At the 10th International Summer School on AI and Big Data, Dr. Josefine Umlauft will give a presentation in the field of Earth and Environmental Sciences and AI.

Talk: Leveraging Earth System Data Cubes for Enhanced Environmental Insights through Machine Learning

The integration of Earth System Data Cubes (ESDCs) with machine learning (ML) techniques represents a promising approach for unlocking valuable insights into complex environmental processes. ESDCs, comprising multidimensional arrays of geodata organized coherently in space and time, offer a rich source of information about the Earth system. Meanwhile, ML algorithms have demonstrated remarkable capabilities in extracting patterns and making predictions from such large and diverse datasets. Utilizing various types of data, such as satellite imagery or seismic records, and use cases, we explore the synergy between ESDCs and ML, highlighting their combined potential to address pressing environmental challenges.

We review recent advancements in the application of ML techniques to analyze ESDCs, encompassing tasks such as gap filling, prediction, clustering, and monitoring of spatio-temporal environmental dynamics. Additionally, we discuss methodological considerations and challenges associated with integrating ML into ESDC workflows, including data preprocessing, model selection, big data handling, and interpretability. Furthermore, we present case studies illustrating successful applications of ESDCs and ML, showcasing their ability to enhance understanding of Earth system dynamics.

Bio

Dr. Josefine Umlauft is the head of the Earth and Environmental Sciences group at ScaDS.AI (Center for Scalable Data Analytics and Artificial Intelligence) in Leipzig. She completed her BSc., MSc. and PhD at Leipzig University, while she spend several months abroad for individual research stays at Tbilisi State University (Georgia), Institut des Sciences de la Terre (Grenoble, France) and Los Alamos National Laboratory (NM, USA).

She is a trained geophysicist specializing in Environmental Seismology and Machine Learning. Her main research interests are on methodological developments for an improved understanding of near-surface environmental processes on the exploration scale. She thereby focuses on natural hazards and mass movements, especially in the cryosphere and mountainous regions, but she also investigates oscillations generated by trees to study their vitality or animals to study their social behavior. Based on highly resolved time series data, Dr. Josefine Umlauft combines signal processing, array techniques), synthetic modelling and machine learning to conduct her research.