Prediction of Solar Cycle 25 Using Simplex Projection Method: A Long-Term Analysis Based on F10.7


Creative Commons License

Gerçeker K., Kılçık A.

Solar Influences on the Magnetosphere, Ionosphere and Atmosphere, Burgas, Bulgaristan, 3 - 07 Haziran 2024, ss.1-8

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Burgas
  • Basıldığı Ülke: Bulgaristan
  • Sayfa Sayıları: ss.1-8
  • Akdeniz Üniversitesi Adresli: Evet

Özet

This study aims to predict the amplitude and time of solar activity for the solar cycle 25 by applying the Simplex Projection method to the 27-day averaged 10.7 cm Solar Radio Flux (F10.7) data. For this purpose, F10.7 data from 1963 to 2019 (up to the end of Solar Cycle 24) were used, and the current values of the Solar Cycle 25 were excluded from the dataset. Consequently, the dataset consists of 746 points (approximately 700 months), with the first 460 points selected as the library set and the remaining 286 points as the prediction set. Simplex Projection is a method that analyzes the complexity and chaotic properties of dynamic systems and can generate predictions. Subsequently, it compares the data points in the prediction set with the library set for each embedding dimension, identifying similar points and producing forward predictions (i.e., similarities) according to the desired time to prediction (Tp). To obtain better predictions and partially eliminate unsuccessful ones, the Mean Absolute Error (MAE) was calculated between the observed first 71 data points of the current cycle (Solar Cycle 25). Predictions with an error of less than 15% were considered successful predictions. When
evaluating successful predictions, a double-peaked pattern is expected for Solar Cycle 25. Accordingly, the first peak of solar activity for Solar Cycle 25 was observed around mid-2023, with the second peak, stronger than the first, anticipated to occur in early 2025 (January 2025), and the minimum of the next cycle expected in the middle or later stages of the year 2030 (July 2030). Finally, the effect of the initial conditions of the library set and the prediction set on the prediction performance is revealed.