The partitioning of the total sediment load of a river into suspended load and bedload: a review of empirical data

The partitioning of the total sediment load of a river into suspended load and bedload is an important problem in fluvial geomorphology, sedimentation engineering and sedimentology. Bedload transport rates are notoriously hard to measure and, at many sites, only suspended load data are available. Often the bedload fraction is estimated with ‘rule of thumb’ methods such as Maddock’s Table, which are inadequately field‐tested. Here, the partitioning of sediment load for the Pitzbach is discussed, an Austrian mountain stream for which high temporal resolution data on both bedload and suspended load are available. The available data show large scatter on all scales. The fraction of the total load transported in suspension may vary between zero and one at the Pitzbach, while its average decreases with rising discharge (i.e. bedload transport is more important during floods). Existing data on short‐term and long‐term partitioning is reviewed and an empirical equation to estimate bedload transport rates from measured suspended load transport rates is suggested. The partitioning averaged over a flood can vary strongly from event to event. Similar variations may occur in the year‐to‐year averages. Using published simultaneous short‐term field measurements of bedload and suspended load transport rates, Maddock’s Table is reviewed and updated. Long‐term average partitioning could be a function of the catchment geology, the fraction of the catchment covered by glaciers and the extent of forest, but the available data are insufficient to draw final conclusions. At a given drainage area, scatter is large, but the data show a minimal fraction of sediment transported in suspended load, which increases with increasing drainage area and with decreasing rock strength for gravel‐bed rivers, whereby in large catchments the bedload fraction is insignificant at ca 1%. For sand‐bed rivers, the bedload fraction may be substantial (30% to 50%) even for large catchments. However, available data are scarce and of varying quality. Long‐term partitioning varies widely among catchments and the available data are currently not sufficient to discriminate control parameters effectively.     

Supply limited sediment transport in a high-discharge event of the tropical Burdekin River, North Queensland, Australia

Interactions between catchment variables and sediment transport processes in rivers are complex, and sediment transport behaviour during high‐flow events is not well documented. This paper presents an investigation into sediment transport processes in a short‐duration, high‐discharge event in the Burdekin River, a large sand‐ and gravel‐bed river in the monsoon‐ and cyclone‐influenced, semi‐arid tropics of north Queensland. The Burdekin's discharge is highly variable and strongly seasonal, with a recorded maximum of 40 400 m³ s^?1. Sediment was sampled systematically across an 800 m wide, 12 m deep and straight reach using Helley‐Smith bedload and US P‐61 suspended sediment samplers over 16 days of a 29‐day discharge event in February and March 2000 (peak 11 155 m³ s^?1). About 3·7 × 10⁶ tonnes of suspended sediment and 3 × 10⁵ tonnes of bedload are estimated to have been transported past the sample site during the flow event. The sediment load was predominantly supply limited. Wash load included clay, silt and very fine sand. The concentration of suspended bed material (including very coarse sand) varied with bedload transport rate, discharge and height above the bed. Bedload transport rate and changes in channel shape were greatest several days after peak discharge. Comparison between these data and sparse published data from other events on this river shows that the control on sediment load varies between supply limited and hydraulically limited transport, and that antecedent weather is an important control on suspended sediment concentration. Neither the empirical relationships widely used to estimate suspended sediment concentrations and bedload (e.g. Ackers & White, 1973) nor observations of sediment transport characteristics in ephemeral streams (e.g. Reid & Frostick, 1987) are directly applicable to this river.     

Bedload transport and bed resistance associated with density and turbidity currents

Turbidity currents in the ocean are driven by suspended sediment. Yet results from surveys of the modern sea floor and turbidite outcrops indicate that they are capable of transporting as bedload and depositing particles as coarse as cobble sizes. While bedload cannot drive turbidity currents, it can strongly influence the nature of the deposits they emplace. This paper reports on the first set of experiments which focus on bedload transport of granular material by density underflows. These underflows include saline density flows, hybrid saline/turbidity currents and a pure turbidity current. The use of dissolved salt is a surrogate for suspended mud which is so fine that it does not settle out readily. Thus, all the currents can be considered to be model turbidity currents. The data cover four bed conditions: plane bed, dunes, upstream‐migrating antidunes and downstream‐migrating antidunes. The bedload transport relation obtained from the data is very similar to those obtained for open‐channel flows and, in fact, is fitted well by an existing relation determined for open‐channel flows. In the case of dunes and downstream‐migrating antidunes, for which flow separation on the lee sides was observed, form drag falls in a range that is similar to that due to dunes in sand‐bed rivers. This form drag can be removed from the total bed shear stress using an existing relation developed for rivers. Once this form drag is subtracted, the bedload data for these cases collapse to follow the same relation as for plane beds and upstream‐migrating antidunes, for which no flow separation was observed. A relation for flow resistance developed for open‐channel flows agrees well with the data when adapted to density underflows. Comparison of the data with a regime diagram for field‐scale sand‐bed rivers at bankfull flow and field‐scale measurements of turbidity currents at Monterey Submarine Canyon, together with Shields number and densimetric Froude number similarity analyses, provide strong evidence that the experimental relations apply at field scale as well.     

