Bibliography



Notice:

  • Clicking on the DOI link will open a new window with the original bibliographic entry from the publisher.
  • Clicking on a single author will show all publications by the selected author.
  • Clicking on a single keyword, will show all publications by the selected keyword.




Modeling of Ultraviolet Aurora Intensity Associated With Interplanetary and Geomagnetic Parameters Based on Neural Networks


AuthorHu, Ze-Jun; Han, Bing; Zhang, Yisheng; Lian, Huifang; Wang, Ping; Li, Guojun; Li, Bin; Chen, Xiang-Cai; Liu, Jian-Jun;
Keywordsconditional generation adversarial network; generalized regression neural network; interplanetary and geomagnetic parameters; neural networks; ultraviolet auroral intensity model
AbstractThe spatial distribution of aurora intensity is an important manifestation of solar wind-magnetosphere-ionosphere energy coupling process, and it oscillates with the change of space environment parameters and geomagnetic index. It is of great significance to establish an appropriate aurora intensity model for the prediction of space weather and the study of magnetosphere dynamics. Based on Ultraviolet Imager (UVI) data of Polar satellite, we constructed two auroral models by using two different neural networks, that is, the generalized regression neural network (GRNN), and the conditional generation adversarial network (CGAN). Input parameters of the models include interplanetary magnetic field, solar wind velocity and density, and the geomagnetic AE index. Output result is the spatial distribution of auroral intensity in altitude adjusted corrected geomagnetic (AACGM) coordinates. The structural similarity index (SSIM) of image quality is used as an evaluation standard of detail similarity between the prediction results of auroral intensity model and corresponding UVI images (complete similarity is 1, dissimilarity is 0, SSIM is generally considered to have good similarity if it is greater than 0.5). Based on the respective training datasets of GRNN and CGAN models, the evaluating results showed that the mean values (standard deviation) of SSIM were 0.5409 (0.0912) and 0.5876 (0.0712), respectively, so the prediction results from both models can restore the auroral intensity distribution of the original images of UVI. In addition, the value of SSIM can increase with the increase of the number of training data. Therefore, more training data will help improve the effectiveness of these models.
Year of Publication2021
JournalSpace Weather
Volume19
Number of Pagese2021SW002751
Section
Date Published
ISBN
URLhttps://onlinelibrary.wiley.com/doi/abs/10.1029/2021SW002751
DOI10.1029/2021SW002751