[R-sig-Geo] [Soc] GRASS module for kriging - call for users [was: R-sig-Geo Digest, Vol 69, Issue 21]

[G. Allegri]

My two cents.
Some of the best interface features I've found in Isatis, one of the
best commercial geostatistical sw I've used, are:

 - interaction with the variogram map to select direction and plot the
ESPs together differentiating the various selected dirs with colors.
It permits to reveal anisitropies structures with a glance.
 - interaction with the variogram model (plotted with the ESP) to fit
it by hand. In practice you can change the model parameters (sill,
nugget, range) by clicking on the plot handles.
 - let the user select pairs from the variogram cloud and visualize
them on the map. It permits, for example, to see where high variances
pairs (linked with a line) are located.

Another useful feature is the visualization of the anisotropy
parameters on the variogram map (the cone). And, last, the possibility
to define visually the neighborhood as circle/ellipse.

Ok, lots of wishes, and I don't know how hard it would be to develop
them. I've just written them to share...

bye,
giovanni


http://www.geovariances.com/software/video-data-investigation-with-isatis-exploratory-data-analysis-ar0353.html

2009/5/28 Anne Ghisla Insubriae <a.ghisla at studenti.uninsubria.it>:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Dylan Beaudette ha scritto:
> [snip]
>> Hi. It is alway nice to see people working GRASS SoC projects, thank you!
>>
>> I would just add-- I think that all efforts to build onto existing
>> libraries would be a good thing. That is to say, try and work with
>> GRASS-R-gstat: this is by far one of the best implementations of
>> geostatistics available. Paul's automap would be another likely
>> candidate .
>
> Hi Dylan, and sorry for late response,
>
> there are two packages that provide kriging in R, and the discussion in
> this thread leads to use one of them to avoid the differences in results
> coming from implementation differences. On your advice I'll choose gstat
> and automap, even if geoR looked very complete.
> Maybe I could offer the user the choice bteween the two packages, in a
> 2.x release of the plugin. This could allow R users keep using their
> preferred algorithms, what do you think?
>
>> The real problem with a single GRASS module for 'kriging' is that the
>> operation requires careful thought and familiarity with the data-- not
>> really something that can be easily generalized to the conceptual
>> basis of a single GRASS module. Perhaps asking Edzer about room for
>> user-interface improvements in the gstat code would yield some good
>> starting points.
>
> Good - I'll ask him soon.
>
>> Cheers,
>> Dylan
>
> cheers,
> Anne
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (MingW32)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iEYEARECAAYFAkoeSaQACgkQrgxAvFOwDdiYdQCeKVwmLg3D1DW4WTHAfficfN4b
> 7wcAn34f0lTzGrgstcWMJX5VqsKUh2ec
> =NmbT
> -----END PGP SIGNATURE-----
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>