Speaker
Description
We develop two artificial intelligence-based models (Models A and B) to predict time-evolving photospheric magnetic fields with an adjustable timestep, ranging from a few seconds to one solar rotation ahead or behind. Model A predicts future magnetic field data using three consecutive radial magnetic field datasets with a 12-hour cadence. Model B reconstructs evolving magnetic fields over the past solar rotation using two sets of three consecutive datasets—one from the present and one from the previous solar rotation—each with a 12-hour cadence. To train and evaluate our models, we use a Pix2PixCC-based architecture and datasets of SDO/HMI vector magnetograms during the solar maximum periods of 2012–2016 and 2021–2023. Models A and B successfully generate magnetic field data corresponding to the input prediction timestep. Based on several evaluation metrics applied to the model outputs, Model A outperforms the persistence model and yields results comparable to the classical surface flux transport model, and Model B shows improved performance over both. We also compare the prediction results using data from NOAA AR 12673 and 13664, which produced intense solar flares and geomagnetic disturbances. Notably, our models can predict emerging magnetic flux several days in advance if their evolution is already captured in the input data. Model B can reconstruct radial magnetic flux that smoothly overlaps in regions near the Sun’s east and west limbs. Finally, we discuss the potential applications of our models in space weather forecasting—particularly in forecasting extreme solar flares—and in revisiting past solar events near the solar limb.
| Do you plan to attend in-person or online? | In-person |
|---|
