Scholarlink

<p>Ultra-rapid global ionospheric maps (GIMs) with low time latency can more quickly monitor and characterize the effects of space weather activities, which is crucial for various scientific applications. However, the number of stations that serve as its data source is relatively small and their stability is insufficient, which may cause certain data gaps in some areas. To improve the accuracy of GIMs in these areas, we introduce virtual vertical total electron content (VTEC) observations from the image inpainting model Aggregated COntextual-Transformation Generative Adversarial Nets (AOT-GAN) into the modeling process. Firstly, the VTEC of real ionospheric pierce points (IPPs) is gridded and imaged by constructing virtual grids. Then the AOT-GAN model is used to complete the image, and the completion results are integrated into the modeling process as virtual VTEC observations to obtain the final ultra-rapid GIMs product (ULTG). Finally, its accuracy is evaluated using three methods and compared with the ULTG_MD product with some station data missing. The GPS dSTEC (differential slant total electron content) assessment results show that compared with the ULTG_MD, the daily RMSE of ULTG has a maximum decrease of 23.28% and an average decrease of 7.52%. Moreover, ULTG can achieve similar performance to IGRG and UPCG in areas where station data are missing. The Jason-3 VTEC assessment results show that compared with the ULTG_MD, the maximum decrease in STD of ULTG is 28.52%, and the average reduction is 12.71%. In the three latitude ranges of 66° ~ 33°, 22° ~ -22°, and -33° ~ -66°, the maximum decreases of STD are 20.90%, 15.68%, and 31.68%, respectively, and the average decreases are 14.30%, 9.43%, and 22.38%, respectively. The performance of ULTG in the ocean region is roughly at the same accuracy level as the IGSG products and is slightly better than other rapid GIMs except for the optimal UQRG. In terms of consistency with IGSG, ULTG is roughly at a level similar to or even slightly better than CORG and UPCG. Incorporating virtual VTEC observations can not only improve the adverse effects of data missing on ionospheric modeling, but also improve the performance of the model in ocean areas and enhance the stability and reliability of the model.</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

Ultra-rapid global ionospheric modeling method incorporating AOT-GAN virtual VTEC observations under partial missing data conditions

  • 期刊论文

  • 发表日期：

    2026年5月15日

查看全文
  • Rong Wang,
  • Yibin Yao,
  • Liang Zhang,
  • Peng Chen,
  • Jian Kong,
  • Liangcai Qiu,
  • Fucai Tang

摘要

Ultra-rapid global ionospheric maps (GIMs) with low time latency can more quickly monitor and characterize the effects of space weather activities, which is crucial for various scientific applications. However, the number of stations that serve as its data source is relatively small and their stability is insufficient, which may cause certain data gaps in some areas. To improve the accuracy of GIMs in these areas, we introduce virtual vertical total electron content (VTEC) observations from the image inpainting model Aggregated COntextual-Transformation Generative Adversarial Nets (AOT-GAN) into the modeling process. Firstly, the VTEC of real ionospheric pierce points (IPPs) is gridded and imaged by constructing virtual grids. Then the AOT-GAN model is used to complete the image, and the completion results are integrated into the modeling process as virtual VTEC observations to obtain the final ultra-rapid GIMs product (ULTG). Finally, its accuracy is evaluated using three methods and compared with the ULTG_MD product with some station data missing. The GPS dSTEC (differential slant total electron content) assessment results show that compared with the ULTG_MD, the daily RMSE of ULTG has a maximum decrease of 23.28% and an average decrease of 7.52%. Moreover, ULTG can achieve similar performance to IGRG and UPCG in areas where station data are missing. The Jason-3 VTEC assessment results show that compared with the ULTG_MD, the maximum decrease in STD of ULTG is 28.52%, and the average reduction is 12.71%. In the three latitude ranges of 66° ~ 33°, 22° ~ -22°, and -33° ~ -66°, the maximum decreases of STD are 20.90%, 15.68%, and 31.68%, respectively, and the average decreases are 14.30%, 9.43%, and 22.38%, respectively. The performance of ULTG in the ocean region is roughly at the same accuracy level as the IGSG products and is slightly better than other rapid GIMs except for the optimal UQRG. In terms of consistency with IGSG, ULTG is roughly at a level similar to or even slightly better than CORG and UPCG. Incorporating virtual VTEC observations can not only improve the adverse effects of data missing on ionospheric modeling, but also improve the performance of the model in ocean areas and enhance the stability and reliability of the model.