IBM and NASA Launch AI Model to Predict Solar Flares

IBM and NASA Launch AI Model to Predict Solar Flares

Surya, an AI model from IBM and NASA, predicts space weather that can disrupt satellites, electricity, and internet.

IBM and NASA have jointly developed an AI model that predicts not Earth’s weather, but space weather. The model, named Surya (Sanskrit for “sun”), is designed to detect solar flares and other intense solar eruptions before they impact our technology.

Protection against Solar Storms

Such “eruptions send charged particles toward Earth and can disrupt satellites, power grids, and even the internet. This also happened in 1859, according to The Register. The eruption was so powerful that telegraph systems caught fire. Today, such a” storm would have more far-reaching consequences because our modern infrastructure is more vulnerable.

Surya was trained on nine years of solar images from NASA’s SDO satellite. The model learns how the sun behaves and can predict where and when a solar flare will occur up to two hours in advance and in high resolution.

Open Source and more Accurate than Ever

According to IBM, Surya increases the accuracy of solar flare detection by 16 percent. It’s also the first AI model that can visualize this prediction. The technology is available open source via Hugging Face, GitHub, and IBM’s TerraTorch. The compiled datasets and benchmarks (SuryaBench) are also freely accessible. This way, the researchers aim to make developing applications for space weather forecasting easier.

read also

IBM and NASA Launch AI Model to Predict Solar Flares