Geological uncertainty is a recognition that our understanding of geology is based on observations that may deviate from reality (Bardossy and Fodor, 2001). The aim of this study is to identify how geostatistics can assist in the modelling of geological uncertainty. Within this study, four different types of geostatistical techniques are considered: Indicator Kriging (IK), Sequential Indicator Simulation (SIS), PluriGaussian Simulation (PGS), and Multiple-Point Simulation (MPS). These techniques are used to analyse the influence of deposit geology on the representation of the estimated resources contained within an iron ore deposit in the Pilbara (Western Australia). Based on pre-existing classification, the deposit is divided into four geological domains: detritals and miscellaneous, Banded Iron Formation/waste, hydrated domain, and mineralised domain. During calculations relating to a domain under consideration, this domain is assigned an indicator value of unity while the other domains receive a value of zero.
Four different comparisons are developed:
1. PGS output is compared for three different lithotype rules.
2. MPS output is compared for variations in grid size.
3. Direct comparison of outputs from all techniques (following post-processing).
4. Output from all techniques is compared to an external block model of the deposit.
Overall, PGS and MPS give the most consideration to the geology of the deposit but, compared to SIS and IK, they require additional knowledge of the deposit characteristics. Post-processing suggests that the complex techniques (PGS and MPS) are generally more precise than simpler techniques (SIS and IK), with more nodes of each simulation reoccurring as the dominant domain across all simulations. If a comparison to an external block model is considered for accuracy, SIS relates most closely to the block model, and generates fewer uncertainties during pre-simulation processing and simulation.
This study supports the notion that geostatistical simulation can assist in the modelling of geological uncertainty.