An Innovative Solar Flare Metadata Collection for Space Weather Analytics

Jinsu Hong, Chetraj Pandey, Anli Ji, Berkay Aydin

Abstract: Space weather events can have a significant impact on electric systems and health, with solar flares being one of the central events in space weather forecasting. However, existing solar flare prediction tools heavily rely on the Geostationary Operational Environmental Satellites (GOES) classification system, using maximum X-ray flux measurements as proxies to label instances. This approach becomes problematic during solar minimum, where background X-ray flux fluctuations lead to false alarms and inaccurate predictions. To address this issue, we propose a new collection of solar flare intensity labels computed from GOES X-ray flux, introducing innovative labeling regimes that incorporate relative increases and cumulative measurements over prediction windows. Our goal is to improve the accuracy of flare prediction methods by reducing false positives and enhancing overall prediction performance. Throughout this paper, we introduce the concept of relative X-ray flux increase and explain how to derive relative X-ray flux increase metadata for generating new labels. Additionally, we present new cumulative indices and data-driven categorical labels designed for active regionbased and full-disk flare prediction models. We then evaluate the effectiveness of our new labels when applied to established solar flare prediction models, demonstrating that they significantly enhance prediction capabilities and complement existing efforts. With our innovative data-driven labels, we aim to enhance flare forecasting capabilities and provide more accurate and reliable predictions for space weather phenomena.