Unlocking Machine Learning in Earth Sciences: Think Tank Event

On Friday, 21 February, the Curtin Frontier Institute for Geoscience Solutions (CFIGS) hosted a community Think Tank event at Curtin University, showcasing the transformative role of machine learning (ML) across Earth and Planetary Sciences.

Bringing together researchers from across the School of Earth and Planetary Sciences and collaborators from government and industry the event explored how ML is reshaping research from deep Earth processes to orbital satellite tracking.

The morning opened with a strategic briefing from Prof Chris Kirkland, outlining CFIGS’ vision to integrate cutting-edge tools and collaborative approaches across the full mining and geoscience value chain. A rapid series of talks followed, highlighting applied ML across diverse domains:

• Dr Taryn Scharf warned of statistical pitfalls in supervised ML;

• Dr Amir Allahvirdi-Zadeh demonstrated deep learning for satellite constellation prediction;

• Dr Hadrien Devillepoix shared innovations in meteorite tracking via drone and crowdsourced ML;

• Piero Sampiao explored mantle isotope patterns through ML clustering;

• Dr Tasman Gillfeather-Clark discussed open file geoscience data from GSWA;

• Emad Al-Hemyari presented deep learning approaches to microseismic detection using DAS;

• A/Prof Petra Helmholz applied ML to urban heat island mapping and speech analysis; and

• Prof Melanie Johnston-Hollitt highlighted Curtin’s broader data science and HPC capabilities.

With over 50 participants, the event fostered new conversations around harmonising data, accelerating discovery, and ensuring scientific rigour in ML workflows. It culminated in networking over lunch, an opportunity to build collaborations that push Earth science into its next computational frontier.

CFIGS continues to support researchers working at the interface of science and application, driving innovation that connects geoscience insight to real-world challenges.