Abstract

Although narrow deposits hold economic and environmental importance, the assessment of their mineral resources with kriging is often challenging due to the shortage of paired data for variogram analysis along the narrow dimension. This study addresses the problem of modeling alumina and silica grades in a layered bauxite deposit using three geostatistical approaches. The first one is the direct approach, where grades are cokriged inside the mineralized layer interpreted by geologists; the second one (indirect approach) substitutes the grades with the layer thickness and accumulations as service variables; the third approach trades the traditional random field representation for a deterministic representation to jointly predict the alumina and silica grades inside the mineralized layer, based on an innovative technique called transitive cokriging. A comparative analysis of the results highlights the strengths and weaknesses of each approach, in terms of data preparation, geological modeling, and mineral resources modeling. The study concludes that the transitive approach is a promising alternative for mapping ore grades in deposits with narrow dimensions, due to a more congenial structural analysis and a reduced smoothing effect in comparison with the direct approach, while it avoids scaling down the representation of the deposit to two dimensions as in the indirect approach.