Abstract Several different methods have been used for the representation of orebody shapes in computer systems including serial-slice, gridded seam, regular block, mathematical functions, solid geometry, and boundary representation models. The accuracy and precison of such models are affected by factors such as errors associated with the original data used, natural variations, errors in data capture and processing, etc. In order to asses the relative importance of such factors it is necessary to quantify differences in accuracy and precision between different model structures. To meet this objective a new technique which relies on statistical study of a series of sample lines, intersecting two or more different orebody models, is proposed. Application of this technique on orebody models at the Neves-Corvo copper-tin deposits in southern Portugal suggested that in certain circumstances regular block models can give very poor estimates of the total volume of materials of different geological types. In addition it was shown quantitatively that shape interpretations close to the plane of drill-sections are better, in terms of accuracy and precision, than interpretations for zones falling between the drill-section planes.
Read full abstract