Prediction of High-Energy Electron Flux at Geosynchronous Orbit using a Neural Network Technique

The radiation belt, where the high-energy particles are predominant in near earth space from the low earth orbit (LEO) to the geostationary orbit (GEO), sometimes causes satellite malfunction.Therefore, the objective of this study is to predict the high-energy electron flux at GEO with the energy above 2 MeV after 24 hours with higher accuracy for the safety satellite operation in terms of the space weather science.In this study the various kinds of solar wind data from satellite observations and ground geomagnetic observation data in 1999 were used for the Recurrent Neural Network (RNN).Prediction results were evaluated by the prediction efficiency, which is derived from both predicted and actual variation data.As a result, the prediction using combined data of solar wind and geomagnetic data shows highest prediction efficiency of 0.72.