Abstract We present a novel method for forecasting key ionospheric parameters using transformer‐based neural networks. The model provides accurate forecasts and uncertainty quantification of the F2‐layer peak plasma frequency (foF2), the F2‐layer peak density height (hmF2), and total electron content for a given geographic location. It supports a number of exogenous variables, including F10.7 cm solar flux and disturbance storm time (Dst). We demonstrate how transformers can be trained in a data assimilation‐like fashion that uses these exogenous variables along with naïve predictions from climatology to generate 24‐hr forecasts with nonparametric uncertainty bounds. We call this method the Local Ionospheric Forecast Transformer. We demonstrate that the trained model can generalize to new geographic locations and time periods not seen during training, and we compare its performance to that of the International Reference Ionosphere using CCIR coefficients.

Abstract We employed zonal wind data from Thermosphere, Ionosphere, Mesosphere Energetics and Dynamics Doppler Interferometer, equatorial electrojet (EEJ) measurements from Jicamarca (12°S, 77°W), and global ionospheric total electron content (TEC) maps to investigate the effects of ultra‐fast Kelvin waves (UFKW) with zonal wavenumbers 2 and 3 propagating eastward (E2, E3) in the equatorial mesosphere on both ionospheric TEC and EEJ signatures, as well as their differences in ionospheric response characteristics compared to E1 waves. Periodic components in zonal wind, EEJ, and TEC are quantified through the least squares spectral fitting. Our findings reveal three distinct categories of UFKW events: Type 1 exhibits both TEC and EEJ responses, Type 2 shows TEC response without EEJ signature, and Type 3 displays neither TEC nor EEJ response. The finding reveals distinct seasonal dependencies in TEC responses: E2 and E3 waves exhibit significant seasonality, whereas E1 waves show negligible seasonal variation. Furthermore, E1 waves demonstrate higher ionospheric response occurrence rates compared to E2 and E3 waves. For E1 waves, shorter periods, larger amplitudes, and longer vertical wavelengths correlate strongly with enhanced ionospheric responsiveness. Conversely, amplitude exerts minimal influence on ionospheric responses for E2 and E3 waves. For E2 and E3 waves temporally coincident with E1 waves, E2 and E3 waves may elevate Type 2 event occurrence among concurrent E1 waves, while E1 waves tend to suppress ionospheric response capability in coincident E2 and E3 waves, increasing Type 3 event prevalence.

An eight-year global look at correlations between total electron content, earthquakes and solar wind

Abstract The generation of traveling ionospheric disturbances (TIDs) near solar terminators has been predicted, and several studies have reported the detection of TIDs associated with sunrise. However, there are also observations that do not show TID signatures at sunrise. We address this issue by investigating false TID detection at sunrise using total electron content (TEC) data from the global navigation satellite system (GNSS) network over the United States and plasma density measurements from the first Republic of China satellite (ROCSAT‐1). The TEC morphology near sunrise, characterized by a rapid TEC increase, creates significant deviations between observed and trend (or filtered) TEC values. The collective pattern of this difference (dTEC) appears as a negative dTEC band aligned with sunrise. While this feature can be interpreted as a signature of large‐scale TIDs (LSTIDs), it may also falsely suggest a high occurrence rate of medium‐scale TIDs (MSTIDs) when dTEC is used as a detection proxy. However, the negative dTEC band at sunrise is not indicative of LSTIDs because the rapid TEC transition at sunrise is caused by photoionization. This interpretation is supported by the detection of a similar dTEC band at sunrise in model outputs that contain no TID information. Furthermore, the distribution of electron density irregularities derived from ROCSAT‐1 showed no evidence of enhanced MSTID activity near sunrise. Therefore, false TID detections near sunrise in GNSS TEC data should not be overlooked.

