An appraisal of the velocity-reversal hypothesis for stable pool-riffle sequences in the river severn,England

The ‘velocity-reversal hypothesis’ is the linchpin for a number of recent conceptual models pertaining to sediment sorting and the maintenance of pool-riffle sequences in gravel-bedded streams. The literature in support of the hypothesis together with published adverse criticism is reviewed. It is concluded that convincing evidence for the ubiquitous occurrence of such a reversal in a range of channel geometries is currently unavailable. Continuity considerations indicate that riffles need to be considerably wider than pools for a reversal in the mean velocity to occur under conditions of subcritical flow, high stage, and stable morphology. These observations are substantiated by a detailed study of the hydraulic geometry of stable pool-riffle sequences in the River Severn, England. Neither the sectionally-averaged velocity nor the near-bed shear velocity is sensibly greater in the pools than over the riffles during bankfull or near bankfull flow. Instead a tendency towards equalization of the values of average hydraulic variables is noted as discharge increases. A detailed investigation of the three-dimensional character of the flow is required to demonstrate whether the entrainment forces within pools can locally exceed those over neighbouring riffles. Unusual behaviour of the energy gradient over riffles during moderate discharges is related to backwater effects as mediated by the spacing of the riffles. The hydraulic data are used to comment on the stability of the test reaches in the context of the development of the River Severn in the vicinity of Shrewsbury.     

The morphology of riffle-pool sequences

The riffle-pool sequence has not been subjected to the same level of intensive research as the meandering planform, although riffles and pools may be a fundamental prerequisite for meandering. The pseudo-cyclic oscillation of the bed in a riffle-pool stream suggests the application of a variety of techniques of spatial series analysis, which provide objective measures of riffle wavelength, and suggest processes capable of explaining riffles and pools and their relationship with meanders. The second-order autoregressive process is suggested as a stochastic process which models the bed-profile oscillation. Velocity pulsations associated with large scale turbulent eddies are probably responsible for accretions and erosions which interact with the flow to maintain these perturbations, so that sections lagged by distances of 2πw are positively correlated. The effect of the riffle-pool sequence on flow geometry is far more significant than the effects of plan geometry or of downstream variations, which supports the view that this feature is a fundamental aspect of channel morphometry. There is a tendency, however, for curved reaches to exhibit reduced variance of roughness, velocity, and water surface slope, which reinforces the minimization hypothesis. The extreme temporal variation between riffle and pool flow characteristics demands that any classificatory scheme uses scale-free and stable measures, and a discriminant analysis using hydraulic exponents represents a convenient summary of the field data.     

Bed forms and bend-arc spacings of some coarse-bedload channels in Upland Britain

Seventy-eight riffle to riffle and 80 bend spacings along eleven coarse-bedload, low sinuosity stream channels in upland Britain have been surveyed. Frequency distributions of these spacings are notably right-skewed. The most common repeating distances between riffles and bend inflections are between 4 and 6 channel widths although spacings up to 20 widths are also present. Riffle and pool locations around bends at different stages of planform development indicate that change is largely through increased sinuosity between two consecutive riffles of an original straight reach. Observed straight segments exhibit alternating riffles and pools evenly spaced at 4-6 widths, and most bends have similarly spaced riffles at their inflections in plan, with the intermediate pool at their apex. However angular deflections between axial lines joining inflections indicate existing sequences of bends did not develop from a single straight reach. Bends which are significantly longer than 4-6 widths are of low sinuosity and represent variability in naturally irregular planforms rather than arcs in advanced stages of meander growth. Adjusted bed topography around such long bends takes two forms. Either a single riffle-pool cycle is present with one or both bed forms being longer than average, or a 4-6 widths spacing is maintained by more than one riffle-pool sequence. Locally, the cross-section characteristics of riffles and pools are also influenced by planform location.     

Simulation of flow over pool-riffle topography: A consideration of the velocity reversal hypothesis

Computer flow simulations using the HEC-2 step-backwater routine are used to demonstrate the effect of systematically varying river channel width, riffle spacing and channel roughness on the shear velocity, section-mean velocity and energy slope in fixed-bed pool-riffle sequences. Initial scaling is obtained by utilizing published information on hydraulic parameters within reaches of the River Severn. Subsequently this restriction is relaxed and the effect of varying parameter combinations within realistic limits is explored. The purpose of this exercise is to isolate those scenarios which may preclude or promote the occurrence of a competence ‘reversal’, such that pools scour at high flow whilst deposition occurs on riffles. It is concluded that rivers in which pools are hydraulically rougher than riffles are likely to demonstrate a competence reversal. For prescribed conditions, the critical discharge at which a reversal occurs is a negative function of riffle spacing and riffle width relative to pool width. Downstream variation in hydraulic roughness also has implications for the phase relationship of shear velocity maxima and minima in relation to the extremes in pool-riffle topography.     

