Masters Theses

Date of Award


Degree Type


Degree Name

Master of Science


Environmental and Soil Sciences

Major Professor

Michael E. Essington

Committee Members

Forbes R. Walker, Carol P. Harden


Sediment is often listed as one of the main contributors to the impairment of surface waters throughout the United States. Sediment source identification is difficult in watersheds with complex combinations of land-uses and non-point sources because of the complexities involved in correlating water quality data, which are relatively easy to collect, to the source of a degrading component. The elemental properties of a particular soil on the landscape may be viewed as a “fingerprint”. A comparison of the elemental fingerprints of potential sources and in-stream sediment may be used to establish sediment source. The objectives of this investigation were to characterize the elemental content of suspended stream sediment and potential sources of sediment in an impaired watershed, Pond Creek watershed in east Tennessee (HUC: TN06010201013), and to use multivariate statistical techniques to identify and quantify sediment sources in the watershed. Potential sediment source samples were collected throughout the watershed and suspended sediment samples at two locations. Subsamples of the <53>μm material and suspended sediment were subjected to total dissolution, HNO3-extraction, and Mehlich 3-extraction. Descriptive statistics suggested that each dataset contained considerable heterogeneity. The source samples were grouped according to land management and position in the landscape. The results of a Kruskal-Wallis rank test and discriminant function analysis indicated that for all three datasets the elemental variability of the samples was not sufficient to differentiate the source and sediment samples and characterize the suspended sediment sources using the initial group definitions. When using all available elemental data from each dataset the groups defined by cluster analysis and canonical discriminant analysis did not match the contents of the initially defined groups. The composition of the clusters varied from one dataset to another, making it difficult to draw conclusions concerning the cluster contents, or to identify sources of suspended sediment. The lack of elemental content variability for differentiating the source and sediment samples and characterizing the suspended sediment sources is likely an artifact of the watershed sampling procedure that was employed, which was directed towards sampling sources likely to be contributing to the suspended sediment load in Pond Creek.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."