ACCOUNTING FOR SEDIMENT IN RIVERS
A tool box of sediment transport and transfer analysis methods
and models to support hydromorphologically-sustainable flood
risk management in the UK
Nick Wallerstein
University of Nottingham
August 2006
FRMRC Research Report UR9
Project Web: www.floodrisk.org.uk
Accounting for Sediment in Rivers
FRMRC Research Report UR9
FRMRC Partners
The FRMRC Partners are:
• University of Bristol
• Heriot Watt University
• HR Wallingford
• Imperial College, London
• University of Lancaster
• University of Manchester
• University of Nottingham
• University of Sheffield
Report Contributors
Project Secretariat
ARP
Directorate of Planning and Academic Services
University of Manchester
Sackville Street,
Manchester
PO Box 88
M60 1QD
Tel: +44 (0)161 306 3626
Fax: +44 (0)161 306 3627
Web: www.floodrisk.org.uk
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
ii
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Summary
It has become increasingly clear that river channel sediment dynamics must be taken into account in
the UK when developing flood alleviation schemes because sediment transport can have an impact on
channel flood capacity. The aim of this Work Package (WP 8.1) within Research Priority Area 8 of the
Flood Risk Management Research Consortium (FRMRC) has therefore been dedicated to developing
new and improving on existing tools and models which quantify sediment continuity in the river
system in terms of reaches which act as sources (degradational reaches), pathways (equilibrium
reaches) and sinks (aggradational reaches).
Initially, it was intended to adapt the Sediment Impact Assessment Model (SIAM) developed in the
USA for UK use but, as the work unfolded, two problems with this intention emerged. First, it soon
became apparent that due to the limited availability of sediment data in the UK applications of SIAM
would be confined to relatively large and well-resourced projects. Second, given the need to support a
wide variety of potential applications, it was impossible for the SIAM tool, or indeed any other single
tool, to meet the diverse needs of the end user community. To address the problem of limited sediment
data availability, the focus of the work switched to development of the River Energy Audit Scheme
(REAS) in parallel with SIAM. To address the second problem, the scope of the work package was
extended to include the compilation of a suite of methods and models with the aim of creating a
versatile ‘toolbox’, rather than a single tool. The six models described in this toolbox are: the Stream
Power Screening Tool; River Energy Audit Scheme (REAS); Sediment Impact Assessment Model,
embedded in HEC-RAS (HEC-RAS/SIAM); Hydraulic Engineering Centre, River Analysis System:
Sediment Transport (HEC-RAS ST); iSIS Sediment; and the Cellular Automaton Evolutionary Slope
and River Model (CAESAR).
These tools span a range of requirements in terms of data input, technical knowledge and costs (time
and money) and generate output resolutions which range from indicative to diagnostic and spatial
scales from whole catchments to short river reaches. Through the use and development of these tools it
has been borne in mind that the results should consolidate and build upon current and new
fundamental ‘blues skies’ research which then enables the generation of user focused measurable
outcomes – the products that will actually be operationalised by the flood risk manager.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
iii
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Document Details
Document History
Version
Date
1
16/06/06
1_0
August 06
Lead Author
Institution
Nick
Wallerstein
University of Nottingham
J Bushell
University of Nottingham
Jeremy Benn Associates
Jacobs Babtie
Jacobs Babtie
US Army Corps of
Engineers, Engineer
Research & Development
Center
Colorado State University
US Bureau of
Reclamation
US Army Corps of
Engineers, Engineer
Research & Development
Center
Institute for Water
Resources, Hydrologic
Engineering Center
Jeremy Benn Associates
The University of Hull
HR Wallingford Ltd
Joint authors
Comments
Colin Thorne
Philip Soar
Andrew Brookes
Duncan Wishart
David
Biedenham
Chester Watson
David Mooney
Charles Little Jr
Stanford Gibson
Tony Green
Tom Coulthard
Formatting for
publication; change
of name from
‘FRMRC RPA 8
WP 8.1 Final
Report.doc’
End user approval: Jim Walker and David Brown, Environment Agency, 8 December 2006
Acknowledgements
This research was performed as part of a multi-disciplinary programme undertaken by the Flood Risk
Management Research Consortium. The Consortium is funded by the UK Engineering and Physical
Sciences Research Council under grant GR/S76304/01, with co-funding from:
• Defra and the Environment Agency through their Joint R&D programme on Flood and Coastal
Erosion Risk Management,
• UKWIR
• NERC
• The Scottish Executive
• Rivers Agency Northern Ireland
The authors wish to give special thanks to Jim Walker and David Brown of the Environment Agency
for their considerable input towards guidance of this project.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
iv
Accounting for Sediment in Rivers
FRMRC Research Report UR9
The authors would also like to thank the project steering committee for their valuable input:
David Corbelli (Environment Agency), Helen Dangerfield (Royal Haskoning), Steve Dangerfield
(Cascade Consulting), John Rees (British Geological Survey), Mike Thorn (Defra – EA), Roger
Bettess (HR Wallingford), Rachel Wignall (Scottish Natural Heritage), Duncan Wishart (Jacobs
Babtie).
Front Cover: The Hawkcombe Stream downstream of Porlock exhibiting serious bed and bank
instability due to imbalance between sediment supply and transport capacity in the fluvial system.
(Photograph by Colin Thorne and Kevin Skinner, University of Nottingham.)
Disclaimer
This document reflects only the authors’ views and not those of the FRMRC Funders. This work may
rely on data from sources external to the FRMRC Partners. The FRMRC Partners do not accept
liability for loss or damage suffered by any third party as a result of errors or inaccuracies in such data.
The information in this document is provided “as is” and no guarantee or warranty is given that the
information is fit for any particular purpose. The user thereof uses the information at its sole risk and
neither the FRMRC Funders nor any FRMRC Partners is liable for any use that may be made of the
information.
©
Copyright 2006
The content of this report remains the copyright of the FRMRC Partners, unless specifically
acknowledged in the text below or as ceded to the funders under the FRMRC contract by the Partners.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
v
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Table of Contents
Title page
i
FRMRC Partners
ii
Summary
iii
Document Details
iv
Table of Contents
vi
1. Accounting for Sediment in Rivers: The FRMRC Toolbox of Methods and Models .......................1
1.1 Overview....................................................................................................................................1
1.1.1 Background...................................................................................................................1
1.1.2 The Needs Case ............................................................................................................2
1.2 FRMRC research on sediment dynamics ...................................................................................4
1.2.1 Aims..............................................................................................................................4
1.2.2 Study Approach ............................................................................................................5
1.2.3 Project Deliverable: User Focused Measurable Outcome ............................................7
1.2.4 Dealing with Uncertainty..............................................................................................9
1.2.5 Issues Affecting Take-up of the Tools........................................................................10
1.3 The next steps...........................................................................................................................11
1.4 References................................................................................................................................12
2. Application of a Stream Power Screening Tool...............................................................................13
2.1 The conceptual basis for the stream power tool.......................................................................13
2.2 A brief history of the use of stream power (total, specific, unit) in sediment analysis and
sediment transport prediction...................................................................................................14
2.3 The origins and evolution of the screening tool itself ..............................................................15
2.4 The current version of the tool and its uses..............................................................................15
2.5 The following are three examples of actual application by Brookes .......................................16
2.6 River Caldew Case Study.........................................................................................................18
2.6.1 Issue ............................................................................................................................19
2.7 Results......................................................................................................................................19
2.7.1 Key Points...................................................................................................................19
2.8 Synthesis ..................................................................................................................................20
2.9 Next stage.................................................................................................................................21
2.10 Dealing with uncertainties and risks associated with using the stream power screening tool .21
2.11 Guidance on its limitations.......................................................................................................22
2.12 Acknowledgement ...................................................................................................................22
2.13 References and further reading ................................................................................................22
3. Sediment Impact Assessment Model (SIAM) .................................................................................25
3.1 Background ..............................................................................................................................25
3.2 SIAM Description ....................................................................................................................25
3.2.1 Model Computation Methodology..............................................................................26
3.3 Data Requirements ...................................................................................................................26
3.3.1 Model Output..............................................................................................................27
3.3.2 Capabilities and Limitations .......................................................................................27
3.4 Incorporating SIAM into HEC-RAS........................................................................................28
3.4.1 SIAM Input.................................................................................................................28
3.4.2 SIAM Output ..............................................................................................................29
3.5 Judy’s Branch Case Study (Watson & Eom 2003) ..................................................................30
3.6 Conclusions..............................................................................................................................32
3.7 References................................................................................................................................32
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
vi
Accounting for Sediment in Rivers
FRMRC Research Report UR9
4. Sediment Transport Computations with HEC-RAS ........................................................................33
4.1 Introduction..............................................................................................................................33
4.2 Methodologies..........................................................................................................................33
4.2.1 Hydrodynamics...........................................................................................................33
4.2.2 Transport Calculations................................................................................................34
4.2.3 Physical Constraints to Erosion and Deposition.........................................................35
4.2.4 Temporal Modifiers ....................................................................................................35
4.2.5 Sorting and Armouring ...............................................................................................35
4.3 Model Testing and Verification ...............................................................................................36
4.3.1 Comparison to Meyer-Peter and Muller Data.............................................................36
4.3.2 Comparison to HEC-6 ................................................................................................36
4.4 Example Application................................................................................................................37
4.5 Range of Appropriate application ............................................................................................42
4.6 Availability and release schedule.............................................................................................42
4.7 Conclusion ...............................................................................................................................42
4.8 References................................................................................................................................43
5. River Energy Audit Scheme (REAS)...............................................................................................44
5.1 Background ..............................................................................................................................44
5.2 Theory ......................................................................................................................................48
5.2.1 The River Energy Audit Scheme ................................................................................48
5.2.2 Data Requirements......................................................................................................48
5.2.3 Specific Stream Power................................................................................................49
5.2.4 Critical Power Based Upon a Grainsize Distribution .................................................50
5.2.5 The REAS Annual Energy Budget .............................................................................51
5.3 Discussion ................................................................................................................................54
5.3.1 Solution for Flow Width and Depth ...........................................................................54
5.3.2 Representing Stream Power in a Variable Cross-section ...........................................55
5.4 Relationship between Total Specific Power and Excess Power ..............................................60
5.4.1 Accounting for Lateral Sediment Inputs.....................................................................61
5.4.2 Channel Junctions and Structures ...............................................................................62
5.4.3 Reach Limits...............................................................................................................63
5.5 Sources of Uncertainty.............................................................................................................63
5.6 Limitations ...............................................................................................................................64
5.7 Case studies..............................................................................................................................65
5.7.1 Hawkcombe Stream....................................................................................................65
5.7.2 River Wharfe ..............................................................................................................72
5.8 Conclusions..............................................................................................................................77
5.8.1 Key Points: Answers to questions posed by the Environment Agency ......................77
5.8.2 General Conclusions...................................................................................................79
5.9 References................................................................................................................................80
6. Use of 1-D Sediment Models...........................................................................................................83
6.1 Background and Conceptual basis ...........................................................................................83
6.1.1 Background.................................................................................................................83
6.1.2 Conceptual Background to 1-D Sediment Models .....................................................83
6.1.3 Available 1-D Models.................................................................................................85
6.1.4 Other General Purpose 1-D Sediment Codes..............................................................86
6.2 Application of 1-d sediment models ........................................................................................87
6.2.1 Range of Applications ................................................................................................87
6.2.2 Recent Applications....................................................................................................87
6.2.3 Likely Applications for which 1-D Sediment Models are Applicable........................88
6.2.4 Issues and Decisions to be Made when Setting up a 1-D Sediment Model................88
6.3 Application of iSIS sediments to study the effect of the number of sections USED ...............88
6.4 Uses and Limitations of 1-D Sediment models........................................................................98
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
vii
Accounting for Sediment in Rivers
FRMRC Research Report UR9
6.5 References and further reading ................................................................................................98
7. The Cellular Automaton Evolutionary Slope And River model (CAESAR).................................101
7.1 Introduction and background to CAESAR ............................................................................101
7.2 Background to cellular models...............................................................................................102
7.3 Theory and model description................................................................................................103
7.4 Application.............................................................................................................................105
7.4.1 Usability....................................................................................................................105
7.4.2 Data Requirements....................................................................................................106
7.4.3 Notes on Other Parameters .......................................................................................106
7.4.4 Grid Cell Size and Resolution ..................................................................................107
7.5 Case studies............................................................................................................................107
7.5.1 Reach mode: Modelling the River Severn nr. Caersws, U.K. ..................................107
7.5.2 Catchment Example: Modelling the River Swale, U.K............................................112
7.5.3 Processing the Results ..............................................................................................113
7.6 Limitations and uncertainty....................................................................................................115
7.7 Discussion and Conclusions...................................................................................................117
7.8 Acknowledgements ................................................................................................................117
7.9 References..............................................................................................................................117
Table of Tables
Table 5.1 Width adjustment factors used in REAS ............................................................................61
Table 5.2 Data Type, availability, cost, processing time and licensing agreements...........................78
Table 6.1 Recent applications of iSIS Sediment.................................................................................87
Table 6.2 Standard Guidelines for spacing of sections in flood models (after Samuels as quoted
in iSIS Manual (undated))...................................................................................................89
Table 7.1 Chronology of CAESAR development.............................................................................101
Table of Figures
Figure 1.1 Over-arching research philosophy of the FRMRC ...............................................................2
Figure 1.2 Relative contributions of interpretational and analytical approaches in different tools
included in the FRMRC Sediment Toolbox .........................................................................6
Figure 1.3 ‘Proof of Concept’ criteria for FRMRC UFMOs..................................................................8
Figure 1.4 Main tools in the toolbox assessed in terms of ‘proof of utility’ ..........................................9
Figure 1.5 Relationship between model complexity, cost and risk of not representing the modelled
system .................................................................................................................................10
Figure 1.6 Balancing management resources, the science of predictive models and acceptance by
society .................................................................................................................................11
Figure 2.1 Tilmore Brook, Hampshire (UK)........................................................................................16
Figure 2.2 Some river restoration projects in southern Britain ............................................................17
Figure 2.3 Slope - bankfull discharge plots for several steep upland catchments in Britain................18
Figure 2.4 Variations of specific stream power along the River Caldew.............................................19
Figure 2.5 River Caldew, Carlisle ........................................................................................................21
Figure 3.1 Input data requirements for SIAM......................................................................................27
Figure 3.2 HEC-RAS interface for SIAM with the bed material input tab ..........................................28
Figure 3.3 SIAM data tabs in the HEC-RAS interface ........................................................................29
Figure 3.4 Source specification dialog and tab in the HEC-RAS interface for SIAM.........................29
Figure 3.5 SIAM output plot of local bed material balance .................................................................30
Figure 3.6 SIAM output table of local balance by grain size...............................................................30
Figure 3.7 Sediment yield reduction along Judy’s Branch (Watson & Eom 2003) .............................31
Figure 4.1 Schematic of quasi-steady flow division ............................................................................34
Figure 4.2 Schematic of 3-layers used in Exner 5 sorting and armouring method ..............................36
Figure 4.3 Meyer-Peter and Muller flume data with HEC-6 and HEC-RAS simulations ...................37
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
viii
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 4.4 Single grain trapezoidal channel with supply slightly exceeding capacity.simulated with
HEC-RAS and HEC-6 ........................................................................................................37
Figure 4.5 Sediment data and methods editor ......................................................................................38
Figure 4.6 Bed gradation template .......................................................................................................38
Figure 4.7 Inflowing sediment boundary condition .............................................................................39
Figure 4.8 Quasi-steady flow editor .....................................................................................................40
Figure 4.9 Sediment analysis window..................................................................................................40
Figure 4.10 Initial and final thalweg profile...........................................................................................41
Figure 4.11 Time series plot of bed elevation at station 4075................................................................41
Figure 5.1 Flow diagram for solution of 1-D channel boundary adjustment using the gradually varied
flow equation time-step approach .......................................................................................47
Figure 5.2 Sediment transfer through the fluvial system .....................................................................48
Figure 5.3 Schematic of variables that must be obtained from the river basin to run REAS...............49
Figure 5.4 Example of cumulative probability plot for 15 minute flow data.......................................51
Figure 5.5 Discharge–frequency plot ...................................................................................................52
Figure 5.6 Schematic diagram of REAS calculation procedure...........................................................53
Figure 5.7 Definition of parameters required to solve Manning’s equation. n = Manning’s roughness
coefficient; and K = conveyance.........................................................................................54
Figure 5.8 Width variation over a berm within the geomorphologically active channel .....................56
Figure 5.9 Percentage error between discharge calculation for a cross-section based upon a single
area and a panelised cross section.......................................................................................58
Figure 5.10 The relationship between channel cross-sectional geometry and specific power–discharge
curves ..................................................................................................................................59
Figure 5.11 Diagramatic representation of the potential impact of an in-channel berm upon excess
stream power.......................................................................................................................59
Figure 5.12 Schematic plot of grainsize against power showing the relationship between grainsize and
critical power and the variation in critical power with increasing flow depth....................60
Figure 5.13 The relationship between critical and excess power over a range of grainsizes .................61
Figure 5.14 Schematic of model system for accounting for channel junctions......................................62
Figure 5.15 Schematic diagram showing potential to override energy balance at structures such as
dams and weirs enabling all energy to be ‘dumped’...........................................................63
Figure 5.16 General Location map of the Hawkcombe stream in Somerset, England...........................65
Figure 5.17 1:50000 OS map of Hawkcombe Stream catchment ..........................................................66
Figure 5.18 Hawkcombe Stream geomorphological reaches .................................................................66
Figure 5.19 Re-profiled culvert on the Hawkcombe Stream under the village of Porlock ....................67
Figure 5.20 Digitised drainage basin areas for each of the 17 reaches ..................................................68
Figure 5.21 Excess energy budgets for the Hawkcombe stream............................................................69
Figure 5.22 HEC RAS long profile of Culvert under Porlock Village. .................................................70
Figure 5.23 Photographs of Reach 10 (8) showing degradational features............................................71
Figure 5.24 Reach 12 (6) showing the stable nature of this channel reach. ...........................................71
Figure 5.25 General location map of the Upper Wharfe showing study reach limits ............................72
Figure 5.26 Reach Map showing location of 60 cross-sections .............................................................73
Figure 5.27 Volume difference for the 60 cross-sections on the Wharfe reach (note that in this instance
1 represents the most upstream cross-section. Negative balances indicate degradation;
positive balances indicate aggradation over the 5 year period............................................74
Figure 5.28 Method use to weight cross-section volume difference values for each 500m reach .........74
Figure 5.29 Weighted volume difference values for each 500m reach with channel long-profile ........75
Figure 5.30 Excess energy budgets for the Hawkcombe Stream ...........................................................76
Figure 5.31 HEC-RAS model long profile plot of the River Wharfe showing the highly variable bed
elevations at the river distances 3000–4000m ....................................................................77
Figure 6.1 Bed Layer Concept (iSIS 2001)..........................................................................................84
Figure 6.2 Bed Updating Options (iSIS 2001). For option 1 the whole section is updated, for option 2
the wetted section only is updated and, for option 3 deposition/ erosion is distributed
across a section according to depth to a user specified exponent .......................................85
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
ix
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 6.3 Gorai River Survey Sections from 3rd year dredging programme. Spacing of survey lines
is approximately 100m for a reach of 32km from the offtake from the Ganges.................90
Figure 6.4 iSIS model schematic with full number of sections (above) and at 4km spacing (below)
used for testing separation of sections ................................................................................91
Figure 6.5 Repeating flow and sediment boundary used in Gorai River section tests .........................93
Figure 6.6 31km Long Section of Gorai from the Gorai Rail Bridge – Full Model with 100m & 200m
section spacing....................................................................................................................93
Figure 6.7 255km Long Section of Gorai River from Ganges to Khulna measured from boat in 1992
(Halcrow 1993) ...................................................................................................................94
Figure 6.8 Long section of Gorai River from Gorai Rail Bridge using 4km section spacing ..............94
Figure 6.9 Gorai River changes in simulated bed elevation over 9 years for different section spacings
– Node GOR120 at Upstream end of reach ........................................................................95
Figure 6.10 Gorai River changes in simulated bed elevation over 9 years for different section spacings
– Node GOR240 crossing section after 2 bends .................................................................95
Figure 6.11 Gorai River changes in simulated bed elevation over 9 years for different section spacings
– Node GOR280 crossing after 3 bends .............................................................................96
Figure 6.12 Gorai River changes in simulated bed elevation ove 9 years for different section spacings
– Node GOR330 middle of fifth bend ................................................................................96
Figure 6.13 Gorai River change in sediment concentration at the start of the model and at the final
crossing ...............................................................................................................................97
Figure 6.14 Gorai River sediment rating curves at Kushtia and Gorai Railway Bridge derived in
different studies and recent sediment measurements (supplied from GRRP study) ...........97
Figure 7.1 Schematic of CAESAR’s operation..................................................................................104
Figure 7.2 Schematic of the scanning algorithm ................................................................................105
Figure 7.3 Files to be used for the River Teifi example.....................................................................107
Figure 7.4 Main CAESAR interface window ....................................................................................108
Figure 7.5 Screen showing water depths (left) and changes in elevation (right: red is erosion, green is
deposition) ........................................................................................................................109
Figure 7.6 Input file, highlighting first discharge input point and the corresponding inundation areas
caused by raising that value from 10 to 120 .....................................................................109
Figure 7.7 Image showing the section of the River Teifi modelled in Figures 7.5–7.7. This illustrates
erosion and deposition within the channel and flooplain, with the lower left hand image
showing the D50. The right hand frame shows changes in the D50 across the cross section
marked on the left hand image, as well as elevation changes that occurred during the
simulation. A more detailed explanation is provided in Van de Wiel et al. (In Press) .....111
Figure 7.8 Selection of catchment mode computation .......................................................................112
Figure 7.9 Initial drainage network for the Swale catchment simulation...........................................113
Figure 7.10 Expanded drainage network, due to rainfall event............................................................113
Figure 7.11 Example of the text output files generated by CAESAR..................................................114
Figure 7.12 Water depths draped over the DEM for the River Teifi example, created in ARC-SCENE114
Figure 7.13 Example of multiple catchment CAESAR set up, to simulate two reaches of the River
Severn, U.K (after Van de Wiel et al., In Press)...............................................................115
Figure 7.14 Comparison of iSIS modelled inundation extents and CAESAR modelled depths and
inundation areas ................................................................................................................116
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
x
Accounting for Sediment in Rivers
FRMRC Research Report UR9
1.
Accounting for Sediment in Rivers: The FRMRC
Toolbox of Methods and Models
Colin Thorne1, Nick Wallerstein2 and Philip Soar3
1.
2.
3.
1.1
Professor, School of Geography, University of Nottingham, Nottingham.
[email protected]
Research Associate, School of Geography, University of Nottingham, Nottingham.
[email protected]
Senior Geomorphologist, Jeremy Benn and Associates, Atherstone, and Industrial
Research Fellow, University of Nottingham. [email protected]
OVERVIEW
1.1.1 Background
Historically, flood alleviation schemes (FAS) in the UK have seldom taken into account the erosion,
transfer and deposition of sediment, or the effects of flood management (through either containing or
promoting flooding) on sediment transfer in the fluvial system. However, it is now recognised that
schemes that disrupt sediment transfer connectivity may require heavy maintenance, either to prevent
sedimentation from compromising the design capacity of the channel for flood conveyance, or to
prevent deterioration or failure of flood defence assets due to fluvial erosion. Further, the impact of
flood alleviation schemes may not only be local but may also trigger degradation and/or aggradation
elsewhere in the fluvial system.
Where river sediment dynamics are considered in flood defence schemes in the UK the ‘Fluvial Audit’
approach (Sear et al., 2003) is currently recommended as a means of characterising the sediment
transfer system and identifying geomorphic reaches as being sediment sources, transfers or sinks.
In this approach, a field and documentary investigation is used to divide the fluvial system into
geomorphic reaches designated as sediment source (scouring), transfer (equilibrium), or sink
(depositional) reaches. The approach rests on detailed field reconnaissance of the watercourse,
performed by an experienced fluvial geomorphologist. While the Fluvial Audit has proven very useful
in river conservation and restoration projects, in its present form it does not support the types of
quantification of the sediment dynamics required to interface effectively with the engineering
components of strategic flood studies and Catchment Flood Management Plans (CFMPs). While a
Fluvial Audit provides essential insights into catchment sediment dynamics, the approach requires
extended inputs by experienced geomorphologists and is of limited utility for strategic planning due to
its qualitative outcome. Also, the Fluvial Audit is not predictive and cannot simulate system response
to new FAS or flood management actions – limiting its utility in options appraisal.
While the qualitative Fluvial Audit is a practical approach that can be applied across a range of
catchment scales, quantitative, numerical modelling of sediment dynamics and morphological
adjustments at the catchment scale remains a difficult challenge that is yet to be solved in the context
of practical applications.
Recognising the practical limitations of existing qualitative and quantitative approaches, a component
of the long-term research aim pursued by the Flood Risk Management Research Consortium
(FRMRC) has been directed at developing new tools to account for sediment in rivers, concentrating
particularly on semi-quantitative and indicative characterisation of the dynamics of the sediment
transfer system and its response to the impacts (intentional or unintentional) of interventions in the
fluvial system made for flood alleviation purposes.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
1
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Conventionally, computation of sediment movement and imbalance is approached through application
of the equations of fluid flow, sediment transport and sediment continuity in hydrodynamic models
with a sediment module – such as iSIS Sediment or HEC-6. However, the resources and data required
to apply these models restrict their use to the reach rather than the catchment scale, while extended run
times mean that they cannot readily be used for the types of long-term simulation required to
investigate sediment movement over long periods or through large alluvial rivers. Also, accurate
sediment modelling demands both specialist training and prior experience on the part of the modeller –
not only in hydrodynamic modelling, but also in the applicability and appropriate use of different
sediment transport equations. At present (2006) there are only a few dozen academics and even fewer
practitioners in the UK who are fully competent and confident in sediment (as opposed to
hydrodynamic) modelling.
It is in the context of the limitations of conventional qualitative and quantitative methods of
accounting for sediment that the FRMRC has sought to assemble a more complete sediment toolbox
that could be useful to practitioners faced with the need to account for sediment and sediment-related
problems in river engineering and management. The approach adopted involves linking fundamental
(blue skies) research to the needs of end users (strategic research). This stems from with the overarching research philosophy of the FRMRC (Figure 1.1), which is intended to function as an industrial
consortium.
FUNDER AND END USER REQUIREMENTS
Project body – FRMRC funded by EPSRC, EA etc.
Cutting edge research and ‘User-Focussed Measurable Outcomes’
Therefore need to show that we are developing sound new scientific theory
EPSRC: ‘Blue Skies’ research:
Must Have PROOF OF CONCEPT to lead to..
Practical tools for EA (and others)
Figure 1.1
Over-arching research philosophy of the FRMRC
Consequently, the research approach has included: assembling some existing methods and attempting
to make them more accessible to practitioners; strategic research to produce new approaches that have
the potential to be useful in a UK context; and, fundamental research to develop the River Energy
Audit Scheme – a new tool that has been designed in close collaboration with end-users in the
Environment Agency.
Not only should the products of this work be genuinely useful in identifying and solving sedimentrelated problems in UK rivers, but also they can provide a window on future research needs to support
improved understanding of sediment dynamics. Hopefully, the findings reported in this document may
help guide development of the next-generation ‘whole-system’ models by indicating how they might
be made capable of including and accounting for sediment in the river system.
1.1.2 The Needs Case
The strategic need to consider sediment at the catchment or regional scales first emerged during the
late twentieth century and led to initiatives to better represent the sediment dynamics, using
approaches that are both reliable and fit the purpose. Approaches were developed in a number of
countries including, notably, Australia, in the form of SedNet:
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
2
Accounting for Sediment in Rivers
FRMRC Research Report UR9
http://www.clw.csiro.au/publications/general2002/managing_regional_water_quality.pdf
and the USA, in the form of Regional Sediment Management:
http://www.wes.army.mil/rsm/
In contrast, the needs case in the UK during the twentieth century was comparatively weak due to the
naturally lower sediment loads associated with UK rivers and the impacts of decades of well-organised
and sustained river management and maintenance designed to suppress not only sediment transfer in
the fluvial system, but, more especially, its morphological outcomes. However, the need to account for
sediment became more pressing during the first years of the twenty-first century due to:
•
•
•
a switch from flood prevention to flood risk management,
new environmental regulation and legislation, and
recognition that to be sustainable river management (including flood risk management) must
work with natural processes rather than against them – including sediment transport.
With respect to flooding, recent, wide acceptance of the need for changes in the way floods are
managed in the UK may be traced to the outcomes of the government’s Foresight Project on Flood and
Coastal Defence (Evans et al., 2004).
http://www.foresight.gov.uk/Previous_Projects/Flood_and_Coastal_Defence/Repor
ts_and_Publications/Project_Outputs/Outputs.htm
At about the same time, the EU Water Framework Directive (WFD) was being carried into law
throughout the UK, with immediate implications for river management, including flood risk
management. The new Controlled Activities Regulations (CAR) for Scotland:
http://www.sepa.org.uk/wfd/regimes/index.htm
reflect SEPA's policy on how river engineering in Scottish rivers will be regulated to ensure that
artificial impairment of their hydromorphology does not prevent them from reaching good ecological
status by 2015. Further information on UK research on hydromorphology may be obtained from
SNIFFER:
http://www.sniffer.org.uk/res_area_water_1.asp
Controlled Activity Regulations currently apply only in Scotland, but given the tight WFD timetable
they probably indicate how the regulatory framework will change in the rest of the UK – within
months rather than years.
Hydromorphological standards in the CAR stress the importance of natural channel form rather than
process, though connectivity in the fluvial system is one of the criteria covered in the morphological
standards and featured in the Morphological Impact Assessment System (MImAS) used to assess
hydromorphological condition:
http://www.sepa.org.uk/wfd/standards/index.htm
The UK Technical Advisory Group (UKtag) website has copies of the technical reports and summaries
of feedback from the consultation exercise on hydromorphology and the WFD:
http://www.wfduk.org/
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
3
Accounting for Sediment in Rivers
FRMRC Research Report UR9
One of the themes in commentary on hydromorphology standards focuses on how to manage sediment
in order to tackle the risks posed by disturbed sediment regimes – a topic which MImAS does not
address particularly well. It follows that when the hydromorphological dimension of WFD and CAR
legislation and implementation are reviewed in a few years’ time, the requirement to first, characterise,
second, understand and third, either avoid disrupting or reconnect natural links in the sediment transfer
system that have been severed by past engineering actions, is likely to become paramount.
The case for adopting a new paradigm in flood management that brings together flood defence, land
drainage, conservation, recreation and ecological functions is eloquently set out in Making Space for
Water (Defra, 2004), a consultation process that quickly moved forward the Foresight Project’s debate
on flood management from consideration of broad principles of sustainable risk reduction to practical
measures capable of merging flood alleviation goals with a range of other multi-functional objectives.
However, it appears that there is still some way to go before the common principles and common
policies that will unite the FRM and WFD communities can be defined.
It has been suggested the Performance-based Asset Management System (PAMS – a suite of methods
and tools that provide an improved decision support approach to maintaining and improving assets in
FRM systems) may provide a suitable vehicle with which to advance integrated channel management
at the appropriate operational level. This is the case because working at the catchment scale by
bringing together Catchment Flood Management Plans (CFMPs) from the FRM side and River Basin
Management Plans (RBMPs) from the WFD side will remain impractical until it is possible to
characterise and model the river’s hydromorphological and ecological systems at the catchment scale.
Irrespective of whether progress is to be made through PAMS at the operation (reach-scale) level or by
combining CFPMs and RBMPs at the catchment (system-scale) level, or both, it is imperative that the
capability to account for sediment becomes routinely available to river managers and their consultants
within a few years and certainly in time for the first review of hydromorphological standards, which
must be completed in 2011. This is the case either to allow reach-scale sediment-related problems to
be considered and solved within the catchment context, or to support the much more ambitious goal of
treating the sediment transfer system in a manner commensurate with treatment of the hydrological
system – that is, through integrated catchment management and planning.
1.2
FRMRC RESEARCH ON SEDIMENT DYNAMICS
1.2.1 Aims
In the light of the research needs identified in the run up to the inception of the FRMRC and restated
above, a Work Package (WP 8.1) in Research Priority Area concerning Geomorphology, Sediments
and Habitats (RPA-8) was tasked with developing a quantitative tool which will build on the existing
qualitative Fluvial Audit to derive an approach capable of:
•
•
•
characterising sediment source, transfer and sink areas on a reach-by-reach basis (where reaches
are defined as geomorphically consistent sub-units of a river drainage network);
representing sediment flux divergences between reaches resulting from differences between the
supply of sediment and their capacity to transport sediment; and
predicting the reach-scale response of the sediment transfer system to structural interventions
and/or management actions undertaken for flood alleviation purposes.
This tool was conceived as being a computer simulation method utilising archive and routinely
collected data readily available to the user, together with limited additional data collection.
