Date of Graduation


Document Type


Degree Name

Master of Science in Biological Engineering (MS)

Degree Level



Biological and Agricultural Engineering


Brian Haggard

Committee Member

Marty Matlock

Second Committee Member

Thad Scott


Applied sciences, Earth sciences, Prioritization, Watershed


The purpose of this study was to prioritize subwatersheds using water quality data and watershed characteristics. Water quality data was provided through studies by the Arkansas Natural Resources Commission, Beaver Water District, Arkansas Water Resources Center, and publicly available USGS gage data. A total of 114 sites across five HUC-8 watersheds were analyzed, including 12 USGS gages. Watershed characteristic data was retrieved from USGS and Arkansas GeoStore geodatabase repositories. A significant linear relationship between baseflow and stormflow nutrient concentrations was established allowing for the use of baseflow concentrations in the prioritization methodology. Pearson correlation, linear regression, classification and regression tree, and change point analysis were used to study relationships between watershed characteristics and four constituents; nitrate-nitrogen, total nitrogen, soluble reactive phosphorus, and total phosphorus. Human disturbance of the landscape, particularly forested area and agricultural production in the riparian buffer were the most significantly correlated with nutrient concentrations. The density of poultry houses within the watershed as well as a combined human disturbance index were also significantly correlated to nutrient concentrations. These relationships were used to develop prioritization methodologies for HUC-12 subwatersheds, ranking them in order of predicted constituent concentration. The first method utilized percent forest within the riparian buffer to separate watersheds by the predicted change point in nutrient concentrations; ultimately, those exhibiting less than 50% forested buffer were identified as a priority. The second method also used significant change points to classify nutrient trends as either high or low, but included multiple metrics: agricultural land use in the riparian buffer, forested riparian buffer, human use index, and poultry house density. Subwatersheds were ranked higher in priority as they increased in the number of predictors indicating high nutrient concentrations. Finally, as a way to corroborate the results, analysis of variance was performed on subwatersheds identified as a priority versus those that were not using available water quality data. Priority subwatersheds contained significantly higher nutrient concentrations. Empirical watershed models and prioritization schemes such as this one may provide a viable alternative to extensive deterministic watershed modeling in watersheds lacking adequate water quality data.