UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Sediment transport and morphological response of a semi-alluvial channel : insights from a Froude scaled… Luzi, David Steven 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2014_september_luzi_david.pdf [ 69.57MB ]
JSON: 24-1.0165910.json
JSON-LD: 24-1.0165910-ld.json
RDF/XML (Pretty): 24-1.0165910-rdf.xml
RDF/JSON: 24-1.0165910-rdf.json
Turtle: 24-1.0165910-turtle.txt
N-Triples: 24-1.0165910-rdf-ntriples.txt
Original Record: 24-1.0165910-source.json
Full Text

Full Text

Sediment Transport andMorphological Response of aSemi-Alluvial ChannelInsights from a Froude Scaled Laboratory ModelbyDavid Steven LuziMSc, The University of British Columbia, 2000A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty Graduate and Postdoctoral Studies(Geography)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2014c© David Steven Luzi 2014AbstractLaboratory physical models have been used in geomorphology for over a century. Physicalmodels are a useful tool for understanding and observing phenomenon that are difficultor impossible to observe in the field. The objective of this study was to understand thelong-term evolution of a semi-alluvial channel in terms of its morphology and sedimenttransport under various scenarios of constant sediment supply and discharge. Specifically,the research aimed to investigate (1) effects of various flows and sediment feed rates onsurface textures and sediment output, (2) relationship between channel storage, and the(3) morphology and sediment transport sediment transport processes and pathways. Theseobjectives were addressed by building a Froude scaled physical model based on the irreg-ular meandering planform of Fishtrap Creek, and conducting ten experiments of varyingtemporal lengths, discharge and feed rates. The model successfully replicated pool-riffleand plane-bed morphologies.The effects on the characteristics of the bed surface and transported sediment underdiffering regimes of discharge and sediment feed were investigated. Scaled formative flowsranging from 2-yr to over 150-yr return period events were employed. The results indicatedthat even with discharges exceeding the 10-yr event, full mobility was not observed. Thisslight but persistent size-selectivity produced long-term aggradation and surface coarsen-ing.The effects of varying sediment supply and discharge in channel storage and morphologywere explored. Results showed that sediment transport rates varied both spatially andtemporally. The variability was more dependent upon changes in channel morphologythan adjustments in the grain size distribution of the surface. Cycles of aggradation-degradation were observed to occur without changes in sediment supply of discharge andthat they tended to occur in periods when sediment output approximately equaled sedimentfeed rates.Lastly, one experiment was selected to describe sediment transport processes and path-ways. Primary information regarding sediment pathways was obtained through the obser-vation of bedload sheet movement and migration during the experiments, as well as throughsubsequent review of videos made during the experiments. The behaviour of bedload sheetsalso shed new information on how sediment sorting through a pool varies.iiPrefaceThis dissertation presents research conducted by David Luzi in collaboration with hissupervisor, Dr. Brett Eaton. David Luzi was the primary investigator and responsiblefor the design of the research and the collection, analysis and interpretation of the data.Dr. Brett Eaton and Dr. Marwan Hassan provided analytical support and timely reviewsthroughout the research and the preparation of this dissertation.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix1 Preamble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2.1 Process Response Models . . . . . . . . . . . . . . . . . . . . . . . . 21.2.2 Sediment Supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.2.3 Discharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121.4 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Experimental Design and Methods . . . . . . . . . . . . . . . . . . . . . . . 142.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.2 Fixed bank - Mobile Bed Experiments . . . . . . . . . . . . . . . . . . . . . 142.2.1 Prototype Stream and Scaling Considerations . . . . . . . . . . . . 142.2.2 Flume Instrumentation and Measurements . . . . . . . . . . . . . . 212.2.3 Experimental Protocols . . . . . . . . . . . . . . . . . . . . . . . . . 252.3 Mobile bank - Mobile Bed Experiments . . . . . . . . . . . . . . . . . . . . 273 Sediment Mobility and Channel Stability: Implications from Froude-scaled Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31ivTable of Contents3.2.1 Sediment Transport Rates . . . . . . . . . . . . . . . . . . . . . . . 313.2.2 Sediment Mobility - Fixed Bank Experiments . . . . . . . . . . . . 373.2.3 Sediment Mobility - Mobile Bank Experiments . . . . . . . . . . . . 433.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504 Spatial and Temporal Patterns in Sediment Transport and Storage . 514.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 514.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534.2.1 Sediment Transport Rates . . . . . . . . . . . . . . . . . . . . . . . 554.2.2 Sediment Transport-Storage Relations . . . . . . . . . . . . . . . . . 604.2.3 Patterns of Channel Adjustments . . . . . . . . . . . . . . . . . . . 634.2.4 Sediment Texture of the Transported Material and Bed Surface . . 704.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 724.3.1 Sediment Transport Patterns . . . . . . . . . . . . . . . . . . . . . . 734.3.2 Transport-Storage Relations . . . . . . . . . . . . . . . . . . . . . . 754.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795 Sediment Transport at the Pool-Riffle Scale: Observations from a Phys-ical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 855.3 Experimental Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . 855.3.1 General Observations . . . . . . . . . . . . . . . . . . . . . . . . . . 855.3.2 Bedload Sheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925.3.3 Observations of sediment transport . . . . . . . . . . . . . . . . . . 965.4 Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 996 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1016.1 Questions, Observations and Future Research . . . . . . . . . . . . . . . . . 104References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106vList of Tables2.1 Experimental design parameters in prototype and model . . . . . . . . . . . 182.2 Experimental conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263.1 Summary statistics related to sediment transport rates . . . . . . . . . . . . 324.1 Summary data of the experimental conditions and results. . . . . . . . . . . 544.2 Transport efficiency for experiments with similar stream discharge and sed-iment feed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554.3 Transport efficiency for experiments with similar stream discharge . . . . . 594.4 Transport efficiency for experiments with similar sediment feed . . . . . . . 594.5 Transport efficiency for experiments with similar stream discharge and sed-iment feed rates, but differing grain size distribution of the feed. . . . . . . 594.6 The D50t and D90t of transported and D50s and D90s surface sediment ispresented. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725.1 Experimental conditions and results . . . . . . . . . . . . . . . . . . . . . . 85viList of Figures2.1 Apparatus used for the fixed bank experiments conducted at UBC , lookingupstream. White arrow indicates flow direction. . . . . . . . . . . . . . . . . 152.2 Prototype and model bed material grain size distributions. . . . . . . . . . . 202.3 Photo of a) sediment exiting the feeder, and b) sediment feeder with sediment. 222.4 Photos of a) converted tipping bucket rain gauge used to introduce saltmixture into the stream, and b) time delayed image of laser cart during ameasurement of the channel. . . . . . . . . . . . . . . . . . . . . . . . . . . 232.5 Photo looking upstream at the mobile bank model. . . . . . . . . . . . . . . 283.1 Experimental results of sediment output plotted against time . . . . . . . . 343.2 Cumulative sediment flux plotted against time . . . . . . . . . . . . . . . . 353.3 Fractional transport ratio diagrams of mobile size classes from a) Exp. 1,b) Exp. 2, and c) Exp. 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.4 Fractional transport ratio diagrams of mobile size classes from a) Exp. 4and b) Exp. 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.5 Fractional transport ratio diagrams of mobile size classes from a) Exp. 6and b) Exp. 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.6 Fractional transport ratio diagrams of mobile size classes from a) Exp. 8and b) Exp. 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423.7 Modified fractional transport ratio diagrams of mobile size classes from fixedbank Exp. 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443.8 Fractional transport ratio diagrams from Exp. 10 . . . . . . . . . . . . . . . 453.9 Fractional transport ratio diagrams of mobile size classes from the mobilebank experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464.1 Variations in sediment transport rates with respect to time. . . . . . . . . . 564.2 Histograms of sediment transport rates (Qb) normalized by the sedimentfeed rate (Qf ). For each theme histograms are shown for data truncated toexclude the initial start-up spikes in transport rates. Q represents streamdischarge. Solid vertical lines represents Qb/Qf = 1 for reference. . . . . . . 574.3 Transport-storage relations between transport rate and sediment storage. . 614.4 Hillshaded DEMs for selected time periods for all experiments. . . . . . . . 65viiList of Figures4.5 Changes in channel storage illustrated through DEM difference maps forExp. 1 to Exp. 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674.6 Changes in channel storage illustrated through DEM difference maps forExp. 6 to Exp. 9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694.7 Particle sizes (D50 and D90) of bed surface and bed material measured atthe outlet for all experiments. . . . . . . . . . . . . . . . . . . . . . . . . . . 714.8 Conceptual model of the probability of hysteresis occurring in the transport-storage-feed relation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 775.1 This figure contains a) flow structure, b) shifting loci of maximum boundaryshear stress and sediment transport pathways, and c) forces acting on abedload particle in meander bends. From Powell (1998). . . . . . . . . . . 845.2 Photo of experiment looking upstream during low flow conditions. Emergentbars and pools are identifiable in the image. . . . . . . . . . . . . . . . . . . 865.3 Plot showing individual velocity measurements for Exp. 6. . . . . . . . . . . 885.4 Large scale photos taken during periods of no flow and overhead photos takenfrom video during the experiment are presented at various time intervals: a)0 minutes, b) 900 minutes, c) 1800 minutes, and d) 2400 minutes. . . . . . . 895.5 a) Cumulative sediment output (g) at 60 minute intervals is shown on theprimary axis over time. b) The average width of active sediment transportof all ten cross sections during the experiment and from four individual crosssections. Note that no observations were made at 240 minutes. . . . . . . . 915.6 Selected cross sections from laser scans of the bed for: a) XS 195; b) XS245; c) XS 295; d) XS 345. Data is presented at 300 minute intervals, whenscans were completed while experiment was not running. . . . . . . . . . . . 935.7 Image showing two bedload sheets emerging from the same pool. As thesheet entered the pool it diverged into two separate sheets. The coarseleading fronts of the two sheets are indicated in the image. The arrowshighlight the direction of movement of the bedload sheet. . . . . . . . . . . 955.8 Flow processes controlling morphology and sediment sorting in a river me-ander. Flow direction is from the lower to upper end of the figure. FromDietrich and Smith (1984). . . . . . . . . . . . . . . . . . . . . . . . . . . . 98viiiAcknowledgements”By the time it came to the edge of the Forest, the stream had grown up, sothat it was almost a river, and, being grown-up, it did not run and jump andsparkle along as it used to do when it was younger, but moved more slowly. Forit knew now where it was going, and it said to itself, ”There is no hurry. Weshall get there some day.” But all the little streams higher up in the Forest wentthis way and that, quickly, eagerly, having so much to find out before it was toolate.” Milne (1928)Numerous people have been involved in this project. My supervisor Brett Eaton en-couraged me to return from the field and to explore the laboratory as another means ofunderstanding fluvial processes. Marwan, although late to the official party, has beenpart of this endeavour since the beginning and has always been open to conversations ongeomorphology.I appreciate Drs. Dan Moore, Peter Ashmore and Sarah Gergel for reviewing the thesis,their comments resulted in a much improved final product.I am thankful for the friendship from my fellow lab mates at the beginning of thisjourney, Ashley Perkins, Jeff Phillips, and Christie Andrews. I would also like to thankAndre Zimmermann for his help with coding Labview, troubleshooting problems and forbeing open and willing to help at many points along the way. Thanks also to Mike, Jen,Jaclyn, Michael, Dylan and Sarah for all of your help in conducting the experiments andsieving sediment.This work could not have happened with out the support of my family. To my wifeMarla, I hope to one day to repay the patience and understanding you have given meduring this journey. To Ariel and Zoe¨, my amazing girls, whose arrival during this projectshowed me that being a father is the most important thing in life.ixChapter 1PreambleWhile sitting on the floodplain of Fishtrap Creek and observing how the channel wasresponding to a recent fire which devastated the riparian zone, I began to think of whathappens to the channel during the 100 years or so between disturbances. During myearlier work at Carnation Creek, I found that, on average, channel disturbances, inferredby ages of log jams, occurred almost every 20 years and that the locations of same agedlog jams were spatially distributed throughout the channel’s length. Therefore, as onereach responded to some channel disturbance, another reach had a decade or two to adjustfrom the previous disturbance, and at another it may have been a century or two. AtCarnation Creek, channel recovery appeared to occur relatively rapidly, and most of thetime in any given reach of the river, the channel remained within its banks, which were inturn influenced by the forest that held them together and the size and quantity of sedimentthat composed the channel bed. This disturbance regime, seemed to be similar to what Iwas currently observing at Fishtrap Creek. From the age of the trees in the floodplain andthe evidence of historical channel activity, Fishtrap Creek appeared to spend most of thetime exhibiting long periods of relative lateral stability.Once I had begun to realize that it was the longer time frame, the one which occurredin between disturbances, that held the more interesting questions and answers to riverbehaviour, I knew I must return from the field and head into the laboratory. Space fortime studies are a common and useful tool in field based geomorphology, but sometimes canintroduce more questions than answers. In the laboratory environment, one can controlthe forcing functions, and with proper scaling can observe morphological responses overmuch longer time scales than is achievable in the field.1.1 IntroductionThe main objective of this project is to provide a better understanding of the pathways, orprocess-form interactions responsible for channel response. Although advancements havebeen made in the understanding of how some channel characteristics react to changes in thegoverning conditions the broader understanding of channel form response is still limited.This limitation is partly due to limits of time and budgets for long-term field programs,but is also due to the use of overly simplistic channels in laboratory environments, wherelonger term channel morphodynamics can be investigated. Understanding the responsemechanisms for mountain gravel-bed streams is of growing importance in regions such11.2. Literature Reviewas British Columbia with large-scale industrial forest operations and where pressure todevelop these systems for small scale power generation and impacts due to transmissionlines, pipelines and mining is increasing.1.2 Literature Review1.2.1 Process Response ModelsThe character and behaviour of a stream at any particular location reflects the net effectof landscape, upstream, and local basin variables, referred to as governing conditions, thatexert control on channel morphology. At the reach scale, the frequency, magnitude andduration of streamflow and the valley slope (which together determine transport capac-ity), the frequency, volume and size distribution of sediment input (i.e. sediment supply),and the boundary materials (including both alluvial and non-alluvial material and veg-etation), determine dependent (response) channel variables, such as width, depth, bedslope, resistance and surface grain size. Engineers, geologists and geomorphologists havelong pursued some methodology to explain and/or predict channel response to changes ingoverning conditions. However, it is important to determine at what temporal scale theinformation needed.The importance of temporal scale in fluvial geomorphology is that it determines whatthe independent and dependent variables are, because at some point, over a long enoughtime span, most variables become dependent (Schumm, 1977). Schumm and Lichty (1965)proposed that three time scales exist in geomorphology:• cyclic or geologic time — which is at the scale of landscape erosion cycles (millionsof years);• graded time — a shorter period than cyclic during which a graded condition ordynamic equilibrium exists (tens to hundreds of years); and,• steady time — during this time a static equilibrium may exist (less than a month).For the purposes of investigating channel response to changes in governing conditions,the graded time scale is the span of interest as at this scale channel morphology is dy-namically dependent upon the specified governing conditions. However, an analysis atthis temporal scale can not simply ignore the historical contingency or conditioning of thechannel in understanding its response. Further, the assumption that the morphology ofthe channel and its governing conditions are ’causally independent of each other’ is notnecessarily true. Short-term, small scale processes can influence morphodynamics over thelonger term and at larger spatial scales (Lane and Richards, 1997).Schumm and Lichty (1965) referred to a graded condition or dynamic equilibrium thatexists during graded time. Mackin (1948) defined a graded condition or a graded stream21.2. Literature Reviewas one that, over time, has adjusted its slope to produce a velocity sufficient to transportthe sediment supplied with the available discharge and channel characteristics. The gradedstream, therefore, is a stream in equilibrium and any change in the governing conditions willresult in adjustments in channel slope to absorb the effect of the change (Mackin, 1948).This definition overemphasized slope and ignored both adjustments in channel storage andtransport capacity (Lisle and Church, 2002) amongst others (for example, width, grainsize and armouring, and sinuosity). Dynamic equilibrium, discussed as early as Gilbert(1880), was defined by Chorley and Kennedy (1971) as a circumstance in which fluctua-tions are balanced about a constantly changing system condition which has a trajectory ofunrepeated average states through time.In the absence of data of a ’sufficient’ nature to develop quantitative predictions ofchannel response to changes in governing conditions, Lane (1955) proposed a very generalexpression:QS ∝ QsD (1.1)where Q is discharge, S is channel gradient, Qs is sediment flux of the bedload, and D issediment size (for Lane it was indexed by the D50). The expression describes the conditionunder which a river is in equilibrium with its imposed water and sediment fluxes. Here,discharge sets the scale of the channel, and together with gradient determines the rate ofenergy expenditure, and the morphology is then determined by the size and quantity ofsediment delivered to the channel (Church, 2006). This equilibrium model then predictsthat an increase in discharge can be offset with a decrease in channel gradient, i.e. degra-dation, or increase in the sediment flux or grain size. It would appear that one limitation inthe relationship is that only Q and Qs are independent variables, as the other variables canall be adjusted by the channel. Thus the equation can be rearranged to have independent(governing conditions) and dependent (response) variables on separate sides:QsQ∝SD(1.2)From above it can be seen that any changes in sediment supply or discharge are com-pensated for by adjustments in the bed slope or surface grain size. As with Mackin (1948),this model does not account for changes in the storage of sedimentSince Lane, many more functional relationships have been proposed based on the typeand direction of change in sediment supply and/or discharge, and the response (expressedqualitatively as either an increase or decrease) of an expanded set of channel responsevariables, which may include meander wavelength (λ), channel sinuosity (L), which areboth essentially slope metrics, mean depth (d), width (w), and the width-depth ratio(Schumm, 1969, 1977; Nunnally , 1985; Simon and Hupp, 1986; Kellerhals and Church,1989; Montgomery and Buffington, 1998). One of the drawbacks of these types of modelsis they can be indeterminate. For example, the Schumm (1969) model indicates that31.2. Literature Reviewdecreasing discharge while increasing sediment supply will produce a decrease in channeldepth and sinuosity but either increase or decrease channel width.One of the main limitations with predicting channel response lies in the number of waysa channel can respond. The primary response modes that have been identified include:• bed slope (Mackin, 1948; Lane, 1955);• channel geometry and planform (Schumm, 1969);• bed texture (Lane, 1937; Dietrich et al., 1989); and,• channel form and roughness (Madej , 2001; Eaton et al., 2004; Gran and Montgomery ,2005).Additionally, the nature of the response is contingent upon both historical and existingconditions (Church, 1995; Brewer and Lewin, 1998; Eaton and Lapointe, 2001; Talbot andLapointe, 2002), channel sedimentology (Simon, 1992; Gaeumann et al., 2005; Bartley andRutherford , 2005), and sediment supply within a reach and from upstream sources (Kasaiet al., 2004; Schuerch et al., 2006). The end result is that similar changes in governingconditions can produce very different channel responses, depending on the local conditionsfound in different reaches (Madej and Ozaki , 1996; Carling , 1988).Although some insight has been gained from process response models, for example bedsurface and slope response, a detailed understanding of reach scale response still remainsillusive. Specifically, understanding how slope, grain size and bar forms adjust to changesin either sediment supply or discharge and how these adjustments are influenced by his-toric conditions of the channel. As there are a variety of channel types in a variety ofenvironments, it is necessary to limit the scope of the research for brevity.In the Pacific Northwest, intermediate-sized forested gravel-bed mountain streams arecritical habitat for many aquatic species including salmonids, they supply clean water fordownstream users, and are increasingly being utilized for small scale hydro generation.Increasing interest has been brought to these streams in British Columbia as companiessearch for systems suitable for small scale hydro, as well as attempting to assess impactsof forestry in a results based regulatory environment, and understanding the long-termeffects of mining. In Washington, declining salmon stocks have renewed interest in habitatimprovements, and thus renewed interest in improving our knowledge of how these systemsrespond to changes in discharge and sediment supply. The focus of this research is onintermediate-sized forested gravel-bed streams.Church (1992) identified intermediate streams as having relative roughness (D/d) of 0.1to 1.0 and channel widths of up to 30 m: the streams of interest here are those with widthsbetween 10 and 20 m. In forested environments, bank strength is increased by root strength(Millar and Quick , 1993), which leads to channel stability. Lateral instability in forestedchannels appears to be related to either changes in the riparian vegetation, such as theresult of logging (e.g. Millar , 2000) or due to in-channel aggradation of sediment (Madej41.2. Literature Reviewand Ozaki , 1996; Brummer et al., 2006). An improved understanding of how responseto changes in governing conditions within the channel, may shed additional light on tomechanisms of lateral activity. Channel response can therefore be restricted to changesof variables within channel itself, such as bed topography and surface texture. Thus, inthis study, attention will be given to changes in slope, depth, and sediment characteristics,which require measurements of changes in velocity, surface grain size distributions, andchannel bedforms.The type of channels investigated here have been further described as being thresholdchannels by Church (2006). In this channel type bedload transport characteristically occursonly at low intensity during high flows. Threshold channels are relatively stable for longperiods of time, except during major floods or following development of in-channel log jams,which may induce local channel aggradation that can lead to lateral channel instabilityand/or avulsions. Additionally, instability may be related to changes in the boundaryconditions either due to riparian disturbance or with an influx of sediment.Intermediate (or threshold) channels can exhibit either plane-bed or pool-riffle mor-phologies (Montgomery and Buffington, 1997). A plane-bed channel is a largely featurelesschannel that lacks significant depositional structures, such as channel bars, and is mostcommon at channel gradients of 0.01 to 0.03. Channel bars represent major storage placesfor bedload and are important flow resistance elements which may evolve in the shortterm in response to changing flow and sediment transport conditions; bars generally donot occur if the flow depth is less than three times the grain size (Church and Jones,1982). Plane-bed channels commonly exhibit an armoured bed surface, which is indicativeof transport capacity being greater the sediment supply (Dietrich et al., 1989). In additionto armouring, bed structures, such as grain imbrication and grain clusters, are commonin these streams, and can also lead to decreased sediment mobility (Lane, 1937; Church,1998). In contrast to plane-bed channels, pool-riffle channels tend to have regularly oc-curring bedforms generated by local flow convergence and divergence and may be eitherfreely formed by cross-stream flow and sediment transport, or forced by channel bends andobstructions (e.g. Lisle, 1986). This morphology tends to be more responsive to changesin sediment supply and discharge than plane-bed channels.A brief review of the existing literature on channel response to changes in sedimentsupply and discharge is provided below.1.2.2 Sediment SupplyA variety of research papers have documented the effects of sediment supply on channelmorphology. The scales of response to changes in sediment supply have been investigatedat the grain scale, the bedform scale and the reach scale, as such the research will bereviewed at these three scales.The response of surface grains to changes in sediment supply has been the focus of muchinterest for over 20 years. To examine grain response to changes in supply, Dietrich et al.51.2. Literature Review(1989) used a plane-bed flume and observed that a 90% reduction in the feed rate resultedin a 32% increase in the median grain size of the surface. Dietrich et al. (1989) proposedq∗, which is the ratio of sediment transport predicted with and without armouring of thebed surface:q∗ = [(τb − τcs)/(τb − τct)]1.5 (1.3)in which q∗ is the transport ratio, τb is the boundary shear stress determined with DuBoysshear stress estimate (ρgdS, where ρ is the density of water and g is gravitational acceler-ation), τcs is the critical shear stress for the surface material, and τct is the critical shearstress of the transported bedload (following Hassan and Church, 2000).If q∗ = 1 the sediment supply is so high that no armouring exists; when q∗ = 0 nosediment is being transported. Lisle et al. (1993) observed similar but stronger texturaladjustment in experiments using a wider flume with alternate bars. They noted that a 90%reduction in the feed rate resulted in a 62% increase in mean surface grain size. In the casewith bars, the bed surface coarsening was accomplished by narrowing of the zone of activebedload transport, accretion of coarse particles onto emerging bar heads, and winnowingof fines in inactive areas of the bed (Lisle et al., 1993). The variability of the active zone insediment transport was previously noted by Gilbert (1914). These observations support theidea that pool-riffle channels may be more sensitive to changes in supply than plane-bedones. This is likely related to both longitudinal and lateral variability (Ferguson, 2003) ofthe bed surface, which may allow a wider range of response as well as concentrating theeffects over a much smaller area of the bed than under plane-bed conditions. The resultsof a numerical model presented in Francalanci et al. (2012) confirmed that the net effectof the variability in the bed surface was an increase in the transport rate over a plane-bed.In addition to coarsening of the bed surface, decreases in sediment supply (or in thecase of Church et al. (1998), the elimination of all sediment supply) may contribute tothe development of bed surface structures. Lane (1937) noted that, over time, the surfaceof a channel can rearrange itself to become more stable. In follow up experiments toChurch et al. (1998), Hassan and Church (2000) investigated low sediment feed rates over apreviously armoured and structured bed and found that increased sediment supply resultedin a reduction in surface armouring but bed structures remained relatively similar. Thissuggests that bed structures may play a larger role in channel stability over a wider rangeof fluctuations in sediment supply than does armouring. It is uncertain at this point whateffect the combination of bed structures and bars would have in terms of channel response.Iseya and Ikeda (1987) and Lisle et al. (1991) observed that the bed surface becamesorted into three distinct textural patches in response to local variations in sediment supply.The zones included a smooth zone of mostly sand over which gravel moved at high velocities,a congested zone of mostly stationary, interacting gravel, and a transitional zone (Iseya andIkeda, 1987). Further Iseya and Ikeda (1987) found that varying the percentages of graveland sand in their flume mixtures resulted in three different bed states, smooth (more61.2. Literature Reviewsand), transitional and congested (more gravel). Field studies by Dietrich et al. (2005)observed much greater transport rates over fine patches than coarse patches. Smith (2004)observed a similar response in flume experiments, and noted that the greatest transportrates occurred when fine patches were longitudinally linked, rather than in distinct patches.At the grain scale, the channel can respond either by:1. armouring;2. developing surface structures; or,3. developing patches.Changes in sediment supply can be accommodated by one or all three of these mech-anisms. In total they operate to stabilize the channel under changing sediment supplyconditions.Sediment supply to channels in mountainous regions is inherently stochastic and sedi-ment supply to channels usually occurs as a complex series of pulses (Benda and Dunne,1997a). Lisle et al. (1997) introduced a single pulse of sediment into a recirculating flumewith migrating alternate bars. The initial response was adjustment in bed slope and no de-tectable differences in bars were identified. Sediment from the pulse appeared to overpassthe bars without being deposited. Madej and Ozaki (1996) documented a large increasein sediment supply at Redwood Creek using 20 years of channel cross sectional changes.Channel response included channel widening, decreases in both pool spacing and depths,and persistently higher sediment yields. High sediment supply inhibits the developmentof armour layers and promotes surface fining (Lisle, 1982; Lisle and Hilton, 1992; Madejand Ozaki , 1996), which drowns out roughness associated with bars (Lisle, 1982; Buffing-ton and Montgomery , 1999b; Kasai et al., 2004), all of which lead to increased transportcapacity (Buffington and Montgomery , 1999a; Lisle and Church, 2002). This suggests astrong linkage between channel response and channel resistance.An additional linkage may exist between channel response and channel storage. Lisleand Church (2002) suggested that strong linkages exist between transport capacity andthe volume of sediment stored in the channel. They proposed that changes in transportcapacity reflect changes in channel planform, geometry, surface grain size, and surfacestructure, all of which have been shown to be directly linked to changes in sediment supply.In channels with either increasing or decreasing rates of sediment supply, two distinctphases of sediment transport exist. Phase I is an aggradation (transport-limited) phasewhere changes in sediment supply cause relatively small changes to sediment transportrate; transport in this phase is non-selective, and is accommodated by changes in storage.Phase II is a degradation (supply-limited) phase and is indicated by reduced