streambank erosion hazard mapping:

concepts, methodology and application on

the venoge river (switzerland)

THÈSE NO 3523 (2006)

PRÉSENTÉE le 21 juin 2006
à la faculté ENVIRONNEMENT NATUREL, ARCHITECTURAL ET CONSTRUIT
Laboratoire d'hydrologie et aménagements
SECTION DES SCIENCES ET INGéNIERIE DE L'ENVIRONNEMENT

ÉCOLE POLYTECHNIQUE FÉDÉRALE DE LAUSANNE

POUR L'OBTENTION DU GRADE DE DOCTEUR ÈS SCIENCES

PAR

John Raymond BECK

B.Sc. in Engineering, University of lowa, Etats-Unis

et de nationalité américaine

acceptée sur proposition du jury:

Prof. A. Buttler, président du jury

Prof. A. Mermoud, Prof. A. Musy, directeurs de thèse
Prof. M. Altinakar, rapporteur
Dr A. Petrascheck, rapporteur
Prof. A. Schleiss, rapporteur

Lausanne, EPFL
2006
 Abstract

Abstract
Streambank erosion hazard mapping has received much less attention than flood inundation mapping in
the past due to the complexity of the task as well as bank protection works that have reduced bank erosion
and unfortunately, the ecological functions of our watercourses at the same time. Damages due to
streambank erosion in some flooding contexts are greater than the flood water damages (Loat and
Petrasheck, 1997). For these reasons, streambank erosion hazard mapping should be an integral part of
flood hazard mapping and methods must be developed to accomplish it. This research proposes a
methodology for mapping streambank erosion hazards based on the directives of the Swiss Federal
Office for Water and Geology (now within the Swiss Federal Office for the Environment). It permits the
calculation of bank failure widths and their probability as opposed to future channel migration paths.
This research also investigates the input data necessary for streambank erosion hazard modeling.
Geomorphological mapping must be the first step to streambank erosion hazard mapping as it permits
the identification of the sediment movements in the catchment. After this step, modeling of streambank
erosion can be undertaken. Geofluvial models that combine hydraulic sediment transport and
geotechnical modeling are well suited for streambank erosion modeling. The model CCHE1D is such a
model and was adapted for the calculation of the streambank erosion hazard on an 8 kilometer reach of
the Lower Venoge River, Switzerland.
CCHE1D performs one dimensional hydraulic calculations. A shear stress correction function based on
channel curvature distributes mean boundary shear stress appropriately to outer and inner bank toes and
the phase lag of maximum toe shear stress compared to the apex of the bend curvature is ensured by a
convolution of upstream shear stresses. Tension cracking was added to the slab failure algorithm due to
its significant effect on bank failure widths. After a bank failure, the cross section shape does not change
which allows the flow conditions to remain the same and in turn allows the probability of failure for the
modeled bank profile to be evaluated.
To gain a better understanding of streambank erosion on the Lower Venoge River, detailed erosion and
flow depth monitoring were done on two 1 kilometer river reaches from November 2003 through
September 2005. These measurements showed the mass failures to be mainly soil falls and cantilever
failures. Measured bank erosion was linearly related to the product of maximum discharge and flood
volume. Bank and bed sediment data were also collected for the study reach.
Eighty-two cross sections were surveyed in 2004 in the 8 kilometer study reach. A new cross section tool
was developed to properly reproduce scour holes in bends. It calculates transverse position and distance,
graphs the cross section to allow identification of bank definition points, linearly interpolates, calibrates
bed topography parameters based on surveyed cross sections, and interpolates with respect to channel
curvature.
Hydrological modeling allowed for the generation of input hydrographs for the period January 1979-
February 2005. This information was combined with historically based low probability floods to
construct three 300 year discharge series.
Flow and erosion measurements allowed for the calibration of roughness and critical shear stress
parameters, respectively, in the CCHE1D model. Where detailed erosion measurements were not
available, past channel migration served as a guide for estimating the critical shear stress. Calibrated
critical shear stress was poorly correlated with measured bank properties indicating the necessity of
Abstract

measuring critical shear stress. A regression equation involving the percentage of fine sand - large silt
and the fraction of non-vegetated bank explained 75% of the variation.
The three 300 year discharge series were simulated with CCHE1D. Bank failures series were output for
each of the banks of the 1149 computational nodes in the study reach. Empirical frequency was used to
determine the bank failure width of a given probability. These bank failure widths and their probabilities
were used to calculate a streambank erosion danger. To further qualify this danger, it was mapped with
a bar proportional to the mean annual simulated erosion rate. The extreme failure width for the entire
reach was determined by multiplying the maximum simulated failure width by a safety factor. The
erosion hazard in straight reaches is too high showing that the shear stress reduction due to the highly
vegetated banks needs to be taken into account better in the shear stress correction function.
This research has demonstrated the feasibility of streambank erosion hazard mapping, although the
quantity of input data necessary is prohibitive. Data acquisition methods must be researched and
improved to reduce costs, and research must be continued to improve understanding of bank failure
processes to be included in geofluvial models.
Keywords: bank erosion, bank failure, bank stability, shear stress, critical shear stress, erosion pin, bank
profiling, cross section interpolation, geofluvial modeling, hydrologic modeling
 Résumé

Résumé
La cartographie des dangers liés à l'érosion des berges d'un cours d'eau a reçu moins d'attention dans le
passé que celle des dangers liés aux inondations. Cette différence est due à la difficulté de la tâche, et à
la présence d'ouvrages de protection, réduisant l'érosion, mais malheureusement aussi les fonctions
écologiques de nos cours d'eau. Dans certains contextes, les dommages dus à l'érosion des berges sont
cependant plus importants que ceux causés par l'eau (Loat et Petrasheck, 1997). De ce fait, la
cartographie des dangers liés à l'érosion des berges devrait faire partie intégrante de la cartographie des
dangers liés aux crues, et des méthodes devraient être développées pour ce faire. Le présent travail de
recherche propose une méthodologie pour la cartographie des dangers liés à l'érosion des berges basée
sur les directives de l'Office Fédérale Suisse de l'Eau et de la Géologie (maintenant intégré à l'Office
Fédérale de l'Environnement). Cette méthodologie permet le calcul des largeurs et probabilités
d'effondrement des berges, et non la position future du cours d'eau. Le présent travail de recherche fait
par ailleurs une revue des données nécessaires pour la modélisation de l'érosion des berges.
La cartographie géomorphologique doit être la première étape dans la cartographie des dangers liés à
l'érosion des berges car elle permet l'identification des mouvements de sédiment dans le bassin versant.
La modélisation de l'érosion des berges peut être entreprise ensuite. Les modèles géo-fluviaux qui
combinent des modèles hydrauliques de transport de sédiments et des modèles géotechniques sont
particulièrement adaptés pour ce faire. Le modèle CCHE1D est un modèle de ce type. Il a été adapté dans
le cadre de cette étude pour le calcul des dangers liés à l'érosion des berges, sur un tronçon de rivière de
8 kilomètres de la Venoge, Suisse.
CCHE1D est un modèle hydraulique unidimensionnel. Basée sur la courbure du cours d'eau, une
fonction de correction des contraintes de cisaillement distribue celles-ci entre les pieds des berges
intérieures et extérieures. Le décalage de la contrainte maximale de cisaillement au pied par rapport à la
courbure maximale est assuré par une convolution des contraintes de cisaillement à l'amont. La fissure
de tension est intégrée à l'algorithme d'effondrement en plan du fait de son effet significatif sur la largeur
d'effondrement. Dans l'approche proposée, la géométrie du profil en travers n'est pas modifiée après un
effondrement de berge. Ceci permet que les conditions d'écoulement restent inchangées et ceci permet
aussi l'évaluation de la probabilité d'effondrement du profil modélisé.
Pour une meilleure compréhension du phénomène d'érosion des berges de la Venoge, des mesures de
hauteurs d'eau et d'érosion ont été effectuées sur deux tronçons de 1 kilomètre, de novembre 2003 à
septembre 2005. Ces mesures ont démontré que les effondrements des berges correspondent
principalement à des chutes de sol et des effondrements en levier. L'érosion mesurée est corrélée
linéairement au produit du débit maximum et du volume de crue. Des données sur les sédiments du lit et
des berges ont également été collectées pour le tronçon d'étude.
Quatre-vingt-deux profils en travers ont été levés en 2004 dans le tronçon d'étude de 8 kilomètres. Un
nouvel outil de traitement des profils en travers a été développé pour reproduire les affouillements du
fond dans les méandres. Cet outil calcule position et distance transversale, dessine le profil en travers
pour l'identification des points de définition des berges, interpole linéairement, cale les paramètres de la
topographie du lit, et interpole en respectant la courbure du cours d'eau.
Une modélisation hydrologique a permis la génération des hydrogrammes d'entrées pour la période de
janvier 1979 à février 2005. Cette information a été combinée avec des crues rares, construites à partir
des crues observées, pour préparer trois séries temporelles de débit d'une durée de 300 ans chacune.
Résumé

Des mesures de hauteur d'eau et d'érosion ont permis le calage de la rugosité et des contraintes de
cisaillement critiques pour le modèle CCHE1D. Dans les zones n'ayant pas bénéficié de mesures
d'érosion, les divagations historiques du cours d'eau ont guidé l'estimation de la contrainte de
cisaillement critique. Les valeurs calées des contraintes de cisaillement critiques ne sont pas bien
corrélées avec les propriétés des berges mesurées, ce qui indique la nécessité de mesurer la contrainte de
cisaillement. Une régression utilisant le pourcentage de sable fin et de limon grossier et la fraction de
berge non couverte de végétation a pu expliquer 75% de la variation, et peut être utilisée pour le tronçon
étudié.
Les trois séries temporelles de débit ont été simulées avec CCHE1D. Les séries d'effondrements pour
chaque berge des 1149 nœuds de calcul ont été produites par le modèle. Pour une probabilité donnée, la
fréquence empirique a été utilisée pour déterminer la largeur d'effondrement. Ces probabilités et leur
largeur d'effondrement ont été utilisées pour calculer le danger lié à l'érosion des berges. Pour qualifier
le niveau de danger, ce dernier a été cartographié avec des lignes proportionnelles à l'érosion annuelle
moyenne simulée. La largeur d'effondrement relative à un événement extrême pour le tronçon a été
déterminée en multipliant l'effondrement maximal simulé par un facteur de sécurité. Le danger d'érosion
est trop élevé dans les tronçons droits, ce qui montre que la réduction des contraintes de cisaillement due
à la végétation des berges doit être mieux prise en compte dans la fonction corrigeant les contraintes de
cisaillement.
Le présent travail de recherche a montré la possibilité de réalisation de cartes des dangers liés à l'érosion
des berges, bien que la quantité des données nécessaires y soit restrictive. Les méthodes d'acquisition des
données doivent être améliorées pour réduire les coûts, et des recherches supplémentaires doivent être
mise en œuvre pour mieux comprendre les processus d'effondrement des berges afin de les inclure
ensuite dans les modèles géo-fluviaux.
Mots-clés: érosion de berge, effondrement de berge, stabilité de berge, contrainte de cisaillement,
contrainte de cisaillement critique, fiche d'érosion, profil de berge, interpolation de profils en travers,
modélisation géo-fluviale, modélisation hydrologique
 Acknowledgements

Acknowledgements
For such a multidisciplinary fieldwork thesis, a considerable number of people were necessarily
involved. Fortunately, I have this section to thank them. I will inevitably forget someone or some aspect.
Please forgive me and know that I appreciated your help.
First of all I thank God, who has given me the opportunity to study his magnificent creation that we will
never finish to contemplate.
I want to thank my thesis directors, Professor André Musy and Professor André Mermoud, who provided
me the opportunity to start, develop, and finish my thesis! The HYDRAM laboratory was an excellent
place to work under the direction of Prof. Musy and still is under the new direction of Prof. Mermoud.
The great environment is also due to my incredible HYDRAM colleagues, who are too numerous to
invoke here, and to whom I wish to thank not only for their provocative discussions, but also for the
kindness and good spirits they brought to work.
Among my HYDRAM colleagues, I wish to thank Richard Metzger, who helped me countless times with
HydroRoute, Delphi, and Mapinfo, and who verified a portion of the draft. Jean-Marc Lance laid the
foundation for this thesis by his erosion work on the Venoge River and also helped out with the draft
verification. Benoît Hingray was always kind to discuss the hydrologic modeling with me. Anne-
Catherine Favre provided excellent advice for statistical analyses. My appreciation also goes out to
Bettina Schaefli, Cécile Picouet, and Colin Schenk for their verification of portions of the draft. I couldn't
have done it without you all!
The amount of fieldwork accomplished during this thesis wouldn't have been possible without the help
of the HYDRAM technical staff. Discussions with Alexis Amiguet, Bernard Sperandio, and Karine
Vernez-Thomas allowed for the fieldwork techniques to be transferred from paper to reality. Cold
weather and getting dirty did not discourage them from getting the measurements event after event!
Gabrielle Schwab will probably never count as many rocks as she did in the summer of 2004. Bernard
Sperandio also helped out with several figures in this document. Thanks to you all and bravo!
This thesis was given a boost by Professor Mustafa Altinakar and Professor Sam Wang when they
provided me the opportunity to work with the CCHE1D code. Dalmo Vieira and Weiming Wu
exchanged countless emails with me during my code modifications and debugging, and Dalmo also
verified a portion of my draft. I met Andrew Simon and Eddy Langendoen during my visit to Oxford,
Mississippi. Their advice and input were greatly appreciated.
Closer to home, Pierre-Yves Gillieron, Ivan Spassov, and Hervé Gontran provided valuable advice and
help on the survey work. Christophe Bonnard and Gilbert Steinmann gave me advice on geotechnical
aspects and fieldwork. Pascal Turberg contributed with geological information. Discussions with Koen
Blanckaert concerning hydraulic theory were insightful. Thank you all for taking the time!
I am grateful to the jury members for the pain they took to judge my work and for their remarks to
improve this document. I appreciated the fruitful discussions with Professor Anton Schleiss and Dr.
Armin Petrascheck throughout the thesis.
The Interreg IIIb NWE project "Strategies and Actions/Implementations for Flood Emergency Risk
Management" partially financed this thesis. I am grateful to the project designers who had the foresight
to see the necessity of including this work in the SAFER project and to the project partners for their
Acknowledgements

fruitful discussions on streambank erosion. Thanks goes out to the SESA and SIT of the Vaud Canton,
the national hydrological service (SHN) of the Swiss Federal Office for the Environment, the Swiss
Federal Statistical Office, and MeteoSwiss for the data they provided.
Lastly, but definitely not least, I thank my family. Damaris, thank you from the bottom of my heart for
the immense support you gave me during these last few years. Lucas, Daniel, and Samuel, you guys were
patient with your Daddy when he was working weekends, and you were fantastic when you made him
forget about work at times! To the rest of my family, thank you for your prayers and encouragement.
 Table of Contents i

Table of Contents

Chapter 1
Introduction .......................................................................................................... 1
1.1 Streambank erosion in the flood damage context .......................................................... 1
1.1.1 Streambank erosion damages ...................................................................................... 1
1.1.2 Streambank erosion mapping and prevention ............................................................. 3

1.2 Thesis objectives ........................................................................................................... 5

1.2.1 Modeling streambank erosion ..................................................................................... 5
1.2.2 Acquisition of streambank erosion data ...................................................................... 6

1.3 Report organization ....................................................................................................... 6

Chapter 2
Streambank Erosion Hazard Assessment and Mapping Methodology ................ 9
2.1 Introduction ................................................................................................................... 9

2.2 Streambank erosion processes ....................................................................................... 9

2.2.1 Catchment processes ................................................................................................. 10
2.2.2 Fluvial processes ....................................................................................................... 12
2.2.3 Bank processes .......................................................................................................... 15
2.2.3.1 Bank erosion......................................................................................................... 15
2.2.3.2 Bank failure .......................................................................................................... 17

2.3 Streambank erosion assessment models ...................................................................... 20

2.3.1 Engineering analysis and conceptual modeling ........................................................ 21
2.3.2 Geofluvial .................................................................................................................. 27

2.4 Proposed streambank erosion hazard methodology .................................................... 29

2.4.1 Streambank erosion hazard assessment and mapping ............................................... 30
2.4.2 CCHE1D model modifications ................................................................................. 33

Chapter 3
The Venoge River............................................................................................... 37
3.1 Introduction ................................................................................................................. 37

3.2 Location of the Venoge catchment .............................................................................. 37

3.3 Geology ....................................................................................................................... 38

3.4 Soil and land use .......................................................................................................... 41

3.5 Topography and the hydrographic network ................................................................ 42

ii Table of Contents

3.5.1 Catchment topography .............................................................................................. 43

3.5.2 Subcatchment limits and stream network ................................................................. 44

3.6 Climate and hydrologic regime ................................................................................... 46

3.7 Channel morphology ................................................................................................... 47

3.7.1 Channel type ............................................................................................................. 48
3.7.2 Planform .................................................................................................................... 50
3.7.3 Longitudinal profile .................................................................................................. 53
3.7.4 Cross section characteristics ..................................................................................... 54

Chapter 4
Venoge River Data for Streambank Erosion Modeling...................................... 57
4.1 Introduction ................................................................................................................. 57

4.2 Experimental setup ...................................................................................................... 57

4.2.1 Erosion monitoring ................................................................................................... 59
4.2.1.1 Possible methods .................................................................................................. 59
4.2.1.2 Chosen methods ................................................................................................... 62
4.2.2 Flow monitoring ........................................................................................................ 64
4.2.3 Measured values ........................................................................................................ 64
4.2.3.1 High flows and stage recorder values................................................................... 64
4.2.3.2 PEEP..................................................................................................................... 67
4.2.3.3 Erosion pin and bank profiler............................................................................... 69

4.3 Historical and remotely sensed channel data ............................................................... 76

4.3.1 Available data ........................................................................................................... 76
4.3.1.1 Historical maps..................................................................................................... 76
4.3.1.2 Remotely sensed data ........................................................................................... 76
4.3.2 Derived information .................................................................................................. 77
4.3.2.1 Lower Venoge River centerline ........................................................................... 77
4.3.2.2 Channel migration and erosion estimates............................................................. 78

4.4 Channel topography .................................................................................................... 79

4.4.1 Cross section data ...................................................................................................... 80
4.4.2 Interpolation in meanders .......................................................................................... 85
4.4.2.1 Development ........................................................................................................ 85
4.4.2.2 Procedure.............................................................................................................. 87
4.4.2.3 Results .................................................................................................................. 88
4.4.3 LIDAR and river topography .................................................................................... 91
4.4.3.1 Linear streamwise interpolation ........................................................................... 92
4.4.3.2 Creation of a DTM with streamwise interpolated cross section points................ 94
4.4.3.3 Results .................................................................................................................. 95
4.4.3.4 Possible future developments............................................................................... 95

4.5 Bed ............................................................................................................................... 97

4.5.1 Necessary input data ................................................................................................. 97
4.5.2 Grain size distribution ............................................................................................... 97
4.5.2.1 Sampling techniques............................................................................................. 97
 Table of Contents iii

4.5.2.2 Analysis techniques .............................................................................................. 99

4.5.2.3 Equivalence of sampling methods........................................................................ 99
4.5.2.4 Fehr conversion of pavement GSD to subpavement GSD ................................... 99
4.5.2.5 GSD’s of the Lower Venoge River .................................................................... 100

4.6 Bank ........................................................................................................................... 102

4.6.1 Input data and methods ........................................................................................... 102
4.6.2 Bank classification and results ................................................................................ 106

4.7 Boundary conditions .................................................................................................. 113

4.8 Summary and conclusions ......................................................................................... 113

Chapter 5
Hydrologic Modeling for Distributed Discharge Input .................................... 117
5.1 Objectives of the hydrologic modeling ..................................................................... 117

5.2 Hydrologic model ...................................................................................................... 117

5.2.1 Elements of the HydroRoute model ........................................................................ 118
5.2.2 Snow storage ........................................................................................................... 119
5.2.3 SOCONT-Alpine .................................................................................................... 120

5.3 Subcatchment delineation .......................................................................................... 121

5.4 Data and modeling periods ........................................................................................ 123

5.4.1 Available data ......................................................................................................... 123
5.4.2 Modeling strategy ................................................................................................... 125
5.4.3 Corrected and derived data ..................................................................................... 126

5.5 Spatial interpolation of data ...................................................................................... 128

5.5.1 Precipitation ............................................................................................................ 128
5.5.2 Temperature ............................................................................................................ 130
5.5.3 Potential evapotranspiration .................................................................................... 131
5.5.4 Accuracy of data ..................................................................................................... 131

5.6 Simulations and results .............................................................................................. 132

5.6.1 Parameter estimation ............................................................................................... 132
5.6.2 Calibration ............................................................................................................... 133
5.6.3 Validation ................................................................................................................ 137

5.7 Distributed hydrographs for streambank erosion modeling ...................................... 137

5.8 Extreme hydrographs and scenarios .......................................................................... 138

5.8.1 Flood volume and duration frequency analysis ...................................................... 139
5.8.2 Venoge River historically based flood hydrographs ............................................... 140
5.8.3 Discharge scenarios ................................................................................................. 141

5.9 Summary and conclusions ......................................................................................... 144

iv Table of Contents

Chapter 6
Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping ....... 147
6.1 Introduction ............................................................................................................... 147

6.2 Model setup ............................................................................................................... 147

6.3 Calibration ................................................................................................................. 150

6.3.1 Water surface elevation ........................................................................................... 150
6.3.2 Bank erosion ........................................................................................................... 150
6.3.3 Critical shear stress estimation ................................................................................ 156
6.3.4 Sediment transport .................................................................................................. 157

6.4 Streambank erosion hazard simulations and danger mapping .................................. 158
6.4.1 Streambank erosion danger map ............................................................................. 158
6.4.2 Streambank erosion hazard results .......................................................................... 163

6.5 Streambank erosion simulation tests ......................................................................... 164

Chapter 7
Conclusions and Future Research Recommendations...................................... 167
7.1 Conclusions ............................................................................................................... 167

7.2 Future research recommendations ............................................................................. 170

References ........................................................................................................ 173

Appendix .......................................................................................................... 183

Appendix 1 Textural triangles of the Vaud plateau ............................................................ 184

Appendix 1.1 Molassic and Morainic particle distributions ............................................. 184
Appendix 1.2 Alluvial soil particle distributions .............................................................. 185

Appendix 2 Map of the Lower Venoge Study Reach ......................................................... 186

Appendix 3 Lower Venoge River and tributary GSD’s ..................................................... 188

Appendix 4 River Enrick streambank classification ........................................................... 189

Appendix 5 Textural triangles of the Lower Venoge River ............................................... 191

Appendix 6 Cumulative measured bank positions changes ................................................ 193

Appendix 7 Streambank erosion danger maps of the Venoge River .................................. 202
 List of Figures v

