Unsupervised classification of simulated magnetospheric regions

Abstract

Abstract. In magnetospheric missions, burst-mode data sampling should be triggered in the presence of processes of scientific or operational interest. We present an unsupervised classification method for magnetospheric regions that could constitute the first step of a multistep method for the automatic identification of magnetospheric processes of interest. Our method is based on self-organizing maps (SOMs), and we test it preliminarily on data points from global magnetospheric simulations obtained with the OpenGGCM-CTIM-RCM code. The dimensionality of the data is reduced with principal component analysis before classification. The classification relies exclusively on local plasma properties at the selected data points, without information on their neighborhood or on their temporal evolution. We classify the SOM nodes into an automatically selected number of classes, and we obtain clusters that map to well-defined magnetospheric regions. We validate our classification results by plotting the classified data in the simulated space and by comparing with k-means classification. For the sake of result interpretability, we examine the SOM feature maps (magnetospheric variables are called features in the context of classification), and we use them to unlock information on the clusters. We repeat the classification experiments using different sets of features, we quantitatively compare different classification results, and we obtain insights on which magnetospheric variables make more effective features for unsupervised classification.

Using Machine Learning to Diagnose Relativistic Electron Distributions in the Van Allen Radiation Belts

“…Machine learning classification methods have been found to be incredibly useful in space physics, including for investigating the different plasma regions in the Earth's magnetosphere (e.g., Breuillard et al (2020); Innocenti et al (2021)) and other planetary magnetospheres (e.g., Cheng et al (2022); Yeakel et al (2022)), identifying solar wind types (e.g., Bloch et al (2020); Amaya et al (2020); Camporeale et al (2017)), and solar wind characteristics (e.g., Bakrania et al (2020b)) and even space weather forecasting (e.g., Smith et al (2020); Maimaiti et al (2019)). Machine learning methods have also increasingly been used to model the highly dynamic radiation belts (e.g., Chu et al (2021); Bortnik et al (2016); Wing et al (2022)).…”

Unsupervised classification of fully kinetic simulations of plasmoid instability using self-organizing maps (SOMs)

“…In Innocenti et al. (2021), it has been used to cluster simulated data, and specifically data from a global magnetospheric simulation. The code used there was the magneto-hydro-dynamic (MHD)-based code OpenGGCM-CTIM-RCM (Raeder 2003), which targets large-scale processes originating from the interaction of the solar wind with the magnetosphere–ionosphere–thermosphere system.…”

“…One rather fundamental question left open in Innocenti et al. (2021) was whether a similar clustering procedure would produce equally meaningful results when applied to smaller-scale processes, kinetic in nature. The MHD simulations intend to reproduce plasma behaviour at scales large enough that certain assumptions can be considered satisfied.…”

“…Hence, the need arises to verify if the clustering procedure described in Innocenti et al. (2021) is robust to using simulation approaches, such as PIC methods, that deliver results directly comparable to observations (Innocenti et al. 2016).…”

A Database of MMS Bow Shock Crossings Compiled Using Machine Learning

“…Furthermore, Kruparova et al (2019) compiled a database of 529 shock crossings using the Cluster spacecraft (www.cosmos.esa.int/web/csa/bow-shock-magnetopause-crossings-2001-2013) with a focus on studying the statistical dependence of the shock velocity on different parameters. Supervised (da Silva et al, 2020;Olshevsky et al, 2021) and unsupervised (Innocenti et al, 2021) machine learning and nonmachine learning based techniques (Jelínek et al, 2012) have been applied to automatically classify the various regions that a spacecraft traverses. This classification can be used to identify shock crossing events.…”