Hi Julia,
initial variogram values just need to be sufficiently reasonable from where
to start fitting.
I know about some known rule-of-thumb:
+ sill: the total variance of your data;
+ range: half the diagonal of your bbox;
+ nugget: the measurement error.
You can also decide whether to keep some of these parameters fixed, instead
of being shifted during model fitting: see arguments in ?fit.variogram!
Make sure you have at least ~30 pairs of points per lag, and 10 lags
minimum.
You might also want to increase the resolution of the lags at short
distances, check the `boundaries` arg of variogram()).
Regarding anisotropy, you can read Notes and Examples in ?vgm!
Piero
On 25 September 2012 10:57, Julia Chacon Labella <juliacha...@gmail.com>wrote:
> Please excuse these basic questions, but I am a bit lost in geostatistics.
>
> I am trying to make a kriging of the soil properties of a plot. I am
> already using gstat with R and I don't have any problems with the scripts
> and so on...but, I don't really know how to choose the nugget, psill and
> range for the variogram model. Moreover, I am not sure about how to include
> the anisotropy in the model, I have already include a linar model in some
> cases, to remove any trend.
>
> I am following the steps recomended in the gstat help, but I am a bit
> befuddled.
>
> Let me show you my scripts (the psill, range and nugget are choosen just to
> make the test).
>
> coordinates(corg)= c("x", "y")
>
> v = variogram(SOC~1, corg)
>
> SSErr.a <- numeric(4)
> names(SSErr.a) <- c("Exponencial","Esférico","
> Gausiano","Lineal")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Exp", range=5, nugget=0.1))
> SSErr.a[1] = attr(v.fit, "SSErr")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Sph", range=5, nugget=0.1))
> SSErr.a[2] = attr(v.fit, "SSErr")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Gau", range=5, nugget=0.1))
> SSErr.a[3] = attr(v.fit, "SSErr")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Lin", range=5, nugget=0.1))
> SSErr.a[4] = attr(v.fit, "SSErr")
> SSErr.a
>
> v = variogram(SOC~x+y, corg)
>
> SSErr.b <- numeric(4)
> names(SSErr.b) <- c("Exponencial","Esférico","Gausiano","Lineal")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Exp", range=5, nugget=0.1))
> SSErr.b[1] = attr(v.fit, "SSErr")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Sph", range=5, nugget=0.1))
> SSErr.b[2] = attr(v.fit, "SSErr")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Gau", range=5, nugget=0.1))
> SSErr.b[3] = attr(v.fit, "SSErr")
> v.fit = fit.variogram(v, vgm(psill=0.18, model="Lin", range=5, nugget=0.1))
> SSErr.b[4] = attr(v.fit, "SSErr")
> SSErr.b
>
> Thanks a lot,
> Cheers.
>
> --
>
> Julia Chacon Labella [PhD candidate]
>
> Área de Biodiversidad y Conservación
>
> Escuela Superior de Ciencias Experimentales y Tecnología
>
> Universidad Rey Juan Carlos
>
> c/ Tulipán, s/n.
>
> E-28933 Móstoles (Madrid)
>
> España
>
>
> --
> Julia
>
> [[alternative HTML version deleted]]
>
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo@r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>
[[alternative HTML version deleted]]
_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-geo