List of Figures
Figure 1.1 Destruction of the roadway leading to the mountain resort in Engelberg, Switzerland. The
village was accessible only by helicopter during a two week period. (from Blick, 2005) 2
Figure 1.2 Undermining of a bridge pillar (from Loat and Petrascheck, 1997) ................................ 3
Figure 1.3 Swiss Federal Office for Water and Geology (Loat and Petrascheck, 1997) directives for
mapping streambank erosion ............................................................................................ 4
Figure 1.4 Streambank intervention map (adapted from Lance and Consuegra, 1998) .................... 5
Figure 2.1 Controls of river morphology and behavior (after Schumm, 2005)................................ 10
Figure 2.2 Helical flow motion in channel bends
(adapted from Hamblin and Christiansen, 1998) ........................................................... 13
Figure 2.3 Critical shear stress as a function of SAR, soil salt concentration, and dielectric dispersion
(from Osman and Thorne, 1988, after Arulanandan et al., 1980 and reprinted with
permission from the American Society of Civil Engineers) ............................................ 16
Figure 2.4 Modes of bank mass failure (from Hey et al., 1991, in Lawler et al., 1997)................... 17
Figure 2.5 Planar failure analysis (from Wu and Vieira, 2002 after Osman and Thorne, 1988) .... 19
Figure 2.6 Erosion intensity, given as D (m), according to the Swiss FOWG guidelines ................ 21
Figure 2.7 Channel evolution model (provided courtesy of Andrew Simon).................................... 23
Figure 2.8 Channel centerline and distances used in upstream influence equations ....................... 24
Figure 2.9 Nominal erosion rate as a function of radius of curvature (adapted from Consuegra et al.,
1999)................................................................................................................................ 25
Figure 2.10 Streambank erosion hazard index based on stream power and bank resistance............ 32
Figure 2.11 Correction factor applied to outer bank toe shear stresses based on the ratio of radius of
curvature over width ....................................................................................................... 34
Figure 2.12 Channel centerline smoothing procedure ....................................................................... 35
Figure 2.13 Bank stability is no longer guaranteed
by the toe angle being lower than the failure angle ........................................................ 36
Figure 2.14 Cross sectional adjustments in the modified CCHE1D model. Toe erosion destabilizes the
bank causing bank failure. Bank geometry does not change although the bank failure
material is added to the virtual reservoir. Subsequent toe erosion empties the reservoir so
a new bank failure can occur. ......................................................................................... 36
Figure 3.1 Venoge catchment topographical limit (background map: CP500 ©1997, Swisstopo) . 38
Figure 3.2 Geologic substrates of the Vaud Canton, Switzerland (adapted from Grattier, 1980)... 39
Figure 3.3 Bedrock and quaternary sediment of the Venoge catchment
(data from Schneider, 2001)............................................................................................ 40
Figure 3.4 Venoge 1997 land-use (data from the Swiss Federal Office of Statistics (OFS)) ........... 42
Figure 3.5 Venoge digital terrain model, subcatchments, and hydrographic network. Relief data is
from the DTM25 ©1995, Swisstopo, and catchment and hydrographic network data come
from GESREAU............................................................................................................... 43
Figure 3.6 Venoge catchment hypsometric curve and slope frequency. ........................................... 44
Figure 3.7 The Venoge River and its tributaries and subcatchments.
Background map: CP500 ©1997, Swisstopo. ................................................................. 45
vi List of Figures

Figure 3.8 Longitudinal profile of the Venoge River and its important tributaries.
Only starting and ending elevations are represented.
Elevations based on the DTM25 ©1995, Swisstopo. ...................................................... 46
Figure 3.9 Seasonal variation of precipitation, potential evapotranspiration (PET calculated with the
Penman-Monteith formula), discharge, and temperature. Data is from the period 1983-
2003. ................................................................................................................................ 47
Figure 3.10 Classification of channel pattern (from Knighton,1998 based on the original work of
Schumm, 1981). (Reproduced by permission of Stanley Schumm and Edward Arnold
(Publishers)). ................................................................................................................... 48
Figure 3.11 Two of the three alluvial zones of the Lower Venoge River. To the left - Lovataire and to
the right - the islands of Bussigny. Orthophotos from 1996 EPFL student field study... 49
Figure 3.12 LIDAR DTM (DTM-AV ©2004, Swisstopo (DV043683)) showing four meander cutoffs in
the Bussigny islands alluvial zone. The 1964 orthophoto shows the 2nd and 3rd meander
cutoffs. ............................................................................................................................. 50
Figure 3.13 Meander planform geometry (adapted from Thorne, 1997) ........................................... 51
Figure 3.14 Distribution of radius of curvature for the Lower Venoge River .................................... 51
Figure 3.15 A view of the sinuosity of the Lower Venoge River. Background: DTM25 © 1995,
Swisstopo. ........................................................................................................................ 52
Figure 3.16 Analysis of mean sinuosity and its variance on the Lower Venoge River. Mean channel
width is considered to be 20 meters. ............................................................................... 53
Figure 3.17 Longitudinal profile of the Lower Venoge River............................................................. 54
Figure 3.18 Reach averaged width for four zones of the Lower Venoge River at discharges
approximating the 0.5, 0.75, 1, and 2 year return period flows...................................... 55
Figure 3.19 Reach averaged depth for four zones of the Lower Venoge River at discharges
approximating the 0.5, 0.75, 1, and 2 year return period flows...................................... 56
Figure 4.1 Erosion scars (stars) observed along the Lower Venoge River. Solid bold lines indicate
bank protection and the dashed line indicates the line approximately 5 meters above the
river. ................................................................................................................................ 58
Figure 4.2 Study reach, new cross sections, water level measurements, and bends where erosion was
monitored. Center background map: CP100 ©1997, Swisstopo. ................................... 59
Figure 4.3 PEEP sensor and its typical installation for measuring bank erosion (from the PEEP user
guide (HydroScientific, 1997)) ........................................................................................ 61
Figure 4.4 Bank profiling system used for the Lower Venoge study................................................. 62
Figure 4.5 Erosion monitoring on the Lower Venoge River. Bottom shows a small monitoring bend
(pin locations marked by arrows). Upper left shows the pin installation. Upper right shows
the bank profiler. ............................................................................................................. 63
Figure 4.6 October 26, 2004 high flow event. Rating curves for the limni17 (left) and limni11 (right)
based on the Ecublens discharges................................................................................... 65
Figure 4.7 Maximum stage recorder measurements. Top left shows recorder limni6. .................... 67
Figure 4.8 In situ calibration of the PEEP sensors during overcast conditions. Eros11 is shown to the
left and Eros17 to the right.............................................................................................. 68
Figure 4.9 The small mass failure of the 17.1 erosion monitoring point measured by the Eros17 PEEP
sensor on April 27, 2005. ................................................................................................ 69
 List of Figures vii

Figure 4.10 Evolution of the erosion measurement field book due to the complex situations
encountered. .................................................................................................................... 70
Figure 4.11 Bank profiling measurement system used (top) and recommended (bottom) ................. 70
Figure 4.12 Relationship of event erosion compared to maximum event discharge, flood volume, and
Qmax*flood volume, and bank failures compared to Qmax*flood volume .................... 71
Figure 4.13 "Carpet" of the soil-root system left after a soil fall ....................................................... 72
Figure 4.14 Observed tension cracks at bend 3.3
during the drawdown of the October 26, 2004 flood ...................................................... 72
Figure 4.15 Tree cantilever failure of bend 7.2 .................................................................................. 73
Figure 4.16 Channel centerlines determined with 1998 orthophotos and 2001/2002 LIDAR data and
2004 cross sections. Background images: SWISSIMAGE © 1998,
Swisstopo (DV012716). ................................................................................................... 78
Figure 4.17 Longitudinal profiles of the Lower Venoge River based on cross sections from 1996 and
2004. Points indicate cross section locations. ................................................................ 81
Figure 4.18 Database file cross section tool for viewing cross sections in dbf format. The tool
interpolates cross sections linearly and with respect to bend curvature. The tool can also
interpolate and smooth LIDAR topographic points to estimate channel geometry. ....... 83
Figure 4.19 Linear cross section interpolation based on master chords between cross section
definition points............................................................................................................... 84
Figure 4.20 Attribution of Manning values and ineffective flow areas to cross section points within a
GIS. Background images: SWISSIMAGE © 1998, Swisstopo (DV012716). .................. 84
Figure 4.21 Calculation of α, a, and K values based on measured cross sections ............................ 86
Figure 4.22 Ten examples of the influence of the parameter alpha
on the upstream weighted curvature ............................................................................... 87
Figure 4.23 Results of the search for optimum parameters for the bed topography model applied to the
Venoge study reach ......................................................................................................... 88
Figure 4.24 Examples of best fit calculated channel bottoms. R2 values are 0.99, 0.63, and 0.36 for the
top, middle, and bottom cross sections, respectively. ..................................................... 89
Figure 4.25 The meander interpolation method can significantly improve interpolated cross sections
in bends such as this one. ................................................................................................ 90
Figure 4.26 Comparison of some 1996 measured cross sections with linear and meander interpolated
cross sections. Improvement is based on the difference of outer bend elevation between
linear and meander cross sections. F=flat, R=right, L=left, and C=center. ................. 91
Figure 4.27 Examples from the Lower Venoge River comparing cross sections extracted from a LIDAR
derived DTM and surveyed cross sections showing the interpolation problems of widening
(top), bed raising (middle), and constriction (bottom).................................................... 92
Figure 4.28 Top - Created cross section points (black) around the channel centerline (light grey).
Dark grey circles show an interpolation line. Bottom - White circles show points that were
attributed a laser point that will be used to interpolate elevation values onto points having
a null elevation. ............................................................................................................... 93
Figure 4.29 Comparison of the LIDAR derived non-river integrated DTM (top left) and corresponding
slopes (bottom left) and the river integrated DTM (top right) and corresponding slopes
(bottom left). LIDAR points come from DTM-AV ©2004, Swisstopo (DV043683). ....... 95
viii List of Figures

Figure 4.30 Approximation of the longitudinal water surface elevation profile with LIDAR data and
comparison with the thalweg based on surveyed cross sections. LIDAR data from DTM-
AV ©2004, Swisstopo. ..................................................................................................... 96
Figure 4.31 The author performing a line sample on the Vaube tributary of the Venoge................ 100
Figure 4.32 Example of a GSD determined with the Fehr line-by-number analysis on the Lower
Venoge River ................................................................................................................. 101
Figure 4.33 Evolution of the characteristic diameters of the bed sediment of the Lower Venoge ... 101
Figure 4.34 Pocket penetrometer (left) and shear vane (right) used during the Lower Venoge bank
investigations................................................................................................................. 105
Figure 4.35 Bank sampling locations and the level of investigation
(Background maps: CP100 © 1997, Swisstopo)........................................................... 106
Figure 4.36 Particle size distributions of the Lower Venoge banks ................................................. 111
Figure 4.37 Soil areas along the Lower Venoge River according to field investigations to be used in
the streambank erosion modeling.................................................................................. 112
Figure 4.38 Sediment transport curves for the estimation of sediment input into the Lower Venoge
River study reach........................................................................................................... 114
Figure 4.39 Estimation of the downstream rating curve for the Lower Venoge River study reach . 114
Figure 5.1 Elements of the HydroRoute model and its graphical interface ................................... 119
Figure 5.2 Snow storage unit used in the SOCONT-Alpine element in the HydroRoute model (adapted
from Hamdi et al., 2005) ............................................................................................... 120
Figure 5.3 SOCONT-Alpine element of the HydroRoute model
(adapted from Hamdi et al., 2005) ................................................................................ 121
Figure 5.4 Delineation of the subcatchments for the hydrologic modeling. ID’s of the subcatchments
and Upper Venoge snow storage units are shown. ....................................................... 122
Figure 5.5 Precipitation gages used in the hydrologic modeling. Polygons show the zone of influence
attributed to each nonrecording gage. Background image is the DTM25 © 1995,
Swisstopo. ...................................................................................................................... 124
Figure 5.6 Climate and discharge stations used in the hydrologic model. The polygons show the zone
of influence for each climate station. Background is the DTM25 © 1995, Swisstopo.. 125
Figure 5.7 Analysis of the effect of elevation on mean annual precipitation to determine precipitation
lapse rates...................................................................................................................... 129
Figure 5.8 Hourly temperature lapse rate (°C/m) calculated with the Dole, Fretaz, and Pully ANETZ
stations for the year 2004 .............................................................................................. 130
Figure 5.9 Estimation of the snowmelt temperature based on the end of the simulated Jura snowmelt
and the snowmelt visible in the 1978 Lussery discharge series .................................... 132
Figure 5.10 Comparison of mean monthly PET calculated with PM
and the temperature-based method ............................................................................... 135
Figure 5.11 Flow duration curve (left) and the observed vs. simulated mean daily flows (right) for the
period 1994-2004. ......................................................................................................... 135
Figure 5.12 Comparison of maximum annual discharge behavior of the best "sum of Nash" of the
temperature-based PET calibration versus observed discharges using a Gumbel (EVI)
frequency analysis ......................................................................................................... 136
 List of Figures ix

Figure 5.13 Gumbel frequency analysis of maximum annual discharges for the station of Ecublens,
observed and simulated ................................................................................................. 136
Figure 5.14 Gumbel frequency analysis of maximum annual discharges for the station of Lussery,
observed and simulated ................................................................................................. 137
Figure 5.15 Simulated and historically corrected discharges .......................................................... 138
Figure 5.16 Mean discharge-duration-frequency curves for the Ecublens Venoge River recording
station ............................................................................................................................ 139
Figure 5.17 Low probability floods generated from historical floods.............................................. 142
Figure 5.18 Example illustrating the insertion of the 25
and 27 year return period floods into the three 300 year series.
One rectangle represents the 236 floods of the 25 year historical period. ................... 144
Figure 6.1 Comparison of observed and simulated water surface elevations for the October 26, 2004
high flow event............................................................................................................... 150
Figure 6.2 Final toe shear stress correction factor (in grey) as a function of the ratio of radius of
curvature to width used in the streambank erosion modeling. Test case curve is the dashed
line. ................................................................................................................................ 151
Figure 6.3 Comparison of total measured and simulated erosion as well as the average absolute %
difference for measured and simulated profile erosion................................................. 155
Figure 6.4 Comparison of two equations for erodibility as a function of critical shear stress. ..... 155
Figure 6.5 Bed elevation changes simulated with the 26 year discharge record ........................... 159
Figure 6.6 Examples of bank failure widths and probabilities for the following cross sections: km
2.693 right (top left), km 10.080 left (top right), km 10.016 (bot. left), and km 2.949 (bot.
right).............................................................................................................................. 160
Figure 6.7 Streambank erosion danger map based on CCHE1D simulations and the FOWG danger
matrix (bottom left inset). Bank protection failure probability
is not taken into account................................................................................................ 162
Figure 6.8 Results of the streambank erosion simulations for the right bank at kilometer 2.693 .. 163
Figure 6.9 Comparison of erosion results when the shear stress correction function does not reduce
the average shear stress in straight sections (left) or reduces it to half (right). The
simulated eroded distances are shown by the lines perpendicular to the channel
centerline. ...................................................................................................................... 165
Figure 6.10 Test case erosion to determine the importance of centerline smoothing. The simulated
eroded distances are shown by the lines perpendicular to the channel centerline. ...... 166
Figure A.1 Textural triangles (FAO) of the Vaud plateau morainic and molassic particle distributions
(adapted from Grattier, 1980)....................................................................................... 184
Figure A.2 Textural triangles (FAO) of the Vaud plateau alluvial soil particle distributions
(adapted from Grattier, 1980)....................................................................................... 185
Figure A.3 Upstream section of the study reach of the Lower Venoge River (background: CP25
©1997, Swisstopo). Distance in meters from outlet is shown by points. 2004 centerline in
black. ............................................................................................................................. 186
Figure A.4 Grain size fractions as determined by the Fehr method for the Lower Venoge, the Ruisseau
de Molomba, the Vaube, and the Arena ........................................................................ 188
Figure A.5 Classification of streambanks used in the project SAFER on the River Enrick
(Jacobs Babtie, 2005).................................................................................................... 189
x List of Figures

Figure A.6 FAO textural triangle representation

of Lower Venoge 2004 level 2 investigation PSD’s ...................................................... 191
Figure A.7 FAO textural triangle representation of Lower Venoge 2004 level 1 investigation PSD’s.
Lines join PSD’s sampled from different depths of a same bank. ................................. 192
Figure A.8 Streambank erosion danger map of the Venoge River (bank protection failure probability
not taken into account) .................................................................................................. 202
 List of Tables xi

List of Tables
Table 2.1 Comparison of geofluvial streambank erosion models capable of simulating cohesive
bank erosion and failures (adapted from ASCE (1998b))............................................... 29
Table 3.1 Venoge catchment land-use and changes from 1985 to 1997 (Data extraction from
GESREAU) ...................................................................................................................... 41
Table 3.2 Lower Venoge River sinuosity and characteristic bend lengths ..................................... 53
Table 3.3 Average width of 4 zones of the Lower Venoge River (c.f. figure 3.15). Results were
developed from HEC-RAS simulations using 1996 cross section data. .......................... 55
Table 4.1 High flow events during the erosion monitoring period. Discharge information is derived
from the Ecublens stage recorder. .................................................................................. 65
Table 4.2 Measured high water marks for the study reach. 1996 report values come from the
HYDRAM Venoge study (Lance et al., 1997). ................................................................. 66
Table 4.3 Measured cumulative bank top change from November 2003 - September 2005. Shaded
cells indicate bank failures (>20 cm). Maximum change was 580 cm. .......................... 74
Table 4.4 Measured cumulative bank bottom changes from November 2003 - September 2005.
Maximum bottom erosion was 444 cm. ........................................................................... 75
Table 4.5 Comparison of 1996 and 2004 cross section widths....................................................... 82
Table 4.6 Cohesion and friction angle values as well as initial values used according to soil
classification.................................................................................................................. 103
Table 4.7 Bank types and stability as evaluated in the SAFER project on the Enrick River (Jacobs
Babtie, 2005) ................................................................................................................. 103
Table 4.8 Field and laboratory tests performed in the detailed level 1 of bank investigation for the
classification of the bank soil.
The right column indicates the method or possible values. .......................................... 107
Table 4.9 Results of level 1 soil analyses for the monitored erosion bends and their USCS
classification.................................................................................................................. 108
Table 4.10 Results of the level 2 and 3 bank investigations on the Lower Venoge River. Root density
is medium with a value of 1 and high with a value of 2. LS is a sandy-silt and SL is a silty-
sand. .............................................................................................................................. 109
Table 5.1 Properties of the subcatchments and elevation bands as identified in figure 5.4 ......... 123
Table 5.2 Precipitation station data used in the hydrologic study. Mean annual precipitation is
given for different periods. Les Bioux and Le Sentier stations are summed together... 124
Table 5.3 Summary of the available hydrological data and simulation periods........................... 126
Table 5.4 Muskingum K values for the Lower Venoge.................................................................. 128
Table 5.5 Calibration objective function results for 25 of the 10000 simulations using Penman-
Monteith PET ................................................................................................................ 134
Table 5.6 Calibration objective function results for 25 of the 10000 simulations using temperature-
based PET...................................................................................................................... 134
Table 5.7 Gumbel frequency analysis results for instantaneous peak discharges, and flood volumes
and durations of the Venoge River for the Ecublens recording station ........................ 140
Table 5.8 Annual peak floods recorded at the Ecublens Venoge River recording station. Events
shaded in grey are those selected for the generation of low probability floods. .......... 141
xii List of Tables

Table 5.9 Insertion of low probability floods into the 300 year record to form three 300 year
scenarios. The 1000 year flood is added to the end of each scenario........................... 144
Table 6.1 Comparison of historical and simulated WSE’s............................................................ 151
Table 6.2 Results of the bank erosion calibration. Calibration of the bank critical shear stress is
based upon groups of measured verticals as shown by areas outlined in solid black. . 152
Table 6.3 Correlation matrix of critical shear stress and soil properties..................................... 156
Table 6.4 Results of a multiple linear regression to explain calibrated critical shear stress....... 157
Table A.1 Cumulative bank position change.
Medium gray shaded cells indicate bank failures (>20 cm). ........................................ 193
 Chapter 1

Introduction

1.1 Streambank erosion in the flood damage context

1.1.1 Streambank erosion damages

The streambank failure on the unique access road to the Engelberg mountain resort in the Swiss Central
Alps on August 22, 2005 illustrates quite well the extreme damages that can be caused by streambank
erosion (figure 1.1). This streambank failure caused the collapse of the roadway foundations which led to
the collapse of approximately 80 meters of roadway. This perturbation of traffic had enormous
consequences as the village of Engelberg was accessible only by helicopter during a two week period.

Although the Engelberg case is an extreme one, streambank failures which perturb traffic are not
uncommon. Roadways and railways are often located along rivers. They are usually protected by bank
protection works, but in extreme floods those works sometimes fail leading to the collapse of the roadway
or railway. Sometimes, it is the foundations of bridges or their pillars which are attacked causing structural
damage to the bridge and significant traffic perturbation or at minimum important reparation costs (figure
1.2).

The actual cost of streambank erosion damages is poorly known. They usually are not differentiated from
flooding damages, and when no floods occur but streambank protection must be put in place, the costs are
often not tallied as flood damage. The other difference compared to flood water damages is that
streambank erosion causes protection work degradation. The protection works must then be restored or

1
2 Introduction

replaced. This is also a hidden streambank erosion damage cost. Even if the streambank protection
doesn’t degrade rapidly, it must be surveyed regularly to ensure its stability.

Figure 1.1. Destruction of the roadway leading to the mountain resort in Engelberg, Switzerland. The
village was accessible only by helicopter during a two week period. (from Blick, 2005)

Considering only land value, Graf (1984) estimated the value of land lost due to streambank erosion over
a 50 year period for the Rillito Creek to be 5 times the floodwater losses estimated by the US Army Corps
of Engineers. "In 1978, the US Army Corps of Engineers estimated that there were over 1,130,000 km
of eroding river banks in the USA, with the resulting annual land loss valued at about $270 million and
with annual bank-protection costs nearly three times that amount" (Wood et al., 2001 citing Blackmar,
1995). In many countries today, the rivers are not allowed to meander anymore, removing the question
of loss of land, but in these countries the pertinence of the cost of bank protection implementation,
maintenance, and replacement as well as the sometimes dramatic damages to structures or transportation
networks comes into play.
 Streambank erosion in the flood damage context 3

Figure 1.2. Undermining of a bridge pillar (from Loat and Petrascheck, 1997)

1.1.2 Streambank erosion mapping and prevention

The Federal Office for Water and Geology (FOWG) in their booklet (Loat and Petrascheck, 1997) on
recommendations to take into account flood hazards in the framework of spatial planning states that
streambank erosion, is, in many cases, the type of hazard causing the most damages. This has led the
Swiss government to require that streambank erosion be included in flood hazard maps. The directives
for mapping the streambank erosion hazard are shown in figure 1.3 (Loat and Petrasheck, 1997). The
map of the streambank erosion hazard is called the streambank erosion danger map. A danger level on a
map aggregates the streambank erosion hazard information. The streambank erosion danger mapping
directives, contrary to the flood inundation directives, are very difficult to evaluate. Knowledge of
historical bank failure widths is often very poor and rarely measured precisely, and their probability is
usually not known at all. Even if a rough estimation of a bank failure width is possible with a relatively
simple calculation, the estimation of the bank failure probability is much more difficult. Once a bank has
failed, the hydraulic flow conditions affecting the bank will be changed so that preceding bank failures
can not be used in the probability calculation of the present bank. This presents a major problem for
Swiss engineers to produce the streambank erosion danger maps according to the Federal Office for
Water and Geology directives.
4 Introduction

Intensity
(Failure 30 yr 100 yr 300 yr extreme
width) Streambank Erosion
strong Danger Level

2m high

medium
medium
low
0.5 m
residual
weak

high medium low

Occurrence Probability (Frequency)
Figure 1.3. Swiss Federal Office for Water and Geology (Loat and Petrascheck, 1997) directives for
mapping streambank erosion

Many other governments have also recognized the importance of mapping streambank erosion. The
Federal Emergency Management Agency (FEMA) of the United States published the "Riverine Erosion
Hazard Areas - Mapping Feasibility Study" (FEMA, 1999). It concluded that mapping of riverine
erosion hazard areas is technologically feasible. Contrary to the Swiss approach, it is recommended to
delineate the "riverine erosion hazard area" (REHA) for a 30-60 year period. The report expresses the
fact that the 30 year estimation will be more reliable, but a 60 year estimation can be drawn up to comply
with Section 577 of the National Flood Insurance Reform Act (NFIRA). The erosion hazard area is
defined in the act as follows:

"Erosion hazard area means, based on erosion rate information and other historical data available, an
area where erosion or avulsion is likely to result in damage to or loss of buildings and infrastructure
within a 60-year period." (section 577 of NFIRA quoted in FEMA, 1999)

This definition of course can be criticized in that it confuses the definitions of hazard and risk.

