Workshop on “Machine Learning in Geophysical Modeling”

September 24–26, 2025
Speinshart Scientific Centre for AI and SuperTech (SSC)

Organizers: Tijana Janjic, Marcel Oliver, and Nadja Ray

This workshop brings together researchers in machine learning with applications in geophysical sciences and geophysical scientists using machine learning in their work.

Topics include:

Wednesday, September 24, 2025

Time Activity
10:00-12:30 Arrival and informal discussion
12:30 Lunch
13:30-14:30 Welcome and introduction of participants
14:30-15:30 Thomas Mortier (Ghent University): Valid prediction intervals for weather forecasting with conformal prediction
15:30 Coffee break
16:00-17:00 Maryam Ramezani Ziarani (KU Eichstätt-Ingolstadt): Parameter estimation and state prediction in a simplified shallow water model
17:00-18:30 Discussion: What is needed for successful application of machine learning in geosciences, and what machine learning scientists need for cooperation
19:00 Dinner

Thursday, September 25, 2025

Time Activity
until 8:30 Breakfast
9:00-10:30 Karoline Rummel (IOW Warnemünde): Real-time estimation of salt intrusion with LSTM networks
Jakub Truszkowski (Chalmers University): Inferring the geographical origin of timber from its chemical properties
Catherine George (KU Eichstätt-Ingolstadt): Uncertainty-aware forecasting in idealized atmospheric models
10:30 Coffee Break
11:00-12:15 Katharina Holube (Hamburg University): Energy transfers between normal modes of the primitive equations in the atmosphere: Towards their quantification using data-driven methods
Kyriakos-Stavros Malisovas (Hamburg University): Using machine learning to understand the behavior of a data assimilation system
Maxim Kirsebom (Bremen University): Ergodic theory and machine learning for ocean drifter data
Cas Decancq (Ghent University): Evaluation of deep learning weather emulators for forecasting extreme heat events
Sebastian Jimenez (Ghent University): Origin identification for commodities
Joshua Stiller (LMU): Conformal prediction for time-dependent PDEs
12:30 Lunch
14:00-15:30 Discussion Sessions
15:30 Coffee break
16:00-17:00 Tobias Heckmann (KU Eichstätt-Ingolstadt): Earth surface landforms, materials and processes: Potential ML applications in geomorphology
Toni Himmelstoss (KU Eichstätt-Ingolstadt): Application of graph-based deep learning to the generation and analysis of geomorphological maps
17:00-18:30 Short Hike
19:00 Dinner

Friday, September 26, 2025

Time Activity
until 8:30 Breakfast
8:30-9:00 Guided tour of the Speinshart Monastery
9:00-10:30 Sergiy Vasylkevych (Hamburg University): Growth of forecast errors in AIFS from the normal modes analysis perspective
Sarah Eberle-Blick (KU Eichstätt-Ingolstadt): Neural networks and the monotonicity method for solving inverse problems: a first approach
Tatsiana Bardachova (KU Eichstätt-Ingolstadt): Towards the assimilation of dual-polarization radar data
Anton Kutsenko (Hamburg University): Using ML for the recovery of energetic spectra from incomplete data, plus possible ocean dynamic predictions
Hao Liang (KU Eichstätt-Ingolstadt): Recovering energy spectra from sparse observations
10:30 Coffee Break
11:00-11:20 Nadja Ray (KU Eichstätt-Ingolstadt): Flow and transport in porous media - accelerating simulations with CNNs
11:20-12:15 Concluding discussion
12:30 Lunch
13:30 Departure