Machine Learning in Earth Science

Climate change, urban sprawl, and unsustainable growth has put many societies in a difficult relationship with the natural world. For many regions in the world one of the most important tasks in the next years will be the damages and casualties of natural disasters. Floods, for example, are among the most frequent and destructive natural hazards in the world that claim many lives and create large economic damages. Modern data-driven approaches provide a unique opportunity to reduce damages and causalities by making it possible to build models that help us with predicting and understanding the variability of natural systems. 

The goal of this research focus is to explore, examine, develope, and evaluate machine learning based approaches that enable us to describe the flow of water on Earth. Specifically, we will focus on advancing data-driven modelling of extremes — such as floods and droughts. These models will, for example, make it possible to provide early flood warnings, predict future droughts, operate hydropower plants more efficiently, or improve access to water.

Members of this research focus will develop strong technical skills with many downstream applications. They will learn to interface across different scientific fields and work in interdisciplinary teams with international partners. We especially invite applicants with a geoscientific background that are interested in machine learning or applicants with a technical background close to machine learning interested in environmental issue. Most importantly, however, we encourage people that are interested in contributing with their curiosity and motivation.