The illustration above, originally presented at the 2014 Victorian Gold Forum in Melbourne, Australia, highlights some typical mineralisation distribution scenarios and conceptual “high geological nugget” issues, together with ways to resolve them and arrive at a better estimate of the global mean grade.
The top left two images show the differences the scale can make. The top right two images show how spatial as well as grade variability make a difference. The bottom left two images show what can happen when there is insufficient drilling. The bottom right two images demonstrate how sparse drilling with top cuts or dense drilling without top cuts might produce a better average.
In high nugget situations, reasonably accurate global grade-tonnage curve estimation at the selective mining unit scale is usually achievable by using low sample number searches to increase variability while maintaining the correct average grade through ordinary kriging.
Globally accurate and locally precise local block estimation, however, is almost impossible at the resource definition and feasibility stages in high nugget situations. Only when multiple holes are available within each block can estimates be locally accurate.
Depending on the situation, locally precise block estimation may not be necessary if reconciliations based on large volumes and large time scales are acceptable.
If you have low confidence in the variogram model you are using because of difficulties interpreting the experimental variogram, then estimates with several variogram models with different nuggets and ranges can be run to examine the sensitivity of your estimate to the variogram model itself.
In summary, the nugget effect isn’t that challenging; we just need to understand the purpose of the model we are creating and how to express the expected local selection variability and or global accuracy correctly for that stage of the project life cycle.