Formation and preservation of open-framework gravel strata in unidirectional flows

Open‐framework gravel (OFG) in river deposits is important because of its exceptionally high permeability, resulting from the lack of sediment in the pore spaces between the gravel grains. Fluvial OFG occurs as planar strata and cross strata of varying scale, and is interbedded with sand and sandy gravel. The origin of OFG has been related to: (1) proportion of sand available relative to gravel; (2) separation of sand from gravel during a specific flow stage and sediment transport rate (either high, falling or low); (3) separation of sand from gravel in bedforms superimposed on the backs of larger bedforms; (4) flow separation in the lee of dunes or unit bars. Laboratory flume experiments were undertaken to test and develop these theories for the origin of OFG. Bed sediment size distribution (sandy gravel with a mean diameter of 1·5 mm) was kept constant, but flow depth, flow velocity and aggradation rate were varied. Bedforms produced under these flow conditions were bedload sheets, dunes and unit bars. The fundamental cause of OFG is the sorting of sand from gravel associated with flow separation at the crest of bedforms, and further segregation of grain sizes during avalanching on the steep lee side. Sand in transport near the bed is deposited in the trough of the bedform, whereas bed‐load gravel avalanches down the leeside and overruns the sand in the trough. The effectiveness of this sorting mechanism increases as the height of the bedform increases. Infiltration of sand into the gravel framework is of minor importance in these experiments, and occurs mainly in bedform troughs. The geometry and proportion of OFG in fluvial deposits are influenced by variation in height of bedforms as they migrate, superposition of small bedforms on the backs of larger bedforms, aggradation rate, and changes in sediment supply. If the height of a bedform increases as it migrates downstream, so does the amount of OFG. Changes in the character of OFG on the lee‐side of unit bars depend on grain‐size sorting in the superimposed bedforms (dunes and bedload sheets). Thick deposits of cross‐stratified OFG require high bedforms (dunes, unit bars) and large amounts of aggradation. These conditions might be expected to occur during high falling stages in the deeper parts of river channels adjacent to compound‐bar tails and downstream of confluence scours. Increase in the amount of sand supplied relative to gravel reduces the development of OFG. Such increases in sand supply may be related to falling flow stage and/or upstream erosion of sandy deposits.     

The response of subaqueous dunes to floods in sand and gravel bed reaches of the Dutch Rhine

Abstract The branches of the River Rhine in the Netherlands, characterized by a sand–gravel bed in the upstream part and a sand bed in the downstream part of the river system, show migrating dunes, especially during floods. In the last 20 years, these dunes have been studied extensively. High-resolution echo-sounding measurements of these dunes, made with single and multibeam equipment, were analysed for three different sections of the Rhine river system during several floods. This analysis was done to quantify the growth, decay and migration rates of the dunes during floods. In addition, the migrating dunes were used to calculate bedload transport rates with dune tracking. The results of dune growth and decay and migration rate are shown to be very different for the various sections during the various floods, and these differences are related to differences in grain size of the bed and to differences in the distribution of discharge over the main channel and the floodplain. The relations are used to show that the growth and migration rate of dunes, and the calculated bedload transport rates during the rising stage of a flood wave can be predicted from the mobility of the bed material with simple power relations.     

