| Presentation Title: | Advancing solar magnetic field extrapolations with physics-informed neural networks |
| Author(s): | R. Jarolim, B. Tremblay, J. K. Thalmann, M. Rempel, A. M. Veronig, M. Molnar, T. Podladchikova
|
Abstract:
While the photospheric magnetic field of our Sun is routinely measured, its extent into the upper atmosphere is typically not accessible by direct observations.
In this presentation, we introduce Physics-Informed Neural Networks (PINNs) for solar magnetic field extrapolations. Our method integrates observational data and the physical force-free magnetic field model to estimate the 3D topology of coronal magnetic fields based on photospheric vector magnetograms.
We demonstrate that our method can provide realistic coronal magnetic field estimates in quasi real-time. The systematic analysis of the time evolution of free magnetic energy and magnetic helicity in the coronal volume, as well as comparison to EUV observations demonstrates the unambiguous relation to the observed flare activity.
We further outline our recent advances where we apply this approach to magnetograms from Hinode/SP, perform multi-height extrapolations with chromospheric magnetograms from the Swedish Solar Telescope, and obtain global force-free extrapolations from SDO/HMI full-disk observations.
PINNs enable realistic magnetic field extrapolations in quasi real-time and have the potential to greatly advance the field of data-driven simulations, accelerate scientific research, and improve our understanding of the solar magnetic field.