极地研究

极地研究 ›› 2024, Vol. 36 ›› Issue (4): 544-555.DOI: 10.13679/j.jdyj.20240049

• 研究论文 • 上一篇    下一篇

扰动条件下极区电离层NmF2预测研究

徐盛,牛月娟,李培豪   

  1. 郑州轻工业大学
  • 收稿日期:2024-04-30 修回日期:2024-08-22 出版日期:2024-12-31 发布日期:2025-01-15
  • 通讯作者: 徐盛
  • 作者简介:徐盛, 男, 1985年生。讲师, 主要从事极区电离层相关研究。E-mail: xusheng@zzuli.edu.cn
  • 基金资助:
    国家自然科学基金项目

Prediction of NmF2 in the polar ionosphere under disturbance conditions

XU Sheng, NIU Yuejuan, LI Peihao   

  1. School of Electronics and Information, Zhengzhou University of Light Industry, Zhengzhou 450000, China
  • Received:2024-04-30 Revised:2024-08-22 Online:2024-12-31 Published:2025-01-15
  • Contact: sheng xu

PDF (PC)

138

摘要： 利用南极中山站和北极Tromso站的电离层长期观测数据，结合时间序列预测模型Prophet、LSTM和多项式回归算法，构建了一种新型机器学习组合模型——PLPR，用于预测地磁扰动条件下极区台站电离层F2层峰值电子密度（NmF2）。结果表明：地磁扰动条件下，PLPR组合模型的预测结果能够较好地反映两个台站NmF2的日变化趋势，其预测精度在极隙区纬度的中山站要优于极光带纬度的Tromso站；与国际参考电离层IRI-2016模型和单一时间序列预测模型Prophet和LSTM相比，该模型具有更好的预测效果。

关键词: 极区电离层, 地磁扰动, F2层峰值电子密度（NmF2）

Abstract: A new prediction model of the ionospheric F2 layer peak electron density (NmF2) during geomagnetic disturbances for 1 hour is developed, using long-term ionospheric observation data from Zhongshan Station in Antarctica and Tromsø Station in Arctic. The machine learning model named PLPR is based on the time series prediction model Prophet, Long Short-Term Memory(LSTM) and polynomial regression. The predictions produced by PLPR can better reflect the daily variations of NmF2 at both stations than can those from other models, and the prediction accuracy at Zhongshan Station is better than that at Tromsø Station. Compared with the international reference ionosphere IRI-2016 model, as well as the single time series prediction model Prophet and LSTM, the PLPR model demonstrates superior performance.

Key words: Polar region ionosphere, Geomagnetic disturbances, F2 layer peak electron density (NmF2)