Saturday, December 13, 2008

Mathematica WeatherData: 17,168 stations

Click the picture to zoom in.

Would you recognize the pattern on this image? Well, yes, the axes are the longitude and latitude and you can see the continents, especially their industrialized subcontinents, on it.

The picture shows 17,168 weather stations across the world. You see that some regions are covered more densely than others. Mathematica 7 users have access to all the detailed current and historical weather data from these stations. 

For example, you can pick a station in Florida and determine when its managers started their barbecue party last Sunday (most U.S. weather stations are cleverly attached to a grill, see

The colors on the picture above don't show any temperature - just random numbers associated with the individual weather station closest to a given point. It was actually pretty easy to draw the Voronoi diagram above:
Mathematica command drawing the picture
If you gain access to Mathematica, you will be able to calculate pretty much everything, find errors, fluctuations, discrepancies, urban heat effects, and any kind of statistics you want.

Sorry, the newest Mathematica 7 is needed.

Average temperature on Thursday (two days ago). Click to zoom in. Note that the Northern Hemisphere seems cooler. It's called "coming winter" (or "summer" in Australian English). You can see that this difference primarily affects the land.

The temperature difference between Dec 11th, 2008 minus Dec 11th, 2005. Note that it's not only the difference between the same seasons but the ENSO index was also identical at these two moments. Unfortunately, the transition to a cool PDO phase is not manifest on these land-based stations. At least, you see the Siberia getting friendlier and Canada freezing over. Click to zoom in. Globally, December 2008 is about 0.15 °C cooler than December 2005.


  1. Wolfram doesn't appear to disclose the provenance of the data they republish, for example to what extent it has been adjusted or may be subject to future retroactive adjustments. Not that this ever happens in climate research ...

  2. Dear fche, well, I think that it only makes sense to ask whether some data have been adjusted if you know what was the old value and what is supposed to be its error.

    I leave these things to whoever manages the stations - and to Wolfram Research. I surely don't have an access to "even more complete" data about these stations, and I don't want to do the thermometer-maintenance job of someone else, so it is my choice to consider the functions in Mathematica to be a trustworthy black box.

    So far it doesn't look like there are errors of 12 deg C affecting the whole continent for a month, like in Hansen's data. ;-) If there are some strange things, the amount of data is sufficient so that you can actually extract some of these man-made errors by a proper statistical analysis.

    So let me say that I view these data as a useful extensive collection of numbers that are hard to get otherwise and that can be useful for various serious hypothesis-checking, not as a gadget to create criticism against someone.

    What I find more annoying is that it takes a lot of time to download all the "horizontal" data (from a few moments but the whole world) that you need for a typical "global" calculation. It seems that the local data (for long intervals) can be obtained much more quickly. It seems that they were thinking about local graphing, not about global maps, when they were designed the database.

    The sea surface temperature are not covered well, either. It would be good if there were UAH/RSS MSU and SST functions, too.

  3. It would be interesting to see also sea related data:

  4. I would suggest to use the term 'developed countries' or 'industrialized country' instead of 'civilized continents' which is quite offensive for those that are not so clear in the map. Some of the places I could easily identify in your map are clearly more developed and have more advanced economies but they are not necessarily 'the civilized ones'.

    Kind regards from a citizen of those 'least developed countries'.