[R-sig-Geo] Universal Block Kriging covariate definition for krige in gstat

[Antonio Manuel Moreno Ródenas]

Thanks a lot Edzer,

I'm not sure that would work.
In that way I would transfer to the kriging function the averaged value of
the covariate in the block. I'm not sure that would make the kriging behave
correctly.

All the points calculated with the prediction inside the block (and later
averaged to give the block kriging prediction) will have as "drift" the
average of the covariate in the block. Instead of getting the correct
spatial variability inside the block (given by the covariate). At first
sight it doesn't seems correct to me. Am I wrong?

kind regards



On 13 January 2016 at 14:43, Edzer Pebesma <edzer.pebesma at uni-muenster.de>
wrote:

>
>
> On 13/01/16 14:16, Antonio Manuel Moreno Ródenas wrote:
> > Hello, I would like to rise a question on the use of predict {gstat},
> >
> > I'm trying to perform the estimation of a spatially distributed variable
> at
> > the support scale of a particular area (Block kriging). I have access to
> an
> > additional variable, it is known that the variable of interest is
> > correlated to the new variable. So I would be interested on updating my
> > estimation by the use of this new information. This could be done by the
> > use of a kriging with external drift (KED), but with a block support
> > (Universal Block kriging). Theoretically this is included in the gstat
> > library as mentioned in the documentation.
> >
> > The issue comes when I try to perform the prediction:
> >
> > blockprediction <-
> predict(gstat(formula=Variabletopredict~additionalVariable,
> > data=Observed, model=vgm), newdata = shapefile)
> >
> > The newdata argument should contain the prediction location. In a normal
> > KED we would include a dataframe with a grid (coordinates in which to
> > predict) and the values of the covariate (additionalVariable). As I'm
> > trying to use a universal block kriging, I understood the newdata should
> be
> > the region in which I'm interested to know the prediction, hence a
> polygon.
> > How could I include in newdata the values of the covariate if its
> > resolution is finer than my block?
> >
> > As far as I know, what block kriging does is to predict point values
> inside
> > the region (which I could specified with the argument sps.args
> > discretization), and later average them. But I don't know how to attach
> the
> > covariate values to the block of interest (shapefile).
>
> maybe by
>
> shapefile = aggregate(additionalVariable, shapefile, mean)
>
> >
> > Thanks in advance,
> > I hope I could explain it properly, but I will give more details if
> > necessary.
> > Kind regards,
> > Antonio
> >
> >       [[alternative HTML version deleted]]
> >
> > _______________________________________________
> > R-sig-Geo mailing list
> > R-sig-Geo at r-project.org
> > https://stat.ethz.ch/mailman/listinfo/r-sig-geo
> >
>
> --
> Edzer Pebesma
> Institute for Geoinformatics  (ifgi),  University of Münster
> Heisenbergstraße 2, 48149 Münster, Germany; +49 251 83 33081
> Journal of Statistical Software:   http://www.jstatsoft.org/
> Computers & Geosciences:   http://elsevier.com/locate/cageo/
> Spatial Statistics Society http://www.spatialstatistics.info
>
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

	[[alternative HTML version deleted]]