Robert Newbury Err Design Notes-2011

1
Designing Pool and Riffle Streams
Lectures and Exercises
River Restoration NW Short Course 2008
Oulette Creek BC riffle at design flow
Robert Newbury PhD PEng

www.newbury-hydraulics.com
www.unb.ca/cri
2
Introduction
Streams are ancient corridors that have existed for thousands of years. We are recent occupants
in these museum pieces, fascinated by their history and attracted by their utility for water supply,
hydro power, agriculture and recreation. But the streams are not set pieces; they grow and shrink
as arterial veins in the watershed. Our changes in the watershed and floodplains alter their size,
shape and functions. Their new ecology may not support fish habitats and higher quality water.
Restoring some of their previous ecology and creating replacement habitats is the objective of
pool and riffle designs. This is not done in isolation. In the same sense as home renovation,
creating pools and riffles is usually stream renovation, limited by a wide range of uses,
ownerships and compliances.
Three riffle design options are considered; back-flooded step channels, meandering run-pool-
riffle channels and isolated riffles that create fish passage pools. A brief review of a few aspects
of river morphology and open channel hydraulics extracted from the CRI field courses are
contained in the first two chapters.
A detailed design for a stepped pool West Coast salmon stream restoration project is presented
in the third chapter.
The application of three open-access programs, HEC-RAS, CHUTE, and FishXing to the design
example is presented in the fourth chapter (see references for download sites).
Appendix A contains a summary of the geometry calculations required to design riffle-pool

sequences and construction procedures.
The fine print: The analyses and designs presented in these notes were approved by all fisheries and water authorities but
no responsibility is taken for their application or interpretation.
3
Contents
Introduction: Stream Restoration 2
1. Review of River Forms and Forces 4
1.1 meanders, pools and riffles

1.2 river bed stability
1.3 hydraulic forces
2. Review of Open Channel Hydraulics 11
2.1 uniform flow

2.2 Chezy and Manning equations
2.3 normal depth
2.4 flow resistance or roughness factors
2.5 backwater curves
2.6 backwater calculations
2.7 rapidly varied local flow
2.8 critical flow
2.9 flow transitions
2.10 specific energy and instream structures
2.11 riffle elevation
2.12 channel constrictions
2.13 pool and riffle profiles
3. Project Design Example: Stepped Pool and Riffle Channel 29
3.1 background: Oulette Creek project

3.2 channel conveyance calculations
3.3 design goals and constraints
3.4 riffle design and construction
3.5 monitoring
4. Programmed Solutions: HEC-RAS, CHUTE, Fish Xing 41
4.1 HEC-RAS
4.2 CHUTE
4.3 FishXing
References 48
Appendix A: Riffle and Pool Geometry Calculation and Construction 49
4
1: Review of River Forms and Forces
1.1 Meanders, Pools and Riffles: The three-dimensional sine wave followed by flowing water
has been observed in glacier ice, the Gulf Stream, and in straight man-made canals (Henderson
1966, Leopold, Wolman and Miller 1964). Theory suggests that the waveform is generated by
turbulent bursts that are scaled to the width and depth of the flow at the channel-forming flood
discharge (Yalin and da Silva 2001). In mobile bed streams, the legacy of the wave is visible after
the flood has passed; pools at the low point of the wave, riffles at the high point of the wave and
meanders that follow the lateral extent of the wave. Two theoretical sine waves expressed as
profile and plan views are shown in Figure 1.
Figure 1: Sine waves generated

as a profile (2 cycles) and a plan
(1 cycle) for a theoretical 10 m
wide channel with a meander
amplitude of 30 m and a pool
depth of 1 m. The 120 m
meander is measured along the
axis of a 1% gradient valley.
distance downstream
Pool, riffle and meander geometries vary widely depending upon the valley materials and man-
made interruptions. It has been observed in many alluvial channels that the average length of a
meander wave is 10 to 14 times the channel width (Figure 2). Because the wave consists of two
pools and two crossover riffles, the average pool and riffle spacing varies from 5 to 7 channel
widths (approximately 2 cycles). The average radius of curvature of natural meander bends in
plan was found to be 2.4 channel widths (Chang 1988). However, whole valley surveys, air photo
reconnaissance and the application of geomorphologic tools are required to establish local
conditions and anomalies in the pattern (see Kondolf and Piegay 2003 for a complete
description).
5
high flow
low flow
hydraulic slope
riffle pool riffle

pool riffle
pool riffle
+ radius of bend
meander length 10 to 14 X channel width
Figure 2: average pool, riffle and meander dimensions observed in some mobile bed streams.
The migration of meandering channels along the valley bottom is affected by the mobility of the
valley materials, riparian vegetation, the inherited valley dimensions of previous rivers and man-
made obstructions. Consequently, uniform meanders that migrate down alluvial valleys thout
distortion are seldom seen. An interesting exercise in understanding meander migration is to
trace an outline of the present river channel from a suitably scaled air photo onto a transparent
sheet. By sliding the sheet upstream and downstream, remnants of former meanders can often
be discovered and subsequently dated by tree ring counts or old mapping. The effects of
channelization and river training works are often discernable as well.
The meander amplitude is highly variable. Differences in valley materials, constrictions such as
bridges or changes in gradient may cause upstream meanders to widen and shorten, often
overtaking downstream meanders and cutting through them to leave “oxbow” lakes on the valley
margins. In rocky channels, the meander may be suppressed altogether but the vertical wave-like
tendency of the flow may be observed as regularly sorted bed deposits in riffles and runs. Even in
the straight and uniform concrete-lined channels, the flow can be observed to oscillate from one
side of the channel to the other, often with a wave length of about 12 times the channel width.
1.2 River Bed Stability: Empirical relationships for the movement of bed materials have been
derived largely from observations in canals and laboratory flumes. Modeling of the forces causing
motion must be simplified, as natural stream patterns and shear stresses are too complex to
model in detail. As a consequence, empirical coefficients to adjust for simplified modeling vary
widely to accommodate so many influences. Assuming uniform flow conditions, an estimate of
6
the shear force (or tractive force) acting on the bed can be derived by assuming that it is equal to
the down-slope component of the force of gravity acting on the flow. In metric units, the tractive
force T (kg/m2) is the product of the density of water (1000 kg/m3), D the depth of flow (m) and S
the slope of the water surface, or
T = 1000 D S (kg/m2)
The empirical relationship between the tractive force and the size of bed material that can be
rolled or skipped along the bed varies widely. In fine-grained materials (<2 mm diameter),
cohesion binds the materials to a break-up threshold before motion occurs, producing a wide
range of observed values. In non-cohesive materials… coarse sands, gravels and cobbles for
example, the observations are less variable and generally contained within the envelope curve
shown in Figure 3 developed by Lane (1955) where
 (diameter at incipient motion cm) = T (kg/m 2).
Figure 3: Relationshi s between the tractive force in a reach and the bed material size in motion (Lane 1955). As
the bed material coarsens (>10mm), the maximum tractive force (kg/m2) at incipient motion is approximately
equal to the bed material diameter (cm).
These observations have also been applied in selecting riprap sizes with a safety factor of 1.5
(US FHA 1988) where riprap  (cm) = 1500 D S. The tractive force exerted by a given flow
compared to the critical tractive force to move a fraction of the bed material is combined with the
duration of flows in historical bed load transport formulas (Du Boys 1879, Bagnold 1966, reviewed
in Kondolf and Piegay 2003).
7
Local erosion adjacent to irregular channel configurations where the flow is concentrated between
boulders or around projecting logs may be related to increased velocities predicted by a flow
"net". The flow net consists of lines representing the path of the flow. Perpendicular to them there
are equipotential lines representing the energy level of the flow. Although the flow is turbulent, the
lines of flow are used to represent theoretically independent "tubes" that conduct discrete portions
of the discharge. There are numerical solutions for plotting flow nets but they are often estimated
by sketching and adjusting the net until all the lines intersect at approximately 90 degrees. For
example, in the side view of a boulder obstruction shown in Figure 4, the flow depth decreases
beside the boulder as the velocity must increase in order to pass the same discharge through the
partially obstructed channel. In front of the boulder, the water rises as the kinetic energy of the
flow has been converted to depth because the velocity is zero at the stagnation point (see Review
2 for definitions).
Figure 4: A flow net of path lines and

