Search

We present a binary hypothesis test for detecting clear sky in auroral all‐sky images based on single‐wavelength keograms. The coefficient of variation c, the ratio of the sample standard deviation to the mean over elevation angle along the meridian, is the test statistic. After image‐correcting keograms and excluding dark sky intervals, detection performance is compared to true conditions as determined by Advanced Very High Resolution Radiometer satellite imagery. The cloud mask, an index of cloud cover, is selected at the corresponding nearest time and location to the site of a meridian spectrograph at Poker Flat Research Range. With training data from 2014 to 2016, theoretical Rayleigh distributions fit to the histograms show a decision threshold of 0.40 could yield an accuracy of about 80%. Separately, we numerically compute the false alarm and missed detection statistics of the greenline 557.7 nm emission and of the redline 630.0 nm emission. We find a threshold of 0.25 for the greenline c maximizes the percent of events correctly identified at 76%. Applied to testing data from 2015 to 2017, the 0.25 threshold yields an accuracy of 68%. Diffuse aurora can have coefficient of variation around 0.2 to 0.5, which would be included by the numerical minimum, but partly excluded by the theoretical model obtained. Numerical results are a few percent worse for the redline emission. Plain Language Summary: Clouds in the sky are a problem for scientists trying to view space beyond. For upper atmospheric scientists, clouds can obscure or scatter auroral light in scientific auroral cameras, making it hard to identify, locate, and track auroral shapes. This paper shows a way to simply and automatically detect clouds using a north‐to‐south line scan of a single color of light from the sky over time, known as a keogram. We compute the ratio of the variation in pixel intensity to the average pixel intensity, for each north‐to‐south scan. Excluding dark sky periods, a large ratio means that the sky is clear, and a small ratio that the sky is cloudy. We find the method works with about a 70% correct rate using red or green auroral light. With this method we can eliminate data during cloudy conditions for any auroral studies that require clear sky conditions. Key Points: Auroral keogram coefficient of variation is used to decide if the sky is cloudy or clear, verified with NOAA satellite data from 2014 to 2017The coefficient of variation distinguishes clear from cloudy auroral conditions with as much as 75% accuracyThis method is intended for subsequent automation of auroral all‐sky image analysis [ABSTRACT FROM AUTHOR]