Data mining for gold

10 Oct 2013

Author: Simon Webster

Photography:

Video:

'Data mining’ methods used to analyse the stockmarket are being applied to huge sets of geophysical data in a bid to better understand the geological evolution of Australia and find mineral deposits in the continent’s vast interior.
Tags 
  1. Category
  2. Class Styles - Category
  3. technology-style
  4. Technology
Creative CommonsWe’d love you to share this content

Australia has lots of data on its geophysical make-up. Thanks to recent breakthroughs in computing, it now also has the means to analyse that data. Combining ‘data mining’ approaches that would ordinarily be used to predict stockmarket movements with recently developed ‘virtual globe’ software, scientists hope to come to a better understanding of how the continent looked up to 1.5 billion years ago, and, in the process, strike gold.

“We are using data mining to improve our reconstructions of the ancient world,” says the project leader, Dietmar Müller, Professor of Geophysics at the University of Sydney. In addition, he hopes the project will reveal ore deposits – such as gold, copper, silver and rare earth elements used in electronic devices – that are currently well hidden.

“Even though there is a set of complex reasons why the mining boom has declined, one of those reasons is that the rate of discovery of new major mineral deposits has declined strongly over the past few decades,” Müller says.

“All the easy-to-find deposits have been found. Eighty per cent of the Australian landscape is covered by a veneer of ‘regolith’, weathered bedrock and sediments, often hundreds of metres and sometimes kilometres thick, obscuring mineral deposits from traditional search methods.”

This regolith layer means mining companies don’t know where to look next. But tectonic plates – the enormous, moving plates that make up the outer shell of the Earth – may hold the answer. About 1.5 billion years ago Australia was made up of at least four smaller continents, separated by plate boundaries. Today, those ancient plate boundaries are in Australia’s interior, and are where minerals are most likely to be found.

“Plate boundaries are the places where all the action happens, geologically speaking,” Müller says. “Plates grind against each other, they get deformed, and sometimes hot fluids comes up through fractures, injecting heat and metalliferous liquids into the system. Plate boundaries are largely the places where ore deposits form.”

A vast amount of geophysical data is available in Australia, thanks to state geological surveys, which are passed on to the national geoscience agency, Geoscience Australia, and made publicly available.

Virtual globe GPlates software, created by Professor Müller’s EarthByte Group at the University of Sydney’s School of Geosciences, will be extended with data mining tools from finance and biosciences. The new combined software tools will be applied to geodata to explore the interactions of tectonic plates through cycles of continental dispersal and collisions. This may reveal ‘tectonic niche environments’ potentially containing significant ore deposits.

“Given the richness of the mineral deposits that have been found in the 20 per cent of the Australian landscape that is not covered by a thick layer of weathered bedrock, we are pretty sure that similar riches are to be found under the remaining 80 per cent,” says Müller.

The conceptual underpinnings of this project were created by the UNCOVER Group of the Australian Academy of Science, which has developed a strategic framework for better understanding Australia’s metallogenic evolution.

Professor Müller’s project is one of three subprojects in the three-year Big Data Knowledge Discovery Project, which is funded by a $4 million grant from the Science and Industry Endowment Fund (courtesy of a gift from Australia’s national science body, the CSIRO), and $8 million from the project’s research collaborators.

The collaborators are the University of Sydney, Macquarie University, Sirca (the Sydney-based non-for-profit developer of the stockmarket data mining software, which is providing its software development expertise), and the project leader, National ICT Australia Ltd (NICTA), Australia’s Information and Communications Research Centre of Excellence.

NICTA is providing expertise in analytics and machine learning, a term closely related to data mining, which relates to computers learning from data and reacting to it.

As well as the geoscience project led by Professor Müller, the Big Data Knowledge Discovery Project also comprises:

  • A study of the physics and mathematics of complex laser systems that may lead to an improvement in the security of optic fibre communications, led by Professor Deb Kane at Macquarie University.
  • A study of how biodiversity arises from the traits of forest plants, which may lead to insights into how climate change will affect ecosystems, led by Professor Mark Westoby at Macquarie University.

“They are all examples of natural science exploration that rely on large-scale data,” says Stephen Hardy, Technology Director for Computational Analytics at NICTA, and leader of the NICTA team involved in the Big Data Knowledge Discovery Project.

The project came about thanks to new computer paradigms that allow the analysis of data sets that were previously too large to tackle, Hardy says.

“This computer paradigm, changing the way big problems are tackled, has allowed new problems to be investigated.”

Each of the three subprojects in the Big Data Knowledge Discovery Project has its own goals. However, the broader aim is to discover data analysis and machine learning techniques that can be used across other sciences.

Paul McCarthy, Director, Strategy and Innovation at Sirca, says the fact that computers can now make sense of the huge amounts of data they can store may have profound consequences.

“This project is working at the heart of two defining technologies of our time that are now reaching a point of maturity: big data and machine learning,” McCarthy says. “Together they hold enormous potential to transform most areas of human endeavour.”