Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Explainability in Solar Energetic Particle Event Prediction: A Global Feature Mapping Approach

Anli Ji, Pranjal Patil, Chetraj Pandey, Manolis K. Georgoulis, Berkay Aydin

Abstract

Solar energetic particle (SEP) events, as one of the most prominent manifestations of solar activity, can generate severe hazardous radiation when accelerated by solar flares or shock waves formed aside from coronal mass ejections (CMEs). However, most existing data-driven methods used for SEP predictions are operated as black-box models, making it challenging for solar physicists to interpret the results and understand the underlying physical causes of such events rather than just obtain a prediction. To address this challenge, we propose a novel framework that integrates global explanations and ad-hoc feature mapping to enhance model transparency and provide deeper insights into the decision-making process. We validate our approach using a dataset of 341 SEP events, including 244 significant (>=10 MeV) proton events exceeding the Space Weather Prediction Center S1 threshold, spanning solar cycles 22, 23, and 24. Furthermore, we present an explainability-focused case study of major SEP events, demonstrating how our method improves explainability and facilitates a more physics-informed understanding of SEP event prediction.

Enhancing Explainability in Solar Energetic Particle Event Prediction: A Global Feature Mapping Approach

Abstract

Solar energetic particle (SEP) events, as one of the most prominent manifestations of solar activity, can generate severe hazardous radiation when accelerated by solar flares or shock waves formed aside from coronal mass ejections (CMEs). However, most existing data-driven methods used for SEP predictions are operated as black-box models, making it challenging for solar physicists to interpret the results and understand the underlying physical causes of such events rather than just obtain a prediction. To address this challenge, we propose a novel framework that integrates global explanations and ad-hoc feature mapping to enhance model transparency and provide deeper insights into the decision-making process. We validate our approach using a dataset of 341 SEP events, including 244 significant (>=10 MeV) proton events exceeding the Space Weather Prediction Center S1 threshold, spanning solar cycles 22, 23, and 24. Furthermore, we present an explainability-focused case study of major SEP events, demonstrating how our method improves explainability and facilitates a more physics-informed understanding of SEP event prediction.

Paper Structure

This paper contains 13 sections, 11 equations, 3 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Sliding Window Multivariate Time Series Forest
  5. Global Explainability
  6. Results and Analysis
  7. Data Collection
  8. Experimental Settings
  9. Evaluation Metrics
  10. Model Performance
  11. Explainability Enhancement
  12. Remarks
  13. Conclusion

Figures (3)

  • Figure 1: Solar Energetic Particle (SEP) event during the Halloween storms, captured by Extreme Ultraviolet Imaging Telescope (EIT) onboard Solar and Heliospheric Observatory (SOHO) from (a) 2003-10-28T07:06 (onset) to (b) 2003-10-28T22:11 (peak). (c) Time series plot occurred on 2003-10-28T11:35 (UT) shown in log scale within GOES P3($\geq$10 MeV), P5($\geq$50 MeV), and P7($\geq$100 MeV) integral proton channels.
  • Figure 2: Performance evaluation of the SEP event prediction model across three different classification scenarios: (a) Strong vs. Weak events; (b) Strong vs. Weak and No-event; and (c) Event vs. No-event at observation windows of 6, 8, and 10 hours and lag windows of 5, 15, 30, 45, 60, 120, and 180 minutes prior to the event onset time. Results were obtained by averaging across 10 bootstrapping runs.
  • Figure 3: Global explainability analysis of feature importance across flux channels (P3, P5, and P7), shown for bootstrap iterations of 1, 10, 100, and 1000 (top to bottom). Each subplot represents the integrated proton flux data across varying energy thresholds, providing insights into the temporal evolution of SEP events.