Resource geologists often find geostatistical algorithms that rely on a single stationary model of spatial continuity inadequate for modelling grades in structurally complex deposits. Current practical methods for dealing with this challenge can be grouped in either coordinate transformations or locally changing spatial continuity parameters. A common feature of both groups is the requirement of a prior model of local anisotropies for the spatial continuity of grades. This model would include local angles of spatial continuity, and for some methods, also the local anisotropy ranges of continuity and other parameters. The local angles model can be built from direct measurements of dips and orientations of the geological structures that control the spatial distribution of grades. However, these measurements are seldom as exhaustive as needed. Thus, the local angles model is often completed by the information provided by geological interpretations in the form of wireframes. The local anisotropy ranges model is more difficult to construct. So, the local anisotropy ratios are often deemed constant and equal to an anisotropy ratio derived from the stationary variogram model. This paper focuses on various practical aspects of the construction of a model of local anisotropy angles from geological wireframes. The topics presented include the methods for obtaining the local anisotropy angles, the motivation and methods to pre-process these angles, and the approaches available to construct a model of local anisotropy angles field for use in mineral resource estimation. Statistics based on the inner product of vectorare presented as simple tools for comparing alternative fields of local anisotropy angles. Kriging with local anisotropy angles is applied to a structurally‐controlled gold deposit in Ghana, and the impact of using pre‐processed angles and alternative angle fields is assessed. Additionally, grade estimation with local anisotropy angles is benchmarked against estimation with ordinary kriging with stationary variograms.