Applications were envisaged to centre on contributing to decision support and options appraisal
processes in the contexts of:
i.
ii.
assessment of river channel stability conditions at the reach and catchment scales,
identification of flood risk management actions that are geomorphologically sustainable, and
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
4
Accounting for Sediment in Rivers
FRMRC Research Report UR9
iii.
design of flood alleviation schemes that avoid disrupting connectivity in the sediment transfer
system.
Initially, it was intended to adapt the Sediment Impact Assessment Model (SIAM) developed in the
USA for UK use but, as the work unfolded, two problems with this intention emerged. First, it soon
became apparent that due to the imitated availability of sediment data in the UK applications of SIAM
would be limited to relatively large and well-resourced projects. Second, given the need to support a
wide variety of potential applications, it was impossible for the SIAM tool, or indeed other any single
tool, to meet the diverse needs of the end user community. To address the problem of limited sediment
data availability, the focus of the work switched to development of the River Energy Audit Scheme
(REAS) in parallel with SIAM. To address the second problem, the scope of the work package was
extended to include the compilation of a suite of methods and models with the aim of creating a
versatile toolbox, rather than a single tool.
In the event, the tools which were selected for inclusion in the toolbox and which are described in this
document are:
•
•
•
•
•
•
Stream power screening tool
River Energy Audit Scheme (REAS)
Sediment Impact Assessment Model – embedded in HEC-RAS (HEC-RAS/SIAM)
Hydraulic Engineering Center – River Analysis System: Sediment Transport (HEC-RAS ST)
iSIS Sediment
Cellular Automaton Evolutionary Slope and River Model (CAESAR)
1.2.2 Study Approach
The approach adopted built on, rather than attempting to replace, the Fluvial Audit. This meant
introducing an analytical basis for assessing and characterising the sediment transfer system while
minimising subjectivity and the need for advanced geomorphic insight judgement on the part of the
end user. However, it must be recognised that due to the scarcity of sediment data for UK rivers and
the complexity of sediment dynamics even in relatively straightforward rivers and streams, the
quantitative outcomes of analytical sediment transport calculations can still usually only be indicative.
Despite this, indicative, quantitative results still provide useful information that readily meshes with
the hydrologic and hydraulic elements of flood alleviation analysis and design. In light of this, the
research performed by the FRMRC recognises that elements of both analysis and interpretation remain
essential in all sediment studies, regardless of the modelling tool employed. Hence, the methods and
models included in the toolbox all start by assuming that the user has a sound qualitative
understanding of the fluvial and sediment systems (gained from a Fluvial Audit or some equivalent
methodology) and they all rely on elements of both interpretation and analysis of the sediment transfer
system, but with varying contributions from these two components of scientific study. Figure 1.2
pictures where each of the tools might lie in the continuum between purely interpretive and fully
analytical approaches.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
5
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Interpretive
Analytical
Fluvial Audit
REAS
HEC-RAS/SIAM
HEC-RAS ST
iSIS Sediment
CAESAR
Figure 1.2
Relative contributions of interpretational and analytical approaches in different
tools included in the FRMRC Sediment Toolbox
The research approach rested on ensuring that the outcomes of sediment analysis using the methods
and models in the toolbox would aid understanding of the interactions between flood defence
infrastructure and sediment dynamics, knowledge of which is vital in assessing the sediment impacts
of existing flood alleviation schemes and for the options appraisal and detailed design of new schemes.
Further, it was intended that the toolbox should also provide the basis for end users to understand, and
therefore account for, the impacts on flood defence infrastructure of changes in the catchment
sediment system and adjustments of the fluvial system that are likely to occur in response to climate
change and socio-economic development in the coming decades.
Finally, it was desired that the toolbox support evaluation of the implications of infrastructure–
sediment interactions for in-channel habitats and the ecosystems they support. The capability to link
infrastructure, morphology, sediments and habitats when managing flood risk in the UK would
facilitate a common approach to satisfying multiple functions as integrated catchment management
gains strength in the coming years and decades.
It was specified at the outset of the FRMRC that the development of new sediment tools would be
carried through to proof of concept for later incorporation into:
•
Guidance and/or Work Instructions used by operations staff in the Environment Agency;
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
6
Accounting for Sediment in Rivers
FRMRC Research Report UR9
•
•
Project Briefs for use by independent consultants; and
Computational modules for use in existing river models such as HEC-RAS and iSIS (both these
models are UK industry standards).
The intention of the research was therefore to develop methods capable of validating and building
upon the insights gained from qualitative assessments of sediment problems and dynamics using the
Fluvial Audit, to provide an indicatively quantified, analytical framework for characterising the current
sediment system, identifying sediment-related problems, and exploring future scenarios of channel
change involving autonomous evolution of the fluvial system in response to climate change or its
response to proposed flood management interventions.
1.2.3 Project Deliverable: User Focused Measurable Outcome
The Inception Report for the FRMRC specifies at least one User Focused Measurable Outcome
(UFMO) will be delivered by each Research Priority Area and states that:
‘These outputs will not only demonstrate direct value for the industrial funders, but also
provide a vehicle for establishing strong links between the research groups and the industrial
funders/users that will be invaluable in assuring appropriate user focus throughout the
remainder of the project.’
The Sediment Toolbox broadly represents the UFMO for RPA-8, with the REAS approach being the
deliverable that is directly derived from research conducted by the FRMRC.
The Inception Report goes on to declare that:
‘To ensure that the outputs are of direct value they will all be subjected to a process to
establish “proof of concept”.’
Within the FRMRC ‘proof of concept’ defines the transformation of a scientific idea or hypothesis into
a tangible methodology, the outcomes of which have been proven through its application to and
agreement with measured data or observed behaviour of some part of the flooding system.
The idea is that, once proven, the concept may be incorporated or developed into a model or procedure
that provides the capability to address issues of uncertainty, accuracy and sensitivity of the flooding
system to changes in flood risk drivers and to provide improved science to underpin risk based
decision making.
The FRMRC Inception Report sets out criteria by which ‘proof of concept’ is to be established. These
are shown below in Figure 1.3.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
7
Accounting for Sediment in Rivers
FRMRC Research Report UR9
PROOF-OF-CONCEPT
PROOF OF THEORY
- Is the sediment transport theory employed in SIAM sound and are
the embedded assumptions directly transferable to real world
applications?
PROOF OF METHOD
- are the computational sequences internally consistent with the
theory and do the results mimic those found in the field?
PROOF OF UTILITY
- Does the tool have the potential to support pre-existing
decision making tools, and are the outputs technically accessible
to the end user
Figure 1.3
‘Proof of Concept’ criteria for FRMRC UFMOs
For each FRMRC UFMO a Lead End User from one of the funding organisations was identified at the
outset of the project. It is the responsibility academic investigators and research collaborators to
present to the Lead End User the evidence necessary to allow him or her to judge whether the criteria
set out in the bullet points above have been met. The Lead End User for RPA-8 is Jim Walker of the
Environment Agency and this document, together with the dissemination Workshop on ‘Accounting
for Sediment in Rivers’ provide the evidence base on which proof of concept will be judged.
In the case of Work Package 8.1, ‘proof of utility’ is crucial to uptake and may be disaggregated into a
number of specific considerations including:
•
•
•
•
Application Scale
Workable Levels of Detail
Input Requirements
Output Deliverables
In this context, it is important to recognise that flood risk management in the UK operates at multiple
scales ranging from:
1.
Strategic scale: large catchments where methods and models may be used for Catchment Flood
Management Plans, Catchment Scale River Habitat Objectives, WFD hydromorphology and
River Basin Management Plans
These applications require low intensity, reach scale data input. Output would necessarily need to be
indicative. Given recognition of climate change and socio-economic change in future, models must be
capable of representing both current conditions and future scenarios.
2.
Operational Scale: applications related to capital works and channel maintenance for flood
defence, land drainage and other functions.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
8
Accounting for Sediment in Rivers
FRMRC Research Report UR9
These applications require high intensity, site specific data input. Output would necessarily need to be
quantitative. Given requirements for achieving good ecological status by 2015 or for improving
ecological potential in heavily modified watercourses, outputs should mesh with models for habitats
and ecology related to biodiversity objectives.
Further, it is fundamentally important that the tools proposed for use must have clear advantages over
current practice in flood risk management in terms of:
•
•
Reduced time cost
Improved decision making
Each of these considerations has been factored into the research performed in WP8.1 and the selection
of methods and models included in the toolbox. Figure 1.4 indicates how the tools are intended to
relate to the criteria for ‘proof of utility’, and this is considered further in the chapters dealing with
each tool in detail, in the remainder of this document.
Figure 1.4
Main tools in the toolbox assessed in terms of ‘proof of utility’
1.2.4 Dealing with Uncertainty
A major tenet of the FRMRC’s ethos is that uncertainty in its research approaches and outcomes must
be recognised and managed appropriately. In the context of dealing with uncertainty in Work Package
8.1, it is pertinent to consider the following statement from Daryl Simons and Rhu-Ming Li:
‘A mathematical model is simply a quantitative expression of a process or phenomenon that is being
studied … with a computer based model a whole array of “what-if” questions can be answered with a
minimum amount of time and effort.’
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
9
Accounting for Sediment in Rivers
FRMRC Research Report UR9
However, no fluvial process can be perfectly observed or understood, let alone predicted. It follows
that, while ‘what-if’ questions can indeed be answered efficiently, the application of any mathematical
model must generate some degree of uncertainty. Uncertainty can be minimised provided that the
governing physical processes are considered in the analysis and the model is properly designed,
calibrated and verified. Model development, validation and application to analysis and design
problems require consideration of the nature of the problem, physical environment, objective of the
study, and the time, human resources and money available to support the work. Since time, human
resources and money are always limited, this will have an impact on the levels of detail and
complexity that can be included in the simulation, and the extent model calibration and validation that
can be performed.
According to Overton and Meadows (1976):
‘…if a highly complex mathematical representation of the system under study is made, the risk of not
representing the system will be minimised, but the difficulty of obtaining a meaningful solution will be
maximised as much data will be required, and mathematical handling in computer model (e.g.
convergence and consistency) and complexity of mathematical processes may even render the problem
formulation intractable. Further, the resource constraints of time, manpower and money may be
exceeded. Conversely, if a greatly simplified model is selected the risk of not representing the physical
system will be maximised, but the difficulty in obtaining a solution will be minimised.’
The conceptual relation between these parameters is shown in Figure 1.5.
Risk of not
Representing
the System
Difficulty in
Obtaining a
solution
RISK
COST
Complexity of Model
Figure 1.5
Relationship between model complexity, cost and risk of not representing the
modelled system
In assembling the methods and models to be included in the sediment toolbox, particular attention was
given to dealing with uncertainty and addressing the issues raised in the quotation from Overton and
Meadows. This helps to explain why a range of methods of models, extending from a very simple
treatment of the sediment transfer system that is easy to apply (Streampower Screening Tool) to an
advanced, transport model that routes sediment by size fraction but is costly and time-consuming to
apply (iSIS Sediment) was included. Only through provision of a suite of approaches that cover a wide
range of complexities and costs can the contrasting needs of the end user community be met.
1.2.5 Issues Affecting Take-up of the Tools
While the Environment Agency have always expected to take up and use the outcomes of WP 8.1, it
was pointed out at a early stage by EA staff on the RPA-8 Steering Panel that accreditation of new
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
10
Accounting for Sediment in Rivers
FRMRC Research Report UR9
computer programs being considered for use by the EA is difficult and there is a 2-year
operationalisation period before executable models can be placed on the Agency’s computers. Clearly,
there will be substantial time lags between achieving ‘proof of concept’, operationalisation and wide
uptake of those newer tools in the FRMRC toolbox, which involve software that is not already
approved for use on EA computers.
It should also be recognised that the technical veracity of a tool is just one of a number of criteria
which will affect its reputation amongst stakeholders and its uptake by end users – which itself may be
strongly influenced by stakeholder attitudes. In practice, a method or model will only be widely taken
up if end users can support and apply the tool within the resource constraints imposed on time, funds
and expertise (the utility of the results is only as good as their interpretation). Even when technical end
users see the need for advanced (a.k.a. expensive) sediment studies, they still have to convince other
important and influential stakeholders who may often be difficult to persuade. For example, resource
managers may not believe the additional costs of sediment assessment to be justified, while politicians
and the wider public are increasingly sceptical about the value of ‘science’ and will also only sanction
a scientific approach if they have the right balance of cognizance (they understand the principles) and
credibility (the model is not so simple that they scoff at it). These considerations impose a final, but
crucial, prerequisite for achieving wide uptake: the need to place the new method or model within the
central zone of the triangle of stakeholder requirements is depicted in Figure 1.6.
Figure 1.6
Balancing management resources, the science of predictive models and acceptance
by society
Stakeholder and end user involvement from the outset of RPA-8 ensured that these considerations
were woven into the weft of the research throughout the project. In fact, in Work Package 8.1 they
dictated the specific objectives for the research, given its bias towards achieving the strategic goals
outlined in the previous sections. The outcome is the toolbox of methods and models that is described
in the remainder of this document. In line with the arguments presented in this Overview, the toolbox
is intended to provide a suite of tools which can be used to assess sediment dynamics at the catchment
scale, making best use of available sediment data, and achieving acceptable levels of technical
veracity, resource effectiveness and stakeholder credibility.
1.3
THE NEXT STEPS
While Work Package 8.1 is coming to a close, research on accounting for sediment will continue for a
further two years under the auspices of other work packages in RPA-8 of the FRMRC. Full details of
the other three work packages may be found at the FRMRC website:
www.floodrisk.org.uk
However, for completeness, a few research lines of particular relevance are outlined here.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
11
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Within Work Package 8.2 (Morphology, Habitat and Infrastructure Interactions, which began in 2004
and ends in 2008) it is planned to test REAS and HEC-RAS/SIAM through applications to trial sites
around the UK. One trial site has already been established on the River Wharfe as part of FRMRCsponsored doctoral research being conducted at the University of Durham by Emma Waterhouse,
under the supervision of Professor Stuart Lane. Additional sites will be selected in collaboration with
end users in the Environment Agency and the River Restoration Centre at Cranfield University. At
some stage during WP 8.2, next generation versions of REAS and HEC-RAS/SIAM will be made
available to selected those end users who are willing to participate in beta-testing the models. This is
an essential step between ‘proof of concept’ and operationalising the models.
In Work Package 8.3 (Contaminated Sediments: Assessing Environment and Public Health Risks,
which began in 2004 and ends in 2008), conducted at the University of Wales Aberystwyth and the
University of Hull, work is underway in developing TRACER – a derivative of CAESAR capable of
becoming a generic tool for mapping metal contamination in valley floor environments by combining
geochemical and geomorphological data to produce GIS-based contamination hazard maps, and using
these to validate TRACER to simulate contaminated sediment dynamics over a range of timescales.
Work Package 8.4 (Sustainable Development of Floodplains and Wetlands, which began in 2004 and
ends in 2008), is concerned with sustainable use of floodplains and includes a monitoring study of
catchment runoff and sediment yield at the Pontbren cooperative in mid-Wales. This project is part of
doctoral studies being performed at the University of Nottingham by Alex Henshaw, under the
supervision of Colin Thorne. It is anticipated that REAS and/or HEC-RAS/SIAM will be applied to
the Pontbren study area to assist in developing a sediment budget for the catchment and exploring the
potential for reduced stocking densities and the introduction of buffer strips and woodlands to reduce
catchment sediment yields. If successful, this application would demonstrate an important use for the
sediment tools in the context of sustainable catchment management.
In addition to the research being performed within the FRMRC, it is important to recognise here that
delivering improvements in the way sediment is taken into account in flood risk management (as well
as other river functions) depends not only on the generation of new science through ‘blue skies’
research or even on production of a successful UFMO and but requires training river scientists,
engineers and policy makers in the need to consider sediment and the means by which to do so. It is,
therefore, highly desirable that the research outcomes of RPA-8 are fed through to e-learning and
university-based courses in a timely fashion.
1.4
REFERENCES
Defra, 2004. Making space for water:
http://www.defra.gov.uk/environ/fcd/policy/strategy.htm
Evans et al., 2004. Foresight Project on Flood and Coastal Defence:
http://www.foresight.gov.uk/Previous_Projects/Flood_and_Coastal_Defence/index.html
Overton, D. E. and Meadows, M. E. 1976. Stormwater Modeling. Academic Press, New York, NY,
358 pp.
Sear, D.A., Thorne, C.R. and Newson, M.D. 2004. Guidebook of applied fluvial geomorphology:
Defra/Environment Agency Flood and Coastal Defence R&D Programme, London, Defra Flood
Management Division, 256pp. (R&D Technical Report FD1914):
http://www.defra.gov.uk/science/project_data/DocumentLibrary/FD1914/FD1914_1147_TRP.pdf
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
12
Accounting for Sediment in Rivers
FRMRC Research Report UR9
2.
Application of a Stream Power Screening Tool
Andrew Brookes1 and Duncan Wishart2
1.
Technical Director. Jacobs Babtie, School Green, Shinfield, Reading, Berkshire RG2 9HL
[email protected]
2.
Senior Geomorphologist. Jacobs Babtie, Fairbairn House, Ashton Lane, Manchester, M33 6WP
[email protected]
This paper introduces the notion of ‘Stream Power Screening’ as a clear and transparent method of
taking an initial decision on how a river channel might be potentially affected by processes of erosion
or deposition. Whilst stream power as a concept has been developed historically by engineers and
sedimentologists, its application as a screening tool has been used more recently by geomorphologists.
This paper (1) outlines the conceptual basis for the tool; (2) provides a history of its application in
sediment analysis and sediment transport prediction; (3) outlines the origins of the screening tool; (4)
explains the current uses and applications of the tool; (5) presents a real case study; (6) looks at
uncertainties and risks; and (7) concludes with guidance on its limitations.
2.1
THE CONCEPTUAL BASIS FOR THE STREAM POWER TOOL
The term ‘stream power’ appears to have been initially introduced by Bagnold in 1966. Bagnold
defined it as the rate of work done by the fluid or the rate of energy loss per unit length of stream.
More recently stream power has been explained as flowing water having the properties of mechanical
power (Rhoads, 1987). An alternative definition is that of McEwen (1994) as the ‘rate of energy
supply at the channel bed which is available for overcoming friction and transporting sediments’.
Work is performed as the potential energy due to elevation is converted to kinetic energy. Most of the
kinetic energy is dissipated in overcoming internal and channel boundary friction. However, a portion
remains available to accomplish geomorphic work by eroding and transporting sediment. Stream
power has a significant influence on many forms and processes (Fonstad, 2003).
The rate of potential energy expenditure per unit length of channel (gross stream power) can be
defined as:
Ω = ρ g Q s Watts per metre (Wm-1); ρ is the specific weight of water; g is acceleration due to gravity;
Q is the discharge; S is the slope
Stream power can also be expressed relative to a unit of stream bed area (specific stream power). It
can be defined as a product of the shear stress τ and average velocity V. That is, ω = τ V or divide the
unit stream power Ω by channel width w:
ω = Ω/w (in units of Watts per square metre, Wm-2).
Geomorphologists have researched spatial variations of stream power. Ferguson (1981) used stream
power to better understand form and behavioural characteristics, in particular channel patterns and
meander dynamics. He produced maps of stream power at bankfull discharge for British rivers and
showed a 1000-fold range of values. Since high runoff and steep slopes go together in Britain the
geographical patterns showed a familiar division between lowlands to the south and east and uplands
to the north and west, though with a tendency for larger rivers in any area to be more powerful.
Dividing by channel width to give specific power almost eliminated this scale effect as many rivers
show a downstream decrease in power as an increase but there is still a huge lowland–upland range. A
substantial minority of British rivers were found to be actively meandering (i.e. with relatively high
stream powers). That is, rivers perceptibly eroding the outsides of bends and depositing point bars at
the insides. An even lesser number of rivers were found to have erosional and depositional activity not
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
13
Accounting for Sediment in Rivers
FRMRC Research Report UR9
concentrated at bends and point bars (ie low sinuosity, braided channels). These also had very high
stream powers. Ferguson (1981) found that active free meandering channels have specific streams
powers in the range 5 to 350 Wm-2 (with a median of about 30 Wm-2) and that feeble rivers have
bankfull powers between 1 and 60 Wm-2 (median = 15 Wm-2).
More recently, Lawler (1992) recognised that little was known about the downstream change in the
hydraulic properties of rivers. In response he devised a model for the spatial distribution of total
stream power. This suggested that total stream power is low in the headwaters of streams and increases
to a mid-catchment peak before reducing in the downstream direction. Since this theoretical paper,
empirical studies by Lecce (1997), Knighton (1999) and Reinfeld et al. (2004) amongst others, have all
shown a mid-catchment peak in total stream power. This work is currently (2004–2007) being
extended by research at the University of Birmingham (by Lawler and Knight) to define and explain
downstream change and variability in river flood power within several dynamic UK catchments. The
project has the following specific objectives:
•
•
•
•
•
•
To quantify and model longitudinal trends in river flood power at the catchment scale in a range
of UK montane, upland, piedmont and lowland catchments, developing a new method based on
established modelling approaches;
Validating the input data and deriving models for key, selected representative catchments, both in
the field and through flood data checking;
Testing for the presence and position of a mid-catchment peak in stream power;
Comparing rational regime theory and the Lawler Model to assess longitudinal distributions of
flood power;
Assessing reach scale variations in friction slope for various flood magnitudes;
Examining reach hydraulics and bank erosion to define critical levels of flood power
Although Magilligan (1992) proposed that geomorphically effective floods typically have unit stream
power values exceeding 350 Wm-2, the great variation in a controlled flood flow conditions
downstream of a dam on the Grand Canyon in 1996 makes such arbitrary thresholds of limited value
(Schmidt et al., 2001). There were areas where energy expenditure was far greater than the threshold
suggested by Magilligan (1992) and other places where the expenditure was far less. Unit stream
powers ranged between 260 and 2150 Wm-2 at 10 rapids during the low discharges of 250 m3/s.
As a concept, stream power has been used to quantify sediment transport, explain channel incision,
channel pattern (e.g. Schumm, 1977), to evaluate the success and failures of river engineering and
river restoration projects (Brookes, 1988; Brookes and Shields, 1996) and riparian habitat
development. Nanson and Croke (1992) classified floodplain types into high, medium and low energy
(stream power).
2.2
A BRIEF HISTORY OF THE USE OF STREAM POWER (TOTAL, SPECIFIC,
UNIT) IN SEDIMENT ANALYSIS AND SEDIMENT TRANSPORT
PREDICTION
This paper examines specifically the application of stream power to sediment analysis and sediment
transport prediction. There is a close history between the concept of stream power and the
development of sediment transport equations. For example, Engelund and Hansen applied both
Bagnold's (1966) stream power concept and the similarity principle to derive a sediment transport
equation. This equation has been used with moderately sorted bed materials having diameters larger
than 0.15 mm. Again, Yang derived an equation to compute concentration of the bed-material
discharge, for sand-bed streams, based on dimensional analysis and the concept of unit stream power.
Yang defined unit stream power as the rate of potential energy dissipated per unit weight of water,
expressed by the product of velocity and slope. Yang, using the same dimensional analysis and
multiple regression methods as used to derive discharge rates in sand-bed streams, also derived an
equation to compute the bed-material discharge concentration, in gravel-bed rivers. The same
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
14
Accounting for Sediment in Rivers
FRMRC Research Report UR9
definition of unit stream power is used in both the sand and gravel transport equations. Further work
by Allen (1977; 1985) found relationships between stream power and bedload transport.
2.3
THE ORIGINS AND EVOLUTION OF THE SCREENING TOOL ITSELF
Screening in environmental management is an important tool. It is used in environmental impact
assessment, for example, as a comprehensive, clear and transparent method of taking a decision
regarding whether a proposal is EIA development or not. In the UK a proposal will be EIA
development if:
•
•
•
It is listed in Schedule One of the EIA Regulations 1999
It is listed in Schedule Two and there are likely to be significant impacts on the environment
It is likely to have significant impacts on a sensitive area or receptor
Risk screening tools are typically applied in risk assessment. For example, in relation to brownfield
development sites risk screening approaches have been developed with the aim of providing a
nationally consistent means of ranking sites that are, or are suspected of being, contaminated, using
readily available information, regardless of location and who is carrying out the assessment. The
purpose of ranking a site is usually so that it may be prioritised for further investigation: only the more
complex examples involving more detailed qualitative or quantitative risk assessment (at a cost).
In a similar way stream power is advocated as a screening tool for river management, in particular to
discern in principle whether a river is likely to erode or deposit in response to intervention. It has been
applied extensively by geomorphologists and ecologists in evaluating the actual or potential success
(or otherwise) of river projects. ‘Stream Power Screening’ is recommended here as an initial decisiontool to provide some indication of whether or not more detailed sediment modelling might be required.
It has been found useful, albeit relatively crudely, to provide written support to intuitive empirical
judgements of whether erosion/sedimentation is likely to be a problem.
Sediment equations could be used to predict stability at a screening level. However stream power is a
fairly rapid means of assessment involving collection of readily accessible variables and does not
involve a decision on which sediment equation might be applicable in a particular circumstance.
Typically ‘bankfull discharge’ is calculated for the affected or potentially affected reach and bankfull
slope approximated in the field by measurement or more crudely by estimation using the contours on a
map of scale say 1:25,000. Hydraulic models such as HEC-RAS and iSIS can also be used to provide
outputs for calculation of stream power.
2.4
THE CURRENT VERSION OF THE TOOL AND ITS USES
The version of the tool currently in use is fairly crude as only limited datasets on river management
problems have been published. Producing plots of bankfull discharge against channel slope for
different channel types (to show stream power thresholds) provides a crude management tool. To date
the emphasis of the application of the tool has been in the context of attempting to pre-empt the
success or otherwise of river engineering projects and/or river restoration per se. For example, when a
meandering river is straightened potential energy of water rushing along the channel is released over a
shorter distance and in a shorter time period. The stream power is increased and the bed resistance is
further overcome thus accelerating erosion. In the absence of further intervention the channel may
widen, deepen and even being to re-meander. Erosion will proceed rapidly until the channel forming
forces come to a balance with other factors that impact on the character of the river (e.g. colonisation
by vegetation, soil structure etc).
For selected rivers in England, Wales and Denmark Brookes (1983; 1987a,b) discovered that the
subsequent adjustment of straightened river channels is strongly related to stream power, with projects
less than 15 to 25 Wm-2 likely to fail due to deposition and those with stream powers in excess of 25 to
35 Wm-2 likely to fail through excessive erosion. Only channels with very high energies regained
some, or all, of their original sinuosity in the absence of subsequent maintenance intervention. Brookes
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
15
Accounting for Sediment in Rivers
FRMRC Research Report UR9
also applied the stream power screening tool to channelised rivers in general (Brookes, 1983) and to
river restoration projects in particular (Brookes, 1990; Brookes and Shields, 1996). Generally lowland
rivers with very low stream powers were found to have a tendency to experience deposition of fine
sediments. However it was recognised that even very weak sediments in lowland river channels may
have a tendency to erode.
Although the datasets were found to have limited application, nevertheless this has been a useful
provisional tool for rivers in the UK (and in Denmark) in being able to pre-empt whether rapid and
adverse adjustment is likely to occur. The broad thresholds for the UK datasets were found to be
broadly in line with the earlier work of Ferguson (1981) which discerned between active and inactive
channels (see Section 1 above).
More recently, at the annual River Restoration Conference in Edinburgh (UK) in 2006 Brookes
‘resurrected’ the stream power screening tool and advocated its wider use in river restoration
management decisions. At this conference several researchers came forward (e.g. from Norway)
saying that they had similar success of developing datasets for their own Regions/areas (pers. comm.).
2.5
THE FOLLOWING ARE THREE EXAMPLES OF ACTUAL APPLICATION
BY BROOKES
Figure 2.1
Tilmore Brook, Hampshire (UK)
Figure 2.1 is the example of the Tilmore Brook, a straightened lowland river channel affected by urban
runoff in southern England, and undergoing severe erosion and affecting riparian properties (Brookes
et al., 2005). The hydrodynamic model allowed computation of stream power at bankfull, found to be
in excess of 40 Wm-2. Two possible desirable management solutions were either to reduce the slope by
re-meandering and recreating something approximating the original course or attenuating peak flows
in the upstream urban catchment. The arrows in Figure 2.1 demonstrate the concept. The limited river
corridor between the edge of the channel and buildings and the lack of available space for attenuation
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
16
Accounting for Sediment in Rivers
FRMRC Research Report UR9
areas within the progressively developed urbanised area meant that both options had to be relatively
rapidly dismissed. The ‘geomorphological’ solution adopted was to raise the bed of the severely
downcut channel bed and to place a series of low step weirs and a new coarse substrate to prevent
future erosion, together with associated bank stabilisation measures. This project (completed in 2004)
has been monitored for the past 2 years and has become increasingly stable as bank vegetation has reestablished and following initial localised adjustment of the channel morphology. The second example
(Figure 2.2) includes a number of river restoration projects with which the author has been directly or
indirectly involved in southern Britain. All of these have been designed with consideration of the
potential for sediment deposition, thereby negating the benefits of re-instated instream features such as
pools and riffles. The chalk river examples are taken from Brookes’s channelised rivers database (see
Brookes, 1983).
Figure 2.2
Some river restoration projects in southern Britain
The River Blackwater in Surrey and Hampshire, for example, involved 20 kilometres of remeandering of a formerly (in places) over-wide and straightened watercourse (Brookes et al., 1998).
Ironically this major project was made possible because the course had to be shifted (and renaturalised) as a consequence of a dual lane road being constructed along the centre of the Blackwater
Valley. Pool and riffle sequences (i.e. topographical highs and lows) were added to the design.
Specific stream powers at bankfull were up to 25 Wm-2, and prone to siltation/sedimentation. To
compensate for this the new channel had a narrowed low flow width. Since the channel also had a
design flood capacity to maintain a multi-stage channel (with a flood berm) was created. However,s a
novel design was applied to the riffles. They were made as steep as possible and with a low flow slot
to try to maintain them clear of fine sediment (thereby increasing the range of habitats in the
watercourse).
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
17
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 2.3
Slope - bankfull discharge plots for several steep upland catchments in Britain
The final example (Figure 2.3) includes a number of steeper catchments in upland Britain. These are
plotted on a graph showing Ferguson’s actively meandering channels (Ferguson, 1981). By
comparison all of these engineering projects have the potential to actively meander (having specific
stream powers of 35 Wm-2 and above) and this is a criterion taken into account in designing the final
solutions to particular problems. Stream power screening has been used on a number of key projects in
the UK to allow an assessment of the potential for bed scour at or near bankfull flows, thereby
assessing the reliability of model results already obtained. Stream power can be calculated relatively
rapidly over long lengths of river channel. Through such work is has been possible to give some
insight (at a strategy level) into the potential for erosion or deposition to support conclusions drawn
from sediment models (such as iSIS-sediment). In some regulated lowland (low energy) rivers it has
been indicated that erosion of gravel beds is unlikely even at bankfull, although other bed materials
(such as sand) may have a potential to move. On some lowland regulated rivers aggradation will occur
even at bankfull events as large amounts of sediment are delivered. In reality, rates of deposition will
vary between sub-reaches and more locally around shoals and hydraulic structures. Locally (for
example below man-made structures such as weirs) stream power is sufficiently high to explain
observed variations in bed material composition.
2.6
RIVER CALDEW CASE STUDY
The River Caldew rises on Skiddaw in the English Lake District and is an upland gravel-bedded river.
There was severe flooding in the town of Carlisle in January 2005, prompting a flood study by the
Environment Agency.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
18
Accounting for Sediment in Rivers
FRMRC Research Report UR9
2.6.1 Issue
•
•
•
•
•
Sediment build-up in the channel within the urban area of Carlisle is thought to contribute to
flood risk
In the past this was managed by dredging and regrading
However the river is now an SAC and SSSI and English Nature do not wish to see heavy channel
maintenance unless it can be demonstrated that there is a clear flood risk associated with
sediment build-up
This analysis is part of a wider geomorphological assessment (Geo-dynamics Assessment) into
the sediment dynamics of the River Caldew over 14 km to determine the current sediment regime
and likely future changes, to determine if future channel maintenance will be required and if so
how much and when
The geomorphology study is part of a wider on-going study into flood management in Carlisle.
2.7
RESULTS
Figure 2.4 (below) shows variations of specific stream power along the River Caldew.
300
1
3
2
4
5
6
URBAN AREA
150
100
Holme Head weir
200
Cummersdale weir (collapsed)
Buckabank weir
Specific Stream Power Wm-2
250
50
35 Wm-2
0
..
1.
..
1.
..
1.
..
1.
..
1.
..
1.
..
1.
..
1.
..
1.
00
95
00
90
00
85
00
80
00
75
00
70
00
65
00
60
00
55
00
50
00
45
00
40
00
35
00
30
00
25
00
20
00
15
00
10
0
50
0
Distance downstream (m)
Figure 2.4
Variations of specific stream power along the River Caldew
2.7.1 Key Points
The specific stream power exceeds 35 Wm-2 in all locations except for a section approximately 1 km
long upstream of Holme Head Weir. Here the channel is widening through erosion.
