Model based nitrate TMDLs for two agricultural watersheds of southeastern Minnesota

In this study, a set of nitrogen reduction strategies were modeled to evaluate the feasibility of improving water quality to meet total maximum daily loads (TMDLs) in two agricultural watersheds. For this purpose, a spatial-process model was calibrated and used to predict monthly nitrate losses (1994-96) from Sand and Bevens Creek watersheds located in south-central Minnesota. Statistical comparison of predicted and observed flow and nitrate losses gave r² coefficients of 0.75 and 0.70 for Sand Creek watershed and 0.72 and 0.67 for Bevens Creek watershed, respectively. Modeled alternative agricultural management scenarios included: six different N application rates over three application timings and three different percentages of crop land with subsurface drainage. Predicted annual nitrate losses were then compared with nitrate TMDLs assuming a 30% reduction in observed nitrate losses is required. Reductions of about 33 (8.6 to 5.8 kg/ha) and 35% (23 to 15 kg/ha) in existing annual nitrate losses are possible for Sand and Bevens Creek watersheds, respectively, by switching the timing of fertilizer application from fall to spring. Trends towards increases in tile-drained crop land imply that attaining nitrate TMDLs in future may require other alternative management practices in addition to fertilizer management such as partial conversion of crop land to pasture.