The EMERALD Model for the Estimation of the Radial Diffusion Coefficients in the Outer Belt

Abstract

Abstract Radial diffusion is one of the dominant physical mechanisms driving acceleration and loss of electrons in the outer radiation belt. Therefore, the accurate estimation of radial diffusion coefficients (DLL) is crucial for detailed radiation belt modeling. In recent years several semi‐empirical (SE) models have been developed for the estimation of radial diffusion coefficients which predominantly rely on parameterizations of the Kp index. However, several studies have suggested that the estimations derived from such models can have large deviations from actual (measurement derived) DLL values. In this work we have used the extensive DLL database created in the framework of the Horizon 2020 SafeSpace project which spans 9 years of hourly DLL calculations to develop a model which uses solely solar wind parameters for the derivation of DLL values. The Electric and MagnEtic RAdiaL Diffusion (EMERALD) model is able to derive simultaneously the magnetic and electric components (DLLB and DLLE, respectively) of the radial diffusion coefficient, and furthermore, provide realistic confidence levels on their estimation, which allows the transition from a deterministic paradigm to a robust probabilistic one. Evaluations on the performance of the EMERALD model are shown by comparing its outputs to the DLL data, and examining the reproduction of various DLL characteristics. Finally, comparisons with widely used SE models are shown and discussed.

Keywords

• neural networks
• radial diffusion
• ULF waves
• radiation belts
• solar wind