Using AI to Discover 3D Multi-element Correlations in Your Deposit
Original blog post: https://minervaintelligence.com/driver-primer-5-overlaps-analysis
In our last blog post, we demonstrated how to upload and manipulate your data in the DRIVER platform.
In this post, we will introduce the ‘Overlap Analysis’ tool and show you how to easily interrogate dozens of elements simultaneously.
Mineral deposits often have complex multi-element signatures, and while multi-variate tools like K-Means cluster analysis and principal component analysis are staples of the exploratory data analysis world, these methods are non-spatial by nature, so it can be challenging to view and interact with the results in 3D. DRIVER’s unsupervised machine learning functionality can be used to create 3D anomaly zones for each of the individual elements that are both high in relative concentration and spatially coherent. Following identification of the zones, the ‘Overlap Analysis’ tool can be used to quickly create 3D Boolean intersection queries for any combination of zones desired.
Multi-attribute overlaps in DRIVER are stored and organised into ‘profiles’, essentially user-created folders that are used to organise the thoughts and hypotheses of the user. To create multi-attribute overlaps, navigate to the ‘Overlaps’ tab and create a new overlap profile. Right-click on the desired profile and select ‘Create overlaps’ (Figure 2).
Multi-select from the created 3D zone objects, or any external data uploaded to the platform (3D domain volumes, external resource estimations or inverted data from a geophysicist). DRIVER will dynamically calculate every theoretically possible combination of the selected attributes in 3D space, along with their corresponding volume. The minimum overlap volume filter may be used to filter the volumes below the selected threshold. Overlap Combinations may be browsed using the drop-down menu, and either saved to the profile or added to the workspace for viewing.
A geologist working on a Porphyry Cu-Au system like Saddle North would naturally be interested in the overlap zones for Cu and Au; however, the functionality offered by DRIVER makes it easy to test how additional elements, such as As, Cd, Hg and Sb, which could potentially pose problems later in the mining or reclamation phases, are distributed in the deposit. For example, the relationship of As to Cu+Au mineralization, is easily tested using DRIVER. In this case, it is evident that As is not strongly enriched in the main mineralized ore zone.
This style of analysis is also powerful for identifying and mapping alteration haloes, exploration vectors and lithology/mineral proxies. At Saddle North, the Au+Cu mineralization is closely associated with potassic alteration. This association may be mapped, rapidly, by proxy, as enriched K, Fe, Au, and Cu. DRIVER’s machine-learning-based method makes it easy to rapidly map the 3D distribution of elemental anomalies in the dataset, and the overlap procedure allows for rapid calculation of where the overlaps are coincident (Figure 6).
In our next DRIVER Primer post we will show you how you can update your drilling data and watch as DRIVER dynamically adjusts all your models to the new inputs.
Figure 6: DRIVER workspace showing the intersection between calculated Fe, K, Au, and Cu for the Saddle North deposit
In the next DRIVER Primer post we will show you how you can update your drilling data and watch as DRIVER dynamically adjusts all your models to the new input data.