Chapter 4

Practical issues roughness,

conveyance and afflux

ABSTRACT
This chapter illustrates the use of the CES-AES methods through a series of practical
examples involving stage-discharge, velocity, backwater and afflux prediction for a
range of channel types (size, shape, cover). Generic guidance on the use of the CES-AES
for different flood risk management activities is provided, with specific examples relating to availability of site information, vegetation maintenance and dredging. A series
of what-if scenarios is explored to demonstrate the sensitivity of the CES-AES outputs to changes in input data (roughness, slope, sinuosity). An approach for estimating
roughness in mountain streams with boulders is presented.
The survey data for the examples are provided in Appendix 3.

4.1 AN OVERVIEW OF THE CES-AES USE IN PRACTICE

The CES-AES software supports practitioners concerned with a range of flood risk
management as well as other wider activities through, for example:

Calculating water levels, flows and velocities for rivers, watercourses and drains;
Providing upper and lower uncertainty scenarios;
Assessing flood or extreme water levels, and the sensitivity of these to channel
adaptation or management options (particularly dredging and plant management);
Assessing the impact of timing and nature of vegetation cutting;
Assessing the impact of blockage due to vegetation or debris;
Understanding the influence of in-stream structures on water levels;
Finding holistic solutions which address both environmental and flood risk
management or land drainage objectives;
Implementing guidance and procedures for channel maintenance.

The use of the CES for channel maintenance and flood risk management activities
is recognised in Sir Michael Pitts review following the 2007 widespread flooding (Pitt,
2008):
7.66 . . . To progress its understanding of how seasonal variation in vegetation
affects the way in which watercourses behave, the Environment Agency has

184 Practical Channel Hydraulics: Roughness, Conveyance and Afux

recently developed a tool called the Conveyance Estimation System (CES), which
will help to deliver an improved maintenance programme. . . .
The Association of Drainage Authorities and Natural Englands Biodiversity
Manual Integrating Wildlife and Flood Risk Management (Buisson et al., 2008)
recognises the potential use of the CES in identifying preferred channel management
techniques:
In choosing the best technique to apply it is essential to examine the effect on
flood conveyance in the specic location and circumstances that the technique
would be used. This will require judgment informed by experience and one of the
available flood risk modelling tools. In many circumstances, the Conveyance Estimation System (www.river-conveyance.net) may provide the information needed.
Modelling allows prediction of the effects of management techniques on conveyance and storage and can identify the additional capacity needed to offset any
reduction in conveyance caused by additional wildlife habitat created, such as a
wider uncut marginal strip of vegetation in the channel.
Table 4.1 provides a summary of some of the core flood risk management activities, the type of hydrodynamic modelling associated with these and advice on the
interpretation of outputs for specic activities. Outputs from the CES-AES may be
used in modelling types (a) through to (f) in the table, which support a range of activities including strategic planning, scheme design, flood forecasting, flood mapping for
regulation and insurance, hydrometry and maintenance.
For all of these activities, the basic building blocks within the CES-AES software
include:

a single cross-section analysis;

a single structure analysis i.e. bridges and culverts;
a backwater profile analysis with no structures present;
a backwater profile analysis with structures present e.g. bridges, culverts; and
the use of the CES-AES within a 1-D modelling environment (see Chapter 5).

4.1.1 Single cross-section analysis

A single cross-section analysis typically involves one or more of the following steps:
1

Data gathering Gather all available data and information about the site, including where possible, survey; water level, discharge and velocity measurements;
water surface slope; photographs; channel and oodplain cover; and any historical
flood data.
Initial stage-discharge analysis In the absence of flow data, apply the CES
methodology using default parameters and the Roughness Advisor unit roughness information. Essential information is the surveyed cross-section geometry
and the longitudinal water surface slope (ideally) or bed slope. Photographs, site
descriptions and UK grid references may be used to assist in selecting the unit
roughness values.

Table 4.1 Flood risk management tasks (adapted from EA/Defra, 2002a and Defra/EA, 2004).
Activity
Type of
modelling

a) 1-D
modelling

b) Routing
models

c) Backwater
models

d) Rating
curves

e) Rating
surfaces

Strategic
planning

Scheme
design

Flood forecasting &

warning

Flood mapping for

regulation/
insurance

Hydrometry

Maintenance

f) Single levels/
flows

Comments/
Use
Evaluate and compare options against criteria
such as effectiveness, efficiency, robustness
and performance. Where small changes in
input parameters effect large changes in water
level, devote additional resources to understanding and potentially reducing the cause.
New schemes ideally based on a calibrated 1-D
model. Alternatively, undertake a cost-benefit
analysis for multiple cases with different
site roughness values.
Examine the sensitivity of flood wave speed to
river resistance using a flood routing model
such as that which is available in ISIS/InfoWorks
RS(CES embedded). Additionally, undertake an
offline analysis to examine where the flood
peaks change significantly with channel roughness.
For broad-scale mapping, inform Agency
Development Control that approach is indicative
not accurate. Examples of CES use in national
flood mapping include the Second Generation
Flood Maps for Scotland (Mc Gahey et al., 2005;
SEPA, 2006) and Northern Ireland (NIRA, 2005).
Typically users have high quality calibration data.
This may be used for model calibration &
extension of ratings.
Detailed local site knowledge (including seasonal
vegetation variations). Use CES to predict flow
and local velocities.

186 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Calibration using stage-discharge measurements If flow data are available,

calibrate the stage-discharge curve by varying the nl values, as the flow rate is
particularly sensitive to this parameter. The top-of-bank markers may also be
varied within a realistic range (e.g. 0.25 m for rivers) to capture the bankfull
variations.
Calibration using stage-discharge and velocity measurement If velocity data are
available, refine the lateral distribution of roughness, in particular, the in channel,
bank and berm nl values. Here, it is necessary to ensure the changes in nl for a given
depth apply to all depths i.e. the calibrated stage-discharge curve is preserved. If
historic flood data are available, this is a useful means of calibrating the top-end
of the rating curve, and establishing the floodplain nl values.
Expert calibration using internal model parameters Expert users may wish to
vary the values of the main channel dimensionless eddy viscosity mc (via the CESAES Graphical User Interface, Chapter 5) and the secondary flow term  (via the
source code, Chapter 5) based on local site knowledge, for example, reducing mc
if bankside vegetation reduces the mixing at the main channel oodplain interface.
Note: this is advocated for research or expert users only as inexperienced application of this multi-parameter calibration approach may lead to wider uncertainty
bounds than alternative calculation methods.
Meandering channels For meandering channels, Steps 1 to 4 are identical, except
there are two additional inputs, the plan form sinuosity, , and the bend orientation i.e. left or right hand bend. The value should only be varied if there is
uncertainty associated with the true value, to gauge the sensitivity for the site. The
bend orientation is a fixed input which is assigned to the top-of-bank markers.

Figure 4.1 provides an example of a simple CES application to a two-stage crosssection located in a straight reach of the River Penk at Penkridge. The effect of nl , and
 on the depth-averaged velocity profile is shown and a small sinuosity is introduced
to illustrate what the affect of Cuv would be. The calibrated nl values are 0.033 and
0.060 for the main channel and floodplains respectively. The default and  models
are used and an example sinuosity of 1.05 (Cuv 1%) is introduced.

4.1.2 Single structure analysis (bridges and culverts)

A single structure analysis typically involves one or more of the following steps:
1

Data gathering Information is required to characterise both the structure and

the channel surrounding it. Channel cross section information is required as in
Section 4.1.1. If the bridge structure has abutments or other modications of the
surrounding channel then these need to be included in the cross section survey.
Ideally, the structure should be surveyed (both upstream and downstream faces)
to provide number of openings and elevation data for springer levels, soft levels
and road or parapet levels. The material and condition of the structure should be
observed, if possible, bearing in mind that the material and condition of the structure at the opening might not always be representative of conditions throughout
its downstream length. For culverts, the entrance structure type and shape should
be recorded. Consideration should be given to the condition of the culvert barrel

Practical issues roughness, conveyance and afux 187

Site:
Measurements:

The River Penk, Penkridge, UK

Velocity / area measurement
from a 56m cableway
0.0015
In-channel weed growth and
some bed movement. Berm and
bankside grass.
Low to moderate relief, Mercia
Mudstone, Boulder Clay and
Sherwood Sandstone, farming &
forestry
40m wide, mc 15m, hbf 1.78m

Bed Slope:
Vegetation:

Catchment:

Channel:

Pre -1990
measurement
location

Input the channel geometry (below), the slope and

initial nl values based on the RA advice.
Calibrate stage-discharge curve based on
roughness and available flow data Fig 1 (Note the
data is pre-1990 i.e. before weir was installed).
Calibrated n values are 0.033 (mc) and 0.060 (fp).
Fig 2-5 show the Ud prediction for depth 1.94m,
stripping out and reintroducing parameters.

Depth (m)

1.9

Bed elevation (m)

Fig 1

2.1

1.7

1.5

1.3

2.5
2.0
1.5
1.0
0.5
0.0

1.1
10

10

15

20

25

30

35

15

20

25

30

35

Discharge (m3s-1)

40

Lateral distance across section (m)

2

1.6

No

1.4

Add

Fig 3

1.8

Depth-averaged velocity (m/s)

Depth-averaged velocity (m/s)

Data

Fig 2

1.8

or

1.2
1
0.8
0.6
0.4
0.2

Data

1.6

Add

1.4

Add

1
0.8
0.6
0.4
0.2

10

15

20

25

30

35

40

10

15

20

25

30

35

40

Lateral distance across channel (m)

Lateral distance across channel (m)

has more effect in main channel, as floodplain has

low depths and large roughness, i.e. less circulations.

introduces lateral dispersion in areas with steep

velocity gradients. Floodplain slows main channel flow.

Fig 4

1.6

Add nl

1.4

Add

Fig 5

1.8

Data

Depth-averaged velocity (m/s)

1.8

Depth-averaged velocity (m/s)

&

1.2

&

1.2
1
0.8
0.6
0.4

Data

1.6

Add nl

1.4

Add Cuv

1.2
1
0.8
0.6
0.4
0.2

0.2
0
0

0
0

10

15

20

25

30

Lateral distance across channel (m)

Vary lateral nl distribution on floodplain, to improve

velocity prediction. Preserve overall flow rate.

Figure 4.1

35

40

10

15

20

25

30

35

40

Lateral distance across channel (m)

Example of introducing small sinuosity of 1.05, as DS

of the bridge, the River Penk appears to bend left.

Example CES applications for the River Penk at Penkridge (Mc Gahey, 2006) (See colour
plate section).

throughout its length, although this may not be known without a full inspection
or CCTV survey.
Initial structure analysis The AES includes certain typical bridge and culvert
types (e.g. arch and beam bridge, pipe and box culvert). However some structures

188 Practical Channel Hydraulics: Roughness, Conveyance and Afux

may not fit exactly into the defined types. Where this is the case, the closest match
should be used. The user defined arch bridge type allows greatest flexibility for
opening shapes. For culverts with multiple barrels, the AES standard types allow
identically sized openings to be specified. It may be possible to model a culvert
with complex openings as a bridge unit if the barrel friction is thought likely to
be the hydraulic control in most circumstances.
Data entry Care is required in adding bridges or culverts to a CES model. There
are a number of geometric and procedural rules enforced by CES-AES. These
include:
a

b
c
d

Openings for bridges and culverts must have valid coordinates within the
channel cross section. This means, for example, that culvert openings should
not extend below ground level and springer levels for arches must be above
ground elevation at a given cross section offset (i.e. arches should not spring
directly from ground level at the side of a cross section).
Road elevations must be contained within the cross section.
The CES river cross sections immediately upstream and downstream of a
bridge or culvert should have three roughness zones.
In the CES-AES backwater module, the upstream openings of a bridge or culvert are placed at the chainage implied by the reach length downstream of the
previous section.
In the CES-AES backwater module, the reach length downstream from the
bridge section to the next CES river section is taken to include the bridge
length.

Cross section analysis The CES-AES software requires a backwater prole to be

calculated in order to generate information on conveyance and flow ratings at the
bridge or culvert opening sections. The Outputs tab of the software requires the
user to check a box Include Structure Outputs from last Afflux Calculation to
include data from the structure cross sections.
Calibration Bridges and culverts can have a single roughness value (expressed
in terms of Mannings n) for the structure, which is combined with the channel
and overbank roughness values at the structure sections when computing the conveyance. Should there be measurements available to assist with calibration then
they may be used as for a normal CES river section. Note, as above, that a backwater prole must be computed to generate conveyance or water level results for
bridges or culverts.

4.1.3 Backwater prof ile analysis (no structures

present)
Backwater profiles are evaluated where there is a downstream control (i.e. known
head and flow). A backwater analysis typically involves the following steps:
1

Data gathering Information (similar to the single cross-section analysis,

Section 4.1.1) for multiple cross-sections is required. An important question is
how far upstream of the control the cross-sections are required. A useful rule

Practical issues roughness, conveyance and afux 189

of thumb (Samuels, 1989b) based on the downstream depth and reach-averaged

longitudinal slope may be used to estimate this. Additional inputs include the distance between each cross-section, the downstream water level and the inflow(s)
to the reach. Additional measurements or historic data may include, for example,
water levels along the reach for a given inflow.
Initial single section analysis Steps 1 to 4 of the single section analysis
(Section 4.1.1) are followed for any cross-sections where (i) measurements are
available and (ii) the water level measurements are not dominated by the downstream control. The adopted/calibrated roughness values may then be assigned
to the remainder of the reach cross-sections in accordance with the available
descriptions.
Backwater analysis The downstream control and inflow to the reach are assigned
and the backwater calculation is implemented, providing the water surface profile
and average velocities along the reach.
Backwater calibration The backwater predictions may then be compared to
water level and velocity measurements along the reach and these may guide further
roughness calibration.

4.1.4 Backwater prof ile analysis (structures present)

The process for backwater profile analysis when a bridge or culvert is included is
the same as above for a reach without structures. As described in Chapter 3, the AES
models the water surface profile through a structure using cross sections supplied by the
user at the structure entrance and exit, these are labelled AE2 and AE3 in the output
files exported by CES-AES following a backwater analysis. Interpolated upstream
and downstream sections are also used where the transition lengths surrounding the
structure require it. These sections AE1 and AE4 are also reported in the output files.

4.2 ESTIMATING AND USING STAGE-DISCHARGE

RELATIONSHIPS AND SPATIAL VELOCITIES

4.2.1 Stage-discharge prediction for the River Trent

at Yoxall, Staffordshire, UK
Aim
To predict the stage-discharge and upper and lower uncertainty scenarios for the River
Yoxall site using the CES-AES software and available site information. The outputs
are compared with measured data to instil confidence in the results.
The Yoxall site on the River Trent
The Yoxall site (Figure 4.2) is located on a gentle bend of the River Trent, in Yoxall,
Staffordshire, immediately downstream of Yoxall road-bridge. The cross-section is
80 m wide, including a 35 m berm, and a cableway for measuring inbank flows is
situated across the main channel. The site is bypassed through two flood relief culverts,
the flows of which are measured under Yoxall Bridge. There are four overbank gauged

190 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.2 Electromagnetic gauging station on the River Trent (a) downstream of Yoxall bridge and
(b) an upstream view showing Yoxall Bridge (Courtesy of the EA).

flows, of which two were recorded before and two after the construction of a training
bank. The inbank flow measurements are generally considered more reliable than
overbank flows (EA, 2005). Measurements from the Environment Agencys gauge
station (Station 28012) include water level, discharge and velocity. The reach-averaged
longitudinal slope at bankfull is 0.001, increasing to 0.00275 at a depth of 0.5 m. The
channel bed material includes gravels and summer weed growth and the floodplains
consist of grasses with occasional bushes.
Roughness and cross-section def inition
Based on the description and photographs, the Roughness Advisor values include
(Figure 4.3):

Channel bed The RA coarse gravel is adopted (nsur = nl = 0.027). This may
be refined to gravel or fine gravel through calibration. The larger gravel option is
incorporated for the initial guess i.e. a conservative approach.
Channel banks The RA height-varying grass (nveg = 0.041) and sand
(nsur = 0.02) is adopted giving the total unit roughness nl = 0.0456.
Floodplain The RA turf (nveg = 0.021) and sand (nsur = 0.02) is adopted giving
nl = 0.029. The reasoning is that bank-side vegetation appears thicker and taller
(i.e. height-varying grass) than the floodplain grass (i.e. turf).

Results for the Yoxall site

Figure 4.4 provides the CES stage-discharge prediction using these RA roughness values and the measured data. The data are fairly scattered but tend to fall within the
upper and lower credible scenarios, defined by the upper and lower roughness values.
The data move away from the curve at high overbank flows which may be attributed
to the reduced accuracy in flow gauging during times of flood together with uncertainty regarding the oodplain vegetation. A further factor may be the variation in
longitudinal water surface slope with depth, which is subject to local topographic

Practical issues roughness, conveyance and afux 191

Figure 4.3 River Trent cross-section, roughness zones and top-of-bank marker locations.

2.5

Depth (m)

1.5
Measu red Data
CES Prediction

CES Upper
0.5
CES Lower
0
0

20

40

60

80

10 0

12 0

14 0

Flow (m3s-1)
Figure 4.4 Stage-discharge prediction for the River Trent at Yoxall.

variations at low flows, 0.001 at bankfull (as used in the CES simulation) and more
closely linked to the overall valley slope at high flows. This issue may be overcome if
the backwater module is used for the reach, as the variations in surface water slope
are implicit. Similarly, where the CES is incorporated in a full 1-D model the depth
specic surface water slope is incorporated. The stand-alone cross-section analysis is
useful in determining and validating the unit roughness values to be adopted for the
remainder of the reach.

192 Practical Channel Hydraulics: Roughness, Conveyance and Afux

4.2.2 Stage-discharge and velocity prediction

for the River Colorado, La Pampas, Argentina
Aim
To predict a stage-discharge curve and velocity distributions for different flow depths
for the River Colorado site using the CES-AES software and available site information.
The outputs are compared to measured (flow and velocity) data throughout the depth
range to instil confidence in the results.
The River Colorado site
The River Colorado provides a useful example of a wide, rigid bed, simple channel with large flows. It is 885 km long, rising in the Andes and owing south-east
across southern central Argentina to the Atlantic Ocean. It marks the northern limit
of Patagonia. It is also a rough boundary between the commercial agriculture to the
north and ranching to the south. The main channel is 60 m wide and 3.6 m deep,
with a reach-averaged longitudinal bed slope of 0.0013. Typical flow rates are of
the order 400600 m3 s1 and the river often overows its banks in Spring. Water
level, discharge and depth-averaged velocity measurements are available; however,
information on channel roughness is limited to the photographic evidence (Figure 4.5).
Roughness and cross-section def inition
Based on the photographs, the Roughness Advisor values include (Figure 4.6):

Channel bed The RA bedrock (nsur = 0.025) and 020% boulder coverage
(nirr = 0.017) are adopted, giving nl = 0.030. Bedrock is assumed due to the
apparent hard cut-out of the channel side-banks (Figure 4.5) and some boulders
are included due to the presence of boulders near the right bank (Figure 4.5c).
Channel banks The RA bedrock (nsur = 0.025) and height-varying grass
(nveg = 0.041) are adopted giving nl = 0.048. Although the vegetation is not
visible on all banks, the added inuence of the height varying grass on the banks
is likely to be minimal in terms of the overall channel width and hence capacity.
Floodplain The RA bedrock (nsur = 0.025) and turf (nveg = 0.021) are adopted
giving nl = 0.033.

Results for the Colorado site

Figure 4.7 shows the CES stage-discharge predictions with these roughness values.
The flow is described reasonably well throughout the range of depths other than near
bankfull (3.6 m), where there is a sharp step or kick in the curve. This is caused
by the embedded secondary flow model, which takes the change in orientation and
magnitude of the secondary flow circulations from inbank to out-of-bank flow into
account, identified in the model by the bank markers (Figure 4.6). The step is sharp as
the change in geometry from in-channel vertical banks to floodplain horizontal banks
is sudden. As survey data were only available for the in-channel section, the floodplain
topography is based on the photographic evidence (e.g. Figure 4.5b). The discrepancy

Practical issues roughness, conveyance and afux 193

Figure 4.5 The River Colorado gauge site showing the instrumentation (a) & (b) and looking
(c) downstream and (d) across the full width of the river (Courtesy of Leticia Tarrab).

Figure 4.6 River Colorado cross-section, roughness zones and top-of-bank marker locations.

between the measured and predicted flows suggests that, in reality, the change from
inbank to out-of-bank is perhaps more gradual.
Figure 4.8 provides the depth-averaged velocity predictions for a range of flow
depths and the corresponding measured data. The CES tends to under-predict the high

194 Practical Channel Hydraulics: Roughness, Conveyance and Afux

5
4.5
4

Depth (m)

3.5
3
2.5
Measured Data
2
CES Prediction

1.5

CES Upper

CES Lower

0.5
0
0

10 0

20 0

300

400

50 0

60 0

70 0

Flow (m3s-1)
Figure 4.7 Stage-discharge prediction for the River Colorado, Argentina.
3

CES 3.20 m
CES 2.90 m
CES 2.49 m
CES 2.28 m
CES 2.04 m
CES 1.90 m
Data 3.20 m
Data 2.90 m
Data 2.49 m
Data 2.28 m
Data 2.04 m
Data 1.90 m
Scaled bed

Depth-averaged velocity (ms-1)

2.5

1.5

0.5

0
0

10

20

30

40

50

60

70

Lateral distance across section (m)

Figure 4.8 Depth-averaged velocity prediction for the River Colorado at various flow depths and
comparison to measured data.

Practical issues roughness, conveyance and afux 195

velocities (channel centre) at large flow depths and slightly over-predict the velocities
at lower depths. This may be a short-coming in the CES approach for adequately
resolving the variation of roughness with depth for this wide channel, where bed
generated turbulence has a dominant role and the channel banks have less influence.
The distribution may be further influenced by the presence of individual boulders (e.g.
Figure 4.5c).

4.2.3 Hierarchical approach to estimating roughness

(and other f low parameters) for the River Main,
County Antrim, UK
Aim
This example is intended to demonstrate the hierarchical approach to estimating
roughness within the CES i.e. no information (use of River Habitat Survey); some
information (use of Roughness Advisor descriptions); and detailed information (use of
photos and measurements to calibrate roughness). The hierarchical approach is based
on stage-discharge predictions. Once the detailed information has been used, additional flow parameters are simulated with the CES including velocity, back-calculated
Manning n, Froude Number and Reynolds Number.
The River Main site
The River Main in County Antrim was introduced in Chapter 2, Section 2.2.2. The
reach of interest was reconstructed and realigned in the 1980s to form a double
trapezoidal channel. The CES is used here to demonstrate its use in predicting various flow parameters at Bridge End Bridge based on available information. Although
this includes survey (Figure 4.9), bed slope (0.00297), roughness, photographs
(Section 2.2.2) and detailed measurements; this example is set-up to illustrate the
RA use with varying degrees of roughness information i.e.:

No information for the main channel (use of RA RHS advice);

Some information for the main channel (use of RA based on descriptions);
Detailed information for the main channel (use of RA including calibrating
roughness values to best-fit measurements and use of photographs).

Roughness and cross-section definition and results for the River Main
i No information
For this case it is assumed there is no information on in-channel cover other than the
quarry stone on the channel banks (0.5 tonne weight, 100200 mm size) incorporated
during the re-design of the channel. Three roughness zones are set-up in the RA:

Channel bed The UK Grid reference for the River Main is 350100 411000 (NI).
Based on this, the RHS suggests the most likely vegetation is filamentous algae
(nveg = 0.015). This is adopted with sand (nsur = 0.02) as a basic substrate giving
nl = 0.025.

196 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.9 River Main cross-section, roughness zones and top-of-bank marker locations.

Channel banks The RA rip-rap option is adopted for the quarry stone (nsur =
0.045) augmented with the RHS suggested filamentous algae, giving nl = 0.047.
Berms Based on the available oodplain information, the RA height-varying
grass (nveg = 0.041) option is adopted with sand (nsur = 0.02) as the basic
substrate. This gives the total unit roughness nl = 0.046.

The RA roughness zones are combined with the cross-section and slope information to predict the stage discharge (Figure 4.10). These include the upper and lower
credible scenarios which are based on the upper and lower roughness values. The measured data falls within these scenarios (which are wide); however the curve based on
the expected roughness falls below the data.
ii Some information
For this case there is additional information on the roughness cover including: no
vegetation present at the channel bed which consists of coarse gravel and thick grass
protruding through the quarry stone on the banks. As before, three roughness zones
are set-up in the RA composed of:

Channel bed The RA coarse gravel is adopted (nsur = nl = 0.027).

Channel banks The RA rip-rap option is adopted for the quarry stone (nsur =
0.045) together with the RA turf (nveg = 0.021) to simulate the thick grass. This
gives the total unit roughness nl = 0.050.
Berms The values are unaltered i.e. a total unit roughness nl = 0.046.

A revised flow prediction is made based on the updated roughness values

(Figure 4.11). The curve based on the expected roughness value still falls below the
measured data albeit closer. A notable difference is the reduced range of uncertainty
i.e. the credible upper and lower scenarios are much narrower. This is largely due
to eliminating the lamentous algae and heavy weed growth, which have considerable associated uncertainties. Some data now fall outside of these scenarios, which is

Practical issues roughness, conveyance and afux 197

1.8
1.6
1.4

Depth (m)

1.2
1
0.8
RA No Information
0.6

Lower No Information
Upper No Information

0.4

Measured Data

0.2
0
0

10

20

30

40

50

60

Flow (m3s-1)
Figure 4.10 Stage-discharge prediction for the River Main with no in-channel information.

1.8
1.6
1.4

Depth (m)

1.2
1
0.8

RA Some Information

0.6

Lower Some Information

0.4

Upper Some Information

Measured Data

0.2
0
0

10

20

30

40

50

3 -1

Flow (m s )
Figure 4.11 Stage-discharge prediction for the River Main with some roughness information.

