[R-sig-Geo] Clarification on cross validation fitted variogram model

Edzer Pebesma edzer.pebesma at uni-muenster.de
Sat Jun 7 19:28:58 CEST 2014

You don't give us enough invormation to give you a generally useful answer.

If fit.variogram() gives you a reasonably fitted variogram model, I see
no point in NOT using exp.mod in krige.cv(). On the other hand, we don't
know how and why the values passed to vgm() were chosen, and someone may
have a reason to argue they are better than the ones fitted by

On 06/06/2014 07:24 PM, Moshood Agba Bakare wrote:
> Hi all,
> Please I need clarification about the cross validation of fitted
> variogram model. I have come across a situation where some use *exp.var
> *in cross validation others use *exp.mod*. As for me, I think *exp.mod
> *is the right to use because the theoretical function is already fitted
> to empirical variogram. Moreso, *exp.mod *has the parameters in terms of
> nugget, sill, and range for the fitted function. Whereas *exp.var *has
> initial starting value of the parameter.
> from the last script, should model equals to *exp.var *or *exp.mod*?
> ## Fit empirical semivariogram using gstat
> empvar<-variogram(yield~1,canocrop.sp,cutoff = 300,width = 5, cressie =
> ## Exponential variogram model
> exp.var<-vgm(psill = psill, model = "Exp", range = range, nugget = nugget)
> exp.mod<-fit.variogram(empvar,model = exp.var)
> #### Cross validation of the variogram models by ordinary kriging
> exp.kcv <- krige.cv <http://krige.cv>(yield ~ 1, canocrop.sp, *model =
> ???*,nmax = 20, nfold=  5)
> Thank you.

Edzer Pebesma
Institute for Geoinformatics (ifgi), University of Münster
Heisenbergstraße 2, 48149 Münster, Germany. Phone: +49 251
83 33081 http://ifgi.uni-muenster.de GPG key ID 0xAC227795

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: OpenPGP digital signature
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20140607/a9fbdbb5/attachment.bin>

More information about the R-sig-Geo mailing list