Estimation of a Drift Coefficient

Let us rewrite the usual universal kriging dichotomy:

where m(x) = a_lf^l(x) is the drift

We wish to estimate the value of one of the drift components (say the one corresponding to the basic drift function number l₀) at the target point by kriging:

The unbiasedness condition implies that:

This leads to the following conditions on the weights:

The optimality condition leads to:

Finally the kriging system is derived: