The Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences is the national centre for research into the geosphere. With around 1,650 employees, we develop a sound understanding of systems and processes of the solid Earth as well as strategies and options for action to counter global change, understand natural hazards and sustainably secure energy and raw material requirements.

Time series, such as seismic data or paleoclimate proxy data (such as stable istopes) from lake deposits, are one of the most common data structures in geoscience. Detecting causal associations and understanding mechanisms based on them is a key challenge for providing insights and predicting behaviors in Earth system. We seek a candidate to develop (1) a causal and equation discovery framework for various time series and spatiotemporal datasets; and (2) a statistical learning model for detecting and predicting phase transition and tipping points in Earth system components and critical thresholds for hazard early warning system. This project will mainly use seismic data for hazard early warning of mass movement events and isotope time series for paleoclimate reconstruction from various sites in Europe and worldwide for identifying regime transitions.

The PhD position will be hosted in the Earth Surface Process Modelling section at the GFZ and supervised by Dr. Hui Tang. In the section, numerical models are developed and used to investigate a wide range of physical, chemical and biological processes and interactions occurring at the Earth’s surface that are driven by tectonic processes and modulated by climate. The successful candidate will work with a team of partners from the Geomorphology section at the GFZ (Prof. Dr. Dirk Sachse) and from the Complex Systems department at the Potsdam Institute for Climate Impact Research (PD Dr. Norbert Marwan). 

Your responsibilities:

• Process and analyse continuous time series data such as seismic data and isotope data
• Develop and improve causal discovery framework for continuous time series
• Develop and improve equation discovery framework (e.g., SINDy) for continuous time series
• Develop a machine learning approach for determining tipping points in Earth system and critical thresholds for early warning system based on causal and equation discovery framework, especially for mass movement
• Build a statistical framework for ensemble simulations deriving thresholds for future climate
• Interact with other members of the Earth Surface Process Modeling Section at weekly group meetings and in other informal ways
• Present results at scientific meetings and in publications in peer-reviewed journals
• Supervise master’s students

Your qualifications:

• Master degree or diploma in Mathematics, Data Science, Machine Learning or Geosciences 
• Programming skills in Python, C, or similar language are necessary
• Familiarity with machine learning and data science is strongly desirable
• Knowledge in seismology or geochemistry is desirable
• Proficiency in spoken and written English