DRAINWAT-BASED METHODS FOR ESTIMATING NITROGEN TRANSPORT IN POORLY DRAINED WATERSHEDS

摘要

Methods are needed to quantify effects of land use and management practices on nutrient and sediment loads at the watershed scale. Two methods were used to apply a DRAINMOD-based watershed-scale model (DRAINWAT) to estimate total nitrogen (N) transport from a poorly drained, forested watershed. In both methods, in-stream retention or losses of N were calculated with a lumped-parameter model, which assumes that N concentration decreases exponentially with residence (or travel) time in the canals. In the first method, daily field outflows predicted by DRAINWAT were multiplied by average N concentrations to calculate daily loads at the field edge. Travel time from the field edge to the watershed outlet was computed for each field for each day based on daily velocities predicted by DRAINWAT for each section of the canal-stream network. The second lumped-parameter method was similar but used predicted annual outflow to obtain annual load at the field edge. The load was transported to the watershed outlet, and the in-stream N loss was determined by using a constant average velocity (obtained by long-term DRAINWAT simulations), independent of season, for the entire canal-stream network. The methods were applied on a 2,950 ha coastal forested watershed near Plymouth, North Carolina, to evaluate daily, monthly, and annual export of nitrogen for a five-year (1996-2000) period. Except for some late spring and hurricane events, predicted daily flows were in good agreement with measured results for all five years (Nash-Sutcliffe coefficient, E = 0.71 to 0.85). Estimates of monthly total N load were in much better agreement (E = 0.76) with measured data than were the daily estimates (E = 0.19). Annual nitrogen load was predicted within 17% of the measured value, on average, and there was no difference ( a = 0.05) between measured and estimated monthly and annual loads. The estimates of annual N loads using travel time with a daily velocity yielded better results than with the constant average velocity. The estimated delivery ratio (load at the outlet/load at the field edge) for total N was shown to vary widely among individual fields depending on their location in the watershed and distance from the outlet. Both of the methods investigated can potentially be used with GIS in predicting impacts of land management practices on total N loads from poorly drained watersheds