Theo Ellingsen: Model Live Diagnostics – an ACCESS-NRI internship project
Authors
Theo Ellingsen (UTAS)
Lachlan Whyborn (ACCESS-NRI)
Charles Turner (ACCESS-NRI)
Mike Tetley (ACCESS-NRI)
Heidi Nettelbeck (ACCESS-NRI)
Paige Martin (ACCESS-NRI)
Introduction and scientific context
Climate models can take months to run on and produce very large datasets, making it difficult to quickly assess whether simulations are progressing correctly while models are still running. As part of an ACCESS-NRI internship, we are developing interactive tools that allow researchers to inspect “live” model output during runtime. ACCESS-NRI’s PhD Internship Program embeds PhD students within ACCESS-NRI to contribute to this infrastructure while gaining experience in research software engineering. In this poster, we present the program itself, work towards the Model Live Diagnostics (MLD) project, and some reflections on the experience thus far.
Methodology/project summary
We describe the structure of the internship program and the plan for the MLD project. This project will improve the ability to monitor, evaluate, and troubleshoot climate model simulations while they are running, helping identify issues earlier, reduce wasted HPC time, and improve the efficiency of model development workflows.
Results
This poster will be presented midway through the internship, so we encourage input from the community on the model outputs (e.g. variables) that would be of most use for this tool to monitor model runs “live”. We will also share reflections on the experience and project so far, including what it looks like to contribute to real research infrastructure as an early-career researcher, and what the internship program offers both to interns and to the broader ACCESS community.
Audience
- Anyone interested in the ACCESS-NRI internship program or in building research software engineering skills
- Anyone who runs ACCESS models
Keywords
Research software engineering, internship, Python, ACCESS, building technical skills