Structures like narrow veins, lenses, seams and dykes generally cannot be modelled directly from the drill data. For these more complex deposits manual selection of data provides a highly flexible and robust way to produce models.

GEOREKA combines implicit techniques with manual editing to achieve the best results. In this example a simplified artificial data set is used to demonstrate different approaches for modelling narrow veins.
Below we also show one of our YouTube videos which shows how a more realistic data set is modelled.

First method

For the first example the hanging wall and footwall points of a near-vertical dyke are extracted. If you have a clean data set you can just extract them straight from the data by specifying the lithology code. However, in real-life data sets are never as clean and the geology not that simple. In those circumstances, manually select the segments from the data that will be used to extract the hanging wall and footwall points.

Next, create a boundary surface from the points to create the hanging wall and footwall. Due to the nature of the implicit modelling approach the surfaces will extent to the model boundary and will generally not pinch out where required. That is where manual correction comes in.
To let the domain of choice (be it vein, dyke or something else)  pinch out where required, one or more pinch lines can be drawn. In this example only one is used to clearly demonstrate the effect. The pinch line is drawn on the dynamic section tool, but could be any surface.
The newly created pinch line can be merged with the hanging wall and footwall points respectively before creating the final boundary surfaces. The end results are two surfaces that intersect at the pinch line. The method described here for a dyke can equally be applied to narrow veins.

With the boundary surfaces corrected the geological domain can be created.
The geological domain can subsequently be used to evaluate interpolation results to obtain a rough estimate of the average grade and volume estimate within the domain.

Second option

If a complete implicit approach is required without manual editing the approach is quite different. Instead of just extracting the hanging wall and footwall, the lithology code of interest is converted to a value, typically an indicator, specifying the inside and outside of the domain to be modeled. For this method the lithology to be modeled will be converted to points with positive values (inside) and other lithologies will be set to negative values (outside). The contact points will be set to zero (boundary). Due to the nature of the structures being quite thin compared to the scale of the project, it often needed to re-sample the lihology data to have sufficient samples inside.

With the lithology data converted to values, it can be modeled using interpolation techniques. Especially the RBF (Radial Basis Function) interpolation is very suited for this. However, for correct interpolation an Anisotropy will be needed.
The resulting interpolation will have positive blocks inside the domain and negative values outside of it. To obtain the shell that represents the domain we just extract it from the interpolation.

Notice how it pinches off as well, but this time there is limited control over where the structure will pinch off.