Scholarlink

<p>Understanding long-term ionospheric total electron content (TEC) variability is crucial for enhancing satellite-based communication and navigation systems, particularly in low-latitude regions where ionospheric dynamics are highly complex. In this study, GNSS-derived TEC data from five low-latitude stations were analyzed during Solar Cycle 24 (2010-2020) and compared with the IRI-2016 and IRI-2020 models. Statistical analyses were performed across diurnal, seasonal, and solar-activity phases, and TEC variations were further predicted using an artificial neural network (ANN)-based NARX model. Strong correlations were observed between TEC and solar activity indices (0.83-0.99), while moderate correlations were obtained with geomagnetic indices (Dst and <InlineEquation ID="IEq1"> <EquationSource Format="MATHML"><math> <mi mathvariant="normal">Σ</mi> </math></EquationSource> <EquationSource Format="TEX">$\Sigma $</EquationSource> </InlineEquation>Kp). The IRI models reproduced the general TEC morphology but exhibited systematic overestimations and underestimations at specific stations, particularly near the equatorial ionization anomaly crest. In contrast, the ANN model demonstrated superior performance, yielding consistently low RMSE values across training, validation, and testing datasets. Entropy analysis revealed slightly enhanced variability during equinox periods, indicating increased ionospheric complexity. These results highlight the limitations of empirical models at low latitudes and demonstrate the effectiveness of data-driven approaches for accurate TEC prediction. This study uniquely integrates long-term multi-station TEC analysis, empirical model evaluation (IRI-2016 and IRI-2020), ANN-based prediction, and entropy-based complexity assessment within a unified framework, providing a comprehensive understanding of ionospheric variability over Solar Cycle&#xa0;24.</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

Low-latitude ionospheric TEC variability during Solar Cycle 24: assessment of IRI models using ANN

  • 期刊论文

  • 发表日期：

    2026年4月17日

查看全文
  • Nitin Dubey,
  • Swati,
  • Sparsh Agarwal,
  • Dhananjali Singh,
  • Devbrat Pundhir

摘要

Understanding long-term ionospheric total electron content (TEC) variability is crucial for enhancing satellite-based communication and navigation systems, particularly in low-latitude regions where ionospheric dynamics are highly complex. In this study, GNSS-derived TEC data from five low-latitude stations were analyzed during Solar Cycle 24 (2010-2020) and compared with the IRI-2016 and IRI-2020 models. Statistical analyses were performed across diurnal, seasonal, and solar-activity phases, and TEC variations were further predicted using an artificial neural network (ANN)-based NARX model. Strong correlations were observed between TEC and solar activity indices (0.83-0.99), while moderate correlations were obtained with geomagnetic indices (Dst and Σ $\Sigma $ Kp). The IRI models reproduced the general TEC morphology but exhibited systematic overestimations and underestimations at specific stations, particularly near the equatorial ionization anomaly crest. In contrast, the ANN model demonstrated superior performance, yielding consistently low RMSE values across training, validation, and testing datasets. Entropy analysis revealed slightly enhanced variability during equinox periods, indicating increased ionospheric complexity. These results highlight the limitations of empirical models at low latitudes and demonstrate the effectiveness of data-driven approaches for accurate TEC prediction. This study uniquely integrates long-term multi-station TEC analysis, empirical model evaluation (IRI-2016 and IRI-2020), ANN-based prediction, and entropy-based complexity assessment within a unified framework, providing a comprehensive understanding of ionospheric variability over Solar Cycle 24.