Block modelling is a popular way to analyse the numeric data that is collected during drilling programs. In DRIVER we create and use block models to view the estimated concentrations (or grade) obtained using kriging or inverse distance interpolation methods. But creating accurate block models (i.e., turning the sparse numeric data into an interpolated grid) is often a time-consuming and technically involved geostatistical process. For this reason, multi-element drilling datasets will rarely have block models made for all the assays available.
In our last blog post, we covered how DRIVER’s automatic anisotropy algorithm can be used to rapidly assess the geostatistical properties of a dataset and determine the optimal anisotropic configuration required for modelling.
In this post we are going to show how those anisotropies are used to inform accurate 3D block model interpolations, processed quickly using DRIVER’s parallel cloud servers.
The example drilling dataset is from a Sn deposit in Morocco. Sn mineralisation is hosted in SW-NE-trending, NW dipping shear zones. Correspondingly, the automatically determined anisotropy ellipsoid is a SW-NE-trending, NW dipping ellipsoid.
To initiate an interpolation, we open the interpolation functions, selecting either ‘inverse distance’ or ‘ordinary kriging.’ We will use ordinary kriging for this example. We can multi-select the anisotropy objects and set off a block model evaluation using a pre-defined 10x10x10 m block grid. DRIVER will use the anisotropy object parameters to perform an anisotropic interpolation; in other words, the contributing samples and their relative weightings will be adjusted to account for the directional bias in the distribution of each element the dataset.
For a 10m block grid, these models took about 5 minutes to process, and they did so in parallel on the cloud. Each model can be viewed in the workspace as either blocks, UVW cross-section planes, or as a mesh wrapped around the visible blocks. The block model for Sn shows a clear directional weighting applied to the distribution of grade, trending SW-NE and dipping moderately to the NW.
Ti (pct) is an element that rarely receives detail geostatistical attention in this sort of deposit, however the results (obtained easily with DRIVER), reveal interesting lithological deposit characteristics. Ti is not strongly enriched in the main Sn ore zone, but it is clearly elevated in a suite of sub-parallel dipping bands in the northern part of the deposit. These bands are mafic dikes that crosscut (and probably dilute) the mineralised Sn ore. The dikes can be visualised in 3D using DRIVER’s instant wireframe feature. By adjusting the display range for Ti to 0.6 to 1.4 pct, and then changing the display mode to ‘mesh’, DRIVER has produced a Ti grade shell that adequately delineates the 3D volume and distribution of the mafic intrusive lithologies.
DRIVER gives you the power to create models of every attribute in your dataset, within minutes. In the next blog post, we will talk about how DRIVER uses machine learning clustering techniques to automatically identify and categorize 3D anomaly wireframes.