Uncertainty analysis in bed load transport prediction of gravel bed rivers by ANN and ANFIS

River flow is a complex dynamic system of hydraulic and sediment transport. Bed load transport have a dynamic nature in gravel bed rivers and because of the complexity of the phenomenon include uncertainties in predictions. In the present paper, two methods based on the Artificial neural networks (ANN) and adaptive neuro-fuzzy inference system (ANFIS) are developed by using 360 data points. Totally, 21 different combination of input parameters are used for predicting bed load transport in gravel bed rivers. In order to acquire reliable data subsets of training and testing, subset selection of maximum dissimilarity (SSMD) method, rather than classical trial and error method, is used in finding randomly manipulation of these subsets. Furthermore, uncertainty analysis of ANN and ANFIS models are determined using Monte Carlo simulation. Two uncertainty indices of d factor and 95% prediction uncertainty and uncertainty bounds in comparison with observed values show that these models have relatively large uncertainties in bed load predictions and using of them in practical problems requires considerable effort on training and developing processes. Results indicated that ANFIS and ANN are suitable models for predicting bed load transport; but there are many uncertainties in determination of bed load transport by ANFIS and ANN, especially for high sediment loads. Based on the predictions and confidence intervals, the superiority of ANFIS to those of ANN is proved.     

Calibration of piezoelectric bedload impact sensors in the Pitzbach mountain stream

Non-invasive techniques to continuously measure the rates or intensities of bedload transport in gravel bed rivers would be quite useful for both research and practice. Since 1986 measurements have been made with piezoelectric bedload impact sensors (PBIS) in several mountain torrents. In 1994 and 1995, a calibration campaign was carried out at a water intake in the Pitzbach mountain stream in Austria, where weighing cells and a water pressure sensor in the settling basin allow sediment transport loads to be determined in 15 minute intervals. Daily averages of PBIS impulses correlate reasonably well with the measured sediment load. Using a linear calibration relation, the relative error between predicted and measured sediment loads is within a factor of about 1.5 to 2. The scatter between PBIS impulses and bedload volumes is much larger if the measurements are averaged over one hour intervals or when the original 15 minute values are considered. A comparison with measurements from the Swiss Erlenbach torrent and from laboratory experiments using sediment particles from the Erlenbach torrent indicates that the proportionality coefficient between PBIS impulses and sediment load appears to depend on site specific conditions, including factors such as flow intensity and sediment properties.     

Shoal formation in the Piscataqua River,New Hampshire

The formation of a shoal was investigated in the Piscataqua River, New Hampshire, which is a well-mixed channel with low freshwater flow and tidal currents up to 2.3 m s^?1. Observations of sediment characteristics, bathymetry, and bottom current were made, and theory was used to predict bedload transport. Sediment sampling showed the bottom material to be coarse sand and gravel, and sidescan sonar revealed large sand waves directed upriver at the shoal. Bottom current measurements were made along transects upriver and downriver of the shoal and downriver of an adjacent deepwater area that was also studied for comparison. Bedload flux inferred from current measurements using the Brown-Einstein theory indicated that transport is generally directed upriver. Sediment budget calculations showed the shoal area to be depositional before, immediately after, and subsequent to a dredging operation at rates of 0.36 m yr^?1, 1.06 m yr^?1, and 0.35 m yr^?1, respectively. Predredge and subsequent rates were consistent with the historical record of removal by dredging at the shoal.     

Gravel transport,gravel harvesting,and channel-bed degradation in rivers draining the southern olympic mountains,Washington, U.S.A.

The potential for gravel extraction to adversely affect anadromous fish habitat in three gravel-bed rivers of southwestern Washington, U.S.A., prompted the need to determine sustainable rates of gravel removal. This was accomplished by evaluating the components of a long-term sediment mass balance for the three rivers. Average annual gravel transport was determined by three independent methods. The closely agreeing results indicate that annual bedload supply decreases downstream through deposition and storage in response to declining gradient and from attrition during transport, as confirmed by laboratory experiments. A survey of gravel-bar harvesting operations indicates that the annual replenishment rate has been exceeded for up to three decades, often by more than tenfold. Analysis of data from nine stream gauging stations over a 55-yr period indicates degradation of about 0.03 m/yr in these reaches and suggests that bed degradation has produced the difference between the replenishment rates and the volumes of gravel harvested from the river beds and bars.     

