Forecasting the Timing of Peak Solar Flare Index in Solar Cycle 25 for Improved Space Weather Prediction

Solar flares, driven by magnetic reconnection on the Sun, pose substantial threats to Earth's technological infrastructure. Accurate forecasting of flare activity, quantified by the Solar Flare Index (SFI), is crucial for mitigating space weather risks. This study leveraged an optimized long short-term memory (LSTM+) neural network, to achieve robust long-term SFI predictions. The LSTM + model incorporated a novel reprediction procedure and fine-tuned parameter optimization, demonstrating high accuracy in hindcasting SFI for Solar Cycles 23 and 24. The validated model predicted a peak SFI for Solar Cycle 25 in January 2025, aligning with historical trends of SFI lagging behind sunspot number maxima. This projection, along with the recent resurgence in sunspot activity, suggests a potential second, higher SFI peak may occur. Incorporating inherent model uncertainties, the maximum SFI for Solar Cycle 25 was estimated to occur between December 2023 and February 2026. These findings contribute to a deeper understanding of solar flare dynamics and provide valuable insights for space weather prediction, enabling proactive measures to protect critical technological systems.