"Exact" is a difficult standard. The krieging will depend, in part, on the size of the surface being used. too fine or too coarse and the results will be less helpful. And while the result of krieging is useful, it's precise utility will depend in part on the size of the cell being examined and the data being used. So it's good, precise, but ...it is a matter of smoothing data on a surface and that always is ... an estimate.

No, the ordinary spatio-temporal kriging predictor is not an exact interpolator. This is because the kriging model is based on the assumption that the sample data points are not perfectly representative of the underlying trend of the data. Thus, kriging will provide an estimate of the underlying trend, rather than an exact interpolation of the sample data points.

Peter Donkor Thank you for your answer. And in the case of spatial kriging, the kriging predictor is an exact interpolator. Am I right ?

Tom Koch Thank you for your answer. Yes it depends of the size of the surface estimated, I agree with you. But if the estimate surface was a point (the exact sample data location) (which makes no sense for a prediction), the estimate will be equal to the observed value ?

The estimate will be based on the observed value interpolated across the surface. So .. the more points the more precise the surface estimate will be.

I hope that helps. Most forget that the size of the area for each point can be set manually as you move across an area.