[GRASS-dev] v.surf.idw broken

GRASS GRASS developer
1

Hi,
I can not get reasonable result (compared with v.surf.rst) from
v.surf.idw using attached input vector points

is it possible, that v.surf.idw is kind of broken?

[xy location]

v.in.ogr dsn=vzorky.shp out=vzorky

v.surf.idw in=vzorky out=dpidw column=dp
v.surf.rst in=vzorky elev=dprst zcolumn=dp

thanks

jachym
--
Jachym Cepicky
e-mail: jachym.cepicky gmail com
URL: http://les-ejk.cz
GPG: http://les-ejk.cz/pgp/JachymCepicky.pgp

(attachments)

vzorky.zip (20.6 KB)

2

On Tue, 13 Apr 2010, Jachym Cepicky wrote:

Hi,
I can not get reasonable result (compared with v.surf.rst) from
v.surf.idw using attached input vector points

is it possible, that v.surf.idw is kind of broken?

Hello,
I can confirm that I can reproduce the strangeness, that it only happens when *not* using the -n flag (i.e. when indexing points) and that it happens when interpolating a relatively small number of points into a region with a relatively high resolution. E.g. for the given data I observe strangeness for a resolution of 10, but it goes away either by specifying the -n flag or decreasing the resolution to 15.

It does indeed look to be caused by my buggy code from 2003, or (perhaps less likely; I'm not sure) some later change to it - specifcally something to do with the "shortlist" method of avoiding an intensive search for nearby sites/points when the number of points is relatively small compared to the region size.

I can't see anything obviously wrong but I'll look into it over the next few days...

Paul

3

Thanks

I can confirm, that using -n flag works

Jachym

2010/4/13 Paul Kelly <paul-grass@stjohnspoint.co.uk>:

On Tue, 13 Apr 2010, Jachym Cepicky wrote:

Hi,
I can not get reasonable result (compared with v.surf.rst) from
v.surf.idw using attached input vector points

is it possible, that v.surf.idw is kind of broken?

Hello,
I can confirm that I can reproduce the strangeness, that it only happens
when *not* using the -n flag (i.e. when indexing points) and that it happens
when interpolating a relatively small number of points into a region with a
relatively high resolution. E.g. for the given data I observe strangeness
for a resolution of 10, but it goes away either by specifying the -n flag or
decreasing the resolution to 15.

It does indeed look to be caused by my buggy code from 2003, or (perhaps
less likely; I'm not sure) some later change to it - specifcally something
to do with the "shortlist" method of avoiding an intensive search for nearby
sites/points when the number of points is relatively small compared to the
region size.

I can't see anything obviously wrong but I'll look into it over the next few
days...

Paul

--
Jachym Cepicky
e-mail: jachym.cepicky gmail com
URL: http://les-ejk.cz
GPG: http://les-ejk.cz/pgp/JachymCepicky.pgp