Guest Lecture

Project Details

Speaker

Speaker

guest lecture highlight climate science machine learning presentation university

Screengrab

Description

The European Centre for Medium-Range Weather Forecasts (ECMWF) is currently integrating machine learning (ML) techniques into its operational weather forecasting system. This is expected to improve the accuracy and reliability of weather forecasts, especially for extreme events. The integration of ML involves developing new algorithms that can efficiently process large amounts of weather data and identify patterns that can be used to make better predictions. The goal is to provide more precise and timely weather forecasts that can help governments, businesses, and individuals make better decisions and prepare for weather-related events. Here's how:

Lecture 18: Guest Lecture - Dr. Jesper Dramsch (ECMWF)

Title: Integrating ML into Operational Weather Forecasts at the ECMWF (by Jesper Dramsch)

Reading

• Düben et al. (2021). Machine Learning at ECMWF: A roadmap for the next 10 years. ECMWF Technical Memoranda.
• ML Roadmap Webinar

Related Readings

• Ben-Bouallegue et al. (2023). Improving medium-range ensemble weather forecasts with hierarchical ensemble transformers. (arXiv preprint)
• Clare et al. (2022). Explainable AI for Bayesian Neural Networks: Toward Trustworthy Predictions of Ocean Dynamics. JAMES.
• Chantry et al. (2021). Machine Learning Emulation of Gravity Wave Drag in Numerical Weather Forecasting. JAMES.
• Data-driven weather forecasting papers:
• FourCastNet (Pathak et al., 2022 / NVIDIA)
• Pangu-Weather (Bi et al., 2022 / Huawei)
• GraphCast (Lam et al., 2022 / DeepMind)

Resources

• Projects
• MAELSTROM project: Machine Learning for Scalable Meteorology and Climate.
• Destination Earth (DestinE) project: creating Earth's digital twin
• WMO S2S Forecast Challenge: code from CRIM team
• Educational Materials
• Machine Learning in Weather & Climate MOOC
• ECMWF Training Course Materials
• Publications (e.g. advances in data assimilation [pdf])
• Software, Jupyter Notebooks, and Examples:
• Inferno Library: a lower-level API for ML inference in operations
• CliMetLab
• Climate Data Store (CDS) Toolbox
• Copernicus Climate Change Service (C3S) Data Tutorials