edzer.pebesma at uni-muenster.de
Wed Jun 4 20:01:23 CEST 2014
On 06/04/2014 07:15 PM, Micha Silver wrote:
> I could use some guidance finishing a co-kriging script. I have rain
> data, and elevation for each rain gauge. I have successfully make an
> ordinary kriging of the rainfall. Now I'd like to try to use the
> elevation as an auxiliary variable to improve the prediction. I have
> made the model variogram with two variables, and the fitted variogram
> using fit.lmc:
> vg.fit <- fit.lmc(vg, g, vgm(700, "Exp", 50000))
> #gstat object 'g' prepared already with the 2 variables
> plot(vg, vg.fit) # (looks OK)
> then I run:
> # grd prepared in advance to match the range of the data
> precip.cokrige <- predict.gstat(vg.fit, newdata=grd)
> I expected to see:
> "Linear Model of Coregionalization.Good"
> [using ordinary cokriging]
> but instead the command returned:
> "Intrinsic Correlation found. Good.
> [using ordinary cokriging]"
> The co-kriging finishes successfully, and the results "look good." But
> what am I getting? What is the difference between Intrinsic Correlation
> and Linear Model of Coregionalization?
Please look up the literature -- IC is a special case of LMC, where each
variogram is a multiple of a basic model. For LMC this is not the case
for the full variogram, but for each variogram component (sub-model:
Nugget, Spherical). If you'd add a nugget component to the model passed
to fit.lmc, you'd get an LMC.
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
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...
Size: 490 bytes
Desc: OpenPGP digital signature
More information about the R-sig-Geo