Simulation with Data of Different Support

The estimates performed by kriging are built, so that each estimate considered individually (without considering the neighboring estimates) is “best.” Best because the estimate minimizes the variance of the estimation error. However, when several estimated locations are considered together, such as an estimated grade model, the estimates do not represent the true spatial variability of the modeled variable. The map of the estimates has less variability than the modeled variable. The variance of the estimates is much lower than the variance of the data. This effect is called smoothing.