Spatial variations in surface sediment structure in riffle–pool sequences: a preliminary test of the Differential Sediment Entrainment Hypothesis (DSEH)

Riffle–pool sequences are maintained through the preferential entrainment of sediment grains from pools rather than riffles. This preferential entrainment has been attributed to a reversal in the magnitude of velocity and shear stress under high flows; however the Differential Sediment Entrainment Hypothesis (DSEH) postulates that differential entrainment can instead result from spatial sedimentological contrasts. Here we use a novel suite of in situ grain‐scale field measurements from a riffle–pool sequence to parameterize a physically‐based model of grain entrainment. Field measurements include pivoting angles, lift forces and high resolution digital elevation models (DEMs) acquired using terrestrial laser scanning, from which particle exposure, protrusion and surface roughness were derived. The entrainment model results show that grains in pools have a lower critical entrainment shear stress than grains in either pool exits or riffles. This is because pool grains have looser packing, hence greater exposure and lower pivoting angles. Conversely, riffle and pool exit grains have denser packing, lower exposure and higher pivoting angles. A cohesive matrix further stabilizes pool exit grains. The resulting predictions of critical entrainment shear stress for grains in different subunits are compared with spatial patterns of bed shear stress derived from a two‐dimensional computational fluid dynamics (CFD) model of the reach. The CFD model predicts that, under bankfull conditions, pools experience lower shear stresses than riffles and pool exits. However, the difference in sediment entrainment shear stress is sufficiently large that sediment in pools is still more likely to be entrained than sediment in pool exits or riffles, resulting in differential entrainment under bankfull flows. Significantly, this differential entrainment does not require a reversal in flow velocities or shear stress, suggesting that sedimentological contrasts alone may be sufficient for the maintenance of riffle–pool sequences. This finding has implications for the prediction of sediment transport and the morphological evolution of gravel‐bed rivers. Copyright © 2012 John Wiley & Sons, Ltd.     

Experimental investigation of the morphodynamic response of riffles and pools to unsteady flow and increased sediment supply

Gravel-bed rivers characteristically exhibit shallow riffles in wide sections and deeper pools where the channel becomes constricted and narrow. While rivers can adjust to changing flow and sediment supply through some combination of adjustments of channel slope, bed-surface sorting, and channel shape, the degree to which riffle-pools may adopt these changes in response to changing flows and sediment supplies remains unclear. This article presents results from a flume experiment investigating how constant- and variable-width channels adjust their morphology in response to changing flow and increased sediment supply. Two flume geometries were used: (1) constant-width and (2) variable-width, characterized by a sinusoidal pattern with a mean width equal to that of the first channel. The variable-width channel developed bed undulations in phase with the width, representing riffle-pools. The experiment consisted of three phases for each flume geometry: (1) steady flow, constant sediment supply; (2) unsteady flow, constant sediment supply; and (3) unsteady flow, doubled sediment supply. Unsteady flow was implemented in the form of repeated symmetrical stepped hydrographs, with a mean discharge equal to that in the steady flow phase. In all phases the bed and sediment supply were composed of a sand/gravel mixture ranging from 1 to 8 mm. In both the straight and variable-width channels, transitioning from steady flow to repeated hydrographs did not result in significant changes in bed morphology. The two channel geometries had different responses to increased sediment supply: the slope of the straight channel increased nearly 40%, while the variable-width channel reduced the relief between bars and pools and decreased the variability in cross-sectional elevation with a slight slope increase. Bar-pool relief varied with repeat discharge hydrographs. Pool elevation changed twice the distance of bar elevations, emphasizing the relevance of pool scour for riffle-pool self-maintenance in channels with width variations.     

