1 CES UserGuide PDF

Defra / Environment Agency
Flood and Coastal Defence R&D Programme
Reducing uncertainty in river flood conveyance, Phase 2
Conveyance Manual
Project Record W5A-057/PR/1

Defra / Environment Agency
Flood and Coastal Defence R&D Programme
REDUCING UNCERTAINTY IN RIVER FLOOD

CONVEYANCE, PHASE 2
Conveyance Manual
Project Record W5A-057/PR/1
Research Contractor: HR Wallingford

Publishing organisation
Environment Agency, Rio House, Waterside Drive, Aztec West, Bristol BS32 4UD
Tel: +44 (0)1454 624400 Fax: +44 (0)1454 624409 Web: www.environment-agency.gov.uk
© Environment Agency September 2004 Product Code: SCHO0904BIHR-E-E
ISBN: 1844323048
The Environment Agency will waive its normal copyright restrictions, and allow this document
(or other item), excluding the logo to be reproduced free of licence or royalty charges in any
form, provided that it is reproduced unaltered in its entirety and its source acknowledged as
Environment Agency copyright.
This waiver is limited to this document (or other item) and is not applicable to any other
Environment Agency copyright material, unless specifically stated. The Environment Agency
accepts no responsibility whatever for the appropriateness of any intended usage of the
document, or for any conclusions formed as a result of its amalgamation or association with any
other material.
The views expressed in this document are not necessarily those of Defra or the Environment
Agency. Its officers, servants or agents accept no liability whatsoever for any loss or damage
arising from the interpretation or use of the information, or reliance on views contained herein.
Dissemination Status
Internal: Released to Regions and Areas
External: Released to Public Domain
Statement of use
This document provides support and training for users of the Conveyance Estimation System
software, the core output for R&D Project W5A-057.
Keywords
Conveyance principles, river flow, numerical model, software training, worked examples
Contract Details
The lead funder for this collaborative project was the Defra / EA Joint Flood and Coastal
Defence R&D Programme. Scottish Executive, Rivers Agency – Northern Ireland, and the
Natural Environment Research Council also contributed funding. Wallingford Software
supported the development of new software.
Research Contractor
This document was produced under R&D Project W5A-057 by
HR Wallingford Ltd, Howbery Park, Wallingford, Oxon OX10 8BA, supported by an expert
advisory group (Professors Garry Pender, Alan Ervine and Donald Knight plus Dr Chris
Whitlow).
Tel: +44 (0)1491 835381 Fax: +44 (0) 1491 832233 Web: www.hrwallingford.co.uk.
Contractor’s Project Manager: Manuela Escarameia
Environment Agency's Project Manager

The Environment Agency's Project Manager: Dr Mervyn Bramley, Engineering Theme Leader
This report is only available in electronic form over the Defra / Environment Agency
webpages for the Joint Flood and Coastal Erosion Risk Management R&D Programme.
R&D OUTPUTS: CONVEYANCE USER MANUAL

- ii -
ABBREVIATIONS
CES Conveyance Estimation System
CG Conveyance Generator
DCM Divided Channel Method
DEFRA Department for Environment, Food and Rural Affairs
EA Environment Agency
FCF Flood Channel Facility
IDB Internal Drainage Boards
LiDar Airborne scanning Laser Altimetry
MDSF Modelling and Decision Support Framework
OS Ordnance Survey
RA Roughness Advisor
RANS Reynolds-Averaged Navier-Stokes
RASP Risk Assessment of Flood and Coastal Defence Systems for Strategic
Planning
RHS River Habitat Survey
SAR Sonar Acoustique Remorque
SP Strategic Programme
TP Targeted Programme
UE Uncertainty Estimator
UK United Kingdom
1D One Dimensional

- iii -
- iv -
SYMBOLS
Latin Alphabet
C Chezy coefficient (m0.5.s-1)

Cuv coefficient for sinuosity
Dr relative depth = h/Hmax (m.m-1)
E specific energy (m)
f local boundary friction factor based on the local substrate/vegetation roughness
g gravitational acceleration (m.s-2)
H local water depth normal to the bed (m)
Hmax maximum depth for a given channel section (m)
h water surface level (m)
Dhc uncertainty in water level due to calibration
Dhq uncertainty in water level due to hydrology
Dhr uncertainty in water level due to roughness
Dhs uncertainty in water level due to survey data
Dhu total uncertainty in water level
DhDx uncertainty in water level due to numerical considerations e.g. grid resolution
K cross-sectional channel conveyance (m3.s-1)
k unit conveyance (m2.s-1)
ks absolute roughness height (m)
n Manning’s roughness coefficient (s.m-1/3)
ne engineering ‘n’ which is all encompassing (s.m-1/3)
nirr unit roughness component due to irregularities (s.m-1/3)
nl local ‘n’ includes local boundary friction losses only (s.m-1/3)
nsur unit roughness component due to surface material (s.m-1/3)
nveg unit roughness component due to vegetation (s.m-1/3)
Q total cross-section discharge (m3.s-1)
q unit flow rate (m2.s-1)
S common uniform gradient
Sf streamwise friction or ‘specific energy’ gradient
So reach-averaged longitudinal bed slope
T fluid temperature (°C)
u streamwise point velocity (m.s-1)
U section average velocity (m.s-1) in Section 2.3.1
Ud depth-averaged streamwise velocity (m.s-1) (RANS approach)
U* shear velocity (m.s-1)
U streamwise velocity component at a given depth (m.s-1)
Vd depth-averaged lateral velocity (m.s-1) (RANS approach)
V lateral velocity component at a given depth (m.s-1))
x streamwise direction parallel to the bed (m)
y lateral distance across section (m) (RANS approach)
z measured perpendicular to the channel bed (m)
Greek alphabet
b coefficient for the influence of lateral and longitudinal bed slope on the bed
shear stress

-v-
G secondary flow term for straight prismatic channels
Gfp secondary flow term for floodplain flows
Gmci secondary flow term for main channel inbank flows
Gmco secondary flow term for main channel out-of-bank flows
l dimensionless eddy viscosity
lmc main channel dimensionless eddy viscosity
n kinematic viscosity of the fluid (m2.s-1)
s sinuosity - thalweg length over the valley length (m.m-1)
r fluid density (kg.m-3)
tb bed shear stress (N.m-2)
c relaxation factor

- vi -
GLOSSARY OF TERMS
Accuracy: Accuracy has two facets. It deals with the precision to which
measurement or calculation is carried out (i.e. the maximum
deviation of the measurement or calculation from the “true”
amount) and with the number of digits of precision which are
carried through a calculation (whether or not these are
meaningful). Potentially, accuracy can be improved by
better technology.
Backwater: Backwater effects occur when sub-critical flow is controlled

by the downstream conditions e.g. presence of an outfall,
bridge constriction, dam etc.
Bankfull: The maximum channel discharge capacity; further discharge

spreads onto the floodplains.
Berms: (i) The space left between the upper edge of a cut and toe of
an embankment to break the continuity of an otherwise long
slope.
(ii) The sharp definitive edge of a dredged channel such as in
a rock cut.
(iii) Natural levee where river deposits sediment
Braided channel: A channel that is divided into two or more channels e.g. flow
around an island.
Conveyance: Channel conveyance is a measure of the discharge carrying

Q
capacity of a channel K m3/s, defined as K = 1
Sf2
Eddy viscosity: Parameter representing the momentum exchange, e (m2/s),
defined in the CES methodology (DEFRA/EA, 2003a) as e
= l U * D which relates viscosity to the bed shear stresses.
Dimensionless eddy Calibration coefficient to account for lateral shear l

viscosity:
Discharge: The volume of water that passes through a channel section

per unit time.
Error: Errors are mistaken calculations or measurements with

quantifiable and predictable differences from the actual
value.
Flow: A general term for the movement of volumes of water at a

speed.
Flow zone: Area in a channel cross-section within which the flow

characteristics represent a particular flow mechanism or

- vii -
selection of mechanisms e.g. secondary flow, shearing flow
etc. (This is different to a roughness zone.)
Irregularity: Channel irregularities represent variations in roughness from
obstructions such as exposed boulders, trash, groynes etc or
channel shape e.g. pools and riffles.
Manning’s ne: An empirical coefficient ne that lumps together all losses i.e.
losses due to local friction, secondary flows, form losses and
lateral shear stresses (Chow, 1959). Derived for use in
medium to large non-vegetated rivers with fully developed
flow profiles.
Manning’s pure nl: A coefficient that approximates the local friction losses due
to bed features only i.e. surface material, vegetation and / or
irregularities.
Maximum depth: The vertical distance of the lowest point of a channel section
from the free water surface
Morpho-types: See “vegetation morpho-types”
Multi-thread channel: A channel that is divided into two or more channels e.g. flow
around an island.
Offset: Distance measured laterally across the channel, taken from

the left-hand side when looking downstream.
RANS: The Navier-Stokes equations are derived from the governing

equations of fluid flow. These equations are termed
“Reynolds Averaged” as the instantaneous turbulent
fluctuations of the velocity and pressure components have
been averaged over small time-scales i.e. small compared
with variations in the bulk flow that are of any practical
importance, but large compared with inter-particle collisions.
The “Reynolds Averaged” Navier-Stokes equations are
100% based on modelling of the physical properties.
Resistance: As for roughness but defined as flow-, form-, frictional or

turbulent etc
Reynolds Number: The Reynolds Number describes the ratio of the inertia
forces to the viscous forces (Re = 4Rv/n and R is the
hydraulic radius). In open channel flow, for Re < 500
laminar flow occurs, Re > 1000, turbulent flow occurs and
for 500 < Re < 1000 transitional flow occurs, i.e. the flow is
characterised by both laminar and turbulent effects.
Riffle: A point at which the stream is relatively energetic due to a

constriction or steep gradient (the counterpart of pool).

- viii -
Roughness: The effect of impeding the normal water flow of a channel
by the presence of a natural or artificial body or bodies,
biotic e.g. vegetation; abiotic / mineral e.g. bank, bed
substrate.
Roughness A roughness component is a category of unit roughness that

components: may be vegetation, surface material or irregularity. All unit
roughness values fall into one of these three components.
Roughness types: The roughness type refers to the portion of the cross-section
where a particular surface cover is likely to be found e.g.
crops on the floodplain. There are three roughness types:
bed, bank and floodplain.
Roughness zones: A roughness zone is a plan area in the river/floodplain

system that is characterised by similar roughness e.g. a field,
bank vegetation, crops with clay below etc. A roughness
zone can only be comprised of three roughness components.
Sinuosity: A measure of a channel’s tendency to meander defined in the

CES as thalweg length over valley length.
Stage: The vertical distance of the free surface from an arbitrary or

defined datum
Study reach: The length of a study reach OR a section under investigation/

measurement
Substrate: A generic term for a substance that underlies another; soil is

the substrate for plants, while bedrock is the substrate for
soil.
Surface material: Surface material encompasses the substrate on the bed, bank
and floodplains. The roughness due to surface material
includes for e.g. sand, gravel, peat, rock etc.
Thalweg: A line connecting the deepest points along a channel length
Top-of-bank bend The top-of-bank bend markers are top-of-bank markers for
markers: meandering channels. They are used to determine the
orientation of the bend through specifying them as inside or
outside of bend.
Top-of-bank markers: Top-of-bank markers are used to incorporate the changes in

secondary flow mechanisms at bankfull into the conveyance
calculation. These are placed on the cross-section at the
intersection between the main channel and floodplain.
Turbulence: Turbulent flow occurs at high Reynolds Number (Re > 1000
for open channel flow) when the flow incorporates an

- ix -
eddying or mixing action, and the inertia forces dominate the
viscous forces. A chaotic and random state of motion
develops in which the velocity and pressure changes
continuously with time, within substantial regions of flow.
The fluid particles are therefore continuously interchanging
and it follows that the fluid properties may be exchanged.
The crucial difference between visualisations of laminar and
turbulent flows is the appearance of eddies on a wide range
of length scales in the turbulent flows.
Uncertainty: Uncertainty comprises (i) natural variability and (ii)

knowledge uncertainty. It arises principally from lack of
knowledge or that of ability to measure or to calculate which
gives rise to potential differences between assessment of
some factor and its “true” value.
Unit conveyance: Conveyance per metre width of channel k (m2/s), defined as
q
k= 1 .
Sf2
Unit discharge: Flow rate per metre width of channel section q (m2/s),
H
defined as q = ò u dz .
0
Unit roughness: Roughness due to an identifiable segment of boundary

friction per unit length of channel.
Vegetation Morpho- These are aquatic and marginal plant species of similar form
types: or function in the riparian corridor. These are not vegetation
types with similar morphological character.

-x-
SUMMARY
Under their joint R&D programme for Flood and Coastal Defence, Defra/Environment
Agency have funded a Targeted Programme of research aimed at obtaining better
predictions of flood water levels. In order to achieve this, advances in knowledge and
understanding made over the past three or four decades in the estimation of river
conveyance are being introduced into engineering practice.
The project relates particularly to water level estimation, leading to a reduction in the
uncertainty in the prediction of flood levels and hence in flood risk, and consequently
facilitating better targeting of expenditure. The project will equally benefit the targeting
of maintenance by providing better estimates of the effects of vegetation and its
management. It is expected that the application of this knowledge from UK engineering
research will have an international impact through improving the methods available to
consultants.
The above objectives will be achieved through three core components of the new
Conveyance Estimation System (CES): the Conveyance Generator, the Roughness
Advisor and the Uncertainty Estimator. A Backwater Module is also available. The
CES is designed so that new knowledge from a parallel Strategic Programme of
research can be integrated into the CES in due course.
The Conveyance Manual provides a guide for the Conveyance Estimation System
software, covering the technical background, user application and providing worked
examples. The Manual is intended to assist the user in the hands-on application of the
software, providing advice where subjective decisions are required and assisting in
interpretation of results. It includes an additional section on the advanced user options
and the Backwater Module.

- xi -
- xii -
CONTENTS
ABBREVIATIONS iii
SYMBOLS v
GLOSSARY OF TERMS vii
SUMMARY iii
1. Introduction 1
1.1 Why do we need a new conveyance system? 1
1.2 What is the CES? 1
1.3 Why Provide a Conveyance Manual? 2
1.4 How do I use the Conveyance Manual effectively? 2
2. Technical Background 3
2.1 Overview 3
2.1.1 What are legacy practices in conveyance estimation? 3
2.1.2 What changes are expected after the CES is recognised as the
industry norm? 3
2.1.3 What will I do that I did not do in legacy practices? 4
2.1.4 What will the CES do? 4
2.1.5 What new opportunities are provided by the CES? 4
2.2 Roughness 5
2.2.1 Why do we need to worry about roughness? 5
2.2.2 Why is there a need for a new approach to estimating roughness? 5
2.2.3 How is the roughness represented? 6
2.2.4 What is a unit roughness nl-value? 6
2.2.5 Why was the decision taken to express roughness in terms of ‘n’? 6
2.2.6 What is a roughness component? 6
2.2.7 How are the component roughness values combined? 6
2.3 Conveyance 7
2.3.1 What is conveyance? 7
2.3.2 Why do we need to know about channel conveyance? 8
2.3.3 Why is there a new approach for estimating channel conveyance? 8
2.3.4 Do I need to have a strong mathematical background to use the CES? 9
2.3.5 What are the different energy loss mechanisms? 9
2.3.6 What is the selected approach in the CES for calculating conveyance? 11
2.3.7 Why was the RANS approach selected? 15
2.3.8 What are the limitations of this method? 16
2.3.9 How are the RANS equations solved? 17
2.3.10 How is my cross-section discretised? 17
2.3.11 Has this approach been tested? 17
2.4 Uncertainty 18
2.4.1 What is uncertainty? 18
2.4.2 Why do I need to account for uncertainty? 18
2.4.3 What factors contribute to the uncertainty in conveyance? 19
2.4.4 What approaches for estimating uncertainty were considered? 20

- xiii -
2.4.5 How is uncertainty represented in the CES? 23
2.4.6 How is the uncertainty calculated in the CES? 25
2.4.7 What are the dominant factors that are considered? 26
2.4.8 Why was this approach selected? 26
2.4.9 What is the effect of calibration? 27
2.5 Backwater 27
2.5.1 What is a backwater profile? 27
2.5.2 Why do I need to know about backwater effects? 27
2.5.3 How is the backwater calculated? 28
2.5.4 What are the Backwater method restrictions? 29
3. Using the CES 31

3.1 Roughness advisor 31
3.1.1 What is the Roughness Advisor? 31
3.1.2 What will the Roughness Advisor provide? 31
3.1.3 What is a roughness zone type? 31
3.1.4 What is a roughness zone? 31
3.1.5 Why do I specify roughness zones? 32
3.1.6 How do I get the roughness values in the Roughness Advisor? 32
3.1.7 Can I put my own roughness values in? 32
3.1.8 Are there minimum and maximum allowable unit roughness values? 33
3.1.9 What are the seasonal variations in vegetation? 33
3.1.10 How are the cutting regimes used? 33
3.2 Conveyance Generator 33
3.2.1 How do I describe the cross-section geometry? 33
3.2.2 How do I represent the roughness zones in my cross-section
geometry? 34
3.2.3 Does the channel cross-section represent the start or mid-point of the
channel reach? 35
3.2.4 What are the top-of-bank markers used for? 35
3.2.5 Do I have to use top-of-bank markers? 36
3.2.6 Where should top-of-bank markers be positioned? 36
3.2.7 How do I model meandering channels? 38
3.2.8 How do I label top-of-bank bend markers on skewed channels? 39
3.2.9 How do I calculate the channel sinuosity? 40
3.2.10 What are typical UK channel sinuosities? 40
3.2.11 How do I model braided channels? 42
3.2.12 How do I model braided channels with different sinuosity values? 43
3.2.13 How do I model non-conveyance storage zones i.e. zones with re-
circulation or stationary water? 44
3.2.14 What is the range of depths for which the conveyance is calculated? 44
3.2.15 Are there any special survey requirements for application of the
CES? 44
3.2.16 What slope do I use? 45
3.3 Uncertainty Estimator 45
3.3.1 How is uncertainty represented in the CES? 45
3.3.2 How should this uncertainty information be interpreted? 45
3.3.3 How should this uncertainty information be used? 46
3.3.4 How ‘certain’ is the uncertainty calculation? 47

- xiv -
3.3.5 How do I ensure the uncertainty is calculated? 47
3.4 Calibration and interpretation of results 48
3.4.1 How do I calibrate my model and what do I change when the data
does not fit? 48
3.4.2 Why do I get jumps in the conveyance curve at sudden geometry
changes e.g. bankfull? 48
3.4.3 Why do I sometimes get large jumps at bankfull depth for high
sinuosities? 49
3.4.4 Why are the uncertainty curves not symmetrical about the expected
value? 50
3.4.5 Why is the uncertainty greater at small depths? 50
3.5 Advanced Options 50
3.5.1 Do I need to alter the advanced options? 50
3.5.2 What effect will increasing the number of depths have? 51
3.5.3 What effect will increasing the number of lateral divisions have? 51
3.5.4 What happens if I change the dimensionless eddy viscosity? 52
3.5.5 What happens if I change the straight secondary flow coefficients? 52
3.5.6 What happens if I alter the temperature? 52
3.5.7 What happens if I alter the relaxation parameter? 52
3.5.8 If I alter an advanced parameter, does the change apply to that cross-
section only? 53
3.5.9 What happens if I set the minimum depth? 53
3.5.10 Why would I set the minimum depth? 53
3.5.11 What does the wall height multiplier do? 53
3.5.12 What happens if I alter the maximum number of iterations? 53
3.5.13 What does the experimental flume option change? 53
3.5.14 What does the convergence tolerance change? 53
3.6 Backwater Module 54
3.6.1 How is the backwater module used? 54
3.6.2 What Backwater options are available? 54
3.6.3 What is the advantage of using the backwater module compared to a
complete 1D modelling software such as iSIS or MIKE 11? 54
4. Worked examples 55
4.1 Worked example of the River Main, Northern Ireland 55
4.1.1 Available user information 55
4.2 Using the Roughness Advisor to generate roughness zones 57
4.2.1 Using the Conveyance Generator to generate a rating curve 63
4.2.2 Calibrating process (depends on data availability) 69
4.2.3 Final result and interpretation of uncertainty 69
4.3 Worked example of the River Blackwater 69
4.3.2 Using the Roughness Advisor to generate roughness zones 72
4.4 Worked example of River Dane 81

