Geologists across the mining industry often face challenges when dealing with datasets that are severely positively skewed. In these situations many decisions need to be made, for instance to top cut or not to top cut? And if applied, what value should be used?
To help geologists in making these decisions, we have developed a tool called the ‘mean and variance top cut diagram’, located inside the Basic Statistics window. The diagram plots the mean against the Coefficient of Variation (COV) at various grade cut offs, allowing the geologist to assess which cut off will yield an acceptable COV and its impact on the mean grade.
To generate this diagram, firstly let’s start with a string file containing the composited sample values from within a domain; this will need to be loaded into the Basic Statistics Window. Database > Analysis > Basic Statistics Window.
Once the Basic Statistics Window has opened, go to File > Load data from string files.
Select your string file and the appropriate d-field and type the name of the element (e.g. Platinum). Additionally if desired, the number of histogram bins can be adjusted. Click Apply.
A histogram and cumulative frequency curve will then be displayed for the element(s) selected above.
Next, generate the mean and variance top cut diagram by going to Display > Mean and variance top cut diagram. The following diagram will then appear:
In the above diagram, the blue line represents the COV at various grade cut offs and should be measured against the axis on the right hand side of the diagram. The red line represents the mean at various grade cut offs and should be measured against the axis on the left hand side of the diagram.
Generally, a dataset with a COV greater than 1.2 demonstrates a mixed population which may benefit from top cutting and a dataset with a COV less than 1 generally does not. Remember the aim of top cutting is to reduce the COV while maintaining the characteristics and distribution of the original population (i.e the mean is maintained and the tail of the population is not unduly cut).
It should also be noted there are other alternative and complementary methods documented across the industry for helping geologists select and decide on the appropriate top cut value, such as percentiles, Sichel’s mean, log-probability plots etc. These other methods which are available in GEOVIA Surpac will be examined in upcoming posts.
