A new artificial intelligence model developed by Microsoft, under the leadership of Greek professor Paris Perdikaris from the University of Pennsylvania, can successfully predict various aspects of Earth’s behavior, from weather patterns to air pollution and ocean waves. The model, named Aurora, is featured in a publication in the journal Nature.

Predicting Earth’s systems is an essential tool for providing timely warnings about extreme events. Traditional forecasts rely on complex models based on decades of data and often require supercomputers and specialized teams to maintain, making them inaccessible to many communities around the world.

Aurora is an AI model trained on over a million hours of geophysical data. Its creation had two main goals: to build a forecasting tool that is both more accurate and computationally much more efficient. By adopting a fundamentally different approach from traditional weather forecasting models, Aurora learns patterns directly from data, identifying complex relationships in historical Earth system data and using them to make predictions.

As noted in the Nature publication, Aurora outperforms existing models in air quality, ocean wave behavior, tropical cyclone tracking, and high-resolution weather forecasting, all while operating at a lower computational cost than current forecasting methods. According to the data, for air quality predictions, Aurora matched or exceeded the performance of the Copernicus Atmosphere Monitoring Service in 74% of targets and was approximately 50,000 times faster. Furthermore, for high-resolution weather conditions, the model outperformed the leading numerical weather model IFS HRES in 92% of targets at 0.1° resolution, showing better performance in extreme events.

“Aurora represents a significant innovation in environmental systems prediction, as it is the first AI model functioning as a unified foundational model capable of adapting to different applications—from high-resolution weather forecasts and air quality predictions to tropical cyclone and ocean wave monitoring. This approach achieves high accuracy at a computational cost thousands of times lower, making advanced environmental forecasting accessible to wider communities globally,” Professor Perdikaris explained to the Athens-Macedonian News Agency.

A key innovation of the model is its ability to be trained on a vast volume of diverse geophysical data and then optimized for specific forecasting tasks—like a powerful brain that can be specialized for different prediction tasks.

As Professor Perdikaris noted, “The Aurora project, during my time at Microsoft Research, was part of my broader research vision to create foundational models for scientific applications that can generalize across various fields and accelerate discoveries.” He added that at the University of Pennsylvania, “my team is extending this vision beyond Earth sciences to various science and engineering applications, creating AI systems that not only predict but also help us understand complex natural phenomena across multiple disciplines.” Similar modeling approaches are being applied by his team in other scientific fields, from materials engineering to biomedical applications.