Presenting the attribute kriging algorithm for automatic domaining and simultaneous estimation

Due to the development of sampling facilities in the mining industry, the spatial reporting of data sets is increasing, more variables are being noted, and broader areas are being covered. Therefore, it is crucial to subdivide the study areas into smaller domains to clarify the computation of the different behaviour of natural phenomena. In the estimation of mineral resources, the process consists of partitioning the mineralised area into several domains defined by the ore grade differences. Defined domains can be considered as a clustering problem. Even so, non-spatial techniques of clustering do not guarantee the spatial continuity of geostatistical data sets. Multivariate spatial clustering methods must therefore be applied, which often indicate the specifics of spatial continuity and heterogeneity. As a result, the Geostatistical Hierarchical Clustering algorithm is proposed. The validation of the non-spatial and spatial clustering techniques is evaluated by a synthetic data set in which the acquired results highlight the necessity of applying the algorithm. The mentioned algorithm is used as a proper tool for automatic domaining in geostatistical data set estimation. Its effect on improving the results derived from the kriging estimator is analysed on the synthetic data set. Consequently, the Attribute Kriging algorithm is introduced for estimating mineral resources.