The above two approaches illustrate two types of streambank erosion hazard maps. One which is related
to the bank failure width due to a single event or the other which maps the river channel planform change
over a number of years. Both maps have their purposes. The Swiss streambank erosion hazard map,
called a streambank erosion danger map, is important for describing how much erosion could happen
during a single event. This information is sometimes roughly known for rapidly moving channels, but
for relatively stable banks it is usually unknown. The REHA is also an important information for flood
risk management and for spatial planners (assuming the erosion hazard is mapped regardless of
vulnerability). The REHA will help the flood manager to decide where bank protection works will be
necessary and the spatial planner where development zones should be avoided.

A third type of streambank erosion hazard map is the streambank erosion intervention map. A prototype
of this map was developed in the Venoge study (Lance and Consuegra, 1998) done by the HYDRAM
laboratory of the EPFL. Figure 1.4 shows an example of this type of map. What is important in this map
is that a stream corridor is defined. This stream corridor in the example of the Venoge study was
 Thesis objectives 5

determined according to the historical locations of the river visible in historical maps, aerial photos, and
Digital Terrain Models (DTM) and according to the confining nature of the river valley. With knowledge
from either a streambank erosion danger map or from observations, a maximum possible channel change
is used as a security distance. Land use maps superimposed on the stream corridor will show the existing
conflicts. An additional security distance, let’s call it the infrastructure security distance, can be added
to the erosion security distance as an extra protection for infrastructure. Thus, spatial planners can use
this map to make appropriate decisions on land use and riverway managers can use the map to decide
when they must react to stop an erosion. This map has the advantage that it does not try to use the perilous
calculation of a 30 year or 60 year possible river channel change, but rather shows the maximum possible
space allowed for channel movement. Usually the stream corridor can be more easily defined than the
possible channel movement during the next 60 years.

Figure 1.4. Streambank intervention map (adapted from Lance and Consuegra, 1998)

1.2 Thesis objectives

Due to the importance of damages caused by streambank erosion during flood events, and the lack of
tools available to calculate bank failure widths and their probability, the primary objective of this thesis
is to develop a methodology to map streambank erosion hazards. This will necessitate the improvement
of existing models. To test such a model, a case study is necessary. The Venoge River is chosen for the
case study and bank erosion data is acquired.

1.2.1 Modeling streambank erosion

Near-bank hydraulic processes require a 3-D representation of the flow field to attempt to estimate the
boundary shear stress that is responsible for the fluvial entrainment of bank materials. This poses of
course the problem of the significant computational time and data necessary for such a model. Two
6 Introduction

dimensional flow models are most often used due to the lateral variation of the flow parameters they
provide. One dimensional flow models have also been used, but a correction to the shear stress
estimation on the outer bank must be applied for realistic erosion-producing shear stresses to be
simulated.

Due to the computational consequences of 2-D and 3-D flow and sediment transport models, the first
objective of this thesis is to adapt an existing 1-D model to simulate erosion and bank failures. This will
require that the complexities of the longitudinal extent of bank erosion and the estimation of the near-
bank shear stress be addressed.

1.2.2 Acquisition of streambank erosion data

The second objective of this research is to explore the practicality of obtaining sufficient data for
simulating streambank erosion with spatially varying data. If streambank erosion models are to be used
for streambank erosion hazard mapping, data collection methods must also be considered and their
optimization must also be carried out.

Besides the importance of estimating the practicality of acquiring data for streambank erosion modeling,
this research could provide a valuable data set for further streambank erosion modeling efforts. In their
article on "River Width Adjustment", the ASCE Task Committee (1998b) cites only three data sets "that
are suitable for use with numerical models of width adjustment". This significant lack of data is a major
hurdle to overcome for developing streambank erosion models.

1.3 Report organization

Chapter 2 of this dissertation starts with a description of the important processes involved in streambank
erosion. It goes on to explore the modeling possibilities available for streambank erosion. Finally, a
methodology for calculating the streambank erosion hazard map based on the model CCHE1D is
presented.

Chapter 3 gives the geomorphological context of the Venoge catchment (geology, land-use, soil,
topography, hydrographic network, climate, and hydrologic regime) and river (channel morphology) as
background information for the case study.

Chapter 4 goes into detail about the data that was obtained for the case study on the Lower Venoge River.
Two experimental reaches were setup on the Venoge to gather bank erosion data. The studied reaches
and equipment are described and the measured values are given. Orthophotos are available for the
Venoge River and their importance is discussed concerning channel planform and erosion measurement.
LIDAR (LIght Detection And Ranging) topographic data was obtained for the Lower Venoge River and
its comparison with and integration with surveyed topographic points is expounded on. The interpolation
of cross section data poses a significant problem for the topographic description of banks. Methods are
developed to improve conventional interpolation. The methods used for gathering information on bed
and bank sediment information are also described. The calculation of boundary conditions (flow depth
 Report organization 7

and sediment input) finish this chapter.

Although infrequent in hydraulic studies, a detailed hydrologic analysis was performed and is reported
in Chapter 5. It is well known that the sequence of flood hydrographs is very important to streambed
fluctuations and streambank erosion. A continuous hydrologic model, HydroRoute, is used to generate
the important tributary hydrographs. This chapter presents in the case of the Venoge how data can be
spatialized and how this spatialization permits the variability of tributary discharges and consequently
the tributary sediment input. By generating low probability floods and combining them with the
historical discharge series, three 300 year discharge series are constructed for the hydraulic simulations.

Chapter 6 describes the detailed streambank erosion modeling done with the model CCHE1D. The
model setup and preparation are explained. The calibration of the model as well as the simulation of bank
erosion and sediment transport with the historical discharges from 1979-2005 is accomplished. Bank
mechanical strength tests were done and are compared to the calibrated bank parameters to determine
the utility of those tests. Simulated and measured event erosion are examined to examine the capability
of the model to reproduce event erosion. A program to permit the frequency analysis of the bank failures
is developed and its use with the simulation results is illustrated. The results are mapped to produce a
streambank erosion danger map. Finally, a hypothetical test is performed to investigate a possible
improvement to the shear stress formulas and to analyze the sensitivity of the model to planform
digitization.

These chapters lead to the final conclusions of this dissertation and also to a section dedicated to the
numerous ideas generated by this thesis for future research.
8 Introduction
 Chapter 2

Streambank Erosion Hazard

Assessment and Mapping
Methodology

2.1 Introduction
The simulation of streambank erosion has until present eluded commercial software packages. This is not
because of its insignificance, but rather because of the complexity of the processes involved. This chapter
will serve to review the processes involved in streambank erosion. Secondly, it will briefly describe
different type of models used in research today. Thirdly, the methodology chosen for the assessment of
streambank erosion hazards of single-threaded, subcritical flow streams is elaborated.

2.2 Streambank erosion processes

Streambank erosion is best referred to as streambank retreat. Bank retreat refers to the combined processes
of bank erosion and bank mass failure. Bank erosion, simply put, occurs when the shear strength of
sediment is overcome by the boundary shear stress exerted by the streamflow. Bank mass failure occurs
when a large mass of bank collapses because the bank soil strength can no longer resist the bank weight
and is often provoked by bank steepening or heightening by bank erosion or bed scour, respectively. Bed
scour can play a significant role and this necessitates the understanding of sediment transport. Catchment
and climate configuration produces the water and sediment discharges and the channel morphology that
in turn produce the boundary shear stresses. The reviewed processes below then will be structured
according to these main themes: catchment, fluvial, and bank processes.

9
10 Streambank Erosion Hazard Assessment and Mapping Methodology

2.2.1 Catchment processes

Schumm’s (2005) diagram (figure 2.1) describes quite well the catchment processes that dictate river
type and reach variability which in turn will dictate the streambank erosion potential of a reach. Thorne
(1997) describes channel form as depending on the driving variables of water and sediment inflow and
boundary characteristics of valley slope and topography, bed and bank materials, and riparian
vegetation. These two conceptualizations of catchment processes are represented in figure 2.1.

Upstream Controls

Geology Humans Climate

Driving variables: water and sediment input

nnel morpholo
Ch a gy

River Type
Local fixed controls Local variable controls
Bank characteristics Floods
Tributaries Reach Variability Vegetation
Active tectonics Accidents
Valley morphology

Bo Baselevel Humans
un
da r
yC
on d
itions Downstream Controls

Figure 2.1. Controls of river morphology and behavior (after Schumm, 2005)

Upstream controls include geology, climate, and humans. These controls interact and produce numerous
river types. Schumm (2005) also includes history and subdivides geology into tectonics and lithology.
History is not included in figure 2.1 because time evolution is, in the author’s opinion, a significant part
of each of the upstream controls. Tectonics and the geological formation is responsible for the catchment
relief and lithology. Uplift or subsidence due to plate activity or isostatic adjustment continues to change
the relief of the earth although at very long time scales. The geological formation of a catchment is often
dependant on many systems: weathering, slope mass movement, river, groundwater, glacier, and eolian
systems. These systems are products of the upstream controls of geology and climate. Weathering is
produced by the interaction of geology and climate. Physical and chemical breakdown of the catchment
rock will be dependant on the climate. Slope mass movement also is a result of climate and geological
controls. A more pronounced catchment relief in friable soil and rock will promote a high downslope
mass movement. Climate plays a very important part in downslope movement due to soil saturation or
by freezing and thawing. Earthquake activity can also be significant in downslope mass movement.
River systems are extremely important in relief development by their sediment erosion and deposition.
 Streambank erosion processes 11

River systems can be divided into three systems: collection (sediment source), transport (transfer), and
dispersal (sediment sink) (Hamblin and Christiansen, 1998). Over geological periods, these three
systems can significantly form a relief. Groundwater systems are intricately related to river systems.
They can also be determinant in forming relief by dissolution of bedrock. This is especially the case of
sinkholes in karst topography. Glacier systems are extremely effective in their eroding and depositing
action. During the last Ice Age, much of the world’s drainage systems were largely modified by the
Pleistocene glaciers. Wind is also responsible for sediment erosion, transport, and deposition, although
Eolian systems are the least effective in landscape transformation.

Besides the significant historical effect of climate on the geological formation of a catchment, it is of
course responsible in connection with the catchment relief for the present hydrological regime of the
catchment. Climate of course produces the precipitation events which dictate the subsurface moisture
conditions and produce the stream discharge events. Connected with the subsurface moisture conditions,
slope mass movement provides sediment input into the drainage network.

Schumm (2005) includes humans also as an upstream control and rightfully so. Graf (2001, in Schumm,
2005) estimated that 79 percent of American rivers have been influenced by humans. Dams and
reservoirs, interbasin diversions, irrigation, channelization, levees, and mining are the most common
human disturbances to natural catchment processes.

Local controls, fixed and variable, create reach variability. Local fixed controls are those that are
geographically fixed. Schumm (2005) lists bedrock, tributaries, active tectonics, and valley morphology
as fixed local controls. Bedrock and resistant alluvium will resist channel erosion. This can maintain
longitudinal profile convexities or cause irregular meander development. Clay plugs resulting from
meander chutes or cutoffs are known for their ability to resist meander migration. Tributaries offer their
water and sediment to the main river which in the case of small tributaries may only have slight
repercussions, but in the case of large tributaries could cause a complete regime change or significant
changes in longitudinal slope, sinuosity, and channel width downstream from the tributary. Active
tectonics can also cause changes to the valley floor. Subsidence due to petroleum or groundwater also
effects the valley floor and is called pseudotectonics by Schumm (2005). Valley morphology provides
the convexities and concavities in the longitudinal profile that are critical to channel form and sediment
transport.

Variable local controls are those that are variable in space and time according to Schumm (2005). He
lists floods, vegetation, and accidents in that category. Including floods could be questionable in that the
upstream controls produces the sediment and water input into the drainage network, but floods can have
significant sediment transport that can vary from flood to flood. In this respect, it is reasonable to include
it as a reach variability control. It is strongly connected to the accident control. Accidents are
unanticipated events such as log jams, ice jams, and natural dams produced by earthquakes and
landslides. The significant, usually temporal, change in channel discharge and sediment discharge can
cause significant channel changes that can take a long time to reestablish. It is well known that vegetation
influences significantly channel width. Hey and Thorne (1986) determined hydraulic geometry relations
that showed that the width of the channel increases with increasing bank vegetation density. Recently,
though, Hession et al. (2003) showed that channel width was greater for forested riparian zones than
12 Streambank Erosion Hazard Assessment and Mapping Methodology

grassy ones. Abernethy and Rutherfurd (1998) indicated that the effectiveness of vegetation on width
control depends on the location of the reach, whether it be upstream, middle, or downstream.

Downstream controls are that of base level and humans (Piégay and Schumm, 2003). Base level is the
lowest point to which a river flows. It is sea level for large rivers, but for smaller rivers could a large lake
or a resistant bedrock. Humans can influence the baselevel by river damming. River damming creates an
artificial baselevel for the river upstream and creates an upstream control for the river downstream.

Besides looking only at the driving and boundary conditions that will dictate the channel form, long
profile, and planform, it is convenient to look at catchment processes in terms of the three river systems
mentioned above: sediment source, transfer, and sediment sink zones. It is fundamental in sediment
transport studies to identify the type of the river system. Sediment source zones are usually supply
limited. Mountainous and hilly headward zones are usually sediment source zones. Bed movement can
be significant in these zones due to the limited sediment supply and sediment bursts caused by slope
failures. Transfer zones are zones in which the sediment transport is in balance. Erosion can occur in
these zones, but deposition will keep up with the erosion. Sediment sink zones are transport limited. The
streamflow does not have a sufficient transport capacity to carry all of the sediment. Alluvial fans and
deltas are the most flagrant sediment sink zones.

2.2.2 Fluvial processes

Natural river flows are extremely complex. Hydraulic research has not yet unraveled the physical laws
that can completely describe river flow. It is beyond the context of this thesis to fully describe what is
known about the physical laws that govern them, but a brief description will be given here to describe
the important fluvial processes acting in streambank retreat.

All river flows, whether it be in straight or bending flow, have secondary flows. These secondary flows
are more pronounced in river bends. Secondary flows can be stress induced (anistropic turbulence) or
skew induced (streamwise curvature) (ASCE, 1998a). Stress induced secondary flows are the reason for
their existence in straight sections, while secondary flows in bends are mostly skew induced. These
strong bend secondary flows are represented in figure 2.2 by the well known helical flow motion in
bends. This flow motion results from the centrifugal force on the water elements in the bend. Those
forces are higher at the surface. They cause the superelevation that is observed in bends. Superelevation
causes a greater pressure on the flow in the outer bottom. This difference of pressure results in a bottom
flow from the outer bank towards the inner bank. These two lateral flows are superimposed on the
streamwise flow to create the helical flow pattern. Blanckaert and Graf (2001) call this helical motion
the center-region cell. As can be seen in figure 2.2, an outer region cell exists. Blanckaert and Graf
(2001) suggest that the outer-bank cell protects the outer bank from the high velocities of the center-
region cell and reduces the turbulent activity. Boundary shear stress is often expressed by mean values,
but that shear stress varies around the mean. These variations are caused by turbulent bursts: sweeps and
ejections. Those bursts have been found to be extremely energetic by Franca (2005). Including equations
to represent those bursts to better characterize the instantaneous boundary shear stress is not yet feasible
for practical engineering application (ASCE, 1998a).
 Streambank erosion processes 13

Center-region cell Outer-region cell

Figure 2.2. Helical flow motion in channel bends (adapted from Hamblin and Christiansen, 1998)

The Navier-Stokes equations are classically used to represent open channel flow. The most well known
form of these equations is the Saint-Venant equations for 1-D open channel flow (e.g. Cunge et al.,
1980). The Saint-Venant Exner equations include sediment transport with the hydrodynamic flow. These
equations use the momentum and continuity equations of the Saint-Venant equations and a continuity
equation for solid transport, leaving the energy slope and sediment discharge to be computed by
empirical formulas (Graf and Altinakar, 2002). As in the Saint-Venant equations, a resistance law, such
as that of Manning, must be used to determine the energy slope.

Sediment transport is initiated by the boundary shear stress that overcomes the critical shear stress of the
sediment. Boundary shear stress, τo, can be estimated by:
τ o = ρgR h S Eq. 2.1

where ρ is the density of water, g is gravity, Rh is the hydraulic radius and S is the energy slope. Other
variations of equation 2.1 exist (ASCE, 1998a) such as the one that replaces Rh with water depth. Critical
shear stress, τc, is often based on Shield’s work (1936) and the well known Shield’s diagram. It can be
estimated by:
τ c = θ c g ( ρ s – ρ )D Eq. 2.2

where θc is the critical dimensionless shear stress known as Shield’s value, ρs is the density of the grain,
and D is the grain diameter. On hydraulically rough beds, θc ranges from 0.03 to 0.06 and is usually taken
to be around 0.45 which is less than the original Shield’s value of 0.06 (Knighton, 1998). The varying
range of Shield’s value is partially due to grain shape and packing. For cohesive sediments, there is no
14 Streambank Erosion Hazard Assessment and Mapping Methodology

unique critical shear stress. This results from the electrochemical binding of the soil particles and the fact
that erosion occurs by aggregate movement and not by particle movement.

Sediment transport modeling often relies on sediment transport capacity equations. The Smart and
Jaeggi formula is often used in Switzerland due to its reported validity domain of (0.04 to 20% slopes)
(Smart, 1984) and is given by the following formula:
d 90 0.2 0.6 0.5
φ = 4 ⎛⎝ -------⎞⎠ S Cθ ( θ – θ c ) Eq. 2.3
d 30

where φ is the dimensionless sediment transport (qb/[g(s-1)d3]0.5, in which qb is the volumetric sediment
discharge per unit channel width, d is the mean grain diameter, and s is the specific weight of sediment,
and d90 and d30 are the grain diameters for which 90% and 30% weight, respectively, of a nonuniform
sample is finer, S is the bed slope, C is a flow resistance factor, and θ is the dimensionless shear stress.
More detailed developments on sediment transport equations can be found in Graf and Altinakar (2002)
and the implementation of other currently used formulas can be found in (Wu and Vieira, 2002).

Johannesson and Parker (1989) proposed a model that combined the flow field, bedload transport, and
bed topography calculations. Their model is important because of the physical explanation given to bend
pools and phase lag via the incorporation of a secondary flow parameter. A simplified equation for bend
topography is given by
h
sin α = K --- Eq. 2.4
r

where K is a coefficient, h is local depth and r is the radius of curvature. After integration this formula
becomes:
h r K
----- = ⎛ ----⎞ Eq. 2.5
h0 ⎝ r 0⎠

where h0 is the centerline water depth, and r0 is the centerline radius of curvature. The value of K has
been approximated as 7 ⋅ tan φ by Engelund (1974, in Hersberger 2003) where φ is the dynamic shear
angle. Bridge (1976, in Hersberger 2003) estimated that K is 11 ⋅ tan φ with values of φ in the range of
0.32 to 0.75 and has proposed other estimates as well (Bridge, 2003).

In recent years, two and three dimensional models of flow and sediment transport have been developed.
One recent example of the use of a 2D model based on the equations developed by Rodi (1993) for
solving flow in curved bends is that of Wu and Wang (2004). They proposed a semi-empirical formula
for the determination of the helical flow intensity. This allowed them to evaluate the dispersion terms of
the depth-averaged 2D momentum equations and suspended-load transport equation. It also allowed
them to evaluate the bedload transport angle. The quantification of the dispersion terms is very important
for the representation of the secondary flow.

In spite of the advances in modeling 2D and 3D flows in curved channels, the variability of natural rivers
poses many problems for modeling (ASCE, 1998a). Adequately representing the topography, roughness,
and sediment variability is impossible for long reaches. Topographic variability occurs from the
changing cross section and planform shapes. Roughness variability occurs due to the distribution of grain
 Streambank erosion processes 15

sizes in the section, but also due to bed forms. Sediment variability occurs due to selective entrainment
and fining processes.

2.2.3 Bank processes

Definitions of bank retreat terms, as defined by Lawler et al. (1997), are:
"Bank erosion: detachment, entrainment and removal of bank material as individual grains or
aggregates by fluvial and subaerial processes.
Bank failure: collapse of all or part of the bank en masse, in response to geotechnical instability
processes.
Bank retreat: net linear recession of bank as a result of erosion and/or failure
Bank erodibility: the ease with which bank material particles and aggregates can be detached,
entrained and removed (normally by flow processes)."

As can be seen in these definitions, bank erosion should not be confused with bank failure and the two
are combined in the term bank retreat. Channel cutoffs are also extremely important in terms of channel
planform change, but since the modeling in this dissertation does not concern middle or long-term
planform changes, it will not be dealt with.

2.2.3.1 Bank erosion

Fluvial

As is the case for the river bed, the force exerted by the boundary shear stress must overcome the
resisting forces of the sediment as given by the critical shear stress. For non-cohesive soils, the critical
shear can be determined by equation 2.2. This equation must be multiplied though by a coefficient, λ,
that takes into account the additional instability caused by the bank slope. It was defined by Lane (1953)
as:
2
tan θ
λ = cos θ 1 – ------------
2
- Eq. 2.6
tan φ

where θ is the bank angle with the horizontal, and φ is the internal friction angle of repose of the bank
material.

Fluvial erosion of cohesive materials is much more complicated. The ASCE Task Committee (1998a)
indicates that in addition to the interparticle, electrochemical bonding that resist the driving forces, the
soil structure due to its formation and soil moisture history influence a soil’s critical shear stress.
Although, critical shear strength testing (see section 4.6.1 for testing methods) must be undertaken to
determine a soil’s value, generally, it is higher for cohesive soils than non-cohesive soils, especially for
higher clay content soils.

For cohesive banks, Arulanandan et al. (1980, in Osman and Thorne, 1988) developed a method for
estimating lateral erosion. Their method consists of:
16 Streambank Erosion Hazard Assessment and Mapping Methodology

1.Estimating the shear stress.

2.Estimating the critical shear
stress. This can be done by
determining the sodium adsorption
ratio (SAR), pore fluid salt
concentration, and dielectric
dispersion. With these values, the
critical shear stress can be
determined with figure 2.3.
3.An erosion rate, εr, can be
determined by the formula,
–4 – 0.13τ c
ε r = 223 ⋅ 10 τ c e Eq. 2.7
where τc is the critical shear stress
in dynes/cm², and εr is in g/
(cm²•min).
4.A lateral erosion rate can then be
determined by:
dW ε τ–τ
Figure 2.3. Critical shear stress as a function of SAR, soil salt -------- = ----r ⎛ ------------c-⎞ Eq. 2.8
concentration, and dielectric dispersion (from Osman and dt γs ⎝ τc ⎠
Thorne, 1988, after Arulanandan et al., 1980 and reprinted where γs is the specific weight of
with permission from the American Society of Civil Engineers) the soil, and dW/dt is the lateral
erosion rate for all τ greater than τc.

Subaerial

The effectiveness of subaerial processes in erosion is poorly known. What is sure is that, whether by
actual erosion or by weakening of the soil structure, subaerial processes can be a significant factor in
erosion. The most commonly cited subaerial processes are freeze-thaw and wetting-drying cycles.
Couper (2003) exposed riverbank samples to repeated freeze-thaw and wetting-drying cycles in the
laboratory and measured erosion. Due to the laboratory conditions, her work can not be used directly to
estimate erosion from subaerial conditions, but she did confirm that banks with higher silt-clay content
have a higher susceptibility to subaerial erosion. Compared to fluvial erosion, the role of subaerial
erosion though is subsidiary (Knighton, 1998). It can be, though, the dominant process, especially in
small drainage basins that do not produce enough shear stress to erode cohesive banks (Lawler et al.,
1997).

