Deep Learning for Space Weather Prediction
Abstract
Through the use of our current fleet of in-orbit solar observatories, we have accumulated a vast amount of high quality solar event data which has greatly helped us to understand the underlying mechanisms of how the Sun works. However, we still lack an accurate and robust system for autonomously predicting solar eruptive events, which are known to cause geomagnetic storms, disturbances in electrical grids, radio black outs, increased drag on satellites, and increased radiation exposure to astronauts. We address the need for a flare prediction system by developing deep neural networks (DNNs) trained with solar data taken by the Helioseismic & Magnetic Imager (HMI) and Atmospheric Imaging Assembly (AIA) instruments onboard the Solar Dynamics Observatory and X-ray flux data taken by the GOES satellites. We describe the architecture of the DNNs trained and compare the performance between different implementations.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2016
- Bibcode:
- 2016AGUFMED41A0792P
- Keywords:
-
- 0805 Elementary and secondary education;
- EDUCATIONDE: 0825 Teaching methods;
- EDUCATIONDE: 0850 Geoscience education research;
- EDUCATIONDE: 0855 Diversity;
- EDUCATION