sedimentmobility, due to increased armouring and form roughness. However, there is no suggestionas to how channel response would be different due to changes in morphology.71.2. Literature ReviewLisle et al. (2000) examined how a channel responds to changes in sediment supply atthe reach scale and found that it is generally the result of interactions between channelform, local grain size, and local flow dynamics that govern bed mobility. Channels withhigh sediment supply have greater areas of their channel involved in full mobility, whereaslow sediment supply channels had greater areas of partial mobility. However, under bothsupply rates, large areas of the channel exhibited little to no mobility. They also found thatareas of finer bed material seem to be the most responsive to sediment supply changes, andthe coarse areas of the channel created during some initial disturbance remain relativelystatic. Two reach scale parameters determined for bankfull stage (modified Shields numberand Q∗) seemed to correlate reasonably well with sediment supply. The modified Shieldsnumber (θ), a ratio of tractive and gravitational forces acting on bed particles, is definedas:θ = pgdSg/g(ρs − ρ)D50s (1.4)where pgdSg represents a modified boundary shear stress (τb) where channel slope S is re-placed with Sg which is the energy slope attributed to grain resistance, calculated followingParker and Peterson (1980), ρs is sediment density, and D50s is the median particle sizeof the bed surface. Shields stress represents the ratio of tractive and gravitational forcesacting on bed particles. Lisle et al. (2000) defined Q∗, modified from Dietrich et al. (1989),as the ratio of the mean predicted transport rate of the reach-averaged median particle sizeof the subsurface material (D50ss) given the armouring measured by the surface particlesize to the transport rate assuming there is no armouring (D50s). Of the two reach scaleparameters, they concluded that Q∗ was the best choice in that it can be directly linked tochanges in sediment supply, and therefore may have greater predictive capability of channelresponse (Lisle et al., 2000). However, they also caution that it is unlikely that either ofthe metrics can predict response from anything smaller than a doubling of sediment supply.A variety of field and experimental studies have investigated the impacts of changes inthe size of the sediment supplied on channel response, more specifically sediment trans-port. Gilbert (1914) noted that the addition of fine sediment increased the mobility ofcoarser grains. Similarly, Jackson and Beschta (1984) increased the amount of sand intheir flume experiments and noted increased rates in gravel transport and instability inpreviously stable riffles. Wilcock et al. (2001) also found increasing gravel transport rateswith increasing sand content, even after the proportion of gravel on the bed had decreased.Their findings seem to indicate that the increased transport capacity due to the sand con-tent can in turn limit the magnitude of channel response to large sediment inputs as thechannel can evacuate the sediment much more quickly (Wilcock et al., 2001). Gran andMontgomery (2005) observed channel recovery following fine lahar deposits from MountPinatubo, and observed the initial preferential transport of the finer material, followed bythe development of bed structures amongst the larger classes and ultimately an increase inboth form roughness and critical shear stress.81.2. Literature ReviewThe response time to changes in sediment supply is also an important issue and dependson a variety of factors. Parker (1990) proposed that the time line for surface response ismuch quicker than other adjustments, and is therefore the primary response to changes insediment supply. In a numerical model of the effects of sediment pulses Cui and Parker(2005) found that the transport capacity of the system, as well as the size of the pulse,determine the recovery time of the system. Cui et al. (2003) examined the effects of differinggrain size in pulse material and found that finer grain sizes resulted in higher transportrates and faster recovery times than coarser material. Similar results were seen in the fieldby Gran and Montgomery (2005).1.2.3 DischargeChannel response to changes in discharge has also been investigated in the field and in thelaboratory, as well as through the use of numerical models. As above, the research will bebroken into three spatial scales (i.e. grain, bedform and reach scales).At the grain-scale Parker et al. (2007a) employed a numerical model to investigatethe effects of a sequence of repeated hydrographs on surface armouring. Their resultssuggested that gravel-bed rivers respond to repeated flood hydrographs by evolving a bedthat changes little in terms of either elevation or surface size distribution as flow varies.As result, nearly all the variation in transport capacity was being absorbed by the bedloadtransport rate and bedload grain size distribution. This conclusion was later verified influme experiments conducted by Wong and Parker (2006). Although both studies variedthe hydrograph shape, the peak discharge was held constant, as was the sediment feedrate, which may have influenced their conclusions. In a field study, Wittenberg and Newson(2005) found that variability of the flow hydrograph was related to the variability of bedstructure, and hence bed stability, and that peak discharges determined the rate of bedmaterial transport, whereas the recession limb regulated the critical patterns of deposition.Additionally, they found that the legacy of the last flood, in terms of the nature and extentof bed clusters, determined the effectiveness of subsequent floods in terms of sedimenttransport (Wittenberg and Newson, 2005). Uncertainty still seems to exist in how flowsalter surface grain sizes.Hassan et al. (2006) investigated the effects of hydrograph shape on the armouringprocesses, and in particular, the effect of flashiness of the flood on armouring develop-ment. They hypothesized that hydrologic regimes characterized by relatively flat, longhydrographs can be associated with conditions that promote the development of armour-ing, whereas regimes characterized by short, peaky floods tend to subdue or destroy thisarmour. Hassan et al. (2006) concluded that sediment supply tends to dominate the de-velopment of bed surface armouring while hydrograph shape plays a secondary, but alsoimportant role.Lewin (1976) found that the range of natural discharge values resulted in a variety ofbedforms and that the channel at any point in time cannot be represented by a single dis-91.2. Literature Reviewcharge value. Jones (1977) found that the speed and magnitude of the change in dischargeresulted in varying types of bedforms.At the reach scale, establishing what discharge is related to equilibrium morphology hasa long history of debate in geomorphology. The earliest work in quantifying a relationshipbetween discharge and reach scale morphology was undertaken in order to design stable,unlined canals in India (Kennedy , 1895; Lacey , 1930; Blench, 1969). This work, referredas regime theory, empirically found that stable canal and river dimensions varied withdischarge in the form:w ∝ Q0.5 (1.5)d ∝ Q0.33 (1.6)Leopold and Maddock (1953) found similar hydraulic geometry relations from dataobtained from gauged rivers in the midwestern United States:w ∝ Q0.5 (1.7)d ∝ Q0.4 (1.8)The exponents found were very similar to the regime equations. The approach ofLeopold and Maddock (1953), was similar to the regime approach, and the results areentirely empirical.Although the original work of Leopold and Maddock (1953) incorporated a range ofdischarge frequencies in developing their relations, subsequent work has focused on theuse of formative discharge (Eaton, 2013). Wolman and Miller (1960) concluded thatequilibrium channel form appears to be related to flows at or near bankfull, rather thanthe rarer floods of unusual magnitude. Ackers and Charlton (1970) further found in theirexperimental channels that a steady discharge, equivalent to bankfull, produces that samemeandering pattern as varying discharge. Carling (1987), however, noted that a thresholdexists between flows that maintain channel form and those that initiate channel change.Pickup and Warner (1976) found that channel form depended on the ability of the eventto erode channel banks, generally with a return period between 4 and 10 years. This ideawas supported by Pizzuto (1994), who found that Powder River expanded and contractedin response to variations in discharge where the relatively infrequent occurrence of rapidchannel expansion, was followed by slow channel recovery. He felt that a satisfactorymodel of fluvial processes should consider the cumulative effects of a wide variety of flowsoperating over many decades.In the field, the observations on the relation between discharge and channel form hasbeen mixed. From an analysis of gravel-bed rivers in Alberta, Bray (1975) found that thebest correlation between discharge and channel geometry was the 2-yr flood. Like Pizzuto(1994), McNamara et al. (2008) found that channel morphology was maintained by largeand infrequent summer flow events. Bartholdy and Billi (2002) found that moderate floods101.2. Literature Reviewwere associated with bend migration, while channel cutoffs were associated with majorfloods, those on a 10-yr to 20-yr recurrence interval. Hooke (2007) found that processthresholds for bank erosion, aggradation and degradation occurred on average several timesa year.Parker et al. (2003) used a flume to investigate the effects of flow variability and waterdiversion on mountain streams. They found that channels with a full hydrograph andno diversion of water tended to reduce the fines content in the bed surface observed atlow flows and increase variability of bed elevation. Surface fines content progressivelyincreased, the surface median grain size of the model gravel decreased, and the variabilityof bed elevation decreased as the degree of diversion became stronger (Parker et al., 2003).This contradicts earlier work by Parker et al. (1982) wherein they found that variations inflow and differential entrainment of bed particles may not be essential to the formation ofa bed surface in some gravel-bed rivers.1.2.4 SummaryMuch research has investigated how channels respond to changes in sediment supply anddischarge. The majority of the sediment supply experiments have tended to focus onlyon the response of the bed surface to imposed changes in supply. The understanding ofchannel response at the bedform or reach scale has been done in the field, where the lackof experimental control limits the ability of the research to isolate the underlying sourceof the response. The majority of the earlier experimental work focused on the effectsof low sediment supply on the bed surface, where recent work has investigated the fateof sediment pulses. However, neither has addressed channel response to either sustainedchanges in sediment supply or variable peaks, as occurs in natural rivers. Additionally,most of the experimental work has been conducted in flumes which, by design, limitsresponse to changes in the surface grain size.Experiments conducted with meandering channels seem more appropriate for largerfloodplain systems, and not to mountain streams. In mountain streams, channels aregenerally straighter and exhibit an irregular meandering or more complex pattern. Channelplanform is dictated more by boundary conditions, such as riparian forests, than by alluvialprocesses. The complexity in channel planform may be an additional factor in determiningchannel response and has not been sufficiently investigated to date.From a review of the literature, it is evident that there is still a need to acquire moreinformation on the response mechanisms available to complex mountain channels. Thelimited research done in these systems has either been in the field, where complex responsesare difficult to interpret or in the laboratory where too much oversimplification to channelplatform has been generally used. Additionally, as pointed out by Madej (2001), very littlework has been done on how bedforms respond, and further on how grain, bedform and reachscale adjustments work together under changes in governing conditions. A fundamentalstudy on mountain channels is required in order to better understand pathways of channel111.3. Objectivesadjustment to changes in governing conditions. This project was designed to address someof these gaps.1.3 ObjectivesForested mountain streams are continually subject to variable flows and sediment loads,and for the most part they remain relatively stable. For the most part, variability in thegoverning conditions can be accommodated within the channel boundaries. Only majordisturbances, those that exceed the capacity of the channel to adjust, lead to channelrelocation or widening. Consequently, the banks of the channels remain relatively static,or at least remain stable around some mean value over time. Most channel responsescenarios rely on discrete step changes in one or more independent variable and ignore thehistory under which the pre-existing channel configuration was developed. This projectwill focus on the temporal and spatial patterns of channel response to changes in sedimentsupply and discharge. Specific objectives of the project were to:• Investigate the effects of varying sediment supply and discharge on surface textureand characteristics of the transported sediment; and,• Investigate the effects of varying sediment supply and discharge on channel storageand morphology, and sediment transport rates.• Investigate sediment transport processes and pathways;Sediment transport and morphological response of a semi-alluvial channel was investi-gated using a Froude scaled physical model. The model was designed using Fishtrap Creekas a field prototype. Long-term channel processes cannot be observed in the field and thuslaboratory experiments offer an opportunity to explore long-term river behaviour.Equilibrium in the laboratory setting is commonly defined as being reached when sedi-ment output equates to sediment feed, at which point many experiments are either stoppedor one of the inputs is changed. One of the drivers in these experiments ended up beingthe pursuit of what happens after equilibrium is reached. Although it was not one of thestated objectives of the research, it ended up being responsible for the long experiments.1.4 Thesis OrganizationThe thesis is organized into three research chapters. Chapter 2 introduces the experimentaldesign and methods. Chapter 3 investigates the degree to which the level of sedimentmobility describes the sediment transport dynamics in a laterally confined stream andcompares these results to experiments conducted in a laterally unconfined stream.In Chapter 4 the spatial and temporal patterns of sediment transport are explored.Additionally, the transport-storage relation proposed by Lisle and Church (2002) is tested121.4. Thesis Organizationin an aggradational environment under conditions of constant flow and sediment supply.Lastly, the response of sediment storage within the channel to changes in sediment supplyand flow is investigated.In Chapter 5, a qualitative approach was undertaken to describe the observations ofsediment transport and transport pathways made during the experiments. Out of the tenexperiments conducted during this project, Exp. 6 was selected to simplify the discus-sion. The visual observations of sediment transport were used to infer flow and sedimenttransport behaviour in the experimental channel.A concluding chapter summarizes the results contained herein.13Chapter 2Experimental Design and Methods2.1 IntroductionThe research presented in this thesis is based on two sets of experiments using physicalmodels to study the processes of sediment transport and bed stability in gravel bed streams.One set was conducted in a stream table with fixed banks having an irregular planformand a mobile channel bed including a wide range of grain sizes. These experiments aredesigned to represent a specific field prototype stream (Fishtrap Creek, British Columbia)that has been extensively studied. Another set of experiments was conducted in a streamtable with mobile bed and banks having a similar range of grain sizes as the first set. Forthese experiments, no specific prototype stream was identified, and they are considered torepresent the morphodynamics of some unspecified, general set of gravel bed rivers. Themajority of the work herein is based on the fixed bank runs.2.2 Fixed bank - Mobile Bed ExperimentsFor the purposes of these experiments a stream table was constructed that was 7 m longand 0.9 m wide. Previous experiments investigating the effects of varying sediment sup-ply and/or discharge on channel morphology have used narrow flumes with smooth ver-tical walls (Parker et al., 1982; Iseya and Ikeda, 1987; Dietrich et al., 1989; Wilcock andSouthard , 1989; Kuhnle and Southard , 1990; Lisle et al., 1993; Wilcock and McArdell , 1993;Hassan and Church, 2000; Hassan et al., 2006; Nelson et al., 2009) or fully alluvial bound-aries (Friedkin, 1945; Schumm and Khan, 1972; Eaton and Church, 2004; Tal and Paola,2007; Braudrick et al., 2009). For these experiments, flume design was based on actualbank alignments for a prototype stream, and therefore exhibited a more complex (i.e., re-alistic) channel alignment than is typical in these types of experiments. A floodplain wasconstructed within the stream table using Styrofoam, into which a channel with irregularbanks was cut; the channel was filled with sediment to a depth of 5 cm (Figure 2.2).2.2.1 Prototype Stream and Scaling ConsiderationsExperimental design for the fixed bank experiments was based on field data collected inFishtrap Creek. Fishtrap Creek is an intermediate size forested gravel-bed stream whichdrains a 158 km2 watershed located in the Interior Plateau region of British Columbia,142.2. Fixed bank - Mobile Bed ExperimentsFigure 2.1: Apparatus used for the fixed bank experiments conducted at UBC , lookingupstream. White arrow indicates flow direction.152.2. Fixed bank - Mobile Bed ExperimentsCanada. Fishtrap Creek has been the focus of a number of studies investigating watershedresponse to an intense forest fire (for example, Eaton et al., 2010a,b). The 2003 McLureforest fire was a high-intensity fire that burned more nearly 70% of the watershed, includ-ing a significant portion of the riparian zone of Fishtrap Creek. In 2004, 11 cross sectionswere established over a 130 m study reach; this was later expanded to 27 cross sections in2006 (Phillips, 2007). In addition to channel cross sections, surface and subsurface sedi-ment samples were collected and a reach survey was conducted which produced a detailedlongitudinal profile and generalized planimetric map. The reach is located immediatelyupstream of a Water Survey of Canada (WSC) hydrometric station (08LB024), which hasoperated more or less continuously since 1971.The basis of physical modelling is founded on the principles of similitude, which requiresa sound understanding of the physical processes and recognition of a model’s capacity toreplicate those processes (Ettema et al., 2000). Similitude is used to relate a processes thatoccurs in the prototype to the model, which is at a different scale. The power of similitudeis that processes at different scales can be described using dimensionless parameters. Themain advantages in physical model investigations are the direct control of specific vari-ables and the possibility to observe and to measure the development of planimetric andelevational patterns (Young and Warburton, 1996).Complete similitude between the model and the prototype requires geometric, kinematicand dynamic similarity. Geometric similarity requires that the linear proportions are keptbetween the model and the prototype and hence that they have the same shape, this isrepresented by:λl =LpLm(2.1)where λl represents the length ratio between the prototype (Lp) and the model (Lm).Kinematic similarity requires that ratios of time and motion are similar, for example theshape of the streamlines of the prototype are replicated by the model at any particulartime. Dynamic similarity necessitates that the ratios corresponding forces in the modeland the prototype are preserved.The fixed bank model was designed using geometrically undistorted movable bed Froudescaling. A movable bed model must reproduce two phase flow (i.e. water and sediment)and can be described by the following seven characteristics parameters: water density (ρ),dynamic viscosity (µ), grain size (D), channel slope (S), hydraulic radius (R - defined asR = A/P , where A is the channel cross sectional area and P is the wetted perimeter),acceleration due to gravity (g), and sediment density (ρs) (Yalin, 1971). Some of theparameters can be combined to produced the shear velocity (u∗ =√gdS), and the immersedspecific weight of sediment (γs = g(ρs − ρ)), leaving a new set of parameter combinationsρ, µ,D, ρs, d, u∗, γs, which produces four dimensionless parameters:162.2. Fixed bank - Mobile Bed ExperimentsΠ1 =ρDu∗µ(2.2)Π2 =ρu2∗γsR; (2.3)Π3 =ρsρ; (2.4)Π4 =RD; (2.5)where the term Π1 represents the grain Reynolds number (Re∗), which relates grain sizeto the thickness of the laminar sublayer, Π2 is the dimensionless Shield number (θ), whichis the ratio of the boundary shear stress to the submerged weight of the characteristic grainsize, Π3 and Π4 represent relative density and roughness, respectively.In undistorted Froude scaled models the time-scales associated with the transport ofwater and sediment are the same (unlike for distorted models), thus avoiding the complexi-ties of the two fluids operating at different time-scales (Yalin, 1971; Young and Warburton,1996). This scaling ensures that the dimensionless Shields number is the same in bothmodel and prototype.The experimental design also ensured that the boundaries of both model and prototypeare hydraulically rough, since the critical Shields number is constant for hydraulically roughboundaries. This condition is maintained so long as the grain Reynolds number is greaterthan about 70 (Yalin, 1971), though some have argued for a lower limit (Ashworth et al.,1994; Peakall et al., 1996; Shvidchenko and Kopaliani , 1998; Moreton et al., 2002). Thegrain Reynolds number is calculated as:Re∗ =D90su∗ν(2.6)where D90s is the grain size at which 90% of the surface sediment is finer and ν is thekinematic viscosity of water. For these experiments, a length scaling ratio (λr) of 1/30 wasselected, this yields Re∗ = 180.The following scaling relations relate the parameters of the model to those of the pro-totype:1. the relative widths (wr = wmodel/wprototype), depths (dr), and grain sizes (Dr) arescaled linearly according to wr = dr = Dr = λr2. the relative slope (Sr) is unity, such that Smodel = Sprototype3. the relative velocity (vr) scales according to vr =√λr4. the relative time (tr) scales according to tr =√λr172.2. Fixed bank - Mobile Bed Experiments5. the relative discharge (Qr) scales according to Qr = λ5/2rThe fixed bank model was constructed in a stream table with an overall slope of 0.016m/m using a geometric scale ratio of 1:30. The design channel width (0.34 m) and depth(0.015 m) were determined by averaging the bankfull channel dimensions from 90 crosssections. See Table 2.1 for the prototype and model design parameters.Table 2.1: Experimental design parameters in prototype and modelValue in Fishtrap Creek Value in ModelDesign Q (m3/s) (Return Period) (L/ s)7.9 (2.8-yr) 1.69. (6-yr) 2.011.8 (11-yr) 2.4W (m) 0.34d (m) 0.015S 0.016D50feed (mm) 1.14D90feed (mm) 3.28Time (1 day) 4.38 hr (5 hrs used)Streamflow: Streamflows at Fishtrap Creek are dominated by snowmelt generatedevents. Based on 40 years of record from the WSC gauge located immediately downstreamof the study area, the estimated daily mean bankfull flood at the study site is 7.9 m3/shaving a return period of 2.8-yr, with a maximum mean daily flow of 14.9 m3/s (1997) overthe period of record. The streamflow return periods were determined using the ConsolidateFrequency Analysis (CFA) program (Pilon and Harvey , 1993). The streamflow data weretested for independence, trend and randomness, and passed at the 5% significance level.The duration of the average snowmelt event is ∼1 month, with the peak of the hydro-graph lasting, on average, one day. The estimated bankfull flow of 7.9 m3/s translates to1.6 L/s in the model. Larger magnitude flows of 2.0 L/s and 2.4 L/s were also used (seeTable 2.1). The estimated prototype flood duration of one day translates to about 5 hrsin the model.Grain Size Distributions: Field measurements made by Phillips (2007) and An-drews (2010) indicate that the median particle size of the subsurface (D50ss) at FishtrapCreek is about 36 mm and the D90ss is about 91 mm (Figure 2.2). Furthermore, it was ob-served that there were differences between the bedload, surface (Ds) and subsurface (Dss)grain size distributions. Surface grain size distribution data (Phillips, 2007; Andrews, 2010)indicates that the surface grain size distributions (D50s = 69 mm and D90s = 170 mm)were much coarser than the subsurface, which is not uncommon in mountain streams. Thebed material load, although not directly sampled, was approximated by a bulk sample182.2. Fixed bank - Mobile Bed Experimentstaken upstream of a hydrometric weir. This sample was much finer than both the surfacesamples and the other subsurface samples (D50ss = 15 mm and D90ss = 61 mm).Pitlick et al. (2008) proposed that sediment in gravel-bed rivers consisted of two popu-lations: one population which represents the subsurface and bedload material and anotherwhich represents the bed surface. Observations at Fishtrap Creek suggest that the bedload,subsurface and bed surface sediments are mostly from the same population, except for thelarger grains. The larger clasts on the bed surface may be the result of processes exogenousto the modern channel or rarely move (Church and Hassan, 2002). The bedload and sub-surface size distributions represent sediment transport via modern alluvial processes; thisis also the case for most of the surface grain sizes. The larger clasts present on the surfaceappear less frequently in the subsurface and may be the result of either higher magnitudeflow events, major disturbances or are of non-alluvial origin (for example, glaciofluvialdeposits). Observationally, these larger clast sizes appear to represent a major stabilitycomponent in Fishtrap Creek. During degradational periods the larger grains act as keymembers establishing stone clusters or can act as a degradational barrier when smaller sizeclasses are selectively transported downstream.The sediment used to create the bed in the model was collected from a sandur deposit;the bed material grain size distribution was sieved to fit a distribution that representedas closely as possible a 1:30 scale of the measured bed material grain size distribution inFishtrap Creek (see Figure 2.2). The bed material grain size distribution for the experi-ments was truncated at 0.177 mm (equivalent to 5.3 mm in Fishtrap Creek) and 8.0 mm(equivalent to 240 mm in Fishtrap Creek). The bed material grain size distribution wasthe same for Exp. 1 to Exp. 7; for Exo. 8 the feed grain size had a distribution equal tothe sediment output trapped during a previous run. The main difference between the twodistributions is in the truncation of the tails (Figure 2.2).Morphologic Characteristics: The average bed slope for the study reach is 0.02m/m overall, and about 0.016 m/m in sections without significant volumes of large wood.According to Montgomery and Buffington (1997) these slopes are within those expected forplane-bed morphology (0.015 to 0.030), but greater than that expected for pool-riffle mor-phology (<0.015). However, these slopes are consistent with forced pool-riffle morphology,which can extend the gradient range of pool-riffle channels from 0.02 to 0.035 (Montgomeryet al., 1995). Since wood was not included in our fixed bank model, we set the channelgradient for the model to the value observed in the field for relatively wood-free sectionsof the stream.Eaton and Giles (2009) suggested that the lateral stability in gravel bed streams likeFishtrap Creek is strongly conditioned by the nature of the riparian vegetation. The abilityof such streams to migrate laterally across their floodplains appears to be directly linked tothe disturbance regime of the forests adjacent to it, with migration events limited to thoseperiods of time during which the root strength of the riparian vegetation has been compro-mised by, for example, wildfire. Eaton and Church (2009) speculate that, between distur-bance events, Fishtrap Creek remains laterally confined within a fixed planform dictated192.2. Fixed bank - Mobile Bed ExperimentsFigure 2.2: Prototype and model bed material grain size distributions.202.2. Fixed bank - Mobile Bed Experimentsby the riparian forest and that the throughput of water and sediment is accommodatedby morphological adjustments within the active channel, not by lateral migration of thechannel; as a result, the use of a fixed bank physical model is appropriate when studyingthe long-term morphodynamics of this kind of river system.2.2.2 Flume Instrumentation and MeasurementsThe experiments employed a sediment feed system, only water was recirculated. Sedimentwas introduced at the upstream end of the stream table using a rotating sediment feeder(Figure 2.3). Prior to entering the channel, pre-mixed sediment was loaded into the feeder(Figure 2.3a). Sediment then exited the rotating sediment feeder into a PVC pipe whichthen exited onto the channel (Figure 2.3b). The volume of sediment exiting the feeder waschecked every 5 hr to ensure that the feed rate remained constant. Water was pumpedinto a head pond then spilled over a broad crested weir before entering the channel.Discharge was measured using an Omega FP-1521 digital flow transmitter, which isessentially an in-line impeller flow meter. Measurements of the flow meter, determinedas the percentage of the actual value, are accurate to plus or minus 5%. Average flowvelocities were measured using a tipping bucket that injected a 10 mL slug of a salt-watersolution into the upstream end of the flume. The slug was introduced into the middleof the channel at the upstream end of the flume to ensure full mixing of the tracer bythe time it reached the probe 5.5 m downstream (Figure 2.4a). Estimation of averagevelocity was calculated using the spatial harmonic mean velocity as described in Waldon(2004). Waldon (2004) found that the harmonic mean provides an unbiased measure ofmean velocity compared to the peak of the pulse which over estimates the velocity and thecentroid which underestimates the mean velocity.All sediment leaving the table was initially collected at the outlet every 15 minutes in 1L plastic sample jars. Trap efficiency problems led to a modification of the outlet; a Helley-Smith sample bag was attached to the outlet and a 0.0125 mm sieve placed underneaththe sample bag to ensure all sediment transported out of system was captured. Collectedsediment was dried and weighed at 15 minute interval. Hourly sediment samples were thencombined from the 15 minute samples and split, the resulting sample was sieved to 1/2 Φintervals. The sediment was then re-mixed to the distribution of the initial feed mixtureand put back into the sediment feeder.Every hour during the experiment, observations of channel morphological developmentand sediment transport pathways were made and sketched onto maps. In addition, watersurface elevations and depths on both sides of the channel were measured at 10 equallyspaced cross sections using a ruler, and were used to map the water surface position duringeach experimental run. An overhead camera monitored a 2 m section of the channeland took images at 3 frames per second (fps). These images were post-processed afterthe experiment was completed to produce a video in which sediment movement could bemore easily observed. This information provided additional verification to the information212.2. Fixed bank - Mobile Bed Experimentsb)a)Figure 2.3: Photo of a) sediment exiting the feeder, and b) sediment feeder with sediment.222.2. Fixed bank - Mobile Bed Experimentsb)a)Figure 2.4: Photos of a) converted tipping bucket rain gauge used to introduce salt mixtureinto the stream, and b) time delayed image of laser cart during a measurement of thechannel.232.2. Fixed bank - Mobile Bed Experimentsrecorded on the hand drawn maps.Every 5 hours the experiment was halted to collect surface sediment samples and surveythe bed of the channel. However, for Exp. 1 surface sediment samples were collected everyhour. Nine surface sediment samples were taken at the same locations in the flume every5 hours. Samples were collected using a square plexiglass plate, side length of 3.7 cm andsurface area of 13.7 cm2, covered with a layer of wet clay, ≈1 cm thick (Diplas and Fripp,1992). The plate was pressed firmly against the bed and dipped into water to release anysediment adhering to the sample due to surface tension effects. A series of pilot studiesdetermined that, while the presence of the plunger produced local erosion and deposition,these effects were generally short-lived. The minimum sampling area was calculated usingthe maximum particle size of the surface material, 0.8 cm, and equations from Diplas andFripp (1992) and Fripp and Diplas (1993). The minimum sampling area was determineto be 64 cm2. To ensure an adequate sample size between six and nine samples weretaken during each sampling period (Diplas and Fripp, 1992; Fripp and Diplas, 1993). Theclay-sediment mixture was rinsed onto a 0.125 mm sieve to remove the clay, the remainingsediment was dried, weighed, and sieved at 1/2 Φ intervals.To survey the channel bed, five red line lasers (3.5V ≈ 4.5 mW 16 mm) were mountedon an instrument cart at 10 cm intervals approximately 0.8 m from the bed surface. AProsilica EC 1280 (1280 x 1040 pixels) firewire video camera was used to photograph eachlaser line from an oblique angle, to obtain cross sectional topographic data with a 1 mmresolution. This system produced a survey that had cross sections spaced at 1 cm intervalsover a 4.50 