Riffle‐pool morphometry and stage‐dependant morphodynamics of a large floodplain river (Vereinigte Mulde,Sachsen‐Anhalt,Germany)

Riffle‐pool sequences are a common feature of gravel‐bed rivers. However, mechanisms of their generation and maintenance are still not fully understood. In this study a monitoring approach is employed that focuses on analysing cross‐sectional and longitudinal channel geometry of a large floodplain river (Vereinigte Mulde, Sachsen‐Anhalt, Germany) with a high temporal and spatial resolution, in order to conclude from stage‐dependant morphometric changes to riffle and pool maintaining processes. In accordance with previous authors, pool cross‐sections of the Mulde River are narrow and riffle cross‐sections are wide suggesting that they should rather be addressed as two general types of channel cross‐sections than solely as bedforms. At high flows, riffles and pools in the study reaches changed in length and height but not in position. Pools were scoured and riffles aggraded, a development which was reversed during receding flows below the threshold of 0·4Q_bf (40% bankfull discharge). An index for the longitudinal amplitude of riffle‐pool sequences, the bed undulation intensity or bedform amplitude, is introduced and proved to be highly significant as a form parameter, its first derivative as a process parameter. The process of pool scour and riffle fill is addressed as bedform maintenance or bedform accentuation. It is indicated by increasing longitudinal bed amplitudes. According to the observed dynamics of bed amplitudes, maintenance of riffle‐pool sequences lags behind discharge peaks. Maximum bed amplitudes may be reached with a delay of several days after peak discharges. Increasing bed undulation intensity is interpreted to indicate bed mobility. Post‐flood decrease of the bed undulation intensity indicates a retrograde phase when transport from pools to riffles has ceased and bed mobility is restricted to riffle tails and heads of pools. This type of transport behaviour is referred to as disconnected mobility. The comparison of two river reaches, one with undisturbed sediment supply, the other with sediment deficit, suggests that high bed undulation intensity values at low flows indicate sediment deficit and potentially channel degrading conditions. It is more generally hypothesized that channel bed undulations constitute a major component of form roughness and that increased bed amplitudes are an important feature of channel bed adjustment to sediment deficit be it temporally during late floods or permanently due to a supply limitation of bedload. Copyright © 2011 John Wiley & Sons, Ltd.     

Simulation of flow geometry in a riffle-pool stream

The nature of the flow in most natural streams is gradually varied rather than uniform. This is particularly true of streams with coarse gravel bed material organized into relatively stable riffle and pool features. In spite of this, there are few applications of the gradually varied flow models (e.g. the Bernoulli equation) to such streams. This paper presents some initial results of a simulation of flow patterns in two riffle-pool reaches, using an open channel flow profile computation method based on an equation defining an energy balance between successive cross-sections separated by an incremental distance.     

Flow in a tightly curving meander bend: effects of seasonal changes in aquatic macrophyte cover

The effects of aquatic macrophytes on flow and turbulence were studied in a tightly curving meander bend. Three field measurement campaigns were carried out within a one year period to capture effects of seasonal changes in macrophyte cover. They comprised three‐dimensional velocity measurements and mappings of vegetation cover and bathymetry. Flow accelerates and converges over the run into an axial pool in a jet‐like flow pattern bordered by outer and inner bank flow separation zones. The jet and widening of the cross‐section anticipate helical flow until the second half of the bend, where an asymmetric pool developed. Submerged vegetation at the riffles preserves the jet at much lower discharges during the summer period by concentrating high momentum fluid near the surface. Plants locally modify the velocity and stress patterns, reduce bed shear stresses, create zones of fine sediment accumulation and reinforce the bed and banks with roots and rhizomes. Plant patches colonising the banks and the point bar confine secondary flow cells laterally and affect shape and magnitude of the transverse flow profiles near their edges. The morphology of the bend was very stable over the observation period and neither bank erosion nor pool scouring occurred. However, fine sediments accumulate within vegetation patches and in the recirculation zones while the remaining open areas tend to erode slightly. With the decay of macrophytes in winter, sediment accumulations are mobilised again and the bathymetry levels, supporting cyclic models of morphologic change in vegetated bends. In the second part of the paper, semi‐empirical models for the three predominant flow types were tested and discussed; velocity and stress models of vegetated mixing layers and plane turbulent jets, and Rozovskii's model for the transverse flow in bends. Copyright © 2012 John Wiley & Sons, Ltd.     

