Metaheuristic Kriging: A New Spatial Estimation Method

Abstract

Kriging is one of the most widely used spatial estimation method. In kriging estimation, weights assigned to the neighboring data are determined by minimizing the estimation error variance (EEV). Due to the minimization of the EEV the variability of the estimation result is lower than the original data. This paper presents the metaheuristic kriging (MK) as a new estimation method which has similar structure with kriging. But unlike kriging MK does not minimize the estimation error variance, instead converges to the EEV minimum which provides MK to increase the variability of the estimation. The MK uses the metaheuristic di erential evolution algorithm in minimization of the EEV which gives names the MK. As a case study, Ordinary kriging (OK) and MK are applied to the Jura data set to estimate the spatial distribution of the Nickel (Ni) content. Results of the estimations are compared. Results shows that metaheuristic kriging over performed to the ordinary kriging in terms of variability of the estimation. The MK can be used any place where kriging is applied due to the variability of the estimation is higher than OK. The parameters used in MK are case speci c so parameter tuning have to be made in the estimations to reach the desired outcomes. This study only exposes the univariate spatial estimation.