Alternative Methods of Interpolating Precision Farming Data.

Interpolation is a mathematical technique which constructs one or more new (unobserved) data points based on observed data points in a neighborhood. In the spatial statistical analysis of precision farming data, commonly used interpolation methods include nearest-neighbor analysis, inverse-distance weighting, inverse-distance-square weighting, and kriging. The interpolation methods differ in how observed values are weighted to obtain the unobserved with the weights being directly determined based on the spatial covariance function of the distance between the observed values.  In this presentation we will compare and contrast these methods in terms of their computing efficiency and prediction accuracy. The use of these interpolation methods is illustrated through two applications: (i) detection of ‘soft’ errors in yield monitor data and (ii) alignment of multi-year data from the same farm field but with different data density.