Abstract:
Bedrock shoals are geomorphic features located in river systems throughout the world. They are commonly found in bedrock channels of the Eastern United States. Previous studies have identified the importance of these features showing that they are areas of high biologic complexity. Shoals can affect sediment transport dynamics and are capable of acting as nitrate sinks due to the vegetation that inhabits them. However, information regarding their distribution and formation is lacking. In this study, shoaled and non-shoaled reaches throughout the length of the Cahaba River, located in central Alabama, were analyzed to determine the factors responsible for their presence. Flow orientation (expressed and measured as perpendicularity), rock type, and confinement ratio were selected as possible contributing variables based on the existing literature on shoals and bedrock bedforms. Each variable was initially tested using exploratory statistical methods (Mann-Whitney U tests) to determine if there were differences between shoaled and non-shoaled sites. The results of the Mann Whitney U test showed that the all three variables were significantly different when located in a shoaled or non-shoaled reach. This information was then used to conduct a binary logistic regression analysis. Results of the logistic regression were in alignment with the Mann-Whitney U tests and suggested that perpendicularity as well as rock integrity, are significant predictors of shoal occurrence, with rock integrity being the most significant (p<.001) and strongest predictor. Confinement ratio, a variable often believed to be a major contributor to stream geometry, was found to be insignificant both as a consolidated variable (p=.727). A predictive equation was formulated based on the results of the logistic regression using the regression coefficients for flow perpendicularity and rock integrity and the y-intercept for the regression model. The resulting predictive equation was used to test several shoaled and non-shoaled sites that were not included in the original dataset. The equation predicted 8 of these sites with greater than 98% probability, one site at 76% and another at 28%. All but one non-shoaled site produced predictive probabilities under 35%, correctly predicting low possibility for shoal occurrence in these areas.
