Solar Activity Predictability Horizons

Abstract

The chaotic component in solar-activity variations, the limited amount of observational statistics, and the use of proxy data impose limitations on long-term predictions. This study uses the well-known reconstructions of total solar irradiance (TSI) from optical observations (from 1610 AD) and cosmogenic radioisotopes (from 7362 BC) to estimate the maximum Lyapunov exponent for 11-year and secular cycles. This is done with the classical algorithm described by Wolf et al. (1985). The maximum Lyapunov exponents turned out to be positive in both cases, thus confirming the presence of the chaotic component in time series. We then estimate the predictability horizon Tp with the well-known error for each reconstruction. It turned out that the predictability horizon is on average 10 ≤ Tp ≤ 22 years for 11-year cycles and on average 40 ≤ Tp ≤ 60 years for secular cycles.