Developing Maximum Contaminant Loads for Georgia: Monitoring and Modelling Single Land Use Watersheds.
James Bradshaw1, David Radcliffe2, Mark Risse1, Mark Bakker1, C.R. Jackson1, Daniel Markewitz1, and K.P. Lichtenstein1. (1) Univ of Georgia, 3111 Miller Plant Science Bldg, Athens, GA 30602, (2) Dept of Crop & Soil Science, Univ of Georgia, 3111 Miller Hall, Athens, GA 30602
The 1972 Clean Water Act initiated the TMDL program to aid states in meeting water quality standards. A TMDL is the maximum amount of a pollutant that can be allowed into a stream in order for it to meet its assigned designated use. For TMDLs to be developed, USEPA requires that sources of contaminants be identified. Streams that do not meet their designated use are listed on the States 303(d) list. Once a stream is listed, a plan must be developed to set a TMDL for that stream. One accepted method of contaminant identification is through the use of watershed-scale models such as the Soil and Water Assessment Tool (SWAT). SWAT has been widely used to model water quality in large watersheds. When modelling a large watershed, it is almost impossible to determine if model parameters used to calibrate SWAT represent real-world values for a specific landuse. One way to determine if parameter values are accurate, is to model small, single landuse watersheds. This approach provides the opportunity to study how landuse and model parameters are related. A research project was initiated to study how model parameter values relate to land use. Flow, sediment, and nutrient data collected from two agricultural watersheds, were used to calibrate SWAT. One watershed (813 ha) is used primarily for poultry and cattle production. The second watershed (289 ha), is used primarily for dairy operations. Base-flow samples were collected weekly and analyzed for total suspended solids (TSS), and total phosphorus (TP).. Storm-flow samples were collected with ISCO automated samplers. These samples were analyzed for TSS and TP. The parameterization method for model calibration outlined by Mulla and Addiscott (1999) was used to calibrate the model. Results from this study may help to determine if parameter estimations during SWAT model calibration are accurate.