What twelve holes are
Twelve holes of 250 m are three kilometres of core through a volume of rock that may be a cubic kilometre or more. Each hole is a line. Between the lines, everything is inferred. A 3D model at this stage is therefore mostly interpretation, constrained at a few places by measurement, and its usefulness depends on keeping those two things distinguishable.
What it can show
- The data in its real geometry. Holes drilled at different angles from different pads are hard to compare on paper sections. In 3D, intervals that looked unrelated may line up along a plane, and intervals that looked continuous may turn out to be on different structures.
- Whether an interpretation is possible. A proposed fault, contact or mineralised zone can be drawn as a surface and checked against every hole. Interpretations that require the geology to do impossible things between holes are found quickly.
- More than one interpretation. Sparse data rarely allows only one reading. Two or three alternatives, built explicitly, show which holes would discriminate between them.
- Other datasets in the same space. Terrain, surface mapping, soil geochemistry and geophysical sections or inversions can be placed with the drilling. A chargeability anomaly that the holes skirted, or a mapped structure that none of them tested, becomes obvious.
- Where the data is not. The untested volume is as important as the tested one. A model can show how far each point is from the nearest sample.
- The next holes. Planned collars, azimuths and dips can be checked against terrain, existing holes and the surfaces they are meant to cross.
What it cannot show
- Grade between holes. Estimating grade away from samples requires knowing how quickly grade varies with distance. That is measured from many pairs of samples at many spacings, and twelve holes do not provide them in the directions that matter.
- A block model worth the name. Software will produce one from any input. Filled from sparse data, it reflects the interpolation settings more than the deposit.
- A tonnage. A volume can be drawn around intersections and multiplied by a density, but the result depends almost wholly on where the interpreter chose to stop. At most this supports a conceptual range, as explained in exploration target versus mineral resource.
- Continuity. Two intersections 200 m apart may be one body or two. The model can show both cases; it cannot choose.
Explicit and implicit modelling
There are two ways to build geological surfaces. In explicit modelling, a geologist digitises outlines on sections and joins them. In implicit modelling, a mathematical function is fitted through the contact points and orientation measurements, and surfaces are extracted from it.
Implicit modelling is fast and repeatable, and it updates when a hole is added. With sparse data it has a known weakness: it always produces a smooth, complete, convincing surface, whether the data constrain it or not. With twelve holes, the geologist's structural measurements and judgement must steer the result, and the model should be read as one of several that fit.
How to judge an early-stage model
- Can you see which parts are measured and which are interpreted?
- Does it honour every hole, including the ones that do not fit the favoured story?
- Are alternative interpretations shown where the data allows them?
- Is the detail in proportion to the data, or does it look like a mine model?
- Can the people who need to see it open it?
Where this fits
Gossan builds models of this kind as a fit-for-purpose 3D model: to the detail the data supports and no further, opened in a browser. As holes are added, the same model is rebuilt through updates on demand.