m length in the middle of the stream table. Capture and processing of thedata was achieved using a combination of Labview and Matlab programs. DEMs weregenerated from the original data using triangular based linear interpolation to 2 mm2 gridcells. These grids were then imported into SAGA GIS and the grid difference module wasused to document changes in channel morphology between surveys. Channel slope wasdetermined by fitting a regression through the mean elevation of each of the 450 crosssections. An example of a scan in progress is provided in Figure 2.4b.Samples of surface grain size were collected before, during and after each experimentusing a square plexiglass plate, side length of 3.7 cm and surface area of 13.7 cm2, coveredwith a layer of wet clay, ≈1 cm thick (Diplas and Fripp, 1992). Samples were taken everyhour. The plate was pressed firmly against the bed and dipped into water to release anysediment adhering to the sample due to surface tension effects. A series of pilot studiesdetermined that, while the presence of the plunger produced local erosion and deposition,these effects were generally short-lived. The minimum sampling area was calculated usingthe maximum particle size of the surface material, 0.8 cm, and equations from Diplas andFripp (1992) and Fripp and Diplas (1993). The minimum sampling area was determineto be 64 cm2. To ensure an adequate sample size between six and nine samples weretaken during each sampling period (Diplas and Fripp, 1992; Fripp and Diplas, 1993). Theclay-sediment mixture was rinsed onto a 0.125 mm sieve to remove the clay, the remainingsediment was dried, weighed, and sieved at 1/2 Φ intervals.242.2. Fixed bank - Mobile Bed Experiments2.2.3 Experimental ProtocolsTen experimental runs were conducted using varying rates of discharge and sediment supply(see Table 2.2). Experimental runs either began from a screeded bed, having a uniformsediment depth and surface grain size distribution, or on a bed inherited from the previousexperiment, referred to as a conditioned bed. Experiments that began with a screededbed are indicated by bold numbers in Table 2.2. Prior to the start of those experimentswith a screeded bed, flows were run at about half the estimated bankfull flow (0.8 L/s)for 2 hours in order to wet up the bed and let an initial bed surface develop. Althoughsediment transport was observed in the channel during this initial phase, no sedimentexited the stream table. As discussed in more detail above, after every 5 hours duringan experimental run, flows were turned off and various measurements were made: at thebeginning of the next run, flows were ramped up incrementally over 15 minutes in order toprevent major morphologic changes from occurring. While minor adjustments did occurduring the shut down and start up period, they were restricted primarily to pool headsand riffle tails. The only deviation from this experimental protocol occurred during Exp.10, which was run in 2 hr increments between bed surveys.The return period of the equivalent prototype discharge in the experiment is providedin Table 2.2. The table also provides the sediment feed rate (Qf ) for each experiment andthe average sediment output over the duration of the experiment (Qb). The D50 of the bedsurface (D50s) and the transported sediment recorded at the outlet (D50t) are given.Equilibrium conditions during the experiments were defined as equality in sediment in-put and output, which is consistent with previous experimental work (for example, Dietrichet al., 1989; Lisle et al., 1993; Eaton and Church, 2004, 2009; Venditti et al., 2010).252.2.Fixedbank-MobileBedExperimentsTable 2.2: Experimental conditionsExp. Q Tra Qf Time Bed Slope Depth τbb Qb D50sc D50td D50s/D50tL\s yrs kg\hr min ×102 cm Pa kg\hr mm mm observedFeed1 1.6 2 3.24 5700 1.53±0.062 1.47±0.080 2.21±0.16 2.72±0.54 2.59 1.15 2.282 1.6 2 1.65 2400 1.65±0.032 1.50±0.076 2.42±0.16 1.37±0.24 3.03 1.12 2.723 1.6 2 1.65 2400 1.51±0.042 1.58±0.051 2.34±0.11 0.76±0.19 2.37 1.04 2.284 2.0 4 1.98 2400 1.52±0.093 1.60±0.067 2.39±0.18 1.54±0.32 2.89 1.16 2.515 2.4 11 2.47 1800 1.62±0.038 1.85±0.025 2.93±0.08 2.16±0.54 3.29 1.06 3.106 2.0 4 1.80 2400 1.62±0.112 1.72±0.045 2.73±0.24 1.23±0.42 2.64 1.03 2.597 2.4 11 1.80 1800 1.61±0.060 1.81±0.041 2.84±0.15 1.74±0.68 3.13 1.13 2.798 2.4 11 1.80 1500 1.66±0.052 1.79±0.020 2.91±0.08 1.86±0.11 3.22 1.14 2.84No feed9 2.4 11 - 900 1.62±0.009 1.99±0.104 3.16±0.17 0.21±0.27 3.12 1.11 3.0910-1 2.6 18 - 120 1.61 2.18 3.45 0.09 3.27 1.64 2.0010-2 2.8 30 - 120 1.60 2.60 4.08 0.14 3.81 1.40 2.7210-3 3.0 55 - 120 1.60 2.43 3.82 0.36 3.60 1.55 2.3310-4 3.2 95 - 120 1.50 2.52 3.70 0.66 3.61 1.55 2.3310-5 3.4 150 - 120 1.46 3.19 4.58 0.72 3.17 1.12 2.61a Tr equals return period of equivalent prototype discharge, estimated using Extreme Value Gumbel distribution.b τb calculated as ρgdS.c D50s is the median particle size of the bed surface.d D50t is the median particle size of the transported bed material collected at the channel outlet.262.3. Mobile bank - Mobile Bed Experiments2.3 Mobile bank - Mobile Bed ExperimentsThe details of the mobile bed experiments have been previously discussed in Eaton andChurch (2004), so only a brief summary of the methods are presented here. The mobilebank experiments were conducted in a 20 m long tilting stream table set at a slope of 0.01m/m (see Figure 2.5). The channel was allowed to freely migrate across a 3 m wide, 12cm deep floodplain composed of material with a grain size distribution identical to thebed and the sediment feed. Each experiment began with a straight rectangular channelcut into the floodplain. The physical model was a generic, 1:32 scale representation ofa moderately steep ( 1%), meandering gravel bed river with a D50 of about 22 mm anda bankfull discharge of ≈ 17 m3/s. A commercially available sand mixture (see Figure2.2) was used to create the floodplain approximately 12 cm deep on the stream table withD50 ≈ 1 mm. This material was collected from a naturally sorted beach deposit, and wasnot modified during excavation on-site, nor in the laboratory.As in the fixed bank experiments, the model grain size was truncated at 0.177 mm.Water flowed onto the centre of the stream table at the upstream end through a trayoriented at 25◦ relative to the stream table centreline, so as to generate an initial bend.A sediment feed unit introduced sediment to the system that had the same grain sizedistribution as the floodplain sediment. All sediment leaving the stream table at thedownstream end was captured in a sediment trap. Trap efficiency was nearly 100% formost size classes, but dropped for the smallest size classes that were carried out of the trapby turbulent eddies.Measurements of the water surface elevation and of the bed topography were madeusing a point gauge. Surveyed cross sections were located at each apex and cross-over tocharacterize the channel bed topography. All results are based on measurements from themiddle half of the stream table, away from any potential inlet and outlet effects (totalstudy length of ≈10 m ).Samples of the bed and bank surface texture were collected following each experimentusing a flexible rubber plate covered with a layer of wet clay. Samples were also taken ofthe bed armour, approximately located at the thalweg, for each cross section.272.3. Mobile bank - Mobile Bed ExperimentsFigure 2.5: Photo looking upstream at the mobile bank model.28Chapter 3Sediment Mobility and ChannelStability: Implications fromFroude-scaled Experiments3.1 IntroductionThe morphology of gravel-bed rivers is primarily a consequence of the erosion, transportand deposition of bed material. Bed material found in mountain gravel-bed rivers mayoriginate from either endogenous or exogenous sources and processes, and consequentlythe grain size distribution of the material may span orders of magnitude. The ability ofthe channel to erode and transport the range of sizes found within the channel determines,in a large part, its stability.Field studies have observed bed material transport to occur in either two or threedistinct phases. Jackson and Beschta (1982) (see also Emmett , 1976; Andrews, 1983)proposed two phases of bedload transport in pool-riffle morphologies: Phase I involves thetransport of fine material during low to moderate flows over a static bed, this material isusually recruited from pools, channel margins and behind in-stream obstructions; Phase IIoccurs during flows high enough to entrain sediment in riffles, and involves the transportof most grain sizes found in the bed material, and is associated with local mobilization ofthe armour layer.Ashworth and Ferguson (1989) redefined the two phases of Jackson and Beschta (1982)based on observations of sediment transport in gravel-bed rivers into three phases. PhaseI involves movement of fine material over a static bed during low flow periods. PhaseII occurs during moderate flows when local bed material is entrained and transported.The existence of this phase suggests that at some flows entrainment is size dependent, inthat smaller grains are more likely to be entrained under these flow conditions than largergrains. Phase III occurs during high flow events and produces conditions similar to theclassical notion of full mobility. Warburton (1992) extended the model of Ashworth andFerguson (1989) to step-pool streams; during his Phase 2 flows were high enough to breakup of the gravel portion of the bed. During Phase 3, flows were high enough to result inthe destruction of boulder structures and step-pool sequences (Warburton, 1992).Wilcock and McArdell (1993, 1997) introduced the concept of partial mobility in con-293.1. Introductiontrast to the classical notion of full mobility. Church and Hassan (2002) and Hassan et al.(2005) modified the existing model to include the concepts of partial (Phase II) and fullmobility (Phase III) to describe the three sediment transport phases. Partial transportoccurs when only a given size range of surface grains are mobilized, and the remainder areimmobile (Wilcock and McArdell , 1993). Haschenburger and Wilcock (2003) later modifiedthis definition of partial mobility to describe conditions in which only portions of the bedsurface are mobile, irrespective of grain size, and other portions remain immobile. Earlier,in his three-phase model, Carling (1988) attributed the partial mobilization of ”smallerframework gravels” to Phase II sediment transport which is associated with flows greaterthan 60% of bankfull; during this phase only a minority of the coarser grains are entrained.Selective transport occurs when the all of the sizes present on the bed surface are presentin the transported bed material, but the transported load is finer (Parker et al., 2007b).A final condition that is related to bed mobility is equal mobility. The hypothesis ofequal mobility was initially proposed by Parker et al. (1982) and Parker and Klingeman(1982) and suggests that in order for a graded stream to move the coarse half of its meanannual load at the same rate as the finer portion, the grain size distribution of the surfacelayer must adjust so that the coarse portion of the surface sediment is overexposed and istherefore more likely to be entrained compared to the finer portion. The result is that thearmouring and over exposure of coarse grains ensures that both portions move throughthe system at the same rate. Parker and Toro-Escobar (2002) referred to this as the’weak form’ of the equal mobility hypothesis; the ‘strong form’ additionally requires thatthe grain size distribution of the bed material load is equivalent to that of the subsurfacewhen averaged over multiple flood events. The ’strong’ form of the hypothesis can be seenas a time averaged phenomenon relating the development of the armour layer to particlemobility.In experiments using scaled models of mountain streams Parker and Toro-Escobar(2002) found evidence for both forms of the hypothesis. Lisle (1995) compared the longterm average grain size distributions of bed material load to the subsurface grain size dis-tributions from data available for 14 reaches of 13 gravel-bed rivers. The data he presentedshowed that 8 of the 14 supported the strong form of the hypothesis; however, in 6 of the14 reaches analyzed the load was finer than the substrate. He observed that these streamswere commonly steep channels with coarse surfaces that presumably limited the annualamount of scour and fill. This lack of scour and fill may have limited the exchange betweenthe surface and subsurface layers during floods, thus invalidating the strong form of thehypothesis (Parker and Toro-Escobar , 2002). Others have also found that transported loadtends to be systematically finer that the bed material (for example Church and Hassan,2002; Gomi and Sidle, 2003; Ryan et al., 2005; Wathen et al., 1995; Whiting et al., 1999;Pitlick et al., 2008; Thompson and Croke, 2008).Pitlick et al. (2008) proposed that sediment in gravel-bed rivers consist of two popula-tions: one population represents the subsurface and bedload material and another whichrepresents the bed surface. Observations at Fishtrap Creek suggest that the bedload, sub-303.2. Resultssurface and bed surface sediments are mostly from the same population, except of the largergrains. The larger coasts on the bed surface may be the results of processes exogenous tothe modern channel or rarely move (Church and Hassan, 2002). The bedload and subsur-face size distributions represent sediment transport via modern alluvial processes. This isalso the case for most of the surface grain sizes. The larger clasts present on the surfaceappear less frequently in the subsurface and may be the result of either higher magnitudeflow events, major disturbances or are of non-alluvial origin (for example, glaciofluvialdeposits). Observationally, these larger coasts sizes appear to represent a major stabilitycomponent in the system. During degradation periods the larger grains act as key membersestablishing stone clusters or can act as a degradation barrier when smaller size classes areselectively transported downstream.In addition to the stability in the channel due to the presence of larger grain coasts, aspreviously mentioned, Eaton and Giles (2009) suggested that the lateral stability of Fish-trap Creek is related to the riparian forest. This lateral stability in the channel is directlylinked to the disturbance regime of the forests adjacent to it. The recurrence interval ofriparian disturbance, which in the interior region of British Columbia is mainly forest fires,occurs approximately every 100 years (Eaton and Church, 2009). This suggests that for amajority of the time Fishtrap Creek remains laterally stable with a fixed planform dictatedby the riparian forest and that the throughput of water and sediment is accommodated bymorphological adjustments within the active channel.The objectives of this chapter are to:1. design a Froude scaled physical model using Fishtrap Creek as a prototype;2. test the degree to which the level of sediment mobility describes the sediment trans-port dynamics in a laterally confined stream for flood flows that typically occur overthe period between disturbances (in Fishtrap Creek this period is approximately 100years); and,3. compare these results with a previous study in which the channel banks were aserodible as the bed.3.2 Results3.2.1 Sediment Transport RatesSummary data for the observed sediment transport rates, calculated at 15 minute intervals,for all of the experiments are presented in Table 3.1. Exp. 1 displayed the highest averagetransport rate, which would be expected as it was also conducted with the highest sedimentfeed rate, the lowest transport rates were associated with the degradation experiments(Exp. 9 and Exp. 10) and Exp. 3. To account for the role of sediment feed rate insediment transport rates, the mean transport rate for each experiment was divided by313.2. Resultsthe respective sediment feed rate (presented as the standardized mean in Table 3.1). Onaverage, Exp. 7 and Exp. 8 operated near equilibrium conditions; Exp. 3 appears to bethe outlier.Plots of sediment transport rates are shown in Figure 3.1 along with the respectivesediment feed rates, where applicable. Included on each plot is the centred moving average,with a width of 180 min. The moving average line provides a clear indication of trendin transport rates over time, as well as when the experiment is in equilibrium with thesediment feed rates. It can be seen that for most experiments sediment transport rateswere lower than the feed rates, indicating aggradation. Exp. 9 and Exp. 10 were theexception as they were degradational experiments. Exp. 8 appears to be in equilibriumfor nearly the entire run time (Figure 3.1h), while Exp. 1 (3.1a), Exp. 2 (3.1b) and Exp.7 (3.1g) experience periods of equilibrium.Table 3.1: Summary statistics related to sediment transport ratesExperiment 1 2 3 4 5 6 7 8 9 10Duration a 5700 2400 2400 2400 1800 2400 1800 1500 900 600Average b 1.34 0.72 0.37 0.76 1.06 0.61 0.86 0.91 0.10 0.48Standardized Mean c 0.84 0.89 0.46 0.78 0.87 0.69 0.97 1.02 - -Max b 3.03 1.81 1.03 3.56 4.22 3.84 5.85 1.85 1.27 2.02Min b 0.18 0.13 0.01 0.16 0.29 0.12 0.21 0.42 0.00 0.07SD b,d 0.48 0.29 0.18 0.56 0.60 0.46 0.63 0.27 0.20 0.43Aggradation e 8.92 3.03 14.11 7.32 4.06 9.44 0.89 -0.97 -3.55 -16.44a. Duration of experiment in minutes.b. Sediment transport rates in g/min-cm.c. Standardized mean was calculated for each experiment as the average sediment transport rate dividedby the sediment feed rate.d. Standard deviation.e. Average aggradation rate in g/min.The sediment output data can also be presented as cumulative differences between sedi-ment feed rates and sediment transport rates (Figure 3.2). Inspection of these plots revealshow significant the aggradation was for most of the experiments. Average aggradation ratesfor each experiment are provided in Table 3.1.Average transport rates for the first 2010 minutes in Exp. 1 were 1.00 g/min-cm whichis 30% less than the sediment feed rate (Figure 3.1a), an average sediment output rateof 1.39 g/min-cm was maintained for the duration of the experiment. From the figure,Exp. 1 appears to approach equilibrium following 2010 minutes; however, it was obviousthat bed elevations were increasing at various locations. From Figure 3.2a it is evidentthat Exp. 1 experienced persistent, but minor aggradation. The overall aggradation ratewas 11.89 g/min, however two breakpoints can be seen in the plot which indicates threedifferent periods of aggradation, separated by periods of relative equilibrium in sedimentflux. The first period lasted 1950 minutes and had an average aggradation rate of 21.07323.2. Resultsg/min. The second period, with a an average rate of 11.66 g/min, lasted between 2130 and2985 minutes. The final period began after 3735 minutes and lasted for the remainder ofthe experiment, during which an average aggradation rate of 9.43 g/min was observed.Exp. 2 (which continued on the bed from Exp. 1) had the same design discharge, butthe sediment feed rate was reduced by 56% (see Figure 3.2b), producing a feed rate thatwas well below the volume of sediment transport rate consistently observed during Exp. 1.However, the average sediment output rate for the experiment was 0.67 g/min-cm, whichis still below the reduced feed rate of 0.80 g/min-cm. Inspection of Figure 3.2b showspersistent aggradation during Exp. 2 (average rate of 4.76 g/min). The experiment washalted once flows began to over-top channel banks.During Exp. 3 flow was the same as Exp. 1 and 2 and the sediment feed rate was thesame as Exp. 2. The conditions were kept the same to explore the role which aggrada-tion played during Exp. 1 on sediment transport rates observed during Exp. 2, or morespecifically the role of the increase in channel slope (i.e. stream power). The averagesediment output rate for Exp. 3 was 0.35 g/min-cm, which is 43% of the sediment feedrate. This led to persistent aggradation, which is evident in Figure 3.2c. The average rateof aggradation during Exp. 3 was 14.79 g/min, compared to only 4.76 g/min during Exp.2. The average aggradation rate during the first 2400 minutes of Exp. 1 was 18.47 g/min.Sediment output rates during the first 2400 minutes of Exp. 1 were 1.06 g/min-cm, whichis three times greater than Exp. 3.Following the results of the first three experiments, discharge was increased to simulatethe 6-yr return period event in the prototype. The sediment feed rate for Exp. 4 wasdetermined by keeping the sediment concentration the same as it was during Exp. 2 and3.After an initial period of degradation, related to erosion of the unarmoured bed surface,the sediment output for Exp. 4 remained below the feed rate for the duration of theexperiment (Figure 3.1d). Sediment output was greater than the sediment feed rate only20 times over the course of the experiment, with only two periods as long as 1 hour whichoccurred at 1575 minutes and 2070 minutes. Persistent aggradation is evident in Figure3.2d, with the overall aggradation rate of 9.58 g/min following the first 60 minutes ofdegradation. A distinct breakpoint can be seen in the plot around 1575 minutes when theaverage aggradation rate changes from 11.89 g/min to 6.26 g/min for the remainder of theexperiment.In Exp. 5 flows were increased to 2.4 L/s to simulate approximately the 11-yr returninterval flood. The feed rate was also increased to keep sediment concentration the sameas it had been during Exp. 4. The transport rate remained elevated above the input ratefor 165 minutes during the first 225 minutes (Figure 3.1e). Sediment output rate exceededthe feed rate 34 times during the course of Exp. 5, 14 times more than had during Exp. 4.Following the initial period of incision, however, aggradation returned to an overall rate of0.55 g/min, which is much lower than the previous experiments. Additionally, the rate isnot consistent after 225 minutes, with breakpoints in the plot being associated with periods333.2. Results012340 200 400 600 800 1000 1200 1400 1600 1800012340 200 400 600 800 1000 1200 1400 1600 1800 2000012340 500 1000 1500 2000 2500012340 200 400 600 800 1000 1200 1400 1600 1800 2000012340 500 1000 1500 2000 2500012340 500 1000 1500 2000 2500012340 500 1000 1500 2000 2500012340 1000 2000 3000 4000 5000 6000012340 100 200 300 400 500 600 700012340 100 200 300 400 500 600 700 800 900 1000(f)-Q =-2.0-L/s4.22-0.80Time-(min)Sediment-Ouput-(g/min-cm)Exp.-1 Exp.-21.600.80(a)-Q =-1.6-L/s-Exp.-3(b)-Q =-1.6-L/s-(c)-Q =-1.6-L/sExp.-6(e)-Q =-2.4-L/s0.88Exp.-75.850.88Exp.-8(h)-Q =-2.4-L/s(i)-Q =-2.4-L/sExp.-51.18Exp.-4(d)-Q =-2.0-L/s0.97(g)-Q =-2.4-L/sExp.-90.88Exp.-10Exp.-10-1Exp.-10-5Exp.-10-4Exp.-10-3Exp.-10-2(j)-Q =-2.6---3.4-L/sFigure 3.1: Experimental results of sediment output plotted against time. Data representsstandardized 15-min average transport rates (g/min-cm). The thicker black line for eachexperiment represents a 180 min centred moving average. For each experiment, the ex-periment number is shown in the top right corner. For Exp. 10, the individual 2 hourexperiments are also identified. The sediment feed rate (horizontal dashed line) is shownfor reference purposes (except for Exp. 9 and Exp. 10 where there was no sediment feed).343.2. Results-10000-8000-6000-4000-2000020000 200 400 600 800 100005000100001500020000250000 500 1000 1500 2000 2500 3000-10000-500005000100001500020000250000 500 1000 1500 2000 2500-10000-50000500015000250000 200 400 600 800 10001200140016001800 2000-10000-500005000100001500020000250000 500 1000 1500 2000 25000100002000030000400000 500 1000 1500 2000 2500020000400006000080 0000 1000 2000 3000 4000 5000 6000Time (min)∑(Qf-Qb)(a) Q = 1.6 L/s Exp. 1 Exp. 2(b) Q = 1.6 L/sExp. 3(c) Q = 1.6 L/s Exp. 4(d) Q = 2.0 L/sExp. 5(e) Q = 2.4 L/sExp. 8(g) Exp. 9(h)Exp. 6(f) Q = 2.0 L/s-10000-500005000100001500020000250000 200 400 600 800 100012001400160018002000 -10000-500005000100001500020000250000 200 400 600 800 1000 1200 1400 1600Exp. 7(g) Q = 2.4 L/s Exp. 8(h) Q = 2.4 L/sExp. 9(i) Q = 2.4 L/s-10000-8000-6000-4000-2000020000 100 200 300 400 500 600 700Exp. 10(j) Q = 2.6 - 3.4 L/s2000010000Figure 3.2: Cumulative sediment flux plotted against time. Data represents the cumulativedifference between sediment input (Qf ) and sediment output (Qb) measured in (g). Foreach experiment, the experiment number is shown in the top right corner.353.2. Resultswhere sediment output rates exceed the sediment feed rate. Similar to Exp. 4, followingthese breakpoint periods average aggradational rates appear to shift, which may indicateminor morphological adjustments enabling the increased transport of sediment.The results from Exp. 6 and 7 are shown in Figures 3.1f and 3.1g. This set of exper-iments was relatively similar to Exp. 4 and Exp. 5. except that the sediment feed ratewas reduced by about 10% compared to Exp. 4 and was held constant for the dischargeincrease in Exp. 7. The result for Exp. 6 was fairly similar to Exp. 4, which exhibitedscour of the un-armoured bed at the start followed by an extended period where sedimentoutput was below the sediment feed rate. However, for Exp. 6 the period following thehigh sediment output rates was followed by an extended period of very low sediment outputrates, average of 0.31 g/min-cm (Figure 3.2f). During this period aggradation in the flumewas concentrated over the areas scoured during the initial 60 minutes of the experiment.For the remainder of the experiment the average sediment output rate was 0.67 g/min/cm.When looking at the cumulative plot for Exp. 6 (Figure 3.2f), it appears similar to the Exp.4 plot, except the breakpoint occurs much earlier (915 minutes). For the duration of theexperiment, after the first 75 minutes, during which sediment output exceeded sedimentinput, the channel aggraded at an average rate of 0.73 g/min. However when the averageaggradation rate is divided into pre and post 915 minutes, the average rates become 1.25g/min and 0.48 g/min, respectively.After the first 15 minutes of the experiment, Exp. 7 experienced a period of aggrada-tion that lasted for an hour and then sediment output rates exceeded feed rates beginningafter 75 minutes (Figure 3.1g). The large spike in the rate of sediment output at 135minutes resulted from a large degradational event which connected two pools on eitherside of the channel. This resulted in a high volume of residual scour, the scoured channelexposed a large area of un-armoured bed sediment. The average sediment output ratefor the remainder of the experiment after 240 minutes was 0.74 g/min-cm. Following thelarge scour event an extended period of relative equilibrium occurred from 195 minutesuntil 615 minutes, during this period slight degradation of the channel occurred at a rateof 0.05 g/min (Figure 3.2g). Over the next 255 minutes a period of rapid aggradationoccurred, with average aggradation rates of 15.65 g/min. Following a brief period of equi-librium, aggradation returned with an average rate of 3.95 g/min for the remainder of theexperiment.The grain size distribution (GSD) of the sediment feed for Exp. 8 was changed from theprevious experiments. Instead of the initial bed mixture, the sediment output from Exp.6 was used to ensure that the feed would be fully mobile (as it was mobile at the lowerflow discharge used in Exp. 6). The result seen in Figure 3.1h shows that equilibrium inthe sediment flux was attained rapidly and maintained for the duration of the experiment.The average sediment output rate for the experiment was 0.91 g/min-cm, slightly greaterthan the sediment feed rate. Inspection of Figure 3.2h also illustrates this, and in factslight degradation is evident during the extent of the run.For Exp. 9 no sediment was fed and the flow rate was kept the same as Exp. 8. The363.2. Resultspeak in sediment output rates occurred during the first 30 minutes and quickly declinedafter that. After 600 minutes, sediment output was less than 1 % of the beginning value(Figure 3.1i). The peaks evident in Figure 3.1i reflect the 5 hour experimental interval,every restart of the experiment produced a pulse of sediment. As the volume of materialoutput from this experiment was so minimal it appears as a relatively large effect. Mostof the sediment moved was related to the entrainment of sediment that settled while flowswere down-ramped from the previous run. In Figure 3.2i the degradational nature of thisrun is clearly evident, the trend fits a logarithmic decay function.Exp. 10 was designed to explore the effects of increased discharge on channel responseand consisted of five 120 minute experiments with no sediment feed. The experiments wereconducted on bed inherited from Exp. 9. Figure 3.1j shows the sediment outputs ratesover time. Sediment output rates increased with each experiment, however, the percentageincrease in output rates declined following Exp. 10- Sediment Mobility - Fixed Bank ExperimentsFollowing Church and Hassan (2002), the relative mobility of individual size classes is pre-sented as the ratio Pi/fi, where Pi is the proportion of the transported load in the ith sizeclass, and fi is the size distribution of the bulk sediment mix. As sediment transport is typ-ically a stochastic phenomenon it was felt that box-plots better display the characteristicsexhibited by the fractional transport ratios. Unity in the plots represents equal mobilityof transport for that size class with respect to its proportion in the bulk sediment mix,or that the transport ratio for that size class is independent of the particle size and thetransport rate of the fraction depends on its proportion in the feed (Church and Hassan,2002; Wilcock and McArdell , 1993).For most of the feed experiments presented below the grain size distribution of the bulksediment mix and the sediment feed are the same. The exceptions are Exp. 8, for whicha modified sediment feed grain size was used, and Exp. 9 and Exp. 10, which had nosediment feed.In Figures 3.3a, 3.3b and 3.3c the smallest three grain sizes (the region greyed out inthe figure) were under-represented in the sediment output due to trapping inefficiency ofthe smallest grains in transport (as discussed above, this continued for Exp. 4 and Exp.5). The region of equal mobility, grains ≥ 0.500 mm and <2.83 mm, is similar for Figures3.3a and 3.3b, even with the reduction in sediment feed. The region of partial mobility,≥ 2.83 mm, is similar for both experiments (Figures 3.3a and 3.3b). The mobility of thethree larger grain size classes within the partial mobility region for Exp. 2 is greater thanExp. 1, especially for grains ≥ 4.00 mm.From Figure 3.3c, it is clear that the range of grains sizes in the region of equal mo-bility is smaller in Exp. 3, grains ≥ 0.500 mm and <2.00 mm, than in the previous twoexperiments. It is interesting to note, however, the classes experiencing equal mobility aremuch more mobile than in the previous experiments. A more marked decline and lower373.2. Results00.511.52P i / f ia) Exp. 1 Qf = 54 g min-100.511.52 b) Exp. 2 Qf = 27.5 g min-1P i / f in = 96n = 400.21 0.42 0.84 1.68 3.36 6.7300.511.52P i / f ic) Exp. 3 Qf = 27.5 g min-1 n = 40Grain size (mm)Figure 3.3: Fractional transport ratio diagrams of mobile size classes from a) Exp. 1, b)Exp. 2, and c) Exp. 3. Box-plots show Pi/fi vs particle size, the size shown in mm isthe arithmetic mean between the retaining and passing sieve sizes (sieves in 1/2 Φ units).The box has lines at the lower quartile, median, and upper quartile of the data. Whiskersextend from each end of the box to the adjacent values in the data. Grey boxes in thefigures indicate partial values due to trap inefficiencies (for Exp. 1 - Exp. 5). Additionally,the number of individual observations in each experiment is included.383.2. Resultsvalues in fractional transport ratios for grains ≥ 2.00 mm is seen on the right hand side of3.3c when compared to either Figure 3.3a or 3.3b.The regions of equal mobility in Figures 3.4a (Exp. 4) and 3.4b (Exp. 5) are verysimilar to those during Exp. 1 and Exp. 2. (see Figure 3.3). The result is surprising giventhe higher discharge used in these experiments. The increased discharge in Exp. 5 onlyresulted in a slight increase in the ratios in the region of partial mobility for grains ≥ 2.00mm and <2.83 mm compared to Exp. 4.For Exp. 6 and Exp. 7 reconfiguration of the sediment trap lead to confidence inthe measured output for the lower grain size classes, thus the greyed out region is nolonger needed in the figures (Figure 3.5). The feed rate (Qf = 30 g min−1) used in theseexperiments was the same for Exp. 6, Exp. 7 and Exp. 8. The same bulk sediment mixturewas used for the feed in Exp. 6 and Exp. 7, and a modified feed mix was employed forExp. 8. No sediment was fed during Exp. 9 and Exp. 10. Discharge in Exp. 6 (2.0 L s−1)was 20 % lower than in Exp. 7 through to Exp. 9 (2.4 L/s inclusive), discharge increasedsteadily at 0.2 L/s increments during the five runs in Exp. 10.With improved trap efficiency, the region of equal mobility extends from grains ≥ 0.177mm to grains <2.00 mm in Figure 3.5a and <4.00 mm in Figure 3.5b. Although the smallersize classes, ≤ 0.354, appear to be partially mobile during a majority of the time. Theresults suggest that for greater than 50% of the measurements, the finer material is beingretained by the bed. It is possible that this material is being trapped behind the coarsergrains that are only partially mobile and are accumulating on the bed surface. The figuresshow the increased mobility of grains under higher flows, but still equal mobility for allgrain sizes was not witnessed. Comparatively, Exp. 4 and Exp. 6 differ negligibly yet thefractional transport ratios are greater for grains ≥ 2.83 mm appears. Sediment feed ratewas 27% lower in Exp. 7 than Exp. 5, yet the mobility of the grains appear to be relativelysimilar.For Exp. 8 the sediment feed was modified compared to the mix used in Exp. 6 andExp. 