Auroral breakup detection in all-sky images by unsupervised learning

Abstract. Due to a large number of automatic auroral camera systems on the ground, the image data analysis requires more efficiency than what a human expert visual inspection can provide. Furthermore, there is no solid consensus on how many different types or shapes exist in auroral displays. We report the first attempt to classify auroral morphological forms by unsupervised learning method on an image set that contains both nightside and dayside aurora. We used six months of full-colour auroral all-sky images captured at a high-arctic observatory on Svalbard, Norway, in 2019–2020. The selection of images containing aurora was performed manually. These images were then input to a convolutional neural network called SimCLR for feature extraction. The clustered and fused features resulted in 37 auroral morphological classes. In the classification of auroral image data with two different time resolutions we found that the occurrence of eight morphological classes strongly increased when the image cadence was high (24 seconds), while the occurrence of 13 morphological classes experienced little or no change with changes in input image cadence. We therefore investigated the temporal evolution of the group of eight "active auroral classes". Time periods for which "active auroral classes" persisted for longer than two consecutive images with maximum cadence of six minutes coincided with ground-magnetic deflections and their occurrence was found clustered around the magnetic midnight. The active auroral onsets typically included vortical auroral structures and equivalent current patterns typical for substorms. Our findings therefore suggest that our unsupervised image classification method can be used to detect auroral breakups in ground-based image datasets with a temporal accuracy determined by the image cadence.