20 décembre 2024
AI-Powered Satellite Data: Mapping Clouds in 3D for Climate Research
Innovation in Earth Observation through Artificial Intelligence
In May 2024, ESA's EarthCARE satellite started its mission to improve understanding of clouds and aerosols in the Earth's climate system. While its first data sets are expected in early 2025, scientists – among them, from ESA's Climate and Long-Term Action Division, our Research Fellow Anna Jungbluth – have pioneered a novel use of AI to generate wall-to-wall 3D cloud profiles from existing, more limited satellite data. This breakthrough bridges the gap between traditional satellite observations and the real-time, global insights needed to address climate uncertainties.
Scope and Focus of the Study
This innovative work, carried out by an international team coordinated by ESA's Φ-lab and Frontier Development Lab (FDL) Europe, combines data from NASA's CloudSat and Europe's Meteosat Second Generation (MSG) missions. The research focuses on applying machine learning techniques to merge CloudSat's vertical cloud profiles with MSG's frequent top-down 2D images to create 3D cloud maps in space and time. This approach opens up new possibilities for climate modelling and decision-making, particularly in the context of urban heat islands and changes in the Earth's energy balance.
Key Achievements of the Study Include
- Matching vertical cloud profiles with MSG imagery to develop AI models.
- Training machine learning algorithms to extend the spatial and temporal coverage of cloud profiles.
- Creating a proof-of-concept 3D cloud mapping system using archived satellite data.
Scientific Highlights
Clouds play a critical role in the Earth's climate system – reflecting sunlight and trapping heat. Uncertainty about their net effect on the planet's energy balance is a critical challenge in climate science. Real-time 3D cloud data could reduce these uncertainties, improving climate predictions and informing policy decisions.
Applications for Future Missions
The integration of advanced AI techniques into Earth Observation not only enhances the utility of historical data, but also lays the groundwork for future missions such as EarthCARE. By generating global 3D cloud profiles, scientists can refine their understanding of the Earth's radiation budget and the impact of cloud dynamics on climate change.
This research, supported by public-private partnerships such as FDL Europe Earth Systems Lab, demonstrates the power of collaboration between ESA, academia and leading AI innovators such as Google Cloud, Scan AI and NVIDIA.