60

198 Practical Channel Hydraulics: Roughness, Conveyance and Afux

plausible as these are not condence intervals or envelopes (Chapter 3, Section 3.3)
they simply indicate the area where the data is likely to fall.
iii Detailed information
For this case there is detailed information available including photographs and measurements (flows and velocities), which can be used to improve the CES simulations.
From Figure 4.11 it is apparent that the RA is under-estimating the roughness as the
measured data are above the curve, and in some instances, above the upper credible
scenario. The photographic evidence (Figure 2.5) suggests that the coarse gravel is
very large with cobbles present. The RA unit roughness value for coarse gravels and
cobbles are 0.027 and 0.035 respectively, a substantial difference. Here, steered by the
measured data and photographic evidence, a value of 0.032 is adopted. Figure 4.12
provides the resulting stage-discharge prediction which follows the measured data
well including the variations at bankfull (0.95 m). The no information curves
are added to illustrate the reduction in uncertainty and improved stage-discharge
predictions where more detailed information is available.
These calibrated roughness values are used to predict the depth-averaged velocities for a range of flow depths (Figure 4.13). The predictions follow the data prole
reasonably well for greater depths; however, at lower depths the main channel velocity
is under-predicted. There appears to be a small degree of skew in the velocity data at
low depths, which may arise from the effects of the upstream bend circulations being
transported downstream. Since the channel bed material is gravel, with no vegetation
present, these higher velocities are plausible but are not captured by the CES where

1.8
1.6
1.4

Depth (m)

1.2
1

Measured Data
RA Detailed Information

0.8

Upper Detailed Information

0.6

Lower Detailed Information

RA No Information

0.4

Lower No Information
0.2

Upper No Information

0
0

10

20

30

40

50

60

Flow (m3s-1)
Figure 4.12 Stage-discharge for the River Main with detailed roughness information (and the no
information case).

Practical issues roughness, conveyance and afux 199

2.5

CES 0.790 m

Depth-averaged velocity (ms-1)

CES 1.200 m
CES 1.365 m

CES 1.785 m
Data 0.790 m
Data 1.200 m

1.5

Data 1.365 m
Data 1.785 m
Scaled bed

0.5

0
0

10

15

20

25

30

35

40

Lateral distance across channel (m)

Figure 4.13 Depth-averaged velocity predictions for the River Main at various flow depths and
comparison to measured data.

the effect of the slightly higher calibrated bank side roughness (nl = 0.050) is present
even at low flows.
Exploration of additional CES f low parameters
The CES enables users to back-calculate an equivalent Manning n resistance parameter.
In modelling practice, a Manning n value is assigned to the channel bed, banks and
berms, which is constant for all flow depths. In the field, measurements show how
Mannings n varies with flow depth (e.g. Chow, 1959), with a greater resistance at
low depths where roughness size is comparable with flow depth. Figure 4.14 provides
the CES back-calculated and measured Manning n values with depth. These show a
reasonable correlation throughout the depth albeit with the predictions 15% lower
than the measured data.
Figure 4.15 shows the CES predicted Froude Number with depth. The Froude
Number is well below 1.0 throughout the depth, indicating subcritical flow conditions, as expected for natural channels. In some instances, supercritical flow may
occur, for example, in steep mountain streams, rapidly varied flow downstream of
structures or localised supercritical flow in parts of the cross-section. Engineers have
also been known to design supercritical channels through cities i.e. steep heavily modied concrete conduits designed to convey floods over shorter distances than the natural
plan form channel prole. For the River Main, the Froude Number increases away
from the channel bed, reecting the higher velocities at greater depths. The shape of

200 Practical Channel Hydraulics: Roughness, Conveyance and Afux

1.8
1.6
1.4

Depth (m)

1.2
1
0.8

Measured Data

0.6

RA_Detailed Information

0.4

Upper Credible Scenario

0.2

Lower Credible Scenario

0
0

0.02

0.04

0.06

0.08

0.1

0.1 2

Manning n
Figure 4.14 CES back-calculated and measured Manning n values for the River Main.

1.8
1.6
1.4

Depth (m)

1.2
1
0.8
RA Detailed Information

0.6

Upper Credible Scenario

0.4

Lower Credible Scenario

0.2

Critical Flow

0
0

0.2

0.4

0.6

0.8

1.2

1.4

Froude Number
Figure 4.15 CES predicted Froude Number with depth for the River Main.

1.6

Practical issues roughness, conveyance and afux 201

1.8
RA Detailed Information

1.6

Upper Credible Scenario

1.4

Lower Credible Scenario

Depth (m)

1.2
1
0.8
0.6
0.4
0.2
0
10

100

1000

10000

100000

1000000

10000000

Reynolds Number
Figure 4.16 CES predicted Reynolds Number with depth for the River Main.

the curve changes as the channel flow moves out of bank onto the berms and again
as it comes into contact with the distant berm edges, as these boundaries inuence the
velocity distribution.
Turbulence occurs at high Reynolds Numbers, where the flow incorporates
a three-dimensional eddying or mixing action. Turbulence is dispersive, diffusive,
chaotic, and contrary to viscosity, it is a property of the flow and not of the fluid
(Reynolds, 1974; Nezu & Nakagawa, 1993). Figure 4.16 shows the CES predicted
Reynolds Number with depth which is of the order 103 to 106 for most of the depth
column i.e. characteristic of fully turbulent flow (Re >103 ). Near the channel bed, the
Reynolds Number drops substantially indicating the potential for laminar flow close
to the boundary (although this is unlikely to occur in natural channels with sizable
roughness elements e.g. cobbles).
This single cross-section analysis is a useful means to establish the calibrated rating
curve and hence the roughness values to be used for the whole reach. These roughness
values are adopted in Section 4.3, where different backwater profiles are explored for
the River Main.

4.2.4 Stage-discharge, velocity and roughness

predictions for the River Severn
at Montford Bridge, UK
Aim
To improve condence in high flow predictions required for a Flood Risk Assessment climate change scenario. This is achieved through simulating stage-discharge

202 Practical Channel Hydraulics: Roughness, Conveyance and Afux

and velocity distributions which capture the measured data at lower and intermediate
flows, with the bed slope playing an important role. The back-calculated Manning n
values are also compared to measured values to improve condence. This is achieved
using the CES software and available site information.
The Montford Bridge site on the River Severn
The Montford Bridge site along the River Severn is introduced in Chapter 2,
Section 2.5.2. To full the requirements of a Flood Risk Assessment climate change
scenario, it is necessary to determine the depth at a flow rate well above the available flow measurements. The scenario is to model the 100 year return period,
Q100 = 330 m3 s1 , +20% flow to satisfy Defras Project Appraisal Guidance on
indicative sensitivity ranges for peak river flows in 50100 years time, giving Q =
400 m3 s1 . The available site information includes survey, flow and velocity measurements. The description of the oodplains is limited to grass-cover together with
some photographic evidence.
Roughness and cross-section def inition
Based on the site description, the RA values include (Figure 4.17):

Channel bed The RA sand (nsur = nl = 0.02) is adopted as the bed material.
Channel banks The RA sand (nsur = 0.02) and medium grass 0.751.0 m (nveg =
0.08) is adopted giving nl = 0.082.
Floodplain The RA sand (nsur = 0.02) and turf (nveg = 0.021) is adopted giving
nl = 0.029.

Results for the River Severn at Montford Bridge

For this site, there are measurements available for the water surface slope at each
depth of flow, with an average value of 0.0002 at greater depths (>6 m) and the slope
decreasing down to as low as 0.00006 for depths of 26 m. The initial CES stage-

Figure 4.17 River Severn cross-section, roughness zones and top-of-bank marker locations.

Practical issues roughness, conveyance and afux 203

Depth (m)

6
Measured Data
5
CES Predictionn
4

Upper scenario

Lower scenario

2
0

50

100

150

200

250

300

350

400

Flow (m3s-1)
Figure 4.18 Stage-discharge prediction for the River Severn (slope = 0.000195).

discharge predictions (Figure 4.18) are based on the value 0.000195 giving reasonable
predictions above 5 m, including the bankfull variations, and reducing the slope to
0.00015 (Figure 4.19) marginally improves the simulation magnitude at lower depths.
The shape of the curve in the shallower region is poor, which may be related to the
seasonal vegetation. Low flows typically occur in the summer months when the channel
vegetation is dense, resulting in reduced conveyance and large flow depths.
Figure 4.20 shows the depth-averaged velocity predictions and measured data for
a range of flow depths. The data are captured reasonably well, other than the very high
velocities in the centre region of the channel. This may indicate that the CES model
is over-representing the lateral shear (i.e. mc is too high), resulting in an increased
retarding effect of the slower floodplain flow on the main channel flow. This parameter
may be altered in the CES-AES software (Chapter 5).
Figure 4.21 shows the CES back-calculated Mannings n values together with
the seasonal data measurements. The CES simulates the average through the scattered
data, capturing the slight increase in roughness where the depth drops from 6.5 m (out
of bank) to 5.5 m (inbank). At lower flow depths, the CES under-predicts the roughness. This is not unexpected due to the uncertainty about the in-channel vegetation in
the summer low-flow months.
As the flow, velocity and roughness information has been reasonably simulated,
the stage-discharge curve (slope = 0.000195) can be extended with reasonable condence to determine the flow depths for the climate change scenario of Q100 + 20% =
400 m3 s1 , giving 8.10 m depth with an uncertainty range of 7.98.4 m. Note that
the uncertainty on water level is typically smaller than that associated with flow, due
to the shape of the rating curve.

204 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Depth (m)

Measured Data

CES Predictionn
4
Upper scenario
3
Lower scenario
2
0

50

100

150

200

250

300

350

400

Flow (m3s-1)
Figure 4.19 Stage-discharge prediction for the River Severn (slope = 0.00015).

1.6

CES 4.75 m
CES 6.45 m
CES 6.92 m
CES 7.81 m
Data 4.75 m
Data 6.45 m
Data 6.92 m
Data 7.81 m
Scaled bed

Depth-averaged velocity (m/s)

1.4

1.2

0.8

0.6

0.4

0.2

0
0

20

40

60

80

100

120

140

Lateral distance across section (m)

Figure 4.20 Depth-averaged velocity predictions for the River Severn at various flow depths and
comparison to measured data (with depth 4.75 m based on slope = 0.00015).

Practical issues roughness, conveyance and afux 205

Depth (m)

6
Measured Data
5
CES Prediction
4
Upper scenario
3

Lower scenario

2
0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Manning n
Figure 4.21 CES back-calculated and measured Manning n values for the River Severn.

4.2.5 Investigating the inf luence of roughness, slope

and sinuosity on stage-discharge for the
River La Suela, Cordoba, Argentina
Aim
To explore a series of what-if scenarios for a simple channel and establish the sensitivity of the CES flow and velocity predictions to roughness, slope and sinuosity. This
makes use of the CES-AES software and the available information for the La Suela
river site and the outcomes are therefore specic to this site.
The River La Suela site
The River La Suela is a small river situated north-east of Cordoba, Argentina, with
approximate dimensions of 25 m wide, 2 m deep and a reach-averaged longitudinal
bed slope of 0.001355. Typical flow rates are in the range 5070 m3 s1 . Available
measurements include water level, discharge and depth-averaged velocity proles. Little is known about the roughness other than it consists of gravels, is alluvial in places
and there is sparse bank-side vegetation.
Roughness and cross-section def inition
Based on the description and a calibration using the unit roughness values, the final
assigned roughness values include (Figure 4.22):

Channel bed the RA coarse gravels (nsur = 0.027) and pools (nirr = 0.020)
giving nl = 0.034.

206 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.22 River La Suela cross-section, roughness zones and top-of-bank marker locations.

Channel banks the RA coarse gravels (nsur = 0.027) and height-varying grass
(nveg = 0.041) giving nl = 0.049.

Results for the River La Suela

Figure 4.23 shows the resulting stage-discharge predictions and the measured data
which provide a reasonable fit due to calibration. Figure 4.24 shows the depthaveraged velocity prediction for 1.52 m depth. For both cases the data fall largely
within the credible upper and lower uncertainty scenarios.
Exploring what-if scenarios for the River La Suela
As little is known about this river, the La Suela provides a useful example to explore
different what-if scenarios for the predicted depths, flows and velocities. The aim is
to improve understanding of how the channel behaves through testing the sensitivity
of the CES flow predictions to changes in the input parameters. For example:
1
2
3
4
5
6
7
8
9
10

What happens if the channel bed roughness is doubled i.e. nl = 0.068?

What happens if the channel bed roughness is halved i.e. nl = 0.017?
What happens if the channel bank roughness is doubled i.e. nl = 0.1?
What happens if the channel bank roughness is halved i.e. nl = 0.025?
What happens if the channel is lined with concrete i.e. nl = 0.02?
What happens if the concrete-lined channel is made rectangular?
What happens if the bed slope is doubled i.e. slope = 0.00271?
What happens if the bed slope is halved i.e. slope = 0.000678?
What happens if the channel is meandering with a sinuosity of 1.5?
What happens if the channel is meandering with a sinuosity of 2.5?

