Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Analysis and Predictive Modeling of Solar Coronal Holes Using Computer Vision and ARIMA-LSTM Networks

Juyoung Yun, Jungmin Shin

TL;DR

This work integrates computer-vision-based detection of coronal holes from SDO/AIA-193 imagery with a hybrid ARIMA-LSTM time-series model to forecast coronal hole areas over a seven-day horizon. The detection pipeline uses NLmeans denoising, binarization, and Laplacian of Gaussian contouring within defined solar regions, yielding area measurements across 3857 days. The forecasting stage combines ARIMA and LSTM models, with automatic ARIMA parameter selection and prediction averaging, to produce robust short-term hole-area forecasts. Overall, the approach advances space-weather forecasting by enabling automated, scalable hole-size predictions that can improve anticipation of solar wind impacts on Earth.

Abstract

In the era of space exploration, coronal holes on the sun play a significant role due to their impact on satellites and aircraft through their open magnetic fields and increased solar wind emissions. This study employs computer vision techniques to detect coronal hole regions and estimate their sizes using imagery from the Solar Dynamics Observatory (SDO). Additionally, we utilize hybrid time series prediction model, specifically combination of Long Short-Term Memory (LSTM) networks and ARIMA, to analyze trends in the area of coronal holes and predict their areas across various solar regions over a span of seven days. By examining time series data, we aim to identify patterns in coronal hole behavior and understand their potential effects on space weather.

Analysis and Predictive Modeling of Solar Coronal Holes Using Computer Vision and ARIMA-LSTM Networks

TL;DR

This work integrates computer-vision-based detection of coronal holes from SDO/AIA-193 imagery with a hybrid ARIMA-LSTM time-series model to forecast coronal hole areas over a seven-day horizon. The detection pipeline uses NLmeans denoising, binarization, and Laplacian of Gaussian contouring within defined solar regions, yielding area measurements across 3857 days. The forecasting stage combines ARIMA and LSTM models, with automatic ARIMA parameter selection and prediction averaging, to produce robust short-term hole-area forecasts. Overall, the approach advances space-weather forecasting by enabling automated, scalable hole-size predictions that can improve anticipation of solar wind impacts on Earth.

Abstract

In the era of space exploration, coronal holes on the sun play a significant role due to their impact on satellites and aircraft through their open magnetic fields and increased solar wind emissions. This study employs computer vision techniques to detect coronal hole regions and estimate their sizes using imagery from the Solar Dynamics Observatory (SDO). Additionally, we utilize hybrid time series prediction model, specifically combination of Long Short-Term Memory (LSTM) networks and ARIMA, to analyze trends in the area of coronal holes and predict their areas across various solar regions over a span of seven days. By examining time series data, we aim to identify patterns in coronal hole behavior and understand their potential effects on space weather.
Paper Structure (11 sections, 8 figures)

This paper contains 11 sections, 8 figures.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methodology
  4. Coronal Hole Detection using Computer Vision
  5. Coronal Hole Analysis
  6. Time Series Prediction using Deep Learning
  7. Results
  8. Experimental Setting
  9. Experimental Results
  10. Discussion
  11. Conclusion

Figures (8)

  • Figure 1: A visual representation of the changes in coronal hole size and position over a 4-day period from December 23, 2022 to December 26, 2022 Nasa. The image illustrates the dynamic nature of coronal holes and their potential impact on space weather events.
  • Figure 2: Overall structure of the predictive modeling of coronal hole areas using computer vision and deep learning. The regions used to determine the coronal hole area on the sun are defined by the binary-based Coronal Hole Detection (BCH) model with AIA-193 Nasa.
  • Figure 3: A depiction of the multi-filters computer vision based detection applied to an original AIA 193 image Nasa, including anti-noising and binary filters to accurately identify and measure coronal holes in the sun.
  • Figure 4: Daily coronal hole data for each region of the sun over 3857 days from January 6, 2011, to August 10, 2021, as detected by the BCH model. The graph shows the variation in coronal hole area over time.
  • Figure 5: The AP index and coronal hole areas in the middle and redzone regions of the sun from June 22, 2021, to August 10, 2021, as detected by the BCH model. The graph shows the correlation between the AP index and coronal hole area in these regions.
  • ...and 3 more figures