This Is AuburnElectronic Theses and Dissertations

Mapping Stream Channel Head Locations in the State of Alabama




Burak, Avcioglu

Type of Degree

Master's Thesis




This study focuses on assessing the utility of the slope-area method for predicting channel head locations in forested areas in Alabama under different physiographic regions, improving this method by evaluating the other landscape parameters related to channel initiation, and lastly comparing the results to the widely used National Hydrographic Dataset. To achieve these objectives, the models developed and evaluated in this study, included three region-specific multi-variable models, a comprehensive state multi-variable model, and the slope-area model. A total of 163 stream channel head locations across three physiographic regions of Alabama, including Southwestern Appalachians (n=51), Piedmont (n=61), and Coastal Plains (n=51), were mapped with extensive field work using a GPS device. The field data were collected in national forests within each physiographic region, namely William B. Bankhead National Forest, Talladega National Forest, and Conecuh National Forest of Alabama, to capture the least disturbed (i.e. reference) conditions. The local slope and the drainage area had an inverse and strong correlation in the Piedmont/Ridge and Valley region (r2=0.71) and the Southwestern Appalachian region (r2=0.61). Among the three physiographic regions, the weakest correlation was observed in the Coastal Plain region (r2=0.45). Overall, calculated reliability and sensitivity indices indicated that model accuracy and reliance were weak to moderate; however, the slope-area method gave a much better estimation on the channel head locations compared to NHD. The region-specific multi-variable models provided the best predictions of stream channel head locations mapped in the field at each region. A comprehensive state-model was less accurate at locating the actual stream channel head locations compared to the region-specific models. However, the comprehensive model was comparable to the slope-area model, showing that it could be applied in regions where no field mapped stream channel head location data is available.