This study evaluates the representation of the life cycle of a radiation fog case study observed at the Cabauw 213‐m tower (Netherlands) facility by the WRF single‐column model, and aims to advance the understanding of the model behaviour, which will assist to set research priorities for the future. First an ensemble of sixteen WRF configurations that vary in parameterization schemes for the PBL, land surface, longwave radiation, and microphysics are evaluated. Next, we perform a sensitivity study to examine which physical process is most crucial in modelling the fog, i.e. soil heat diffusivity, the CO2 concentration (representing clear‐sky longwave radiation), the vapour diffusion to droplets, and the turbulent mixing. Subsequently, we study whether these perturbations can improve the model representation, and on the other hand whether they can explain the model behaviour of the sixteen ensemble members. Results are displayed in process diagrams. We find that behaviour of the ensemble can be explained by a variations in the soil heat diffusivity and the turbulent mixing. However their sensitivities orient in approximately the same direction, and as such, errors in the formulation of the boundary‐layer scheme can be hidden by compensating errors in the land‐surface scheme. In addition, we find that simultaneous perturbations in the soil heat diffusivity and turbulent mixing do not result in the same results as superpositioning of the individual perturbations.