7, the mobility plots for Exp. 8 are shown in Figure 3.6a. In order to maintainconsistency in the presentation of results between experiments the GSD of the initial feedmixture was used to scale the output from Exp. 8. The figure illustrates that there is aclear under-representation of the smaller size classes in the output, this is partly due todifferences in the GSD of the sediment feed, but it is also due to a reduced proportionof these grains remaining on the surface. The region of equal mobility is similar to thatshown in Figure 3.5b, although it can arguably be extended to include grains <4.00 mm,as this size class is close to unity. The grains sizes in the partial transport regime, those ≥4.00 mm (Figure 3.6a), appear to be more mobile than those in Figure 3.5b. Interestingly,the variability in the observations has reduced between Exp. 7 and Exp. 8, as representedby width of the boxes. The results suggest that using the modified feed can increase themobility grains, although only moderately.Figure 3.7 shows the fractional transport ratios from Exp. 8 using the GSD of the feedinstead of that of the original bulk sediment mix. When viewed from this reference point393.2. Results00.511.52P i / f ia) Exp. 4 Qf = 33 g min-1      Q = 2.0 L s-10.21 0.42 0.84 1.68 3.36 6.7300.511.52b) Exp. 5 Qf = 41.2 g min-1      Q = 2.4 L s-1Grain size (mm)P i / f in = 40n = 40Figure 3.4: Fractional transport ratio diagrams of mobile size classes from a) Exp. 4 andb) Exp. 5. See Figure 3.3 for definitions.403.2. Results00.511.52P i / f ia) Exp. 6  Q = 2.0 L s -100.511.52 b) Exp. 7  Q = 2.4 L s -1P i / f in = 40n = 300.21 0.42 0.84 1.68 3.36 6.73Grain Size (mm)Figure 3.5: Fractional transport ratio diagrams of mobile size classes from a) Exp. 6 andb) Exp. 7. The two experiments have the same feed rate (Qs = 30 g min−1) but dischargewas increased in Exp. 7, see Figure 3.3 for definitions.413.2. Results0. 0.84 1.68 3.36 6.731.51.00.502.0Grain size (mm)P i / f iP i / f ib) Exp. 9  Q = 2.4 L s -1     No Feed n = 15a) Exp. 8  Q = 2.4 L s -1     Modied Feed n = 25Figure 3.6: Fractional transport ratio diagrams of mobile size classes from a) Exp. 8 andb) Exp. 9. Exp. 8 was conducted with the same feed rate as Exp. 6 and Exp. 7 butemployed a modified feed. Exp. 9 used no feed. See Figure 3.3 for definitions.423.2. Resultsit suggests that full mobility occurred for all grain sizes of the sediment feed. Grains <.707mm were overrepresented in the sediment output when referenced to the feed, suggestingthat these sizes were entrained from the bed during the run. However, this effect was alsoinfluenced by increased bed scour. The over representation of grains ≥ 2.83 mm, indicatesthat these grains were being transported in a greater proportion than in the feed, and weretherefore also being entrained from the bed. This may also reflect increased bed scour, butit is also attributable to increased entrainment from the bed surface.Exp. 9 was a no feed, degradational experiment. Figure 3.6b shows that the fractionaltransport ratios are near full mobility for all size classes. For grains <0.707 mm in mostof the samples these size classes are over represented in the sediment output. This is likelythe result of entrainment of grains from the subsurface layer following pool scour eventsduring the run and partial breaching of the armour layer. Grain sizes ≥ 0.707 mm and<1.00 mm are near unity for half of the samples. Grains ≥ 1.00 mm and <2.83 mm areonly partially mobile for most of the time. A region of partial transport is evident forgrains >5.6 mm, although equal mobility occurs greater than 25% of the time. It shouldbe noted that mobility of the larger grains >4.00 mm increased over the duration of theexperiment.The partial transport rates for Exp. 10 are shown in Figure 3.8. It is clear from all of theexperiments that during the first 60 min of each run all size classes were partially to fullymobile. Many of the experiments experienced a higher proportion of the smaller and largersize classes being transported. By the second hour most of the experiments tended towardequal mobility of the grains (Figure 3.8). The exception being seen in Figure 3.8c, whichreflects the formation of a large pool near the downstream end of the channel. The scourassociated with pool development exposed the subsurface material, which has the sameGSD as feed, to entrainment. Given the proximity of the pool to the outlet, all erodedmaterial went directly to the trap. These experiments clearly reflect armour breachingduring the first hour of the experiment, but also reflect a relatively rapid recovery in thearmour layer during the second hour.3.2.3 Sediment Mobility - Mobile Bank ExperimentsIn Figure 3.9 the fractional transport ratios are again plotted as a function of particlesize for the mobile bank experiments. The experiments are divided into stable and un-stable channels (Figure 3.9). The data for the stable channels comprise 22 samples from12 different experiments. The unstable data consist of 9 samples from 7 different exper-iments. It should be noted that stable and unstable channels may have existed duringthe same experimental run. For the stable channel the pathway to stability was mainlyaccomplished through slope adjustments, with negligible surface coarsening (Eaton andChurch, 2004). The unstable channels exhibited accelerated bank erosion, slope reductionand rapid aggradation, that would have eventually led to braiding if the experiments werenot stopped.433.2. Results0.21 0.42 0.84 1.68 3.36 6.7300.511.522.533.54Grain Size (mm)P i / f fExp. 8  Q = 2.4 L s -1 n = 25Figure 3.7: Modified fractional transport ratio diagrams of mobile size classes from fixedbank Exp. 8. This figure slightly modifies the fractional transport ratio (Pi/fi) to (Pi/ff )by employing the feed grain size distribution (ff ) rather than the subsurface as in previousplots. See Figure 3.3 for definitions.443.2. Results0.1 1.0 10012345P i  /  f ia) Exp. 10-1 Q = 2.6 L s-1   0.1 1.0 10012345b) Exp. 10-2 Q = 2.8 L s-10.1 1.0 10012345P i  /  f ic) Exp 10-3  Q = 3.0 L s-10.1 1.0 10012345Grain size (mm)d) Exp 10-4  Q = 3.2 L s-10.1 1.0 10012345Grain size (mm)P i  /  f ie) Exp. 10-5 Q = 3.4 L s-160 min120 minFigure 3.8: Fractional transport ratio diagrams from Exp. 10 with no sediment feed andincreasing discharge: a) Exp. 10-1, b) Exp. 10-2, c) Exp. 10-3, d) Exp. 10-4, e) Exp.10-5. All experiments were run for 120 min. For these experiments the fractional transportratios are presented in terms of Pi/fi. See Figure 3.3 for definitions.453.2. Results00.511.52P i / f ia) Stable0.21 0.42 0.84 1.68 3.36 6.7300.511.52P i / f ib) Unstablen = 22n = 8Grain size (mm)Figure 3.9: Fractional transport ratio diagrams of mobile size classes from the mobile bankexperiments for a) stable and b) unstable periods. See Figure 3.3 for definitions.463.3. DiscussionThree regions of transport can be identified in the plots from the mobile bank exper-iments (Figure 3.9). For both Figures 3.9a and 3.9b size fractions <0.500 mm are underrepresented in the output with respect to the bulk mix. This result is due to trap efficien-cies (discussed above), thus no conclusions can be reached about the behaviour of thesesize classes, although it is assumed that they would plot near unity.Grain sizes ≥ 0.500 mm and <2.00 mm in the stable samples are equally mobile for amajority of the observations (Figure 3.9a). For the unstable periods grains sizes ≥ 0.500mm and <5.60 mm exhibit equal mobility, median ratio values were either equal to orgreater than unity (Figure 3.9b). A value of unity indicates equal mobility of these grainsizes or that these sizes are transported at proportions similar to their presence in thesediment mix. The median values for grain sizes between 0.500 mm and 2.00 mm appearto be higher in the stable experiments compared to the unstable experiments, indicating arelatively higher mobility for these size classes during stable periods.The right side of Figure 3.9a shows a steady decline in the fractional transport ratioof larger size classes, ≥ 2.00 mm, following the sizes that exhibit equal mobility. For thelargest size class, >5.6 mm, for most experiments no grains were observed at the outlet(Figure 3.9a). It should be noted that this does not negate the potential for movementwithin the channel, just that it was not recorded at the outlet.This steady decline in the transport ratios indicates the decreasing mobility of grainswith reference to their presence in the bed, a condition referred to as partial transport. InFigure 3.9b, however, the region of equal mobility is much greater than observed duringstable periods, and extends up to grain sizes <5.6 mm. A sharp drop in the mobility forgrains ≥ 5.6 mm is seen, however, in comparison to the stable channels its mobility in theunstable channels is much greater.3.3 DiscussionIn general, the experiments highlight the usefulness of prototype scaling as it can giveinsight into the frequency and magnitude of channel response and time-scales of channelevolution. The combination of a field prototype and Froude scaled physical models allowsone with a discharge record to use emergent channel features and be able to contextualizeresults found in the lab to those in the field. The scaling of time and sediment transport hasproved difficult to numerically model due to gaps in our understanding of how to accountfor evolving beds. Much historic information has been gleaned from previous field basedstudies, physical models offer another geomorphic tool that can be used alongside fieldbased studies to develop improved understanding of geomorphic processes (for example,Yalin, 1971; Davies and Lee, 1988; Warburton and Davies, 1994; Hassan and Church, 2000;Eaton and Church, 2004; Gran et al., 2006; Madej et al., 2009; Pryor et al., 2011). The useof realistic physical models is standard practise outside of academia to inform hydraulicdesign and to understand potential impacts that an proposed or existing infrastructure473.3. Discussionmay produce. Their use may shed more insight into river behaviour than the standardflume set-up, upon which much of our understanding of river processes has been built.The experiments demonstrated that slight size selective transport can occur at a rangeof transport rates and discharges, and that it is governed by the particle size of the feed,and not the feed rate. The decline in Pi/fi ratios for the larger grain size classes hasbeen observed in both field (Powell et al., 2001; Church and Hassan, 2002; Thompson andCroke, 2008) and flume (Wilcock and Southard , 1989; Wilcock and McArdell , 1993, 1997;Hassan and Church, 2000; Hassan et al., 2006) experiments. The mobility of the larger sizeclasses present on the bed surface is an important determinant of channel stability. Exp. 10illustrated that once the coarser size classes approached equal mobility the channel becameunstable. The results from the mobile bank experiments also illustrated that unstablechannels are represented by more mobile coarse fractions than were stable channels. Thissuggests that determining at what discharge the D90s or D95s of the bed surface becomesmobile can determine at what flow magnitude the channel may potentially become unstable.An additional factor in the stability of the fixed bank experiments is the morphology ofthe channel. A link between the stability of bars, for example, is likely tied to the stabilityof the larger grains on the bed. An additive effect is suggested to exist between bed surfacearmouring and channel bedforms, which has a net positive increase on the stability of bothand likely the channel overall.The degree of size selective transport was found to vary with the applied discharge andwith the grain size distribution of the sediment feed. The experiments demonstrated thatbeds aggraded at a slow rate due to a persistent (but slight) selective deposition/transportof bed material. The rate of the aggradation would typically go unnoticed in most field en-vironments. The timescales at which the rates of aggradation observed in these experimentwould occur in the field is nearly 100 years, and is on the same scale as the disturbanceregime. In cases where the channel is not laterally constrained, slow aggradation ratescoupled with surface coarsening would induce lateral channel movement, as was evident inthe New Zealand experiments. The experiments also demonstrated that the channel doesnot aggrade when the sediment was slightly finer, Exp. 8.Church and Hassan (2005) wondered whether in channels that exhibit partial and se-lective transport if the coarse grain sizes would continue to accumulate. These experimentssuggests that they do, and that this accumulation results in slow, persistent channel aggra-dation. Similar persistent aggradation has been observed in the experiments of Pryor et al.(2011), Recking et al. (2009) and Braudrick et al. (2009), although they avoided long-termaggradation by reducing the proportion of coarse feed. Although, this rate of aggradationhas yet to be observed in the field, the rates of aggradation observed here suggest that itwould be difficult to identify with annual surveys as it is within the measurement errorof most channel surveys. Church and Hassan (2005), also suggested two mechanisms bywhich this phenomenon could be avoided in natural channels:1. large floods that are of sufficient magnitude to mobilize the large material and there-483.3. Discussionfore establish a long term equilibrium between sediment supply and output. Thissuggests that equal mobility is a long term phenomenon, and would not be observedat shorter, more dynamic, time scales.2. abrasion that reduces the size of larger clasts and enables the transport of this ma-terial out of the reach.In these experiments, even under flow conditions with magnitudes greater than the 10-yr flood, aggradation was still observed. This suggests that in these types of channels, flowswith magnitudes above the 10-yr flood event are required to mobilize the larger clasts inthe channel or that weathering in situ is required to maintain conservation of mass. Theseresults also may explain the discrepancy of some of the channels studied by Lisle (1995),that appeared to deviate from equal mobility. The bedload data was not sampled from aflow large enough, that equal mobility would be expected to occur.Church and Hassan (2005) attributed the attrition of grains to weathering processesduring storage within the channel (for example, Jones and Humphrey , 1997) as opposedto abrasion during transport (for example, Sklar et al., 2006). An additional mechanismof weathering is suggested here. This mechanism is related to the lateral activity of thechannel. Disturbance of the riparian zone, such as from forest fires in the case of FishtrapCreek, can result in temporary periods of increased lateral activity. A direct result oflateral activity can be channel avulsions, which results in the abandonment of the formerchannel into a new channel that is cut into the floodplain. This process can result in twooutcomes:1. the sediment in the abandoned channel is exposed to increased weathering activity,physical and biological, and particularly freeze-thaw weathering.2. the sediment in the new floodplain channel has been sufficiently weathered during itssequestration to the floodplain environment, that grain sizes have decreased enoughto now be transported by lower magnitude flow events.Over time it is anticipated that the former floodplain sediment will be transporteddownstream and replaced by sediment from upstream sources as the channel moves aboutits floodplain, and the cycle will repeat. Even moderate channel erosion into the floodplainmaterial can lead to a decreased sediment size. As Exp. 8 has shown, even a moderateincrease in the proportion of the more mobile fractions in the feed, can increase the mobilityof the larger classes.The aggradation observed here produced a coarse surface, which suppressed pool-riffledevelopment and led to a plane-bed like morphology. This pathway to plane-bed channelsdiffers from that proposed by Montgomery and Buffington (1997), as plane-bed channelsare degradational in nature. As observed during these experiments, and by the authorin various field settings, plane-bed channels can develop in forested streams even during493.4. Conclusionsperiods of net aggradation. Plane-bed morphologies can also temporarily exist in reachesthe have received landslide deposits that bury the existing morphology in material notreadily erodible by normal flows. This morphology may also be characteristic of channelsthat are unable to erode their banks (due to, for example, riparian forests) which wouldprevent the exchange of sediment between the channel and the floodplain. For channelswith erodible banks, the lower mobility of the coarser size fractions has a less rapid effect onthe bed surface texture since sediment is constantly being exchanged between the channeland floodplain, and may be counter-balanced by weathering-related changes in the sedimentsize distribution that can occur in the interval between deposition and re-entrainment. Inthese cases, the channel does not aggrade due to coarse sediment deposition. Furthermore,when the largest grains do become fully mobile, the bank erosion rate accelerates, channelsinuosity increases, thereby decreasing the reach-average energy gradient and ultimately thebed material transport capacity. This feedback intensifies the morphodynamic instabilityand ultimately is likely to result in a change in channel morphology from a single-threadchannel to a braided one.3.4 ConclusionsThe experiments were used to investigate sediment mobility during various sediment feedand flow regimes. In general, the largest size fractions were under-represented in transport.This under representation led to channel aggradation. However, when sediment fed intothe system was equivalent to sediment that was output from the system, no aggradationwas observed. In fact, the introduction of this feed actually increased the mobility of someof the larger previously less mobile size classes.Aggradation due to partial mobility of selective transport can potentially be avoidedin channels by increased lateral activity. Lateral activity can result in the re-activation ofweathered channel sediment in the floodplain in exchange for the coarser sediment respon-sible for the aggradation. The observed aggradation produced plane-bed like morphologies,which creates another pathway for the creation of this channel type.Full mobility only appears to be characteristic of unstable/degrading systems in gravel-bed rivers. Partial mobility in these experiments led to aggradation, not a capacity lim-itation but competence limitation in the ability to move the largest grain sizes. Theseexperiments indicate that stability in mountain channels may hinge on the mobility of thelargest grain size class less than the D90 in our experiments. Mobility of this size class maybe an indication of current or future channel instability.50Chapter 4Spatial and Temporal Patterns inSediment Transport and Storage4.1 IntroductionA proper accounting of sediment transport in gravel-bed rivers is important for fluvial geo-morphology, channel assessments, river engineering, sediment budgeting, landscape evolu-tion models, fisheries habitat assessments, and river rehabilitation projects. The samplingof sediment transport data is both financially and temporally cumbersome, and is usu-ally only feasible under selected circumstances (for example, Wilcock , 2001a). Thus mostestimates of sediment transport are made using a variety of transport equations. Theseequations generally contain a combination of theoretical and empirical components, thatare mostly based on averages of local hydraulic and sedimentological variables. Whenapplied to gravel-bed rivers these equations tend to either under- or over-estimate the ob-served values by an order of magnitude or more (Carson and Griffiths, 1987; Gomez andChurch, 1989; Lenzi et al., 1999; Bravo-Espinosa et al., 2003; Barry et al., 2004; Pitlicket al., 2008; Recking , 2010; Nitsche et al., 2011).Measurements of sediment transport rates in field and laboratory environments havebeen observed to fluctuate both spatially and temporally even under steady state condi-tions (Hoey , 1992; Kuhnle, 1996). The causative mechanisms behind this variability havebeen attributed to flow regime (Nordin, 1985; Lisle, 1989; Nash, 1994), sediment regime(Dietrich et al., 1989; Hoey and Sutherland , 1991; Lisle et al., 1993; Gintz et al., 1996;Benda and Dunne, 1997b; Gomi and Sidle, 2003; Venditti et al., 2010), the passage of bedforms (Reid et al., 1985; Iseya and Ikeda, 1987; Lisle et al., 2000; Recking et al., 2009;Nelson et al., 2009), and the evolution of bed topography and bed structures (Laronne andCarson, 1976; Komar , 1987; Kuhnle and Southard , 1988; Church et al., 1998; Papanicolaouand Schuyler , 2003; Nelson et al., 2010). In essence, the variability in transport rates hasbeen linked to the variability in sediment supply and storage.To overcome the limitations of the current transport theory, Lisle and Church (2002)proposed to shift the focus away from hydraulic related parameters and to focus on sed-iment supply and storage instead. In order to achieve this, they focused on transportcapacity, which they see as being a mediator between sediment supply, transport and stor-age. Lisle and Church (2002) proposed that transport capacity is not a fixed quantity, asis frequently assumed, but varies according to the volume of sediment stored within each514.1. Introductionsediment reservoir and the bed state. They defined a sediment reservoir as homogeneousreach of the channel and its associated floodplain that has the potential to store sedimentand can be accessed by the channel under the current hydroclimatic regime. Using datafrom field and flume studies they asserted that the transport capacity of each sedimentreservoir was a unique positive function of the volume of sediment stored within that reser-voir; storage volume influences sediment mobility and availability though surface texture,channel gradient and availability of sediment stored in the floodplain. They proposed twophases in the transport-storage relation for degrading experimental channels: Phase I,comparable to transport-limited conditions, is associated with conditions of high sedimentsupply and transport rates, non-selective transport and weak armouring; Phase II, compa-rable to supply-limited conditions, is characterized by selective transport, bed armouringand transport rates that decrease as sediment storage decreases. Lisle and Church (2002)recognized that their understanding of the transport-storage was limited as only degradingsystems were investigated.Flume experiments by Pryor et al. (2011) were conducted in order to investigate thetransport-storage relation over periods of both aggradation and degradation. They re-ported that full aggradation-degradation cycles exhibited counter-clockwise hysteresis or acyclic pattern in the transport-storage relation, and thus a unique relation between sedi-ment transport and sediment storage was not observed. Similar observations were made byMadej et al. (2009) in their flume experiments and by Hassan et al. (2007) using long-termfield data. Together these newer results concluded that the hysteresis occurred during dis-equilibrium conditions and that the cycles are a result of an external perturbation, suchas a change in sediment supply or flow characteristics. However, this hysteretic effect mayin fact be an inherent response in sediment transport rates to the evolution of the bedsurface and within channel morphological adjustments to changes in sediment supply. It ispossible that a non-linear relationship between transport rates and sediment storage maybe more the norm than not.In light of these recent results, Lisle (2012) re-visited the earlier model and modifiedit to include two general scenarios: Scenario 1 - a state of dynamic equilibrium betweentransport capacity and sediment storage, a common relation at a given stage of sedimentstorage in the channel is the same whether the channel is aggrading or degrading; Scenario2 - a state of transient equilibrium, where increased transport rates are experienced duringdegrading conditions and depressed transport rates occur during aggrading conditions fora given sediment stage. Lisle (2012) concluded that the conditions represented by Scenario2 were caused by short-term variations in supply. The results of experiments conducted byPryor et al. (2011) and Madej et al. (2009) support Scenario 2.All of the experiments under which the transport-storage relations have been developedand/or tested have been conducted using variable sediment input rates to induce channeldegradation of short-term periods of either aggradation or degradation. It may be arguedthat the variability of sediment input may artificially inhibit Scenario 1 from occurring.Most of the previous experiments cited above, only observed Scenario 1 during degrada-524.2. Resultstional phases of the experiment. The experiments presented here are a further attempt toelucidate the relation between transport capacity and sediment storage. As with the pre-vious experiments, intermediate size forested gravel-bed rivers were the targeted prototypefor experimental design.Intermediate size forested gravel-bed rivers are subject to variable flows and sedimentloads, yet they remain relatively stable, at least in the lateral sense. For the most part,variability in the governing conditions is accommodated within the channel boundaries.Only major disturbances, those that exceed the capacity of the channel to adjust, lead tochannel relocation or widening. Consequently, the banks of the channels remain relativelystatic or are stable around some mean value over time. Channel planform in these en-vironments appears to be dictated more by boundary conditions, such as riparian forestsand non-alluvial boundaries, than by alluvial processes. Most experiments to date havebeen conducted in straight relatively narrow flumes with no consideration to the channelplanform. The experimental channel used here, was designed based on the planform ge-ometry of a mountain gravel-bed river to explore the effects of irregular channel walls andto generate results comparable to the field.The objectives of this chapter are to:1. explore spatial and temporal patterns of sediment transport;2. investigate the transport-storage relation under steady state forcing conditions; and,3. investigate the response of the channel in terms of sediment storage in changes insediment supply and flow conditions.4.2 ResultsResults are presented for the following topics: sediment transport rates, sediment transport-storage relations, patterns of channel adjustments, and the sediment texture of bed surfaceand transported material. In order to explore each topic area, the experiments are orga-nized according to themes. The first theme explores the effects of the initial conditions ofthe channel, with initial conditions beginning on a screeded bed and discharge and feedrates being similar (Exp. 4 versus Exp. 6) or different initial conditions with similar dis-charge and feed rate (Exp. 2 versus Exp. 3). Secondly, the effects of stream dischargewith constant sediment feed rate and texture (Exp. 6 versus Exp. 7) is explored. For thesecond theme, the sediment feed rates are varied while the stream discharge and sedimentfeed texture are held constant (Exp. 1 versus Exp. 2 and Exp. 3, Exp. 4 versus Exp. 6,and Exp. 5 versus Exp. 7). Lastly, the effects of a change in the sediment feed texturewhile holding flow discharge and sediment feed rate constant is explored (Exp. 8 versusExp. 7).Summary data of the experimental conditions and results are presented in Table 4.1.534.2.ResultsTable 4.1: Summary data of the experimental conditions and results. For some of the values that are averages the standarddeviation has been included in brackets.Exp.a Time Discharge Sediment Averageb Averagec Averaged Averagee D50s Cycle Maximum Length Number(min) (L/s) Feed Rate Velocity Slope Depth τb (mm) Number Storage of of(g/s) (m/s) (m/m) (cm) (Pa) Difference Cycle Cycles(g) (min)1 5700 1.6 0.91 0.32 0.0153 1.47 (0.08) 2.21 (0.15) 2.58 (0.35) 1 648.9 105 12 225.3 60 13 886.1 285 34 733.6 195 15 4336.7 1680 116 666.0 195 17 750.1 165 12 2400 1.6 0.45 0.31 0.0164 1.50 (0.08) 2.44 (0.08) 3.03 (0.19) 1 453.7 135 22 1044.5 510 53 2226.0 840 33 2400 1.6 0.45 0.30 0.0152 1.58 (0.05) 2.31 (0.12) 2.37 (0.31) No cycles - - -4 2400 2.0 0.55 0.37 0.0152 1.60 (0.07) 2.35 (0.13) 2.89 (0.20) 1 677.2 225 22 1486.1 270 15 1800 2.4 0.67 0.38 0.0158 1.85 (0.03) 2.90 (0.09) 3.29 (0.14) 1 1541.6 195 12 762.2 240 23 648.8 225 14 897.2 105 16 2400 2.0 0.50 0.35 0.0163 1.72 (0.05) 2.71 (0.27) 2.64 (0.22) 1 426.5 135 17 1800 2.4 0.50 0.39 0.0161 1.81 (0.04) 2.86 (0.17) 3.13 (0.31) 1 623.8 105 12 6616.7 1470 13 436.1 375 54 846.1 315 25 288.4 60 16 177.7 105 18 1500 2.4 0.50 0.40 0.0168 1.79 (0.02) 2.92 (0.07) 3.22 (0.13) 1 794.0 1125 142 67.2 105 39 900 2.4 0 0.37 0.0162 1.99 (0.10) 3.24 (0.26) 3.12 (0.55) No cycles - - -a Experiments with bold numbers indicate experiment started on hand screeded bed, and non bolded numbers identify experiments conducted on bedinherited from preceding experiment.b Average velocity determined using the mean of hourly velocity averages, standard deviation is included.c Following Eaton and Church (2009), average slope determined using a linear regression of the mean bed elevation at each cross section and the downvalley distance of each cross section.d Average depths were determined using continuity.e Average shear stress (τb) was calculated using τb = ρgdS544.2. Results4.2.1 Sediment Transport RatesFor all of the experiments in which the sediment feed grain size distribution was the same asthe bed material grain size distribution, the sediment output was consistently lower than thesediment input, regardless of the actual supply rate or the imposed stream discharge (Figure4.1). This effect has been quantified by calculating a bed material transport efficiency term,Teff , for which:Teff =QbQf(4.1)The value of Teff varies significantly between experiments with the same inputs andinitial conditions. For example, Exp. 4 and Exp. 6 had identical flows and nearly identicalsediment feeds, and both began from the same featureless initial bed state. However, due todifferences in the channel morphology, evident in the different reach-average flow velocities,the transport efficiencies were about 20% higher for Exp. 4 (see Table 4.2). Differences inchannel response during the initial phase of each experiment were also observed, Exp. 6experienced more scour than Exp. 4 which lead to a much longer period where transportrates were much lower than the feed rates (Figures 4.1d and 4.1f), up to 900 minutes forExp. 6. Histograms of Teff , calculated for each 15 minute sampling period (Figure 4.2c),show that values of Teff fluctuated greatly during both experiments, with most observa-tions ranging 0.50 to 1.1. The histograms show data from the experiments starting at 120min. in order to remove the high transport rates at the onset of the experiments. A two-sample Kolmogorov-Smirnov (K-S) goodness-of-fit test (p = 0.05) was used to determineif the transport efficiencies from the two experiments were drawn from the same under-lying continuous distribution. The result of the test was that these two distributions aresignificantly different from each other. It is interesting to note, that when comparing Teffvalues after 900 minutes the distributions were no longer statistically different.Table 4.2: Transport efficiency for experiments with similar stream discharge and sedimentfeed.Experiment Q Qf Qb/Qf 95% Confidence Range4 2.0 0.55 0.74 0.69 – 0.796 2.0 0.50 0.62 0.57 – 0.672 1.6 0.45 0.89 0.83 – 0.953 1.6 0.45 0.46 0.42 – 0.49Similarly, Exp. 2 and 3 had identical flows and sediment feed rates, the difference wasthe initial conditions at the beginning of the experiment. Exp. 2 was initiated using thefinal bed morphology created during Exp. 1, while Exp. 3 was initiated on a featureless554.2. Results0120 400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 6000012340 200 400 600 800 1000 1200 1400 1600 1800012340 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 24000120 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400012340 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500012340 200 400 600 800 1000 1200 1400 1600 1800012340 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 24000120 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 24001 3 4 21  Time (min)Transport Rates (g/s)a. Exp. 1b. Exp. 2c. Exp. 3d. Exp. 4e. Exp. 5f. Exp. 6g. Exp. 7h. Exp. 8i. Exp. 9012340 100 200 300 400 500 600 700 800 900Q = 1.6 L/sQ = 1.6 L/sQ = 1.6 L/sQ = 2.0 L/sQ = 2.4 L/sQ = 2.0 L/sQ = 2.4 L/sQ = 2.4 L/sQ = 2.4 L/s25 611 2 31 21 2 31 2 3 4 5 6 7Figure 4.1: Variations in sediment transport rates with respect to time. The sediment feedrates are indicated by the dotted lines. For most of the experiments there are additionallines and numbers indicating the occurrence, duration, and number of cycles in the sedimenttransport-storage relations (refer to Table 4.1 for more details).564.2. Results0510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.10510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.10510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1a) Variable Q - Exp. 6 (2.0 L/s) Exp. 7 (2.4 L/s) (Qf = 0.015 g/cm/s)b) Variable Qf - Exp. 1 (0.026 g/cm/s) , Exp. 2, Exp. 3 (0.013 g/cm/s) (Q = 1.6 L/s)0510152025300.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20510152025300.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2c) Variable Qf - Exp. 4 (0.016 g/cm/s) and Exp. 6 (0.015 g/cm/s) (Q = 2.0 L/s)d) Variable Qf - Exp. 5 (0.016 g/cm/s) and Exp. 7 (0.015 g/cm/s) (Q = 2.4 L/s)e) Variable GSD - Exp. 7 and Exp. 8 (Qf = 0.015 g/cm/s) (Q = 2.4 L/s)Qb/QfFrequency0510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 0510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.10510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1Exp. 1: 0-2400 0510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.10510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 0510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.10510152025300.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1Exp. 6: 900-2400 Exp. 7: 900-1800Exp. 2: 0-2400 Exp. 1: 3300-5700Exp. 3: 0-2400Exp. 4: 120-2400 Exp. 6: 120-2400Exp. 5: 195-1800 Exp. 7: 195-1800Exp. 7: 165-1800 Exp. 8: 165-1800Figure 4.2: Histograms of sediment transport rates (Qb) normalized by the sediment feedrate (Qf ). For each theme histograms are shown for data truncated to exclude the initialstart-up spikes in transport rates. Q represents stream discharge. Solid vertical linesrepresents Qb/Qf = 1 for reference. 