A comparative evaluation of Gaussian Markov random model based simulation and sequential Gaussian simulation for orebody modelling

摘要

This paper presents two conditional simulation algorithms: sequential Gaussian simulation (SGSIM) and Gaussian Markov random (GMR) model based simulation. A comparative evaluation of these two algorithms was performed. The study results revealed that although the Gaussian Markov random model based algorithm was computationally faster; however, the sequential Gaussian algorithm performed fairly well particularly in variogram reproduction. The poor performance of the Gaussian Markov random model might be due to the apparent lack of fit for the Gaussian Markov random model to the full Gaussian model. Finally, the SGSIM algorithm has been investigated in the Nome placer gold deposit as it provided better performance in comparison with the other algorithm. The simulated realizations generated in the Nome deposit although reproduce the shapes of the histogram and variogram appropriately; however, the variance and sill values increased highly. This is due to the influences of some extreme values of the conditioning data.