[P-EM12] 太陽地球系結合過程の研究基盤形成

[PEM12-P04] Improved real-time three-dimensional ionospheric tomography based on GNSS and ionosonde observations

*斎藤 享¹、河上 晃治²、Ssessanga Nicholas³、山本 衛²、西岡 未知⁴ (1.国立研究開発法人海上・港湾・航空技術研究所電子航法研究所、2.京都大学生存圏研究所、3.オスロ大学、4.情報通信研究機構)

キーワード：GNSS電離圏トモグラフィー、リアルタイム電離圏監視、イオノゾンデデータ同化GNSSトモグラフィー

Three-dimensional (3-D) ionospheric tomography based on GNSS network observations of the ionospheric total electron content (TEC) is a powerful tool to investigate behaviors of the ionosphere. Real-time 3-D ionospheric tomography over Japan based on GEONET (GNSS Earth Observation Network) has been implemented and operated since March 2016 (Saito et al., NAVIGATION, 2017). The 3-D ionospheric density over Japan is estimated every 15 minutes with about 6 minutes latency. The performance has been validated by different independent observations by ionosonde, incoherent scatter radar, and GNSS occultation by low earth orbit satellites. While the peak density estimated by the tomography was in good agreement with other observations, the ionospheric peak height was sometimes overestimated, when the peak height is low. Ssessanga et al. (GPS Solutions, 2021) proposed to assimilate ionosonde observations around Japan to GNSS tomography and improved the accuracy of the estimated peak height. Negative density in tomography solutions has also been avoid by using the log-normal distribution as the statistics of the electron density. In this study, the new tomography technique with ionosonde assimilation were brought into the real-time tomography system.
The real-time ionosonde data from four ionosonde stations in Japan (Wakkanai, Kokubunji, Yamagawa, and Okinawa) are used to assimilating into the GNSS tomography. Auto-scaled ionosonde parameters, hpF2 and foF2 are obtained as soon as the ionosonde observation and analysis finishes to estimate ionospheric peak density (NmF2) and height (hmF2). NmF2 and hmF2 are assimilated with the results of the original tomography results.
Preliminary comparison with the ionospheric density profile obtained by incoherent scatter observations by the MU radar showed that the improved real-time tomography provided the ionospheric density profile closer to the MU radar results than the current real-time tomography without ionosonde assimilation.
Mainly due to improved hardware which is still a normal desktop workstation and mathematical library, the process time has been reduced to about 2 minutes, which could realize real-time tomography analysis every 5 minutes.
While the preliminary results are promising, tuning of parameters involved in the tomography process is necessary to further improve the results. The parameters to be examined include the constraint about smoothness between adjacent voxels, ambiguity of ionosonde and TEC, correlation length in the assimilation, and weight factor between the original tomography and assimilation.