The river is a wandering gravel-bed river with channel change (large scale erosion and deposition)
concentrated in discrete locations. In geomorphological terminology these are often referred to as
Sedimentation Zones. Sediment deposition and bank erosion are concentrated in the locations where
stream power is relatively low. Although seeming counterintuitive, this is because the sections with
high power act as efficient sediment transport sites, and channel erosion is prevented by factors such
as tree lining. However, where trees have collapsed or some other factor has promoted local erosion,
the high stream power causes bank erosion that widens the channel. This progressively reduces the
stream power and encourages sediment deposition. However, because the stream power is still
relatively high, reworking of the channel deposits and localised bank erosion occurs. The bank erosion
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
19
Accounting for Sediment in Rivers
FRMRC Research Report UR9
is actually encouraged by the sediment within the channel as it concentrates flow against the river
banks resulting in a trend towards increasing sinuosity and channel migration. In addition, occasional
flows greater than bank full also contribute to the reworking of sediments in these active zones.
Six key sites of interest along the study reach are highlighted on Figure 2.4:
1.
2.
3.
4.
5.
6.
2.8
This is an area of pronounced channel widening and sediment deposition, with active bank
erosion.
This is an area of recent bank erosion due to the breakdown of tree lining during two recent
floods (January 05 and October 05). Here the channel is beginning to increase its width and
sinuosity. It is anticipated that this will lead to a gradual reduction in stream power and increased
sedimentation which will then promote further increases in bank erosion through flow deflection.
This is an area of channel widening, involving significant bank erosion. The channel in this
location appears to be adjusting to past channel modification (gravel extraction and channel
regrading).
This is an area of pronounced sediment deposition immediately downstream of the former
Cummersdale Weir. The high stream power immediately upstream of the weir stems partly from
weir collapse, but also due to channel confinement by bank protection designed to stop channel
migration which has effectively narrowed the channel. It is likely that the high stream powers in
this location will promote continued channel change (erosion and deposition) locally.
This an area of low channel gradient, caused by the presence of a high weir. The weir is
approximately 160 years old and gradual adjustment of the channel to this bed level constraint
has caused a pronounced lowering of channel gradient upstream. Recent high magnitude floods
have increased the sediment supply to this section of channel, which because of the low stream
powers in this area, has formed several large gravel bars. Flow deflection around these channel
deposits has resulted in some bank erosion. It is anticipated that bank erosion will continue but at
a relatively low rate, due to the low stream powers. This area is acting as a relatively efficient
sediment sink which is limiting the supply of sediment into the urban area.
This is an area of long term sediment accumulation in the river channel which has been a cause of
significant concern to the Environment Agency. The sediment is deposited around and
downstream of the piers of two disused railway bridges. This sediment has however stabilised as
a result of vegetation colonisation. Consequently the active channel width has been reduced and
the Stream Power is relatively high. In general the channel through the urban area of Carlisle has
readjusted to realignment, widening and deepening which took place during the 18th and 19th
centuries through the more recent formation of berms which have reduced the active channel
width. The river channel appears to have adjusted to a more stable state in the urban area. This is
supported by the Stream Power results which reveal relatively limited potential for sediment
deposition. Existing sediment accumulations are localised and concentrated around channel
structures, where small scale hydraulic effects promote localised sediment deposition.
SYNTHESIS
The River Caldew is characterised by an alternating pattern of dynamic and stable reaches. The 10 km
section of river upstream from Carlisle has 5 reaches where stream powers are low; these areas are
characterised by coarse sediment storage with localised bank erosion. A sediment budget analysis
reveals that in these reaches bank erosion mainly results in fine sediment supply with some coarser bed
sediments. The relatively low stream power in these reaches reduces the effectiveness of sediment
transfer downstream and therefore limits the wider impact of the erosion in these locations.
This preliminary analysis suggests that dynamic reaches along the lower Caldew could be allowed to
evolve naturally. The fine sediment released by bank erosion does not lead to sedimentation in the
urban area (this is coarse sediment). Channel widening and lateral migration causes a net storage of
coarse sediment rather than an increase in coarse sediment supply. This view is supported by the
stream power results. Actions such as bank protection can result in restriction of the active channel
width, leading to an increase in stream power and a reduction in the effectiveness of sediment storage.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
20
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Ultimately such activities can result in a switch the dominant function of the reach toward efficient
downstream sediment transfer, rather than storage. Ultimately this could increase the volume of
sediment entering the urban section of the channel, potentially reducing the capacity of the channel
and increasing flood risk.
2.9
NEXT STAGE
A 1-D iSIS Sediment Model of the reach will be produced. The stream power screening will be
augmented by the sediment modelling but it will also provide a useful check on the results of the
model. The two approaches are inter-dependent, each informing the other.
Figure 2.5
2.10
River Caldew, Carlisle
DEALING WITH UNCERTAINTIES AND RISKS ASSOCIATED WITH USING
THE STREAM POWER SCREENING TOOL
Together with a calculated stream power value it is also important to have knowledge of sediment type
to give a broad indication of whether a channel has the potential for erosion or deposition. This will
lead to completion of projects that should require less ongoing maintenance or that have been designed
to adjust in an adaptive fashion. Thresholds of critical stream power depend on the bed and bank
materials. That is, more powerful rivers can erode more resistant materials and vice versa.
Several authors have highlighted the importance of the size of bed material in relation to stream power
(e.g. Fonstad, 2003). An increase in the silt/clay content of the banks may provide narrower than
expected channels at a particular location, with higher than predicted stream power values (Fonstad,
2003). The river channel has been described as a ‘jerky conveyor belt’ of sediment movement,
removing material eroded from hillslopes. This description highlights the complexities involved in
fluvial sediment transport. Whilst there is a good understanding of the physics and concepts of particle
entrainment and motion, the complexity and diversity of river channels has led to a poor application of
this knowledge to prediction. Therefore the selection and use of the correct sediment transport
approximation will be vital to the success of a model. A key issue is the choice of model – often a
particular model is only robust when compared to its own dataset (Gomez and Church, 1989).
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
21
Accounting for Sediment in Rivers
FRMRC Research Report UR9
2.11
GUIDANCE ON ITS LIMITATIONS
The tool should be applied to alluvial channels with discernible floodplains. It should be recognised as
a relatively simple screening tool, perhaps to support intuitive judgement of how a river channel might
behave prior to, during, or following intervention and to aid a decision on whether more sophisticated
(and costly) sediment modelling is required.
So far there have been very limited datasets produced and it is however important not to take stream
power thresholds (e.g. 35 Wm-2) developed from these prescriptively (i.e. not to make management
decisions in a ‘cookbook’ fashion). Whilst it is bona fide for river managers/researchers to quote
published work, the limitations of that work must be adequately stated. Authors in Australia, USA,
Japan and elsewhere in the UK have quoted Brookes’ work in undertaking their own interpretations of
stream power (see reference list). There are a number of health warnings related to this:
•
•
•
Brookes did not develop specific single thresholds for the rivers that he studied in the UK and
Denmark. Rather (and in confirming the earlier work by Ferguson 1981) he found that there was
a ‘grey area’ between approximately 25 and 35 Wm-2 where the river channels studied could not
be classed as either eroding or depositing;
Brookes’ datasets and interpretations are specific to the downstream impacts of river
channelisation, adjustments occurring within channelised reaches (especially straightened
channels) and river restoration projects. He has never advocated the use of the tool to anticipate
adjustments in other environments and to, for example, other types of incised or incising
channels;
The largest limitation of the published work by Brookes is that bed and bank materials (i.e.
sediment types) are most certainly significant variables likely to determine whether a channel
erodes or not.
Brookes has recommended that river managers in other parts of the world develop their own datasets
peculiar to particular sediment types/regions/ problems. Datasets could be extended fairly rapidly
using high resolution spatial data (e.g. Worthy, 2005). A challenge of looking at variation within a
catchment, or along a reach, is that data of the highest possible resolution show the real and underlying
complexity of fluvial processes that point observations fail to predict. The lack of correlation between
predicted and observed stream power values has sometimes been attributed to complexities of stream
morphology at the local scale (e.g. Worthy, 2005).
2.12
ACKNOWLEDGEMENT
The views expressed in this paper are those of the authors and not necessarily those of their employer,
Jacobs Babtie. If you wish to use any of this material in publications please contact Dr Andrew
Brookes beforehand ([email protected])
2.13
REFERENCES AND FURTHER READING
Allen, J.R.L. (1977) The plan shape of current ripples in relation to flow conditions, Sedimentology, v.
24, pp. 53-62
Allen, J.R.L. (1985) Principles of Physical Sedimentology, George Allen and Unwin, London
Bagnold, R.A. (1966) An approach to the sediment transport problem from general physics.,
Professional Paper. United States Geological Survey, 422
Brookes, A. (1983) River Channelization in England and Wales: Downstream consequences for the
channel morphology and aquatic vegetation, PhD Thesis, University of Southampton, UK
Brookes, A. (1987) The distribution and management of channelized streams in Denmark., Regulated
Rivers, 1- 3-16
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
22
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Brookes, A. (1987) River channel adjustments downstream from channelization works in England and
Wales, Earth Surface Processes and Landforms, 12: 337-351
Brookes, A. (1990) Restoration and enhancement of engineered river channels: some European
experiences, Regulated Rivers: Research and Management, 5: 45-56
Brookes, A. (1988) River Channelization: perspectives for environmental management., John Wiley
and Sons, Chichester, UK
Brookes, A. and Shields, F.S. (eds) (1996) River Channel Restoration: Guiding principles for
sustainable projects., John Wiley and Sons, Chichester, UK
Brookes, A., Downs, P. and Skinner, K. (1998) Uncertainty in the Engineering of Wildlife Habitats.,
Journal of the Chartered Institute of Water and Environmental Management., 12, 25-29
Brookes, A., Chalmers, A. and Vivash, R. (2005) Solving and urban erosion problem: Tilmore Brook,
Hampshire., Journal of the Chartered Institute of Water and Environmental Management.,
Cohen T.J. and Brierley G.J (2000) Channel instability in a forested catchment: a case study from
Jones Creek, East Gippsland, Australia., Geomorphology, 109-128(20)
Colby, B.R (1964) Discharge of sands and mean-velocity relationships in sand-bed streams.,
Professional Paper. United States Geological Survey, 462A- 47pp
Engelund, F. and Hansen, E. (1967) A Monograph on Sediment Transport. Technisk Forlag,
Copenhagen, Denmark
Ferguson, R.I (1981) Channel form and channel changes, In Lewin, J. (ed) British Rivers, George
Allen and Unwin, London. Chapter 4, 90-125
Fonstad, M.A. (2003) Spatial variation in the power of mountain streams in the Sangre de Cristo
Mountains, New Mexico., Geomorphology, 55, 75-96
Gilvear, D.J. and Bradley, S. (1997), Geomorphic adjustment of a newly constructed ecologically
sound river diversion on an upland gravel bed river; Evan Water, Scotland, Regulated Rivers, 13, 2, 113.
Gomez, B. and Church, M. (1989) An assessment of bed load sediment transport formulae for gravel
bed rivers, Water Resources
Knighton, A.D. (1999) Downstream variation in stream power, Geomorphology, 29, 293-306.
Lawler, D.M.. (1992) Process dominance in bank erosion systems, In Carling, P.A. and Petts, G.E.
(eds), Lowland Floodplain Rivers: Geomorphological Perspectives, John Wiley, Chichester, 117-143.
Lawler, D.M.. (1995) The impact of scale on the processes of channel-side sediment supply: a
conceptual model, In Osterkamp, W.R. (eds), Effects of scale on the interpretation & management of
sediment & water quality, International Association of Hydrological Sciences Publications No. 226,
175-184
Lecce, S.A. (1997) Nonlinear downstream changes in stream power on Wisconsin's Blue River,
Annals of the Association of American Geographers, 87(3), 471-486
Magilligan, F.J. (1992) Thresholds and the spatial variability of flood power during extreme floods,
Geomorphology, 5:373-390
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
23
Accounting for Sediment in Rivers
FRMRC Research Report UR9
McEwen, L.J. (1994) Channel planform adjustment and stream power variations on the middle River
Coe, Western Grampian Highlands, Scotland, Catena, 21, 357-374
Morris, P.H. & Williams, D.J. (1997) Exponential longitudinal profiles of streams, Earth Surface
Processes & Landforms, 22, 143-163.
Nakamura, F., Sudo, T., Kameyama, S and Jitsu, M (1997) Influence of channelization on discharge of
suspended sediment and wetland vegetation in Kushiro Marsh, Northern Japan., Geomorphology, 18:
279-289
Petts, G.E and Calow, P (eds) ( 1996) River Flows and Channel Forms, Blackwell Science, Oxford,
UK, 262pp
Rana, S.A., Simons, D.B. and Mahmood, K. (1973) Analysis of sediment sorting in alluvial channels,
Journal of the Hydraulics Division – Proceedings of the American Society of Civil Engineers, 99,
1967-1980
Reinfelds, I., Rutherford, I.D and Bishop, P (1995) History and effects of channelization on the
Latrobe River, Victoria., Australian Geographical Studies, 33, 60-76
Reinfelds, I. Cohen, T. Batten, P. and Brierley, G. (2004) Assessment of downstream trends in channel
gradient, total and specific stream power: a GIS approach, Geomorphology, 60, 403-416
Rhoads, B.L. (1987) Stream power terminology, Professional Geographer, 39(2), 189-195
Schmidt, J., Parnell, R., Grams., P.E., Hazel, J.E., Kaplinski, M., Stevens, K. and Hoffnagle, T. (2001)
The 1996 Controlled flood in Grand Canyon: flow, sediment transport and geomorphic change,
Ecological Applications, 11 (3), 657-671
Schumm, S.A. (1977) The Fluvial System., John Wiley and Sons, New York, 338pp
Urban, M.A. and Rhoads, B.L (2003) Catastrophic human-induced change in stream channel planform
and geometry in an agricultural watershed, Illinois, USA.,Annals of the Association of American
Geographers. 93 (4): 783-796.
Worthy, M. (2005) High resolution total stream power estimates for the Cotter River, Namadgi
National Park, Australian Capital Territory; Regolith 2005 – Tens Years of the Centre for Resource
and Environmental Studies, CRC LEME, Australian National University, Canberra, Australia, 338343
Yang, C.T. (1972) Unit Stream power and sediment transport., American Society of Civil Engineers.,
Journal of the Hydraulics Division., 98. 1804-1826
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
24
Accounting for Sediment in Rivers
FRMRC Research Report UR9
3.
Sediment Impact Assessment Model (SIAM)
David S. Biedenharn1, Stanford A. Gibson2, Charles D. Little, Jr.3, David M. Mooney4, Colin R.
Thorne5, Nick P. Wallerstein6 and Chester C. Watson7
1.
2.
3.
4.
5.
6.
7.
3.1
Research Hydraulic Engineer, U.S. Army Corps of Engineers, Engineer Research Development
Center, Coastal and Hydraulics Laboratory, 3909 Halls Ferry Road, Vicksburg, MS 39180,
[email protected]
Research Hydraulic Engineer, U.S. Army Corps of Engineers, Institute for Water Resources,
Hydrologic Engineering Center, 609 Second Street, Davis, CA 95616-4687,
[email protected]
Research Hydraulic Engineer, U.S. Army Corps of Engineers, Engineer Research and
Development Center, Coastal and Hydraulics Laboratory, 3909 Halls Ferry Road, Vicksburg, MS
39180, [email protected]
Hydraulic Engineer, U.S. Bureau of Reclamation, Denver Technical Service Center, Building 67,
P.O. Box 25007, D-8540, Denver, CO 80255-0007, [email protected]
Professor, School of Geography, University Park, University of Nottingham, Nottingham NG7
2RD, England, [email protected]
Post Doctoral Research Fellow, School of Geography, University Park, University of
Nottingham, Nottingham NG7 2RD, England, [email protected]
Professor, Colorado State University, Engineering Research Center, Fort Collins, CO 80523,
[email protected]
BACKGROUND
As water resource projects become more complex, there is a growing emphasis on the ability to
implement effective regional sediment management. A common goal of many regional sediment
management projects is the reduction of sediment loading from the watershed. This is often
accomplished with rehabilitation features such as grade control structures, bank stabilisation, drop
pipes, dams, and land treatments. While these features are often implemented to reduce sediment
yields to downstream areas, the spatial and temporal impacts of these features on the sediment regime
of the system are far from straightforward, and often result in unexpected morphologic changes in the
channel system. Therefore, the challenge in regional sediment management projects is to select the
appropriate sediment management features that produce the desired reductions in sediment delivery
while minimising the disruption to the stability of the channel systems. To facilitate this decisionmaking process, the Sediment Impact Assessment Model (SIAM) has been developed to provide for
rapid assessment of the impact of sediment management activities on sedimentation trends. SIAM is
viewed as a screening tool for the assessment of multiple rehabilitation alternatives, particularly in the
reconnaissance and feasibility phases of a project. It provides a framework to combine sediment
sources and computed sediment transport capacities into a model that can evaluate sediment
imbalances and downstream sediment yields for different alternatives. The development of SIAM
includes the incorporation of the model into the ‘Hydraulic Design’ module of the Hydrologic
Engineering Center’s River Analysis System (HEC-RAS), which is an on-going task at the time of this
writing. The implementation of SIAM into HEC-RAS will allow users to utilise the popular, widely
used hydraulic modelling system for stream network development and data entry for SIAM
applications. In addition, sediment impact assessments with SIAM can easily be conducted for systems
where existing HEC-RAS models are available. A summary of SIAM and related application has been
presented by Gibson & Little (2006), and much of the following paragraphs are taken from that
reference.
3.2
SIAM DESCRIPTION
SIAM was initially developed through the Corps of Engineers Engineer Research Development Center
(ERDC) research activities conducted with Colorado State University on channel stability in small
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
25
Accounting for Sediment in Rivers
FRMRC Research Report UR9
watersheds. David M. Mooney, who was at that time a Ph.D. candidate with Professor Chester C.
Watson as his advisor, wrote the original computer code for SIAM as part of his research (Mooney
2006 – in progress). The objective of model development was to create a tool that would combine
sediment, hydrologic, and hydraulic information for a channel network and determine the average
annual sediment budget for the system. In general, SIAM performs reach average sediment transport
computations by grain size class, and integrates the computed transport rates with flow duration
information to compute an average annual sediment transport capacity in tons per year. Computed
average annual sediment transport capacity is compared with the average annual inflowing sediment
load to evaluate sediment continuity for the reaches in the system. This provides the means to assess
the potential impact of local alterations in sediment regime on channel stability.
3.2.1 Model Computation Methodology
SIAM treats a stream network as a series of user-defined sediment reaches. Sediment reaches are
typically delineated based on observed locations of significant geomorphic change such as tributary
locations, changes in channel gradient, planform and geometry, and shifts in sediment composition.
Computations of sediment supply and transport are conducted on a reach-by-reach basis and are
representative of the average annual conditions for each reach.
In addition to reach-based computations, SIAM sediment computations are also conducted by grain
size class. The sediment gradations are divided into fractions with a single representative grain
diameter, and sediment transport and supply calculations are conducted independently for each
fraction. This accounting by grain size allows the fate of specific size sediments to be observed
throughout the system.
The grain size accounting also allows the tracking of wash material and bed material within the
system. SIAM determines whether sediments within a system are wash material or bed material based
on a user-defined wash load threshold diameter. Changes in the wash load threshold diameter permit
sediment that is wash material in one reach to transition into bed material in a downstream reach, and
vice versa. The wash load threshold diameter is typically determined following Einstein (1950). The
value of this model feature is illustrated by considering a channel where the upstream reach is very
steep and the channel bed material is correspondingly very coarse, but the downstream reach is
significantly less steep and the bed material is much finer. Coarse sands may be included in the wash
material of the upstream reach due to a larger wash load threshold diameter. In the downstream reach
that is less steep, the wash load threshold diameter is smaller, thus the coarse sand will transition into
the bed material. The coarse sand load would have little morphological impact on stability of the
upstream reach as wash material, but would have much more impact on the downstream reach as bed
material. This demonstrates how modification of a sediment source by a given management practice
could have little effect on channel stability in one reach but have significant effect on stability in
reaches farther downstream.
The original version of SIAM used sediment property records to represent bedrock, cohesive, sand,
gravel, or mixed substrates under armoured or un-armoured conditions. The cohesive control and
armouring capabilities are not available in the HEC-RAS SIAM at this time.
3.3
DATA REQUIREMENTS
The SIAM process requires developing input records for each sediment reach that describe bed
material composition, sediment properties, hydrology, hydraulics, and sediment loading from local
sources (Figure 3.1). The bed material records define the percentage of sediment present in the channel
bed for each grain size class. Sediment property records are used to set the threshold between wash
material and bed material, and to select the sediment transport function. Hydrology records define the
discharges and corresponding durations that are representative of an average annual hydrologic cycle.
The hydrology records are populated with discharge values corresponding to each flow profile in the
HEC-RAS model. Hydraulic records list the reach-averaged HEC-RAS hydraulic parameters of depth,
area, velocity, hydraulic radius, wetted perimeter, top width, friction slope, and roughness for each
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
26
Accounting for Sediment in Rivers
FRMRC Research Report UR9
flow profile. The local source records define the sediment sources and corresponding loadings from
channel and watershed sources such as eroding channel banks, gullies, upland surface erosion, and
point sources such as sand and gravel mining operations.
Sediment Reaches
Global Data Tables
Bed Material
{
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
Hydrology
{
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
Hydraulics
{
{
Figure 3.1
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
Local Sources
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
Loading Templates
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
~~~~~~~~~~
Input data requirements for SIAM
3.3.1 Model Output
The SIAM outputs consist of local bed material balance, average annual transport capacities, bed
material and wash material supplies, and local sediment supply totals for each sediment reach. Local
bed material balance is defined as the difference in the bed material supply and the average annual
transport capacity for a sediment reach. A negative local balance indicates excess transport capacity
and thus erosion potential for a reach, whereas a positive local balance indicates excess supply and
potential for deposition. All output is listed by grain size class as well as total values for each sediment
reach.
3.3.2 Capabilities and Limitations
SIAM provides sediment managers an intermediate assessment tool between qualitative evaluations
and comprehensive mobile boundary numerical models. Funding constraints and limited resources
often preclude the wide-scale use of complex numerical models in a study. With SIAM, users can
rapidly evaluate the effectiveness of proposed sediment management techniques and identify which
techniques may be candidates for more detailed investigation, thus providing cost savings to the
project. Also, the data input structure of SIAM allows individual sediment sources to be easily entered
and/or modified, allowing the user to quickly alter sediment loadings to reflect various sediment
management techniques. The incorporation of SIAM into HEC-RAS makes the model available to the
engineering community in a familiar format with continuing user support.
SIAM provides a tool for assessing sediment continuity for a single, defined condition. Channel
geometry is not updated based on erosion or deposition, so the results are only indicative of a single
channel configuration for the entire period of record being analysed. Since SIAM is a reach-based
model that uses reach-averaged parameters and produces reach-averaged results, information on
specific locations of erosion/deposition cannot be determined.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
27
Accounting for Sediment in Rivers
FRMRC Research Report UR9
3.4
INCORPORATING SIAM INTO HEC-RAS
Since SIAM directly utilises HEC-RAS outputs, it was advantageous to incorporate SIAM directly
into the HEC-RAS framework. This enables a SIAM user to conduct the entire study in a single
program, and utilise existing HEC-RAS models, which commonly exist, as a foundation for new
SIAM evaluations. The combination of these programs also facilitates integrated output and results
analysis capabilities.
3.4.1 SIAM Input
SIAM can be accessed through the ‘Hydraulic Design’ menu in HEC-RAS versions 3.2 and later. The
user interface populates the HEC-RAS schematic and provides a series of tabs under which the
pertinent SIAM input data can be entered (Figure 3.2). Sediment reaches are specified and appropriate
bed materials, sediment properties, and sources are attributed to each sediment reach. Most of these
data sets can have a one-to-one relationship with the reaches (e.g. each sediment reach has a bed
material gradation) or can be specified once and applied globally (e.g. a single sediment properties
designation, including transport function and fall velocity method can be applied to all reaches).
The five major templates for input data are depicted in Figures 3.2 through 3.4. Each sediment reach
must have a bed gradation to compute proportional grain fractions for transport capacity computations
(Figure 3.2). HEC-RAS will populate the hydrology dialog (Figure 3.3a) with the flows corresponding
to the sediment reach for each specified profile. The user then associates a duration with each profile
to distribute this hydrologic record over a statistically average year. Next a few basic sediment
properties including a transport function are specified on the sediment properties tab (Figure 3.3b).
The hydraulics tab is automatically populated by HEC-RAS (Figure 3.3c). A single set of hydraulic
parameters is associated with each reach for each profile. HEC-RAS computes weighted averages of
hydraulic parameters by prorating the value at each cross section by the length of the associated
control volume as a percentage of the total reach length. Finally, users specify sediment sources
(Figure 3.4). Annual load by grain size is entered for each source, which can then be associated with
multiple sediment reaches and modified by means of a multiplier. Following complete specification of
the input data, the ‘Compute’ button will write a SIAM input file and launch the stand-alone program.
Figure 3.2
HEC-RAS interface for SIAM with the bed material input tab
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
28
Accounting for Sediment in Rivers
FRMRC Research Report UR9
(a) hydrology
(b) sediment properties
(c) hydraulics
Figure 3.3
SIAM data tabs in the HEC-RAS interface
Figure 3.4
Source specification dialog and tab in the HEC-RAS interface for SIAM
3.4.2 SIAM Output
SIAM currently generates a binary output file that HEC-RAS reads and makes available through
several user output options. A range of tables and graphs are available for analysis of results following
a computation. The primary SIAM output is ‘Local Balance’, which reports magnitude of the average
annual tendency of a reach to fill or scour. The local bed material balance plot for two alternatives,
reported by reach, is depicted in Figure 3.5. HEC-RAS can also report SIAM output in graphical or
tabular form by grain size as shown in Figure 3.6. Deficits and surpluses reported in tabular form are
colour coded to quickly identify expected aggradation or degradation, with any computed intervals
falling within a user-specified ‘equilibrium tolerance’ reported in a third colour (Figure 3.6). Since the
local balance can be driven by reach length, a user can also select output by ‘Normalized Local
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
29
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Balance’ which translates reach deficits or surpluses into tons/year/linear foot of channel. This output
is more directly related to aggradation and degradation, and allows more general comparison of impact
between reaches. Other output options include grain-specific transport potentials, sediment source and
supply information, and breakdowns of wash and bed material.
Figure 3.5
SIAM output plot of local bed material balance
Figure 3.6
SIAM output table of local balance by grain size
3.5
JUDY’S BRANCH CASE STUDY (WATSON & EOM 2003)
Judy’s Branch is a tributary to Cahokia Canal, which is located near Glen Carbon, Illinois (and
Madison County), and is located across the Mississippi River from St. Louis, Missouri. Cahokia Canal
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
30
Accounting for Sediment in Rivers
FRMRC Research Report UR9
drains into Horseshoe Lake through an overflow feature at Horseshoe Lake State Park. The Judy’s
Branch watershed area is 8.64 sq mi and the total stream length of the project encompasses
approximately 14.5 mi.
The East St. Louis and Vicinity, Illinois Flood Protection Project was authorised through
Congressional action in 1965 and in 1974. The Water Resources Development Act of 2000 again
modified the project authorisation. The principal goal of the presently authorised project is to identify
potential improvements that will enhance habitat quality and sustainability, while also providing
incidental ecosystem services, such as flood-damage reduction. A specific sediment-reduction goal is
to reduce sediment yield by 70%. Two choices that are included in a range of alternatives for
controlling sediment delivery to downstream wetlands are: (1) excavate a large basin to retain
sediment, or (2) control the sediment sources upstream. A large sediment basin would require
continuing maintenance to excavate the accumulating sediment and would require a disposal site.
Control of the upstream sediment sources provides the opportunity to enhance instream aquatic and
riparian habitat, and reduces the dependence on long-term maintenance. In addition, as the watershed
urbanises, larger flood peaks will exacerbate channel instability.
A number of options were proposed to control the upstream sediment sources and thereby reduce the
sediment loadings to the downstream reaches of Judy’ Branch by 70%. These included vegetative
filter strips (VFS), sediment basins (SB), and drop structures (DS). SIAM was utilised to assess the
impacts of these alternatives. The network of the Judy’s Branch model consists of 48 sediment reaches
with a maximum of three levels of tributaries. Sediment reaches were created at tributary junctions and
at points of significant change in hydrology. Figure 3.7 shows the sediment yield reductions for the
various alternatives as determined by SIAM. The goal of the project was to reduce sediment loads at
station JB-34 by 70%. As shown in Figure 3.7, this would require the implementation of all three
measures (VFS, SB, and DS). SIAM not only provided a rapid assessment method to evaluate the
existing conditions and the three sediment control alternatives, but was also used to assess the potential
channel stability problems resulting from future urbanisation. This was accomplished by modifying
the flow duration data in SIAM to reflect increased runoff. The results indicated continued
urbanisation of the watershed would cause future channel stability problems, and that a system of drop
structures (grade control) would be needed to ensure long-term stability of the channel system.
Sed Yield-Existing
Sed Yield-with SB
Sed Yield-with SB & VFS
Sed Yield-with SB VFS & DS
% Reduction in Sed Yield
35000
Sediment Yield (tons/yr)
30000
100%
90%
80%
25000
70%
60%
20000
50%
15000
40%
30%
10000
20%
5000
JB - 14ENab
JB - 1314
JB - 1112
JB - 911
JB - 89
JB - 78
JB - 67
JB - 56
JB - 45
JB - 34
JB - 23
JB - 12c
JB - 12b
JB - 12a
0
10%
0%
---> U/S
Figure 3.7
Sediment yield reduction along Judy’s Branch (Watson & Eom 2003)
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
31
Accounting for Sediment in Rivers
FRMRC Research Report UR9
3.6
CONCLUSIONS
SIAM is being implemented into HEC-RAS with the aim of providing planners and designers an
easily usable means of integrating sediment continuity concepts into stream rehabilitation and best
management practices. It has the capability to be a very effective and easily applied tool for evaluating
sediment management alternatives on a watershed scale where the application of complex numerical
routing models may be impractical. The incorporation of SIAM into the proven and user-friendly
environment of HEC-RAS greatly enhances the option for managers to address the impacts of
sediment supply and transport in an expedient and cost-effective manner.
3.7
REFERENCES
Einstein, H.A. 1950. The bed-load function for sediment transport in open channel flows. U.S.
Department of Agriculture, Soil Conservation Service, Technical Bulletin No. 1026.
Gibson, S.A. & Little, Jr., C.D. 2006. Implementation of the Sediment Impact Assessment Model
(SIAM) in HEC-RAS. In Proceedings of the joint federal interagency conferences 2006 (3rd Federal
Interagency Hydrologic Modeling Conference and 8th Federal Interagency Sedimentation
Conference), Reno, NV, April 2-6.
Mooney, D.M. 2006. Sediment impact assessment methods. Dissertation, Department of Civil
Engineering, Colorado State University, Fort Collins, CO, In progress.
Watson, C.C. & Eom, M. 2003. Rehabilitation design of Judy’s Branch. Final Report to the U.S. Army
Corps of Engineers Ecosystem Restoration Project, St. Louis District, St. Louis, MO, and U.S. Army
Corps of Engineers, Engineer Research and Development Center, Vicksburg, MS, November, 117 p.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
32
Accounting for Sediment in Rivers
FRMRC Research Report UR9
4.
Sediment Transport Computations with HECRAS
Stanford A. Gibson1
1.
4.1
Research Hydraulic Engineer, U.S. Army Corps of Engineers, Institute for Water Resources,
Hydrologic Engineering Center, 609 Second Street, Davis, CA 95616-4687,
[email protected]
INTRODUCTION
Sediment routing and mobile boundary simulations have been employed for years in support of
traditional applications including dredging prediction, reservoir sedimentation and engineered channel
stability and, more recently, channel restoration and dam removal. In 1976, Tony Tomas and U. S.
Army Corps’s of Engineers’ Hydrologic Engineering Center (HEC) developed the mobile boundary
model HEC-6 which has been an industry standard for these applications ever since. This DOS
program remains widely utilised while other popular HEC hydrologic and hydraulic models (HEC-1,
HEC-2, and UNET) have been eclipsed by more powerful and user friendly products (e.g. HEC- HMS
and HEC-RAS). Recently, however, many of the core capabilities of HEC-6 have been incorporated
into Hydrologic Engineering Center’s River Analysis System (HEC-RAS), leveraging the robust,
existing, hydrodynamic capabilities in RAS and providing helpful user interfaces for one dimensional
sediment transport modelling.
The initial version of this sediment model will utilise a wide range of hydraulic capabilities existing in
HEC-RAS to compute a series of steady flow profiles in order to develop hydrodynamic parameters
for sediment transport. Hydraulic computations are ‘explicitly coupled’ with transport, erosion,
deposition, bed mixing and cross section change computations. The result is a continuous
simulation of cross section change as sedimentation processes respond to the inflowing watersediment hydrograph. Sediment computations in HEC-RAS utilise one dimensional, cross-section
averaged, hydraulic properties from RAS’ hydraulic engine to compute sediment transport rates and
update the channel geometry based on sediment continuity calculations. The initial objective of model
development involves replicating HEC-6 functionality within the HEC-RAS framework. Once these
capabilities are available new features and model advancements will be implemented.
