Abstract
The singular spectrum analysis (SSA) is a relatively new method for decomposing time series into separable and interpretable series (i.e., trend, periodic, and irregular components). Here, the ionospheric variability in the peak electron density (NmF2) observations provided by FORMOSAT-3/Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC) radio occultation measurements during 2009–2019, which completely covers solar-cycle 24, was examined using the nonparametric method (SSA). This study identified different temporal patterns inherent in the COSMIC data, which can be employed to reconstruct/model noise-free time series. Based on a proper choice of SSA parameters, it was observed that SSA can be employed to extract various components of these datasets. Additionally, the results showed that the slowly varying component (trend) was largely influenced by solar activity and closely followed the behavior of the sunspot cycle. The SSA method demonstrated seasonal patterns (the semiannual component). Based on the patterns, the reconstructed series for each of the selected components were separately modeled with a corresponding fit function using the 10.7 cm solar proxy, which prior be smoothed by Spline fitting. It was observed that the polynomial function efficiently modeled the trend component with R2 = 0.95, while the harmonic function adequately represented the semiannual component with R2 = 0.97. To validate the proposed model, the data were estimated for the time interval from January 2020 to August 2022, which represented the ascending phase of the ongoing solar-cycle 25. The results revealed that the modeled data correlated with the observed COSMIC-2 data with a high correlation coefficient (0.95). These findings indicate that the underlying model can effectively describe the behavior of the electron density during the investigated period, and the SSA method is particularly powerful for analyzing nonstationary time series and capable of identifying sub-time series (trends along with seasonal periodic components).
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have