RSS temperatures for May 2015 in pictures

Posted: June 10, 2015 by tchannon in weather

UAH seem late this month, RSS is out. Nothing unusual.


Figure 1, Global mean computed from data used exactly matches official.

What follows is the works for RSS.

I’ll be explaining weighted later on. The point is attempting an honest representation.


Figure 2


Figure 3


Figure 4


Figure 5

The real plots are PDF where pan and zoom are available. This is more important for forensic on certain other datasets where outliers can be drilled down.


Figure 6, for spectacular if not much real interest, unweighted TLT so you can see actual figures, no rivers, still low resolution coastlines, can do far higher, not clickable, PDF here (warning 8MB file)

Oh and you wondered what I have been doing? This stuff does not grow on trees, takes many hours of intense work. Some eyeopeners will be following.

Sniff… RSS all layers, UAH 5.6 and 6 all layers, Hadcrut 3, 4, Crutem, 2, 3, 4, HadSST, 2, 3, Giss250, Giss1200, Vose, Met Office night marine (yes the new one) and probably a few more.

Other projections, yes. Not rotatable, yet, a hard problem with vector plotting.

Post by Tim

  1. Stephen Richards says:

    UAH is 0.27. Came out yesterday because Roy Spencer was away.

  2. tchannon says:

    Always the way Stephen 🙂
    I checked at the data site before opening mouth, site is usually updated later than the figure announcement and so it is. The datasets will appear in a day or so.

    Keep an eye on where timestamps and other changes need a close watch.

  3. tchannon says:

    Articles like this tend to have a chilling effect on the Talkshop, on any place, so why?

    Well, there actually is detail in figure 1 meriting comment, look at the Himalayas. RSS avoid the region yet still it is leaking, not unusual (Andes do too, doesn’t risk Antarctica). As I have pointed out on a number of occasions RSS pushes too hard near the surface so there is a high noise level and artefacts showing.
    All climatic datasets are riddled with artefacts from Nyquist violation too, as though none involved comprehend.

    The temperature does not jump around as these datasets kid on, sometimes so bad that up is down.

  4. tallbloke says:

    Lovely work Tim, congrats on getting it all coded and working. The colour scale is a masterpiece.

    There appears to be anticorrelation between lower strat and lower trop. Needs thinking about.

  5. Paul Vaughan says:

    Is neutrality best represented by green?

    On a purely instinctive level, I perceive the green as demanding complex cognitive processing, whereas black is not perceived …and thus does not compete for processing bandwidth devoted to interpreting signal (blue vs. red).

    I’ve noticed that sometimes white is substituted for black. I regard this as an online spillover from our historical reliance on white paper. That spillover isn’t necessary, but it’s hard to fight that kind of cultural momentum once established. White is at least the second least distracting intermediary.

    When I stare at the images with green for an extended period of time, trying to consciously overcome the instinctive distraction they cause me, I never reach a point where I feel satisfied with the connection of the information to my conscience. The resulting impression left on my mind is a strong will (perhaps not quite a compulsion) to re-express the information in a more effortlessly assimilable form.

    This reminds me of something I’ve seen in a past workplace where all quantitative analysis was carried out using data tables with no graphics whatsoever. It would take people weeks to (barely) interpret (if they succeeded at all) what could be interpreted instantly & instinctively even with rudimentary visualization. It was convenient to some of the manipulative higher-ups that clear vision was thus compromised, as it afforded ample opportunity to supplant the obvious realities with nebulous dreams.

    That’s of course an extreme example, but I’ve always been a fan & supporter of the direction struck by Bill Cleveland (“Visualizing Data”). Without good visualization as a sober diagnostic, statistical analysis drifts uncorrected towards the philosophically abstract, compromised and corrupted by biased null models arising out of nothing more than inspiration & artistic fancy. The illusion of objectivity is algebraically hinged to the dreamy null. In a society governed by p-values, dreamers take control by modifying nulls.

    I do see the potential to play head games with a green null…

    In summary I’m a fan of dark & heavenly (black & white) nulls.
    I advocate going to such extremes for clearer vision!


  6. A C Osborn says:

    Lovely work.
    As an aside are you familiar with the work of Xmetman at

    He has extensive files and programs based on UK weather as well as work on world wide stats.

  7. A C Osborn says:

    Are RSS and UAH using the same baseline now?

