[R-sig-Geo] Kriging prediction variances on gstat

[Carlo Cavalieri]

Hi all,

I would like to know how the kriging variance is computed by gstat, because on my analysis I need to use a cell-specific covariate depended scale parameter rather than a unique one.
If the overall scale of the origins variance is fixed by the sill of the theoretical variogram model used to make the interpolation I can divide the estimated kriging variance by it and then multiply by the cell specific value, but I wanted to make sure that the sill value (partial sill + nugget) is used.

Thank you
Carlo
	[[alternative HTML version deleted]]