Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automatic Identification of Magnetospheric Regions using Supervised Machine Learning Models

Narges Ahmadi, Robert Ergun, Xiangning Chu, Alex Chasapis, Victoria Wilder

TL;DR

The paper tackles the manual bottleneck of magnetospheric region identification by introducing a hybrid CNN–RF framework that processes ion energy spectrograms and scalar plasma parameters from MMS data. It achieves 99% accuracy at a 3-minute cadence and enables boundary detection, all with a small labeled dataset and low computational demands. The approach advances automated, scalable analysis of space plasma environments and can be adapted to other missions with similar data products. This has practical implications for space weather studies and near-real-time data processing.

Abstract

We present an automated approach for identifying magnetospheric regions using supervised machine learning techniques applied to Magnetospheric MultiScale mission data. Our method utilizes ion energy spectra, total magnetic field, total ion temperature, and spacecraft position data to classify five distinct plasma environments: solar wind, magnetosheath, inner magnetosphere, plasma sheet, and lobe regions. The approach combines a convolutional neural network (CNN) for analyzing ion energy spectrogram data with a Random Forest classifier for scalar plasma parameters. The CNN method employs 2D convolution to identify spatial and temporal patterns in the ion energy spectrogram treated as image-like data, while the Random Forest model processes averaged magnetic field, temperature, and position parameters. Our hybrid model achieves 99% accuracy on test dataset with an F1 score of 0.99, providing reliable automated region identification at 3-minute temporal resolution. This lightweight approach requires minimal manual data labeling and can be readily applied to other magnetospheric missions with similar data products.

Automatic Identification of Magnetospheric Regions using Supervised Machine Learning Models

TL;DR

The paper tackles the manual bottleneck of magnetospheric region identification by introducing a hybrid CNN–RF framework that processes ion energy spectrograms and scalar plasma parameters from MMS data. It achieves 99% accuracy at a 3-minute cadence and enables boundary detection, all with a small labeled dataset and low computational demands. The approach advances automated, scalable analysis of space plasma environments and can be adapted to other missions with similar data products. This has practical implications for space weather studies and near-real-time data processing.

Abstract

We present an automated approach for identifying magnetospheric regions using supervised machine learning techniques applied to Magnetospheric MultiScale mission data. Our method utilizes ion energy spectra, total magnetic field, total ion temperature, and spacecraft position data to classify five distinct plasma environments: solar wind, magnetosheath, inner magnetosphere, plasma sheet, and lobe regions. The approach combines a convolutional neural network (CNN) for analyzing ion energy spectrogram data with a Random Forest classifier for scalar plasma parameters. The CNN method employs 2D convolution to identify spatial and temporal patterns in the ion energy spectrogram treated as image-like data, while the Random Forest model processes averaged magnetic field, temperature, and position parameters. Our hybrid model achieves 99% accuracy on test dataset with an F1 score of 0.99, providing reliable automated region identification at 3-minute temporal resolution. This lightweight approach requires minimal manual data labeling and can be readily applied to other magnetospheric missions with similar data products.

Paper Structure

This paper contains 14 sections, 7 figures, 1 table.

Table of Contents

  1. Introduction
  2. Data and Methodology
  3. The MMS Mission
  4. Training Data Generation and Labeling
  5. Data Preprocessing
  6. Hybrid Model Design
  7. Convolutional Neural Network for Flux Data
  8. Random Forest for Scalar Parameters
  9. Model Integration and Prediction
  10. Results
  11. Model Performance Metrics
  12. Boundary Identification
  13. Operational Application
  14. Discussion and Future Directions

Figures (7)

  • Figure 1: Magnetospheric regions corresponding to normalized ion energy spectrogram data and associated normalized scalar parameters.
  • Figure 2: Summary of combined CNN and random forest models.
  • Figure 3: Summary of loss and accuracy for CNN model. The accuracy is 98% on the the validation data.
  • Figure 4: Confusion matrix applied to test dataset.
  • Figure 5: Classification report on test dataset. The hybrid model shows 99% accuracy on the test dataset with an average F1 score of 0.99.
  • ...and 2 more figures