|Chin, David - UNIVERSITY OF MIAMI|
|Sakura-Lemessy, Donna - ALBANY STATE UNIV.|
|Gay, Paige - UNIVERSITY OF GEORGIA|
Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: January 14, 2009
Publication Date: January 30, 2009
Citation: Chin, D.A., Sakura-Lemessy, D., Bosch, D.D., Gay, P. 2009. Watershed-Scale Fate and Transport of Bacteria. Transactions of the American Society of Agricultural and Biological Engineers. Vol 52(1):145-154. Interpretive Summary: The leading cause of impairment of rivers and streams in the United States is excessive levels of pathogen-indicator bacteria, and the most common indicator bacteria in freshwater bodies are fecal coliform and Escherichia coli. Computer simulation models are a useful tool for examining the processes which lead to bacteria within freshwater sources and methods which can reduce their delivery to these sources. However, the study of bacteria within streams and particularly computer simulation of their delivery to streams is a fairly new science with high uncertainty in the process. Over the 7-year period from 1996 to 2002, the USDA-ARS Soil Water Assessment Tool (SWAT) model provided a more accurate representation of fecal-coliform concentrations in a south Georgia stream than did the Hydrological Simulation Program - Fortran (HSPF) model which is commonly used by the USEPA for risk assessment. These results will provide guidance on the best methods to be used for simulating bacterial concentrations in freshwater streams, and more importantly helping to identify the potential sources of pathogen-indicator bacteria to these streams.
Technical Abstract: The added dimensionality provided by using multiple models to predict the fate and transport of bacteria at the watershed scale were investigated. Both HSPF and SWAT were applied to the 15.6 km2 catchment K of the Little River Experimental Watershed (LREW) in Georgia. Over the 7-year period from 1996 to 2002, SWAT provided a more accurate description of fecal-coliform concentrations with a NSE of 0.73 compared to 0.33 for HSPF. For this watershed, the SWAT model provided more accurate estimates of fecal-coliform concentrations at the watershed outlet than did HSPF. Based on this comparative analysis, it can be inferred that elevated levels of fecal coliform in the receiving stream are primarily due to natural sources in the riparian areas of the watershed. This insight could not be achieved by using only the HSPF model, which indicated a much greater contribution from ground-water and agricultural non-point sources. A model-averaging approach in which a weighted-average of the HSPF and SWAT predictions are used to predict streamflow and bacteria concentrations in the receiving stream provided only marginal improvements over the SWAT model alone.