Without polarization measurements, two researchers reconstruct vector magnetograms close to those of Hinode, and the model is transposed to other instruments.
The vector magnetic field on the surface of the Sun is the basic data for understanding and predicting solar activity, but obtaining it is expensive: the polarization of light in several bands must be measured and then reversed these measurements. Ruoyu Wang and David Fouhey propose a shortcut with MAGiDiff, a method based on denoising scattering models that estimates these magnetograms from ultraviolet and extreme ultraviolet images.
The model takes as input a stack of images from the AIA instrument at the Solar Dynamics Observatory and learns to produce the vector field as measured by Hinode's spectropolarimeter. The problem is by nature ambiguous, since UV intensity only indirectly provides information about magnetism; the generative approach makes it possible to sample plausible solutions rather than a single answer.
Robust from one solar cycle to the next
According to the authors, MAGiDiff faithfully reproduces Hinode's ground truth, generates plausible active regions over the entire disk, remains valid despite the polarity reversal of the hemispheres between two cycles and can be refined for STEREO/EUVI or GOES-R/SUVI. They specify that it does not replace a dedicated instrument, but opens the way to estimates where no polarization measurements exist.
Read the full article on the original source
Go to the source →