4.2
METHODOLOGIES
4.2.1 Hydrodynamics
Flow specification for sediment transport computations currently follows the ‘quasi-steady’ flow
approach of HEC-6. An event or period of record is approximated by computing a series of steady
flow profiles (Figure 4.1). HEC-RAS uses each steady flow profiles to develop transport parameters
for every each cross section. Durations are assigned to each profile to define the temporal extent of the
associated hydrodynamics and to rout sediment movement. Usually, however, bathymetry updates are
required more frequently than the flow increment duration, so a computational time step must be
specified as well. The geometry file is updated and new steady flow hydrodynamics are computed at
the beginning of each computational time step.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
33
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 4.1
Schematic of quasi-steady flow division
4.2.2 Transport Calculations
Six different transport functions are currently available in HEC-RAS: Ackers and White (1973),
Englund-Hansen (1967), Laursen (1958), Myer-Peter-Muller (1948), Toffaleti (1968), and Yang
(1972). The Wilcock equation (Wilcock and Crowe, 2003), a surface based, sand and gravel equation
which includes hiding functions and computes gravel transport as a function of sand transport is also
under development.
HEC-RAS facilitates computations of graded beds by dividing the sediment gradation curve into up to
20 discrete, editable, size classes. RAS calculates transport capacity by computing independent
transport potential for each size class, as if it were the only material in the channel, and then weighting
the transport potential computed for each bin by the relative abundance of that gain class in the active
layer such that:
n
Tc = ∑ β j T j
j =1
where: Tc=Total transport capacity; n=number of grain size classes; βj=% of active layer composed of
material in grain size class ‘j’; Tj=Transport potential computed for 100% of the material grain class
‘j’.
The sediment continuity (Exner) equation can then be solved over the control volume associated with
each cross section, computing from upstream to downstream. The Exner equation is:
(1 − λ p ) B
∂η
∂q
=−
∂t
∂x
where η is bed elevation; B is the width of the control volume; q is volumetric transport rate, λp is bed
porosity.
Bed elevation rises and falls in response to sediment supply deficit or surplus in the control volume,
the positive or negative difference between capacity and supply. HEC-RAS solves the Exner equation
separately for each grain size adding material to or removing material from the active layer. At the end
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
34
Accounting for Sediment in Rivers
FRMRC Research Report UR9
of each computational time step, aggregation or degradation is translated into a uniform bed change
over the entire wetted perimeter of the cross section. RAS updates cross-sectional station-elevation
information and performs new hydraulic computations before computing transport capacity for the
next sediment routing iteration.
4.2.3 Physical Constraints to Erosion and Deposition
Physical constraints can preclude the entire sediment surplus or deficit computed by the Exner
equation from translating directly into aggregation or degradation in a given time step. RAS currently
follows HEC-6 in applying temporal erosion and deposition modifiers as well as sorting and
armouring routines to augment the simple continuity computations.
4.2.4 Temporal Modifiers
Solving the Exner equation translates 100% of computed surplus or deficit immediately into
deposition or erosion. This does not reflect actual physical processes, however, as both deposition and
erosion are temporal phenomena. Therefore, HEC-RAS applies time dependent modifiers to the
surplus or deficit calculated at each cross section. Deposition efficiency is calculated by grain size
based on the computed fall velocity and the expected centre of mass of the material in the water
column based roughly on Toffeletti’s concentration relationships (Vanoni, 1975). The deposition rate
as the ratio of sediment surplus that translates into deposition in a given time step is defined as:
Deposition Rate =
Vs (i ) ⋅ ∆t
De (i )
where Vs(i) is the settling velocity for particle size I; De(i) is the effective depth for sediment size i
(e.g. the midpoint of the depth zone in which transport is expected for the grain class); and, ∆t is the
duration of the computational time step (USACE (1993) and Thomas (1994)).
A similar relationship was implemented to limit erosion temporally. RAS uses a ‘characteristic length’
approach from HEC-6 which includes the assumption that erosion takes a distance of approximately
30 times the depth to fully develop. Therefore, in cases where capacity exceeds supply, the
capacity/supply discrepancy is multiplied by an entrainment coefficient (Ce) to compute actual erosion
allowable in a time step. The entrainment coefficient is:
C e = 1.368 − e
−L
30⋅ D
where L is the length of the control volume; and D is the effective depth (USACE (1993), Thomas
(1994)).
As the length of the control volume goes to thirty times the depth, the coefficient approaches unity and
erosion approaches the full amount of computed deficit.
4.2.5 Sorting and Armouring
The other major process considered in the computation of continuity is potential supply limitation as a
result of bed mixing processes. Currently HEC-RAS employs Exner 5, a ‘three layer’ algorithm from
HEC-6 to compute bed sorting mechanisms. Exner 5 divides the active layer into two sub-layers,
simulating bed coarsening by removing fines initially from a thin cover layer. During each time step,
the composition of this cover layer is evaluated and if, according to a rough empirical relationship, the
bed is partially or fully armoured, the amount of material available to satisfy excess capacity can be
limited.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
35
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 4.2
Schematic of 3-layers used in Exner 5 sorting and armouring method
A simplified, ‘two layer’, active bed approach (with the Toro-Escobar et al. (1996) exchange
increment method) has also been added to facilitate the surficial character of the Wilcock method.
Currently this is designed for gravel transport with the active layer thickness set to the d90 of the layer
but HEC is currently evaluation other methods for determining active layer thickness which may have
broader application.
4.3
MODEL TESTING AND VERIFICATION
4.3.1 Comparison to Meyer-Peter and Muller Data
Several tests have been conducted to evaluate these methodologies in HEC-RAS. First, HEC worked
with Tony Thomas to simulate one of the original Meyer-Peter and Muller (MPM) experiments (1948)
with HEC-RAS and HEC-6T. Since the MPM transport function was derived from these experiments,
they can be simulated with an expectation of reproducing the end result without the standard problems
of transport function uncertainty. In the MPM experiment, a constant flow was run through a flat bed
flume with a constant rate of gravel feed (grain size diameter 28.5mm) until it reached a stable,
equilibrium slope of about 0.0081. This slope is plotted in Figure 4.3 with the equilibrium bed profiles
computed by HEC-6T and HEC-RAS. There was very good agreement between the physical data and
both numerical models.
4.3.2 Comparison to HEC-6
There are several settings between HEC-6 and HEC-RAS that can produce divergent results (e.g. fall
velocity method, hydraulic radius vs. hydraulic depth, and friction slope methods). In general, if these
settings are harmonised, HEC-RAS does a reasonably good job replicating HEC-6 (e.g. Figure 4.3).
However, sometimes small hydrodynamic differences can result in divergent sediment results. In the
example depicted in Figure 4.4, Yang (1972) was applied to a trapezoidal channel with a single grain
size material. Small differences in how HEC-6 and HEC-RAS compute water surface profiles resulted
in a minor difference in calculated transport capacity (~0.56%). However, since supply was only
slightly larger than capacity, this small capacity discrepancy translated into a 6% difference in total
aggradation. Therefore the bed profiles diverge, despite very small calculated differences. It is of note
that the computational differences implemented in RAS, though minor, are considered improvements
by HEC and, therefore, the sediment responses to these are considered to be improvements as well.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
36
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Comparison of MPM Flume Data, and Simulations with HEC RAS and HEC 6T
B ed E levatio n (ft)
1.2
Equilibrium Slope from MPM flume (d=28.5 mm) - 24hrs
1.0
Simulated Equilibrium Slope for HEC 6T (d=28.5 mm)- 24hrs
Simulated Equilibrium Slope for HEC RAS (d=28.5 mm)- 24hrs
0.8
0.6
0.4
0.2
0.0
0
20
40
60
80
100
120
140
Flume Station (ft)
Figure 4.3
Meyer-Peter and Muller flume data with HEC-6 and HEC-RAS simulations
Elevation (ft)
0.6
0.4
0.2
0
-0.2
-0.4
Orriginal Bed Profile
HEC RAS: t=1 days
HEC RAS: t=6 days
HEC6: t=1 day
HEC6: t=6 days
-0.6
0
100
200
300
400
500
600
700
800
900
1000
Station (ft)
Figure 4.4
4.4
Single grain trapezoidal channel with supply slightly exceeding capacity.simulated
with HEC-RAS and HEC-6
EXAMPLE APPLICATION
In order to demonstrate data requirements and interface features an example application follows. The
site is a flood damage reduction project in the United States that has a history of aggressive deposition
which reduces the level of protection offered by the project. Data and results are provided for
demonstration purposes only.
A good sediment transport model begins with a good hydraulics model. The channel geometry must be
specified by a series of cross sections and the steady flow hydraulics should be carefully analysed for a
range of flows sufficient to encompass those expected in the sediment model. While channel
bathymetry and, thus, hydraulics will change over the course of the simulation it is advantageous to
evaluate ineffective flow areas, bridge hydraulics and other hydrodynamic phenomena of the system
before adding complexity with sediment simulations. Guidelines on setting up an HEC-RAS geometry
file and steady flow application can be found in the Applications Manual that accompanies the
program when it is installed.
Once satisfied with the hydraulics, sediment boundary conditions must be specified. These include
definition of the sediment control volumes, bed gradations, and inflowing sediment load at the
upstream and lateral boundaries. A sediment control volume is defined for each cross section by
defining lateral movable bed limits and an erodible bottom (Figure 4.5). The example application
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
37
Accounting for Sediment in Rivers
FRMRC Research Report UR9
includes a mix of concrete and natural channel bed so some cross sections have specified maximum
elevations (that correspond to the concrete and which the model will not erode below) while others
have significant associated erodible depths.
This screen also provides several drop-down boxes which can be used to select transport method,
sorting approach and fall velocity algorithms. The same editor allows the user to select from a list of
bed gradations pre specified with the screen pictured in Figure 4.6. If bed gradation is known only at
specific points (e.g. sample locations) channel composition between these points can either be defined
by interpolation or be associating gradation templates with multiple cross sections. Interpolation is
appropriate if smooth transitions are expected. However, if bed gradation is typified by extended
reaches of similar properties separated by abrupt transitions associating known data with similar cross
sections may be preferable. Both approaches are implemented for different portions of the example
application in Figure 4.5.
Figure 4.5
Sediment data and methods editor
Figure 4.6
Bed gradation template
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
38
Accounting for Sediment in Rivers
FRMRC Research Report UR9
The final sediment boundary condition required is the load entering the model either at the upstream
end, laterally due to tributaries or point sources, or by a measured or simulated distributed overland
wash off load. Loads can be associated with flows using a flow-load rating curve or can be entered as a
time series. Additionally, in the absence of load information an Equilibrium Load option is available
as a rough estimate which will compute an inflowing load at equilibrium with the hydrodynamics and
bed composition of the upstream cross section for each flow. In the example a single upstream load is
specified in the form of a multi-point flow-load rating curve (Figure 4.7). Each flow is associated with
a load and a fractional breakdown of the load by grain size.
Figure 4.7
Inflowing sediment boundary condition
Next flow boundary conditions must be specified including downstream stage, system temperature as
well upstream or lateral flow series (Figure 4.8). A hydrograph is broken down into discrete intervals
that can be reasonably represented by a period of steady flow. Larger and rapidly changing flows will
correspond to smaller durations to capture flow changes as depicted in Figure 4.5. Additionally,
smaller computational increments are used for larger flows since the rate of bed chance (and thus the
need to update the geometry) is largest for these flows. For longer flow records a tool is available to
automatically associate computational increments with flow records based on the magnitude of the
flow. Specifying computational increments that are too large is the most common cause of model
instability.
Finally, a plan must be constructed in the Sediment Analysis dialog. A sediment plan file (*.pxx)
consists of a geometry file (*.gxx), a quasi-steady flow file (*.qxx) and a sediment file (*.zxx). The
sediment analysis window can be used to organise different flow, geometry or sediment scenarios into
different plans and store results separately. After computational and output parameters are specified
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
39
Accounting for Sediment in Rivers
FRMRC Research Report UR9
under the Options menu and a simulation time window is set the Compute button on this dialog
launches the program.
Figure 4.8
Quasi-steady flow editor
Figure 4.9
Sediment analysis window
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
40
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Results become available for each time step as soon as they are completed so the user does not have to
wait until the computations finish to view results. Spatial and time series plots for a wide range of
variables can be retrieved from the View menu on the main HEC-RAS dialog. Initial and final bed
profiles are depicted in Figure 4.10. Spatial data can also be stored as an animation. A time series plot
of bed elevation at station 4075 is shown in Figure 4.11. Output represents the results of four years of
synthetic hydrology developed from the frequency information available from this system.
Figure 4.10 Initial and final thalweg profile
Figure 4.11 Time series plot of bed elevation at station 4075
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
41
Accounting for Sediment in Rivers
FRMRC Research Report UR9
4.5
RANGE OF APPROPRIATE APPLICATION
When considering the application of HEC-RAS to any engineering analysis, whether or not the mobile
bed capabilities are employed, it is imperative to recognise the range of applicability of a one
dimensional model. Issues of dimensionality are exacerbated in sediment analysis as transport
processes are laterally heterogeneous. Application of a one dimensional model to a river sedimentation
study invokes lateral averaging of bed dynamic processes and erosion or deposition is applied by
volume averaging over the entire wetted control volume. HEC-RAS also only accounts for vertical bed
movement, so applications involving lateral migration or bed failure may be poorly characterised in
this framework.
Additionally, while quasi-steady flow approximations are often appropriate, careful thought should be
given to the implications of approximating a hydrograph with a series of backwater profiles. This may
be an inappropriate approximation for steep or ‘flashy’ systems.
It is also important to stress that mobile bed modelling results are extremely sensitive to the transport
equation selected and results for different equations (even a subset deemed ‘applicable’ to a given
application) can differ by an order of magnitude. Therefore, calibration and sensitivity is particularly
imperative in developing a mobile bed model.
Finally, since HEC-RAS currently routes sediment by solution of the continuity equation rather than
an advective or advective-diffusive (ADE) scheme that would tie transport to water velocity, careful
consideration should be given to cross section spacing and sediment movement to assure that sediment
transport does not significantly exceed water velocity. One of the implications of the dimensionality
and non-ADE solutions is that HEC-RAS is most appropriately applied to sediments in the sand and
gravel range. While rough algorithms are employed to approximate cohesive transport, erosion and
deposition, care should be employed when attempting to apply RAS to cohesive applications,
particularly in fundamentally multi-dimensional applications (e.g. estuaries).
Despite the numerous caveats, many decades of meaningful and instructive one-dimensional, mobile
bed modelling by experienced engineers has demonstrated that the approach employed by HEC-RAS
be applied successfully to a range of engineering problems, particularly in non-cohesive, riverine
environments. However, the user must be cognizant of the limitation listed above, and apply the model
within this framework.
4.6
AVAILABILITY AND RELEASE SCHEDULE
Mobile bed capabilities will be available in HEC-RAS Version 4.0. The model is currently undergoing
alpha testing by a small group of U.S. Corps of Engineers users and minor fixes are being made. A
beta test version which will also include SIAM and a temperature model should be publicly available
shortly. HEC-RAS is public domain software and is internationally available for download on the
HEC website: http://www.hec.usace.army.mil/
4.7
CONCLUSION
HEC-RAS now has basic sediment transport capabilities. RAS utilises quasi-steady hydrodynamics
and one of several transport equations to solve the sediment continuity equation. Sediment surpluses
and deficits are modified with temporal and physical constraints and translated into bed aggradation
and degradation. After each computational time step the RAS geometry file is updated based on bed
elevation changes for the hydrodynamics and sediment potential computations to use during the next
time step. The model has generally performed well in testing against HEC-6 and flume data, but can
differ slightly from HEC-6 in certain conditions due to minor differences in hydraulics. RAS includes
a convenient user interface to specify the necessary data for a sediment analysis and a wide range of
available outputs for analyzing a simulation.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
42
Accounting for Sediment in Rivers
FRMRC Research Report UR9
4.8
REFERENCES
Ackers, P., and White, W. R. (1973) ‘Sediment Transport: New Approach and Analysis,’ Journal of
Hydraulics Division, American Society of Civil Engineers, Vol. 99, No. HY11, pp. 2040-2060.
Armanini, A. (1992) ‘Variations of Bed and Sediment Load Mean Diameters Due to Erosion and
Deposition Processes,’ Dynamics of Gravel Bed Rivers, Edited by P. Billi, R. G. Hey, C. R. Thorne,
and P, Tacconi, pgs. 351-359.
Englund, F. and Hansen, E. (1967) A Monograph on Sediment Transport in Alluvial Streams Teknisk
Vorlang, Copenhagen.
Laursen, E. M. (1958) ‘Total Sediment Load of Streams,’ Journal of Hydraulics Division, American
Society of Civil Engineers, Vol. 84, No. HY 1, pp. 1530:1- 1530:36.
Little and Mayer (1972) The Role of Sediment Gradation on Channel Armoring. Publication ERC0672, Georgia Institute of Technology: School of Engineering, 105p.
Meyer-Peter, B. and Muller, T. (1948) ‘Formulas for Bed Load Transport,’ Report on Second Meeting
of International Association for Hydraulics Research, Stockholm Sweden, pp. 39-64.
Toffaleti, F. B. (1968) Technical Report No. 5. A Procedure for Computation of Total River Sand
Discharge and Detailed Distribution, Bed to Surface, Committee on Channel Stabilization, U.S. Army
Corps of Engineers.
Thomas, W. A. (1994) Sedimentation in Stream Networks, HEC-6T User’s Manual, Mobile Boundary
Hydraulics Software, Inc., Clinton, MS.
Toro-Escobar, C. M., Parker, G., Paola, C. (1996) ‘Transfer Function for the Deposition of Poorly
Sorted Gravel in Response to Streambed Aggradation,’ Journal of Hydraulic Research, 34:1, 35-53.
US Army Corps of Engineers (1993) Scour and Deposition in Rivers and Reservoirs: HEC-6 User’s
Manual, Hydrologic Engineering center, Davis, CA.
V. A. Vanoni (1975) Sedimentation Engineering, ASCE Manuals and Reports on Engineering
Practice-No. 54, 745 p.
Wilcock, P. R. and Crowe, J. C. (2003) ‘Surfaced-based Transport Model for Mixed-Size Sediment’,
Journal of Hydraulic Engineering, Vol. 129, No. 2, pgs 120-128.
Yang, C. T. (1972) ‘Unit Stream Power and Sediment Transport,’ Journal of Hydraulics Division,
American Society of Civil Engineers, Vol. 98, No. HY10, pp. 1805-1826.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
43
Accounting for Sediment in Rivers
FRMRC Research Report UR9
5.
River Energy Audit Scheme (REAS)
Nicholas P. Wallerstein1 and Philip Soar2
1.
2.
5.1
Post Doctoral Research Associate. University of Nottingham, Nottingham.
[email protected]
Senior Geomorphologist. Jeremy Benn and Associates, Atherstone, and Industrial
Research Fellow, University of Nottingham. [email protected]
BACKGROUND
Until recently, sediment dynamics were rarely taken into account when comparing options for flood
defence schemes. However, imbalances in sediment supply and transport capacity may lead to
progressive reduction in the standard of protection provided by a scheme. Potential problems include,
for example, channel bed degradation which may undermine flood defence assets reducing their
condition and increasing the risk of failure under load, while bed aggradation can significantly reduce
conveyance and make overtopping more likely. In addition, flood defence infrastructure may impact
on the sediment dynamics in a reach with, for example, enlargement of a channel resulting in a
tendency for sediment deposition. Sediment dynamics in turn impact on river morpho-dynamics and
the diversity and quality of river and floodplain habitats, which are increasingly recognised as key
factors in achieving good ecological status, as required under the European Water Framework
Directive (WFD) (The European Parliament and the Council of the European Union 2000).
There is, therefore, a need for the planners and designers of flood defence schemes to predict and
account for sediment dynamics and their impacts when appraising options for flood risk management.
In England and Wales, the Environment Agency is tasked with coordinating flood risk management at
the system scale through the development of Catchment Flood Management Plans (CFMPs)
(Environment Agency 2004). Their statement of intent is that measurable modelling outcomes should
be risk based and reach national targets for flood control, but also targets for channel habitat diversity
as stressed in the Water Framework Directive.
At present in the UK, routine assessments of sediment-related problems within a wider, catchment
context are performed using the qualitative Fluvial Audit (Environment Agency 2004, Sear et al.
2003). In this approach, a field and documentary investigation is used to divide the fluvial system into
geomorphic reaches designated as: source (degradational); transfer (equilibrium); or receptor
(aggradational) reaches. The approach involves detailed field reconnaissance of the watercourse
(Thorne 1998), performed by an experienced fluvial geomorphologist. While the Fluvial Audit has
proven very useful in river conservation and restoration projects, in its present form it does not support
quantification of the sediment dynamics required to interface effectively with the engineering
components of strategic flood studies and CFMPs. Certainly, Fluvial Audits provide useful insights
into catchment sediment dynamics, but the approach requires extended inputs by experienced
geomorphologists and is of limited utility for strategic planning due to its somewhat subjective
technical approach and qualitative outcome.
At the other end of the scale sediment dynamics are encapsulated within routing models such as HEC6 (HEC 1976) and iSIS (iSIS 2006) which solve the St Venent Equations for gradually varied unsteady
flow coupled with a sediment transport equation and the sediment continuity equation.
It is instructive to briefly describe the equations of motion and sediment continuity here for reference
purposes. Flood routing is defined as the process of tracing the movement of a flood wave, using the
equation for unsteady, gradually varied flow for which the governing equations are: Conservation of
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
44
Accounting for Sediment in Rivers
FRMRC Research Report UR9
mass and conservation of momentum. These equations are known as the St Venent equations. The
continuity equation is defined by:
∂Q ∂A
=
=0
∂x ∂t
(1)
where Q = discharge; A = cross-sectional area; x = longstream direction; and t = time increment.
The momentum equation is derived from the fact that F = M /dt (F = force, M = mass). We therefore
get:
So − Sf =
∂ (d) 1 ∂  Q 2  1 ∂Q
+

+
∂x gA ∂x  dx  gA ∂t
(2)
where So = channel bedslope; Sf = channel friction slope; and d = flow depth.
The first term on the left-hand side = gravitational force, and the second term = shear force. On the
right hand side the first term represents the pressure gradient, the second the convective acceleration,
the third the local acceleration. Equation (2) is valid for irregular cross-sections and it states that the
friction slope is equal to the channel bottom slope minus the sum of the pressure gradient and
convective and local acceleration gradients.
The sediment continuity equation in its simplified form is defined as (Richards 1982):
∂z 1 ∂q s
=
+i
∂t γ s ∂x
(3)
where z = bed elevation (m); t = time interval (s); γs = bulk unit weight of sediment (Nm-3); qs = unit
sediment load expressed as dry weight (tonnes); x = long stream interval; and i = lateral sediment input
(tonnes).
Other forms of the Exner equation have variously been defined as (Simons & Li 1982):
∂PZ
Q s ∂CA
+
+ (1 − λ)
= qs
∂x
∂t
∂
(4)
where Qs = total sediment transport rate in volume per unit time; the concentration C = Qs / Q, Q =
channel discharge; λ = sediment porosity; P = wetted perimeter; Z = sediment depth (eroded /
aggraded); and qs = lateral sediment inflow or outflow.
Also see (Julien 1996):
TEii ∂q txi
∂z i
=
∂t (1 − λ) ∂x
(5)
where zi = depth change due to grain size i; qtxi = unit sediment transport in the downstream direction
for a given size fraction; and TEi = Trapping efficiency factor:
TEi i =
Coi − Ci
= 1 − e − Xω i /hV
Coi
(6)
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
45
Accounting for Sediment in Rivers
FRMRC Research Report UR9
where Coi = upstream sediment concentration over reach length x; at the location x = 0 for size fraction
i; Ci = sediment concentration in reach x for size fraction i; x = reach length; ωi = settling velocity for
size fraction; h = flow depth; and V = mean flow velocity.
To solve the St Venent Equations solutions are made for time steps specified by an inflow, or series of
inflow hydrographs and are the solved 1) by the method of characteristics, which has largely been
replace by; 2) finite-difference methods. The above approach for solution of bed aggradation and
degradation for the passage of a hydrograph(s) is employed by commonly used models such as iSIS
(iSIS 2006) and HEC-6 (US Army Corps of Engineers). A simpler approach for solving longstream
channel bed level adjustment is to solve the equations for gradually varied flow for a reach, for a given
discharge which generates given channel boundary hydraulic parameters that enable the solution of an
appropriate sediment transport equation. This can then be used to solve the sediment continuity
equation to adjust the channel boundary. The procedure is then repeated for the next time step solving
gradually varied flow for the new boundary. The gradually varied flow equation is defined by:
dd S 0 − S e
=
dx 1 − Fr 2
(7)
where Se = energy slope; and Fr is the Froude number. Se can be defined by (using Manning’s
equation):
 QnP 2/3 
Se =  5/3 
 A

2
(8)
where n = Manning’s roughness coefficient. The Froude Number is defined as:
2
v
q2
Fr =
= 3
gd gd
2
(9)
where v = mean flow velocity; and q = unit discharge.
The calculation procedure for this simplified type of approach is shown in Figure 5.1.
Constructing and running these models can be highly data and time intensive and the considerable
input of resources, coupled with long run times, limits their application to studies of reach-scale
scour/fill rather than issues related to catchment scale sediment transfers and imbalances. Also, as such
models require a considerable level of numerical approximation uncertainty in the results can be very
difficult to track, with errors propagating though the simulation due to incomplete representation of
physical processes and numerical problems associated with convergence, consistency and stability
(Versteeg & Malalasekera 1995). Finally, sediment models rely on sediment transport equations that
characteristically predict sediment load to within +/- 50% about 60% of the time (Yang 1996). This
makes selecting an appropriate sediment transport equation crucial to producing a successful
simulation and there are no universally accepted rules concerning which equation is suitable for a
particular river environment.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
46
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.1
Flow diagram for solution of 1-D channel boundary adjustment using the
gradually varied flow equation time-step approach
The intention of the research presented here is therefore to develop a quantitative method capable of
validating and building upon the insights gained from qualitative assessments of sediment dynamics,
such as the Fluvial Audit but which does not have the pitfalls of more complex numerical schemes. It
was recognised that the model should be a simple, yet robust, tool which fills the gap between
empirical, qualitative and complex, sediment transport modelling approaches. Further objectives were
that the model must be generically applicable, amenable to use by river scientists and engineers with a
modest amount of specialised training, applicable within the resources available for typical strategic
catchment studies and capable of producing robust results in situations with very limited data
availability – albeit accepting a greater degree of uncertainty. The last and perhaps most important
objective is that the output must be easily understood by those responsible for river management and
planning at the catchment scale. To this end, in the latter half of the FRMRC project a working version
of the model will be tested by typical end users in the Environment Agency.
The approach adopted here is aimed at a broad representation of a catchment scale potential for
channel adjustment, it does not purport to represent local conditions in detail. The development of a
numerical model at the whole catchment scale would be very time, labour, and resource intensive,
commodities which the EA and its contractors simply do not have free rein to use. Thus in the
approach adopted here numerical modelling schemes are replaced by an alternative reliable indicator
of the river’s potential to perform geomorphological work through expending energy which thus
enables the transport of sediment. By invoking this approach the authors are not proposing that simpler
or more complex models generate a worse representation of a river system, simply that the approach
outlined below is robust and generates a good first approximation of the potential for
geomorphological adjustment.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
47
Accounting for Sediment in Rivers
FRMRC Research Report UR9
5.2
THEORY
5.2.1 The River Energy Audit Scheme
The method adopted, termed the River Energy Audit Scheme (REAS), uses predicts sediment Sources
(Scour), Pathways (Transfers) and Sinks (Deposition) (Figure 5.2) over a period of years by estimating
the difference in time-integrated specific stream power between consecutive reaches to indicate
potential continuity or imbalance in the sediment transfer system.
Figure 5.2
Sediment transfer through the fluvial system
The theoretical justification for this approach stems from the concept of specific stream power
proposed by Bagnold (1966), in which he defined specific stream power as a measure of the flow’s
ability to perform work on its boundary. The rate of doing work through sediment transport is defined
by Bagnold in terms of the specific stream power multiplied by an efficiency factor. Bagnold’s
approach has been further developed by several authors to predict sediment movement and
geomorphological adjustments in rivers (for example; Langbein 1964, Yang 1971, Chang 1979, Graf
1983, Lawler 1992, Magilligan 1992, Molnar & Ramirez 1998, Knighton 1999). The approach
developed does not attempt to predict sediment transport or route sediment (or energy) through an
event hydrograph. Instead, it calculates the balance or imbalance between the specific stream power
available in a reach in a year (that is the annual stream energy in KJyr-1) with that in the next reach
downstream.
5.2.2 Data Requirements
With respect to data requirements, there are only five input variables required to run REAS for each
study reach in the river system (see Fig 3):
1.
2.
3.
4.
5.
Representative bed material particle size(s);
Representative flow duration curve;
Representative channel cross-section;
Representative bed slope; and
Representative roughness values for the channel and floodplain.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
48
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.3
Schematic of variables that must be obtained from the river basin to run REAS
The manner in which these variables are employed is discussed in the next section.
5.2.3 Specific Stream Power
It must be stressed at this point that REAS is not a sediment routing model, but instead generates
indicative, quantitative trends in river energy, and therefore the ability for the flow to perform work
through sediment transport, during an average year in which runoff is similar to that associated with
the long term flow duration curve. The specific power concept was first applied by Rubey (1933) and
Knapp (1938), and later by Velikanov (1955). It was again suggested by Bagnold (1956) and Bagnold
(1966). Bagnold (1966) defined available stream power as:
•
Rate of doing work = available power – unutilised power
Or
•
Rate of doing work = available power x efficiency
Stream power per unit bed area (ω) is expressed in Wm-2 and is termed Specific Stream Power. It is
defined by:
ω=
γQSe
= τ oV
W
(10)
where γ = bulk unit weight of water (Nm-3); Q = discharge (m3s-1); Se = energy slope (mm-1); W =
geomorphologically active channel top width (m) for the discharge in question; τo = bed shear stress
(Nm-2); and V = mean flow velocity (ms-1).
Bagnold (1966) used this measure of energy expenditure to produce a rational equation of bedload and
suspended load sediment transport derived from first principles. The relation he generated is given by:
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
49
Accounting for Sediment in Rivers
FRMRC Research Report UR9
u
 e
i = ω  b + 0.01 
V
 tanα
(11)
where ω= specific power without the gravity term = ρw QSe /w or ρwdVSe, ρw = density of water
(approx. 1000 kgm-3); d = flow depth, eb = bedload efficiency factor; u = depth averaged flow velocity;
V = sediment fall velocity; tanα = ratio between normal and tangential force on a grain = which for
most grains where the angle of repose is 33 degrees = 0.63; and:
i=
ρs − ρ w
j
ρs
(12)
where i = sediment transport load in units of submerged weight (Nms-1), j = dry mass = 0.62 for sand
density of 2650; and ρs = density of sediment (kgm-3).
We therefore have an equation which has four unknown factors: eb, tanα, u and V. The derivation of
these factors by rational and experimental procedures can be found in Bagnold (1966). Bagnold’s
(1966) sediment transport equation is given here for completeness but actual sediment transport loads
are not calculated in REAS for the reasons outlined in the Background.
In the model adopted here Se in Equation (10) can usually be approximated by the bed slope (So) for
channels with moderate to low slopes. Note that in Bagnold’s (1966) version of Equation (10) he used
water density (ρw: kgm-3) in place of bulk unit weight, γ. The bulk unit weight of water is used in most
derivations of specific power. This has often been a source of error in calculations of sediment load
using Bagnold’s transport functions (Bagnold 1977, 1980) i.e. as Bagnold did not use g (where g =
gravitational constant: approx. 9.81 ms-2) and because γ = ρw × g results can be out by a factor of 10
(Ferguson 2005). The geomorphologically active channel width is defined as the area of channel bed
and banks beneath the flow top width for a given discharge which is within the bankfull channel crosssection (Richards 1982), i.e. it does not extend onto the left hand and right hand floodplain if flow
depth becomes higher than the bankfull elevation.
To calculate an excess specific stream power (ωe), which is that actually available to perform work on
the channel boundary by moving sediment, it is necessary to subtract the critical specific stream power
for mobilising the bed (ωc) from the available specific stream power:
ωe = ω − ωc
(13)
Bagnold (1980) defined critical specific stream power as:
 (12d ) 
1.5
ω c = 290D m log 

 Dm 
(14)
where Dm = mean grain size in the size distribution representing the bed of the channel (m); and d =
flow depth (m).
Ferguson (2005) provides a full derivation of the terms, but essentially, the term 290Dm1.5 is derived
from the Shields equation and the log term is derived from the ‘law-of-the-wall’ velocity profile
equation (Richards 1982).
