Re: ordinary kriging

[Edzer Pebesma <edzer@pangea.Stanford.EDU>]

In addition to Konstantin's remark and as an elaboration
on Kor's remark I would like to add the following:

Ordinary kriging is supposed to reproduce measured values
only where the prediction location exactly matches the
observation location. Check this for your grid and data
locations. If your variogram has no nugget, than the
interpolated surface will go through the data points;
if your variogram has a nugget, the interpolated surface
will be smoother and be discontinuities at data points.

Best regards,
--
Edzer