Machine Learning–Driven Prediction of the S4 Scintillation Index and Geomagnetic Storms using GNSS-RO Data

Fattahi, Delaram; Barzegar, Sahar; Khoshsima, Masoud; Karami, Saber; Jalili, Hadi; Aghayari, Jamal

doi:10.22034/jsst.2025.1585

Machine Learning–Driven Prediction of the S4 Scintillation Index and Geomagnetic Storms using GNSS-RO Data

Document Type : Original Research Paper

Authors

Delaram Fattahi ¹

Sahar Barzegar ²

Masoud Khoshsima ³

Saber Karami ³

Hadi Jalili ³

Jamal Aghayari ³

¹ Space Physics group, Institute of Geophysics, University of Tehran, Tehran, Iran

² Space Physics Group, Institute of Geophysics, University of Tehran, Tehran, Iran

³ Iranian Space Research Center, Tehran, Iran

10.22034/jsst.2025.1585

Abstract

This study applies the Random Forest (RF) machine learning (ML) algorithm to GNSS radio occultation (GNSS-RO) data for two critical space weather forecasting tasks: the prediction of ionospheric amplitude scintillation (S4 index) and the classification of geomagnetic storm occurrence. For S4 index prediction, an RF regression model was trained on a comprehensive feature set derived from interplanetary magnetic field components, geomagnetic indices, solar radio flux, and historical S4 statistics. The model's architecture is optimized, and Recursive Feature Elimination with Gini Importance (Mean Decrease Impurity) method is applied to identify the most predictive features. For storm detection, an RF classifier was trained on parameters including Total Electron Content (TEC), S4 indices, and key solar and geomagnetic variables to distinguish between storm and non-storm conditions. The optimized S4 prediction model achieved high precision with a Mean Absolute Error of 0.007263 and an R² score of 0.517352. Feature selection via Gini Importance significantly improved model efficiency, increasing the Adjusted R² by 50.5%. The geomagnetic storm classifier demonstrated a critical strength of recall of 0.99 for storm events, ensuring missing only 1% of actual storms. Analysis of feature importance confirmed the model's physical validity, correctly identifying the Kp index, F10.7 solar flux and Dst index as the primary drivers for prediction, which aligns with established solar-terrestrial physics. The results demonstrate the high potential of machine learning, specifically RFs, for precise space weather forecasting

Keywords

Machine Learning

Random Forest

Scintillation Index

GNSS RadioOccultation

Geomagnetic Storm

Subjects

space weather prediction

[1] S. Taran, N. Alipour, K. Rokni, S. H. Hosseini, O. Shekoofa, and H. Safari. “Effect of geomagnetic storms on a power network at mid latitudes,” Advances in Space Research, vol. 71, no. 12, pp. 5453-5465, 2023, https://doi.org/10.1016/j.asr.2023.02.027.

[2] E. Saiz et al., “Geomagnetic response to solar and interplanetary disturbances,” Journal of Space Weather and Space Climate, vol. 3, 2013, Art. no. A26, https://doi.org/10.1051/swsc/2013048.

[3] S. Priyadarshi, “A review of ionospheric scintillation models,” Surveys in geophysics, vol. 36, no. 2, pp. 295–324, 2015, https://doi.org/10.1007/s10712-015-9319-1.

[4] J. Vil`a-Valls, P. Closas, C. Fern´andez-Prades, and J. T. Curran, “On the mitigation of ionospheric scintillation in advanced gnss receivers,” IEEE Transactions on Aerospace and Electronic Systems, vol. 54, no. 4, pp. 1692–1708, 2018, https://doi.org/10.1109/TAES.2018.2798480.

[5] G. DeFranceschi, L. Spogli, L. Alfonsi, V. Romano, C. Cesaroni, and I. Hunstad, “The ionospheric irregularities climatology over Svalbard from solar cycle 23,” Scientific Reports, vol.9, 2019, Art. no. 9232, https://doi.org/10.1038/s41598-019-44829-5.

[6] V. Sreeja, “Impact and mitigation of space weather effects on GNSS receiver performance,” Geoscience letters, vol. 3, no. 1, 2016, Art. no. 24, https://doi.org/10.1186/s40562-016-0057-0.

[7] ESA, “Space weather study results,” 2016. [Online]. Available: https://esamultimedia.esa.int/docs

[8] M. A. Kelly, J. M. Comberiate, E. S. Miller, and L. J. Paxton, “Progress toward forecasting of space weather effects on UHF SATCOM after operation Anaconda,” Space Weather, vol. 12, no. 10, pp. 601–611, 2014, https://doi.org/10.1002/2014SW001081.

