Considering the length of this thread and the various issues mentioned, it seems that I need to update the manual of r.stream.extract a bit. I will try to address the issues:
r.stream.extract is designed to work with a DEM only or with a DEM and a flow accumulation map as input. This flow accumulation map can be created with any method and any software, tested with r.watershed, r.terraflow, TAS (now Whitebox), and SAGA. The methods tested are D8, MFD, and D-inf. Some modules/software need a hydrologically conditioned DEM without sinks as input (have a look at r.hydrodem in GRASS 7 addons), and not the raw DEM. In this case the modified DEM used to calculate flow accumulation must be used as input for r.stream.extract if a flow accumulation map is also provided. In case of r.terraflow, the filled DEM produced by r.terraflow must be used as input. The computational region should be aligned to the DEM in order to get accurate results. Aligning to the resolution of the DEM is not enough, i.e. you should use g.region align=DEM instead of g.region res=XXX -a. r.stream.extract should work if only a subregion is to be analysed by setting the computational region accordingly and/or creating a MASK e.g. for a specific basin. The flow accumulation map used as input to r.stream.extract can be modified, e.g. by multiplication with a weighing map. In this case valid accumulation values should not be converted to NULL, but zero is fine. The weighing map should not be used to mask out certain areas, instead it should be used to weigh the original accumulation map according to slope, valley type, rainfall, etc. The module will exit with a fatal error if for a given cell there is a NULL accumulation value provided, but the DEM is non-NULL. In this case a MASK must be created for the subregion covered by the accumulation map. HTH, Markus M On Sat, Apr 7, 2012 at 11:33 PM, Rich Shepard <[email protected]> wrote: > On Sat, 7 Apr 2012, Helmut Kudrnovsky wrote: > >> if the (sub-)watershed are already given, maybe the easiest way to define >> the region extent for raster processing of your interest is using g.region >> [1]: > > > Helmut, > > It's been long enough that I totally forgot about this. I don't need a > mask if I set the region to the basin of interest. > > Thanks very much for the reminder! > > Rich > > > _______________________________________________ > grass-user mailing list > [email protected] > http://lists.osgeo.org/mailman/listinfo/grass-user _______________________________________________ grass-user mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/grass-user