574.2. Resultsinitial bed. This resulted in much lower Teff values for Exp. 3. Only on a few occasionsdid sediment transport rates (Qb) exceed feed rates (Qf ) during Exp. 3, this occurrencewas more frequent during Exp. 2 (Figure 4.1b and 4.1c). The histograms of Teff for Exp.2 and 3 are both visually and statistically different (based on a K-S test): while Teff variesbetween 0.8 and 1.2 during most of Exp. 2, it almost never exceeds 0.8 during Exp. 3(Figure 4.2b). The cause of this significant discrepancy is attributed to differences in thehistorical contingency of the two experiments. Exp. 2 was conducted on the bed developedduring Exp.1, except Qf was two times greater. Compared with the initial conditions ofExp. 3 (a flat screeded bed), Exp. 2 began on a bed with a higher slope, thus higherstream power, and well developed morphology, in which sediment transport pathways andhydraulic efficiencies had already been established. These two initial conditions resultedin only minor morphological and textural adjustments during Exp. 2 to accommodatethe reduction in Qf . Further evidence of differences in transport efficiency due to initialconditions, is seen by comparing the histograms of Exp. 1 which illustrate how differentTeff is when comparing the two temporal periods plotted in Figure 4.2b.When experiments with similar discharges (but different sediment feed rates) are com-pared, no clear effect of sediment feed rate on the transport efficiency emerges (see Table4.3). Two experiments were conducted at the equivalent of bankfull flow: in Exp. 1(Qf = 0.91 g/s), the system was able to transport 0.76 g/s with a Teff = 0.83; however,when Qf was reduced nearly 50% for Exp. 2, Teff increased to 0.89, the increase was notstatistically significant. When only the data between 3300 and 5700 minutes for Exp. 1are examined the Teff becomes 0.93, suggesting that feed difference and resultant mor-phological changes resulted in a slightly improved Teff during the later part of Exp. 1, asdiscussed above. Furthermore, the distributions of 15 minute estimates of Teff for bothexperiments are not statistically different (see Figure 4.2b, Exp. 1 from 3300 to 5700 min).Exp. 1 and Exp. 2 had different initial conditions, thus the conclusions reached in thiscomparison cannot be entirely attributed to sediment feed rates alone. However, Exp. 1and Exp. 3 had similar initial conditions and discharges, and differed only in the volumet-ric rate of sediment feed. In Exp. 3 (Qf = 0.45 g/s), the system was able to transport0.21 g/s with a Teff = 0.46; thus, a 50% reduction in the sediment feed rate resulted innearly a 55% reduction in transport efficiency, the decrease was statistically significant.The distributions of 15 minute estimates of Teff for both experiments are not statisticallydifferent (see Figure 4.2b, Exp. 1 from 0 to 2400 min).For flows near the 10-yr recurrence interval (i.e. Exp. 5 and 7 in Table 4.3), a similarpattern results: a decline in feed rate produced an increase in Teff that is not statisticallysignificant. As for the first pair, the histograms of Teff estimates for Exp. 5 and 7 (Figure4.2d) are visually similar (and a K-S test shows that they are not statistically different).The start-up conditions between Exp. 5 and Exp. 7 also appear to be a little different, theinitial scour period during Exp. 7 was delayed by 100 minutes when compared to Exp. 5(Figure 4.1e and 4.1g), possibly due to differences in armouring at the end of Exp. 4 andExp. 6. At the end of Exp. 4, 17 600 g had accumulated in the channel, compared to 22584.2. Results650 g for Exp. 6, which resulted in a higher slope at the end of Exp. 6 (0.0178) comparedto Exp. 4 (0.0158).Table 4.3: Transport efficiency for experiments with similar stream dischargeExperiment Q Qf Qb/Qf 95% Confidence Range1 1.6 0.91 0.84 0.81 – 0.872 1.6 0.45 0.89 0.83 – 0.953 1.6 0.45 0.46 0.42 – 0.495 2.4 0.67 0.78 0.72 – 0.847 2.4 0.50 0.88 0.81 – 0.94Only one set of experiments had similar sediment feed rates but different discharges,the average Teff shows that higher values are associated with higher stream discharge(Table 4.4). The histograms of Teff seem to show that Teff becomes more variable as Qincreases, since the spread of the histograms increases and the relative size of the peaksdecreases (Figure 4.2a); but there are not enough replicate experiments to decide thematter. Additionally, the experiments had different initial conditions.Table 4.4: Transport efficiency for experiments with similar sediment feedExperiment Q Qf Qb/Qf 95% Confidence Range6 2.0 0.50 0.62 0.57 – 0.677 2.4 0.50 0.88 0.81 – 0.94Table 4.5: Transport efficiency for experiments with similar stream discharge and sedimentfeed rates, but differing grain size distribution of the feed.Experiment Q Qf Qb/Qf 95% Confidence Range7 2.4 0.50 0.88 0.81 – 0.948 2.4 0.50 1.03 0.97 – 1.09One set of experiments had similar discharge and sediment feed rates but different grainsize distribution of the feed, the average Teff shows that higher values are associated withfeed of Exp. 8 (Table 4.4). The histograms of Teff seem to show that distribution of Teffbecomes more narrow in Exp. 8, since the spread of the histograms decreases in Exp. 8594.2. Resultsand is wider in Exp. 7 (Figure 4.2e); not enough replicate experiments were conducted toextrapolate to a general case.4.2.2 Sediment Transport-Storage RelationsTo further explore the interactions among flow, sediment feed, and channel morphology weanalyzed sediment transport-storage relations. Sediment transport-storage relations weredeveloped for our experiments using transport rates and cumulative storage volume; thesediment feed rate is also provided for each experiment in Figure 4.3. Changes in storagevolume were calculated from differences between the sediment feed rate and sedimentoutput.Hysteresis loops in the transport-storage relations were observed in a number of ex-periments (Figure 4.3). These loops or cycles, appear to be the result of periods wheretransport rates out of the reach are below the rates into the reach, which then leads toan increase in the volume of sediment stored in the channel. This aggradational period isfollowed by a period of higher transport rates and bed degradation and a return to theprevious lower transport regime completes the loop. Following the terminology of Smith(2004), Pryor et al. (2011) and Hassan et al. (2007) the hysteresis loops will be referredto as aggradation-degradation cycles. Hassan et al. (2007) further differentiated betweenmajor and minor cycles of aggradation and degradation using a volumetric threshold. Inthese experiments major cycles were cycles that lasted longer than 1000 minutes or had amaximum storage difference within the cycle ≥ 1000 g. In addition to a volumetric thresh-old, a temporal threshold was added to identify a cycle that had was clearly important,however sediment availability may have limited large sediment movement and thus thevolumetric threshold was not met. Throughout the experiments, many cycles containedmultiple smaller cycles within them, making it difficult to identify where one cycle beganand another ended. For reporting purposes the values presented in Table 4.1 representtotal values of the entire cycle (which may or may not have been classified as being major),if multiple cycles occurred and were able to be differentiated the total number of cycleswas also noted.The transport-storage relations of Exp. 6 and 7 (Figure 4.3f and 4.3g respectively),the experiments had the same feed rate, illustrate the effects of varying discharge on therelation. A comparison of the two figures reveals a clear difference between the numberand scale of the transport-storage cycles. High transport rates at the beginning of Exp.6 reflect channel degradation; a partial loop occurred in the relation at the start of Exp.6. The initial degradation at the start of the experiment was followed by an aggradationalperiod, which is reflected by the relatively constant transport rate (Figure 4.3g); this issimilar to Phase 1 of Lisle and Church (2002). The transport-storage relation appears tobecome less stable as the transport rate approached the feed rate, starting at around 12 000g of storage. The variability in the transport-storage relation persisted for the remainderof the experiment. The only complete aggradation-degradation cycle in Exp. 6 occurred604.2. Results00.511.522.53-6000 -4000 -2000 0 2000 4000 6000 800000.511.522.5-3000 0 3000 6000 9000 12000 15000 1800000.511.522.50 1000 2000 3000 4000 5000 6000 7000 80000. 3000 6000 9000 12000 15000 18000 21000 24000 27000 30000 33000 36000 390000. 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000 550001 231 2 3 54600.511.522.5-5000 0 5000 10000 15000 20000 250001 1 00.511.522.533.5-7000 -6000 -5000 -4000 -3000 -2000 -1000 0 1000 20001 2 3 4 5 6 Cycle 4 -2000 -1500 -1000 -500 0 5001 2 -3000 -2500 -2000 -1500 -1000 -500 0Transport Rate (g/s)Sediment Storage (g)a. Exp 1b. Exp 2c. Exp 3d. Exp 4e. Exp 5f. Exp 6g. Exp 7h. Exp 8i. Exp 900. -2900 -2700 -2500 -2300 -2100 -19007Q = 1.6 L/sQ = 1.6 L/sQ = 1.6 L/sQ = 2.0 L/sQ = 2.4 L/sQ = 2.0 L/sQ = 2.4 L/sQ = 2.4 L/sQ = 2.4 L/sDegradationAggradationDegradationAggradationDegradationAggradationDegradationAggradationDegradationAggradationDegradationAggradationDegradationAggradationDegradationAggradationDegradation2 3 1 2 Figure 4.3: Transport-storage relations between transport rate and sediment storage. Feedrates are indicated by the dotted lines. Arrows show the temporal direction of the relations.The numbers refer to cycle number during a given experiment, refer to Table 4.1 for moreinformation. A more detailed look at cycle 4 of Exp. 7 is provided as an inset.614.2. Resultsaround 16 000 g (Cycle 1, Figure 4.3f).Contrary to Exp. 6, multiple aggradation-degradation cycles are evident in the transport-storage relation of Exp. 7 (Figure 4.3g). Six cycles in total were identified, one major andfive minor. Cycle 2, the largest, lasted 1 470 minutes and represented a maximum storagechange of 6615 g during the cycle. The remaining five minor cycles all occurred within thefirst cycle and ranged in duration from 60 to 375 minutes (Table 4.1). The main differencebetween Exp. 6 and Exp.7 is the net transport through the channel, Exp. 7 had a higherTeff , which is likely the result of the higher discharge and the historical contingency. Bythis we mean that Exp. 7 needed to make only minor modifications to the pre-existingmorphology in order to attain equilibrium in sediment transport rates and had begun onan initial channel slope which was greater than the initial slope of Exp. 6.Figures 4.3a, 4.3b, and 4.3c show the transport-storage relations for Exp. 1, 2, and3. In Exp. 1, seven separate cycles were identified; of these only one cycle (Cycle 5) wasclassified as being major (Figure 4.3a). Cycle 5, which consisted of eleven minor cycles,lasted for 1680 minutes and had a maximum storage difference of 4337 g, the seven othercycles lasted between 60 minutes and 285 minutes and displaced 225 to 886 g of sediment(Table 4.1). The interesting feature of Figure 4.3a is the long period of aggradation(1440 minutes) during which no cycle occurred. For Exp. 2, three cycles were identified(Figure 4.3b). Two of these cycles were classified as being major because maximum storagedifferences during each cycle was greater than 1000 g. The differences were 1045 g and2226 g for Cycle 2 and 3, respectively. Exp. 3 exhibited no cycles at all (Figure 4.3c).This suggests that there must have been a transport capacity limitation. Comparing Exp.3 to Exp. 1 one can postulate that it was not until the channel stored at least 35 000g of material that the transport rate equalled the feed rate. Channel aggradation led toincreased channel slope and filled the available sediment reservoirs, which in turn increasedtransport capacity. This would also explain the result in Exp. 2 as a greater slope andchannel storage already existed.The transport-storage relations for Exp. 4 and 6 are shown in Figure 4.3d and 4.3f, thetwo experiments had the same flows and relatively similar feed rates. The transport-storagerelation for Exp. 4 mainly reflects the aggradational nature of this experiment; transportrates remained relatively constant or flat as channel storage increased. Only two cycles canbe observed in Figure 4.3d, Cycle 2 is a major cycle as it was the result of a nearly 1500 gstorage difference. Exp. 6 (Figure 4.3f) was mainly aggradational. The relations for bothexperiments look very similar and the first cycle occurred at a relatively similar time andvolume of sediment storage (15 000 g).Figure 4.3i, 4.3g and 4.3e show the transport-storage relations for Exp. 9, 7, and 5. Theno-feed experiment (Exp. 9) shows an exponential relationship between transport rate andvolume of sediment storage with declining transport rates associated with the depletion ofstored sediment available for transport due to armouring and winnowing, similar to PhaseII of Lisle and Church (2002) (Figure 4.3i). As discussed above, the relationship betweenthe transport rate and storage for Exp. 7 is complex, with a constant transport rate624.2. Resultsassociated with aggradational periods, and hysteresis during periods of quasi equilibrium(Figure 4.3g). Exp. 5 shows a similar relationship between transport and storage as Exp.7, except that periods of equilibrium occurred less frequently and for shorter duration.Exp. 5 exhibited three aggradation-degradation cycles (Figure 4.3e). Cycle 1 lasted825 minutes, and was the largest cycle in terms of volume of material (4595 g). Flumeobservations showed that the degradational phase of this cycle was related to the additionalscouring of a large pool located at the downstream end of the flume. On the other hand, theaggradational phase of the cycle was dominated by the growth of a bar located downstreamof the same pool. The remaining two cycles were related to deposition and erosion withina pool and bar. The range of sediment input for these three experiments offers additionalinsight into the transport-storage relation. Exp. 9 had no cycles which suggested that thechannel was not in equilibrium and that upstream sediment supply is needed for the cyclesto occur, as internal supply is not sufficient These results indicate that the channel mayhave achieved brief periods of temporary dynamic equilibrium during Exp. 5, as indicatedby the cycles, whereas Exp. 7 exhibited longer periods of dynamic equilibrium.The transport-storage relation for Exp. 8 shows a system in equilibrium (Figure 4.3f).Two cycles of aggradation-degradation were identified. Cycle 1 began after 75 minutes andlasted 1125 minutes and 790 g were displaced, nearly fourteen cycles occurred within cycle1. Cycle 1 is obviously the largest cycle during Exp. 8 but it is relatively small comparedto the large cycle in Exp. 7. The transport-storage relation for Exp. 8 suggests relativelylittle change in channel morphology occurred during Exp. 8 compared to Exp. 7.In general, the experiments indicate that cycles occurred when transport rates weresimilar to the feed rates. The cycles tended represent periods of elevated transport rates,which lead to a decrease in the sediment stored in the channel, followed by lowered transportrates, during which sediment stores were replenished. Cycles exhibited a very complexrelation between transport and storage, and not just simple hysteresis. Cycle 4 of Exp. 7(inset of Figure 4.3f) clearly indicates that multiple transport rates can be associated witha single storage value.4.2.3 Patterns of Channel AdjustmentsIn this section patterns of channel morphological adjustments which occurred during theexperiments are examined. Observations of channel adjustments were made hourly withhand sketched maps and every 5 hours using the DEMs generated from the laser scans ofthe bed. Figure 4.4 shows hillshaded DEMs from the experiments and includes selectedDEMs from all of the experiments. This section will be organized in a different manner thanthe previous sections. In the previous section it was shown that historical contingenciesplay an important role in morphological development. Thus, in order to chronologicallyshow the patterns of channel adjustment, the experiments will be presented in the orderthat they occurred.Exp. 1 and Exp. 2 were run sequentially on the same bed, the experiments investigated634.2. Resultsthe changes in sediment feed rate. Exp. 1 was the longest running experiment and hadthe highest sediment feed rate (0.91 g/s), stream discharge was the same (Q = 1.6 L/s) forExp. 2 but the feed was significantly lower (0.45 g/s) . The hillshaded DEMs (Figure 4.4)provide context for the channel conditions at the start of each of the two experiments. Exp.1 began on a bed exposed to 2 hours of conditioning flows only, sediment lobes are evidentin the DEM as is the uniformity of the bed at this stage. Exp. 2 started on the bed left byExp. 1 (Exp. 1 5700 min in Figure 4.4), pools and riffles are clearly evident in the DEM.The difference DEMs illustrate the patterns of channel adjustment at selected intervalsduring each of the experiments (Figure 4.5). The first 300 minutes in Exp. 1 resulted inscouring of pools and aggradational areas associated with bar development. The location ofpools are relatively well distributed throughout the length of the channel. For the majorityof the experiment aggradational areas were dominant. Degradation during the experimentwas mainly the result of the removal of material associated with collapsed bar faces andpool scour. Exp. 2 had behaved similarly to the final 900 minutes of Exp. 1. The patternof channel adjustments during Exp. 2 appear to follow the template inherited from Exp.1. It is likely that the reduction of sediment supply was not entirely completed during theduration of the experiment (the experiment was halted as flows overtoppped the banks).Inspection of the final hillshade DEMs show that the adjustments during Exp. 2 wererelatively minor when the two DEMs from Exp. 1 are compared (Figure 4.4).644.2.ResultsExp. 1 0 5700 Exp. 2 2400 Exp. 3 0 2400 Exp. 4 0 2400 Exp. 5 1800 Exp. 6 0 2400 Exp. 7 1800 Exp. 8 1500 Exp. 9 900 Flow DirectionElevation in mm  Figure 4.4: Hillshaded DEMs for selected time periods for all experiments. DEMs for the initial, screeded bed after 2 hoursconditioning, and final DEM of each experiment are shown. In addition to the experiment number, the experimental time inminutes for each DEM is included. The scale of the plots represents relative elevation in mm, an arbitrary datum was used.654.2. ResultsExp. 3 was conducted using similar governing conditions as Exp. 2 except that it wasstarted on a new bed (Figure 4.4). The difference DEMs for Exp. 3 are provided in Figure4.5. During the first 300 minutes channel scour associated with the development of poolswas the primary adjustment mechanism. Following this initial period of activity, aggrada-tional processes dominated. The final DEM in Figure 4.4 for Exp. 3 (2400 minutes) showsthat the channel pattern established during the first 300 minutes remained throughout theexperiment, with minor adjustment.The next set of comparative experiments are Exp. 4 and Exp. 5. Exp. 4 (Q = 2.0L/s and Qf = 0.55 g/s) began on a new bed and had lower stream discharge and sedimentfeed rates than Exp. 5 (Q = 2.4 L/s and Qf = 0.68 g/s). The experiments, however, hadsimilar sediment concentrations.The hillshaded DEMs show a similar configuration (i.e.,thelocation of pools and riffles) as the earlier experiments (Figure 4.4). The downstream poolis substantially larger in Exp. 4 and Exp. 5 than in the previous, lower flow, experiments.Additionally, the upper portion of the channel, exhibits a more plane-bed morphology thanwas seen in the previous experiments. The first 300 minutes of Exp. 4 was predominantlydegradational, as the flow incised into the previously undisturbed bed and a pool-rifflemorphology developed (Figure 4.5). After this period, patterns of channel adjustment inExp. 4 were focused around these areas up until the last five hours, when a pool locatedalong the left bank half way down the channel began to fill in. This appears to be theresult of the extension of plane-bed morphology downstream.As with Exp. 4, the first 300 minutes of Exp. 5 shows degradation and scour of poolsinherited from Exp. 4 (Figure 4.5). For the remainder of the experiment, most of theactivity occurred in the lower portion of the flume. The hillshaded DEM shows how theplane-bed morphology extended further downstream during Exp. 5 than it had by the endof Exp. 4.664.2.Results1800-1500Flow DirectionScale units = mm2 300-0 1200-900 2100-1800 4800-4500  300-0 Exp. 1 Exp. 23900-3600 5700-5400  600-300  900-600  1200-900 1500-1200  2400-2100  300-0 Exp. 3 600-300  900-600  2400-2100  300-0  900-600 1500-1200 1800-1500 2400-2100  300-0 1200-900 1500-1200Exp. 4 Exp. 5Figure 4.5: Changes in channel storage illustrated through DEM difference maps for Exp. 1 to Exp. 5. Scale units are in mm2,and negative values represent degradation (orange-red) and positive values aggradation (blue). Each individual map represents thedifference between a DEM generated following one 5 hour run subtracted by a DEM generated from the previous 5 hour run. Forexample, in Exp. 2 the difference between the DEMs generated after 300 min and at 0 min is labeled 300-0. Flow direction iscommon for both experiments.674.2. ResultsThe comparison between Exp. 6 and 7 examines differences in discharge, which in-creased from 2.0 L/s in Exp. 6 to 2.4 L/s in Exp. 7, sediment feed rates were held constant(Qf = 0.50 g/s). Additionally, Exp. 7 began on a bed which had been conditioned duringExp. 6. Within the first 15 minutes of Exp. 6, the formation of incipient bars beganto occur as downstream migrating tongues of finer sediment were prevented from furtherdownstream movement possibly due to interactions with channel banks (some of which areevident in Figure 4.4). After 300 minutes, three pools are evident in the DEM (representby the orange-red colour in Figure 4.6), the upper part of the channel exhibited plane-bedlike morphology and the lower half resembled pool-riffle morphology. Over the next 1 500minutes the channel evolved by increasing lateral variability, pools and bars became moreprominent and as such became a more dominant influence on flow patterns and patterns ofsediment transport. Pool locations for the duration of the experiment appeared to remainin place although their pool dimensions (length, width and depth) varied overtime. Animportant observation was that the alignment of most pools varied as well, with the poolinlet and outlet angles varying over time. The orientation of the inlet and outlet were ini-tially determined by the channel banks and over time the orientation evolved in responseto pools that developed upstream and downstream (see Exp. 6 2400 min Figure 4.4).684.2.Results 300-0 600-300 900-600 1200-900 1800-1500Flow Direction 300-0 900-600 1500-1200 2400-21002100-1800Scale units = mm2Exp. 6 Exp. 7Exp. 8 Exp. 9 300-0 600-300 900-600 1200-900  300-0 600-300 900-6001500-1200Figure 4.6: Changes in channel storage illustrated through DEM difference maps for Exp. 6 to Exp. 9. Scale units are in mm2,and negative values represent degradation (orange-red) and positive values aggradation (blue). Each individual map represents thedifference between a DEM generated following one 5 hour run subtracted by a DEM generated from the previous 5 hour run. Flowdirection is common for both experiments.694.2. ResultsThe higher stream discharge in Exp. 7 capitalized on the pre-existing flow pathwaysestablished during Exp. 6 and the resultant scour along these routes is evident in theExp. 7 300-0 difference map (Figure 4.6). Surface structures and armouring were observedin the upper section of the flume. The lower section exhibited pool-riffle morphology andresponded immediately to the higher flows with deepening of the pools and degradation andaggradation on some of the bars. The first 300 minutes showed the greatest change as thebed and morphology adjusted to the increased flows. Morphological change dramaticallyslowed after the first 5 hours as evident in the 600-300 difference map in Figure 4.6, andthe channel appeared to quickly accommodate the higher flows. After 600 minutes theupstream end of the flume degraded, and the downstream end aggraded as pools werein-filled and the complexity diminished. After 900 minutes, and for the remainder ofthe experiment, relatively little morphological changes occurred. During Exp. 7 mostof the adjustments were surface coarsening and the extension of plane-bed morphologydownstream the channel, although a large pool remained at the lower end of the observationarea (Figure 4.4).Exp. 8 started on the bed inherited from Exp. 7; discharge and feed rates were thesame, but the size distribution of the feed was different. The difference maps in Figure 4.6for Exp. 8 are a much lighter shade than the previous two experiments, which indicatesthat there was less scour and fill during this experiment. Very small differences are evidentat the end of Exp. 8 as compared to the DEM end of Exp. 7 (Figure 4.4).This trend of limited morphological activity continued for Exp. 9, a no feed experimentwhich continued on the bed from Exp. 8. During the first 300 minutes of Exp. 8, littlechange is evident in the difference maps (Figure 4.6). Even by the end of Exp. 9 relativelylittle change is evident in the hillshade DEMs (Figure 4.4).4.2.4 Sediment Texture of the Transported Material and Bed SurfaceTemporal patterns in bed surface texture and sediment output to changes in sedimentsupply, flow conditions and within channel storage, are shown in plots of the D50 and D90of both the bed surface and the bedload material as trapped at the outlet (Figure 4.7).For comparison, D50f and D90f of the sediment feed mixture is included. For Exp. 1 boththe bed surface and outlet are presented at 60 minutes intervals, for the remainder of theexperiments the bed surface samples are presented at 300 minute intervals and sedimentoutput at 60 minute intervals. A summary of the average values for each experiment isprovided in Table 4.6, also included is the average armour ratio calculated using the D50and D90.In each experiment the transported D90t is relatively similar to the surface D50s, exceptfor Exp. 9 (Figure 4.7). Exp. 9 was a degradation experiment and had no sediment feed,it was also conducted on a bed that had a well developed armour layer and exhibited fairlyentrenched plane-bed morphology. Deviations between the D50s of the surface and thetransported D90t likely reflect armour breaching conditions and thus mobilization of larger704.2. Resultsa. Exp. 1b. Exp. 2c. Exp. 3D i (mm)d. Exp. 4e. Exp. 5f. Exp. 6g. Exp. 7h. Exp. 8i. Exp. 90510 500 1000 1500 2000 2500D50 surface D90 surface D50 transport D90 transport012345670 500 1000 1500 2000 2500012345670 200 400 600 800 1000 1200 1400 1600 1800 2000012345670 500 1000 1500 2000 2500012345670 200 400 600 800 1000 1200 1400 1600 1800 2000012345670 100 200 300 400 500 600 700 800 900 1000012345670 200 400 600 800 1000 1200 1400 1600012345670 1000 2000 3000 4000 5000 6000012345670 500 1000 1500 2000 2500012345670 500 1000 1500 2000 2500Time (min)D90D50D90D50D90D50D90D50D90D50D90D50D90D50D90D50D90D50Figure 4.7: Particle sizes (D50 and D90) of bed surface and bed material measured at theoutlet for all experiments.714.3. Discussiongrain sizes. Further the D50t of the transported sediment in Exp. 9 was not different fromthe D50f of the feed, neither was there a difference in the relative values of the D90.Table 4.6: The D50t and D90t of transported and D50s and D90s surface sediment arepresented. The D50f is 1.14 mm and D90f is 3.28 mm for the sediment feed. The averagearmour ratio (AR) using the D50 and D90 are also provided.Experiment D50t D90t D50s D90s AR D50 AR D901 1.15 2.76 2.58 5.16 2.26 1.532 1.12 2.89 3.03 5.69 2.65 1.733 1.04 2.25 2.37 4.80 2.07 1.464 1.14 2.80 2.89 5.37 2.53 1.645 1.06 2.67 3.29 5.62 2.88 1.716 1.03 2.52 2.64 5.27 2.30 1.617 1.13 2.96 3.13 5.81 2.73 1.778 1.14 2.98 3.22 6.18 2.81 1.889 1.11 3.43 3.12 6.05 2.73 1.84Generally, the results suggest that the response of the bed may not be as dependentupon sediment supply and discharge, but on the range of sediment sizes supplied to thechannel. The D50t for all of the experiments did not differ greatly from that of the initialsediment feed, the larger range in values observed in the D90t suggest that this may bea better metric for evaluating channel responses. The armour ratios observed in theseexperiments is well within the range of those observed in the field, suggesting that thisprocess was replicated for these experiments.4.3 DiscussionThe nine experiments presented above were designed to explore the spatial and temporalpatterns of sediment transport. In addition, channel response to changes in sedimentsupply and flow was observed through changes in sediment storage. The experiments weregrouped into themes based on differences in either discharge, sediment supply or sedimentfeed grain size. The first theme explored the effects of the initial conditions of the channel,with initial conditions beginning on a screeded bed and discharge and feed rates beingsimilar (Exp. 4 versus Exp. 6) or different initial conditions with similar discharge andfeed rate (Exp. 2 versus Exp. 3). Secondly, the effects of stream discharge with constantsediment feed rate and texture (Exp. 6 versus Exp. 7) is explored. For the second theme,the sediment feed rates were varied while the stream discharge and sediment feed texturewere held constant (Exp. 1 versus Exp. 2 and Exp. 3, Exp. 4 versus Exp. 6, and Exp. 5724.3. Discussionversus Exp. 7). Lastly, the effects of a change in the sediment feed texture while holdingflow discharge and sediment feed rate constant was explored (Exp. 8 versus Exp. 7).4.3.1 Sediment Transport PatternsSediment transport for all nine experiments exhibited both spatial and temporal variabilitywithin each experiment even under conditions of constant sediment feed and flow discharge.The experiments were designed to investigate the effects of formative flows and time onsediment transport patterns. The design provided a unique opportunity to observe andmeasure long-term patterns in sediment transport.Differences in sediment transport patterns emerged when the experiments within a spe-cific theme were compared. Exp. 6 and Exp. 7 explored the effects of increased dischargeand found that a 20% increase in discharge resulted in a 35% increase in sediment transportrates, after the initial scour period was excluded. The elevated transport rates in Exp. 7were associated with channel scour in pools and along the channel thalweg. The increasein sediment transport rates is more reflective of the increase in sediment availability, thanto the changes in transport capacity due to the flow increase. Channel scour resulted inincreased transport rates due to: 1) increased availability of sediment for transport and, 2)increased efficiencies in the bed shear stress distribution due to increased lateral variability(Ferguson, 2003; Francalanci et al., 2012). After the initial scour period, minor adjust-ments in the size distribution of the surface material associated with individual bedformsand surface structuring, transport rates declined. The surface grain size adjustments ef-fectively acted to stabilize the new morphological configuration, allowing an efficient yetstable morphology to develop.Previous experiments have indicated the importance of adjustments of surface texturein regards to sediment transport (for example, Dietrich et al., 1989; Buffington and Mont-gomery , 1999b), with little or no mention of morphological adjustments. The processappears to be more complex in channels with a more complex planform when compared tostraight flumes. The experiments were conducted using formative flows (i.e. bankfull andabove), in pursuit of understanding the morphological development of this type of channel.What was observed under these flow and transport conditions, was that morphological ad-justment was the primary response mechanism, after the surface layer was breached, andtextural changes were secondary. Typically, the higher transport events which dominatedtransport rates were associated with channel unit changes, rather than textural changes.Channel unit adjustment was manifested by pool/bar development as well as the devel-opment of surface structuring in the straighter upstream section of the flume. After theinitial morphological adjustment period had slowed down, the adjustment never stoppedoccurring as pool and bar dimensions generally evolved throughout the experiments; sur-face material size adjustments were a secondary mechanism and appeared to adjust inresponse to altered flow velocities associated with newly formed morphology (for example,topographic steering Nelson (1988)).734.3. DiscussionThe importance of morphological adjustment can be seen by examining the spatialdifferences between Exp. 6 and Exp. 7 (Figure 4.6). The location of scour and fill isslightly different between the two experiments, and the magnitude of the changes appearsgreater in the plots. The deeper scour in Exp. 7 is associated with channel banks, especiallyat the downstream end of the channel and along the left bank. This bank-channel couplingwas not as pronounced during Exp. 6. Thus, during high flows it appears that the flowcan be temporarily attracted to non-erodible banks, which results in deep scour holes. Itis interesting to note that a similar scour feature was observed in Fishtrap Creek, whenthe creek encountered a bedrock outcrop along its left bank. We have also observed thisbehaviour in the field in many channels where the flow gets keyed into some non-erodiblebank feature and an associated deep scour hole develops. These experiments suggest thatthis behaviour is at least initially associated with some high magnitude flow event, greaterthan 10 years in this experimental case. This may suggest that above some magnitude flowevent within-channel transport processes are unable to accommodate the associated flowenergy and that the boundary conditions (such as hardened banks) are attractors becausethey can dissipate the additional energy.The DEM difference maps can be used to highlight the spatial and temporal variabilityof in-channel sediment reservoirs (Figures 4.5 and 4.6). For example, inspection of thedifference DEMs of Exp. 2 (Figure 4.5) reveals a variable pattern of channel aggradationand degradation both laterally and longitudinally during the course of the experiment.Patterns of bar building (blue areas) and pool scouring (orange-red areas) vary in bothmagnitude and space. Similar differences can be seen in the Exp. 5 maps, especially whencomparing the the top 1/4 and lower 1/4 of the channel. The behaviour suggests thatin-channel sediment reservoirs are closely linked to individual morphologically units, andthat this may be a more important driver of in-channel adjustments to changes in sedimentsupply than reach scale information. This may also be the scale at which transport-capacityis determined.An additional observation can be made regarding discharge magnitude and channelmorphology. Examination of the hill shade DEMs from Exp. 1 to Exp. 3 in Figure 4.4,reveals the development of a pool-riffle morphology throughout the entire experimentalchannel. A pool can be clearly identified in the Exp. 3 2400 minute DEM. When dischargeis increased (for example, Exp. 5 1800 minute DEM in Figure 4.4) the upper, straightersection of the channel exhibits plane-bed morphology. I believe that this change representsthe more dominant role that channel alignment and planform have in channel morphologicaldevelopment as flood stage increases. This suggests that during higher magnitude floodevents, channel morphology is increasingly dependent upon hydraulic forcing due to theboundary conditions and bank alignment than either armouring or morphologic units.The effects of changes in sediment supply were examined by comparing the results fromExp. 