Overview of Operational and Research Ocean Models

Across the week you’ve built up what ocean models are, how they’re constructed, and their trade-offs. In this final lecture, Emanuela Clementi brings it all together by surveying the models actually used around the world – and how modellers choose between them.

The lecture is organised around a few key distinctions:

• Operational versus research models – operational models run routinely (daily or more often) to deliver real-time forecasts, so they must be stable, efficient, and reliable. Research models are built for scientific study, often more complex or experimental. The defining difference is that operational models prioritise robustness and timeliness
• A hierarchy of scales – global models simulate the whole ocean (~1/12°, ~8-10 km) and provide the large-scale context and boundary conditions for others; regional models (~1-10 km) refine specific basins, forced at their edges by global output; coastal models reach from kilometres down to tens of metres, handling ports, estuaries, storm surge, and flooding, with complex bathymetry and wetting-and-drying
• Downscaling – the strategy that links these scales, using a global model’s output as the boundary conditions for a finer regional or coastal model
• The major models in practice – including NEMO (the backbone of many European operational services), ROMS (coastal and regional, terrain-following), HYCOM (global, hybrid coordinates), MITgcm (flexible, non-hydrostatic, research), FVCOM and CROCO (unstructured or high-resolution coastal work)
• The challenges that remain – the high cost of high resolution, limited observations to constrain small scales, the need for good parameterizations, and the reminder that more resolution doesn’t help if there aren’t enough observations to support it

The takeaway pulls the week together: ocean models operate across global, regional, and coastal scales; global models set the context while regional and coastal models refine the detail; downscaling bridges them; and every design is a trade-off between resolution, physics, and computational cost. There is no single best model – only the right one for the job, which is exactly what you’ll weigh up in this week’s activities.