Cheng Wang

The bottomside morphological parameters of the ionospheric F2 layer, the thickness parameter B 0 and the shape parameter B 1, are crucial for describing electron density profiles and monitoring space weather. However, existing climatological models and indirect ionosonde approaches have limitations in real‐time applicability and direct analytical fitting, struggling to meet the high‐precision, continuous representation needs of three‐dimensional ionospheric modeling. We propose a direct joint inversion framework based on ionosonde electron density data, combining robust nonlinear least‐squares (NLS) initialization with modern gradient optimization algorithms, including gradient descent, Adam, AdaGrad, and limited‐memory BFGS (L‐BFGS). The convergence performance of these algorithms was systematically compared using global ionosonde observations from 2014 (high solar activity) and 2020 (low solar activity). Results demonstrated that the NLS‐initialized L‐BFGS approach performs best overall in inversion accuracy, convergence stability, and computational efficiency across various conditions. These optimized parameters were then used to quantitatively evaluate the ABT‐2009 empirical model, revealing pronounced spatiotemporal variability in its performance. During high solar activity, the model exhibits systematic biases at low and high latitudes. Conversely, during low solar activity, prediction dispersion increases in summer mid‐latitudes. Persistent error concentration in low‐latitude and polar regions highlights the deficiencies of empirical models in capturing regional variability and dynamic physical processes. This high‐precision framework accurately retrieves ionospheric parameters, effectively diagnosing existing model deficiencies and laying a solid data foundation for future three‐dimensional ionospheric modeling.

The ionosphere has significant impacts on wireless communication, satellite navigation, and space weather. Based on hundreds of global GNSS observation stations, we calculate the rate of TEC index (ROTI). Through the analysis of ROTI, we establish a global ionospheric disturbed index (IDI) to directly characterize ionospheric disturbances. Daily IDI values from 1998 to 2023 are then calculated. Furthermore, we establish a multiple linear regression model to fit the IDI, using data from Ap, F_10.7, the upper atmospheric wind field, Kp, SYM‐H, and Dst indices. We analyze the contribution of each index to the IDI and classify the disturbance levels of the IDI time series, calculated by the multiple linear regression model for the period from 1981 to 2023, into five levels, ranging from severe disturbance to quiet. Next, the Lomb‐Scargle periodogram method is applied to perform spectral analysis on the IDI time series, extracting the annual, monthly, and daily periodic components of the IDI. Respectively, the annual periods are 1, 5.7, 6.6, 8.2, and 11.5 years; the monthly period is 6 months; and the daily periods are 9 and 26.5 days. Regarding the spatiotemporal features of IDI, our analysis reveals that temporally, the variations in IDI align with those of ROTI. Spatially, IDI exhibits stronger representation effectiveness in high‐latitude regions compared to mid‐ and low‐latitude regions. Finally, an investigation is conducted into the differences in the accuracy of standard point positioning under various conditions, including high and low solar activity years, different severities of geomagnetic storms, and across a range of IDI levels.

The ionosphere poses a significant source of error in satellite‐based navigation systems for aviation and radio communication applications. Accurate estimation of the total electron content (TEC) can effectively mitigate the impact of such errors. However, constrained by observational techniques, the acquisition of global ionospheric TEC in practical applications relies heavily on high‐precision forecasting products. In this study, we construct a global ionospheric forecasting model based on the global ionospheric TEC products disseminated by the International GNSS Service (IGS) and deep learning. We incorporate an attention module to extract global spatiotemporal features from historical ionospheric data and employ these features to predict the TEC values over the next 24 hr. Additionally, we select a long short‐term memory (LSTM) model and a ConvLSTM model as baseline models for comparison, conducting experiments under varying solar activity conditions. The experimental results demonstrate that RA‐ConvLSTM model outperforms the other two models in quantifying the performance of the models. During high solar activity years, the bias and Root Mean Square Error (RMSE) of RA‐ConvLSTM model are −0.0298 TECU and 3.8980 TECU, respectively, while during low solar activity years, these values are 0.0905 TECU and 1.5059 TECU, marking a notable improvement over the comparative models. Furthermore, by contrasting the precision of the three forecasting models during geomagnetic storms, the RA‐ConvLSTM model exhibits the least fluctuations in accuracy, indicative of a higher degree of stability in its forecasting outcomes.