
Machine Learning Classification of Interplanetary Coronal Mass Ejections Using Satellite Accelerometers
Abstract
Space weather phenomena is a complex area of research as there are many different variables and signatures that are used to identify the occurrence of solar storms and Interplanetary Coronal Mass Ejections (ICMEs), with inconsistencies between databases and solar storm catalogues. The identification of space weather events is important from a satellite operation point of view, as strong geomagnetic storms can cause orbit perturbations to satellites in low-earth orbit. The Disturbance storm time (Dst) and the Planetary K-index (Kp) are common indices used to identify the occurrence of geomagnetic storms caused by ICMEs, among several other signatures that are not consistent with every storm. Moreover, specific instrumentation is needed for solar storm and space weather phenomena, which can be costly and technically difficult for small and nano-satellite applications. This thesis demonstrates the capability of a new signature for identification and characterization of ICMEs, through the use of satellite accelerometer data from the Gravity Recovery and Climate Experiment (GRACE) satellite, and machine learning techniques. Utilizing pre-existing satellite instrumentation, this research proposes the use of accelerometers for future space weather monitoring applications. Four binary classification algorithms have been explored: Random Forest, Support Vector Machine, Extremely Randomized Trees, and Logistic Regression. It is proposed that a binary classification model can differentiate between a solar storm caused by an ICME versus a period of quiet geomagnetic activity, using only the accelerometer data of a satellite. Of the four architectures, the tree-based machine learning models performed the best, with accuracy scores over 80%.