Prediction of dissolved reactive phosphorus losses from small agricultural catchments: calibration and validation of a parsimonious model

摘要

Eutrophication of surface waters due to diffuse phosphorus (P) lossescontinues to be a severe water quality problem worldwide, causing the lossof ecosystem functions of the respective water bodies. Phosphorus in runoffoften originates from a small fraction of a catchment only. Targetingmitigation measures to these critical source areas (CSAs) is expected to bemost efficient and cost-effective, but requires suitable tools.Here we investigated the capability of the parsimoniousRainfall-Runoff-Phosphorus (RRP) model to identify CSAs in grassland-dominatedcatchments based on readily available soil and topographic data. Aftersimultaneous calibration on runoff data from four small hilly catchments onthe Swiss Plateau, the model was validated on a different catchment in thesame region without further calibration. The RRP model adequately simulatedthe discharge and dissolved reactive P (DRP) export from the validationcatchment. Sensitivity analysis showed that the model predictions were robustwith respect to the classification of soils into "poorly drained" and"well drained", based on the available soil map. Comparing spatialhydrological model predictions with field data from the validation catchmentprovided further evidence that the assumptions underlying the model are validand that the model adequately accounts for the dominant P export processes inthe target region. Thus, the parsimonious RRP model is a valuable tool thatcan be used to determine CSAs. Despite the considerable predictive uncertaintyregarding the spatial extent of CSAs, the RRP can provide guidance for theimplementation of mitigation measures. The model helps to identify thoseparts of a catchment where high DRP losses are expected or can be excludedwith high confidence. Legacy P was predicted to be the dominant source forDRP losses and thus, in combination with hydrologic active areas, a high riskfor water quality.