Application and Interpretation of Deep Learning for Identifying Pre-emergence Magnetic Field Patterns

Abstract

Abstract Magnetic flux generated within the solar interior emerges to the surface, forming active regions (ARs) and sunspots. Flux emergence may trigger explosive events—such as flares and coronal mass ejections, and therefore understanding emergence is useful for space-weather forecasting. Evidence of any pre-emergence signatures will also shed light on subsurface processes responsible for emergence. In this paper, we present a first analysis of EARs from the Solar Dynamics Observatory/Helioseismic Emerging Active Regions dataset using deep convolutional neural networks (CNN) to characterize pre-emergence surface magnetic field properties. The trained CNN classifies between pre-emergence line-of-sight magnetograms and a control set of nonemergence magnetograms with a true skill statistic (TSS) score of approximately 85% about 3 hr prior to emergence and approximately 40% about 24 hr prior to emergence. Our results are better than a baseline classification TSS obtained using discriminant analysis (DA) of only the unsigned magnetic flux, although a multivariable DA produces TSS values consistent with the CNN. We develop a network-pruning algorithm to interpret the trained CNN and show that the CNN incorporates filters that respond positively as well as negatively to the unsigned magnetic flux of the magnetograms. Using synthetic magnetograms, we demonstrate that the CNN output is sensitive to the length scale of the magnetic regions, with small-scale and intense fields producing maximum CNN output and possibly a characteristic pre-emergence pattern. Given increasing popularity of deep learning, the techniques developed here to interpret the trained CNN—using network pruning and synthetic data—are relevant for future applications in solar and astrophysical data analysis.