PredRNN is a spatiotemporal prediction model based on ST-LSTM units, capable of simultaneously extracting spatiotemporal features from ionospheric Total Electron Content (TEC). However, its internal convolutional operations require large kernels to capture low-frequency features, which can easily lead to model over-parameterization and consequently limit its performance. Although some studies have employed wavelet transform convolution (WTConv) to improve feature extraction efficiency, the introduced noise interferes with effective feature representation. To address this, this paper proposes a denoising wavelet transform convolution (DWTConv) and constructs the DWTPred-Net model with it as the key component. To systematically validate the model’s performance, we compared it with mainstream models (C1PG, ConvLSTM, and ConvGRU) under different solar activity conditions. The results show that both MAE and RMSE of DWTPred-Net are greatly reduced under all test conditions. In high solar activity, DWTPred-Net reduces RMSE by 13.81%, 6.19%, and 9.28% compared to the C1PG, ConvLSTM, and ConvGRU, respectively. In low solar activity, the advantage of DWTPred-Net becomes even more pronounced, with RMSE reductions further increasing to 19.39%, 11.51%, and 16.10%, respectively. Furthermore, in additional tests across different latitudinal bands and during geomagnetic storm events, the model consistently demonstrates superior performance. These multi-perspective experimental results collectively indicate that DWTPred-Net possesses obvious advantages in improving TEC prediction accuracy.

Total Electron Content (TEC) maps allow the evaluation of the state of the ionosphere. There are many providers/sources of worldwide or regional TEC maps for the continuous monitoring of the ionosphere, which employ different GNSS monitoring networks for data acquisition, TEC calculation or interpolation methods for generating the maps, or different spatial and temporal resolutions and coverage. How reliable are TEC maps over Brazil? We employed TEC maps from four different providers for 2022–2024, in the growing phase of the current solar cycle 25. Seasonality is also taken into account. A systematic comparison of TEC maps over Brazil was performed using correlation and similarity analysis between maps of different sources. Significant differences were found. Even for the same source there are differences in the density of monitoring stations according to the region. An example of bubble signature in TEC maps is also analyzed. Ground truth validation of TEC is performed by comparing TEC point values extracted from the maps with values derived from a set of GNSS stations over Brazil. As a result, no TEC maps of these sources were deemed reliable, due to low spatial and/or temporal resolution, low monitoring station density, or inadequate interpolation scheme.

Abstract This study, using the peak electron density of Ionospheric F‐region from the Global‐scale Observations of the Limb and Disk reports, for the first time, a unique phenomenon: the rapid reversal of the intensity of the Equatorial Ionization Anomaly (EIA) crests between the hemispheres during the main phase of 23 April 2023, geomagnetic storm. The Prompt Penetration Electric Field amplified the intensity of both EIA crests. However, the enhancement at the southern crest began to decay within an hour, while the northern crest began to strengthen. Thermosphere‐Ionosphere‐Electrodynamics General Circulation Model simulations indicate that trans‐equatorial wind played a key role in these variations. Wavelet periodograms of ground‐based Total Electron Content measurements confirmed the presence of Traveling Ionospheric Disturbances (TIDs). Storm‐induced winds and TIDs likely changed the altitude of the F‐peaks and the recombination rates between crests, and plasma transport by trans‐equatorial winds also contributed to this rare phenomenon.

This study presents a comprehensive analysis of pre- and co-seismic ionospheric disturbances associated with the 2023 Ms6.2 Jishishan earthquake by leveraging the unique observational strengths of BDS, particularly its high-orbit satellites. A multi-parameter space weather index was employed to effectively isolate seismogenic signals from geomagnetic disturbances, confirming that the main shock occurred during geomagnetically quiet conditions. Statistical analysis of 41 historical earthquakes (Mw ≥ 5.5) reveals that 47.2% were associated with detectable Total Electron Content (TEC) anomalies. An inverse correlation between earthquake magnitude and anomaly detectability within a 31-day window suggests prolonged precursor durations for larger events may produce longer-duration precursory signals, which challenge conventional detection methods. The synergistic capabilities of BDS Geostationary Earth Orbit (GEO) and Inclined Geosynchronous Orbit (IGSO) satellites were demonstrated: GEO satellites provide unprecedented temporal stability for continuous TEC monitoring, while IGSO satellites enable high-resolution spatial mapping of Co-seismic Ionospheric Disturbances (CIDs). The detected CIDs propagated at velocities below 1.6 km/s, consistent with acoustic gravity wave (AGW) mechanisms. A case study during a geomagnetically active period further reveals modulated CID propagation characteristics, indicating potential coupling between seismic forcing and space weather. Our findings validate BDS as a powerful and precise tool for ionospheric seismology and provide critical insights into Lithosphere–Atmosphere–Ionosphere Coupling (LAIC) dynamics.