5.2.4 Critical Power Based Upon a Grainsize Distribution
The approach is further developed here by replacing an estimate of the critical specific stream power
based on a single mean grain size with a formulation that better represents the range of grain sizes
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
50
Accounting for Sediment in Rivers
FRMRC Research Report UR9
present on the bed. This is achieved by multiplying the critical power for the median grainsize in each
size class (1024mm, 724mm, etc) by its frequency of occurrence, and then summing all values. Thus,
for a given depth, excess power ωe’ is equal to:
n
ωe ' = ∑ [Pi (ω − ωci )]
(15)
i =1
where Pi = proportion (from 0 to 1) of that size class in the whole sample; and ωci = critical specific
stream power for a given grainsize class.
ωci is calculated using Equation (14) in which Dm is replaced by Di (m), the median grainsize in each
class. The justification for the formulation and use of Equation (15) is that use of a single particle size
such as the Dm might not be a suitable parameter alone given a highly variable bed sediment gradation.
Thus Equation (15) offers a more integrated approach which is preferable where sediment grainsize
distribution data is available. It is recognised that for some applications of this model a full grain size
sample may be too time-consuming and costly to collect and a single representative grain size for each
reach may be all that is available. An option for using either a single grain size value or a full
distribution has therefore been made available in the model.
Note that the recommendation is to use the size distribution of the surface sediment layer on the bed as
research has shown that sediment transport equations best fit empirical data when this layer is used
(Parker 1990, Wilcock & Crowe 2003).
5.2.5 The REAS Annual Energy Budget
The annual energy budget for a reach must take into account the full range of discharges acting on the
channel over a period of years. This is done by integrating ωe’ with a representative flow duration
curve, adopting the procedure set out by Biedenharn et al. (2001). This is performed by taking the
entire flow record available for the reach in question (15 minute or mean daily values (note that this
may produce slightly different results)), ranking the values from lowest to highest and producing a
cumulative probability curve (Figure 5.4).
Figure 5.4
Example of cumulative probability plot for 15 minute flow data
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
51
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.5
Discharge–frequency plot
The discharges are then split into ranges (usually 25–35 classes) and their frequency determined
(Figure 5.5).
If gauged data are not available for the reach in question, or the discharge record is short, surrogate
methods have to be used. These include:
1.
2.
3.
Scaling discharge from a gauge elsewhere in catchment. This can be achieved through simply
scaling by the ratio of drainage areas;
Scaling discharge from a donor catchment with similar hydrological and morphological
characteristics; or
Calculation using hydrological modelling packages; for example HEC-1 (HEC 1998).
Integrating for the expected range of discharges yields:
n
 n

ω e = ∑ Fj ∑ [Pi (ω − ω ci )]
j=1 
i =1

(16)
where Fi = frequency (from 0 to 1) of each discharge class.
Note that discharge, width and depth in Equation (16) are adjusted to the values appropriate for each
discharge class.
The difference in excess annual specific stream power between consecutive reaches gives the balance
for a reach (∆ωe):
∆ω e = ω e(r = n) − ω e(r =n +1)
(17)
where the subscripts r = reach; and n = reach number from 1 to nt (where nt = total number of reaches).
Reaches are numbered sequentially from upstream to downstream. Equation (17) is then multiplied by
the gravitational constant because Bagnold’s formulation of Equations (10) and (14) do not
incorporate g. This value is then multiplied by the number of seconds in a year to give a measure of
annual excess energy in KJyr-1 at a single point in a cross section.
Equation (17) is comparable in essence to a sediment budget calculation except that the sediment
transport rate has been replaced by a term representing excess specific stream power. Thus, if ∆ωe is
greater than zero then the reach has the potential to aggrade (it is likely to be a sediment sink), while if
∆ωe is less than zero then the reach has the potential to degrade (it is likely to be a sediment source).
Figure 5.6 shows a flow diagram of the basic REAS computations.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
52
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.6
Schematic diagram of REAS calculation procedure
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
53
Accounting for Sediment in Rivers
FRMRC Research Report UR9
5.3
DISCUSSION
5.3.1 Solution for Flow Width and Depth
Because REAS uses a range of discharges a flow depth and channel top width must be defined for
each flow. Manning’s flow resistance equation is used for this purpose (Chow 1959):
5
1 A 3 12
Q=
Se
n P 23
(18)
where n = Manning’s roughness coefficient; A = channel cross-sectional area (m2); and P = channel
wetted perimeter (m).
Values for Manning’s n can be obtained from, for example, Barnes (1967) or Chow (1959).
It is acknowledged that in a simple model such as this obtaining the energy gradient may prove to be
impossible due to the wide spacing of cross sections which would prevent generation of a gradually
varied flow profile. This source of uncertainty will be small in channels with low bedslopes due to the
fact that the energy slope will approach that of the bedslope. In higher gradient streams or channels
with very variable bedslopes the source of error caused by the approximation of energy slope through
the use of bedslope is acknowledged as potentially a significant source of uncertainty.
To solve Manning’s equation a cross-sectional representation of the channel for the reach in question
must be defined. This requires a survey of what is considered to be the geomorphologically active
channel and, if out of bank flows are thought to be significant, the floodplain as well. It is recognised
that field survey data will probably be the most costly input to obtain when using the REAS and the
longstream spacing of cross-sections may have to be significant. This will inevitably reduce the
resolution of the model in terms of the number of reaches that can be defined. To solve Equation (18)
flow depth must be adjusted iteratively to solve for A and P until the value of the right hand side of
Equation (18) converges to the value of the left hand side. Iterative techniques available include the
Newton-Raphson and secant methods. The normal flow depth for a given discharge is solved by
apportioning flow between the main geomorphologically active channel and floodplain areas (left and
right) (Figure 5.7). If the lateral topography of the floodplain is unknown, the model has the option for
the user to assign an estimate of the lateral floodplain slope. This ensures that large flows are not
contained within vertical walls either side of the main channel which would otherwise overestimate
flow depth. Such a problem can occur in 1-D flow models such as HEC-RAS if the floodplain is not
extended either side of the channel. A value of n must be prescribed for the channel and floodplains,
along with the limits of the geomorphologically active channel (Figure 5.7). The term active channel is
used here to describe that part of the channel within which the excess stream power acts to perform
significant geomorphic work through transporting sediment.
Figure 5.7
Definition of parameters required to solve Manning’s equation. n = Manning’s
roughness coefficient; and K = conveyance
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
54
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Of significant importance in the calculation of in-channel power is how to account for channel area
and wetted perimeter at those discharges that involve overbank flow. It is assumed here that a ‘glass
wall’ extends vertically above the bank markers, thus defining the maximum width and discharge used
in Equation (10). Flow depth within the bank markers above the bankfull depth is used but,
importantly wetted perimeter is not extended vertically as there is no actual physical boundary. This
last assumption grossly simplifies floodplain–channel flow interaction as there will be significant
momentum exchange at this location. Methods are available to account for momentum loss (for
example; Thorne & Soar 2000, Thorne & Soar 2001), but they are considered to be beyond the scope
of this model as it is truly only one dimensional and thus designed to be simple in its representation of
flow processes. Indeed many far more complex 1-D flow models such as HEC-RAS and HEC-6 do not
incorporate any momentum exchange calculations.
To solve Equation (18) for the full channel section it is normal to sum conveyances (K) for each
section (Chow 1959):
n
Q = ∑ K i Se
1
(19)
2
i =1
where Ki = conveyance for panel i. Ki is defined as:
Ki = AiR i
2
3
1
ni
(20)
where Ai = cross-sectional area for panel i; Ri = hydraulic radius for panel i; and ni = Manning’s
roughness coefficient for panel i.
5.3.2 Representing Stream Power in a Variable Cross-section
Flow depth for a given discharge must be defined to solve Equation (14). Two approaches were
identified:
1.
2.
Based on hydraulic depth, and;
Based on panels.
For case 1 Equation (11) becomes:
 (12d h ) 
1.5
ω c = 290D i log 

 Di 
(21)
where dh = hydraulic depth (m) = A / W.
This seems to be a reasonable and simple assumption on initial inspection but it can be a misleading
interpretation in variable cross-sections.
Imagine a geomorphologically active channel cross-section (within the in-channel markers) that has an
irregular cross-section (Figure 5.8). At low flows, when W = W1 an increase in discharge will cause an
increase in power (Eq. 10) because the rise in W is proportionally small compared to the rise in Q and
critical power. Thus, excess power will increase with stage. Now, however consider the case where
flow depth increases to W2 due to an increase in discharge. It is possible at this stage that the rise in W
is disproportionately greater than the numerator in Equation (10), causing excess specific stream
power for a given discharge to actually fall or even go to zero, even though the shear stress applied to
the bed below W1 has increased. In reality this cannot be the case as in the inner low flow channel
(W1) particles will still be mobile, and to a greater degree because Q has increased. Thus, the
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
55
Accounting for Sediment in Rivers
FRMRC Research Report UR9
mathematical definition in Equation (21) does not tally with the process which would actually occur in
nature.
Figure 5.8
Width variation over a berm within the geomorphologically active channel
However it is considered that a fall in specific stream power for an increase in stage is tenable because
excess power is a width averaged variable and must therefore take into account the full channel width,
but powers falling to zero cannot be justified. An alternative approach has therefore been investigated
and is the preferred option in REAS.
A better approximation of depth can be gained by sub-dividing the active channel cross-section into
panels of equal width and with average panel depth dz. Panels can be generated by interpolating areas
between each node generated from the survey of a reach cross-section. For this case the excess specific
power equation becomes:
ρ Q S
ωe = w m e −
W
290D i
1.5


 12  N m
 N m log   + ∑ logd z 
 D i  z =1


N
(22)
where Qm = discharge for those panels where the bed is mobile; Nm = number of panels where the bed
is mobile; and N = total number of panels.
The reason for not showing a division of the total power in Equation (22) by N is that the summed
division of ρwQmSe in the total stream power term by the width for each individual panel and then
division by the number of panels N, is exactly same as the use the whole width. That is:
 γQ mzSe 
dW  γQ mSe
z =1
=
N
W
N
∑ 
(23)
where dW = individual panel width.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
56
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Under this approach, for a given discharge and water surface elevation, the depth for each panel can be
determined. Next, a check is made to see if bed particles would be mobile (each size class in the
distribution is considered as discussed in Section 2.4) for a given flow depth. It would appear in the
first instance that this could be achieved by simply calculating total stream power for a panel and
subtracting its corresponding critical power using Equations (10) and (14). However this is actually
not possible because of a phenomenon first recognised by Garbrecht & Brown (1991). If discharge is
calculated based upon a single panel for an entire cross-section using Manning’s equation a given
depth and width will be determined. However if a cross-section is panelised and a conveyance is
calculated for each, the implicit assumption is made that the channel elements have frictionless walls
and no lateral momentum exchange, thus resulting in a lateral velocity gradient. This approach violates
the postulate of a single mean velocity across the section for the application of Manning’s formula.
The result is that a panelised approach will overestimate conveyance, and therefore discharge and
consequently flow depth for each panel, as compared with the calculation made using an integrated
cross-section:
n
K ≠ ∑ Ki
(24)
i =1
This inequality is shown for a range of variables in Figure 5.9, showing that, as the number of panels
by which the channel is divided into increases, so the error increases between the sum of panelised
discharges and a conveyance based upon a single section. The figure also shows the relationship
between channel width to depth ratio and percentage error. This error may result in conveyance overestimation in flow models where the user has the option of dividing the cross-section of a channel into
a significant number of panels each with their own roughness coefficient.
An alternative approach has therefore had to be adopted. This has been achieved by simply using a
form of the Shields entrainment function (Yang 1996):
θ=
τo
γ(G s − 1)D i
(25)
where θ = Shields parameter; τo = bed shear stress = γRSo (Nm-2); R = hydraulic radius (m); and, Gs =
specific gravity of the sediment.
With some rearrangement of Equation (25) we get:
R=
θ(G s − 1)D i
So
(26)
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
57
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.9
Percentage error between discharge calculation for a cross-section based upon a
single area and a panelised cross section
To distinguish between mobile and immobile panels, Equation (26) is solved using a critical Shields
value. Here, a value of 0.04 is used as this was the value used by Bagnold (1980) in his derivation of
Equation (14). When Equation (26) is solved using a critical Shields value, R is equal to Rc, the critical
hydraulic radius. Rc for a particular panel can then be compared with the actual hydraulic radius
computed for the panel. If Rc > R the panel will be immobile at the given flow and given grain size.
When all panels are checked in this manner, Nm can be determined. The critical specific power is then
calculated using Equation (19). Note that N is the total number of panels including those where Rc > R.
Once this has been done the total discharge for the mobile panels (Qm) can be calculated using
Manning’s equation. Now we can calculate total specific power based upon the mobile panels using
the first term on the right hand side of Equation (22).
Using this scheme excess specific stream power can fall if the width suddenly increases, but it should
never fall to zero. An example of the functional relationship between discharge and excess power for
the Hawkcombe data-set is shown in Figure 5.10.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
58
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.10 The relationship between channel cross-sectional geometry and specific power–
discharge curves
It can be seen in this diagram that the cross-section for Reach 4 has a distinct berm and this has an
impact on the discharge–power curve where, as the flow initially passes over the berm, the sudden
increase in width results in a fall in power. As discharge for this larger width increases again so the
upward trend between discharge and power starts again. The trend for Reach 2 is monotonic because
the cross-sectional form is reasonably uniform. This concept is outlined in diagrammatic form in
Figure 5.11.
Figure 5.11 Diagramatic representation of the potential impact of an in-channel berm upon
excess stream power
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
59
Accounting for Sediment in Rivers
FRMRC Research Report UR9
5.4
RELATIONSHIP BETWEEN TOTAL SPECIFIC POWER AND EXCESS
POWER
When Equation (22) is applied to a single or whole range of grain sizes present in the cross-section a
relationship is formed between total specific stream power and excess specific stream power. A
schematic of the generalised form of this relationship is shown in Figure 5.12.
Total
available
power ω
ωe
Flow depth
ωc
threshold
ωc
Grainsize
Figure 5.12 Schematic plot of grainsize against power showing the relationship between
grainsize and critical power and the variation in critical power with increasing
flow depth
A worked example is shown in Figure 5.13. Here the relationship between total and critical power is
shown for a nominal discharge of 100 m3s-1. Depth for this discharge has been calculated using
Manning’s equation based upon an idealised, rectangular channel with a slope of 0.0001, a Manning’s
n coefficient of 0.035 and a channel width of 5m. The total power which is calculated using the above
parameters is indicated by the dashed horizontal line in Figure 5.13. An example of critical power (4
Wm-2) and excess power (2.2 Wm-2) values for a sediment size of 0.02m is shown in the figure.
The relationship that is developed shows that, as grain size increases, so critical power also increases
thus reducing excess power (ωe). The functional relationship between the critical power and grain size
is formed by a power-based curve. It can be seen that for a limiting value of 0.035m, excess power
falls to zero. The power form of the critical curve is explained by the fact that the term, 290D1.5 in
Equation (14) is a power function. What is also apparent from Equation (14) is that, for the log term,
as depth rises so critical depth will also increase relative to grain size. This means that for a greater
flow depth the critical curve on Figure 5.13 should shift upward from, and parallel to, the critical
specific power curve for a lower depth,, as indicated in Figure 5.12. However manipulation of depth
values over a wide range shows that the position of the critical power curve varies only fractionally.
This can be explained because the log term in Equation (14) is based upon the log10 of d/D and
therefore even for a great range of depths the difference in the log term will be almost minimal (this
can be illustrated by, for example, the fact the for d = 2, log10 d = 1 and, for d = 100, log10 d = 2). This
has implications for the academic development of critical power equations as, for example, hiding
functions such as that suggested by Ferguson (2005) are, in part, based upon the log term.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
60
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.13 The relationship between critical and excess power over a range of grainsizes
5.4.1 Accounting for Lateral Sediment Inputs
Lateral adjustment of a channel through expansion or contraction is a further expression of power
expenditure and can be conceptualised through the adjustment of width in the specific power equation,
i.e. if width increases specific power falls and vice versa. Therefore an adjustment in power should be
accounted for. This presents a difficulty in conceptualisation as to how to apportion power gain or loss
because we are computing power balances based upon channel excess power which does not mesh
with the gravitational force which is required to overcome the cohesive forces which hold a bank
together. There has been a significant amount of research into channel width adjustment in bends and
downstream through the channel network, with associated predictive equations (Hooke 1979, Nanson
& Hickin 1986, MacDonald 1991, Lawler et al. 1999) but these are expressions of lateral adjustment
where channel width ultimately does not change, the channel merely erodes on one bank and accretes
on the other at the same rate. What is required in this model is prediction of changes in power due to
widening, say perhaps caused by a steepening in the channel gradient following channelisation or a
narrowing of the channel due, for example, to the construction of confining levees. It has therefore
been conceived that, given the risk based setting for the model, width adjustment could be placed in
the context of a nominal scale risk factor between 0 and 0.1 for channel erosion (widening) and 0 and
2 for channel accretion (narrowing). This factor is then multiplied by the total excess specific power.
This scale bank adjustment factor is shown in Table 5.1. The scale is, of course, to some extent
arbitrary and based upon the scale given assumes that the channel width can only double or halve.
Table 5.1 Width adjustment factors used in REAS
Widening Descriptor
None
Minor
Intermediate
Moderate
Major
Extreme
% yr-1 × ωe
0
0.9
0.7
0.5
0.3
0.1
Narrowing Descriptor
None
Minor
Intermediate
Moderate
Major
Extreme
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
% yr-1 × ωe
0
1.1
1.3
1.5
1.7
1.9
08/08/06
61
Accounting for Sediment in Rivers
FRMRC Research Report UR9
One issue which may seem to prevent the invocation of such an approach is the fact that bank erosion
may be due to mass failure and not particle entrainment and is thus not directly due to changes in
power. This problem can be overcome if we consider the fact that even if a bank fails through mass
failure the removal of the failed material, and therefore ultimately the possibility of further bank
adjustment depends upon fluvial entrainment of the failed material. This phenomenon is known as
Basal Endpoint Control (Carson & Kirkby 1972).
5.4.2 Channel Junctions and Structures
Channel Junctions
Because REAS is envisaged to be utilised over long channel networks the issue of accounting for
power differentials at channel confluences must be addressed. This can simply be handled by
subtracting the specific power of the reach downstream of the confluence from the summed powers of
the threads upstream of the confluence. There is one proviso here which relates to reach limits,
described above, in that in order to achieve the balance across junctions reach limits for the trunk
stream and the downstream end of the tributaries must all converge at the junction itself. This is shown
diagrammatically in Figure 5.14.
Figure 5.14 Schematic of model system for accounting for channel junctions.
Structures
If long reaches of channel through a catchment are to be analysed there are likely to be structures
present such as bridges, culverts, dams, and weirs. It is recommended that structures not be
incorporated into the averaged geometry of a reach if the reaches are significantly longer than the
influence of the structure (for example scour at a bridge contraction). However if it is felt that a
structure has a significant impact on channel dynamics over a long distance and its influence will be
felt upstream or downstream or both there may be a case for making a structure a reach in its own
right. One simple way to account for a structure is given here:
If there is a weir or gravel trap which is empty and the range of influence of the backwater from the
weir lip is set as a reach, the automatic forcing of the specific power in this reach to zero will mean
that any specific energy balance will remain the same as that for the incoming reach, i.e. if the balance
is positive upstream, the balance in the trap will be the same indicating the potential for the trap to
store sediment. In the same manner, if specific power in the weir reach is set to zero the balance
downstream will be negative, i.e. the reach will likely degrade, which seems like a reasonable
assumption.
The option to force a reach specific power to be zero is enabled in the model. This option is shown
diagrammatically in Figure 5.15.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
62
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.15 Schematic diagram showing potential to override energy balance at structures such
as dams and weirs enabling all energy to be ‘dumped’.
5.4.3 Reach Limits
A major issue with employing REAS is that of where to locate reach limits. There are three
alternatives as we see them:
1.
2.
3.
Limits based upon a qualitative Fluvial Audit;
Limit based upon the River Habitat Survey (RHS); and
Limits Predicted by analysis of the output trend data.
It was originally envisaged that a Fluvial Audit would be used to locate reach limits and then in turn be
used to calibrate the output from REAS. In this manner REAS is simply a scenario modelling tool in
that once the model has been calibrated to current conditions it can be used to assess future scenarios
that involve changes in the input parameters. The model in this form is in essence operated like any
other, more complex scheme, i.e. the model is run using available data and best practice but is then
adjusted through calibration to fit a set of measured conditions.
An alternative is to use a fixed reach length scheme such as the one currently employed by the EA, for
its River Habitat Survey (EA 2003), where the river in question is divided into 500m reaches for
analysis. This approach is well used and understood be EA personnel and may be the only option in
many locations as Fluvial Audits are limited.
This is not what one would hope for as a scientist, we would wish our models be predictive without the
need for what is essentially circular reference.
However, there is an alternative more scientifically ‘pleasing’ option which may be available in this
instance which would make REAS predictive in its location of reach boundaries. To achieve this case
REAS is run for every cross-section in the surveyed system but not split into pre-defined reaches for
which composite cross-sections are made. Trends are then looked for in the data which suggest self
similarity, i.e. reaches that have similar characteristics in terms of net aggradation, degradation or
relative equilibrium. The statistical method for achieving this is known as Global Boundary Hunting
(Davis 2002) which uses a moving window to divide a sequence into a number of zones which are
internally homogeneous as possible and as distinct as possible from adjacent zones. This approach will
be explored in the future.
5.5
SOURCES OF UNCERTAINTY
The following statement made in the RPA 9 Risk & Uncertainty Tools and Implementation Report
(Pappenberger et al. 2005) should be borne in mind when assessing this model, ‘…an increase in
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
63
Accounting for Sediment in Rivers
FRMRC Research Report UR9
complexity will introduce more model data and parameter requirements, where both data and
parameters may be uncertain’. It was exactly with this observation in mind that REAS was developed.
The REAS model structure is that of simple equations which combine to form an expression (see Eq.
22 & Figure 5.5) that calculates specific stream power differentials, measured in KJyr-1. However
uncertainties are still present.
Primary sources of uncertainty associated with this model are:
•
•
•
•
•
•
•
•
•
5.6
Uncertainty as to the QUALITY of data collected. For example one contractor may survey
channel geometry in a slightly different form to another unless a stringent protocol has been laid
out. Error may therefore creep in, for example, in headwater streams where one survey crew
might consider a large boulder to be a component of the channel whereas another might not. This
observation is made in the RPA 9 Risk and Uncertainty Tools and Implementation Report,
‘…topography is often seen as the factor with the least uncertainty … (but) … small error in
floodplain topography can have significant effects on flood inundation results’;
Overestimation in calculated flow depth using Manning’s equation simply as a result of
increasing the number of cross-section panels used (see Section 3.2);
Potential for misrepresentation of flow duration characteristics by using too few / too many bins
in the flow frequency curve (see Section 2.5);
Significant variation in in-channel specific stream power as a result of changing the position of
channel bank markers;
Significant variation in flow depth with small variations in Manning’s n (note that the model flow
conditions may well be developed in a non-calibrated form where data is limited);
Length of flow record from which the flow duration curve is developed may change the form of
the curve. It is however important to note that, ‘…quality of data is sometimes more important
than the length of the record’ (see RPA 9 Risk and Uncertainty Tools and Implementation
Report, 2005). This will certainly be the case if the flow data for a particular REAS reach comes
directly from a gauge close by as compared to that from a surrogate catchment;
The coefficients in Equation (14) are approximations as is the use of a Shields parameter of 0.04
(Eq. 24).
The use of a bedslope if an energy slope cannot be obtained, while making the model much more
data friendly, may significantly alter flow depth calculated using Manning’s equation (Eq. 18).
This impact is most likely to be significant in steep channel environments; and
The use of a single grainsize to represent bed roughness instead of a whole distribution, while
also allowing for limitations in data availability, will alter the computed excess specific power.
LIMITATIONS
The following are the main limitation of REAS:
•
•
•
Although REAS is designed to be used for long reaches of river network, and ideally whole
catchments, the input of sediment from the catchment surface itself cannot be accounted for. This
is because the energy required to move sediment into the channel is external to the river itself and
is therefore not accounted for.
REAS is not a sediment routing model. Given the scope of the model outlined in the Background
this in itself is not a problem but there is an issue of misunderstanding by the end user and
academic community of what the model actually does calculate.
Data Availability. A potential major limitation is that of the availability of sediment grainsize
distributions for each cross-section. With this in mind REAS has the option of selection of a
single representative grainsize which may be determined through no more analysis than a visual
assessment. A second data limitation is that of the availability of a full flow duration record. This
can, in most instance, be overcome by using data from a surrogate catchment, although the level
of uncertainty will be increased.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
64
Accounting for Sediment in Rivers
FRMRC Research Report UR9
5.7
CASE STUDIES
5.7.1 Hawkcombe Stream
Site
A pilot data-set comprising the variables required to run the model has been collected from the
Hawkcombe Stream, in Somerset, England (Figure 5.16). Figure 5.17 shows a 1:50000 map of the
catchment. This is a steep, upland catchment with a relief of 436m over a length of about 8km. Base
flows are generally less than 1 m3s-1, with maximum discharges of up to 10 m3s-1. The channel has a
boulder, cobble and gravel bed and is generally armoured. HEC-RAS modelling of the Hawkcombe
suggests that at high discharges the flow is supercritical in many places, indicating that the stream has
a high energy environment. In 2002, a Fluvial Audit of the Hawkcombe was performed (Thorne et al.
2002) and the main stream was divided into 17 reaches (Figure 5.18) characterised as sediment sources
(degradational), transfers (dynamic equilibrium), exchanges (in essence dynamic equilibrium), and
sinks (aggradational).
Figure 5.16 General Location map of the Hawkcombe stream in Somerset, England
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
65
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.17 1:50000 OS map of Hawkcombe Stream catchment
Figure 5.18 Hawkcombe Stream geomorphological reaches
In 2004, the lower 4.5km of the Hawkcombe was surveyed in detail. One hundred and twenty seven
cross-sections were surveyed, located at major breaks of slope and positions of significant channel
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
66
Accounting for Sediment in Rivers
FRMRC Research Report UR9
geomorphological change and fifty four bed material samples were taken. Bed material was analysed
using size-by-number for particles between 1025mm and 8mm, sieving and size-by-weight between
8mm and 0.063mm, and using a Coulter Counter below 0.063mm.
The Hawkcombe has several hydraulic structures along it, including bridges, mill weirs, gravel traps,
bed sills and bank protection works. It flows through an 83m long culvert, beneath the village of
Porlock. This culvert has required maintenance de-silting in the past as sediment deposition has
occurred, reducing its capacity to convey floods. In 2005, the invert of the culvert was re-profiled to
reduce sediment trapping in a scheme designed by Royal Haskoning Consulting Engineers (Figure
5.19).
Figure 5.19 Re-profiled culvert on the Hawkcombe Stream under the village of Porlock
Data Analysis
Initial trials of REAS were conducted using the Hawkcombe dataset using the current 5.5 version of
the MS Excel based model. Discharge for each reach was scaled from a donor catchment, the River
Horner by a direct scaling based upon the drainage basin area for each reach. Reach drainage basin
area were derived by very careful manual interpretation and digitising, as automated upstream area
calculation was deemed too unreliable. The Area of Flowstrip method was used judging from the
terrain data, to join the stream at that reach, from either side, as seen by the coloured polygon areas in
Figure 5.20) (screenshot from Arcview GIS). Thirty five discharge classes were developed from the
flow frequency relationship.
Other variables input to the model were: a lateral floodplain slope of 0.0001, a floodplain Manning’s n
estimate of 0.6 and a channel Manning’s n estimate of 0.035. These values were selected purely based
upon observational judgement. The full bed material sediment grainsize distributions were used. The
option in REAS to use a width adjustment factor was not used as information on bank characteristics
was sparse. While there is the option in the current version of REAS to import iSIS and RAS geometry
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
67
Accounting for Sediment in Rivers
FRMRC Research Report UR9
files this was not done as it was felt that a fair trial of the model using an in-built interpolated
weighting process would provide a good test bed for representing the cross-sections for each reach
(see Figure 5.28). All structure based cross-sections were stripped from the each reach as they were
deemed to cause only local effects except the culvert under Porlock where conditions are known (the
culvert aggrades due to poor invert geometry). Trial runs were performed with both the full grainsize
distribution and with a calculated reach based D50 to enable comparison.
Figure 5.20 Digitised drainage basin areas for each of the 17 reaches
As the REAS output will be used in the context of risk it was felt that differences in excess annual
energy in KJyr-1 should be expressed in dimensionless form. The output from the Hawkcombe is
therefore presented in Figure 5.21 using a scale between +1 and -1, where values between 0 and +1
indicate increasing potential for sediment storage, values between 0 and -1 indicate increasing
potential for sediment scour, and values close to 0 indicate channel stability. The Fluvial Audit
prediction bars in Figure 5.21 are all set to +1 / -1 as the direction of predicted change in terms of
aggradation or degradation is on an observational scale thus if Figure 5.18 is examined there are only 1
degradational (source) and 4 aggradational (sink) reaches identified. The rest are considered to be in
some form of equilibrium. For the case of equilibrium reaches no bar is shown on the graph. Note that
on Figure 5.21 the reach numbering is in reverse order to that on Figure 5.18, with Reach 17 located at
the downstream end of the surveyed channel. This was adopted in REAS as it is the scheme preferred
by the EA. All references to REAS reach numbers in Figure 5.21 are in brackets.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
68
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Dimenstionless Specific
Stream Power
1.3
1.0
0.8
0.5
0.3
0.0
-0.3
-0.5
1
2
3
4
5
6
7
8
9
10 11
12 13 14 15 16 17
-0.8
-1.0
-1.3
Reach (17 = downstream)
Predicted: No Distribution
Predicted: Distribution
Fluvial Audit
Figure 5.21 Excess energy budgets for the Hawkcombe stream
The agreement between the observed geomorphological adjustment and that predicted by REAS is
good for the four cases of observed potential aggradation, these being reaches 1 (17), 5 (13), 7 (11)
and 4 (14). Of particular significance in the correspondence here is the fact that reach 5 (13) which
represents the culvert under Porlock predicts aggradation and this is a known active process in this
location. Although other reaches are indicated to be ‘transfer’ or ‘exchange’ in Figure 5.18, the reach
descriptions suggest some vertical adjustment. Reach 1 (17) is observed to be aggradational where the
channel is avulsing onto a marsh area at the seaward end of the channel. Reach 2 (16) is described as
being generally stable but with some small knickpoints and REAS predicts some minor degradation.
Reach 3 (15) is anomalous as it is evidently degrading but REAS predicts mild aggradation and again
Reach 4 (14) is observed to be stable but minor degradation is predicted by REAS. Reach 5 (13) is, as
mentioned above, in very strong agreement. Figure 5.22 shows a HEC RAS model of the culvert that
comprises Reach 5 (13). Notice the broken back and pressurised flow at the WS 30 profile.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
69
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.22 HEC RAS long profile of Culvert under Porlock Village.
Reach 6 (12) is observed to be in equilibrium and REAS suggests this also. Reach 7 (11) has a series
of gravel traps in place to prevent excess sediment from entering the culvert under Porlock. This reach
therefore has an aggradational trend and this is strongly predicted by REAS. Reach 8 (10) is
anomalous as REAS predicts strong degradation whereas observationally channel appear to be stable.
Again Reach 9 (9) is observed to be stable but REAS predicts a strong aggradational trend. Reach 10
(8) is given as equilibrium on Figure 5.13 but the description describes a degradational trend due to the
failure of a small dam in 2002. REAS predicts this trend. Figure 5.23 shows a series of photographs
from Reach 10 (8) in which the collapsed weir can be seen.
Reach 11 (7) shows evidence of mild degradation in the field and this is picked up by REAS. Reach 12
(6) is described as being a bedrock channel and therefore stable and REAS suggests this also. Figure
5.24 presents a series of photographs of this reach which clearly show the stable nature of the bed and
banks.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
70
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.23 Photographs of Reach 10 (8) showing degradational features.
Figure 5.24 Reach 12 (6) showing the stable nature of this channel reach.
Reach 13 (5) is described as degrading slightly and again REAS suggests this trend. Reach 14 (4)
contains a series of gravel traps and therefore has an aggradational trend. The direction of this trend is
predicted by REAS but only weakly so. Reach 15 (3) is observed to be relatively stable and again
REAS suggests this. Finally Reach 16 (2) is observed to be in equilibrium but REAS predicts a mild
aggradational trend. Reach 17 (1) cannot predict a balance in REAS because there is no power value
upstream of this reach.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
71
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Overall in terms of aggradational, degradational or equilibrium trends REAS and the Fluvial Audit
agree to varying degrees in 11 of the 16 reaches calculated in REAS, a strong correlation. The reason
for incorrect prediction in five reaches warrants more in-depth exploration of the model properties
especially as reaches 8 (10) and 9 (9) have such strong inverse trends.
It must be borne in mind however that a direct comparison of the Fluvial Audit and REAS predictions
may show misalignment because the Fluvial Audit is an observation of CURRENT conditions whereas
REAS gives a prediction of potential ONGOING and FUTURE trends. This is what makes REAS a
powerful tool which can be used to assess the potential geomorphological development of reaches.