- xv -
4.4.2 Using the Roughness Advisor to generate roughness zones 83
5. References 91
Tables
Table 2.1 Potential consequences of uncertainty in flood conveyance 19
Table 2.2 Summary of Uncertainty users and their needs 24
Table 2.3 Backwater lengths for some UK Rivers (Samuels, 1989) 28
Table 3.1 Use of uncertainty information for different user groups 46
Table 3.2 Allowable ranges for the advanced options 51
Table 4.1 River Main cross-section 14 geometry 56
Table 4.2 Final Roughness Advisor (.rad file) output for the River Main 63
Table 4.3 River Blackwater cross-section geometry 71
Table 4.4 Final Roughness Advisor (.rad file) output for the River Blackwater 76
Table 4.5 River Dane cross-section 30 Geometry 83
Figures
Figure 2.1 Example of components for substrate and vegetation contributing to
local boundary friction 7
Figure 2.2 Energy loss mechanisms in a straight compound channel for three
depths of flow 10
Figure 2.3 Flow mechanisms associated with straight overbank flow in a two-stage
channel (Shiono and Knight, 1991) 10
Figure 2.4 Flow mechanisms associated with in a two-stage meandering channel
(Ervine et al., 2001) 11
Figure 2.5 Contributions from the secondary flow terms with increasing sinuosity 14
Figure 2.6 Lateral distribution of calibration parameters for a typical two-stage
channel 15
Figure 2.7 Cross-section discretisation 17
Figure 2.8 Upper and lower uncertainty bands in the RASP analysis 21
Figure 2.9 The Monte Carlo approach 23
Figure 2.10 Representation of uncertainty for a rating curve 25
Figure 2.11 Effect of calibration 27
Figure 2.12 Backwater calculation method from cross-section 1 to 3 29

- xvi -
Figure 3.1 Plan view of roughness zones 32
Figure 3.2 Example of cross-sections that cannot be modelled within the
Conveyance Generator 34
Figure 3.3 Setting extreme / edge co-ordinates with maximum values as a
function of maximum depth 34
Figure 3.4 Cross-sections with the lateral distribution of roughness zones 35
Figure 3.5 Variation of secondary flow term for straight channels 35
Figure 3.6 Variation of secondary flow term for floodplains with different
elevations 36
Figure 3.7 Positioning of top-of-bank markers for single channels and isolated
floodplains 36
Figure 3.8 Positioning of top-of-bank markers for asymmetric channels 37
Figure 3.9 Positioning of top-of-bank markers for channels with berms 37
Figure 3.10 Top-of-bank markers for low elevation floodplains 38
Figure 3.11 Typical observed depth-averaged velocity profiles in meandering
channels and top-of bank bend marker positions 39
Figure 3.12 Top-of-bank markers for skewed channels 40
Figure 3.13 Definition sketch for sinuosity 40
Figure 3.14a Sinuosity values for (a) the Flood Channel Facility at HR Wallingford,
(b) the Glasgow University small-scale flume, (c) the River Dane in
Cheshire, (d) Kleinschmidt Creek (Photo courtesy of Land and Water
Consulting), (e) incised meanders, Utah and (b) Wood River, Southern
Nebraska 41
Figure 3.14b Plan view of channels with increasing sinuosity 42
Figure 3.15 Addition of flow in braided channels (DEFRA/EA, 2003a) 42
Figure 3.16 Top-of-bank markers for a braided channel 42
Figure 3.17 Top-of-bank marker positions for deep flow in a braided channel 43
Figure 3.18 Top-of-bank bend markers for multi-thread channels with different
sinuosities 43
Figure 3.19 Top-of-bank bend and straight markers for multi-thread channels with
different sinuosities 43
Figure 3.20 Non-conveyance zone specification – single units only 44
Figure 3.21 Unusual ‘jumps’ in stage-discharge curves reflect observed data e.g.
River Severn, Montford 49
Figure 3.22 Discontinuity in stage-discharge curves in meandering channels reflect
the observed data e.g. Flood channel Facility Phase B20/21 observations 50
Figure 4.1 River Main in County Antrim, Northern Ireland (a) the side-banks with
rip-rap and (b) a gentle meander 55
Figure 4.2 Plan view of the River Main indicating the selected cross-section 14 56

- xvii -
Figure 4.3 River Main cross-section 14 geometry 57
Figure 4.4 Create a new roughness file 57
Figure 4.5 Create three roughness zones: bed cover, bank cover and floodplain
cover 58
Figure 4.6 Providing roughness zone description: Coarse gravel 58
Figure 4.7 Three roughness component options 59
Figure 4.8 Bed material component with gravel selected 59
Figure 4.9 Bed material component now populated 60
Figure 4.10 Providing roughness zone description: Rip-rap 60
Figure 4.11 Bank material component with rip-rap selected 61
Figure 4.12 Providing roughness zone description: Heavy weed growth 61
Figure 4.13 Floodplain ground material component with height varying grass
selected 62
Figure 4.14 Floodplain ground material component with sand selected 62
Figure 4.15 Create a new conveyance file 63
Figure 4.16 Select River Main roughness file 63
Figure 4.17 Cross-section description 64
Figure 4.18 Select ‘properties’ for the now populated first cross-section 64
Figure 4.19 Cross-section data for cross-section 14 65
Figure 4.20 Roughness zone allocation 66
Figure 4.21 The River Main rating curve for Section 14 67
Figure 4.22 Comparison of the CES stage-discharge prediction to the measured data 67
Figure 4.23 Alter the number of calculation depths in the Advanced Options 68
Figure 4.24 Improved CES prediction from increasing the number of depths in
Advanced Options 68
Figure 4.25 Upper and lower uncertainty bands for the conveyance calculation 69
Figure 4.26 Vegetation on the River Blackwater (Hampshire, UK) in (a) May 1996
downstream from Colebrook Bridge (b) March 1995 downstream from
Colebrook Bridge and (c) May 1996 upstream from the footbridge 70
Figure 4.27 Plan view of the River Blackwater indicating Gauge site 3 71
Figure 4.28 River Blackwater cross-section geometry – Gauge site 3 72
Figure 4.29 Roughness zones and descriptions for the River Blackwater 72
Figure 4.30 Select time series for the submerged fine-leaved plants 73
Figure 4.31 Expected time series for the submerged fine-leaved plants (0.6-1.2m
depth) 74
Figure 4.32 Bed cover zone unit roughness values 74
Figure 4.33 Emergent reeds characteristic of the bank vegetation 75

- xviii -
Figure 4.34 Roughness zones for the River Blackwater 75
Figure 4.35 Cross-section details for Gauge site 3 on the River Blackwater 76
Figure 4.36 River Blackwater cross-section at Gauge site 3 illustrating the left
inside of bend top-of-bank marker 77
Figure 4.37 River Blackwater roughness zone allocation 78
Figure 4.38 Select the date for the CES conveyance prediction 78
Figure 4.39 Comparison of the CES predicted conveyance to the measured data for
February 1995 79
June 1995 79
December 1999 80
Figure 4.42 The conveyance predictions for Gauge site 3 on the River Blackwater
with the uncertainty bands shown 81
Figure 4.44 River Dane vegetation 81
Figure 4.43 The 6km reach of the River Dane with cross-section 30 identified 82
Figure 4.45 River Dane cross-section 30 geometry 83
Figure 4.46 Roughness zone definition and description 84
Figure 4.47 Select the Time series for emergent reeds 85
Figure 4.48 Emergent reed time series for a given year 85
Figure 4.49 Manually overwrite the suggested roughness based on available
information 86
Figure 4.50 Final roughness zone value in the .rad file 86
Figure 4.51 River Dane channel description 87
Figure 4.52 River Dane cross-section with straight top-of-bank markers 87
Figure 4.53 River Dane roughness zone allocation 88
Figure 4.54 River Dane CES rating curve prediction compared to the measured data 89
Figure 4.55 The River Dane stage-discharge curve after calibration 90
Figure 4.56 Final stage-discharge curve for the River Dane 90

- xix -
- xx -
1. INTRODUCTION
1.1 Why do we need a new conveyance system?
The Environment Agency for England and Wales identified the need to reduce the
uncertainty associated with flood level prediction through incorporating the recent
research advances in estimating river and floodplain conveyance. Existing methods for
conveyance estimation that are available within 1D Hydrodynamic modelling software,
e.g. ISIS, MIKE11, HECRAS, HYDRO-1D, are based on some form of the Manning
Equation, first published in 1890. With the substantial improvement in knowledge and
understanding of channel conveyance that has taken place over the past twenty years,
there is a need to make these more advanced techniques available for general use in
river modelling. A key component of the new conveyance system is the increased
knowledge on river resistance from a diverse set of sources, covering different types of
vegetation and surface material (bed, bank and floodplain) for the fluvial system. The
desire to reduce the uncertainty for the predicted water levels has prompted the need to
understand and quantify the uncertainty associated with the methodology and model
inputs. The new conveyance system is therefore designed to capture the advances in all
three components i.e. the diverse roughness knowledge, the improved conveyance
estimation and the quantification of uncertainty.
1.2 What is the CES?
The Conveyance Estimation System or ‘CES’ is a software tool that enables the user to
estimate the conveyance or carrying capacity of a channel. This capacity is based on
the ability of the channel to resist the flow, driven by gravitational acceleration in terms
of channel gradient, through surface friction and channel morphology i.e. cross-section
shape and planform sinuosity. The CES includes a component termed the ‘Roughness
Advisor’, which provides advice on this surface friction or ‘roughness’, and a
component termed the ‘Conveyance Generator’, which determines the channel capacity
based on both this roughness and the channel morphology. In addition, the CES
includes a third component, the ‘Uncertainty Estimator', which provides some
indication of the uncertainty associated with the conveyance calculation. The primary
outputs from the CES components are:
· Roughness Advisor: roughness values

· Conveyance Generator: stage-conveyance relationship
· Uncertainty Estimator: upper and lower bands for the stage-conveyance
relationship.
These three components are essential for determining site-specific stage-conveyance

curves. The CES has a further module for calculating backwater curves in reaches
upstream of a control point i.e. a cross-section where the flow and / or depth are known.
In addition to the stand-alone CES software, the CES will be available as open source to
incorporate in any one-dimensional hydrodynamic modelling software. As part of this
research programme, the CES has been incorporated into iSIS Flow.

1
1.3 Why Provide a Conveyance Manual?
The Conveyance Manual Provides background knowledge and advice, enabling the user
to obtain the maximum benefit from the CES modelling tool and supplements its ‘Help’
menu.
The Conveyance Manual describes the three key components of the Conveyance
Estimation System (CES): the Roughness Advisor, the Conveyance Generator and the
Uncertainty Estimator. It gives an overview of the hydraulic equations and fundamental
channel flow processes, with emphasis on their relevance to successful model
application and usage. It provides advice on the complete CES modelling process such
that users can:
· optimise the use of interdisciplinary knowledge on channel resistance that is

available through the Roughness Advisor
· describe the channel and floodplain system such that all flow processes / energy
losses are ‘captured’, thus optimising the Conveyance Generator predictions
· make informed decisions based on sound hydraulic knowledge and understanding
· understand and interpret information provided by the Uncertainty Estimator
· work through examples of the complete modelling process within the CES.
1.4 How do I use the Conveyance Manual effectively?
The Conveyance Manual is structured through a Question and Answer approach and all
questions are listed in the contents for ease of use. The Conveyance Manual comprises
four sections:
· Section 1 provides a brief introduction to the Conveyance Manual

· Section 2 provides a technical description of the complete conveyance
methodology, including background; the selected approaches for estimating
roughness, conveyance, uncertainty and backwater profiles; the key equations and
their range of applicability and the method limitations
· Section 3 describes the use of the CES software at component level, providing
advice on how best to describe the channel-floodplain system to achieve desired
results. This does not include a software ‘button-clicking’ guide, which is available
within the online Help
· Section 4 provides worked examples of the complete conveyance methodology for
typical channel sections.
Thus to effectively use the Conveyance Manual, it is necessary to identify the required
task. First time users of the CES should read the technical background and calculation
details [Section 2]. Section 3, in conjunction with the online help facility, is essential
for the hands-on application of the CES software and Section 4 is useful as training
material.

2
2. TECHNICAL BACKGROUND
This section provides the technical background to the roughness approach, the
conveyance calculation and the method for estimating uncertainty.
2.1 Overview
2.1.1 What are legacy practices in conveyance estimation?
When building an open-channel model, the modeller schematises the real system into a
series of analogous objects of hydraulic units and then builds a preliminary model by
pooling together these units with their appropriate data. These units interconnected in
the modelling dataset are an analogous image of the real system. Legacy conveyance
estimation is embedded in this modelling procedure (not an independent modelling
activity) through tasks carried out (a) externally by modellers and (b) internally by open
channel modelling software tools. The modeller’s tasks are in two steps:
1. The modeller transforms the real system into a series of cross-sections represented
in terms of x and y co-ordinates, where x co-ordinates are referred to as offsets and y
co-ordinates to spot-heights or elevations.
2. Each offset is associated with further assigned information, where the modeller
decides the role that the particular strip of cross-section between two consecutive
offsets plays in the make-up of conveyance. This is in the form of: (i) assigning
particular roughness values; (ii) schematising the cross-section into panels and
identifying them by an appropriate labels; (iii) identifying dead-storage zones; and
(iv) assigning pre-calculated relative path lengths. Either the assigned roughness
values are obtained through calibration exercises or they are selected using
published homologous data.
The model built in the above fashion is normally embedded within the 1D
hydrodynamic model without any provision to treat this task as an independent activity.
Thus, conveyance is estimated according to the particular conveyance estimation
method, such as the Manning equation, which is embedded in the 1D-hydrodynamic
software tool. These tools do not encourage the modeller to study conveyance
independently, yet this process provides a great deal of insight.
2.1.2 What changes are expected after the CES is recognised as the industry
norm?
Legacy conveyance estimation (Section 2.1.1) is bound to change following the uptake
of the CES, starting with changes in schematisation. The envisaged schematisation
procedure is as follows:
· Schematise the whole system into reaches and schematise each reach into plan
form areas termed “roughness zones” (Section 3.1.3).
· Abstract data by assigning appropriate unit roughness values to each zone through
application of the Roughness Adviser software tool (Section 3.1), and estimating
sinuosity (Section 3.2.13) along the thalweg of the channel. The outcome is
assigning three different roughness unit values to each roughness zone and a
sinuosity value to each portion of the modelled reach where the sinuosity is

3
unchanging. Areas of static storage or re-circulation zones are defined as non-
conveyance zones (Section 3.2.9).
· A cross-section framework is then incorporated to define the cross-section shape
through the offset and elevation. Top-of-bank markers (Section 3.2.4) are assigned
which are subtly different to previously used “panel markers”. Here, where the
cross-section passes through plan form roughness, sinuosity or non-conveyance
zones, these values are assigned to that portion of the cross-section.
· Run the model and obtain results using the CES. This is a new culture, which will
enable the modeller to gain an insight to the role played by the conveyance of the
channels in the overall hydraulic performance of the system. This is particularly
useful in design and flood forecasting and a new ground in maintenance and
navigation.
· Monitor and improve the performance of the model. This is a new culture and
it is important to realise the CES models do not represent absolute truth and in fact
there is no such thing as perfect modelling results. The modellers have to
continually assess the performance of the model and feedback the insight gained
into the model performance to peer user groups to facilitate knowledge retention.
Improving model must be a continual process.
2.1.3 What will I do that I did not do in legacy practices?
The conveyance estimation tasks using the CES are not embedded in the model building
activities anymore. This task is now modularised and mimics the model building
activities through the procedure outlined above. The modellers are provided with a
flexible interactive tool with a great deal of freedom. Their results become dependent
of their assumptions and therefore it becomes an important task to keep a record of all
of their assumptions. As a result of this new freedom, defensibility of the results
becomes a more important issue than before. Owing to this defensibility issue
sensitivity tests gain a new meaning and precisely for this reason, uncertainty estimation
is provided as the integral component of the CES (Section 2.4). The key difference to
previous modelling is that an insight is gained into the role of conveyance in the overall
hydraulic performance of the system, in an uncertain background.
2.1.4 What will the CES do?
The CES transforms research findings of the last 2-3 decades into working tools in a
modular form encouraging a new culture in modelling open-channel hydraulics.
Provided that the modeller schematises the system as above, assigns the unit roughness
values, allocates top-of-bank markers and inputs sinuosity values, all the other activities
will be carried out internally by the CES.
2.1.5 What new opportunities are provided by the CES?
· Treating the conveyance estimation as a modelling activity mimicking 1D-

hydrodynamic modelling
· The ability to gain an insight into the role of conveyance in the overall system
performance
· The modellers will gain new freedoms and open up the doors for a range of uses

4
· Defensibility of the CES modelling will become an issue and to this end both
quality assurance and understanding inherent uncertainties will be of new
significance
· The CES will be used in new ways by designers and flood forecasters and the CES
will cultivate a new breed by becoming tools of navigation and channel
maintenance.
2.2 Roughness
2.2.1 Why do we need to worry about roughness?
Channel roughness provides the primary resistance to the pressure imbalance or force,
which effects the fluid flow. Understanding and describing the nature of this resistive
force is therefore essential in determining the water level, since a greater resistance will
result in a slower flow and higher water level. The channel roughness is the resistance
due to the local boundary friction only and is therefore best quantified through
interpretation of the surface roughness [e.g. substrate, vegetation, irregularities] into an
associated energy loss.
2.2.2 Why is there a need for a new approach to estimating roughness?
Existing methods for describing roughness are based on semi-empirical formulae [e.g.
Chezy (1768), Manning (1890)] which are not based on rigorous physics and include
roughness coefficients that are increased to account for channel losses due to shape
effects. Chow (1959) provides the most thorough set of roughness values for various
surface materials to date. Other substantial contributions include Hicks and Mason
(1993, 1998) and Barnes (1967), which cover roughness values, descriptions and
photographs specific to New Zealand and America.
The new approach to estimating conveyance (See Section 2.3) provides a method
whereby the energy “loss”1 mechanisms e.g. lateral shear, transverse currents and
boundary friction are treated individually. The channel section is divided into a number
of elements, through vertical slicing, and the contributions to the total conveyance are
determined within each slice. To this end, it is necessary to express the roughness in
terms of a unit roughness value, which may vary laterally across the section, and is only
dependent on the local boundary friction.
In recent years, much research and data has provided the basis for a thorough review of
river roughness. This Roughness Review (DEFRA/EA, 2003b) includes over 700
references and draws upon data for natural and man-made materials, obstructions,
floodplain cover and aquatic vegetation. This review has sourced all unit roughness
values that are in a format that is appropriate to the conveyance calculation, together
with descriptive advice and a substantial photographic database. In addition,
information on vegetation morpho-types derived from the national data set obtained
through the national survey of river habitats (River Habitat Survey, Raven et al 1998) is
included.
1
The term energy “loss” is used here to describe the transfer between energy states, rather than actual physical losses,
since energy within a closed system is conserved. This loss may be interpreted as a loss to the overall energy
contributing to the streamwise cross-sectional conveyance through e.g. the development of transverse vorticity.