  8. tchannon says:

    Yes ACO, xmetman does a lot of attractive articles, he is though accepting the interpretation of others,the only way to cover a lot of ground. My tack is very different and successful in real life, I usually go for ignored detail where from that things might appear. This is bottom up not top down. Different.

  9. tallbloke says:

    Paul V: I think the idea here i that since green is in the middle of the clour spectrum, it is a good colour to choose to represent the middle between extremes. Tim’s colour scheme has continuity. Put your hand on the grass, it’s rarely warm or cold.

  10. tchannon says:

    Paul, quite a subject. Since Rog has commented, this is not ganging up, I’m just going to explain, no right or wrong.

    I’m very aware of culture differences, white hair and wrinkles arrive by ignoring this or longevity.

    You are part of the American technicolor culture where what to other culture might seem extreme, blinding, unsubtle is much of life. No doubt the converse. People vary. I get blinded by many things which to me are just a meaningless jumble.

    There is no one size, hence the popularity of what some call themes.

    The colour system I am using is newly written code here, goes the whole length, the processing uses full blown CIE xyY encoded colour, is aware of RGB, HSI and XYZ as well. The assumption is sRGB display with PC gamma and luminant (err… D50 maybe). I can easily change things but postscript devices such as printer drivers should do this as necessary.
    (don’t know if they still do, US computer monitors are set to a different illuminant (white colour))

    I’ve including breakpoints so the palette can be arbitrarily large, could be dozens of colours and breakpoints for all the software cares. Changing palette is trivial, making a new one very simple, add a new named entry in a table. Name goes on command line.

    In this case the scheme is trivial[*], deliberately simple and linear. This is possible because xyY disconnects brightness, xy define the colour axis, Y controls brightness, hence the very smooth graduation, which is linear interpolation between defined breakpoints, no need for fancy laws.

    The rational used involves various compromises and subtlety which won’t be obvious.

    You would be perfectly correct in stating people don’t like green, the least popular primary colour, you won’t find many green cars, green clothes, green much at all. We have to live with a fixed spectrum and human sense thereof, animals see differently and humans vary a little. (CIE work, origins might be of considerable interest, good history)

    Next up as not liked is also a major colour, yellow, won’t find much of that around either, you will though see shades, off green, off yellow and so on.

    The limits I have chosen are deliberately not saturated. the reason is subtle and specific to the problem. Bright red for example is seen by the eye as highly dominant but the work is about removing all dominance, firstly by mapping a sphere to flat as equal area but human perception is able to see a tiny bright dot overriding the actual area significance and that is the reason for the cosine weighting against latitude which also happens to be the primary function in computing a global average (weighted mean). In combination with muting a dominant colour the intent is evenness. This doesn’t mean I am right it’s just my best judgement given what I know.
    Context matters, this is about global

    The green is also muted, the exact colour, way the whole thing works is trying to avoid those particularly “yuk” yellow greens.

    You mention a white background, yes, uninteresting and customary. The alternative is?
    I prefer something off white, just haven’t had time to focus on a small detail.

    Things can be changed.

    The ideal? We set up a web site which serves images or persuade someone with the time to produce an interactive display program. This kind of thing is strongly influential on the standards people expect, lead by walking, show me.

    * trivial palette? The entire gamut comes from this.
    type5 = { {25,25,212},
    {124,205, 124},
    segs = {0, 0.5, 1 },
    kind = ‘RGB’

  11. Paul Vaughan says:

    No right & wrong, for sure.

    Taste, preference, use in mind.
    e.g. marketing (success determined by sales) vs. art contest (aesthetics) vs. … whatever else.

    I look forward to seeing the off-white (cream?) version whenever.

    Interesting work for sure — lots of tedious programming, but very rewarding too.

    I couldn’t resist taking another shot at null hypotheses …the dreamer’s ticket to control in p-valued democracy = null manipulation. (There’s a good reason why serious journal’s are beginning to ban it …’s always based on assumptions and they’re malleable…..)

    Keep up the good work Tim.


  12. Paul Vaughan says:

    There are plenty of color schemes out there that are ordered according what we might call “mysterious” “logic” …so I should acknowledge: even if it doesn’t meet my personal preference, the color scheme Tim has chosen has the benefit of having an intuitive gradient.

    The things I would do if I had time for programming…