equipotential lines around an emergent
boulder.
The change in velocity from the upstream channel to the flow adjacent to the boulder is inversely
proportional to the ratio of the size of the sides of the flow net cells. This is because the same
portion of the flow is assumed to pass through the smaller net cell. If the velocity of the
8
approaching flow is known, the velocity adjacent to the boulder may be predicted by the change
in cell size. This method cannot be used in a zone where the flow separates to form eddies
beside and behind the boulder.
A relationship between the velocity and the size of material transported is required to evaluate the
bed stability in this case because the average slope of the channel used in the tractive force
equation does not describe the local conditions. A compilation of empirical observations is shown
in Figure 5 (ASCE 1967).
Figure 5: Velocity and streambed material size in transport observed for finer grained bed materials (ASCE 1967).
The transport relationships with velocity, depth and slope are also useful in predicting bed load
movement. The duration of these factors may be compared to the range of bed material size to
estimate their stability and rate of transport. In designing structures and stable channels, the bed
stability may be predicted by selecting rip-rap materials using conservative assumptions of the
tractive force and factors of safety. For example, this is the method used to predict stable riprap
sizes for rock spillways or riffles in the program CHUTE (CSIRO 2003).
1.3 Hydraulic Forces: For very large materials in the stream such as isolated boulders, root
wads, weirs, logs, and occasionally machines, the moving forces exerted on the object must be
analyzed individually. One approach is to isolate the object in a free body diagram to solve for the
net downstream force Fx exerted by the flow. This is equal to the difference between the
upstream and downstream static pressures and the change in momentum of the flow (Chow
1959), or
Fx =  q C (Vup – Vdown) / 9.81
9
where Fx = the force exerted by the flow in the x direction (kg),  = the density of water at the
earth’s surface (1000 kg/m 2), q = the discharge that would have flowed through the cross-section
of the submerged object perpendicular to the flow (m3/s) and (Vup – Vdown) is the net change in
velocity upstream and immediately downstream from the body. In the case of an impermeable
body Vdown = 0. The value of C, the drag coefficient, depends upon the turbulence of the flow and
the streamlining of the body. For large blunt objects, the drag coefficient approaches 1. For
streamlined shapes such as a trout body it may be as little as 0.015 (see Vogel, 1994 for an
interesting discussion of this complex number and its effect on flow patterns and water
pressures). The force exerted on a blunt impermeable object that is submerged on both the
upstream and downstream sides to the same level can be approximated by rounding off and
reducing the equation to:
Fx = 100 Ax V2
where Ax is the submerged cross-sectional area blocking the flow (m 2) and V is the approaching
velocity (m/s). For example, for a 1.5 m diameter boulder fully submerged in a 3.0 m/s flow, the
force exerted would be Fx = 100 ( 0.752) 3.02 = 1590 kg. The frictional force resisting motion
between the streambed and the bottom of the boulder (assuming it is embedded enough not to
roll) may be estimated as the submerged weight of the boulder times the coefficient of friction
between the boulder and the bed. In this case, if the boulder had a specific weight of 2.67
(granite) and the coefficient of friction was a maximum of 1, the resisting force would be
approximately (2670 – 1000) ( 1.53 / 6) (1) = 2951 kg. The boulder would not slide but it may roll
or fall into a downstream scour-hole. Porous sedimentary rocks that are much lighter than granite
may slide in the 3.0 m/s flow.
Another interesting application of the momentum equation estimates the depth of flow at which a
wading person will be swept away (Figure 6). The cross-sectional area exposed to the flow
depends on the width of the wader B (front or side exposed to the flow) and the depth of the flow.
Standard sloppy waders are approximately 0.4 m wide on the face side and 0.2 m wide on edge.
The force exerted by the flow is Fx = 100 B (depth) (V2). The frictional force that resists being
swept away depends upon the weight and height of the wader (W t kg in air and Ht m) and the
coefficient of friction  between the soles of the wader and the stream bed where
Fresistance = W t ( 1 – depth/Ht). As most experiments in being swept downstream are uncontrolled,
there is little data on the coefficient of friction between rubber and rock. A useful substitute may
be a value of  = 0.6 for rubber tires on wet pavement.
A wader would be swept away at the depth at which the two forces are equal or:
depth away (m) =  W t / (100 B V2 +  W t/Ht)
10
For example, a 1.8 m tall wader weighing 80 kg stepping sideways across a 2 m/s flowing stream
in rubber-soled waders would be swept away at 0.45 m. Aluminum cleats would help the wader
stay in place by increasing the coefficient of friction between the soles and river bed. Carrying
more weight above the water line is possible but a field-tested velocity and safe depth chart for
your own weight, height and type of waders is a better idea.
Figure 6: Some wader stability factors.
11
2: Review of Open Channel Hydraulics

“Remember in discoursing on the flow of water to adduce first experience and then reason.”
Leonardo da Vinci
The distribution and variation of flow in natural channels is affected by meander curves,
obstructions on the channel bed and abrupt transitions in the channel geometry. Uniform flow
occurs when a constant discharge flows at a fixed or “normal” depth governed by friction, the
channel geometry and the slope of the bed. Because there is no slip between the flow and the
channel boundaries, the velocity of the flow varies from zero at the wetted perimeter to a
maximum near the centre of the cross-section. In rugged channels where rapids, pools and
cross-sectional changes occur abruptly, the velocity distribution and depth is governed by the
momentum of the flow as well as the channel geometry. These are the conditions of rapidly-
varied flow. As the discharge decreases the flow may change from uniform to rapidly varied
conditions, making the prediction of low flow velocities and depths difficult. Direct observations
and reference river data are usually required.
2.1 Uniform Flow: In analyzing uniform flow it is assumed that all of the energy losses in the flow
can be modeled as frictional drag on the wetted channel boundaries (Chow 1959). With these
assumptions, a “pseudo-flow”, really a flux of water through the cross-section with a uniform
velocity distribution, may be analyzed as a sliding block as outlined by the wetted perimeter in
Figure 7.
Figure 7: The wetted perimeter of a trapezoidal block of water at the bankfull stage in the uniform channel of
Black Creek ON is outlined in black.
12
At a constant velocity, the down-slope component of gravity acting on the block of flow is equal to
the frictional resistance forces acting on the wetted perimeter of the block shown schematically in
Figure 8.
Figure 8: The sliding block analysis in a uniform channel used to predict the average velocity in the Chezy and
Manning equations. In effect, the flow is the net flux of water in the channel, disregarding the actual velocity
variations in pools, riffles, backwater zones and eddies. The assumption of a single velocity for the whole section
or slices of the section limits the distinction of river habitat hydraulics.
For a block of flow of length L with a cross-sectional area A and a wetted perimeter of P, the
steady state forces may be written as:
g A L sin  = T P L
where  = the density of water on the earth’s surface (1000 kg/m 3), g = gravitational acceleration
(9.81 m/sec2), A = the cross-sectional area of the flow (m 2), L = the length of the flowing block of
water (m), P = the wetted perimeter of the flow (m),   the slope angle of the channel bed and
water surface and T = the unit area frictional force exerted by the stream bed on the flowing water
(Newtons/m2). This may be simplified as
T = g R S (N/m2)
where R is the hydraulic radius (m) = the cross-sectional area divided by the wetted perimeter
and S = the slope of the water surface and channel bed. S is approximately equal to sin  in
mildly sloping rivers. In some engineering notations for wide shallow channels, T may be divided
by g to estimate the tractive force of the flow on the channel bed as:
T = 1000 D S (kg/m2)
2.2 Chezy and Manning Equations: In 1769, Antoine Chezy was

