Predicting Solar Flares Using CNN and LSTM on Two Solar Cycles of Active Region Data

Abstract

We consider the flare prediction problem that distinguishes flare-imminent active regions that produce an M- or X-class flare in the future 24 hours, from quiet active regions that do not produce any flare within $\pm 24$ hours. Using line-of-sight magnetograms and parameters of active regions in two data products covering Solar Cycle 23 and 24, we train and evaluate two deep learning algorithms -- CNN and LSTM -- and their stacking ensembles. The decisions of CNN are explained using visual attribution methods. We have the following three main findings. (1) LSTM trained on data from two solar cycles achieves significantly higher True Skill Scores (TSS) than that trained on data from a single solar cycle with a confidence level of at least 0.95. (2) On data from Solar Cycle 23, a stacking ensemble that combines predictions from LSTM and CNN using the TSS criterion achieves significantly higher TSS than the "select-best" strategy with a confidence level of at least 0.95. (3) A visual attribution method called Integrated Gradients is able to attribute the CNN's predictions of flares to the emerging magnetic flux in the active region. It also reveals a limitation of CNN as a flare prediction method using line-of-sight magnetograms: it treats the polarity artifact of line-of-sight magnetograms as positive evidence of flares.

31 pages, 16 figures, accepted in the ApJ

References in corpus (18)

• The Helioseismic and Magnetic Imager (HMI) Vector Magnetic Field Pipeline: SHARPs -- Space-weather HMI Active Region Patches
• Issues in Stacked Generalization
• Solar Flare Prediction Using SDO/HMI Vector Magnetic Field Data with a Machine-Learning Algorithm
• Toward Reliable Benchmarking of Solar Flare Forecasting Methods
• A Comparison of Flare Forecasting Methods, I: Results from the "All-Clear" Workshop
• Solar Flare Prediction Model with Three Machine-Learning Algorithms Using Ultraviolet Brightening and Vector Magnetogram
• Predicting Solar Flares Using a Long Short-Term Memory Network
• Predicting Solar Flares Using SDO/HMI Vector Magnetic Data Product and Random Forest Algorithm
• Properties of sunspots in cycle 23: I. Dependence of brightness on sunspot size and cycle phase
• SMARPs and SHARPs: Two Solar Cycles of Active Region Data
• Evaluating (and Improving) Estimates of the Solar Radial Magnetic Field Component from Line-of-Sight Magnetograms
• How good is the bipolar approximation of active regions for surface flux transport?
• Fine-grained Solar Flare Forecasting Based on the Hybrid Convolutional Neural Networks
• Reliable Probability Forecast of Solar Flares: Deep Flare Net-Reliable (DeFN-R)
• Supervised convolutional neural networks for classification of flaring and nonflaring active regions using line-of-sight magnetograms
• Modelling and Interpreting The Effects of Spatial Resolution on Solar Magnetic Field Maps
• A Framework for Designing and Evaluating Solar Flare Forecasting Systems
• Ensemble Forecasting of Major Solar Flares: Methods for Combining Models

Cited by in corpus (7)

• Solar Flare Index Prediction Using SDO/HMI Vector Magnetic Data Products with Statistical and Machine Learning Methods
• Explaining Full-disk Deep Learning Model for Solar Flare Prediction using Attribution Methods
• Solar Active Region Magnetogram Image Dataset for Studies of Space Weather
• Explainable Deep Learning-based Solar Flare Prediction with post hoc Attention for Operational Forecasting
• Deep learning reconstruction of sunspot vector magnetic fields for forecasting solar storms
• Solar Flare Prediction Using Long Short-term Memory (LSTM) and Decomposition-LSTM with Sliding Window Pattern Recognition
• ALEXIS: Recreating the X-Ray Emission from the Full Solar Disk as a Linear Combination of Discrete Regions in the Extreme Ultraviolet and Soft X-Rays