1, Exp. 2 and Exp. 3 and Exp. 5 and Exp. 7. The sediment feed rate in Exp. 2 andExp. 3 was half of that in Exp. 1, this resulted in a 46% reduction in transport rates inExp. 2 and a 72% reduction rates during Exp. 3. When transport efficiency was used as744.3. Discussionthe metric a different picture emerges, Exp. 2 had the highest transport efficiency (0.84),followed by Exp. 1 (0.78) and Exp. 3 (0.44). As Exp. 2 inherited the bed from Exp. 1, itis likely that the gain in sediment transport efficiency is related to increased channel slope.Transport efficiency during the last 1500 minutes of Exp. 1 is 0.81, which is more similarto Exp. 2. This similarity is likely the result of two morphologic changes in the channel ofExp. 1, the first is increased channel slope, but the second is increased transport efficiencydue to the establishment of sediment pathways, the result of morphologic forcing. Theresults for Exp. 3, however, cannot be attributed to slope alone. The start conditionsfor Exp. 1 and Exp. 3 were similar, so the experiments should have produced relativelysimilar results. When the first 2400 minutes of Exp. 1 are examined transport rates are0.60 g/s with an efficiency of 0.65 which is still more than double the results observed forExp. 3.The feed rate for Exp. 5 was 34% higher than Exp. 7, but the discharge for theseexperiments was 20% higher than Exp. 4 and Exp. 6, the beds upon which Exp. 5 andExp. 7 were conducted. The result was that the overall transport rate for Exp. 5 was only16% higher than Exp. 7, this increased to 24% when the initial scour event was removedfrom both experiments. Interestingly, the transport efficiency was actually reduced by11%, from 0.84 in Exp. 7 to 0.75 in Exp. 5. These results indicate the importance ofhistorical contingencies in sediment transport patterns. Differences in the initial conditionscan produce unique transport pathways that can result in different transport efficienciesand unique morphological characteristics that can be promoted under differing governingconditions. This validates the complexity in responses observed when trying to understandchanges in the governing conditions and channel characteristics. It does suggest that abetter understanding of conditions prior to the disturbance may help in improving ourunderstanding of channel response pathways.The role of changes in the grain size of sediment feed did not produce dramatic changesin the morphology of the channel. The result may have been different if the experimentswere both started on screened beds. Comparing Exp. 7 and Exp. 8 the results indicatethat sediment transport equilibrium was reached relatively rapidly during Exp. 8, usingthe alluvium of Exp. 6. As to be expected Exp. 8 resulted in relatively minor changes inchannel morphology, with the majority of channel adjustment occurred in the lower tenthof the flume.4.3.2 Transport-Storage RelationsThe primary outcome of Lisle and Church (2002) was the identification that transportcapacity is not a static value. In their paper they supported this outcome using bothexperimental and field evidence from degrading channels and found that a unique, pos-itive relation existed between sediment transport rates and storage. Later experimentsconducted by Madej et al. (2009) and Pryor et al. (2011) looked at this transport-storagerelation over episodes of degradation and aggradation, and found that in some instances754.3. Discussioncycles in the relation occurred. Therefore a unique relation between transport and storagecould not be supported given their experimental conditions.Lisle (2012) amended the initial model and proposed two general scenarios: Scenario 1represents a dynamic equilibrium between transport capacity and sediment storage; and,Scenario 2 both the channel response and transport capacity vary as sediment storagechanges due to a transient disequilibrium related to changes in the rate of sediment supply.The conditions observed here display both scenarios. Under Scenario 1 transport ratesappear to represent a dynamic equilibrium during the aggradational phase, similar tothose reported in Madej et al. (2009) and Pryor et al. (2011). Scenario 2 conditions werealso observed here, under constant sediment feed rates and flows, which contradicts thehypothesis that the cycles are due to changes in the rate of sediment supply. From ourexperiments we suggest that the cycles tend to occur when rates of sediment output aresimilar to sediment input rates. This could also occur, as evident in the results of Madejet al. (2009) and Pryor et al. (2011), when episodes of the degradation and aggradationoccur such that the new sediment output rates are similar to the new sediment feed rates.During this cross-over period between the two episodes, sediment output can temporarilyequal sediment input and a hysteretic event could occur.I propose that the probability of hysteresis occurring in the transport-storage relationis directly related to the difference between sediment feed and sediment output. As thetransport rate nears the feed rate the probability of cycles occurring increases; this isrepresented graphically in Figure 4.8.An additional feature of these hysteretic events emerged from our experiments; thatis the temporal legacy of these cycles. As the experiments were Froude scaled the lengthof these cycles can be converted from experimental time to time in the prototype. Thisconversion results in a cycle duration time of over 2 years for a small cycle in Exp. 1 (60minutes in model) to over 50 years for a larger cycle in Exp. 7 (1470 minutes). Thesecycles can occur over periods that are generally much greater than common periods of fieldinvestigation, and therefore the evidence of such cycles is likely not frequently observed.Examining annual cross-sectional data from a long-term dataset in Carnation Creek, Has-san et al. (2007) found cycles greater than 10 years in duration in reaches where in-channelstorage is predominantly associated with large woody debris. The potential duration ofcycles in the transport-storage relation further reaffirms the importance of considering vari-ability of transport capacity in landscape evolution models. More importantly, if assuminga static value, depending on where in the cycle the current channel is, may have a strongeffect on values measured in the field and then applied to a model.Hassan et al. (2007) concluded that the cycles depended more on the rate of sedimentsupplied to the channel, bedform dynamics and the supply of LWD than on hydraulicforcing. The LWD component aside, these experiments tend to partially support thisconclusion, with the addendum that conditions must exist such that the rate of sedimentsupplied to the reach is nearly equal to the rate exiting the reach.Partial cycles in the transport-storage relation were observed at the beginning of Exp.764.3. DiscussionFeed - Output Probability of Hysteresis 0 Degrading Aggrading Figure 4.8: Conceptual model of the probability of hysteresis occurring in the transport-storage-feed relation. The plot shows the probability of hysteresis occurring vs the feed-output (F-O); when F-O > 0 the system is aggrading, when F-O < 0 the system is degrad-ing. 774.3. Discussion4, Exp. 5 and Exp. 6 (see Figure 4.3d, 4.3e, 4.3f), the appearance of these cycles is inresponse to the initial degradation followed by form resistance and channel armouring. Theinitial flows in these experiments, whether they began on a conditioned bed or not, resultedin scour of the bed as the armour has either not been formed (Exp. 4 and Exp. 6) or wasformed under lower discharge conditions (Exp. 5). Once the armour layer has adjusted tothe new flows, bed scour lessens and the channel begins to aggrade, however, this resultsin transport rates much lower than the initial transport rates. This difference in transportrates between the armoured and un-armoured bed, resulted in the partial cycles evidentduring these three experiments.Only Exp. 9 and Exp. 3 did not experience cycles in the transport-storage relation.In both experiments it appears to be due to sediment feed rates. In Exp. 9 there was nosediment feed and the channel remained degradational throughout the experiment. Therelatively high transport rates at the beginning of the experiment represent an abundantsupply of in-channel sediment was initially available for transport, as this sediment wasremoved, transport rates declined. Exp. 3 had sediment feed, but transport rates neverapproached feed rates (see Figure 4.1c). In both experiments equilibrium was not reached,this is in contrast to the conclusions reached by Pryor et al. (2011), who determined thatthe cycles they had a observed were a disequilibrium phenomena. In these experimentsdisequilibrium appeared to have negated the occurrence of cycles.Full cycles in the transport-storage relation were observed for all of the remaining ex-periments. Close inspection of Figure 4.1 reveals that all of the cycles occur when thechannel is in a state of dynamic equilibrium, or where transport rates fluctuated aroundthe feed rate. During minor cycles the additional sediment is easily accommodated by avail-able storage reservoirs within the channel causing minor alterations in transport capacity.Hassan et al. (2007) associated small-scale aggradation-degradation cycles to the growthand shifting of lateral bars, similarly the minor cycles observed during these experimentswere associated with minor morphologic adjustments. These adjustments included the ex-pansion and contraction of bars, as well as the scouring and deposition in pools. Whereasmost of the adjustments for Hassan et al. (2007) were hypothesized to be associated withwood accumulations. Coarsening and structuring of the surface layer also played a part inthe development of minor cycles observed. Although the direct mechanism is more relatedto minor morphologic adjustments, the state of the surface layer obviously determines theavailability of sediment to be eroded.Major cycles seem to reflect an initial perturbation, usually degradational, followed byrecovery. For example, in Exp. 7 the perturbation was the initial period of degradation atthe beginning of the experiment. For Pryor et al. (2011) complete cycles of aggradation-degradation were related to external perturbations introduced by the experimenters in theform of changes in the sediment feed rate, similar results and outcomes were observed byMadej et al. (2009). These experiments clearly illustrate that no external perturbation tothe channel is required, and that these major cycles can be induced endogenously understeady state forcing conditions.784.4. ConclusionsThe cycles seem to represent a period of internal morphologic adjustment to the forcingconditions. In the experiments of Madej et al. (2009) and Pryor et al. (2011) this morpho-logical adjustment was manifested by changes to channel pattern from a single-threadedchannel to braided, and back to single-thread. However, in these experiments where thechannel was laterally constrained this did not occur, instead storage reservoirs within thechannel stored and released sediment in response. The releases represent perturbations inthe system, and are hypothesized to represent limitation in storage capacity. Althoughadjustments to the texture of the bed surface are more frequently discussed, it is appar-ent from these experiments that the bedform level of response if the more dominant andimportant. It is these features that give the most elasticity of the channel to exogenousperturbations, the surface armour offers a limited response to changes in sediment supplybut channel bars offer a much greater mechanism to accommodate changes in supply andcan internally regulate these changes by the storage and release of sediment in response tothe supply of sediment relative to channel flows.Lisle and Church (2002) suggested that experiments in which only the sediment feedrate is varied over-represent the adjustment of armouring to variations in sediment supply.This stems from Wilcock (2001b) who wrote that in order to replicate a realistic armouringresponse in a flume, varying sediment size, feed rate as well as flow is difficult. Added toWilcock’s statement is that realistic armouring also required realistic planform. Channelarmouring is not only dependent upon sediment size, feed rate and flow, but on morphologicvariability which creates complex hydraulic environments.4.4 ConclusionsThe experiments conducted herein focused on channel forming conditions, so higher flowswere employed. The results of these conditions showed that sediment transport ratesvaried both spatially and temporally. This variability was found to be more dependentupon changes in the morphology of the channel, rather than adjustments in the surfacegrain size distribution. The primary self-adjusting mechanism in the experimental channelwas to adjust channel form. Adjustments to bed surface texture, tended to be a short-term response and the development of patches appeared to develop more in response tothe channel topography rather than vice-versa.Sediment transport efficiency of the experimental channel was strongly influenced bymorphology and historical contingencies. Although, no replicate tests were conductedhigher transport efficiency occurred when discharge was greater or the grain size distribu-tion of the feed was finer, all other governing conditions remaining equal.The transport-storage relation was investigated under constant forcing conditions. Itwas found that sediment transport capacity was not constant and did change as the volumestored in the channel changed. As others have observed, hysteresis cycles were observedto occur in a number of experiments. Unlike previous studies, the cycles observed here794.4. Conclusionstended to occur when transport rates were approximately equal to the feed rates (i.e.,when the channel was near equilibrium). Cycles of aggradation-degradation were observedwithout changes in either sediment supply or flow discharge. It was concluded that thetransport-storage relation does not offer a unique relation between transport rates andsediment storage, thus limiting its usefulness in the development of an improved sedimenttransport model.The results suggest that the idea of sediment reservoirs is more complex than envi-sioned and that these reservoirs are transitional and multiple reservoirs may exist withinhomogeneous reaches and they may all have different time signatures in terms of their re-sponse. This difference is dependent upon flows, sediment supply, and local adjustment tothe topography, which in addition to the previous parameters is dependent upon boundaryconditions (such as bank stability, bank form and alignment (important in most forestedlandscapes). The importance of the capacity of these reservoirs is shown in these results;however the capacity evolves with the morphology of the channel, so a simple relationshipcould not be identified.Further attention to the role of topographic evolution within channels and their asso-ciated effect on sediment transport rates is suggested. It is believed that the topographicsignature of the channel plays a dominant role not only in hydraulic, and the distribu-tion of shear stresses, but on the development of in-channel surface patches and sedimentreservoirs.80Chapter 5Sediment Transport at thePool-Riffle Scale: Observationsfrom a Physical ModelIn this chapter a less formal style has been taken. During the course of the experimentsa number of phenomena were observed that were not anticipated at the experimentaldesign stage. Consequently appropriate data were not collected. Some of the qualitativeobservations that were made include hand drawn maps and videos created from digitalphotography of the channel. This allowed me to make inferences in a manner similar tothat proposed by Drake et al. (1988) in their study of bedload transport at Duck Creek.Similarly, I use these data to elucidate the processes of sediment movement and sorting inthe experimental channel.The general topic of this chapter is the movement of sediment through the experimentalchannel. Specifically, how sediment moves through a pool-riffle unit and the dynamics ofsediment movement as bedload sheets. A brief introduction to the literature regardingthese topic areas is presented next.5.1 IntroductionThe stochastic nature of sediment transport rates in gravel-bed rivers has been observedin both field (Ehrenberger , 1931; Einstein, 1937; Cudden and Hoey , 2003; Hassan andChurch, 2001; Hoey and Sutherland , 1991; Jackson and Beschta, 1982; Madej and Ozaki ,1996; Nanson, 1974; Powell et al., 2001; Reid et al., 1985; Wooldridge and Hickin, 2005)and laboratory (Dietrich et al., 1989; Iseya and Ikeda, 1987; Kuhnle and Southard , 1988;Lisle et al., 1993; Kuhnle et al., 2006; Nelson et al., 2009; Recking et al., 2009). In gravel-bed rivers, part of this stochasticity has been attributed to the migration of bedload sheets(Ashmore, 1991; Bennett and Bridge, 1995a; Drake et al., 1988; Gomez , 1983; Kuhnleand Southard , 1988; Kuhnle and Willis, 1998; Whiting et al., 1988; Madej et al., 2009;Nelson et al., 2009; Pryor et al., 2011). Whiting et al. (1985) initially coined the termbedload sheets to describe their observations of bed material being organized into waveswith distinct coarse fronts. The coarse fronts were only 1 or 2 grains in height and metresin length. A similar phenomenon, referred to as diffuse gravel sheets, had been previously815.1. Introductionidentified in the field by Smith (1974) and Hein and Walker (1976), who described thesesheets as incipient bar features (see also Prestegaard , 1987), the sheets were observedto only move during peak flows, and once the flow decreased the sheets were deposited,sometimes leading to the formation of diagonal bars (Hein and Walker , 1976).Whiting et al. (1988) suggested that bedload sheets develop in moderately to poorlysorted sediment potentially as a consequence of coarse and fine grain interaction. Thedifferentiation of bed material into coarse fronts and fine tails was thought to lead toincreased mobility of the sediment as a result of smoothing, exposure and collision (Iseyaand Ikeda, 1987). Seminara et al. (1996) attributed the growth of bedload sheets to grainsorting. Following Iseya and Ikeda (1987), Whiting et al. (1988) proposed that bedloadsheet migration in mixed-size sediment is related to a ”catch and mobilize” process inwhich the coarse grains in transport interact with the bed and either slow down or stop.The finer grains in transport pass over and fill in the interstices between the coarse grains,which in turn reduces roughness and increases the drag on the coarse grains, leading totheir re-entrainment.Bedload sheets appear to be directly related to sediment availability in the channel.Dietrich et al. (1989) observed that with decreasing sediment supply the spatial segrega-tion of the bed varied. During high feed runs, bedload sheets dominated the surface whichmanifested in alternating zones of congested (coarse), smooth (fine), and transitional (in-termediate) zones, as described by Iseya and Ikeda (1987). As sediment feed was reduced,bedload sheets became less frequent and distinct, coarse and inactive zones expanded, andthe zone of active sediment transport became a narrow longitudinal fine textured zone.Two recent papers have brought renewed attention to bedload sheets, Recking et al.(2009) focused on the production and migration of bedload sheets, and Nelson et al. (2009)focused on the response of bedload sheets to reductions in sediment supply. Recking et al.(2009) conducted 20 experiments under constant discharge and feed conditions, with vary-ing mixtures of uniform sediments, and at slopes that ranged from 0.8 to 9%. They foundthat bedload sheet production and migration was associated with variations in bed slope,bed fining and paving, and bedload. Recking et al. (2009) proposed that bedload sheetsresulted from vertical and longitudinal grain sorting that results in episodic increases inthe transport rate efficiency of the coarser sized fractions of the bed.Nelson et al. (2009) presented results from two sets of flume experiments, one conductedin the late 1980s at the University of Tsukuba (Dietrich et al., 1989; Kirchner et al.,1990) and another conducted at the UC Berkeley with the purposes of understanding theresponse of bed surface patchiness to reduction in sediment supply. Sediment patches arehomogeneous areas of similar sized sediments on the bed surface of a channel, that mayresult from variations in shear stress, sediment transport and topographic sorting (Paolaand Seal , 1995). Nelson et al. (2009) extended the terminology of bar types from Seminara(1998) to classify patches into three types: free patches, which are zones of sorted materialthat can move freely, bedload sheets being an example; forced patches, which are areas ofsorting forced by topographic controls; and, fixed patches, which are immobile as a result825.1. Introductionof localized coarsening and can remain persistent through time. In general, they founda direct link between sediment supply and the distribution of free and fixed patches, aswell the dynamics of bedload sheets. Nelson et al. (2009) attributed substantial increasesin sediment flux due to the passage of bedload sheets, and suggest that the migration ofthe sheets can be the primary cause of the observed short-term fluctuations in sedimenttransport rates.For the experiment presented here, the focus was on observations of bedload sheetmovement through complex channel morphology, where stable equilibrium conditions werenot observed. In addition to gaining insight into the behaviour and downstream movementof these features, their movement downstream allowed visualization of sediment transportpaths. In particular, we focussed on the evolution of sediment transport patterns througha pool and riffle sequence over time.How sediment moves and is sorted through pools and riffles is strongly dependent onthe varying shear stress induced by diverging and converging flow conditions related to thedownstream variability in bed topography. A number of models have been developed todescribe the relationships between the distributions of boundary shear stress with channeltopography and the spatial pattern of bedload transport and grain size in a meanderingchannels.The following brief description of flow structure in a curved channel comes from Powell(1998). In a channel meander, channel curvature results in a depth-dependent, centrifugalforce acting in the flow-transverse direction forcing surface water toward the outer bankleading to higher water surface elevation than the inner bank, where the water surfaceelevation is lowered (super-elevation). The difference between the centrifugal and pressuregradient forces results in outward flow on the surface where centrifugal forces are greater,and an inward flow on the bed where pressure gradient forces are greater (Figure 5.1a).This difference results in the helical secondary circulation. As flow progresses through ameander bend the forces change and this results in the zone of maximum boundary shearstress changing from the inside of the upstream bank to the outside of the downstream bank(Figure 5.1b). This is further enhanced in channels with pool-bar topography, because italters the near-bed flow velocity, increases the cross-stream variation in shear stress, andpromotes rapid shifting of the zone of maximum bed shear stress Robert (2003). The pathof a particle through a meander bend responds to the forces acting upon it and results ina pattern of sediment sorting that results in coarser sediment moving down the transverseslope of the bar toward the pool as it moves through the meander. How sediment movesthrough a bend will depend on the relative magnitude of the drag due to the inward-actingsecondary flow and the outward-acting gravitational force (Figure 5.1c).The objectives of this chapter are to:1. describe the general patterns of sediment transport in a complex flume during a singleexperimental run; and2. elucidate how bedload sheets interact with the bed morphology in a complex flume.835.1. Introductiona b c Super-elevated water surface Outer bank cell Characteristic secondary velocity profiles Secondary flows Outward shoaling flow across point bar Flow direction 4 metres Fine gravel Coarse sand Fine sand Region of maximum shear stress Fl Lift force Fd Drag Force Fg Gravity force Fgx Cross-stream component of gravity Figure 5.1: This figure contains a) flow structure, b) shifting loci of maximum boundaryshear stress and sediment transport pathways, and c) forces acting on a bedload particlein meander bends. From Powell (1998).845.2. ExperimentThe approach taken to meet these objectives is essentially qualitative in nature. Theseobjectives emerged from observations made during the experiments, consequently the de-sign of the experiments was not made with them in mind. Thus the data required toquantitatively present these objectives was not collected.5.2 ExperimentOnly the results from Exp. 6, will be discussed in this chapter. The experiment was con-ducted under conditions of constant flow (2.0 L/s) and sediment feed rate (0.50 g/s). Thehydraulic characteristics from this experiment are provided in Table 5.1, as are estimatesof the bedload sheet velocity, based on analysis of videos taken during the run.Table 5.1: Experimental conditions and resultsTime U a Slope d b Fr c Qb D50s (D90s)d D50t (D90t)e τb θ Bedload SheetVelocityf(min) (cm/s) (m/m) (cm) (g) (mm) (mm) (Pa) (cm/s)300 35.4 0.015 1.67 0.87 9,024 2.70 (5.00) 0.88 (2.12) 2.44 0.056 0.24-1.11600 35.6 0.015 1.66 0.88 2,953 2.35 (4.74) 1.16 (2.61) 2.44 0.064 0.22-0.69900 34.5 0.015 1.71 0.84 3,071 2.51 (4.80) 1.12 (2.57) 2.54 0.063 0.26-0.891200 34.5 0.016 1.71 0.84 7,046 2.82 (5.38) 1.12 (2.77) 2.65 0.058 0.21-0.991500 34.2 0.017 1.73 0.83 6,686 2.70 (5.23) 1.00 (2.50) 2.82 0.065 0.21-0.631800 33.4 0.017 1.77 0.80 6,381 3.02 (6.31) 1.04 (2.58) 2.95 0.060 0.15-0.992100 33.0 0.017 1.79 0.79 6,952 2.51 (5.09) 0.95 (2.41) 3.02 0.074 0.22-0.772400 34.7 0.018 1.70 0.85 7,232 2.49 (5.58) 0.95 (2.58) 2.97 0.074 0.27-1.46a. U is the average spatial harmonic velocity.b. d is the average depth, calculated using continuity d = Q/(wU).c. Fr = U/√gd, where U is the velocity, g is gravitational acceleration, and d is the average depth.d. D50s (D90s) represents the 50th (90th) percentile of the bed surface grain size distribution.e. D50t (D90t) represents the 50th (90th) percentile of the transported material grain size distribution.f. Bedload sheet velocity was determined from videos. The minimum and maximum velocity observedduring a 5 hour are given.5.3 Experimental Observations5.3.1 General ObservationsA total of 648 individual velocity measurements were made during the experiment, whichis approximately 16 per hour. The average velocity for the entire experiment was 34.4cm/s, all individual 300 min intervals were well within one standard deviation of the mean.855.3. Experimental ObservationsFlow Bar Pool Figure 5.2: Photo of experiment looking upstream during low flow conditions. Emergentbars and pools are identifiable in the image.865.3. Experimental ObservationsVelocity appeared to decrease from the onset of the experiment up until until 2100 minutes(Figure 5.3). The channel aggraded throughout the experiment, as a result channel slopeincreased from 0.015 to 0.018. Average channel depth, determined using the continuityequation, was approximately 1.7 cm for the experiment. The Froude number was alsoconstant for most of the experiment, at around 0.84. Volumetric sediment output peakedduring the first 300 minutes of the experiment, then decreased for the following 600 minutes,and then remained relatively constant for the remaining 1500 minutes.Changes in the surface grain size distribution were relatively minor over the course ofthe experiment. The bulk of the adjustments in grain sizes were made during the first 5hours; recall that the D50 and D90 of the initial bed sediment and sediment feed was 1.14mm and 3.28 mm respectively. The average armouring ratio for the experiment was 2.31for the D50 and 1.61 for the D90. For the transported material the average D50t was 1.03mm, which is slightly smaller than the D50 of the feed, and the average D90t was 2.52,which is significantly finer than the feed.During the conditioning phase of the experiment, small lobes or tongues of fine sed-iment were observed. These features appear to be similar to those described previouslyby Ashmore (1991). Figure 5.4a shows a colour close up of one such lobe, that was stillvisible after the experiment had stopped. The lobe is delineated in the photo based on itfiner texture relative to the surrounding sediment. In the smaller scale image, the blackand white one, additional fine sediment depositional areas can be identified upstream anddownstream of the location of the close-up image. In Figure 5.4a the location of four of theten cross sections monitored during the experiment have been identified. Micro bedformfeatures that had developed during the first 300 min did not seem to be of sufficient sizeto influence channel hydraulics, even at the local scale. At this flow stage, the planformof the channel appeared to be the dominant influence on flow characteristics. The channelat this stage was fairly uniform, no pools had developed and the fine depositional areaswere not topographically significant. Figure 5.6 shows selected cross sections following theconditioning flow, 0 min, and only minor cross section variability is evident. The bed sur-face sediment, other than the areas of fine deposition, remained fairly similar to the initialcondition.At the start of the experiment active sediment transport was observed to be occurringover 90% of the channel width at some of the observed cross sections. The presence ofsediment transport occurring in the active areas was identified by observing either sedimenttransport as bedload sheets or traction transport.Within the first 60 minutes of the experiment, channel bars had developed and poolshad begun to form. The morphological development of the channel was partly associ-ated with incipient bar features that had been identified following the conditioning run.The downstream progression of some of the sediment lobes had been halted by local flowconditions primarily due to channel hydraulics associated with the planform. Using theterminology of Nelson et al. (2009), free patches had become fixed patches. Some of thefiner free patches that became fixed, remained fine for long periods. The width of active875.3. Experimental ObservationsTime (min)Average Velocity (cm/s)0510152025303540450 500 1000 1500 2000 2500Figure 5.3: Plot showing individual velocity measurements for Exp. 6.885.3. Experimental Observationsc) 1800 mind) 2400 mina) 0 minb) 900 minFlowfbfbfbfbfbb ppfbbfbfbfbp pppfbpppbbbsbsbsbsbsXS 195XS 245XS 295XS 345XS 195XS 245XS 295XS 345XS 195XS 245XS 295XS 345XS 195XS 245XS 295XS 345Figure 5.4: Large scale photos taken during periods of no flow and overhead photos takenfrom video during the experiment are presented at various time intervals: a) 0 minutes,b) 900 minutes, c) 1800 minutes, and d) 2400 minutes. Channel features identified on themaps have been transferred onto the smaller scale images, the following acronyms wereused: fb = fine bar; b=bar; p=pool; bs = bedload sheet. Bedload sheets when present onthe maps were added as dotted lines on the image. The location of cross sections (XS 195,245, 295 and 345 presented in Figure 5.6) discussed is shown on overhead photo a).895.3. Experimental Observationstransport was 100% at some of the cross sections, but on average represented 62 % of thechannel width.After 300 minutes some minor cross sectional adjustment had occurred, except forFigure 5.6d where the channel aggraded. Aggradation of the channel at this cross sectionwas the result of a developing lateral bar, which also resulted in a decrease in the widthof active transport by over 50% at this cross section (Figure 5.5b). Other cross sectionsdid not experience such a reduction and overall active transport remained high relativeto channel width (78%). The lateral topographic change at this cross section resulted inconcentration of the flow, and consequently sediment transport..As morphological complexity of the channel increased, the width of active sedimenttransport zone decreased. By 720 minutes, a large bar had developed in the upper sectionof the flume along the right bank. The development of the bar modified local hydraulicswhich caused the concentration of sediment transport. Although