[9] T. E. Humphreys, M. L. Psiaki, B. M. Ledvina, A. P. Cerruti, and P. M. Kintner, “Data-driven testbed for evaluating gps carrier tracking loops in ionospheric scintillation,” IEEE Transactions on Aerospace and Electronic Systems, vol. 46, no. 4, pp. 1609–1623, 2010, https://doi.org/10.1109/TAES.2010.5595582.

[10] Y. U. Jiao, D. Xu, C. L. Rino, Y. T. Morton, and C. S. Carrano, “A multifrequency GPS signal strong equatorial ionospheric scintil lation simulator: Algorithm, performance, and characterization,” IEEE Transactions on Aerospace and Electronic Systems, vol. 54, no. 4, pp. 1947 1965, 2018, https://doi.org/10.1109/TAES.2018.2805232.

[11] A. D. O. Moraes, E. Costa, E. R. De Paula, W. J. Perrella, and J. F. G. Monico, “Extended ionospheric amplitude scintillation model for GPS receivers,” Radio Science, vol. 49, no. 5, pp. 315–329, 2014, https://doi.org/10.1002/2013RS005307.

[12] J. A. Klobuchar, “Ionospheric time-delay algorithm for single-frequency GPS users,” IEEE Transactions on Aerospace and Electronic Systems, vol. 3, pp. 325–331, 1987, https://doi.org/10.1109/TAES.1987.310829.

[13] R. Prieto-Cerdeira, R. Orús-Pérez, E. Breeuwer, R. Lucas-Rodriguez, and M. Falcone, “Performance of the Galileo single-frequency ionospheric correction during in-orbit validation,” GPS World, vol. 25, no. 6, pp. 53–58, 2014.

[14] Y. Yuan, N. Wang, Z. Li, and X. Huo, “The BeiDou global broadcast ionospheric delay correction model (BDGIM) and its preliminary performance evaluation results,” Navigation, vol. 66, no. 1, pp. 55–69, 2019, https://doi.org/10.1002/navi.292.

[15] S. Basu, K. M. Groves, S. Basu, and P. Sultan, “Specification and forecasting of scintillations in communication/navigation links: Current status and future plans,” Journal of atmospheric and solar-terrestrial physics, vol. 64, no. 16, pp. 1745–1754, 2002, https://doi.org/10.1016/S1364-6826(02)00124-4.

[16] K. M. Groves et al., “Equatorial scintillation and systems support,” Radio Science, vol. 32, no. 5, pp. 2047–2064, 1997, https://doi.org/10.1029/97RS00836.

[17] S. Aol, S. Buchert, and E. Jurua, “Ionospheric irregularities and scintillations: A direct comparison of in situ density observations with ground-based L-band receivers,” Earth Planets Space, vol. 72, no. 1, 2020, Art. no. 164, https://doi.org/10.1186/s40623-020-01294-z.

[18] J. A. Pappoe, A. Yoshikawa, A. Kandil, A. Mahrous, “A machine learning approach combined with wavelet analysis for automatic detection of Pc5 geomagnetic pulsations observed at geostationary orbits,” Advances in Space Research, vol. 74, no. 12, pp. 6277-6291, 2024, https://doi.org/10.1016/j.asr.2023.11.001.

[19] J. A. Pappoe, Y. Yoshikawa, A. Kandil, and A. Mahrous, “Machine learning techniques for estimation of Pc5 geomagnetic pulsations observed at geostationary orbits during solar cycle 23,” Journal of Atmospheric and Solar-Terrestrial Physics, vol. 260, 2024, Art. no. 106258, https://doi.org/10.1016/j.jastp.2024.106258.

[20] H. Farooki et al., “A machine learning approach to understanding the physical properties of magnetic flux ropes in the solar wind at 1 au,” The Astrophysical Journal, vol. 961, no.1, 2024, Art. no. 81, https://doi.org/10.3847/1538-4357/ad0c52.

[21] S. Halawa, M. Alansaari, M. Sharif, A. Alhammadi, and I. Fernini, “Ionospheric scintillation forecasting using machine learning,” in IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium, Athens, Greece, 2024, pp. 5635-5639.

[22] S. Nasurudiin, A. Yoshikawa, A. Elsaid, and A. Mahrous, “Prediction of ionospheric scintillations using machine learning techniques during solar cycle 24 across the equatorial anomaly,” Atmosphere, vol. 15, no. 10, 2024, Art. no. 1213, https://doi.org/10.3390/atmos15101213.

[23] Y. H. Zhan, Z. Y. Xing, T. Xu, Q. H. Zhang, F. Oiao, and Y. Wang, “Prediction of Northern Polar ionospheric scintillation Using the RF and LSTM machine learning methods,” in 14^th International Symposium on Antennas, Propagation and EM Theory (ISAPE), Hefei, China, 2024, pp. 1-4.