charged with predicting the flow in a proposed water supply canal for
Paris. His observations of the velocity of wax balls used as floats in the
Seine River and nearby canals suggested that T varied with the square of
the velocity. Consequently V2 was substituted for T to derive his famous
13
velocity equation V = C R1/2 S1/2 where C = Chezy's flow resistance factor. Chezy’s C is a
combination of constants that provides a frictional resistance index peculiar to a river and stage.
In later studies by Manning (1890), Chezy's C was replaced by Manning's channel roughness
factor n where C = R1/6 / n, recognizing that the resistance to the flow generally decreases as the
depth increases in an open channel. The Manning’s version of Chezy’s equation generally used
in North America for the uniform flow velocity in an open channel became:
V = R2/3 S1/2
n
where V = the average velocity (m/sec), R = the hydraulic radius (m), S = the slope of the channel
and n = the Manning roughness factor (with implied units of T/L1/6, Chow 1959). The discharge Q
(m3/s) is the net flux of water passing through the wetted channel cross-section A (m 2). This is
also known as the continuity equation (Q = VA) in the sense that the same discharge can be
conducted with different combinations of velocity and cross-sectional area.
2.3 Normal Depth: Manning’s or Chezy’s equations may also be used to predict the depth of
flow in the channel for a given cross-section and discharge. This is defined as the “normal depth”
governed by the slope, cross-section and roughness of the channel. If the minimum channel
cross-section required for the discharge is optimized, the solution produces a semi-circular cross-
section that is commonly constructed as a trapezoidal canal section. This “optimal” cross-section
is often far narrower than that observed for the same discharge in wider alluvial channels.
Consequently care must be taken to limit the tractive force exerted on previously deposited bed
materials of natural channels and alluvial fans.
2.4 Flow Resistance or Roughness Factors: The difficulty of applying either the Chezy or
Manning equations in wide and diverse natural channel lies in the approximations that were made
to simplify the flow. The single roughness factor (Chezy’s C or Manning’s n) must accommodate
all of the empirical relationships with velocity distribution, turbulence, and surface roughness in
one constant. Three common prediction methods for estimating the all-encompassing Manning’s
n coefficient are:
1) Bed Material Approach for Smooth Channels: In reaches with smooth beds where the depth of
flow is much greater than the median size of the bed surface materials (5X or more), Manning’s n
has been related to the bed material diameter in laboratory studies by Strickler (Chow 1959) as:
n = .04 (50 )1/6
where 50 is the median diameter (m) of the largest sizes of particles on the stream bed that
protrude into the flow (Chang 1988). A “Wolman” random pebble count of surface materials is
often used to sample these surface materials (see Harrleson et al 1994). This limits the upper
value of the roughness estimate to n = .04 for most river depths and bed materials.
14
The roughness value under low flow conditions and in rugged channels could be considerably
higher. As the depth of flow decreases, the “relative roughness” ratio of the depth to the bed
material decreases and a larger portion of the flow is directly impacted by the protruding materials
(Figure 11).
2) Composite n Approach for Complex Channels: Factors in addition to the bed material size may
be used to modify the Manning’s n estimate (Chow 1959). To apply this approach, a basic value
nb (Table 1) is chosen for the channel type and adjusted for meandering after small increments
are added for other resistance features as:
n = m (nb + n1 + n2 + n3 + n4)
where: nb = the base value chose for the channel type (Table 1),
n1 = addition for bed irregularities ranging from .001 for channels with mildly
slumping banks to .02 for irregular bedrock channels,
n2 = addition for changes in the channel cross-section ranging from .001 for
channels with shifting flows to .015 for tortuous channels,
n3 = addition for obstructions ranging from .005 for channels with isolated objects
to .05 for channels where obstructions occupy 50% of the cross-section,
n4 = addition for vegetation ranging from .002 for low grasses in the channel that
are less than one third of the depth of flow to 0.1 in channels where small trees,
cattails or brush obstruct the flow, and
m = varies from 1.15 for mildly meandering channels (channel length to valley
length ratios from 1.2 to 1.5) to 1.3 for fully meandering channels (channel
length to valley length ratios greater than 1.5).
Channel or Median Size of Base Value of

Floodplain Surface Bed Material mm Manning’s nb
concrete .012 - .018

firm soil, grassed .025 - .032
coarse sand 1–2 .026 - .035
gravel 2 – 64 .028 - .035
cobble 64 – 256 .030 - .050
boulder 256 + .040 - .070
Table 1: Common base values of Manning's n (Chow 1959)
The utility of the composite n approach is often limited by the data available in smaller streams
and rivers. Where there are hydrometric surveys of the channel at different discharges, it can be
useful in assigning factors for the variation in observed n values. These can be transferred to
other channels in similar settings but only with some caution and field evidence.
15
3) Pictorial Guide Approach: Roughness coefficients may be chosen by comparing the geometry,
bed materials and appearance of the design channel to index channels that have observed
roughness values (Barnes 1967 USA, Hicks and Mason 1991 New Zealand). The New Zealand
guide provides roughness values for a range of discharge. Similarly, a local roughness guide can
be compiled using surveyed streams and water level gauges. For example in the sample reach of
Chapman Creek BC shown in Figures 9 and 10 the n value varies from 0.15 at low flow to 0.055
at the annual median flood level (Figure 11).
Figure 9: roughness reference reach in Chapman Creek, BC (discharge 2.4 m 3/s, n = 0.15).
Figure 10: roughness reference reach in Chapman Creek, BC (discharge 53 m3/s, n = 0.055).
16
Figure 11: roughness variation with discharge in the steep cobble-boulder channel of Chapman Creek BC
(gradient 3%).
2.5 The Special Case of Gradually Varied “Uniform” Flow in Backwater Curves: The
“normal” depth of flow Dn and velocity Vn are governed by the frictional forces acting on the
channel boundaries and the downslope component of the gravitational force in a stream reach.
However, the normal depth may be altered locally by obstructions or changes in the channel
dimensions. This will affect the depth and velocity of the flow in adjacent reaches as they adjust
to meet the depth in the altered reach. The curving water surface in the adjustment zone between
the reaches is known as the backwater curve. Backwater curves for a variety of flow conditions
have been classified with a letter and number grade in open channels (Chow 1959). Two
commonly encountered backwater curves in mildly sloping channels are shown in Figure 12. The
M1 curve is formed when an obstruction at the downstream end of the reach such as a weir,
anchor ice or a log jam increases the normal depth D n to Do. An M2 curve is formed when the
depth of flow at the downstream end of the reach decreases to less than the normal value, for
example as the flow approaches a rapid or falls. The curves extend upstream from the
downstream control point and approach the normal depth asymptotically. The computed length of
the curve L is determined by the accuracy required. It is commonly considered terminated when
the depth has recovered to 95 or 98 percent of the normal depth.
17
Dn
M1 M2
Do
Do
So So
Figure 12: M1 and M2 backwater curves in mildly sloping channels caused by changes in the depth of flow at a
controlling section at the downstream end of a reach.
2.6 Backwater Curve Calculations: The approximate shape of the curve can be determined by
dividing the backwater reach into small segments which are short enough to approximate the
curve as a series of straight lines. A single segment is shown in Figure 13. The energy loss
gradient Se is determined by calculating the gradient at each end of the segment using Manning’s
equation for an assumed increment in depth and averaging the values. The distance from datum
to the energy gradient is equal to the elevation + depth + kinetic energy of the flow V2/2g.
2
V1 /2g Seav. Se L
2
V2 /2g
D1
D2
SoL
L
Figure 13: The hydraulic dimensions of a short segment of a backwater curve reach.
For simple channels measured above a constant datum throughout the segment,
V12/2g + D 1+ S L 2
o = V /2g
2 + D + 2S Le
where So is the slope of the channel, Se av. is the average slope of the energy gradient and L is
the length of a segment for a small increment in depth. Re-arranging the terms for the simple
channel equation to solve for the length of the segment :
L = (V22/2g + D 2) - (V 12/2g + D 1)
So - Se av.
An M1 backwater curve for a reach of Battle Creek above the Alberta/Saskatchewan border is
shown in Figure 14. The curve was divided into the 5 straight-line segments for assumed
increments in depth (Table 2). Landslide debris observed by Sauchyn and Lemmen (1996)
18
entered the channel while the stream was flowing at a discharge of 11.4 cum/s in a 2 m deep
entrenched section of the reach (Figure 15). In this example the debris raised the depth of flow
from 0.9m to 1.72m.The estimated backwater effect extends over 140m to the point where the
depth has recovered to within two percent of the normal depth.
station D A V H R Se Se av. So - H L L
2
m m m/s m m Se m m m
0 1.72 10.3 1.10 1.78 1.09 .0013 0

2 .00165 .00735 0.20 27.2
1 1.50 9.00 1.27 1.58 1.00 .0020 27.2
.00250 .00650 0.17 26.2
2 1.30 7.80 1.46 1.41 0.91 .0030 53.4
.00395 .00505 0.16 31.7
3 1.10 6.60 1.73 1.25 0.80 .0049 85.1
.00625 .00275 0.10 36.4
4 0.95 5.70 2.00 1.15 0.72 .0076 121.5
.00810 .0009 0.02 22.2
5 0.91 5.46 2.09 1.13 0.70 .0086 143.7
Table 2: M1 backwater curve step calculations for a 1 m deep landslide fill in Battle Creek while it is flowing at
11.4 m3/s in a reach above the AB/SK border. In the table D = depth, A = cross-sectional area, V = velocity, H =
specific energy (D + V2/2g), R = hydraulic radius, Se = the energy line gradient, Se av. = the average energy line
gradient in the segment, So = slope of the channel bed (0.009), H = change in specific energy in the segment, L
= segment length, L = total length.
143.7 m
Dn = 0.90 m M1
0.91m
0.95m
1.10m
1.30m 1.72m
1.50m
landslide
22.2m 36.4m 31.7m 26.2m 27.2m

5 4 3 2 1 0 stations
Figure 14: Backwater curve depth and distance segments calculated in Table 2 for a 1 m landslide fill in Battle
Creek below Police Point, AB.
19
Figure 15: Landslide debris infilling the upper reaches of Battle Creek near Police Point SK (photograph D.
Sauchyn 1998).
The backwater calculations in natural rivers with a wide variety of cross-sections, roughness
values and alignments can be tedious. There are public access programs like the HEC-RAS
(USACE) that can conveniently provide solutions and graphic outputs (see Section 4.1).
2.7 Rapidly Varied Flow: Rapidly varied hydraulic conditions describe the velocity and trajectory
of flow where the geometry is shaped by the flow boundaries and by the momentum of cells of
water moving within one another. For example, this is the condition in whitewater reaches where
rapid flows penetrate downstream quiescent flows entraining air and generating noise as the air
bubbles rise and break on the surface. Vertical and horizontal eddies in the slower receiving flow
are driven by the faster flow. These hydraulic “compartments” of flow conditions are utilized by
many aquatic plants and organisms. Benthic insects seek the unique accelerating flows around
and over cobbles where there is maximum light penetration and plant growth on the bed of the
stream. Fish find passage and refuge in eddies and vortices that have no solid boundaries but are
part of the structure of the flow.
Contrary to the uniform flow equation assumptions, a surprising amount of water flows across and
upstream in the cross-section. In the reach of Chapman Creek shown in Figure 16 over half of the
flow is moving upstream in 4 large backeddiies.
20
Figure 16: In this short reach at the head of a pool, 1/2 of the flow directions are upstream or across the channel
in two back eddies on the sides of the channel and a “horseshoe vortex” formed behind the boulder. Aeration
occurs as the rapid flow penetrates the slower-moving pool water, carrying air under the surface that re-emerges
as noisy bubble-popping white water.
Energy losses in rapidly varied flow are primarily from contraction, expansion and deceleration.
These conditions occur over such a short reach that effects of frictional losses on the boundaries
are comparatively small. To accommodate the turbulent energy loss using uniform flow
equations, surrogate roughness factors and expansion or contraction losses must be applied.
The hydraulics of man-made structures can be adapted to at least partially describe the diverse
structure of the flow. This requires the division of the energy of the flow into kinetic and potential
components that will describe the state of the flow.
Potential and kinetic energy components are expressed in an equation named for Daniel
Bernoulli (1700-1782) in recognition of his founding hydraulic concepts by Euler (1707-1783):
E = Z + D + V2/2g (Bernoulli's equation)
Applied to stream channels, Z = the elevation of the channel bed above a datum (m),
D = depth of flow (m), V2/2g = the kinetic energy of the flow (m), g = the acceleration due to
gravity (9.8 m/sec2 ), V = the mean velocity (m/sec) and E = the total energy of the flow at that
point in the channel. The specific energy of the flow H (m) is the sum of the only the depth and
kinetic energy at that point in the channel:
H = D + V2/2g.
21
Three categories of velocity and depth combinations or “states of flow” can occur for the same
discharge (Figure 17) but with different specific energies:
 subcritical state (typical of most mildly sloping natural channels, high H),
 critical state (occurring at riffle crests, in flow over boulders or in steeply sloping
streams at high discharge, minimum H), and
 supercritical state (waterfalls and short vertical drops below boulders and
ledges, high H).
Figure 17: Three possible categories for combinations of velocity and depth at the same discharge. The state of
the flow is also described by the Froude Number FR = V / (gd)1/2.
The minimum specific energy H for a given discharge occurs when the depth of flow is “critical”.
This occurs when the depth of flow in a steep channel or over an obstruction such as a weir or
riffle crest is 2/3 of H. Consequently, the kinetic energy at the “critical” overflowing velocity V c is
1/3 H. This is an important state of flow with a FR of 1.0 that it is the benchmark against which
other states of flow are described.
The transition from the subcritical to supercritical state occurs smoothly as the accelerating flow is
drawn forwards into the faster flowing zone. The turbulent decelerating transition from the
supercritical to subcritical state is the opposite. In the case of plunge pools, the fast flow
penetrates the slower flow or pool as a coherent torrent. Air carried below the surface emerges to
form white water and sound. The shear force of the torrent spins turbulent cells in the slower flow
forming backeddiies and vortices.
2.8 Critical Flow: Critical flow may occur in a steeply sloping stream at higher discharges but it is
most often observed on the top of an obstacle where water is falling freely over the surface, for
example, as water flows over a boulder or weir and falls freely down the lower side. On top of the
22
obstacle, it flows with the critical velocity at the critical depth with the minimum specific energy
required for that particular discharge. As V c2 /2g is 1/3 H and Dcis 2/3 H
Vc = (g Dc )½.
By combining equations, the critical depth and specific energy may be expressed in terms of the
discharge and width of the flow as:
Dc = ( Q2 / gW 2 )1/3 and Hc = 1.5 Dc = 1.5 ( Q2 / gW 2 )1/3
where Q = discharge (m 3/s) and W = the width of the flow (m). The value (gD)1/2 is also the
velocity of a shallow wave caused by a disturbance in a still pond of with a depth of D.
Consequently standing waves of disturbance often form or slowly oscillate upstream and
downstream in sections flowing at the critical velocity because their upstream progress is trapped
in place by the coincident downstream velocity (Figure 18).
Figure 18: Standing waves on the edge of a riffle crest in the critical velocity zone. The smooth supercritical flow
zone occurs in the central throat of the crest.
2.9 Flow Transitions: The state of the flow at a specific habitat location and the state of the
surrounding flows can provide useful insight for partitioning the hydraulic structure of stream
habitats. For example, in the simplified flow structure created by the weirs and rapids shown in
Figure 19, the state of the flow shifts from (1) sub-critical to (2) critical to (3) supercritical and then
back to sub-critical through a hydraulic jump (4).
23
Figure 19: Changing states of flow in man-made and natural structures (H and h represent depth).
In the accelerating transition 1 to 2, the flow is drawn smoothly over and around the obstruction.
The glassy surface and smooth flow lines are often mistaken for laminar flow but that occurs only
in the first few millimeters or so next to the channel boundaries. In the decelerating transition, the
condition is exactly the opposite. The shallow rapid flow penetrates the deeper and slower flow,
sweeping air into the water and causing internal rollers and back-eddies within the flow. The
transition occurs abruptly in the form of a hydraulic jump. The position of the jump, for example on
a back-flooded riffle, may be predicted by calculating the upstream or downstream backwater
profiles and searching for the condition at which:
y1 / y2 = ½ ((1+ 8 FR12)1/2 – 1)
where y1 and y2 are the upstream and downstream depths and FR1 is the Froude number of the
upstream flow (Chow 1959). This calculation may be easily programmed using HEC-RAS and
CHUTE.
24
There are several types of hydraulic jumps, varying from a simple wave form to a rapidly re-
circulating near-vertical face depending on the upstream Froude number and downstream depth
(Chow 1959). The transitions between fast and slow states of flow re repeated many times in
riffles and rapids (Figure 20). It is important to duplicate these conditions in built riffles and rapids
to create continuous threads of flow and resting areas for fish passage over a range of
discharges.
Figure 20: Critical overflows, hydraulic jumps, subcritical platforms and multiple changes of flow states in rocky
rapids exiting an upstream lake (the pre-diversion channel of the Burntwood River, MB).
2.10 Specific Energy and Instream Structures: All of the combinations of depth and velocity at
a given discharge and channel cross-section may be expressed in a specific energy curve where
the depth of flow D is plotted against the specific energy of the flow H (Figure 21). The point of
minimum energy occurs at the critical depth where the curve reverses slope between the
subcritical zone and supercritical zone, in this case at a depth of 1.0 m. If the normal depth is
greater than the critical depth in mildly sloping natural channels, there is some room for the
manipulation of the local cross-section without causing backwater effects. In the specific energy
curve shown in Figure 21, a low weir or riffle may be placed in the stream that raises the bed
locally by the height Z without causing a backwater effect.
25
Figure 21: All combinations of depth and velocity that can occur for a given discharge are shown in this specific
energy diagram for Chapman Creek BC. The minimum specific energy occurs at the critical depth.
2.11 Riffle Elevation: The critical flow and specific energy relationships may be used to estimate
the channel capacity when designing in-stream works that obstruct a small portion of the flow. For
example: a riffle was added to a reach of Chapman Creek BC to maintain a shallow pool at low
flow next to a fish hatchery. The average cross-section in the reach is 1.6 m deep and 21 m wide
(Figure 22). The median flood discharge of 68 m 3/s (50% annual frequency) just fills the channel
to the floodplain or bankfull stage. The question was, how deep a pool can be created behind the
riffle without increasing the natural flood frequency that would affect the hatchery grounds and
nearby trailer park? To meet this condition the height of the riffle must be less than or equal to
the difference between the specific energy of the normal depth in the channel and the minimum
specific energy required to pass the same discharge at the critical depth.
At the bankfull stage in the pool above the riffle at a discharge of 68 m 3/s the average velocity is
Q/(W x D) or 68 / (21 x 1.6) = 2.02 m/s. The specific energy is D + V2/2g = 1.6 + ((2.02)2/2x9.81)
= 1.6 + 0.21 = 1.81 m. This is greater than the specific energy required to pass the same flow at
26
the critical depth. The critical depth at 68 m 3/s = (Q2/gW 2)1/3 or (682/9.81(21)2)1/3 = 1.07 m.
The critical velocity = (gDc)1/2 = (9.81 x 1.07)1/2 = 3.24 m/s. The critical specific energy = 1.07 +
(3.24)2/ (2x 9.81) = 1.07 + 0.54 = 1.61 m. Therefore the maximum allowable height of the riffle Z
= 1.81 – 1.61 = 0.2 m for the target conditions. The depth and specific energy as the flow as it
approaches and passes over the riffle is shown in Figure 23.
Figure 22: A rock riffle was added to Chapman Creek to retain spawning gravels and maintain water depths. The
bankfull depth at the site is 1.6m. The height of the riffle crest is 0.2 m.
Figure 23: Schematic profile of the change in state of subcritical to critical flow in an open channel as it passes
over a riffle crest in Chapman Creek without increasing the flood stage. In more detailed analysis small
contraction and frictional losses will alter the dimensions slightly.
Note that meeting the bankfull stage with the median flood is only one option for the design. Other
stages and discharges may be selected for habitat or property protection. In many cases
27
channels that have been straightened or dredged become deeply entrenched and the design
objective is to raise the flood stage so that the floodplains are re-accessed by median floods.
2.12 Channel Constrictions: The critical flow analysis may be applied to determine how narrow
the stream can be made without causing local flooding as well, for example by intruding bridge
abutments, flow retarding bars, groynes or boulder clusters (Figure 24).
Figure 24: Schematic plan and profile of the minimum width between bridge piers that can be attained without
increasing the upstream energy line and water level at the bankfull flood stage in a normally subcritical channel.
In this case, the width of the stream may be narrowed locally until the specific energy of the
critical flow equals the normal specific energy in the reach. For example, if the specific energy for
1.6 m deep normal bankfull conditions in Chapman Creek is 1.81 m, the critical depth available
will be 2/3 (1.81) = 1.21 m. The critical velocity will be (9.81 x 1.21)1/2 = 3.45 m/s. To pass the
bankfull flow of 68 m3/s without backflooding, the width required will be 68 / (1.21 x 3.45) = 16.3m.
If contraction losses are minimized with a smooth transition between widths, the stream can be
narrowed locally without causing flooding by (21 – 16.3) = 4.7 m. The addition of the bridge piers
or boulder clusters of this width counter-intuitively causes the water level to decrease to less than
the normal stage in the narrower section because of the greater kinetic energy requirements of
the critical flow.
2.13 Pool and Riffle Profiles: The restoration of a pool and riffle profile to channelized streams
requires both rapidly varied and uniform flow analyses. Several flow patterns are possible
28
depending upon the steepness of the channel, the riffle heights and the pool lengths (Chanson
1994). Two common cases that occur in mildly sloping channels are shown in Figure 25.
Figure 25: Mixed hydraulic conditions in stepped or pool and riffle channels.
Although the height of the riffle may be set by target flood levels or pool depths as discussed in
the previous section, there are several options for the spacing of the riffles. Observations in
meandering streams suggest that the energy losses in the shallowly sloping pool segment are
approximately equal to the losses in the steep riffle segment (Leopold, Wolman and Miller 1964).
It is possible to decrease the pool velocities and potential bank erosion by shifting more of the
losses to the riffle section. For a given riffle height and average gradient this can be accomplished
by reducing the riffle spacing. If the spacing is close enough, the riffles will back-flood from the
downstream crest to a selected depth on the upstream riffle face creating a stepped channel that
drops from pool to pool (Case A, Figure 25). In contrast, the riffle spacing may be increased to
create a run-pool-riffle sequence where the distance between riffles is sufficient to allow the flow
to attain normal depth before encountering the upstream riffle (Case B, Figure 25).
A summary the calculations, dimensions and methods required to design and construct a pool
and riffle profile is given in Appendix A.
29
3. Salmon Project Design Example