A comparison between the use in REAS of a single D50 and a full grainsize distribution (indicated on
Figure 5.21 by ‘Predicted: No Distribution’ and ‘Predicted: Distribution’ shows that there is an
average of 20% difference between the use of a full distribution and a single D50 value. It is interesting
to note that in six out of the sixteen reaches the single D50 value produces higher relative balances than
the full grainsize distribution while in the remainder of cases the opposite is true and in one case,
Reach 6 (12), the direction of trend is different between the D50 value and the distribution. This points
to interesting dynamics between the calculation of excess power based solely on Equation (14) as
compared to using Equation (15).
5.7.2 River Wharfe
Site
The River Wharfe between Hubberholme and Starbotton was chosen for this study due to the
extensive cross-sectional data available from the University of Durham, Department of Geography.
Figure 5.25 below shows the location of Wharfedale in England, the extent of the catchment from the
source within the Yorkshire Dales National Park to the confluence with the River Ouse near the town
of Ryther and also shows the upper and lower limits of the study reach. A more detailed map of the
study reach is shown in Figure 5.26. The reach has a catchment area of 72 km2, and a reach length of
5.6km. Rainfall is approximately 1720–2000 mm per year. The catchment geology is that of
Limestone and Millstone Grit. The morphology is classic U-shaped valleys which were historically
glaciated. The relief is therefore steep and soils thin. Land-use is predominantly pasture and hay
meadows. Engineering works have been performed recently with the capacity of channel increased by
a series of levees, gravel shoal removal and a gravel trap installed just downstream of Hubberholme.
Figure 5.25 General location map of the Upper Wharfe showing study reach limits
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
72
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.26 Reach Map showing location of 60 cross-sections
Data has been collected by the University of Durham and consists of repeat topographic surveys of 60
cross-sections in April and December between December 2001 and April 2005 with an average
number of points surveyed per cross-section of 106 and average spacing of 0.32 m. Cross-section
locations represented local morphology. Sediment samples were taken at 16 locations and measured
using sieving analysis. Samples were taken for both surface and active layer. Discharge was scaled
from a weir located downstream of the study reach a Flint Mill which measures mean daily discharges.
Data Analysis
Because repeat surveys had been made for each cross-section over a number of years this enabled a
more quantitative assessment of the predictive capability of REAS as compared with the Hawkcombe
Stream analysis in that the calculation of volume differences in cross-sectional area between
December 2001 and April 2005 will give some indication, albeit over a very short period in
geomorphological terms as to whether a reach was aggrading or degrading. This analysis was achieved
in the following manner:
1.
2.
3.
4.
5.
Rectify the cross-section end points measured in Dec. 01 with those measured in Apr. 05;
Fit panels to the cross-section that best represent the channel cross-sectional geometry (about 100
panels seemed to be enough);
Select a discharge for the cross-section and calculate the area in each panel for Dec. 01 and Apr.
05 using the alpha method for Manning’s equation;
Adjust the discharge and recalculate the water surface such that the normal depth was above the
level of the channel top markers; and
Calculate the area of each panel for Dec. 01 and Apr. 05, subtract one from the other and sum the
total for the whole cross-section to give the unit volume difference over that time period.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
73
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.27 shows the results for each of the 60 cross-sections. Note that some sections have balances
of zero. This is due to the fact that these sections were screened out when long-profile plots indicated
suspect bed-levels. Also note that cross-sections are numbered from upstream to downstream as this is
the format preferred by the EA.
Figure 5.27 Volume difference for the 60 cross-sections on the Wharfe reach (note that in this
instance 1 represents the most upstream cross-section. Negative balances indicate
degradation; positive balances indicate aggradation over the 5 year period
To make a clear assessment of the measured volume difference in each reach a scheme was adopted to
apportion the relative importance of each cross-section value relative to its distance between adjacent
cross-sections. The method employed to do this is shown in Figure 5.28.
Figure 5.28 Method use to weight cross-section volume difference values for each 500m reach
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
74
Accounting for Sediment in Rivers
FRMRC Research Report UR9
The results of this weighting process are shown in Figure 5.29 onto which are superimposed the long
profile, reach limits and sediment sample location.
Figure 5.29 Weighted volume difference values for each 500m reach with channel long-profile
In order to assess the modelling capability of REAS the approach taken in this case was to use it as a
PREDICTIVE tool in that reaches were not pre-assessed by qualitative Fluvial Audit, but the river
reach simply divided up into lengths of 500m from the upstream end. Average reach–interpolated
cross-sections and bed slopes were generated using the REAS routine which uses the approach shown
in Figure 5.28. Thirty five flow–frequency classes were generated using drainage basin area scaling
from the Flint Mill gauge located downstream of the study reach. The model was tested using a full
grainsize distribution obtained from the surface layer of the channel bed; however, grainsizes are only
in the range 8–512 mm and this, as discussed below, has a significant impact on model results. REAS
was also tested using a single calculated D50 value. It is worth noting at this point that both the surface
and active layer grainsizes show a downstream fining trend in this reach of the Wharfe.
The output from REAS is shown in Figure 5.30. The comparison of these results with the measured
unit volume changes indicates a significant difference. The weighted volume difference in Figure 5.29
show that, in general, the whole reach is aggrading, in short the flows do not have the competence to
carry the imposed load from upstream of this reach. Output from REAS indicates almost no potential
for aggradation or degradation in Reaches 3 and 6 to 10 (Reach 1 cannot be computed because there is
no upstream power value). This does not negate the results however as essentially the almost-zero
excess power values for the whole reach show, as in Figure 5.21, that the reach as a whole cannot
carry the imposed load, thus comparative balances between Reaches are almost zero.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
75
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Dimensionless Specific
Stream Power
1
0.75
0.5
0.25
0
-0.25
1
2
3
4
5
6
7
8
9
10
-0.5
-0.75
-1
Reach
Predicted: No Distribution
Predicted: Distribution
Figure 5.30 Excess energy budgets for the Hawkcombe Stream
The key anomalies in Figure 5.30 are the balances for Reaches 4 and 5. There is a simple explanation
for these results which stems from the generation of slope input data. This is because the bed profile in
Reach 3 has a bedslope, which, when based upon the vertical elevation difference between the two end
cross-sections, is actually zero. It is evident that a zero slope cannot be used in the calculation of
specific power as this forces power to zero (see Eq. 10) a result that cannot be possible if the river is
actually flowing. This leads to the point that the use of a bedslope rather than an energy gradient in the
calculation of power may produce unrealistic results for certain channel bed slopes. Therefore in the
case of the Wharfe a HEC-RAS model of the total reach was used to provide an energy gradient rather
than a bedslope for each REAS reach. The long-profile output from this model is shown in Figure
5.31. Even for the highest discharge run (the discharges used were the same as developed for the flow–
frequency input data for the REAS model) the energy slope for Reach 3 was exceedingly variable
ranging from, in the upstream direction for consecutive cross-sections, 0.000004, 0.000000, 0.029761
and 0.000001. As a result total excess power in Reach 4 is high, and thus produces the strong negative
trend. This has the knock-on effect of generating a strong positive trend in Reach 5 because; once
again here the slope, and therefore excess power, is very low. The highly variable bed elevation, and
therefore bedslopes in the HEC model of the Wharfe between distances 3000 and 4000m upstream
brings to light a crucial point regarding REAS. This is that reach-averaged conditions need to be used
for a REAS reach and reaches need to be of a significant length to avoid picking up local effects such
as bed elevation variations through pools and riffles and around meander bends which may distort the
prediction of the potential for the reach to adjust.
It is also worth noting from these results that when using the single D50 grainsize value for Reaches 3,
and 6 to 10 the power values (not the balances) calculated for each reach in KWyr-1 are zero, i.e. the
critical powers are greater than total powers, whereas when the full distribution is used the power
values are non-zero, but are still exceedingly small. This may well be due to the fact that because the
range of sediment grainsize data was only in the range 8–512 mm critical powers are high whereas it is
quite likely that if a distribution which included material finer than 8mm were used there would be
greater transport potential, i.e. the river flowing over very coarse material and does not have the
competence to carry this but may well transport material finer the 8mm through the reach. This result
suggests that using a single D50 value in environments where critical powers are high may result in the
model not picking up the fact that there is some transport of finer material. However this may not be a
significant issue as such results indicate that the channel is only likely to move fines which are likely
to not play a part in shaping the channel form.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
76
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 5.31 HEC-RAS model long profile plot of the River Wharfe showing the highly variable
bed elevations at the river distances 3000–4000m
5.8
CONCLUSIONS
5.8.1 Key Points: Answers to questions posed by the Environment Agency
The aim of this research has been to build upon the Fluvial Audit by developing a quantitative tool that
identifies the trends in sediment Sources, Pathways and Sinks (or Sources, Transfers and Receptors in
EA parlance). The model is envisaged to be used by the Environment Agency and its contractors for
flood analysis which is carried out through CFMPs.
In order for the model to be utilised it must show clear advantages in terms of reduced time and cost
and improved decision making. We believe this has been achieved because REAS adds a quantitative
element to CFMP development. It is quick to run and requires a relatively low level of technical
knowledge and is adaptable to data availability. It is therefore CHEAPER to use than numerical
models and more CONSISTENT then visual observation.
Answers to other key questions posed by the Environment Agency are:
•
•
•
Why: REAS has been developed to help with the assessment of flood risk management
conducted by the Environment Agency. It is designed to bridge the gap between the qualitative
Fluvial Audit and more complex numerical models such as iSIS.
Who: It is envisaged that REAS will be used predominantly by the Environment Agency at the
Strategic Scale for development of Catchment Floodplain Management Plans.
Workable levels of detail: REAS is NOT a sediment routing model but instead shows general
trends in the potential for the river to perform geomorphological work on a reach-by-reach basis
integrated over a flow duration curve.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
77
Accounting for Sediment in Rivers
FRMRC Research Report UR9
•
•
•
Application Scale: The model is aimed at the catchment scale with trends shown over relatively
long reaches. Inspection of these trends will enable the targeting of more intense analysis using
numerical models at key sites where flood defence issue need to be assessed.
Output: the output from REAS is given in the forms of power differentials between reaches in
KJyr-1 which indicate whether a reach has the potential to aggrade or degrade.
Input Requirements:
Five variables are required to run REAS:
1. Representative reach cross-section
2. Reach slope
3. Representative reach grainsize distribution or single grainsize
4. Flow duration curve
5. Reach averaged roughness (Manning’s n)
Because it is envisaged that the end user of REAS will predominantly be the Environment Agency and
its sub-contractors it is instructive to look at the potential sources of data that are available for running
REAS, their cost, processing time and licensing agreements, as this will strongly influence model
uptake. Table 5.2 shows examples of data sources which are envisaged that can be sourced for each of
the five input variables.
Table 5.2 Data Type, availability, cost, processing time and licensing agreements
Data Type
Available Datasets
Cost
Flow
Low Flows 2000
Continuous gauge
Pre-existing HEC / iSIS
geometries. 10% of
lowland UK has HEC
survey
Commissioned survey
DEM (1:10000)
Expensive
Low
Moderate
Cross Sections
Expensive
Processing
Time
High
Moderate
Moderate
Moderate –
High
Expensive
Slope
Grainsize
Roughness
•
DEM (1:10000)
Lidar
Pre-existing HEC / iSIS
geometries
Commissioned survey
OS Profile data
GeoRHS (c. 400 sites) /
RHS
New surveys
Existing RAS / iSIS
GeoRHS/RHS (17000
sites, 25% catchment
coverage)
Fluvial Audit
Moderate
Free to EA
Moderate
Licensing Agreement
CEH owned
EA owned
Mostly EA owned
Consultancy
Get mapping plc
Moderate –
High
Moderate
Moderate
Get mapping plc
EA owned
Mostly EA owned
Expensive
Low
Consultancy
Moderate
Moderate
Free to EA
Low
DigiMap or
OS License
EA owned
Very Expensive
Low
Free to EA
High
Low
Very low
Consultancy
EA owned
EA owned
Free to EA
Low
EA Owned
Format: REAS is currently coded in the form of a series of MS Excel 2003 spreadsheets within
one file with VBA macros running the computations. It was decided to use an Excel interface
because this is a program which is licensed to the EA and has been established as a tool that can
be used by its staff. Also the EA has a two year run-in time for executable (.exe) programs which
would thus mean that, if it were not in a VBA format, it could not be used by them for beta
testing within the lifetime of the FRMRC project. Appendix A presents the User Manual for the
current version of REAS (ver. 5.5).
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
78
Accounting for Sediment in Rivers
FRMRC Research Report UR9
5.8.2 General Conclusions
Accounting for sediment dynamics in river channel networks must become a key component of flood
risk management planning. To date the approaches for accounting for sediment in analysis of the
fluvial system are generally based either on complex numerical models or qualitative interpretation of
the results of stream reconnaissance. Both of these approaches have benefits but also considerable
pitfalls in that highly trained modellers and geomorphologists are required to use such tools. Time and
financial constraints are also a significant factor. A need for an approach which lies between these two
techniques has therefore been identified in the UK as a key resource that will be used to facilitate river
flood risk management within the context of meeting statutory requirements for ecological status and
goals for habitat preservation and enhancement. A novel approach, the River Energy Audit Scheme
(REAS), is therefore being developed as part of research performed by the Flood Risk Management
Research Consortium (FRMRC). The approach is based on an audit of excess specific stream power in
which excess reach-averaged specific stream power is used as a surrogate for sediment transport to
give an excess specific stream power budget between consecutive reaches in the channel network.
Excess power available to perform work on the channel boundary is defined as total specific stream
power minus critical specific stream power, using Bagnold’s (1980) formulation of both parameters.
Only five input variables are required for the calculations: a flow frequency curve; channel crosssectional geometry; a bed slope; a roughness coefficient and a representative bed material grainsize or
grain size distribution. The approach given here has is mind the fact that geometry data available to the
UK Environment Agency and its contractors has widely spaced cross-sections. Also a single
representative grainsize can be used to solve Equation (14) if a full distribution is not available.
The approach for generating a flow duration curve has been described along with specific solutions for
calculating flow top width and depth using Manning’s equation. Interesting properties have been found
in the relationship between the variables, showing that excess specific stream power can fall even as
discharge rises. The reason for this apparent inconsistency is explained by the manner in which flow
depth is derived. This leads to the justification of one particular solution (Eq. 22). The Shields
equation is used to predict panel depth because of the phenomenon that calculating discharge using
panels rather than using an integrated cross-section will overestimate conveyance and therefore flow
depth for a given discharge. The functional form of the relationship between total power and critical
power is discussed and indicates the importance of previously unconsidered factors in the derivation of
critical power equations.
It is worth considering that a criticism which might be levelled at this approach needs to be addressed,
that being that it suffers from the same problems as complex numerical models but by not using
sediment transport relationships provides less reliable results. This statement may be counteracted by
the following observation made by Beven (2002): ‘In science where unique local characteristics so
affect local observations as to make the application of small-scale physical theory intractable at the
larger scale we must look to alternative larger scale descriptions and must recognise that there may be
many possible descriptions that provide equally good predictions of the observations.’ We believe
therefore in the light of this statement that REAS is a credible approach and will sit well between the
Fluvial Audit and models such as iSIS.
It is important to understand that a direct comparison between the Fluvial Audit and REAS predictions
may be slightly contradictory because the Fluvial Audit is an observation of CURRENT conditions
whereas REAS gives a prediction of potential ONGOING and FUTURE trends. This makes REAS a
powerful tool which can be used to assess the potential geomorphological development of river
reaches.
Finally we must consider the ‘proof of concept’ outlined in Chapter 1 (see Figure 5.3). The levels of
Proof of Concept with regard to REAS are presented below; the answers to each of the proof are
given. We believe therefore given the case stated below that Proof of Concept has been achieved for
REAS in RPA 8, WP 8.1.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
79
Accounting for Sediment in Rivers
FRMRC Research Report UR9
5.9
REFERENCES
Bagnold, R.A. 1955. Some flume experiments on large grains but little denser than the transporting
fluid, and their implications. Inst. Civil Engineers Proc., Pt 3.
Bagnold, R.A. 1966. An approach to the sediment transport problem from general physics. United
States Geological Survey Professional Paper 4221.
Bagnold, R.A. 1977. Bed load transport by natural rivers. Water Resources Research 13: 303-312.
Bagnold, R.A. 1980. An empirical correlation of bedload transport rates in flumes and natural rivers.
London: Royal Society of London Proceedings A405: 369-473.
Barnes, H.H. JR. 1967. Roughness Characteristics of Natural Channels. U.S. Geological survey
water-supply paper 1849 Washington: United States government printing office. 213 pp.
Beven, K. 2002. Towards an alternative blueprint for physically based digitally simulated hydrological
response modelling systems. Hydrological Processes 16: 1-17.
Biedenharn, D.S., Thorne, C.R., Soar, P.J., Hey, R.D. & Watson, C.C. 2001. Effective discharge
calculation guide. International Journal of Sediment Research 16(4): 445-459.
Carson, M.A. & Kirkby, M.J. 1972. Hillslope form and Process. Cambridge: Cambridge University
Press,. 475 pp.
Chang, H.H. 1979. Minimum stream power and river channel pattern. Journal of Hydrology 41: 303327.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
80
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Chow, V.T. 1959. Open-channel hydraulics. New York: McGraw-Hill.
Davis, J. C. 2002. Statistics and data analysis in geology. (3rd ed). New York, Chichester: John Wiley
& Sons Ltd. 638 pp.
Environment Agency. 2003. River Habitat Survey in Britain and Ireland: Field guidance manual.
River Habitat Survey Manual: 2003 version. Environment Agency. 136 pp.
Environment Agency. 2004. Catchment Flood Management Plans: Volume i – Policy Guidance.
Bristol: Environment Agency.
Environment Agency. 2005. Sustainable flood and coastal management; Final draft technical repro.
Part 1: Handbook.: Bristol: Environment Agency.
Ferguson, R.I. 2005. Estimating critical stream power for bedload transport calculations in gravel-bed
rivers. Geomorphology 70: 33-41.
Garbrecht, J. & Brown, G.O. 1991. Calculation of total conveyance in natural channels. Journal of
Hydraulic Engineering 117(6): 788-798.
Graf, W.L. 1983. Downstream changes in stream power in the Henry Mountains, Utah. Annals of the
Association of American Geographers 73(3): 373-387.
HEC, 1976. HEC-6: Scour and deposition in rivers and reservoirs: Users Manual. California:
Hydrologic Engineering Center. US Army Corps of Engineers. Computer Program 723-G2-L2470.
HEC, 1998. Flood Hydrograph Package: User’s Manual. California: California: U.S. Army Corps of
Engineers Hydraulic Engineering Center. CPD-1A.
Hooke, J.M. 1979. An analysis of the processes of river bank erosion. Journal of Hydrology 42, 39-62.
iSIS, 2006. iSIS Flow / Hydrology. Wallingford: Wallingford Software Ltd. and Halcrow Group Ltd.
Julien, P.Y. 1998. Erosion and Sedimentation. Cambridge: Cambridge University Press. 280 pp.
Knapp, R. T. 1938. Energy balance in stream flows carrying suspended load. Am. Geophys. Union
Trans. P. 501-505.
Knighton, D.A. 1998. Fluvial Forms & Processes: A New Perspective. London: Arnold. 383 pp.
Knighton, D.A. 1999. Downstream variation in stream power. Geomorphology 29: 293-306.
Langbein, W.B. 1964. Geometry of river channels. Journal of Hydraulic Engineering 90: 301-312.
Lawler, D.M. 1992. Process dominance in bank erosion systems. In P.A. Carling & G.E. Petts (eds).
Lowland Floodplain Rivers: Geomorphological Perspectives: 117-143. London: John Wiley & Sons
Ltd.
MacDonald, T.E. 1991. Inventory and analysis of stream meander problems in Minnesota. M. S.
Thesis, Univ. of Minnesota, Minneapolis, MN.
Magilligan, F.J. 1992. Thresholds and spatial variability of flood power during extreme floods.
Geomorphology 5: 373-390.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
81
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Molnar, P. & Ramirez, J.A. 1998. An analysis of energy expenditure in Goodwin Creek. Water
Resources Research 37 (7): 1819-1829.
Nanson, G.C. & Hickin, E.J. 1986. A statistical analysis of bank erosion and channel migration in
western Canada. Geological Society of America Bulletin 97, 497-504.
Pappenberger, F., Harvey, H., Beven, K., Hall, J., Romanowicz, R. & Smith, P. 2005. Risk &
uncertainty tools and implementation. Flood Risk Management Research Consortium Report.
Research Priority Area 9. 111pp.
Parker, G. 1990. Surface-based bedload transport relation for gravel rivers. Journal of Hydraulic
Research, IAHR 119 (11): 417-436.
Richards, K.S. 1982. Rivers: Form and Process in alluvial channels. London: Methuen & Co. Ltd. 361
pp.
Rubey, W.W. 1933. Equilibrium conditions in debris-laden streams. Am. Geophys. Union Trans., 14th
Ann. Meeting. P 497-505.
Sear, D.A., Newson, M.D. & Thorne, C.R. 2003. Guidebook of Applied Fluvial Geomorphology.
Swindon: Defra R&D Technical Report FD1914, Environment Agency R&D Dissemination Centre,
WRc. ISBN 0-85521-053-2..
Simons & Li (1982). Engineering Analysis of Fluvial Systems. Fort Collins, Colorado: Simons, Li
Associates. 80522.
The European Parliament and the Council of the European Union. 2000. Directive 2000/60/EC.
Thorne, C.R. 1998. Stream Reconnaissance Handbook: Geomorphological Investigation and Analysis
of River Channels. Chichester: John Wiley & Sons Ltd, UK.
Thorne, C.R. & Soar, P.J. 2000. Analysis of channels with compound channels for channel restoration
design. Final report submitted to United States Army. London: Research Development and
Standardisation Group – UK. Contract No. N68171-00-M-5506, Project No. W90C2K-8913-EN01.
Thorne, C.R. & Soar, P.J. 2001. Performance of channels with compound channels for channel
restoration design. Final Report submitted to United States Army. London: Research Development and
Standardisation Group-UK. Contract No. N68171-01-M-5483, Project No. W90C2K-9125-EN01.
Thorne, C.R., Skinner, K.S. & Priestnall, G. 2002. Geomorphology of the Hawkcombe Stream: CDROM. Submitted to the Environment Agency, Wessex Region. Nottingham: Nottingham University
Consultants Limited.
Velikanov, M. A. 1955. Dynamics of channel flow – v. 2, Sediments and the channel (3rd ed.): Moscow
State Publishing House for Tech. – Theoretical literature p. 107-120 (In Russian).
Versteeg, H.K. & Malalasekera, W. 1995. An introduction to computational fluid dynamics: The finite
volume method. London: Prentice Hall, an imprint of Pearson Education. 257 pp.
Wilcock, P.R. & Crowe, J.C. 2003. Surface-based transport model for mixed-size sediment. Journal of
Hydraulic Engineering 129(2): 120-128.
Yang, C.T. 1971. Potential energy and stream morphology. Water Resources Research 7(2): 311-322.
Yang, C.T. 1996. Sediment Transport: Theory and Practice. London: McGraw-Hill. 396 pp
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
82
Accounting for Sediment in Rivers
FRMRC Research Report UR9
6.
Use of 1-D Sediment Models
Tony Green1
1.
6.1
Senior Hydraulic Engineer, Jeremy Benn and Associates. Wallingford.
[email protected]
BACKGROUND AND CONCEPTUAL BASIS
6.1.1 Background
The development of 1-D computational models of river flow since the 1960s (e.g. Sogreah 1963)
firmly established the technique for analysis and planning of flood protection works. Since that time,
in the UK a significant proportion of the main rivers have had at least part of their length modelled
using commercial packages such as iSIS, HEC-RAS or Mike11 and the current Environment Agency
flood mapping budget of around £10m per year contributes further to the stock of available models.
This represents a significant investment in survey, calibration and assembly of models and a valuable
resource not only of information but people experienced in use of these models.
The hydraulic parameters needed for a sediment transport calculation are common to those used in the
hydraulic model calculations and thus the existing model packages are suited for adaptation to
incorporate a sediment analysis. The commonly used codes such as iSIS, Mike 11 and SOBEK all
have additional modules to incorporate a sediment flux and morphological adjustment of sections and
other programs primarily originating from the US such as HEC-6, GSTARS Fluvial 12 and now HECRAS have similar capabilities. It can thus be argued that the 1-D approach that builds on existing
information and existing models rather than continually reinventing the wheel ought to be the first
point of call for any analysis that is to use quantitative information on sediment flux.
In practice the amount of sediment modelling in the UK is much smaller than more routine hydraulic
modelling although there is increasingly a need to consider the sediment regime in studies under the
requirements of the Water Framework Directive. There are practical and theoretical difficulties in the
application of 1-D models that are not addressed in the program user manuals (these concentrate
primarily on data input and mechanics of running a model) but are understood by a small corps of
users. As suggested by Hayter (2002) experience shows that it is relatively easy to create a 1-D
sediment model in one of the standard packages but the interpretation of results requires at least as
much knowledge as use of multi-dimensional models.
Recognising the utility of 1-D sediment modelling it is thus proposed that, without being overly
prescriptive for sediment studies that may widely vary in requirements, there is a need to compile
common issues and potential solutions in use of 1-D sediment models and that the FRMC Package 8.2
is a suitable vehicle to achieve this. iSIS is the most widely used hydraulic modelling package in the
UK and is thus the most suitable sediment model to consider although other packages have similar
requirements and the applied research will be applicable including potentially use of the latest HECRAS.
6.1.2 Conceptual Background to 1-D Sediment Models
It is not necessary to go into detail of the equations used in hydraulic and sediment transport
computations as these are well documented in user manuals (for example iSIS 1999). The sediment
simulation is based on calculation of sediment transport rates and an accounting of erosion and
deposition using the concept of layers of sediment with ‘well mixed’ distribution of sediment size. The
basic balance of sediment is simply a mass balance expressed as the Exner equation:
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
83
Accounting for Sediment in Rivers
FRMRC Research Report UR9
(1 − λ ) W
δz ∂G
+
=0
∂t ∂x
(1)
where l = bed porosity; W
= water surface width; z = bed elevation; t = time; G = sediment
transport rate (m3/s); x = distance in flow direction (m).
The layer concept is explained Figure 6.1. Accounting for active and parent layers is important
especially for gravel bed rivers where there is a wide grading of sediment and surface layers may be
significantly coarser than the parent material. Bank material is not specified separately so is assumed
to be the same as bed material.
Figure 6.1
Bed Layer Concept (iSIS 2001)
Updating of sections is carried out during the hydraulic simulation and there are a number of options
in iSIS Sediment as shown in Figure 6.2 for how this is done which may have an impact on the results
obtained.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
84
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 6.2
Bed Updating Options (iSIS 2001). For option 1 the whole section is updated, for
option 2 the wetted section only is updated and, for option 3 deposition/
erosion is distributed across a section according to depth to a user specified
exponent
Sediment transport formulae for gravel, sand and silt fractions may be separately specified by size
fraction. Boundary conditions specific to sediment simulations that must be set are sediment inflows
which may be specified in a number of ways taking account of inflow grading, the bed size gradings
(which can vary along a reach). In the standard version of iSIS the sediment transport equations
available are Engelund Hanson (1967), Ackers and White (1973), Ackers and White (1993) and
Westrick Jurashek (1985) for fine silts. Other standard formulae are available in the research version
of the code (iSIS 2001). The implementation of the sediment transport calculation utilises the sectional
hydraulic properties calculated for the hydraulic computation. This may cause complications where
there are divided channels or floodplain flows and the threshold of movement is a significant
influence.
6.1.3 Available 1-D Models
iSIS Sediment
iSIS Sediment originated in 1994 as part of the iSIS collaborative venture between Halcrow and
HR Wallingford. The earlier code developed at HR Wallingford is described by Bettess and White
(1981). The initial use of the iSIS Sediments model was for simulation of siltation in large irrigation
canals such as found in Pakistan but with the addition of graded sediments and improved accounting
for layering effects, its use and capabilities rapidly become wider. In 2001 further extensions were
developed and testing was carried out during a research project at Herriot Watt and Glasgow
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
85
Accounting for Sediment in Rivers
FRMRC Research Report UR9
University (Shvidchenko et al. 2001). This research version has not totally been incorporated in the
standard code.
To be able to run iSIS Sediment, a licence for the iSIS hydrodynamic model is needed but there is
currently no additional charge for the addition of the Sediment module. The model can be used in a
number of ways including long term simulation of bed evolution or for engineering studies of, for
example, the effectiveness of interventions such as removal of gravel or modification of the channel to
improve sediment conveyance.
Advantages and features of the model include use of the iSIS interface for model preparation,
simulation and display of results, coupled simulation with a fully hydrodynamic model. The sediment
calculations may use a number of sediment transport formulae, graded or single size sediment
simulations and multiple layers. The user may specify different bed composition at any cross section
and a different gradation for the inflow of sediment. Looped, branched and tidal channels may be used
and a range of bed updating and dredging options are available. Limitations include the need to edit
the sediment file in a text format, sediment transport calculations based on composite section
properties, a restricted set of iSIS units can be used excluding for example interpolated sections and
fixed roughness formulation based on Manning’s n.
HEC-6 / HEC-RAS
Although used by UK consultants overseas for simulation of dam impacts, to date the practical use of
the HEC-6 for the UK has been limited due to the lack of a modern interface, support and the need to
specifically set up data files for its use.
Advantages of the system include use of an effective depth and effective width calculation is used that
preserves the area* D2/3 relationship for irregular cross sections and sediment calculation is confined to
the defined channel. The new version of HEC-RAS incorporating sediment transport and section
change is the subject of another paper at this workshop but this offers a new opportunity to use
existing models in the UK particularly in steeper streams where use of iSIS can be problematic or
where HEC-RAS models are already in use. One difference to iSIS is that the water surface simulation
is based on steady backwater analysis rather than a full hydrodynamic simulation.
6.1.4 Other General Purpose 1-D Sediment Codes
The Mike11 and SOBEK 1-D sediment codes do not appear to have been used on UK projects
although there is increasing interest in multi-dimensional modelling using these other parts of these
packages.
Sediment modelling using Mike11 in Bangladesh utilised idealised sections rather than sections as
surveyed. This method of model construction is a technique rarely used in the UK and clearly has
some advantages though also has disadvantages. The effects of the pool riffle sequence which be only
intermittently sampled with widely spaced cross sections will for example be removed.
The SOBEK code has recently been used for simulations of gravel beds on the Rhine branches in the
Netherlands. SOBEK previously had a facility to add a pseudo 2-D component of bed changes around
a meandering bend similar to that described in Chang (1988) for Fluvial 12 though this has now been
dropped. Such enhancements of 1-D codes are not being taken further as they offer limited usefulness
over 2-D or 3-D modelling due to the additional data requirements and developments of multidimensional models giving better predictions.
Other US codes such as GSTARS and EFDC-1D have specific features such as an algorithm for bank
stability simulation in the case of GSTARS and an improved ability to simulate overbank fluxes and
consolidation of sediments in the case of EFDC-1D.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
86
Accounting for Sediment in Rivers
FRMRC Research Report UR9
6.2
APPLICATION OF 1-D SEDIMENT MODELS
6.2.1 Range of Applications
For the UK, 1-D sediment models are relatively suitable because:
1.
2.
3.
Banks are relatively stable so adjustment is to bed levels rather than river length
Reach scale problems are becoming increasingly of concern (i.e. river restoration and hydromorphology requirements of the WFD)
there are a large amount of existing models and potential users
Difficulties arise due to the lack of sediment data and it is almost inevitable that some survey would be
required for a study. Often there is only a meagre resource to obtain this data. The next section
illustrates the range of recent models (at the workshop we should be able to extend this list).
6.2.2 Recent Applications
The range of application of 1-D sediment models in the UK or by UK Consultants using iSIS
Sediments is progressively increasing and is illustrated by the list of known recent applications given
in Table 6.1 which ranges from flood defence design studies to research and post flood event studies.
Table 6.1 Recent applications of iSIS Sediment
River /Model
Tees Barrage
Eden (Cumbria)
HR Flume
Organisation
Cranfield Uni
EA (Jim Walker)
Glasgow/Herriot
Watt
Type of study
Academic
Maintenance
Academic
Colorado
Glasgow/Herriot
Watt
Glasgow/Herriot
Watt
Academic
Pat Feeder Canal
Halcrow
Construction
Marala Ravi Canal
Halcrow
Feasibility
Ganges/Gorai
Halcrow
Master planning
Lower Thames
Halcrow
Feasibility
Mekong
Halcrow
Master planning
Belize City
Halcrow
Maintenance
Rothes Morayshire
Haskoning
Feasibility
Glossop Brook
JBA
Post flood /
feasibility
Tidal Dee
Harberton Ford
JBA
Nottingham Uni/
JBA
Nottingham Uni/
JBA
Feasibility
Feasibility
East Fork
Hawkcombe
Stream
Academic
Feasibility
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
Comment/Success
Not known
Yes comparative studies published
Yes using research version of code.
