Prominent advances in the field of artificial intelligence during the past decade and the breakthrough of deep learning would be useful for investigating ionospheric weather using ground and space-based ionospheric sensors data. The significance of deep learning algorithms needs to be assessed in forecasting the low latitude ionospheric disturbances (delays) for the global positioning system (GPS) signals. Total electron content (TEC) data sets prepared by taking advantage of GPS satellite radio frequency (RF) signals. This letter provides the application of deep learning models, long short-term memory (LSTM), gated recurrent unit (GRU), and a hybrid model that consists of LSTM combined with convolution neural network (CNN) to forecast the ionospheric delays for GPS signals. The deep learning models implemented using the vertical TEC (VTEC) time-series data estimated from GPS measurements over Bengaluru, Guntur, and Lucknow GPS stations. The LSTM-CNN model performs well when compared to other ionospheric deep learning forecasting algorithms with minimum root-mean-square error (RMSE) of 1.5 TEC units (TECUs) and a high degree of $R^{2} = 0.99$ .

Implementation of Hybrid Deep Learning Model (LSTM-CNN) for Ionospheric TEC Forecasting Using GPS Data

Long Short-Term Memory and Gated Recurrent Neural Networks to Predict the Ionospheric Vertical total electron Content

The global ionosphere map (GIM) is not capable of serving precise positioning and navigation for single frequency receivers in Australia due to sparse International GNSS Service (IGS) stations located in the vast land. This study proposes an approach to represent Australian total electron content (TEC) using the spherical cap harmonic analysis (SCHA) and artificial neural network (ANN). The new Australian TEC maps are released with an interval of 15 min for longitude and latitude in 0.5° × 0.5°. The validation results show that the Australian Ionospheric Maps (AIMs) well represent the hourly and seasonally ionospheric electrodynamic features over the Australian continent; the accuracy of the AIMs improves remarkably compared to the GIM and the model built only by the SCHA. The residual of the AIM is inversely proportional to the level of solar radiation. During the equinoxes and solstices in a solar minimum year, the residuals are 2.16, 1.57, 1.68, and 1.98 total electron content units (TECUs, 1 TECU = 1016 electron/m2), respectively. Furthermore, the AIM has a strong capability in capturing the adequate electrodynamic evolutions of the traveling ionospheric disturbances under severe geomagnetic storms. The results demonstrate that the ANN-aided SCHA method is an effective approach for mapping and investigating the TEC maps over Australia.

Modeling Australian TEC Maps Using Long-Term Observations of Australian Regional GPS Network by Artificial Neural Network-Aided Spherical Cap Harmonic Analysis Approach

In this paper, the prediction model for ionospheric total electron content (TEC) based on Long Short-Term Memory (LSTM) deep learning network and its performance are discussed. The input parameters of the model are previous values of daily TEC, solar radio flux at 10.7 cm parameter of 81 day moving average (
 $\overline{F107\_81}$
 ), sunspot number (SSN), geomagnetic Kp index, and disturbance storm time (Dst) index, and the outputs are TEC values for the target day. TEC data from January 1, 2001 to December 31, 2016 were used in this study. The dataset almost covers most of the years of the last two solar cycles (23, 24), and it is separated as 81.3% for training, 6.2% for validation, and 12.5% for testing. At BJFS IGS station (39.61° N, 115.89° E), LSTM yielded good TEC estimates with an RMSE of 4.07 TECU in 2001, it was 33% and 48% lower than the RMSE observed in TEC prediction using BP and IRI-2016 models, respectively. In the year of low solar activity (2016), the RMSE predicted by LSTM was 1.78 TECU, it provided 30% and 54% lower RMSE for TEC prediction than for BP and IRI-2016 models. Under the condition of magnetic storm, the LSTM TEC predictions are more consistent with the corresponding IGS Global Ionospheric Maps (GIMs) TEC than TEC predictions by BP and IRI-2016 models. LSTM can better grasp the influence of different external conditions on TEC. Seventeen grid points along 120° E meridian in latitude range from 80° S to 80° N were selected to further study the performance of LSTM model in different latitude. Results show that the prediction accuracy of LSTM is better than that of BP at different latitudes, especially at low latitudes. The performances of the two models are highly correlated with latitude and solar activity, and are both better than that of IRI-2016.

Ionospheric TEC prediction using Long Short-Term Memory deep learning network

The ionosphere plays an important role in satellite navigation, radio communication, and space weather prediction. However, it is still a challenging mission to develop a model with high predictability that captures the horizontal-vertical features of ionospheric electrodynamics. In this study, multiple observations during 2005–2019 from space-borne global navigation satellite system (GNSS) radio occultation (RO) systems (COSMIC and FY-3C) and the Digisonde Global Ionosphere Radio Observatory are utilized to develop a completely global ionospheric three-dimensional electron density model based on an artificial neural network, namely ANN-TDD. The correlation coefficients of the predicted profiles all exceed 0.96 for the training, validation and test datasets, and the minimum root-mean-square error of the predicted residuals is 7.8 × 104 el/cm3. Under quiet space weather, the predicted accuracy of the ANN-TDD is 30%–60% higher than the IRI-2016 at the Millstone Hill and Jicamarca incoherent scatter radars. However, the ANN-TDD is less capable of predicting ionospheric dynamic evolution under severe geomagnetic storms compared to the IRI-2016 with the STORM option activated. Additionally, the ANN-TDD successfully reproduces the large-scale horizontal-vertical ionospheric electrodynamic features, including seasonal variation and hemispheric asymmetries. These features agree well with the structure revealed by the RO profiles derived from the FORMOSAT/COSMIC-2 mission. Furthermore, the ANN-TDD successfully captures the prominent regional ionospheric patterns, including the equatorial ionization anomaly, Weddell Sea anomaly and mid-latitude summer nighttime anomaly. The new model is expected to play an important role in the application of GNSS navigation and in the explanation of the physical mechanisms involved.

/pdf/application-of-a-multi-layer-artificial-neural-network-in-a-54jk3lydnu.pdf

Application of a Multi-Layer Artificial Neural Network in a 3-D Global Electron Density Model Using the Long-Term Observations of COSMIC, Fengyun-3C, and Digisonde

D. SreenivasaRao

Papers

Performance evaluation of neural network TEC forecasting models over equatorial low-latitude Indian GNSS station