sediment output remainedrelatively flat, the variability in the width of active sediment transport appeared to be adirect result of increased morphological complexity.After 900 minutes, the zone of active transport was mostly confined along the channelcentreline (Figure 5.4b). The bed surface appears finer in this image compared to the 0min image, although this is not reflected in the surface samples (Table 5.1), fine sedimentpatches along the left and right banks downstream of the channel curves are also evident.The development and growth of bars upstream of XS 195 limited the downstream movementof sediment, and hence the width of active transport following a peak after 800 minutes(Figure 5.5b). As a result of bar formation, highlighted in Figure 5.4b along the upstreamright bank, the zone of maximum sediment transport was directed from the right bankdownstream toward the left bank, which lead to pool development at XS 245, which thendeflected the zone downstream to the right bank, at XS 295, where an additional pooldeveloped. Bedload sheets were mapped and their position of movement is indicated inthe figure, and remained in the mid-channel position. Channel cross sectional change after900 min are most evident at XS 195, 245 and 295 (Figure 5.6a - c).Following 1800 minutes a pool adjacent to the left bank had developed upstream of XS195, sediment exited the pool and traversed the channel to a pool located at XS 195 (Figure5.4c). The width of the active transport zone declined after 1400 minutes (Figure 5.5b)to 35%. It is interesting to note that although sediment feed rates, sediment output, andthe surface grain sizes following the first 300 min were fairly constant, the zone of activetransport continually evolved which is directly related to the morphology of the channel.In Figure 5.4c the pool at XS 295 scoured to the bottom of the flume, water depth wasgreater than 6 cm. Sediment that exited the pool took one of two paths Figure 5.4c. Theorigin of the sediment that exited the pool determined which path the sediment wouldmore likely take: the mid-channel path was mainly sediment the came from the erodingtransverse bed slope of the pool, related to additional lateral pool scour and growth; and,the path closer to the right bank was sediment that mainly came from upstream. In Figure5.6 maximum pool depths occurred at both XS 195 (4 cm) and XS 295 at 1800 minutes.905.3. Experimental Observations01000200030004000500060000 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400Sediment Output (g)Time (min) 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400Active Transport Width (cm)Time (min)AverageXS 10XS 40XS 70XS 90a)b)Figure 5.5: a) Cumulative sediment output (g) at 60 minute intervals is shown on theprimary axis over time. b) The average width of active sediment transport of all tencross sections during the experiment and from four individual cross sections. Note that noobservations were made at 240 minutes.915.3. Experimental ObservationsAfter 2040 minutes a similar situation occurred with the pool located at XS 195, that istwo separate sheets exited the pool at the same time.After 2400 minutes a bar developed between XS 245 and XS 345 along the left bank(Figure 5.4d). The presence of the bar altered the path of bedload sheet migration as thedimensions of the pool at XS 195 expanded. As sediment emerged from the pool it fannedout over the downstream riffle, with the finer sediments remaining near the right bank andin the middle of the channel, and the coarse material moving in between these two zones.The colour picture in Figure 5.4d) shows this result clearly, with a fine sediment ribbonclose to the right bank, and another ribbon seen in the mid-channel. Coarse sedimentmovement was associated with the bedload sheets as well as traction transport in betweenthe bedload trails visible in the photo. The width of active transport declined after 1800minutes to an average low of 20% at 1980 minutes, by 2400 minutes it had risen to 45%(Figure 5.5b). Infilling of pools is evident in Figure 5.6, for XS 195 to XS 295.Overall, the experimental design successfully produced pool-riffle morphology in theflume. The use of fixed irregular banks created complex morphology and patterns ofsediment transport, not unlike those observed in nature. Surface coarsening of the channeloccurred relatively early on in the experiment, although morphological evolution of thechannel continued throughout. This result suggests that the development and interactionof morphologic units is more important to channel form and sediment transport processesthan channel armouring, especially under high flow conditions, similar to the conditionsproduced in these experiments.The importance of bedload sheets was also highlighted by this experiment. The pres-ence of sheets not only aided in identifying sediment transport pathways, but was also anindicator of in-channel sediment supply and morphological change. Bedload sheets will bediscussed further in the following section.5.3.2 Bedload SheetsMost observations of bedload sheets in flumes have been made using straight walled flumeswith plane bed morphology (for example, Iseya and Ikeda, 1987; Dietrich et al., 1989;Kuhnle and Southard , 1988; Nelson et al., 2009; Recking et al., 2009). It is possible thatthese conditions promote the abrupt segmentation of smooth versus congested zones (Iseyaand Ikeda, 1987). Topographically complex conditions with lateral cross sectional diversity,due to the presence of channel bars and pools, introduced more complex behaviour inbedload sheets than observed previously. Bedload sheets were observed to grow and theirmigration rate increase in pools or topographically convergent areas, generally sediment inthese areas is more mobile during higher flow events. In topographically divergent areas,such as riffles, bedload sheets can sometimes lose velocity and disperse and other timesnot.Bedload sheet migration velocity was highly variable during the experiment. Table5.1 shows the range in observed velocities for each 300 minute period. On average the925.3. Experimental Observations1201401601802002202402600 100 200 300 400 500 600 700120140160180200220240260300 400 500 600 700 800 900120140160180200220240260100 200 300 400 500 600 700 8001201401601802002202402600 100 200 300 400 500 600Distance from Left Bank (mm)Elevation (mm)a) XS 195b) XS 245c) XS 295d) XS 3450 min300 min900 min1800 min2400 minFigure 5.6: Selected cross sections from laser scans of the bed for: a) XS 195; b) XS 245; c)XS 295; d) XS 345. Data is presented at 300 minute intervals, when scans were completedwhile experiment was not running.935.3. Experimental Observationsobserved velocities for bedload sheet migration ranged from 0.15 - 1.46 cm/s in this experi-ment. Comparatively, Recking et al. (2009) found velocities from 0.6-1.4 cm/s; Kuhnle andSouthard (1988) measured 0.5 - 1 cm/s; Iseya and Ikeda (1987) measured 0.36-1.13 cm/s;and, Madej et al. (2009) found velocities of 1 - 2 cm/s. Recking et al. (2009) suggestedthat variability in his experiments was partly attributable to channel slope. Variabilityin the velocities observed here appeared to be partly due to the availability of sediment,with high velocities and frequency at the beginning of the experiment and during periodsof pool scour and growth, and channel morphology. During periods of high sediment avail-ability, bedload sheet velocities were high. It should be noted that migration velocitieswere determined based on a limited observation window, between XS 195 and XS 245 inFigure 5.4. Similar to Kuhnle et al. (2006) coarser grains appeared to travel faster thanthe finer fractions of the bedload sheet. It would be expected that bedload sheet migrationvelocities would be high in an area of convergent flow (i.e., a pool) compared to an area ofdivergent flow (a riffle). The variability may also suggest that bedload sheets interact withthe zone of maximum shear stress (assumed to be approximately where with the highestconcentration of sediment in transport was observed), and when the sheet is in the zonevelocities are higher than when they are not.Figure 5.7 is an image from the overhead camera during the experiment. In the up-stream edge of the image a single bedload sheet can be seen as it entered the pool, nearthe right bank in the image (as indicated by the thicker arrow). The upstream sheet canbe identified in the image by its smoother texture and lighter hue of grey, in contrast tothe coarser textured, darker tone surface of the non bedload sheet bar toward the leftbank. At the downstream end of the pool, two bedload sheets exit the pool. The bedloadsheet adjacent to the right bank followed an existing ribbon of fine sediment, evident inthe photo. The sheet was relatively narrow, less than 5 cm wide, and the sediment thatcomposed it appeared slightly finer than that which entered the pool. The bedload sheetnear the left bank emerged from the pool onto a mid-channel riffle. The coarse-surfacedriffle caused the sheet to diverge as the particles within the sheet moved through the riffle.The result of the riffle, diverging flow, and divergence of the sheet resulted in a width of15 cm. The coarse and finer sediments interacted with the coarse bed sediment as thebedload sheet moved through the riffle, however the majority of the sediment maintaineddownstream momentum. It was difficult to quantify in the video if the interaction of thebedload sheet with the riffle actually resulted in lower velocity for the upper bedload sheet,however, it did appear to. The two bedload sheets converged prior to entering the nextpool downstream.Recking et al. (2009) proposed a model of periodical bedload production, in whichaggradation on the bed continued until a critical slope was reached. Once the slope wasattained it results in increased gravel mobility and strong local erosion, producing a bedloadsheet. Here, sheet production was observed to occur during morphological scouring events.For example, in Figure 5.7, as the pool expanded laterally, the sediment from the erodingface, exited the pool as a bedload sheet. This suggests that sediment supply may be a more945.3. Experimental ObservationsPoolFlowFigure 5.7: Image showing two bedload sheets emerging from the same pool. As the sheetentered the pool it diverged into two separate sheets. The coarse leading fronts of the twosheets are indicated in the image. The arrows highlight the direction of movement of thebedload sheet.955.3. Experimental Observationscritical component of bedload sheet production. Even local sources such as pool scour, ora bank erosion event in a field setting, could result in bedload sheet production.Bedload sheets are commonly associated with fluctuations in bedload transport, pre-vious researchers have suggested that the large, short-term variations in sediment flux areassociated with bedload sheets (Nelson et al., 2009). Although, the resolution of the datapresented here were not collected at a scale fine enough to delineate individual bedloadsheet pulses in the output data, it is clear that morphological adjustments, scour and fill,were the predominant cause of large short-term variability in sediment flux.Bedload sheets have been categorized as bedforms (Whiting et al., 1988; Bennett andBridge, 1995a; Kuhnle et al., 2006) and/or patches (Nelson et al., 2009). Some haveobserved dunes developing from bedload sheets (Kuhnle and Southard , 1988; Whiting et al.,1988; Wilcock , 1992). In this experiment gravel bars developed from bedload sheets, thismay be similar to what Ashmore (1991) observed in his experiments. This suggests thatbedload sheets may be pseudo bedforms. In these experiments, bedload sheets appear tobe more of a sediment transport pathway, in which they develop in response to excess localsediment supply in order to move the excess sediment downstream in a rapid manner.5.3.3 Observations of sediment transportThe observations of sediment transport and bedload sheet movement during the experi-ment and in the videos, allowed visualization of sediment transport pathways. From theseobservations, inferences into flow and sediment transport through the channel were made.Models of flow and sediment transport and sorting through bends differ primarily in howthey account for the relative magnitude and importance of the components of the down-stream and cross-stream force balance in channels (Whiting , 1997). Whiting (1997) foundthat these distinctions were related to differences in either flow stage or magnitude of to-pographic relief, with the convective acceleration term being more important in channelswith greater topographic relief and lower stages. Topographic relief evolved during theseexperiments and the stage was relatively high; it is anticipated that aspects of both modelswill be observed.Figure 5.8 illustrates the flow processes and sediment pathways and sorting through ameander following Dietrich and Smith (1984). In the Bridge (1992) model the cross-streamcomponent of particle weight is balanced by the fluid drag from secondary circulation,suggesting a static balance of forces as the particle travels through the bend. For Bridge’smodel the coarse sediment follows the zone of maximum shear stress downstream throughthe bend, but no net cross-stream transport occurs, therefore particle size differentiationdoes not occur through the meander. Dietrich and Smith (1984) suggested that sedimenttransport through a meander is the result of a dynamic balance of forces. Shoaling flowfrom the bar causes outward sediment transport that forces the zone of maximum bedloadtransport rate to track the outward-shifting zone of maximum shear stress (Figure 5.8).The outward-directed flow over the bar balances the spatial variation in shear stress with965.3. Experimental Observationsthe convergent sediment transport (Powell , 1998). This implies that as coarse and fineparticles move through the meander they cross paths due to inward-acting secondary flow,which moves fine particles up the transverse bed slope, and the outward-acting gravitationalflow, which moves coarse particles down the transverse bed slope toward the pool.Both models were developed for equilibrium conditions, and thus assumptions of stablemorphology were made. In this experiment the pool at which sediment movement wasobserved was not in equilibrium and changed during the course of the experiment. Ad-ditionally, the morphology of the experimental channel is not equivalent to the meanderpattern described in either of the two models. Most importantly, the bar adjacent to thepool was a lateral bar and not a point bar, which may have implications for the magnitudeof the outward shoaling flow across the bar.If we can assume that observations of bedload sheet movement through a pool are repre-sentative of sediment pathways and sorting, then this experiment can offer new insight intothis process. From observations made during this experiment three particle pathways andsorting outcomes were identified. These outcomes appeared to depend upon the morphol-ogy of the pool and its hydraulic connection to upstream and downstream morphologicalunits. The three outcomes are:1. sediment entered the pool, travelled along the transverse bed slope and exited withlittle to no evidence of sorting;2. sediment entered the pool and exited pool with evidence of sorting; and,3. sediment entered the pool, travelled through the maximum pool depth and exitedpool with little evidence of sorting;The first item represents sediment movement through the pool in a manner similarto the scenario described by Bridge (1992). This outcome appeared to be predicatedby pool dimensions that created conditions where gravitational and convective forces werebalanced. Coarse particles were visually tracked as they entered the pool and they appearedto maintain a consistent elevation through the pool. The relation of the coarser grains tothe finer grains did not appear to be altered as the sheet travelled through the pool. Animportant factor appeared to be the angle at which the bedload sheet entered the pool.If the downstream trajectory of the sheet was parallel to the bar face and the overallorientation of the pool, its movement through the pool was not altered. This suggests thatthe morphology was conducive to the balance of forces between shoaling and secondaryflows.The second item represents sediment movement and sorting through the pool similarto that described by Dietrich and Smith (1983), where the shoaling induced outward flowand secondary flows sort the sediment particles as they moved through the pool. This wasobserved when the transverse bed slope of the bar was at a low angle and the pool wasgenerally wider and more fully developed. The bedload sheet shown in Figure 5.4d) is an975.3. Experimental ObservationsFigure 5.8: Flow processes controlling morphology and sediment sorting in a river meander.Flow direction is from the lower to upper end of the figure. From Dietrich and Smith (1984).985.4. Discussion and Conclusionsillustration of this. The cross-section of the pool at 2400 min can be seen in Figure 5.6a,the combination of a deep pool and relatively gentle transverse bed slope encouraged theoutward shoaling flow. The magnitude of the secondary flow circulation is also reflectedby the change in the downstream angle of the bedload sheet trajectory as it entered thepool relative to the angle it had when it entered the pool.Lastly, the bedload sheet entered the pool as a coherent line and travelled through thedeepest part of the pool and exited with very little sorting or alteration of its downstreamtrajectory. Figure 5.4c provides an example of this, the angle of the bedload sheet relativeto downstream flow is unaltered as it travels through the pool-bar unit. The cross-section at1800 minutes (Figure 5.6a) shows a deep pool with a steep transverse bed slope. The angleat which the bedload sheet entered the pool (Figure 5.4c) took the sediment through thecentre of the pool. In addition to the magnitude of the transverse bed slope, the orientationof the zone of maximum sediment transport relative to the downstream orientation of thepool appears to influence how sediment moves through the pool-bar unit.5.4 Discussion and ConclusionsA complex channel planform was used to replicate pool-riffle morphology in a mountaingravel-bed river. Experimental observations, hand drawn maps, still images and videoswere used to describe sediment transport processes and pathways during Exp. 6. The de-velopment of an experimental channel under conditions of constant discharge and sedimentfeed was discussed and linkages were made between sediment transport and the morpho-logical development of the channel. The width of active sediment transport was foundto be variable throughout the experiment, but generally declined overtime. The reduc-tion in width was related to increased lateral variability in the channel, which appearedto concentrate sediment movement into a smaller percentage of the channel width. Thisreduction in the width of active transport was not due to reduced sediment transport ashad been identified previously (for example, Lisle et al., 1993). The reduction suggests thatthe channel became more efficient in transporting sediment over time. Similar volumes ofmaterial were being moved through the channel, but it occurred over a relatively smallerportion of the channel. Channel cross sections evolved from a simple featureless plane-bedmorphology to a topographically complex system exhibiting characteristics of pool-rifflemorphology.Previously, observations of bedload sheets in laboratory settings have been mainly doneunder simplified morphologic conditions: this experiment allowed observations of bedloadsheet movement and production to be made in a topographically and planimetrically com-plex environment. Unfortunately, the frequency of sediment sampling at the output pre-vented direct measurement of variability in sediment output rates due to bedload sheetmigration.Flow and sediment transport characteristics were inferred from observations of bedload995.4. Discussion and Conclusionssheet movement in the experimental channel. Existing models of sediment movement andsorting through a channel are based on the assumption of equilibrium conditions, whichmay or may not occur in many gravel-bed rivers in glaciated regions. The pool-bar unitpresented here was not in equilibrium and changed over the course of the experiment.The development of the outward-directed shoaling flow and the inward-directed secondarycirculation were dependent on pool dimensions and the angle of the transverse bed slope.The angle of the transverse bed into the pool was an important topographic characteristicthat influenced both flow and sediment sorting. Additionally, the relative orientation ofthe incoming bedload sheet relative to the downstream orientation of the pool appeared toinfluence sediment movement and sorting through the pool.100Chapter 6ConclusionsThe aim of this research was to investigate long-term channel development and responseunder conditions of constant discharge and sediment supply. A Froude scaled physicalmodel was constructed using Fishtrap Creek, a gravel-bed stream in the interior of BritishColumbia, as a field prototype. Eight experiments were conducted holding sediment feedrate and water discharge constant within each experiment, but varying the magnitudesbetween experiments. A ninth experiment was conducted with no sediment feed and waterdischarge held constant. A final experiment was conducted to see how sediment mobilityresponded to increased discharge with no sediment feed. Design flows used in the exper-iments were based on conditions in the field prototype, corresponding to flow conditionshaving return periods ranging from the 2-yr to greater than 150-yr. As the purpose ofthe experiments were to investigate long-term channel response, a sediment-feed protocolwas employed. A novel approach to the design of the flume was undertaken in order tofocus the experiments on intermediate-sized gravel-bed rivers in forested environments. Anirregular meandering planform was built to reflect the planform observed in the prototype.The flume design successfully produced both pool-riffle and plane-bed morphologies, bothof which are present in the prototype.Specifically, this research investigated:• the effects of varying sediment supply and discharge on bed surface and transportedsediment characteristics;• the effects of varying sediment supply and discharge on channel storage and channelmorphology; and,• sediment transport processes and pathways.In Chapter 3 the effects on the characteristics of the bed surface and transportedsediment under differing regimes of discharge and sediment feed were investigated. Exper-iments focused on channel formative flows employing discharges ranging from the 2-yr toover 150-yr return period events. The results indicate that even with discharges exceedingthe 10-yr flood full mobility was not observed. This slight but persistent size-selectivityproduces long-term aggradation and surface coarsening, as the larger grain sizes are lefton the surface. This size-selectivity occurred when the sediment feed had the same grainsize distribution as the bed material. These results are compared with a reanalysis ofpreviously published data from a generic model of small gravel-bed streams with erodible101Chapter 6. Conclusionsbed and banks. Those channels exhibit a similar size-selectivity, which seems to be criticalfor maintaining stable channel banks, with full mobility potentially being related to a shiftfrom a stable single thread channel that does not aggrade to an unstable channel that doesaggrade over time.In Chapter 4 the effects of varying sediment supply and discharge on channel storageand morphology were explored. The results showed that sediment transport rates variedboth spatially and temporally. This variability was found to be more dependent uponchanges in the morphology of the channel, rather than adjustments in the surface grainsize distribution. The primary self-adjusting mechanism in the experimental channel wasto adjust channel form. Adjustments to bed surface texture, tended to be a short-termand immediate response to the high flows. The development of surface patches appearedto be in response to hydraulics associated with channel topography.The transport-storage relation originally presented by Lisle and Church (2002) andlater by Madej et al. (2009), Pryor et al. (2011) and Lisle (2012) employing declining orvariable sediment feeds rates was investigated. Similar to the conclusions of Lisle andChurch (2002), sediment transport capacity was found to be variable and that it changedas the volume of sediment stored in the channel changed. Madej et al. (2009) and Pryoret al. (2011) observed hysteresis cycles in the relationship between sediment transport rateand sediment storage, which they attributed to morphological responses to changes in thefeed rate. Unlike these studies, the analysis presented here indicates that these cyclesoccurred when transport rates were approximately equal to the feed rates (i.e., when thechannel was near equilibrium). Cycles of aggradation-degradation were observed to occurwithout changes in either sediment supply or flow discharge. It was concluded that thetransport-storage relation does not offer a unique relation between transport rates andsediment storage, thus limiting its usefulness in the development of an improved sedimenttransport model.The results suggest that the idea of sediment reservoirs is more complex than Lisle andChurch (2002) had originally envisioned and that these reservoirs are transitional. Addi-tionally, multiple reservoirs may exist within homogeneous reaches as they may scale downto the morphologic unit. Each reservoir may also have a have different time signature interms of how responsive they are to changes in either flow or sediment supply. This differ-ence is dependent upon flows, sediment supply, and local adjustment to the topography,which in addition to the previous parameters is dependent upon boundary conditions (suchas bank stability, bank form and alignment (important in most forested landscapes).For Chapter 5 a less formal approach was taken to describe the sediment transportprocesses and pathways observed during the experiments. As similar patterns emergedduring each experiment, only information from one of the experiments, Exp. 6, was usedfor this chapter. A variety of qualitative data sources were used, including experimentalnotes, hand drawn field maps, width measurements of sediment transport, colour imagesduring dewatered periods, and experimental videos.The dominance of bedload sheets as a process by which sediment was transported102Chapter 6. Conclusionsthrough the reach is one of the key findings in the chapter. Bedload sheets appear to bethe conveyor belts of sediment transport. Observations of the movement of bedload sheetsthrough a pool and riffle unit provided further insight in how they behave in more topo-graphically complex settings. For example, single bedload sheets were observed to enter apool and two bedload sheets emerged. The appearance of multiple sheets emerging fromthe pool was linked to an additional source of sediment within the pool itself; divergenceof the sheet as it exited the pool was influenced by the presence of a riffle, and the sheetsappeared to follow the margins of the riffles. When bedload sheets encountered rifflesdirectly, they would disaggregate as the particles moved through the coarser surface andreform into a sheet when downstream of the riffle.During periods of high sediment mobility the zone of active sediment transport ap-peared to occupy the entire cross section. During these periods the majority of sedimentwas being transported in bedload sheets, the movement individual particles was still ob-served outside of the areas occupied by bedload sheets. The area of the channel in whichactive transport was observed was not simply a function of sediment supply, increasedmorphological complexity. The development of a pool-riffle morphology confined sedimenttransport pathways to the lower elevation portions of the channel cross section.Ranges in bedload sheets migration velocities observed in Exp. 6 were found to besimilar to those of previous experiments (Iseya and Ikeda, 1987; Kuhnle and Southard ,1988; Madej et al., 2009; Recking et al., 2009). Bedload sheet production seems entirelydependent on sediment availability, which can also include in-channel scour events. Thenotion that bedload sheets are in themselves a bedform (Whiting et al., 1988; Bennett andBridge, 1995b; Kuhnle et al., 2006) or represent mobile patches (Nelson et al., 2009) needsto be revisited. The observations here suggest that they may actually be a how channelstransport sediment downstream, as they appeared to be a robust method to transportexcess sediment out of the reach more rapidly than normal transport processes.Observations of sediment movement through a pool-riffle unit confirmed aspects ofsediment sorting models of Dietrich and Smith (1984) and Bridge (1992). From theseexperiments the importance of pool dimensions, depth and transverse bed slope of the bar,and local channel hydraulics play an important role in determining not only how sedimententers a pool, but the path it takes as it moves through it and the effectiveness of sortingprocesses within it. Although not measured in these experiments, pool dimensions appearto determine the relative magnitude of secondary circulation to the primary flow direction,with stronger secondary circulation resulting in more sediment sorting as it moves throughthe pool. Additionally, the relative downstream velocity and angle of sediment enteringthe pool relative to the hydraulics within the pool influenced how or if sediment was sortedas it moved through.1036.1. Questions, Observations and Future Research6.1 Questions, Observations and Future ResearchThe experiments provided an opportunity to observe channel evolution under high flowconditions. Whether or not the experimental channels are representative of natural envi-ronments may be debatable, but they are channels in and of themselves, and their mor-phology evolves under the same physics as ”real” rivers.The following itemized list, not in any particular order, provides some additionalthoughts, research questions and approaches that emerged as this research was being con-ducted.• Channel stability in gravel-bed rivers is likely the result of an interaction between thechannel boundaries, armouring of the bed surface and morphological units present.Holding the channel boundaries constant, the relationship between the stability ofchannel units and the effects of armouring can be investigated in experiments withuniform and graded sediments.• The stability of the pool-riffle morphology in gravel-bed rivers may also represent anextension of the jammed state theory, similar to the application made to step-poolchannels by Church and Zimmermann (2007). In a jammed state within granularmaterial, individual grains form force chains that resist movement under directedforce due to grain-on-grain structural arrangements and/or strong frictional binding.For step-pool channels, the force chain is established across the channel. The relativescale of flow depth (D/d) to these grains in step-pool channels is small. In pool-rifflechannels the relative depth of water to individual grains is much greater. However,it is suggested that channel bars might behave in a similar manner; it is a matterof seeing the bar as a structural arrangement of grains. Bars jam up in the channelin which the stored sediment resists movement, in the sense that they create theirown hydraulic environment that can sustains their form over time. This hydraulicenvironment links bars adjacent in both the upstream and downstream directions.• Additional research on the role of topographic evolution within channels and itsassociated effect on sediment transport rates is suggested. It is believed that thetopographic signature of the channel plays a dominant role not only in channel hy-draulics, and the distribution of shear stresses, but on the development of in-channelsurface patches and sediment reservoirs. Linking physical models with numericalmodels is likely the most productive method to accomplish this task.• Improved techniques to investigate bedload sheet production, migration and depo-sition should be undertaken, Experiments specifically designed with this in mindshould be conducted. The design approach should utilize a more realistic channelplanform as the morphological complexity of the channel influences the behaviourof the sheets. Understanding this linkage is critical to further our understanding1046.1. Questions, Observations and Future Researchtheir behaviour. Questions to explore could include: What are the relative roles ofsediment supply, morphological complexity and surface sediment size in determiningbedload sheet migration velocity? Can bedload sheets develop in beds with uniformgrain sizes?• The same intensity of research that was applied to surface armouring should beapplied to channel morphologic units, such as bars and pools. This would likelyrequire numerical and physical models. The physical models should be designed withthis topic in mind, and therefore would require modified flume designs.• Additional research should be undertaken regarding the effects of Froude scaling onthe sediment size distribution. What are the implications of truncating the smallersize classes? What effect does the truncation have on channel morphodynamics?• The importance of the largest grains in determining channel morphodynamics hasbeen shown to be important in this research. The link between the larger grains,channel bedforms and channel stability in gravel-bed rivers merits additional research.