.

0.337 Powered by WordPress