5
In short, the new approach to estimating roughness is necessary to make full use of this
increased interdisciplinary knowledge on channel roughness and aquatic vegetation, and
the unit roughness is essential to the conveyance calculation approach.
2.2.3 How is the roughness represented?
The roughness is represented in terms of the local unit roughness nl.
2.2.4 What is a unit roughness nl-value?
A unit roughness nl represents the roughness due to an identifiable segment of boundary

friction within the channel section. This is different to Manning’s n, which is applied to
whole regions of the cross-section. The roughness magnitude is equivalent to a
Manning’s n that has been stripped of all the energy losses due to lateral shear,
secondary flows and sinuosity, and is hence based entirely on the local boundary
friction. All the remaining energy losses are incorporated through the other parameters
[See Section 2.3].
2.2.5 Why was the decision taken to express roughness in terms of ‘n’?
The use of the Manning n is widespread and to date, most resistance advice,
photographs and summation approaches in the literature are expressed in terms of ‘n’
(Barnes, 1967; Chow, 1959; Cowan, 1956; Hicks and Mason, 1993 and 1998). The
Roughness Advisor is therefore based on an n rather than a Darcy f or Chezy C, to
maintain this user familiarity and confidence. This was a critical decision made by the
Project User Consultative Group and Expert Advisory Board to receive wider user
acceptance.
2.2.6 What is a roughness component?
The local unit roughness values (nl ) in the Roughness Advisor can be composed of
three component roughness values. These include boundary friction due to:
· bed, bank or floodplain surface material (e.g. sand, clay, bedrock)

· vegetation (e.g. grass, reeds, water lillies)
· irregularities (e.g. groynes, urban trash, pools and riffles).
2.2.7 How are the component roughness values combined?
The unit roughness at a point in the channel section can comprise up to three roughness
components. These are combined to get the total unit roughness at a point in the section
through,
2
[
nl = nsur 2
+ nveg 2
+ nirr ] 1
2
2.1
where nsur, nveg and nirr are the unit roughness values due to surface material, vegetation
and irregularity respectively. These are associated with a depth of 1m, which was
selected as a representative depth of flow for UK rivers. Other methods (e.g. Cowan,
1956) of combining individual roughness values were considered, but the “root sum of
the squares” approach was adopted since:

6
· it highlights the contribution of the largest roughness component
· the roughness is squared before being combined since the energy loss is related to
the square of the local velocity
· more complex methods would not be backed-up by sufficient data.
This approach assumes that the resistance “loss” mechanisms are mutually independent
and hence the total resistance can be expressed as the sum of the individual resistances.
Previous experimental work (Einstein and Banks, 1950) supports this theory.
Figure 2.1 provides an example where the portion of the cross-section Dy contains two
roughness components, surface material (pebbles) situated below some vegetation
(grass). The values for each component are used in (2.1) to determine the total local
boundary friction or nl value at that location in the cross-section i.e.
2
[
nl = nsub pebbles
2
+ nveg grass
( 2
+ nirr =0 )]1
2
2.2
[
e.g. nl = 0.027 2sub + 0.030 veg
2
]
1
2
= 0.040 which is a smaller value than direct summation
value of 0.057.
Dy
nsub (pebbles) nveg (grass)
Figure 2.1 Example of components for substrate and vegetation contributing to

local boundary friction
2.3 Conveyance
2.3.1 What is conveyance?
Conveyance is a quantitative measure of the discharge capacity of a watercourse (Evans

et al, 2001). It relates total discharge to a measure of the gradient or slope of the
channel. Original definitions of conveyance assume uniform flow (with no spatial or
temporal variation) in regularly shaped channels with constant gradient. The definition
is:
1
Q = KS 2
2.3

7
where K (m3/s) is the conveyance, Q (m3/s) is the discharge and S is the common
uniform gradient. In idealisation there is no ambiguity about this slope since the water
surface slope, the bed slope and the so-called “energy” or “friction” slope all coincide.
For non-uniform geometry, local variations in the section shape influence the water
velocity and therefore the balance between the potential and kinetic energy of the flow,
the specific energy, E (m) of the flow is commonly defined by:
U2
E = h+ 2.4
2g
where g (m2/s) is the acceleration due to gravity, h (m) is the water surface level of the
flow and U (m/s) is the section average velocity. In this case, the slope to be used in the
definition of conveyance is the magnitude of the gradient, Sf, in the downstream (x-)
direction of the specific energy
dE
Sf = - 2.5
dx
2.3.2 Why do we need to know about channel conveyance?
The ability of a channel to convey water directly influences the water levels. Resistance
to flow results in smaller velocities and greater depths. Water level prediction is
essential for flood management tasks such as flood-forecast warning and flood risk
mapping. Flood management practice relies on reliable water level predictions to make
informed decisions regarding infrastructure design and operation, and more importantly,
emergency evacuation planning. Flood levels are also necessary for strategic planning
i.e. evaluating and comparing options against criteria, hydrometric users and channel
maintenance i.e. timing, scheduling and prioritisation of dredging and vegetation
cutting.
2.3.3 Why is there a new approach for estimating channel conveyance?
Existing methods for estimating conveyance in 1D hydrodynamic models are based on

the Manning equation. This empirically derived equation, first published in 1890, is not
based on rigorous physics and provides unreliable results for sudden changes in area or
wetted perimeter with depth, when shape effects are ignored. When applied as a
Divided Channel Method i.e. the flow rate calculated separately for the floodplain and
main channel zones and subsequently summed; the lateral shearing and consequent
momentum transfer between the vertical divisions or ‘slices’ is ignored. The Manning
equation thus represents all the energy losses though the well-known ‘n’ value (Chow,
1959), lumping all the physical flow processes such as shearing and momentum transfer
into one ‘catch-all’ parameter, with little understanding and interpretation of the
independent energy “loss” mechanisms.
More recently, there have been advances in the understanding and modelling of the
various flow processes and the associated energy losses. Research into the exact nature
of these mechanisms such as local friction, turbulence due to lateral shearing and the
presence of secondary flows, together with an ever increasing database of observed flow
parameters, has provided the facility for testing and validating these new approaches.
Significant contributions include those of Chang (1983, 1984), Ervine and Ellis (1987),

8
Shiono and Knight (1989), Ackers (1992, 1993), Bousmar and Zech (1999) and Ervine
et al (2000). For further details see Evans et al (2001) and DEFRA/EA (2003a).
2.3.4 Do I need to have a strong mathematical background to use the CES?
No. The equations described in this section are all coded within the CES and they
require no user manipulation. The section is provided to enable better understanding
and description of the conveyance inputs (see Section 3.2) such that the use of the CES
is optimised.
2.3.5 What are the different energy loss mechanisms?
The energy “losses” arise through the development of vortex structures on a variety of
length scales. Once vorticity is created, its rotational energy cascades down in length
scale into increased turbulence intensity, until it dissipates as heat through viscosity.
The streamwise translational kinetic energy is thus transferred in part to rotational
kinetic energy, which no longer contributes to the streamwise channel conveyance
capacity. It is possible to identify situations in which the vorticity is increased e.g. these
vortex structures may arise from:
· Boundary friction caused by the resistance due to surface roughness

· Turbulence due to lateral shearing in regions with steep velocity gradients e.g.
adjacent to a flow boundary layer; and at the floodplain main channel interface,
where the water is flowing faster in the main channel than on the floodplains
· Transverse currents developing in regions of steep velocity gradients. These
secondary currents vary with relative depth i.e. they change their orientation as the
flow moves from in-bank to out-of-bank flow; and sinuosity i.e. the secondary
flows expand into the bend in a helical-type shape and then gradually dissipate
beyond the bend
· Water moving from the floodplain region into the main channel, which is also
subject to expansion losses. The velocity i.e. both magnitude and direction of the
flow is important here, as high velocity results in a greater shear as the floodplain
flow passes over the main channel flow, driving the vorticity structures. In
meandering channels, this direction may retard or enhance the main stream
rotations. Contraction losses may occur as water passes from the main channel into
the floodplain region.
Additional “losses” may arise due to changes in form in non-prismatic channels i.e.
channels characterised by rapid changes in channel section over short reaches. Figures
2.2-2.4 illustrate these vortex structures and the region where they are likely to occur.

9
Bed friction
Secondary flows
Lateral shear zones
Terraced floodplain
Floodplain Main channel
Figure 2.2 Energy loss mechanisms in a straight compound channel for three
depths of flow
Figure 2.3 Flow mechanisms associated with straight overbank flow in a two-
stage channel (Shiono and Knight, 1991)

10
Figure 2.4 Flow mechanisms associated with in a two-stage meandering channel
(Ervine et al., 2001)
2.3.6 What is the selected approach in the CES for calculating conveyance?
The calculation approach is based on the depth-integrated Reynolds Averaged Navier-

Stokes (RANS) equations for flow in the streamwise direction. It extends the original
Shiono-Knight Method SKM (Shiono and Knight, 1989) for straight prismatic channels,
to include the more recent approach of Ervine et al (2000) for meandering channels.
The unit flow rate distribution is determined laterally across the section from
(DEFRA/EA, 2003a),
¶ é æ f ö 2 ¶ æ q öù (1.015 - s ) (s - 1.0 ) C ¶ é q 2 ù
1
fb q 2
gHSo - + êl H ç ÷ q ç ÷ú = G + uv ê ú
8H 2 ¶y ê
ë è 8 ø ¶y è H øú
û
0.015 0.015 ¶y ëê H úû
2.6a
(I) (II) (III) (IV)
fb q 2 ¶ é æ f ö 2 ¶ æ q öù
1
¶ éq2 ù
gHS o - + ê l H ç ÷ q ç ÷ú = C uv ê ú 2.6b
8H 2 ¶y êë è8ø ¶y è H øúû ¶y ë H û
where:
g gravitational acceleration (9.807 m/s2)
q streamwise unit flow rate (m2/s)
H local depth normal to the channel bed (m)
So reach-averaged longitudinal bed slope
y lateral distance across the channel (m)
b coefficient to account for influence of local bed slope on the bed shear stress

11
s reach-averaged sinuosity
f local friction factor
l dimensionless eddy viscosity
G secondary flow parameter – straight channels
Cuv meandering coefficient.
Equation 2.6a is applicable for a sinuosity s of 1.0 for inbank flow and 1.0 £ s £ 1.015
for overbank flow. Equation 2.6b is applicable for a sinuosity greater than 1.0 and
greater than 1.015 for inbank and overbank flow respectively. The lateral unit flow rate
distribution can then be integrated to find the total cross-section flow rate Q (m3/s), and
hence the total cross-section conveyance K (m3/s), from
K=
Q
»
ò q dy 2.7
1 1
Sf 2 So 2
where the reach-averaged longitudinal friction slope Sf is approximated by the reach-

averaged longitudinal bed slope So. The dependence on four experimentally determined
coefficients, f, l, G and Cuv implies that the method is semi-empirical.
The terms in Equation 2.6a and b represent:
I) An approximation to the variation in hydrostatic pressure along the reach (Sf » So)
II) energy losses due to boundary friction effects
III) turbulence losses due to shearing between the lateral layers
IV) turbulence losses due to secondary currents
Term (I), the source or hydrostatic pressure term, is evaluated from the gravitational
acceleration g, the local depth H and the longitudinal bedsope So, as an approximation
to Sf.
Term (II), the boundary friction term, and term (I) comprise the primary balance of
momentum in Equation 2.6a and b. Simplification of these two terms results in the
well-known Darcy-Weisbach friction formula. Term (II) is evaluated from the local
depth H, the local friction factor f and the bed slope coefficient b given by,
(
b = 1 + S y2 ) 1
2
2.8
where Sy is the transverse bed slope. The lateral distribution of f is based on the local nl
value provided from the roughness calculation (Section 2.2). This nl value is converted
to an equivalent ks value, through the rough turbulent Colebrook-White Law at a
reference depth of 1m (based on typical UK river depths), with coefficients suitable for
open channel flow,
é 1 ù
ê H 6 ú é 1 ù
ê 8 g nl ( -2.03) ú ê ú
ëê 8 g n ( -2. 03 ) ûú
k s = 12.27 H 10 ë û
= 12.27 ´ 10 l
2.9

12
This ks value is similar to Nikuradse’s sand grain dimension, and in the case of
vegetation, it can be interpreted as an equivalent length scale of the turbulent eddies that
are generated. The reasons for selecting the Colebrook-White equations are
documented elsewhere DEFRA/EA (2003a).
The method for resolving the lateral distribution of f is dependent on the ks/H ratio. The
reason for this is that the Colebrook-White law is not applicable at high ks/H ratios,
since the bracketed term (Equation 2.10) approaches one and the logarithm of one is
zero. Further details in DEFRA/EA (2004/5 in preparation). Three categories are
defined:
1. ks/H < 1.66
Here, f is solved from the rough turbulent Colebrook-White Law for natural rivers,
1 é ks ù
= -2.03 log ê ú 2.10
f ë12.27 H û
and from the full Colebrook-White Law (i.e. rough and smooth components) for
experimental flumes,
1 é ks 3.09n ù
= -2.03 log ê + ú 2.11
f ëê12.27 H 4q f úû
2. 1.66 ks/H £ 10
Here, f is solved through a Power Law approximation,
8 ks
f = 2.12
41.3015 H
3. ks/H > 10
The Power Law approximation (category 2, Equation 2.12) is restricted at a

maximum ks/H ratio of 10, i.e.
8 ks 8
f = = (10) = 1.94 2.13
41.3015 H 41.3015
Term (III), the lateral shear term, has a relatively small impact on the total discharge Q.
It is evaluated from the local friction f, the local depth H and the dimensionless eddy
viscosity l. The lateral distribution of l is given by (Abril, 2002),
(
l = lmc - 0.2 + 1.2 Dr-1.44 ) 2.14
where the main channel value, lmc, is 0.24 (DEFRA/EA, 2003a) and the relative depth
Dr is defined as the local depth H over the maximum depth. For further details
regarding the eddy viscosity model see DEFRA/EA (2003a).

13
Term (IV), the secondary flow term, changes between Equation 2.6a and b, since the
nature of the secondary flow varies with relative depth and planform sinuosity.
Equations 2.6a and b are solved for the unit flow rate q rather than the depth-averaged
velocity Ud. This is due to the strong continuity properties of q (Samuels, 1989).
Expressing the complete secondary flow term (right hand side of 2.3a and b) in terms of
the depth-averaged velocities gives,
¶H UV( ) d
2.15
¶y
where U and V are the streamwise and lateral velocity components respectively, and the
suffix ‘d’ represents depth-averaging. Since the lateral velocity is an additional
unknown, the Conveyance Generator provides a “closure” for this term based on two
previous approaches: that of Shiono and Knight (1989) for straight prismatic channels,
where a secondary flow term G is defined as
G=
¶
¶y
[ (
H rUV )]
d 2.16
and that of Ervine et al (2000) for meandering channels,
UV = C uvU d2 2.17
Here, the secondary currents are related to the depth mean velocity by a coefficient Cuv,
which is applied within the main channel only. The meandering model (Equation 2.17)
introduces a non-symmetrical effect on the lateral depth-averaged velocity distribution. The
conveyance methodology combines these two approaches such that for straight, transitional and
fully meandering channels, the secondary flow term is composed of G, a linear combination G
and Cuv, and the Cuv model respectively (Figure 2.5). The sinuosity s is defined as the thalweg
length over the valley length.
Secondary
flow term Legend:
straight
transition meandering total

Cuv
G
1.0 1.015 1.030 1.045 Sinuosity s
Figure 2.5 Contributions from the secondary flow terms with increasing
sinuosity
The inbank (i) and overbank (o) distributions of G are given as (Abril, 2002),
Gmc ( 0) = 0.15 Hr gS o Gmc (i ) = 0.05 Hr gS o G fp ( o ) = -0.25 Hr gS o 2.18a,b,c
For the case of uneven main channel bank elevations, a linear combination of Equation
2.18a and b is applied for depths between the two top-of-bank elevations, in the main

14
channel region. Equation 2.18a,bandc are only applied in the shear layer i.e. regions of
mixing between the flood plain and main channel. In practice, these typically extend
across the entire main channel region (Samuels, 1988) and over a portion of the flood
plain. The flood plain extent is approximated based on the lesser of the main channel
width and ten times the bankfull depth. For straight channels Cuv = 0. For inbank flow
in channels of sinuosity greater than 1.0, Cuv is given by (DEFRA/EA, 2003a),
C uv = 4.3274s 2 - 7.8669s + 3.5395 [1.0 < s £ 2.5 i.e. 0% < Cuv £ 11%] 2.19
For overbank flow in channels of sinuosity between 1.0 and 1.015 the secondary flow
term is a linear combination of G and Cuv. For overbank flows in channels of sinuosity
greater than 1.015 the secondary flow term Cuv is given by (DEFRA/EA, 2003a),
C uv = 7.1659s - 6.6257 [1.015 < s £ 2.500 i.e. 0.6% < Cuv £ 11%] 2.20
The expected distributions of calibration parameters f, l, G and Cuv from terms (II) to
(IV) are represented in figure 2.6 for a typical two-stage channel.
f æk ö
f = fç s ÷
èHø
l
l = f (lmc, Dr)
G Gfp (-ve) Gmc (+ve) Gfp (-ve)
Cuv Varies with s
Top of
bank
Figure 2.6 Lateral distribution of calibration parameters for a typical two-stage

channel
2.3.7 Why was the RANS approach selected?
After an extensive literature review (Evans et al, 2001; DEFRA/EA, 2003a), two key
approaches to estimating conveyance were identified:
· The energy loss approach: Division of the channel cross-section into physically
identifiable units, representing different flow mechanisms, and then summing the
energy loss contributions (Ervine and Ellis, 1987; Shiono et al, 1999)
· The RANS approach: Depth-integration of the Reynolds Averaged Navier-Stokes
(RANS) equations for flow in the streamwise direction (Shiono and Knight, 1990;
James and Wark, 1992; Ervine et al, 2000).
After a thorough review and assessment (DEFRA/EA, 2003a), the RANS approach was
selected because:

15
· The RANS approach has a sound physical basis as it is derived from the original
Navier-Stokes equations for fluid flow
· This approach incorporates most of the energy losses outlined in Section 2.3.5. It
includes boundary friction, lateral shearing and momentum exchange i.e. secondary
flows and expansion / contraction of flow between the main channel and
floodplains. However, form losses are not considered. These can be represented in
a one-dimensional hydrodynamic model through discrete energy “losses” e.g. the
Bernoulli loss unit in ISIS
· The energy loss hierarchy is considered. The lateral distributions of f, l and G vary
with depth of flow, and the meandering coefficient Cuv varies with both depth of
flow and sinuosity. The result is that the contributions of the different energy loss
terms are weighted according to when the relevant loss mechanism is likely to
occur
· Despite having four calibration parameters, f, l, G and Cuv, there are existing
models that can be adapted to represent the lateral distributions of these terms
· The method can be applied to any river cross-section or planform geometry,
including multi-thread meandering two-stage channels. Although an analytical or
numerical solution is required, the implementation is feasible within computer
code, without requiring detailed knowledge by the user
· The method allows for changes in local roughness values across the section,
enabling representation of localised bed features such as trees, boulders, artificial
lining etc.
· The method is applied within a cross-section framework and is thus readily
interpreted for use within existing 1D hydrodynamic modelling software
· The primary output is the conveyance and flow rate for a given water level. In
addition, the lateral distribution of the unit flow rate, depth-averaged velocity, bed
shear stress and shear velocity is provided. The velocity and bed shear stresses are
useful for vegetation maintenance and sediment transport respectively
· The method has been widely tested against both experimental and real river
observations, with encouraging results.
The Project Expert Advisory Panel reviewed and approved the scientific basis and
application of this method.
2.3.8 What are the limitations of this method?
· The conveyance equations are based on the depth-integration of the RANS equation
for flow in the streamwise direction. These equations are inappropriate at high
sinuosities, where the main channel velocity vector changes direction above
bankfull depth, such that it is aligned with the mean floodplain direction. The use
of depth-averaging is inappropriate in this instance
· Energy losses due to form change are not considered i.e. rapid changes in cross-
section between consecutive river sections. In 1D-river models, these losses can be
represented through discrete energy “losses” e.g. the Bernoulli loss unit in ISIS
· There are four calibration parameters, f, l, G and Cuv, which may provide
difficulties. E.g. Which parameter is not performing adequately? Which parameter
should be altered to achieve calibration? Are the parameters mutually independent?
Are the parameters accurately depicting the energy loss mechanisms for which they
are prescribed?

16
· The method will take longer computationally than the previous ‘Manning’
equation, as an iterative solution is required for each depth of flow. The solution of
Equation 2.6a and b has been designed to converge nearly quadratically, thus
reducing the number of iterations to less than 7
· The method is difficult for the ‘casual’ user to understand in detail.
2.3.9 How are the RANS equations solved?
Equations 2.3a and b are non-linear, elliptic, second order partial differential equations.
These are approximated numerically, using linear finite elements; a method well suited
to the solution of elliptic equations. The cross-sectional area of flow represents the
solution domain, which is discretised laterally into a number of elements, and the
variable q is replaced with piecewise linear approximations. The solution to the
resulting system of discrete equations is generated through an iterative procedure,
linearised in the correction term Dqn, to update unit flow rate qn+1 from the known qn
value from the previous iteration. The fixed boundary condition, q = 0, is prescribed at
the boundary nodes at the edges of the flow domain. The iteration procedure is
designed to converge nearly quadratically. For further details regarding the numerical
scheme, see DEFRA/EA (2003a).
2.3.10 How is my cross-section discretised?
The cross-section is discretised laterally into 100 divisions. The default number of
depths for which the conveyance calculation is solved for is 25 (Figure 2.7).
y1y2y3y4y5 ……………. y100
Depth 25
Depth 1
Figure 2.7 Cross-section discretisation
2.3.11 Has this approach been tested?
The conveyance methodology has been tested against a range of data, covering small-
and large scale experimental flume measurements, purpose-made real river
measurements and Section 105 and MDSF (Modelling and Decision Support
Framework) models accompanied by real river observations. The data sets were
carefully selected (DEFRA/EA, 2002) to represent a range of channel types (rural /
urban), channel morphologies (widths, gradients, aspect ratios), flow conditions and
surface cover. The results were encouraging and showed an improvement on previous
methods (DEFRA/EA, 2003a; DEFRA/EA, 2004/5 in preparation), particular in
capturing the flow variations near the bankfull depth.

