Analysis Algorithm for Sky Type and Ice Halo Recognition in All-Sky Images

Abstract

Abstract. Halo displays, in particular the 22 ∘ halo, have\nbeen captured in long time series of images obtained from total sky imagers\n(TSIs) at various Atmospheric Radiation Measurement (ARM) sites. Halo\ndisplays form if smooth-faced hexagonal ice crystals are present in the\noptical path. We describe an image analysis algorithm for long time series\nof TSI images which scores images with respect to the presence of\n22 ∘ halos. Each image is assigned an ice halo score (IHS) for\n22 ∘ halos, as well as a photographic sky type (PST), which\ndifferentiates cirrostratus (PST-CS), partially cloudy (PST-PCL), cloudy\n(PST-CLD), or clear (PST-CLR) within a near-solar image analysis area. The\ncolor-resolved radial brightness behavior of the near-solar region is used\nto define the discriminant properties used to classify photographic sky type\nand assign an ice halo score. The scoring is based on the tools of\nmultivariate Gaussian analysis applied to a standardized sun-centered image\nproduced from the raw TSI image, following a series of calibrations,\nrotation, and coordinate transformation. The algorithm is trained based on a training set for each class of images. We present test results on halo\nobservations and photographic sky type for the first 4\xa0months of the year\n2018, for TSI images obtained at the Southern Great Plains (SGP) ARM site. A\ndetailed comparison of visual and algorithm scores for the month of March\xa02018 shows that the algorithm is about 90\u2009% reliable in discriminating the\nfour photographic sky types and identifies 86\u2009% of all visual halos\ncorrectly. Numerous instances of halo appearances were identified for the\nperiod January through April 2018, with persistence times between 5 and 220\u2009min. Varying by month, we found that between 9\u2009% and 22\u2009% of\ncirrostratus skies exhibited a full or partial 22 ∘ halo.