The paper presents the results of modeling of spatial and temporal perturbations of the thermosphere during a strong meteorological disturbance. The modeling was performed using the Global Self-Consistent Model of the thermosphere, ionosphere, and protonosphere (GSM TIP). The impact of tropospheric/stratospheric sources on the thermosphere during dissipation of acoustic and internal gravity waves, generated in the meteorological storm region, was considered in GSM TIP by specifying an additional thermal source. The results of modeling of ionospheric effects of the meteorological storm in October 2017 have shown that the action of a local additional source of heating of the thermosphere leads to perturbations of the thermosphere and ionosphere parameters both directly above the source region and at a significant distance from it. In additional heating of the thermosphere, a decrease is observed in the total electron content (TEC) values, reaching 20 % in the daytime compared to a meteorologically quiet day. To the south and east of the source region, there are positive TEC perturbations with relative amplitudes 5–10 % during the daytime. The physical processes determining the ionospheric response directly in the source region are conditioned by heating of the thermosphere and its influence on changes in the neutral composition and circulation of the neutral wind. The TEC perturbations in the regions remote from the source region are determined by dynamic processes, which lead to the eastward transport of plasma and displacement of ionospheric perturbations to low latitudes.

The paper presents the results of modeling of spatial and temporal perturbations of the thermosphere during a strong meteorological disturbance. The modeling was performed using the Global Self-Consistent Model of the thermosphere, ionosphere, and protonosphere (GSM TIP). The impact of tropospheric/stratospheric sources on the thermosphere during dissipation of acoustic and internal gravity waves, generated in the meteorological storm region, was considered in GSM TIP by specifying an additional thermal source. The results of modeling of ionospheric effects of the meteorological storm in October 2017 have shown that the action of a local additional source of heating of the thermosphere leads to perturbations of the thermosphere and ionosphere parameters both directly above the source region and at a significant distance from it. In additional heating of the thermosphere, a decrease is observed in the total electron content (TEC) values, reaching 20 % in the daytime compared to a meteorologically quiet day. To the south and east of the source region, there are positive TEC perturbations with relative amplitudes 5–10 % during the daytime. The physical processes determining the ionospheric response directly in the source region are conditioned by heating of the thermosphere and its influence on changes in the neutral composition and circulation of the neutral wind. The TEC perturbations in the regions remote from the source region are determined by dynamic processes, which lead to the eastward transport of plasma and displacement of ionospheric perturbations to low latitudes.

The 27-day oscillation in total electron content (TEC) is analysed by means of world maps of TEC. The TEC maps are derived from measurements of the ground receiver network of the Global Navigation Satellite System (GNSS) and are provided by the International GNSS Service (IGS). The observed 27-day oscillation in TEC is mainly due to the 27-day solar rotation period, which induces a 27-day oscillation in extreme ultraviolet radiation (EUV) of the Sun. Analysing the time interval from 2003 to 2020, cross-correlation of the 27-day oscillation of the solar MgII-index of the Solar Radiation and Climate Experiment (SORCE) and the 27-day oscillation in TEC shows an average time delay of about 1.1 days for the ionospheric response with respect to the solar EUV variation. The average correlation coefficient of the solar and the ionospheric variation is 0.85. The cross-correlation of the 27-day oscillation in solar radio flux F10.7 and the 27-day oscillation in TEC gives a time lag of about 1.3 days and an average correlation coefficient of 0.78. The world maps of the amplitude of the 27-day oscillation in TEC are discussed for the TEC data from 1998 to 2024. Finally, TEC composites are derived for F10.7 enhancement events and geomagnetic storms.

Abstract Ionospheric data assimilation is the art of combining imperfect data with incomplete models to estimate the state of the ionosphere. The three most common data types are ionosonde measurements, Ground‐to‐GNSS (Global Navigation Satellite System), TEC (Total Electron Content) measurements, and RO (Radio Occultation) TEC measurements. Despite the ubiquitous use of these measurement types, scant research exists on the relative merits of each measurement type. This study evaluates the impact of assimilating all possible combinations of these three measurement types. To do this, we simulate representative data for all three measurement types using an electron density truth model, and then ingest all possible combinations of data in separate assimilation runs. Since we assimilate ground TEC in an absolute and relative sense, this yields 11 combinations. The performance of each assimilation run is assessed by how well each analysis replicates the truth model's vertical TEC (vTEC), critical plasma frequency of the F2 layer (foF2), the height at which it occurs (hmF2) and HF propagation metrics. When considering vTEC, fof2, and hmF2, we find that absolute ground TEC data is the most useful for specifying vTEC and that Radio Occultation data is the most useful when specifying foF2 and hmF2. Somewhat surprisingly, we find that adding absolute ground TEC can worsen predictions of foF2 and hmF2. Our analysis of HF propagation shows that ionosonde and RO data are quite valuable, and that ingesting ground TEC in a relative sense is better than absolute, regardless of what additional data (RO, Ionosonde) is present.