Among freeze-thaw activity, ice needles are probably the most effective eroding agent. They can lift and
incorporate bank material during their growth, thus weakening or eroding the bank (Lawler et al., 1997).
Wetting-drying cycles can produce swelling and shrinking that induce cracks which will increase bank
erodibility (ASCE, 1998a).

Vegetation

Vegetation, generally, reduces bank erosion. This is largely due to (ASCE, 1998a): root systems that
physically retain soil particles, the reduction of near-bank velocities, and the dampening of peak
 Streambank erosion processes 17

turbulence to reduce shear stresses. Erosion of well vegetated banks is reduced by one to two orders of
magnitude as reported by the ASCE Task Committee (1998a) according to many studies.

Seepage

Within fluvial processes, erosion by the main river current is the most frequent eroding agent, but water
within the bank can also be responsible through seepage (ASCE, 1998a). Strong seepage often occurs in
stratified banks in layers of higher permeability. If this seepage is strong enough, it can entrain soil
particles. This process is called piping. After significant piping, mass wasting (failure) of the upper bank
soil is likely to occur.

2.2.3.2 Bank failure

Bank mass failure occurs when the driving gravitational force overcomes the resisting force provided by
soil friction and cohesion, and by root reinforcement. Bed scour and/or bank toe erosion often causes
bank instability. Different bank failure modes exist depending on bank soil, stratification, height, and
vegetation conditions (figure 2.4). These bank failure modes are difficult to clearly observe in nature
because of the combination of bank weakening effects and fluvial scour (Thorne, 1982).

Tension crack
Bulging
Erosion

α α
Before Before Before Before

After After After After

(a) Shallow failure (b) Planar failure (e) Rotational failure with weak zone (f) Massive rotational failure/landslide
- shallow bank angle - steep or vertical bank angle - failure surface dictated by position of weak zone - erosion of river bank threatens stability of whole valley side
- usually in non-cohesive banks - frequently (but not always) in non-cohesive banks - see also comments for type (d) - very large volume of slipped material
- failure nearly parallel to slope at α = φ’ - water table/channel water level usually low - failure nearly parallel to slope at α = φ’
- water seepage from bank can substantially reduce stable α relative to bank height - tension cracks up valley side, bulging above toes, or
- vegetation will normally help stabilise against failure noticeable movement are signs of potential failure
> 0.3 m
Cohesive Desiccation
Water Cohesive
table Sand/
> 1.0 m Sand/
gravel
gravel
Before Before Before Before

After After After After

(c) Planar/slab failure (d) Rotational failure in homogeneous material (g) Failure of composite bank (in tension) (h) Failure of composite bank (as beam)
- steep or near vertical banks - usually on moderately high or steep banks - occurs only where upper cohesive layer overlies - occurs as type (g)
- deep tension cracks - usually in cohesive material erodible sand/gravel - failure with upper soil in tension, followed by rotation
- failure occurs by sliding and/or toppling - tension cracks reduce stability particularly when water filled - failure by tension of lower part of overhanging block - after failure block usually remains intact with vegetation
- failure more likely if crack fills with water - significantly affected by position of water table towards river
- little affected by groundwater table - failure may extend beyond toe, see also type (e) - failure can also be by shear

Figure 2.4. Modes of bank mass failure (from Hey et al., 1991, in Lawler et al., 1997)

Mass failure is significantly different for cohesive and non-cohesive soils. Failures will be shallow for
non-cohesive banks (figure 2.4 a, b), while those in cohesive banks can be deep. This is mainly due to
the fact that shear stress increases less quickly than shear strength, while in cohesive banks it is the
opposite, promoting deeper failures (Terzaghi et al., 1996). The non-cohesive failure shown in figure 2.4
(a) occurs by the dislodgement and/or avalanching of the sand or gravel particles. It is often induced by
toe erosion. Planar failure as shown in figure 2.3 (b) can occur in non-cohesive and cohesive soils. Planar
18 Streambank Erosion Hazard Assessment and Mapping Methodology

and planar/slab failures (figure 2.4 b and c) are very common for steep bank slopes. In medium to high
bank slopes, rotational slips have been observed to occur (figure 2.4 d, e, f). These rotational slips have
a curved base which depends on the bank stratigraphy (especially weak layers) and this failure surface
can pass through the bank slope, toe, or below it. Tension failures in overhanging/cantilevered banks are
also very common. This often occurs in stratified banks. This stratification can occur because of a
composite bank soil configuration as in figure 2.4 (g) and (h), but can also occur simply by a root
reinforcement of the upper soil layer.

Stability analysis for the various modes of bank failure exist, although planar failure is the stability
analysis that has been most commonly implemented into streambank erosion models described in
literature. Rotational slip failures are at the crossroad between landslide hazard assessment and
streambank erosion hazard assessment. The multiple factors (see Hutchinson, 1995 for a listing of those
factors) involved in deep seated rotational slip failures precludes their analysis within a long river reach
streambank erosion hazard analysis. The detailed lithographic information needed as well as the
hydrogeologic and geotechnical modeling necessary are net yet feasible within a streambank erosion
hazard model. Optimally, regional streambank erosion hazard mapping or regional landslide hazard
mapping should detect potentially unstable slopes along rivers. These slopes should then be analyzed by
geotechnical means, more or less sophisticated, depending on the risk involved. Overhang tension failure
analysis formulas exist (Thorne and Tovey, 1981, Langendoen, 2000), but have not been implemented
in streambank erosion hazard models. The hydraulic analysis of such complex cross sections is usually
not incorporated in hydraulic models.

Since planar failure analysis is common for low to medium cohesive steep streambanks, and will be used
in the modeling within this thesis, planar failure analysis will be described in more detail here. For steep
cliffs in brittle granular soils exhibiting tension cracking, slips are planar requiring an analysis such as
the Culmann wedge analysis (Selby, 1993). This analysis determines the factor of safety, Fs, as the
resisting force, Fr, divided by the driving force, Fd (figure 2.5). The bank is considered stable if Fs is
greater than 1. The weight of the failure block, Wt, is the driving force, given by:
2
γ s ⎛ H – y d H' 2 ⎞
F d = W t sin β = ---- ⎜ ----------------- – ------------⎟ sin β Eq. 2.9
2 ⎝ tan β tan α⎠

where β is the failure plane angle, γs is the unit weight of the soil, H is the bank height, yd is the tension
crack depth, H’ is the non-eroded bank height, and α is the bank slope. The resisting force is derived
from a shear strength equation. The importance of taking into consideration saturated or unsaturated
conditions has been demonstrated by many authors (Simon et al., 2000 for example) and thus the
appropriate resisting force equation is:
b
Fr = Lc' + N tan φ' + ψL tan φ Eq. 2.10

where L is the length of the resisting plane, c’ is the effective cohesion, N is the normal stress due to the
weight of the soil block, φ’ is the effective friction angle, ψ is the matric suction, and φb is the increase
in shear strength associated with an increase in matric suction. The length of the resisting plane can be
found to be (H-yd)/sinβ and N = Wtcosβ. Matric suction can be estimated as the negative pore water
pressure by γ(zs-zg) where zs is the elevation of the soil and zg is the elevation of the groundwater table
 Streambank erosion processes 19

assuming hydrostatic pore water pressure distribution. The matric suction can be negative in the case of
a high water table representing the case of high pore water pressure. Groundwater modeling is possible
but has not been incorporated into present-day streambank erosion hazard models. Usually, the
groundwater table is assumed to be the same as the water level in the river. Simon et al. (2000) note that
"the value of φb is generally between 10° and 20° and increases with the degree of saturation". They
evaluated its value with linear regression after having determined the apparent cohesion, ca, which is
c’+ψtanφb, by in situ shear strength testing. The failure plane angle and the tension crack depth must
still be determined for the application of equation 2.10. Failure plane angle, β, can be determined by the
formulation of Osman and Thorne (1988):

⎧ H 2 yd 2 ⎫
β = 0.5 ⎨ atan ⎛⎝ -----⎞⎠ ⎛⎝ 1 – ⎛ -----⎞ ⎞⎠ tan α + φ ⎬ Eq. 2.11
H' ⎝ H⎠
⎩ ⎭

This equation uses the hypothesis that the most dangerous failure angle occurs when the developed
cohesion is maximum, which is criticized by Langendoen (2000). He implemented a search algorithm to
search for the failure plane which minimizes the factor of safety, Fs. Maximum tension cracking depth,
yd, is not well known. It has been estimated by Taylor (1948) as:
2c' φ
y d = ------- tan ⎛⎝ 45 + ---⎞⎠ Eq. 2.12
γ 2

Darby and Thorne (1994) reported improvements in the determination of the crack depth with their new
formula, but it uses the tensile shear strength of the soil which is difficult to determine and to which they
partly attribute the mediocre results of their calculated tension crack locations compared to observed
ones.

Figure 2.5. Planar failure analysis (from Wu and Vieira, 2002 after Osman and Thorne, 1988)

Vegetation effects

The effect of vegetation on bank failures is poorly known. It has both positive and negative effects
according to the ASCE task committee (1998a) on river width adjustment. They listed the positive
effects as:
20 Streambank Erosion Hazard Assessment and Mapping Methodology

1. an increase in shear strength due to the tensile shear strength of root systems for low banks in which
the vegetation root systems cross the shear failure plane
2. a decrease in soil moisture levels due to canopy interception and increased evapotranspiration
3. buttressing or abutment of slopes

and the negative effects as:

1. a weight surcharge
2. wind loading of the vegetation that can cause a significant destabilizing moment
3. wedging and prying in soil cracks and fissures to decrease soil strength.

Basal endpoint control

Mass failures come to rest at the base of the bank and are known as basal sediments. The rate of
entrainment by streamflow of these sediments will determine the medium and long term retreat rate of a
bank and will determine the frequency of bank failures. Thorne (1982) described the three states of basal
endpoint control as follows:

1. Impeded removal: Basal accumulation is important and the rate of entrainment is slow. Bed
elevation will increase to promote lower banks resulting in lower bank supply rates to promote
progression towards the second state.
2. Unimpeded removal: Bank supply rates are in balance with the entrainment rates. The bed level
remains stable.
3. Excess basal capacity: Fluvial entrainment is rapid, promoting frequent bank failures. Bed lowering
will result until the bank height is significant enough to supply greater volumes of bank material.
The rate of supply will increase promoting a progression to the second state.

These three states could be associated with stream reaches that are aggrading, in dynamic equilibrium,
and degrading, respectively.

Wood et al. (2001) showed that failed soil blocks that come to rest at the base of the bank develop
cohesion with the substrate to aid in their resistance to fluvial entrainment. This means that large failed
blocks will have to be eroded particle by particle until they become small enough for the streamflow to
able to entrain the entire cohesive aggregate.

Sections 2.2.2 and 2.2.3 summarize the fluvial processes and bank processes responsible for bank retreat.
More details, especially on bank processes, can be found in Thorne (1982), Lawler et al. (1997), and
ASCE (1998a). The book section by Thorne (1982) is particularly interesting because of the field
observations enumerated.

2.3 Streambank erosion assessment models

This section will serve to briefly review the state of the art in streambank erosion modeling. This has
largely been done already by the ASCE Task Committee (1998b) on river width adjustment, especially
 Streambank erosion assessment models 21

on numerical modeling, but additions to the models they present will be shown here to emphasize models
that can be practically put into use for streambank erosion hazard mapping. The Federal Emergency
Management Agency (FEMA, 1999) has also produced a guide on "Riverine Erosion Hazard Areas".
Their document describes geomorphological, engineering, and mathematical models. Although, this
document claims riverine erosion hazard mapping to be feasible, it shows that mathematical modeling
has rarely been used to calculate riverine erosion hazards and the document is vague about the analyses
involved in geomorphological and engineering analysis. In general, geomorphological analysis is
described as the qualitative and sometimes quantitative understanding of the discharge and sediment
movements in the river reach of interest based on field observations and the use of historical information.
Most studies have used orthophotos to calculate mean erosion rates (see chapter 4.3 for details on use of
remotely sensed data). Engineering analysis usually draws on geomorphological analysis and goes
beyond it by quantifying discharge, sediment discharge and bank retreat with empirical formulas. This
section will present conceptual models and engineering analysis as well as geofluvial models.
"Geofluvial" refers to models that couple flow and sediment-routing models and treat bankline
adjustments mechanistically (ASCE, 1998b).

It must be remembered that the goal of the

modeling will be to calculate the
streambank erosion hazard. This hazard as
D
3.5 cm
thesis
defined by the Swiss Federal Office for
Water and Geology (Loat and Petrscheck,
1997) is the combination of the probability DFOWG
3.0 cm
of occurrence and intensity. In terms of
streambank erosion, this is interpreted to
be the intensity of a bank failure (figure
Figure 2.6. Erosion intensity, given as D (m), according to
2.6), given in meters, and its probability. If
the Swiss FOWG guidelines
the guidelines of the FOWG are strictly
followed, the erosion intensity is determined by the distance of the erosion perpendicular to the bank
face, or in the case of a channel change, the depth of the newly eroded channel. This distance is not very
meaningful, though, and it makes more sense to consider it to be the bank failure width, the distance
between the original bank top and the bank top after failure, at least in the case of non-braided rivers.

2.3.1 Engineering analysis and conceptual modeling

Richardson meander bend migration rate model (Richardson, 2002)

This model uses results issued from a 1D hydraulic model such as HEC-RAS. Average bottom shear
stress from the 1D results is corrected according to planform geometry to produce an estimation of the
outer bank shear stress. The average shear stress and outer bank shear stress are then used in sediment
transport relations. The difference in the average sediment transport and the outer bank sediment
transport is assumed to come from bank erosion. Then based on bank height, a bank erosion rate is
calculated. Richardson calibrated the sediment transport relations according to measured bedload and
was able to verify the bank erosion rates against historical data. This model is intended as a preliminary
22 Streambank Erosion Hazard Assessment and Mapping Methodology

assessment for determining bend migration rates. The model is limited at present because channel
curvature and slope are not incorporated into the 1D hydraulic model and thus channel curvature and
slope have to be updated periodically after non-negligible changes have occurred due to migration.

This model obviously makes the assumption that the river bed is stable and that deposition at bend inner
banks keeps up with outer bank erosion. This model does not take into account the effect of the resistance
of the bank material via an excess shear stress erosion equation, as in equation 2.8, and does not use bank
instability relations to determine failure widths. A relationship between bend migration rate and the
probability and widths of bend failures would have to be determined to use this model.

Rosgen streambank erosion potential method (Rosgen and Silvey, 1996)

Rosgen is well known for his stream classification system that has been widely applied. In his book that
describes stream classification (Rosgen and Silvey, 1996), he also has presented a streambank erosion
potential method. He proposes the calculation of bank erosion potential by ranking five different criteria
from 1 to 10 and summing the ranks to determine a bank erosion hazard index (BEHI). Those criteria
are: the ratio of bank height to bankfull water depth, the ratio of rooting depth to bank height, the root
density, the bank angle, and the surface protection. The sum of the ranks of the five categories are also
increased or decreased depending on the bank composition and stratigraphy. The BEHI must be
compared to the near bank shear stress index (NBS) to find an erosion rate. The NBS is the ratio of the
near bank shear stress to the average bed shear stress. A linear relationship can be sought for the relation
of the erosion rate to the NBS for each BEHI category, preferably by calibration. From field
observations, the BEHI for all banks within a reach can be determined and an erosion rate then calculated
for each bank.

This model again calculates an erosion rate and not bank failure widths and probabilities, as in the
Richardson model. It is also dependant on calibration.

Evaluation of potential instability in alluvial channels (Simon and Downs, 1995)

This method proposes site evaluations to characterize alluvial channel stability. For a state agency, it
gives recommendations on the hydraulic, including bridges, geomorphological, and vegetative variables
to be collected at a site evaluation for introduction into a state database. Enough site evaluations to
characterize the rivers throughout the region must be planned. The site evaluation should include the
status of channel evolution. Simon and Downs (1995) encourage the use of the Simon and Hupp (1986)
channel evolution model (figure 2.7). Their model is important because of the characteristic forms and
geobotanical evidence they suggest that characterize each stage of evolution, which can help the
observer to correctly identify channels in evolution and stabilized channel systems. They suggest that
the variables collected at site evaluations should be put into a Geographical Information System (GIS)
and ranking of the variables in terms of channel instability be done. A sum of the ranks of the different
variables can then be done to calculate a channel instability index. Such an index, in combination with
the vulnerability of land-use, can be used to prioritize reaches for further investigation. Further
investigation can then be based on numerical alluvial modeling, channel evolution models (figure 2.7),
regime equations (discussed further in chapter 3.7.4), or empirical relations to predict future channel
changes.
 Streambank erosion assessment models 23

This method is essentially the geomorphological analysis that precedes engineering analysis and/or
numerical modeling. Here again, the information in the geomorphological analysis alone is insufficient
to calculate bank failure widths and their probability.

Stage I. Sinuous, Premodified hc = critical bank height

h<hc
= direction of bank or
h bed movement

Stage IV. Degradation and

Stage II. Channelized Stage III. Degradation Widening
h<hc h<hc h>hc
floodplain terrace

h h h

slumped material

Stage V. Aggradation and Widening Stage VI. Quasi Equilibrium

h>hc h<hc
terrace terrace
bank
h
h bankfull
slumped
material
aggraded material aggraded material

Stages I, II
primary
knickpoint Stage III
Stage IV
top bank Stage V
precursor plunge
knickpoint pool direction of Stage VI
flow
secondary
knickpoint
oversteepened reach aggradation zone aggraded material

Figure 2.7. Channel evolution model (provided courtesy of Andrew Simon)

A GIS-based approach to mapping probabilities of river bank erosion (Winterbottom and Gilvear, 2000)

This method is quite attractive because it calculates the probability that a land cell adjacent to the river
will retreat. The method was originally developed by Graf (1984). Winterbottom and Gilvear (2000)
reused and adapted the method to study bank retreat on the regulated River Tummel, Scotland. The
original method (Graf, 1984) specifies that the probability that the i-j cell, pij, will be eroded is:

⎛ n ⎞

p i, j = f ⎜⎜ d l, d u, ∑ r⎟⎟ Eq. 2.13

⎝ t=1 ⎠
24 Streambank Erosion Hazard Assessment and Mapping Methodology

where dl is the lateral distance to the river, du is the upstream distance to the river, and r is the recurrence
interval of the high flows (1 through n) during the period for which the probability is to be evaluated. In
the case of a meandering stream, it would seem obvious that another variable should be present in
equation 2.13, that of the presence of the cell on the exterior or interior of a river bend. Winterbottom
and Gilvear (2000) recognized the other weakness of the Graf model: it must be applied to homogeneous
banks. They added a variable in equation 2.13: erc, which they called the erosion risk class. Based on
observations, they classified the banks of their reach into five erosion risk classes. They also discarded
the upstream distance to the river variable as it did not improve the probability calculation. Winterbottom
and Gilvear (2000) then calibrated the following equating using historic discharge records and river
channel changes:

⎛ n ⎞

log 10pi, j = log 10a0 + b1 log 10dl + b 2 log 10erc + b 3 ⎜ log 10 ∑ r⎟⎟
⎜ Eq. 2.14
⎝ t=1 ⎠

where a0, b1, b2, and b3 are calibration coefficients.

This model makes many assumptions. The river needs to be in dynamic equilibrium for such a model to
be applied. It should also be applied only to homogenous flow reaches. The recurrence interval variable
could be improved in the author’s opinion by considering the volume of water flow above a critical
discharge during a period. This model deserves further exploration in the author’s opinion as it is capable
of producing a probability of bank retreat for a given flood or floods.

Howard and Knutson (1984) and Venoge river meandering models

Howard and Knutson’s (1984) model was developed to explore

conditions to explain river meandering. A first approximation of
meander migration can be based on calibrated power law equations in
which migration rate is a function of the ratio of the radius of curvature
to the streamwidth, Rc/w, such as those of Hickin and Nanson (1975).
s They showed that lateral migration on the Beatton River tended to be
greatest when Rc/w is approximately 2.9. Such a relationship, though,
is insufficient to produce realistic meandering, as meanders translate
not only laterally, but also downstream. Howard and Knutson (1984)
ζ suggest that the actual migration rate for a bend must be determined
not only by the nominal migration rate, such as that of Hickin and
Nanson (1975), but also by an upstream weighting of the nominal
migration rates. They express this in the following equation:
d ∞ ∞ –1
point s
R 1 ( s ) = ΩR 0 ( s ) + Γ ∫ R0 ( s – ζ )G ( ζ ) dζ ∫ G ( ζ ) dζ Eq. 2.15
0 0

where R1(s) and R0(s) are the adjusted and nominal migration rates,
Figure 2.8. Channel centerline
and distances used in upstream respectively, at the location s (measured downstream from the
influence equations upstream starting point), R0(s-ζ) is the nominal erosion rate at a
 Streambank erosion assessment models 25

distance ζ upstream from s, Ω and Γ are weighting parameters, and G(ζ) is an upstream weighting
function (c.f. figure 2.8 for distance definitions). The upstream weighting function can be expressed as
an exponential decay equation:
– αζ
G( ζ) = e Eq. 2.16

where α is the decay parameter. Using this form of weighting necessitates a value of Ω=-1 in equation
2.15. Howard and Knutson (1984) propose relationships to determine α and Γ, as functions of hydraulic
parameters.

The Howard and Knutson model is limited by the fact that banks must be homogenous if the same
migration rate function is to be used throughout the channel network and the flow must also be
homogenous if the same values of α and Γ are to be used throughout the channel network, and that
erosion rates are determined rather than bank failure widths and their probabilities.

The Venoge model (Consuegra et al., 1999)

developed further Howard and Knutson’s
model. Rather than using two power
functions as did Hickin and Nanson (1975),
it is possible to approximate a bend
Nominal erosion rate (m/year)

migration rate function with a gamma type

function as follows (Consuegra et al., 1999):
b
( b – 1 ) –k2 X ( s )
R0 ( s ) = k1 X ( s ) e Eq. 2.17

where X(s) is defined as Rc(s)-Rc0(s), Rc(s)

is the radius of curvature in meters at point
s, Rc0(s) is the radius of curvature for which
no erosion occurs at point s, and k1, k2, and
Radius of curvature b are parameters that define the shape of the
function as shown in figure 2.9. If the
Rc0 Rcm Rc00 variation of R between point s and its
0
upstream point is assumed to be linear, an
Figure 2.9. Nominal erosion rate as a function of radius analytical solution can be found for the
of curvature (adapted from Consuegra et al., 1999)
integral in the numerator of equation 2.15:
s d s
– αζ – αd – αζ
∫ R0 ( s – ζ )G ( ζ ) dζ = ∫ ( aζ + c )e dζ + e ∫ R0 ( s – ζ )e ( dζ ) Eq. 2.18
0 0 d

where s is the distance to the upstream starting point, d is the distance to the point just upstream from the
point s (c.f. figure 2.8) , and a and c are the coefficients determined by a linear regression between the
26 Streambank Erosion Hazard Assessment and Mapping Methodology

values of R0(s) and R0(s-d). The spacing of channel points should be such that the linear regression
estimation is acceptable. The first term of the right hand side can be determined analytically as:
d d
– αζ
–e a
dζ = ------------- ⎛⎝ aζ + --- + c⎞⎠
– αζ
∫ ( aζ + c )e α α
Eq. 2.19
0 0

This equation allows the influence of the erosion at the point upstream from s to be calculated. Equation
2.18 recovers the influence of all other upstream points and diminishes their influence according to the
distance between point s and its upstream point and then combines it with the result of equation 2.19.
Equation 2.18 is important computationally as it eliminates the need to calculate the complete integral
from 0 to s at each channel point. Equations 2.18 and 2.19 were determined in the framework of the
Venoge modeling (Consuegra et al., 1999). Furbish (1991) treated the convolution of dimensionless
curvature integral of Parker and Andrews (1986) in a similar way as expressed in equation 2.18.