Exploring these scenarios provides some insight into the input parameters which
are driving the predictions, for example, the importance of survey versus roughness.
This may help direct the effort when obtaining improved local information. For a more
detailed CES sensitivity analysis of numerous experimental and natural channels, see
Latapie (2003) and Mc Gahey (2006).

Practical issues roughness, conveyance and afux 207

2.5

Depth (m)

1.5

Measured Data
1
CES Prediction
CES Upper

0.5

CES Lower
0
0

10

20

30

40

50

60

70

80

90

10 0

Flow (m3s-1)
Figure 4.23 Stage-discharge prediction for the River La Suela.

CES 1.52 m
Data 1.52 m

Depth-averaged velocity (ms-1)

1.8

Scaled bed
1.6

CES Upper
CES Lower

1.4
1.2
1
0.8
0.6
0.4
0.2
0
0

10

15

20

25

30

35

40

Lateral distance across section (m)

Figure 4.24 Depth-averaged velocity prediction for 1.52 m flow depth showing the uncertainty
scenarios (QCES = 38 m3 s1 ; Qdata = 34 m3 s1 ).

208 Practical Channel Hydraulics: Roughness, Conveyance and Afux

i Results for what-if 1 to 6

What-if scenarios 1 to 6 are designed to improve understanding of the influence of
channel roughness on flow within the channel. This includes the magnitude of the
roughness (e.g. double or half the calibrated roughness value, concrete-lined) and the
distribution of the roughness (e.g. the importance of bank cover versus bed cover).
The outcomes are peculiar to the La Suela River and should only be considered as
indicative for other channels of similar size, shape and roughness.
Figure 4.25 shows the CES predicted stage-discharge, measured data and the
revised stage-discharge predictions based on scenarios 1 to 6. Intuitively, any increase
in roughness results in a decrease in flow and vice versa for all depths. To this end, the
results of scenarios 1 and 2 are both shown in solid lines and the results of scenario
3 and 4 are both indicated with dashed lines. Some observations can be made:

bed roughness has a greater influence on flow than a change in bank roughness;
a change in bed roughness influences the curve throughout the depth range;
a change in bank roughness has little influence at low flow depths and increasing
influence with increasing depth of flow;
concrete-lining (nl = 0.02 for the banks and floodplain) increases the flow capacity
of the channel to a similar magnitude as scenario 2 (nl bed = 0.017) as may be
expected;
the rectangular shape appears to improve conveyance; however this is only an
artefact of the change in area for a given flow depth (here, the change in area is
only zero at 1.52 m);
there is a large amount of scatter in the measured data, particularly at flow depths
of 1.3 to 1.5 m, and none of the predicted curves capture this.

Depth (m)

1.5

Measured Data
1

CES Prediction
Scenario 1
Scenario 2
Scenario 3

0.5

Scenario 4
Scenario 5
Scenario 6

0
0

10

20

30

40

Flow

50

60

70

(m3s-1)

Figure 4.25 Scenario 1 to 6 stage-discharge prediction for the River La Suela.

80

Practical issues roughness, conveyance and afux 209

Figure 4.26 shows the corresponding predicted velocities for scenarios 1 to 4 for
a 1.52 m depth together with the measured data. The observations include:

the change in bank roughness influences the velocity profile close to the channel
banks;
the point at which the altered velocity due to the change in bank roughness returns
to the original velocity profile mid-stream may differ for different degrees of bank
roughness;
the change in bed roughness has a significant impact on the mid-stream velocity
profile which weakens towards the channel banks.

Figure 4.27 shows the predicted velocities for scenarios 5 and 6 for 1.52 m depth
(i.e. zero change in area) together with the measured data and scaled rectangular
geometry. The observations include:

increased velocities due to the concrete lining for both scenarios;

change in velocity profile as a result of the change in channel shape.

The main findings for the River La Suela are that altering bed roughness has a
more significant impact on the channel conveyance than altering the bank roughness;
and that concrete-lining improves the conveyance capacity from that of a natural
channel as expected. These outcomes may well differ for very narrow channels where
the channel banks have a more influential role on the mid-stream velocities. Similarly,
for very wide channels, it is likely the channel banks will have little influence.

CES 1.52 m

Depth-averaged velocity (ms-1)

Data 1.52 m
Scaled bed

2.5

Scenario 1
Scenario 2
Scenario 3

Scenario 4

1.5

0.5

0
0

10

15

20

25

30

35

40

Lateral distance across section (m)

Figure 4.26 Scenario 1 to 4 depth-averaged velocity predictions for 1.52 m flow depth (QCES =
38 m3 s1 ; Qdata = 34 m3 s1 ; Scenarios Q1 = 22 m3 s1 ; Q2 = 69 m3 s1 ; Q3 = 38 m3 s1 ;
Q4 = 41 m3 s1 ).

210 Practical Channel Hydraulics: Roughness, Conveyance and Afux

CES 1.52 m

Depth-averaged velocity (ms-1)

Data 1.52 m
Scaled bed

2.5

Scenario 5
Scaled new bed
2

Scenario 6
Scaled depth

1.5

0.5

0
0

10

15

20

25

30

35

40

Lateral distance across section (m)

Figure 4.27 Scenario 5 and 6 depth-averaged velocity predictions for 1.52 m flow depth (QCES =
38 m3 s1 ; Qdata = 34 m3 s1 ; Scenarios Q5 = 65 m3 s1 ; Q6 = 69 m3 s1 ).

ii Results for what-if 7 and 8

What-if scenarios 7 and 8 are designed to improve our understanding of the influence
of longitudinal bed slope on channel flow capacity. The scenarios involve doubling
(slope = 0.00271) and halving (slope = 0.000678) the bed slope. Figure 4.28 provides
the resulting stage-discharge distributions. From this it is apparent that obtaining the
correct bed slope is imperative to determining the flow rate (30% > % change >
+45%, Table 4.2).
The calculation of conveyance using the CES approach requires the bed slope as
an input as it solves for the unit flow rate (Chapter 3). As conveyance is independent
of slope, the conveyance is calculated by dividing the final flow rate by the slope
1/2
K = Q/So . Figure 4.29 provides the calculated conveyance for the CES prediction
and scenario 7 and 8 as well as the values obtained from the measured data. Here, it is
apparent that the conveyance has little dependence on slope and is therefore a robust
measure of the channel capacity. For further details on slope testing see Mc Gahey
(2006).
iii Results for what-if 9 and 10
What-if scenarios 9 and 10 are designed to improve our understanding of the influence
of sinuosity on in-channel velocity profiles. The River La Suela cross-section is located
in a straight portion of the reach and therefore has a sinuosity of 1. Scenarios 9 and
10 involve increasing the sinuosity in a left-bearing bend to 1.5 and 2.5 respectively.
Figure 4.30 provides the depth-averaged velocity profiles for these. On the right or
outside bend there is no change in velocity profile. On the left or inside of the bend
the velocities are reduced and the reduction corresponds to the magnitude of the

Practical issues roughness, conveyance and afux 211

2.5

Depth (m)

1.5
Measured Data
1

CES Prediction
Scenario 7

0.5
Scenario 8
0
0

10

20

30

40

50

60

70

80

90

10 0

Flow (m3s-1)
Figure 4.28 Scenario 7 and 8 stage-discharge prediction for the River La Suela.

Table 4.2 Summary of predicted and measured flow rates for 1.52 m and 1.75 m flow depths.

Measured
Original CES
Scenario 1 double bed roughness
Scenario 2 halve bed roughness
Scenario 3 double bank roughness
Scenario 4 halve bank roughness
Scenario 5 concrete-lined channel
Scenario 6 concrete-lined rectangular
Scenario 7 double bed slope
Scenario 8 halve bed slope
Scenario 9 sinuosity 1.5
Scenario 10 sinuosity 2.5

@ 1.52 m depth

@ 1.75 m depth

Flow rate
(m3 s-1 )

Change in flow
rate (%)

Flow rate
(m3 s-1 )

Change in flow
rate (%)

34
38
22
69
38
41
65
69
55
28
39
37

11
0
43
80
2
7
70
79
44
28
2
4

49
50
30
86
48
54
84
86
71
36
50
47

1
0
40
74
3
8
69
73
43
28
1
5

sinuosity. This is typically reversed for overbank flows, with the higher main channel
velocities observed in the outside of the bend (as observed in the Flood Channel Facility
data). For this case, the impact on the overall flow rate is small (<5% for both cases,
Table 4.2).

212 Practical Channel Hydraulics: Roughness, Conveyance and Afux

2.5

Depth (m)

1.5
Measured Data
1

CES Prediction
Scenario 7

0.5
Scenario 8
0
0

50 0

1000

150 0

2000

250 0

300 0

Conveyance (m3s-1)
Figure 4.29 Scenario 7 and 8 conveyance prediction for the River La Suela.

CES 1.52 m

Depth-averaged velocity (ms-1)

Data 1.52 m
Scaled bed

2.5

Scenario 9
Scenario 10
2

1.5

0.5

0
0

10

15

20

25

30

35

40

Lateral distance across section (m)

Figure 4.30 Scenario 9 and 10 depth-averaged velocity predictions for 1.52 m flow depth
(QCES = 38 m3 s1 ; Qdata = 34 m3 s1 ; Scenarios Q9 = 39 m3 s1 ; Q10 = 37 m3 s1 ).

Practical issues roughness, conveyance and afux 213

90

Scenario Results 1.75 m

80

CESCalibrated Flow 1.75 m

Scenario Results 1.52 m

70

CESCalibrated Flow 1.52 m

60
50

Predicted flow
(m3s-1)

40
30
20
10

.5

.5

f2
nu

Si
10

9S

in

uo

os

ity
o

sit
yo

slo

f1

pe

e
ed
alv
eb
8H

be
ble
ou

on

cre

7D

ed
lin
te-

lo
p
ds

gu
tan
rec

ch
ed
lin
te-

cre
on
6C

5C

lar

l
ne
an

ne
gh
ou
kr

an
alv
eb

4H

ba
ble
ou
3D

ss

s
es
hn
ug
ro

nk

ed
alv
eb

2H

1D

ou

ble

be

dr

ro

ou

ug

gh

hn

ne

es

ss

Scenarios

Figure 4.31 Summary of predicted and measured flow rates for 1.52 m and 1.75 m flow depths.

iv Summary flow results all scenarios

Table 4.2 provides a summary of the predicted and measured flows for Scenarios 1 to
10 at flow depths of 1.52 m and 1.75 m. It also indicates the percentage difference in
flow rate relative to the calibrated CES flow rate. Figure 4.31 provides a corresponding
visual impression of these results.
From this, it is apparent that Scenarios 1, 2, 5, 6, 7 and 8 have the largest influence
on the flow rate for the La Suela i.e. altering the bed roughness, concrete-lining the
channel and altering the bed slope. Scenarios 3, 4, 9 and 10 have the least impact i.e.
altering the bank roughness and the sinuosity. Although these outcomes are specific to
the River La Suela and are likely to be different for other channel types (e.g. irregular,
compound, narrow/wide, vegetation, boulders, alluvial etc.), they serve to show the
qualitative changes that might be expected.

4.2.6 Application of the CES to a mountain stream

with boulders
Aim
CES stage-discharge simulations are carried out for two mountain rivers, the River
Waiwakaiho (New Zealand) and the River Tomebamba (Ecuador). These are used
to demonstrate the use of the CES for mountain streams with boulders and to propose an alternative roughness law to the current CES prediction of f (Chapter 3,

214 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Colebrook-White law) where large boulders are present. The alternative approach is
based on that of Abril and Knight (2004).
Approach for boulders
Abril and Knight (2004) proposed an alternative approach for evaluating boulder
roughness in mountain rivers based in part on the work of Ramette (1992), where,
2


12Hmc
fmc = 8 5.75 log
3d90 + r

(4.1)

and d90 (m) is the sediment dimension, r is a coefficient representing the bed form
roughness and fmc is evaluated from,


f = fmc 0.669 + 0.331D0.719
r

(4.2)

Hmc is taken as the maximum local cross-section depth and r is zero for the channels
considered.
The CES supports simulations for a wide range of channel types e.g. different section and plan form shapes, sizes, vegetation types, substrate cover, etc. An
important component of the calculation is the ability to adapt the model as appropriate to the particular site characteristics. For example, where the roughness varies
signicantly within a cross-section, the model allows for a description of the local
roughness values and these are taken into account in the calculation. However, there
are some areas which are more challenging, where the basic science is still emerging,
for example:
1
2
3

vegetation which changes its behaviour (e.g. form, resistance) as a function of

velocity and depth;
alluvial channels where the bed roughness is characterised by bed form growth
and wash-out;
very large channels (e.g. Alto Parana, Argentina, flows upwards of 20,000 m3 s1 )
where the roughness laws developed for small channel and pipe flow may not be
applicable;
steep mountain streams with large boulders comparable with flow depth.

