High-Resolution Fog Detection over a Mountainous Lake Basin

Dense fog occurs frequently on the surface of complex mountainous areas in southern China, seriously affecting human life, industrial production, and transportation. To mitigate the risks brought by dense fog weather events, we take the Poyang Lake basin as an example and propose a new fog detection method with high spatiotemporal resolution. The method includes the following steps. Firstly, we use the deterministic eight-neighbor (D8) algorithm embedded in ArcGIS 10.2 to calculate the boundaries of the three-level subbasins of Poyang Lake and divide them into multiple groups according to the boundaries. Then, we utilize the Fengyun-4A (FY-4A) meteorological satellite data and the digital elevation models (DEMs), and set remote sensing thresholds in groups based on the cumulative distribution of the Gaussian function. Finally, we piece together the fog identification results of each group to obtain the final product. The study evaluates the effectiveness of the method for 48 heavy fog cases in the Poyang Lake basin from 2021 to 2023, using the probability of detection (POD), false alarm ratio (FAR), and critical success index (CSI). The results show that this method exhibits appreciable ability in identifying fog and can distinguish fog from other surface features, such as land surfaces, clouds, and low-level mist. The average POD, FAR, and CSI from 2021 to 2023 are 0.719, 0.109, and 0.660, respectively. The performance of this method shows obvious monthly variations, and the scores generally present a unimodal feature from September to June of the following year. The optimal detection periods for spring, autumn, and winter are 0600–0900, 0700–1000, and 0800–1030 local time (LT), respectively. Among the fog-prone seasons, detection performance is best in autumn, followed by winter, and worst in spring. In addition, this method enables the observation of dynamic changes, boundary textures, and other detailed variations in fog areas.