INTEGRATING SITE-SCALE AND REGIONAL MODELING STUDIES TO UNDERSTAND GROUNDWATER FLOW AND NITRATE TRANSPORT IN THE BEAR CREEK WATERSHED

Riparian buffers have been proposed to minimize agricultural impacts and improve water quality along streams in Iowa. The hypothesis of improved groundwater quality through buffers is being tested in the Bear Creek watershed, a 7,656 ha agricultural watershed in central Iowa. Multi-species riparian buffers, consisting of 20- to 40-m-wide strips of switchgrass, shrubs, and trees up to 14 years old, have been installed along 7 km sections of the watershed. Two sets of wells were used at the site scale to assess the nitrate-N removal efficiency of the buffers. At the three primary sites (Risdal North, Risdal South, and Strum sites), 41 monitoring wells have been sampled monthly for nitrate-N, chloride, and dissolved O₂ since 1996. Despite differences in geology and seasonal fluctuations, these sites all showed a decline in nitrate-N concentrations along groundwater flow paths to the creek. Multilevel piezometers were installed at seven locations within the watershed in 2001. Data collected during 2001-02 showed that five of seven sites consistently removed from 25 to 99.5 percent of nitrate-N in groundwater. In all cases, nitrate removal was favored by groundwater flow toward the creek, small groundwater velocities, and long groundwater residence times. The potential for these groundwater conditions to occur in the entire watershed and region (233,000 ha) was investigated using a regional, two-dimensional, analytic element model (GFLOW 2000) coupled with a parameter estimation code (UCODE). Calibrated model results indicate that groundwater flow is dominated by the South Skunk River, a regional discharge area to the west. The model predicts that Bear Creek may occur under losing, gaining, or flow-through conditions throughout its extent and that more groundwater discharge occurs into the east side of the creek than into the west side. Model-calculated, groundwater discharges for 90-m length of buffers were 47 m³/d at the Risdal North site, 41 m³/d at the Risdal South site, and 110 m³/d at the Strum site – similar to results obtained from previous field studies. Our investigation suggests that while site-scale studies are critical to understanding the process of nitrate-N removal in the hydrogeologic environments found along the creek, groundwater flow modeling can identify buffer areas with little or no “hydraulic potential” to improve groundwater quality.