Spatio-temporal Variability of the Middle and Upper Atmosphere: Characterization of Multi-Source Data, Aggregation, and Use Cases for Artificial Intelligence

The MATRIA project (Real-Time Middle and Upper Atmosphere Using Artificial Intelligence) aims to
develop and evaluate proof-of-concepts in artificial intelligence for the real-time
forecasting of atmospheric fields (winds, temperature, density) on a global scale across the entire
0–150 km column. Led by an IRD-CEA-ONERA consortium, it is part of the
dual-use (civil/defense) exploratory research framework of the ASTRID call (Specific Support for
Research of Defense Interest), addressing strategic challenges related to understanding
the atmospheric environment for forecasting, infrasonic monitoring, and decision support. However, current approaches to monitoring consist of fusing real-time data up to the stratosphere with climatological data (seasonal trends) in the upper layers. To meet operational surveillance requirements, a tool must be developed to forecast atmospheric conditions in real time up to the lower thermosphere (~150 km). Artificial intelligence could help address this need.