IBM and NASA develop digital twins of the sun to predict future solar storms

The most sun complex mystery could soon be resolved thanks artificial intelligence. 20. August, IBM and NASA announced the launch of Suria, a Foundation model For the sun. Having Been Trained on Large Datasets Of Solar Activity, This Ai Tool Aims To Deepen Humanity’s Understanding of Solar Weather and Accurately Predict Solar Flares-Bursts Of Electromagnetic Radiation Emitted by Our Star That Threaten Both Astronauts In Orbit And Communication infrastructure on Earth.

Suria is trained with nine years of data collected by NASA Osovoritorator Solar Dynamics (SDO), an instrument that has been eagle from 2010. years, taking high resolution images every 12 seconds. SDO takes the observations of the sun on various various electromagnetic wavelengths to estimate the temperature of the stars layers. Precise measurements of the sun’s magnetic field is also required – essential data for understanding how energy moves through the star and for predicting solar storms.

The historical, interpretation of this huge amount of diverse and complex data was a challenge to heliophysics. To deal with this challenge, IBM says If Suria developers used SDO data to create digital twins of the sun-dynamic virtual star replica that is updated when new data is taken and which can be manipulated and easily studied.

The process began to combine different data formats that are inserted into the model, allowing him to process them consistently. The transformer of the vision of long range and architecture that allows for a detailed analysis of very high resolutions and identifying relations between their components, regardless of their distance.

The performance of the model is optimized using a mechanism called spectrum gira, which reduces the use of memory up to 5 percent filtering noise in data, increasing the quality of processed information.

More precisely predictions in less time

Its developers say that this design gives Suria a significant advantage: unlike other algorithms that require a large labeling of data that they fed, Suria can directly learn directly from data. This allows you to quickly adapt to different tasks and delivers reliable results in less time.

During testing, Suria showed its versatility in integrating data from other instruments, such as Probe Parker Solar Both solar and heliospheric observatory (SOHO), two other spacecraft that watch the sun. Suria also showed that it is effective in various predictive functions, including forecasting work activities and solar wind speed.

According to IBM, traditional prediction models can only predict balance only one hour before signals detected in certain regions of the sun. In contrast, “Suria provided a two-hour guidance using visual information. In early modeling, the team said that 16 percent improving the classification of solar palette, marked improvements based on existing improvements in a statement.

NASA emphasizes that, although the model is designed to study heliophysics, its architecture is adaptable to different areas, from planetary science to the observation of the country. “By developing the Foundation Trained in NASA Heliophysical data, we facilitate the analysis of solar behavior with unprecedented speed and precision,” Kevin Murphy, director of Data Science Kevin Murphy, statement. “This model allows the wider understanding of the way solar activity affects critical systems and technologies that we all rely here on earth.”

The risk that sets an abnormal solar activity is not smaller. The main solar storm could directly affect global telecommunications, collapse of electricity networks and harass GPS navigation, satellite operations, internet connections and radio transmits.

Andres Munoz-Jaramillo, a solar physicist in the southwest research institute in San Antonio, Texas and a leading scientist on the project, emphasized that the goal of Suria is to maximize guidance time for these possible scenarios. “We want to give the country the longest possible time of guidance. The hope is that the model has learned all critical processes behind the evolution of our stars through time so we can extract activating insight.”

This story originally appeared Wired Spanish and has been translated from Spanish.