Visualising 20 years of Climate Change

I have been experimenting with different ways to visualise triangulated global temperature data in 3D. Here are the monthly anomaly values for the last 20 years. Of course you can only see one side of the earth at a time, so I also slowly rotate it. The monthly variability is much more striking than a slow overall warming trend.

The temperature colour scale used in all animations is this one.

This is the latest data for March 2017 showing explicitly the spherical triangulation. The underlying data I use is GHCN V3 station data normalised to 1961-1990 combined with HSST3 ocean data.

There is also a youtube video of 137 years starting in 1880. The distorted earth shape in early years is caused by poor sampling.


I think that the best way to view these results on the Web is probably to use WebGL. I have been putting this off as from my past experience Javascript can quickly become a can of worms. However I may eventually to delve into this as WebGL allows to interact directly with 3-D data.

This entry was posted in climate science, Internet and tagged . Bookmark the permalink.

3 Responses to Visualising 20 years of Climate Change

  1. Nick Stokes says:

    “I think that the best way to view these results on the Web is probably to use WebGL.”
    That was my experience. I described my early learning process here.

    It’s an interesting presentation, though I had trouble reading the dates on your animation. Partly the moving, but also the red on black. I find a non-linear colour scale useful, otherwise you end up with pale oceans where the resolution is often best. I use mostly tanh, also artan. Colour=tanh(temp), with appropriate scaling.

    • Clive Best says:


      I am going to look closely at what you have already done. I already stole your continents coastline coordinates in order to draw them in 3D, by converting from spherical coordinates! That was much easier than IDL who use shape files. Where did you find the data?

      I don’t think there is any correct solution for the colour scheme, because it is too subjective. It depends whether you want warming to appear dangerous (like NASA), or not that bad yet (like me). Ideally I’d like to build an interface where a user can select different colour schemes.

      • Nick Stokes says:

        I use the map outline that comes with R. I’ve looked elsewhere, but that’s the best compromise I can find between size and accuracy. There is a more complete set that has the names of the various entities, but it gets very detailed.

Leave a Reply