The following pages now illustrate topic 4 above further, using data from two
mountain rivers.
The River Waiwakaiho site
Stage and discharge measurements were taken at station SH3 along the River
Waiwakaiho over the nine year period, January 1980 to 1989. The Waiwakaiho serves
a catchment area of 58 km2 , has a mean annual flood of 327 m3 s1 and an average
flow rate of 0.63 m3 s1 . The observations were made at three cross-sections along a
fairly straight 100 m reach, each approximately 40 m wide and 2 m deep. The water
surface slope varied from 0.0091 at low depths to 0.0176 at large depths. The bed

Practical issues roughness, conveyance and afux 215

comprises cobbles and boulders, some as large as 2 m in diameter (Figure 4.32), and
the banks consist of boulders with occasional scrub. The observed Manning n values
vary with depth in the range 0.047 to 0.180.
River and cross-section def inition for River Waiwakaiho site
Based on this description the assigned roughness values include (Figure 4.33):

Channel bed The RA bedrock (nsur = 0.025) and boulders >50% (nirr = 0.045)
giving nl = 0.051.
Channel banks The RA cobbles 64256 mm (nsur = 0.035).
Channel high banks The RA sand (nsur = 0.02).

The River Tomebamba site

The River Tomebamba is a natural mountain river located in the Southern Andean
region of Ecuador, another tributary to the Paute River. Water level, discharge,
velocity and roughness measurements were taken along the Tomebamba River at

Figure 4.32 River Waiwakaiho at SH3 looking (a) upstream and (b) downstream along the reach
(Hicks & Mason, 1998).

Figure 4.33 River Waiwakaiho cross-section, roughness zones and top-of-bank marker locations.

216 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Monay. The channel is 25 m wide and the reach-averaged longitudinal bed slope
is 0.0176. Here, the river bed consists of large boulders, approximately 1.3 m in
diameter, and the Manning n values, back-calculated from discharge and water level
measurements, range from 0.08 to around 0.15 (Abril & Knight, 2004).
River and cross-section def inition for River Tomebamba site
Based on this description the assigned roughness value for the whole section is
(Figure 4.34):

Channel Cover The RA bedrock (nsur = 0.025) and boulders >50% (nirr =
0.045) giving nl = 0.051.

Results for the River Waiwakaiho and the River Tomebamba

Figure 4.35 provides the CES predicted stage-discharge based on the aforementioned
RA values. These are labelled as uncalibrated and it is clear the CES roughness is too
small, resulting in the flow capacity being over-predicted. The roughness values for
the two rivers were increased to better simulate the data. The Waiwakaiho calibrated
channel bed roughness is nl = 0.5 (2 m boulders) and the Tomebamba calibrated
channel roughness is nl = 0.25 (1.3 m diameter boulders). These calibrated curves
(dashed lines) fall closer to the data; although the curve shape of the data is not captured. The reason for this is most likely related to the use of the Colebrook-White law
in areas where the boulder size is comparable with depth and different flow mechanisms are taking place. Note that a unit roughness value of 0.25 corresponds to a ks
value of 7.4 m, substantially larger than the 12 m boulders.
The boulder approach of Abril and Knight (2004) is also used to simulate the
data. Here, the initial d90 is taken as 2 m and 1.3 m for the Waiwakaiho and
Tomebamba. The curves were then calibrated using the sediment dimension and the
final d90 s are 2.1 m and 2 m for the Waiwakaiho and Tomebamba. Figure 4.35 shows
how the shape of the curve follows the data more closely than the Colebrook-White
approach.

Figure 4.34 River Tomebamba cross-section, roughness zones and top-of-bank marker locations.

Practical issues roughness, conveyance and afux 217

2.5
2.3
2.1

Depth (m)

1.9
1.7
1.5

Waiwakaiho: Data
Waiwakaiho: Boulder approach
Waiwakaiho: CES calibrated
Waiwakaiho: CES uncalibrated
Tomebamba: Data
Tomebamba: Boulder approach
Tomebamba: CES calibrated
Tomebamba: CES uncalibrated

1.3
1.1
0.9
0.7
0.5
0

10

20

30

40

50

60

70

80

Flow (m3s-1)
Figure 4.35 Stage-discharge predictions for the River Waiwakaiho and the River Tomebamba using
different roughness calculations.

3.5

Waiwakaiho: Data
Waiwakaiho: Boulder approach
3

Waiwakaiho: CES calibrated

Tomebamba: Data

Depth (m)

2.5

Tomebamba: Boulder approach

Tomebamba: CES calibrated

1.5

0.5
0

0.05

0.1

0.15

0.2

0.25

0.3

Manning n
Figure 4.36 Back-calculated Manning n values for the River Waiwakaiho and the River Tomebamba
using different roughness calculations.

218 Practical Channel Hydraulics: Roughness, Conveyance and Afux

At flow depths greater than 2 m, both approaches provide poor predictions

which may be due to the particular boulder layout at the site or, more fundamentally, that the boulder formulae may only be applicable for a limited range e.g. flow
up to the level of the sediment diameter, d90 . If the boulder approach is adopted
in the CES, a transitional roughness rule between the two approaches should be
explored.
Figure 4.36 provides the back-calculated Manning n values for both approaches.
As before, the boulder approach best captures the shape of the data.

4.3 USE OF BACKWATER MODULE FOR ESTIMATING

WATER LEVELS ALONG THE RIVER MAIN
Aim
This example demonstrates the use of the CES-AES backwater calculation to calculate
water levels along a reach of the River Main with:
i
ii

an inflow of 60 m3 s1 and a downstream control of 36 mAD; and

an inflow of 10 m3 s1 with normal depth conditions at the downstream end.

For condition (i), the channel section 600 m along the reach needs to convey the
100 year return period storm (Q100 = 60 m3 s1 ) below a water level of 38 mAD,
despite the new downstream control at 36 mAD.
For condition (ii), the water levels 1400 m along the reach need to be maintained above 35 mAD for flows above 10 m3 s1 , to ensure the land drainage pumps
can operate. Below this flow, the land drainage pumps are automatically turned off.
This is explored for the 12 month period prior to installation of the new 36 mAD
downstream control. Thereafter, the water levels will be sufcient to ensure pump
operation.
The River Main reach
A 1.6 km reach of the River Main in County Antrim, Northern Ireland, is used
for this example (see Chapter 2, Section 2.2.2 for a full description). The inflow to
the reach is just downstream of Lisnallan weir and the reach extends to Gracehill
Bridge. The cross-sections are located at approximately 50 meter intervals and the
topographical survey data includes channel and oodplain information, measured at
right angles to the main flow direction, and measured as far across the oodplain as
possible.
Bridge End Bridge is located about 400 m downstream of Lisnafillan weir and
the roughness values for the reach are based on the calibrated values for this site
(Section 4.2.3).
Note 1: This reach of the River Main has a series of fish groyne structures (approximately every 500700 m) which control the depths at low flows. These are excluded
from the current CES-AES model but could readily be introduced as additional
reach-averaged cross-sections with short reaches or as downstream controls.
Note 2: This reach of the River Main includes 2 bridge structures. These are
excluded from the current example but may be introduced using the AES.

Practical issues roughness, conveyance and afux 219

Results for the River Main

An initial run is undertaken to assess the water levels along the reach at a more common
flow rate of 40 m3 s1 . Figure 4.37 provides the model output for this initial run, with
inflow 40 m3 s1 , and a downstream level of 36 mAD. The surface prole is more-orless parallel to the ground prole in the upstream reach and in the downstream reach
it approaches the controlled level of 36 mAD. The velocity prole indicates the higher
velocities in the upstream reaches (in the region 1.252.0 ms1 ) and lower velocities
(<1.25 ms1 ) further downstream due to the control. The water levels approximately
600 m and 1400 m downstream of Lisnallan weir are 37.7 mAD and 36.1 mAD
respectively.
Figure 4.38 provides the model outputs for condition (i). The downstream water
level is controlled at 36 mAD (assuming no drowning) and the 100 year return period
storm, Q1 00 = 60 m3 s1 , is set as the inflow to the reach. The water level 600 m downstream of Lisnallan weir is 37.9 mAD, which is lower than the required 38 mAD for
condition (i). The lower and upper credible scenarios (dashed lines) give water levels of
37.7 mAD and 38.0 mAD respectively at this location. As these are close to the required
water level of 38 mAD, some additional resource may be invested to better understand
any sources of uncertainty. If the water levels are subsequently considered uncertain,
possible options may be explored in the CES e.g. improved vegetation maintenance
through altering roughness, dredging through altering channel cross-sections etc.
Figure 4.39 provides the model outputs for condition (ii). The inflow to the reach
is 10 m3 s1 and normal depth conditions are present at the downstream end. These
correspond to a downstream flow depth of 34.3 mAD. The water surface profile

40

2.5

Expected Water Level

39

Bed

37

1.5

36

35

Velocity (ms-1)

Level (mAD)

Velocity

38

34

0.5
33

32

0
0

200

400

600

800

1000

1200

1400

1600

1800

Distance along reach (m)

Figure 4.37 Long-section of the River Main reach showing the water surface and velocity profiles for
a 40 m3 s1 inflow to the reach and 36 mAD downstream control.

220 Practical Channel Hydraulics: Roughness, Conveyance and Afux

40

2.5

Expected water level

Bed

39

Upper/lower credible scenarios

2

Velocity

37

36

1
35

-1

1.5

Velocity (ms )

Level (mAD)

38

34

0.5
33

32

0
0

200

400

600

800

1000

1200

1400

1600

1800

Distance along reach (m)

Figure 4.38 Long-section of the River Main reach showing the water surface, upper and lower credible
scenarios and velocity profiles for a 60 m3 s1 inflow to the reach and 36 mAD downstream
control.

40

2.5

Expected Water Level

39

Bed

37

1.5

36

35

Velocity (ms-1)

Level (mAD)

Velocity

38

34

0.5
33

32

0
0

200

400

600

800

1000

1200

1400

1600

1800

Distance along reach (m)

Figure 4.39 Long-section of the River Main reach showing the water surface and velocity profiles for
a 10 m3 s1 inflow to the reach with normal depth at downstream end (34.3 mAD).

Practical issues roughness, conveyance and afux 221

follows the ground profile more closely as would be expected at low flows, where the
effect of the local longitudinal gradient is more dominant. The velocities throughout
the reach are in the range 0.5 to 1.5 ms1 as there is no downstream control to reduce
these at the downstream end. The velocity peaks (2.1 ms1 ) at 800 m downstream,
which is most likely a result of the local topography i.e. bed level and cross-section
shape. At 1400 m downstream of Lisnafillan weir, the expected, upper and lower
water levels are all 34.6 mAD. This is well below the required 35 mAD for the drainage
pumps and it will be necessary to explore different solutions for this e.g. turn the land
drainage pump off at a higher flow rate, introduce a temporary downstream control
of 35 mAD prior to the planned 36 mAD control.
4.4 ESTIMATING AFFLUX AT BRIDGES

4.4.1 Field scale verif ication of bridge backwater

analysis at Pea Creek, Alabama
Aim
To predict the backwater associated with bridge afflux at a field site where comparisons can be made with survey data (available at http://ms.water.usgs.gov/publications/
backwater_HA/HA-608.html).
Pea Creek
The field site is a heavily vegetated floodplain near Louisville, Alabama. It is one of
the field sites that was included in the USGS study of backwater at bridges, begun in
1969, that collected detailed topographic and water level surveys plus flow data for
35 floods at 11 sites in the southern USA. This data set remains one of the primary
sources of field observations for the study of bridge afflux. Measurements from the
Pea Creek site were made for two flood events in December 1971 and published in
the Hydrologic Investigations Atlas HA-608 (Colson et al., 1978). The study area is
shown in Figures 4.40 and 4.41.
Roughness def inition
The Pea Creek is a tributary to the swamp and creek systems of southern Alabama.
Floodplains in the region are densely covered with trees (Figure 4.42 shows typical
floodplain cover). Underlying substrates are typically silt, sand and clay deposits.
Channels are incised into the broad floodplains and so the Roughness Advisor has
been used to set unit roughness zones for channel and floodplain zones.

Channel bed The RA silt with trailing bank side plants is adopted, resulting in
a unit roughness of nl = 0.053.
Floodplain The RA floodplain heavy stands of trees (with depths below branches)
is adopted on sand substrate, with resulting nl = 0.102. The reasoning is that the
entire floodplain is densely wooded, but there are unlikely to have been significant
trailing leafy branches for the vegetation types observed in the field, especially on
the date of the modelled event.

222 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.40 Pea Creek site map. Grey box shows extent of study area. Dotted lines show approximate
along-stream locations of USGS cross section survey data, section 3 is downstream. Road
bridge is situated between sections 4 and 5 (After USGS, 1979).

Figure 4.41 Pea Creek wooded floodplain and road bridge (Photo: USGS).