17
2.4 Uncertainty
2.4.1 What is uncertainty?
One definition of uncertainty is as follows:
“Uncertainty is a general concept that reflects our lack of sureness about something or
someone, ranging from just short of complete sureness to an almost complete lack of
conviction about an outcome”
Uncertainty arises principally from lack of knowledge or of ability to measure or to

calculate and gives rise to potential differences between assessment of some factor and
its “true” value. Understanding this uncertainty within our predictions and decisions is
at the heart of understanding risk. Within uncertainty we are able to identify:
· knowledge uncertainty arising from our lack of knowledge of the behaviour of the
physical world
· natural variability arising from the inherent variability of the real world
· decision uncertainty which reflects the complexity of our social and organisational
values and objectives.
However, this classification is not rigid or unique. For example, uncertainty on weather
or climate will be taken as “natural variability” within flood risk management but as
“knowledge uncertainty” in the context of climate simulation. It is helpful also to
consider the differences between accuracy, error and uncertainty. Accuracy and error
differ from uncertainty as defined above but limitations in accuracy or the possibility for
human error will contribute to the overall uncertainty.
· Accuracy has two facets. It deals with the precision to which measurement or
calculation is carried out (i.e. the maximum deviation of the measurement or
calculation from the “true” amount) and with the number of digits of precision
which are carried through a calculation (whether or not these are meaningful).
Potentially, accuracy can be improved by better technology.
· Errors are mistaken calculations or measurements with quantifiable and
predictable differences from the actual value.
2.4.2 Why do I need to account for uncertainty?
Consideration of uncertainty within the decision process attempts to provide bounds to

our lack of sureness and thereby provides the decision-maker with a better context in
which to take a decision. Through investigation of the sources of uncertainty, this type
of analysis enables the engineer or decision-maker to identify the uncertainties that most
influence the final outcome and focus resources efficiently. By understanding the
sources and importance of uncertainty within our decisions, we should be able to make
better and more informed choices.
The effects of uncertainty in the estimation of conveyance differ with the various
process undertaken by the Agency. The potential consequences of uncertainty and
typical current methods of mitigation for the uncertainty in conveyance are listed in
Table 2.1 below. The sensitivity of project decisions to uncertainty in conveyance
estimation needs to be established as strategic decisions made early in the project life

18
cycle can have far reaching consequences but it is at this early stage that uncertainties in
information and data are greatest.
Table 2.1 Potential consequences of uncertainty in flood conveyance
Agency Typical mitigation

Catchment-wide consequence of uncertainty
Activity strategy
Strategic Inadequate framework for long-term view of flood risk at a Undertake sensitivity
Planning catchment scale, leading to sub-optimal or inadequate testing to bound the
strategies and solutions. Potential unexpected long-term expected results
increase in flood impacts on people and properties.
Flood defence Under capacity of defences leading to potential failure below Undertake sensitivity
design the design standard or over capacity potentially leading to analyses and add a
morphological problems or lower economic return than freeboard to allow for
planned; over estimation of capacity of defences leading to under capacity
lack of implementation of schemes due to excessive cost.
Maintenance Inadequate or excessive maintenance activities, possibly Maintain to a defined
unnecessary disruption to aquatic and riparian habitats or programme
insufficient capacity of the watercourse leading to increased
flood risk.
Real time Inaccuracies in the threshold trigger levels, under- (over-) Implement real-time
forecasting estimation of lead times, inexact inundation extent, and updating procedures
incorrect retention times of floods.
Hydrometry - Incorrect discharges with potentially large errors, influences Undertake sensitivity
Rating curve flood forecasting and statistical estimation of flood flows for analyses
extension design, impacts upon cost-benefit assessment and decisions
to promote flood defence schemes.
Regulation Indicative Flood Mapping (IFM) in error – inadequate tool Give “health warnings”
for planning and information on possible flood risk, on use of IFM
inadequate (or over necessary) development control, loss of
professional and public confidence in the Agency’s technical
abilities.
There is a close relationship between uncertainty and risk in that the greater the
uncertainty the greater the potential of the project or maintenance activity of not
achieving its objective. This is linked to the confidence on the performance of the
scheme or process to meet its intended objectives. Thus, optimisation of performance
and the confidence with which performance can be achieved is linked inexorably with
understanding and controlling uncertainty.
2.4.3 What factors contribute to the uncertainty in conveyance?
The uncertainties in conveyance estimation are principally due to natural variability and
knowledge uncertainty. Two contributions are normally recognised in knowledge uncertainty;
these are:
· Process model uncertainty in any deterministic approach used in the assessment

· Statistical uncertainty arising from the selection and fitting of statistical
distributions and parameters from data within the assessment.
In the Conveyance Estimation System no use is made of statistical modelling and so

issues of statistical uncertainty do not arise. However, in the overall estimation of flood
risk there will be an important component of statistical uncertainty arising from any
hydrological estimation procedures for flood flow.

19
There are several important components of process model uncertainty in the estimation
of river and flood plain conveyance.
· process model uncertainty arising from the selection and approximation of

physical processes and from parameterisation made in the definition of
conveyance
· representation of topography through the density of discrete survey points and
interpolation rules between them, i.e. the difference between the profile of the river
and of the physical features on the flood plain as represented in the calculations and
the river in “real” life
· uncertainty related to data accuracy from the limitations of the survey methods
used
· uncertainties arising from parameter estimation – particularly the experience and
expertise of the modellers who set up and calibrate computational models of river
flows
· uncertainties from the model calculation methods, approximations and rules; these
are the domain of traditional numerical analysis.
The contributions to uncertainty from natural variability include:
· seasonality of plant growth from variation in temperature, light and nutrients (and
hence resistance of vegetation in channels)
· “memory” in the system e.g. arising from vegetation or river bed conditions being
different after a high flow from the conditions before the event or through episodic
input of sediments into the river system washed off the land surface
· secondary influences of temperature on water viscosity and hence Reynolds’
number and flow and sediment transport capacity of channels.
2.4.4 What approaches for estimating uncertainty were considered?
There are several approaches to the assessment of uncertainty, depending upon the type
and extent of information available and the type of decision to be made.
The Italian Flag concept
This involves a decision process where the results are given a green light for reliable
results, a white light for some uncertainty or a red light for poor results. Each parameter
is given a colour rating and the overall colour is then assessed. This method is limited
as it is based on subjective decisions, cannot be quantified and within the white area, the
level of uncertainty cannot be determined, e.g. closer to green or red.
Scenario Assessment and Climate Impact Analysis
The Intergovernmental Panel on Climate Change (IPCC) is overseeing the international

research effort on assessment of potential climate change, its environmental and social
impacts and policy responses. This is an area where complex interactions abound and
there is significant uncertainty inherent in the “predictions”. The IPCC has issued
guidance on the treatment and interpretation of uncertainty in the context of climate
modelling, impact analysis and consequent decisions (IPCC, 2001). Climate change
impact analysis is conducted in terms of “scenarios” which set as inputs atmospheric

20
emissions to GCM climate simulations as described in the IPCC (2001) Third
Assessment Report. These scenarios are not quantitative predictions about future
conditions but are described as “projections” which enable a qualitative assessment to
be made on the potential direction and magnitude of impacts. It is a truism that the only
definite statement which can be made about the scenarios is that they all have zero
probability as accurate forecasts of the future.
The Environment Agency (2003) guidance on climate change recommends approaches

to include the potential effects of climate change in flood risk management. The
guidance suggests the following categories based on the UKCIP02 high, medium and
low relative levels of confidence in the results of climate model scenarios
· Projection – confidence in the direction and approximate magnitude of change

· Contingency – realistic and consistence national allowance to the upper end of the
range of “possibilities”
· Sensitivity – “What-if” tests but with no confidence in particular scenarios of
change.
Like the Italian Flag concept, this categorisation in scenario assessment provides no
information on the probability of any of the results, but gives a framework for
judgement of the degree of “belief” which may be given to the results.
RASP and flood defence planning
The RASP methodology takes a tiered approach to assessing flood risk (usually
expressed in terms of expected economic damage). The amount of resources invested
in data acquisition and analysis reflect the importance of the decision(s) that the risk
assessment is being used to inform.
A simple, yet explicit method of representing uncertainty at all stages of the flood risk
assessment is applied. The uncertainty is represented by probabilistic description where
appropriate and through the identification of upper and lower “bounds” of the uncertain
quantities where such a detailed description is not appropriate, see Figure 2.8. For
further details on the three levels see DEFRA/EA (2004/2005 in preparation).
upper
central
lower
Water
level or
damage
Flow/ loading
Figure 2.8 Upper and lower uncertainty bands in the RASP analysis
Due to the limited nature of the original high level analysis there is unlikely to be a
mechanism with which uncertainty in conveyance estimation can be incorporated into
the analysis. However, it could make a contribution to the inundation assessment in the
High Level Plus methodology currently under development. Improved understanding

21
and assessment of uncertainty in river conveyance will provide an opportunity to
improve outputs from the Intermediate and Detailed levels of the RASP methodology.
Currently, uncertainty associated with the hydrodynamic modelling process and with
the data used therein is considered very crudely if at all. The RASP intermediate and
detailed level methodologies have been designed not to be limited to specific models.
Improvements to the risk assessment can therefore be obtained with minimal disruption
to the methodology and its data requirements.
Combined Distribution Analysis
Where the quantity of interest is generated from a sum of the factors which are
independent and each have a normal distribution, the combined uncertainty as measured
by the overall variance, σt, can be calculated from:
st = {å s } i
2 1/ 2
2.21
where σi is the standard deviation of the distribution of the i-th factor. However, if the
quantity, F, of interest is a more general function of the factors, yi, which are still
assumed independent with a normal distribution, and the uncertainty in the individual
factors cause only small changes to the quantity of interest, then a first order variation
assessment is feasible. We may define the overall influence of the uncertainties as
approximately
1/ 2
ìï æ æ ¶F ö ö
2
üï
s t = íå çç çç ÷÷s i ÷
÷ ý 2.22
ïî è è ¶y i ø ø ïþ
In this expression all second and higher order terms e.g.
¶2F
s is j 2.23
¶y i ¶y j
and cross-correlations have been ignored. The PDF of F will be approximately a

Normal Distribution in these circumstances. Where F is not a simple linear function
then Equation 2.22 does not hold exactly but is sometimes used as an estimate, its value
as an estimate decreases as the non-linearity in F or cross-correlation between the yi
become more significant.
Despite these theoretical difficulties the approach of Equation 2.19 to the estimation of
the uncertainty in the “result” of a process is recommended in the Environment Agency
(2000) guidance on determination of freeboard. It is also the basis of the work of
Samuels (1995) on combining uncertainties in flood level assessments from model
calibration, numerical error, roughness estimation and topographic survey.
The Monte Carlo approach
This involves determining the uncertainty in water level H from randomly sampling the
uncertainty in the various factors which influence H. The value of H is obtained from

22
taking sample values of each of the parameters that influence it e.g. channel and flood
plain Manning’s n (or each component to the unit Manning’s n), bed level, yi, at each
point across the section (if sediment processes are important) etc. The variability of
each parameter is described through an associated probability distribution function
(PDF) and possibly with cross-correlation relationships. The output H is obtained for a
large number (potentially thousands) of randomly selected samples from each
distribution using cross-correlation to obtain conditional probabilities if the factors are
not independent. The frequency distribution for H is then compiled from the
realisations obtained and possible a new PDF fitted. See Figure 2.9 below.
H = F (nc , n f yi ,...)
Combine many samples to
generate output PDF
Random sample
of each PDF
PDF for H PDF for nc PDF for nf PDF for yi
Figure 2.9 The Monte Carlo approach
The advantage of this method is that an uncertainty distribution is obtained based on the
sum effect of the individual PDFs and the method can be extended to add in the
influence of more factors, such as the variability in the flood flows from hydrological
assessment. The method is rigorous if the PDF is known for each of the input
parameters, and can be used to estimate the variability of the results from assumptions
about the input PDFs where these are not known.
However, the approach is computationally intensive where there are many parameters to
test and if the method is used to assess the variability of the result to assumptions on
unknown distribution. If the approach is set within a larger analysis (such as a
computational river model and incorporating hydrological PDFs) then the method my
not be feasible for practical use.
2.4.5 How is uncertainty represented in the CES?
A key decision in developing advice and estimates of uncertainty was on how the
information should be presented. Several approaches are possible as described above
but some require considerable additional information and analysis to generate the
results. Thus an appropriate starting point was to consider the users of the CES and
how they might incorporate information on uncertainty into their decisions. This led to
some requirements of the method in practice, it should:

23
· express uncertainty in water level
· link to and inform other Environment Agency - Defra risk approaches
· be understandable to the user to facilitate better, more informed decisions in the
situations they are responsible for
· be as simple as possible in presentation
· not require significant computational resources
· ideally not require any additional input data
· be credible by being based on existing scientific methods for uncertainty where
possible.
The potential direct and indirect users of the CES information have different needs for
information on uncertainty in the decisions and plans they need to make. The project
team’s review of needs of different users identified the potential use, importance and
form of presentation of information on uncertainty as shown in Table 2.2 below.
Table 2.2 Summary of Uncertainty users and their needs
User community Use of uncertainty Importance Suitable presentation of

uncertainty
Strategic Planning Evaluating / comparing options High Upper and lower estimates
against criteria such as
effectiveness, efficiency,
robustness, performance
Scheme Design Change in water level for given Medium to high Upper and lower estimates
flow rate (moving towards Potentially probabilistic
probabilistic optimisation for information for a full risk-
describing uncertainty) based design appraisal of a
major scheme
Maintenance Timing, scheduling and Low to medium Upper and lower estimates or
prioritisation of vegetation nothing
cutting and dredging
Flood Forecasting Minimise false alarm and Low in current Possibly an upper estimate of
and Warning missed opportunities practices flood level (and early
Aid operational decisions estimate of arrival time)
Hydrometry Flow rate for a given depth Medium (but Upper/lower estimates
(influences other stakeholders) should be high) Probabilistic
Regulation, Water levels, flood outlines, High Upper/lower estimates
insurance indicative flood risk mapping,
There is a need for quantitative information on uncertainty so that the “Italian Flag”
representation cannot be used in isolation. The approach finally adopted to respond to
the user requirements combined elements of the other approaches described in Section
2.4.4 above. In terms of presenting quantitative information on uncertainty the
important choices are whether to:
· recommend the use of Monte-Carlo testing based

· assign probabilities to the information given on uncertainty.
It was decided not to recommend either of these because:

24
· Monte Carlo testing requires considerable computation, particularly if the testing is
embedded in a large flow simulation model
· the underlying information is not available for the probability distribution of river
resistance values which is required by Monte Carlo resting
· the information on river resistance is not homogeneous in origin
· the effects of residual process model uncertainties are inherently not quantifiable in
any probabilistic manner
· some of the information in the roughness advisor is derived from expert opinion
rather than field measurement.
Thus making statements on probability would give an appearance of confidence in the

quantification of information, which is partly subjective in origin.
However, it is important that the CES includes some quantitative statement on the
effects of uncertainty even if probability (or likelihood or frequency) cannot be given.
This has been achieved by illustrating the uncertainty in the relationship between water
level and discharge through presenting a central estimate and credible upper and lower
values of water level for a given flow rate (Figure 2.10). The stage discharge curve is
usually expected to lie within the upper and lower scenario bands, which may not be
symmetrical about the mean. These bands must not be interpreted as minimum /
maximum envelopes, but rather ‘soft’ boundaries within which the ‘true’ value is
“likely” to occur.
Dh
Q Q
Figure 2.10 Representation of uncertainty for a rating curve
2.4.6 How is the uncertainty calculated in the CES?
The overall uncertainty is based on a pragmatic analysis similar to that of Samuels

(1995). The uncertainty in water level Dhu is given by the “root sum of the squares” of
the uncertainties in water level due to different parameters i.e.
[ 2 2 2 2 2
D h u = D h r + D h s + D h c + D h D x + D h q ..... ] 1
2
2.24

25
where the subscripts are:
r for roughness where little or no calibration data exist
s for survey data depending upon the method of survey
c for calibration and extrapolation away from the calibrated state
Dx for numerical error (grid resolution and physical method)
q for flow estimation (outside the use of the CES)
and … represents any other commensurate factor.
The assumptions behind this type of formulation are that the sources of uncertainty are
mutually independent and that they are roughly normally distributed. The numerical
uncertainty ∆h∆x is small (order of magnitude smaller than the other parameters) and
well controlled because the grid can be refined to ensure this i.e. cross-section spacing.
The flow estimation or uncertainty due to the hydrometric analysis ∆hq is outside the
scope of the CES, but experience shows that this is a key factor in scheme design,
contributing to at least 40% of the overall uncertainty. The remaining factors are thus
∆hr, ∆hs and ∆hc.
2.4.7 What are the dominant factors that are considered?
The dominant factor to be used in the analysis of the estimation of conveyance is the
unit roughness as distributed across the section. A thorough review of the sensitivity of
water level to both the a-priori chosen and calibration parameters identified roughness
as having the most significant contribution to uncertainty in conveyance. The
parameters considered were roughness, elevation of the floodplain and main channel,
cross-chainage, planform sinuosity, temperature, longitudinal bed slope and position of
top-of-bank markers. The uncertainty of the input survey data depends largely on the
method of measurement e.g. measured, estimated, ground survey, River Habitat Survey
(RHS), airborne scanning laser altimetry (LiDAR), aerial photography, Synthetic
Aperture Radar (SAR) etc. The a-priori chosen considered were the dimensionless
eddy viscosity l, the secondary flow coefficient G and the number of lateral divisions
for integration. Other possible sources of uncertainty include non-modelled processes
such as form losses, vertical accelerations through sudden local changes in depth,
energy dissipation from sediment movement over mobile channel beds and lateral
variations in surface water level around bends.
2.4.8 Why was this approach selected?
The approach was selected after careful consideration of the:
· method requirements (Section 2.4.5)

· the available approaches that could be applied within the CES environment
(Section 2.4.4)
· the users and their needs (Table 2.2).
The use of the central estimate with the credible upper and lower values ties in well
with the accounting of uncertainty in the RASP system used in flood defence planning
and also the practice of sensitivity testing usually undertaken as part of scheme design.

26
The concentration of the scenarios on the influence of uncertainty in river roughness
(Section 2.4.7) is justified in terms of the sensitivity analysis undertaken in the testing
phase of the CES research (See DEFRA/EA, 2004/2005 in preparation).
2.4.9 What is the effect of calibration?
Calibration of the flow resistance alters the uncertainty in the estimation of conveyance.
Calibration data for flow resistance should supersede the generic, non-site specific
information in the Roughness Advisor. Calibration may result in a different value for
the central estimate of water level for the calibration flow rate and there is no guarantee
that a single roughness value will achieve a perfect calibration for water levels and flow
rates observed on different occasions (due to natural variability). However, once the
model has been calibrated, ∆hr and ∆hs are likely to decrease, narrowing the range of
uncertainty, see Figure 2.11 but the exact PDFs for the calibrated and uncalibrated cases
will be difficult to determine.
PDF
Reduced uncertainty due to

calibration
Calibrated
Uncalibrated
E(h)calibrated E(h)uncalibrated Water Level, h
Figure 2.11 Effect of calibration
2.5 Backwater
2.5.1 What is a backwater profile?
A backwater profile is present where there is a disturbance to the normal depth of flow.
The effect of the disturbance is felt upstream and the backwater length is the distance
upstream before normal depth is re-established. Typically backwater profiles occur
upstream of structures such as reservoirs, weirs and bridges.
2.5.2 Why do I need to know about backwater effects?
· For determining the upstream influence of works in a river channel. The backwater
length may extend far upstream, potentially causing inundation in areas far
upstream of the study reach. Table 2.3 provides some typical UK backwater
lengths

27
· The backwater length and profile effects the location, number and crest level of
flow measurement structures such as weirs. Flow gauging stations should not be
located within the reach influenced by backwater effects
· The backwater profile is useful for the operation of land drainage pumps. These
pumps are switched on and off when the water level drops below the intake level,
causing increased wear and tear. Thus knowledge of the backwater profile will
ensure that these pumps remain operational for long periods of time.
Table 2.3 Backwater lengths for some UK Rivers (Samuels, 1989)
River Location Depth (m) Slope So Backwater length

Avon Pershore 4.2 0.46 6.90
Cleddau Haverfordwest 2.5 1.70 1.00
Lagan Dunmurry 2.5 1.20 1.40
Lagan Lisburn 3.0 0.29 7.20
Lagan Banoge 2.0 1.30 1.10
Little Ouse Lakenheath 2.5 0.05 35.00
Rhymney Llanedeyrn 2.0 2.20 0.64
Severn Shrewsbury 4.5 0.50 6.90
Severn Tewkesbury 8.5 0.10 59.50
Soar Loughborough 3.5 0.25 9.80
Stour Christchurch 4.5 1.0 3.20
Taff Cardiff 6.0 2.5 1.70
Wharfe Otley 2.0 1.8 0.76
2.5.3 How is the backwater calculated?
The backwater is calculated iteratively, starting from a downstream control or cross-

section of known depth and flow rate; and moving upstream applying energy balances
at each consecutive section. Here, the backwater effect is present until normal depth is
achieved at some backwater length L, upstream from this downstream control. Where
either the flow rate or the flow depth is known, the backwater approach may be used to
calculate the corresponding normal depth of flow or discharge respectively. In this
latter case, the backwater profile is essentially a normal depth profile upstream of this
known quantity.
At the control point the flow Q (m3/s) and depth h0 (m) is known. The downstream
cross-section conveyance for this depth of flow, K0 (m3/s), is estimated with the new
CES calculation method (Section 2.2). The water surface slope S is approximated from,
1
æ Q ö 2
S = çç ÷÷ 2.25
è K0 ø
The section average velocity V0 is calculated from.
Q
V0 = 2.26
A0
The total downstream energy level H0 is determined from,

28
V02
H 0 = h0 + 2.27
2g
The surface water slope is extrapolated to predict the energy level H1 for the upstream
cross-section i.e.
H 1 = H 0 + SD 2.28
where D is the streamwise chainage between the two cross-sections. The upstream
depth of flow h1 is then solved for iteratively i.e.
· guess h1
· calculate the cross-section area A1 (by-product of the CES calculation)
· calculate the velocity V1 (Equation 2.26)
V2
· calculate the energy level H1 from H 1 = h1 + 1
2g
· if H1 (step above) is equal to H1 (Equation 2.28) stop, if not update h1 and repeat.
The conveyance K1 is estimated for this cross-section and the procedure is repeated
moving progressively further upstream (Figure 2.12).
Energy SD
WL V 12
2g V o2
H1 2g
Ho
Bed h
3 h
2
1
0
Figure 2.12 Backwater calculation method from cross-section 1 to 3
2.5.4 What are the Backwater method restrictions?
· The method is for sub-critical flow only

· For sub-critical flow where the channel bed is steep, the calculation may result in a
negative upstream energy head (and hence a negative depth) being calculated
· The method is limited to application in single reaches i.e. no branches, confluences,
junctions, loops or spills into floodplain storage areas
· The method only applies to steady flow conditions i.e. the flow rate is constant with
time.

