• Clicking on the title will open a new window with all details of the bibliographic entry.
  • Clicking on the DOI link will open a new window with the original bibliographic entry from the publisher.
  • Clicking on a single author will show all publications by the selected author.
  • Clicking on a single keyword, will show all publications by the selected keyword.

Found 3 entries in the Bibliography.

Showing entries from 1 through 3


A dynamical model of equatorial magnetosonic waves in the inner magnetosphere: A machine learning approach

Abstract Equatorial magnetosonic waves, together with chorus and plasmaspheric hiss, play key roles in the dynamics of energetic electron fluxes in the magnetosphere. Numerical models, developed following a first principles approach, that are used to study the evolution of high energy electron fluxes are mainly based on quasilinear diffusion. The application of such numerical codes requires statistical models for the distribution of key magnetospheric wave modes to estimate the appropriate diffusion coefficients. These waves are generally statistically modelled as a function of spatial location and geomagnetic indices (e.g. AE, Kp, or Dst). This study presents a novel dynamic spatiotemporal model for equatorial magnetosonic (EMS) wave amplitude, developed using the Nonlinear AutoRegressive Moving Average eXogenous (NARMAX) machine learning approach. The EMS wave amplitude, measured by the Van Allen Probes, are modelled using the time lags of the solar wind and geomagnetic indices as inputs as well as the location at which the measurement is made. The resulting model performance is assessed on a separate Van Allen Probes dataset, where the prediction efficiency was found to be 34.0\% and the correlation coefficient was 56.9\%. With more training and validation data the performance metrics could potentially be improved, however, it is also possible that the EMS wave distribution is affected by stochastic factors and the performance metrics obtained for this model are close to the potential maximum.

Boynton, R.; Walker, S.; Aryan, H.; Hobara, Y.; Balikhin, M.;

Published by: Journal of Geophysical Research: Space Physics      Published on: 06/2021

YEAR: 2021     DOI:

magnetosonic waves; Machine learning; NARMAX; Van Allen Probes

A combined neural network- and physics-based approach for modeling plasmasphere dynamics

AbstractIn recent years, feedforward neural networks (NNs) have been successfully applied to reconstruct global plasmasphere dynamics in the equatorial plane. These neural network-based models capture the large-scale dynamics of the plasmasphere, such as plume formation and erosion of the plasmasphere on the nightside. However, their performance depends strongly on the availability of training data. When the data coverage is limited or non-existent, as occurs during geomagnetic storms, the performance of NNs significantly decreases, as networks inherently cannot learn from the limited number of examples. This limitation can be overcome by employing physics-based modeling during strong geomagnetic storms. Physics-based models show a stable performance during periods of disturbed geomagnetic activity, if they are correctly initialized and configured. In this study, we illustrate how to combine the neural network- and physics-based models of the plasmasphere in an optimal way by using data assimilation. The proposed approach utilizes advantages of both neural network- and physics-based modeling and produces global plasma density reconstructions for both quiet and disturbed geomagnetic activity, including extreme geomagnetic storms. We validate the models quantitatively by comparing their output to the in-situ density measurements from RBSP-A for an 18-month out-of-sample period from 30 June 2016 to 01 January 2018, and computing performance metrics. To validate the global density reconstructions qualitatively, we compare them to the IMAGE EUV images of the He+ particle distribution in the Earth s plasmasphere for a number of events in the past, including the Halloween storm in 2003.This article is protected by copyright. All rights reserved.

Zhelavskaya, I.; Aseev, N.; Shprits, Y;

Published by: Journal of Geophysical Research: Space Physics      Published on: 02/2021

YEAR: 2021     DOI:

plasmasphere; plasma density; neural networks; data assimilation; Kalman Filter; Machine learning; Van Allen Probes


Empirical modeling of the plasmasphere dynamics using neural networks

We propose a new empirical model for reconstructing the global dynamics of the cold plasma density distribution based only on solar wind data and geomagnetic indices. Utilizing the density database obtained using the NURD (Neural-network-based Upper hybrid Resonance Determination) algorithm for the period of October 1, 2012 - July 1, 2016, in conjunction with solar wind data and geomagnetic indices, we develop a neural network model that is capable of globally reconstructing the dynamics of the cold plasma density distribution for 2<=L<=6 and all local times. We validate and test the model by measuring its performance on independent datasets withheld from the training set and by comparing the model predicted global evolution with global images of He+ distribution in the Earth\textquoterights plasmasphere from the IMAGE Extreme UltraViolet (EUV) instrument. We identify the parameters that best quantify the plasmasphere dynamics by training and comparing multiple neural networks with different combinations of input parameters (geomagnetic indices, solar wind data, and different durations of their time history). The optimal model is based on the 96-hour time history of Kp, AE, SYM-H, and F10.7 indices. The model successfully reproduces erosion of the plasmasphere on the night side and plume formation and evolution. We demonstrate results of both local and global plasma density reconstruction. This study illustrates how global dynamics can be reconstructed from local in-situ observations by using machine learning techniques.

Zhelavskaya, Irina; Shprits, Yuri; c, Maria;

Published by: Journal of Geophysical Research: Space Physics      Published on: 10/2017

YEAR: 2017     DOI: 10.1002/2017JA024406

inner magnetosphere; Machine learning; Models; neural networks; plasmasphere; Van Allen Probes