Machine Learning Applications to Planetary Magnetospheric Reconnection Signatures

Abstract:

Magnetic reconnection is a regular occurrence for planetary magnetospheres immersed in the variable interplanetary magnetic field (IMF) where nearby magnetic field lines of opposing polarity can breach and reform into a more stable structure. This continual process releases significant amounts of energy and allows mass to transfer about the planetary magnetosphere and between it and the solar wind. Within planetary magnetotails, signatures of these reconnection events can be identified via in situ observations of the magnetic field as characteristic deflections in the north aligned component of the magnetic field. The most common approach to identifying these signatures is through manual and semi-automated means, however these methods are typically slow and heavily reliant on human verification. As an era of data abundance is approached, these previous methods become dated and unsustainable. Here, we present a fully automated, supervised learning machine learning model to identify signatures of reconnection with Cassini magnetic field measurements of Saturn's planetary magnetosphere. This model utilizes a previously created catalogue of Kronian magnetotail reconnections as created using semi-automated identification by Smith et al., 2016 for training. Utilizing this method attains an event identification accuracy of 99% for the year of 2010, and a Heidke skill score of 0.84. From this model, a full cataloguing and examination of magnetic reconnection events in the Kronian magnetosphere across Cassini's near Saturn lifetime (~13 years) is now possible.