29
30
3. USING THE CES
This section provides modelling advice for the three core components of the CES: the
Roughness Advisor, the Conveyance Generator and the Uncertainty Estimator, as well
as the use of advanced options, advice on model calibration and interpretation of
conveyance curves with unusual characteristics.
3.1 Roughness advisor
3.1.1 What is the Roughness Advisor?
The Roughness Advisor is a component of the CES software, which provides roughness
advice based on channel descriptions, photographs and grid references. The roughness
values are based on the findings of an extensive literature review (DEFRA/EA, 2003b).
Information is included on surface materials [e.g. natural: bedrock, cobbles, gravel,
sand, silt, clay, peat, earth, firm soil and manmade: sheet piling, stone block, hazel
hurdles, gabion, concrete, rip-rap, wood piling etc], vegetation morpho-types,
vegetation categories due to human intervention [e.g. grass, hedges, trees, shrubs and
crops: wheat, sorghum, sunflower, corn, cotton etc] and irregularities [urban trash,
groynes, pools and riffles, obstructions]. For the vegetation, the expected seasonal
variations are available.
3.1.2 What will the Roughness Advisor provide?
The Roughness Advisor provides total unit roughness values, comprised of up to three
component roughness values: surface material, vegetation and irregularities (Sections
2.1.6 and 2.1.7). Mid, upper and lower total unit roughness values are provided, based
on the mid, upper and lower unit roughness values for each component. This range of
roughness values reflects what is expected within these natural systems. Each total unit
roughness value is assigned to a zone (Section 3.1.4), since cross-sections passing
through plan areas or ‘zones’ with similar roughness characteristics will use the same
total unit roughness value.
3.1.3 What is a roughness zone type?
Each roughness zone has a type. Three roughness zone types have been established
within the Roughness Advisor: bed, bank and floodplain. These are designed to
encourage user awareness of the different roughness components that are typically
found in these regions e.g. crops are more likely to occur on the floodplains than on the
channel banks or bed. These zone types have no influence on the calculation process
and are purely descriptive.
3.1.4 What is a roughness zone?
A roughness zone is a plan area depicting a region of similar roughness characteristics

i.e. similar total unit roughness values. Figure 3.1 shows a plan view of a meandering
channel situated in a reach with straight floodplains. The plan area has been divided
into roughness zones e.g. on the floodplain there are areas with reeds and sand; bushes
and coarse sand; crops and grass which each represent a roughness zone. Similarly, the
main channel can be divided up into however many sections are required to describe the
local roughness properties.

31
A
C
crops
bushes &
coarse sand
reeds & grass
sand
A
C
Figure 3.1 Plan view of roughness zones
3.1.5 Why do I specify roughness zones?
Roughness zones are specified within the Roughness Advisor. The user can specify
different names for each zone e.g. ‘field’ may indicate a zone with grass and sand. The
three component roughness values contributing to the total unit roughness or zone value
can then be selected (further details in CES online help). Naming the roughness zone
e.g. ‘field’ is useful for the Conveyance Generator application (Section 3.2), as the parts
of the reach cross-sections that pass through the ‘field’ zone can be termed ‘field’
roughness. This prevents the need to define roughness zones for each cross-section,
when consecutive cross-sections may have similar roughness values.
3.1.6 How do I get the roughness values in the Roughness Advisor?
To effectively use the Roughness Advisor (RA), it is preferable to have some

information or knowledge of the reach of interest e.g. channel description, photographs,
grid co-ordinates. If a roughness description is available, the user can navigate through
the different roughness components and select the unit roughness values. If a site
photograph is available, this can be used to identify the surface cover by comparison to
the RA photographs. If the grid reference is known, this can be used to obtain expected
vegetation morpho-types.
In the absence of any survey data or channel description, the RA provides advice using
the national data set obtained through the national survey of river habitats (River
Habitat Survey, Raven et al 1998). This survey was based on a 10 by 10 km square grid
and 25 items were assessed at each river cross-section, spaced at 50m intervals, in 500m
long river segments. This enables the user to enter the UK grid reference of the study
reach and receive advice on expected in-channel and bank-side aquatic vegetation.
3.1.7 Can I put my own roughness values in?
The Roughness Advisor enables the user to manually overwrite the roughness values.
This is essential for the calibration process. The values entered should be interpreted as
equivalent unit roughness values, not the all-encompassing Manning n values (Section
2.2.4).

32
3.1.8 Are there minimum and maximum allowable unit roughness values?
The CES allows a minimum total unit roughness of 0.0109. This corresponds to an
equivalent roughness size ks of 0.1 mm. Roughness values which are smaller than this
minimum value are outside the range of practical application in river engineering.
Although the user may overwrite the individual unit roughness values such that the
combined zone value falls below 0.0109, this value will be altered within the CES
software to the minimum value of 0.0109.
There is no upper limit on the unit roughness, however the conversion of the total unit
or zone roughness nl to the roughness size ks (Section 2.3.6) restricts ks to an upper limit
of 12.27m. This large roughness size, which in most instances is greater than the water
depth, can be interpreted as the size of the horizontal eddies generated from large
obstructions e.g. boulders, tree stumps. This large value is extremely unlikely to be
reached in any practical engineering applications.
3.1.9 What are the seasonal variations in vegetation?
The Roughness Advisor provides seasonal variations in the unit roughness for the
vegetation morpho-types (where there is available data). This enables the user to enter
the exact date for which they want the roughness predictions, and receive expert advice
on the expected biomass, percentage cover and roughness values for that date.
3.1.10 How are the cutting regimes used?
The Roughness Advisor provides the facility to implement cutting regimes e.g. the
percentage vegetation cutback and the associated date. It will then advise on the
expected post-cutting growth pattern. This enables the user to consider different cutting
regimes (timing, amount) and their effect on the total conveyance.
3.2 Conveyance Generator
3.2.1 How do I describe the cross-section geometry?
The cross-section geometry is described in terms of the offset distance y (m) from the
left bank and elevation z (mAOD). The offset distance should increase from a
minimum value on the left bank to a maximum value on the opposite (right) bank. E.g.
the cross-sections in Figure 3.2 cannot be solved for within the Conveyance Generator.
The prescribed elevations need not have maximum values at the extreme ends i.e. the
channel banks. The Conveyance Generator introduces two additional co-ordinates at
each end, with identical offsets to the extreme / channel edge values and elevations set
at 1.5 times the maximum elevation relative to the bed level within the cross-section
(Figure 3.3).
For the case of cross-sections with rapidly rising banks on the floodplain edges, where
the maximum bank elevation is far greater than the expected maximum flood level, this
maximum calculation depth may be unreasonably high. This will result in the
conveyance being calculated at high depths, beyond the range of expected water levels,
and fewer conveyance calculation depths in the area of interest e.g. main channel / low
overbank flows. There are three available for improved predictions:

33
· Exclude these high elevation survey points from the cross-section in the analysis.
This will ensure the depth range for the conveyance calculation is spread over the
region where the flow is likely to occur
· Reduce the wall height multiplier (Advanced Options, Section 3.5.11). In this
instance, the value may be set at less than 1.0. It should not be reduced below 0.2
· The number of calculation depths can be increased (Advanced Option, Section
3.5.2).
Figure 3.2 Example of cross-sections that cannot be modelled within the

Conveyance Generator
z
Two additional co-ordinates
set at the channel edges
Maximum elevation or ‘z’ value 0.5 * Maximum elevation
Lateral distance across section, y
Figure 3.3 Setting extreme / edge co-ordinates with maximum values as a

function of maximum depth
3.2.2 How do I represent the roughness zones in my cross-section geometry?
After describing the roughness zones within the Roughness Advisor, the portions of the
cross-sections passing through these zones can be assigned the associated roughness
values. E.g. Figure 3.4 provides cross-sections A-A, B-B and C-C that correspond to
Figure 3.1, indicating the roughness zone with the appropriate shading.

34
A-A
B-B
C-C
Figure 3.4 Cross-sections with the lateral distribution of roughness zones
The extent of the roughness zone within a cross-section is described through provision
of the offset distance at which the roughness zone starts. These offset distances need
not coincide with the survey points, i.e. offset and elevation that describe the section
geometry. This enables the user to have any number of roughness zones within a cross-
section. If a roughness zone is so narrow such that the extent falls within one hundredth
of the total cross-section chainage, this roughness zone may not be identified. The user
can alter the advanced options (Section 3.5) to increase the number of channel divisions
to ensure the roughness contribution is considered in the conveyance calculation.
3.2.3 Does the channel cross-section represent the start or mid-point of the
channel reach?
The conveyance calculation is applied to a cross-section that is representative of a

channel reach. The streamwise location within the reach being represented i.e. start,
mid-way or end, is irrelevant to the conveyance calculation within the CES. Should the
rating curve be implemented within a 1D hydrodynamic model, e.g. ISIS, this will
depend on the cross-section definition according to the reach discretisation within the
respective software.
3.2.4 What are the top-of-bank markers used for?
The top-of-bank markers are used in the conveyance calculation to determine the nature
and extent of the secondary flows. The secondary flow processes are different in the
main channel and floodplain, and they vary with relative depth. The energy loss
coefficients are allocated according to the flow regions (Section 2.3.5 and 2.3.6), and
these can only be identified through top-of-bank markers. Figure 3.5 illustrates the
variation of the secondary flow term in a straight two-stage channel. For meandering
channels, see Section 3.2.7.
Gfp Gmco Gfp
Gmci
Figure 3.5 Variation of secondary flow term for straight channels

35
Where the top-of-bank markers are not at the same height, the conveyance calculation
approximates the cross-over point for the main channel secondary flow term from the
average elevation of the two floodplains (Figure 3.6).
Gmco Gfp
Gfp
trans. Gmci®Gmco
Gmci
Figure 3.6 Variation of secondary flow term for floodplains with different
elevations
3.2.5 Do I have to use top-of-bank markers?
Top-of-bank markers must always be used.
3.2.6 Where should top-of-bank markers be positioned?
The placing of top-of-bank markers is partially subjective. They should be placed in the
part of the channel bank that is considered the intersection between the main channel
and floodplain. Top-of-bank markers are always placed in pairs. There should never be
an odd number of top-of-bank markers in a given cross-section. Advice is provided for
special cases.
No floodplains present
In the instance where only a main channel is present i.e. no floodplains, the top-of-bank
markers should be placed at the channel edges to indicate that the region between these
markers has secondary flows that are characteristic of inbank flow. If a floodplain is
not connected to the main channel, and is therefore being treated as a separate channel,
the top-of-bank markers should be placed similarly to those of a main channel. The
secondary flows will not be characteristic of those that are found on floodplains, as
there is no main channel - floodplain interaction. Figure 3.7 provides an example of
what is and is not considered acceptable.
Gmci
4
Gfp
7
Gmco
7
Gmci
Figure 3.7 Positioning of top-of-bank markers for single channels and isolated
floodplains

36
Asymmetric channels
For asymmetric channels, the top-of-bank marker on the side with the floodplain is
positioned in the usual manner. For the opposite bank, it is suggested that this top-of-
bank marker be placed at the channel edge (Figure 3.8). This will result in the main
channel secondary flows changing from inbank to out-of-bank flows mid-way between
the floodplain height and the opposite bank. This is suggested, as the presence of one
floodplain should only generate 50% of the secondary flows typical of channels with
two floodplains.
trans. Gmco®Gmci Gfp

4
Gmci
Gmco
trans. Gmco®Gmci
Gfp 7
Gmci
Gmco Gfp 7
Gmci
Figure 3.8 Positioning of top-of-bank markers for asymmetric channels
Multiple berms
Channels with multiple berms have more complex flow patterns. The user should make
a subjective decision based on the typical bankfull water depth in the channel to
determine where the top-of-bank markers should be placed. Figure 3.9 provides
suggested placements for (i) a main channel berm, (ii) a main channel with berms and a
distant floodplain, and (iii) a floodplain berm.
4
Gmci
Gfp trans. Gmco®Gmci

4
Gmci
Gfp trans. Gmco®Gmci 4

Gmci
Figure 3.9 Positioning of top-of-bank markers for channels with berms

37
Floodplains with a lower elevation than the main channel banks
For channels with elevations lower than the main channel banks, the conveyance
capacity of the floodplains should ideally be ignored until spillage occurs. The
conveyance calculation is designed to consider all portions of the cross-section at a
given depth and hence this does not occur. If the cross-section is treated as a whole
(Option 1, Figure 3.10), the top-of-bank markers should be placed at the top of the main
channel banks or ‘spills’. The results should be used with caution, as the conveyance
capacity for depths below this spill depth will be over-estimated due to the additional
floodplain capacity. An alternative is to model the floodplains as separate channels, and
only add the floodplain conveyance to the main channel conveyance above the spill
depth (Option 2, Figure 3.10). For 1D-river modellers, a further alternative is to model
the floodplains as storage, e.g. “reservoir” units in ISIS, with the main channel banks as
lateral spill units (Option 3, Figure 3.10). For all three options, the placing of the main
channel top-of-bank markers is similar.
Option 1
Gfp Gmco Gfp (4)

Gmci
Option 2
Floodplain channel Lateral spills Floodplain channel

Gmco
4
Gmci
Option 3
Lateral spills Floodplain storage

Floodplain storage
Gmco
4
Gmci
Figure 3.10 Top-of-bank markers for low elevation floodplains
3.2.7 How do I model meandering channels?
When modelling meandering channels, the key difference is the nature and orientation
of the secondary flows. The user specifies the orientation of the meander bends through
the use of inside of bend (I) and outside of bend (O) top-of-bank markers, as well as
providing the planform reach-averaged channel sinuosity. This information is necessary
for the conveyance calculation, as it provides the extents within which the Cuv
meandering coefficient (Section 2.3.6) will operate (blocked areas in Figure 3.11). The
nature of the bend will influence the skew direction of the velocity profile; i.e. the
velocity is skewed towards the outside of the bend for inbank flow, towards the inside
of the bend for overbank flow, and it dips downwards in the cross-over region between
consecutive bend apexes (Figure 3.11).

38
Ud
I O AA
AA
A A I O
BB
O I
B B
C BB
O I
C
CC
I O
Distance across section
Figure 3.11 Typical observed depth-averaged velocity profiles in meandering

channels and top-of bank bend marker positions
Cross-sections in meandering reaches should ideally be taken at the bend apexes, or as

close to them as possible. The top-of-bank markers are then similar in position to the
straight top-of-bank makers, but additionally labelled inside or outside of bend. If a
cross-section is situated at the cross-over point, and thus the inside / outside of bend is
not apparent, the section should still be modelled as a meandering section and not a
straight section, since the meandering effects are still present downstream of the bend.
The cross-over point should assume the inside / outside of bend labels of the previous
bend apex. For detailed studies of velocity profiles, the closer the meandering cross-
section to the bend apex, the more accurate the resulting velocity predictions.
3.2.8 How do I label top-of-bank bend markers on skewed channels?
For channels with long reaches which are skewed relative to the floodplain (Figure
3.12), the channel should be modelled as a meandering reach, as the sinuosity value is
greater than 1.0 (Section 3.2.9). The inside / outside of bend markers should be labelled
as downstream of a bend i.e. similarly to cross-over regions (Section 3.2.7). In Figure
3.12, if the skewed reach were downstream of a bend, the orientation of would have to
be such that the inside of bend (I) was on the left and the outside of bend (O) was on the
right.

39
Ud
D D I O
E E I O
Distance across section

Figure 3.12 Top-of-bank markers for skewed channels
3.2.9 How do I calculate the channel sinuosity?
The channel sinuosity s is a measure of the degree to which a channel meanders. It is

defined here as the thalweg length TL over the valley length VL i.e. the length of the
channel relative to the length of the floodplain (Figure 3.13). A sinuosity of 1.0
indicates a straight channel, and a sinuosity greater than 1.0 indicates a channel with
some degree of meandering. A value less than 1.0 is meaningless. The sinuosity should
be calculated on a bend-by-bend basis.
TL
sinuositys =
VL
Radius of curvature
Meander / Meander
Valley length VL amplitude
Thalweg length TL
Figure 3.13 Definition sketch for sinuosity
3.2.10 What are typical UK channel sinuosities?
The sinuosity of UK channels varies between 1.0 (straight) and 2.0 (heavily
meandering). Channels of sinuosity greater than 2.0 are highly unlikely. Channels with
sinuosity greater than 2.5 are not catered for in the CES, as this is outside the range of
any known UK (or other) river. Figure 3.14 provides some examples of different
sinuosity values.

40
(a) s = 1.37 (b) s = 1.57
(c) s = 1.8 (d) s = 1.4
(e) s = ~2 (f) s = ~2
Figure 3.14a Sinuosity values for (a) the Flood Channel Facility at
HR Wallingford, (b) the Glasgow University small-scale flume, (c)
the River Dane in Cheshire, (d) Kleinschmidt Creek (Photo courtesy
of Land and Water Consulting), (e) incised meanders, Utah and (b)
Wood River, Southern Nebraska

41
Sinuosity:
1.0930
1.1107
1.3700
1.5710
1.7723
Figure 3.14b Plan view of channels with increasing sinuosity
3.2.11 How do I model braided channels?
The conveyance calculation handles multi-thread or braided river sections through

treating each parallel channel as a separate unit and evaluating the flow for each. These
individual channel flows can hence be summed to find the total flow rate (Figure 3.15).
The assumption is that the parallel channels have similar bed/friction slopes and
temperature for a given water surface elevation. The top-of-bank markers are placed as
outlined in Section 3.2.6, with each braided section treated as an individual channel
(e.g. Figure 3.16).
Q6 Qtop depth = Q6
Q4 Q5
Qmid-depth = Q4 + Q5
Qbottom depth = Q1 + Q2 + Q3
Q1 Q2 Q3
Figure 3.15 Addition of flow in braided channels (DEFRA/EA, 2003a)
Island
Gmci Gmci®Gmco Gfp
4
Gmci
Figure 3.16 Top-of-bank markers for a braided channel
The top-of-bank markers are always placed in pairs. For multi-thread channels with
deep flows, there may be multiple top-of-bank markers acting in a given cross-section
(Figure 3.17).

42
Gfp Gmco Gfp Gmco Gfp
4
Gmci Gmci
Figure 3.17 Top-of-bank marker positions for deep flow in a braided channel
3.2.12 How do I model braided channels with different sinuosity values?
Braided channels with different sinuosities are modelled in a similar manner to that
outlined in Section 3.2.11 with the additional requirement of specifying the orientation
for the bend i.e. inside (I) or outside (O) of bend (Figure 3.18).
common floodplain
Figure 3.18 Top-of-bank bend markers for multi-thread channels with different
sinuosities
For the combination of straight and meandering braids, the top-of-bank markers are
labelled intuitively (Figure 3.19), with top-of-bank bend markers on the meandering
portion and top-of-bank straight markers on the straight portion. The secondary flows
on the common floodplain area between two adjacent braids of different sinuosity are
based on an average effect from both the adjacent channels.
common floodplain
S
Figure 3.19 Top-of-bank bend and straight markers for multi-thread channels
with different sinuosities

43
3.2.13 How do I model non-conveyance storage zones i.e. zones with re-circulation
or stationary water?
Non-conveyance zones are indicated in the software by replacing the roughness zone
with a non-conveyance marker. As with the roughness zones, the extent of the non-
conveyance zone is given by the offset at the start of the zone. The conveyance
calculation, i.e. the solution of the unit flow rate q at each point in the cross-section, will
be carried out in the usual manner with a high unit roughness in the non-conveyance
portions. For the evaluation of the total cross-section flow rate Q, only the sections
outside of the non-conveyance portions are included. The result is that the non-
conveyance areas are effectively storage volumes that exert some resistance on the
principal flow, but do not contribute to the section conveyance.
Entire channel sections cannot be defined as non-conveyance zones since this implies
complete blockage / storage. Non-conveyance zones cannot be specified adjacent to
each other. In this instance, a single, larger non-conveyance zone should be specified
(Figure 3.20).
1 2
7
Non-conveyance zones
single 4
unit
Figure 3.20 Non-conveyance zone specification – single units only
If a non-conveyance zone is sufficiently narrow such that the extent falls within one
hundredth of the total cross-section chainage, this non-conveyance zone may not be
identified. The user can alter the advanced options (Section 3.5) to increase the number
of channel divisions to ensure the non-conveyance contribution is considered in the
conveyance calculation.
3.2.14 What is the range of depths for which the conveyance is calculated?
The conveyance is calculated for 25 evenly distributed depths, ranging from the lowest
cross-section elevation to the highest cross-section elevation. The latter is determined
as 1.5 times the maximum depth above the lowest elevation (Section 3.2.1). The
minimum depth and number of depth intervals can be altered in the advanced options
(Section 3.5.2 and 3.5.10).
3.2.15 Are there any special survey requirements for application of the CES?
Existing practice promotes the cross-section surveys at cross-over sections for

meandering channels. In the CES, the conveyance predictions are more accurate than
current practice at bend apexes. It is therefore suggested that surveys be taken at bend
apexes or near to bend apexes where possible.

