Abstract: |
In mineral resource evaluation a careful analysis and assessments of the geology, assay data and structural data is performed. One critical question is where to position the exploration boreholes that render it possible to classify as much of the deposit as possible as a measured or indicated resource. Another important question is what method to use when analyzing the grade in the collected material. For the deposit we consider, a challenge of this type is to assess whether one should analyze the collected core samples with accurate and expensive XRF equipment or the less accurate and less expensive XMET equipment.
A dataset of 1,871 XMET and 103 XRF observations is available, along with relevant explanatory variables. At the 103 sites where XRF data is acquired, 103 XMET measurements are also available.
We first derive estimates of the regression and covariance parameters of a Gaussian random field model for the log XMET and log XRF data. Next, the model is used to predict the decisive grade parameter on block support. To improve the predictions, the mining company has planned to drill and collect 265 core samples along new boreholes. The associated reduction in prediction variance, with XRF or XMET data collection, is studied. Moreover, we compute the value of the XRF or XMET information using the statistical model, the expected development costs and revenues. The value of information is a useful diagnostic here, comparing the actual price of the XRF or XMET data with its added value. |
