There’s gold in this – the development of new CET grid analysis software

By Graham Chandler

With 26 operating mines and projects across five continents, Barrick Gold is a minerals industry leader. Most of Barrick’s African gold mines sit within a 100-kilometre radius around Lake Victoria in Tanzania, East Africa. It’s a prolific gold region where Barrick continues to search its landholdings for new reserves.
And that’s where the new CET grid analysis software has worked so well and become an indispensable tool for the company.
“The most relevant application to date has been on airborne gravity data in the Lake Victoria Goldfields to assist in mapping dislocations and gradient changes commonly associated with greenstone gold deposits,” said Matthew Hope, Barrick’s project geophysicist for Africa/Eurasia.
It is the identification and mapping of these features that makes the new CET grid analysis tool so useful to explorers. The modules are essentially automatic interpretation tools that provide a first pass lineament detection on gridded/image data.
“Exploration teams today have a lot of data to sift through and don’t have a long time to do it,” said Louis Racic, Geosoft’s product management director. “At one time people would spend weeks or months completing interpretations - we don’t have that luxury anymore.”
The clever and sophisticated algorithms behind these tools were created by the Geophysics and Image Analysis Group of The University of Western Australia (UWA)’s Centre for Exploration Targeting (CET), part of the University’s School of Earth and Environment. The group is supported through a partnership between UWA, Curtin University of Technology and the mineral exploration industry. Since 2006, the team has been focusing on developing new techniques to enhance and automatically detect features of interest from geoscientific datasets.
While the UWA team are clearly the mathematical theorists behind the initiative, creation of the new modules was a joint effort with partners Barrick and Geosoft.
“We saw an opportunity to support an educational initiative and it was a good fit with our mission of empowering explorers,” said Racic.
“The University of Western Australia is well regarded in the exploration community, especially in Australia. And we can benefit from them as they can benefit from us.”
The other partner, Barrick Gold, agrees - they were in fact instrumental in getting the project underway.
“The significance of the grid analysis tools was first recognised through our association with the External Advisory Group for CET,” said Barry Bourne, chief geophysicist-global exploration at Barrick Gold.
The company provided already-analysed datasets for testing of the algorithms.
“Initial processing of potential field data with the CET grid analysis tools showed a good correlation with known structure and geology,” said Bourne. Bourne said it wasn’t just a single interpretative situation that identified the need for the new grid analysis tool.
“It was more a case of having access to new layers of information to assist with the interpretation of potential field data,” he said.
“We offered critical feedback on the outputs, [which] varied from comments on the association of outputs with known geology/structure to the provision of GIS-compatible output for integration with other exploration data.”
In the CET group at UWA, Associate Professor Eun-Jung Holden teamed up with Professor Mike Dentith and Dr Peter Kovesi to develop the methodology used for the software. In 2009 when the project hit its stride, Holden coordinated and led the team, assisted by research associate Shih Ching Fu, to develop the software.
Holden explained the outcome and the three basics of what the mathematical processes do:
“The algorithms provide methods to enhance local discontinuities within data by analysing local textures; to locate laterally continuous regions of discontinuity by finding texturally complex line features or finding data edges; and to vectorise the axes of resulting discontinuity regions,” she said.
Holden said the base algorithms are well-established methods in the computer vision community, some of which had been developed by team member Dr Peter Kovesi.
“These algorithms are combined and adapted for geophysical processing in CET grid analysis,” she said.
The software provides generic tools that are broadly applicable. Barrick’s Bourne believes the new software will find use in most exploration situations with potential field or other image analysis requirements.
“It is not a standalone tool but a supplement to traditional image outputs,” he said.
“The efficiency of the CET grid analysis tools is one of its major strengths. From a user perspective the code is relatively easy to understand and quick to run. Several iterations can be achieved in short periods of time on most potential field datasets.”
UWA’s Holden says the new processes have only become possible more recently.
“A main challenge in developing geophysical image processing tools is in dealing with the size and scale of the data while preserving a reasonable execution time,” she said.
“Often geophysical datasets are larger than datasets used in other fields, and clients do not have access to high powered hardware.”
Clearly the specific algorithms contained in this new tool all provide significant gains for the interpretation process. But the research doesn’t stop there. “Image processing research is a rapidly moving field, and CET intends to transfer and extend these advances into the field of geophysical image processing,” said Holden.

How the new algorithms work their magic
The CET Grid Analysis software provides tools for grid texture analysis, lineament detection, edge detection, and thresholding to coax out trends from geophysics datasets and facilitates the following:
1. Texture analysis-based image enhancement: by highlighting local intensity variations, this method enhances regions of discontinuity within aeromagnetic/gravity datasets. Regions of magnetic/gravity discontinuity correspond with, and can reveal, lithological boundaries, faults and dykes critical to understanding an area’s geology.
2. Discontinuity structure detection: takes the texture analysis output and finds the skeletal structure of the regions of the magnetic/gravity discontinuity. The output is a set of a binary skeletal line segments that belong to each of the discontinuity regions, clearly showing the changes of orientations and offsets within the structures. This process emulates the traditional manual drawing of interpretive lines along the discontinuity region.
The algorithms will be useful in exploring for most kinds of mineral deposits, in particular gold and base metal mineralisation.