44
3.2.16 What slope do I use?
The conveyance calculation requires an estimate of the reach-averaged longitudinal

friction slope. This can be approximated from the water surface slope if measured data
is available, or alternatively from the bed slope as measured on Ordnance Survey maps.
The slope is incorporated in the equation to evaluate the total flow rate Q (m3/s). This
flow rate is then divided by the square root of this slope S to provide the conveyance K
(m3/s), i.e. rearranging Equation 2.1 gives:
Q
K= 1
3.1
S 2
There is thus a weak dependence of conveyance on slope. Testing (DEFRA/EA, 2003a)

has shown that provided the slope is approximated within 100% (i.e. a factor of two) of
the actual value, the predicted conveyance will fall within 3% of the actual value.
3.3 Uncertainty Estimator
3.3.1 How is uncertainty represented in the CES?
The uncertainty in water level for a given flow rate is represented through upper and
lower bands. The stage discharge curve is usually expected to lie within these upper
and lower bands, which may not be symmetrical about the mean. These bands should
not be interpreted as minimum / maximum envelopes, but rather ‘soft’ boundaries
within which the ‘true’ value is likely to occur.
3.3.2 How should this uncertainty information be interpreted?
The central estimate represents the “best” assessment without any particular definition
of “best”. It is important to note that the upper and lower values
· are not necessarily equally distributed about the central estimate (i.e. not just ± x%)
· are not upper or lower bounds (i.e. it is possible for some real cases to lie outside
the range of the scenarios)
· are not for any specific confidence limits (e.g. ± 95%)
· are not simply related to the standard deviation of measured data.
Instead they represent a practical choice of a degree of variability to include in the

analysis for flood risk management applications, based upon the range of values found
in the compilation of data on river roughness.
Indeed, the variability of modelling results is the primary driver for the emerging
culture of understanding uncertainty and undoubtedly synonymous with it. This culture
is at its early stages and as such the methodologies have not yet been translated to more
reliable design procedures. Consider the following cases:
· There is a close analogy between travel time of a journey between the points A and
B. If all the factors influencing the journey can be controlled, the travel time can be
estimated accurately. In reality, this time can be affected by a wide range of
factors, including the number of crossings between the points, rush-hour traffic, the

45
weather conditions, etc. Undoubtedly, travel times can be managed better by a
matrix of different journey times in terms of daily, weekly and seasonal cycles,
versus the probability of different factors affecting the travel time at each
subdivision of these cycles. Each one of us builds a tacit knowledge of travel times
of our journey and manages our journeys accordingly. Understanding variability in
estimated conveyances, i.e. their inherent uncertainties, is currently analogous to
the variability to travel times in the analogy. Knowing the variability of
conveyance in quantitative terms is a new culture and will undoubtedly influence
decision-making.
· Civil engineers are well familiar with elastic and plastic design of structures.
Ignoring plastic designs will result in inefficient designs but plastic designs require
a more thorough understanding of the properties of design materials. Traditional
rule-of-thumb selections of roughness values followed by deterministic modelling
are analogous to elastic design procedures and neglecting plastic designs.
However, the emerging culture of uncertainty is analogous to understanding plastic
design procedures but not using the potentials of plastic design. Undoubtedly, there
is a learning period ahead of us.
3.3.3 How should this uncertainty information be used?
Guidance on the interpretation and use of the uncertainty information will vary for
different user groups. Table 3.1 provides an initial approach for incorporating this
information into the decision making process. As there is little previous knowledge or
experience in the use of this information, this should be considered as “suggested
advice” rather than strict guidelines.
Table 3.1 Use of uncertainty information for different user groups
User community Calibration data available No calibration data available

Strategic Planning Where data is available, the bands are Carry out calculations with upper and
likely to be small. The main planning lower estimates. If the bands are very
should be based on the central estimates. wide, i.e. where cost decisions would be
different for the upper or lower band
compared to the central estimate, use the
central estimate but devote additional
resources to understanding the cause of
uncertainty and hence potentially reduce
it.
Scheme Design New schemes should always be based Examine the cost / design implications
on a calibrated model. In general, more for the upper, lower and central
data implies narrower uncertainty band estimates. Undertake a cost-benefit
widths and hence reduced freeboard analysis to understand the potential
design. impact of the site roughness.
Maintenance Maintenance users are likely to have For the case of no data and hence wide
detailed local knowledge of the local bands, the seasonal model should be
vegetation growth and cutting regimes. considered over one or two seasons to
Potentially, they should able to improve establish where maintenance is critical
/ update the existing growth models to to flood levels.
suit their local region and thus reduce
the associated uncertainty. In addition,
local knowledge of seasonal variations
may inform as to curve shifts e.g. late /
early summer growth.

46
Table 3.1 Use of uncertainty information for different user groups (continued)
User community Calibration data available No calibration data available

Flood Forecasting Examine the sensitivity of flood wave
and Warning speed to river resistance using a flood
routing model such as that which is
available in iSIS Flow. For the online
system, base this on the central estimate.
Additionally undertake an offline
analysis / test to examine the uncertainty
and ascertain where the flood peaks
change significantly with channel
roughness.
Hydrometry Hydrometry users generally have
detailed water level and flow
measurements providing high quality
calibration data. They can therefore
provide localised resistance calibrations
to narrow the potential uncertainty bands
as well as using this knowledge to
advise other users on the potential
widths to which these bands may be
narrowed. They should refer to the
report on extension of rating curves for
extending these calibrated curves for
predictions of flood levels.
Regulation, Where data is available, the bands are For flood risk mapping, need to review
insurance likely to be small. the information to decide whether to
publish the central estimate or the more
conservative option. For both options,
an appropriate description should be
used e.g. suggest that the user obtain
more local information to improve
estimates. I.e. indicate where there is
scope to improve the results through the
use of new data / information. In
addition, inform the Agency
Development Control that the approach
used is a “broad brush” indicative flood
plain mapping rather than detailed
accurate modelling.
3.3.4 How ‘certain’ is the uncertainty calculation?
The uncertainty calculation is based on the upper / lower roughness values provided in
the Roughness Advisor. These values are based on expert advice from a multi-
disciplinary team of specialists in aquatic vegetation and surface roughness. The
calculation provides some indication of the confidence in the predicted water level, thus
reducing the uncertainty previously associated with these predictions. This calculation
is never entirely ‘certain’ or correct.
3.3.5 How do I ensure the uncertainty is calculated?
The CES calculates the uncertainty for all model runs. No additional user information
is required, as the uncertainty is based on the upper / lower roughness values provided
by the Roughness Advisor.

47
3.4 Calibration and interpretation of results
3.4.1 How do I calibrate my model and what do I change when the data does not
fit?
Model calibration is fairly subjective. It depends on the channel characteristics (simple

/ compound, straight / meandering), the available information, the quality of the data,
availability of stage-flow calibration data and the influence of backwater effects.
For calibration within the CES environment, it is suggested that the internal model
parameters such as the dimensionless eddy viscosity l and the secondary flow
parameters Gmci, Gmco and Gfp are not altered. These parameters have been subjected to
rigorous testing and are set at the optimum performance values.
The calibration in straight channels is therefore based on altering the roughness zones
i.e. roughness magnitudes and extents; and altering the positioning of the top-of-bank
markers to best capture the bankfull secondary flow transition characteristics. In
meandering channels, in addition to the above, the magnitude of the sinuosity can be
altered within a reasonable range for the site concerned.
The water levels should be most sensitive to variations in roughness, while the top-of-
bank markers and sinuosity magnitude effect the bankfull depths and main channel flow
respectively.
If calibration is not achieved, other alternatives include re-evaluating and checking the
input data e.g. longitudinal reach-averaged bed slope and survey data.
3.4.2 Why do I get jumps in the conveyance curve at sudden geometry changes
e.g. bankfull?
At bankfull, the secondary flow model changes from representing the secondary
circulations characteristic of inbank flow, to those characteristic of out-of-bank flow
(Section 2.3.6). The energy transfer mechanisms change and in some instances this
results in a ‘jump’ in the stage-discharge curve at bankfull (e.g. Figure 3.21). This
physical phenomenon can be observed in real channels e.g. River Severn at Montford
Bridge (DEFRA/EA, 2003a). Varying the top-of-bank marker positions can alter the
‘jump’ level.

48
8.0
7.5 CES prediction

Measured data
10
7.0
Stage (m)
Bed level (m)

6.5 6
6.0 2
0
-10 10 30 50 70 90 110 130 150
Lateral distance across channel (m)

5.5
150 200 250 300 350 400
3
Discharge (m /s)
Figure 3.21 Unusual ‘jumps’ in stage-discharge curves reflect observed data e.g.
River Severn, Montford
3.4.3 Why do I sometimes get large jumps at bankfull depth for high sinuosities?
At bankfull, the meandering secondary flow model alters from the inbank model to the
out-of bank model. This is to account for the change in expected skew of the velocity
profile and the change in magnitude of the associated energy losses (Section 2.3.6). The
curve may have sharp reductions in conveyance which have been observed in the Flood
Channel Facility Phase B stage-discharge observations (Figure 3.22).

49
0.30
CES Prediction
Measured data
0.25
0.35
0.20
Depth (m)
0.30
0.25
0.15 0.20
Bed level (m)

0.15
0.10
0.10 0.05
0.00
0 1 2 3 4 5 6 7 8 9 10
Lateral distance across channel (m)

0.05
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
3
Discharge (m /s)
Figure 3.22 Discontinuity in stage-discharge curves in meandering channels

reflect the observed data e.g. Flood channel Facility Phase B20/21
observations
3.4.4 Why are the uncertainty curves not symmetrical about the expected value?
The uncertainty prediction is based on the roughness upper / lower estimates. These
values are as observed in nature and are not always evenly distributed about the
expected value.
3.4.5 Why is the uncertainty greater at small depths?
The conveyance calculation considers the variation of roughness with depth. The unit
roughness value or equivalent roughness size ks, is large relative to the water depth H at
low flows, i.e. a high ks/H ratio. The change in water level due to a small change in
roughness is therefore large. This directly effects the uncertainty, which is based on
channel roughness.
3.5 Advanced Options
3.5.1 Do I need to alter the advanced options?
For general use, the advanced options need not be altered.
For specific / research cross-section analysis e.g. velocity, bed shear stress profiles at a
particular depth of flow, the advanced options may be implemented to improve local
parameter variations and energy loss representation. It is strongly advised that Section
2.3.6 be consulted prior to this.

50
For the advanced options, the more widely used options are available in the user
interface and the less widely used options can be accessed in the source code (Table
3.2). These are described in more detail in the remainder of the section.
Table 3.2 Allowable ranges for the advanced options
Advanced Options Default value Allowable range User Source

Interface Code
Temperature 15°C >0 4
Number of depth intervals 25 100 4
Minimum depth used in Lowest bed Any depth below maximum 4
calculation elevation entered elevation
Lateral main channel eddy 0.24 0.1-0.5 4
viscosity lmc
Number of vertical segments 100 500 4
used in integration
Relaxation parameter 1 0.5 £ Relaxation £ 1.5 4
Convergence tolerance 0.001 0.001 – 1.000 4
Maximum number of iterations 20 50 4
Wall height multiplier 1.5 10 4
Experimental flume Off On / off 4
Secondary flow main channel 0.05 ± 0.25 4
inbank coefficient
Secondary flow main channel 0.15 ± 0.25 4
out-of-bank coefficient
Secondary flow floodplain -0.25 ± 0.25 4
coefficient
3.5.2 What effect will increasing the number of depths have?
The conveyance is calculated at 25 equally spaced horizontal divisions and is thus not
related to the cross-section geometry or user input co-ordinates. If more depth intervals
are required e.g. to fully capture the bankfull effects, these can be set in the advanced
options. Increasing the number of depths will increase the computational time.
3.5.3 What effect will increasing the number of lateral divisions have?
The lateral integration includes 100 equally spaced elements regardless of the geometry
and the number of user input co-ordinates. For multi-thread channels, each braid
includes 100 elements. This number is based on the testing of multiple channel
geometries, to determine the minimum number of elements necessary for evaluating the
total flow rate Q within a given accuracy. At 100 elements, the solution of Q is within
0.5% of the solution achieved with 1000 elements.
The number of integration elements can be altered in the advanced options, e.g. if the
section is particularly wide or more detail is required for the local unit flow rate, shear
stress, velocity etc.
The facility to weigh the lateral distribution of these elements is available e.g. increased
number of panels for rapid variations in local geometry. This option involves altering
the source code.

51
3.5.4 What happens if I change the dimensionless eddy viscosity?
The dimensionless eddy viscosity l represents the amount of lateral spread due to
boundary or shear layer effects. In the literature (Knight and Shiono, 1990; Knight and
Shiono 1991; James and Wark, 1992), values are typically in the range 0.07
(experimental facilities) to 0.5 (natural channels). The model is based on the main
channel value, lmc, which can be altered from the default 0.24 in the advanced options.
Values less than 0.1 may cause convergence difficulties. Values greater than 0.5
represent a large amount of lateral momentum exchange, and are highly unlikely.
It is advised that this parameter is not altered other than for academic purposes.
3.5.5 What happens if I change the straight secondary flow coefficients?
Altering the straight secondary flow coefficients, Gmci, Gmco and Gfp (Section 2.3.6)
requires access to the source code. The default values are 0.05, 0.15 and –0.25. These
can be altered within the range ±0.25. The effect is to uniformly increase / decrease the
energy losses in the appropriate flow region (Section 3.2.4).
3.5.6 What happens if I alter the temperature?
Altering the temperature from the default 15 °C will have little impact. For natural
cross-sections, the CES analysis does not use the temperature. For experimental flumes,
the temperature is used to determine viscosity n for the Colebrook-White equation, i.e.
(Raudkivi, 1998),
n = ( 1.741 - 0.0499T + 0.00066T 2 ) ´ 10 -6

3.2
3.5.7 What happens if I alter the relaxation parameter?
The default relaxation parameter c is set at 1.0. This value may be varied between 0.5<
c £1.5, and always less than 2.0. The relaxation is used to update the correction
parameter Dq in the iteration process,
q n +1 = q n + cDq n 3.3
where n is the iteration number i.e. it affects the value of successive approximations to
the actual solution. The relaxation should never be zero, as this will eliminate the
iterative procedure i.e. successive q guesses will be identical (qn = qn+1). A negative
relaxation is not advised, as this eliminates the sign of the suggested correction from the
previous iteration e.g. if Dq = –0.001m2/s, and the relaxation is –1, this implies a
correction of +0.001m2/s, which will not support convergence. Altering the relaxation
will affect the number of iterations taken to reach the solution. The relaxation does not
affect the results i.e. the actual water level predictions.
The relaxation is altered internally for the rare case where Dqn < 0 and |Dqn| ³ qn , to
ensure that the updated qn+1 is positive. This is critical for numerical stability, as the
unit flow rate appears within the log function in the Colebrook-White friction law. It is
also physically meaningless as it implies upstream flow.

52
3.5.8 If I alter an advanced parameter, does the change apply to that cross-
section only?
Yes. The software is designed such that the advanced parameters for each cross-section
can be described and applied individually.
3.5.9 What happens if I set the minimum depth?
The minimum depth enables the user to enter a specific depth for the first and lowest of
the 25 depths that are solved for. The range of calculation depths will extend from this
minimum depth to the maximum depth (defined Section 3.2.1).
3.5.10 Why would I set the minimum depth?
This is a useful tool if the user is interested in the flow parameters for a specific depth.
The default calculation always provides the results for 25 equally spaced depths that
may not coincide with the depth of interest.
3.5.11 What does the wall height multiplier do?
The wall height multiplier represents the amount that the maximum cross-section depth
is multiplied by to determine the maximum depth for calculation (Section 3.2.1). The
default value is 1.5. This can be altered to any number greater than 1.0. A value greater
than 1.5 would imply that very large flows and depths are anticipated. If both low and
very high flows are of interest, then both wall height multiplier and the number of depth
intervals should be increased.
3.5.12 What happens if I alter the maximum number of iterations?
The default maximum number of iterations is 20. This represents the maximum
allowable iterations of Equation 2.6a and b to solve q within the required tolerance of
0.001m2/s. The solution is designed to converge nearly quadratically, and is therefore
unlikely to take more than 5-7 iterations. In rare instances, e.g. sections with rapidly
varying local roughness, the solution may take longer to converge. Increasing the
number of iterations to 25 or 30 may marginally improve the results.
3.5.13 What does the experimental flume option change?
For modelling experimental flumes, this option should be selected. The Conveyance
Generator will then use the full Colebrook-White equation (Section 2.3.6), i.e. both
smooth and rough laws, for determining the lateral distribution of f. The distribution of
f is thus dependent on both the roughness size and the localised velocity, as is expected
for laboratory flume flow.
3.5.14 What does the convergence tolerance change?
The convergence tolerance is used in the iterative calculation for determining how close
the approximation to the unit flow rate q (m2/s) is to the previous iteration. Once the
required tolerance is achieved (default 0.001 m2/s), the solution is assumed to have
converged. In rare cases where the river is extremely wide, this tolerance may never be
achieved. It is thus possible to increase the tolerance up to a value of 1.0 to ensure a

53
solution is reached. Increasing this tolerance for channels with standard dimensions
will result in poor model predictions.
3.6 Backwater Module
3.6.1 How is the backwater module used?
Once the roughness zones and two or more cross-sections in the reach have been
described, the backwater ‘tab’ should be selected. Here, the absolute or relative
chainage information may be entered together with the downstream known flow depth
and / or water level. Selecting the ‘Long section’ tab will initiate the backwater
calculation.
3.6.2 What Backwater options are available?
The option is available to omit the kinetic energy term. This is equivalent to a simple
Manning equation, as currently used by the spreadsheet-based calculation for the
Internal Drainage Boards. The only difference here is that the conveyance K (m3/s) is
based on the new CES calculation.
Where levels are known at cross-sections located within the reach as well as the
downstream end, the reach may be divided into two sections and the calculation is re-
started at the split between the two reaches i.e. the control point.
3.6.3 What is the advantage of using the backwater module compared to a

complete 1D modelling software such as iSIS or MIKE 11?
The backwater calculation is useful for assessing backwater effects in simple channel
reaches with no detailed structures such as reservoir storage, dual bridges, complex
weirs etc. Typical application would be for drainage channels subject to steady flows,
where the conveyance capacity is essentially dominated by roughness, and there is a
downstream section of known flow rate and / or water depth. The backwater module
provides the facility to approximate the backwater profile without the use of these more
complex software packages.

