Item

Abstract

The vast majority of solar cycle predictions concentrate on predicting the evolution of the sunspot number. This is largely motivated by the fact that the most severe geomagnetic storms in near-Earth space are driven by coronal mass ejections, whose evolution follows the sunspot number. However, geomagnetic activity is actually mostly driven by high-speed solar wind streams originating from solar coronal holes, which reach their peak in the declining phase of the sunspot cycle. Because of this, the solar cycle variation of geomagnetic activity is vastly different from the cycle variation of sunspots. Despite its practical importance to different manifestations of space weather, there have been surprisingly few attempts to predict the solar cycle variation of geomagnetic activity.Here we aim to predict the geomagnetic activity cycle, as depicted by the homogenized version of the geomagnetic aa index. We separate the aa index into two components: (1) its baseline corresponding to the minimum aa level at the start of each solar cycle, and (2) the remaining cyclic variation. We parameterize the cyclic component using a combination of Gaussian curves representing the distinct activity peaks separated in time. We discuss the prediction of the aa cycle model parameters and the baseline level using past values of the aa index and sunspot numbers, as well as the predicted sunspot numbers. We also estimate the performance of our method using a cross-validation methodology.