Abstract
Ground-based cloud detection at nighttime is achieved by using cameras, lidars, and ceilometers. Despite these numerous instruments gathering cloud data, there is still an acknowledged scarcity of information on quantified local cloud cover, especially at nighttime. In this study, a digital camera is used to continuously collect images near the sky zenith at nighttime in an urban environment. An algorithm is developed to analyze pixel values of images of nighttime clouds. A minimum threshold pixel value of 17 is assigned to determine cloud occurrence. The algorithm uses temporal averaging to estimate the cloud fraction based on the results within the limited field of view. The analysis of the data from the months of January, February, and March 2015 shows that cloud occurrence is low during the months with relatively lower minimum temperature (January and February), while cloud occurrence during the warmer month (March) increases.