Stepped Pool and Riffle Channel for Salmon Spawning and Over-wintering
Habitats (Oulette Creek BC) 1
3.1 Background: The

meandering natural channel of
Oulette Creek was straightened
and diverted in 1978 to create
a large dry-land log sorting
yard in Howe Sound BC on the
alluvial fan of the stream. The
600 m long trapezoidal
diversion channel was stepped
with alternating single log and
single row boulder drop
structures every 30.5 m (Figure
26). The additional shear stress
in the shortened channel
rapidly eroded the fan
materials. The bed elevation
stabilized 1.1 m below the
constructed grade in two
seasons, collapsing or under-
cutting the drop structures
(Figure 27). Coho salmon
spawning riffles and rearing
pools were destroyed. Pink and
chum spawning platforms in
the lower reaches were in-filled
with large cobbles transported
by the shorter and steeper Figure 26: The Oulette Creek diversion immediately after
channel (Figure 28). construction in 1978.
1The Oulette Creek restoration project (1994) was funded, built and monitored with the help of Rob Ladan (Terminal
Forest Products Ltd.), Grant McBain & Jim Wilson (Canada DFO) and Dave Bates (FSCI Biological Consulting).
30
A small hatchery was built at the site as a habitat loss compensation measure but fish returns to
the cobble-bed stream were insignificant and it was abandoned in 1988.
Figure 27: The Oulette diversion channel in 1982, four years after construction. Single
“digger” log and rock row drop structures contributed to the rapid down-cutting.
Figure 28: The uniform boulder and cobble channel of lower Oulette Creek in 1993, fifteen
years after it was constructed and prior to the addition of spawning pools and riffles.
In 1998, rock riffles were constructed at each of the former drop structure locations in the now
hardened channel (Figure 29). The riffles were designed to form a step channel that would
restore access to the floodplains at the median annual flood flow. This option reduces the shear
stress in the channel at higher flows, creates over-wintering pools and allows spawning size
gravels to be deposited above the riffles.
31
Figure 29: A larg boulder riffle under construction in the Oulette Creek diversion channel in 1994.
3.2 Channel Conveyance Calculations: Surveys undertaken in 1994 in the reach immediately
above the diversion found that the undisturbed bankfull channel width and depth were 7 m and
0.7 m (Table 3). The diverted reach had stabilized at a width of 9m with an average depth of 1.8
m. There was no evidence of flood flows reaching the level of the now elevated floodplain in the
entrenched diversion reach.
Table 3: Oulette Creek BC survey summary. The natural channel measurements were made in the undisturbed
reach immediately upstream from the diversion.
The natural channel bankfull velocity and discharge estimates using an estimated roughness
coefficient of 0.04 (Hicks and Mason 1991, reference channel 9065) are:
V = (0.7)2/3 (0.03)1/2 / 0.04 = 3.4 m/s

