Spectral reflection curves from 58 mine tailings samples were grouped into 11 unique spectral categories using a cluster analysis technique. A good relationship was found between the spectral categories and Munsel colors (hue and value) but only 70% of all groups could be separated on the basis of a single chemical parameter (Fe). It appears that the overall chemistry has a profound influence on spectral reflection, with total Fe, Al, Si and C playing a dominant part. The parameter interactions are complex and this makes it difficult to predict chemical concentrations from spectral reflection values. Cluster analysis offers a means to predict several parameters and allows an examination of their interaction. However, the prediction accuracy of this technique needs to be improved.

Date of this Version