Direkt zum Inhalt | Direkt zur Navigation

Personal tools

You are here: Home » Community » Announcements » Events » Joint IS-ENES3/ESiWACE2 Virtual Workshop on New Opportunities for ML and AI in Weather and Climate Modelling

Joint IS-ENES3/ESiWACE2 Virtual Workshop on New Opportunities for ML and AI in Weather and Climate Modelling

last modified Apr 09, 2021 11:57 AM
Mar 16, 2021 03:00 PM to Mar 18, 2021 06:30 PM (Europe/Vienna / UTC100)
Add event to calendar

Organised jointly by the H2020 projects IS-ENES3 and ESiWACE2, the workshop was held virtually from the 16th to 18th March 2021 (3.5h per day). 

The aim of this workshop was to bring together climate scientists and experts from academia and industry to share knowledge and experience and to identify new opportunities in the areas of machine learning, artificial intelligence and big data techniques for Weather and Climate Modelling.

The workshop was organised around three sessions:

- 16 March (Day 1): Views from Domain Science

- 17 March (Day 2): ML/AI Software Technologies

-18 March (Day 3): High performance, Infrastructure and Big data Challenges

The final agenda is available here.

The participants' personal data used for the organisation of the workshop was covered by the following privacy policy.

 Recorded talks

Find the recording of the talks on the IS-ENES3 Youtube channel, under the event playlist !


  • Session 1 : Views from Domain Science
New approaches based on ML for a range of climate prediction problems, Emily Shuckburgh (U. Cambridge)
Philosophy and Targeted Applications of ML/AI Techniques for Climate Risk Analytics at Jupiter, Luke Madaus & Steve Sain (Jupiter Intel.)
The optimization dichotomy: Why is it so hard to improve climate models with machine learning, Stephan Rasp (ClimateAI)
Improving convection parameterizations with a library of large-eddy simulations, Zhaoyi Shen (Caltech)
Stochastic Super-Resolution for Convective Regimes using Gaussian Random Fields, Rachel Prudden (Met Office Inf. Lab)
Subseasonal Forecasts of Opportunity Identified by an Explainable Neural Network, Kirsten Mayer (CSU)
Using transfer learning and backscattering analysis to build stable, generalizable data-driven subgrid-scale models: A 2D turbulence test case, Pedram Hassanzadeh (U. Rice)
  • Session 2: ML/AI Software technologies
Stochastic machine learning for atmospheric fields with generative adversarial networks, Jussi Leinonen (MeteoSwiss)
Causal discovery in time series with unobserved confounders, Andreas Gerhardus (DLR Jena)
Estimating stochastic closures using sparsity-promoting ensemble Kalman inversion, Jinlong Wu (Caltech)
Deep Learning on the sphere for weather/climate applications, Gionata Ghiggi and Michaël Defferrard (EPFL)
Deep learning-based remote sensing for infrastructure damage assessment, Thomas Chen (AMSE)
Leveraging physics information in neural networks for fluid flow problems, Akshay Subramaniam (NVIDIA)
Sub-seasonal forecasting with a large ensemble of deep-learning weather prediction models, Jonathan Weyn (U. of Washington)
  • Session 3: High performance, Infrastructure and Big data challenges
Scaling Up Deep Learning Workloads - A Data-Centric View, Tal Ben-Nun (ETHZ)
Radar QPE and Machine Learning, Micheal Simpson (NOAA)
Using ML at the Edge to Improve Data Gathering, Pete Warden (Google)
An Overview of ML and AI on Arm Based HPC Systems for Weather and Climate Applications, Phil Ridley (Arm)
Machine-Learned Preconditioners for Linear Solvers in Geophysical Fluid Flows, Jan Ackmann (U. Oxford)
You do you. How next-gen data platforms can stop weather and climate scientists from being software engineers and other perversions, Theo McCaie (MO Informatics Lab)
3D bias correction with deep learning in the Integrated Forecasting System, Thorsten Kurth (NVIDIA)