PhD Position in Machine Learning for Satellite Gravimetry

The Chair of Space Geodesy invites applications for an exciting PhD opportunity focused on advancing research in satellite gravimetry for hydrology using machine learning. The GRACE mission has transformed the way we can monitor surface mass changes in the Earth system. By applying machine learning methods to terrestrial water storage anomalies derived from GRACE data, we aim to improve their spatial resolution and improve predictions of flooding events.

The PhD position is embedded in the SNSF/DFG project HiGrav: High-resolution gravity fields for better flood forecasting. The team behind HiGrav includes researchers from ETH Zurich (lead), University of Bern, GFZ Helmholtz Centre for Geosciences and TU Braunschweig.

The project aims to improve flood forecasts using satellite data from the GRACE mission. First, daily solutions are calculated to enhance the temporal resolution of gravity field solutions from GRACE. The spatial resolution is improved through machine learning and a combination with hydrological simulations. From this data, a high-resolution moisture index is generated, serving as an early warning indicator for flood-prone catchments. Finally, this approach is integrated into existing flood forecasting models and tested for various catchments, including regions in Lower Saxony.

HiGrav has significant societal benefits, as it enhances the accuracy of flood warning systems, helping to reduce damage and loss of life caused by flooding. Additionally, the improved GRACE data aids in better monitoring of droughts and groundwater conditions, supporting sustainable decision-making in these areas.

The selected PhD candidate will join an innovative research team to explore cutting-edge techniques in satellite gravimetry.
The primary objective is to develop and apply novel strategies based on AI/machine learning (ML) for downscaling GRACE data and for predicting floods. Specific tasks include:

• Designing and training ML models for downscaling daily GRACE data using high-resolution hydrological simulations
• Uncertainty quantification of downscaled GRACE data with probabilistic machine learning
• Prediction of GRACE gravity field solutions based on AI/ML
• Flood warning based on GRACE data and AI/ML

In addition to research responsibilities, the candidate will contribute to teaching activities and the supervision of student projects.

Required Qualifications:

• A Master's degree (MSc) in geodesy, hydrology, geosciences, geophysics, computer science, physics, or a closely related discipline
• A strong foundation in mathematics, physics, data analysis, and programming
• Excellent communication and collaboration skills, with fluency in English (spoken and written)

Preferred Qualifications (Not Mandatory):

• Background in satellite gravimetry, with experience in GRACE data or products
• Experience in hydrology, with a focus on floods
• Knowledge of machine learning methods, frameworks, and applications
• A demonstrated ability and willingness to learn new technologies and adapt to evolving research challenges

• A full-time 3-year PhD position at ETH with conditions and benefits defined here
• Research at a world-leading academic institution as part of an innovative project
• A dynamic and international team that embraces fresh perspectives
• A supportive and inclusive work environment that fosters professional growth and development
• Opportunities to attend conferences and collaborate with international experts
• Flexible working hours with the option for part-time home office
• An attractive workplace surrounded by nature with quick access to the city center

The position is set to begin on September 1, 2025.