Considered methods

• closest observation method

  $\hat{Z}(s_0) = Z(s_{closest})$

• inverse distance weighting method (IDW)

  $\hat{Z}(s_0) = \sum_{i = 1}^{n}(Z(s_i) \cdot w_i)$

  • with $w_i = d_i^{-\beta} / \sum_{i = 1}^{n}d_i^{-\beta}$
  • optionally replace $n$ by $k < n$: use only the $k$ nearest locations

• nearest neighbours method:

  • special case of IDW with $\beta = 0$ and $k < n$ so that $w_i = 1 / k$