The Study on Modeling of Global Plasmaspheric Hiss Amplitude Based on Deep Learning Algorithm

Plasmaspheric hiss waves make great significance on the loss of electrons in Earth's radiation belts. The prediction and reconstruction for global evolution plasmaspheric are critical to investigate dynamic process belt. In this study, realistic variation model is established based deep learning technology. Using Light Gradient Boosting Machine Permutation Importance methods, we first select optimal indices corresponding historical time series associated with waves. Then geomagnetic considered as input neural network model, samples observed from Van Allen Probe A utilized training validation. result indicates that can roughly reproduce integrated wave amplitude during storm. Based analyze a storm event. addition, find performance has been greatly improved addition situ density an auxiliary parameter, it provides new idea establishing higher accuracy other characteristic parameters.