The Venoge model (Consuegra et al., 1999) also considers meander cutoffs by calculating a cutoff
distance, dcs, expressed as:
d(s)
dc s = k 3 W -------------------------- Eq. 2.20
max [ d ( s ) ]

where k3 is another calibration parameter, W is the average river width, and d(s) is the channel
displacement at point s. The ratio of d(s)/max[d(s)] weights the current river displacement as compared
to the maximum displacement. This allows the meander cutoff distance to be greatest when the meander
is actively progressing. If dcs intersects the downstream channel, a cutoff occurs and the channel
computational nodes are changed accordingly.

Consuegra et al. (1999) also attempted to resolve the problem of associating meander bend migration
rates to event migration rates. They hypothesized that a proportionality coefficient, k100, could be
multiplied against a mean meander migration rate to determine the migration distance due to a flood.

The Venoge model has 6 variables to calibrate: k1, k2, b, α, k3, and k100. These variables must be
calibrated for each homogeneous bank area. Twelve homogeneous bank areas were determined in the
Venoge study. The model needs to know the channel centerline (or thalweg) at different historical
periods. Good topographic maps or orthophotos must be available and preferably at long time periods.
Observations of bank failures must be at hand to give an appropriate value to k100 or orthophotos before
and after a significant event must be available for its calibration.

This model has a significant number of variables to calculate and the calibration procedure becomes
difficult if there are a wide variety of bank types. This model as in the other reviewed models assume
that flow conditions do not change. This model has attempted to calculate a bank failure width for high
flows. More work should be done to refine the formulation that translates meander migration rate into
event migration rates.

Lancaster and Bras (2002) river meandering model

This research model conceptualizes the calculation of the near bank shear stress responsible for bank
migration. A lateral force increment generated in meanders is calculated as a function of the mean depth,
 Streambank erosion assessment models 27

mean downstream velocity and the changing bed topography. Mean depth and downstream velocity are
based on continuity and Manning equations. Bed topography is estimated with a transverse bed slope
that depends on bend curvature, friction, and the median bed particle size. The lag between the channel
curvature and the secondary flow development, downstream lag (or phase lag) is estimated. The near
bank shear stress is then calculated according to the lateral force increment, phase lag, bank height, and
a dissipation length variable. They relate their dissipation length variable to the effect of bank roughness.
They calculate bank migration then as a function of an erodibility coefficient and the near bank shear
stress.

This model has not been applied to a real river case study with spatially varying data. It could be
imagined though to use it in connection with a 1D model, such as proposed in the Richardson model
above. The model also assumes a river in dynamic equilibrium such that mean flow conditions do not
change with time. Here again bank migration is calculated and not bank failure distances.

Use of conceptual models for determining the probability of bank failure width

The above models conceptualize bank retreat and/or channel changes. The geomorphological analysis
of the Simon and Downs method is a good starting point for determining channel instabilities. In
combination with their model, a catchment geomorphological analysis should be done. In Switzerland,
this catchment geomorphological analysis is called the phenomena map (Loat and Petrascheck, 1997).
A phenomena map, or geomorphological map, will indicate sediment sources and localize erosion or
aggradation problems (Kienholz and Krummenacher, 1995). This map in conjunction with a GIS
database based on the Simon and Downs method will give the elements necessary to determine if a river
reach is in dynamic equilibrium or if it is undergoing channel evolution. The other conceptual models
can then be used to estimate bank migration in situations of dynamic equilibrium.

The Venoge model (Consuegra et al., 1999) is the only model that tried to estimate an event bank retreat.
Their conceptualization of the event bank retreat should be further investigated. If the relationship
between mean annual bank retreat can be better correlated with event bank retreat, this will allow a more
widespread use of the above conceptual models to determine the probability of a bank failure width.

2.3.2 Geofluvial
With the incapacity of the above conceptual models to calculate bank failure widths and their probability,
it is necessary to turn to fluvial and bank failure mechanistic models (geofluvial) models to try to make
this calculation. The ASCE Task Committee (1998b) on river width adjustment reviewed twelve
numerical river width adjustment models. Of those twelve models, four models incorporate fluvial
entrainment and cohesive bank failures: Darby and Thorne ((1996) and Darby et al., 1996), RIPA
(Mosselman, 1992 and Mosselman, 1998), STREAM2 (Borah and Bordoloi, 1989), and WIDTH
(Osman, 1985 and Osman and Thorne, 1988). Since the writing of the ASCE review article, three other
important models have been developed: CONCEPTS - Conservational Channel Evolution and Pollutant
Transport System (Langendoen, 2000), CCHE1D (Wu and Vieira, 2002), and a modified version of
RIPA (Darby et al., 2002), called RIPA2 here.
28 Streambank Erosion Hazard Assessment and Mapping Methodology

The difference between RIPA and RIPA2 is that the bank erosion and failure mechanisms of the Osman
and Thorne (1988) model replace the previous bank erosion algorithms of RIPA.

The models CONCEPTS and CCHE1D are quite similar in their basic equations and algorithms. Both
models are 1D unsteady models providing the possibility to simulate flows with the full dynamic
equations or the diffusive wave approximation. Both models can handle the following hydraulic
structures: bridge crossings, culverts, drop structures, and a generic structure based on a rating curve.
None of the models in the ASCE (1998b) review implement hydraulic structures. Manning values as
specified by the user provide flow resistance estimations in CONCEPTS and CCHE1D. These two
models are 1D so secondary flow or lateral shear is not taken into account.

Sediment transport in CONCEPTS and CCHE1D is different due to some of the equations which are
different. Both models treat the 1D non-equilibrium sediment transport of cohesionless beds, although
CONCEPTS can also be applied to beds with cohesion. CONCEPTS uses 13 predefined classes for
sediment transport and assigns an appropriate sediment transport capacity formula to each of the
predefined classes. CCHE1D can have a variable number of sediment classes and the user can choose
among the SEDTRA module, Wu, Wang, and Jia’s formula, a modified Acker and White formula, or a
modified Engelund and Hansen formula for the sediment transport capacity calculation. Both models
treat suspended load and washload in combination with bedload. Both models employ mixing layer
models so sorting in the 1D direction can occur.

Bank erosion and failure in CONCEPTS is more developed than it is in CCHE1D. Both models estimate
fluvial entrainment based on excess critical shear and both models inject eroded sediment into the
sediment continuity equation as a lateral input. Both models verify bank stability based on planar failure
analysis. The CONCEPTS manner though is more sophisticated, though, in that pore-water pressure, and
hydrostatic confining pressure are taken into account. CONCEPTS also searches for the failure plane
that minimizes the factor of safety. CONCEPTS injects failed bank material into the sediment continuity
equation immediately while CCHE1D uses a virtual reservoir to release these basal sediments at the rate
of fluvial entrainment. As in all of the models in the ASCE review article, CONCEPTS and CCHE1D
do not deposit on the inner bank and do not permit layered banks.

All of the geofluvial models mentioned above incorporate fluvial entrainment and planar failure
analysis, although the planar failure analysis of CONCEPTS includes pore-water pressure and confining
pressure, and can calculate a non-toe bank failure. The Darby-Thorne model and the WIDTH models are
the only models to incorporate rotational failures. The chosen models only treat cohesive banks. No
existing model treats both cohesive and non-cohesive banks. The only model that tries to explicitly treat
the longitudinal extent of a bank failure is that of the Darby and Thorne (1996) model. Based on
probability distribution functions of bank parameters, they calculate a probability of bank failure. When
the probability of bank failure is greater than 0, they attribute a bank failure volume that corresponds to
the calculated probability of failure and the reach length.

The comparison of the four ASCE reviewed models along with RIPA2, CONCEPTS and CCHE1D in
terms of hydraulic flow and sediment transport capabilities is given in table 2.1.
 Proposed streambank erosion hazard methodology 29

Discharge Secondary Lateral Flow resistance Hydraulic

Model Planform Dimension variation flow shear Friction factor formula structures
Darby and Stepped time and
Thorne straight Quasi 2D hydrograph No Yes space variable Strickler No
arbitrary Stepped
RIPA single thread 2D hydrograph Yes No constant Specified No
arbitrary Stepped
RIPA2 single thread 2D hydrograph Yes No constant Specified No
Stepped
STREAM2 straight 1D hydrograph No No constant Specified No
Stepped time and
WIDTH straight 1D hydrograph No No space variable Strickler No

CCHE1D straight 1D Unsteady No No constant Strickler Yes

CONCEPTS straight 1D Unsteady No No constant Strickler Yes

Streamwise Transverse Bed Suspended

Model Dimension flux difference flux difference load load Sorting Bed material
Darby and
Thorne Quasi 2D Yes Yes Yes Yes Yes sand
sand and
RIPA 2D Yes Yes Yes No No gravel
sand and
RIPA2 2D Yes Yes Yes No No gravel
sand and
STREAM2 1D Yes No Yes Yes Yes gravel
sand and
WIDTH 1D Yes No Yes Yes No gravel
sand and
CCHE1D 1D Yes No Yes Yes Yes gravel
cohesive and
CONCEPTS 1D Yes No Yes Yes Yes non-cohesive

Bank failure Lateral

Model type Pore-water Non-toe failure extent
Darby and
Thorne planar/curved No No Yes

RIPA planar No No No

RIPA2 planar No No No

STREAM2 planar No No No

WIDTH planar/curved No No No

CCHE1D planar No No No

CONCEPTS planar Yes Yes No

Table 2.1. Comparison of geofluvial streambank erosion models capable of simulating cohesive bank
erosion and failures (adapted from ASCE (1998b))

2.4 Proposed streambank erosion hazard methodology

Stream managers need to know what streambank failures might happen suddenly during a flood event
and the channel migration that is likely to occur during the next 30-60 years. Flow conditions in natural
streams are such that only in the simplest streams will channel migration models have a high success rate
30 Streambank Erosion Hazard Assessment and Mapping Methodology

of predicting the future channel planform. Channel migration models should output results in terms of
the probability of a future channel placement. All uncertain parameters within a channel migration model
should be varied according to their Probability Distribution Function (PDF), possibly in the framework
of Monte-Carlo simulations, and the resulting channel planforms be used to calculate probabilistic future
channel planforms. Knowledge of those PDF’s is often poor, though, and sometimes important factors
are not taken into account. For example, the influence of woody debris on channel migration is very
significant, but this aspect has hardly been treated in channel migration studies. A tree that falls into a
river causes small damming or important flow diversions. A fallen tree can also reduce near bank shear
stresses and inhibit future channel migration. The effects of trees should also be incorporated into
channel migration models in terms of the probability that a fallen tree will influence channel migration.
Cutoffs are also extremely important for channel migration models.

Although knowledge of the future channel migration is beneficial, a truly probabilistic approach of
producing future channel planform in natural streams for a period of 30-60 years should in most cases
produce a wide variety of possible channel planforms. In this thesis, it is proposed to follow the logic of
the Swiss Federal Office for Water and Geology and to estimate the probability of a given bank failure
width to characterize the streambank erosion hazard (Loat and Petrasheck, 1997). Following this logic
is also helpful to land-use management. This is because of the importance of knowing what bank failures
could occur during a flood event so as to ensure significant setback of vulnerable objects. This logic of
land-use management has been shown in the streambank erosion intervention map in figure 1.4. The
problem, though, to follow this logic is that it is unknown how to calculate the probability of a given
bank failure width. This is understandable because of the near bank flow conditions that change with
time. Such non-stationarity makes it difficult to apply frequency analysis to bank failures to be able to
produce a bank failure width probability. Darby and Thorne (1996) calculated the probability of failure
of a bank considering only bank parameter uncertainty. Their probability is for a specific moment only
and does not consider the entire possible flow conditions.

A methodology is proposed here for assessing and mapping the streambank erosion hazard for single-
threaded streams with subcritical flow. This methodology will use the CCHE1D model, although
adaptations to the model will be needed. A collaboration with the National Center for Computational
Hydroscience and Engineering (NCCHE) was possible to adapt the CCHE1D model to the needs of this
thesis. CONCEPTS was also considered for the modeling, but since the source code was not available
for modifications, it was not chosen. The CCHE1D adaptations necessary for this thesis will be also be
outlined.

2.4.1 Streambank erosion hazard assessment and mapping

1. Catchment geomorphological mapping (Kienholz and Krummenacher, 1995) is the starting point. A
stream manager must understand the sediment and water flow through the stream reach. This will
permit detection of reaches which are most likely aggrading, degrading, or in equilibrium. All
locations where bank erosion or bed aggradation or degradation are observed are noted. Such
observations help also to determine the type of bank erosion processes occurring. Historical maps
and orthophotos are critical for understanding past channel migration. The geomorphological
 Proposed streambank erosion hazard methodology 31

mapping should be done in combination with the Simon and Downs (1995) channel instability
method. Variables for evaluating a channel instability index should be decided upon for entry into a
GIS. The results of geomorphological analysis and the channel instability index ranking can be used
to draw up a preliminary streambank erosion hazard map that outlines the maximum expected
channel movement during an extreme flood and potential unstable bank slopes.
2. Although geomorphological mapping is the recommended starting step, a possible regional model is
outlined here. The idea behind such a method would be to run it to prioritize stream reaches for
geomorphological mapping. The method would involve the characterization of the flow by stream
power and bank resistance by available information as follows:
• Stream power (Ω=γQS) or unit stream power (ω=γQS/w), where γ is the specific weight of water,
Q is the discharge, S is the channel slope, and w is the channel width, can be used to characterize
the flow condition of the river. It can be classified into three classes of high, medium, and low
stream power. For important variations of stream power, which will probably reflect
aggradational or degradational zones, the stream power rank could be increased to the next
higher category to reflect the possible channel instability. Approximations for discharge exist for
most rivers, or can be estimated by relationships that estimate discharge as a function of
catchment area. The bankfull discharge, or the one or two year return period discharge, can be
used. Slope and channel width can be estimated from an appropriately interpolated digital terrain
model (DTM) (see chapter 4.4 for more details).
• Bank resistance can be determined from available information. Minimally, bank height can be
estimated from a high resolution DTM. If surface and terrain LIDAR elevation points are
available, it is expected that the difference in point density along river banks can give an
estimation of the vegetation density and consequently the root reinforcement of the banks that
will resist fluvial erosion. Information on bank soils is rare. Geological information is more
common though and if available, categories of stability could be assigned to the geological units
that surround the stream. With the available bank information, a ranking system could combine
the information into a bank resistance index (similar to the BEHI of Rosgen and Silvey (1996)).
The bank resistance indices could then be classified into high, medium, or low bank resistance.
• A streambank erosion hazard index is then determined according to a hazard level matrix as
shown in figure 2.10.
This methodology has not been tested so its pertinency can not yet be guaranteed. Once again, it is
recommended to do geomorphological analysis and mapping throughout the region so that such a
regional method is not needed. This method is shown though as a possibility for prioritizing the
geomorphological mapping as its cost over a large region is considerable.
3. Based on geomorphological analysis and mapping, a decision must be made on where detailed
streambank erosion hazard mapping must be conducted. If a large rotational streambank failure
could occur or a planar failure on a high bank, it is suggested that a detailed geotechnical study be
performed. Such a failure falls more into landslide hazard assessment than streambank erosion
hazard assessment. For planar/slab failures on low to medium high banks (less than 15-20 meters
high), a geofluvial model can be used.
4. Input data for the geofluvial model must be collected. This includes:
• channel planform data
• cross section data including roughness
32 Streambank Erosion Hazard Assessment and Mapping Methodology

• bank data
• bed data
• tributary and boundary condition sediment and discharge input
Details on the data to collect and methods are given in chapters 4 and 5.

Bank
Resistance
Streambank Erosion
low Hazard Index
high

medium
medium
low

high

low medium high

Stream Power
Figure 2.10. Streambank erosion hazard index based on stream power and bank resistance

5. Simulations are done with a geofluvial model. Calibration of water depth should be done with
historically observed flood depths. Bedload transport parameters can be adjusted to reproduce
historical changes in bed level changes. Calibration of bank parameters can be done to reproduce
approximate historical channel migration. Simulations to produce bank failure widths is tricky. The
problem is that a frequency analysis of the bank failures will not be pertinent if the cross section and
planform change significantly altering the near bank shear stress. In the adaptations to the CCHE1D
model, it will be seen in the next section that cross section width and planform will be kept constant
to avoid non-stationarity associated with these changes. Bed level change can be allowed though to
try to take into account bed fluctuations during a flood. Simulations can be done with a long
historical period or with a generated discharge series.
6. An extreme flood is simulated so as to calculate the maximum bank failure widths expected.
7. The probability of a given failure width can then be evaluated with the simulated bank failure
widths. Failure width probability where bank protection exists involves the multiplication of the
probability of the bank protection failure and the failure width probability. Bank protection failure
probability depends on the protection work type and its condition and should be evaluated by
experts. Its foundation depth relative to the present or potential future bed depth is a very helpful
information for the expert in evaluating the stability of the work.
8. Mapping a streambank erosion hazard in terms of bank failure width and probability is difficult due
to the many small failures that occur. A very fine resolution map is necessary to show very small
bank failures. To overcome this while respecting the danger level calculation of the Swiss Federal
Office for Water and Geology directives (figure 1.3), it is proposed to map the danger level with a
line or polygon in the danger color (the most critical danger level according to figure 1.3) and to
 Proposed streambank erosion hazard methodology 33

map it to a specified factor of the mean annual erosion rate. This mean annual erosion rate can be
determined by integrating the frequency distribution of the bank failure width. The multiplication
factor will be determined by the mapper according to the readability of the map. By using the mean
annual erosion rate, the user can interpret the danger level in terms not only of instantaneous bank
failure width, but also in terms of the mean annual migration rate. Besides the mapping of the danger
level, the danger map should also show a maximum failure distance as calculated in step 6. The
distance to be mapped for a particular bank will be based on the maximum failure value throughout
the reach.

2.4.2 CCHE1D model modifications

CCHE1D (version 3.0) is a one-dimensional model. For the calculation of fluvial erosion in a
meandering river this raises the problem of the calculation of the bank shear stresses. A correction
method based on channel planform is used. Streambank erosion and bank failures originally
implemented in CCHE1D are for the purpose of determining channel widening in straight channels due
to bed instabilities. For this reason, channel planform is not kept track of. Input of the channel planform
coordinates will need to be implemented. The bank failure algorithm of the current CCHE1D model does
not take into account tension cracks. Tension cracks have a very significant effect on bank failure widths
so they must be implemented in the bank failure algorithm. The current version of the model does not
output bank failure widths. This also must be implemented. CCHE1D is described in detail by its user
and technical manuals, quick-start guide, and capabilities report that can be found on the internet (http:/
/www.ncche.olemiss.edu/index.php?page=cche1d_documents). Only the modifications to the model
will be described here.

Correction of bank shear stresses

Correction of the bank shear stresses is based on the Venoge planform modeling (Consuegra et al., 1999)
that was adapted for the correction of bank shear stresses in the Seymaz, Switzerland project (Consuegra
et al., 2000). Assuming bank migration rate is proportional to the excess bank shear stresses (Hanson,
1990), nominal and adjusted migration rates are replaced by the shear stress exerted at the bank toe in
equation 2.15 to give:
∞ ∞ –1

τ 1 ( s ) = Ωτ 0 ( s ) + Γ ∫ τ 0 ( s – ζ )G ( ζ ) dζ ∫ G ( ζ ) dζ Eq. 2.21
0 0

where τ0 is the nominal bank toe shear stress and τ1 is the planform adjusted bank toe shear stress. In
initial trials of this shear stress correction, using the Venoge planform modeling values for Ω=-1 and Γ=2
forced unrealistic shear stresses in straight sections. More realistic shear stress values in straight sections
occur with values of Ω=0 and Γ=1.
34 Streambank Erosion Hazard Assessment and Mapping Methodology

Nominal bank toe shear stress is based on a correction value that resembles the function shown in figure
2.9. The function implemented in CCHE1D calculates a shear stress correction factor with the following
equation:
b(s )
( b ( s ) – 1 ) – k 2 ( s )X ( s )
c b ( s ) = k 5 ( s ) + k 1 ( s )X ( s ) e Eq. 2.22

where cb(s) is the shear stress bend correction factor, k1(s), k2(s), and b(s) are the shape parameters, k5(s)
regulates the correction factor to be applied in straight river reaches, and X(s)=Rc(s)/w(s). The
parameters k1, k2, b, k5 are initially set at 0.785, 0.0845, 1.7, and 1 for all cross sections, respectively,
although they can be different for each cross section. This function is graphed in figure 2.11. Nominal
outer bank shear toe stress, τOB,0, at point s, is then calculated as:
τ OB, 0 ( s ) = k 4 c b ( s )ρgR h ( s )S ( s ) Eq. 2.23

where k4 is a coefficient that takes into account the bank angle, set to 0.75 for this thesis, and the other
variables are those defined in equations 2.1 and 2.22. Nominal inner bank shear toe stress, τIB,0, at point
s, is calculated as:

τ IB, 0 ( s ) = k 4 ( 2k 5 ( s ) – c b ( s ) )ρgR h ( s )S ( s ) : cb ≤ 2k 5 ( s )
Eq. 2.24
τ IB, 0 ( s ) = 0 : cb > 2k 5 ( s )

The inner and outer bank shear stresses are attributed to the left or right cross section side according to
the planform and then left and right nominal toe bank shear stresses are weighted according to the
upstream values with equation 2.21 to calculate the adjusted toe bank shear stress.

2
1.9
1.8
Correction factor, C b

1.7
1.6
1.5
1.4
1.3
1.2
1.1
1
0 1 2 3 4 5 6 7 8 9 10
Rc/w

Figure 2.11. Correction factor applied to outer bank toe shear stresses based on the ratio of radius of
curvature over width
 Proposed streambank erosion hazard methodology 35

Channel centerline input

To calculate the radius of curvature needed in the above equations, the model must have the coordinates
of each cross section center. The center of the channel is defined here as the middle point between the
bank tops. This definition is adopted because it is much easier to approximate the channel centerline
when digitizing it from orthophotos than the channel thalweg. The procedure for producing the
coordinates of the channel centerline, based on the procedure used during the Venoge model study, are
as follows:

1. Digitization of the channel centerline based on remotely sensed data, usually orthophotos.
2. Smoothing of the channel centerline based on the following algorithm:
• A downstream channel centerline point that is closer than a minimum distance is eliminated.
• An intermediate channel centerline point is added between a channel point and the following
downstream point when the distance is greater than a maximum allowed distance (figure 2.12).
This is done by calculating the point 1 based on the curcumscribed circle through points i, j, and
k, calculating point 2 based on the curcumscribed circle through points j, k, and l, and then
averaging these two points to get the new intermediate point to be placed in between points j and
k.
• The channel centerline points are looped
line
through over and over until all channel n ter
e
lC
points are within the maximum allowed n ne
a
distance. Ch
3. Channel longitudinal distance is calculated i
with an appropriate GIS tool and points
representing the channel centerline are
extracted. In this study 1 meter spacing was j 2
chosen. new intermediate point
4. Points representing real cross sections are
1 k
selected by the GIS. Other intermediate nodes
between real cross sections must also be
selected so that the computational network for
bank erosion sufficiently represents possible
longitudinal failure distances. Through GIS l
scripting, in this study Mapbasic® was used in
connection with Mapinfo® Professional®,
other intermediate nodes are selected so that a
Figure 2.12. Channel centerline smoothing
maximum distance between nodes is procedure
respected.
5. Each point is given a cross section identifier and those identifiers and their coordinates are exported
from the GIS.

