[GRASS-user] Some questions about DEMs

Hi

I have a couple of questions about DEMs, and how they work in GRASS. My application is hydrological and hydraulic analysis, and at the moment I’m investigating using r.sim.water to generate floodlines.

Usually, for this sort of analysis, I would have 5m contours for the catchment and a detailed survey (i.e. spot heights, sometimes with triangles) for sections of river. So my first question is, what is the best way to generate a DEM from this information? I know that I can use r.surf.contour to create a DEM from the contours, and I’d assume that something like v.surf.rst would be best for the points, but I need a single DEM for the analysis.

The second problem is that I don’t need a highly detailed DEM for the catchment, I only need detail along the watercourses - i.e. where I have the detailed survey. If I create two DEMs at different resolutions, can I patch them without losing detail?

My third question is about creating a mask. If I’m working along the coast, or if I’m creating a detailed DEM along the watercourse, I only want raster data where there is base data (i.e. contours or points). Is there a way to quickly generate a mask for this?

Regards
David

W dniu 09.07.2010 12:29, David Townshend pisze:

I have a couple of questions about DEMs, and how they work in GRASS. My
application is hydrological and hydraulic analysis, and at the moment I'm
investigating using r.sim.water to generate floodlines.

Usually, for this sort of analysis, I would have 5m contours for the
catchment and a detailed survey (i.e. spot heights, sometimes with
triangles) for sections of river. So my first question is, what is the best
way to generate a DEM from this information? I know that I can use
r.surf.contour to create a DEM from the contours, and I'd assume that
something like v.surf.rst would be best for the points, but I need a single
DEM for the analysis.

For heterogenous input data I'd try the nn algorithm in nnbathy, through r.surf.nnbathy (http://grass.osgeo.org/wiki/Addons#r.surf.nnbathy).

The second problem is that I don't need a highly detailed DEM for the
catchment, I only need detail along the watercourses - i.e. where I have the
detailed survey. If I create two DEMs at different resolutions, can I patch
them without losing detail?

You can patch 2 rasters of different resolution, but not to loose information, the output raster map needs to be of the resolution of the more detailed input raster map. This will result in a flat squares + smooth valleys surface.

My third question is about creating a mask. If I'm working along the coast,
or if I'm creating a detailed DEM along the watercourse, I only want raster
data where there is base data (i.e. contours or points). Is there a way to
quickly generate a mask for this?

v.hull + v.to.rast?

Maciek

--
Maciej Sieczka
http://www.sieczka.org

Maciek wrote:

For heterogenous input data I'd try the nn algorithm in
nnbathy, through r.surf.nnbathy (http://grass.osgeo.org/wiki/Addons#r.surf.nnbathy).

It is great for sparse points, but for contour lines + hydrological
modeling it is not the best. See nn valley staircase effect in
http://grass.osgeo.org/wiki/Contour_lines_to_DEM

regards,
Hamish

W dniu 10.07.2010 02:31, Hamish pisze:

Maciek wrote:

For heterogenous input data I'd try the nn algorithm in
nnbathy, through r.surf.nnbathy (http://grass.osgeo.org/wiki/Addons#r.surf.nnbathy).

It is great for sparse points, but for contour lines + hydrological
modeling it is not the best. See nn valley staircase effect in
  http://grass.osgeo.org/wiki/Contour_lines_to_DEM

Interpolating from contour lines has always been a nightmare for me :). What is the best approach in your opinion?

Maciek

--
Maciej Sieczka
http://www.sieczka.org

In the end I combined the contours and points and used r.fillnulls to
generate a 5m DEM. It seemed to run quite a bit faster than
r.surf.contours and gave fairly good results. Not perfect, but given
the data I'm starting with, its not bad either.

My next problem is getting r.sim.water working. Does anyone have any
experience with this? I have a quite a few questions about it:
- What exactly do the diffusion terms and weighting factor do?
- What is the best way of simulating hydraulic controls such as
bridges or weirs (is it even possible)?
- Is there a correlation between the time used for iterations (niter)
and the storm duration? If not, then how do I determine it?
- How applicable is this model to flow in rivers?
- I want to try to use this model for floodline generation. Does
anyone know of any work that has already been done on this, or is the
model totally unsuitable?

I had hoped to figure some of this out by myself, but whenever I try
running the simulation I get empty output. Can anyone help?

David