Integration of Dual Border Effects in Resource Estimation: A Cokriging Practice on a Copper Porphyry Deposit

Abstract

Hierarchical or cascade resource estimation is a very common practice when building a geological block model in metalliferous deposits. One option for this is to model the geological domains by indicator kriging and then to estimate (by kriging) the grade of interest within the built geodomains. There are three problems regarding this. The first is that sometimes the molded geological domains are spotty and fragmented and, thus, far from the geological interpretation. The second is that the resulting estimated grades highly suffer from a smoothing effect. The third is related to the border effect of the continuous variable across the boundary of geological domains. The latter means that the final block model of the grade shows a very abrupt transition when crossing the border of two adjacent geological domains. This characteristic of the border effect may not be always true, and it is plausible that some of the variables show smooth or soft boundaries. The case is even more complicated when there is a mixture of hard and soft boundaries. A solution is provided in this paper to employ a cokriging paradigm for jointly modeling grade and geological domains. The results of modeling the copper in an Iranian copper porphyry deposit through the proposed approach illustrates that the method is not only capable of handling the mixture of hard and soft boundaries, but it also produces models that are less influenced by the smoothing effect. These results are compared to an independent kriging, where each variable is modeled separately, irrespective of the influence of geological domains.

Revisited Bayesian Sequential Indicator Simulation: Using a Log-Linear Pooling Approach

“…The reason is that some of the weights obtained by solving the simple kriging system via Equation (3) receive negative or high values. These particular weights lead to producing some negative values or very high values for estimated indicators, which are more than 1 [24][25][26]. Then, in the fourth step, the deviations in probabilities for order relation should be rectified [26].…”

“…These particular weights lead to producing some negative values or very high values for estimated indicators, which are more than 1 [24][25][26]. Then, in the fourth step, the deviations in probabilities for order relation should be rectified [26]. In this respect, one can utilize the bounds correction, signifying that all negative estimated values are set to 0, and all high values (more than 1) are set to 1.…”