Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images
IEEE Transactions on Pattern Analysis and Machine Intelligence
A clipped latent variable model for spatially correlated ordered categorical data
Computational Statistics & Data Analysis
Iterative algorithms for fitting a linear model of coregionalization
Computers & Geosciences
| Hi-index | 0.00 |
The plurigaussian model is currently used for simulating geological domains (facies) in petroleum reservoirs and mineral deposits, with the aim of assessing the uncertainty in the domain boundaries and of improving the geological controls in the characterization of quantitative attributes. This paper discusses the main aspects of the model and provides a set of computer programs to perform its inference and conditional simulation. Two types of conditioning information are allowed: hard data for which one has an exact knowledge of the actual domain at sample locations, and soft data consisting of inequality constraints on the local domain proportions (probabilities of occurrence) at control points chosen by the mining or reservoir geologist. An application to a Chilean porphyry copper deposit is finally presented, in which three Gaussian fields are used to simulate the spatial distribution of five mineralogical domains: gravels, leached capping, oxides, primary and secondary sulfides. The model is constructed so as to honor the topological contacts between mineralogical domains, their spatial continuity, the information logged at exploration drill holes, as well as the vertical proportion curves that indicate the mineralization profile with depth.