• Further research into the variability of sediment supplied to a system and its influenceon channel responses. Depending on the geology and geologic history of a basin, achannel in mountainous areas can receive a wide range in the calibre of materialdelivered to the channel. Coupled numerical and physical model experiments couldbe set up to explore these effects.Research in gravel-bed rivers has been mainly focused on the grain scale for the last 30years, it is suggested that this scale may not be entirely appropriate for understanding themorphology of rivers at the scale of interest to most practitioners.105ReferencesAckers, P., and F. G. Charlton (1970), Meander geometry arising from varying flows,Journal of Hydrology, 11 (3), 230–252.Andrews, C. (2010), A stream in transition: short-term morphodynamics of Fishtrap Creekfollowing wildfire, Master’s thesis, The University of British Columbia.Andrews, E. D. (1983), Entrainment of gravel from naturally sorted riverbed material,Geological Society of America Bulletin, 94 (10), 1225–1231.Ashmore, P. (1991), Channel morphology and bed load pulses in braided, gravel-bedstreams, Geografiska Annaler. Series A. Physical Geography, 73 (1), 37–52.Ashworth, P., J. Best, J. Leddy, and G. Geehan (1994), The physical modelling of braidedrivers and deposition of fine-grained sediment, in Process models and theoretical geomor-phology, edited by M. Kirkby, pp. 115–139, John Wiley and Sons.Ashworth, P. J., and R. I. Ferguson (1989), Size-selective entrainment of bed load in gravelbed streams, Water Resources Research, 25 (4), 627–634.Barry, J. J., J. M. Buffington, and J. G. King (2004), A general power equation for pre-dicting bed load transport rates in gravel bed rivers, Water Resources Research, 40 (10),W10,401, doi:10.1029/2004WR003190.Bartholdy, J., and P. Billi (2002), Morphodynamics of a pseudomeandering gravel barreach, Geomorphology, 42 (3), 293–310.Bartley, R., and I. Rutherford (2005), Re-evaluation of the wave model as a tool for quan-tifying the geomorphic recovery potential of streams disturbed by sediment slugs, Geo-morphology, 64 (3), 221–242.Benda, L., and T. Dunne (1997a), Stochastic forcing of sediment routing and storage inchannel networks, Water Resources Research, 33 (12), 2865–2880.Benda, L., and T. Dunne (1997b), Stochastic forcing of sediment supply to channel net-works from landsliding and debris flow, Water Resources Research, 33 (12), 2849–2863.106ReferencesBennett, S., and J. Bridge (1995a), An experimental study of flow, bedload transportand bed topography under conditions of erosion and deposition and comparison withtheoretical models, Sedimentology, 42 (1), 117–146.Bennett, S., and J. Bridge (1995b), The geometry and dynamics of low-relief bed formsin heterogeneous sediment in a laboratory channel, and their relationship to water flowand sediment transport, Journal of Sedimentary Research, 65 (1), 29–39.Blench, T. (1969), Mobile-Bed Fluviology, The University of Alberta Press, Edmonton.Braudrick, C., W. Dietrich, G. Leverich, and L. Sklar (2009), Experimental evidence forthe conditions necessary to sustain meandering in coarse-bedded rivers, Proceedings ofthe National Academy of Sciences, 106 (40), 16,936–16,941.Bravo-Espinosa, M., W. Osterkamp, and V. Lopes (2003), Bedload transport in alluvialchannels, Journal of Hydraulic Engineering, 129 (10), 783–795.Bray, D. (1975), Representative discharges for gravel-bed rivers in Alberta, Canada, Jour-nal of Hydrology, 27 (1), 143–153.Brewer, P., and J. Lewin (1998), Planform cyclicity in unstable reach: complex fluvialresponse to environmental change, Earth Surface Processes and Landforms, 23 (11), 989–1008.Bridge, J. (1992), A revised model for water flow, sediment transport, bed topography andgrain size sorting in natural river bends, Water Resources Research, 28 (4), 999–1013.Brummer, C. J., T. B. Abbe, J. R. Sampson, and D. R. Montgomery (2006), Influenceof vertical channel change associated with wood accumulations on delineating channelmigration zones, Washington, USA, Geomorphology, 80 (3), 295–309.Buffington, J., and D. Montgomery (1999a), Effects of hydraulic roughness on surfacetextures of gravel-bed rivers, Water Resources Research, 35 (11), 3507–3521.Buffington, J. M., and D. R. Montgomery (1999b), Effects of sediment supply on surfacetextures of gravel-bed rivers, Water Resources Research, 35 (11), 3523–3530.Carling, P. (1987), Bed stability in gravel streams, with reference to stream regulation andecology, in River Channels: Environment and Process, edited by K. Richards, p. 321347, Blackwell, Oxford, UK.Carling, P. (1988), The concept of dominant discharge applied to two gravel-bed streams inrelation to channel stability thresholds, Earth Surface Processes and Landforms, 13 (4),355–367.107ReferencesCarson, M. A., and G. A. Griffiths (1987), Bedload transport in gravel channels, Journalof Hydrology (NZ), 26 (1), 1–151.Chorley, R., and B. Kennedy (1971), Physical Geography - A Systems Approach, 370 pp.,Prentice Hall International Inc.Church, M. (1992), Channel morphology and typology, in The Rivers Handbook, edited byP. Callow and G. Petts, pp. 126–143, Blackwell Science.Church, M. (1995), Geomorphic response to river flow regulation: case studies and time-scales, Regulated Rivers Research and Management, 11 (1), 3–22.Church, M. (1998), Think globally, learn locally: broadening prespectives of the earth,Geoscience Canada, 25 (4), 171–182.Church, M. (2006), Bed material transport and the morphology of alluvial river channels,Ann. Rev. Earth Planet. Sci., 34, 325–354.Church, M., and M. Hassan (2005), Upland gravel-bed rivers with low sediment transport,in Catchment Dynamics and River Processes. Mediterranean and Other Climate Regions,edited by C. Garcia and R. Batalia, pp. 141–168, Elsevier.Church, M., and M. A. Hassan (2002), Mobility of bed material in harris creek, WaterResources Research, 38 (11), 1237, doi:10.1029/2001WR000753.Church, M., and D. Jones (1982), Channel bars in gravel-bed rivers, in Gravel-bed Rivers,edited by R. Hey, J. Bathurst, and C. Thorne, pp. 291–338, John Wiley & Sons Ltd.Church, M., and A. Zimmermann (2007), Form and stability of step-pool channels: Re-search progress, Water Resources Res, 43, W03,415, doi:10.1029/2006WR005037.Church, M., M. A. Hassan, and J. F. Wolcott (1998), Stabilizing self-organized structuresin gravel-bed stream channels: Field and experimental observations, Water ResourcesResearch, 34 (11), 3169–3179.Cudden, J. R., and T. B. Hoey (2003), The causes of bedload pulses in a gravel channel: theimplications of bedload grain-size distributions, Earth Surface Processes and Landforms,28 (13), 1411–1428.Cui, Y., and G. Parker (2005), Numerical model of sediment pulses and sediment-supplydisturbances in mountain rivers, Journal of Hydraulic Engineering, 131 (8), 646–656.Cui, Y. T., G. Parker, T. E. Lisle, J. Gott, M. E. Hansler-Ball, J. E. Pizzuto, N. E. All-mendinger, and J. M. Reed (2003), Sediment pulses in mountain rivers: 1. Experiments,Water Resources Research, 39 (9), 1239, doi:10.1029/2002WR001803.108ReferencesDavies, T. R. H., and A. I. Lee (1988), Physical hydraulic modelling of width reductionand bed level change in braided rivers, Journal of Hydrology (NZ), 27 (2), 113–127.Dietrich, W., and J. Smith (1984), Bed load transport in a river meander, Water ResourcesResearch, 20 (10), 1355–1380.Dietrich, W., P. Nelson, E. Yager, J. Venditti, M. Lamb, and L. Collins (2005), Sedimentpatches, sediment supply, and channel morphology, in River, Coastal and EstuarineMorphodynamics: RCEM 2005, edited by G. Parker and M. Garcia, pp. 79–90.Dietrich, W. E., and J. D. Smith (1983), Processes controlling the equilibrium bed mor-phology in river meanders, in River Meandering, pp. 759–769, ASCE.Dietrich, W. E., J. W. Kirchner, H. Ikeda, and F. Iseya (1989), Sediment supply andthe development of the coarse surface layer in gravel-bedded rivers, Nature, 340 (6230),215–217.Diplas, P., and J. B. Fripp (1992), Properties of various sediment sampling procedures,Journal of Hydraulic Engineering, ASCE, 118 (7), 955–970.Drake, T., R. Shreve, W. Dietrich, P. Whiting, and L. Leopold (1988), Bedload transportof fine gravel observed by motion-picture photography, Journal of Fluid Mechanics, 192,193–217.Eaton, B. (2013), Hydraulic geometry: empirical investigations and theoretical approaches,in Treatise on Geomorphology, vol. 9, edited by J. Shroder and E. Wohl, Academic Press.Eaton, B. C., and M. Church (2004), A graded stream response relation for bedload dom-inated streams, Journal of Geophysical Research - Earth Surface, 109 (F03011), doi:10.1029/2003JF000,062.Eaton, B. C., and M. Church (2009), Channel stability in bed load-dominated streamswith nonerodible banks: Inferences from experiments in a sinuous flume, Journal ofGeophysical Research - Earth Surface, 114 (F01024), doi: 10.1029/2007JF000,902.Eaton, B. C., and T. R. Giles (2009), Assessing the effect of vegetation-related bankstrength on channel morphology and stability in gravel bed streams using numericalmodels, Earth Surface Processes and Landforms, 34 (5), 712–714.Eaton, B. C., and M. F. Lapointe (2001), Effects of large floods on sediment transport andreach morphology in the cobble-bed Sainte Marguerite River, Geomorphology, 40 (3–4),291–309.Eaton, B. C., M. Church, and R. G. Millar (2004), Rational regime model of alluvial channelmorphology and response, Earth Surface Processes and Landforms, 29 (4), 511–529.109ReferencesEaton, B. C., C. A. E. Andrews, T. R. Giles, and J. C. Phillips (2010a), Wildfire, morpho-logic change and bed material transport at Fishtrap Creek, British Columbia, Geomor-phology, 118, 409–424.Eaton, B. C., R. D. Moore, and T. R. Giles (2010b), Forest fire, bank strength and channelinstability: the ’unusual’ response of Fishtrap Creek, British Columbia, Earth SurfaceProcesses and Landforms, 35 (10), 1167–1183.Ehrenberger, R. (1931), Direct bedload measurements at the Danube near Vienna andtheir results to date, Die Wasserwirtschaft, 34, 1–9.Einstein, H. A. (1937), The calibration of the bedload trap used in the rhine,Schweuzerische Bauzeitung, 110, 29–32.Emmett, W. W. (1976), Bedload transport in two large, gravel-bed rivers, Idaho andWashington, in Proceedings of the 3rd Intra-Agency Sedimentation Conference, pp. 4–101 – 4–114.Ettema, R., R. Arndt, P. Roberts, and T. Wahl (2000), Hydraulic modeling: concepts andpractice, no. 97 in Task Committee on Hydraulic Modeling, Environmental and WaterResources Institute, manual of practice no. 97 ed., 390 pp., ASCE Publication.Ferguson, R. I. (2003), The missing dimension: effects of lateral variation on 1-d calcula-tions of fluvial bedload transport, Geomorphology, 56 (1-2), 1–14.Francalanci, S., S. Solair, M. Toffolon, and G. Parker (2012), Do alternate bars affectsediment transport and flow-resistance in gravel-bed rivers?, Earth Surface Processesand Landforms, 37 (8), 866–875.Friedkin, J. F. (1945), A laboratory study of the meandering of alluvial rivers, Tech. rep.,US Waterways Engineering Experimental Station.Fripp, J., and P. Diplas (1993), Surface sampling in gravel streams, Journal of HydraulicEngineering, 119 (4), 473–490.Gaeumann, D., J. Schmidt, and P. Wilcock (2005), Complex channel responses to changesin stream flow and sediment supply on the lower Duchesne River, Utah, Geomorphology,64 (3-4), 185–206.Gilbert, G. (1880), Report of the Geology of the Henry Mountains, 170 pp., US GovernmentPrinting Office.Gilbert, G. K. (1914), The transport of debris by running water, no. 86 in U.S. GeologicalSurvey Professional Paper 105, 154 pp., U.S. Geological Survey.110ReferencesGintz, D., M. Hassan, and K. Schmidt (1996), Frequency and magnitude of bedload trans-port in a mountain river, Earth Surface Processes and Landforms, 21 (5), 433–446.Gomez, B. (1983), Temporal variations in the particle size distribution of surficial bedmaterial: the effect of progressive bed armouring, Geografiska Annaler., 65 (3–4), 183–192.Gomez, B., and M. Church (1989), An assessment of bed load sediment transport formulaefor gravel bed rivers, Water Resources Research, 25 (6), 1161–1186.Gomi, T., and R. Sidle (2003), Bed load transport in managed steep-gradient head-water streams of southeastern Alaska, Water Resources Research, 39 (12), 1336, doi:10.1029/2003WR002440.Gran, K., and D. Montgomery (2005), Spatial and temporal patterns in fluvial recoveryfollowing volcanic eruptions: Channel response to basin-wide sediment loading at MountPinatubo, Philippines., Geological Society of America Bulletin, 117 (1–2), 195–211.Gran, K., D. Montgomery, and D. Sutherland (2006), Channel bed evolution and sedimenttransport under declining sand inputs, Water Resources Research, 42 (10), W10,407,doi:10.1029/2005WR004306.Haschenburger, J., and P. Wilcock (2003), Partial transport in a natural gravel bed channel,Water Resources Research, 39 (1), 1020–1029, doi:10.1029/2002WR001532.Hassan, M., R. Egozi, and G. Parker (2006), Experiments on the effect of hydrographcharacteristics on vertical grain sorting in gravel bed rivers, Water Resources Research,42 (9), doi:10.1029/2005WR004707.Hassan, M., B. Smith, D. Hogan, D. Luzi, A. Zimmermann, and B. Eaton (2007), Sedimentstorage and transport in coarse bed streams: scale considerations, in Gravel-Bed RiversVI: From Process Understanding to River Restoration, vol. 11, edited by H. Habersack,H. Piegay, and M. Rinaldi, pp. 473–496, Elsevier.Hassan, M. A., and M. Church (2000), Experiments on surface structure and partial sedi-ment transport on a gravel bed, Water Resources Research, 36 (7), 1885–1895.Hassan, M. A., and M. Church (2001), Sensitivity of bed load transport in harris creek:Seasonal and spatial variation over a cobble-gravel bar, Water Resources Research, 37 (3),813–825.Hassan, M. A., M. Church, T. E. Lisle, F. Brardinoni, L. Benda, and G. E. Grant (2005),Sediment transport and channel morphology of small, forested streams, Journal of theAmerican Water Resources Association, 41 (4), 853–876.111ReferencesHein, F., and R. Walker (1976), Bar evolution and development of stratification in thegravelly, braided, Kicking Horse River, British Columbia, Canadian Journal of EarthScience, 14 (4), 562–570.Hoey, T. (1992), Temporal variations in bedload transport rates and sediment storage ingravel-bed rivers, Progress in Physical Geography, 16 (3), 319–338.Hoey, T., and A. Sutherland (1991), Channel morphology and bedload pulses in braidedrivers: a laboratory study, Earth Surface Processes and Landforms, 16 (5), 447–462.Hooke, J. M. (2007), Spatial variability, mechanism and propagation of change in an activemeandering river, Geomorphology, 84 (3), 277–296.Iseya, F., and H. Ikeda (1987), Pulsations in bedload transport rates induced by a lon-gitudinal sediment sorting: a flume study using sand and gravel mixtures, GeografiskaAnnaler. Series A. Physical Geography, 69 (1), 15–27.Jackson, W. L., and R. L. Beschta (1982), A model of two-phase bedload transport in anoregon coast range stream, Earth Surface Processes and Landforms, 7 (6), 517–527.Jackson, W. L., and R. L. Beschta (1984), Influences of increased sand delivery on themorphology of sand and gravel channels, Water Resources Bulletin, 20 (4), 527–533.Jones, C. (1977), Effects of varying discharge regimes on bed-form sedimentary structuresin modern rivers, Geology, 5 (9), 567–570.Jones, L. S., and N. F. Humphrey (1997), Weathering-controlled abrasion in a coarse-grained meandering reach of the Rio Grande: implications for the rock record, GeologicalSociety of America Bulletin, 109 (9), 1080–1088.Kasai, M., T. Manutani, and G. Brierley (2004), Channel bed adjustments following majoraggradation in a steep headwater setting: Findings from Oyabu Creek, Kyushu, Japan,Geomorphology, 14 (4), 199–215.Kellerhals, R., and M. Church (1989), The morphology of large rivers: characterizationand management, Can. Spec. Publsl Fish. Aquat. Sci, 106, 31–48.Kennedy, R. G. (1895), The prevention of silting in irrigation canals, in Proc. Inst. of Civ.Engrs., vol. 119, pp. 281–290.Kirchner, J., W. Dietrich, F. Iseya, and H. Ikeda (1990), The variability of critical shearstress, friction angle, and grain protrusion in water-worked sediments, Sedimentology,37 (4), 647–672.Komar, P. D. (1987), Selective grain entrainment by a current from a bed of mixed sizes:a reanalysis, Journal of Sedimentary Petrology, 57 (2), 203–211.112ReferencesKuhnle, R. (1996), Unsteady transport of sand and gravel mixtures, in Advances in FluvialDynamics and Stratigraphy, edited by P. Carling and M. Dawson, pp. 183–201, JohnWiley.Kuhnle, R., and J. Willis (1998), Statistics of sediment transport in Goodwin Creek, Jour-nal of Hydraulic Engineering, 124 (11), 1109–1114.Kuhnle, R., J. Horton, S. Bennett, and J. Best (2006), Bed forms in bimodal-sand-gravelsediments: laboratory and field analysis, Sedimentology, 53 (3), 631–654.Kuhnle, R. A., and J. B. Southard (1988), Bed load transport fluctuations in a gravel bedlaboratory channel, Water Resources Research, 24 (2), 247–260.Kuhnle, R. A., and J. B. Southard (1990), Flume experiments on the transport of heavyminerals in gravel-bed streams, Journal of Sedimentary Research, 60 (5), 687–696.Lacey, G. (1930), Stable channels in alluvium, Proceedings of the Institution of Civil Engi-neers, 229, 259–384.Lane, E. (1937), Stable channels in erodible material, Transactions of the American Societyof Civil Engineers, 63, 123–142.Lane, E. (1955), Design of stable channels, American Society of Civil Engineers, Transac-tions, 120, 1234–1260.Lane, S. N., and K. S. Richards (1997), Linking river channel form and process: time,space and causality revisited, Earth Surface Processes and Landforms, 22 (3), 249–260.Laronne, J., and M. Carson (1976), Interrelationships between bed morpholog and bed-material transport for a small gravel-bed channel, Sedimentology, 23 (1), 67–85.Lenzi, M., V. D’Agostino, and P. Billi (1999), Bedload transport in the instrumented catch-ment of the Rio Cordon: Part i: Analysis of bedload records, conditions and thresholdof bedload entrainment, Catena, 36 (6), 171–190.Leopold, L. B., and T. Maddock (1953), The hydraulic geometry of stream channels andsome physiographic implications, U.S. Geol. Surv. Prof. Pap., 252.Lewin, J. (1976), Initiation of bed forms and meanders in coarse-grained sediment, Geo-logical Society of America Bulletin, 87 (2), 281–285.Lisle, T. (1995), Particle size variations between bed load and bed material in naturalgravel bed channels, Water Resources Research, 31 (4), 1107–1118.Lisle, T. (2012), Transport capacity, bedrock exposure, and process domains, in Gravel-bedRivers: Processes, Tools, and Environments, edited by M. Church, P. M. Biron, andR. A. G., first ed., John Wiley & Sons Ltd.113ReferencesLisle, T., F. Iseya, and H. Ikeda (1993), Response of a channel with alternate bars toa decrease in supply of mixed-size bed load: A flume experiment, Water ResourcesResearch, 29 (11), 3623–3629.Lisle, T., J. Pizzuto, H. Ikeda, F. Iseya, and Y. Kodama (1997), Evolution of a sedimentwave in an experimental channel, Water Resources Research, 33 (8), 1971–1981.Lisle, T. E. (1982), Effects of aggradation and degradation on riffle-pool morphology in nat-ural gravel channels, northwestern California, Water Resources Research, 18 (6), 1643–1651.Lisle, T. E. (1986), Stabilization of a gravel channel by large streamside obstructions andbedrock bends, Jacoby Creek, northwestern California, Water Resources Research, 18,1643–1651.Lisle, T. E. (1989), Sediment transport and resulting deposition in spawning gravels, northcoastal California, Water Resources Research, 25 (6), 1303–1319.Lisle, T. E., and M. Church (2002), Sediment transport-storage relations for de-grading, gravel bed channels, Water Resources Research, 38 (11), 1219, doi:10.1029/2001WR001086.Lisle, T. E., and S. Hilton (1992), The volume of fine sediment in pools - an index ofsediment supply in gravel-bed streams, Water Resources Bulletin, 28 (2), 371–383.Lisle, T. E., H. Ikeda, and F. Iseya (1991), Formation of stationary alternate bars in asteep chanel with mixed-size bed load: A flume experiment, Earth Surface Processesand Landforms, 16 (5), 463–469.Lisle, T. E., J. M. Nelson, J. Pitlick, M. A. Madej, and B. L. Barkett (2000), Variability ofbed mobility in natural, gravel-bed channels and adjustments to sediment load at localand reach scales, Water Resources Research, 36 (12), 3743–3755.Mackin, J. H. (1948), Concept of the graded river, Geological Society of America Bulletin,59 (5), 463 – 512.Madej, M. (2001), Development of channel organization and roughness following sedimentpulses in single-thread, gravel bed rivers, Water Resources Research, 37 (8), 2259–2272.Madej, M., and V. Ozaki (1996), Channel response to sediment wave propagation and move-ment, redwood creek, california, usa, Earth Surface Processes and Landforms, 21 (10),911–927.Madej, M. A., D. Sutherland, T. E. Lisle, and B. Pryor (2009), Channel responses tovarying sediment input: a flume experiment modeled after Redwood Creek, California,Geomorphology, 103 (4), 507–591.114ReferencesMcNamara, J., J. Oatley, D. Kane, and L. Hinzman (2008), Case study of a large summerflood on the north slope of Alaska: bedload transport, Hydrology Research, 39 (4), 299–308.Millar, R. G. (2000), Influence of bank vegetation on alluvial channel patterns, WaterResources Research, 36 (4), 1109–1118.Millar, R. G., and M. C. Quick (1993), Effect of bank stability on geometry of gravel rivers,Journal of Hydraulic Engineering-ASCE, 119 (12), 1343–1363.Milne, A. (1928), The House at Pooh Corner, 1988 print ed., Dutton’s Children Books.Montgomery, D., J. Buffington, R. Smith, K. Schmidt, and G. Pess (1995), Pool spacingin forest channels, Water Resources Research, 31 (4), 1097–1105.Montgomery, D. R., and J. M. Buffington (1997), Channel-reach morphology in mountaindrainage basins, Geological Society of America Bulletin, 109 (5), 596–611.Montgomery, D. R., and J. M. Buffington (1998), Channel processes, classification, andresponse, in River Ecology and Management, edited by R. Naiman and R. Bilby, pp.13–42, Springer-Verlag, New York.Moreton, D., P. Ashworth, and J. Best (2002), The physical scale modeling of braidedalluvial architecture and estimation of subsurface permeability, Basin Research, 14 (3),265–285.Nanson, G. C. (1974), Bedload and suspended-load transport in a small, steep, mountainstream, American Journal of Science, 274 (5), 471–486.Nash, D. (1994), Sediment-transporting discharge from magnitude-frequency analysis,Journal of Geology, 102 (1), 79–95.Nelson, J. M. (1988), Mechanics of flow and sediment transport over non-uniform erodiblebeds, Ph.D. thesis, University of Washington.Nelson, P., J. G. Venditti, W. E. Dietrich, J. W. Kirchner, H. Ikeda, F. Iseya, and L. Sklar(2009), Response of bed surface patchiness to reductions in sediment supply, Journal ofGeophysical Research - Earth Surface, 114 (F02005), doi:10.1029/2008JF001,144.Nelson, P., W. Dietrich, and J. Venditti (2010), Bed topography and the developmentof forced bed surface patches, Journal of Geophysical Research, 115, F04,024, doi:10.1029/2010JF001747.Nitsche, M., D. Rickenmann, J. Turowski, A. Badoux, and J. Kirchner (2011), Evaluationof bedload transp[ort predictions using flow resistance equations to account for macro-roughness in steep mountain streams, Water Resources Research, 47, W08,513, doi:10.1029/2011WR010645.115ReferencesNordin, C. F. (1985), The sediment loads of rivers, in Facets of hydrology, edited by J. C.Rodda, pp. 183–204, Wiley.Nunnally, N. R. (1985), Application of fluvial relationships to planning and design ofchannel modifications, Environmental Management, 9 (5), 417–426.Paola, C., and R. Seal (1995), Grain-size patchiness as a cause of selective deposition anddownstream fining, Water Resources Research, 31 (5), 1395–1407.Papanicolaou, A., and A. Schuyler (2003), Cluster evolution and flow-frictional characteris-tics under difference sediment availablilities and specific gravity, Journal of EngineeringMechanics, 129 (10), 1206–1219, doi:10.1061/(ASC)0733-9399(2003)129:10(1206).Parker, G. (1990), Surface-based bedload transport relation for gravel rivers, Journal ofHydraulic Research, 28 (4), 417–436.Parker, G., and P. C. Klingeman (1982), On why gravel bed rivers are paved, WaterResources Research, 18 (5), 1409–1423.Parker, G., and A. W. Peterson (1980), Bar resistance of gravel-bed streams, Journal of theHydraulics Division-ASCE, 106 (Hy10), 1559–1575, american Society of Civil Engineers.New York, NY, United States. 1980. Serial.Parker, G., and C. M. Toro-Escobar (2002), Equal mobility of gravel in streams: The re-mains of the day, Water Resources Research, 38 (11), 1264, doi:10.1029/2001WR000669.Parker, G., S. Dhamotharan, and H. Stefan (1982), Model experiments on mobile, pavedgravel bed streams, Water Resources Research, 18 (5), 1395–1408.Parker, G., C. M. Toro-Escobar, M. Ramey, and S. Beck (2003), Effect of floodwaterextraction on mountain stream morphology, Journal of Hydraulic Engineering, 129 (11),885–895.Parker, G., M. A. Hassan, and P. Wilcock (2007a), Adjustment of the bed surface sizedistribution of gravel-bed rivers in response to cycled hydrographs, in Gravel Bed Rivers6: From Process Understanding to River Restoration, vol. 11, edited by H. Habersack,H. Piegay, and M. Rinaldi, pp. 241 – 290, Elsevier.Parker, G., P. Wilcock, C. Paola, W. Dietrich, and J. Pitlick (2007b), Physical ba-sis for quasi-universal relations describing bankfull hydraulic geometry of single-threadgravel bed rivers, Journal of Geophysical Research - Earth Surface, 112, F04,005, doi:10.1029/2006JF000549.116ReferencesPeakall, J., P. Ashworth, and J. L. Best (1996), Physical modelling in fluvial geomorphol-ogy: principles applications and unresolved issues, in The scientific nature of geomor-phology: Proceedings of the 27th Binghamton Symposium in Geomorphology, edited byB. Rhoads and C. Thorn, pp. 221–253, John Wiley & Sons Ltd.Phillips, J. (2007), Post-fire dynamics of a gravel bed stream: Fishtrap Creek, BritishColumbia, Msc, The University of British Columbia.Pickup, G., and R. F. Warner (1976), Effects of hydrologic regime on magnitude andfrequency of dominant discharge, Journal of Hydrology, 29 (1/2), 51–75.Pilon, P., and D. Harvey (1993), CFA - Consolidated Frequency Analysis version 3.1,Environment Canada, Surveys and Information Systems Branch.Pitlick, J., E. Mueller, C. Segura, R. Cress, and M. Torizzo (2008), Relation betweenflow, surface-layer armoring and sediment transport in gravel-bed rivers, Earth SurfaceProcesses and Landforms, 33 (8), 1192–1209.Pizzuto, J. (1994), Channel adjustments to changing discharges, Powder River, Montana,GSA, 106 (11), 1494–1501.Powell, D. M. (1998), Patterns and processes of sediment sorting in gravel-bed rivers,Progress in Physical Geography, 22 (1), 1–31.Powell, M., I. Reid, and J. Laronne (2001), Evolution of bed load grain size distributionwith increasing flow strength and the effect of flow duration on the caliber of bed loadsediment yield in ephemeral gravel bed rivers, Water Resources Research, 37 (5), 1463–1474.Prestegaard, K. (1987), Sediment tarnsport and depostion zones in braided streams, inErosion and Sedimentation in the Pacific Rim, no. 165 in IAHS Publ., pp. 411–412.Pryor, B., T. Lisle, D. Montoya, and S. Hilton (2011), Transport and storage of bed materialin a gravel-bed channel during episodes of aggradation and degradation: a field and flumestudy, Earth Surface Processes and Landforms, 36, 2028–2041, doi:10.1002/esp.2224.Recking, A. (2010), A comparison between flume and field bed load transport data andconsequences fro surface-based bed load transport prediction, Water Resources Research,46, WO3518, doi:10.1029/2009WR008007.Recking, A., P. Frey, A. Paquier, and P. Belleudy (2009), An experimental investigation ofmechanisms involved in bed load sheet production and migration, Journal of GeophysicalResearch, 114, F03,010, doi:10.1029/2008JF000990.117ReferencesReid, I., L. Frostick, and J. Layman (1985), The incidence and nature of bedload trans-port during flood flows in coarse-grained alluvial channels, Earth Surface Processes andLandforms, 10, 33–44.Robert, A. (2003), River Processes: An Introduction to Fluvial Dynamics, 214 pp., Arnold.Ryan, S., L. Porth, and C. Troendle (2005), Coarse sediment transport in mountain streamsin Colorado and Wyoming, USA, Earth Surface Processes and Landforms, 30, 269–288.Schuerch, P., A. Densmore, B. McArdell, and P. Molnar (2006), The influence of landslidingon sediment supply and channel change in a steep mountain catchment, Geomorphology,78 (3), 222–235.Schumm, S. (1977), The Fluvial System, 338 pp., John WIley & Sons.Schumm, S., and R. Lichty (1965), Time, space, and causality in geomorphology, AmericanJournal of Science, 263, 110–119.Schumm, S. A. (1969), River metamorphosis, Journal of the Hydraulics Division-ASCE,95 (HY1), 255–273.Schumm, S. A., and H. R. Khan (1972), Experimental study of channel patterns, GeologicalSociety of America Bulletin, 83, 1755–1770.Seminara, G. (1998), Stability and morphodynamics, Meccanica, 33, 59–99.Seminara, G., M. Colombini, and G. Parker (1996), Nearly pure sorting waves and forma-tion of bedload sheets, Journal of Fluid Mechanics, 312, 253–278.Shvidchenko, A., and Z. D. Kopaliani (1998), Hydraulic modeling of bed load transport ingravel-bed Laba River, Journal of Hydraulic Engineering, 124 (8), 778–785.Simon, A. (1992), Energy, time, and channel evolution in catastrophically disturbed fluvialsystems, Geomorphology, 5 (3), 345–372.Simon, A., and C. Hupp (1986), Channel evolution in modified Tennessee channels, inProceedings of the Fourth Federal Interagency Sedimentation Conference: March 24-27,1986, Las Vegas, Nevada, vol. 2, pp. 71–82, Subcommittee on Sedimentation of theInteragency Advisory Committee on Water Data, US Government Printing Office.Sklar, L., W. E. Dietrich, E. Foufoula-Georgiou, B. Lashermes, and D. Bellugi (2006), Dogravel bed river size distributions record channel network structure?, Water ResourcesResearch, 42 (W06D18), doi:10.1029/2006WR005,035.Smith, B. J. (2004), Relations between bed material transport and storage during aggrada-tion and degradation in a gravel bed channel, Master’s thesis, Humboldt State University.118ReferencesSmith, N. (1974), Sedimentology and bar formation in the Upper Kicking Horse river, abraided outwash stream, The Journal of Geology, 82 (2), 205–223.Tal, M., and C. Paola (2007), Dynamic single-thread channels maintained by the interactionof flow and vegetation, Geology, 35 (4), 347–350.Talbot, T., and M. Lapointe (2002), Modes of response of a gravel bed river to mean-der straightening: The case of the Sainte-Marguerite River, Saguenay Region, Quebec,Canada, Water Resources Research, 38 (6), 1073, doi:10.1029/2001wr000324.Thompson, C., and J. Croke (2008), Channel flow competence and sediment transportin upland streams in southeast austalia, Earth Surface Processes and Landforms, 33,329–352.Venditti, J., W. Dietrich, P. Nelson, M. Wydzga, J. Fadde, and S. L. (2010), Effect ofsediment pulse grain size on sediment transport rates and bed mobility in gravel bedrivers, Journal of Geophysical Research, 115, 2003–2012, doi:10.1029/2009JF001418.Waldon, M. (2004), Estimation of average stream velocity, Journal of Hydraulic Engineer-ing, 130 (11), 1119–1122.Warburton, J. (1992), Observations of bed load transport and channel bed changes in aproglacial mountain stream, Arctic snd Alpine Research, 24 (3), 195–203.Warburton, J., and T. Davies (1994), Variability of bedload transport and channel mor-phology in a braided river hydraulic model, Earth Surface Processes and Landforms, 19,403–421.Wathen, S., R. Ferguson, T. Hoey, and A. Werritty (1995), Unequal mobility of gravel andsand in weakly bimodal river sediments, Water Resources Research, 31 (8), 2087–2096.Whiting, P., W. Dietrich, L. Leopold, and L. Collins (1985), The variability of sedimenttransport in a fine-gravel stream (abs), in International Fluvial Sedimentology Confer-ence, 3rd, p. 38, Colorado State University Press.Whiting, P., W. Dietrich, L. Leopold, T. Drake, and R. Shreve (1988), Bedload sheets inheterogeneous sediment, Geology, 16 (2), 105–108.Whiting, P., J. Stamm, D. Moog, and R. Orndorff (1999), Sediment-transporting flows inheadwater streams, GSA Bulletin, 111 (3), 450–466.Whiting, P. J. (1997), The effect of stage on flow and components of the local force balance,Earth Surface Processes and Landforms, 22, 517–530.119ReferencesWilcock, P. (1992), Experimental investigation of the effect of mixture properties on trans-port dynamics, in Dynamics of Gravel-bed Rivers, edited by P. Billi, R. Hey, C. Thorne,and P. Tacconi, pp. 109–139, John Wiley & Sons Ltd.Wilcock, P. (2001a), Toward a practical method for estimating sediment-transport ratesin gravel-bed rivers, Earth Surface Processes and Landforms, 26 (13), 1395–1408, doi:10.1002/esp.301.Wilcock, P. (2001b), The flow, the bed, and the transport: interaction in flume and field,in Gravel-bed Rivers V, edited by M. Mosley, pp. 181–220, New Zealand HydrologicalSociety, Wellington, New Zealand.Wilcock, P., and B. McArdell (1993), Surface-based fractional transport rates: mobilizationthresholds and partial transport of a sand-gravel sediment, Water Resources Research,29 (4), 1297–1312.Wilcock, P., and B. McArdell (1997), Partial transport of a sand/gravel sediment, WaterResources Research, 33 (1), 235–245.Wilcock, P., and J. Southard (1989), Bed load transport of mixed size sediment: fractionaltransport rates, bed forms, and the development of a coarse bed surface layer, WaterResources Research, 257, 1629–1641.Wilcock, P., S. Kenworthy, and J. Crowe (2001), Experimental study of the transport ofmixed sand and gravel, Water Resources Research, 37 (12), 3349–3358.Wittenberg, L., and M. D. Newson (2005), Particle clusters in gravel-bed rivers: an exper-imental morphological approach to bed material transport and stability concepts, EarthSurface Processes and Landforms, 30, 1351–1368.Wolman, M. G., and J. P. Miller (1960), Magnitude and frequency of forces in geomorphicprocesses, Journal of Geology, 68, 54 – 74.Wong, M., and G. Parker (2006), One-dimensional modeling of bed evolution in a gravelbed river subject to a cycled flood hydrograph, Journal of Geophysical Research, 111,doi:10.1029/2006JF000478.Wooldridge, C., and E. Hickin (2005), Radar architecture and evolution of channel barsin wandering gravel-bed rivers: Fraser and Squamish Rivers, British Columbia, Canada,Journal of Sedimentary Research, 75, 844–860.Yalin, S. M. (1971), On the formation of dunes and meanders, in Proceedings of the 14thCongress of the International Association of Hydraulic Research, Paper C13, pp. 1–8.Young, W., and J. Warburton (1996), Principles and practice of hydraulic modeling ofbraided gravel-bed rivers, Journal of Hydrology, New Zealand, 35 (2), 175–198.120


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items