Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/1/2011
Publication Date: 5/1/2011
Publication URL: http://hdl.handle.net/10113/50050
Citation: Zuercher, B.W., Flanagan, D.C., Heathman, G.C., Frankenberger, J.R. 2011. Evaluation of the AnnAGNPS model for atrazine prediction in NE Indiana. Transactions of the ASABE. 54(3):811-825. Interpretive Summary: Agriculture is extremely important in providing the food and fiber needs of our society as economically as possible. However, some aspects of modern agriculture, such as use of chemical pesticides, can impact off-site natural resources, such as fresh water supplies. In northeastern Indiana, the herbicide atrazine is used to economically control weeds during the production of corn. Unfortunately, atrazine has also been found in the drinking water supply for the city of Fort Wayne, and the water treatment plant there must regularly test and treat their intake water during the spring of each year to reduce the contaminant level. Use of computer simulation models can help us to assess the amount of atrazine and other agricultural chemicals that can be transported in runoff into ditches and rivers. These models can also be used to determine how various land management practices, such as conservation tillage, buffer strips, tile drainage, etc. can affect and alter the movement of chemicals. In this study we applied the AnnAGNPS (Annualized Agricultural Non-Point Source pollution) model to two watersheds of different size in the drainage basin of the St. Joseph River (the river that supplies water to Ft. Wayne, Indiana). It was possible to match fairly well the amount of predicted stream flow with that measured. When modeling the amount of atrazine herbicide in the water, it was not initially possible to predict it well, however after making a correction to the AnnAGNPS model itself we were then able to successfully calibrate the model to match observed data in the smaller of the two watersheds. In the larger watershed, the model matched the observed trends but was not able to accurately match the peak atrazine concentrations in the ditch runoff water, though this may have partially been due to the quality of the observed data. Further work with this model is needed for it to be a useful tool to assess management practice effects on water quality. This research impacts conservation agency personnel, extension agents, other researchers, farmers, landowners, and the public affected by water pollution from agriculture.
Technical Abstract: The Annualized Agricultural Non-Point Source (AnnAGNPS) pollution model was developed for simulation of runoff, sediment, nutrient, and pesticide losses from ungauged agricultural watersheds. Here, the model was applied to the 707 km^2 Cedar Creek Watershed (CCW) and the 45 km^2 Matson Ditch Sub-Catchment (MDS), which are predominantly (>85%) agricultural, with major crops of corn and soybeans. Atrazine herbicide is of significant concern, as the St. Joseph River is the source of drinking water for the city of Fort Wayne, Indiana with Cedar Creek being the main tributary. Major objectives were to evaluate the ability of AnnAGNPS to simulate runoff and atrazine concentrations in uncalibrated, calibrated, and validation modes. Data sources for the model inputs included USGS Digital Elevation Model for topography, NRCS spatial SSURGO soils data, and the USDA-NASS cropland data layer. Observed flow data for CCW were available from a USGS gauging station and atrazine data was from the Saint Joseph River Watershed Initiative (SJRWI) database. Flow and atrazine concentration data were available from a National Soil Erosion Laboratory (NSERL) water quality sampling site for the MDS. In an uncalibrated mode, flow discharge predictions by AnnAGNPS were satisfactory at the CCW scale, but could be improved through calibration. Flow discharge for both CCW and MDS could be well matched with observed during model calibration and validation. AnnAGNPS predictions of atrazine concentrations in runoff water were very poor and it was impossible to improve the results through any type of calibration. Inspection of the model source code revealed a unit conversion error in the runoff value being input to the pesticide routine, which when corrected greatly improved the results. The corrected AnnAGNPS model code could be satisfactorily calibrated and validated for predictions of atrazine concentrations in the MDS, but not in the CCW.