Killer electrons (having energies exceeding 100,00,000 eV) in the Van Allen radiation belts are highly influenced by Sun-driven changes to the near-Earth environment. Such changes can lead to increased numbers of these high-energy electrons which in turn pose severe risks to spacecraft residing amongst the radiation belts. As such, predicting how populations of high-energy electrons vary with time and space due to various phenomena is of upmost importance.

 

We took 4 years' worth of data from NASA's Van Allen Probe space mission, specifically counts of so-called relativistic electrons at a range of energies and pitch angles (the angle of an electron's velocity to the local magnetic field), and applied a number of machine learning techniques of it in order to cluster the distribution of electrons in the radiation belts based on their energy and pitch angle. We found that, though there were a number of different distribution shapes in pitch angle-energy space, there were three common shapes observed; flattop, butterfly, or pancake (see Figure 1). We then studied how these distributions changed in time and space over the years, including times of significant activity in the near-Earth space such as during geomagnetic storms, as presented in Figure 2 where "L" is the L-shell, a measure of distance from the planet, at which these distributions were observed.