H52A-01
Field Monitoring of Bedrock Channel Erosion and Morphology
We present field measurements of erosional morphology and fluvial incision into Navajo Sandstone bedrock along a short human-perturbed channel reach in the Henry Mountains, Utah. Bedrock rivers flow in self-formed channels and form diverse erosional morphologies. The parameters that collectively define channel morphology (e.g. width, slope, bed roughness, bedrock exposure, sediment size distribution) all dynamically adjust in poorly understood ways to imposed fluid and sediment fluxes. The picture of erosion that emerges from our field monitoring is consistent with laboratory flume experiments conducted under sediment-starved transport conditions. We find that erosion is a sensitive function of the evolving bed topography because of feedbacks between the turbulent flow field, sediment transport, and bottom roughness. To facilitate highway construction over Swett Creek in the early 1970s, the Utah Department of Transportation created a natural experiment by filling in part of the original canyon and routing flow through a culvert and blasted bedrock slot. The vertical-walled, blasted channel has an upper reach (the ''flume'', length ~80 m, slope ~0.022, width ~5 m) and a shorter downstream reach that steeply slopes into the original channel (the ''flume mouth'', length ~17 m, slope ~0.18). The field-scale flume thus constructed provides an excellent opportunity to observe the morphology of a bedrock channel that has eroded from a well constrained initial geometry over a known amount of time. Ephemeral channel flow only occurs during spring snowmelt and flash floods. We monitor water depth, and we have a limited number of direct bedload transport measurements. In addition, we have installed bolts in the bedrock and are currently monitoring active erosion and morphological changes in the channel reach. Bedrock incision into the Navajo Sandstone occurs mostly by abrasion, with dramatic sculpted forms apparently carved by the finer bedload that becomes locally suspended in intense vortices. As has been observed previously in other bedrock channels, erosional morphology varies with slope. The lower slope ''flume'' has longitudinal grooves, significant sediment cover and modest rates of erosion. Erosion in the higher slope ''flume mouth'' is not spatially uniform but is focused to form a narrow inner channel. We measured approximately 45 cm of vertical incision into Navajo Sandstone in the inner channel from the 2005 season of exceptional snowmelt flow. When transported sediment becomes focused in the center of the channel without depositing, an inner channel forms simply because abrasion only happens where sediment energetically impacts the bed. Positive feedback occurs between the sediment preferentially transported in topographic lows and the further abrasion of topographic lows. The inner channel in particular has potholes and sculpting; erosional bedforms such as these add roughness to the channel. Complementary laboratory experiments that also resulted in inner channel formation suggest that sidewall friction and tortuous flow conditions in the inner channel reduce the shear stress available to transport sediment, acting as the negative feedback that ultimately limits inner channel incision. Inner channels may often form where the sediment load imposed on the reach is well below the transport capacity of the flow, such as where the local slope is higher.
H52A-02
Numerical Simulation of Meandering Evolution
Abstract: Evolution of meandering channels, a complex morpho-dynamic process, has been the focus of research challenging geomorphologists and river engineers for decades. The evolution of channel meandering is the result of the complex interactions between flow, sediment transport, and bank erosion. A numerical model including a depth-averaged two-dimensional hydrodynamic flow model, a sediment transport model, and a bank erosion model was developed to simulate the evolution of meandering channel from a low to high sinuosity. The sediment transport model calculates both bed load and suspended load assuming equilibrium sediment transport. Bank erosion simulation consists of two interactive processes: basal erosion and bank failure. The mass conservation equation where basal erosion and bank failure are considered source terms, was solved to obtain the rate of bank erosion. Then, this model was applied to simulate the evolution of free meandering channels in laboratory experimental settings. The model successfully replicated the evolution processes of meandering channel from downstream translation, lateral extension, and then follows by upstream and downstream rotation when sinuosity approaches 3.7. Plots of meandering planforms illustrated the evolution of meandering planform is a resultant of 1) redistribution of primary flow momentum due to curved channel geometry, 2) secondary flow generated by channel curvature and growing point bars, and 3) the material input from bank failure.
