[en] We examined the consequences of the spatial heterogeneity of atmospheric ammonia (NH₃) by measuring and modelling NH₃ concentrations and deposition at 25 m grid resolution for a rural landscape containing intensive poultry farming, agricultural grassland, woodland and moorland. The emission pattern gave rise to a high spatial variability of modelled mean annual NH₃ concentrations and dry deposition. Largest impacts were predicted for woodland patches located within the agricultural area, while larger moorland areas were at low risk, due to atmospheric dispersion, prevailing wind direction and low NH₃ background. These high resolution spatial details are lost in national scale estimates at 1 km resolution due to less detailed emission input maps. The results demonstrate how the spatial arrangement of sources and sinks is critical to defining the NH₃ risk to semi-natural ecosystems. These spatial relationships provide the foundation for local spatial planning approaches to reduce environmental impacts of atmospheric NH₃. -- Highlights: •Local farm inventory provided field-level emissions for high resolution modelling. •Model-derived concentrations were compared against intensive spatial measurements. •Spatial arrangement of NH₃ sources and sinks is critical to environmental impact. •Average national emission factors were not appropriate for an NH₃ risk assessment. •Modelling at 1 km resolution did not capture the full spatial variability of NH_3. -- Fine scale resolution modelling to reproduce the spatial heterogeneity of atmospheric NH₃ concentrations and deposition is critical for NH₃ risk assessment on sensitive ecosystems