AI advances unlock 3D cloud mapping from satellite data

Written by Copernical Team Tuesday, 17 December 2024 06:48

font size decrease font size increase font size
Print
Email

Write a comment

Paris, France (SPX) Dec 13, 2024

ESA's EarthCARE satellite, launched in May 2024, is approaching the end of its commissioning phase, with its first data release on clouds and aerosols anticipated early next year. Meanwhile, an international team of scientists has leveraged artificial intelligence to transform existing satellite data into groundbreaking 3D cloud profiles. Clouds are central to Earth's climate system, influ

AI advances unlock 3D cloud mapping from satellite data
by Erica Marchand
Paris, France (SPX) Dec 13, 2024

ESA's EarthCARE satellite, launched in May 2024, is approaching the end of its commissioning phase, with its first data release on clouds and aerosols anticipated early next year. Meanwhile, an international team of scientists has leveraged artificial intelligence to transform existing satellite data into groundbreaking 3D cloud profiles.

Clouds are central to Earth's climate system, influencing both cooling and warming processes. High, thin clouds tend to trap heat radiating from Earth's surface, contributing to warming, while low, thick clouds reflect sunlight, aiding cooling. Despite their critical role, the precise impact of clouds on Earth's energy balance remains uncertain, particularly in the context of climate change.

Understanding potential changes in cloud behavior and their future impact on the climate is urgent. Real-time, global 3D cloud data is seen as a key tool for addressing these uncertainties, improving climate forecasts, and guiding decision-making.

Historically, NASA's CloudSat mission has provided valuable vertical cloud profiles but lacked frequent revisits. In contrast, geostationary satellites like Europe's Meteosat Second Generation (MSG) offer frequent observations but are limited to a two-dimensional, top-down perspective.

To overcome these limitations, ESA F-lab and FDL Europe coordinated a study utilizing advanced machine learning techniques. By combining archived CloudSat profiles with MSG imagery from 2010, the team successfully created a method to generate continuous 3D cloud profiles globally.

Anna Jungbluth of ESA's Climate and Long-Term Action Division explained, "We carefully aligned the measured CloudSat profiles with images from MSG. This helped us understand how the 'view from top' and the corresponding cloud profiles were related. We then trained machine learning models to understand this mapping and derive cloud profiles from the 2D imagery. This allowed us to extend the CloudSat profiles in both space and time."

The results were presented at the Neural Information Processing Systems conference in Canada, demonstrating how artificial intelligence can draw new insights from historical satellite observations. Animated visualizations showcased the AI model's ability to predict vertical cloud profiles across large areas, even without direct CloudSat tracks, enabling the creation of 3D cloud maps over time.

Michael Eisinger from ESA's Climate and Long-Term Action Division and the EarthCARE project team highlighted the implications for the future: "EarthCARE has already given us some very promising preliminary data. Our work generating these 3D cloud profiles lays the foundation for exploiting EarthCARE from a different angle. These new AI methods promise to maximize EarthCARE's scientific potential and integrate its data into comprehensive global models that will push the boundaries of climate science."

As EarthCARE continues to generate data, researchers aim to refine and expand the use of these AI-driven methods, unlocking new insights into cloud dynamics and their role in Earth's climate.