Sediment pathways and transport rates through a tide-dominated entrance, Rangaunu Harbour, New Zealand

Bed load sediment traps were deployed at two sections across channels in Rangaunu Harbour entrance. Traps were inspected and emptied by divers at hourly intervals through both spring and neap tidal cycles for a total of 292 trap deployments. Current velocities were measured simultaneously with the trap inspections. Transport is concentrated in sandy megaripple fields on the channel banks and sub-tidal platforms flanking the channels. There, transport is almost continuous throughout the tidal cycle, increasing with flow velocity but lagging by approximately one hour. The channel floors are lined with shell-gravel lag across which bedload transport rates are low and discontinuous. Tidal asymmetry produces a net seaward transport through the channel troughs and a net landward transport across the channel banks and flanking sub-tidal platforms. Sediment leaving the harbour recirculates in anticlockwise gyres across the ebb-tide delta to re-enter the harbou and maintain the supply of sand to the megaripple field. Transport during spring tides is typically 25–30 times that during neaps. Predictions of transport rates, from a method developed by Black & Healy utilizing the Yalin bedload equation, produced transport rates similar to the traps over sand beds. Transport over shell lag surfaces appears independent of near-bed velocity and more dependent on the passage of ribbons of sand across the lag surface.     

Particle over-passing on depth-limited gravel bars

An experimental channel is used to examine the transport of mixed sand and gravel bedload over the crestal platform of ‘hump-back’ bars and along the top of planar gravel sheets. Hydraulic processes result in the simultaneous transport of cobbles and pebbles over a static closely packed bed consisting of like-sized and finer particles. For prescribed conditions, flat upper-stage plane sand-beds develop over the crestal location with pebbles rolling easily over the sandy bed. At the brinkpoint, flow separation ensures effective segregation of the gravel from the sand. Over the slip-face the deposition rate of the sand is insufficient to fill fully the interstices within the gravel foresets before rapid deposition of gravel further advances the bed-form. Consequently, distinctive vertical assemblages of open-work and closed contact framework gravels could be generated as another bar migrates over, and preserves, the initial structure. In respect to the observed mechanisms of sorting over the bars, a mathematical expression is developed to explain the critical conditions allowing coarse particle mobility over planar sand or gravel beds under upper-stage plane-bed conditions on the crestal platform. The model then is used to ascertain whether the depositional environment ascribed to certain facies in the Bunter Pebble Beds, described in a recent publication, is appropriate given the distinctive facies assemblages generated in this experiment and the known hydrodynamic control of the particle-segregation process.     

Supply‐limited bedform patterns and scaling downstream of a gravel–sand transition

A distinct suite of sand bedforms has been observed to occur in laboratory flows with limited sand supply. As sand supply to the bed progressively increases one observes sand ribbons, discrete barchans and, eventually, channel spanning dunes; but there are relatively few observations of this sequence from natural river channels. Furthermore, there are few observations of transitions from limited sand supply to abundant supply in the field. Bedforms developed under limited, but increasing, sand supply downstream of the abrupt gravel–sand transition in the Fraser River, British Columbia, are examined using multi‐beam swath‐bathymetry obtained at high flow. This is an ideal location to study supply‐limited bedforms because, due to a break in river slope, sand transitions from washload upstream of the gravel–sand transition to bed material load downstream. Immediately downstream, barchanoid and isolated dunes are observed. Most of the bedform field has gaps in the troughs, consistent with sand moving over a flat immobile or weakly mobile gravel bed. Linear, alongstream bedform fields (trains of transverse dunes formed on locally thick, linear deposits of sand) exhibit characteristics of sand ribbons with superimposed bedforms. Further downstream, channel spanning dunes develop where the bed is composed entirely of sand. Depth scaling of the dunes does not emerge in this data set. Only where the channel has accumulated abundant sand on the bed do the dunes exhibit scaling congruent with previous data compilations. The observations suggest that sediment supply plays an important, but often overlooked, role in bedform scaling in rivers.     

Spatio-temporal variability of bedload transport rate: analysis and 2D modelling approach

Field measurements to calibrate numerical bedload formulae are largely missing. Measurements using a Large Helley-Smith sampler were performed over a period of five years in the large Alpine Drau River, Austria. Our results reveal the high spatio-temporal variability of bedload transport rates. Commonly used bedload predictors poorly describe measured transport rates. Temporal and localised cross-sectional variation in bedload transport rates are observed in short time frames. To obtain significant mean values, the measurement period has to be extended to cover the existing bedload transport periodicity. The discrepancy between bedload transport measurements and simulation is partially explained by local hydraulic variations. The results can be improved, particularly for verticals where most of the bedload occurs, by relating measured transport rates to local hydraulic parameters. The incorporation of local cross-sectional parameters demonstrates the utility of 2D bedload models and their greater predictive power over similar 1D models.     

