Simulation of statistical distributions in the space of parameters of the solar wind and interplanetary magnetic field using artificial neural networks

Extended statistical and cross-correlational analyses of the data time series of the parameters of the solar activity (the Wolf numbers, the radio-emission flux F10.7, and the global magnetic field of the Sun), the solar wind (velocity, density, and temperature), and the interplanetary magnetic field are performed. All the parameters involved have distributions that, as a whole, are close to the lognormal distributions. Integro-cross-correlational analysis with a floating averaging period showed that the maximum pair correlation is observed for all parameters for the averaging period of the order of one year. The attempt of complex simulation and forecast of annual smoothed values of the parameters with the use of artificial neural networks leads us to conclude that a rapid degeneration of the forecast occurs at a time scale of the order of one year. This is due to the high degree of chaotization of the distributions of statistical parameters, which are characterized by rather high informational entropy, lying usually in the range 0.8-0.9.