Noise Robust ML Sunspot Forecasts for Cycle Aware Space Weather Operations in the Cycle 25 Era

Noise Robust ML Sunspot Forecasts for Cycle Aware Space Weather Operations in the Cycle 25 Era

Esraa Zeki Mohammed, Ahmed Ghareeb, Mohammed Rashad Baker

Abstract. Accurate monthly solar activity forecasting is vital for satellite operations and ionospheric modeling. We developed a lightweight, interpretable machine-learning framework using the complete SILSO record (1749–2024) to predict the 13-month smoothed sunspot number (SSN) for solar cycles 20–24 and the rising phase of cycle 25. We tested linear regression, tree-based models, their stacked ensemble, and a symbolic regressor, assessing robustness by injecting Gaussian noise (σ = 0, 1, 3 SSN) in to training data. Across 75 experiments, median performance was r2 = 0.86 and RMSE< 31 SSN. The framework delivers deep-learning-class accuracy with negligible computational overhead and full transparency, offering an operationally practical tool as solar cycle 25 approaches its peak. Keywords
Solar Cycle Prediction, Machine Learning, Symbolic Regression, Stacked Ensemble, Space Weather Forecasting, Noise Robustness, Satellite Drag Mitigation

Published online 7/20/2026, 4 pages
Copyright © 2026 by the author(s)
Published under license by Materials Research Forum LLC., Millersville PA, USA

Citation: Esraa Zeki Mohammed, Ahmed Ghareeb, Mohammed Rashad Baker, Noise Robust ML Sunspot Forecasts for Cycle Aware Space Weather Operations in the Cycle 25 Era, Materials Research Proceedings, Vol. 69, pp 271-274, 2026

DOI: https://doi.org/10.21741/9781644904251-48

The article was published as article 48 of the book CEAS – AIDAA Conference 2025

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

References
[1] K. Petrovay, “Solar cycle prediction,” Living Reviews in Solar Physics, vol. 17, no. 2, 2020. https://doi.org/10.1007/s41116-020-0022-z
[2] L. A. Upton and D. H. Hathaway, “Solar Cycle Precursors and the Outlook for Cycle 25,” J. Geophys. Res.: Space Physics, vol. 128, no. 11, e2023JA031681, 2023. https://doi.org/10.1029/2023JA031681
[3] NOAA/NWS SWPC, “NOAA forecasts quicker, stronger peak of solar activity,” Dec. 19, 2023.
[4] J. J. Love, D. A. Jonas, and A. Kelbert, “Mapping a Magnetic Superstorm: March 1989 Geoelectric Hazards in the United States,” Space Weather, vol. 20, no. 5, 2022. https://doi.org/10.1029/2021SW003030.
[5] W. Nie et al., “The Mechanism for GNSS-Based Kinematic Positioning Degradation at High Latitudes Under the March 2015 Great Geomagnetic Storm,” Space Weather, vol. 20, no. 6, 2022. https://doi.org/10.1029/2022SW003132
[6] V. Lanabere, Y. Y. Podladchikova, E. I. Gordeev, A. Wintoft, and M. Palmroth, “Characterizing the distribution of extreme geoelectric field gradients in North America,” J. Space Weather Space Clim., vol. 14, 2024, Art. 12. https://doi.org/10.1051/swsc/2024012.
[7] X. Su, B. Liang, S. Feng, Y. Cai, W. Dai, and Y. Yang, “Solar cycle prediction using a combinatorial deep learning model,” Mon. Not. R. Astron. Soc., vol. 527, no. 3, pp. 5675–5682, 2024. https://doi.org/10.1093/mnras/stad3451
[8] H. Zhu, W. Zhu, and M. He, “Predicting Solar Cycle 25 Using an Optimized Long Short-Term Memory Model Based on Sunspot Area Data,” Adv. Space Res., vol. 71, no. 8, pp. 3521–3534, 2023. https://doi.org/10.1016/j.asr.2023.02.025
[9] WDC-SILSO, “Sunspot Number graphics (daily, monthly, 13-month smoothed),” accessed Sept. 2025. [Online]. Available: https://www.sidc.be/SILSO/ssngraphics
[10] J.-V. Rodríguez, V. M. Sánchez Carrasco, I. Rodríguez-Rodríguez, A. J. Pérez Aparicio, and J. M. Vaquero, “Hemispheric Sunspot Number Prediction for Solar Cycles 25 and 26 Using Spectral Analysis and Machine Learning Techniques,” Solar Physics, vol. 299, no. 116, 2024. https://doi.org/10.1007/s11207-024-02363-2
[11] D. Sierra-Porta, M. Tarazona-Alvarado, and D. D. Herrera Acevedo, “Predicting sunspot number from topological features in spectral images I: Machine learning approach,” Astronomy and Computing, vol. 48, p. 100857, 2024. https://doi.org/10.1016/j.ascom.2024.100857
[12] V. Kumar, “Enhancing Solar Cycle 25 and 26 Forecasting with Vipin-Deep-Decomposed-Recomposed Rolling-window (vD2R2w) Model on Sunspot Number Observations,” Solar Physics, vol. 299, no. 147, 2024. https://doi.org/10.1007/s11207-024-02389-6
[13] M. S. Miesch et al., “Solar Cycle Prediction at NOAA’s Space Weather Prediction Center,” Space Weather, 2025. https://doi.org/10.1029/2025SW004444