Raster headers & unsupervised classification

[Marc Pelletier]

At 05:31 PM 12 17, 1999 +0000, you wrote:
>You see that "G_get_c_raster_row" is used. This indicates (I think so)
>that i.cluster only works with integer maps. This means you have
>to use either r.mapcalc or r.quant to make integer maps from your float
>maps.
>
>Hope this helps
>
> Markus Neteler
>
>

Actually Markus, once I got the beta 5 version it works fine on my floating
point rasters but still fails on the integer ones! That's ok because I
really wanted to work with the floating point data.

But my next question is...

Does the clustering and classification implicitly assume that all the data
varies from 0 to 255? I ran i.cluster and i.maxlik on a set of 4 rasters
where one represented magnetic data, which varied from 60000 to 61500
gammas, and the other 3 were radioelement concentrations which varied from
0 to 10,15 and 25. The result was totally overwhelmed by the magnetics. The
K, Ur and Th radioelements are strongly correlated with each other and have
little or no correspondence with the magnetics.

Thanks

Marc Pelletier
---------------------------------------------------------------------------
Data Donkey - Geophysical Data Processing, Consulting and Software Development
               Creator of the POWER TOOLBOX utilities for OASIS montaj
contact: mailto:marc at datadonkey.com  tel (306) 931-6853 or
http://www.datadonkey.com