Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Extreme Solar Flare Prediction Using Residual Networks with HMI Magnetograms and Intensitygrams

Juyoung Yun, Jungmin Shin

TL;DR

This work addresses extreme solar flare prediction by leveraging HMI intensitygrams to locate sunspots and extracting corresponding magnetogram patches for training a ResNet50 classifier on a binary QA vs MX task. The two-stage approach demonstrates that magnetogram data provide superior features for deep learning than alternative AIA images, achieving a test accuracy of about 0.755 on MX-class prediction. The study highlights the critical role of magnetic-field information in flare forecasting and discusses data-scarcity as a limitation, calling for larger, purpose-built solar-imaging datasets to enhance generalization and pre-flare timing forecasts. Overall, the method advances space weather forecasting with practical implications for mitigating the impacts of extreme solar activity.

Abstract

Solar flares, especially C, M, and X class, pose significant risks to satellite operations, communication systems, and power grids. We present a novel approach for predicting extreme solar flares using HMI intensitygrams and magnetograms. By detecting sunspots from intensitygrams and extracting magnetic field patches from magnetograms, we train a Residual Network (ResNet) to classify extreme class flares. Our model demonstrates high accuracy, offering a robust tool for predicting extreme solar flares and improving space weather forecasting. Additionally, we show that HMI magnetograms provide more useful data for deep learning compared to other SDO AIA images by better capturing features critical for predicting flare magnitudes. This study underscores the importance of identifying magnetic fields in solar flare prediction, marking a significant advancement in solar activity prediction with practical implications for mitigating space weather impacts.

Extreme Solar Flare Prediction Using Residual Networks with HMI Magnetograms and Intensitygrams

TL;DR

This work addresses extreme solar flare prediction by leveraging HMI intensitygrams to locate sunspots and extracting corresponding magnetogram patches for training a ResNet50 classifier on a binary QA vs MX task. The two-stage approach demonstrates that magnetogram data provide superior features for deep learning than alternative AIA images, achieving a test accuracy of about 0.755 on MX-class prediction. The study highlights the critical role of magnetic-field information in flare forecasting and discusses data-scarcity as a limitation, calling for larger, purpose-built solar-imaging datasets to enhance generalization and pre-flare timing forecasts. Overall, the method advances space weather forecasting with practical implications for mitigating the impacts of extreme solar activity.

Abstract

Solar flares, especially C, M, and X class, pose significant risks to satellite operations, communication systems, and power grids. We present a novel approach for predicting extreme solar flares using HMI intensitygrams and magnetograms. By detecting sunspots from intensitygrams and extracting magnetic field patches from magnetograms, we train a Residual Network (ResNet) to classify extreme class flares. Our model demonstrates high accuracy, offering a robust tool for predicting extreme solar flares and improving space weather forecasting. Additionally, we show that HMI magnetograms provide more useful data for deep learning compared to other SDO AIA images by better capturing features critical for predicting flare magnitudes. This study underscores the importance of identifying magnetic fields in solar flare prediction, marking a significant advancement in solar activity prediction with practical implications for mitigating space weather impacts.
Paper Structure (11 sections, 4 figures)

This paper contains 11 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methodology
  4. Training Deep Neural Networks with Magnetogram
  5. Sunspot Detection for Patch Extraction
  6. Results
  7. Experimental Setting
  8. Training Results of Deep Neural Networks
  9. Prediction Results
  10. Discussion
  11. Conclusion

Figures (4)

  • Figure 1: The two-phase process for extreme solar flare prediction. Phase 1 involves training a deep neural network (ResNet) with HMI magnetogram (HMIB) data to classify extreme and non-extreme events. Phase 2 includes detecting sunspots on HMI intensitygram (HMII) images, extracting corresponding magnetic field patches from HMIB, and using the trained model to predict extreme solar flare occurrences.
  • Figure 2: The process of detecting sunspots using HMI intensitygrams. From left to right: the original HMI intensitygram (HMII), the binarized HMII, detected sunspots on the HMII, and the corresponding sunspot detection mapped onto the HMI magnetogram (HMIB).
  • Figure 3: Comparison of test accuracy for different satellite image datasets using the SDOBenchmark dataset SDOBenchmark. The model used is ResNet-50 he2016deep pre-trained on ImageNet deng2009imagenet for predicting two flare classes, QA and MX.
  • Figure 4: Flare Prediction Results. (Left) HMI magnetogram (HMIB) showing predicted MX class flare region. (Right) Corresponding AIA-131 image showing the flare. Timeline is 2022-01-16 00:00:00.