Appropriate Use of Catchment Models for Water-Quality Target Setting and Land-use Management

Management of land use and the setting of water-quality targets on the whole-catchment scale are typically constrained by a lack of observational data on which to base decisions. Typically, GIS-based models are used to interpolate and extrapolate a limited observational data set to cover an entire catchment contaminant generation and transport models. We examine the impact of the intrinsic uncertainty in pseudo-data on predictions made using the catchment sediment and nutrient model, SedNet/ANNEX, and the consequences of this uncertainty for land-use management planning. The use of pseudo-data to estimate bulk soil properties (% clay, % phosphorus, % nitrogen) exhibited both a high bias and high standard deviation when compared with direct measurements. The high bias limits the usefulness of model predictions for setting quantitative targets. The high standard deviation requires model results to be aggregated over scales considerably larger than the intrinsic model grid to facilitate comparison of contaminant export rates between spatial units. Assessment of competing land management scenarios is not severely impacted by model uncertainty provided the interpretation of results is limited to relative changes in contaminant export.