54
4. WORKED EXAMPLES
This section provides three worked examples of the CES application. For each
example, a cross-section in the channel reach is selected and the CES is used to produce
a rating curve. These channels cover a range of cases, which include:
· A straight compound channel with some vegetation and substrate: The River Main
· A meandering compound channel with seasonal variation in roughness (due to
cutting/ wash out): The River Blackwater
· A straight channel with wide floodplains: River Dane.
The intention of these worked examples is to provide a hands-on application of the CES
software to a real river case, rather than a ‘button clicking’ guide. Thus only the first
example provides detailed screen shots.
4.1 Worked example of the River Main, Northern Ireland
4.1.1 Available user information
The first step in analysis is to collate all the available information e.g. channel
description, survey data, site photographs and any other relevant material.
The River Main is located in County Antrim, Northern Ireland. It rises in the Antrim
Plateau about 6km south-east of Ballymoney, County Antrim. It flows south for the
whole of its 50km course through Culleybackey, Galgorm and Randalstown to Lough
Neagh where it discharges at Mainwater Foot. The 800m study reach is in the Galgorm
area. Between 1982 and 1986 the channel was reconstructed and realigned to form a
double trapezoidal channel from Lisnafillan Weir to the junction with the Braid River.
The final dimensions include a top width of 14m, a total width inclusive of floodplains
of 27.3-30.4m, and bankfull depth ~0.9-1.0m. The floodplains slope towards the main
channel with a 1:25 gradient. The reach-averaged longitudinal bed slope is 1:520.
The river bed consists of coarse gravel with a d50 of 10-20mm. The main channel side
slopes consist of quarry stone / rip-rap (0.5 tonne weight, 100-200mm size) and the
berms are covered with heavy weed growth (Figure 4.1). The cross-section for this
example is taken in a straight section of the reach, near Bridge End Bridge (Figure 4.2).
(a) (b)
Figure 4.1 River Main in County Antrim, Northern Ireland (a) the side-banks
with rip-rap and (b) a gentle meander

55
0 50 100m
Bridge End
Bridge
Factory
Figure 4.2 Plan view of the River Main indicating the selected cross-section 14
The offset (y) and elevation (z) for cross-section 14 is given in Table 4.1 and the cross-
section geometry is illustrated in Figure 4.3. The bed level 0.0 position corresponds to a
datum of 36.39mAOD.
Table 4.1 River Main cross-section 14 geometry
Offset (y) m Elevation (z) m

0.00 4.54
5.30 1.41
13.5 0.92
14.4 0.00
26.6 0.00
27.6 0.97
35.7 1.38
40.8 4.89

56
5.0
4.5
4.0
3.5
Bed elevation (m)
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0 5 10 15 20 25 30 35 40 45
Lateral distance across channel - offset (m)
Figure 4.3 River Main cross-section 14 geometry
4.2 Using the Roughness Advisor to generate roughness zones
Once the available data has been collated, the CES application can occur. Launch the
CES software and select the file / new roughness file option (Figure 4.4). Save the
roughness file with an easily recognisable name e.g. River Main.
Figure 4.4 Create a new roughness file
Select and name the number of zones that the channel reach is likely to have. Consider
Figure 4.1. The floodplain/berm vegetation appears uniform and can therefore be fully
described by one roughness zone. Call it ‘floodplain cover’. Similarly, the channel
banks have a uniform cobbled roughness, call it ‘bank cover’. The channel description
mentions that the channel bed consists of coarse gravel with a d50 size of 10-20mm.
Call this ‘bed cover’. Three potential roughness zones have been identified and these
can be named in the RA (Figure 4.5).
For the ‘bed cover’ roughness zone, select the properties tab.

57
Figure 4.5 Create three roughness zones: bed cover, bank cover and floodplain
cover
Give the roughness zone a description, e.g. coarse gravel (Figure 4.6) and then select the
components tab.
Figure 4.6 Providing roughness zone description: Coarse gravel
This tab provides the three roughness components that may contribute to the total unit
roughness for that zone. The fields for these zones will be the default zero, as they have
yet to be populated. At this stage, there is no information regarding inbank vegetation
or irregularities for the River Main. Select the bed material option and ‘edit component’
(Figure 4.7).

58
Figure 4.7 Three roughness component options
The bed material options will appear (Figure 4.8). The categories for gravel include
‘gravel’ and ‘coarse gravel’, and each is accompanied by a suggested sediment size and
a photograph. In this instance, the inbank sediment size of 10-20mm for the River Main
gravel is similar to that of the ‘gravel’ rather than the ‘coarse gravel’ option.
Figure 4.8 Bed material component with gravel selected

59
Select the ‘gravel’ option and the ‘Bed material’ line should now be populated with a
unit roughness value, as well as minimum and maximum values (Figure 4.9). The total
unit roughness is also provided. As there are no vegetation or irregularity component
roughness values present, the ‘Bed Cover’ roughness zone is now complete.
Figure 4.9 Bed material component now populated
The Bank Cover roughness zone can now be selected. This zone can be described as
‘rip-rap’ (Figure 4.10). As with the previous roughness zone, only the bank material is
selected. In this instance the component roughness is that of rip-rap and this can be
readily selected from the bank material options (Figure 4.11). The bank material option
for this zone should now be populated.
Figure 4.10 Providing roughness zone description: Rip-rap
Finally, the ‘Floodplain Cover’ roughness zone can be edited. For this roughness zone,
both the vegetation and floodplain material (i.e. substrate material below the vegetation)
are likely to contribute to the total unit roughness. First provide a roughness zone
description e.g. ‘heavy weed growth’ (Figure 4.12) and then select the vegetation
option. Here, none of the options appear to match ‘heavy weed growth’ (Figure 4.13).
However, based on the photographs, the ‘height varying grass’ component for grass less
than 0.75m high is a possible choice. This is a purely subjective decision based on the
photographic advice, and may differ from user to user. Since there is substrate below
this grass, which is likely to be sand or fine gravel, an additional roughness component
due to ground material is included for this (Figure 4.14). Both sand and fine gravel

60
have virtually identical unit roughness values, so either component can be selected. The
unit roughness values for the grass and sand are 0.041 and 0.02 respectively. The
resulting total unit roughness value is 0.045618 (Equation 2.1) i.e.
[
nl = 0.0412 + 0.022 ]
1
2
= 0.045618 4.1
This value should only be recorded to three significant figures i.e. 0.0456 due to the
accuracy in determining the individual unit roughness values.
Figure 4.11 Bank material component with rip-rap selected
Figure 4.12 Providing roughness zone description: Heavy weed growth

61
Figure 4.13 Floodplain ground material component with height varying grass
selected
Figure 4.14 Floodplain ground material component with sand selected

62
The roughness zone allocation is now complete and three distinct zones (Table 4.2) have
been defined. The River Main roughness .rad file should be saved.
Table 4.2 Final Roughness Advisor (.rad file) output for the River Main
Roughness zone Roughness Vegetation Substrate Irregularity Total unit

Description component component component roughness
Bed cover Coarse gravel Gravel: 0.025 0.0250
Bank cover Rip-rap Rip-rap: 0.045 0.0450
Floodplain cover Heavy weed Grass: 0.041 Sand: 0.020 0.0456
growth and sand
4.2.1 Using the Conveyance Generator to generate a rating curve
The roughness file is now complete and ready for use within the conveyance
calculation. In the CES select the file / new Conveyance Generator file option (Figure
4.15). Save the conveyance .GEN file with an easily recognisable name e.g. River
Main.
Figure 4.15 Create a new conveyance file
When prompted under the ‘General’ tab, browse to the River Main .rad roughness file
and select it (Figure 4.16). This will ensure the roughness zones are available when
describing the cross-section roughness.
Figure 4.16 Select River Main roughness file
Select the ‘Cross-sections’ tab and ‘Add section’. Complete the cross-section details
such as providing a section name ‘Section14’, a section description ‘Bridge End
Bridge’, the reach-averaged longitudinal bed slope (0.0019 for the River Main) and the

63
sinuosity which is 1.0 as Section 14 is located in a straight portion of the reach (Figure
4.17).
Figure 4.17 Cross-section description
The first line of the cross-sections tab should now be populated with this information
(Figure 4.18), and ‘properties’ can be selected.
Figure 4.18 Select ‘properties’ for the now populated first cross-section
The cross-section information that has been entered should appear under the ‘General’
tab. Select the ‘Section Data’ tab and enter the cross-section offsets / cross-chainages
and the corresponding elevations (Table 4.1). The top-of-bank markers should be
selected. The drop-down menu provides six options: Left Inside, Left Outside, Left
Straight, Right Inside, Right Outside, Right Straight. The top-of-bank markers should
always be entered in pairs. Note that Left and Right banks are defined looking

64
downstream. The Straight, Inside and Outside represent straight channels, inside of
bend and outside of bend (see Section 3.2.6-3.2.8) top-of-bank markers. Section 14 is
located in a straight reach, and thus Left Straight and Right Straight are the appropriate
top-of-bank markers. The top-of-bank position is easily identified for this trapezoidal
shaped channel and the position is indicated on the section (Figure 4.19). This
corresponds to the standard secondary flow model (Figure 3.5).
The sinuosity field is only populated if the cross-section includes multiple channels with
varying sinuosity i.e. to indicate the sinuosity for each braided contribution. For this
case, the sinuosity is 1.0 for the whole section.
Figure 4.19 Cross-section data for cross-section 14
Select the ‘Roughness Zones’ tab and enter the cross-chainage at which the roughness
zone begins, i.e. from left to right, and the relevant roughness zone from the drop down
menu (Figure 4.20).
Once the roughness zones have been allocated, it is possible to generate the rating
curve. Select the ‘Outputs’ tab. In the ‘Output options’ drop down menu, the default is
conveyance, however other options can be calculated such as the stage-area, stage-flow,
stage velocity etc. Generate a conveyance curve using the default option (Figure 4.21).
The actual conveyance values can be observed in the adjacent window. The ‘default
plus min and max’ option will provide the upper and lower estimates for the uncertainty
calculation. As the date for he observed data is unknown, the ‘Specific date’ options are
not considered. The ‘Interrogator’ provides the facility to calculate the exact flow,
conveyance, area etc for a specified depth.

65
Figure 4.20 Roughness zone allocation
Select ‘Output to file’ and provide an output filename e.g. ‘output section 14’. This
output can be opened in Microsoft Excel to enable a comparison to the observed data
(Figure 4.22). For Section 14 the initial CES prediction compares well to the measured
data. From the shape of the curve, it is observed that the CES has calculated the flow
for very few depths within the range of the measured data. This has occurred since the
cross-section description has high elevations at the extreme ends. To increase the
number of depths for which the flow is calculated, select the ‘Advanced Options’ tab.
The default of number of depths is 25. This can be increased to for example 100
(Figure 4.23). Recalculate the rating curve and write the results to a file called ‘output2
section 14’. Figure 4.24 shows the new prediction compared to the data. The variation
at bankfull is captured with the increased number of calculation depths. The top-of-
bank markers were located at depths of 0.92m and 0.97m, hence the slight ‘kink’ in the
curve can be observed at 0.95m. This ‘kink’ represents the losses due to the change in
secondary flow mechanisms typically occurring at bankfull. The observed data appears
to emulate this trend.

66
Figure 4.21 The River Main rating curve for Section 14
1.8
CES prediction
1.6
Measured data
1.4
1.2
Stage m
0.8
0.6
0.4
0.2
0
0 10 20 30 40 50 60 70
Discharge m3/s
Figure 4.22 Comparison of the CES stage-discharge prediction to the measured
data

67
Figure 4.23 Alter the number of calculation depths in the Advanced Options
1.8
CES prediction
1.6
Measured data
1.4
1.2
Stage m
0.8
0.6
0.4
0.2
0
0 10 20 30 40 50 60 70
3
Discharge m /s
Figure 4.24 Improved CES prediction from increasing the number of depths in
Advanced Options

68
4.2.2 Calibrating process (depends on data availability)
For this example, the initial run compared well to the data and calibration was not
necessary.
4.2.3 Final result and interpretation of uncertainty
The final rating curve can be generated with the upper and lower bands for uncertainty
i.e. ‘Default plus upper and lower values’ option. These are plotted in Figure 4.25. For
this case all the measured data falls within the upper and lower uncertainty bands. The
uncertainty calculation is based on the upper and lower roughness values, and thus
accounts for any variation in the specified local roughness from its ‘true’ value.
2
CES central estimate
1.8
Measured data
1.6
Credible upper / lower bands
1.4
1.2
Stage m
0.8
0.6
0.4
0.2
0
0 10 20 30 40 50 60
3
Discharge m /s
Figure 4.25 Upper and lower uncertainty bands for the conveyance calculation
4.3 Worked example of the River Blackwater
The River Blackwater is situated in Hampshire in the South of England. It serves a

catchment area of approximately 40 km2 that responds to local storms.
In 1993, the construction of a major trunk road along the River Blackwater valley
resulted in the need to relocate a length of that river. This provided the opportunity to
the then UK National Rivers Authority to design the new channel and undertake a
multidisciplinary study of an environmentally acceptable channel design. This 3km
reach of the River Blackwater has therefore been extensively monitored in terms of flow
parameters, vegetation growth patterns and channel deformation.
The channel was reconstructed as a doubly meandering two-stage trapezoidal channel,

with a main channel and floodplain sinuosity of 1.18 and 1.05 respectively. It is
capable of carrying the 100year flood, 4.3 m3/s, and an upstream spillway ensures that

69
the flow does not exceed 4m3/s. The study reach is 520 m long, the main channel width
is 4.25 m and the average berm width is 4.9 m (Sellin and van Beesten, 2002). The
reach-averaged longitudinal bed slope is 0.00044. The inbank vegetation builds up in
the summer (Figure 4.26) and is washed out in Autumn, while the out-of-bank
vegetation is cut in October. The channel bank vegetation is emergent reeds with a silt
substrate and the bed consists of silt and fine gravel. The berms are covered in heavy
grass / weed growth.
The measured flow parameters available for calibration are stage, discharge and change
in Manning’s n with season. These measurements achieve good accuracy for high
flows, but are less accurate at low flows.
The selected cross-section (Figure 4.27) coincides with Gauge 3 of the five gauges and
is located at the bend apex. Stage-discharge measurements for three events are
available: mid-winter, late winter and mid-summer. The conveyance curve is thus
determined for all three cases to assess the impact of the vegetation growth and wash
out. The vegetation cutting has negligible effect as this occurs in late summer. Figure
4.28 and Table 4.3 provide the cross-section geometry.
(b)
(a) (c)
Figure 4.26 Vegetation on the River Blackwater (Hampshire, UK) in (a) May
1996 downstream from Colebrook Bridge (b) March 1995
downstream from Colebrook Bridge and (c) May 1996 upstream
from the footbridge

70
Gauge 3 situated at bend apex
Flow direction
Figure 4.27 Plan view of the River Blackwater indicating Gauge site 3
Table 4.3 River Blackwater cross-section geometry

34.43 63.60
37.13 63.58
39.10 63.50
41.10 62.93
43.84 62.93
47.29 62.89
50.58 62.76
52.63 62.74
53.30 62.16
53.90 61.84
55.02 61.80
56.28 61.83
57.26 61.82
57.90 61.83
58.20 62.35
59.19 63.15
59.93 63.89
60.97 63.92
61.88 63.90
63.39 63.94
64.45 64.03
65.13 63.95

71
64.5
64
63.5
Elevation (m)
63
62.5
62
61.5
34 39 44 49 54 59 64
Figure 4.28 River Blackwater cross-section geometry – Gauge site 3
4.3.2 Using the Roughness Advisor to generate roughness zones
Launch the CES software and create a new Roughness .rad file called the ‘River
Blackwater’. Define two roughness zones, ‘Bed cover’ and ‘Berm and bank cover’ of
type bed and bank respectively. As the measured data is inbank, it is not essential to
define a floodplain zone for this calibration.
Select the properties tab for the Bed cover. Give it a description e.g. ‘inbank vegetation
and silt’. Similarly select the properties tab for the Bank and berm cover and provide a
description e.g. ‘Emergent reeds and silt’ (Figure 4.29).
Figure 4.29 Roughness zones and descriptions for the River Blackwater

72
For the Bed Cover roughness zone, select ‘properties’ and then the ‘components’ tab
and highlight the vegetation option. Select ‘Edit component’. This will bring up the
bed vegetation options. From the description there is an indication that there is much
inbank vegetation i.e. the unit roughness is likely to be high. Of the available options,
the most suitable would be the ‘submerged fine-leaved’ option for water depth of 0.6-
1.2m (Figure 4.30). This corresponds to the typical depth of flow in the Blackwater and
the inbank vegetation is submerged, as observed in the photograph. Select the ‘Time
series’ option and activate the time series for any one-year period (Figure 4.31). This is
necessary since the measurements are taken at three different times of year, December,
February and June. Activating this time series will allow the conveyance curve to be
calculated for a particular date i.e. using the unit roughness associated with that date.
The vegetation field should now be populated with the expected, maximum and
minimum unit roughness values. Highlight the bed material option and select ‘Edit
component’. Select the sand option as the general bed description refers to silt and
gravel and this provides an intermediate option. There are no irregularities present
hence this option is ignored. The Bed cover roughness zone is now complete (Figure
4.32). The total unit roughness is given by:
[
nl = 0.12 + 0.02 2 ]
1
2
= 0.10198 4.2
Highlight the Bank and berm cover zone and select ‘Edit component’. Highlight the
vegetation option and consider the available options. From the Blackwater description,
the banks have been characterised as having emergent reeds. Select this option (Figure
4.33) and ‘time series’. Activate the emergent reeds time series.
Figure 4.30 Select time series for the submerged fine-leaved plants

73
Figure 4.31 Expected time series for the submerged fine-leaved plants (0.6-1.2m
depth)
Figure 4.32 Bed cover zone unit roughness values

74
Figure 4.33 Emergent reeds characteristic of the bank vegetation
Select the silt option for the bank material as no mention is made of gravel on the banks.
There are no irregularities and hence the Bank and berm cover zone is now complete.
The data field for this should be appropriately populated (Figure 4.34). The final
selection is summarised in Table 4.4. Save the River Blackwater .rad file.
Figure 4.34 Roughness zones for the River Blackwater

75
Table 4.4 Final Roughness Advisor (.rad file) output for the River Blackwater
Roughness Roughness Vegetation component Substrate Irregularity Total unit

zone Description component component roughness
Bed cover Inbank vegetation Submerged fine-leaved Sand: 0.020 0.102
and silt (0.6-1.2m): time-series
Bank and Emergent reed Emergent reeds: time- Silt: 0.018 0.151
berm cover and silt series
Open a new Conveyance Generator .GEN file and call it ‘River Blackwater’. Browse to
the River Blackwater .rad roughness file to activate the roughness zones defined in
Section 4.2.2. Create a new cross-section and under the ‘General Date’ tab enter the
cross-section name, description, bed slope and sinuosity as Gauge site 3, Bend Apex,
0.00044 and 1.18 (Figure 4.35).
Figure 4.35 Cross-section details for Gauge site 3 on the River Blackwater
Select the Section Data tab and enter the cross-section co-ordinates. Consider the plan
view of the River Blackwater (Figure 4.36). The flow is from left to right. Looking
downstream, the left top-of-bank marker is located on the inside of the bend and the
right top-of-bank marker is located on the outside of the bend. In the ‘Bank marker’
field select these options (Figure 4.36).

76
Figure 4.36 River Blackwater cross-section at Gauge site 3 illustrating the left
inside of bend top-of-bank marker
Select the ‘Roughness Zones’ tab and allocate the previously defined roughness zones
(Figure 4.37). Since most of the data is available for the inbank flow, select the
‘Advanced Options’ and 100 depths i.e. to increase the number of depths that are solved
for within the inbank region. Select the ‘Ouputs’ tab and ‘Specific date’ option. It is
now possible to generate a conveyance curve for the three applicable dates: February
1995, June 1995 and December 1999. The exact date within the month can be selected,
however in this case the 15th is selected, as the exact date for the actual events is
unknown (Figure 4.38).
The River Blackwater is an example of a river reach that is subject to backwater effects.
For this reason, a cross-section analysis for discharge prediction that is based on the
local bed slope rather than the local friction slope is likely to give incorrect predictions.
To facilitate comparison to the data, the measured conveyance is determined from the
measured flow divided by the square root of the local measured surface water slope
(Equation 2.7). This is possible since the depths for each flow reading were also taken
at gauge sites 1, 2, 4 and 5. This example is typical of where the CES is used to predict
the conveyance rather than the discharge.
The resulting conveyance predictions are given in Figures 4.39-4.41 for the February,
June and December events respectively.

