Properties of relativistic electron precipitation: a comparative analysis of wave-induced and field line curvature scattering processes

We analyze the properties of relativistic (>700 keV) electron precipitation (REP) events measured by the low-Earth-orbit (LEO) POES/MetOp constellation of spacecraft from 2012 through 2023. Leveraging the different profiles of REP observed at LEO, we associate each event with its possible driver: waves or field line curvature scattering (FLCS). While waves typically precipitate electrons in a localized radial region within the outer radiation belt, FLCS drives energy-dependent precipitation at the edge of the belt. Wave-driven REP is detected at any MLT sector and L shell, with FLCS-driven REP occurring only over the nightside–a region where field line stretching is frequent. Wave-driven REP is broader in radial extent on the dayside and accompanied by proton precipitation over 03–23 MLT, either isolated or without a clear energy-dependent pattern, possibly implying that electromagnetic ion cyclotron (EMIC) waves are the primary driver. Across midnight, both wave-driven and FLCS-driven REP occur poleward of the proton isotropic boundary. On average, waves precipitate a higher flux of >700 keV electrons than FLCS. Both contribute to energy deposition into the atmosphere, estimated of a few MW. REP is more associated with substorm activity than storms, with FLCS-driven REP and wave-driven REP at low L shells occurring most often during strong activity (SML* < −600 nT). A preliminary analysis of the Solar Wind (SW) properties before the observed REP indicates a more sustained (∼5 h) dayside reconnection for FLCS-driven REP than for wave-driven REP (∼3 h). The magnetosphere appears more compressed during wave-driven REP, while FLCS-driven REP is associated with a faster SW of lower density. These findings are useful not only to quantify the contribution of >700 keV precipitation to the atmosphere but also to shed light on the typical properties of wave-driven vs FLCS-driven precipitation which can be assimilated into physics-based and/or predictive radiation belt models. In addition, the dataset of ∼9,400 REP events is made available to the community to enable future work.