[GRASS-user] elevation values of watershed basins

[Dylan Beaudette]

On Thursday 08 February 2007 07:56, temiz wrote:
> Dylan Beaudette wrote:
> > On Thursday 01 February 2007 08:27, temiz wrote:
> >> Dylan Beaudette wrote:
> >>> On Wednesday 31 January 2007 05:14, Markus Neteler wrote:
> >>>> temiz wrote on 01/31/2007 06:53 PM:
> >>>>> hello
> >>>>>
> >>>>> Can you make a suggestion about how I can find the highest and lowest
> >>>>> elevation points of all watershed basins
> >>>>> as vector points ?
> >>>>>
> >>>>> I mean, I have watershed map as vector. I need it's table includes
> >>>>> like "cat", "highest elevation value", "lowest elevation value"
> >>>>>
> >>>>> How can I do that ?
> >>>>
> >>>> If I understand your question correctly, you can do this easily with
> >>>>
> >>>> v.rast.stats
> >>>> Description:
> >>>>  Calculates univariate statistics from a GRASS raster map based on
> >>>> vector objects
> >>>>
> >>>> The help page contains a similar example.
> >>>>
> >>>> cheers
> >>>> Markus
> >>>
> >>> If you have problems with this approach (no reason to suspect
> >>> problems), I have found that 'starspan' coupled with GRASS to be an
> >>> excellent raster+vector reporting tool.
> >>>
> >>> main idea:
> >>>
> >>> start grass
> >>> starspan --vector location/mapset/vector/yourvector \
> >>> --raster location/mapset/cellhd/your_raster \
> >>> --stats output.csv min max avg mode (...)
> >>>
> >>> Output is saved to a CSV file, not quite as nice as v.rast.stats saving
> >>> results back to the attribute table though... The nice thing about this
> >>> approach is that the raster and vector sources can be any GDAL-readable
> >>> data type. For example, I use starspan to compute raster statistics
> >>> within a given radius of a set of points. The raster files are all
> >>> stored in GRASS, and the points are stored in PostGIS.
> >>>
> >>> Cheers,
> >>
> >> /*thank you
> >> */
> >
> > Hi Ahmet,
> >
> >> /*I have never met  */starspan so far. I am looking at its home page and
> >> it will certain to be very useful.
> >> I will work on it.
> >
> > Be sure to, it is a simple compile from source code if you have both GDAL
> > and GEOS installed with their include files.
> >
> >> As I see, you are a soil scientist, so you will understand me better.
> >>
> > :) possibly...
> >>
> >> The logic of my question is based on finding slopes' length as landslide
> >> susceptibility parameter ( I accept slope is waterhed area).
> >
> > so: landslide susceptibility ~ f( slope length ) --> this is a simple
> > raster-based operation: calculate the upslope contributing area for each
> > cell of a DEM. Check the literature for ideas.
> >
> > I am not entirely sure I understand what you are trying to do... Are you
> > trying to estimate a landslide potential for each watershed boundary
> > polygon?
>
>      I am going to build the model using logistic regression ( in R -
> Stat). slope  length  is not only parameter.
>      -- existing landslides are response variable
>      -- rock type,slope angle, slope shape, aspect, slope length (**
> ??**)  are the explatory variables.
>      I have already constructed all maps except slope length.
>      I am going to use watershed basin as slope unit in one model. So
> you are right, each watershed
>      basin ( slope)  I will generate unique landslide potential value.
> something like you see, landslide is
>      a slope movement so I have to say, this slope is unstable but
> adjacent slope is stable)
>
>     Unfortunately, while r.terraflow generates very good and realistic
> watershed basin map, it does not generate
>     half basin map. ( as far as I see r.watershed's basin is not as good
> as r.terraflow's basin)
>

Hi,

Unfortunately I cannot comment on the output from r.watershed vs. 
r.terraflow -- you will need to talk to the developers about this.

Good luck,

Dylan


> >> Firstly, I used PostGis's extent function and got width and length of
> >> polygons (say watershed or  landslide as polygon). But this
> >> polygon's longest dimension doesn't fit dimension in downslope
> >> direction, which is the slope length I am looking for. So, if I found
> >> max and min values of elevation of the polygons, I could calculate
> >> slopes' length.
> >
> > This does not sound like a realistic solution, as the extent() function
> > will return a minimum bonding box -- which provides the maximum 'length'
> > and 'width' of some polygon, but not much useful information.
> >
> >> /*v.rast.stats */gives max and min values of elevation of the polygons(
> >> /*v.rast.stats*/ has  still been running for most 20 hours).
> >> maybe from these values I generate vector points containing min & max
> >> values .
> >
> > Do you have a number of polygons, and the specifications for your DEM:
> > cell size, region size? You may have hit some upper boundary for
> > v.rast.stats. Starpan seems to scale very well, so you may not run into
> > this problem.
> >
> >> I will appreciate if you supply your ideas and recommendations
> >>
> >> regards
> >
> > Sure. With a clearer statement of the problem, and some information on
> > your source data it may be possible to come up with a workable solution.
> >
> > Cheers,
>
> --
> Ahmet Temiz

-- 
Dylan Beaudette
Soils and Biogeochemistry Graduate Group
University of California at Davis
530.754.7341