Building a 3D earth model conditioned by the spatial statistics of petrophysical data

Characteristics of the earth’s heterogeneity are inherent in the physical properties measured from the target rock mass. Hence, investigation of the spatial variability of measured rock properties, for example porosity, permeability, density, and metal concentration, are useful for understanding the spatial changes in characteristics of the rock mass. Given that most data in the earth sciences are not exhaustive, heterogeneity in a rock mass at locations with no available data can be constrained by information at sampled data locations. Geostatistical tools are used to model spatial variability of physical properties at such unsampled locations. This study presents the application of geostatistical techniques to build a 3D earth model constrained by existing large database information that includes density logs measured from core samples of 32 boreholes, lithology, mineral concentration, gravity, magnetic and electromagnetic data. The database constituted part of an integrated study to delineate the lateral distribution of a Zn-Pb-Ag sulfide deposit. Density values are often correlated to the chemical constituent of the rock. In a sulfide setting, despite the potential influence of geologic processes in altering rock densities, sulfide mineralized rock samples have high densities relative to host rocks. Heuristic density models can then be built to honour any existing hard data and can be further tailored to accommodate the spatial correlation with other existing interdependent (cokriging) rock properties such as metal concentration. The cokriging process is based on an established hierarchy between the property variables e.g. Geology (felsic/mafic) → Density (Primary variable) → Zn%(Secondary variable). The 3D earth model honours not only the 3D statistics of the various properties considered, but also existing hard data (logs) in existing boreholes. Such an approach for modeling the spatial distribution of density and metal concentration provides a better assessment for inferred resource estimates. We used this geostatistical approach to locally characterize the 3D distribution of the rock mass at the Nash Creek exploration project, New Brunswick.