Three-dimensional (3D) geological models are needed for minerals exploration and to plan for mining below the surface. These 3D models show the predicted geometry and distribution of different rock types.
To create them, geologists traditionally relied on manual logging of drill samples and lab assay results. Working through this process to create a single model is time consuming, with the accuracy of the result dependent on the experience of the geologist.
With the development of a plethora of new measurement sensors delivering greater levels of data, CSIRO mathematical geologist, June Hill, says mining companies soon found "they had a whole lot of numbers instead of rock types".
What was needed was a solution that delivered more objective results and could speed up the 3D model building process.
Launching Data Mosaic – a new software tool
To help with this, CSIRO developed Data Mosaic – a software tool that automates the integration and interpretation of numerical drillcore analyses and downhole data from different geophysical and geochemical sources.
The software helps geologists assign rock types to the numbers, by domaining and classifying the different rocks along the length of a core.
And because it uses a consistent algorithm, rock types from within and between nearby cores can be matched, which assists in predicting what lies between drillholes and in generating 3D models of what's underground.
Data Mosaic is a web-based application (app). Using the app, geological models can be updated as new information becomes available.
Data Mosaic will be useful to exploration and mining geologists alike.
Interpreting data to help geologists domain and classify rock types
Across any distance (above the molecular scale) the composition of rock continually changes. Rock types vary from major changes – such as volcanic to sedimentary rock – to subtle variations in chemical composition. These changes are reflected in the measured characteristics or variables used by geologists, such as X-ray fluorescence or gamma wireline data.
Numerical data of one or more variables collected along a drillcore or downhole can be loaded into the app. Then, guided by a geologist as to the scale at which differences become significant, the software can detect boundaries between rock types and employ machine learning to classify the rock types that are important to exploration or mining.
The strength of the boundary is reflected in the scale of change and in the number of variables in which it is detected. Strong boundaries are more likely to reflect regional changes and weaker ones more localised differences. Data Mosaic allows a geologist to model rock types on a broad scale and then zero down to finer scale details in areas of interest.
The most obvious benefit of the new app, Dr Hill says, is that it significantly speeds up analysis. A broadscale sweep of a drillcore can be done in a matter of hours rather than days.
App to speed up geological logging with robust results
Data Mosaic also increases the repeatability and robustness of the results.
"When manually assigning labels to rocks, different geologists of different experience will call the same rock type different things. But in a digital system, such as Data Mosaic, everything has to be consistent."
That means, you can have six different geologists of varying experience out at exploration sites and treat their interpretations with a lot more confidence.
Barrick Gold Corporation's senior manager for resource geochemistry, Natalie Caciagli – who is trialling the new software – says, for that reason alone, Data Mosaic "could well be a game-changer in using downhole survey data".
"An additional use of the mosaic plot is that it shows up where the data is highly variable," CSIRO group leader, Mark Pearce, says.
"Understanding the variance, will help with sampling."
Data Mosaic is being made available through CSIRO via a subscription service. Companies will be able to buy access for a period of time, from a few weeks to ongoing.