Land-use management and parameterization as sources of uncertainty in hydrologic models: A Maumee River Watershed case study

The use of hydrological models in water management and policy has grown with increasing demand for scientifically credible solutions to rising environmental concerns. However, difficulty in quantifying uncertainty is a key limitation for interpreting model results. Uncertainties associated with parameters and data inputs are commonly reported. While some studies reported the relative effects of specific farm management implementation, the type and timing of farm field management operations remain some of the most uncertain data inputs and have been poorly studied in the context of model uncertainty. This study aims to assess the relative role of two potential drivers of uncertainty: 1) assumptions made for farm management; and 2) model parameterization through analysis of a Soil and Water Assessment Tool (SWAT) model of the Maumee River Watershed. We identified a suite of model simulations representing management practices of known importance to the region, and we identified a set of commonly calibrated model parameters and a set of prescribed value combinations representing the range of plausible values for these parameters. SWAT was run over each unique combination of parameter sets, and management realizations. Model outputs were compared with observations to quantify and attribute uncertainty to management inputs and parameterization. We examined the sensitivity of modeled outlet-level discharge and nutrient loading and found that parameterization and management were large contributors to uncertainty across all water quality outputs examined. Furthermore, model uncertainty in discharge was dominated by parameterization, while uncertainty in nutrient loading was dominated by management inputs. Based on the results, we suggest that when developing models for informing decision making, management practices that are implemented using the best available spatial and temporal data likewise undergo a management implementation sensitivity analysis and that these results are reported in the context of uncertainty, similar parameter uncertainty standards.