H52A-03
Linking Meander Initation to Instability in the Cross-Sectional Sediment Transport Field
Recent theoretical and experimental work on reach-scale channel dynamics suggests that stream channel response to changes in the discharge or sediment supply regime produce adjustments that are consistent with the optimality criteria (or extremal hypotheses) proposed by rational regime modellers. However, a general form of the optimality criteria based on the maximization of system-scale flow resistance permits multiple channel responses to the same imposed change, and is thus inadequate for predicting channel response, absent additional information. The way forward is to identify the geomorphic processes that produce various possible channel responses, and to develop an understanding of the conditions under which these processes may or may not be effective. Using a simple analytic model relating local sediment transport capacity to variance in the transverse shear stress distribution I develop a physically based conceptual model of the initiation of meandering in straight, bedload dominated streams as a result of a feedback mechanism. This corresponds to a channel response that is dominated by changes the energy slope, which appears to be dominant in channels having relatively erodible banks and transporting their bed material load at conditions near the threshold for entrainment. The feedback maximizes the cross sectional shear stress variance and, in order to achieve stability, minimizes the energy slope at repeated locations along the channel. These locations develop into pools in a fully developed meandering channel; they represent attractor states wherein sediment continuity is satisfied using the least possible energy expenditure per unit length of channel. Between two successive pools, a stream occupies a metastable, higher energy state (corresponding to a riffle) that requires greater energy expenditure per unit length of channel to transport the same volume of sediment. The model links processes at the scale of a channel width to adjustments of the channel sinuosity and slope at the scale of a channel reach. The analysis supports the contention that the reach-scale extremal hypotheses employed by rational regime models are mathematical formalisms that permit a one-dimensional theory to describe the three dimensional dynamics producing stream morphology.
H52A-04
Hydrologic and Geomorphic Controls on the Downstream Transport of a Wave of Fine Sediment, Clark Fork River, Montana
The floodplain of the Clark Fork River between Warm Springs and Garrison, Montana, is contaminated with fine mine tailings deposited during several large floods approximately 100 years ago. While perhaps unlikely given the vagaries of local politics, if left alone and given enough time, the channel of the Clark Fork River would naturally rework the contaminated sediment and eventually remove most of it from the valley. Based on the thickness of the contaminated sediment deposits, the tailings raised the level of the floodplain on the order of 10 to 30 cm within the meander belt of the river, changing the bankfull capacity of the channel. A large tailings basin constructed around 1918 and enlarged periodically until about 1950 controls flow from approximately 70 percent of the contributory watershed at the upstream end of the study reach. Both the tailings themselves and the tailings basin have affected the hydraulic and hydrologic characteristics of the system. The time scale over which the natural removal of the contaminated sediment would occur must be controlled both by the rate at which the channel migrates laterally and, to the extent that the channel currently rebuilds point bars to a new, potentially lower bankfull elevation, by changes in the discharge and sediment transport characteristics of the system. The model presented here assumes a decoupling between the volume of sediment eroded from the floodplain due to bank migration and the volume deposited in point bars or on the floodplain during flood events, allowing channel geometry and bankfull capacity to evolve over time in response to historic changes in a) floodplain elevation and b) flood frequency. The model can be used both to identify the time scale for natural attenuation of the contaminated sediment wave as it travels through the floodplain and to assess in a general sense the long term evolution of a typical channel cross section. It can also predict the general pattern of cross section adjustment that might be expected if the layer of contaminated tailings were artificially removed as part of site remediation.
