Advancing Machine Learning for Space Weather Forecasting: Integrating Solar Observations, Geomagnetic Storm Prediction, and GNSS Ionospheric Forecasting

Space weather forecasting remains a major challenge in heliophysics, as geomagnetic storms continue to pose significant risks to satellite operations, power grids, and global navigation satellite system (GNSS)-based positioning. The complex and nonlinear coupling between solar activity and Earth's magnetosphere-ionosphere system limits the accuracy of traditional physics-based models. Ionospheric total electron content (TEC) irregularities, often induced by geomagnetic storms, degrade GNSS signal quality, underlining the critical need for ionospheric modeling in space weather impact assessment. The rapid growth of deep learning and other machine learning (ML) methods has created major opportunities for data-driven forecasting, where solar remote sensing data from missions such as the Solar Dynamics Observatory (SDO) and GNSS-based ionospheric observations from ground and space networks can be jointly analyzed. Recent advances in physics-informed and physics-augmented ML frameworks promise more interpretable and operationally relevant predictions, yet challenges remain related to explainability, generalization, and the incorporation of physical constraints into ML models. [JS1.1]This Research Topic aims to advance accurate and timely forecasting of geomagnetic storms, ionospheric dynamics, and broader space weather phenomena through the integration of machine learning and physics-based approaches. The objectives include bridging solar physics, heliospheric science, GNSS remote sensing, and artificial intelligence to improve operational space weather forecasting capabilities. Specific goals include: (1) demonstrating the value of solar EUV and X-ray imagery, in situ solar wind measurements, and GNSS-derived ionospheric data as inputs for ML-based forecast systems; (2) developing physics-informed machine learning frameworks that embed domain knowledge into their structure; (3) extending prediction horizons for geomagnetic indices such as Dst and Kp; and (4) advancing four-dimensional (4D) ionospheric electron density and TEC modeling for GNSS positioning accuracy and satellite drag applications. This Research Topic seeks to accelerate progress toward reliable, end-to-end AI-driven pipelines for space weather research and operations.This Research Topic focuses on the intersection of heliophysics, ionospheric science, and artificial intelligence, emphasizing both methodological innovation and operational application. We particularly encourage interdisciplinary submissions coupling physics-based constraints with data-driven learning, leveraging GNSS data as both an ionospheric sounding tool and a validation benchmark. To gather further insights into data-driven and hybrid modeling of Sun-Earth intersections, we welcome original research articles, reviews, and methods articles addressing, but not limited to, the following themes: (1) Deep learning architectures (LSTM, transformers, CNNs) for geomagnetic storm and Dst/Kp forecasting(2) Solar EUV and X-ray image-driven prediction models (3) Physics-augmented ML pipelines for space weather (4) 4D ionospheric electron density and TEC modeling using GNSS observations (5) GNSS signal degradation prediction during geomagnetic storms (6) Prediction of solar energetic particle events, solar flares, and coronal mass ejections (7) Uncertainty quantification in AI-based space weather forecasts (8) Operational integration of AI-driven models within space weather service frameworks. Topic editors: Andong Hu, Yu Hong, Kefei Zhang Source: https://www.frontiersin.org/research-topics/81735/advancing-machine-learning-for-space-weather-forecasting-integrating-solar-observations-geomagnetic-storm-prediction-and-gnss-ionospheric-forecasting