OSNACA PROJECT
Carl runs the OSNACA Project at the University of Western Australia. It is an open source project where all types of different hypogene ore samples from around the world have been analysed for a standard suite of elements at Bureau Veritas. Consistent 65-element data for over a thousand samples have been used to create a model that quantifies multielement metal signatures for each sample. One product from the model is a simplified "Magmato-Hydrothermal Space" where the distribution of varying metal signatures can be mapped (figure opposite). All other existing descriptions of metal signature are qualitative; e.g., Au-As-Sb-Ag-Te orogenic gold mineralisation and cannot be mapped with respect to each other. Industry applications of the OSNACA transform, thus far, are twofold.
- For a sample provided by a client, it is possible to find out which samples OSNACA database plot closest to that unknown sample. Typically, a sample or samples, is taken from a new body of mineralisation. An appropriate element suite, but not necessarily the full OSNACA suite, is required to create an accurate model for comparison.
- It is possible to model consistently analysed data for thousands of samples from a single ore deposit, using the OSNACA transform. That allows domaining of the deposit into bodies of like metal signature. Conventional deposit-scale investigations have only ever been constructed on the basis of grade before the development of the OSNACA transform. An OSNACA classified deposit-scale dataset is useful for near-mine exploration and to define different ore types for geometallurgy. Examples are provided below
To download the publicly available data and for other resources related to the project, go to:
https://www.cet.edu.au/project/the-osnaca-project/ or come and see Carl to discuss how the OSNACA transform can be used to advance your project.
Golden mile Kalgoorlie
Average Profiles After K-Means Clustering
Northern Star kindly made data for a section through the Golden Mile public, but redacted Au. Without Au to define grade (and be part of the signature), a weighted sum of pathfinder elements was used as a proxy for grade. The OSNACA transform was then calculated for each mineralised sample.
High Grade: Highest Mo Population - all samples
The averaged profiles above are for five populations that were determined by K-means clustering, an algorithm that isolates groups of proximal (similar) samples in multi-dimensional space. Populations are named according to distinctive features on their profile and their "grade" according to their absoulte pathfinder
Base Metals: Cu-Zn-Cd-(Pb)
element abundance (see definition of "grade" below).
Each population has a reasonably consistent set of individual profiles, particularly when compared to other populations below.
Lowest Grade: Lowest Te Highest As-Sb-W
Ag
Low Grade: Te-(As-W)
"Grade"
In the absence of Au data, a proxy for grade has been calculated which is the sum of squares for each element score PRIOR to scaling. In the final output, OSNACA scores are scaled so that the coordinates for each sample are independent of grade.
High-grade and low-grade are named according to their position on this plot, high grade standing out clearly from the rest.
Na/Al versus K/Al Molar Ratio Plot
The green and blue populations can be predicted to have a sericite-paragonite-chlorite alteration assemblage, whereas the other groups will have a sericite-paragonite-albite assemblage. In that second group, the red and pink populations are more biased towards the sericite node.