MAELSTROM dissemination workshop (28 March) and Machine Learning Workshop (29 March - 1 April)

Monday, 28 March
MAELSTROM Workshop: Updates on MAELSTROM
8:30 AM 8:50 AM
Introduction
Speaker: Peter Dueben (ECMWF)

Virtual

20m
8:50 AM 9:10 AM
WP1 Summary on Applications and Data Sets
Speaker: Bing Gong (JSC)

Virtual

20m
9:10 AM 9:30 AM
WP2 Summary on Software Tools
Speaker: Fabian Emmerich (4-cast)

Virtual

20m
9:30 AM 9:50 AM
WP3 Summary on Hardware Benchmarks
Speakers: Andreas Herten (JSC) , Daniele Gregori (E4 Computer Engineering SPA)

Virtual

20m
9:50 AM 10:30 AM
Possibility for general discussion

Virtual

40m
10:30 AM 11:00 AM
Coffee break

Virtual

30m
EuroHPC Partner Project Talks
11:00 AM 11:20 AM
TimeX
Speakers: Giovanni Samaey (KU Leuven) , Martin Schreiber (Technical University of Munich)

Virtual

20m
11:20 AM 11:40 AM
Deep-Sea
Speaker: Estela Suarez (JSC)

Virtual

20m
11:40 AM 12:00 PM
Red-Sea
Speaker: Nikos Xrysos (FORTH)

Virtual

20m
12:00 PM 12:30 PM
Discussions between MAELSTROM and the speakers

Virtual

30m
12:30 PM 1:30 PM
Lunch break

Virtual

1h
External Science Talks
1:30 PM 2:00 PM
Time-Consistent Downscaling of Atmospheric Fields with Generative Adversarial Networks
Speaker: Jussi Leinonen (MeteoSwiss)

Virtual

30m
2:00 PM 2:30 PM
Pangeo: An Open Source Ecosystem for Data-Intensive Science
Speaker: Ryan Abernathey (Columbia University)

Virtual

30m
2:30 PM 3:00 PM
Deep Learning for Earth Sciences in the HPC Context
Speaker: Thorsten Kurth (NVIDIA Corporation)

Virtual

30m
3:00 PM 3:30 PM
Discussions between MAELSTROM and the speakers

Virtual

30m
3:30 PM 4:00 PM
Coffee break

Virtual

30m
4:00 PM 5:30 PM
Possibility for further discussions between the working groups

Virtual

1h 30m
Tuesday, 29 March
Machine Learning Workshop: Session 1: Machine learning for emulation of model components
Chair: Matthew Chantry (ECMWF)
9:40 AM 10:00 AM
Welcome and introduction
Speakers: Florian Pappenberger (ECMWF) , Peter Dueben (ECMWF)

Virtual

20m
10:00 AM 10:20 AM
Climate-Invariant, Causally Consistent Neural Networks as Robust Emulators of Subgrid Processes across Climates
Speaker: Tom Beucler (University of Lausanne)

Virtual

20m
10:20 AM 10:40 AM
How AI/ML interpenetrate into Weather Forecast: NN emulator for radiation parameterization and Retrieval similar weather condition using satellite images
Speaker: Hyesook Lee (Director of AI Weather Forecast Research Division)

Virtual

20m
10:40 AM 11:00 AM
ClimateBench: A benchmark dataset for data-driven climate projections
Speaker: Duncan Watson-Parris (University of Oxford)

Virtual

20m
11:00 AM 11:20 AM
Why and how to learn end-to-end subgrid closures for atmosphere and ocean models?
Speaker: Julien Le Sommer (IGE, UGA/CNRS/IRD/G-INP)

Virtual

20m
11:20 AM 11:40 AM
Coffee break

Virtual

20m
11:40 AM 12:40 PM
Poster session

Virtual

1h
12:40 PM 2:00 PM
Lunch break

Virtual

1h 20m
2:00 PM 2:20 PM
Developing an emulator of an Ocean GCM
Speaker: Rachel Furner (University of Cambridge)

