On a mountaintop better known for sunrise selfies, UH Mānoa scientists are quietly teaching a computer to unravel the Sun. Their new artificial intelligence tool peels back the star’s tangled magnetic field in three dimensions, promising sharper predictions for space weather that can scramble satellites, power grids and communications on Earth.
The algorithm, nicknamed the Haleakalā Disambiguation Decoder, figures out whether magnetic fields are pointing toward or away from Earth and estimates the height of overlapping solar layers. That lets researchers tease out the vector electric currents they have long struggled to measure. Tied directly to high resolution data from the Daniel K. Inouye Solar Telescope on Haleakalā, the work is a rare homegrown leap in a field with global consequences, turning island telescope pixels into practical forecasting tools.
UH team publishes a physics informed AI for solar magnetism
According to University of Hawaiʻi at Mānoa, researchers at the Institute for Astronomy, led by postdoctoral researcher Kai Yang, built a physics informed machine learning system that blends observed data with one basic rule of magnetism: solar magnetic fields close into loops. Using that rule, the system resolves the 180 degree azimuthal ambiguity that plagues solar measurements and estimates the height of different solar layers at the same time. The team’s findings were published in The Astrophysical Journal.
How the Haleakalā Disambiguation Decoder works
The method enforces the divergence free condition for magnetic fields in three dimensions while it maps optical depth to geometric height. In tests on magnetohydrodynamic simulations, that combination produces self consistent 3D vector magnetic fields and electric current estimates. As outlined in the project’s materials and the arXiv preprint, the team trained and validated their neural networks on large synthetic datasets designed to mimic DKIST spectropolarimetric measurements, then showed the method can recover field directions in quiet Sun, plage and sunspot regions.
Why it matters for DKIST and space weather
The Daniel K. Inouye Solar Telescope is set to deliver the high resolution spectropolarimetric data this algorithm was built to digest. That pairing should enable faster and more detailed 3D reconstructions of magnetic structures that can trigger solar flares and coronal mass ejections, turning space weather forecasts into less of a cosmic guessing game. The National Solar Observatory says DKIST’s mission is to observe magnetic fields on the Sun and improve space weather forecasting, an outcome that could help utilities, satellite operators and communications providers prepare for disruptive events before they hit.
What is next for the team
Tests on detailed computer models show the tool reliably recovers horizontal field orientation and estimates current density, and the researchers say they expect similar performance when real DKIST data start flowing, nudging forecasts toward earlier warnings, according to University of Hawaiʻi at Mānoa. The news release also links to a short explainer video that walks through the technique and its implications: University of Hawaiʻi at Mānoa.
Beyond the technical upgrade, the project underlines Hawaiʻi’s role in frontline solar research. Telescopes on Haleakalā and Institute for Astronomy teams in Honolulu are building tools that plug into national observatories and forecasting systems. The SPIn4D project materials and preprints are being shared with the broader solar community so DKIST users around the world can generate faster, more accurate 3D maps of the Sun.
