EUVIMEDEUVIMED entdecken
Crossref

A Hybrid Data Assimilation of GNSS Ground‐Based VTEC and Radio Occultation Ionospheric Measurements Using Kalman Filter and Deep Learning

M. J. Wu, P. Guo, N. F. Fu, Y. Gao, J. C. Xue, B. Z. Pang, Y. H. Cai

Journal of Geophysical Research: Space Physics · 2026-05 · 131 (5)

DOI: 10.1029/2025ja034893 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Geophysical Research: Space Physics, online erschienen am 21.05.2026. Verfasst von M. J. Wu und 6 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The ionosphere plays a critical role in various geodetic techniques, including Global Navigation Satellite Systems (GNSS), Very Long Baseline Interferometry (VLBI), and associated applications. While the wealth of data from extensive ground‐based GNSS and spaceborne observation projects such as radio occultation facilitates high‐resolution data‐driven ionospheric modeling, significant challenges persist in the design of forward models and update of covariance matrix in Kalman Filter‐based assimilation, especially under demanding timeliness constraints. This study proposes a hybrid assimilation approach which applies hourly convolution gated recurrent unit (ConvGRU) forecast as a forward model during Kalman Filtering, based on ionospheric vertical total electron content (VTEC) observations from global ground‐based GNSS stations, follow‐on Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC‐2), and Yunyao commercial radio occultation constellation. Validations were conducted using the following independent data sets: GNSS observations, Global Ionospheric Maps (GIM), and JASON‐3 ionospheric measurements. The results demonstrate that the hybrid Kalman Filter and deep learning forecast framework can continuously correct the background state and rapidly achieve a stable, physically plausible error covariance; validation with GIM grids and JASON‐3 altimetry VTEC shows standard deviation of 4.7 TECU and 5.57 TECU, respectively, in December 2024; while assimilating multi‐source data yields better VTEC estimates within recursive procedures, utilizing a single data set can sustain model accuracy without significant degradation; the hybrid data assimilation has excellent performance in constructing ionospheric VTEC analysis even under disturbed geomagnetic conditions.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Wu, M. J., Guo, P., Fu, N. F., Gao, Y., Xue, J. C., Pang, B. Z., & Cai, Y. H. (2026). A Hybrid Data Assimilation of GNSS Ground‐Based VTEC and Radio Occultation Ionospheric Measurements Using Kalman Filter and Deep Learning. Journal of Geophysical Research: Space Physics, 131(5). https://doi.org/10.1029/2025ja034893

Vancouver

Wu MJ, Guo P, Fu NF, Gao Y, Xue JC, Pang BZ, et al. A Hybrid Data Assimilation of GNSS Ground‐Based VTEC and Radio Occultation Ionospheric Measurements Using Kalman Filter and Deep Learning. Journal of Geophysical Research: Space Physics. 2026;131(5). doi:10.1029/2025ja034893

BibTeX
@article{Wu2026,
  author = {Wu, M. J. and Guo, P. and Fu, N. F. and Gao, Y. and Xue, J. C. and Pang, B. Z. and Cai, Y. H.},
  title = {A Hybrid Data Assimilation of GNSS Ground‐Based VTEC and Radio Occultation Ionospheric Measurements Using Kalman Filter and Deep Learning},
  journal = {Journal of Geophysical Research: Space Physics},
  year = {2026},
  volume = {131},
  number = {5},
  doi = {10.1029/2025ja034893}
}
Kostenlos starten →