Virtual

20m
2:20 PM 2:40 PM
An Online-Learned Neural Network Chemical Solver for Stable Long-Term Global Simulations of Atmospheric Chemistry
Speaker: Makoto Kelp (Harvard University)

Virtual

20m
2:40 PM 3:00 PM
Neural-network parameterization of subgrid momentum transport learned from a high-resolution simulation
Speaker: Janni Yuval (MIT)

Virtual

20m
3:00 PM 3:40 PM
Coffee break

Virtual

40m
3:40 PM 4:00 PM
Machine learning for gravity waves in a climate model
Speaker: Steven Hardiman (Met Office)

Virtual

20m
4:00 PM 4:20 PM
Physics-Informed Learning of Aerosol Microphysics
Speaker: Paula Harder (Fraunhofer Institute ITWM)

Virtual

20m
4:20 PM 4:40 PM
Neural network emulation of precipitation and condensation processes in FV3GFS
Speaker: Noah Brenowitz (Allen AI)

Virtual

20m
4:40 PM 5:00 PM
Towards physics-based machine learning for land surface modeling: The case of land-atmosphere interactions
Speaker: Andrew Bennett (University of Washington)

Virtual

20m
5:00 PM 5:15 PM
Coffee break

Virtual

15m
5:15 PM 6:00 PM
Discussion with all speakers - moderated by Matthew Chantry

Virtual

45m
Wednesday, 30 March
Session 2: Machine learning for forecasts from now-casting to seasonal
Chair: Peter Dueben (ECMWF)
9:30 AM 9:50 AM
High-Resolution Solar Power Nowcasting by Deep Learning: How to Extract Features from Historic Time Series, Remote Sensing, and Numeric Weather Prediction Models to achieve Optimized Machine Learning Forecasts
Speaker: Petrina Papazek (ZAMG)

Virtual

20m
9:50 AM 10:10 AM
Improving sub-seasonal forecasts by correcting missing teleconnections using ANN-based post-processing
Speaker: Chiem van Straaten (KNMI)

Virtual

20m
10:10 AM 10:30 AM
Interpretable Deep Learning for Probabilistic MJO Prediction
Speaker: Antoine Delaunay (Ecole Polytechnique / Imperial College London )

Virtual

20m
10:30 AM 10:50 AM
Convolutional neural networks for skillful global probabilistic prediction of temperature and precipitation on sub-seasonal time-scales
Speaker: Nina Horat (Karlsruhe Institute of Technology, Department of Economics, Institute of Econometrics and Statistics)

Virtual

20m
11:00 AM 11:30 AM
Break

Virtual

30m
Session 2a: Machine learning for forecasts from now-casting to seasonal
Chair: Shraddha Gupta (ECMWF)
11:30 AM 11:50 AM
Identifying relevant large-scale predictors for sub-seasonal precipitation forecast using explainable neural networks
Speaker: Niclas Rieger (Centre de Recerca Matemática (CRM))

Virtual

20m
11:50 AM 12:10 PM
Ensemble forecast of the Madden Julian Oscillation using a stochastic weather generator based on analogs of Z500
Speaker: Meriem Krouma (ARIA Technologies; Laboratoire des Sciences du Climat et de l’Environnement, UMR 8212 CEA-CNRS-UVSQ, IPSL \& Université Paris-Saclay)

Virtual

20m
12:10 PM 12:30 PM
Improving the prediction of the Madden-Julian Oscillation of the ECMWF model by post-processing
Speaker: Riccardo Silini (Universitat politecnica de Catalunya)

Virtual

20m
12:30 PM 12:50 PM
Sub-Seasonal Probabilistic Precipitation Forecasting using Extreme Learning Machine
Speakers: Kyle Hall (International Research Institute for Climate & Society) , Nachiketa Acharya (Pennsylvania State University)

Virtual

20m
1:00 PM 3:00 PM
Lunch break

Virtual

2h
Session 2b: Machine learning for forecasts from now-casting to seasonal
Chair: Nikolaos Mastrantonas (ECMWF)
3:00 PM 3:20 PM
Opportunistic mixture model for post-processing S2S temperature and precipitation forecasts using convolutional neural networks
Speaker: David Landry (Computer Research Institute of Montreal)

