The evolution and propagation of chorus waves by a machine learning-based model

Abstract

The Earth’s inner magnetosphere is a complex, non-linear, interconnected system driven by external solar wind driving and internal processes. A neural network approach has been proposed to reconstruct and predict the complexity of the inner magnetospheric environment [Bortnik 2016], including the cold plasma density [Chu et al., 2017a; b], the plasma waves such as whistler-mode chorus, and hiss waves, EMIC and ULF waves [Bortnik et al., 2018], and electron fluxes [Chu et al., 2021; Ma et al., 2022]. While the electron density and fluxes can be accurately modeled in these studies, the ML-based models of the plasma waves are usually biased due to the too-often-too-quiet problem both in numerical simulations and observations [Ma et al., 2018; Camporeale et al., 2019; Guo et al., 2021]. We present the machine learning (ML) based empirical models of the wave environment for the inner magnetosphere. The ML-based wave model used a neural network approach, which takes the solar wind parameters and geomagnetic indices as its input parameters and provides a global reconstruction and prediction of these wave environments in the inner magnetosphere below L~7. The model performance has been validated and tested on out-of-sample data sets which have never been ‘seen’ by the model, thereby demonstrating that the model provides reliable and stable predictions when the too-often-too- quiet problem is solved. We will conduct event and statistical analysis using the ML-based reconstruction of the wave environment for a list of geomagnetic activities, including geomagnetic storms and substorms. The temporal and spatial evolution of the wave environment is investigated, which cannot be done using in-situ observations or statistically-averaged models. The results show how machine learning techniques might be used to help achieve new discoveries in magnetospheric physics and advance state-of-the-art space weather prediction. 269