Concentrations of phosphorus (P), the main limiting nutrient in freshwater ecosystems, need to be reduced, but this is difficult due to high spatial and temporal variations and limited resources. Reliable targeting of critical source areas, such as erosion-prone fields and parts of fields, is necessary to improve the cost efficiency of mitigation measures. We used high-resolution (2 m × 2 m) distributed modelling to calculate erosion risk for a large area (202 279 km2) covering > 90% of Swedish arable land. Comparison of model results with independent farmers' observations in a pilot catchment showed high spatial agreement. The modelled worst case scenario produced reasonable quantitative results comparable to measured 90th percentile values of suspended sediment (SS) loads at both field and small catchment scale (R2 = 0.81, p < 0.001). Overall, loads of SS, especially during extreme episodes, strongly governed losses of unreactive P and total P at both field and catchment scale.
Keywords: Critical source areas; Distributed modelling; Erosion; High-resolution; Phosphorus.