The alluvial architecture of a suspended sediment dominated meandering river: the Río Bermejo,Argentina

The alluvial architecture of fine‐grained (silt‐bed) meandering rivers remains poorly understood in comparison to the extensive study given to sand‐bed and gravel‐bed channels. This paucity of knowledge stems, in part, from the difficulty of studying such modern rivers and deriving analogue information from which to inform facies models for ancient sediments. This paper employs a new technique, the parametric echosounder, to quantify the subsurface structure of the Río Bermejo, Argentina, which is a predominantly silt‐bed river with a large suspended sediment load. These results show that the parametric echosounder can provide high‐resolution (decimetre) subsurface imaging from fine‐grained rivers that is equivalent to the more commonly used ground‐penetrating radar that has been shown to work well in coarser‐grained rivers. Analysis of the data reveals that the alluvial architecture of the Río Bermejo is characterized by large‐scale inclined heterolithic stratification generated by point‐bar evolution, and associated large‐scale scour surfaces that result from channel migration. The small‐scale and medium‐scale structure of the sedimentary architecture is generated by vertical accretion deposits, bed sets associated with small bars, dunes and climbing ripples and the cut and fill from small cross‐bar channels. This style of alluvial architecture is very different from other modern fine‐grained rivers reported in the literature that emphasize the presence of oblique accretion. The Río Bermejo differs from these other rivers because it is much more active, with very high rates of bank erosion and channel migration. Modern examples of this type of highly active fine‐grained river have been reported rarely in the literature, although ancient examples are more prevalent and show similarities with the alluvial architecture of the Río Bermejo, which thus represents a useful analogue for their identification and interpretation. Although the full spectrum of the sedimentology of fine‐grained rivers has yet to be revealed, meandering rivers dominated by lateral or oblique accretion probably represent end members of such channels, with the specific style of sedimentation being controlled by grain size and sediment load characteristics.     

东江沙卵石河床浅滩-深潭序列水沙演变特征

东江是珠三角网河区入汇河流之一,属少沙河流,其上游浅滩-深潭序列密集分布,河床结构的演变特性尚不清晰。选取东江浅滩-深潭序列河段进行现场河床质与推移质测量,结合试验得到不同流量和坡降中水流与河床在无上游来沙补给中的互馈机理。试验结果表明：浅滩与深潭交替造成沿程平均流速不连续变化,浅滩中部在洪水流量的塑造下容易产生紊动能峰值。无上游来沙条件下,序列中粗化层的形成、破坏过程交替与循环发生,其中静态分选是主导因素且导致沿程级配不连续。流量与坡降的增大促进浅滩-深潭序列发育,同时使得深潭与浅滩的纵向长度缩短;有上游来沙条件下推移质输沙率增大,无来沙条件下输沙率减小。     

Direct numerical simulation of bedload transport using a local, dynamic boundary condition

ABSTRACT Temporally and spatially averaged models of bedload transport are inadequate to describe the highly variable nature of particle motion at low transport stages. The primary sources of this variability are the resisting forces to downstream motion resulting from the geometrical relation (pocket friction angle) of a bed grain to the grains that it rests upon, variability of the near‐bed turbulent velocity field and the local modification of this velocity field by upstream, protruding grains. A model of bedload transport is presented that captures these sources of variability by directly integrating the equations of motion of each particle of a simulated mixed grain‐size sediment bed. Experimental data from the velocity field downstream and below the tops of upstream, protruding grains are presented. From these data, an empirical relation for the velocity modification resulting from upstream grains is provided to the bedload model. The temporal variability of near‐bed turbulence is provided by a measured near‐bed time series of velocity over a gravel bed. The distribution of pocket friction angles results as a consequence of directly calculating the initiation and cessation of motion of each particle as a result of the combination of fluid forcing and interaction with other particles. Calculations of bedload flux in a uniform boundary and simulated pocket friction angles agree favourably with previous studies.     