H52A-05
Continuum Statistics of the Bed Topography in a Sandy River
Temporal and spatial variabilities in the bed geometry of sandy rivers contain information about processes of sediment transport that has not been fully appreciated. This is primarily due to a disparity between the dynamic nature of the sediment-fluid interface and the relatively static methods of surveying bed elevation, e.g. single profiles or point measurements. High resolution topographic data is paramount to understanding the dynamic behavior of sandy beds. We present and analyze a data set collected on a 2cm x 2cm grid at 1 minute intervals and with a vertical precision of ~1mm. This was accomplished by using Lambert-Beer's Law for attenuation of light to transform low-altitude aerial photographs into digital elevation models. Forty successive models were generated for a 20 m by 30 m section of channel bottom of the N. Loup River, Nebraska. To calculate the average, whole bed translation rate, or celerity, cross-correlations between a reference bed topography and its proceeding configurations were determined. Time differences between models were related to the shift lengths that produced correlation maxima for each model pair. The result is a celerity of ~3.8cm/s with a correlation coefficient of 0.992. Bed topography also deforms while it translates, and this can be seen as a secular decrease of correlation maxima. The form of this decrease in correlation is exponential, and from it an interface half-life is defined. In this case, the bed had become extensively reorganized within ~40 minutes, the time necessary to translate the bed one wavelength of the dominant roughness element. Although the bed is continuously deforming, its roughness is statistically stationary. Essentially, a mean roughness is maintained as the bed creates new realizations of itself. The dynamic nature of the whole bed and similarly transient behavior of individual elements suggests the utility of a holistic approach to studying the feedback between bed topography, fluid flow, and sediment transport. Furthermore, it raises questions about the usefulness of detailed analysis of flow and transport over individual forms.
H52A-06
Comparison of Sediment-Transport and Bar-Response Results From the 1996 and 2004 Controlled-Flood Experiments on the Colorado River in Grand Canyon
The sediment-transport paradigm for the regulated Colorado River in Grand Canyon (GC) prior to the 7-day 45,000 ft3/s 1996 controlled-flood experiment was that, under normal releases from Glen Canyon Dam, tributary-supplied sand would accumulate in the channel over multiple years and could then be transferred from the channel bed to eddies during controlled floods, increasing both the area and volume of eddy sandbars. Work conducted during and after the 1996 flood indicates that this paradigm was based on assumptions that were either false or only partially true (Rubin et al., {\it EOS}, 2002). First, sand did not accumulate in the channel over years. Second, sand deposited at higher elevations in eddy sandbars during the 1996 flood was derived mostly from the lower parts of these bars (not from the channel bed) resulting in bars that were smaller in area and volume (although they did contain more sand at higher elevations). Tributary inputs of sand were low in the year preceding the 1996 flood and dam releases were moderate to high. Thus, the 1996 flood was conducting during a period when the Colorado River in GC was relatively sand-depleted. The design of the 2004 controlled-flood experiment was to: (1) keep dam releases relatively low (<10,000 ft3/s) in Sept.-Nov. 2004 to allow accumulation and retention of new tributary sand in the channel, and (2) after retention of >800,000 metric tons of new sand in Marble Canyon (the first 99 km of GC), release a 60-hr flow of 41,000 ft3/s from the dam to transfer this sand from the channel bed into the eddies. More sand, silt, and clay were present in Marble Canyon (MC) during the 2004 experiment than during the 1996 experiment. At the lower end of MC, suspended-silt & clay concentrations were 3x higher and suspended-sand concentrations were ~30% higher than during the 1996 flood. Furthermore, during the 2004 flood, sand concentrations were higher in the upstream half of MC than in the downstream half of MC. In contrast, during the 1996 flood, sand concentrations likely increased downstream throughout MC. The spatial pattern in concentration during the 2004 flood resulted from effective retention of tributary-supplied sand in the upstream half of MC during the lower releases preceding the flood. The response of the eddy sandbars during the 2004 flood correlates with this observed spatial pattern in sand concentration. About 2/3 of the bars surveyed in the upstream half of MC were larger than they were after the 1996 flood, whereas only 1/3 of the bars surveyed in the downstream half of MC were larger in both area and volume than they were after the 1996 flood. In contrast, less sand was present in GC downstream from MC during the 2004 flood than during the 1996 flood. At 2 gaging stations located 43- and 260-km downstream from MC, sand concentrations were ~30% lower than during the 1996 flood. As in MC, the response of eddy sandbars in this downstream reach also reflects the difference in sand concentration between the two experiments, with fewer surveyed bars being larger after the 2004 flood than after the 1996 flood. Therefore, it appears that the 800,000 metric tons of new sand in retention prior to the 2004 flood was sufficient to result in substantial increases in sandbar area and volume in only the first 50 km (i.e., the upstream half of MC) of the 400-km long reach of the Colorado River in GC.
