Analysis of Geoeffective Impulsive Events on the Sun During the First Half of Solar Cycle 24

Abstract

A coronal mass ejection (CME) is an impulsive event that emerges rapidly from the Sun. We observed a quiet Sun without many spectacular episodes during the last decade. Although some fast halo and partial halo CMEs had taken place, among them was the backside CME on 23 July 2012. In this work, we verify the link between the variability of solar-wind, heliospheric and geomagnetic parameters and the transmission grid failures registered in southern Poland during 2010 – 2014 when many geomagnetic storms appeared, caused by halo and partial halo CMEs. We aim to apply three machine learning methods: Principal Components Analysis, Self-Organizing Maps, and Hierarchical Agglomerative Clustering to analyze sources on the Sun and the impacts of the intense geomagnetic storms in the first half of Solar Cycle 24. The conducted analyzes underline the importance of solar-wind proton temperature and point out other solar-wind and geomagnetic parameters independently indicated by all the methods used in this study.

Review of Geomagnetically Induced Current Proxies in Mid-Latitude European Countries

“…A hypothesis was formulated that failures from Clusters ( 4)-( 6) may have their source in sudden changes in the geomagnetosphere caused by a short-term variability of the Sun [16]. The authors used various methods of statistical inference [16,34,50], time series analysis [16,34], machine learning, such as principal component analysis (PCA) [50], neural networks (self-organizing maps (SOMs)) [34,50], and hierarchical agglomerative clustering (HAC) [50]. The regression analysis showed that the average annual number of failures from Groups (4)- (6) in the period from January 2010 (around the solar activity minimum) to July 2014 (solar maximum) displayed an upward trend.…”

“…However, other solar wind parameters or heliosphere characteristics also have an affect [34]. Gil et al [50], using cross-correlation, revealed particular delays in the increase of the power grid failures and changes in the geomagnetic, heliospheric, and solar parameters. Those delays were taken into account in the PCA, SOM, and HAC analyses.…”

“…Studies of the situation in Poland were based on several machine learning methods, such as principal component analysis, neural networks, and hierarchical agglomerative clustering. The authors showed that the solar wind parameters have the most-substantial connections with all the considered types of electric grid failures [50]. Section 4 contains the summary of the presented study.…”