A Spatial Analysis of Streambank Heterogeneity and its Contribution to Bank Stability

Streambank erosion is a function of fluvial detachment and geotechnical failure mechanisms working in combination to cause bank retreat. It is generally agreed that bank stability is dependent on both types of erosion; however, few studies have attempted to correlate the driving and resisting forces between the two. It has been proposed that: (1) streambanks possess a spatial structure and dependence of non-erodible resistant structures such as root masses and rocks; (2) streambanks naturally “armor” themselves from fluvial erosion with a combination of hard points and resistive soil; and (3) the stability of the streambank can be predicted by the amount of composite fluvial resistance, thereby connecting fluvial resistance and geotechnical stability. These hypotheses were tested through extensive field analysis, spatial statistical methods, and multivariate statistics. Eighteen streambanks sites with cohesive sediment structures in the Eastern Tennessee Ridge and Valley Ecoregion were sampled for in-situ erodibility and critical shear stress as well as the spatial distribution of nonerodible hard points. Using a combination of nearest neighbor, join count, and indicator variogram statistics it was discovered that streambanks in this region possess natural clusters of these non-erodible structures. However, the study was unsuccessful in determining the cause of spatial dependency of non-erodible clusters between various banks, as not all banks exhibit this quality. Multiple linear regression was utilized to compare all streambanks, with the results indicating that the density of these clustered hard points displays a positive linear relationship with the critical shear strength of the surrounding soil matrix, suggesting a coordination between the two and possible natural armoring of the composite bank structure. Additionally, the USDA Bank Stability and Toe Erosion Model (BSTEM) was utilized to determine the relative stability of these streambanks. With this information, both discriminate analysis and multiple linear regression were applied to determine that stable streambanks are controlled by a combination of the respective stream power, average soil critical shear strength, standard deviation of the critical shear strength, bank height, and cluster density of non-erodible structures. This information helped derive relevant interactions for stable stream bank analysis including possible stream restoration and site investigation applications.