Channel and oodplain roughness at bridge sections The AES bridge calculations use Mannings n to represent flow resistance, rather than the unit
roughness parameter of the CES (see Section 3.2.5). The bridge unit requires a three
panel roughness specication (left overbank, main channel and right overbank),

Practical issues roughness, conveyance and afux 223

Figure 4.42 Choctawhatchee/Pea River channel and vegetation (Photograph: Alabama Clean Water
Partnership. Reproduced with the permission of StormCenter Communications, Inc.;
http://wsfa.envirocast.net/index.php?pagename=ow_watershed_choctawatchee_river).

separated by left and right bank markers. Although the Mannings n and unit
roughness are conceptually different quantities, for a short, straight, relatively
uniform reach the distinction between the CES and the conveyance calculations
based on Mannings equation is less important. Hence for the short, straight reach
within the bridge the unit roughness values have been adopted.
Bridge roughness The AES calculates energy losses associated with flow acceleration in the transition reaches upstream and downstream of the bridge, plus
energy loss associated with friction within the structure. The bridge at Pea Creek
is a deck bridge supported by 16 timber pile bents with centres at 4.57 m spacing.
It is assumed that each support is made of 4 wooden piles arranged in a 2 2
pattern. The AES treats piers as if they extend continuously from the upstream
face to the downstream face of the bridge, which is a common type of structure in
the UK. This is unlikely to be the case for the Pea Creek bridge, where the user of
more complex hydraulic modelling software would be likely to choose a method
based on conservation of momentum to represent form and drag losses around
the bridge piers. However, in the simpler AES, only changes in the cross sectional
flow area and frictional energy losses are accounted for, and so an effective value

224 Practical Channel Hydraulics: Roughness, Conveyance and Afux

of roughness is needed. In this case, the CES oodplain value has been adopted on
the basis that, as a crude approximation, dense tree stands on the oodplain may
have create similar resistance to flow as closely spaced the timber pile supports
within the channel.
Cross section def inition
Floodplain and bridge cross section data are as published by Colson et al. (1978). The
floodplain is in general around 300400 m wide, but contracts to about 75 m at the
bridge.

River cross sections CES cross sections are set up using USGS channel survey data at the locations shown in Figure 4.40. A simple roughness zonation is
adopted, with the RA roughness zones for oodplain and channel used, as shown
in Figure 4.43 for Section No. 5, which is located approximately 77 m upstream
of the bridge (upstream face).
Bridge section The AES bridge section is shown in Figure 4.44. The bridge
length (upstream to downstream faces) is 8 m and the overall valley gradient is
very shallow, hence the AES option to use the same cross section and structure
profile for both upstream and downstream faces is selected. The soffit elevation
is given as 111.8 m above datum and the road elevation as 112.3 mAD.

Results for Pea Creek bridge

Conveyance rating curves for the modelled reach are shown in Figure 4.45. The bridge
section is clearly visible showing reduced conveyance at a given elevation owing to the
obstruction to flow within the bridge section combined with the frictional resistance.
There is, as expected, a sharp change in the curve between soffit and road level, with
little change in conveyance. Above the road level, the increase in conveyance with
elevation at the bridge section is roughly parallel with the curves for unobstructed
sections.
The USGS Hydrologic Investigations Atlas HA-608 includes detailed measurements for an event on 21 December 1972 (Table 4.3), where the discharge, measured

Figure 4.43 Cross-section for Pea Creek site for Section No. 5.

Practical issues roughness, conveyance and afux 225

Figure 4.44 Pea Creek AES bridge section.

Figure 4.45 CES-AES conveyance curves for Pea Creek.

by velocity area gauging from the bridge, was 50.5 m3 s1 . The corresponding level of
the flood outline surveyed by the USGS at Section No. 3 is 108.5 mAD. This provides
a downstream boundary condition for the backwater analysis.
The CES-AES backwater analysis produces water surface profiles as shown in
Figure 4.46. The afflux is clearly visible in the comparison of the profile for the reach
including the bridge with a second profile where the bridge structure is removed from
the model. The modelled water levels can be compared with the surveyed water levels
from the flood outline of 21 December 1972.
The surveyed water levels are derived from ground elevations at the mapped flood
outline. In some cases, these are uncertain because the outline was between survey
points, with no knowledge of the ground elevations in between. In addition, there are

226 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Table 4.3 Surveyed and modelled flood levels for the 21 December 1972 event at Pea Creek.
Section

Flood level from maximum flood outline

Modelled flood level

3
4
Bridge
5
6

108.50
109.05 (uncertain)
109.23
109.53
109.93

108.50 (downstream boundary condition)

108.68
109.29
109.53
109.60

Figure 4.46 Backwater profiles for Pea Creek.

numerous surveyed water levels, assumed to be from water marks on trees, which show
uncertainty in identifying a cross section average flood water level. Given the uncertainty in the estimate of flood levels from the measurements, it seems that the CESAES backwater prole offers a good approximation of the water surface prole in
the vicinity of the bridge. In particular, the comparison at Section No. 5, close the
location of the afflux, shows very close agreement. Figure 4.47 shows the water surface profiles calculated for the minimum and maximum credible roughness values
generated by RA.
This example gives details of one field data test of the AES. Further comparisons
between afflux data modeled using AES and field measurements by USGS (1978) have
been given by Mantz and Benn (2009).

Practical issues roughness, conveyance and afux 227

Figure 4.47 Backwater profile for Pea Creek for minimum, default and maximum roughness cases.

4.4.2 Field scale bridge backwater analysis

on the River Irwell, UK
Aim
To predict the backwater associated with bridge afflux at a field site on a river in
northern England.
Holme Bridge on the River Irwell
The River Irwell flows through Salford close to the city of Manchester in northern
England. The Irwell has numerous small tributaries, some of them originating as
steep watercourses owing into small, upland oodplains before joining upstream
of Salford. The case study is a twin arched masonry bridge downstream of the
town of Rawtenstall. The oodplain and channel gradient is 0.003 for the reach
containing the bridge. The oodplain is about 150 m wide and the main channel approximately 25 m wide. The channel is incised at the bridge. The bridge is
one of 40 included in a detailed hydrodynamic river model. There are no direct
observations available for flows and water levels at the bridge site, but the river is
gauged nearby allowing for condent estimates of design flow rates and corresponding water levels are available from the hydrodynamic model for comparison. The
upstream and downstream faces of the bridge are shown in Figures 4.48(a) and (b)
respectively.

228 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.48(a) Holme Bridge, upstream (Photo: JBA Consulting/Maltby Land Surveys) (See colour
plate section).

Figure 4.48(b) Holme Bridge, downstream (Photo: JBA Consulting/Maltby Land Surveys) (See colour
plate section).

Practical issues roughness, conveyance and afux 229

Roughness def inition

The reach is an upper branch of the Irwell. The valley has shallow soils underlain by
bouldery deposits and bedrock. The floodplain around the bridge is rough grassland
with sparse tree cover. The river bank is similar in character to the floodplain. Cross
sections upstream and downstream of a bridge in CES-AES require three roughness
panels (representing main channel and floodplains on the left and the right). The
Roughness Advisor has therefore been used to set roughness zones for the main channel
and the floodplain.

Channel bed The RA coarse gravel with riffles is adopted, resulting in a unit
roughness of nl = 0.031.
Floodplain The RA floodplain light brush with trees is adopted, with resulting
nl = 0.060.
Channel and oodplain roughness at bridge sections The AES bridge calculations use Mannings n to represent flow resistance, rather than the unit roughness
parameter of the CES. The bridge unit requires a three panel roughness specication (left overbank, main channel and right overbank), separated by left and
right bank markers. Although the Mannings n and unit roughness are conceptually different quantities, for a short, straight, relatively uniform reach the
distinction between the CES and the conveyance calculations based on Mannings
equation is less important. Hence for the short, straight reach within the bridge
the unit roughness values have been adopted.
Bridge roughness The AES adopts Mannings n values to represent the friction
within the arch. In this case a value of n = 0.020 is adopted appropriate to an
unglazed masonry or brick surface in good condition.

Cross section def inition

Floodplain and bridge cross section data were collected by survey.

River cross sections CES cross sections are set up using the channel survey data at
the locations shown in Figure 4.49. A simple roughness zonation is adopted, with
the RA roughness zones for floodplain and channel used, as shown in Figure 4.50
for the CES sections located approximately 123 m upstream of the bridge and
89 m downstream.
Bridge section The AES bridge section is shown in Figure 4.51. The bridge length
(upstream to downstream faces) is 3.5 m. The cross section and structure profiles
are the same for both upstream and downstream faces. The two arch openings
differ slightly in shape but the soffit elevation is approximately 156 m above datum
for both. The road elevation is set to equal 157.5 mAD (in fact representing the
solid parapet).

Results for Holme Bridge

Conveyance rating curves for the modelled reach are shown in Figure 4.52. The bridge
section is clearly visible showing reduced conveyance at a given elevation owing to the

230 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.49 Typical bankside and floodplain vegetation around the study reach (Photo: JBA Consulting/
Maltby Land Surveys).

obstruction to flow within the bridge section combined with the frictional resistance.
There is, as expected, a sharp change in the curve between soffit and road level,
with changes in conveyance above road level roughly parallel with the curves for
unobstructed sections.
Three design flow conditions were analysed using CES-AES, corresponding to
flood flows for the return periods of T = 2, 10 and 100 years (or annual excedance
probabilities of 50%, 10% and 1%, respectively). The downstream boundary conditions were taken from outputs of the large hydrodynamic river model that included
the Holme Bridge. This was a model built using the ISIS Flow software package, with
Holme Bridge represented using the arch bridge method of Brown (1988).
The CES-AES backwater analysis produces water surface profiles as shown in
Figure 4.53 for each of the three design flows. In each case, the plot shows the water
surface profile computed for the model including the bridge (heavy lines) and also a
profile computed after removing the bridge structure from the model. The afflux is
hence clearly visible by comparing each pair of profiles.
The afflux is clearly seen to increase with increasing flow rate and hence blockage
ratio, as the water level rises into the arch openings. At low water levels, the arch
bridge has little effect on the flow because only the relatively slender pier causes an
obstruction and the bridge structure contributes relatively little to the frictional energy
losses through the section. When water rises above the level of the lowest arch springer

Practical issues roughness, conveyance and afux 231

Figure 4.50 CES channel cross sections (a) 123 m upstream and (b) 89 m downstream of bridge
section.

Figure 4.51 Holme Bridge AES bridge section.

then the structure begins to have an effect on the water surface profile. The water
surface just rises to soffit level in the modelled 100-year flow condition and so the
friction losses caused by the bridge structure are large.
There are no field measurements available to test these results, but it is of interest
to compare them with the outputs of the ISIS hydrodynamic model for the same design
flow conditions. Table 4.4 shows the flow and water level at the downstream section,
the water levels computed at the bridge section in the ISIS simulation and the results

232 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.52 CES-AES conveyance curves for Holme Bridge.

Figure 4.53 Backwater profiles for Holme Bridge for three design flow conditions or return period
T = 2, 10 and 100 years. For the profiles labelled with an asterisk (*) the bridge was
removed from the model.

Practical issues roughness, conveyance and afux 233

Table 4.4 Surveyed and modelled flood levels for T = 2, 10 & 100 years at Holme Bridge.
Return period
(years)

Flow rate
(m3 s1 )

Downstream
water level
(mAD)

ISIS bridge
section water
level (mAD)

CES-AES water
level, downstream
bridge face (mAD)

CES-AES water
level, upstream
bridge face (mAD)

2
10
100

45
71
117

153.72
154.24
155.07

154.20
154.72
155.60

154.41
154.95
155.78

154.54
155.10
155.96

from CES-AES for the downstream and upstream bridge faces. The ISIS bridge unit
is based on an empirical dimensionless correlation analysis of afflux, calibrated with
laboratory data, and of the same general form as the similarity model used in the
AES analysis of transition lengths and energy loss coefficients. This model contains no
bridge length scale and is therefore unable to represent the backwater profile through
the bridge waterway.
The comparison shows that the CES-AES water levels are similar to those computed using ISIS, but slightly higher. A discrepancy of this size can be judged in the
light of extensive tests against scaled laboratory data (Atabay, 2008) that have shown
the ISIS ARCH bridge unit to have uncertainties in the range of approximately 0.5
to 1.0 meters when estimating an afflux of scale 0 to 3.0 meters.

4.5 ESTIMATING AFFLUX AT CULVERTS

4.5.1 Shallow culvert backwater analysis

in a long reach
Aim
To demonstrate backwater analysis for a channel reach including a culvert.
Culvert site
The example is based on the River Main reach described in earlier examples. The reach
is 750 m long, and is initially configured for backwater analysis without a structure
in place. There are 19 river sections, labelled from 67 (downstream) to 87 (upstream)
and a 20 m long culvert is added 560 m downstream. Results are illustrated for a
number of example culvert configurations.
Cross section and roughness def inition
The channel cross sections immediately upstream and downstream of the culvert are
shown in Figure 4.54. The channel upstream and donwnstream has RA unit roughness
values nl = 0.102 for the floodplain, nl = 0.045 for the river banks and nl = 0.025
for the main channel. The culvert roughness, expressed in terms of Mannings n, is
n = 0.010.

234 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.54 Downstream channel and structure cross sections for a pipe culvert in the River Main
reach.

