Simulation of statistical distributions in the space of parameters of the solar wind and interplanetary magnetic field using artificial neural networks
Is. Veselovskii et al., Simulation of statistical distributions in the space of parameters of the solar wind and interplanetary magnetic field using artificial neural networks, SOL SYST R, 34(2), 2000, pp. 116-123
Extended statistical and cross-correlational analyses of the data time seri
es of the parameters of the solar activity (the Wolf numbers, the radio-emi
ssion flux F10.7, and the global magnetic field of the Sun), the solar wind
(velocity, density, and temperature), and the interplanetary magnetic fiel
d are performed. All the parameters involved have distributions that, as a
whole, are close to the lognormal distributions. Integro-cross-correlationa
l analysis with a floating averaging period showed that the maximum pair co
rrelation is observed For all parameters for the averaging period of the or
der of one year. The attempt of complex simulation and forecast of annual s
moothed 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 cha
otization of the distributions of statistical parameters, which are charact
erized by rather high informational entropy, lying usually in the range 0.8
-0.9.