AI for Earth and Sustainability Science | AI for Good Discovery - Forecasting and understanding bird migration with process-guided deep learning

The seasonal migrations of billions of birds are among the most spectacular natural phenomena on Earth, but they are increasingly threatened by the impacts of human activities, such as light pollution, climate change, and collisions with wind turbines and aircraft. To effectively protect migration systems, we need models that can accurately forecast large-scale movements and help us better understand how birds respond to environmental conditions. This talk explores the benefits and challenges of process-guided deep learning, combining ecological principles with neural networks to model bird migration across continents based on weather radar data. In the first part, we present a migration forecast model that augments a Eulerian movement model with flexible neural network components to predict bird fluxes across time and space. In the second part, we showcase how this hybrid forecast model, when interpreted carefully, can provide novel insights into the decision-making of migrating birds across diverse environments. The talk concludes with a broader perspective on the potential of process-guided deep learning in ecology and Earth science, highlighting applications where domain knowledge is partial yet essential for building models that are accurate, robust, and interpretable. Learning Objectives: By the end of the session, participants will be able to: Describe the major threats to bird migration and the importance of large-scale migration modeling. Explain the principles of process-guided deep learning and how ecological knowledge can be integrated into neural network models. Interpret outputs from a hybrid Eulerian-neural network model to assess bird flux patterns across continents. Critically evaluate the benefits and limitations of combining ecological theory with AI for robust and interpretable ecological forecasting. Propose potential applications of process-guided deep learning in other ecological or Earth system contexts where domain knowledge is partial. Speakers: Fiona Lippert Researcher, Space Research Organisation Netherlands (SRON) Moderators: Alexander Brenning Professor, Friedrich Schiller University Jena AI for Good is identifying innovative AI applications, building skills and standards, and advancing partnerships to solve global challenges. AI for Good is organized by ITU in partnership with over 50 UN partners and co-convened with the Government of Switzerland. Register now for the AI for Good Global Summit 2026! Go free or VIP. https://aiforgood.itu.int/summit26/ Join the Neural Network! 👉https://aiforgood.itu.int/neural-network/ The AI for Good networking community platform powered by AI. Designed to help users build connections with innovators and experts, link innovative ideas with social impact opportunities, and bring the community together to solve global challenges using AI. 🔴 Watch the latest #AIforGood videos! https://www.youtube.com/c/AIforGood/videos 📩 Stay updated and join our weekly AI for Good newsletter: http://eepurl.com/gI2kJ5 🗞Check out the latest AI for Good news: https://aiforgood.itu.int/newsroom/ 📱Explore the AI for Good blog: https://aiforgood.itu.int/ai-for-good-blog/ 🌎 Connect on our social media: Website: https://aiforgood.itu.int/ X: https://twitter.com/AIforGood LinkedIn Page: https://www.linkedin.com/company/26511907 LinkedIn Group: https://www.linkedin.com/groups/8567748 Instagram: https://www.instagram.com/aiforgood Facebook: https://www.facebook.com/AIforGood Disclaimer: The views and opinions expressed are those of the panelists and do not reflect the official policy of the ITU.