Validation and implications of an energy‐based bedload transport equation

A recently developed bedload equation (Abrahams & Gao, 2006) has the form i_b = ωG^3˙4, where i_b is the immersed bedload transport rate, ω is the stream power per unit area, G = 1?θ_c/θ, θ is the dimensionless shear stress and θ_c is the associated threshold value for the incipient motion of bed grains. This equation has a parsimonious form and provides good predictions of transport rate in both the saltation and sheetflow regimes (i.e. flows with low and high θ values, respectively). In this study, the equation was validated using data independent of those used for developing it. The data represent bedload of identical sizes transported in various steady, uniform, fully rough and turbulent flows over plane, mobile beds. The equation predicted i_b quite well over five orders of magnitude. This equation was further compared with six classic bedload equations and showed the best performance. Its theoretical significance was subsequently examined in two ways. First, based on collision theory, the parameter G was related to the ratio of grain‐to‐grain collisions to the total collisions including both grain‐to‐grain and grain‐to‐bed collisions, P_g by P_g = G², suggesting that G characterizes the dynamic processes of bedload transport from the perspective of granular flow, which partly accounts for the good performance of the equation. Moreover, examining the ability of two common equations to predict bedload in gravel‐bed rivers revealed that G can also be used to simplify equations for predicting transport capacities in such rivers. Second, a simple dimensionless form of the equation was created by introducing B = i_b/ω. The theoretical nature of the term B was subsequently revealed by comparing this equation with both the Bagnold model and two commonly used parameters representing dimensionless bedload transport rates.     

The decrease in shear stress and increase in transport rates subsequent to an increase in sand supply to a gravel-bed channel

It is generally accepted that a gravel-bed river will aggrade if the supply of sediment to the river is increased. In a series of flume experiments using constant discharge and gravel feed rate, sand feed rates were increased to 6.1 times that of gravel. The slope of the bed decreased with increasing sand supply, indicating that the increased sediment load could be transported at the same rate due to a decrease in shear stresses. These results extend previous experiments to a wider range of boundary conditions. A recent surface transport model is used to predict the changes in bed composition and transport using the same sediment supply composition and feed rates as in the laboratory experiments. This model reasonably predicts a decrease in the reference shear stresses of the sand and gravel fractions as the sand supply is increased. An increase in sand supply can increase the mobility of gravel fractions in the stream bed, which can lead to bed degradation and preferential evacuation of these sediments from the river.     

The morphodynamics of fluvial sand dunes in the River Rhine, near Mainz, Germany. I. Sedimentology and morphology

The dynamics of large isolated sand dunes moving across a gravel lag layer were studied in a supply‐limited reach of the River Rhine, Germany. Bed sediments, dune geometry, bedform migration rates and the internal structure of dunes are considered in this paper. Hydrodynamic and sediment transport data are considered in a companion paper. The pebbles and cobbles (D₅₀ of 10 mm) of the flat lag layer are rarely entrained. Dunes consist of well‐sorted medium to coarse sand (D₅₀ of 0·9 mm). Small pebbles move over the dunes by ‘overpassing’, but there is a degree of size and shape selectivity. Populations of ripples in sand (D₅₀ < 0·6 mm), and small and large dunes are separated by distinct breaks in the bedform length data in the regions of 0·7–1 m and 5–10 m. Ripples and small dunes may have sinuous crestlines but primarily exhibit two‐dimensional planforms. In contrast, large dunes are primarily three‐dimensional barchanoid forms. Ripples on the backs of small dunes rarely develop to maximum steepness. Small dunes may achieve an equilibrium geometry, either on the gravel bed or as secondary dunes within the boundary layer on the stoss side of large dunes. Secondary dunes frequently develop a humpback profile as they migrate across the upper stoss slope of large dunes, diminishing in height but increasing in length as they traverse the crestal region. However, secondary dunes more than 5 m in length are rare. The dearth of equilibrium ripples and long secondary dunes is probably related to the limited excursion length available for bedform development on the parent bedforms. Large dunes with lengths between 20 m and 100 m do not approach an equilibrium geometry. A depth limitation rather than a sediment supply limitation is the primary control on dune height; dunes rarely exceed 1 m high in water depths of ≈4 m. Dune celerity increases as a function of the mean flow velocity squared, but this general relationship obscures more subtle morphodynamics. During rising river stage, dunes tend to grow in height owing to crestal accumulation, which slows downstream progression and steepens the dune form. During steady or falling stage, an extended crestal platform develops in association with a rapid downstream migration of the lee side and a reduction in dune height. These diminishing dunes actually increase in unit volume by a process of increased leeside accumulation fed by secondary dunes moving past a stalled stoss toe. A six‐stage model of dune growth and diminution is proposed to explain variations in observed morphology. The model demonstrates how the development of an internal boundary layer and the interaction of the water surface with the crests of these bedload‐dominated dunes can result in dunes characterized by gentle lee sides with weak flow separation. This finding is significant, as other studies of dunes in large rivers have attributed this morphological response to a predominance of suspended load transport.