H52A-07
Influence of Glen Canyon Dam on Fine-Sediment Storage in the Colorado River in Marble Canyon, Arizona
Glen Canyon Dam has caused a fundamental change in the distribution of fine-sediment storage in the 99-km reach of the Colorado River in Marble Canyon, Grand Canyon National Park, Arizona. The two major storage sites for fine sediment (i.e., sand and finer material) in this canyon river are lateral recirculation eddies and the main-channel bed. We use a combination of methods, including direct measurement of sediment storage change, measurements of sediment flux, and comparison of the grain size of sediment found in different storage sites relative to the supply and that in transport, in order to evaluate the change in both volume and location of sediment storage. The analysis shows that the bed of the main channel was an important storage environment for fine sediment in the pre-dam era. In years of large seasonal accumulation, ~50% of the fine sediment supplied to the reach from upstream sources was stored on the main-channel bed. In contrast, sediment budgets constructed for two short-duration, experimental releases from Glen Canyon Dam indicate that ~90% of the sediment discharge from the reach during each release was derived from eddy storage, rather than from sandy deposits on the main-channel bed. These results indicate that the majority of the fine sediment in Marble Canyon is now stored in eddies, even though they occupy a small percentage (~17%) of the total river area. Because of a 95% reduction in the supply of fine sediment to Marble Canyon, future high releases not timed with substantial tributary inputs will potentially erode sediment from long-term eddy storage, resulting in continued degradation in Marble Canyon.
H52A-08
A New Method for Identification of Tributary Sediment Sources using Hydroacoustics
Identification of tributary sediment sources is important in geomorphologic studies because different tributaries within the same basin may deliver sediment with widely varying properties. In particular, tributaries may deliver sediment with very different grain-size distributions due to differences in lithology between tributary drainages. These differences have important implications for understanding the link between tributary sediment supply and main channel morphology. Hydroacoustic instrumentation has become popular in recent years for the study of sediment-transport processes in both marine and fluvial environments. Because suspended material scatters and attenuates acoustic energy, transducers designed to record the backscattered energy can be used to infer the suspended sediment concentration. The transducer records the energy received from direct backscattering from the particles, which is reduced by transmission losses that occur as the wave travels through the medium. These transmission losses are composed of: 1) geometrical spreading, 2) attenuation of energy by the fluid, and 3) attenuation of energy by the suspended particles. The backscatter and particle attenuation are functions primarily of the wave frequency, concentration of particles, and size of particles. Thus, for a known frequency and particle size, it is possible to invert the acoustic signal to determine particle concentration. Here, we present a new method that takes advantage of the relationship between particle attenuation and particle size in order to identify tributary sediment sources. The study site is on the Colorado River below Glen Canyon Dam, where a sideward-looking acoustic instrument has been bank-deployed since August 2002. Suspended-sediment samples were collected and a relationship was developed between suspended fine material (silt and clay) and particle attenuation (R2=0.98). However, substantial deviations occur from this relationship during flooding events from certain tributaries. These deviations can be directly traced to differences in the grain-size distribution of samples from these tributaries versus samples from the main supplier of fine sediment (the Paria River). They are relatively easy to recognize because the relationship between attenuation and grain size is highly non-linear (i.e. small changes in grain size lead to large changes in attenuation). Recognition of these deviations is critical in the context of continuous monitoring for a sediment budget, because the acoustic calibration predicts highly erroneous concentrations during these tributary flooding events which must be corrected using manual samples or other in-situ instrumentation (e.g. optical transmissometers/grain-size analyzers).