[GRASSLIST:2878] Re: visible segments in the DEM

Roger Miller rgrmill at rt66.com
Thu Jan 10 08:26:35 EST 2002

On Thursday 10 January 2002 06:08, Steve Trick wrote:
> Dear grasslisters,


Your description sounds like the effect you would get from representing a 
topographic surface with integer values.  That happens when you use a raster 
map of type CELL.

I think that v.surf.rst is supposed to produce a map with floating point 
values, so that shouldn't be your problem.  Is it possible that your input 
elevations are integer values, or that you are doing something to the map 
after it is calculated by v.surf.rst that converts the floating point values 
to integers?

Roger Miller
Lee Wilson and Associates

> I am conducting a project looking at lines of sight in and around a
> river valley.  My problem regards generating a satisfactory DEM on
> which to base the analysis.  I have been using v.surf.rst to generate
> a surface from vectorised contours but I always get visible segments
> in the DEM, which translate as unnatural 'steps' on the surface
> model.  I have tried increasing the minimum number of points used for
> interpolation, npmin, but to get good results, a value of around 2000
> instead of the default 200, the calculation time gets into the realm
> of weeks rather than hours and this is impractical.
> The problem mainly occurs in flat areas of the terrain, such as
> floodplains, where the contours are few and far between, and I guess
> the program can't get enough points for interpolation.  Can anybody
> suggest a way of getting a 'smoother' DEM, either by adjusting the
> parameters of v.surf.rst or using another interpolation program? I
> have tried r.surf.contour but find this is quite crude and it tends
> to truncate the tops of hills.  I have looked at some of the other
> grass modules but find not all of them are implemented in GRASS
> 5.pre2 which I am using.
> Any comments welcome.
> Thanks in advance,
> Steve
> -----------------------------------
> Steven Trick BA MSc
> School of History and Archaeology
> Cardiff University
> PO Box 909
> Cardiff
> Wales
> CF10 3XU
> email: TrickSG1 at cf.ac.uk
> mob: 07799897107
> http://www.cardiff.ac.uk/hisar/people/st/

More information about the grass-user mailing list