Machine learning emulator for physics-based prediction of ionospheric potential response to solar wind variations

Abstract

Abstract Physics-based simulations are important for elucidating the fundamental mechanisms behind the time-varying complex ionospheric conditions, such as ionospheric potential, against unprecedented solar wind variations incident on the Earth’s magnetosphere. However, carrying out an extensive parameter survey for comprehending the nonlinear solar wind density dependence of the ionospheric potential, for example, requires state-of-the-art global magnetohydrodynamic (MHD) simulations, which cannot be executed efficiently even on large-scale cluster computers. Here, we report the performance of a machine-learning based surrogate model for estimating the ionospheric potential outputs of a global MHD simulation, using the reservoir computing technique called echo state network (ESN). The trained ESN-based emulator demonstrates exceptional speed in conducting the parameter survey, which can lead to the identification of a solar wind density dependence of the ionospheric polar cap potential. Finally, we discuss future directions including the promising application for space weather forecasting. Graphical Abstract

Keywords

• Ionospheric potential
• Machine learning
• Magnetohydrodynamics