Bank erosion and failure algorithm changes

1. Toe erosion is always lateral. In the previous version toe erosion could also deepen the bed.
36 Streambank Erosion Hazard Assessment and Mapping Methodology

Depending on the bed geometry, this caused oversteepening of the bank in some cases.
2. In the unmodified version of CCHE1D, banks are considered stable if any lower portion of the bank
has an angle between the toe and a bank point that is lower than the failure angle, β (figure 2.13).
This condition has been eliminated so that bank stability calculations always progress up to or
beyond the top of the bank.
3. Tension cracking has been implemented. Tension crack depth is determined by equation 2.12. If the
predicted depth is greater than half the bank height, the tension crack depth is limited to half the
bank height.
4. The failure angle is no longer calculated
bank top
according to equation 2.11. Failure angles
are computed at 10 cm intervals starting at
20 cm from the bank top until a minimum
factor of safety is found. The failure angle
giving the smallest factor of safety is
retained.
5. Bank failures are output from the model.
failure
52.3°angle, β
The location given by the cross section
identifier and the bank side, the bank toe angle
43.2°
failure distance, and the time are output. bank toe
Other hydraulic variables, beyond the
Figure 2.13. Bank stability is no longer guaranteed by
standard model output variables, can also
the toe angle being lower than the failure angle
be output for analysis purposes, such as
mean or maximum event toe shear stress.
6. No cross sectional change occurs
t=0 Virtual Reservoir 1
due to bank failure (figure 2.14). empty
Initial bank or bed erosion is
allowed to steepen or heighten
the bank. If it is significant
Toe Erosion Bank Failure
enough, the bank will eventually
fail. In the unmodified CCHE1D,
the bank geometry is then 2 Toe Erosion 3 empty
changed according to the failure.
For this thesis, bank geometry is
left unchanged. The bank is left
New Failure
unstable. The bank will not fail
Figure 2.14. Cross sectional adjustments in the modified
again, though, until the virtual
CCHE1D model. Toe erosion destabilizes the bank causing bank
tank that holds the failed failure. Bank geometry does not change although the bank
sediment is empty. In this way, failure material is added to the virtual reservoir. Subsequent toe
the virtual tank models the basal erosion empties the reservoir so a new bank failure can occur.
sediments. Until the basal sediments are eroded away, another bank failure can not occur. This
conceptualization is appropriate for a discharge controlled bank retreat. If channel migration is to be
studied, then cross section updating would be necessary. But in the case where we want to calculate
the probability of failure for the modeled bank, it makes sense to adopt this modeling procedure to
keep the cross sectional and planform conditions the same.
 Chapter 3

The Venoge River

3.1 Introduction
The Venoge River in the Vaud Canton in Switzerland was chosen for streambank erosion monitoring. This
choice was not arbitrary as the Venoge River is one of the closest rivers to the EPFL and has been the
object of an important hydrologic and hydraulic study conducted by the HYDRAM laboratory from 1996-
1998 (Lance et al., 1997, Lance and Consuegra, 1997, and Lance and Consuegra, 1998). More important
than the proximity of the river to the EPFL, the Venoge River is allowed to meander through three major
alluvial zones and partially in a few other reaches. Streambank erosion occurs in an observable manner
each year giving the possibility to record significant erosion during the time of a PhD thesis.

This chapter will give a brief introduction to the Venoge catchment and the morphology of the Lower
Venoge River.

3.2 Location of the Venoge catchment

The Venoge catchment is located in the west of Switzerland (figure 3.1). The topographic drainage basin
has an area of 238 km². The upper catchment to the northwest is dominated by the karstic Jura mountains
while the majority of the catchment lies on the Swiss Plateau. The catchment outlet is on Lake Geneva.

37
38 The Venoge River

0 20 40 km

Venoge Catchment

Reproduced with permission from swisstopo (BA068334)

Figure 3.1. Venoge catchment topographical limit (background map: CP500 ©1997, Swisstopo)

3.3 Geology
Geologic history of the Venoge catchment

To understand the type of sediments that are found in the Venoge catchment, and especially in the bed
and the banks of the Venoge River, it is important to understand the geologic history of the catchment
and the type of rock and deposits encountered. The majority of the catchment lies on the Swiss Molasse.
This sandstone plateau is the result of the erosion of the Alps during its formation during the Oligocene
period. The sediment accumulation has a depth of 2.5 km (Grattier, 1980).

The Stampien formation of the Swiss Molasse (figure 3.2) is largely made up of lacustrine sediments,
while sandstone proportions are greater in the Aquitanien and the Burdigalien formations due to the
increased marine influence during these epochs (Grattier, 1980).

During the Pliocene, the folded Jura mountain belt detached from the Alps. As the folds separated from
the Alps, the space in between was left for the formation of the Swiss Plateau (Collet, 1955). During the
Quaternary, the Rhone glacier invaded the plateau, eroding the layers of Molasse. The moraines of the
Venoge catchment are essentially those left from the Würm glacial period (Grattier, 1980).

After the glacial periods, during a period of tundra, the landscape was transformed by the eroding action
of water and wind. The resulting layers of restructured silts forms what is the present day soil of the
 Geology 39

Venoge catchment. Basins were filled up with colluvial material and formed marshes. Morainic material
was picked up and deposited to form the alluvial zones of the lower plains (Grattier, 1980).

Venoge
catchment

BURDIGALIEN
AQUITANIEN
STAMPIEN (pudding stone)
LIMESTONE-JURA
OVERLAP (transported molasse)
PRE-ALPS

Figure 3.2. Geologic substrates of the Vaud Canton, Switzerland (adapted from Grattier, 1980)

Characteristics of the geologic deposits

Figure 3.3 shows a description of the geologic deposits based on the Swiss geologic maps (1:25’000).
40 The Venoge River

Lower Venoge
River
Upper Venoge and
Venoge Tributaries

Figure 3.3. Bedrock and quaternary sediment of the Venoge catchment (data from Schneider, 2001)

Grattier (1980) described the deposits in five categories: cretaceous Jura limestone, oligocenic molasse,
glacial deposits, silt and decarbonated colluvium, and alluvium and carbonated colluvium. The
following is a summary of the characteristics of each of these deposits as described by Grattier (1980):

1. Cretaceous Jura limestone: These limestones are very hard, clastic, and cut into compartments by
their faults. Fragmentation is slow, although the clasts have favored a karstic erosion.
2. Oligocene molasse: The molasse is a feldspatic-micaceous sandstone cemented by calcite or clayey-
calcite. In terms of its petrography, it is rich in quartz and contains a lot of debris from crystalline
and carbonated rocks, micas, chlorites, and other varied heavy minerals. In terms of its facies, the
molasse is composed of alternating benches of marl-marly sandstone and sandstone. As mentioned
above, the proportion of sandstone usually increases with decreasing age. The molasse is usually
rich in silt. Marly molasse is especially rich in fine silt particles. The proportion of calcite varies
from bench to bench; it is less frequent in the clayey molasse (red molasse). The marls are often rich
in calcite.
3. Glacial deposits: Morainic deposits are quite variable in their spatial distribution and thickness. All
of the degrees of mixes are found between the restructured molasse and the deposits of the Rhone
and Jura. In terms of petrography, the gravels and pebbles are dominated by the limestones, and the
sands and associated heavy minerals come mainly from the molasse. The three main facies are the
 Soil and land use 41

bottom, fluvio-glacier, and the superficial moraine. The bottom moraine is usually rich in silt and
poor in gravel. The fluvio-glacier moraine has lenses of sand and gravel that were deposited in the
front of the glacier. The superficial moraine is an intermediate moraine between the bottom moraine
and the fluvio-glacier in which there is a lot of gravel and the different particle proportions are
relatively equal. Figures A.1 and A.2 in the appendix shows textural triangle representations of the
particle distribution of various samples taken from molassic and morainic deposits from the Vaud
canton.
4. Silts from restructuring and decarbonated colluvium: These silts have very little gravel and were
formed from the restructuring of older deposits. Their composition is more silty for the Stampien
derived formations, and more sandy-silt when derived from the Aquitanien formations. This type of
deposit occurs in a very limited way (either in its extension or its depth).
5. Alluvium and carbonated colluvium: These deposits are post-glacial. They are usually finer except
near streams. In the alluvial plains, a finer silt layer, with varying degrees of organic matter, often
covers a courser layer of sands and gravels. In the marsh basins, the succession from bottom to top is
often a glacial clay, lacustrine lime, peat, and finally an organic clayey silt. Laterally, the finer
alluvial deposits give way to bottom slope colluvium.

3.4 Soil and land use

The land-use of the Venoge catchment is dominated by cultivated fields (figure 3.4). Forests dominate
the slopes of the Jura mountains. The catchment is essentially rural as the percentage of urbanized
surface is only 8.6%, and of that only 3.5% was considered impermeable in the 1997 mapping.

Table 3.1 shows the percentages of the different land-use categories for 1985 and 1997 as extracted from
the GESREAU1 database. There is very little change from 1985 to 1997, and those being the conversion
of some farmland to urban development.

Land-use category 1997 1985

Urban impermeable 3.5 3.2

Urban permeable 5.1 4.7

Forests 32.8 32.9

Cultivated Fields 51.1 51.5

Prairie, Pasture 6.8 7.1

Vineyards 0.3 0.3

Wetlands, lakes 0.2 0.2

Scree 0.1 0.1

Table 3.1. Venoge catchment land-use and changes from 1985 to 1997 (Data extraction from
GESREAU)

1. GESREAU is the integrated water management database of the Vaud canton in Switzerland. See http:/
/hydram.epfl.ch/gesreau/menu.htm for more information.
42 The Venoge River

Land-use
Urban impermeable
Urban permeable
Forest
Cultivated fields
Pasture
Vineyards
Wetlands, lakes
Rocks (scree)

Venoge subcatchment
Venoge catchment
Lower Venoge River

0 5 10

kilometers

Figure 3.4. Venoge 1997 land-use (data from the Swiss Federal Office of Statistics (OFS))

Prétorian (1994) summarizes the soils of the Venoge catchment to be generally silt or fine silt. On
average, the particle size distribution contains 15% clay, 28% fine silt, 23% large silt, and 35% sand. The
pH varies from 5.3 to 7.8, the percentage of calcium from 0 to 20%, and the organic material from 2.5
to 3.8%. In general, the soils of the Venoge are strongly erodible with a factor K of Wischmeier between
0.2 and 0.4. The most erodible zones are the Upper Senoge, and on the Venoge the areas of Eclépens,
Lussery and downstream from Cossonay (percentage of silt greater than 60% and little organic matter).

3.5 Topography and the hydrographic network

The limits of the Venoge catchment as determined within the GESREAU database are shown in figure
3.5. The total surface is 238 km2. The maximum elevation within the catchment, 1680 m, is at the Mont-
Tendre. The Lake Geneva typical water surface elevation at the outlet is 372 m. There are three relatively
distinct topographic zones of the catchment (figure 3.5). The upper catchment formed by the Jura
mountains has several sources connected to the karst system. In the middle catchment, water from the
Jura sources and from the marshes at the foot of the Jura then runs through the impermeable molassic
plateau where the waters of the Veyron join the waters of the Venoge at the Tine de Conflens. Finally,
the Venoge flows through its alluvia in the lower catchment.
 Topography and the hydrographic network 43

Elevation
m.a.s.l.
1680
1584

1356

1032

784

555
370
Venoge
subcatchment
Venoge
hydrographic
0 3 6 network
kilometers

Reproduced with permission from swisstopo (BA068334)

Figure 3.5. Venoge digital terrain model, subcatchments, and hydrographic network. Relief data is
from the DTM25 ©1995, Swisstopo, and catchment and hydrographic network data come from
GESREAU.

3.5.1 Catchment topography

The hypsometric curve of the Venoge catchment (figure 3.6) shows a basin in the monadnock phase
according to the descriptions of Strahler (in Maxey, 1964). The monadnock phase is characterized by a
resistant rock body, in this case the Jura mountains, and a relatively subdued relief, the Swiss molassic
plateau. The mean elevation of the catchment is 700 m.a.s.l. and the median is 600 m.a.s.l. The mean
slope is 10.9% while the median is 7.1% (figure 3.6).

The coefficient of compactness of Gravelius (Musy and Higy, 2004) indicates the effect of the form of
the basin on the streamflow. It is defined as:
P
K G = 0.28 • ------- Eq. 3.1
A

where P is the catchment perimeter (smoothed) and A is the catchment area. For a round basin, KG is
approximately 1, while for a very long rectangular basin KG is around 1.6 (Musy and Higy, 2004). The
smoothed Venoge perimeter is approximately 76 km. Using this value and the catchment area, the value
of KG is 1.4. This demonstrates that the Venoge is a relatively long basin. Such basins typically produce
attenuated hydrographs, longer concentration times, and lower peak discharges.
44 The Venoge River

1700 42
Elevation
Slope(%)
1500 36

1300 30
Elevation (m.a.s.l.)

Slope (%)
1100 24

900 18

700 12

500 6

300 0
0 20 40 60 80 100
Basin above elevation (%), Slopes above elevation (%)

Figure 3.6. Venoge catchment hypsometric curve and slope frequency.

3.5.2 Subcatchment limits and stream network

The GESREAU subcatchments are shown in figure 3.5 and the main subcatchments are shown in figure
3.7. The Veyron is the principal subcatchment of the Venoge collecting waters from the marshy plain at
the foot of the Mont-Tendre slopes. The Upper Venoge collects the remainder of the waters issued from
the Jura karst. In his report on the hydrologic behavior of the Venoge catchment, Prétorian (1994),
estimated the real Venoge catchment to be 15.5 km2 greater than the topographic catchment according
to hydrogeologic studies. The limit of the Venoge catchment from Mont-Tendre to the northern most
point of the catchment should most likely be farther northwest stealing a part of the water from the Orbe
catchment. There seems to be three levels of sources that feed the Venoge: 630-720 m.a.s.l., 850-950,
and 1000-1150 as visible in the 1:25’000 geological maps of the Jura slopes, while the Veyron is only
fed by the bottom level of sources at around 700-720 m.a.s.l. During low summer flow, the water table
falls below 700 m.a.s.l. so that the Malagne et the Etremble become dry and only the Morand feeds the
Veyron. The lower sources of the Upper Venoge continue to flow during low summer flows.

When the Veyron (86.6 km2) joins with the Upper Venoge (43.6 km2), combined they form 130 km2 of
the 238 km2 of the Venoge catchment. During the inter-glacial period Riss-Würm, these two streams
flowed through the Sarraz canyon (Custer, 1935). Post-Würm, due to moraines that filled the passages
(Sarraz et Mormont) toward the Nozon of the Orbe catchment and the regressive erosion that eroded the
 Topography and the hydrographic network 45

lower Venoge valley after the lowering of Lake Geneva, the Lower Venoge River was able to capture
the Veyron and the Upper Venoge River (Custer, 1935).

The Lower Venoge valley is elongated and receives no major tributary. The Venoge essentially flows
through its alluvium. Besides the areas surrounding the Venoge (intermediate catchments in figure 3.7),
the other significant tributaries of the Lower Venoge River are the Molomba (13.5 km2), the Senoge
(22.5 km2), the Vaube (6.7 km2), and the Arena (7.3 km2). The lengths, positions, and slopes of the
above mentioned tributaries are indicated in the longitudinal profile of the Venoge and its tributaries in
figure 3.8. As can be seen on the graph, the only torrential river slopes occur on the Jura tributaries of
the Upper Venoge.

Voualève
Voualève

Upper
Upper Venoge
Venoge

Morvaz
Morvaz
Les Mousses
Les Mousses
Chergeaule
Chergeaule
Veyron
Veyron Venoge
Venoge

Gebre
Gebre

MORAND (Le)
MORAND (Le) Molomba
Molomba

Lamponex
Lamponex
Malagne
Malagne

Etremble
Etremble
Senoge
Senoge

Vaube
Vaube

Arena
Arena
0 3 6

kilometers

Figure 3.7. The Venoge River and its tributaries and subcatchments.
Background map: CP500 ©1997, Swisstopo.

The drainage density is indicative of the type of runoff that occurs in the basin. Those with permeable
structures will have a low value, while those that occur in cohesive materials will usually have a high
value due to the important superficial runoff (Musy and Higy, 2004). The drainage density is defined as:

∑L
D d = ---------- Eq. 3.2
A

where L is the length of an individual stream and A the area of the basin. Of course, depending on how
the beginning of the stream is defined, the resulting value can easily change by 2 or 3 times. The length
of the stream network for the Venoge basin according to the GESREAU system is 208 km, giving a value
46 The Venoge River

of 0.9 for the drainage density. This value is very low indicating the important infiltration that occurs in
the karstic and morainic formations. With such an important infiltration, a long response time is
expected.

1200

Venoge

Chergeaule
1100 Veyron
Tributary end points

Mousses
1000

900

Malagne
Elevation (m.a.s.l.)

800

Etremble
Cressonniere

Morvaz

Lamponex
Voulalève
700

Molomba
600 Senoge bre
Gè
Arena
Vaube

500

400

300
0 10000 20000 30000 40000 50000
Distance to Lake Geneva (m)

Figure 3.8. Longitudinal profile of the Venoge River and its important tributaries. Only starting and
ending elevations are represented. Elevations based on the DTM25 ©1995, Swisstopo.

3.6 Climate and hydrologic regime

The climate of the Swiss plateau is considered to be humid temperate, with a continental tendency. The
climate of the Venoge catchment deviates from a typical continental climate due to the effect of the Jura
mountains. The ratio of summer precipitation to winter precipitation is approximately 1 for the Jura
mountains and at the foot of the Jura there can be more winter rainfall than summer rainfall. This
explains the small variation in monthly rainfall in figure 3.9. This figure equally shows the water deficit
that can occur in the summer months due to an evapotranspiration potential that is greater than the
rainfall. Mean annual Venoge precipitation is approximately 1250 mm, varying from 950 mm towards
Chavornay (inland from the Jura and the lake) to 1800-2000 at Mont-Tendre (see chapter 5 for more
details). Temperature variation over the catchment is represented in figure 3.9 by the Changins and Dole
monthly mean temperatures. Although slightly outside of the Venoge catchment, they represent
relatively well the elevation effect felt in the Venoge catchment. The Changins mean annual temperature
is 10.1 °C for an elevation of 430 meters and the Dole mean annual temperature is 3.6 °C for an elevation
of 1670 meters. The mean discharge in terms of depth is 600 mm. Assuming no other rainfall losses
(deep percolation, hydrogeologic losses,...), this would give an ETR of approximately 650 mm. This is
higher than what is indicated by the Swiss hydrological atlas (approximately 600 mm) (Spreafico, 1992),
 Channel morphology 47

but is realistic due to the important effect of the Joran winds that descend from the Jura mountains
(Primault, 1972).

Avg. Precip. PET - PM Pully Q Ecublens

160 Changins Temp. Dole Temp. 30

140 25
Precipitation, PET, Discharge (mm)

120 20

100 15

Temperature (°C)
80 10

60 5

40 0

20 -5

0 -10
1 3 5 7 9 11
Month

Figure 3.9. Seasonal variation of precipitation, potential evapotranspiration (PET calculated with the
Penman-Monteith formula), discharge, and temperature. Data is from the period 1983-2003.

The hydrologic regime of a catchment can be defined by its variation of discharge. Usually this variation
is examined by calculating each monthly discharge coefficient which is the mean monthly discharge
divided by the mean annual discharge. In the Hydrologic Atlas of Switzerland (Spreafico et al., 1992),
16 regime classifications are proposed. The Venoge is classified as a "nivo-pluvial jurassien" regime.
This classification is based on the Venoge Lussery discharge data collected from 1949-1978. Its name is
indicative of the discharge generating mechanisms of the catchment, the Jura snow and rain. The typical
nivo-pluvial jurassien regime has a peak monthly discharge in April and a minimum between July and
October. Within this classification, the Venoge has the most marked difference between high spring flow
and low summer flow. Figure 3.9 shows the discharge variation for the Venoge based on measured
values (Ecublens 1983-2003) near the outlet. The difference between the high flows in March and low
flows in August is even greater than that of the Venoge at Lussery, and thus no longer follows very well
the classification of nivo-pluvial jurassien. The classification of pluvial jurassien is better, but does not
yet capture the significant August low flow (monthly discharge coefficient of 0.26).

3.7 Channel morphology

In the above sections, the Venoge catchment has been described in its entirety. In this section, only the
channel morphology of the last 25 kilometers of the Lower Venoge River, the river reach of interest, will
48 The Venoge River

be considered. Channel morphology is dependant on the driving variables of water and sediment input
and the boundary conditions (valley slope and topography, bed and bank material, and riparian
vegetation). The resulting channel form is characterized by its longitudinal profile, cross sectional
geometry, and its planform (Thorne, 1997).

3.7.1 Channel type

Leopold and Wolman (1957) defined channel forms as ranging from straight, single threaded channels
through to multithread, braided systems. They classified a straight stream as having a sinuosity of less
than 1.1, a slightly sinuous stream between 1.1 and 1.5, and a meandering stream greater than 1.5.
Besides these three main categories of straight, meandering, and braided streams, a forth classification
has been deemed necessary by Nanson and Croke (1992), that of anastomoses streams. These streams
are low energy compared to braided streams which are usually high energy.

The Lower Venoge, as

demonstrated by the
orthophotos in figure 3.11,
can be classified as a 3, 7,
or 8 stream (figure 3.10).
Some areas of the Lower
Venoge do exhibit middle
bars so that the 3rd
classification is sometimes
appropriate, although the
Lower Venoge is mainly a
meandering stream
according to the 7th
classification. Meandering
must also be classified as
active or passive (Thorne,
1997) or anchored,
entrenched, or free
(Bravard and Petite, 1997).
The Lower Venoge was
mostly free, although
confined by the valley
sides, but in the past
century has been mostly
fixed in place by bank
protection. This leaves a
minority of reaches to be
Figure 3.10. Classification of channel pattern (from Knighton,1998 based
on the original work of Schumm, 1981). (Reproduced by permission of truly freely meandering.
Stanley Schumm and Edward Arnold (Publishers)).
 Channel morphology 49

Chute cutoff

Figure 3.11. Two of the three alluvial zones of the Lower Venoge River. To the left - Lovataire and to
the right - the islands of Bussigny. Orthophotos from 1996 EPFL student field study.

In figure 3.11, especially in the left picture, the Venoge exhibits compound and multibend loops.
Another important feature of the Venoge morphology, is the meander cutoffs that occur. In the left
picture, we see quite distinctly a meander cutoff that appears to be a chute cutoff. Cutoffs are either neck
cutoffs, in which two bends of a loop meet so as to cutoff the meander, or chute cutoffs, in which
significant overbank flow will erode a path (often following depressions) to cutoff the bend. Figure 3.12
shows four meander cutoffs that occurred in the Bussigny islands alluvial zone, three of which occurred
50 The Venoge River

since 1964. The 1st cutoff is very small and occurred between 1996 and 2002. The second and third
cutoffs occurred during the period 1964-1996, and the timing of the fourth one is unknown. The first
three seem most likely to be chute cutoffs, while the fourth one is less clear. The orthophoto of figure
3.12 shows the important downstream planform modifications that occur after a meander cutoff, in
addition to the significant slope change that occurs.
Reproduced with permission from swisstopo (BA068334)

1996

Reproduced with permission

from swisstopo (BA068334)

0 100 200

meters

Figure 3.12. LIDAR DTM (DTM-AV ©2004, Swisstopo (DV043683)) showing four meander cutoffs in
the Bussigny islands alluvial zone. The 1964 orthophoto shows the 2nd and 3rd meander cutoffs.

