Three-dimensional fog forecasting in complex terrain

Fog in complex terrain shows large temporal and spatial variations that can only be simulated with a three-dimensional model, but more modifications than simply increasing the resolution are needed. For a better representation of fog, we present a second-moment cloud water scheme with a parametrization of the Kohler theory which is combined with the mixed-phase Ferrier microphysics scheme. The more detailed PAFOG microphysics produce many differences to the first-moment Ferrier scheme and are responsible for the typically low liquid water content of fog. The inclusion of droplet sedimentation in the Ferrier scheme cannot reproduce the results obtained with PAFOG, as there is a large sensitivity to the sedimentation velocity. With explicitly predicted droplet number concentrations, sedimentation of cloud water can be modelled with variable fall speeds, which mainly affects the vertical distribution of cloud water and the end of the fog's life cycle. The complex topography of the Swiss Alps and their surroundings are used for model testing. As the focus is on the model's ability to forecast the spatial distribution of fog, cloud patterns derived from high-resolution MSG satellite data, rather than few point observations from ground stations, are used. In a five-day period of anticyclonic conditions, the satellite-observed fog patterns showed large day-to-day variations from almost no fog to large areas of fog. This variability was very well predicted in the three-dimensional fog forecast. Furthermore, the second-moment cloud water scheme shows a better agreement with the satellite observations than its first-moment counterpart. For model initialization, the complex topography is actually a simplifying factor, as cold air flow and pooling dominate the more uncertain processes of evapotranspiration or errors in the soil moisture field. Copyright (C) 2010 Royal Meteorological Society