[GRASS-user] Strategies for using v.surf.rst (getting rid of those segmentation boxes)

[Markus Neteler]

On Fri, Jan 18, 2008 at 2:53 AM, Jonathan Greenberg
<greenberg at ucdavis.edu> wrote:
> Grassers:
>
>        I'm using v.surf.rst, and I'm wondering what the best strategies for
> getting rid of those rectangular segmentation boxes are, while also not
> having the high local peaks around lone sample points?  Thanks!

I had the same problem and worked around it but avoiding segmentation.
Example:

6921 input points
...
WARNING: Points are more dense than specified 'DMIN'--ignored 5575
points (remain 1346)

Now we can calculate:
6921-5575=1346.

For smooth connection of segments, npmin > segmax.

In my interpolation, I had the problem that I got errors like
  ERROR: segmentation parameters set to invalid values: npmin = 300,
segmax = 850
  for smooth connection of segments, npmin > segmax (see manual)
ERROR: input failed!

In the v.surf.rst/v.vol.rst, there is a MAXPOINTS value defined (surf.h).
MAXPOINTS simply says how many input points are allowed to be computed
without segmentation. I had to increase the value of MAXPOINTS in surf.h to
1346 (or a bit more) and then to use segmax=1346 to avoid segmentation.
Tricky but works.
The idea of the authors is to change v.surf.rst/v.vol.rst to a more dynamic
management of this value, the question is how to implement that.

Summary: avoid segmentation if possible.

Markus