[GRASS-user] r.watershed: pretty flat areas

[Markus Metz]

Hi all,

r.watershed in trunk r42236 has a new option to beautify flat areas,
activated with a new -b flag, works with both SFD and MFD. The method
is adapted from Garbrecht & Martz (1997).

The justification for the new option is that some regard straight flow
lines in flat areas as an issue, e.g.

Nardi et al (2008), Hydrogeomorphic properties of simulated drainage
patterns using digital elevation models: the flat area issue
http://dx.doi.org/10.1623/hysj.53.6.1176

For r.watershed, this is not that much of an issue because it does not
create flat areas by sink filling. For naturally flat areas like lakes
and broad rivers, it is not possible to determine the "correct"
drainage direction only from a DEM, information about river and lake
bottom would be needed to more accurately determine flow direction,
but DEMs only show water surface (apart from bathymetry). Therefore I
chose to describe the new option as "Beautify flat areas" and not as
"correct flat areas". The new option is currently only implemented for
the all in memory mode, and it requires more memory and more time.
Contrary to other implementations, this new one modifies elevation
values only temporarily to adjust drainage direction, flow is
accumulated using the original elevation values. I'm not really
convinced about this option, but since methods to modify flat areas
appear regularly in the literature, I added one. If there is no strong
support for the new option, I would remove it again. I guess it's
mostly useful to produce pretty maps.

Markus M