2026/03/24 by Piyush Teeloku, Zheqi Chen, Peter Taylor +1
Earth and Planetary Sciences · Environmental Science · #Oceanographic and Atmospheric Processes #Oil Spill Detection and Mitigation #Tropical and Extratropical Cyclones Research
paper · doi:10.1080/07055900.2026.2643835
openalex publication_date 2026/03/24 · openalex created_date 2026/03/25 · openalex updated_date 2026/08/01
In recent years, machine learning (ML) has gained popularity in the field of weather forecasting, particularly in areas where numerical weather prediction (NWP) models face challenges. One such area is fog prediction. Reduced visibility due to fog events poses serious challenges to transportation and public safety. Accurate representation of microphysics processes, radiation, boundary layer turbulence, and air–surface interaction is crucial in fog prediction. Traditional NWP models face limitations in accurately forecasting fog due to those complex processes. In this study, we propose a post-processing approach that combines the Weather Research and Forecasting (WRF) model forecasts with a machine learning classifier to improve fog prediction by distinguishing between fog and no-fog conditions. Using 12 years of data (2012–2023) for St John’s, Newfoundland and Labrador, and Yarmouth, Nova Scotia, Canada, the approach was tested on independent 2024 observations. Compared to forecasts based solely on liquid water content from WRF, the ML model achieved higher skill, with F1-score improvements of 13% at St John’s and 18% at Yarmouth. According to the results obtained, this approach demonstrates the potential of ML techniques to enhance operational fog forecasting capabilities.