Abstract The prediction of fog using numerical weather forecasting models is a continuing challenge due to the combined influence of processes at different spatial and temporal scales as well as nonlinear interactions between them, especially in coastal regions. The focus of this work is on a specific mixing‐fog event observed during the “Toward Improving Coastal Fog Prediction” field campaign at the Ferryland field site, when a cold‐frontal air mass arriving from the northeast approached the Downs peninsula, Newfoundland, Canada. Simulations with the Weather Research and Forecasting model using a Large Eddy Simulation option (WRF‐LES) were used to study physical processes at different scales contributing to the life cycle of fog. The model was thoroughly evaluated by conducting simulations with a suite of available model options and comparing results with observations, and the best set of options selected were used for simulations of the present field case to reveal the formation, evolution, and dissipation of fog. Our analysis suggests four factors that play a role on the fog formation: (i) synoptic scale – advection of cold moist air over shallow warmer coastal waters creates a shallow marine boundary layer capped by inversion; (ii) mesoscale – low cloud formation strengthens the inversion by releasing latent heat at the top of the cloud, thus generating convective instability and downward mixing in the cloud layer; this further enhances the descent of the lower boundary (base) of the cloud layer; (iii) local scale – near‐surface turbulence induced by the collision of cold denser air mass with an orographic barrier can promote mixing, cooling of the air and fog initiation; and (iv) microscale – mixing of near‐saturated with saturated air and decrease in temperature drive the water vapor condensation. All described processes have significant roles and play together to provide favorable conditions to patchy fog/mist formation for the described case.