Skip to main navigation Skip to search Skip to main content

Short-term prediction of ionospheric total electron content based on deep neural network models

  • Fangzhou Wu
  • , Di Yao
  • , Changjun Yu*
  • *Corresponding author for this work
  • School of Information Science and Engineering, Harbin Institute of Technology Weihai
  • Northeastern University China

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This study utilizes ionospheric TEC data from Englewood, Florida (26.95°N, 82.35°W) in April and September 2024 in the United States. Transformer, CNN-LSTM and LSTM neural network model are used to generate predictions and evaluate the predictive performance of the neural network models. The research results prove that CNN-LSTM neural network model proves effective in predicting various ionospheric parameters and demonstrates notable improvements in accuracy compared to traditional LSTM neural network models. The Transformer model exhibits a significant enhancement in prediction accuracy over LSTM and CNN-LSTM, albeit at the cost of a notable increase in computational time.

Original languageEnglish
Title of host publicationInternational Conference on Remote Sensing, Surveying, and Mapping, RSSM 2026
EditorsFei Meng, Hongquan Song
PublisherSPIE
ISBN (Electronic)9798902325345
DOIs
StatePublished - 19 May 2026
Externally publishedYes
Event2026 International Conference on Remote Sensing, Surveying, and Mapping, RSSM 2026 - Chongqing, China
Duration: 16 Jan 202618 Jan 2026

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume14237
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Conference

Conference2026 International Conference on Remote Sensing, Surveying, and Mapping, RSSM 2026
Country/TerritoryChina
CityChongqing
Period16/01/2618/01/26

Keywords

  • CNN-LSTM
  • LSTM
  • Transformer
  • ionospheric irregularities
  • short-term forecast

Fingerprint

Dive into the research topics of 'Short-term prediction of ionospheric total electron content based on deep neural network models'. Together they form a unique fingerprint.

Cite this