NASA and IBM Launch AI Model 'Surya' to Predict Space Weather

NASA and IBM have launched an advanced open-source AI model called ‘Surya', designed to understand solar flares, predict space weather, and protect critical technological infrastructure on Earth. The model has been trained on nine years of high-resolution solar data obtained from NASA’s Solar Dynamics Observatory (SDO). This AI model offers a 16% improvement in accuracy in predicting solar flares and can predict them up to two hours in advance.

Protecting Against Solar Storms

‘Surya’ is the first foundational AI model for solar physics, helping to protect satellites, GPS navigation, power grids, and telecommunication systems from damage caused by solar storms.

Key Features of the ‘Surya’ Model

High-Resolution Data

‘Surya’ was trained on data from NASA’s SDO, which provides high-resolution images and magnetic field measurements of the sun every 12 seconds. This includes ultraviolet and extreme-ultraviolet images of the solar system and maps of the solar surface’s motion and magnetic field.

Space Weather Prediction

This AI model predicts solar flares, solar wind speed, and the emergence of active regions on the solar surface. It helps predict space weather events, such as solar flares and coronal mass ejections, which can impact satellites, power grids, and communication systems.

Improved Accuracy

Preliminary results show that ‘Surya’ is 16% more accurate in solar flare classification and provides visible predictions two hours in advance for the first time.

Open-Source Availability

‘Surya’ has been made available open-source on Hugging Face, allowing global researchers and developers to use it for solar weather research and other applications. It will also assist researchers in benchmarking for solar weather prediction.