Culvert structure
The structure is modelled as a pipe culvert, with a circular concrete inlet set into a
headwall (CIRIA type A) and square inlet edge. The openings are placed on alignment
with the deepest part of the channel. The barrel diameter is 4.5 m.
Results
Conveyance curves are obtained by running a backwater analysis for the reach and
plotting the CES-AES output curves, including the results of the afflux calculations
(Figure 4.55). It can be seen that, as expected, the conveyance through the culvert
is restricted at a given elevation because of the reduction in available flow area. The
pipe culvert displays the expected reversal of the conveyance curve as the water level
elevation rises into the arch formed in the upper half of the pipe. Above road level, the
conveyance increases rapidly with elevation as the channel cross section area expands
rapidly above the road.
Water surface proles are shown in Figure 4.56 for two different downstream
boundary conditions, in each case comparing the proles obtained with and without the culvert. Figure 4.56(a) shows a case where oodplain flow is assumed at the
downstream boundary. Here, the culvert overtops the road and the resulting combined pressure and weir flow results in an afflux upstream of approximately 2 m.
Figure 4.56(b) shows a flow just within bank at the downstream section. In this case

Practical issues roughness, conveyance and afux 235

Figure 4.55 Conveyance curves for a pipe culvert in the River Main reach. Solid line shows culvert
conveyance at inlet. Dark dashed line shows conveyance in the channel downstream of
the culvert, light dashed line upstream.

there is free surface flow through the barrel of the culvert and a smaller afflux of
approximately 1 m. Note that there is an increase in velocity through the culvert barrel,
as shown by the velocity traces in the CES-AES long section output (Figure 4.57).

4.5.2 Exploratory culvert design and maintenance

calculations in CES-AES
Aim
To demonstrate analysis of a trial culvert design. The analysis includes hypothetical
scenarios for blockage and deterioration of the culvert barrel condition. The calculations are presented to illustrate an exploratory analysis for a hypothetical culvert
design.
Culvert site
The example illustrates simple, exploratory trial calculations for a culvert crossing of
a minor road. The required length of the culvert is 12 m and the project requires a
box section culvert design. The culvert site is located between two river cross sections
in a uniform reach with natural downstream channel control and an average slope of
approximately 0.005. The analysis therefore assumes normal depth flow conditions in
the reach. The CES-AES model has only three sections. The downstream river section
is labelled 1 and the upstream section is labelled 4. Sections 2 and 3 are contained
within the culvert unit in CES-AES.

(a)

(b)
Figure 4.56 Water surface proles for pipe culvert in the River Main reach (a) out of bank flow =
75 m3 s1 , downstream water level = 36 mAD, (b) in bank f low = 19.7 m3 s1 ,
downstream water level = 34.5 mAD. Horizontal axis is distance along stream in meters.

Practical issues roughness, conveyance and afux 237

Figure 4.57 Long section backwater profile for reach with 20 m long pipe culvert, showing increased
velocity through the culvert barrel.

Table 4.5 Design flow estimates.

Return period (years)

Flow (m3 s1 )

10
25
50
100

3.5
4.6
5.5
8.6

Hydrological analysis has provided design flows for four return periods, as shown
in Table 4.5. The area upstream is farmland although there is a car park at the edge of
the floodplain. The design should keep any increase in flood water levels to no more
than 0.10 m for the 10-year return period flow.
Cross section and roughness def inition
The channel cross section downstream of the culvert is shown in Figure 4.58. The
main channel comprises medium sized cobbles and so the RA unit roughness values
nl = 0.102 is adopted. The floodplain is well grazed pasture and so the RA turf
vegetation with no irregularities is adopted with unit roughness nl = 0.021. The CES
river cross sections upstream and downstream of the culvert unit have to include three
roughness zones, and so there is no separate bank roughness. The banks are assumed

238 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.58 Downstream channel cross section. Bed level is 9.25 m.

to have similar roughness to the main channel owing to a combination of substrate

and some trailing bank side vegetation.
Culvert structure trial calculations
An estimate of the depth of flow in the channel through the culvert site is obtained from
the Conveyance Generator outputs for Section 1. Figure 4.59 shows the flow rating
curve for the downstream section. The water level is 10.25 mAD, which corresponds
to a flow depth of 1.0 m. The section averaged velocity rating is shown in Figure 4.60.
The velocity corresponding to the flow depth of 1.0 m is obtained directly using the
outputs interrogator as 1.40 ms1 .
An approximate trial size for the culvert barrel is determined as follows. The
section averaged velocity in the culvert should be greater than the channel velocity to
discourage excessive siltation. An increase of 20% is desired in this case, giving a design
velocity of 1.7 ms1 . The discharge is equal to the cross section averaged downstream
velocity multiplied by the flow area. The area required to convey the design flow with
a velocity of 1.7 ms1 is therefore 3.5/1.7 = 2.1 m2 . The depth of flow in the culvert
should be similar to the normal depth in the channel, which is 1.0 m. A suitable
vertical dimension for the barrel would be 1.3 m, allowing for 0.3 m freeboard. With
an assumed flow depth of 1.0 m, the required width of the box section would be 2.1 m.
Hence the trial culvert barrel dimensions will be a span of 2.1 m and a rise of 1.3 m.
The culvert barrel will be pre-fabricated concrete. Mannings roughness coefficient
n would typically be in the range 0.012 to 0.015 for a new concrete culvert. To allow
for some minor degradation in use a design value of 0.018 is adopted. The structure is
modelled in AES as a box culvert, with a rectangular concrete inlet including headwall
and wingwalls and square inlet edges. A similar structure is shown in Figure 4.61
during high flow conditions. The culvert opening data is shown in Figure 4.62.
Exploratory analysis of the trial design
The trial design is analysed by applying the design flow conditions and generating a
backwater profile. Results for the 10 year return period flow are shown in Figure 4.63.

Practical issues roughness, conveyance and afux 239

Figure 4.59 Conveyance generator flow rating curve for the downstream section, showing downstream water level for a design flow for the 10-year return period of 3.5 m3 s1 .

Figure 4.60 Conveyance generator velocity rating curve for the downstream section and Interrogator, showing section averaged velocity for a water level of 10.25 mAD.

240 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.61 Culvert inlet during high flow conditions. (Photo: John Riddell).

Figure 4.62 Culvert trial opening data in AES.

It can be seen that the culvert causes only a very small increase in water levels upstream
of approximately 0.03 m, much less than the stipulated value of 0.1 m. Figure 4.63 also
shows water surface profiles for the credible uncertainty bounds around the default
roughness, as encoded within CES-AES. The range of uncertainty about water levels is
much greater than the indicated change in water levels resulting from the introduction
of the culvert, confirming the trial design.

Practical issues roughness, conveyance and afux 241

Figure 4.63 Water surface profiles showing best estimate and uncertainty bounds for the trial design,
with and without culvert for downstream water level 10.25 m.

Figure 4.64 shows water surface profiles for the 10 year, 50 year and 100 year
design flows, both with and without the culvert. At higher flows, the culvert clearly has
a greater impact on upstream water levels. For the 50 year design flow the upstream
water level just reaches the soffit and a large afflux is produced. For the 100 year
flow, the structure is overtopped, flooding the road. The analysis of these larger flow
events helps in understanding the robustness of the design. For example, should there
be a possibility of development of housing or retail property in the area upstream of
the road crossing then the performance of the culvert under higher flow conditions
becomes much more important.
Over time there could be deterioration of the condition of the culvert barrel leading
to an increase in flow resistance and hence a possible increase in the afflux upstream.
An increased roughness value of Mannings n = 0.035 has been applied in the AES to
test this scenario for future culvert condition. It is also possible that blockage could
occur during a high flow event. A simple way to approximate the effects of blockage
in AES is to adjust the structure cross section data to reflect a reduction in flow area
through the culvert. One such scenario is illustrated in Figure 4.65 where the culvert
barrel and opening dimensions have been reduced to a span of 1.6 m and a rise of
1.0 m, which could represent a combination of obstructions within the barrel and
floating debris at the inlet. This is a highly simplified representation, but useful for an
indicative analysis within the CES-AES.
The impacts of the two culvert condition scenarios are shown in Figure 4.66,
which plots rating curves derived from CES-AES backwater profiles for a position

242 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Figure 4.64 Water surface profiles for trial design for design flows with return periods of 10, 50 and
100 years (Q10, Q50 and Q100, respectively).

Figure 4.65 Possible blockage scenario represented by a reduction in culvert barrel dimensions.

50 m upstream of the culvert entrance. It can be seen that the modelled scenario
for poor barrel condition (increased friction) causes an increased afflux compared to
the original trial design, but that this effect diminishes once the structure becomes
drowned. This is because the change in barrel friction is less significant once weir flow
ensues over the road level. However, the blockage case causes a large increase in water
levels for all design flows as a result of the much greater total obstruction to flow in
the channel.
Note that the example presents a simplified approach and should not be relied
upon as a basis for design calculations. It is intended to demonstrate use of CES-AES

Practical issues roughness, conveyance and afux 243

Figure 4.66 Flow rating curves for a position 50 m upstream of the culvert entrance showing two
culvert condition scenarios.

as a simple tool to aid understanding of a proposed culvert structure. More detailed

calculations are required for a design study.

4.6 DEALING WITH VEGETATION AND MAINTENANCE

OF WEEDY RIVERS
Channels provide an important habitat for plants, which in turn provide a range of useful functions such as creating sheltered areas with reduced velocities for fauna, altering
the temperature, light penetration and oxygen concentration to promote a variety of
species and encouraging siltation. These plants are a vital source of shelter and food
for fish, invertebrates and some birds. Where water spills overbank, the drying and
wetting of the berms promotes vegetation growth of wetland plants. Vegetation also
plays an important role in preventing scour and protecting bed and banks through the
binding action of the roots. Channel maintenance is, therefore, multi-faceted, incorporating a range of requirements such as providing sufcient capacity to convey flood
flows; reducing the seasonal cutting requirements (and hence expenses) where possible; utilising the bank and scour protection function of the vegetation and promoting
the natural habitat.
Guidance on channel management provides current best practice for vegetation
cutting (e.g. EA, 1998a&b; EA, 2003; CAPM, 1997; Buisson et al., 2008). This
typically involves clearing the vegetation in part of the river, for example, cutting

244 Practical Channel Hydraulics: Roughness, Conveyance and Afux

along the channel centre or one of the banks. These variations in pattern give rise to
spatial variations in channel roughness, and the roughness is also linked to the density,
type, height and stiffness of the vegetation (Fisher, 2001; Defra/EA, 2003b).
There are a number of existing methods for evaluating the reduced conveyance
due to the presence of vegetation (e.g. Cowan, 1956; Whitehead et al., 1992; Gordon
et al., 1992; Fisher, 2001). These include approaches such as reducing the crosssection or bankside Manning n value according to a simple rule, adapting the Manning
equation to account for a reduced flow area due to blockages as well as more complex approaches which aim to relate plant resistance to the biomechanical properties
of plants such as form, stiffness, shape and drag (e.g. Petryk & Bosmajian, 1975;
Kouwen & Li, 1980; Naden et al., 2006) or to velocity and depth (e.g. Palmer, 1945;
Ree & Palmer, 1949; Garton & Green, 1983; Naden et al., 2004; Statzner et al.,
2006). In most cases, the main channel and oodplain sections are treated as single
units, and the output is an average velocity for each region. As the capacity for aquatic
vegetation to support life may be quantied (e.g. invertebrates, Wright et al., 2002),
it is possible to estimate the impact to populations of improved conveyance through
habitat loss.
The CES conveyance methodology provides scope to describe the local variations
in vegetation roughness and it resolves the local depth-averaged velocities and flow
depths with due consideration of lateral shearing and boundary layer development.
This information can be used to advise on optimum cutting regimes, vegetation
patterns, plant types and percentage cut and provides increased condence in the
prediction of the overall channel ow capacity. This is illustrated through examples:

River Cole in Birmingham exploration of different cutting regimes;

River Avon in Enford exploration of different cutting regimes to satisfy both
ecological and flood risk objectives;
River Hooke in Maiden Newton exploration of channel deepening.

4.6.1 Exploration of cutting regimes

for the River Cole
Aim
The River Cole site is used to explore the impact on flow and depth-averaged velocity
for different vegetation cutting regimes at a flow depth of 1.4 m. These include:

Scenario 1: no cutting i.e. vegetation across the entire channel;

Scenario 2: vegetation cut in the channel centre (50% of existing vegetation at
cross-section location in Summer);
Scenario 3: vegetation cut on the right bank (70% of existing vegetation at crosssection location in Summer); and
Scenario 4: vegetation cut on both banks (70% of existing vegetation at crosssection location in Summer).

A comparison is made with the results produced using one of the existing
best practice approaches (EA, 1997), the HR Wallingford (Whitehead et al., 1992)
method.

Practical issues roughness, conveyance and afux 245

HR Wallingford approach
The empirically derived HR Wallingford (HRW) approach is for inbank flow only. It
involves evaluating an increased Manning n value, ntotal , to allow for the additional
retardance due to vegetation in the channel. This is evaluated from:

ntotal = nclear + 0.0239

Kw
UR


(4.3)

where nclear is the Manning n value when the channel is clear, Kw is the fraction of the
surface area covered by vegetation and U is the average velocity of the clear section.
This ntotal is then used in the Manning equation to evaluate the flow capacity and
velocity with vegetation.
The River Cole site
The River Cole in Birmingham is a straight trapezoidal channel, which was over
widened for flood control purposes. It consists of a mainly gravel bed with some silt
especially amongst the emergent vegetation. In the summer, it has submerged and oating channel vegetation in the main channel and emergent vegetation on both banks.
The emergent vegetation traps silt and creates berms, which cause self-meandering
within the channel in the Summer. Pool rife sequences create some variation in
depth along the channel and urban-debris is common (Figure 4.67). The average water
surface slope is 0.00174.

