Temporal Modeling and Prediction of Global Ionospheric Total Electron Content: A Comparative Study of Deep Learning Approaches

Xinran Bu, Zheng Gong, Kunlin Yang, Yifei Chen, Yufei Zhang, and Yang Liu

Peer Reviewed

Abstract:	The ionosphere serves as an important part of the Earth's atmosphere and significantly impacts radio wave propagation and global navigation satellite systems (GNSS). The total electron content (TEC) is a key parameter for studying ionosphere morphology. Traditional TEC prediction methods perform well when dealing with steady or periodic variations but have limitations in face of the highly dynamic variability of the ionosphere. To address the problem, this work focuses on TEC prediction using deep learning models: Transformer, CNN-BiLSTN-Attention, and PatchTST. The PatchTST introduces a segmentation strategy to decompose TEC sequences into localized patches, enabling efficient learning of multi-scale ionospheric variations through self-attention mechanisms. We compare the performance of three deep learning networks for TEC prediction. The results show that the PatchTST model outperforms the other models in prediction accuracy, adaptability, and computational efficiency, with the smallest prediction error(RMSE=2.12TECU), the degree of concordance(R-square=0.844), exhibiting strong robustness and reliability. However, CNN-BiLSTM-Attention performs optimally during space weather and high solar activity periods due to its hybrid architecture combining spatial feature extraction and temporal memory. This work provides new perspectives for related fields, which is of great significance for the optimization of the GNSS system, the improvement of positioning accuracy, and the monitoring and prediction of space weather. Keywords—TEC, Self-Attention, Transformer, PatchTST
Published in:	2025 IEEE/ION Position, Location and Navigation Symposium (PLANS) April 28 - 1, 2025 Salt Lake Marriott Downtown at City Creek Salt Lake City, UT
Pages:	1161 - 1169
Cite this article:	Bu, Xinran, Gong, Zheng, Yang, Kunlin, Chen, Yifei, Zhang, Yufei, Liu, Yang, "Temporal Modeling and Prediction of Global Ionospheric Total Electron Content: A Comparative Study of Deep Learning Approaches," 2025 IEEE/ION Position, Location and Navigation Symposium (PLANS), Salt Lake City, UT, April 2025, pp. 1161-1169.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In