Visual Explanation of a Deep Learning Solar Flare Forecast Model and Its Relationship with Physical Parameters

Abstract:

In this study, we have presented visual explanation of a deep learning solar flare forecast model and its relationship with physical parameters of solar active regions (ARs). For this, we use full-disk magnetograms at 00:00 UT from Solar and Heliospheric Observatory/Michelson Doppler Imager (1996 May - 2010 December) and Solar Dynamics Observatory/Helioseismic and Magnetic Imager (2011 January - 2017 June), physical parameters from Space-weather HMI Active Region Patch (SHARP), and Geostationary Operational Environmental Satellite X-ray flare data. Our deep learning flare forecast model based on the Convolutional Neural Network (CNN) predicts “Yes or No" of daily flare occurrence for C-, M-, and X-class. We interpret the model using two CNN attribution methods (guided backpropagation and Gradient-weighted Class Activation Mapping, Grad-CAM) which provide quantitative information on explaining the model. Major results of this study are as follows. First, we successfully apply our deep learning models to the forecast of daily solar flare occurrence with TSS=0.65 and ApSS=0.61, without any preprocessing to extract features from data. Second, thanks to the attribution methods, we find that a result of the model is mainly determined by the areas near the polarity inversion lines of ARs. Third, the ARs with high Grad-CAM values produce more flares than those with low Grad-CAM values. Fourth, 9 SHARP parameters such as total unsigned vertical current, total unsigned current helicity, total unsigned flux, and total photospheric magnetic free energy density are well correlated with Grad-CAM values.