Abstract A modelling approach is presented that appears to have the potential to provide reliable assessments of pollution concentration in coastal zones and complex terrain, where the assumptions behind current operational air quality models often are inadequate. With the use of a numerical mesoscale model (NMM), physically consistent flow fields can be predicted, thereby providing higher spatial and temporal resolution in the meteorological fields than would be available from a limited number of observation points. These predictions are used to calculate mean trajectories of pollutant parcels, as well as to provide quantitative estimates of pollution concentration using two techniques. One technique, most relevant for point and line sources, uses mean and fluctuating velocities as derived from the mososcale model in order to estimate the spread of pollutant, while the second, which is applicable mainly to area sources, utilizes the advection-diffusion equation. Considering the scarcity of meteorological...