The simulations were run with the fully kinetic Particle-In-Cell code SMILEI. The data presented are normalized using ion-scale quantities. The magnetic field and density are normalized to arbitrary values B0 and n0, respectively. We choose B0 and n0 such that the density and magnetic field are equal to 1on the flowing side of the layer (in our simulations: the right side). The masses and charges are normalized to the proton mass mp and charge e, time is normalized to the inverse of the proton gyrofrequency ωci-1 = mp /eB and length to the proton inertial length δ= cpi, where c is the speed of light and ωpi =(n0e2/mpε0)1/2 is the proton plasma frequency. Velocities are normalized to the ions’ Alfvén velocity vAl = δiωci
The simulation is initialized with a single layer where density and magnetic field (directed along the x -direction) vary along the y direction. This layer is contained in the (x, y) plane in a 2-D domain of size (xmax,ymax) = (1280, 256)δi. There are nx = 25600 and ny = 10240 cells in the x and y directions, corresponding to a grid resolution of ∆x = 0.05 δi and ∆y = 0.025 δi. Their are four populations (electrons, plume ions, magnetosheath ions and hot magnetosphere ions) with initially 50 macro-particles per cell and per population loaded using Maxwellian distributions. Plasma moments and electromagnetic forces are calculated using second-order interpolation. The time step is calculated using a Courant–Friedrichs–Lewy condition, which in our simulations turns out to be ∆ = 8.4 × 10-4 ωci-1, and the total simulation time is 800 ωci-1. A reduced mass ratio mi/me = 25 is used for computational reasons.


Simulations of the impact of a plasmaspheric plume on magnetic reconnection. On this picture, the plasmaspheric plume (dense plasma in orange) has yet to reach the reconnection layer. However, the reconnection is already well-developed and even formed a plasmoid, that is evacuated by the exhaust.