[24] S. Trachuentong, P. Supnithi, L. Min Min Myint, and S. Saito, “Ionospheric scintillation prediction using decision tree and rainforest techniques,” in 10^th International Conference on Engineering, Applied Sciences, and Technology (ICEAST), Louang Phabang, Laos, 2024, pp. 89-92, https://doi.org/10.1109/ICEAST61342.2024.10553652.

[25] T. Dittmann, Y. Jade Morton, and H. Chang, “Machine learning classification of ionosphere and RFI disturbances in spaceborne GNSS GNSS-RO measurements,” in International Technical Meeting of The Institute of Navigation, Long Beach, California, 2025, pp. 161-176, https://doi.org/10.33012/2025.20007.

[26] L. Kuruva, M. R. Avula, and D. S. Achanta, “ML based model for detection of ionospheric scintillations using PolaRx5S data,” Space, Aerospace and Defence Conference (SPACE), Bangalore, India, 2024, pp. 678-681, https://doi.org/10.1109/SPACE63117.2024.10668168.

[27] A. Das, A. Das Gupta, and S. Ray, “Characteristics of L-band (1.5 GHz) and VHF (244 MHz) amplitude scintillations recorded at Kolkata during 1996–2006 and development of models for the occurrence probability of scintillations using neural network,” Journal of Atmospheric and Solar-Terrestrial Physics, vol. 72, no. 9-10, pp. 685–704, 2010, https://doi.org/10.1016/j.jastp.2010.03.010.

[28] L. F. C. Rezende et al., “Survey and prediction of the ion-ospheric scintillation using data mining techniques,” Space Weather, vol. 8, no. 6, 2010, Art. no. S06D09, https://doi.org/10.1029/2009SW000532.

[29] O. Carvalho, P. A. A. da Silva de Almeida Nava Alves, R. Y. C. Cueva, and A. B. Filho, “Nowcasting of amplitude ionospheric scintillation based on machine learning techniques,” in IEEE Transactions on Aerospace and Electronic Systems, vol. 58, no. 6, pp. 4917-4927, Dec. 2022, https://doi.org/10.1109/TAES.2022.3188741.

[30] A. M. Darya, A. A. Al-Owais, M. M. Shaikh, and I. Fernini, “Amplitude scintillation forecasting using bagged trees,” in IGARSS 2022-2022 IEEE International Geoscience and Remote Sensing Symposium, Kuala Lumpur, Malaysia, 2022, pp. 2275-2278, https://doi.org/10.1109/IGARSS46834.2022.9883380.

[31] A. Atabati, I. Jazireeyan, M. Alizadeh, and R. B. Langley, “Prediction of ionospheric scintillation with ConvGRU networks using GNSS ground-based data across South America,” Remote Sensing, vol. 16, no, 15, 2024, Art. no. 2757, https://doi.org/10.3390/rs16152757.

[32] P. Tian, B. Yu, H. Ye, X. Xue, J. Wu, and T. Chen, “Estimation model of global ionospheric irregularities: An artificial intelligence approach,” Space Weather, vol. 20, no. 9, 2022, Art. no. e2022SW003160, https://doi.org/10.1029/2022SW003160.

[33] P. Tian, B. Yu, H. Ye, X. Xue, J. Wu, and T. Chen, “Deep learning insights into ionospheric Sporadic E irregularities under different solar activity conditions,” Journal of Geophysical Research: Machine Learning and Computation, vol. 1, no. 4, 2024, Art. no. e2024JH000279, https://doi.org/10.1029/2024JH000279.

[34] G. Biau and E. Scornet, “A random forest guided tour,” TEST, vol. 25, no. 2, 2016, pp. 197-227, https://doi.org/10.1007/s11749-016-0481-7.

[35] M. Büyükkeçeci and M. C. Okur, “A comprehensive review of feature selection and feature selection stability in machine learning,” Gazi University Journal of Science, vol. 36, no. 4, pp. 1506-1520, 2023, https://doi.org/10.35378/gujs.993763.

[36] T. Mueller, A. Segin, C. Weigand, and R. Schmitt, “Feature selection for measurement models,” International Journal of Quality & Reliability Management, vol. 40, no. 3, pp. 777-800, 2023, https://doi.org/10.1108/IJQRM-07-2021-0245.

[37] R. Dunne et al., “Thresholding Gini variable importance with a single-trained random forest: An empirical Bayes approach,” Computational and Structural Biotechnology Journal, vol. 21, pp. 4354-4360, 2023, https://doi.org/10.1016/j.csbj.2023.08.033.

[38] N. G. Pardeshi and D. V. Patil, “Applying Gini importance and RFE methods for feature selection in shallow learning models for implementing effective intrusion detection system,” in International Conference on Applications of Machine Intelligence and Data Analytics (ICAMIDA), Aurangabad, India, 2022, https://doi.org/10.2991/978-94-6463-136-4_21.