Nighttime fog and low stratus detection under multi-scene and all lunar phase conditions using S-NPP/VIIRS visible and infrared channels

A scheme for satellite remote sensing is proposed to detect nighttime fog and low stratus (FLS) by combining visible, mid-infrared, and far-infrared channels. The S-NPP/VIIRS dataset and ERA5 reanalysis data are primarily used, and a comprehensive threshold system is established through statistical analysis, simulation calculations, and sensitivity experiments. 98 cases of nighttime FLS occurring from 2012 to 2020 in China, the United States, and surrounding areas are selected for algorithm validation, utilizing the global surface meteorological observations as comparison data. Preliminary results from the analysis of four typical cases indicate that the algorithm is temporally suitable for all lunar phase conditions from new moon to full moon at night, and spatially applicable to various types of underlying surfaces. The accuracy evaluation results of 14,378 satellite-ground matching samples further show that the algorithm has high accuracy overall, with a POD of 0.86, CSI of 0.81, and FAR of 0.06. The accuracy is highest in winter, lowest in summer, and intermediate in spring and autumn. The missed detections and false alarms predominantly occur at the edge of clouds, which may be caused by parallax and time difference between satellite and ground observations.