Fixing blinds to a wall—it’s a commonplace problem, but there are many approaches depending on the environment and technology available.
If the wall is brick, you’ll need a hammer drill and rawl plugs. If wood, an ordinary drill and wood screws will do. Plasterboard calls for spring-loaded toggles. Metal or plastic, perhaps a glue, but adhesive technology changes so quickly you would need to spend time to find out about the latest glues. Perhaps it would be better to go back to one of the older techniques…
Exploration geophysicists are often confronted with the same issues. They collect reams of data, but how best to analyse them to tease out what they want?
And how can they find out about the latest and most effective approaches without wasting too much time and effort?
That is just the problem the Common Framework for Inference (CoFI) has been set up to solve. It is one of the projects progressed by InLab (www.inlab.edu.au) short for Inference Laboratory, a multi-sided research centre run jointly by CSIRO and the Australian National University (ANU).
Almost all geophysical exploration problems are issues of inference and inversion. What geologists are usually looking for is under the ground and cannot be seen directly. So, its presence has to be inferred indirectly from measuring what is on the surface.
We make similar inferences in medicine when using scanning technology.
A 3D image of a displaced joint, for example, can be generated by computerised tomography (CT)—using a computer to stitch together thousands of measurements of x-rays passing through the affected area, and how they are absorbed or diverted.
“Applying inversion is often somewhat harder in mining and exploration”, says Professor Sambridge, “because you cannot surround a subterranean ore body with sensors.”
The tools provided by CoFI represent techniques for putting data together in a similar fashion to a CT scan to build a model of what is beneath the surface of the Earth.
Leading edge analysis software
CoFI is a collection of the latest methods for analysing geophysical data and more broadly any geo-data.
“It’s a tool bag of methods either we have invented or built ourselves, or have borrowed and adapted from other fields like maths, physics, statistics or computer science,” says Professor Malcolm Sambridge of ANU’s Research School of Earth Sciences, who founded InLab together with Dr Andrew Valentine (ANU) and Dr Juerg Hauser of CSIRO’s Deep Earth Imaging Future Science Platform.
“The motivation is to educate people about what they can infer from their data and allow them to try out different approaches, with minimal effort once they have setup their problem in CoFI," says CSIRO's Dr Hauser.
So geophysicists and others can use CoFI to explore leading edge data analysis methods and, by tinkering with the tools, how to use them.
“It can save a huge amount of project development time,” says ANU Professor Sambridge.
The tools are relevant for a wide range of applications for example the detection and delineation of critical mineral resources and the characterisation and monitoring of ground water resources.
One of the first applications of CoFI to a field data set is centred around the delineation of mineralised zones at the Century Deposit in North West Queensland, halfway between Mt Isa and the Gulf of Carpentaria.
In addition, CoFI is already being used as a teaching tool in graduate courses at the ANU.
The complexity and sophistication of state-of-the-art methodologies mean that it is no longer possible for the same person to be an expert both in understanding data analysis tools and in geophysical exploration.
The latest techniques combine advances in machine learning, applied mathematics, model parametrisation, data collection, forward modelling and predictive applications.
But CoFI makes the latest techniques available, together with examples of how they can be used.
“There’s now no excuse for resorting to the methods of decades ago,” says Professor Sambridge.
InLab is committed to open-source software and CoFI can be downloaded freely.
CoFI also runs in the cloud and so the only thing needed to gain access and run inversions is a web browser.
Because of the collaboration between CSIRO and the ANU, its breadth of examples is poised to grow even more in the future.
The key is collaboration
The ANU researchers bring expertise in seismology on a grand scale. They explore the structure of the whole Earth, whereas CSIRO is concerned with solving the problems of industry, a much finer scale.
While their approaches overlap, the two sides bring different attitudes, approaches and tools which widen the analytical perspective.
“For us industrial problems are hard,” says Professor Sambridge.
“We are interested because they present a challenge and are good tests of our algorithms.”
But researching and providing all this needs support. For the past two years, the two organisations have provided the initial resources to launch InLab and its practical outreach in CoFI.
"While we have made good progress computational frameworks like CoFI that are of broad benefit to Academia, Industry and Government don’t emerge overnight, long term funding is absolutely critical," says Dr Hauser.
Mining and exploration companies can support research at InLab on problems of particular interest to them. The InLab team is also exploring the possibility of a membership system. Companies would pay an annual fee that would not only give them access to the workshops but also to the research students who are working in InLab.
“Early access to such students is of great value to industry,” says Professor Sambridge.
“In the US, it’s often the number one reason that companies join an industry consortium. They find it a cheap way of recruiting the best, fresh talent.”
So far, while CoFI itself has been accessible online free of charge, InLab has been organising and charging for workshop-seminars to educate the exploration industry on the development and application of the tools available on the website.
The first workshop was in Adelaide in November 2022 and a second is being scoped for Perth late 2023.
You can express interest and suggest possible topics using this contact form.