A multifactor analysis of parameters controlling solar wind ion flux correlations using an artificial neural network technique

Abstract

Solar wind plasma and magnetic field observations from multiple spacecraft can be used to separate temporal and spatial variations and to determine the accuracy of predictions of solar wind conditions near Earth based on distant-spacecraft measurements. The study of correlations between the ion fluxes measured by three spatially separated spacecraft (IMP 8, WIND and INTERBALL-1) was one of the first steps in this direction. This paper describes a complex multifactor analysis of different physical, geometrical, and statistical parameters that control such correlations (considered separately and in combination). A linear-regression and an artificial neural network techniques are used for this analysis. The analysis is applied to an extensive array of correlation coefficients for the ion flux in the solar wind and provides estimates of the relative significance of the factors that control these correlation coefficients. The study shows that the most influential parameters are the solar wind density and the standard deviations of solar wind density, solar wind velocity and interplanetary magnetic field. This set of parameters permits us to develop a sufficiently accurate (with a relative error of less than a few per cent) quantitative model for the correlation between the ion fluxes measured on two spatially separated spacecraft.