Am Samstag, 1. Dezember 2007 00:08:08 schrieb Hamish: > Frank Broniewski wrote: > > I am experimenting with r.in/out.xyz. As far as I understand, r.out.xyz > > take a cell center coordinate and writes this and the associated z value > > into something(stout/file). > > Now I have recognised while importing the output with r.in.xyz that there > > is a different cell count with different rasters while the region is the > > same. > > [Glynn answered this probably comes from r.slope.aspect] > > > I am sorting the output by the z value in order to create classes of > > equal area > > perhaps 'r.univar -e percentile=' for percentile in n*(100/num classes) > makes this much easier? > > > Hamish > > > > > ___________________________________________________________________________ >_________ Get easy, one-click access to your favorites. > Make Yahoo! your homepage. > http://www.yahoo.com/r/hs
I tried the percentile option from r.univar already, but this only works good for normal distributed data. I got quite nice (but not perfect) results for my terrain model, but really no good results for slope and other similar distributed data with a single very high peak. Counting cells for equal area is more robust towards the z-value than any statisticial method I tried so far, although I always have trouble when it comes to statistics in general ;-) I am just writing a perl script for this, so that things can be automated as much as possible. r.slope.aspect seems really to blame for the null values. I just overlayed the slope map over the terrain map with the d.rast overlay option and there is indeed a 1-cell wide boundary. I would have never figured that out. Thanks alot. Frank _______________________________________________ grass-user mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/grass-user
