Report – 4th COMMODORE workshop in Trieste by Hugo Lévy

Last September, I participated in the 4th COMMODORE workshop: the COMmunity for the numerical MODeling of the glObal, Regional and coastal ocEan – the kind of convoluted acronym that we all love. After Paris, Hamburg and Boulder, the 2026 edition – my first – was hosted by the ICTP in Trieste, Italy, overlooking the Adriatic Sea. Beyond this charming setting, I was quite excited about the scientific programme and its strong focus on numerical techniques for ocean models. Over the years, I have come to realise that this type of workshop is probably the most relevant kind of scientific gathering for me, mainly because of its manageable size (a little over 50 participants in the present case) and sufficiently narrow scope. These workshops also give me, as a presenter, the opportunity to receive external and often spot-on feedback on my own work.

As anticipated, getting there was not so easy. No longer keen on flying for business trips, I travelled by train from Bremen. The long and somewhat hectic journey was made much more enjoyable thanks to Steffen Maaß, another TRR 181 member who was also attending the workshop and travelling with me. We had planned to arrive two days early so that we could enjoy the area and the Adriatic Sea a little, as summer was coming to an end.

The workshop itself lived up to expectations. There were many familiar faces and just as many new encounters over the four intensive days. The topics covered in the oral presentations and posters ranged from the subtle yet crucial details of numerical schemes and how they are implemented in ocean GCMs, to subgrid-scale parameterisations and the use of machine learning in ocean models. As one prominent participant put it, this was a great setting for discussing all the “nerdy” stuff that keeps us entertained as scientists. I particularly enjoyed learning about the history of model development, thanks to the many contributions from people who have been working in this field for a long time now.

Each day ended with an open discussion session on specific issues, such as the perspective of the Global South on ocean modelling, or taking stock of where we currently stand in terms of Machine Learning in ocean modelling. In my opinion, the topic that sparked the most discussion was, unsurprisingly, AI and its now-ubiquitous use across all fields of science — ours being no exception. We openly shared our experiences with AI, discussing both the opportunities it creates and the concerns it raises, and speculated on what the (near) future of our field might look like. It was insightful to hear other scientists' perspectives on what feels to me like a major upheaval in the way we work. To be continued…

On the last day, I gave a talk myself on eddy parameterisation over the Antarctic continental slope. I explored whether traditional approaches are well suited to this region, looking both at their underlying assumptions and at how well they perform in ocean models. The talk connected naturally with some of the earlier presentations. It also followed nicely from the discussion session the previous evening, which focused on whether traditional mesoscale parameterizations are still needed in eddy-permitting models, given recent advances in kinetic energy backscatter and numerical methods.

After one last coffee break, it was already time to wrap up the workshop. We did so in a friendly atmosphere, with a walk along the seashore (including a quick dip for some of us), followed by dinner at a large pizzeria in the city centre of Trieste. Overall, I am very grateful to TRR181 for supporting this trip. I came back with the feeling that I had learnt new things, gained some fresh ideas and found extra motivation to continue my research!