Using artificial neural networks for comparison of the 09 March 2012 intense and 08 May 2014 weak storms


KÖKLÜ K.

Advances in Space Research, 2022 (Peer-Reviewed Journal) identifier

  • Publication Type: Article / Article
  • Publication Date: 2022
  • Doi Number: 10.1016/j.asr.2022.07.067
  • Journal Name: Advances in Space Research
  • Journal Indexes: Science Citation Index Expanded, Scopus, Academic Search Premier, Aerospace Database, Artic & Antarctic Regions, Communication Abstracts, Compendex, INSPEC, MEDLINE, Metadex, Civil Engineering Abstracts
  • Keywords: Artificial neural network (ANN) model, Mathematical modeling, Solar wind parameters (SWP), Zonal geomagnetic indices (ZGI)

Abstract

© 2022 COSPARInterplanetary parameters investigations are the keyspace of space research and are in the center of the relationship to the Sun and Earth. The investigations gain meaning by modeling various solar wind parameters (SWP) and zonal geomagnetic indices (ZGI). This essay, firstly, touches on the variables utilizing the classical approach and secondly, discusses them with an artificial neural network model (ANN). The classical approach is based on the relationship between SWP (E, v, P, T, N, Bz)-ZGI (Dst, Kp, AE, ap) and factor analysis. The ANN estimates the ZGI by SWP. While the SWP are independent variables (input), the ZGI are dependent variables (output) in the model. The ANN employs the backpropagation algorithm specified as the Scaled Conjugate Gradient (trainscg). Two different geomagnetic storms (May 08, 2014, weak storm, and March 09, 2012, strong storm) were handled as a problem. The ANN estimates the ZGI of the weak (Dst = −46) and the strong (Dst = −145 nT) storms with high accuracy. While the model validation performance evaluated with Mean Square Error (MSE) displayed as an absolute average total error.