Solar storms are increasingly threatening for the earth. NASA wants to prepare with a “digital twin” of the sun

The sun, that star that gives us life, also has an unpredictable character and potentially destructive. So much so that A large solar storm It could return to us technologically to the stone age in the blink of an eye. To avoid this, NASA and IBM They have joined forces To create a model that simulates the behavior of the sun. And how could it be otherwise, it has been generated with artificial intelligence.

A authentic twin of the sun to understand it better. The union of forces of these two institutions has resulted in the creation of Surya, an artificial intelligence Designed to act as a digital twin of the sun and anticipate their violent outbursts with precision that with the models that are currently used cannot be achieved.

Training an AI with the heart of a star. The challenge of predicting the space climate is undoubtedly a great challenge For scientists. To build Surya, the engineers turned to an inexhaustible data source: the Solar Dynamics Observatory (SDO) NASA Hala. For nine years, this probe has State watching the sun without restcapturing images of very high resolution every 12 seconds in different wavelengths and measuring its complex magnetic field.

Once you have all this information is where artificial intelligence comes into action to be able to organize and interpret them for experts. That is why the first thing Surya did is standardize all data formats to be able to process them together.

Intelligent filtering of all this information. Once the data was unified, the next step was to use a long -range vision transformer, an architecture capable of analyzing gigantic images to identify patterns and relationships between points of solar activity, regardless of how far they are from each other.

But he did not stay here, because thanks to a mechanism known as ‘spectral door’, the system was able to filter the ‘noise’ of the data to reduce memory use and improve the quality of the information with which it worked. Therefore, researchers were removed a lot of work to have to label all the images, causing it to adapt rapidly.

More precision and twice as much time to react. The results of the initial tests are very promising. Until now, Traditional models barely gave us an hour in advance before a solar eruption. With Surya it has been shown to be able to launch a reliable warning two hours in advance, doubleing the humanity preparation window.

But it is not only faster, but also accurate. The IBM and NASA team recorded a 16% improvement in precision when classifying solar rashes compared to the models used right now. Something that is also thanks to the ability to integrate information from other missions such as Parker solar probe or the Soho Observatory.

An open tool for the science of the future. Far from saving this powerful tool in a key drawer, IBM and NASA have made it available to the entire scientific community. Surya is now available on platforms such as Hugging Face, GITHUB or even the Terratorch library.

Kevin Murphy, NASA scientific data director, is clear: “We facilitate the analysis of the complexity of our star’s behavior with unprecedented speed and precision. This opens the door to a better understanding of the impact of solar activity on the systems on which our daily life depends.”

The goal is for the Earth to be prepared. Although we see the central star of our system as harmless, the reality is that at any time this sensation can change. In this way, preparation and anticipation is fundamental and for the moment all hopes are put in this model of where it is possible to learn from the processes behind the evolution of the sun with the aim of having a greater amount of information.

Images | Javier Miranda

In Xataka | How the Solar System was formed: So that the Earth was born, a star had to die