Virtual

20m
3:20 PM 3:40 PM
Deep learning augmented numerical weather prediction digital twin experiments for global precipitation forecasting
Speaker: Manmeet Singh (University of Texas)

Virtual

20m
3:40 PM 4:00 PM
Improving medium-range ensemble forecasts with transformers
Speaker: Jonathan Weyn (Microsoft)

Virtual

20m
4:00 PM 4:20 PM
Forecasting Global Weather with Graph Neural Networks
Speaker: Ryan Keisler

Virtual

20m
4:20 PM 4:40 PM
Deep learning weather prediction: epistemology and new scientific horizons
Speaker: Dale Durran (University of Washington, Seattle)

Virtual

20m
4:40 PM 5:00 PM
Coffee break

Virtual

20m
5:00 PM 5:45 PM
Discussion with all speakers - moderated by Peter Dueben

Virtual

45m
Thursday, 31 March
Session 3: Machine learning for feature detection and user applications
Chair: Mihai Alexe (ECMWF)
10:00 AM 10:20 AM
A Machine Learning Approach to Stochastic Downscaling of Precipitation Forecasts
Speaker: Lucy Harris (AOPP at Oxford University)

Virtual

20m
10:20 AM 10:40 AM
Causal deep learning for studying the Earth system: soil moisture-precipitation coupling in ERA5 data across Europe
Speaker: Tobias Tesch (Forschungszentrum Jülich)

Virtual

20m
10:40 AM 11:00 AM
Rainfall scenarios from AROME-EPS forecasts using autoencoder and climatological patterns
Speaker: Arnaud Mounier (Météo-France)

Virtual

20m
11:00 AM 11:20 AM
Utilizing self-learning capability of a deep neural network and continuous monitoring of geostationary satellite to understand clouds structure and organization.
Speaker: Dwaipayan Chatterjee (Institute for Geophysics and Meteorology,University of Cologne)

Virtual

20m
11:30 AM 1:00 PM
Lunch break

Virtual

1h 30m
Session 3a: Machine learning for feature detection and user applications
Chair: Jesper Dramsch (ECMWF)
1:00 PM 1:20 PM
ML-based fire hazard model trained on thermal infrared satellite data
Speaker: Julia Gottfriedsen (Ludwig Maximilian University of Munich (LMU), OroraTech)

Virtual

20m
1:20 PM 1:40 PM
Photographic Visualization of Weather Forecasts with Generative Adversarial Networks
Speaker: Christian Sigg (MeteoSwiss)

Virtual

20m
1:40 PM 2:00 PM
Identifying Lightning Processes in ERA5 Soundings with Deep Learning
Speaker: Gregor Ehrensperger (Department of Mathematics, University of Innsbruck)

Virtual

20m
2:00 PM 2:20 PM
Deep Learning for the Verification of Synoptic-scale Processes in NWP and Climate Models
Speaker: Julian Quinting (Karlsruhe Institute of Technology)

Virtual

20m
2:30 PM 3:00 PM
Coffee break

Virtual

30m
Session 3b: Machine learning for feature detection and user applications
Chair: Mihai Alexe (ECMWF)
3:00 PM 3:20 PM
Latent space, feature space and the global domain – how ozone research can benefit from explainable machine learning
Speaker: Clara Betancourt (Jülich Supercomputing Centre)

Virtual

20m
3:20 PM 3:40 PM
Learning from Noisy Class Labels for Earth Observation
Speaker: Begüm Demir (TU Berlin)

Virtual

20m
3:40 PM 4:00 PM
Generative Adversarial Networks for Extreme Super-Resolution and Downscaling of Wind Fields at Convection-Permitting Scales
Speaker: Nic Annau (Environment and Climate Change Canada, University of Victoria)

Virtual

20m
4:00 PM 4:20 PM
A machine learning correction model for the warm bias over Arctic sea ice in atmospheric reanalyses
Speaker: Lorenzo Zampieri (National Center for Atmospheric Research (NCAR))

