Deposit Zoning based on the spatial distribution of ore grade using self-organizing map clustering algorithm (Case study: Choghart iron deposit)

Deposit zoning is one of the significant issues in the field of modelling, evaluation and exploitation planning in the mining sector. In ore modelling, the site is divided into zones based on the physical features affecting mineralization or the spatial distribution of ore grade. In this research, self-organizing map (SOM) has been used for three-dimensional zoning of the mineral deposits. The cluster validity indices has been applied to define the optimal number of zones. The proposed algorithm verified using the data of Choghart iron deposit. The Clustering validation indices were executed based on assay data (iron and phosphorus) of exploratory boreholes and optimal number was resulted in two zones. The SOM clustering algorithm was utilized to determine the confine of each zone and assigning samples within the two zones. As a result, the output of SOM and K-means algorithms illustrated that about 90% of the data was similarly assigned to the same cluster. In SOM clustering, the discrimination surface of the two zones has the northeast-southwest orientation with the southeast slope, while the K-means algorithm determined the surface with east-west orientation and south slope. The resulting surface of SOM is in accordance with the dimensional and directional properties of the structural features (especially fault system) in Choghart iron deposit.