A mesoscale meteorological model (MM5) coupled with an advanced land surface model (PX LSM) is used in this study to model high-resolution (2 km) dry deposition velocity of ozone over the Swedish west coast, together with a newly revised dry deposition parameterization for air-quality models with emphasis on non-stomatal resistance. The important air-surface exchange processes for air quality (surface fluxes of heat, moisture and momentum) are also simulated by this model. The modelled subgrid scale variability of the dry deposition velocity and its dependence on land use, terrain height and synoptic conditions are investigated. It is found that a systematic difference in the deposition velocity modelled by different resolutions exists, and the difference varies diurnally and daily. The subgrid scale variation is significant, which has a clear impact on the area-averaged deposition velocity. The deposition velocity depends strongly on land use and weather conditions, but not on topography for the area studied. Meteorological conditions at subgrid scales play an important role in determining the deposition velocity. It is also concluded that the dry deposition velocity simulated in this study is reasonable, and that a 6-km resolution would be practically good enough to resolve the inhomogeneity of the surface properties for dry deposition studies in this area. The variation range of dry deposition velocity over different land use categories and the corresponding resistances are outlined. Moreover, the difference in the estimated dry deposition velocitiy between the methods using fractional land use and using dominant land use is compared. (c) 2005 Elsevier Ltd. All rights reserved.