NIGHT TIME CLOUDS DETECTION ABOVE PERMATAPINTAR OBSERVATORY USING ALL-SKY IMAGERY

Abstract

We proposed a method to detect clouds in a suburban class night sky for the larger purpose of astronomical site testing. A 'threshold criterion' approach was adopted to discriminate between the pixels representing the foreground clouds from the pixels representing the background sky in a single all-sky image. This method was developed based on all-sky images captured at the PERMATApintar Observatory (PpO) in Selangor (2°55'02" N, 101°47'17" E), where the night sky has been categorised as a suburban class night sky. The night sky conditions were divided into three categories depending on the cloud cover: clear, partially cloudy, and overcast skies. Samples of all-sky images for each night sky condition were selected and respective histogram images were generated. These samples were then used to inductively derive the threshold criterion based on the skewness and peak values of the image's histogram. This sky/cloud threshold will enable pixels representing foreground clouds to be discriminated from the pixels representing the background sky under each type of night sky conditions. Our work found that the night sky over PpO requires two thresholds to accurately distinguish between cloud and sky pixels due to the sky glow effect. The first threshold is based on the peak value of the image's histogram. If an image's histogram has a peak value ≥ 80, then the image is considered a clear sky. Otherwise, the image is considered cloudy or overcast sky if the peak value is < 80.