77
Figure 4.37 River Blackwater roughness zone allocation
Figure 4.38 Select the date for the CES conveyance prediction

78
63.2
63
62.8
Stage m
62.6
62.4
Feb CES prediction

62.2
62 Feb data 1995
61.8
0 20 40 60 80 100 120 140 160
Conveyance m3/s
Figure 4.39 Comparison of the CES predicted conveyance to the measured data
for February 1995
62.8
62.7
62.6
62.5
62.4
Stage m
62.3
62.2
June CES prediction

62.1
62
June data 1995

61.9
61.8
0 10 20 30 40 50 60
Conveyance m3/s
for June 1995

79
63.4
63.2
63
62.8
mAOD
62.6
62.4
Dec CES prediction
62.2
Dec data 1999
62 Uncertainty bands
61.8
0 20 40 60 80 100 120 140
Conveyance m3/s
for December 1999
Comparison of the CES predicted conveyance to the observed conveyance suggests a

good fit for the February and June data. This is encouraging, as it suggests that the
vegetation time series has captured the true growth variation. The December prediction
over-predicts the conveyance to some extent. This may be for a number of reasons:
· The December data should be associated with a little vegetation as is expected in

winter, yet the measured data appears to be subject to a similar resistance to the
June data
· The December event reaches greater depths, thus increasing the floodplain extent.
The floodplain vegetation growth is possibly not as low in winter as the berm
vegetation. Possibly this vegetation is not cut
· The December data is from a different year, 1999, and the cross-section geometry is
likely to have changed since the 1995 cross-section survey
· The December event is associated with higher flows over a longer period, providing
the typical rising and falling limbs observed in unsteady flow models. This effect
is not simulated in the CES stand-alone software. For this analysis, a full
hydrodynamic model is recommended.
Improving the December prediction through calibration is not considered here since the
poor prediction may be a result of any of the listed factors (Section 4.2.3) or a
combination of them. The February and June data predictions are considered adequate.
Figure 4.42 shows the final predictions for February and June 1995 with the associated
uncertainty bands. These are plotted on the same set of axis to illustrate the effect that

80
the seasonal variation has on the conveyance curve. All the data falls within the range
of the uncertainty bands.
63.4
63.2
63
62.8
mAOD m
62.6
Feb CES prediction

62.4
June CES prediction
62.2
Uncertainty bands
62 June data 1995
Feb data 1995

61.8
0 20 40 60 80 100 120 140
Conveyance m3/s
Figure 4.42 The conveyance predictions for Gauge site 3 on the River
Blackwater with the uncertainty bands shown
4.4 Worked example of River Dane
The River Dane rises in the Peak District, flowing west to join River Wheelock, and
then north-west towards Northwick in Cheshire. The upstream catchment area is 407
km2 with a mean annual flood discharge of approximately 78 m3/s and a 1:100 year
flood of the order of 161 m3/s. The 6 km study reach with cross-section 30 identified, is
shown in Figure 4.43.
Figure 4.44 River Dane vegetation

81
Figure 4.43 The 6km reach of the River Dane with cross-section 30 identified
The vegetation consists of thick growth over the channel edges, which is less dense over
the floodplains (Figure 4.44).
The river channel itself is strongly meandering with sinuosity 1.8, however, section 30
is located in a straight portion. The channel widths are typically 25m and the river
depth / bankfull depth is approximately 5m. The river cross-section is also
approximately trapezoidal with side slopes at 30°-40° (Figure 4.45, Table 4.5) and the
longitudinal slope of the floodplain is in the range 1/900 to 1/1300 although the slope of
the river channel itself is of the order 1/2000.
The existing flood data has been taken at the gauging station, which coincides with
cross-section 30 at Chainage 25. Some stage-discharge data for this cross section has
been taken from the period January to July 1995. The maximum flood during this period
is 107.64 m3/s when the flood level was 16.66m AOD.

82
Table 4.5 River Dane cross-section 30 Geometry

-145 17.5
-145 17
4.5 16.19
12 16.19
15 15.28
16 14.25
17 13.51
18 13.45
19 12.47
24 12.04
28 12.29
30 13.44
32 14.76
34 15.1
38 16.76
45.5 16.76
388 16.4
388 17.5
30
28
26
24
Elevation (m)
22
20
18
16
14
12
10
-200 -100 0 100 200 300 400
Figure 4.45 River Dane cross-section 30 geometry
4.4.2 Using the Roughness Advisor to generate roughness zones
Launch the Roughness Advisor software and save the .rad file as the ‘River Dane’.
Create three new roughness zones termed Floodplain cover, Channel edge cover and
Inbank cover which are defined as roughness types: floodplain, bank and bed
respectively. For each roughness zone, provide a description e.g. thick grass on the
floodplain, thick bush adjacent to the channel and sand with some vegetation in the
channel (Figure 4.46).

83
Figure 4.46 Roughness zone definition and description
Select the properties tab for the Floodplain cover roughness zone. The floodplains are
covered with grass and the substrate is likely to be sand. Select the ‘0.75-1m grass’
option for the vegetation component and ‘sand’ for the ground material. There are no
apparent irregularities on the floodplains. The total unit roughness for the floodplain
zone should be 0.0824621.
Select the properties tab for the Channel edge cover roughness zone. The channel edges
are characterised by thick growth (Figure 4.44). The flow measurements were made
between January and July i.e. from mid- winter when the vegetation is likely to be
scarce through to the height of summer when the vegetation is likely to be flourishing.
For the vegetation, select the emergent reeds, as this is the most appropriate option of
the three available. Select the ‘time series’ option (Figure 4.47) and activate the time
series for any one-year period (Figure 4.48). It is clear from this that the variation in the
unit roughness from January to July is 0.01-0.15 – quite a large range! Close this option
with the time series still activated. This will enable the calculation of the rating curve in
the Conveyance Generator at any specified date within the time series. For the bank
material, select the ‘sand’ option. No irregularities are present on the floodplain. The
final unit roughness for the Channel edge cover roughness zone should be 0.151327.
This results from the combined unit roughness at the height of summer and it will
change for different times of year.

84
Figure 4.47 Select the Time series for emergent reeds
Figure 4.48 Emergent reed time series for a given year

85
Select the properties tab for the Inbank cover roughness zone. Inbank, the channel
consists of sand and the scarce vegetation at the edges. For the bed material select sand.
For the vegtation, select ‘Trailing bank-side plants’. Note that the unit roughness value
varies between a minimum of 0.00 and a maximum of 0.10, with an expected value of
0.05. Since the vegetation inbank is scarce, this value is likely to be on the low.
Manually overwrite the expected value with a 0.03 (Figure 4.49). This should provide a
combined unit roughness of 0.360555 for the Inbank cover roughness zone.
Figure 4.49 Manually overwrite the suggested roughness based on available

information
The final roughness values for the three zones are provided in Figure 4.50. Save the
.rad file.
Figure 4.50 Final roughness zone value in the .rad file

86
Once the roughness file is complete, open a new Conveyance Generator file. Browse to
the River Dane .rad file to activate the newly created roughness zones. Add a new
cross-section and call it e.g. Section 30. The description could be ‘straight’, since it is
located in a straight portion of the reach. Define the longitudinal reach averaged bed
slope 1:2000 (0.0005) and set the sinuosity as 1.0 for a straight channel (Figure 4.51).
Figure 4.51 River Dane channel description
Enter the cross-section co-ordinates and define the left and right panel markers (Figure
4.52).
Figure 4.52 River Dane cross-section with straight top-of-bank markers

87
The roughness zones can now be allocated. For the initial run, the section is divided
into the main floodplain portions, the channel edge sections ±15m and the main channel
(Figure 4.53). This division is based purely on the photograph and description and may
be altered for the calibration.
Figure 4.53 River Dane roughness zone allocation
Select the ‘Outputs’ tab to generate the rating curve. This rating curve will correspond
to that day’s date, as listed in the panel. The date can be altered to reflect the likely time
of measurement. Set this to 4/11/2003 i.e. mid-way between winter and summer.
Export the results to an output file and open in Microsoft Excel to facilitate comparison
to the measured data (Figure 4.54). The stage-discharge results indicate a reasonable fit
for a blind attempt. The resistance is slightly over-predicted just near bankfull.

88
6
Measured Data
5
CES prediction
4
Stage (m)
0
0 20 40 60 80 100 120 140
3
Discharge (m /s)
Figure 4.54 River Dane CES rating curve prediction compared to the measured
data
As there is data available, it is possible to try and calibrate a better fit for the rating
curve. The region where the data is above the curve is inbank. Increasing the inbank
resistance will create a worse error at low flow and a better fit at higher flows. Alter the
user defined vegetation roughness value to 0.025 i.e. lower than the previous 0.03. As
the data is mainly inbank, it may be worthwhile to increase the number of depths for the
calculation to ensure that sufficient points in this region is calculated. Select the
Advanced options tab and change the number of depths to 50 (default is 25).
The calibrated rating curve (Figure 4.55) provides the expected improvement through
lowering the inbank resistance. It is unlikely to improve this calibration further without
increasing the number of roughness zones across the section. This is left to the user as
an optional exercise.

89
6
Measured data
5
CES prediction
Calibration
4
Stage (m)
0
0 20 40 60 80 100 120 140
3
Discharge (m /s)
Figure 4.55 The River Dane stage-discharge curve after calibration
Once calibrated the final stage-discharge curve with upper and lower uncertainty bands
can be compared to the data (Figure 4.56). The width of the uncertainty bands
encompasses all the data points. The upper band is particularly high for the inbank
flow, which is likely since the expected roughness value was over-written manually
with a lower roughness in this region.
Measured data
7
Uncertainty bands
6 Calibrated
5
Stage (m)
0
0 20 40 60 80 100 120 140 160 180 200
3
Discharge (m /s)
Figure 4.56 Final stage-discharge curve for the River Dane

90
5. REFERENCES
Ackers, P. 1992. Hydraulic Design of Two-Stage Channels, Proc. ICE Journal for
Water, Maritime and Energy, Paper no. 9988, Vol. 96, Dec., pp 247-257.
Ackers, P. 1993. Stage-discharge Functions for Two-Stage Channels: The Impact of

New Research, Journal Institute Water and Environmental Management, Vol. 7, No. 1,
February, pp 52-61.
Ackers, P. 1993. Flow Formulae for Straight Two-Stage Channels, Journal of Hydraulic
Research, IAHR, Vol. 31, No. 4, pp 509-531.
Abril, J. B. 1997. Numerical River Modelling of Turbulent Flow, Sediment Transport

and Flood Routing using the Finite Element Method, PhD Thesis, January, School of
Engineering, The University of Birmingham, England, UK.
Abril, J. B. 2001. Updated RFMFEM Finite Element Model based on the SKM for
Depth-averaged River Flow Simulation, EPSRC Technical Report GR/R54880/01.
Abril, J. B. 2002. Overbank Flood Routing Analysis Applying Jointly Variable

Parameter Diffusion and Depth-Averaged Flow Finite Element Models, Proc. River
Flow 2002, Louvain, Belgium, compiled by Y. Zech and D. Bousmar.
Abril, J. B. and Knight, D.W. 2002. Stage-discharge prediction for rivers in flood
applying a depth-averaged model, submitted to Journal of Hydraulic Research, IAHR.
Barnes, H. H. 1967. Roughness Characteristics of Natural Channels, US Geological

Survey, Water-Supply Paper No. 1849, pp214, Washington DC.
Bousmar, D. and Zech, Y., 1999. Momentum Transfer for Practical Flow Computation
in Compound Channels, Jnl. of Hydr. Eng., ASCE, Vol. 125, no. 7, July, pp 696-706.
Bousmar, D. and Zech, Y. 1999. Momentum Transfer for Practical Flow Computation
in Compound Channels, Journal of Hydraulic Engineering, ASCE, Vol. 125, No. 7,
July, pp 696-706.
Bousmar, D. and Zech, Y. 2002. Periodical Turbulent Structures in Compound

Channels, Proc. River Flow 2002, Louvain, Belgium, compiled by Y. Zech and D.
Bousmar.
Chang, H. H. 1983. Energy Expenditure in Curved Open Channels, Journal of

Hydraulic Engineering, Vol. 109, No. 7, pp 1012-1022.
Chang, H. H. 1984. Regular Meander Path Model, Journal of Hydraulic Engineering,

Vol 110, No. 10, October, pp 1398-1411.
Chezy, A. 1768. In: On the Origin of the Chezy Formula by C. Herschel, Journal
Association of Engineering Societies, Vol. 18, pp 363-368.
Chow, V. T. 1959. Open Channel Hydraulics, McGraw-Hill Book Company, US.

91
Colebrook, C. F. and White, C. M. 1937. Experiments with Fluid Friction in Roughened
Pipes, Proc. of the Royal Society, A161, pp 367-379.
Cowan, W. L. 1956. Estimating Hydraulic Roughness Coefficients, Agricultural

Engineering, Vol. 37, pp 473-475.
DEFRA/EA, 2002. Reducing Uncertainty in River Flood Conveyance, Interim Report 1:

Data Mining, Project W5A- 057, HR Wallingford Ltd., United Kingdom.
DEFRA/EA, 2003a. Reducing Uncertainty in River Flood Conveyance, Interim Report

2: Review of Methods for Estimating Conveyance, Project W5A- 057, HR Wallingford
Ltd., United Kingdom.
DEFRA/EA, 2003b. Reducing Uncertainty in River Flood Conveyance, Roughness

Review, Project W5A- 057, HR Wallingford Ltd., United Kingdom.
DEFRA/EA, 2004/5. Reducing Uncertainty in River Flood Conveyance, Interim Report

3: Testing of Conveyance Methods in 1D River Models, Project W5A- 057,
HR Wallingford Ltd., United Kingdom.
Einstein, H. A. and Banks, R. B. 1950. Fluid Resistance of Composite Roughness,

Transactions of the American Geophysical Union, Volume 31, No. 4, pp 603-610.
Evans, E. P., Pender, G., Samuels, P. G. and Escarameia, M. 2001. Reducing

Uncertainty in River Flood Conveyance: Scoping Study, R&D Technical Report to
DEFRA / Environment Agency, Project W5A- 057, HR Wallingford Ltd., United
Kingdom.
Ervine, D. A. and Ellis, J. 1987. Experimental and Computational Aspects of Overbank

Floodplain Flow, Trans. Royal Society of Edinburgh: Earth Sciences, Vol. 78, pp 315–
475.
Ervine, D. A., Babaeyan-Koopaei, K. and Sellin, H.J. 2000. Two-dimensional Solution

for Straight and Meandering Overbank Flows, Journal of Hydraulic Engineering,
ASCE, Vol. 126, no. 9, September, pp 653-669.
Hicks, D. M. and Mason, P. D. 1998. Roughness Characteristics of New Zealand

Rivers, NIWA, Christchurch, 329pp.
HR Wallingford, 2002. Risk Performance and Uncertainty in Flood and Coastal

Defence - A Review. Report SR 587 (Second Draft). HR Wallingford.
Intergovernmental Panel on Climate Change (2001). Climate Change 2001, Impacts,

Adaptation and Vulnerability, Cambridge University Press.
James, C. S. and Wark, J. B. 1992. Conveyance Estimation for Meandering Channels,

HR Wallingford Ltd. Report SR 329, pp 1-91.

92
Knight, D. W., 1981. Some field measurements concerned with the behaviour of
resistance coefficients in a tidal channel, Estuarine, Coastal and Shelf Science, Vol. 12,
pp 303-322.
Knight, D. W. and Abril, B. J. 1996. Refined Calibration of a Depth-Averaged Model

for Turbulent Flow in a Compound Channel, Proc. Inst. of Civil Engineers Water,
Maritime and Energy, Paper 11017, Vol. 118, Sept., pp 151-159.
Manning, R. 1890. On the Flow of Water in Open Channels and Pipes, Transactions of
the Institute of Civil Engineers of Ireland, Vol. 20, pp 161-207.
National Research Council (NRC). 2000. Risk Analysis and Uncertainty in Flood
Damage Reduction Studies. Washington, D.C.: National Academy Press.
Nikuradse, J. 1933. Translates to: Laws of Flow in Rough Pipes, Verein deutscher
Ingenieure, Forschubgsheft, No. 361, Berlin.
Raven, P. J., Holmes, H. T. H., Dawson, F.H., Fox, P. J. A., Everard, M., Fozzard, I.
and Rouen K. J. 1998. River Habitat Quality The Physical Character of Rivers and
Streams in the UK and the Isle of Man, River Habitat Report No. 2, Environment
Agency, pp 90.
Raudkivi, A. J. 1998. Loose Boundary Hydraulics, A.A. Balkema, Rotterdam,

Netherlands.
Samuels, P. G. 1985. Modelling of River and Flood Plain Flow using the Finite Element
Method, PhD Thesis, Department of Mathematics, University of Reading, Hydraulics
Research Report no. SR61, November.
Samuels, P. G. 1988. Lateral Shear Layers in Compound Channels, International

Conference on Fluvial Hydraulics.
Samuels, P. G. 1989. The hydraulics of Two-stage Channels – Review of Current

Knowledge, Conference of River Engineers, University of Loughborough, HR
Published Paper no. 45.
Samuels, P. G. 1989. Some Analytical Aspects of Depth Averaged Flow Models, Intnl.
Conf. Hydraulic and Environmental Modelling of Coastal, Estuarine and River Waters,
Bradford, England, 19-21 September.
Samuels, P. G. 1989. Backwater lengths in rivers, Proc. Inst. of Civil Engineers, Part 2,
Vol. 87, pp 571-582.
Samuels, P. G. 1995. Uncertainty in Flood Level Prediction, XXVI Biannual Congress

of the IAHR, HYDRA2000, London, Sept 1995, Volume 1 of Proceedings published by
Thomas Telford.
Samuels, P. G., Bramley, M. E. and Evans, E. P. 2002. Reducing Uncertainty in

Conveyance Estimation, Paper presented at River Flow 2002, Université Catholique
Louvain, September 2002, Papers published by Balkema & Sons.

93
Sellin, R. H. J. and van Beesten, D. P. 2002. Berm Vegetation and its Effect on Flow
Resistance in a Two-Stage River Channel: An Analysis of Field Data, Proceedings of
the River Flow 2002 Conference, Belgium, Zwets and Zeitlinger, Lisse, ISBN 90 5809
509 6, Volume 1, pp 319-327.
Shiono, K. and Knight, D. W. 1989. Two-Dimensional Analytical Solution for

Compound Channel, Proc. 3rd Intl. Symposium on Refined Flow Modelling and
Turbulence Measurements, Universal Academy Press, pp 591-599.
Shiono, K. and Knight, D. W. 1991. Turbulent Open-channel flows with Variable Depth
across the Section, Journal of Fluid Mechanics, Vol. 222, pp 617-646.
Shiono, K. and Knight, D. W. 1990. Mathematical Models of Flow in Two or Multi

Stage Straight Channels, International Conference on River Flood Hydraulics, (Ed.
W.R.White), Wiley and sons Ltd., Paper G1, pp229-238.
Shiono, K. and Muto, Y. 1998. Complex Flow Mechanisms in Compound Meandering

Channels with Overbank Flow, Journal of Fluid Mechanics, Vol 376, pp 221-261.
Shiono, K., Muto, Y., Knight, D. W. and Hyde, A. F. L. 1999. Energy Losses due to
Secondary Flow and Turbulence in Meandering Channels with Overbank Flow, Journal
of Hydraulic Research, IAHR, Vol. 37, No. 5, pp 641-664.

94

1 CES UserGuide PDF

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

1 CES UserGuide PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 CES UserGuide PDF

Uploaded by

Copyright:

Available Formats

Defra / Environment Agency

Flood and Coastal Defence R&D Programme

Reducing uncertainty in river flood conveyance, Phase 2

Project Record W5A-057/PR/1

REDUCING UNCERTAINTY IN RIVER FLOOD

Project Record W5A-057/PR/1

Research Contractor: HR Wallingford

Environment Agency's Project Manager

R&D OUTPUTS: CONVEYANCE USER MANUAL

DCM Divided Channel Method

DEFRA Department for Environment, Food and Rural Affairs

FCF Flood Channel Facility

IDB Internal Drainage Boards

LiDar Airborne scanning Laser Altimetry

MDSF Modelling and Decision Support Framework

RANS Reynolds-Averaged Navier-Stokes

RHS River Habitat Survey

SAR Sonar Acoustique Remorque

R&D OUTPUTS: CONVEYANCE USER MANUAL

C Chezy coefficient (m0.5.s-1)

R&D OUTPUTS: CONVEYANCE USER MANUAL

R&D OUTPUTS: CONVEYANCE USER MANUAL

Backwater: Backwater effects occur when sub-critical flow is controlled

Bankfull: The maximum channel discharge capacity; further discharge

Conveyance: Channel conveyance is a measure of the discharge carrying

Dimensionless eddy Calibration coefficient to account for lateral shear l

Discharge: The volume of water that passes through a channel section

Error: Errors are mistaken calculations or measurements with

Flow: A general term for the movement of volumes of water at a

Flow zone: Area in a channel cross-section within which the flow

R&D OUTPUTS: CONVEYANCE USER MANUAL

Morpho-types: See “vegetation morpho-types”

Offset: Distance measured laterally across the channel, taken from

RANS: The Navier-Stokes equations are derived from the governing

Resistance: As for roughness but defined as flow-, form-, frictional or

Riffle: A point at which the stream is relatively energetic due to a

R&D OUTPUTS: CONVEYANCE USER MANUAL

Roughness A roughness component is a category of unit roughness that

Roughness zones: A roughness zone is a plan area in the river/floodplain

Sinuosity: A measure of a channel’s tendency to meander defined in the

Stage: The vertical distance of the free surface from an arbitrary or

Study reach: The length of a study reach OR a section under investigation/

Substrate: A generic term for a substance that underlies another; soil is

Thalweg: A line connecting the deepest points along a channel length

Top-of-bank markers: Top-of-bank markers are used to incorporate the changes in

R&D OUTPUTS: CONVEYANCE USER MANUAL

Uncertainty: Uncertainty comprises (i) natural variability and (ii)

Unit roughness: Roughness due to an identifiable segment of boundary

R&D OUTPUTS: CONVEYANCE USER MANUAL

R&D OUTPUTS: CONVEYANCE USER MANUAL

R&D OUTPUTS: CONVEYANCE USER MANUAL

3. Using the CES 31

R&D OUTPUTS: CONVEYANCE USER MANUAL

R&D OUTPUTS: CONVEYANCE USER MANUAL

R&D OUTPUTS: CONVEYANCE USER MANUAL

R&D OUTPUTS: CONVEYANCE USER MANUAL

R&D OUTPUTS: CONVEYANCE USER MANUAL