Abstract
The physical properties of Enceladus plume particles can shed light on the processes responsible for driving the moon's geological activity. Cassini's Visual and Infrared Mapping Spectrometer (VIMS) recorded near-infrared spectra of the plume for three Enceladus orbits around Saturn in 2017 that show variations reflecting changes in the typical plume particle size over various timescales. We translate these spectra into information about the plume-particle's size distribution using a machine learning model trained on Mie-theory predictions for the light scattered by various tenuous particle populations. This algorithm considers multiple realizations of random noise on top of both the observed and predicted spectra in order to obtain more stable estimates of the size distribution parameters. These models reveal that the typical particle size may decrease with increasing altitude, but this stratification is only detectable when Enceladus is far from its orbital apocenter and the plume is less active. The average particle size also appears to increase as the orbital phase increases after Enceladus passed through its apocenter during the observations on Jun 18th and Aug 2nd, but not during the observation on Aug 28th. Secondly, the maximum particle size in the plume appears to be elevated on Aug 2nd, which may be due to a highly collimated jet that was active only on that date. These patterns might indicate that there are different sources for the particles and their intensity may be changing over time.