Parker possibly best performing
transport equation
Modified Parker gave best agreement
for degradation downstream of dam
Meyer Peter Muller and Parker gave
reasonable agreement to channel
evolution
Predicted changes in bed level over
time
Predicted changes in bed level over
time
Predicted changes in bed level over
time for Barrage
Additional units to allow for spill
incorporated during study
Training of riparian professionals as
part of decision support system
Concern re maintenance of tidal
drainage system
Reasonable agreement with observed
and used for comparative predictions
Difficulty in reproducing observed
sediment accumulation in gravel trap
during floods due to steep river.
Sandy bed tidal river
Comparison of options during
feasibility study
Derivation of options for improving
sediment continuity
08/08/06
87
Accounting for Sediment in Rivers
FRMRC Research Report UR9
River /Model
Gorai
Fall River
Colorado
Organisation
Nottingham
University
Nottingham
University
Type of study
Academic
Comment/Success
Tests on cross section separation
Academic
Unable to reproduce effects of massive
sediment influx due to limited field
data.
6.2.3 Likely Applications for which 1-D Sediment Models are Applicable
From the experience above an indication of where the use of a 1-D model is likely to be successful can
be inferred.
Comparison of options for feasibility studies for flood schemes has been successful in a number of
cases though for Glossop Brook difficulties in obtaining a reasonable calibration of sediment
accumulations in a large flood limited the usefulness of the technique. The movement of the massive
sediment influx to the Fall River following a dam break also was difficult to simulate possibly due to
insufficient field data.
Gradual changes due to a barrage or sediment trap also seem to be successful for canals and rivers.
The research cases suggested that for the gravel bed rivers and flume tests simulated additional
sediment transport formulations such as Parker (1990) and Meyer-Peter Muller (1948) would be
worthwhile implementing in the standard code.
6.2.4 Issues and Decisions to be Made when Setting up a 1-D Sediment Model
There are a number of key questions that need to be answered when starting a modelling study such as:
1.
2.
3.
4.
5.
6.
7.
8.
Sediment grading, how much data to obtain and complexity to use in the model
The range of sediment sizes to be used in the simulation
Sediment Inflows – often lack of data
Sediment transport equations to use
What flow regime should be simulated/ Run period to use
Extent to be simulated
Cross section spacing needed to represent the sediment processes concerned
What changes to an existing hydrodynamic model are needed and is simplification required for
sediment simulations such as inbank by reducing extended sections or continue using spills
9. Representation of sediment traps, bridges and culverts and out of bank flows around these
10. Run parameters such as choice of layer thickness and mixed or sorted sediment and method of
section updating
The experienced modeller has knowledge of how previous cases have been simulated and where
pitfalls may lie due to software limitations and due to weaknesses in process knowledge such as
sediment transport when flow spills over bank. For the novice more guidance than is currently
available would be useful. The next chapter considers section spacing in meandering channel as an
example of how research could help improve the use of 1-D sediment modelling. Examples and further
case studies would also be useful.
6.3
APPLICATION OF ISIS SEDIMENTS TO STUDY THE EFFECT OF THE
NUMBER OF SECTIONS USED
Although there is some guidance available on the choice of section spacing in hydrodynamic models
of floods (Samuels 1990) there is no similar guidance for sediment modelling. There will clearly be
different levels of detail required for different studies. For example, when modelling a long reach of
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
88
Accounting for Sediment in Rivers
FRMRC Research Report UR9
several hundred kilometres it would be impractical to have the very close spacing that might be
desirable over a short reach. The guidelines for flood modelling are:
1.
2.
3.
4.
Cross sections should not be more than 20xB apart, where B=top width;
Cross sections should be not more than 1/2s apart where s=the mean slope of the river;
Cross sections should be not more than 2D/s apart where D=typical river depth or characteristic
bankfull depth in the case of a flood model;
Where the mean flow velocity exceeds 1m/s the cross sectional area should not vary by more
than 35% between sections.
In the case of the Gorai River in Bangladesh, a large distributary river from the Ganges, the top width
is 250–350m, average depth at a reference bank full flow is 10m and slope is 0.000035. Hence, these
guidelines suggest cross section spacing given in Table 6.2. Depth and width vary significantly with
flow in the Gorai River, but for high in bank flows fairly widely spaced sections of 5–7km could be
used for a flood model according to these guidelines.
Table 6.2 Standard Guidelines for spacing of sections in flood models (after Samuels as quoted
in iSIS Manual (undated))
Criteria
1
2
3
4
Formula
20B
1/2s
2D/s
Should not vary by >35% area
Calculated Spacing
5-7km
14km
57km
-
Generally, the Bangladesh Water Development Board (BWDB) survey the Gorai to fixed monumented
stations (though some have been lost) with a spacing of about 5km, which would seem to be adequate
for flood modelling. For this study, however, the recently surveyed sections taken for the dredging and
river monitoring studies were available for one season and these have a spacing of 100m in the upper
reach and 200m in the lower. The locations of the sections surveyed in October 2000 are shown in
Figure 6.3. These were surveyed towards the end of the monsoon season, but prior to redredging. The
section spacing is, by normal standards and in relation to the guidelines for flood modelling, very close
and so it allows more confidence to be placed in the accuracy with which morphological features such
as deep scour holes and the extent of outer bank scour pools are represented in the survey data. Some
of the sections in the lower part of the river were taken parallel to each other and not necessarily
perpendicular to the river. Where sections were too oblique to the downstream direction to be useful,
they were not used in the model.
The response of the upper part of the Gorai near the offtake from the Ganges is complex and depends
largely on the inflow of water and sediment from the main river. For the purposes of testing section
spacing, the first few kilometres of the Gorai were, therefore, excluded from consideration and an iSIS
Sediments model was constructed starting at the Gorai Railway Bridge and extending downstream to
the limit of the number of sections permitted in an iSIS Sediment Model (which at the take of this
work was 400 sections). This resulted in the iSIS model shown in Figure 6.4.
The model covered a number of bends and a reasonable length of the river (32km) to allow assessment
of the changes that occur during an annual discharge cycle and to illustrate how modelling these
changes is influenced by cross section spacing.
A program was written to generate versions of the model with successively fewer cross sections
spaced at wider and wider intervals, up to a spacing of about 4km (with just 9 sections) compared to
the original 250 sections spaced at 100m to 200m intervals. The ‘trimmed’ models had common
starting and ending sections to minimise the effect of boundary conditions on the results. A schematic
of the trimmed model with sections spaced at 4km shows a dramatic reduction in the level of detail
compared to that in the original model (Figure 6.4). It should be borne in mind that the spacing of
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
89
Accounting for Sediment in Rivers
FRMRC Research Report UR9
sections in this, the most reduced model is actually the best that could be developed from a standard
BWDB river survey.
Figure 6.3
Gorai River Survey Sections from 3rd year dredging programme. Spacing of
survey lines is approximately 100m for a reach of 32km from the offtake from the
Ganges
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
90
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Gorai Railway Bridge
Individual Sections
QH Boundary at
end of model
Figure 6.4
iSIS model schematic with full number of sections (above) and at 4km spacing
(below) used for testing separation of sections
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
91
Accounting for Sediment in Rivers
FRMRC Research Report UR9
The iSIS Sediments model was run using a synthetic repeating pattern of wet season/dry season
discharges based on the flows observed in the year 2000. This was done to allow the sections to reach
a condition of dynamic equilibrium rather than the continual state of change due to the year on year
differences in run off that would be present in a natural record. The duration of low flows over the dry
season is cut short as very little sediment movement takes place during that period because flows go
down to only 50–100m3/s, and without dredging the Gorai ceases to flow at all. Sediment
concentrations at the inflow boundary were based on the sediment rating curve derived from field
measurements. The sediment concentration at the Gorai Railway Bridge does not increase markedly at
high flows, the exponent in the rating (1.234) indicating that sediment concentration is only a weak
power function of discharge. The Engelund Hanson formula was used for sediment transport
calculations as this has generally been found to give reasonably reliable results in Bangladesh. No
calibration factor was applied.
The long profile of the full model appears similar to the echo sounding trace (Figure 6.7) taken in 1992
(Halcrow 1993), which at the time seemed to show far greater variation in depth than was thought
physically likely. The long section for the model with a with 4km section spacing is shown in Figure
6.8, demonstrating that clearly much less detail will be obtained from the model with 4km spacing,
particularly regarding the impact of changes in depth around a bend. It is clear that a simple echo
sounder survey of the type described above may thus add useful knowledge on depth variation in
bendways in a large river such as the Gorai.
The full and reduced models were each run for the equivalent of 9 years and results are given in
Figures 6.9–6.12 in terms of simulated minimum bed levels. Results are presented for four locations,
giving covering a range of planform positions relative to bend apices (as noted in the figure captions).
Intra-season changes are generally greater during the first few years of the simulation as cross sections
are scoured and filled and the system adjusts to the imposed water and sediment inflow regimes and
transport processes. Variation is more limited at meander crossings, such as Node 280 (Figure 6.11)
than within bends (Figure 6.12). At the upper reach of the model, bed levels seem to have settled down
to a seasonal cycle and the river may be considered to be in dynamic equilibrium. The changes in the
downstream bends are less pronounced than those upstream and it is likely that the sediment regime
imposed from upstream influences the intra-seasonal bed changes. Sediment concentrations at the
upstream and downstream limits to the model (Figure 6.13) suggest sediment concentrations at the
Gorai Rail Bridge site might be influenced by upstream sediment influxes that do not transfer all the
way down the river. In fact, changes in sediment concentration with discharge lower down the river
are much greater than those measured at the Railway Bridge and, consequently, imposed at the
upstream boundary. This pattern also emerges in the ‘sparse’ model with sections at 4km intervals,
suggesting that the transport of sediment in the Gorai varies much more with discharge than is
suggested by measurements at the Gorai Rail Bridge gauge site.
There is evidence from other studies on the Gorai and Ganges (DHI 1996) and the Gorai Restoration
studies (Haskoning 2001) to support the contention that the sediment rating relationship can vary
significantly with location and may not be representative in places such as bridges where it is easier to
measure which may be constricted. As shown in Figure 6.14, the Kushtia sediment gauge site (which
is more constricted than the Gorai Rail Bridge site due to bank protection groynes) has a steeper
sediment rating curve to that indicated by observed sediment concentrations in 1996–7 (DHI 1996)
and by SWMC for the Gorai River Restoration Project in 1999–2000 (Haskoning 2001).
Comparison of results for the reduced models shows that intra-seasonal changes in simulated variation
of bed levels and the eventual equilibrium levels are both affected by the distance between cross
sections. The simulations reported show that a spacing of less than 400m is necessary to reproduce the
local bed level fluctuations revealed by the most detailed model. This would imply the need for cross
sections to be spaced at less than 2xB, compared with the spacing of 20xB suggested for flood
modelling, if the model is to reproduce the morphological changes that occur annually. While in some
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
92
Accounting for Sediment in Rivers
FRMRC Research Report UR9
applications it may be unnecessary to model such changes, another significant finding is a tendency for
models with fewer sections to evolve towards lower bed levels, if roughness is kept constant. Whilst
increasing the roughness during a calibration phase might rectify this problem, because the roughness
is translated into a shear stress at the bed then the sediment transport rate would be expected to
increase unjustifiably as a result.
Figure 6.5
Repeating flow and sediment boundary used in Gorai River section tests
Figure 6.6
31km Long Section of Gorai from the Gorai Rail Bridge – Full Model with 100m &
200m section spacing
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
93
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 6.7
255km Long Section of Gorai River from Ganges to Khulna measured from boat
in 1992 (Halcrow 1993)
Figure 6.8
Long section of Gorai River from Gorai Rail Bridge using 4km section spacing
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
94
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 6.9
Gorai River changes in simulated bed elevation over 9 years for different section
spacings – Node GOR120 at Upstream end of reach
Figure 6.10 Gorai River changes in simulated bed elevation over 9 years for different section
spacings – Node GOR240 crossing section after 2 bends
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
95
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 6.11 Gorai River changes in simulated bed elevation over 9 years for different section
spacings – Node GOR280 crossing after 3 bends
Figure 6.12 Gorai River changes in simulated bed elevation ove 9 years for different section
spacings – Node GOR330 middle of fifth bend
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
96
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 6.13 Gorai River change in sediment concentration at the start of the model and at the
final crossing
Sediment Rating Curves - Gorai
1,000,000
Sediment Discharge (tonnes/day)
100,000
10,000
1,000
100
10
10
100
1,000
10,000
Discharge (m3/s)
SWMC 90-98(before dredging)
SWMC 98-99(after dredging)
FAP4 66-67
FAP24 66-70
FAP24 (Kushtia)
BWDB (GRB)
GRC (Kushtia)
SWMC (Kushtia)
Figure 6.14 Gorai River sediment rating curves at Kushtia and Gorai Railway Bridge derived
in different studies and recent sediment measurements (supplied from GRRP
study)
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
97
Accounting for Sediment in Rivers
FRMRC Research Report UR9
6.4
USES AND LIMITATIONS OF 1-D SEDIMENT MODELS
The range of uses of 1-D models is illustrated by the examples listed in Chapter 2. Given a practical
need for a quantitative sediment study 1-D sediment modelling should be a distinct possibility for a
relatively small budget especially where hydrodynamic models already exist.
Limitations to the technique include:
1.
2.
3-D effects of sediment transport such as local scour at bridges are not accounted for
sediment calculations for compound channels and overbank flows may be less accurate due to the
averaging of parameters in a !-D simulation
3. Layers of sediment are assumed homogeneous, i.e. sorting around meander bends does not occur
4. A fixed Mannings N is used and this is applied to both channel resistance and bed shear stress
used in sediment transport calculations
5. The effect of dunes on resistance and sediment mixing is not included explicitly
6. Simulation of gravel traps which are small relative to cross section spacing elsewhere can be
difficult within the model
7. Steep rivers and high Froude numbers may affect model stability and sediment transport
8. Simulation of armouring effects is dependent on criteria selected for layer thickness
9. Sediment transport formulations for gravel bed such as Peter Meyer Muller or Parker as
implemented in the research version of the code may not be in the standard program
10. Bank erosion effects are not well represented in the standard model
11. Selection of the technique for updating of sections can influence results
12. Interpretation of results may depend on user experience.
Many of these limitations also apply to other modelling techniques and are a function not only of the
sediment model limitations but the limitations in practical knowledge of sediment processes. Although
uncertainty associated with sediment calculations is high and this influences the uncertainties in a 1-D
model, this is associated with measurements as well as predictions and performance of the model as a
predictive tool within a widely stochastic environment may be adequate. The modelling technique is
also stronger for comparison of options which is frequently the main requirement for a feasibility
study. Bradley et al. (1998) consider applicability and limitations of sediment transport in gravel bed
rivers and conclude that it is advisable to have reasonable calibration and verification data to develop a
good predictive tool. They also believed that with insufficient data a good deal of judgement is needed
by the modeller.
In Chapter 3 an example of some of the complexities and data requirements of 1-D Sediment
modelling were presented. In the UK where gravel bed rivers dominate, the data limitation is more
typically associated with knowledge of the bed sediment grading and fluxes and similar tests on model
inputs can be expected to inform the 1-D Sediment modeller on parameter selection so that the user
base can be widened and better confidence in results can be obtained. For the next stage of the FRMC
it is thus proposed that tests should be carried out using typical UK datasets to provide guidance and
case studies on the use of 1-D Sediment models including iSIS Sediments and potentially HEC-RAS
sediment simulation if this is now available. Links with data available on bed material from the River
Habitat Survey and GeoRHS should also be explored.
6.5
REFERENCES AND FURTHER READING
In addition to program manuals, a useful reference for practical application of 1-D sediment models is
that prepared by HEC and is available as an appendix to the HEC-6 manual at
www.hec.usace.army.mil/software/legacysoftware/hec6/documents/td13man.
This publication covers some US experience in collection of data from different sources, calibration
and desirable sensitivity tests.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
98
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Ackers, P. and White, W.R. 1973. Sediment transport: new approach and analysis. Journal of the
Hydraulics Division American Society of Civil Engineers 99 (HY11)
Ackers, P. 1993. Sediment transport in open channels: Ackers and White update Proceedings of the
Institution of Civil Engineers Water, Maritime & Energy 101 2041-2060.
Bettes, R. & White, W.R. 1981. Mathematical simulation of sediment movement in streams. Proc ICE
part 2 Vol 71 Sept 1981
Bradley, J.B., Williams, D.T., Walton, R. 1998. Applicability and limitations of sediment transport
modeling in gravel bed rivers. In Gravel-bed rivers in the environment, Klingeman, Beschta, Komar,
and Bradley, eds., Water Resources, Highlands Ranch, Colo.
Chang, H.H. 1988. Fluvial processes in river engineering. John Wiley & Sons.
Delft Hydraulics, Danish Hydraulic Institute, 1996a River Survey Project Special Report No 10
Morphology of Gorai Off-take. Project Report under FAP24 for Government of Bangladesh, Water
Resources Planning Organisation & European Commission. 158p
Delft Hydraulics, Danish Hydraulic Institute, 1996b. River Survey Project Special Report No 18
Sediment rating curves and balances. Project Report under FAP24 for Government of Bangladesh,
Water Resources Planning Organisation & European Commission. 60p
Engelund, F. and Hansen, E. 1967. A monograph on sediment transport in alluvial streams. Teknisk
Vorlag, Copenhagen, 65p
Halcrow, DHI, EPC & SSL 1983. Southwest Areas Water Resources Management Project FAP4 Final
Report Volume 3 Morphological Studies. For UNDP, ADB, Bangladesh Flood Plan Coordination
Organisation.
Haskoning, 2001. Gorai River Restoration Project. Report prepared for Bangladesh Water
Development Board and World Bank Global Environment Fund.
Heyter, E. 2002. Overview of type 2 sediment transport models. Paper presented at Sediment
Stability Workshop. New Orleans January 22-24 2002. US EPA, US Corps of Engineers. and US
Navy.
iSIS, 1999. iSIS Sediment manual. Halcrow and HR Wallingford
iSIS, 1999(2). iSIS Flow Manual. Halcrow and HR Wallingford
iSIS, 2001. iSIS Sediment manual (research Version). Halcrow, HR Wallingford. University of
Glasgow and Herriot Watt University.
Meyer-Peter, E. & Muller, R. 1948. Formulas for bedload transport Proc 2nd IAHR Meeting
Stockholm.
Parker, G. 1990. Surface-based bedload transport relation. J. Hydraulic Research 28(4)
Samuels, P.G. (1990). Cross-section Locations in 1-D Models –International Conference on River
Flood Hydraulics, White WR (ed), John Wiley.
Schvidchenko, A.B., Pender, G., Hoey, T.B. 2001. Using iSIS Sediment for simulating graded
sediment behaviour in rivers. River Basin Management Ecology and the Environment v50.WIT
Southampton
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
99
Accounting for Sediment in Rivers
FRMRC Research Report UR9
SOGREAH, 1963. Mathematical Model of the Mekong Delta. 7 Volumes UNESCO.
Walker, J. 2001. The use of sediment modeling techniques to address the differing needs of
management on the River Eden Cumbria UK.Journal of the Chartered Institution of Water and
Environmental Management 15 No 4 Westrich, B. and Jurashek, M. 1985. Flow transport capacity for
suspended sediment. Presented at the 21st Congress IAHR, Melbourne, Australia
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
100
Accounting for Sediment in Rivers
FRMRC Research Report UR9
7.
The Cellular Automaton Evolutionary Slope And
River model (CAESAR)
Tom Coulthard1 and Marco Van De Wiel2
1.
2.
7.1
Geography Department, University of Hull, U.K. [email protected]
http://www.coulthard.org.uk
University of London, Ontario, Canada. [email protected]
INTRODUCTION AND BACKGROUND TO CAESAR
CAESAR is a two dimensional flow and sediment transport model. It can simulate morphological
changes in river catchments or reaches, on a flood by flood basis, over periods up to several thousands
of years. It was initially designed to simulate the geomorphic response of river catchments to changes
in climate and/or land cover. Its purpose was to address the debate as to whether changes in climate or
anthropogenic changes in land cover had led to changes in UK river behaviour over the Holocene. In
order to answer these research questions the model had several requirements:
1.
2.
3.
4.
To simulate the relevant time scales (10’s to 10 000 years – Holocene).
Cover suitable spatial scales, e.g. a long reach of river or a catchment
To incorporate the relevant processes and parameters that were operating over the above scales
(e.g. hydrologicial, fluvial and slope processes)
To simulate the whole catchment. There are many feedbacks and interactions between processes
operating within a river catchment, and it is important to try and include this within the model.
This remit was initially part of a PhD project, and since then the model has grown, both in
sophistication and application. CAESAR has proved to be very flexible, and to date has been applied
to over 20 different catchments and reaches on scales ranging from 500m to 500km2, and over time
scales from individual floods to 10 000 years. A chronology of the model development is shown in
Table 7.1.
Table 7.1 Chronology of CAESAR development
Date
Research Topic
Associated Publications
1995-1998
PhD research and development, application to Cam Gill Beck,
Starbotton, Yorkshire Dales. Examining interactions of climate
and land cover change on a small upland catchment (4km2)
Coulthard et al., (1996, 1997,
1998a, b, c, 1999, 2000, 2001,
2002)
1998-1999
part of the NERC LOIS project, modelling the Holocene
development of the main northerly tributaries of the Yorkshire
Ouse (Rivers Swale, Nidd, Ure, Wharfe)
Coulthard and Macklin (2001,
2003) Coulthard et al., (2005, In
Press)
1999
Development of the TRACER addition to CAESAR to
simulate the movement of contaminated sediment through river
systems. (applied to the River Swale, UK.)
Coulthard and Macklin (2003)
2002-2005
Further development of CAESAR (mainly by Marco Van deWiel), including the addition of suspended sediment and
improved flow routing schemes
Van de Wiel et al., (In Press),
Coulthard et al., (In Press)
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
101
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Date
Research Topic
Associated Publications
2003-2005
Application to 7 reaches of 4 river catchments (Rivers Severn,
Teifi, Dyfi and Dee) working with Dr Gez Foster
Brewer et al, (2005), Coulthard
et al., (In Press)
2006
Development of lateral erosion routine (with Marco Van deWiel’s assistance)
Coulthard and Van de Wiel
(2006)
2006 –
2008
Further development of TRACER components as part of the
EPSRC FRMRC program
2005 – ?
Development by others workers including; Joe Wheaton
(UWA) developing ooCAESAR; Clare Cox (Cambridge)
working with a Matlab version of the routing code; Katharine
Welsh (Liverpool) applying CAESAR to the Annecy
catchment, France; NIWA (New Zealand) developing and
applying CAESAR to solve contemporary management issues
in braided channels.
7.2
BACKGROUND TO CELLULAR MODELS
CAESAR is one of a new genre of cellular fluvial models that have also been termed ‘reduced
complexity’ models. They have developed partly to fill the gap between complex 2- and 3-D CFD
approaches that are often too complex to be readily applied to large areas over timescales greater than
a single flood event, and coarse resolution landscape evolution models that simulate the development
of landscapes over thousands of years.
Cellular models in geomorphology can be defined as representing the modelled landscape with a grid
of cells, and that the development of this landscape is determined by the interactions between cells (for
example fluxes of water and sediment) using rules based on simplifications of the governing physics
(Nicholas, 2005). In fluvial geomorphology, cellular models use simplified or ‘relaxed’ versions of the
complex flow equations used in CFD models. This allows a substantial increase in speed of operation,
which in turn enables them to be applied to long reaches and large catchments over ‘useful’ time
scales. Importantly, the increase in computational speed and simplicity also allows these models to
include sediment transport processes between cells, meaning that morphological change can also be
modelled.
The first of these cellular models was the braided river model of Murray and Paola (1994). This
simulated the development of a braided river by routing water discharge over a grid of cells
representing the channel and braidplain according to local variations in bed slope. Erosion of these
cells was then carried out according to simple discharge-dependent erosion rules, and the eroded
material was transported to adjacent cells again according to bed slope. Their simple flow model
allowed divergent and convergent flow, and importantly the width of channels was represented across
one or more cells. There were no calculations of depth, momentum or velocity yet the model produced
qualitatively realistic braided patterns. Importantly, it reproduced the dynamic behaviour of a braided
channel with the downstream and lateral migration of bars and channels. By simplifying (sometimes
grossly) the laws of physics, Murray and Paola (1994) recreated the basic conditions that cause a river
to braid: laterally unconstrained flow, mobile bed material and erodible banks. This simple model
represented a paradigm shift in both how we look at braided rivers and how we model them. For
fluvial models, it indicated that perhaps we do not have to pursue reductionist approaches by trying to
simulate every process operating within a river channel in great detail. It also raised the possibility of
simulating the general behaviour of fluvial systems using a far simpler approach. This ‘experimental’
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
102
Accounting for Sediment in Rivers
FRMRC Research Report UR9
approach is important for researchers, as often a qualitative understanding of the dynamics of a system
is more important than a quantitative solution.
Following their 1994 paper, Murray and Paola (1997) carried out an extensive review of their model
and subsequently integrated a simple vegetation growth model to examine how stabilising the braid
plain with vegetation would alter the channel pattern (Murray and Paola, 2003). The Murray and Paola
model also inspired the development of a series of similar cellular models. Coulthard et al. (1996,
1998) developed a cellular automaton model of river catchment evolution that was further developed
into the CAESAR model described in more detail later (Coulthard et al., 2000, 2002, 2005). This
model built upon the flow routing methodology developed by Murray and Paola (1994, 1997) by
including a calculation of flow depth, a more detailed representation of sediment transport using
multiple grain sizes, and hillslope processes (e.g. landsliding and soil creep). CAESAR has been
applied to a range of river catchments and reaches (4 to 40 km2) with grid cell sizes ranging from 2 m
by 2 m to 50 m by 50 m. Thomas and Nicholas (2002) developed a cellular model of braided rivers
(termed CRS) that used a flow model that built upon and refined the Murray and Paola method. They
applied this to a 470 by 230 m reach of the Aroca River, New Zealand with 1 m grid cells, and
favourably compared the simulated inundation extents and flow velocities to results from a 2D CFD
model of the same reach (Hydro2de). Cox et al. (2005) has also compared and reviewed the flow
routing capabilities of the Murray and Paola method, the CRS and CAESAR models. There is also a
series of reduced complexity flood inundation models based upon the Lisflood model (Bates and De
Roo, 2000), which uses kinematic wave equations to route a wave of water down the main river
channel, then where banks are overtopped uses a cellular algorithm to route flow across the floodplain.
Nicholas (2005) has outlined the principles and issues of cellular modelling in fluvial geomorphology,
commenting that cellular models represent ‘one of the most important advances in fluvial
geomorphology over the past decade’. However, Nicholas (2005) notes that there are technical issues,
such as flow routing algorithms that tend to concentrate flow disproportionately, and many difficulties
with validation. Nevertheless, Nicholas (2005) recognises the significant potential for multi-scenario
‘what if’ modelling and the capability for simulating extended time scales that can, for example, allow
the effects of climate change on fluvial geomorphology to be modelled. On a more philosophical note,
Nicholas (2005) comments that cellular models may also encourage open mindedness when
developing models and hey can challenge reductionist ideals.
As noted above, these recent model developments provide considerable potential to simulate
morphological change in river catchments and reaches over pertinent time and space scales (e.g. 1–100
yrs and 1–100 km2). The defining qualities of these cellular models are their ‘medium’ time and space
scales, their inclusion of erosion, deposition and morphological change and that within the model the
channel is treated as one or more cells wide.
7.3
THEORY AND MODEL DESCRIPTION
A brief description of the CAESAR model is provided here, but for more detailed information, readers
are referred to Coulthard et al. (2002) and Van De Wiel (In Press). CAESAR is a cellular model that
uses a regular mesh of grid cells to represent the river catchment studied. Every cell has properties of
elevation, water discharge and depth, vegetation cover, depth to bedrock and grain size. It is based
upon the cellular automaton concept, whereby the repeated iteration of a series of rules on each of
these cells determines the behaviour of the whole system. CAESAR has a set of rules for a
hydrological model, hydraulic model (flow routing), fluvial erosion and deposition and slope erosion
and deposition. For every model iteration, cell properties (e.g. elevation) are updated according to the
rules, and the interaction between an individual cell and its neighbours. For example, the amount of
fluvial erosion in a cell may depend upon the depth of water in the cell and the slope between that cell
and its neighbours. A schematic of the models operation is shown in Figure 7.1.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
103
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 7.1
Schematic of CAESAR’s operation
The following paragraph briefly describes how the model operates running through the hydrological,
hydraulic, fluvial and slope erosion models. CAESAR uses an hourly rainfall record as the input for a
hydrological model (based upon TOPMODEL, Beven and Kirkby, 1979), which may be altered to
represent the hydrological effects of different vegetation covers. The output from the hydrological
model is then routed through the catchment using a scanning multiple flow algorithm that sweeps
across the catchment in four directions (from north to south, east to west, west to east and south to
north, see Figure 7.2). CAESAR can also run in reach mode (see below) where instead of inputting a
rainfall data set, a discharge can be inputted at one or a series of points, and then routed across the
catchment with the same scanning algorithm. In each scan, flow is routed to the three down slope
neighbours (as per Murray and Paola, 1994), but if the total flow is greater than the subsurface flow,
the excess is treated as surface runoff and a flow depth is calculated using an adaptation of Manning’s
equation. The maximum depth calculated for the cells over all four scans is then recorded. Any flow
that is not removed from the basin remains for the following iteration, allowing hollows to fill up and
also any flow ‘trapped’ in meanders remains, enabling complex channel patterns (such as braids and
meanders) to be simulated. For all cells with a flow depth, fluvial erosion and deposition is calculated
using the Wilcock and Crowe equation (Wilcock and Crowe, 2003). This is applied to 11 grain size
fractions (from 1 to 256 mm) that are integrated within a series of active layers (Hoey and Ferguson,
1994) that allows surface armouring to develop as well as a limited stratigraphy. Over the course of
long simulations, this importantly allows previously deposited finer sediment (as on a floodplain) to
become future erosion sources. Limited slope processes are also included, with mass movement when
a critical slope threshold is exceeded, together with soil creep. These allow material from slopes to be
fed into the fluvial system as well as the input from landslides (both large scale and small – e.g. bank
collapse). After the fluvial erosion/deposition and slope process amounts are calculated, the elevations
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
104
Accounting for Sediment in Rivers
FRMRC Research Report UR9
and grain size properties of the cells are updated simultaneously. A variable time step is utilised
(operating between 10-6s and 104s) that restricts erosion to 10% of the local slope, preventing
computational instability. Therefore, despite being complex in operation, CAESAR only requires the
simple inputs of topography (a DEM), an hourly rainfall record and a land cover record to drive a
sequence of erosion, deposition and landscape evolution. This is then used to simulate individual
floods, responding to both local hydraulic responses from runoff events, as well as cumulative inputs
(or deficits) arriving from up-catchment that may themselves have been triggered by previous
conditions.
1
2
Direction of main
valley floor slope
3
4
Figure 7.2
7.4
Schematic of the scanning algorithm
APPLICATION
7.4.1 Usability
CAESAR is coded in Visual C#, and runs as a windows program on Windows NT, 2000 and XP. No
programming experience is required in order to use it, and example files can be loaded and the
program successfully run in minutes. Applying it to different data sets requires the capability to
manipulate and edit DEM files, and the user would require some basic knowledge about data
manipulation using (for example) Excel. The source code for CAESAR is openly available for
download under the terms of a GNU licence which prevents it from being sold for profit. The code is
presently more than 10,000 lines, but there is much duplication, and users with medium programming
skills should be quite capable of editing and altering it for their own purpose.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
105
Accounting for Sediment in Rivers
FRMRC Research Report UR9
7.4.2 Data Requirements
CAESAR can be run in two modes; a catchment mode, with no external fluxes or inputs aside from
rainfall; and a reach mode with one or more points where water and sediment are inputted to the
system.
Catchment Mode Data Requirements
For the catchment mode, CAESAR requires an hourly rain data set. Ideally, the study catchment
should have such a rainfall record as well as a gauged point or outlet. If this is the case, the
hydrological model can be calibrated, so modelled discharges match the field observed discharges for
a given flood. However, if this is not available nearby rain data can be used and there are ranges of
example settings from which the hydrological model can be parameterised. CAESAR also requires a
raster DEM (not TIN) for the catchment, and editing and correcting the DEM is an important part of
preparing for a CAESAR simulation. The model can cope with a wide range of DEM resolutions, and
has been applied with DEMs of grid cell sizes ranging from 1m to 100m. Some DEM’s can be applied
in their raw form, but often the data contains errors which can cause the model significant problems,
for example an erroneous series of cell elevations can cause an obstruction across a valley floor. It is
therefore recommended that DEMS are first processed to remove any sinks or pits, and to ensure that
the drainage network follows a straightforward descent to the exit point. This can be carried out simply
using the freely available ARC-HYDRO extensions toolkit for ARC-GIS 8.x and 9. CAESAR is also
set up so that the exit point from the DEM must be on the right hand edge of the map. This may
require the DEM being rotated in order to correctly align the exit point. Furthermore, the model will
not route water or allow water to exit from no data cells (those with a value of -9999) so these must be
removed from the right hand side of the DEM. CAESAR accepts data in an ascii raster format which
consists of a 6 line header, followed by the grid cell elevations in rows and columns. This is in the
same format as data exported from ARC-VIEW, ARC-GIS etc. using the RasterAscii command or
equivalent.