Sediment routing hypothesis for pool‐riffle maintenance

This paper provides comprehensive evidence that sediment routing around pools is a key mechanism for pool‐riffle maintenance in sinuous upland gravel‐bed streams. The findings suggest that pools do not require a reversal in energy for them to scour out any accumulated sediments, if little or no sediments are fed into them. A combination of clast tracing using passive integrated transponder (PIT) tagging and bedload traps (positioned along the thalweg on the upstream riffle, pool entrance, pool exit and downstream riffle) are used to provide information on clast pathways and sediment sorting through a single pool‐riffle unit. Computational fluid dynamics (CFD) is also used to explore hydraulic variability and flow pathways. Clast tracing results provide a strong indication that clasts are not fed through pools, rather they are transported across point bar surfaces, or around bar edges (depending upon previous clast position, clast size, and event magnitude). Spatial variations in bedload transport were found throughout the pool‐riffle unit. The pool entrance bedload trap was often found to be empty, when the others had filled, further supporting the notion that little or no sediment was fed into the pool. The pool exit slope trap would occasionally fill with sediment, thought to be sourced from the eroding outer bank. CFD results demonstrate higher pool shear stresses (τ ≈ 140 N m^–2) in a localized zone adjacent to an eroding outer bank, compared to the upstream and downstream riffles (τ ≈ 60 N m^–2) at flows of 6 · 2 m³ s^–1 (≈ 60% of the bankfull discharge) and above. There was marginal evidence for near‐bed velocity reversal. Near‐bed streamlines, produced from velocity vectors indicate that flow paths are diverted over the bar top rather than being fed through the thalweg. Some streamlines appear to brush the outer edge of the pool for the 4 · 9 m³ s^–1 to 7 · 8 m³ s^–1 (between 50 and 80% of the bankfull discharge) simulations, however complete avoidance was found for discharges greater than this. Copyright © 2013 John Wiley & Sons, Ltd.     

Detecting multi-scale riverine topographic variability and its influence on Chinook salmon habitat selection

Quantifying geomorphic conditions that impact riverine ecosystems is critical in river management due to degraded riverine habitat, changing flow and thermal conditions, and increasing anthropogenic pressure. Geomorphic complexity at different scales directly impacts habitat heterogeneity and affects aquatic biodiversity resilience. Here we showed that the combination of continuous spatial survey at high resolution, topobathymetric light detection and ranging (LiDAR), and continuous wavelet analysis can help identify and characterize that complexity. We used a continuous wavelet analysis on 1-m resolution topobathymetry in three rivers in the Salmon River Basin, Idaho (USA), to identify different scales of topographic variability and the potential effects of this variability on salmonid redd site selection. On each river, wavelet scales characterized the topographic variability by portraying repeating patterns in the longitudinal profile. We found three major representative spatial wavelet scales of topographic variability in each river: a small wavelet scale associated with local morphology such as pools and riffles, a mid-wavelet scale that identified larger channel unit features, and a large wavelet scale related to valley-scale controls. The small wavelet scale was used to identify pools and riffles along the entire lengths of each river as well as areas with differing riffle-pool development. Areas along the rivers with high local topographic variability (high wavelet power) at all wavelet scales contained the largest features (i.e., deepest or longest pools) in the systems. By comparing the wavelet power for each wavelet scale to Chinook salmon redd locations, we found that higher small-scale wavelet power, which is related to pool-riffle topography, is important for redd site selection. The continuous wavelet methodology objectively identified scales of topographic variability present in these rivers, performed efficient channel-unit identification, and provided geomorphic assessment without laborious field surveys.     

Morphology of riffle–pool sequences in the River Severn,England

Despite the occurrence of riffle–pool sequences in many rivers there are few data concerning riffle–pool unit morphology. Of many criteria proposed to identify riffle–pool units, only three methods can be regarded as objective and robust. These are the ‘zero‐crossing’, the ‘spectral analysis’ and the ‘control‐point’ methods. In this paper statistics are developed using the first two of these methods to describe the streamwise morphology of 275 riffles and 285 pools which form a continuous 32·1 km reach of the bed of the River Severn in Shropshire, England. Yalin's theoretical relationship between the average riffle:pool unit length ( λ _p) and channel width ( W ), λ _p = 3 W , applies to the River Severn. Reach‐average riffle height ( H ) is a constant proportion of bankfull depth ( h ); typically H ≅ 0·16 h . Riffle height is a positive function of riffle length. Pool depth is a positive function of pool length. However, both riffle length and pool length increase more rapidly than the bed‐level amplitude, such that long riffles or pools are relatively ‘flat’. As channel gradient reduces, bedforms flatten and become more asymmetric as riffle stoss sides and the proximal slope of pools lengthen at the expense of riffle lee sides and pool distal slopes. The statistical relationships between riffle steepness (H/L) and water depth are similar to those for equilibrium subaqueous dunes. The Severn data are consistent with Yalin's theoretical analysis relating riffle bedform length (L _r) to water depth, i.e. L _r = α2π h, wherein α ≅ 1 for steep near‐equilibrium bedforms but α ≅ 2 to 3 as the relative depth decreases and riffles become long, low features. Theoretical consideration and turbulence data indicate that the frequency of coherent turbulent‐flow structures associated with the riffle–pool mixing length in the Severn should be of the order of 50 to 100 s. The morphological similarity of the steepest River Severn riffles with dunes raises intriguing questions with respect to self‐similar, convergent organization of periodic alluvial bedforms and to bedform dynamic classification particularly. Copyright © 2000 John Wiley & Sons, Ltd.     