Q = 7 x 0.7 x 3.4 = 17 m 3/s
The bankfull discharge estimate is approximately equal to the median annual flood peak based
on a nearby gauging station adjusted exponentially by the ratio of drainage areas (Figure 30).
32
Figure 30: The annual flood frequency curve for Oulette Creek BC based on a nearby gauging station. The depth
of the median annual flood corresponded to either the vegetation trim-line or the floodplain level in the
undisturbed reach.
3.3 Design Goals and Constraints: The loss of spawning and overwintering areas is
widespread throughout the Strait of Georgia streams. As one of the first restoration demonstration
streams, Oulette Creek has had two design iterations to re-create target habitats for coho, pink
and chum salmon. The habitat goals identified by fisheries biologists were:
1. Coho spawning riffles must have sufficient head loss to allow water movement through the
gravels that deposit in the tail of the pool approaching the riffle crest,
2. pools greater than 1 m deep must be maintained for over-wintering Coho fry,
3. back-eddies with large wood and boulder covers are required for safe fry habitat, and
4. shallow spawning platforms with smaller gravel sizes are required for chum and pink salmon in
the lowest reaches.
33
To re-establish access to the floodplain, the riffle crests were elevated along the profile to the
original 1978 level (Figure 31).
Figure 31: Pool and riffle profile design constructed in the diverted reach of Oulette Creek BC. In this option the
pool and riffle dimensions are adjusted at each site to backflood and pass the 50% frequency flood at the
bankfull stage. The sample calculations are based on the average cross-section and slope.
Design Iterations: In the first three years (1994-1996) stable uniform spawning platforms were
established above all the riffles. In 1997 the upper riffle crests were adjusted to deepen and
narrow the flow. A new central torrent scoured 1.0 to 1.5 m deep pools and formed two strong
back-eddies on either channel border below the riffles (Figure 32). The lower two riffles were not
modified and continued to maintain shallow spawning platforms (Figure 33).
Constraints: By the year 2000, large cobbles and boulders from the steep upstream reach began
to fill the upper pool in the restored reach. The pool the cobbles and boulders were removed from
the pool and a small landing was constructed on the adjacent floodplain for periodic maintenance.
The accumulation of coarse materials in the upper reaches of the alluvial fan would have normally
caused an avulsion of the stream into an adjacent channel. However, because the location is
fixed in place with dykes the deposition is now confined to the channel. Unless it is removed
every 5 years or so before it enters the lower reaches it will eventually bury the riffles and force
the flood flows over the dykes. This is the recurring cost of keeping the stream in one location.
34
Figure 32: The upper Oulette Creek riffles were deepened and narrowed to concentrate the flow and scour a deep
central pool that is used by coho and steelhead fry as over-wintering habitat.
Figure 33: The lower Oulette Creek riffles with level crests allowed spawning platforms to deposit used by pink
and chum salmon in the shallow upstream pool. The broad exit flows do not form a scour hole below the riffle.
Spawning chum salmon are near the riffle crest and in the upper left portion of the photograph.
3.4 Riffle Design
35
a) flood capacity and height: Calculation of the allowable riffle height is based on the energy
line elevation when a bankfull flow of 17m 3/s is just entering the floodplain at 1.8 m in depth. The
bankfull velocity is Q / (w x d) = 17 / (9 x 1.8) = 1.05 m/s. The specific energy of the bankfull flow
is 1.8 + (1.05)2 / 2 (9.81) = 1.86 m. For a full width 9 m riffle, the critical depth of flow on the crest
will be
Dc9 = (Q2 / gW 2)1/3 = ((172) / 9.81(9)2)1/3 = 0.72 m
The specific energy of the critical flow on the riffle crest is 1.5 x 0.72 = 1.08m. Consequently, the
height of the 9 m wide riffle is 1.86 – 1.08 = 0.78 m at the bankfull discharge. This forces the
median flood to just reach the floodplain level as shown in Figure 34.
Figure 34: Oulette Creek flowing at 17 m3/s over a 9 m wide riffle. The upstream pool just approaches the
floodplain level.
When the upper riffles were narrowed to 6 m to form a central torrent, the critical depth on the
riffle crest increased to 0.94 m where
Dc6 = (Q2 / gW 2)1/3 = ((172) / 9.81(6)2)1/3 = 0.94 m
The specific energy of the narrowed flow is 1.5 x Dc =1.4 m. The allowable height of the central
crest was therefore 1.86 – 1.4 = 0.46 m, assuming again that the maximum depth of the
36
approaching flow must not exceed the bankfull condition. The critical velocity in the central torrent
= (9.81 x 0.94)1/2 = 3 m/s.
Graphical solutions for both riffle types are illustrated in the specific energy diagram (Figure35).
2.80
Oulett eCreek BC
2.60
specific energy curves
9 m and 6 m c rest widths
2.40
2.20
1.80
depth m
1.60
1.40
1.20
0.80
0.60
6m
0.40
9m
0.20
Q = 1 7 cum/ s
0
0 0.20 0.40 0.60 0.80 1 1.20 1.40 1.60 1.80 2 2.20 2.40 2.60 2.80 3
0.46 m
specific energy m
0.78 m
Figure 35: Specific energy curves for the Oulette Creek channelized reach (9m width, solid line and 6m wide,
dashed line). The corresponding riffle heights are 0.78 and 0.46 m resp.
The torrent of flow emerging from the narrowed riffle scours the upper pool and drives two large
backeddiies on either side of the channel. These were covered over with logs and debris to
provide shade and cover from predators, in this case primarily kingfishers.
The Reynolds number of the central torrent at the design discharge is V D /  or 3.0 x 0.94 / 1.3 x
10-6 = 2.17 x 106 (10oC water). Torrents with high Reynolds numbers entering pools have been
observed to cause fully-developed back eddy circulation and local gravel deposits. For example,
the distribution of Reynolds numbers and flow structures in the rapidly varied flow in Chapman
Creek (Figure 16, Section 2.7) are shown in Figure 36.
37
HORSESHOE VORTEX
Re = 80,000
Fr = 0.15
ROCK
Fr = 1
RETURNEDDY
Figure 36: Coherent structures created within the flow by torrents and back-eddies (Chapman Creek BC).
b: geometry and spacing: For convenience, diagrams and the equations to lay out the design
for construction are summarized in Appendix A. The dimensions for the Oulette riffles designed
with a 30.5 m spacing and 0.20 m of back-flooding are summarized in Table 4 on the following
page.
38
RIFFLE HEIGHT Oulette Creek

BC
Q design discharge m3/s 17
D depth of flow approaching riffle m 1.8
W average width of flow m 9
V approaching velocity m/s Q/WD 1.05
v2/2g velocity head (kinetic energy of 0.06

approaching flow) m
H specific energy m D + V2/2g 1.86
Dc critical depth of flow m Dc = (Q2 / gW 2)1/3 0.72

2
vc /2g critical velocity head m Vc = (g Dc)1/2 0.36
2
Hc critical specific energy m Hc = Dc + vc /2g 1.08
RH riffle height above channel bed m 0.78
DIMENSIONS
SB channel slope .03
SRU slope of upstream riffle face 0.5 (2:1)
SRD slope of downstream riffle face 0.1 (10:1)
RU distance of heel to crest m RU = RH / (SRU + SB) 1.47
RD distance of crest to toe m RD = RH / (SRD – SB) 11.14
YD height of bed at the crest above toe m YD = RD (SB) 0.33
total drop in chute m YD + RH 1.11
SPACING
L pool length with no back-flooding m L = RH / SB 26
BF height of back-flooding on upstream riffle 0.20

