Effect of solar dynamics parameters on the formation of substorm activity

Abstract

An algorithm for retrieving the AL index dynamics from the parameters of solar-wind plasma and the interplanetary magnetic field (IMF) has been developed. Along with other geoeffective parameters of the solar wind, an integral parameter in the form of the cumulative sum Σ[N*V²] is used to determine the process of substorm formation. The algorithm is incorporated into a framework developed to retrieve the AL index of an Elman-type artificial neural network (ANN) containing an additional layer of neurons that provides an “internal memory” of the prehistory of the dynamical process to be retrieved. The ANN is trained on data of 70 eight-hour-long time intervals, including the periods of isolated magnetospheric substorms. The efficiency of this approach is demonstrated by numerical neural-network experiments on retrieving the dynamics of the AL index from the of solar wind and IMF parameters during substorms.