📅Date: 24-25 June 2025
📍Location: University of Reading, UK (In-person & Remote Options)
⚒️Organized by: NCAS, NOC, BAS, Met Office, and ECMWF
🔎Contact: t.m.wilder at reading.ac.uk or a.f.sommer at reading.ac.uk
About the Workshop
We are excited to announce the Machine Learning for Ocean Modelling Workshop, bringing together researchers to explore the evolving role of AI and ML in ocean science, share advancements, and discuss future directions. The workshop will feature keynote talks, short presentations, and posters on key themes such as hybrid modelling, benchmarking, and observational data integration. The event aims to foster collaboration, map UK Ocean AI activities, and identify challenges and opportunities in this rapidly growing field.
Themes & Sessions
Main Sessions:
AI for Ocean Model analysis, Postprocessing and Downstream Uses
Description: This session will explore how ML can enhance the analysis, postprocessing, and downstream applications of ocean model outputs. Topics may include bias correction, feature extraction, emulation, uncertainty quantification, and linking model outputs to decision-making processes in climate science, marine ecosystem, and operational oceanography. Discussion will focus on leveraging ML to improve the usability, accuracy, and interpretability of ocean model data, as well as identifying new ways ML can be used to refine and enhance process understanding.
🎯Target question: How can ML-driven approaches enhance the analysis, refinement, and application of ocean model outputs for scientific and operational use cases?
Integrating Ocean Processes and Hybrid Approaches in ML
Description: This session will explore methods to embed process-based constraints, conservation laws, and ocean dynamics into machine learning models, as well as hybrid approaches that combine machine learning with traditional numerical ocean models. Topics may include physical-informed ML, data assimilation, ML-driven parameterisation, model correction, and surrogate modelling techniques to enhance both accuracy and computational efficiency. Discussion will focus on improving interpretability, stability, and reliability in AI-driven ocean process representation and prediction.
🎯Target Question: How can we best integrate ocean processes and hybrid approaches into ML models to enhance generalisation, stability, and predictive skill in ocean forecasting and climate simulation?
Benchmarking AI for Ocean Modelling: Methods, Metrics, and Challenges
Description: To ensure robust AI applications in ocean science, we need clear benchmarking methodologies. This session will focus on performance metrics, standard datasets, reproducibility challenges, and the role of intercomparison studies for AI in ocean modelling. It will also explore how benchmarking can be used to assess ML contributions to improving process representation in ocean models.
🎯Target Question: How do we define success in AI for ocean modelling, and what benchmarks are needed to evaluate different approaches?
AI for Ocean Observations (OSSEs) & AI for Prediction
Description: This session will cover AI applications in ocean observing systems and predictive modelling, including AI-driven Observation System Simulation Experiments (OSSEs), data fusion, and real-time forecasting. It will also address how AI can enhance ocean monitoring and improve the representation of physical and biogeochemical processes in both models and observational frameworks.
🎯Target Question: How can AI best be used to improve ocean monitoring and forecasting in both scientific and operational settings?
Breakout Sessions:
- AI-Only Ocean Modelling: Hype or Future Reality?
- Building an Ocean ML Community: Skills, Training, and Collaboration
- Mapping UK Ocean AI Activities: Research, Industry, and Collaboration
- Trustworthy and generalisable AI for ocean modelling
Workshop Outcomes
Mapping UK Ocean AI activities across academia, research centers, and industry. Position paper articulating key challenges in Ocean AI. Network building for future collaborations and funding proposals.
Call for Participation
We invite researchers, students, industry professionals, and policymakers to contribute through talks, posters, and discussions.
📢 Submit your application by: 12th May
🔗 Apply here
A list of current in-person attendees can be found here