Refuge habitats for fishes during seasonal drying in an intermittent stream: movement,survival and abundance of three minnow species

Drought and summer drying can be important disturbance events in many small streams leading to intermittent or isolated habitats. We examined what habitats act as refuges for fishes during summer drying, hypothesizing that pools would act as refuge habitats. We predicted that during drying fish would show directional movement into pools from riffle habitats, survival rates would be greater in pools than in riffles, and fish abundance would increase in pool habitats. We examined movement, survival and abundance of three minnow species, bigeye shiner (Notropis boops), highland stoneroller (Campostoma spadiceum) and creek chub (Semotilus atromaculatus), during seasonal stream drying in an Ozark stream using a closed robust multi-strata mark-recapture sampling. Population parameters were estimated using plausible models within program MARK, where a priori models are ranked using Akaike’s Information Criterion. Creek chub showed directional movement into pools and increased survival and abundance in pools during drying. Highland stonerollers showed strong directional movement into pools and abundance increased in pools during drying, but survival rates were not significantly greater in pools than riffles. Bigeye shiners showed high movement rates during drying, but the movement was non-directional, and survival rates were greater in riffles than pools. Therefore, creek chub supported our hypothesis and pools appear to act as refuge habitats for this species, whereas highland stonerollers partly supported the hypothesis and bigeye shiners did not support the pool refuge hypothesis. Refuge habitats during drying are species dependent. An urgent need exists to further understand refuge habitats in streams given projected changes in climate and continued alteration of hydrological regimes.     

The challenge of modeling pool–riffle morphologies in channels with different densities of large woody debris and boulders

Studies on pool morphologies include reports of over 80% or 90% of pools being associated with structural controls and large obstructions that include boulders, bedrock outcrops and large woody debris (LWD). A Monte Carlo simulation approach and developmental computer model was created to predict pool formation, spacing and the percentage length covered by pools, riffles, scour holes and runs based on input data that include channel slope, width, the number of small and large boulders, and the number of 10–30 cm, 30–60 cm and >60 cm pieces of wood. The statistical‐empirical model is founded on the idea that boulders, bedrock outcrops and large woody debris provide a physical framework that then controls local water‐surface slopes, velocity patterns and the locations of pools and riffles. The spacing values of individual types and sizes of obstructions are modeled as log‐normal distributions with separate distributions for each obstruction type. Pools are assigned different probabilities of development depending on the obstruction type. Pool and riffle lengths used to create the subsequent morphology follow their own slope‐dependent, log‐normal trends. A minimum distance develops between successive pools because of the backwater and turbulent conditions needed for pool formation. The total number and spacing of pools, riffles and scour holes thus reflects the number and locations of obstructions and characteristics of the pool–riffle couplet. The simulation model accurately captures the number of pools in the modeled data range at 65% of all the verification field sites, and 86% of the verification field sites with a more limited range of width and slope characteristics. Lower levels of prediction capabilities are associated with modeled numbers of scour holes and log jams. The model accurately mimics some statistical attributes of pool spacing, and future versions of the model could be developed to improve overall predictive capabilities. Copyright © 2011 John Wiley & Sons, Ltd.     

