[en] With machine learning entering into the awareness of the heliophysics community, solar flare prediction has become a topic of increased interest. Although machine-learning models have advanced with each successive publication, the input data has remained largely fixed on magnetic features. Despite this increased model complexity, results seem to indicate that photospheric magnetic field data alone may not be a wholly sufficient source of data for flare prediction. For the first time, we have extended the study of flare prediction to spectral data. In this work, we use Deep Neural Networks to monitor the changes of several features derived from the strong resonant Mg II h and k lines observed by the Interface Region Imaging Spectrograph. The features in descending order of predictive capability are: the triplet emission at 2798.77 Å, line core intensity, total continuum emission between the h and k line cores, the k/h ratio, line width, followed by several other line features such as asymmetry and line center. Regions that are about to flare generate spectra that are distinguishable from non-flaring active region spectra. Our algorithm can correctly identify pre-flare spectra approximately 35 minutes before the start of the flare, with an AUC of 86% and an accuracy, precision, and recall of 80%. The accuracy and AUC monotonically increase to 90% and 97%, respectively, as we move closer in time to the start of the flare. Our study indicates that spectral data alone can lead to good predictive models and should be considered an additional source of information alongside photospheric magnetograms.

[en] A three-dimensional picture of the solar atmosphere’s thermodynamics can be obtained by jointly analyzing multiple spectral lines that span many formation heights. In Paper I, we found strong correlations between spectral shapes from a variety of different ions during solar flares in comparison to the quiet Sun. We extend these techniques to address the following questions: which regions of the solar atmosphere are most connected during a solar flare, and what are the most likely responses across several spectral windows based on the observation of a single Mg ii spectrum? Our models are derived from several million IRIS spectra collected from 21 M- and X-class flares. We applied this framework to archetypal Mg ii flare spectra and analyzed the results from a multiline perspective. We find that (1) the line correlations from the photosphere to the transition region are highest in flare ribbons. (2) Blueshifted reversals appear simultaneously in Mg ii, C ii, and Si iv during the impulsive phase, with Si iv displaying possible optical depth effects. Fe ii shows signs of strong emission, indicating deep early heating. (3) The Mg ii line appears to typically evolve a blueshifted reversal that later returns to line center and becomes single peaked within 1–3 minutes. The widths of these single-peaked profiles slowly erode with time. During the later flare stages, strong red-wing enhancements indicating coronal rain are evident in Mg ii, C ii, and Si iv. Our framework is easily adaptable to any multiline data set and enables comprehensive statistical analyses of the atmospheric behavior in different spectral windows.

[en] Spectral lines allow us to probe the thermodynamics of the solar atmosphere, but the shape of a single spectral line may be similar for different thermodynamic solutions. Multiline analyses are therefore crucial, but computationally cumbersome. We investigate correlations between several chromospheric and transition region lines to restrain the thermodynamic solutions of the solar atmosphere during flares. We used machine-learning methods to capture the statistical dependencies between six spectral lines sourced from 21 large solar flares observed by NASA’s Interface Region Imaging Spectrograph. The techniques are based on an information-theoretic quantity called mutual information (MI), which captures both linear and nonlinear correlations between spectral lines. The MI is estimated using both a categorical and numeric method, and performed separately for a collection of quiet Sun and flaring observations. Both approaches return consistent results, indicating weak correlations between spectral lines under quiet Sun conditions, and substantially enhanced correlations under flaring conditions, with some line-pairs such as Mg ii and C ii having a normalized MI score as high as 0.5. We find that certain spectral lines couple more readily than others, indicating a coherence in the solar atmosphere over many scale heights during flares, and that all line-pairs are correlated to the GOES derivative, indicating a positive relationship between correlation strength and energy input. Our methods provide a highly stable and flexible framework for quantifying dependencies between the physical quantities of the solar atmosphere, allowing us to obtain a three-dimensional picture of its state.