The flow velocity in a river is three-dimensional (3D), turbulent, and varies in time and space. Capturing this variability in field measurements to support studies of river processes has proven particularly challenging. While originally developed to measure discharge, boat-mounted acoustic Doppler current profilers (ADCP) are increasingly used in field studies to quantify flow features including mean velocity, boundary shear stress, and sediment motion. Two survey procedures are typically employed with an ADCP. Moving-vessel (MV) measurements provide spatially-rich velocity data while temporally-rich data are obtained with fixed-vessel (FV) procedures. Given the relative ease of MV measurements, recent work has focused on developing MV procedures that produce comparable results to FV measurements. At the present, results of this work are inconclusive. Additionally, there is a lack of reported data and procedures for FV measurements. This work seeks to develop techniques to present 3D velocity data obtained in natural rivers in a unified framework. This framework is based on a stream-fitted coordinate system defined by the flow direction at a cross section and allows for 3D velocity to be decomposed into streamwise, spanwise, and vertical components. Procedures are developed to assure that the velocity profiles measured at fixed locations are (1) not negatively impacted by the inevitable motion of the ADCP, (2) statistically stationary, and (3) of sufficient record length to determine the mean velocity. The coordinate system allows time-averaged velocity from FV procedures to be compared with spatially-averaged velocity from MV vessels. Significant differences are found between the two survey procedures, particularly for secondary velocity components. Ultimately, integrating results of the two survey procedures leads to an improved representation of the mean flow field. The techniques are applied to data obtained on a study reach on the lower Roanoke River, located in eastern North Carolina. / Ph. D.