Computational Earth Laboratory

The Computational Earth Laboratory is a shared RSES capability for high-performance computing, numerical modelling, inverse methods, data analysis and open research software. We bring together the tools and expertise required to investigate Earth processes that are difficult to reproduce in the laboratory or observe directly. Its core computational platforms include G-ADOPT, Underworld and ACCESS ocean and climate models.

Computational Earth science

Our work connects physical models with observations across the solid Earth, surface environment and fluid envelope. Applications include mantle convection and plate dynamics, lithospheric deformation, earthquakes, seismic imaging, geodetic deformation, glacial isostatic adjustment, sea-level change, ocean and climate dynamics, continental groundwater, and the analysis of large Earth-observation datasets.

High-performance computing

Many of these problems require simulations at a scale that exceeds desktop computing. The laboratory supports the development, scaling and reproducible use of parallel workflows across RSES, ANU and national computing systems. RSES computing resources include the TerraWulf Compute Cluster, which remains a distinct facility with its own technical specifications and access arrangements.

Models, data and inverse methods

Numerical models are powerful tools for understanding Earth-system processes and projecting future change, and can also be combined with observations through data assimilation and inverse methods. We develop adjoint methods, optimisation workflows and data-assimilation approaches that estimate uncertain Earth properties and past conditions from satellite, seismic, geological and geodetic observations. These methods allow us to ask how well a proposed model explains the observations, and what additional measurements would reduce uncertainty.

Research software and computational platforms

The laboratory connects and supports open RSES research software platforms. G-ADOPT, Underworld and ACCESS are core platforms, together with workflows for data processing, model construction, visualisation, optimisation and reproducibility.

G-ADOPT provides scalable finite-element modelling and adjoint optimisation for geoscientific inverse problems. Underworld is a parallel particle-in-cell finite-element platform for geodynamics, designed to model large deformation and history-dependent Earth materials. Its current generation, Underworld3, expresses physical models as symbolic mathematics in Python and executes them efficiently from laptops to high-performance computing systems. ACCESS (Australian Community Climate and Earth System Simulator) is Australia’s national Earth system modelling framework which enables research on the ocean, climate and cryosphere.

These platforms serve different but complementary research problems. Together they support modelling of mantle and lithospheric dynamics, tectonics, earthquakes, viscoelastic deformation, fluid and thermal transport, ocean circulation, sea ice, climate variability and data-constrained Earth-system processes. The laboratory supports software that is open, documented, benchmarked and reproducible, so that it can be used and extended beyond a single project or research group.

Training and collaboration

We support students, researchers and technical staff who use computation in Earth science. The laboratory brings together researchers who develop and use computational methods across RSES. It provides a focus for training in numerical methods, high-performance computing, research software and reproducible workflows, including the development and use of G-ADOPT and Underworld. Students and researchers can contribute through scientific applications, software development, documentation, benchmarks and training activities.