[GRASS-user] Precipitation color table?

[Hamish]

Luigi Ponti wrote:
> I just ran into this page
> http://geography.uoregon.edu/datagraphics/color_scales.htm
> 
> that includes, among others, precipitation color tables. I
> don't know if that can be useful.
> 
> The page of this Lab also provides an interesting paper titled "End of 
> the Rainbow" which elaborates on why one should not use continuously 
> varying color schemes (and absolutely no rainbow color table). The way 
> to go seems to be banded color schemes -- a color scheme with equally 
> sized bands of constant color [1].
...
> [1] Borland D, Taylor II RM (2007) Rainbow color map (still) considered 
> harmful. IEEE Computer Graphics and Applications, 27, 14-17.

In matters of human perception the individual experience is not subject
to hard rules. Of course it is important to think about how the color-
blind will see things and how it will look printed in greyscale, etc.

An interesting subject to me, thanks for the link/look forward to reading
it. It reminds me that I still need to read "How to lie with maps":
  http://www.eyrie.org/~eagle/reviews/books/0-226-53421-9.html
(in the spirit of "How to lie with statistics")

It is really quite amazing/scary how much changing the colors affects
the interpretation and is able to hide/highlight features. After spending
a fair bit of time with the issue I personally feel it is most honest to
use a color scale which can be described as a continuous mathematical
function. e.g. log()+BCYR or something directly based on the univariate
stats of the data. -> Take the biased "what looks nicest" human out of
the picture. Choosing arbitrary band limits on a continuous value dataset
just compounds the opportunities for arbitrary human influence. The
statistical equivalent is to slowly vary the number bins in a histogram
while the peaks seemingly double and halve.


...so it is up for debate :)


> As I side note I was wondering how to draw such a color scheme (i.e. 
> with equally sized bands of constant color) in GRASS.

declare the number twice, e.g.

r.colors color=rules << EOF
0 red
100 red
100 yellow
200 yellow
200 cyan
300 cyan
300 blue
400 blue
EOF


Hamish