🗓️Machine Learning for Ocean Modelling Workshop

Details of the schedule are below. Over the two days, we have three sessions with keynote and oral presentations, a dedicated poster session, a practical session and breakout groups.

The schedule is provisional and subject to change.

Last updated 29th June

🕘Day 1 - Tuesday 7 July 2026

10:00 - 10:20 | Arrival

  • Tea and coffee

10:20 - 10:35 | Opening remarks

  • Introduction to the workshop and key themes

10:35 - 11:20 | Keynote talk

Rosie Lickorish: “What Geospatial Foundation Models can do for Ocean Modelling”

11:20 - 12:40 | Oral session 1 - ML for analysis and downstream applications

20 minute (15 minutes for presenting + 5 minutes for questions) talks

11:20 - 11:40 Thomas Prime: Physically Informed Neural Network to infer δš⁸O distributions from routinely observed oceanographic variables
11:40 - 12:00 Mojtaba Masoudi: The Geometry of Embedded Representations for Long-Tailed Classification
12:00 - 12:20 Yavor Kostov: Transient irreversibility of southern extratropical water mass anomalies under CO2 forcing
12:20 - 12:40 Ollie Tooth: OceanDataStore: Accelerating Machine Learning & Model Validation with Cloud-Native Ocean Data

12:40 - 13:30 | Lunch

13:30 - 14:00 | Practical session: Introduction

14:00 - 15:30 | Practical session:

  1. Anemoi demo
  2. Hybrid ML modelling
  3. ML for newbies (notebook tutorials)

15:30 - 16:00 | Tea and coffee

16:00 - 16:10 | Posters: 1 minute introduction slides

16:10 - 17:30 | Breakout discussions

  • Topics and questions TBC

17:30 | Close

19:00 | Optional evening meal

🕘Day 2 - Wednesday 8 July 2026

09:00 - 09:15 | Arrival

09:15 - 11:00 | Oral session 2 - ML for observations, modelling and data assimilation

20 minute (15 + 5) talks

09:15 - 09:35 Gian Giacomo Navarra: “Seasonal Variability of Organic Carbon Flux and Remineralization Inferred With a Bayesian Physics-Informed Neural Network”
09:35 - 09:55 Jozef Skakala: “How to emulate a highly complex marine ecosystem model with deep learning”
09:55 - 10:00 Comfort break
10:00 - 10:20 Rachel Furner: “Developing a data-driven 3d global ocean model at ECMWF”
10:20 - 10:40 Daniel Holmberg: “Njord: A Probabilistic Graph Neural Network for Ensemble Ocean Forecasting”
10:40 - 11:00 Thomas Gardner: “FRAME-FM: a discussion on a joint EDS framework for making AI model training more approachable”

11:00 - 11:30 | Tea and coffee

11:30 - 12:30 | Poster session

Posters should be A0 portrait maximum please, to fit the poster boards (so A1 landscape is fine etc.).

James While “Machine learned ocean prediction at the Met Office: The SPrAI project”
Jonathan Coney “Interpolating sparse ocean observations using machine learning”
Miriam North Ridao “Towards Machine-Learned Subgrid Parameterisations in a Double-Gyre Quasi-Geostrophic Model”
Thomas Wilder “Towards model-independent machine learning parameterisations of mesoscale eddies”
Ben Timmermans “Towards estimation of impacts of ocean surface wave breaking on long term climate using recent wave hindcasts, emulators and climate models”

12:30 - 13:15 | Lunch

13:15 - 14:00 | Keynote talk

Pavel Perezhogin: “Data-driven approaches for parameterizing ocean mesoscale eddies”

14:00 - 15:00 | Oral session 3 - Integrating ocean processes and hybrid physics-ML modelling

20 minute (15 + 5) talks

14:00 - 14:20 Fay Luxford: “A Physics-Informed Graph Neural Network Surrogate for Nearshore Wave Modelling”
14:20 - 14:40 Marcus Juniper: “Lightweight approaches for real-time observation driven bias correction”
14:40 - 15:00 Niraj Agarwal: “Skillful Global Ocean Emulation and the Role of Correlation-Aware Loss”

15:00 | Closing discussion

All times are in BST (GMT +1).