Predictive ARIMA Model for coronal index solar cyclic data

Abstract

Many solar activities, e.g. sunspots, Mg II indices, coronal index of solar activity, etc., are produced by solar magnetic fields that affect both, the solar atmosphere and heliosphere, including the terrestrial environment. The atmosphere of the sun is recognized by photosphere, as a visible solar disk and layers above it: chromosphere, transition zone and corona. The nature of solar activity can be, in the emission solar corona, recognized by the coronal index of solar activity (CI). Its dataset in the period 1944–2009 will be used to test the ARIMA model. Based on the minimum AIC and BIC values, ARIMA(1,0,1) model is selected for each CI cycle and total dataset. The quality of the selected model is endorsed by (RMSE, MAPE, & FPE). The behaviour of the observed (original) CI cycle with predictive CI cycle shows a strong correlation, of more than 0.8 between actual and predictive CI cycles, which is a predictive power of ARIMA(1,0,1) model. Further model persistency estimated by Hurst exponent (smoothness) with the help of fractal dimension (complexity). All predictive CI cycles show persistence and positively correlated, the total dataset equation is given as X(t) – 0.977(t-1) = Z(t) - 0.546(t-1). • The purpose of the paper to understand the search of model for coronal index solar data. • By using of model predict the CI cycles. • Check the appropriateness of predicted cycle with original cycle by selected model. • After all persistency of the original and predictive data obtained by Hurst & fractal dimension.