Modeling of CME and CIR driven geomagnetic storms by means of artificial neural networks

Abstract

A model of geomagnetic storms based on the method of artificial neural networks (ANN) combined with an analytical approach is presented in the paper. Two classes of geomagnetic storms, caused by coronal mass ejections (CMEs) and those caused by corotating interaction regions (CIRs), of medium and week intensity are subject to study. As the model input, the hourly solar wind parameters measured by the ACE satellite at the libration point L1 are used. The time series of the Dst index is obtained as the model output. The simulated Dst index series is compared with the corresponding observatory data. The model reliabilty is assessed using the skill scores, namely the correlation coefficient CC and the prediction efficiency PE. The results show that the model performance is better for the CME driven storms than for the CIR driven storms. At the same time, it appears that in the case of medium and weak storms the model performance is worse than in the case of intense storms.