[GRASS-user] Streams under r.watershed

[Hamish]

Kurt Springs wrote:
> I have generated some watershed maps some time ago.  I intend to do it  
> again from my PhD project.  I was hoping someone could explain what  
> the various maps generated mean.  The Streams maps especially seem to  
> be trying to tell me something as do the Basins.  I tried generating a  
> legend for streams but I don't know what the numbers mean and the  
> colors seem to be repeating themselves.

We are lucky in that for once the help page is well developed.

from the man page:
       basin
              Output map: Unique label  for  each  watershed  basin.
              Each  basin will be given a unique positive even inte-
              ger.  Areas along edges may not  be  large  enough  to
              create an exterior watershed basin.  0 values indicate
              that the cell is not  part  of  a  complete  watershed
              basin in the current geographic region.

       stream
              Output map: stream segments.  Values correspond to the
              watershed basin values.


So stream segment cat numbers correspond to the basins that they bisect.
If a stream segment falls within watershed #36, its cat will be 36.

The dominant stream in the watershed will split the basin into two
half.basin*s, "The right-hand  side  cell  of  the  watershed
              basin  (looking upstream) are given even values corre-
              sponding to the values in basin.  The  left-hand  side
              cells  of  the  watershed  basin  are given odd values
              which are one less than the  value  of  the  watershed
              basin."

[*] the option has been renamed to half_basin in GRASS 7


Basins and Streams maps produced by r.watershed use the same color table.
If there are less than 10000 watersheds it is unlikely that colors will
exactly repeat. If there are >10000, the colors are created by
G_make_random_colors() and at most 1024 colors are used(?).


If you have non connecting tributaries you wish to label, connect contiguous segments with r.cost or v.net.iso. (search the archives for
exact recipe)

You can create a custom color table with r.colors if you prefer.


Hamish

ps- Are you testing Markus Metz's new version of r.watershed? It is
supposed to be a zillion times faster.
  http://thread.gmane.org/gmane.comp.gis.grass.devel/29685/