This study investigates high-latitude ionospheric mesoscale irregularities associated with energetic particle precipitation and magnetosphere-ionosphere-thermosphere coupling processes within the auroral oval using ground-based Global Navigation Satellite Systems (GNSS) total electron content (TEC) measurements. This scale size is much larger than those associated with GNSS scintillations, which range from sub-kilometers (small scales) to > 10 km (large scales). Analyzing 15 yr of data from 2010 to 2024, we characterize, for the first time, the climatology of enhanced intensity of ionospheric mesoscale irregularities at high latitudes. The observed intensity of irregularities in GNSS TEC fluctuations can serve as a proxy for the dynamic behavior of the auroral oval which varies with magnetic local time, longitude, latitude, season, solar activity cycle, geomagnetic disturbances, and hemisphere. The spatial distribution of the irregularity is oval-shaped and therefore this pattern is named as “irregularity oval”; the morphology of the irregularity oval is generally aligned well with the known variations of auroral oval established by using other technologies. While the primary goal has been to document systematically these irregularity long-term observations, future work will focus on the development of a novel GNSS TEC-based “irregularity oval” model.

The structure of the construction and algorithm of the complex for predicting the noise immunity of P-band satellite communications systems in conditions of strong ionospheric disturbances has been developed based on the simulation of the results of GPS-monitoring of the total electronic content of the ionosphere, causing dispersion distortions, fading and intersymbol interference of received signals. The basis for solving this problem is the development of a methodology for determining the dependence of the probability of erroneous reception of satellite communications signals on the average signal-to-noise ratio at the receiver input, the parameters of the transmitted signals, the results of GPS-monitoring of the average value and small-scale fluctuations in the total electronic content of the ionosphere and the propagation angles of radio waves in satellite communications and navigation systems. In accordance with the stages of the methodology, the structure of the construction and the algorithm of the complex for predicting the noise immunity of satellite communication systems in the P-frequency range during strong disturbances of the ionosphere, which are accompanied by the occurrence of frequency-selective fading, intersymbol interference and dispersion distortions of the received signals, have been developed. Experimental results are presented for predicting the noise immunity of a P-band satellite communications system based on GPS-monitoring of the ionosphere using a GPStation-6 receiver and simulating strong ionospheric disturbances by increasing the average value and fluctuations of the total electronic content by 1...2 orders of magnitude.

Abstract When estimating ionospheric Total Electron Content (TEC) using Global Navigation Satellite System (GNSS) observations, one of the significant error sources is the mapping error introduced by slant to vertical TEC conversion and vice versa. A single-layer Mapping Function (MF) based on a thin-shell assumption of the Earth’s ionosphere is commonly used for TEC conversion. However, the accuracy of single-layer MF is susceptible to the inaccurate fixing of the ionospheric single-layer height. In order to find a mapping approach less sensitive to the choice of ionospheric effective height we defined a multi-layer ionosphere mapping function and investigated its performance in comparison with the single-layer model. We found that the multi-layer MF outperforms the single-layer MF when computing GNSS receiver Differential Code Biases (DCBs) especially at low latitude and equatorial regions where ionosphere is highly dynamic and difficult to model. When compared with the International GNSS Services (IGS) products, we found that the mean receiver DCB estimation is improved (closer to benchmark) by about 0.14 – 0.27 ns and 0.30 – 0.78 ns during days in 2019 and 2023, respectively. We found that the receiver DCB estimation improves for about 66–87% receivers. This is also reflected in Global Ionosphere Maps (GIMs) showing better performance for the multi-layer MF when comparing with IGS GIMs. Our investigation using GNSS observations onboard Low Earth Orbiting (LEO) satellites shows that the multi-layer MF can be successfully applied in computing satellite and receiver DCBs accurately.