m
Istep interval between crests with back- Istep = L – (BF / SB) + RD 30.5
flooding m
Irun interval between crests with run and pool Irun = L + RD + run na
m
Table 4: Oulette Creek pool and riffle geometry calculation summary for the 9m wide crest (Figure 33).
3.5 Construction: Cobble and boulder sizes for the riffle crest and downstream face were based
on the conservative assumption that the critical flow would follow the upper segment of the riffle
39
slope at the bankfull stage. The maximum depth (0.94 m) and velocity (3 m/s) occur in the
narrowed riffle. The sloping face of the riffle was set at 1:10, producing a tractive force of 1000 x
.94 x 0.1 = 94 kg/m2 equivalent to a 94 cm diameter riprap size. Consequently large boulders up
to 1m in diameter were embedded in the upper parts of the riffle to create and back-up the crest
(Figure 29).
Forces exerted on the boulders placed on the riffle crest to narrow the flow (Figures 32 and 37)
were estimated using the momentum equation. The cross-section of the bankfull flow blocked by
1.0 m diameter boulders placed on the narrow riffle crest is approximately (π x 0.5 2) = 0.79 m2.
The force exerted by the flow is approximately 100 x A x V2 or 100 x .79 x (1.05)2 = 87 kg. Since
a 1m diameter submerged boulder weighs approximately 875 kg, the coefficient of sliding friction
would have to be exceedingly low before it would slide over the riffle. This is unlikely if the
boulders are nested together on the crest.
Figure 37: Scoured pools and back-eddies caused by the torrent formed between large boulders placed on the
riffle crests in the Coho spawning and rearing reach of Oulette Creek.
3.6 Monitoring: Annual biophysical and topographic surveys were undertaken in the re-
constructed reach of Oulette Creek for 7 years. The surveys have shown that approximately
equal areas of pool and riffle habitats have been maintained below the narrowed riffles (Figure
38). In response, the biomass of fish in the reach increased dramatically with 3X increases in
Coho fry and 8X increases in Cutthroat trout (Figure 39). Several dozen spawning pink and chum
salmon have been observed each year above the lower riffles (Figure 33).
40
RIFFLES
POOLS
Figure 38: The distribution of riffles and pools in the Oulette Creek diversion channel before and after the
addition of riffles in 1994.
Figure 39: The response of fish biomass to the creation of a pool and riffle profile in Oulette Creek.
41
4. Programmed Solutions
Programs can add detail and facilitate design iterations to riffle–pool designs but the basic inputs
must be determined using the hydraulic calculations and the criteria selected for the project.
Three programs applied to the Oulette Creek riffles and pools are discussed below.
4.1 HEC-RAS: Critical water elevations in a riffle and pool profile can be estimated in the HEC-
RAS open channel flow program. To operate the program, cross-sections in the study reach must
have a common datum and reference distance. There are several input protocols:
1. cross-section locations are entered from upstream to downstream
2. cross-section data is entered from the left looking downstream in an ascending order
of distances
3. the top of bank must be selected for each cross-section
4. the reach length to the next cross-section downstream must be specified
5. Manning’s roughness values must be selected for each portion of the cross-section
and each length of the reach
6. modeled discharges must be set at the upper end of a reach
7. expansion and contraction energy loss coefficients based on a fraction of the kinetic
energy change between sections must be specified
8. controlling water levels above and below the study reach must be specified with known
values, rating curves or as the normal depth in the approaching and leaving channels.
Output profiles for a two riffle one pool sequence using the Oulette Creek geometry calculations
are shown in Figure 40. The upper profile is predicted for a low discharge of 0.5 m 3/s. The lower
profile shows the water surface at the design bankfull flow of 17 m 3/s. The dotted red line shows
the where the critical depth would occur in the sequence, in this case over the riffle crests. The
dotted green line shows the floodplain level that is just attained by the bankfull discharge in the
pool above the riffles.
42
Figure 40: HEC-RAS water surface profile solutions for a typical pool-riffle sequence in Oulette Creek BC.
Only the control cross-sections are shown in Figure 40. Additional cross-sections are required to
obtain a more detailed water surface profile that shows where the hydraulic jump will occur
between the riffle crest and the downstream pool as shown in Figure 41.
The actual flow conditions will be more variable than those modeled with the uniform flow
assumptions. However, useful indicators of the change in channel stability and habitat conditions
may be determined by comparing the existing conditions with alternative designs. For example, at
the design flow of 17 m 3/s the modeled shear stress in the channel was 180 N/m 2 (18.3 kg/m2)
implying that bed materials up to 18.3 cm cobble size are swept through the channel at the
median flood level. In the detailed riffle pool model (Figure 41) the shear stress in the pool
segment at the median flood varies from 37 N/m 2 at the toe of the riffle to 12 N/m 2 above the
downstream riffle in the deepest part of the pool, implying that gravels between 3.7 cm and 1.2
cm may be deposited at the same flood stage. This was observed in the lower Pink and Chum
salmon spawning riffles on Oulette Creek (Figure 33).
43
Figure 41: Detailed HEC-RAS water surface profile for a pool riffle sequence in Oulette Creek BC showing the
location of the hydraulic jump.
4.2 CHUTE: The water surface profile, shear stress and suggested stable rock sizes are
computed in the Australian rock spillway design program “CHUTE”. A range of discharges can
be modeled based on the spillway geometry shown in Figure 42.
Figure 42: Dimensions of a rock spillway in the program CHUTE (CSIRO 2003)
44
The entry and output data for an Oulette Creek riffle for the DOS version of CHUTE are
summarized in Table 5. The downstream depth is approximately the pool elevation above the
downstream riffle minus the upstream riffle toe elevation.
ROCK CHUTE DESIGN
Effective drop (m) 1.11 Unit discharge (cumec/m) 1.89

Chute drop (m) 1.11 Critical depth (m) 0.71
Chute length (m) 11.1 Downstream depth (m) 1.27
Chute slope 0.100 Relative density of rock 2.50
Apron rise (m) 0.00 Angle of repose of rock (deg) 40
Apron length (m) 1.0
Apron slope 0.000 Safety factor 1.00
OVERALL PROPERTIES. NOTE - energy heads relative to downstream bed

level.
Upstream energy head (m) 2.18 Critical depth (m) 0.71
Downstream energy head (m) 1.38 Normal depth (m) 0.39
Energy loss - friction (m) 0.60 D50 (mm) 480
Max design bank angle(deg) 24.0 Mannings n 0.035
JUMP PROPERTIES
Position (m) 10.0 Upstream depth (m) 0.42
Energy loss (m) 0.20 Downstream depth (m) 1.13
D50 based on conditions at jump, as follows:-

Depth (m) 0.42
Energy slope 0.0837
WATER SURFACE PROFILE

Posn Bed elev Depth Velocity Energy Energy Conj.
(m) (m) (m) (m/s) Head(m) Slope Depth(m)
Chute 0.0 1.11 0.71 2.67 2.18 0.0142 0.72

Chute 0.1 1.10 0.65 2.90 2.18 0.0187 0.78
Chute 0.2 1.09 0.63 3.01 2.18 0.0211 0.81
Chute 0.4 1.07 0.60 3.15 2.17 0.0247 0.84
Chute 0.7 1.04 0.58 3.27 2.17 0.0278 0.87
Chute 0.9 1.02 0.56 3.36 2.16 0.0306 0.89
Chute 1.1 1.00 0.55 3.44 2.15 0.0332 0.91
Chute 2.2 0.89 0.50 3.75 2.11 0.0441 0.98
Chute 3.3 0.78 0.48 3.97 2.05 0.0530 1.02
Chute 4.5 0.67 0.46 4.13 1.99 0.0605 1.05
Chute 5.6 0.56 0.44 4.25 1.92 0.0668 1.08
Chute 6.7 0.44 0.43 4.35 1.84 0.0722 1.10
Chute 7.8 0.33 0.43 4.43 1.76 0.0767 1.11
Chute 8.9 0.22 0.42 4.50 1.67 0.0805 1.12
* JUMP * 10.0 0.11 0.42 4.55 1.58 0.0837 1.13
Chute 10.0 0.11 1.14 1.66 1.39 0.0029 0.41
Chute 11.1 0.00 1.27 1.49 1.38 0.0020 0.35
Apron 11.3 0.00 1.27 1.49 1.38 0.0020 0.35
Apron 11.5 0.00 1.27 1.49 1.38 0.0020 0.35
Apron 11.7 0.00 1.27 1.49 1.38 0.0020 0.35
Apron 11.9 0.00 1.27 1.49 1.38 0.0020 0.35
Apron 12.1 0.00 1.27 1.49 1.38 0.0020 0.35
Table 5: Data and analysis summary of an average Oulette Creek riffle using the program CHUTE.
45
4.3 FishXing: The FishXing program usefully combines open channel and culvert hydraulics with
swimming performance data for selected fish species. Performance data may also be added by
the user.
The input geometry is shown schematically in Figure 43. The program tests the success of
passage for the selected species, culvert dimensions and alternate tailwater options for a range
of discharges. The solution for a typical backflooded highway culvert near Oulette Creek BC
(Figure 44) is shown in the following sections.
Figure 43: Schematic of FishXing geometry inputs.
Figure 44: backflooded highway box culverts near Oulette Creek BC.
46
The inputs for the culverts and performance criteria for Coho salmon are shown in Figure 45.
Figure 45: Langdale-Oulette FishXing inputs.
The program output profile without backflooding is shown in Figure 46. Fish passage is blocked at
a low flow of 0.5 m3/s and at the maximum discharge of 17 m 3/s.
Figure 46: Langdale-Oulette fish barrier (perched culvert).
47
The amount of backflooding required for fish passage

