Analyzing globally distributed EO data to quantify and predict the impact of climate change on snow cover and snowmelt dynamics

• Applicant: Professorin Dr. Claudia Künzer
• Subject Area: Physical Geography ; Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing ; Methods in Artificial Intelligence and Machine Learning
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 522760169

Terrestrial snow cover is a large-scale indicator of climate change influencing Earth system interactions on a global scale. Mountain and polar regions play a crucial role in the context of climate change as they serve as key indicators of environmental shifts, contribute significantly to global weather patterns, and house vital ecosystems whose resilience is essential for our planet. Large-scale studies based on low and medium resolution earth observation (EO) data indicate that climate change is strongly impacting the snow regime in cold and polar regions, especially as a response of delayed snow onset and earlier snowmelt mainly controlled by temperature. But to monitor snow cover dynamics in regions with complex terrain (e.g. mountains) the World Meteorological Organization defined the requirement of a daily 100 m resolution snow cover data product not yet available. Furthermore, high cloud cover and polar night cause gaps in the snow cover time series not allowing a daily data product. In this project (embedded in the SOS research unit), we aim to exploit high-volume EO data archives from multi-sensor remote sensing data and fuse them based on novel AI developments in computer vision to overcome the limitations of temporal and spatial resolution. The exploitation of multi-modal EO data (EO data in different formats, from different sensors and imaging techniques) with AI-based methods requires in-depth domain knowledge as well as extensive skills in data science and IT systems, creating high entry barriers for geoscience researchers. Therefore, we will develop concepts and building blocks (i.e., model-driven algorithms, platform components, and architectures) to access multi-modal EO data, extract land cover dynamics from multi-modal EO data by training an AI-based land cover classification model, and provide tools for trend detection and forecasting land cover dynamics in a changing climate. This increases productivity and supports interdisciplinary research by providing the foundations for building specialized platforms offering semantically enriched domain-specific components for composing complex workflows. For this project in particular, the components will be combined to a domain-specific processing pipeline to infer high-resolution snow parameters from multi-modal EO data in order to enable more accurate conclusions about climate change impacts on snow and snowmelt dynamics in cold and polar regions.

• DFG Programme: Research Units
• Subproject of: FOR 5696: SOS: Serverless Scientific Computing and Engineering for Earth Observation and Sustainability Research
• Co-Investigator: Dr. Celia Baumhoer