The Sun constantly sends out a stream of charged particles known as the solar wind, which can affect satellites, astronauts, and systems on Earth. To better predict and understand this solar wind, we developed an artificial intelligence model that learns from over a decade of solar images collected by NASA’s Solar Dynamics Observatory (SDO).

Our model uses image sequences from the Sun—including ultraviolet light and magnetic field data—and combines them with the position of NASA’s Parker Solar Probe (PSP), which travels through the solar wind and collects direct measurements. By combining this information, the model builds a 3D representation of how solar wind flows from the Sun into space.

We also use expert-labeled data to help the model recognize different types of solar wind. The goal is to train the model to predict important solar wind properties like speed, density, and magnetic strength, which are critical for space weather forecasting.

This work shows how AI can help turn huge amounts of solar data into useful tools for understanding and forecasting the behavior of the Sun. It was developed as part of the Heliolab FDL 2025 research program.