Abstract During the Van Allen Probes era, several multi-MeV (>4 MeV) electron flux enhancements were observed. The cause of electron acceleration up to multi-MeV remains an ongoing science topic. In this study, we focus on examining the relationship between phase space density (PSD) radial profile shapes and the occurrence of multi-MeV electron flux enhancement events. This will determine which process (local acceleration or radial diffusion) is dominant in producing multi-MeV electron flux enhancements at a specific L*. Growing peaks in PSD radial profiles are associated with the local acceleration (i.e., a wave-particle interaction) of multi-MeV electrons. For each growing peak in PSD, we determined the L* where the local acceleration occurs for each respective electron energy. Similarly, we also identify which PSD profiles are related to acceleration via radial diffusion profiles. Both sets of profiles are compared with the Van Allen Probe-A observed multi-MeV electron flux enhancements. Results show that both mechanisms (local acceleration and radial diffusion) can facilitate multi-MeV electron acceleration, however each mechanism has a preferable L* region where it is the dominant acceleration process.

Abstract We study the geomagnetic storm of 9 October 2012, where it had been generally accepted that the resulting prominent outer radiation belt electron acceleration throughout the storm is due to whistler-mode chorus waves. This storm has been studied previously by two-dimensional Fokker–Planck numerical simulations with data-driven quasi-linear (QL) diffusion rates. However, possible nonlinear (NL) resonant interaction effects on electron flux dynamics haven't been looked at yet. This study aims to fill this gap by demonstrating that theory-informed rescaling of QL diffusion rates accounting for contributions of NL resonant interactions helps to reproduce better observed increase of electron fluxes by diffusion simulations. We use machine learning, uncertainty quantification (UQ), physics-perturbed ensemble of VERB simulations and Van Allen Probes observations to identify optimal rescaling of quasi-linear diffusion rates.