Neural Network to Emulate Numerical Simulations of the Sun and Infer Synthetic Observations for Data Assimilation
Abstract
Satellites and ground-based observatories probe the Sun's photosphere and atmosphere and are key in studying solar activity. Meanwhile, numerical models have attempted to bridge the gap between the physics of the solar interior and such observations. However, there are physical quantities relevant to solar activity that can be modeled but that cannot be directly measured and must be inferred. For example, direct measurements of plasma motions at the photosphere are limited to the line-of-sight component. Recently, neural network computing has been used in conjunction with numerical models of the Sun to be able to recover the full velocity vector in photospheric plasma of the Quiet Sun. We used satellite observations as input in a fully convolutional neural network to generate instantaneous synthetic plasma motions, i.e. plasma motions that reflect the physics of a model but are made to look as if they were observed by a specific instrument. A parallel technique could then be invoked to eventually be able to derive the plasma velocity vector maps of the Active Sun and, by extension, other physical quantities of interest that can not yet be measured directly at the photosphere or anywhere else in the solar atmosphere.
- Publication:
-
Solar Heliospheric and INterplanetary Environment (SHINE 2019)
- Pub Date:
- May 2019
- Bibcode:
- 2019shin.confE..30T