Hierarchical Indicator Simulation

Indicator simulation allows specification of different continuity of high and low values through the use of grade thresholds. The sequential implementation of indicator simulation has been widely used, but the problem of order relation deviations remains a significant drawback. A new procedure for "Hierarchical Indicator Simulation" is introduced that avoids order relations and offers greater flexibility than the conventional sequential approach. The new implementation simulates all locations at the largest threshold and then considers decreasing thresholds in turn. All nodes with grades that are above the previous threshold are excluded from the domain of the current threshold and therefore, they are not used as conditioning data. The probability of being above or below the threshold is calculated using simple indicator kriging using all conditioning data in the domain, thus the covariance structure is preserved. The theoretical framework of this new approach is fully developed. Practical implementation details are discussed. Application to a copper deposit shows the advantages and drawbacks of the method. Finally, the possibility of incorporating multiple-point statistics is discussed.