Feature Reduction on Sym-H Index Image Using Principal Component Analysis Approach

Abstract

Geomagnetic storms pose significant risks to technological systems on Earth. One of the ways to identify the level of a storm is from the Sym-H plot images. The fewer features used for image interpretation, the simpler and more efficient the analysis becomes. In this study, we applied Principal Component Analysis (PCA) to the Sym-H index images, initially consisting of seven statistical features. Through PCA, this study managed to reduce these features to just two principal components, capturing over 98% of the total variance in the first two components, thereby retaining essential information while simplifying the dataset. This reduction not only simplifies the visualization and interpretation of the Sym-H plot images but also retains the critical information necessary for understanding geomagnetic storm dynamics. By focusing on these two principal components, we can effectively present and analyse the essential patterns and behaviours of geomagnetic activity during storm events. The findings highlight the potential of PCA to enhance space weather forecasting and improve the resilience of technological infrastructure against solar storm impacts.