Predicting the Energy Spectra of Solar Energetic Particles with a Machine Learning Regression Algorithm

Abstract

Solar energetic particles (SEPs) are a major source of space radiation, especially within the inner heliosphere. These particles, originating from solar flares and coronal mass ejections (CMEs), propagate primarily along interplanetary magnetic fields. The energy spectra of SEP events are crucial for assessing radiation effects and understanding the acceleration and propagation mechanisms in their source regions. In this study, we employed a decision tree regression algorithm with cost complexity pruning to predict SEP energy spectra, including peak flux and integral fluence spectra. This approach uses only solar flares, CMEs, and solar wind data as input parameters and demonstrates strong performance to accurately predict SEP spectra. This method holds significant real-time application value for monitoring and forecasting radiation risks in both deep space and near-Earth environments.