Virtual

20m
4:30 PM 4:45 PM
Coffee break

Virtual

15m
4:45 PM 5:30 PM
Discussion with all speakers - moderated by Mihai Alexe

Virtual

45m
Friday, 1 April
Session 3c: Machine learning for feature detection and user applications
Chair: Mihai Alexe (ECMWF)
9:00 AM 9:20 AM
Spatially coherent postprocessing of cloud cover and precipitation forecasts using generative adversarial networks
Speaker: Stephan Hemri (University of Zurich)

Virtual

20m
9:20 AM 9:40 AM
Improvements of the Adriatic Deep-Learning Sea Level Modeling Network HIDRA
Speaker: Marko Rus (Faculty of Computer and Information Science, Visual Cognitive Systems Lab, University of Ljubljana)

Virtual

20m
9:40 AM 10:00 AM
Spatio-temporal Forecasting of Meteorological Visibility over Northwest of Morocco using Long short-term memory (LSTM) network.
Speaker: Badreddine Alaoui ((1) Hassania School of Public Works, LaGeS/ MoNum ; (2) Direction Générale de la Météorologie, CNRM)

Virtual

20m
10:00 AM 10:30 AM
Break

Virtual

30m
Session 4: Machine learning tools and high-performance computing
Chair: Peter Dueben (ECMWF)
10:30 AM 10:50 AM
The European Destination Earth project and its potential for boosting machine learning
Speaker: Peter Bauer (ECMWF)

Virtual

20m
10:50 AM 11:10 AM
Convergence of HPC and Data Science at the Edinburgh International Data Facility
Speaker: Mark Parsons (EPCC, The University of Edinburgh)

Virtual

20m
11:10 AM 11:30 AM
Improving the radiative scheme with machine learning on an heterogeneous cluster
Speakers: Rémi Druilhe (Atos) , Christophe Bovalo (Atos)

Virtual

20m
11:30 AM 11:50 AM
Physics-Aware Downsampling with Deep Learning for Scalable Flood Modeling
Speaker: Niv Giladi (Technion - Israel Institute of Technology)

Virtual

20m
12:00 PM 1:00 PM
Lunch break

Virtual

1h
1:00 PM 1:20 PM
Improving rare events predictions by oversampling tabular data with a mix of categorical and continuous variables by generative adversarial networks
Speaker: Alla Sapronova (StormGeo)

Virtual

20m
Session 4a: Machine learning tools and high-performance computing
Chair: Florian Pinault (ECMWF)
1:20 PM 1:40 PM
Generative machine learning for extreme climate scenarios
Speaker: Bianca Zadrozny (IBM Research)

Virtual

20m
1:40 PM 2:00 PM
Use of Machine learning for the detection and classification of observation anomalies
Speaker: Mohamed Dahoui (ECMWF)

Virtual

20m
2:00 PM 2:20 PM
Building Digital Twins of the Earth for NVIDIA’s Earth-2 Initiative
Speakers: Karthik Kashinath (NVIDIA Corporation) , Jaideep Pathak (NVIDIA)

Virtual

20m
2:30 PM 3:00 PM
Coffee break

Virtual

30m
3:00 PM 3:20 PM
Exploring the Use of Machine Learning and Remote Sensing for Traffic Map Generation at Large Scale
Speaker: Taha Alfaqheri (4EI)

Virtual

20m
3:20 PM 3:40 PM
Tropical Extreme Weather Event Management and Climate Adaptation via Supervised Computer Vision-based Algorithms
Speaker: Thomas Chen (Academy for Mathematics, Science, and Engineering)

Virtual

20m
3:40 PM 4:00 PM
Accelerating computational fluid dynamics with deep learning
Speaker: Stephan Hoyer (Google Research)

Virtual

20m
4:00 PM 4:15 PM
Coffee break

Virtual

15m
4:15 PM 5:00 PM
Discussion with all speakers - moderated by Florian Pinault

Virtual

45m

Please select a timezone to show adjusted event times accordingly.