Colorbar for categorical data

I have a spatial, categorical dataset that I have plotted up using pcolor. The dataset contains latitude, longitude and vegetation type data. Vegetation type is an arbitrary number between 1 and 19 that corresponds to a vegetation label.

I am looking for a way to create a colorbar "legend" for this data.

At the moment, when I use pcolor and add a colorbar, I get a bar that goes from 1 - 19 with all the different colours in between. This colorbar means nothing because the data are not ordinal.

I am hoping to find a way to make little boxes for each individual color, and put a label next to the box. Something like the maps of Australia seen on this site http://www.environment.gov.au/node/22149

Is there a way to code this in Matlab, given that I have 19 different vegetation types, and putting labels next to the colorbar would be too busy?

Thanks for your help.

Answer 1

[Image Analyst]

Have you seen the demo for colorbar where they change the labels from numbers to some custom words? Try that:

surf(peaks(30))
colorbar('YTickLabel',...
  {'Freezing','Cold','Cool','Neutral',...
   'Warm','Hot','Burning','Nuclear'});

[Image Analyst]

By the way, did you know pcolor() does not plot every element in your array? Try this:

pcolor(randi(255,3,3)) % Display 3 by 3 array (hopefully)

and see how many tiles get painted for this 3 by 3 array. Were you expecting a 3 by 3 array? Surprise!!! Then replace pcolor with image() or imshow().

[Claire]

Thanks for your help.

The problem with that solution is that it only works when you have a small number of categories. I have 19 in total, and when you try to label the colorbar directly, it just becomes a big mess.

I need to find another way to do this that works for a large number of labels.

[Image Analyst]

If you're going to have 19 segments over a screen size, or a certain height on the paper, then the only way is to have the font size be smaller. Do you not consider your example to be a mess? Why not? Because the letters are separated? Well they are small font sizes - it's a tradeoff.