From field to stream: Rapid runoff through agricultural tile drainage systems within the Minnesota River basin

Sediment loads in the Minnesota River have been linked to post-settlement rapid sedimentation of Lake Pepin, a naturally dammed lake within the Mississippi River channel in southern Minnesota. Nutrient loads in midwestern streams have been linked to the hypoxia problem in the Gulf of Mexico. Agriculture is the dominant land use, and the prevalence of agricultural drainage systems within this glaciated terrain provides direct pathways for rapid surface and subsurface runoff of peak events. Three sampling stations in the Minnesota River basin monitored paired surface-subsurface drainage discharges, and automatic samplers collected water samples during snowmelt and storm events. Samples were analyzed for sediment, and for dissolved cation and anion concentrations, including nitrate- and nitrite-nitrogen and total phosphorus. Drainage responded to recharge events (heavy rainfall and snowmelt) within minutes to hours, and dye traces indicated macropores provided direct connection of the soil surface to subsurface drainage. Peak sediment concentrations in runoff preceded peak discharges, in a non-linear hysteresis pattern. Normalized storm hydrographs revealed consistent water quality-discharge relationships, enabling model estimation of unsampled peak events and total annual loads. Surface runoff into tile inlets carried about 10% of the water, 40% of the sediment, 45% of the phosphorus and 10% of the nitrogen, while subsurface runoff carried about 90% of the water, 60% of the sediment, 55% of the phosphorus and 90% of the nitrogen in annual combined surface-subsurface flow. Peak events were important to annual loading: 62% of the water and 71% of the sediment loading occurred in the first 24 hours of discharge response; 77--97% of annual sediment yield in annual total runoff was derived from loading during events. Moldboard tillage and removal of vegetated buffers at one sampling station increased peak sediment concentrations from 3,000 to 80,000 mg/l in spring runoff, producing seven times greater annual sediment loads than conservation methods employed by the landowners. Runoff loads displayed a scale effect; a 24-ha area with 10 upgradient surface inlets produced much greater loading than three smaller, average 3.5-ha areas with one surface inlet each. Using observed sediment delivery ratios, sediment loads from drained cropland would be equivalent to over 75% of the annual sediment load to Lake Pepin. Using published in-stream nutrient loss rates, the nutrient loads from drained cropland would be equivalent to 5--7% of the annual nitrogen load to the Gulf of Mexico.