Scientists Use Deepfake AI Imagery to Understand Sun’s Atmosphere

AI News


This article has been reviewed in accordance with Science X’s editorial processes and policies. Editors emphasized the following attributes while ensuring content authenticity:

fact-checked

authoritative source

calibrate






An AI-generated deepfake image of the same corona rain with sharper resolution.Credit: Luke McMullan / NASA Solar Dynamics Observatory

Solar scientists are using deepfake AI imagery to unlock the mysteries of the sun’s atmosphere. The study, which will be presented at the National Astronomical Congress this week, was part of a collaboration between the Universities of Northumbria and Bern.

For more than 80 years, solar physicists have sought to understand how and why the upper layers of the Sun’s atmosphere (the corona) are unexpectedly hotter than those closer to the surface. Scientists have narrowed down the possible causes to two. Heating due to dissipation of waves in the plasma, or heating due to energetic recombination of magnetic field lines. There is evidence that both are happening, but the amount each process contributes to the overall heating is still unknown.

The key to solving this mystery seems to lie in a beautiful phenomenon known as “corona rain.” This is a loop of cold plasma that protrudes into the upper layers of the Sun’s atmosphere and dips into it. Identifying this rain is crucial to furthering our understanding of the underlying thermodynamics of the Sun. “Rain” appears to occur only by reconnecting magnetic field lines. If scientists could figure out how much corona rain falls on the sun, they could figure out how this unexpected heating cycle works.

To know how much it rains, we need to observe it separately from the myriad other solar materials. Most of the observations of solar rain are made by the Atmospheric Imaging Assembly (AIA) onboard NASA’s Solar Dynamics Observatory. However, the rain in these images is often obscured by hotter material. A surrogate image taken by NASA’s Solar Observatory, the Interfacial Area Imaging Spectrometer (IRIS), shows the rain more clearly, but has a limited field of view. Many AIAs with the sharp resolution of IRIS require Goldilocks image sets.



An original image of the coronal rain plasma projection taken by the Atmospheric Imaging Assembly (AIA) onboard NASA’s Solar Dynamics Observatory.Credit: Luke McMullan / NASA Solar Dynamics Observatory



An original image of the coronal rain loop taken by the Atmospheric Imaging Assembly (AIA) on NASA’s Solar Dynamics Observatory. Shows blurring due to hot material.Credit: Luke McMullan / NASA Solar Dynamics Observatory



A deep fake image of coronal rain generated by AI after learning a set of high resolution images from IRIS. Coronal rain loops are more clearly visible.Credit: Luke McMullan / NASA Solar Dynamics Observatory









To solve this, Northumbria University researcher Luke McMullan trains an AI machine learning algorithm to study high-resolution IRIS images, and then enhances richer, lower-quality AIA images to Created a “deepfake” that will allow astronomers to understand the extent of the corona. Rain falls into the Sun’s atmosphere, after which the mystery of its unusual thermal layers is revealed.

“We are living in a golden age of solar research,” said Luke McMullan, the project’s principal investigator. “Not only do we now have access to higher-resolution images of the solar atmosphere than ever before, but the rapid development and implementation of machine learning techniques in parallel with these observations has helped the community for decades. We hope that this collaboration of observation and machine learning will deepen and become a major tool in our scientific arsenal.”



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *