International Reference Ionosphere Research Articles

Mid- and High-Latitude Electron Temperature Dependence on Solar Activity in the Topside Ionosphere through the Swarm B Satellite Observations and the International Reference Ionosphere Model

This study focuses on the open question of the electron temperature (Te) variation with solar activity in the topside ionosphere at mid- and high latitudes. It takes advantage of in situ observations taken over a decade (2014–2023) from Langmuir probes on board the low-Earth-orbit Swarm B satellite and spanning an altitude range of 500–530 km. The study also includes a comparison with Te values modeled using the International Reference Ionosphere (IRI) model and with Millstone Hill (42.6° N. 71.5° W) incoherent scatter radar observations. The largest Te variation with solar activity was found at high latitudes in the winter season, where Te shows a marked decreasing trend with solar activity in the polar cusp and auroral regions and, more importantly, at sub-auroral latitudes in the nightside sector. Differently, in the summer season, Te increases with solar activity in the polar cusp and auroral regions, while for equinoxes, variations are smaller and less clear. Mid-latitudes generally show negligible Te variations with solar activity, which are mostly within the natural dispersion of Te observations. The comparison between measured and modeled values highlighted that future implementations of the IRI model would benefit from an improved description of the Te dependence on solar activity, especially at high latitudes.

PyIRI: Whole‐Globe Approach to the International Reference Ionosphere Modeling Implemented in Python

AbstractThe International Reference Ionosphere (IRI) model is widely used in the ionospheric community and considered the gold standard for empirical ionospheric models. The development of this model was initiated in the late 1960s using the FORTRAN language; for its programming approach, the model outputs were calculated separately for each given geographic location and time stamp. The Consultative Committee on International Radio (CCIR) and International Union of Radio Science (URSI) coefficients provide the skeleton of the IRI model, as they define the global distribution of the maximum useable ionospheric frequency foF2 and the propagation factor M(3,000)F2. At the U.S. Naval Research Laboratory, a novel Python tool was developed that enables global runs of the IRI model with significantly lower computational overhead. This was made possible through the Python rebuild of the core IRI component (which calculates ionospheric critical frequency using the CCIR or URSI coefficients), taking advantage of NumPy matrix multiplication instead of using cyclic addition. This paper explains in detail this new approach and introduces all components of the PyIRI package.

Empirical Models of foF2 and hmF2 Reconstituted by Global Ionosonde and Reanalysis Data and COSMIC Observations

AbstractThe F2‐peak plasma frequency (foF2) and the height of the F2 peak (hmF2) are two of the most important parameters for any ionospheric model, as well as radio propagation studies and applications. In this study, we have developed empirical models to capture the most significant variations of foF2 and hmF2. The derived empirical models (referred to as the USTC models within this study) are specified through global ionosonde and reanalysis data based on the International Reference Ionosphere (IRI) Consultative Committee on International Radio (CCIR) method and Constellation Observindg System for Meteorology, Ionosphere, and Climate (COSMIC) observations based on the empirical orthogonal function analysis, respectively. The USTC models are validated against the IRI CCIR model prediction. The comparison results revealed that the empirical foF2 model performs better in capturing the foF2 variations than the IRI CCIR model, which can overcome the underestimation of the IRI CCIR model at low latitudes. Although the IRI CCIR model overestimation at middle latitudes is addressed by the empirical hmF2 model, the visible differences between the model predictions and ionosonde observations still exist at low latitudes, which could be attributed to the significant difference between COSMIC and ionosonde hmF2 measures.

A Simplified Method of True Height Analysis to Estimate the Real Height of Sporadic E Layers

AbstractThe sporadic E (Es) layer virtual height and the ordinary wave critical frequency are routinely measured ionogram parameters, used to characterize the Es altitude occurrence and intensity. It has become common practice to take the real height to be about equal to by assuming that signal propagation delays in the E region plasma below the layer are small and can be neglected. Although this applies for nighttime, during daytime it may overestimate significantly. The present paper relies on true height analysis theory to devise a simplified method and propose an algorithm that can estimate Es real heights reasonably well. The method relies on and ionosonde measurements and E region electron density profiles obtained from the International Reference Ionosphere model. The algorithm is applied to a typical set of Digisonde observations to compute and examine real height variations and functional dependencies. Whereas  ≃  at nighttime, during daytime there are notable  −  differences taking values less than 10 km for most of the observed layers. During the early morning and early afternoon hours, however, when weak layers appear at upper heights, the virtual to real height differences become larger reaching 20–25 km. The method proposed here for the estimation of can be easily applied to improve the accuracy of the results of sporadic E layer studies.

Prediction of ionospheric TEC using a GRU mechanism method

In this paper, we propose a gate recurrent unit (GRU) method to forecast ionospheric the total electron content (TEC) at different latitudes and longitudes. This method aims to address the issues of low accuracy in traditional neural networks and the challenges faced in training long-term time series. We construct a GRU model using TEC data from the European Orbit Determination Center (CODE) for the years 2016 to 2018, geomagnetic indices, and solar activity indices (16 points along the latitude and longitude lines of 120°E and 30°N). Comparative experiments are carried out with the international reference ionosphere (IRI) model and the long short-term memory (LSTM) model. The results show that the GRU model outperforms other models in terms of accuracy. On geomagnetic quiet days, the 5-day predicted average residual can reach 0.51 TECU, while on storm days it can reach 0.72 TECU. The root mean square error (RMSE) for these days is 0.64 TECU and 1.01 TECU, respectively. The correlation coefficients are 0.98 and 0.93, respectively. The RMSE of the GRU model ranges from 0.60 TECU to 2.62 TECU for both geomagnetic quiet days and storm days. In terms of TEC forecast accuracy in 2018, the RMSE of the GRU model is 1.6 TECU, which is smaller than that of the IRI model and LSTM model. These results demonstrate that the GRU model performs better than the IRI model and LSTM model.

F1 Region Ion Composition in Svalbard During the International Polar Year 2007–2008

AbstractIons in the F region ionosphere at 150–400 km altitude consist mainly of molecular NO+ and , and atomic O+. Incoherent scatter (IS) radars are sensitive to the molecular‐to‐atomic ion density ratio, but its effect to the observed incoherent scatter spectra is almost identical with that of the ion temperature. It is thus very difficult to fit both the ion temperature and the fraction of O+ ions to the observed spectra. In this paper, we introduce a novel combination of Bayesian filtering, smoothness priors, and chemistry modeling to solve for F1 region O+ ion fraction from EISCAT Svalbard IS radar (75.43° corrected geomagnetic latitude) data during the international polar year (IPY) 2007–2008. We find that the fraction of O+ ions in the F1 region ionosphere is controlled by ion temperature and electron production. The median value of the molecular‐to‐atomic ion transition altitude during IPY varies from 187 km at 16–17 MLT to 208 km at 04–05 MLT. The ion temperature has maxima at 05–06 MLT and 15–16 MLT, but the transition altitude does not follow the ion temperature, because photoionization lowers the transition altitude. A daytime transition altitude maximum is observed in winter, when lack of photoionization leads to very low daytime electron densities. Both ion temperature and the molecular‐to‐atomic ion transition altitude correlate with the Polar Cap North geomagnetic index. The annual medians of the fitted transition altitudes are 14–32 km lower than those predicted by the International Reference Ionosphere.

Modeling Ionospheric TEC Using Gradient Boosting Based and Stacking Machine Learning Techniques

AbstractAccurately predicting and modeling the ionospheric total electron content (TEC) can greatly improve the accuracy of satellite navigation and positioning and help to correct ionospheric delay. This study assesses the effectiveness of four different machine learning (ML) models in predicting hourly vertical TEC (VTEC) data for a single‐station study over Ethiopia. The models employed include gradient boosting machine (GBM), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM) algorithms, and a stacked combination of these algorithms with a linear regression algorithm. The models relied on input variables that represent solar activity, geomagnetic activity, season, time of the day, interplanetary magnetic field, and solar wind. The models were trained using the VTEC data from January 2011 to December 2018, excluding the testing data. The testing data comprised the data for the year 2015 and the initial 6 months of 2017. The RandomizedSearchCV algorithm was used to determine the optimal hyperparameters of the models. The predicted VTEC values of the four ML models were strongly correlated with the GPS VTEC, with a correlation coefficient of ∼0.96, which is significantly higher than the corresponding value of the International Reference Ionosphere (IRI 2020) model, which is 0.87. Comparing the GPS VTEC values with the predicted VTEC values based on diurnal and seasonal characteristics showed that the predictions of the developed models were generally in good agreement and outperformed the IRI 2020 model. Overall, the ML models used in this study demonstrated promising potential for accurate single‐station VTEC prediction over Ethiopia.

Prediction of ionospheric total electron content over low latitude region: Case study in Ethiopia

This study mainly focuses on the prediction of the variability of the Total Electron Content (TEC) affecting space and ground-based technological infrastructures vulnerable to extreme space weather events. The impact of solar wind conditions (SWC) can be quantified by measuring the ionosphere’s TEC. We used a machine learning (ML) algorithm based on the Support Vector Machine (SVM), Long Short-Term Memory (LSTM) Neural Network (NN), and International Reference Ionosphere 2016 (IRI 2016) models to predict the hourly TEC over Ethiopia during quiet and storm geomagnetic conditions. We considered TEC data from the Global Positioning System (GPS) stations at ADIS, CURG, and NEGE during the years 2013 to 2016. The results indicated that the SVM model predicted the hourly TEC variation with a root mean square error (RMSE) between 2.136 and 7.923 Total Electron Content Unit (TECU) under different ionospheric conditions. For the LSTM model, the RMSE between the predicted and observed TEC lies between 1.483 and 2.527 TECU. But the empirical IRI 2016 models’ RMSE from GPS TEC is in the range of 4.777 and 14.519 TECU. The statistical study points out the LSTM has the highest prediction accuracy and is the most robust for accurate prediction of ionospheric TEC in comparison to the SVM and IRI 2016 models throughout both storm and quiet periods. The result can be further improved with a lower RMSE (high correlation) by utilizing auxiliary data and increasing the number of stations employed in the study. This research adds fresh insights to deep learning applications in space weather forecasting in the region of interest.

Assessment of the accuracy of IRI-2016 model at the equatorial-, low-, mid-, and high-latitudes ionosphere along 30–45oE meridian lines during minimum (2009) and maximum (2013) phases of solar cycle 24

This study assesses the accuracy of the International Reference Ionosphere IRI-2016 model prediction at the equatorial-, low-, mid-, and high-latitude ionosphere along 30–45oE meridian during minimum (2009) and maximum (2013) phases of solar cycle 24. We used raw TEC derived from GPS observation data from GPS receivers within ∼±70o latitudinal band, along the African/European longitudinal sector. We compared the GNSS-TEC data after the Plasmaspheric Electron Content (PEC) has been excluded with those derived from the two components of the IRI-2016 model. The percentage contributions of PEC to GNSS-TEC was minimum in June solstice, and maximum in December solstice with the highest values at the high-latitude regions, but with a clear hemispherical asymmetry; manifesting the winter anomaly effect. Higher TEC values were recorded in the equatorial/low-latitude regions than in the high- and mid-latitude regions. Overall, in both hemispheres, the accuracy of prediction of the IRI-2016 model was higher in the high- and mid-latitude regions than in the equatorial/low-latitude regions. The accuracy of the model was higher during solar minimum than during solar maximum. Both options of IRI-2016 performed better at nighttime than during daytime. IRI-NeQuick option generally performed better than the IRI-Corr option. TEC values were higher in the southern hemisphere than in the northern hemisphere.

Performance evaluation of ionospheric models over equatorial Indian region

An extensive year-long evaluation of Total Electron Content (TEC) obtained from NCAR Thermosphere-Ionosphere Electrodynamics General Circulation Model (TIE-GCM) simulations along with TEC obtained from International Reference Ionosphere (IRI-2020) is carried out with respect to observations made by a GPS TEC receiver over the dip equatorial location of Thiruvananthapuram (8.5° N, 77° E, 0.5° magnetic), in the Indian region during solar minimum period of 2005–2006. The TIE-GCM model and IRI model were run for the period of 2005 (November December) and till October 2006 for each day, and the hourly values of TEC were compared to GPS TEC observations. Overall the temporal and seasonal characteristics of TEC over Thiruvananthapuram were simulated reasonably well by the TIE-GCM model and IRI model. The skill scores of normalized Root Mean Square Error (nRMSE), Prediction Efficiency (P.E), and Ratio between maximum and minimum values were estimated for both models to evaluate their performance. The mean percentage difference between the measured and the simulated TEC was found to be around −25% for IRI, indicating underestimation of the observations, and between −40% and −50% for TIE-GCM again indicating underestimation. A comprehensive analysis revealed that there is good conformity between the modeled and measured values for the whole observation period, with a correlation coefficient R-value of 0.8 and 0.9 for the TIE-GCM and IRI models respectively. This study brings out the competence of the two models in simulating the temporal and seasonal pattern of variability of ionospheric TEC over the equatorial Indian region.

Ionospheric Error Models for Satellite-Based Navigation—Paving the Road towards LEO-PNT Solutions

Low Earth Orbit (LEO) constellations have recently gained tremendous attention in the navigational field due to their larger constellation size, faster geometry variations, and higher signal power levels than Global Navigation Satellite Systems (GNSS), making them favourable for Position, Navigation, and Timing (PNT) purposes. Satellite signals are heavily attenuated from the atmospheric layers, especially from the ionosphere. Ionospheric delays are, however, expected to be smaller in signals from LEO satellites than GNSS due to their lower orbital altitudes and higher carrier frequency. Nevertheless, unlike for GNSS, there are currently no standardized models for correcting the ionospheric errors in LEO signals. In this paper, we derive a new model called Interpolated and Averaged Memory Model (IAMM) starting from existing International GNSS Service (IGS) data and based on the observation that ionospheric effects repeat every 11 years. Our IAMM model can be used for ionospheric corrections for signals from any satellite constellation, including LEO. This model is constructed based on averaging multiple ionospheric data and reflecting the electron content inside the ionosphere. The IAMM model’s primary advantage is its ability to be used both online and offline without needing real-time input parameters, thus making it easy to store in a device’s memory. We compare this model with two benchmark models, the Klobuchar and International Reference Ionosphere (IRI) models, by utilizing GNSS measurement data from 24 scenarios acquired in several European countries using both professional GNSS receivers and Android smartphones. The model’s behaviour is also evaluated on LEO signals using simulated data (as measurement data based on LEO signals are still not available in the open-access community; we show a significant reduction in ionospheric delays in LEO signals compared to GNSS. Finally, we highlight the remaining open challenges toward viable ionospheric-delay models in an LEO-PNT context.

Interaction of global electron content with the Sun and solar wind during intense geomagnetic storms

Assessment of solar and solar wind parameters driving the ionosphere model is essential for prediction of the ionospheric weather. In the present paper impact of the different solar, interplanetary and geomagnetic parameters on the global electron content (GEC) during intense space weather storms is investigated. Hourly GEC values are calculated from JPL global maps of total electron content GIM-TEC from 1995 to 2023. The sample comprises 90 intense storms from 1995 to 2023 associated with monthly peak of the weighted accumulation of the geomagnetic Apo(τ, t) index exceeding 90 nT. The 27 day weighted accumulation of the solar sunspot numbers SSN2(τ), solar radio flux F10.7(τ), the solar hydrogen emission Lyman_α(τ) and the composite magnesium MgII(τ) indices are explored as precursors of GEC enhancements. As distinct from the positive ionosphere storm, the solar wind speed Vsw, the solar wind electric field Ey, merging electric field Em and Apo(τ, t) indices proved to be effective as potential drivers of the negative GEC depletion. The positive and negative dGEC deviations from hourly GEC are produced by subtracting a quiet reference GECav averaged during 24h prior the storm normalized by GECav. The hourly storm profiles Vsw(t), Em(t), Ey(t), Apo(τ, t), Dst(t), GEC(t) and dGEC(t) were reduced by method of superposed epochs. The zero epoch t0 = 0 was taken at the peak Apo*(τ, t0) and the storm time lasted for 48h from −12h prior t0 and 35h afterwards. The best correlation of the positive storm dGECp amplitude is obtained with MgII(τ) and the negative storm dGECn with Em* and Apo*(τ, t0) which are used to derive characteristics of five key points of storm-time dGEC(t) model: 1 – onset of the storm profile t1 = t(dGECp); 2 – the amplitude dGECpmax and its time t2(dGECpmax); 3 – the time of transition t3(dGEC = 0) from the positive to negative (±) GEC storm; 4 – minimum negative disturbance dGECnmin and its time t4(dGECnmin), 5 – the end of the storm profile t5(dGECn). Analytical model of dGEC(t) is derived with Epstein step functions fitting 5 key points. Deviations dGEC(t) are inverted to GEC(t) using quiet reference pre-storm GECav. The model is validated for three intense storms on 26–28 February, 23–25 March and 23–25 April 2023. The results show improvement of dGEC forecast with RMS error reduced from 45 to 80% compared to results produced by the international reference ionosphere−plasmasphere model IRI-Plas.

Modeling Total Electron Content and Critical Frequency at F1 and E Layer Boundary

This work investigates the variation of the total electron content TEC and the critical frequency fo in the boundary zone of the F1 and E layers at the low-latitude in the ionosphere. This study takes place at the Ouagadougou station (12.4°N and 358.5°E), in West Africa during the quiet geomagnetic activity of solar cycle 23. Ionosphere is the upper layer of the Earth's atmosphere ionized mainly by solar X- and UV-rays, extending from around 80km altitude up to 1000km [1] [2]. Ultraviolet light from the sun ionizes the atoms and molecules in the Earth's upper atmosphere [3]. For this study we use the 2016 version of International Reference Ionosphere (IRI) model. The quiet periods of maximum and minimum phase of solar cycle 23 are considered [4] [5]. From this study, it emerges that at the E and F1 layer boundary zone, TEC and fo increase during the day as solar irradiance increases and decrease as solar irradiance decreases.

Topside Electron Density Modeling Using Neural Network and Empirical Model Predictions

AbstractWe model the electron density in the topside of the ionosphere with an improved machine learning (ML) model and compare it to existing empirical models, specifically the International Reference Ionosphere (IRI) and the Empirical‐Canadian High Arctic Ionospheric Model (E‐CHAIM). In prior work, an artificial neural network (NN) was developed and trained on two solar cycles worth of Defense Meteorological Satellite Program data (113 satellite‐years), along with global drivers and indices to predict topside electron density. In this paper, we highlight improvements made to this NN, and present a detailed comparison of the new model to E‐CHAIM and IRI as a function of location, geomagnetic condition, time of year, and solar local time. We discuss precision and accuracy metrics to better understand model strengths and weaknesses. The updated neural network shows improved mid‐latitude performance with absolute errors lower than the IRI by 2.5 × 109 to 2.5 × 1010 e−/m3, modestly improved performance in disturbed geomagnetic conditions with absolute errors reduced by about 2.5 × 109 e−/m3 at high Kp compared to the IRI, and high Kp percentage errors reduced by >50% when compared to E‐CHAIM.

Ion Density Climatology Based on FPMU Measurements on Board the International Space Station

AbstractWe performed, for the first time, a season‐dependent geomagnetically quiet‐time climatology of mid‐ and low‐latitude ion densities during low and moderate solar flux conditions using Floating Potential Measurement Unit (FPMU) observations aboard the International Space Station (ISS) from 2008 to 2019. Our daytime observations indicate that the main characteristics of the equatorial ionization anomaly (EIA) at ∼400 km are consistent with those from other in situ and remote sensing probes. The FPMU daytime densities are also generally in good agreement with the corresponding results from the International Reference Ionosphere (IRI). However, the IRI does not reproduce the mid‐solar flux evening low‐latitude measured densities. In this period, the EIA exhibits strong longitude‐dependent crest‐to‐trough ratios and asymmetries due to the pre‐reversal enhancement (PRE) of the zonal electric field and thermospheric neutral winds. Our data also show strong structuring of the daytime and nighttime plasma densities. This includes a bulge, discussed for the first time here, in the EIA southern crest in the South Atlantic sector during the December solstice and equinox data, which we suggest being generated by the transport of plasma from the Weddell Sea anomaly (WSA). We also highlight and show the evolution of a midlatitude summer nighttime anomaly (MSNA) during the June solstice data in the North Atlantic sector. Our results give new insights into these two anomalies, where we show that they are stronger with increasing solar flux levels and that they last until the early morning. These latter results are not consistent with those from previous studies.

CLIMF2: A climatological model of the ionospheric F2 layer. Part 1: Model methodology

The ionospheric F2 layer is of key importance to systems dependent on the propagation of radio waves through the ionosphere, including over-the-horizon (OTH) radar, high frequency (HF) communications, global navigation satellite systems (GNSS), and low-frequency radio astronomy. The currently accepted F2 layer model, the International Reference Ionosphere (IRI), does not provide the variance in the parameters used to describe the layer and furthermore, was developed from a limited dataset. The newly developed model, CLIMF2, can be used in conjunction with IRI to overcome these limitations, and thereby improve applications such as the climatological modelling of OTH radar performance. CLIMF2 is an empirical climatological model of the F2 layer parameters derived from a global network of vertical incidence sounder (VIS) observations, consisting of 267 sites with data from the 1950s to 2021. CLIMF2 has the added functionality of providing the variance in the foF2 and hmF2 parameters.

Validation of global ionospheric models using long-term observations of pulsar Faraday rotation with the LOFAR radio telescope

Broad band pulsar radiation can be effectively used to monitor the properties of the magneto-ionic media through which it propagates. Faraday rotation calculated from polarised pulsar observations provides an integrated product of electron densities and the line-of-sight component of the magnetic field in the intervening plasma. In particular, a time-variable effect mainly associated with the rapidly changing column density of the Earth’s ionosphere and plasmasphere heavily dominates the observed Faraday rotation of pulsar radiation. In this work, we aim to carry out a performance test of three GNSS-based models of the ionosphere using observations of PSR J0332+5434 taken with the LOw Frequency ARray (LOFAR). As it was shown in Porayko et al. (Month Not Roy Astron Soc 483(3):4100–4113, 2019. https://doi.org/10.1093/mnras/sty3324. arXiv:1812.01463), the conventional single layer model (SLM), which assumes that the ionosphere is a thin slab at a fixed effective height, is not capable of fully accounting for the ionospheric Faraday rotation in pulsar data. The simplified physics of the SLM is upgraded within IRI-Plas (International Reference Ionosphere and Plasmasphere) extended SLM and the dual-layer voxel TOmographic Model of the Ionosphere (TOMION), both of which partially account for the thickness and vertical dynamics of the terrestrial plasma. Although the last two improve the reconstruction of the ionospheric Faraday rotation, none of the considered models completely purge the observed residual variations. With this study, we show that the long term LOFAR observations of Faraday rotation of pulsars provide an excellent tool to test and improve models of the magneto-ionic content of the Earth’s atmosphere.

Variation of Electron Density in the D-Region Using Kunming MF Radar under Low Solar Activity

So far, the least is known about the D-region ionosphere out of the entire ionosphere due to the lack of a conventional detecting method and continuous data accumulation. Medium frequency (MF) radar is an important conventional tool for understanding the D-region ionosphere by measuring the electron density (Ne) within the height range of 60–90 km. To investigate the statistical variation of the D-region, especially at the mid-low latitude area, this study presents the statistical variations in the D-region Ne with the solar zenith angle (SZA), season, and altitude observed by Kunming MF radar (25.6° N, 103.8° E) under low solar activity (2008–2009). The diurnal variation of Ne behaves like typical diurnal changes, which are closely consistent with the SZA. The outstanding feature, the diurnal asymmetry phenomenon, significantly appears in different seasons and at different altitudes. The Ne has obvious semi-annual characteristics, and is larger in summer and fall and the smallest in winter. Compared to other seasons, the variation in the Ne with altitude is the most stable in summer. Due to the impacts of the highest SZA, the value of Ne in winter is the smallest, with a maximum value of less than 300 electrons/cm3, and the largest in summer and fall, with a maximum of 472 electrons/cm3. Particularly, the peaks of Ne above 76 km do not always appear at the time when the SZA is the smallest (at noon). Both the simulations by the International Reference Ionosphere (IRI2016) and observations using MF radar present a strong positive correlation with solar radiation. Meanwhile, it cannot be ignored that there were still large differences between the simulations and observations. To quantitatively analyze the differences between the observations and simulations, the observed value was subtracted from the simulated value. The results show that the maximum value between them was up to 350 electrons/cm3, and the minimum difference appeared at around 72 km, with a value less than 100 electrons/cm3. However, below 66 km, the observations were larger than the simulations, which were, on the contrary, above 76 km.

Evaluating F2-region long-term trends using the International Reference Ionosphere (IRI) model: is this a feasible approximation for experimental trends?

Abstract. The International Reference Ionosphere (IRI) is a widely used empirical ionospheric model based on observations from a worldwide network of ionospheric stations. Therefore, it would be reasonable to expect it to capture long-term changes in key ionospheric parameters, such as foF2 and hmF2 linked to trend forcings like greenhouse gas increasing concentrations and the Earth's magnetic field secular variation. Despite the numerous reported trends in foF2 and hmF2 derived from experimental data and model results, there are inconsistencies that require continuous refinement of trend estimation methods and regular data updates. This ongoing effort is crucial to address the difficulties posed by the weak signal-to-noise ratio characteristic of ionospheric long-term trends. Furthermore, the experimental verification of these trends remains challenging, primarily due to time and spatial coverage limitations of measured data series. Achieving these needs for accurate detection of long-term trends requires extensive global coverage and high resolution of ionospheric measurements together with long enough periods spanning multiple solar cycles to properly filter out variations of shorter terms than the sought trend. Considering these challenges, IRI-modeled foF2 and hmF2 parameters offer a valuable alternative for assessing trends and obtaining a first approximation of a plausible global picture representative of experimental trends. This work presents these global trend patterns, considering the period 1960–2022 using the IRI-Plas 2020 version, which are consistent with other model predictions. While IRI explicitly takes into account the Earth's magnetic field variations, the increase in the concentrations of greenhouse gases appears indirectly through the Ionospheric Global index (IG) which is derived from ionospheric measurements. F2-region trends induced by the first mechanism should be important only around the magnetic equator at the longitudinal range with the strongest displacement, and it should be negligible out of this region. Conversely, trends induced by the greenhouse effect, which are the controversial ones, should be dominant away from the geomagnetic equator and should globally average to negative values in both cases, i.e., foF2 and hmF2. Effectively, these negative global means are verified by trends based on IRI-Plas, even though not for the correct reasons in the hmF2 case. In addition, a verification was performed for more localized foF2 trend values, considering data from nine mid-latitude stations, and a reasonable level of agreement was observed. It is concluded that the IRI model can be a valuable tool for obtaining preliminary approximations of the Earth's magnetic-field-induced long-term changes in foF2 and hmF2, as well as of experimental trends only in the foF2 case. The latter does not hold for hmF2, even if the trends obtained are close to the expected values.

Forecasting Ionospheric foF2 Using Bidirectional LSTM and Attention Mechanism

AbstractThe critical frequency of ionospheric F2 layer (foF2) is an important ionospheric characteristic parameter. In this paper, a deep learning model based on Bidirectional long short‐term memory (BiLSTM) and attention mechanism is implemented for predicting the foF2 parameter. The inputs of models are the foF2 of globally available ionospheric ionosonde stations, geographic longitude and latitude, world time (UT), geomagnetic activity index, and solar activity index from 2015 to 2017. The superiority of the model is analyzed from different latitudes, seasons, and geomagnetic conditions. The results show that the prediction performance of the Bidirectional long short‐term memory model based on attention mechanism (BiLSTM‐Attention) is better than other models. The performance of the prediction model is optimal at high latitudes. The root mean square error (RMSE) and correlation coefficient (R) of the BiLSTM‐Attention model are 0.539 MHZ and 0.908 MHz at high latitudes, respectively. In terms of RMSE, it is 25.243%, 18.209%, and 11.203% lower than those of the international reference ionosphere (IRI), LSTM, and BiLSTM models, respectively. The prediction results of the four seasons show that the models are more applicable in winter. Compared with the IRI model, the RMSE of the BiLSTM‐Attention model in spring, summer, autumn, and winter is reduced by 24.344%, 21.181%, 25.058%, and 30.948%, respectively. The prediction effect of the BiLSTM‐Attention model is improved in the magnetic quiet period, the magnetic moderate period and the magnetic storm period. Also, the improvement effect is more obvious in the magnetostatic day, and the RMSE is reduced by 27.462% compared with the IRI model.