A Data-Driven Approach to Characterizing the Structure and Processes of Saturn's Magnetosphere

The Cassini mission spent 13 years in orbit around Saturn collecting a wealth of data about its global magnetic and plasma environment, its rings and moons, which shifted our understanding of the complex nature of the Saturnian system. The edge of Saturn's magnetosphere, called the magnetopause (MP), and beyond that the bow shock (BS) are of particular interest to planetary scientists due to the role these boundaries play in plasma energization and transport via processes like turbulent heating, wave-particle interaction and magnetic reconnection. A catalogue of crossings is needed to study these fundamental processes. However, a very challenging aspect is the manual identification of thousands of boundary crossings made by the Cassini spacecraft, particularly the precise transition across the magnetopause current layer.

My work aims to automate the detection of these boundaries using time series analysis of multi-instrument datasets, minimum variance analysis of magnetic field data, moments of energy distribution functions and other physical knowledge of Saturn's environment. In addition, these techniques are ensembled with data-intensive algorithms in machine learning such as XGBoost, convolutional neural networks with uncertainty quantification via supervised learning for more efficient detection of BS and MP crossings. With this standardized crossing list, interesting structures and processes of the magnetosphere can be studied. For example, the relation between electron heating at Saturn's MP and magnetic reconnection, characterization of mirror mode instability in Saturn's magnetosheath, improving the search of rare 'cushion' regions near Saturn's MP of which only 5 have been found to date, in contrast to a persistent occurrence near Jupiter's dayside MP.