Hi!
I needed to be able to modify the power of the distance part of the idw
algorithm (the beta). Well I did it, and I've attached the diff.
just apply it in the v.surf.idw directory with the -p0 argument. It will
modify main.c
--Wolf
--
<:3 )---- Wolf Bergenheim ----( 8:>
v.surf.idw.power.patch (3.21 KB)
I needed to be able to modify the power of the distance part of the
idw algorithm (the beta). Well I did it, and I've attached the diff.
often refered to as the "coefficient of friction", 1/d^Cf
A value of 2 gives the smoothest result, but values greater than two
will help to constrain open boundary effects. A compromise is to run at
2.5 or run the module twice, once for Cf=2 and once for Cf=3. Then use
r.series or r.mapcalc to average the results. Values at real data points
should be the same, only the interpolated values are tweaked, so legal.
For giggles, other radial basis functions can be inserted too. (log() etc)
Hamish
(I have an interesting paper to publish on this at some point)
On 06/12/05 16:31, Hamish wrote:
I needed to be able to modify the power of the distance part of the
idw algorithm (the beta). Well I did it, and I've attached the diff.often refered to as the "coefficient of friction", 1/d^Cf
A value of 2 gives the smoothest result, but values greater than two
will help to constrain open boundary effects. A compromise is to run at
2.5 or run the module twice, once for Cf=2 and once for Cf=3. Then use
r.series or r.mapcalc to average the results. Values at real data points
should be the same, only the interpolated values are tweaked, so legal.
Thanks. That was new to me (well almost everything in GIS is new to me).
For giggles, other radial basis functions can be inserted too. (log() etc)
LOL. That could be fun. Perhaps I could add another parameter
'radial_basis', or something, which could take arguments like 'log'
'power' and then add the parameter 'base' that would be used as the base
of the log operation.
--Wolf
Hamish
(I have an interesting paper to publish on this at some point)
I'm looking forward to reading it (:
--
<:3 )---- Wolf Bergenheim ----( 8:>