Comparing the Estimates Obtained from Ordinary and Robust Kriging

Ordinary kriiing (OK) is an interpolation algorithm mostly used in earth sciences applications where it provides satisfactory results. Robust kriging (RK) is an alternative algorithm which at- tempts to minimise the problems caused by the existence of outlier values within a given data set. Robust kriging essentially consists of an editing step prior to running the traditional OK method. This study compares the interpolation results obtained by OK and RK. Two data sets were used. The first data set consists of 140 samples mimicking a drilling survey named cluster.dat. An additional feature of this sample data is that its parent population, the reference data set, is available, thus permitting direct quantitative assessment of the effectiveness of a given estimation technique. The second data set consists of a grid sampled at 1x1 m. The estimates obtained by OK and RK using cluster.dat were compared to the reference data set. RK provided better results in terms of quality of the estimates than OK