[GRASSLIST:5520] Re: r.describe (and r.report, r.stats) on floating point maps.

[Eric G. Miller]

On Mon, Feb 10, 2003 at 11:58:39AM +0100, Aldo Clerici wrote:
> Hi GRASS users (and developers),
> there's something strange in the results of r.describe (and r.report,
> r.stats) when applied to a floating point map derived from a transformation
> of an integer values map. The output is a list of range classes instead of
> the actual floating point classes. An example (Spearfish data):
> 
> r.mapcalc "landuse.1 = landuse * 1.0"  (or r.mapcalc
> "landuse.1=float(landuse)"
> r.describe landuse.1
> * 1.000000-1.027451 1.988235-2.015686 .........7.972549-8.000000
> 
> As the original map contain integer values from1 to 8, I suppose the correct
> result should be 8 floating numbers from 1.0 to 8.0.
> Please note:
> a) The same  results are obtained with r.stats and r.report, but d.what.rast
> output the correct values (so the problem is not in the r.mapcalc).
> b) I applied the operation to other maps  (elevation.dem for example) with
> the same results.

"Categories" for floating point rasters are described via 
ranges of values.  Just because the ranges are not whole numbers does
not mean the underlying raster has values other than whole numbers.
If you want to change the ranges, see the 'nsteps' argument, and also,
r.quant.

-- 
echo ">gra.fcw at 2ztr< eryyvZ .T pveR" | rot13 | reverse