Empowering weather and climate forecast

Boot Camp 2023
Reading, UK
8-10 Nov 2023

Introductory lectures

What it was about

The goal of this three-day Boot Camp was to provide training to use machine learning (ML) in weather and climate applications on High-Performance Computing systems (HPC). The participants got an introduction to the scientific domains of meteorology, ML, and HPC, and got to explore the applications developed in the MAELSTROM project that include:
  • The use of citizen observations and social media data for better weather predictions
  • The use of neural network emulators for faster forecast models
  • The post-processing of ensemble and local weather forecasts
  • The prediction of large-scale weather patterns to support energy production
Intense work in small breakout groups

Who ran it

The event was hosted by the European Centre for Medium-Range Weather Forecasts (ECMWF) at their site in Reading, UK. The tutors came from all MAELSTROM partners:
  • European Centre for Medium-Range Weather Forecasts
  • Eidgenössische Technische Hochschule Zürich (ETH Zürich)
  • 4Cast GmbH & Co. KG
  • Meteorologisk Institutt
  • E4
  • Universite de Luxembourg
  • Forschungszentrum Jülich
The MAELSTROM Boot Camp was organised back-to-back with the second MAELSTROM Dissemination Workshop. Most visitors took the opportunity and attended both events.
Crowded ECMWF council chamber

Who attended it

The target participants were master and PhD students, as well as early career scientists, with a basic background either in meteorology or machine learning or both. Some limited knowledge of Python and standard ML frameworks (e.g. TensorFlow or PyTorch) was encouraged. Also, we recommended that participants were also familiar with Jupyter Notebooks and git, though this was not a requirement.

There was no registration fee.

Well deserved: after-Bootcamp pub with comfort food & drinks

Full agenda

Wednesday, 8 November 2023
Morning session
09:00 to 09:30 A high-level introduction into MAELSTROM, Earth sciences, machine learning, and high performance computing
Presentation slides
Peter Düben (ECMWF)
09:30 to 10:30 A closer look into high performance computing
Presentation slides
Andreas Herten (JSC)
10:30 to 11:00 Coffee break
11:00 to 12:00 A closer look into machine learning for weather and climate applications
Presentation slides
Peter Düben (ECMWF)
12:00 to 12:30 Q&A
12:30 to 13:30 Lunch break
Afternoon session
Group photo
13:30 to 14:30 Group A: A demo of performance models
Karthick Panner Selvam (University of Luxembourg)
13:30 to 14:30 Group B: A shallow entry into deep learning and other machine learning methods
Presentation slides
Peter Düben (ECMWF), Ana Prieto Nemesio (ECMWF)
14:30 to 15:00 Organisation into groups
15:00 to 18:00 Free work in groups

Coffee / tea available at 15:30

19:00 to 21:00 Boot Camp pub dinner
Thursday, 9 November 2023
Morning session
09:00 to 09:30 Progress update of groups
09:30 to 11:00 Free work in groups
11:00 to 11:30 Coffee break
11:30 to 12:30 Free work in groups
12:30 to 13:30 Lunch break
Afternoon session
13:30 to 15:30 Free work in groups
15:30 to 16:00 Coffee break
16:00 to 17:30 Free work in groups
17:30 to 18:00 Progress update of groups
Friday, 10 November 2023
Morning session
09:00 to 11:00 Free work in groups

Including coffee break at 10:30

11:00 to 13:00 Summary of results and close