Abstract:The monitoring network for a river system is designed to provide information about water quantity and quality. The development of Watershed Protection Plans and Total Maximum Daily Loads require systematic monitoring of waterbodies. In this study, optimum water quality monitoring networks were selected to assess E. coli loads in the Guadalupe River and San Antonio River basins. A Genetic Algorithm (GA) was applied to select monitoring stations using the mean annual E. coli flux from the Spatially Referenced Regression Model on Watershed Attributes (SPARROW). The objectives of the GA were to minimize the number of monitoring stations, include large values of the mean annual E. coli flux, and minimize uncertainty of the flux estimations. Constraints related to the monitoring of critical locations were included in a multi-objective optimization problem. The SPARROW model was applied to the optimized GA solution sets, which were compared using the objective values and statistical indices. The best GA-generated alternative set adequately represented the San Antonio River basin, in good agreement with a previous study conducted using only SPARROW. The application of the GA ensured the inclusion of the monitoring stations with large values of E. coli flux, which reflected high-risk areas within the watershed.

The two main rivers of southeast Texas: Guadalupe and San Antonio have shown high temporal increase in bacteria concentration during the last decade. The SPAtially Referenced Regression On Watershed (SPARROW) attributes model, developed by the U.S. Geological Survey (USGS), has been applied to predict the fluxes and concentrations of contaminants in unmonitored streams and to identify the sources of these contaminants. This model identifies every reach as a basic network unit to distribute the sources, delivery, and attenuation factors. The model is data intensive and implements nonlinear regression to solve the parsimonious relations for describing various watershed processes. This study explored watershed and hydrological characteristics (land uses, precipitation, human and animal population, point sources, areal hydraulic load and drainage density, etc.) as the probable sources and delivery mechanisms of waterborne pathogens and their indicator (Escherichia coli [E. coli]) in the Guadalupe and San Antonio River basins. The effect of using various statistical indices for model selection on the final model's ability to explain the various E. coli sources and transport processes was also analyzed.