Plasmasphere model using deep learning
Abstract
The deep learning is powerful tool to recognize images, sentences and audio as well as auto driving. In space sciences, the deep learning is used recently for predictions of non-linear process of phenomena occurring in the space science. We have applied to interpolate the observation data with the deep learning, because the observation data in the space are not still enough to understand the physical processes. We have developed a new plasmasphere model based on Japanese satellite data by use of deep learning. Our developed plasmasphere model estimates the electron densities at altitudes from 300 km to 40000 km in different latitude, longitude and solar and magnetic activities. The model includes memory effects of plasmasphere structure and dynamics affecting with geomagnetic disturbance. The estimated densities by the plasmasphere model show some differences with those of IRI model at 2000 km altitude. The plasmasphere model is connecting to an ionosphere model using deep learning, which has been already developed by us. Our final goal is to construct an ionosphere and plasmasphere model using deep learning (IPDL), and to understand the physical processes of various phenomena occurring in the ionosphere and plasmasphere of the earth.
- Publication:
-
43rd COSPAR Scientific Assembly. Held 28 January - 4 February
- Pub Date:
- January 2021
- Bibcode:
- 2021cosp...43E.798W