Abstract

A critical challenge in evaluating mineral resources is to model the uncertainty of geological boundaries and to classify the block model into domains of similar rock or ore types. The aim of this study is to deal with the problem of simulating geological boundaries in the Golgohar iron deposit based on two datasets extracted from exploration boreholes and blast holes. In the first step, contact analysis tools are used to identify the nature of geological boundaries. Next, a modified version of the direct sampling algorithm is proposed to simulate geological domains based on a training image and local probabilities derived from a vertical proportion matrix. The purpose of using the modified direct sampling is to control the local proportion of each geological domain in the distance-based multiple-point geostatistical approaches. A set of 100 conditional realizations of two ore types are generated. The reconciliation between the realizations produced with the proposed algorithm and actual data values is provided by checking the reproduction of single-, two- and multiple-point statistical parameters and the connectivity of geological boundaries. The proposed method proves to be a useful variant of the direct sampling algorithm for simulating the spatial layout of geological domains when there is a strong difference in the domain proportions between the hard data and the TI.