Qc Dc Hc
may be tested by calculating the rating curve for a
0.1 0.02 0.03
downstream riffle and adjusting its crest elevation
0.2 0.04 0.06
relative to the culvert elevation until the passage is
0.3 0.05 0.08
successful at a selected discharge. The rating curve
0.4 0.06 0.09
may be estimated using the critical velocity relationship
0.5 0.07 0.11
given in discussed in Review Section 2:
1 0.11 0.17
Hc = 1.5 ( Q2 / gW 2 )1/3
2 0.17 0.26
assuming the pool velocities are so low that the
5 0.32 0.48
approaching velocity head correction and flow
10 0.51 0.77
contraction losses may be ignored (Table 6). If not, the
15 0.66 0.99
detailed water surface profile must be calculated from
17 0.72 1.08
the downstream riffle crest.
20 0.80 1.20
Table 6: 9m wide Oulette riffle rating curve
With a crest elevation of 13.8m, fish passage is possible at low flow and at the median flood
(Figure 47).
Figure 47: Langdale-Oulette fish passage with backflooding Oulette riffle.
48
References
ASCE. 1967. River Hydraulics (USCE Technical and Engineering Design Guide 18).
Washington.
Barnes, H.H. 1967. Roughness Characteristics of Natural Channels. USGS Water Supply Paper
1849. Washington DC.
Chang, H.H., 1988. Fluvial processes in river engineering. Wiley.
Chanson, H. 1994. Hydraulic design of stepped cascades, channels, weirs and spillways.
Pergamon.
Chow, Ven Te. 1959. Open-channel Hydraulics. McGraw-Hill.
CSIRO, Australian Commonwealth Scientific and Industrial Research Organization. 2003.

Guidelines for the Design of Rock Chutes.
Harrelson, C.C., C.L. Rawlins and J.P. Potyondy. 1994. Stream channel reference sites: an
illustrated guide to field techniques. USDA Technical Report RM-245.
Henderson, F.M. 1966. Open Channel Flow. MacMillan.
Hicks, D.M. and P.D. Mason. 1991. Roughness characteristics of NZ rivers. WRS. Wellington NZ.
Kellerhals, R. and M. Church. 1989. The morphology of large rivers: characterization and
management. in D.P. Dodge (ed.) Proc. of the international large river symposium. Canadian
Fisheries and Aquatic Sciences Special Pub. 106), Ottawa.
Kondolf, G.M. and H. Piegay 2003. Tools in Fluvial Geomorphology. Wiley.
Lane, E. 1955. Design of stable channels. ASCE Trans. 120:1234-1279.
Leopold, L.B., M.G. Wolman, and J.R. Miller. 1964. Fluvial Process in Geomorphology. Freeman
CA.
Sauchyn, D.J. and D.S. Lemmen 1996. Impacts of landsliding in the Cypress Hills, Saskatchewan
and Alberta; in Current Research 1996-B:7-14. Geological Survey of Canada, Ottawa ON.
Schumm, S.A. 1977. The Fluvial System. Wiley, NY.
USCE. 1982. HEC-2 Water surface profiles, users manual. US Corps of Engineers, DC.
Vogel, S., 1981 (1994 2nd Ed.). Life in Moving Fluids. Willard Grant, Boston.
Yalin, M.S. and A.M.F. da Silva. 2001. Fluvial Processes. IAHR Monograph. Delft.
Programs:
Chute: www.toolkit.net.au/cgi-bin/WebObjects/toolkit
FishXing: www.stream.fs.fed.us/fishxing
HEC-RAS: www.hec.usace.army.mil/software
Sources: text extracted from CRI Stream Restoration Design Notes (Newbury, R. 2007)
www.unb.ca/cri/Courses.html
49
APPENDIX A
50
51
Pool and Riffle Calculation Sheet

STREAM
RIFFLE HEIGHT
Q design discharge m3/s
D depth of flow approaching riffle m
W average width of flow m
V approaching velocity m/s Q/WD
v2/2g velocity head (kinetic energy of

approaching flow) m
H specific energy m D + V2/2g
Dc critical depth of flow m Dc = (Q2 / gW 2)1/3

2
vc /2g critical velocity head m Vc = (g Dc)1/2
2
Hc critical specific energy m Hc = Dc + vc /2g
RH riffle height above channel bed m
DIMENSIONS
SB channel slope
SRU slope of upstream riffle face
SRD slope of downstream riffle face
RU distance of heel to crest m RU = RH / (SRU + SB)
RD distance of crest to toe m RD = RH / (SRD – SB)
YD height of bed at the crest above toe m YD = RD (SB)
total drop in chute m YD + RH
SPACING
L pool length with no back-flooding m L = RH / SB
BF height of back-flooding on upstream riffle m
Istep interval between stepped crests m Istep = L – (BF / SB) + RD
Irun interval between run & pool crests m Irun = L + RD + run

g = 9.81 m/s2
52
53

Robert Newbury Err Design Notes-2011

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Robert Newbury Err Design Notes-2011

Încărcat de

Drepturi de autor:

Formate disponibile

1

Designing Pool and Riffle Streams

Lectures and Exercises

River Restoration NW Short Course 2008

Oulette Creek BC riffle at design flow

Robert Newbury PhD PEng

Appendix A contains a summary of the geometry calculations required to design riffle-pool

Introduction: Stream Restoration 2

1. Review of River Forms and Forces 4

1.1 meanders, pools and riffles

2. Review of Open Channel Hydraulics 11

2.1 uniform flow

3. Project Design Example: Stepped Pool and Riffle Channel 29

3.1 background: Oulette Creek project

4. Programmed Solutions: HEC-RAS, CHUTE, Fish Xing 41

Appendix A: Riffle and Pool Geometry Calculation and Construction 49

1: Review of River Forms and Forces

Figure 1: Sine waves generated

riffle pool riffle

meander length 10 to 14 X channel width

Figure 4: A flow net of path lines and

Figure 6: Some wader stability factors.

2: Review of Open Channel Hydraulics

2.2 Chezy and Manning Equations: In 1769, Antoine Chezy was

Channel or Median Size of Base Value of

concrete .012 - .018

Table 1: Common base values of Manning's n (Chow 1959)

0 1.72 10.3 1.10 1.78 1.09 .0013 0

22.2m 36.4m 31.7m 26.2m 27.2m

E = Z + D + V2/2g (Bernoulli's equation)

Dc = ( Q2 / gW 2 )1/3 and Hc = 1.5 Dc = 1.5 ( Q2 / gW 2 )1/3

3. Salmon Project Design Example

3.1 Background: The

V = (0.7)2/3 (0.03)1/2 / 0.04 = 3.4 m/s

3.4 Riffle Design

Dc9 = (Q2 / gW 2)1/3 = ((172) / 9.81(9)2)1/3 = 0.72 m

Dc6 = (Q2 / gW 2)1/3 = ((172) / 9.81(6)2)1/3 = 0.94 m

RIFFLE HEIGHT Oulette Creek

D depth of flow approaching riffle m 1.8

W average width of flow m 9

V approaching velocity m/s Q/WD 1.05

v2/2g velocity head (kinetic energy of 0.06

Dc critical depth of flow m Dc = (Q2 / gW 2)1/3 0.72

RH riffle height above channel bed m 0.78

SB channel slope .03

SRU slope of upstream riffle face 0.5 (2:1)

SRD slope of downstream riffle face 0.1 (10:1)

RU distance of heel to crest m RU = RH / (SRU + SB) 1.47

RD distance of crest to toe m RD = RH / (SRD – SB) 11.14

YD height of bed at the crest above toe m YD = RD (SB) 0.33

total drop in chute m YD + RH 1.11

L pool length with no back-flooding m L = RH / SB 26

BF height of back-flooding on upstream riffle 0.20

1. cross-section locations are entered from upstream to downstream

3. the top of bank must be selected for each cross-section

4. the reach length to the next cross-section downstream must be specified

6. modeled discharges must be set at the upper end of a reach

Effective drop (m) 1.11 Unit discharge (cumec/m) 1.89

OVERALL PROPERTIES. NOTE - energy heads relative to downstream bed

D50 based on conditions at jump, as follows:-

WATER SURFACE PROFILE

Chute 0.0 1.11 0.71 2.67 2.18 0.0142 0.72

Figure 43: Schematic of FishXing geometry inputs.