Reach Mode Data Requirements
For the reach model, CAESAR also requires a DEM file in the same format. Again, it is worth taking
time to ensure that there are no errors in the DEM. Sometimes, there are individual cells, or groups of
cells that may require editing or removing, and for this purpose a useful program called RasterEdit
(created by Marco Van de Wiel) is available from the CAESAR website. As for catchment mode, the
main flow must exit from the right hand edge of the DEM. In reach mode an additional file is required
that contains the water and sediment inputs for the reach. These are stored in an ascii file with the time
step in the first column, water discharge in the second, and the inputs for the separate grainsize
fractions (in m3 for the time step) in the 6th to 14th column. This file is in the same format as one of the
catchment output files (see later). CAESAR can also be run in both catchment and reach mode, so for
catchments that also contain a point source (for example a major tributary) the model can take both
rainfall and point inputs.
7.4.3 Notes on Other Parameters
CAESAR also requires information on the grainsize distributions for the catchment. It presently takes
up to 9 different grainsize fractions and can cope with both bedload and suspended load fractions. The
model operates using a variable time step controlled by the amount of erosion and deposition
occurring within the catchment. A parameter (erodelimit) is set which represents the maximum amount
of erosion or deposition that can happen within any one time step. If this amount is exceeded, the
model halves the time step and repeats erosion calculations until it is below this limit. This ensures
numerical stability (as too great a time step can lead to excessive amounts of erosion and deposition)
and allows the model to have long time steps (up to 1 hour) during periods of quiescence (e.g. low
flows) yet have small time steps (<0.1s) during floods or periods of erosive activity.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
106
Accounting for Sediment in Rivers
FRMRC Research Report UR9
7.4.4 Grid Cell Size and Resolution
CAESAR can accept any grid cell size in the DEM (though all cells must be the same size) and has
been used with DEM’s from 1m to 100m cells. However, choice of grid cell size is important, as there
are significant compromises to be made between the area that can be modelled, the resolution, and the
time it takes the model to run. CAESAR can run with up to 2 million grid cells, but is probably best
suited to applications with 250,000 to 500,000 cells. Quite simply, the smaller the number of grid
cells, the faster the model will operate. This is particularly important as increasing the resolution
linearly, results in an exponential increase in the number of grid cells. Furthermore, the erodelimit
parameter – or the amount that can be eroded or deposited on a cell per iteration – can be contingent
on grid cell size. Changes in cell elevation represent changes in local slopes, and a 0.1m change with
1m cells equals a 10% change in slope, yet a 0.1m change in 10m cells equals a 1% change. Thus
increasing the grid cell size of the DEM that is being modelled results in a greater than exponential
increase in computational time, as changes between grid cells result in less severe alterations in slope.
These resolution issues are also contingent upon the time that is required to be modelled. If a single
flood is to be simulated, then this can be carried out at a higher spatial resolution that may (for
example) take a day to run. If 100 years are to be simulated, this period may contain 300+ floods, and
so take 300 days to complete. Examples of model run times include 2 months to simulate 10,000 years
on a 800 by 200 cell DEM (50m resolution) of the River Swale, U.K. There are many ways in which
the model speed can be increased, including parallelisation by dividing a catchment into sub
catchments and running these sub catchments simultaneously on separate machines. The output from
these sub catchments can then be fed into one central model.
7.5
CASE STUDIES
In the following section, two examples shall be demonstrated. Firstly an example of CAESAR
operating in reach mode, simulating erosion and deposition on a 2km section of the River Teifi, near
Lampeter, U.K. Secondly, a catchment example for the River Swale, U.K. Both examples are using
sample data that can be downloaded from http://www.coulthard.org.uk/downloads/visualcaesar.htm
7.5.1 Reach mode: Modelling the River Severn nr. Caersws, U.K.
This example is for a c.2km by 2km section of the River Teifi, immediately downstream from
Lampeter, Wales. The DEM is at 10m resolution and created from downgraded LiDAR data. To run
this example, download the relevant files which include the main program file (Caesar 5.1.exe) and the
sample data files teifi.zip and swale.zip. Unzip and place them all in one directory.
The three files to be used for the Teifi example are highlighted in Figure 7.3. They are whole9.txt (the
DEM file) input2.txt the water and sediment input data file and teifi1.xml – the configuration file.
Figure 7.3
Files to be used for the River Teifi example
To run the CAESAR model itself, double click on the Caesar 5.1.exe file and after a few seconds the
main start up page of CAESAR will come up (Figure 7.4). This contains a series of tabs, that contain
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
107
Accounting for Sediment in Rivers
FRMRC Research Report UR9
groups of checkboxes and buttons that allow you to alter the parameters for the model. The
explanation for all these parameters is too long for this brief guide, but will be prepared in to a
technical appendix for the model. In short, each tab contains the parameters for different sets of
processes and parameters – files, hydrology, the DEM, input points etc.
Figure 7.4
Main CAESAR interface window
Click the config file menu and open, then navigate to the folder where the downloaded files are, and
open the configuration file teifi1.xml. This will load in a set of default parameters for the Teifi
example. If you change any of the parameters within the set up screens, you can save them as a
configuration file, which allows the easy set up of the model.
To run the model simply click the large load data button, then when the data has loaded, the start
button to the right will be highlighted. Click this and the program will start. It takes a few seconds to
initialise the data and to go through some initial scans – in order to define the area where the hydraulic
model will operate. After this wait, figures at the base of the window should start to change. These
indicate the number of iterations the model has carried out, the modelled time, the water discharge
exiting the model (at the right hand edge) and the volume of sediment being eroded for each time step.
Whilst the program is running, the menu at the top of the screen allows the user to save the
configuration parameters as a config file, and also to change the display options. There are two menu’s
of display options, that allow the user to display a shaded relief map of the DEM, water depth, erosion
and deposition that has occurred since the model started, grainsize, shear stress, velocity etc.. Figure
7.5 illustrates the Teifi reach example showing the DEM in the background with water depth (a) and
erosion and deposition (b). The final menu group controls the files that are saved by the model. The
elevations, grainsize data, water depths and other parameters (that are listed on the menu) are
automatically saved every 100 iterations. This allows the model to be re-started from the point at
which it was last running (e.g. if the computer is switched off) by using the elevation and grainsize
data as starting data for the new run. Processing the output data is covered further on in this section.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
108
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 7.5
Screen showing water depths (left) and changes in elevation (right: red is erosion,
green is deposition)
To simulate a flood, we can alter the discharge in the input file. Open the input file (called input2.txt)
in a text editor (e.g. Notepad) and edit the second number in the first line from 10 to 120 – and re-run
the model again. This alters the input discharge for the first period of the models run from 10 to 120
m3s-1 which causes the channel to flood, as shown in the right hand frame of Figure 7.6. This
demonstrates how the input file could be a gauged record and a series of flood events could easily be
run through the model. The parameters in the start up screen allow the user to specify many options,
including changing grainsizes, slope failure thresholds, alterations to the flow model, how shear stress
is calculated, which input files are used, what and when output files are saved. There is also the option
to record the visual output as an .avi movie file.
Figure 7.6
Input file, highlighting first discharge input point and the corresponding
inundation areas caused by raising that value from 10 to 120
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
109
Accounting for Sediment in Rivers
FRMRC Research Report UR9
This example demonstrates how rapidly CAESAR simulates channel and floodplain flow patterns, as
well as how it models fluvial erosion and deposition. If model runs are continued, then the erosion and
deposition patterns can change, and provide results as illustrated in Figure 7.7, where there has been
erosion and deposition within the channel, the deposition of overbank fines and the development of
channel bed armour.
CAESAR is capable of modelling larger reaches and in more detail. This basic example is to
demonstrate how the model is relatively straightforward to set up and run, and to give an indication of
what sort of results it can generate in reach mode.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
110
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 7.7
Image showing the section of the River Teifi modelled in Figures 7.5–7.7. This
illustrates erosion and deposition within the channel and flooplain, with the lower
left hand image showing the D50. The right hand frame shows changes in the D50
across the cross section marked on the left hand image, as well as elevation changes
that occurred during the simulation. A more detailed explanation is provided in
Van de Wiel et al. (In Press)
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
111
Accounting for Sediment in Rivers
FRMRC Research Report UR9
7.5.2 Catchment Example: Modelling the River Swale, U.K.
This example is for the a 40 by 10km catchment area of the River Swale, Yorkshire Dales. For this
much larger area the DEM is of 800 by 200 50m grid cells. This example runs with the other sample
files downloaded in the zip file swale.zip. The files used to drive the catchment simulation are
different, instead of an input file, the simulation uses an hourly rainfall data set, that comprises of a
text file with the hourly rainfall rate in mm on each line. To run this example, again double click on
the Caesar 5.1.exe file and from the config-file menu open the configuration file swale1.xml. As this
run is operating in catchment mode, the user has to uncheck to reach mode box and check the
catchment mode box, as shown in Figure 7.8.
Figure 7.8
Selection of catchment mode computation
Then, as for the previous example, click load data, and then start when the start button becomes
highlighted. This will take longer to start than the previous reach example, and this is due to the DEM
containing 4 times the number of grid cells. When the model begins to operate, a small part of the
drainage network should appear, as shown in Figure 7.9. This is only a fraction of the total network,
and examining the water discharge figures (3rd box from the left on the bottom info panel) there is only
a discharge of 1.3m3s-1. This is very small for a 400km2 catchment, and this is due to the hydrological
model only having received a small volume of rainfall. The catchment is in effect in drought
conditions. After approximately 23 days, a significant rainfall event occurs in the rainfall data set, and
this causes the discharge to increase, and the drainage network to expand, as shown in Figure 7.10.
This is a very modest event for a catchment of this size (only 20m3s-1) but it illustrates how the
hydrological model interacts with the catchment drainage network.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
112
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 7.9
Initial drainage network for the Swale catchment simulation
Figure 7.10 Expanded drainage network, due to rainfall event
As per the reach example, when running in catchment mode CAESAR can produce output for
elevations, surface grainsize, water depths etc.. selected using the options in the save options menu.
Importantly, as mentioned previously, both of these examples are rotated so that the main water
outflow is at the far right hand edge of the DEM.
7.5.3 Processing the Results
CAESAR generates data in two forms. Tables of ascii data that can be readily imported into ARC-GIS
for the spatial data (e.g. elevation, water depth, grainsize) and an output file containing the water and
sediment discharges output by the model at the right hand edge of the DEM. The format of the data is
illustrated in Figure 7.11, and comprises of a 6 line header then the spatial data (e.g. elevations as in
Figure 7.11) as the value for each grid cell arranged in rows and columns. This is simply imported into
ARC-GIS or ARC-View using import to ascii options (asciiraster) and can then be manipulated or
displayed in any way required, as shown in Figure 7.12. Typical analyses include cutfill (subtracting a
DEM from one time from a DEM from another) that indicates spatial patterns of erosion and
deposition.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
113
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 7.11 Example of the text output files generated by CAESAR
Figure 7.12 Water depths draped over the DEM for the River Teifi example, created in ARCSCENE
By checking a box on the opening screen, CAESAR can also output the water and sediment discharges
into a text file at a specified interval (e.g. hourly or daily). This allows the user to plot up the data and
record the modelled hydrograph, as well as the sediment discharges for all the nine grainsizes. This
data is output in the same format as the data used as input files when CAESAR runs in reach mode, as
shown in Figure 7.6. This allows CAESAR to be set up in a unique way, where a catchment CAESAR
run can generate water and sediment discharges for a river catchment at a coarse resolution (e.g. 50m).
These data can be saved as an output file, then be fed into a higher resolution reach mode CAESAR
model to simulate (for example) a study reach, or area of interest. This is illustrated in Figure 7.13,
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
114
Accounting for Sediment in Rivers
FRMRC Research Report UR9
where there is a schematic of how 6 CAESAR models were set up to feed into one reach model for the
Upper Severn near Caersws. The advantage of this method is that the modelling task is split up or
parallelised into 6 separate runs that can be run (partly) simultaneously on separate machines, thus
decreasing run time.
Figure 7.13 Example of multiple catchment CAESAR set up, to simulate two reaches of the
River Severn, U.K (after Van de Wiel et al., In Press)
7.6
LIMITATIONS AND UNCERTAINTY
CAESAR is designed as an experimental tool for scientific research and hypothesis testing. It is based
on known physical relationships (e.g. flow approximations and sediment transport rules) but the
accuracy of these, and how they are effected by their application in a 2-D model is largely unknown.
This is compounded by difficulties in validating the results of CAESAR. Some components can be
relatively easily validated. The flow model, for example, can be compared against measured flood
levels or against inundation extents from conventional 1-D flow models such as HEC or iSIS. Such a
comparison is shown in Figure 7.3, where inundation areas predicted by CAESAR are projected over
those from iSIS. Figure 7.14 shows there are some significant differences, but also more detail as
provided by a 2-D model. Some areas (A) show where CAESAR has channelled more flow into the
northerly tributary, whereas in iSIS the channels are modelled as separate streams, where there is no
interchange of flow between channels. However, it must be noted that CAESAR is not designed to be
a flood inundation modelling package, the correct routing of flood waters is only required in order to
carry out fluvial erosion, deposition and subsequent changes in channel and floodplain morphology.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
115
Accounting for Sediment in Rivers
FRMRC Research Report UR9
A
Figure 7.14 Comparison of iSIS modelled inundation extents and CAESAR modelled depths
and inundation areas
However, whilst flow depths and inundation areas can readily be compared between models and with
field data, suitable data for erosion and deposition are more difficult to find. There are few continuous
bedload data sets, with most measurements being averaged over weeks, months or years. One example
is where sediment yields from CAESAR runs from the Waitaki braided river, New Zealand compare
favourably to those measured over a 20 year period (Coulthard et al., In press).
One possibility is to compare results to data that show changes in planform. For example, sequences of
aerial photographs or historical maps can be used to show meander development (e.g. Hooke, 1984;
Braga and Gervasoni, 1989). However, the use of planform data can be limited by the frequency of
aerial photographs or map editions – with each only representing a single moment in time, which may
not always be representative. Repeat topographic surveys could also provide an ideal method for
measuring morphological changes over time, but these can be labour intensive, and expensive in the
case of remotely sensed (e.g. LiDAR) data. Similarly, they can also be restricted by the frequency of
survey. However, even with detailed topographic surveys it is difficult to be precise about what the
initial boundary conditions were. In particular the initial bedload and sub-surface grain size
distribution can have a significant effect on the levels of incision and subsequent deposition that may
occur.
An alternative for validation is to look at the longer, more ‘blurred’ sedimentological record.
Modelling longer time scales (100’s to 1000’s of years), Coulthard and Macklin (2001) and Coulthard
et al. (2005) compared CAESAR model outputs to histograms of 14C dated flood units from the UK.
This is a good example of retro-validation; they simulated the past 9000 years and compared the
results to the present day stratigraphic record. This technique is ideal for longer-term studies, but is
hampered by the temporal and spatial resolution of dated flood units. 14C can at best date within 50
year margins, and spatially, each date only represents one point within a catchment (Coulthard et al.,
2005). Coulthard and Macklin (2003) also use a comparison between modelled and field measured
heavy metal contaminated sediment patterns as a method for model validation.
However, a limitation of fluvial models in general, which is certainly applicable to CAESAR, is that
for natural environments their heterogeneity presents a major problem. For example, changes in bed
roughness, differences in sediment inputs to a reach, changes in water inputs, fluctuations in climate
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
116
Accounting for Sediment in Rivers
FRMRC Research Report UR9
and sediment delivery, vegetation changes can all influence the behaviour of a river system. These
variations and uncertainties are hard, if not impossible, to measure in order to drive a model, or indeed
to replicate within a numerical model.
7.7
DISCUSSION AND CONCLUSIONS
CAESAR occupies somewhat of a unique space in fluvial modelling. It has the capability to simulate
timescales that are useful to engineers, researchers of fluvial systems (1–100 years) and to simulate
flooding and morphological change of pertinent spatial scales (from 2 km reaches to 400+ km2
catchments). As such, it has great potential to be used for answering research questions and practical
applications for river management. The ethos of CAESAR development is very much one of openness.
The code is freely available, and limited support is offered via a discussion board that can be found on
the CAESAR web site. The use of the model for a diverse range of applications is encouraged, and
CAESAR is presently forming significant parts of research projects and programs for researchers
around the world.
However, this also means that the development of the code can be somewhat ad-hoc and sporadic, and
that there is not the level of support available for it as there might be for a commercial package. But, it
is free! Furthermore, because of its development principally for research hypothesis testing, it has not
been extensively validated. The fluvial components appear to be working well, and other models of its
type (e.g. CRS by Thomas and Nicholas, 2002) have been shown to perform as well as if not better
than more conventional modelling packages. But the erosion and deposition components are especially
difficult to validate, as we have such limited field data against which to compare the model runs.
Furthermore, considerable uncertainty surrounds the sediment transport equations used to drive the
model, they are far from ideal, and no generic equation has yet been developed (for further discussion
see Coulthard et al., (in Press)).
At present it would therefore be unwise to base too much on the absolute figures (e.g. sediment yields)
generated from CAESAR. However, it is a valid tool to look at relative changes, to see for example
whether increased flood magnitudes would cause more erosion or not. Indeed, experience of using
CAESAR has shown that its greatest robustness is in simulating general patterns of erosion and
deposition. For example it can consistently demonstrate which zones or areas of a river are more likely
to be eroding and incising or depositing and possibly unstable. It should also be noted that even if
results from the model are difficult to validate, it does give us a numerical way of assessing river
systems, without the taint of human interpretation. One of the difficulties of geomorphic assessment of
river corridors is that repeat surveys may be carried out by a different person, who may pay attention
to different details. The model can provide a synoptic viewpoint. In summary, CAESAR may contain
uncertainties, but offers capabilities that few other techniques have.
There are numerous potential future applications for CAESAR, and its cellular framework makes it
especially suitable for the inclusion of biological parameters, for example to look at fish habitats. A
further application has been the inclusion of vegetation growth models, to look at the roles of
vegetation stabilising banks and bar surfaces (Coulthard et al., In Press). It has not been extensively
used for engineering purposes, and we feel that this is certainly an area in which CAESAR could
usefully be applied and developed.
7.8
ACKNOWLEDGEMENTS
The development of CAESAR has been substantially funded by NERC. We would also like to thank
Dr Gez Foster, Jeremy Walsh, Joe Wheaton and Clare Cox who have all contributed to the
development of the CAESAR model. CAESAR is freely available for download at http://www.
coulthard.org.uk
7.9
REFERENCES
Bates, P.D. and De Roo, A.P.J., 2000. A simple raster-based model for flood inundation simulation.
Journal of Hydrology, 236: 54-77.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
117
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Braga, G. and Gervasoni, S., 1989. Evolution of the Po River: an example of the application of historic
maps. In: G.E. Petts (Editor), Historical Change of Large Alluvial Rivers: Western Europe. Wiley,
Chichester, pp. 113 – 126.
Beven, K.J. and Kirkby, M.J. 1979. A physically based variable contributing-area model of catchment
hydrology. Hydrological Science Bulletin, 24 (1), 43-69.
Brewer, P.A., Coulthard, TJ, Davies, J.R., Foster, G.C., Johnstone, E., Jones, A.F., Macklin, M.G, &
Morgan, C.G. (2005), Flooding related research in Wales: some recent developments. In Basset, M.G.,
Deisler, V.K. and Nichol, D. (eds), Urban Geology in Wales: 2. National museum of Wales, 229-238.
Coulthard, T.J. 2001. Landscape evolution models: a software review. Hydrological Processes, 15:
165-173.
Coulthard, T.J., Hicks, D.M. & Van De Wiel, M.J. (In Press) Cellular modelling of river catchments
and reaches: Advantages, limitations and prospects. Geomorphology
Coulthard, T.J., Kirkby, M.J., Macklin, M.G. 1998. Non-linearity and spatial resolution in a cellular
automaton model of a small upland basin. Hydrology and Earth System Sciences 2, 257-264.
Coulthard, T.J., Kirkby, M.J. and Macklin, M.G. 1999. Modelling the impacts of Holocene
environmental change in an upland river catchment, using a cellular automaton approach. In: A.G.
Brown and T.A. Quine (Editors), Fluvial Processes and Environmental Change. John Wiley,
Chichester, UK, pp. 31-46.
Coulthard, T.J., Kirkby, M.J. and Macklin, M.G. 2000. Modelling geomorphic response to
environmental change in an upland catchment. Hydrological Processes, 14: 2031-2045.
Coulthard, T.J., Lewin, J. and Macklin, M.G. 2005. Modelling differential catchment response to
environmental change. Geomorphology, 69: 222-241.
Coulthard, T.J. and Macklin, M.G. 2001. How sensitive are river systems to climate and land-use
changes? A model-based evaluation. Journal of Quaternary Science, 16(4): 347-351.
Coulthard, T.J. and Macklin, M.G. 2003. Modeling long-term contamination in river systems from
historical metal mining. Geology, 31(5): 451-454.
Coulthard, T.J., Macklin, M.G. and Kirkby, M.J. 1996. A cellular automaton fluvial and slope model
of landscape evolution. In: R.J. Abrahart (Editor), Proceedings of the 1st International Conference on
Geocomputation, University of Leeds, 17-19 September 1996, Leeds, UK, pp. 168-185.
Coulthard, T.J., Macklin, M.G. and Kirkby, M.J. 2002. A cellular model of Holocene upland river
basin and alluvial fan evolution. Earth Surface Processes and Landforms, 27: 269-288.
Coulthard, T.J. and Van De Wiel, M.J. 2006. A cellular model of river meandering. Earth Surface
Processes and Landforms, 31(1).
Cox, C., Brasington, J. and Richards, K. 2005. Predicting reach scale flow patterns using reduced
complexity cellular schemes. EGU General Assembly, Vol 7. EGU05-A-01646
Hoey, T and Ferguson, R. 1994. Numerical simulation of downstream fining by selective transport in
gravel bed rivers: Model development and illustration . Water resources research, 30, 7, 2251-2260.
Hooke, J.M. 1984. Changes in river meanders: a review of techniques and results of analysis. Progress
in Physical Geography, 8: 473-508.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
118
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Murray, A.B. and Paola, C. 1994. A cellular model of braided rivers. Nature, 371: 54-57.
Murray, A.B. and Paola, C. 1997. Properties of a cellular braided stream model. Earth Surface
Processes and Landforms, 22: 1001-1025.
Murray, A.B. and Paola, C. 2003. Modelling the effect of vegetation on channel pattern in bedload
rivers. Earth Surface Processes and Landforms, 28: 131-143.
Nicholas, A.P. 2005. Cellular modelling in fluvial geomorphology. Earth Surface Processes and
Landforms, 30: 645-649.
Thomas, R. and Nicholas, A.P. 2002. Simulation of braided flow using a new cellular routing scheme.
Geomorphology, 43: 179-195.
Van De Wiel, M.J., Coulthard, T.J., Macklin, M.G. and Lewin, J. In Press. Embedding reach-scale
fluvial dynamics within the CAESAR cellular automaton landscape evolution model. Geomorphology
Wilcock, P.R. and Crowe, J.C. 2003. Surface-based transport model for mixed-size sediment. Journal
of Hydraulic Engineering, ASCE, 129(2): 120-128.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
119
Accounting for Sediment in Rivers
FRMRC Research Report UR9
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
120
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Appendix
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
121
Accounting for Sediment in Rivers
FRMRC Research Report UR9
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
122
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Appendix A
REAS Version 5.5 User Manual
Operating System Requirements
REAS runs in MS Excel 2003 with VBA macros driving the computations. MS Excel
2003 is therefore required.
REAS Operation
Data Input Sheets
There are five data manipulation sheets in Excel REAS:
1)
2)
3)
4)
5)
Calculations
Input Data
Sediment
Options
Output
Input Flow Data: Calculations Worksheet
Go to sheet ‘Calculations’
In cell D2 enter the total drainage basin area upstream of the gauging station from
which flows are to be generated for the river reaches in the REAS model. This
drainage basin area could be that for a gauge on the same river as the study reaches
or can be that for a gauge on another river which is to be used as a surrogate. Note
that all discharges in REAS are currently scaled to a given REAS reach based solely
on drainage basin area.
Flow data fed into the model must, at this time, be a .txt file located in the same folder
as the REAS model. Flow data should be a single column of numbers (15 min or
mean daily data) with no headers.
Press the Calc Flows button:
A box will appear asking the user whether they wish the model to Read New Flow
Data. Click ‘Yes’.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
123
Accounting for Sediment in Rivers
FRMRC Research Report UR9
A box will then appear with the header Data Input and a request underneath to
‘Enter filename’. In the text box type the name of the .txt file. The default is
allflow.txt. Click OK. The program will then read in the .txt data file. This may take
some time depending on the length of the flow record. The egg timer cursor symbol
appears until processing is complete.
Once processing is complete an alert box will appear which indicates the number of
samples in the allflow.txt file and that processing has completed. Click OK.
A text box will then appear with the header Classes. In the input box enter the
number of flow-frequency classes you wish to have computed. The default is 35.
Click OK.
A text box will then appear with the header Class Intervals. Here a choice must be
made between the use of arithmetic classes (enter 0) or logarithmic classes (enter 1).
Enter a value and click OK.
A text box will then appear with the header Smoothing. This gives the user the
option to smooth the classes. Enter a smoothing value (1, 3, 5, 7 or 9). The default is
1. Click OK.
A text box will then appear with the header Min Discharge for PDF (Probability
Density Function). The default discharge is the minimum. Enter a discharge value or
‘m’ for minimum. m is the default. Click OK.
The computations will then be made and the results entered in Columns A and B.
Column A has the header Q and displays the discharge for each class. Column B has
the header F(%) and displays the frequency of each discharge class.
Through the computation of the flow-frequency classes REAS generates two
temporary files – CPDF.tmp and PDF.tmp. These files should appear in the same
folder as the REAS Excel spreadsheet. These files can be ignored except if the user
clicks on ‘No’ in the Flow Data pop-up box. If No is selected the computation process
will begin by reading the default PDF.tmp file located in the same folder as Excel
REAS.
Figure 1 shows a screen shot of the Calculations worksheet
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
124
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 1. Calculations worksheet showing discharge and frequency columns.
Input Geometry and Roughness Data: Input Data Worksheet
In this sheet channel geometry data is entered. Enter data in the following manner.
Entries must be precisely in the cell format described here:
For each reach:
Cell:
A: header River. Enter river name.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
125
Accounting for Sediment in Rivers
FRMRC Research Report UR9
B: header Reach. This refers to a number given to the entire REAS model for the
channel in question. This is just a reference number for the user so any value can be
entered (must be numeric).
C: header Section. This refers to the number of a REAS reach. Reaches should be
numbered 1, 2, 3 etc, from top to bottom. 1 must refer to the most upstream reach.
D: header Distance. Under this header the reach length of a REAS reach should be
entered.
E: header Station (m). Under this header cross-section geometries point should be
entered (x values) from left to right. Values can be negative.
F: header Elevation (m). Under this header elevations corresponding to each station
should be entered.
G: header LB Station (m): Under this header a Station value must be entered which
corresponds to the Left Bank top marker. i.e. the location of the channel – floodplain
interface. Values must correspond to a value under column E.
H: header LB Station (m): Under this header a Station value must be entered which
corresponds to the Right Bank top marker. i.e. the location of the channel – floodplain
interface. Values must correspond to a value under column E.
I: header LFP n: Under this column a Manning’s n value should be entered for the left
floodplain. This column can be left blank if the user wishes to use the default n value
which can be selected in the Options worksheet.
J: header Channel n: Under this column a Manning’s n value should be entered for
the main channel. This column can be left blank if the user wishes to use the default
n value which can be selected in the Options worksheet.
K: header RFP n: Under this column a Manning’s n value should be entered for the
right floodplain. This column can be left blank if the user wishes to use the default n
value which can be selected in the Options worksheet.
L: header S. Under this header a slope in mm-1 should be entered for the REAS
reach (either bedslope or energy gradient).
M: header dx (m): Under this header a representative bedmaterial grainsize for the
reach should be entered. A value for the D50 would be appropriate. Note that the
value should be entered in metres not millimetres.
N: header DB Area (km2). Under this header a drainage basin area for the reach in
question should be entered. The drainage basin area should correspond to the
downstream end of the reach.
Data Entry Format:
•
All values should be numeric except Column A.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
126
Accounting for Sediment in Rivers
FRMRC Research Report UR9
•
•
•
All values should be entered in the first row of each set of reach data (except
that Station and Elevation will continue below the first row).
Subsequent reaches should be entered with the data in the row which follows
immediately after the last row of the geometry data for the previous reach.
All fields must contain a value.
Figure 2 shows a screen shot of the Input Data worksheet.
Figure 2. Input Data worksheet.
Input Sediment Data: Sediment Worksheet
If the user has grainsize distribution data for each reach this can be entered in the
worksheet Sediment.
In this sheet each grainsize class is shown in Column A (phi scale) and B (grain
size mm (upper bounds)) (Wentworth Scale). In Columns C onward the percentage
finer values for each grainsize should be entered for each reach from 1 to n, i.e. most
upstream reach to downstream, and from 100% finer to 0% finer. Blank cells above
100% should be filled with 100, and blank cells below 0% should be filled with 0.
In row 39 under each grainsize distribution column a calculated D50 value is displayed
that can be used in the Input Data worksheet. Figure 3 shows a screen shot of the
Sediment input worksheet.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
127
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Figure 3. Sediment data input worksheet.
Options: Options worksheet
In this worksheet several REAS options can be chosen from:
•
•
•
•
•
Variable grainsize: Cell B1 type either ‘y’ to use the full grainsize distribution
entered in the sheet Sediment, or ‘n’ to use the single grainsize value entered
in the Input Data worksheet
Channel width Adjustment:: In cell B3 enter ‘y’ or ‘n’ to select whether a
channel width adjustment factor is to be used. For Widening select a value
corresponding to Severe, Moderate or Minor from cells B5 to B7. For
Narrowing select a value corresponding to Severe, Moderate or Minor from
cells B9 to B11.
Import RAS data: A HEC RAS geometry file (.g0) representing the entire river
reach can be imported by selecting typing ‘y’ into cell B14 and typing the
geometry filename in cell B15 (including extension). Then click on the Import
button.
Import ISIS data: An ISIS geometry file representing the entire river reach can
be imported by selecting typing ‘y’ into cell B17 and typing the geometry
filename in cell B18 (including extension). Then click on the Import button.
Selection of Geometry input data sheet: It is possible for the user to set up
as many geometry data worksheets as they wish for a given river as long as
the geometry data contains the same number of reaches as represented in the
Sediment and Calculations worksheet. In cell B22 enter a sheet name. This
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
128
Accounting for Sediment in Rivers
FRMRC Research Report UR9
name must be entered exactly as it appears on the data input worksheet tab.
NOTE: A GEOMETRY DATA INPUT SHEET NAME MUST BE SPECIFIED.
•
•
•
Selection of Panels: In cell B24 the number of panels into which the
cross-section is divided in order to calculate power using the REAS
panelisation approach should be entered. The default is 100 panels.
Floodplain Slope: In cell F1 the user can select how REAS handles flows
which are deeper than in in-channel depth. If ‘1’ is typed REAS will extent
the channel vertically with a ‘Glass Wall’ above the channel left and right
hand bank markers. Options 2 and 3 are currently not active.
Default Manning’s n values: Default Floodplain and Channel Manning’s
n values for each REAS reach can be entered in cells F3 and F4
respectively.
A screenshot of the Options worksheet is shown in Figure 4.
Figure 4. Options worksheet
Output Results: Output Worksheet
To perform the REAS Specific Power calculations click on the Calc Watts
button in the Output worksheet. In Column A which has the header Reach,
reach numbers are given from 1 to n, i.e. from most upstream reach to
downstream. In column C which has the header KWh the calculated Specific
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
129
Accounting for Sediment in Rivers
FRMRC Research Report UR9
Stream Power in KWh is displayed for each reach. In column D which has the
header Local Balance the power differential for each reach is displayed. Row
2 should have a value of 0 entered as a power differential cannot be computed
for this reach.
When the Calc Watts button is pressed a pop-up box appears. Here a value
of’ ‘0’ can be entered if the original number of panels are to be used (that is no
panels, the entire cross-sectional area is used to calculated total power and
critical power) or ‘1’ if the REAS method of panelisation is used. The number
of panels REAS uses are those specified in the Options worksheet. Enter a
value and click OK.
When REAS has completed computation a pop-up box will appear stating
‘Processing Complete’. Note that in the Calculations worksheet in Columns F
onwards the Specific Powers for each discharge listed in column A are
presented in the same row for each reach (1, 2 etc, from left to right).
The local balance output can now be plotted from Column D in the Output
sheet.
Figure 5 shows a screenshot of the Output worksheet.
Figure 5. Output worksheet showing local balance graph.
UR9_accounting_for_sediment_in_rivers_WP8_1_v1_0.doc
08/08/06
130