Three‐dimensional flow structure and channel change in an asymmetrical compound meander loop,Embarras River,Illinois

The planform dynamics of meandering rivers produce a complex array of meander forms, including elongated meander loops. Thus far, few studies have examined in detail the flow structure within meander loops and the relation of flow structure to patterns of planform change. This field‐based investigation examines relations between three‐dimensional fluid motion and channel change within an elongated, asymmetrical meander loop containing multiple pool–riffle structures. The downstream velocity field is characterized by a high‐velocity core that shifts slightly outward as flow moves through individual lobes of the loop. For some of the measured flows this core becomes submerged below the water surface downstream of the lobe apexes. Vectors of cross‐stream/vertical velocities indicate that skew‐induced helical motion develops within the pools near lobe apexes and decays over riffles where channel curvature is less pronounced. Maximum rates of bank retreat generally occur near lobe apexes where impingement of the flow on the outer channel bank is greatest. However, maximum rates and loci of bank retreat differ for upstream and downstream lobes of the loop, leading to increasing asymmetry of loop geometry over time—a finding consistent with experimental investigations of loop evolution. Copyright © 2003 John Wiley & Sons, Ltd.     

Habitat preference and locomotive traits of the Japanese mitten crab Eriocheir japonica within the stream unit of the freshwater river environment

Preferences by the Japanese mitten crab Eriocheir japonica of three mesoscale environments within a river unit (pools, riffles and runs) were investigated in a freshwater river by pot fishing and hand-net collecting during the fall. In addition, E. japonica’s life modes during each growth stage were elucidated by comparing data obtained by hand-net collecting into three size groups (20 mm < CW, 20 mm ≤ CW < 40 mm, and 40 mm ≤ CW). Both the number of crabs caught per pot and per 10 m stretch sampled along the river by hand-net collecting indicated that mean catch values varied in the following manner runs < pools < riffles, and that this variation was large in riffles and small in pools and runs. Only large crabs of 40 mm ≤ CW could be collected by pot fishing; in contrast, small juveniles were only caught by hand-net collecting. Results of the hand-net collecting indicated that small juveniles (CW < 20 mm) preferred to be in riffles, while the mid-size class crabs (20 – 40 mm CW) presence in the riffles decreased and once they became adults (40 mm ≤ CW) they chose to be in pools. In all size classes, runs were not a preferred choice. These results suggest that the small juveniles are oriented by their positive rheotaxis. Their movement is probably affected by flow regimes, creating large variations in their density in the riffles. In contrast, adults can freely move within a stream unit and use pools as their resting site, thus they are steadily captured in large numbers with a small variation in the pools by pot fishing. However, the adults can actively wander around riffles and runs, searching for animal food, and are attracted to the riffles by bait odors depending on the water flow. Therefore, the most effective catching method is to set the pots in riffles when the river is swollen.     

Models of stream bed topography at the reach scale

A simple model of channel geometry is presented which focuses on the way that a single form variable may be autocorrelated with upstream values of that variable and which is amenable to analysis using time series methods. Stream bed topography at the local scale is the variable of interest. Bed height series were obtained for four stream lengths located at different positions along a mountain stream to test the hypothesis that, as control variables change spatially, the form of the model will also change. The Box-Jenkins models describing the series are dominated by autoregressive terms and indicate a non-random variation of bed height. An increase in the order and parameter magnitude of the models is associated with an increase in discharge and a decrease in bed material size, suggesting that the structure of the model reflects the adjustability of the stream bed and, specifically, the extent of riffle-pool development, although channel gradients may be too high for a well-defined riffle-pool sequence to be developed. Changes in the structure of models obtained for bed height and other form variables may indicate the variable ability of a stream to adjust its channel morphology.     

Reenvisioning velocity reversal as a diversity of hydraulic patch behaviours

Past research investigated the surpassing of mean velocity at riffle cross sections by that at pool cross sections for flows up to bankfull, termed ‘velocity reversals’, to understand one mechanism by which riffle–pool relief is maintained. This study reenvisioned the classic velocity reversal by documenting stage‐dependent changes to the locations of peak velocity without cross sections. Instead, the dynamics of peak velocity patches were considered for flows spanning 0.2 to 22 times bankfull discharge through the use of a high‐resolution DEM and two‐dimensional hydrodynamic modelling. A remarkable diversity in peak velocity patch behaviour was found across discharges, including gradual expansion and shifting as well as abrupt disappearance and emergence relative to the low‐flow patch locations. These behaviours blended together to varying degrees to produce many reversals in peak velocity across morphological units, but it took substantially higher than bankfull discharge for peak velocities to move from riffles and chutes to fast glides and pools. The discharges at which reversals occurred among morphological units were significantly higher for the valley‐confined reach than for the anastomosing reach. Copyright © 2016 John Wiley & Sons, Ltd.