Plurigaussian modeling of geological domains based on the truncation of non-stationary Gaussian random fields

Abstract

The plurigaussian model is used in miningengineering, oil reservoir characterization, hydrology andenvironmental sciences to simulate the layout of geologicaldomains in the subsurface, while reproducing their spatialcontinuity and dependence relationships. However, thismodel is well-established only in the stationary case, whenthe spatial distribution of the domains is homogeneous inspace, and suffers from theoretical and practical impedi-ments in the non-stationary case. To overcome these lim-itations, this paper proposes extending the model to thetruncation of intrinsic random fields of orderkwithGaussian generalized increments, which allows reproduc-ing spatial trends in the distribution of the geologicaldomains. Methodological tools and algorithms are pre-sented to infer the model parameters and to constructrealizations of the geological domains conditioned toexisting data. The proposal is illustrated with the simula-tion of rock type domains in an ore deposit in order todemonstrate its applicability. Despite the limited number ofconditioning data, the results show a remarkable agreementbetween the simulated domains and the lithological modelinterpreted by geologists, while the conventional stationaryplurigaussian model turns out to be unsuccessful.