Computer models of
Earth’s magnetic field.

The work includes setting up the model, running it, checking the output and analysing the results.

I used ARCHER2 and ARC4 to run sets of spherical geodynamo simulations: computer models of the processes that generate Earth’s magnetic field.

I configured the software, submitted and monitored runs, restarted simulations and managed large output files. I then converted those files into a form that could be analysed in Python and MATLAB.

• Maintain and extend modelling code obtained from version-controlled repositories.
• Prepare machine-specific build environments and runtime parameters.
• Submit groups of simulation runs through the computing system’s job scheduler, manage restarts and monitor storage use.
• Transfer and organise multi-gigabyte outputs using Linux, shell scripts and rsync.

• Convert raw NetCDF and statefile outputs into Gauss coefficients, which describe the simulated magnetic field.
• Read, trim and organise the output files using shell scripts, Python and MATLAB.
• Exclude the initial adjustment period and unstable parts of a simulation before interpreting its behaviour.
• Apply the existing QPM comparison methods using Cromwell and GGF100k data, and collect the results in a common simulation dictionary.

• Calculate magnetic and kinetic energy budgets, ohmic dissipation and ohmic fraction.
• Examine how the calculated quantities change through time and calculate averages for stable parts of each run.
• Compare the results from multiple simulation runs.
• Document the processing steps and use the same file structures so new simulations can be analysed in the same way.