Figure 4.67 The River Cole site in Birmingham showing the channel vegetation, pools and riffles and
urban trash (Defra/EA, 2004).

246 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Roughness and cross-section def inition

The Roughness Advisor values for the as is case i.e. Scenario 1 are as follows
(Figure 4.68):

Channel bed The RA coarse gravel 2064 mm (nsur = 0.027), submerged fineleaved plants (medium) (nveg = 0.1) and pools (nirr = 0.2) are adopted giving
nl = 0.106.
Channel banks The RA coarse gravel 2064 mm (nsur = 0.027) and emergent
reeds (nveg = 0.15 but taken as 0.09 average value as do not appear very thick)
are adopted giving nl = 0.094.
Floodplain The RA sand (nsur = 0.02) and height-varying grass (nveg = 0.041)
are adopted giving nl = 0.046.

Results for the River Cole site

Scenario 1 is essentially the as is case i.e. if no vegetation cutting takes place.
Figure 4.69 provides the CES predicted stage-discharge and the measured data. There
are only a few low flow measurements and one high flow (16.7 m3 s1 @ 1.49 m)
measurement available. The curve does not capture the local variations of these measurements, but it does pass from the low flows to the high flows. The upper and
lower uncertainty scenarios are wide, which is indicative of the uncertainty associated
with the submerged and emergent vegetation roughness information. Figure 4.70 provides the back-calculated Manning n values and the corresponding Manning n values
from the measured flows. Although the detail is not captured, the trend of the CES
curve moves from lower Manning n values at high flows to higher Manning n values
at low flows.
The CES flow predictions for scenarios 1 to 4 at a depth of 1.4 m are summarised
in Table 4.6 and the corresponding depth-averaged velocity profiles are shown in
Figure 4.71.
For the HRW method, Kw is taken as 0.70, 0.36, 0.52 and 0.34 for scenarios 1
to 4 respectively. This is derived from the percentage of the cross-section covered by
vegetation for each case at a flow depth of 1.4 m. The nclear is taken as 0.061, giving

Figure 4.68 River Cole cross-section, roughness zones and top-of-bank marker locations.

Practical issues roughness, conveyance and afux 247

1.6
1.4

Depth (m)

1.2
1
0.8
0.6
CES prediction

0.4

Data
0.2

CES Upper & Lower

0
0

10

12

14

16

18

Flow (m3s-1)
Figure 4.69 Stage-discharge prediction for the River Cole with no channel cutting and comparison to
data (data from Defra/EA, 2004).

1.6
CES prediction

1.4

Data
1.2

Depth (m)

CES Upper & Lower

1
0.8
0.6
0.4
0.2
0
0

0.05

0.1

0.15

0.2

0.25

0.3

Manning n
Figure 4.70 Back-calculated Manning n values for the River Cole and comparison to data (data from
Defra/EA, 2004).

248 Practical Channel Hydraulics: Roughness, Conveyance and Afux

Table 4.6 Summary of flow predictions using the CES and HRW approach at depth 1.4 m.
Percentage
difference relative to
Scenario 1 (%)

Flow
(m3 s-1 )
Vegetation cutting scenario

CES

HRW

CES

HRW

Percentage
difference in CES
HRW value (%)

Scenario 1: no cutting
Scenario 2: main channel cut
Scenario 3: right bank cut
Scenario 4: right left bank cut

11.4
12.8
12.1
12.8

11.4
16.5
12.6
14.2

0
44
10
24

0
12
6
13

0
23
4
9

2.5

3.0

HRW Scenario 1: Q = 11.4

HRW Scenario 2: Q = 12.8
HRW Scenario 3: Q = 12.1

2.5

HRW Scenario 4: Q = 12.8

2.0

CES Scenario 1: Q = 11.4

CES Scenario 2: Q = 16.5

Velocity (ms-1)

CES Scenario 4: Q = 14.2

1.5

Cross-section geometry
1.5

1.0

Elevation (m)

2.0

CES Scenario 3: Q = 12.6

1.0

0.5
0.5

0.0

0.0
0

10

12

14

16

Lateral distance across channel (m)

Figure 4.71 Depth-averaged velocity predictions for the River Cole using the CES and the HRW
approach.

ntotal as 0.078, selected to ensure the flow rate for the HRW method and the CES
method are identical for Scenario 1, the as is case, i.e. QHRW = QCES = 11.4 m3 s1 .
The HRW flow rates are provided in Table 4.4 and the average velocities are shown
in Figure 4.71.
The following is observed:

Vegetation type the CES enables a user to select the vegetation roughness associated with the particular plant species (e.g. emergent reeds), including the time of
year (e.g. June) and upper and lower scenarios providing improved condence
in the outputs. Existing methods such as HRW method do not relate the Manning
n values to vegetation morphotypes.

Practical issues roughness, conveyance and afux 249

Vegetation location The CES enables users to identify the exact location in
the cross-section where the vegetation is present e.g. banks, main channel. This
information is used to evaluate the local unit flow rate and velocities (Figure 4.71)
across the section providing improved confidence in the final overall flow rate.
The HRW approach assumes a single Mannings n value for the entire cross-section
regardless of different vegetation types and locations within the section and hence
average velocities are provided (Figure 4.71). The plan form cover is accounted
for through a simple coefficient, Kw .
Vegetation cutting, location, nature & timing The CES enables users to select
the precise location where vegetation is cut, the nature of the cutting (e.g. cut 70%
of the vegetation at the location) and the timing of the cut (e.g. June) which inuences the resulting unit roughness value. This additional detail provides improved
condence in the results.
Flow predictions The CES predicted flow capacities are up to 20% larger
than those calculated using the HRW method and they provide a clear hierarchy of the cutting regimes with a main channel cut as the most favourable in this
case (44% increased flow capacity) followed by cutting on both channel banks
(24% increased flow capacity). This information can be considered together with
associated costs to ascertain the preferred option.
Velocity predictions The CES approach provides information on the local velocity variations in the cross-section which is not provided by the more traditional
approaches. This information may be used to inform, for example, habitat design,
risk to people inclusive during maintenance activities, scour potential.

Further analyses for the River Cole may involve taking the optimum cut, channel
bed, and determining the optimum time of year for the cut (current results pertain to
a June cut).
To prove the potential added value of the CES outputs outlined above, it is recommended that this information is used in practice to inform a vegetation cutting regime,
and the flow capacity should be monitored and compared to that achieved through
previous practice.

4.6.2 Exploration of different cutting regimes

for the River Avon
Aim
Different cutting regimes are explored for the River Avon to determine the improvement in conveyance with vegetation removal and the potential impact on invertebrate
population (Wright et al., 2002). This demonstrates how the trade-off between flood
risk management (e.g. greater flow and reduced levels) and ecological (e.g. increased
habitat and hence life supported) objectives may potentially be explored.
The River Avon site, roughness and cross-section def inition
The Avon comprises a simple channel, 9.2 m wide, with a bankfull depth of 0.6 m
and an average bed slope of 0.0005. The bed consists of gravels and the in-channel

250 Practical Channel Hydraulics: Roughness, Conveyance and Afux

section has an abundance of water crowfoot (Ranunculus pencillatus pseudouitans)

vegetation (Figure 4.72), a submerged fine-leafed plant. The assigned Roughness
Advisor values are (Figure 4.73):

Gravel The RA gravel 720 mm (nsur = 0.025) is adopted.

Ranunculus & gravel The RA gravel 720 mm (nsur = 0.025) and submerged
fine-leaved plants (nveg = 0.10) are adopted giving nl = 0.103.

Results for the River Avon

Discharge predictions (Figure 4.74) are made for 0, 10, 20, 30, 40, 50, 60, 70, 80,
90 and 100% vegetation coverage, with the gravel coverage at the channel centre
for each. The dashed lines show the uncertainty scenarios, which are increasing for
increasing vegetation coverage, as this is the primary source of uncertainty in the
roughness values. The number of invertebrates supported by the vegetation cover is
shown. These are based on densities of Ranunculus pseudouitans (Wright et al.,
2002) and are calculated for a 20 m reach. Combining the flow, invertebrate and
vegetation information enables different what ifs to be explored. For example, if
the vegetation is cut-back from the existing 42% cover to say, 20%, this provides a
25% increase in bankfull flow capacity but this is at the expense of a 30,000 drop
in total invertebrate life supported. These values are, of course, indicative and other
inuences should also be considered, for example, there may be a further trade-off

Figure 4.72 River Avon site at Enford, Pewsey, Wiltshire (OHare et al., 2008; Courtesy Centre for
Ecology and Hydrology).

Practical issues roughness, conveyance and afux 251

Figure 4.73 River Avon cross-section, roughness zones and top-of-bank marker locations for 50%
vegetation present (survey data OHare et al., 2008).

350,000

2.5

250,000
2
200,000
1.5
150,000
1
100,000

CES flow rate (m3s-1)

Total invertebrates supported

300,000

0.5

50,000

0
0

10

20

30

40

50

60

70

80

90

0
100

% Vegetation cover
Figure 4.74 Change in number of invertebrates supported and total flow rate with percentage
vegetation cover (Mc Gahey et al., 2008).

between vegetation abundance and species diversity. This information can be further enriched through, for example, exploration of different cutting options such as
removal of growth on a particular bank and the impact on water levels and flow
capacity.

252 Practical Channel Hydraulics: Roughness, Conveyance and Afux

4.6.3 Exploration of channel deepening

for the River Hooke
Aim
The River Hooke is used to explore the impact of dredging on local flow depths
and depth-averaged velocity proles, assuming no downstream controls. The existing
channel has gravel and water crowfoot scattered throughout the section. The maintenance option will involve removal of the water crowfoot and deepening of the channel
section along the central margin.
The River Hooke site, roughness and cross-section def inition
The site is found in the village of Maiden Newton, in the Frome catchment, directly
upstream of a LOCAR monitoring point, where discharge and other environmental parameters are recorded at high resolution. The channel vegetation is dominated
by water crowfoot (Ranunculus pencillatus pseudouitans, Figure 4.75). Directly
upstream of the sampling area is a vegetated mid-channel bar, seen on the left of
the photograph. The site has been subject to ooding in the past and a flood relief
channel has been built just upstream of the mid-channel bar. Water level, velocity,

Figure 4.75 River Hooke site at Maiden Newton (OHare et al., 2008; Courtesy Centre for Ecology
and Hydrology).

Practical issues roughness, conveyance and afux 253

vegetation and roughness measurements are available. The water level slope is 0.0108
at flow depths of 0.32 m.
The as is assigned roughness values are:

Substrate only The RA gravel 720 mm (nsur = 0.025)

Shallow with vegetation The RA gravel 720 mm (nsur = 0.025) and June value
for submerged water crowfoot or fine-leaved plants in shallow water 0.20.6 m
(nveg = 0.25) are adopted giving nl = 0.251.

Results for the River Hooke

Figure 4.76 shows the measured and predicted depth-averaged velocities for the River
Hooke. There is much scatter amongst the data, which is not unexpected considering
the scattered nature of the vegetation. In the channel centre (chainage 2.5 m5 m) the
measured velocities are low reflecting the influence of the upstream mid-channel bar.
The CES velocity profile may be improved if the precise location of the vegetation is
defined within the CES cross-section. For this case, the average vegetation values for
June in shallow (0.2 m0.6 m) flow are adopted, and the simulated flow rate is within
2% of the measured data.
The cross-section is then altered to simulate channel deepening (Figure 4.77) with
only gravel present in the deepened section. The resulting CES predicted flow capacity

0.8

Channel bed
1.8

0.7

Measured velocity data

1.6

CES velocity prediction

0.6

0.5

1.2

0.4

1
0.8

0.3

Velocity (ms-1)

Elevation (m)

1.4

0.6
0.2
0.4
0.1

0.2
0

0
0

Distance across channel (m)

Figure 4.76 Depth-averaged velocity predictions for the River Hooke at flow depth 0.32 m giving a
QCES = 0.42 m3 s1 and Qdata = 0.41 m3 s1 (data OHare et al., 2008).

254 Practical Channel Hydraulics: Roughness, Conveyance and Afux

0.8

Channel bed
1.8

Deepended channel bed

0.7

CES velocity deepened channel

1.6

CES velocity prediction

0.6

0.5

1.2
1

0.4

0.8

0.3

Velocity (ms-1)

Elevation (m)

1.4

0.6
0.2
0.4
0.1

0.2
0

0
0

Distance across channel (m)

Figure 4.77 Depth-averaged velocity predictions for the River Hooke at flow depth 0.32 m before
and after channel deepening and vegetation removal.

at a depth of 0.32 m is 1.8 m3 s1 , a 330% increase on the CES flow prediction for
the as is channel section. The velocities in the left-hand margin are, as expected,
substantially higher reaching 0.7 ms1 . These velocities may be used to inform scour
potential and habitat change for the reach.