A Probability-Based Daytime Algorithm for Sea Fog Detection Using GOES-16 Imagery

Fog is a hazardous weather event that can endanger navigation, aviation, and transportation. While human has several limitations in detecting and forecasting offshore fog, satellite remote sensing offers cost-effective images. In this study, a probability-based daytime sea fog detection algorithm, applied to geostationary operational environmental satellite (GOES) 16 satellite data over the Grand Banks offshore Eastern Canada, is presented and compared with the National Oceanographic and Atmospheric Administration (NOAA)'s Low Instrument Flight Rules (LIFR) probability map. Initially, clear-sky and ice cloud classes were delineated in the GOES-16 image and then the remaining pixels were assigned a fog probability by conducting small droplet proxy, spatial homogeneity, and temperature difference tests. Moreover, a green band was linearly interpolated using the first three bands of GOES-16 images to generate pseudotrue color composites. The resulting maps were evaluated both during an extended sea fog event and using several statistical measures. The average detection probability for the observed advection fog events was 66% for the proposed method, while that for NOAA's LIFR map was 38%. Furthermore, by thresholding the generated maps at the probability of 60%, the false alarm rate, probability of detection, hit rate, and Hanssen-Kuiper skill score were 0.09, 0.77, 0.83, and 0.68, respectively. The proposed method is operationally being used in this region to detect and monitor sea fog, facilitating safe navigation and aviation. This is the first study that uses GOES-16 for daytime fog detection and discusses a satellite-based solution for fog modeling in Grand Banks, NL.