Modeling nitrate loading from watersheds to coastal waters of the Northumberland Strait
University of New Brunswick
The Canadian province of Prince Edward Island has experienced declining conditions in many estuaries for more than a decade. Nitrates, derived from intensive land-based agricultural activities, are thought to be the major driving force for these degrading conditions. In this research, a spatially lumped model was developed to estimate annual nitrate loads and concentrations coming from PEI watersheds. Nitrate attenuation, based on the width of riparian buffer zones, and the transport delay due to groundwater residence time are both accounted for in the model. A two-reservoir flow module is used to differentiate rapid flow from delayed flow, and these two flow paths are assigned nitrate concentrations that reflect the land-use composition of the watershed. To analyze the uncertainty of the results, key model parameters are assigned probability distribution functions. The model is then run several times with different sets of parameters chosen by a Latin hypercube sampling method. Twelve watersheds, with long-term monitoring data, were used for model calibration (R2=0.91), while 118 different watersheds were used for model verification (R2=0.82). The simulation results for the time period of 1996 to 2012 indicate a good agreement with observed average annual concentrations, especially for agricultural watersheds having catchment areas larger than 6 km2. Overall, it is concluded that lumped parameter models can be accurate and useful tools for simulating annual nitrate loadings from such watersheds, when detailed temporal and spatial agricultural land-use data are available. In PEI, watershed-based nitrate loadings were found to mainly derive from agricultural land, especially land in potato production, and reductions in loading will have to address nitrate leaching from such areas.