(ORDO NEWS) — Scientists have developed a deep learning method for a neural network to obtain information about the latent turbulent motion based on the analysis of observations of the Sun.
Tests on three different datasets have shown that horizontal motion information can be obtained from temperature and vertical motion data processing.
This method will be useful to solar physicists, and can also be used in other areas of physical science, including plasma physics, nuclear physics, and fluid dynamics.
Although the Sun plays an important role as a renewable energy source and a natural example of a fusion reactor, the study of our star is limited by data collection capabilities.
While obtaining information about the temperature and vertical motion of the solar plasma is quite reliable, collecting data on the horizontal motion of the plasma is significantly difficult.
To solve this problem, in a new paper, a team led by Ryohtaroh T. Ishikawa of the National Astronomical Observatory of Japan created a neural network model and ran three different sets of simulated turbulent plasma data with it.
After training, the neural network was able to correctly extract horizontal motion data based on the vertical plasma motion and temperature data.
When the team tried to move from analyzing large sections of horizontal turbulent motion to smaller ones, a decrease in forecast accuracy was observed. Currently, the authors of the article are trying to solve the scaling problem in order to make their method universal.
This method may be used in the future to analyze high-resolution solar observations made with the SUNRISE-3 balloon-mounted telescope, as well as to study plasmas obtained in the laboratory, similar to those studied by nuclear physicists searching for new sources. energy.
—
Online:
Contact us: [email protected]
Our Standards, Terms of Use: Standard Terms And Conditions.