|Abstract:||Confluences, locations where two rivers flow into one another, are characterized by the mixing of water, dissolved nutrients, chemicals, pollutants, and sediment. The process of mixing at confluences is inherently related to patterns of confluent flows, yet the complex nature of confluence hydrodynamics has limited efforts to generalize how mixing occurs at these locations in river systems. Although traditional in-stream measurements of three-dimensional velocities using hydroacoustic instruments provide valuable information on confluence hydrodynamics, such measurements are relatively limited in spatial resolution. State-of-the-art high-resolution velocity-measurement techniques based on analysis of low-level imagery obtained by cameras positioned above the water surface have the potential to document in detail complex patterns of at confluences. Thus far, however, no studies have applied these emerging image-based methods to characterize flow at confluences, or have examined how imagery-based velocity measurement techniques might be enhanced by the acquisition of imagery using small unmanned aerial systems (sUAS). The objectives of this dissertation are to: 1) investigate the accuracy of LSPIV for characterizing two-dimensional patterns of surface flow at stream confluences and 2) to use in-stream measurements of three-dimensional velocity, temperature, and turbidity to characterize mixing patterns and rates at confluences and to examine how mixing at confluences varies with changes in controlling factors. The dissertation research is organized into four distinct investigations focusing on flow and mixing at river confluences.
The first study investigates the potential for using large-scale particle image velocimetry (LSPIV), an imagery-based velocity measurement technique typically employed in laboratory settings or in simple, uniform flows in the field, for improving understanding of complex two-dimensional flow at a river confluence. This study develops a stationary, river channel-spanning camera mount onto which a small action camera is anchored. The camera records the movement of inexpensive, recycled landscape mulch on the surface of the water, and compares the resultant LSPIV-derived velocity with near-surface acoustic velocity measurements. The accuracy of the LSPIV compares favorably to the acoustic measurements, yet LSPIV can be used to obtain velocity over a large spatial extent. The chapter then focuses on applying the strengths inherent in the high spatial and temporal resolution afforded by LSPIV to characterize complex two-dimensional flow structures at a river confluence. The results of this chapter confirm that LSPIV can be a cost efficient and effective supplement to traditional studies in regions of complex flow, and on its own can be used to better understand aspects of flow at confluences such as wake-like flow and shear-layer dominated flow along the confluence shear layer.
The second study extends the analysis of LSPIV methodology at confluences by exploring the potential benefits or disadvantages of using sUAS to obtain LSPIV imagery. This chapter specifically focuses on the potential capability of sUAS to measure complex mean flow and quasi-instantaneous snapshots of flow structure in river confluences with strong two-dimensional velocity gradients. In this study, LSPIV results from a mobile tripod, a channel-spanning stationary camera mount, and sUAS are compared. The accuracy of each method is compared to near-surface acoustic velocity measurements, and all methods are found to be accurate in comparison to the in-stream measurements. Although the sUAS is not fixed and moves slightly in three dimensions when hovering over the water surface, movement rapidly converges to a net of zero within tens of seconds even in relatively windy conditions. The results of this study indicate that mean velocities obtained with sUAS-derived LSPIV are just as accurate as with fixed methods, but can be a substantial improvement because of increased locational flexibility of the field of view. In addition, this study explores the effect of total sUAS image distortion, and confirms that image distortion is not meaningful when flying within about 20 m of the water surface and thus extensive image rectification is not required. Finally, this study investigates the conditions under which quasi-instantaneous snapshots of flow structures can be recorded with sUAS-derived LSPIV. Results reveal that, while more challenging than recording mean velocities, snapshots of flow structure can be obtained using sUAS when the flow structure is many times larger than the LSPIV interrogation area, and if the velocity signal that defines the flow structure is substantially larger than any apparent velocity signal caused by LSPIV error (such as spare seeding or camera movement).
The dissertation’s third chapter builds upon the first two be applying the developed LSPIV and sUAS methodology to produce a study of highly-detailed hydrodynamic mapping at two river confluences. Flow at two river confluences under high and low momentum ratios are investigated in unprecedented spatial detail. Results of this investigation reveal similarities between flow at each confluence and the standard conceptual model of flow at confluences, yet also finds important differences between the field cases and expectations derived from the standard conceptual model. This investigation exposes details about how the presence and location of the expected hydrodynamic zones changes with changing momentum ratio and confluence morphology and confirms that both variables have strong controls on confluence hydrodynamics. Results of this research unequivocally confirm the presence of wake-like flow at confluences at low momentum ratio and shear-layer flow at high momentum ratio, but also show that flow within the stagnation zone and shear layer at one confluence was not strongly controlled by momentum ratio. Detailed hydrodynamic maps for each confluence under each flow condition are produced, which can be directly compared to conceptual, computational, or laboratory models of confluent flow. The dense array of velocity measurements afforded by sUAS-derived LSPIV are combined with in-stream measurements to demonstrate the advancement of knowledge of flow at confluences driven by the application of these new techniques.
The fourth and final investigation assess mixing dynamics at three confluences with distinct external (i.e. geometrical and morphological) characteristics. This study uses detailed measurements of three-dimensional velocity and spatially coincident measurements of temperature and turbidity to determine how patterns of mixing respond to velocity patterns. This research builds upon the growing body of work on mixing at confluences by obtaining these detailed measurements over a suite of flow conditions at each confluence. This study also assesses mixing rates with a formula based on tracer variance that can be compared among different sites and flow conditions. The results of this study indicate that channel-scale secondary flow drives momentum transfer along the mixing interface between flows, and therefore is the dominant control on mixing. The dominant form of secondary flow is coherent helical cells driven by flow streamline curvature and the associated pressure gradient, although this study also shows that in cases without extensive helical flow cells substantial mixing (up to 40%) can occur at one of the confluences. This investigation also confirms that mixing appears to be positively correlated with momentum ratio, negatively correlated with flow scale (e.g. depth), and might be affected by density differences in some cases. This study provides the groundwork for future detailed studies of mixing at these sites supported by thorough analytical and computational investigation. In conclusion, the results of the research presented in this dissertation improve understanding of flow and mixing at river confluences under a suite of external (geometrical and morphological) and internal (hydrodynamic) controls, while simultaneously advancing LSPIV and sUAS methodologies. This dissertation also provides a foundation for ongoing and future computational modeling of flow at confluences, and offers the potential for future comparisons among field, laboratory, and computational work using mixed methodology approaches that yield high-resolution data in the field.