Reconstruction of Plasma Electron Density From Satellite Measurements Via Artificial Neural Networks

Abstract This chapter presents a new approach to reconstruction of plasma electron density from satellite measurements in an automated fashion using artificial neural networks. We design a feedforward neural network to derive the upper-hybrid resonance frequency, f uhr , from satellite measurements that is subsequently used to calculate the electron density. In previous studies, the reconstruction of f uhr was either performed manually or by semiautomated techniques. In this study, we use 2.5 years of electric and magnetic field measurements collected with the Electric and Magnetic Field Instrument Suite and Integrated Science (EMFISIS) instrumentation suite of the Van Allen Probes mission to train, validate, and test a neural network. We then apply it to more than 4 years of EMFISIS data and produce the publicly available electron density data set. We describe the aspects of neural network design and implementation and perform analysis of the obtained electron density distribution.