Speaker
Description
Climatological modeling of ionospheric plasma convection, driven by parameters such as solar wind velocity and density, interplanetary magnetic field (IMF) components, and geomagnetic indices, has been extensively investigated over the past decades. Interest in this topic continues to grow, both for its scientific importance, owing to the complex plasma processes it reveals, and for its practical relevance to satellite operation, space weather forecasting, and GNSS scintillation mitigation.
In this context, the Super Dual Auroral Radar Network (SuperDARN), operating for over four decades, provides one of the most comprehensive observational datasets for high-latitude ionospheric convection. In particular, electric potential maps can be derived at a 2-minute cadence using SuperDARN observations. This dataset provides a unique opportunity to investigate high-resolution spatial and temporal variations in ionospheric dynamics and to relate them to changes in upstream solar wind and IMF conditions.
In this study, we employ Diffusion Models, a deep learning technique, to forecast ionospheric convection in the Southern Hemisphere up to 20 minutes in advance. Our model is trained on SuperDARN-derived electric potential maps from 2014 to 2024, using as input the same solar wind parameters employed in one of the most robust SuperDARN empirical convection models (Cousins & Shepherd 2010): IMF Bx, By, Bz, and solar wind speed (Vsw).
Model performance is evaluated through the reconstruction of full 2D potential maps and validated by comparing the cross polar cap potential (CPCP), i.e., the maximum potential difference through the polar cap, between actual SuperDARN observations and Diffusion Model forecast outputs.
This study was carried out in part within the Space It Up project funded by the Italian Space Agency, ASI, and the Ministry of University and Research, MUR, under contract n. 2024-5-E.0 - CUP n. I53D24000060005.
| Do you plan to attend in-person or online? | In-person |
|---|