Abstract This study statistically investigates the distribution of the coseismic ionospheric disturbances (CIDs) in the north and south directions and the feature of the north‐south asymmetry of the CID amplitude using total electron content measurements for 13 earthquakes with magnitude 6.7–8.8 in the Chile subduction zone. The relative CID amplitude in the south direction is generally smaller than that in the north direction, but the propagation range of the CID in the south direction is basically consistent with that in the north direction. The north‐south amplitude ratio increases with decreasing epicentral distance. The larger CID amplitude in the north direction and the dependence of the north‐south amplitude ratio on the epicentral distance are found to be mainly associated with the effect of the geomagnetic field. Furthermore, the empirical relationship between the relative CID amplitude and the earthquake magnitude is suggested to consider geomagnetic field‐neutral orientation factor (GNF) for improved accuracy. It is also found that the north‐south amplitude ratio increases with the earthquake magnitude, which cannot be fully explained by the combined effect of GNF, north‐south asymmetry of the seismic energy release, and satellite observation geometry.

Ionospheric total electron content model over the Meridian Zone (120°E) based on the CNN-BiGRU algorithm

Abstract The ionosphere, which strongly varies in time and space, is one of the largest error sources for microwave measurements, if not accounted for accurately. Especially for single‐frequency observations, precise ionospheric models are needed to correct for the ionospheric delays. This study explores the utility of Very Long Baseline Interferometry (VLBI) and the VLBI Global Observing System (VGOS) in determining ionospheric delays essential for precise microwave measurements. By comparing Vertical Total Electron Content (VTEC) derived from colocated S/X VLBI and VGOS stations during simultaneous sessions, our analysis reveals good agreement between these two systems and Global Ionosphere Maps (GIM). The results indicate that VGOS yields more stable VTEC estimates, with a smaller RMS increase in differences to the GIM, from 2.48 TECU during solar quiet to 4.19 TECU during solar active conditions, compared to S/X VLBI, which increases from 2.32 to 5.60 TECU. This suggests that VGOS, due to its broader frequency coverage and higher observation density, enables more accurate ionospheric monitoring during active phases. An analysis of long‐term VLBI data indicates good agreement with GIM and corresponds closely to the yearly and solar‐cycle periods. Based on the data used, the bias in the VTEC estimates as derived from S/X VLBI and GIM is −1.5 TECU during solar quiet periods and −4.3 TECU during solar active periods. In relation to VGOS, the bias relative to GIM is −2.2 TECU and −3.4 TECU for the same respective conditions. Overall, S/X VLBI and VGOS can contribute to enhancing our understanding of space weather effects on microwave techniques.

Regional variability and multiscale dynamics of ionospheric total electron content during the intense geomagnetic storm of May 10, 2024, in central Mexico

Abstract The annular solar eclipse of 21 June 2020, provided a valuable opportunity to examine the ionospheric response to celestial events. This study analyzes variations in Vertical Total electron content (VTEC) over the equatorial region using data from the UQRG Global Ionospheric Map (GIM), ground‐based GPS‐TEC measurements, and equatorial electrojet (EEJ) strength from magnetometers near the eclipse path. Significant VTEC reductions were observed during the eclipse. The early decline began in East Africa and South Asia during the morning hours, while in the Western Pacific region, the reduction occurred in the late afternoon, coinciding with the onset of the eclipse. Around local noon, a delayed decrease was detected at stations located in Southeast and East Asia. 22%–53% VTEC reduction was recorded during the eclipse's main phase, with effects persisting from 35 min to over 8 hr post‐eclipse. Post‐eclipse variations in dusk sector suggest local electrodynamical effects. The study found no significant impact on EEJ strength between E to E, and no substantial counter equatorial electrojet was detected. Conducted during the monsoon season and the longest day of the year, observed VTEC reductions likely result from eclipse‐induced pressure changes and cooling effects. GIM visualizations showed that dVTEC% decreased by up to 40% globally. The findings indicate that VTEC decreases are not strongly correlated with obscuration percentage, highlighting complexity of ionospheric responses. This study enhances understanding of eclipse‐driven ionospheric variability, emphasizing the role of photoionization, recombination, geographic location, and local time, with implications for space weather forecasting and ionospheric modeling.