3.7.2 Planform
Being a meandering river, largely single thread, it is standard practice to describe the Venoge planform
by its sinuosity. In theory (figure 3.13), defining meander planform geometry is simple, but in practice
is much more difficult because of asymmetric planform patterns (figure 3.11). For simple bends in
natural freely meandering zones of the Venoge, point bar spacing is in the range of 50-80 meters so that
the wavelength is in the range of 100-160 meters. The sinuosity index is defined as the real channel
length divided by the straight line. For the simple Venoge bends, it is in the 1.1 to 1.2 range. If the straight
line distance that is used is the valley centerline, then for the Lower Venoge, the sinuosity is 1.5. This is
a result of multiple loop bends and meandering that crosses the valley bottom. Figure 3.14 shows the
distribution of the radius of curvature of the Lower Venoge River. The median radius of curvature is 170
meters.
 Channel morphology 51

To express the ranges in

sinuosity possible over
various reach lengths,
Lancaster and Bras
(2002) developed a new
sinuosity-based method
for analyzing planforms
of meandering channels.
Their method consists of
calculating a mean
channel sinuosity and its
variance for every pair
of digitized channel
Figure 3.13. Meander planform geometry (adapted from Thorne, 1997)
points at a given channel
length. As the channel
length increases, they suggest that the mean and variance will reveal two or three characteristic channel
lengths. The first characteristic length they propose is that of the single bend length which is revealed by
the mean sinuosity that increases less rapidly. The second and third characteristic lengths are found by
peaks in the variance. They suggest that a first peak in the variance will determine the characteristic
length of a long simple or compound bend loops. A second peak in the variance will determine the
characteristic length of multibend loops.

100

90

80

70
Percentage Smaller

60

50

40

30

20

10

0
1 10 100 1000 10000 100000 1000000
Radius of Curvature (m )

Figure 3.14. Distribution of radius of curvature for the Lower Venoge

River
52 The Venoge River

This new sinuosity-based method was

programmed in the Delphi® programming
language to analyze a point file in text
format issued from a digitized river
centerline. The analysis was done on the
Lower Venoge from the start of zone 1
(figure 3.15) to its outlet. Table 3.2 shows
the characteristic lengths for the 1996 and
1839 Lower Venoge River which are
derived from the information in figure 3.16.
It is somewhat surprising to see that the
sinuosity of the Lower Venoge River has
changed little from 1839 to 1996 with
respect to the upstream channelization. The
multibend characteristic length is shown to
be longer though in 1996. The 1839
multibend characteristic length is similar to
the freely meandering Roujarde alluvial
zone so it is assumed that the channelization
of the Lower Venoge caused the change in
the multibend characteristic length. The
analysis was also done for the Roujarde
alluvial zone (figure 3.15), the most freely
meandering of the alluvial zones. This zone
shows simple bends which are shorter
compared to the entire Venoge, although the
compound and multibend lengths are
comparable to the natural 1839 state. The
sinuosities though are much greater. The
Reproduced with permission
greater sinuosity, especially for the from swisstopo (BA068334)
multibends, reflects the space the Venoge
Figure 3.15. A view of the sinuosity of the Lower Venoge
has to move laterally in the Roujarde area. River. Background: DTM25 © 1995, Swisstopo.
In other sections of the Venoge valley, the
river is more confined by the valley side walls. Based on the sinuosity of 1.2 and the length z of 4.5w for
the Roujarde zone, the mean wavelength, L, is 7.5w. This is lower than the value suggested by Richards
(1982 in Thorne, 1997) of L=12.34w for a meandering river, but is in line with a sinuosity of 1.2 (Chorley
et al., 1984 in Thorne, 1997).
 Channel morphology 53

1996 1839

Lower Venoge River Zone 1 to Lake Roujarde Alluvial Zone Lower Venoge River

length (m) sinuosity length (m) sinuosity length (m) sinuosity length (m) sinuosity

simple 100-120 1.11 100-120 1.15 80-90 1.19 120-140 1.15

compound 260 1.24 260 1.33 230 1.55

multibend 560 1.44 560 1.53 460 1.88 460 1.41

Table 3.2. Lower Venoge River sinuosity and characteristic bend lengths

2.0 0.25
mean
1.8 mean1839
variance
1.6 variance1839 0.20

1.4
mean sinuosity

1.2 0.15

variance
1.0

0.8 0.10

0.6

0.4 0.05

0.2

0.0 0.00
1 10 100 1000
channel length (number of channel widths)

Figure 3.16. Analysis of mean sinuosity and its variance on the Lower Venoge River. Mean channel
width is considered to be 20 meters.

3.7.3 Longitudinal profile

The longitudinal profile is an extremely important feature in geomorphological analysis. It is often the
breaks in slope that cause the aggradational or degradational tendency of a stream. Its form will also give
an idea of its equilibrium status. As can be seen in figure 3.17 (or in more detail in figure 4.17), the Lower
Venoge River has a concave form with the only significant slope breaks being those imposed by instream
weirs. This concave shape indicates most likely a stable slope, which is largely confirmed by the few
changes in the thalweg elevation from cross sections taken in 1996 and 2004.
54 The Venoge River

460 0.05
1996 2004 slope 2004
450 0.045

440 0.04
0.035
elevation (m.a.s.l.)

430
0.03
420

slope
0.025
410
0.02
400
0.015
390 0.01
380 0.005
370 0
0 5000 10000 15000 20000 25000
distance from outlet (m)

Figure 3.17. Longitudinal profile of the Lower Venoge River

3.7.4 Cross section characteristics

Cross section characteristics are usually expressed in terms of the bankfull channel width and depth. At-
a-station hydraulic geometry is often used to express the width, depth, and velocity variation with respect
b f m
to discharge: W = aQ , D = cQ , and U = kQ (Leopold and Maddock, 1953), in which a, b, c,
f, k, and m are the coefficients of the power laws. Leopold and Maddock (1953) also introduced the idea
of downstream hydraulic geometry suggesting that downstream geometry (W, D, and U) can be
determined based on the evolution of the mean or bankfull discharge. There is considerable variability
in the at-a-station hydraulic geometry (HG) coefficients, and this even for the same stream reach. This
has driven various authors (Jowett, 1998, Lamouroux and Capra, 2002, in Stewardson, 2005) to make
attempts at defining reach hydraulic geometry based on reach average values. Stewardson (2005)
concludes that hydraulic geometry relations using reach averaged width and depth, as well as the
coefficient of variation of width and depth and velocity provide an efficient description.

Reach averaged width and depth as well as the standard deviation were determined for the four zones
shown in figure 3.15 and are given in table 3.3. These results were developed from HEC-RAS
simulations using cross sections from the HYDRAM Venoge study (Lance et al., 1997) and an initial
HEC-RAS project developed during a postgraduate study (Rakotondranaly, 2001). As can be seen in
table 3.3, bridge cross sections were excluded in the estimation of a reach averaged width and depth. The
results of table 3.3 also excluded cross sections which had overbank flow. Four flows approximating the
0.5, 0.75, 1, and 2 year return period flows were simulated. Figures 3.18 and 3.19 graphically show the
reach averaged widths and depths of table 3.3 and show possible HG relations fixing the b and f
coefficients at 0.5 and 0.37 (Hey, 1997), respectively. It can be seen that zone 4 of the Lower Venoge
River distinguishes itself from the three other zones.
 Channel morphology 55

number number used standard average standard

excluding excluding deviation hydraulic deviation
zone bridges overbank flow Q (m3/s) avg width (m) width depth (m) depth
25 23.1 18.9 6.8 1.10 0.31
21 26.9 17.5 5.4 1.22 0.33
20 30.8 17.2 4.2 1.30 0.34
1 31 19 36.2 18.1 4.7 1.37 0.37
18 24.9 16.9 6.7 1.28 0.38
17 29 17.7 7.1 1.34 0.39
17 33.2 18.5 7.3 1.39 0.37
2 21 16 39 19.3 7.4 1.46 0.37
16 27.6 19.8 8.3 1.33 0.40
15 32.2 19.2 5.6 1.42 0.44
12 36.8 19.1 4.4 1.45 0.52
3 22 12 43.2 20.0 4.8 1.54 0.53
54 29.3 22.2 7.7 1.51 0.33
52 34.2 23.6 8.0 1.57 0.32
45 39.1 25.3 8.8 1.62 0.32
4 70 29 46 26.4 8.0 1.66 0.32

Table 3.3. Average width of 4 zones of the Lower Venoge River (c.f. figure 3.15). Results were
developed from HEC-RAS simulations using 1996 cross section data.

28.0

26.0

24.0

22.0
width (m)

20.0

18.0
zone1
16.0 zone2
zone3
14.0 zone4
a=3.2,b=0.5
12.0
a=4,b=0.5
10.0
20 25 30 35 40 45 50
discharge (m³/s)

Figure 3.18. Reach averaged width for four zones of the Lower Venoge River at discharges
approximating the 0.5, 0.75, 1, and 2 year return period flows. Lines show possible parameters for the
b
relation W = aQ .
56 The Venoge River

1.80

1.70

1.60
hydraulic depth (m)

1.50

1.40

zone1
1.30
zone2

1.20 zone3
zone4
1.10 c=0.375,f=0.37
c=0.42,f=0.37
1.00
20 25 30 35 40 45 50
discharge (m³/s)

Figure 3.19. Reach averaged depth for four zones of the Lower Venoge River at discharges
approximating the 0.5, 0.75, 1, and 2 year return period flows. Lines show possible parameters for the
f
relation D = cQ .
 Chapter 4

Venoge River Data for

Streambank Erosion Modeling

4.1 Introduction
Streambank erosion modeling is a complex natural process that involves many parameters. Some data
were readily available for the Lower Venoge River, but other data had to be collected. This chapter will
explain how the experimental erosion zone was determined, what data was available, and the data that was
collected to characterize the streamflow and streambank erosion, cross sections, bank and bed sediments,
and boundary conditions of the study reach.

4.2 Experimental setup

An initial survey of the Lower Venoge River (from Lussery to the lake) was undertaken to determine the
actively eroding banks, note the position of bank protection works, and determine a zone where erosion
could be monitored (figure 4.1). Observations indicated that the alluvial zone "Iles de Bussigny" (islands
of Bussigny) would work well for erosion monitoring as well as a zone relatively free to meander near the
town of Denges (figure 4.2). Significant erosion occurs in the alluvial zones of the Roujarde and the
Lovateire, but these zones are less accessible and farther away. The study reach was thus determined to be
from the Moulin de Choc, Bussigny to the reach just downstream from Denges. This zone is ideal in that
it will provide three different type of zones for the erosion modeling. The Bussigny islands alluvial zone
is free of bank protection and moving frequently. The intermediate zone is mostly fixed by bank
protection. It will be interesting to see how the streambank erosion modeling will handle a zone with

57
58 Venoge River Data for Streambank Erosion Modeling

minimal data. The third detailed erosion monitoring zone at the end of the study reach is also interesting
because of the freely eroding banks which are mixed in between some banks that are protected.

Moulin de Choc

Iles de Bussigny

Roujarde

Denges

Lovateire

Figure 4.1. Erosion scars (stars) observed along the Lower Venoge River. Solid bold lines indicate
bank protection and the dashed line indicates the line approximately 5 meters above the river.
 Experimental setup 59

Cross section

7 Erosion
monitoring point
and bend num-
ber
Stage recorder
Maximum
stage recorder
Centerline 2004

Reproduced with permission

' Swisstopo
from swisstopo (BA068334)

Figure 4.2. Study reach, new cross sections, water level measurements, and bends where erosion was
monitored. Center background map: CP100 ©1997, Swisstopo.

4.2.1 Erosion monitoring

Due to the limited knowledge concerning the erosion of the Venoge, especially on the processes
occurring, it was decided that erosion monitoring should be undertaken. A review of possible methods
to use was undertaken before choosing the methods to be applied.

4.2.1.1 Possible methods

A review of possible methods for "the measurement of river bank erosion and lateral channel change"
was done by Lawler in 1993. A summary of his review with some additional comments based on recent
advances is given here.

Measurement techniques can be divided into three classes based on the timescale that is measured
(Lawler, 1993):

1. long timescale (10-10’000 years): sedimentological evidence, botanical evidence, and historical
sources
2. medium timescale (0.5-20 years): planimetric resurvey and repeated cross-profiling
3. short timescale (event-10 years): erosion pins, the Photo-Electronic Erosion Pin (PEEP) system,
bank profiling, and terrestrial photogrammetry and laser scanning
60 Venoge River Data for Streambank Erosion Modeling

Since the objective of the streambank erosion monitoring is to use the data for the calibration of erosion
and bank failures, it is necessary to use methods adapted to a short timescale. The following paragraphs
give details on short timescale methods:

1. Erosion pins: Lengths of rod are inserted into the bank leaving a small portion visible.
Measurements are made before and after discharge events or on a certain schedule. Measurements
are made with a caliper or ruler depending on the precision required. The frequency of the
measurements depends on the purpose of the study. Pins are usually in lengths of 25-50 cm with a
diameter of 2 - 6 mm. They should be non-rusting and strong enough to resist bending due to debris
flow and/or installation in hard material. The advantages of the system are its suitability in a wide
range of fluvial environments, its simplicity, cheapness, and the precision possible. The
disadvantages include the representativeness of point measurements for spatial erosion assessment,
bank swelling or contraction that can causes difficulties in reading interpretation, pin movement or
loss due to dislodging by debris or mass failures, and the effect of the pins on the bank conditions
(turbulence around pins, reinforcement in cohesive soils, disruption of gravel imbrication). Erosion
pins must be used in a network to produce meaningful measurements. These networks should have a
longitudinal spacing of 1-5 meters and use two or more pins in a vertical profile (Thorne’s
recommendation (1981) in Lawler, 1993) to capture the spatial variations of erosion.
2. PEEP sensor: Lawler et al. (1997) refer to the PEEP sensor as a "self-reading erosion pin". It is a
"relatively simple optoelectronic device consisting of an array of photovoltaic cells (photodiodes)
connected in series and enclosed within a transparent acrylic tube. It outputs an analogue millivolt
signal directly proportional to the length of the tube which is exposed to light" (HydroScientific,
1999) (figure 4.3). To eliminate the problem of varying luminosity during the day and also due to
turbid water conditions, the sensor uses reference photodiodes that are placed at the beginning and
end of the measurement zone. Using the outer reference photodiode measurement, the Photovoltaic
Ratio, Rpp (Hydroscientific, 1999), can be calculated:
Rpp ( % ) = ( V cs ⁄ V rc ) × 100 Eq. 4.1
where Vcs is the voltage of the cell series and Vrc is the outer reference cell voltage. The PEEP user
guide suggests then the use of the linear regression formula:
L = c + d × Rpp Eq. 4.2
to calculate the exposed length, L, in mm of the sensor. Sensors come in two different lengths, 50 cm
and 66 cm. The user guide suggests the calibration of the coefficients c and d in situ. Otherwise,
factory values of c and d can be used. Besides the photodiodes, there are also thermistors
incorporated in recent versions of the sensor (figure 4.3) so that important erosion or mass failure
can be detected during the night. This detection works when the inner thermistor registers a different
temperature value due to the soil around it. The advantage of such a system is of course the
information that can be gained on the erosion process by following it continually. The disadvantage
of the system is the installation necessary and the possible loss of the apparatus in the case of a large
bank failure.
3. Bank profiler: A third option mentioned by Lawler (1993) as a medium timescale option, the bank
profiler (figure 4.4), could also be used at a shorter time scale for rapidly eroding banks. The
horizontal and vertical stadia rod provide datums which can be used to measure down and over to
 Experimental setup 61

record the bank position. The difference in bank position at a certain depth will give the erosion at
that depth. To overcome the inaccuracies possible due to poor positioning of the horizontal stadia
rod, Lawler (1993) suggests using a graduated profile frame which can be inserted into receiving
sockets set in the bank. To ease the vertical and horizontal measurements, he then suggests the use
of a double rack and pinion system for the measuring rod. The advantages of this type of system is
that no flow disturbance is created and that there is no risk of pin loss. The disadvantage of course is
the loss in accuracy, although this can be largely overcome with the relocatable bank profiler of
Lawler. Another disadvantage that can occur can be the difficulties associated with maintaining the
datums. Finally, the measurements are more time-consuming than erosion pins and the profiling
equipment is more cumbersome for transport.
4. Terrestrial photogrammetry and laser scanning: These two measurement techniques are grouped
together because the desired output, a DTM of the bank, is the same. Terrestrial photogrammetry
has been used in a few cases already (Lawler, 1993), but no terrestrial laser scanning examples for
measuring bank erosion were found in literature. Airborne laser scanning data has already been used
for determining bank erosion (Thoma et al., 2005), but would not be practical for the event scale nor
for precise measurements of erosion at the bend length. Laser scanning has the advantage of the
number of points that can be obtained very quickly. With an appropriate filter, the bank vegetation
could be removed leaving the points that can be interpolated to produce a DTM of the bank. The
disadvantage of the system is the cost of the laser scanner and the setup time necessary for installing
precise reference points where the scanner tripod can be setup. In some cases, vegetation and trees
might make it nearly impossible to implement, but this is the case for all of the methods.

Figure 4.3. PEEP sensor and its typical installation for measuring bank erosion (from the PEEP user
guide (HydroScientific, 1999))
62 Venoge River Data for Streambank Erosion Modeling

4.2.1.2 Chosen methods

Stadia Rod Direction stake
Level Reference stake
Erosion pins were chosen due
to their simplicity of use. Base line

Knitting needles, as suggested Level

by Lawler (1993) of 40 and 50
cm lengths were used to begin Plumb line and
measuring tape
Folding Meter

with. The heads on the needles,

which Lawler found practical
for numbering, were found to
be impractical because of the
added difficulty for measuring
and also due to the heads
busting when the needle had to
be hammered into the bank.
Inox rods 50 cm long of 5 mm
diameter proved to be more Figure 4.4. Bank profiling system used for the Lower Venoge study
practical.

Two to four erosion pins were put in each vertical profile and a minimum of three vertical profiles were
used for each measurement group. Each measurement group is called a bend, although sometimes a real
bend encompasses two or three erosion monitoring bends. The twenty erosion monitoring bends can be
seen in figure 4.2. One hundred and eighty pins were installed in the 12 bends in the Bussigny reach and
120 were installed in the Denges reach. Installation of the 300 pins was done on November 11, 12, 19,
and 20 of 2003. Figure 4.5 shows a small bend of three verticals each with 3 pins. The pins were only
possible to find by putting marker stakes on the top of the bank.

Fortunately, a 15 year return period flood occurred on January 15, 2004, wiping out nearly half of the
erosion pins. This allowed for the bank profiling method to be adopted where necessary. Bank profiling
was done for all measurement points during the course of the spring of 2004 in the case of future pin
losses so that pins could be reinstalled if necessary. The bank profiling method adopted was less precise,
but was suitable for measuring mass wasting. The combination of using pins in slowly eroding bends
and the bank profiler in bends where rapid change was occurring due to mass wasting proved to be
sufficient. Figure 4.5 shows how the bank profiling system was setup. The first operator (he is not in
place in figure 4.5 because he is taking the picture) assures the level based on a reference stake that is at
ground level. A second stake, closer to the bank, that is also pounded to ground level gives the direction
of the measurement. The second operator then measures down the plumb line a specified distance (the
same from visit to visit) and measures over to read a distance indicating the bank position.

Besides the erosion pins and the bank profiler, two PEEP sensors were installed in the spring of 2004
(location of big stars (together with stage recorder) in figure 4.2). Their installation was very difficult
due to roots and gravel encountered during boring of the horizontal and vertical conduits. Their
installation was also complicated by the need to place the datalogger in a safe place, to avoid a bank
failure that might drop the datalogger into the river and destroy it, yet in a place that erodes enough to
 Experimental setup 63

make it worth the installation. The datalogger was also used to record water depths from a submersible
pressure transducer.

Figure 4.5. Erosion monitoring on the Lower Venoge River. Bottom shows a small monitoring bend
(pin locations marked by arrows). Upper left shows the pin installation. Upper right shows the bank
profiler.
64 Venoge River Data for Streambank Erosion Modeling

4.2.2 Flow monitoring

Besides the erosion monitoring, maximum stage recorders and continuous stage recorders were used to
monitor flow depths (small and big stars in figure 4.2 respectively). The planned use of the water depth
information resulting from these recorders is for the calibration of Manning coefficients to ensure that
mean velocities, and indirectly mean shear stresses, are relatively correct throughout the detailed study
area.

Maximum stage recorders were fabricated in-house by the HYDRAM technician, Alex Amiguet. A
metallic ruler is suspended by the cap of the galvanized tube. On this ruler is a red tape, which loses its
color as water touches it. The cap of the stage recorder tube was surveyed so that its position and
elevation were known. The resulting measurements could then be transformed to an absolute water
surface elevation. The red tape was changed of course after each high flow event.

Continuous stage recorders used submersible pressure transducers and were connected to HydroMadd
dataloggers. The pressure transducer was placed in an appropriate galvanized tube and its bottom rested
on a cross at the bottom of the tube. The elevation of the cross was determined by surveying, allowing
the resulting measurements to be transformed to absolute water surface elevation. Recording was done
every 15 minutes.

Besides the installed stage recorders, the federal continuous stage recorder data for the station of
Ecublens, Les Bois (figure 4.2), was available and extremely helpful. A rating curve for this station is
maintained by the National Hydrological Service (SHN) of the Swiss Federal Office for Water and
Geology (FOWG). Flow, water level, and temperature evolution can be followed on the internet as it is
updated generally every two hours (http://www.bwg.admin.ch/service/hydrolog/f/2432.htm). Digital
data was available from the SHN.

4.2.3 Measured values

4.2.3.1 High flows and stage recorder values

The continuous stage recorders, limni11 and limni17, recorded water depths from 26.03.2004 to
14.08.2005. Holes in the measured values occurred during the following periods: 01-03.05.2004, 19-
21.05.2004, 15.06.2004-02.07.2004, 18-20.07.2004, 15-30.09.2004, 19.11.2004-21.12.2004, 06-
20.01.2005, 28.05.2005-17.06.2005, and 03.07.2005-29.07.2005. Besides those measurement holes,
limni17 measurements had another hole during the period 03-17.08.2004 and measurements deviated
from the calibrated values during the period 01.03.2005-17.06.05.

The high flow events during the erosion monitoring period are given in table 4.1. The event length and
volume are based on 9 m³/s starting and ending discharges. The high flow events of 14.01.2004 and
26.10.2004 were particularly strong and can be evaluated at 11 and 4 year return periods, respectively,
according to the Gumbel distribution shown in figure 5.13. The hydrograph of the 14.01.2004 high flow
event is shown in figure 5.15, while that of the 26.10.2004 event is shown in figure 4.6. Due to the
 Experimental setup 65

relatively small discharge inputs between the limni11, limni17, and the Ecublens stage recorders,
approximate rating curves for limni11 and limni17 could be produced using the Ecublens discharges of
the 26.10.2004 event (figure 4.6).

Q peak Event Volume Erosion measuring

Start date (m³/s) length (h) (m³) Date peak date Note
11,12,19,20.11.2003 installation pins
28.11.2003 4.48 28.11.2003 04.12.2003
12.01.2004 77.793 256.6 2.36E+07 14.01.2004 28.01.2004 followed event 14-19.01.2004
11.02.04-10.03.04 change of pins to stadia rod
14.03.2004 12.157 20.0 7.80E+05 14.03.2004 18.03.2004
20.03.2004 11.169 51.3 1.92E+06 21,22.03.2004 24.05.2004 snow melt; two peaks
16.07.2004 10.908 0.3 9.73E+03 16.07.2004
24.07.2004 10.495 0.5 1.68E+04 24.07.2004 25.07.2004
19.08.2004 19.237 38.1 1.93E+06 20.08.2004 30.08.2004
19.10.2004 14.655 43.9 1.92E+06
26.10.2004 61.412 139.5 1.11E+07 26.10.2004 09.11.2004 followed event 26-29.10.2004
19.12.2004 25.253 35.1 1.92E+06 20.12.2004 17.01.2005
12.02.2005 22.334 69.3 3.96E+06 13.02.2005 28.02.2005
24.03.2005 14 129.5 4.78E+06 27.03.2005 11.04.2005 snow melt
24.04.2005 20 126.0 6.05E+06 25.04.2005 26.09.2005

Table 4.1. High flow events during the erosion monitoring period. Discharge information is derived
from the Ecublens stage recorder.

