Global Plasmaspheric Refilling Rates Inferred From Deep‐Learning Electron Density Reconstructions

Abstract

As a key process in magnetosphere‐ionosphere coupling, plasmaspheric refilling is a long‐standing focus in space physics. In this study, we develop a global electron density (GEO‐NE) model based on a CNN‐Transformer architecture, and use the model to investigate plasmaspheric electron density refilling rates. Using global electron density maps reconstructed by the model, we perform statistical analyses of refilling rates at different temporal resolutions. The 8‐hr averaged results are consistent with many previous studies, and the median can be fitted by . We further apply our model to an approximate flux‐tube‐following analysis at 30‐min resolution, in which typical two‐stage refilling features and diurnal difference along the simulated trajectory are observed. Our approach overcomes the temporal‐resolution limitations of earlier studies and provides a new pathway for investigating multiscale plasmaspheric refilling, which is helpful for establishing the high‐resolution modeling of magnetosphere‐ionosphere coupling.