Machine Learning Applications to Kronian Magnetospheric Reconnection Classification

Abstract

<p>Signatures of magnetic reconnection in Saturn's magnetotail are identified in magnetometer observations by characteristic deviations in the northward component of the magnetic field. These magnetic deflections are caused by travelling plasma structures created by reconnection rapidly passing over the observing spacecraft. The identification of these reconnection signatures has long been performed by eye, and more recently through semi-automated methods, however these methods are often limited through a required human verification step. Here, we present a fully automated, supervised learning machine learning (ML) model to identify evidence of reconnection in Cassini MAG (magnetometer) observations of the Kronian magnetosphere, constructed using the Smith et al, 2016. reconnection catalogue which contains hundreds of examples of plasmoids, travelling compression regions and dipolarizations. This ML model is capable of rapidly identifying reconnection events in large time-span Cassini datasets, tested against the full year of 2010 with a high level of accuracy (99\%), true skill score (0.97), and Heidke skill score (0.85). From this ML model, a full cataloguing and examination of magnetic reconnection in the Kronian magnetosphere across Cassini's near Saturn lifetime is now possible.</p>

Kronian Magnetospheric Reconnection Statistics Across Cassini’s Lifetime

“… Model of magnetic reconnection in a planetary current sheet. From this form of reconnection, various structures are created: dipolarizations, plasmoids and traveling compression regions which are detectable by in‐situ spacecraft through their unique magnetic deflections (adapted from Garton et al., 2021). These figures are described in a Kronocentric Solar Magnetospheric coordinate system.…”

“…Garton et al. (2021) (G21) applied neural network ML methods to Cassini magnetometer data, utilizing the Smith et al. (2016) (S16) catalog as a training set, to create a Kronian magnetospheric reconnection classifier.…”

Forecasting the Probability of Large Rates of Change of the Geomagnetic Field in the UK: Timescales, Horizons, and Thresholds

“…Neural network models have been shown to perform admirably at classification, regression and forecasting tasks in the fields of space plasma physics and space weather (e.g., Bakrania et al., 2020; Bloch et al., 2021; Bortnik et al., 2016; Clausen & Nickisch, 2018; Garton et al., 2021; James et al., 2020; Lethy et al., 2018; McGranaghan et al., 2020; Wintoft et al., 2017; Zhelavskaya et al., 2021). For space weather forecasting in particular, models can be structured to have a “memory” of the preceding solar wind conditions (Bhaskar & Vichare, 2019; Kugblenu et al., 1999).…”

“…We note that the input for this model is not a matrix of shape , but instead is a flattened input of shape , similar to the approach of Garton et al. (2021).…”