Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Automatic Detection of Interplanetary Coronal Mass Ejections in Solar Wind In Situ Data

Hannah T. Rüdisser, Andreas Windisch, Ute V. Amerstorfer, Christian Möstl, Tanja Amerstorfer, Rachel L. Bailey, Martin A. Reiss

TL;DR

The work reframes automatic ICME detection in solar wind in situ data as a time-series segmentation task and implements a ResUNet++-based pipeline to identify ICME boundaries. Using Wind data from 1997–2015 (with 33 features) and year-stratified cross-validation, the method achieves a True Skill Statistic of about 0.64, with start/end time MAEs of roughly 2h56m and 3h20m, while training about 20× faster than a prior baseline. It maintains reasonable performance on reduced feature datasets from Wind, STEREO-A, and STEREO-B, indicating generalization challenges tied to catalog inconsistencies. The approach offers a fast, scalable tool for automated ICME detection and can be extended to detect other solar wind structures like CIRs, supporting space weather forecasting and large-scale data analyses.

Abstract

Interplanetary coronal mass ejections (ICMEs) are one of the main drivers for space weather disturbances. In the past, different approaches have been used to automatically detect events in existing time series resulting from solar wind in situ observations. However, accurate and fast detection still remains a challenge when facing the large amount of data from different instruments. For the automatic detection of ICMEs we propose a pipeline using a method that has recently proven successful in medical image segmentation. Comparing it to an existing method, we find that while achieving similar results, our model outperforms the baseline regarding training time by a factor of approximately 20, thus making it more applicable for other datasets. The method has been tested on in situ data from the Wind spacecraft between 1997 and 2015 with a True Skill Statistic (TSS) of 0.64. Out of the 640 ICMEs, 466 were detected correctly by our algorithm, producing a total of 254 False Positives. Additionally, it produced reasonable results on datasets with fewer features and smaller training sets from Wind, STEREO-A and STEREO-B with True Skill Statistics of 0.56, 0.57 and 0.53, respectively. Our pipeline manages to find the start of an ICME with a mean absolute error (MAE) of around 2 hours and 56 minutes, and the end time with a MAE of 3 hours and 20 minutes. The relatively fast training allows straightforward tuning of hyperparameters and could therefore easily be used to detect other structures and phenomena in solar wind data, such as corotating interaction regions.

Automatic Detection of Interplanetary Coronal Mass Ejections in Solar Wind In Situ Data

TL;DR

The work reframes automatic ICME detection in solar wind in situ data as a time-series segmentation task and implements a ResUNet++-based pipeline to identify ICME boundaries. Using Wind data from 1997–2015 (with 33 features) and year-stratified cross-validation, the method achieves a True Skill Statistic of about 0.64, with start/end time MAEs of roughly 2h56m and 3h20m, while training about 20× faster than a prior baseline. It maintains reasonable performance on reduced feature datasets from Wind, STEREO-A, and STEREO-B, indicating generalization challenges tied to catalog inconsistencies. The approach offers a fast, scalable tool for automated ICME detection and can be extended to detect other solar wind structures like CIRs, supporting space weather forecasting and large-scale data analyses.

Abstract

Interplanetary coronal mass ejections (ICMEs) are one of the main drivers for space weather disturbances. In the past, different approaches have been used to automatically detect events in existing time series resulting from solar wind in situ observations. However, accurate and fast detection still remains a challenge when facing the large amount of data from different instruments. For the automatic detection of ICMEs we propose a pipeline using a method that has recently proven successful in medical image segmentation. Comparing it to an existing method, we find that while achieving similar results, our model outperforms the baseline regarding training time by a factor of approximately 20, thus making it more applicable for other datasets. The method has been tested on in situ data from the Wind spacecraft between 1997 and 2015 with a True Skill Statistic (TSS) of 0.64. Out of the 640 ICMEs, 466 were detected correctly by our algorithm, producing a total of 254 False Positives. Additionally, it produced reasonable results on datasets with fewer features and smaller training sets from Wind, STEREO-A and STEREO-B with True Skill Statistics of 0.56, 0.57 and 0.53, respectively. Our pipeline manages to find the start of an ICME with a mean absolute error (MAE) of around 2 hours and 56 minutes, and the end time with a MAE of 3 hours and 20 minutes. The relatively fast training allows straightforward tuning of hyperparameters and could therefore easily be used to detect other structures and phenomena in solar wind data, such as corotating interaction regions.
Paper Structure (16 sections, 11 figures, 9 tables)

This paper contains 16 sections, 11 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Data
  3. In situ Data
  4. Catalogs
  5. Machine Learning and Pipeline
  6. Preprocessing
  7. Architecture of the Model
  8. Training
  9. Postprocessing
  10. Validation
  11. Results
  12. Full Wind Dataset
  13. Reduced Wind, STEREO-A and STEREO-B Datasets
  14. Discussion
  15. Summary
  16. ...and 1 more sections

Figures (11)

  • Figure 1: Solar wind in situ data from the Wind spacecraft located at the Lagrangian Point L1, showing an ICME. The solid vertical lines delimitate the event, including shock, sheath and magnetic cloud. From top to bottom: magnetic field amplitude and components, solar wind velocity, proton density, proton temperature.
  • Figure 2: Comparison of the number of events listed in different catalogs: chi, nieves, nguyen and HELIO4CAST Moestl2020.
  • Figure 3: Block diagram of the model architecture.
  • Figure 4: Confusion matrix visualizing the results of the full Wind dataset.
  • Figure 5: Comparison of the ground truth label and the predicted labels for the full Wind dataset in the time period of April 2012 to July 2012. The top panel shows the magnetic field and its components, the second panel shows the ground truth label, the third panel depicts the label predicted by our pipeline, and the label predicted by nguyen is shown in the bottom panel.
  • ...and 6 more figures