Characterizing Ultraviolet Bursts in a Solar Coronal Hole using Machine Learning Techniques

The Interface Region Imaging Spectrograph (IRIS) provides simultaneous imaging and spectral data of the chromosphere and transition region (TR) with great spatial, spectral and temporal resolution. A Solar coronal hole (CH) spectral observation by IRIS is studied for non-Gaussian Si IV profiles that fit the characteristics of UV burst events. Machine learning techniques such as k-means clustering and principal component analysis were applied on the data-set for classifying observed profiles into distinct groups. The least frequent representative profiles (RP), in terms of cluster size, managed to identify profiles that is characteristic for UV bursts events, i.e. complex, non-Gaussian shaped Si IV 1394 Å and Si 1403 Å profiles. It was found that UV burst like events are ubiquitous in a CH atmosphere but with less probability than in an Active Region (AR) atmosphere. The UV burst events were found to be associated with small-scale jet structures and compact brightenings. The simultaneous study of spectral profiles of Si IV 1394 Å, Si IV 1403 Å, C II 1335 Å, C II 1336 Å, Mg II k 2797 Å, Mg II h 2804 Å, and Mg II triplet 2798.7 Å and 2798.8 Å showed signatures of mass flows and intermittent heating in different layers of the Solar atmosphere that were associated with UV bursts. The obtained results suggest that UV bursts-like events in a CH condition as observed by IRIS are similar to the features of the TR explosive events observed by previous generation instruments. The IRIS UV bursts in CH are very much alike those observed in the AR, i.e the presence of complex Si IV spectral profiles that deviates strongly from a smooth Gaussian-line. The main difference is that the Ni II λ 1393 Å blend on the blue wing of Si IV 1394 Å are not as frequent and evident in CH spectra due to the low abundance of the cool chromospheric plasma in the CH environment as compared to the AR. Apart from the blend feature, these events are very much alike which conclude a close association between these categories of events