Scholarlink

<p>This study presents a near-real-time framework for detecting tsunami-induced ionospheric disturbances using detrended total electron content (dTEC) from the NASA JPL GNSS-based GUARDIAN system. The methodology is demonstrated for the Mw 8.8 Kamchatka earthquake of 29 July 2025 using GUARDIAN GNSS stations located in Japan. GUARDIAN processes streaming dual-frequency GNSS observations to estimate slant TEC, mitigate cycle slips, compute ionospheric pierce points at 350&#xa0;km altitude, and extract short-period ionospheric variability. An unsupervised Temporal Convolutional Network (TCN) was trained to model nominal dTEC behavior, with anomalies identified from prediction residuals exceeding statistically defined thresholds and temporal persistence criteria. Physical consistency was validated using phase-synchrony analysis between neighboring station–satellite links and spatial proximity constraints. After false-alarm mitigation, the framework achieved an F1-score of 93.9% with a recall of 91.8%. Detected disturbances propagated at ~ 230–234&#xa0;m s<sup>−1</sup> across multiple GNSS constellations, consistent with tsunami phase speeds inferred from bathymetric models. Spectral analysis confirmed signatures characteristic of atmospheric gravity waves. Notably, coherent ionospheric perturbations were detected approximately 71&#xa0;min prior to tsunami arrival at the Japanese coastline, highlighting the potential of GNSS-based deep learning methods for near-real-time tsunami early warning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Near–real-time detection of tsunami-induced traveling ionospheric disturbances using multi-constellation GNSS and a temporal convolutional network

  • 期刊论文

  • 发表日期：

    2026年5月5日

查看全文
  • Zeeshan Haider,
  • Jianguo Yan,
  • Jean-Pierre Barriot

摘要

This study presents a near-real-time framework for detecting tsunami-induced ionospheric disturbances using detrended total electron content (dTEC) from the NASA JPL GNSS-based GUARDIAN system. The methodology is demonstrated for the Mw 8.8 Kamchatka earthquake of 29 July 2025 using GUARDIAN GNSS stations located in Japan. GUARDIAN processes streaming dual-frequency GNSS observations to estimate slant TEC, mitigate cycle slips, compute ionospheric pierce points at 350 km altitude, and extract short-period ionospheric variability. An unsupervised Temporal Convolutional Network (TCN) was trained to model nominal dTEC behavior, with anomalies identified from prediction residuals exceeding statistically defined thresholds and temporal persistence criteria. Physical consistency was validated using phase-synchrony analysis between neighboring station–satellite links and spatial proximity constraints. After false-alarm mitigation, the framework achieved an F1-score of 93.9% with a recall of 91.8%. Detected disturbances propagated at ~ 230–234 m s−1 across multiple GNSS constellations, consistent with tsunami phase speeds inferred from bathymetric models. Spectral analysis confirmed signatures characteristic of atmospheric gravity waves. Notably, coherent ionospheric perturbations were detected approximately 71 min prior to tsunami arrival at the Japanese coastline, highlighting the potential of GNSS-based deep learning methods for near-real-time tsunami early warning.