70 3.5
Q Ecublens
60 H - limni17 3
H - limni11
50 2.5
Discharge (m³/s)

40 2

Depth (m)
30 1.5

20 1

10 0.5

0 0
05.10 10.10 15.10 20.10 25.10 30.10 04.11 09.11 14.11
Date

70 70
Q Ecublens Q Ecublens
60 Q rating curve 60 Q rating curve

50 50
Discharge (m³/s)
Discharge (m³/s)

40 40

30 30

20 20

10 10

0 0
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3
Depth (m) Depth (m)

Figure 4.6. October 26, 2004 high flow event. Rating curves for the limni17 (left) and limni11 (right)
based on the Ecublens discharges.
66 Venoge River Data for Streambank Erosion Modeling

The maximum stage recorders also provided important water depth readings. Those readings are shown
in figure 4.7. It was possible with the maximum stage recorders to record not only the maximum event
values, but also to measure the water surface at the time of visit. This extended considerably the number
of measurements, especially for the stage recorders that were placed relatively high and recorded only
two or three high flow events.

The stakes marking each erosion measuring point also served as maximum stage recorders. High water
marks during the October 26, 2004 flood were found on many of the stakes and their water surface
elevations could be determined by the reference stake elevations. Besides the October 2004 high water
marks, some high water marks were available from the January 2004 flood as well as others from the
HYDRAM Venoge study (Lance et al., 1997). Those values are given in table 4.2.

distance max water

to outlet depth 1996 thalweg wse 1996
(m) point id discharge wse 2004 date report 1996 report report
10323 360 50 2.5 394.76 397.26
10282 361 55 3.6 393.26 396.86
10104 limni1 57 396.365 26.10.2004
10053 2.1 57 396.14 26.10.2004
9804 4.1 60 395.61 26.10.2004
9728 limni6 60 395.49 26.10.2004
9635 7.1 60 395.18 26.10.2004
9625 7.4 60 395.03 26.10.2004
9583 DELAISSE1 78 395.014 14.01.2004
9549 8.3 60 395.21 26.10.2004
9269 limni-eros11 60 394.31 26.10.2004
9087 DELAIS3 78 393.898 14.01.2004
8213 371 55 4.6 387.64 392.24
7771 390 50 2.8 388.19 390.99
7313 412 55 1 388.57 389.57
7297 413 60 3.5 385.62 389.12
5353 460 70 4.4 381.47 385.87
5241 461 50 3.5 381.65 385.15
4910 limni Ecublens 60 384.47 26.10.2004
3510 15.2 60 381 26.10.2004
3330 limni14 60 380.65 26.10.2004
3275 20.2 60 380.45 26.10.2004
3005 470 55 3.7 376.28 379.98
2963 471 60 3.3 376.93 380.23
2859 16.4 60 379.93 26.10.2004
2734 limni-eros17 60 379.74 26.10.2004

Table 4.2. Measured high water marks for the study reach. 1996 report values come from the HYDRAM
Venoge study (Lance et al., 1997).
 Experimental setup 67

Maximum Stage Recorder Limni1

top tube cap top ruler ruler Elevation 0
396.406 0.045 396.361 1.500 394.861
o bser- WSE mark Ecublens
time o f visit vatio n (m) WSE (m) Ecublens Q Date Qmax H mark (m) (m) Qmax

15.04.04 14:48 1.88 394.53 2.4

21.05.04 08:44 1.98 394.43 1.3
02.06.04 10:45 1.60 394.81 4.7
15.06.04 13:51 2.04 394.37 0.8
19.07.04 14:20 0.5 16.07.04 0.16 395.02 10.9
17.08.04 11:22 1.80 394.61 3.1 24.07.04 0.17 395.03 10.5
30.08.04 13:30 1.90 394.51 2.4 20.08.04 0.63 395.49 19.2
26.10.04 14:20 0.25 396.16 47.2
27.10.04 14:30 0.62 395.79 30.9
28.10.04 14:20 1.00 395.41 18.2 26.10.04 1.50 396.37 61.4
03.11.04 13:10 1.57 394.84 5.3
21.12.04 15:44 1.46 394.95 7.2 20.12.04 0.79 395.65 25.3
31.01.05 10:04 1.83 394.58 2.7
22.02.05 16:15 1.75 394.66 3.6 13.02.05 0.70 395.56 22.3
23.03.05 14:00 1.40 395.01 8.3 0.10 394.96
05.04.05 12:25 1.74 394.67 3.7 27.03.05 0.41 395.27 14.0
25.04.05 12:30 25.04.05 0.63 395.49 20.0
29.04.05 12:30 1.31 395.10 8.9 29.04.05 0.19 395.05 8.90
12.05.05 13:15 1.86 394.55 2.9

Maximum Stage Recorder Limni6

top tube cap top ruler ruler Elevation 0
395.997 0.045 395.952 1.500 394.452
o bser- WSE mark Ecublens
time o f visit vatio n (m) WSE (m) Ecublens Q Date Qmax H mark (m) (m) Qmax

15.04.04 15:00 2.4

21.05.04 08:54 2.49 393.51 1.3
02.06.04 10:30 2.18 393.82 4.5
30.08.04 15:15 2.3 20.08.04 0.09 394.54 19.2
Observation

28.10.04 15:00 1.61 394.39 17.9 26.10.04 1.04 395.49 61.4

22.02.05 15:50 2.30 393.70 3.6 20.12.04 0.24 394.69 25.3

Maximum Stage Recorder Limni14

top tube cap top ruler ruler Elevation 0
380.624 0.045 380.579 1.500 379.079
H mark

o bser- WSE mark Ecublens

time o f visit vatio n (m) WSE (m) Ecublens Q Date Qmax H mark (m) (m) Qmax

15.04.04 16:00 2.3

21.05.04 09:36 2.20 378.43 1.3
02.06.04 09:40 2.00 378.63 3.3
28.10.04 16:30 1.41 379.21 17.3 26.10.04 1.57 380.65 61.4
22.02.05 15:20 3.5 20.12.04 0.50 379.58 25.3

Figure 4.7. Maximum stage recorder measurements. Top left shows recorder limni6.

4.2.3.2 PEEP
The PEEP measurements have the same measurement periods and missing value periods as that of the
accompanying stage recorder. Before installation, laboratory and field calibrations were conducted. The
68 Venoge River Data for Streambank Erosion Modeling

laboratory calibrations gave calibrated c and d values (equation 4.2) very different than those suggested
in the PEEP manual (HydroScientific, 1999). After in situ calibration (figure 4.8), it was clear that the
laboratory luminosity played a role in the poor laboratory calibration. The in situ calibration for the two
PEEP sensors gave a good linear fit, and c and d values are similar to those of the PEEP manual.

300 300

y = 2.0394x + 30.171
250 y = 2.1663x + 27.521 250 2
2 R = 0.9985
R = 0.9954

exposed length (mm)

exposed length (mm)

200 200

150 150

100 100

50 50

0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
RPP RPP

Figure 4.8. In situ calibration of the PEEP sensors during overcast conditions. Eros11 is shown to the
left and Eros17 to the right.

Typical measured values from the PEEP sensor are shown in figure 4.9. It is clear that the calculated
exposed length varies too much throughout varying daily luminosity and is not practical for identifying
precisely the evolution of fluvial erosion. It can be helpful though to determine the moment of a mass
failure. A mass failure at the Eros17 sensor was detected on 27.04.2005. What could be assumed to be
fluvial erosion by an erosion pin is clearly a small mass failure because the water level is below the
sensor level and the length quickly jumps from approximately 56 mm to 125 mm. This sensor was
installed among grass and tree roots. It is assumed that the root reinforcement to the soil failed causing
the 6 cm mass failure. The final exposed length when demounted on 14.11.2005 was 15 cm on top and
18 cm on the bottom. The last calculated exposed length was 15 cm in August 2005 corresponding
correctly to the observed value. The difference between the top and bottom exposed length would
suggest a buttressing type added stability to the bank. The middle erosion pin of bend 17.1 is
approximately 1 meter downstream. A 6.5 cm "failure" occurred between 17.01.2005 and 28.02.2005
and then a 3 cm "failure" between 28.02.2005 and 11.04.2005. The 9-10 cm of erosion occurred at both
spots, but at different times, confirming the important spatial variability of erosion and mass failure as
reported by Lawler (1993).

The Eros11 PEEP sensor showed deposition during the measuring period. As the deposition seems to
have occurred during several events, it is difficult to determine from the measured values the exact
timing of the deposition. The initial exposed value was 65 mm and the final calculated value was 36 mm
in August 2005. The sensor was found completely covered when demounted on 14.11.2005.

The experiences with the PEEP sensors allow the following conclusions:
• the PEEP sensor is difficult to install, especially in rooty, gravelly soils
• the installation process will most likely weaken the neighboring soil matrix during boring
 Experimental setup 69

• luminosity variation causes variation in the calculated exposed length which makes it nearly
impossible to detect fluvial erosion
• when submerged the calculated exposed length was incorrect
• the sensor is useful for detecting mass failures

mV int. mV series mV ext. Length (mm) WSE (m)

180 379
Sensor Elevation = 378.76
160 378.9

Water Surface Elevation (m.a.s.l.)

140
mV or length exposed (mm)

378.8
120
378.7
100
378.6
80
378.5
60
378.4
40

20 378.3

0 378.2

-20 378.1
24.04

25.04

26.04

27.04

28.04
date

Figure 4.9. The small mass failure of the 17.1 erosion monitoring point measured by the Eros17 PEEP
sensor on April 27, 2005.

4.2.3.3 Erosion pin and bank profiler

The erosion pin and bank profiling measurements were very time consuming. Three hundred
measurements took between 2 and 2.5 days for two technicians. Installation of the 300 pins took 4 days
and initial bank profiling of the 300 measurement points took also 4 days for two technicians.

The field notebooks and EXCEL note keeping evolved as the situations encountered became increasing
complex. The first field book (left extract, figure 4.10) was designed simply for noting the erosion pin
measurements. After the January 14, 2004 flood, bank profiling was begun for actively eroding banks
resulting in a second field book in which lines necessary for the bank profiling were incorporated (middle
extract, figure 4.10). Finally, a third field book was necessary to better track reference stake and baseline
changes (right extract, figure 4.10). The meaning of the different measurements are shown in figure 4.11.
This third system assumed no reference stake y, rod/stake y, and bank top y changes. These changes did
occur though occasionally and had to be handled case by case. An improved note keeping system is
shown in figure 4.11 in which boxes are available for all possible x, y measurements. Notice also that
arrival and departure measurement boxes are available. This is necessary in the case of reference stake
movement and for erosion pin changes when they must be pounded in or partially pulled out depending
on the erosion/deposition evolution. Notice that erosion pin measurements can easily be noted on the
bank profiling field book (as seen for bend 2.2 in the right table in figure 4.10). In the x column, the
erosion pin length is noted. Other measurement boxes are left empty.
70 Venoge River Data for Streambank Erosion Modeling

2.2 Date depth 10.03.04 Date depth 30.08.04

Date 11.11.03 04.12.03 2.2 rem arque: 2.2 rem arque:

top 2 2 bank top -23 111 dist. m es.
bottom 2 2.2 top -50 5.5 rod/stake

2.3 bottom -100 4 ref x

Date 11.11.03 04.12.03 2.3 rem arque: bank top -23 235
top 2 2 dist. m eas. 0 270 top -50 5.5
m id top 2 2 bank top -70 180
bottom 2 2.6 top -90 70
m id top -130 61 bottom -100 24
bottom -160 89 bed -200
bed -260 260 2.3 rem arque:
dist. m es. 270
rod/stake 0
ref x 0
bank top -61 190
top -90 72
m id top -130 51

bottom -160 88
bed -260 273

Figure 4.10. Evolution of the erosion measurement field book due to the complex situations
encountered.

distance measure reference x

date
y
bank top arrival depart.
Base line
x x rod / stake dist. mes.
y
rod/stake
top
ref x
Plumb line

mid. top
direction stake bank top
reference stake
mid. bot. top
bed

bottom direction stake mid top

mid bot
bottom
bed

3-4 m stadia rod reference stake date

bank top mark
direction stake 1
y arrival depart.
y
x x x y x y
meas stake reference system 1
meas stake
2

direction stake 2 bank top

te m

top
Plumb line

ys
es

mid top
nc

measure stake
mid bot
er e
r ef

bottom
bed

Figure 4.11. Bank profiling measurement system used (top) and recommended (bottom)
 Experimental setup 71

The complete bank retreat measurements are reported in appendix 6. The cumulative bank top position
changes are given in table 4.3 and that of the bottom position changes are shown in table 4.4. Notes on
bank erosion and bank mass failure are included in table 4.3.

As is expected and witnessed in tables 4.3 and 4.4 is that erosion and bank failures are not directly
correlated with maximum discharge or the volume of the flood. This is assumed to be largely due to the
predisposition of a bank to fail (basal sediment condition and bank soil-vegetation heterogeneities) as
well as the flow conditions which change due to the influence of neighboring failed bank soil and
vegetation. It is expected to find correlation between the event flood volume and discharge and reach
bank erosion. Correlation of reach bank erosion, represented by the sum of all of the average measured
profile erosion, is compared against maximum event discharge, flood volume (based on 9 m³/s start and
end), and the product of Qmax and flood volume (figure 4.12). The graphs show important variability
for the lower discharge events, but that the linear relationship relating erosion and the product Qmax-
flood volume explains quite well the quantity of bank erosion. The number of bank failures is also well
correlated with the product Qmax-flood volume.

7000.0 0.0549x
7000.0
Erosion y = 88.314e Erosion y = 0.0003x - 280.59
2 2
Expon. (Erosion) R = 0.7518 Linear (Erosion) R = 0.933
Sum of measured erosion by event (cm)

6000.0 6000.0
Sum of measured erosion by event (cm)

5000.0 5000.0

4000.0
4000.0

3000.0
3000.0

2000.0
2000.0

1000.0
1000.0
0.0
0.0 0.0E+00 5.0E+06 1.0E+07 1.5E+07 2.0E+07 2.5E+07
0 20 40 60 80 100 -1000.0
3 3
Qmax (m /s) Volume (m )

7000.0 60
Erosion y = 3E-06x + 148.88 Events y = 2E-08x + 5.1339
2 2
Linear (Erosion) R = 0.9913 Linear (Events) R = 0.9174
Sum of measured erosion by event (cm)

6000.0
50
Number of bank failures by event

5000.0
40

4000.0
30
3000.0

20
2000.0

1000.0 10

0.0 0
0.0E+00 5.0E+08 1.0E+09 1.5E+09 2.0E+09 0.0E+00 5.0E+08 1.0E+09 1.5E+09 2.0E+09
6 6
Event Qmax*Volume (m /s) Event Qmax*Volume (m /s)

Figure 4.12. Relationship of event erosion compared to maximum event discharge, flood volume, and
Qmax*flood volume, and bank failures compared to Qmax*flood volume

The bank mass failures observed on the Lower Venoge monitoring sites are slab (block) failures, soil
falls, planar slides, and cantilever failures (table 4.3). Slab failures and soil falls are the most common.
They occurred or could occur on all of the monitored bends except for bend 18, where there was planar
72 Venoge River Data for Streambank Erosion Modeling

sliding of the upper glacial till on a fine silt layer. Soil fall and slab failures are difficult to distinguish
between because it is unknown to what degree the toe is being undercut before failure. Bank failure
mechanisms other than slab failures are noted in table 4.3. Cantilever failures are also frequent, but
serious cantilever failures due to trees toppling into the river are a minority of the failures simply due to
the frequency of tall trees near the banks.

The most frequent bank mass

failure mechanism observed was
that of slab failure underneath a
vegetated surface (figure 4.13). If
the root reinforcement of the
vegetated surface was sufficient,
a "carpet" of the vegetated
surface would remain after slab
failure to protect the upper part of
the bank. This sometimes caused
negative bank retreat
measurements for the top
Figure 4.13. "Carpet" of the soil-root system left after a soil fall measurement zone. The sequence
of events was often the following:
toe erosion, slab failure
accompanied by the incomplete fall of the vegetation zone producing a "carpet" protection of the upper
bank, failure of the upper bank "carpet" after weakening of the soil-root system by subaerial processes
or fluvial erosion. The failed sediments of the bloc failure and upper bank "carpet" mostly deposit at the
bank toe and must be eroded away before a new slab failure can occur.

Another process observed in the slab

failures was that of tension cracking (figure
4.14). Tension cracking was only observed
during high flows or during the following
days. The tension cracking occurred close to
the bank and multiple tension cracks
occurred. It is assumed these tension cracks
are critical in determining the width of the
slab that fails. It is also assumed that the
tension cracking will increase infiltration
promoting a more rapid failure (Abramson
et al., 2002).

Banks are considered to have failed if the

bank top measurement exceeded 20 cm
Figure 4.14. Observed tension cracks at bend 3.3 during (table 4.3). Twenty centimeters was decided
the drawdown of the October 26, 2004 flood as the criteria because of measurement
uncertainty associated with the bank top
measurement. Slab failures caused bank top retreat of 20 to 250 cm, while tree toppling cantilever caused
 Experimental setup 73

bank retreat of up to 350 cm. Of course the bank retreat amount in such a case depends on the rooting
system of the tree. Bank overhang was observed to be as much as 85 cm.

The tree cantilever failures bring up the question of the effect of vegetation on bank stability. Vegetation
has positive and negative effects on bank stability. Those effects are relatively well documented (ASCE,
1998a, Abramson et al. 2002). An indirect effect of tree cantilever failures, though, is that of the failure
on the flow field. A tree cantilever failure was observed at bend 7.2 on 30.08.2004 (table 4.3 and figure
4.15). Rapid bank retreat occurred throughout bend 7 during the preceding year. No bank retreat
occurred during the year after the failure despite the important 26.10.2004 high flow. There is no
question of the positive effect of the bend 7.2 tree cantilever on the reduction the eroding shear stresses.
Daniels and Rhoads (2004) measured the effect of outer bend woody debris on the flow field and showed
that the maximum turbulent kinetic energy shifted away from the outer toe.

Figure 4.15. Tree cantilever failure of bend 7.2

The observed Lower Venoge bank retreats are largely fluvial erosion dominated. The slab failures seem
to be very small and possibly multiple during a single event. It is the rate of erosion of the bank toe and
the basal sediments that seem to dictate the bank retreat rate. The erosion rate of the monitored bends is
evident with the bottom erosion measurement. As can be seen in table 4.4, some bends will quickly
evacuate basal sediments while others do not have sufficient shear stresses to evacuate them.

Piping is another form of erosion. Soil water exfiltration was seen in bends 9 and 10 and some piping is
assumed to have occurred in these bends although it is assumed that no significant erosion occurred
because of the piping. The exfiltration in these banks reflects the sometimes saturated nature of these
banks which is probably more important than the piping itself.
74 Venoge River Data for Streambank Erosion Modeling

date
bend 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05 notes
1.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
1.20 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
1.30 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -6.00 -6.00 -6.00
2.10 0.00 0.00 0.00 0.00 0.00 0.00 3.00 3.00 3.00 3.00 3.00
2.20 0.00 0.00 0.00 0.00 -2.00 -20.00 -7.00 46.00 46.00 46.00 46.00
2.30 0.00 0.00 0.00 0.00 -10.00 -10.00 -10.00 -10.00 -10.00 -10.00 -10.00
2.40 0.00 31.00 31.00 36.00 36.00 36.00 36.00 38.00 41.00 46.00 46.00
2.50 0.00 83.00 83.00 83.00 83.00 83.00 83.00 83.00 83.00 83.00 83.00
2.60 0.00 83.00 83.00 83.00 83.00 83.00 88.00 88.00 88.00 93.00 103.00
3.10 0.00 124.00 124.00 127.00 127.00 128.00 143.00 147.00 147.00 147.00 147.00
3.20 0.00 132.00 137.00 142.00 142.00 142.00 142.00 142.00 137.00 137.00 137.00
3.30 0.00 175.00 175.00 175.00 175.00 175.00 278.00 313.00 313.00 313.00 313.00
4.10 0.00 0.00 0.00 5.00 5.00 5.00 88.00 88.00 88.00 88.00 88.00
4.20 0.00 0.00 65.00 65.00 65.00 65.00 110.00 110.00 110.00 110.00 110.00
4.30 0.00 0.00 0.00 -10.00 -10.00 0.00 35.00 35.00 35.00 50.00 50.00
5.10 0.00 96.00 111.00 106.00 106.00 106.00 134.00 136.00 131.00 173.00 177.00
5.20 0.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00
5.30 0.00 4.00 4.00 4.00 4.00 4.00 4.00 -1.00 -4.00 -4.00 -4.00
6.10 0.00 0.00 8.00 18.00 18.00 18.00 18.00 18.00 18.00 18.00 43.00
6.20 0.00 0.00 0.00 15.00 15.00 15.00 15.00 0.00 35.00 35.00 70.00
6.30 0.00 90.00 95.00 90.00 95.00 95.00 125.00 165.00 165.00 175.00 175.00
6.40 0.00 127.00 135.00 130.00 130.00 130.00 130.00 133.00 135.00 140.00 200.00
6.50 0.00 81.00 81.00 96.00 96.00 96.00 96.00 96.00 96.00 96.00 116.00
6.60 0.00 157.00 162.00 162.00 162.00 162.00 232.00 277.00 277.00 282.00 282.00
6.70 0.00 160.00 160.00 160.00 160.00 160.00 380.00 380.00 380.00 380.00 380.00
6.80 0.00 160.00 165.00 175.00 175.00 175.00 247.00 247.00 252.00 270.00 270.00
6.90 0.00 144.00 144.00 144.00 144.00 144.00 159.00 159.00 259.00 259.00 259.00
6.91 0.00 91.00 91.00 96.00 96.00 96.00 121.00 121.00 121.00 121.00 121.00
6.92 0.00 67.00 72.00 82.00 82.00 82.00 92.00 127.00 132.00 132.00 132.00
7.10 0.00 104.00 169.00 204.00 204.00 314.00 319.00 314.00 324.00 329.00 329.00
7.20 0.00 190.00 231.00 281.00 281.00 501.00 506.00 516.00 521.00 block failure, tree cantilever
7.30 0.00 129.00 154.00 249.00 249.00 254.00 254.00 254.00 256.00 254.00 264.00
7.40 0.00 79.00 99.00 141.00 141.00 134.00 134.00 144.00 154.00 154.00 154.00
7.50 0.00 204.00 243.00 253.00 253.00 253.00 248.00 253.00 253.00 258.00 258.00
7.60 200.00 275.00 281.00 281.00 281.00 281.00 290.00 290.00 290.00 290.00
8.10 0.00 10.00 10.00 15.00 15.00 15.00 70.00 70.00 70.00 85.00 85.00
8.20 0.00 0.00 10.00 10.00 10.00 10.00 10.00 5.00 5.00 60.00 75.00
8.30 0.00 183.00 183.00 183.00 183.00 183.00 183.00 183.00 183.00 183.00 209.00 block failure, tree cantilever
9.10 0.00 150.00 150.00 150.00 150.00 380.00 380.00 380.00 380.00 380.00 piping, block failure, tree cantilever
9.20 0.00 150.00 148.00 148.00 148.00 148.00 150.00 150.00 145.00 145.00
9.30 0.00 150.00 152.00 152.00 152.00 152.00 152.00 152.00 152.00 152.00
10.10 0.00 0.00 0.00 0.00 0.00 0.00 3.00 3.00 3.00 3.00 3.00
10.20 0.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00 -11.00
10.30 0.00 13.00 13.00 13.00 17.00 17.00 1.00 1.00 1.00 1.00
10.40 0.00 23.00 123.00 123.00 138.00 163.00 262.00 263.00 263.00 263.00 263.00 sandier, piping, block failure
11.10 0.00 0.00 0.00 0.00 0.00 0.00 15.00 15.00 15.00 15.00 15.00
11.20 0.00 85.00 85.00 93.00 93.00 93.00 93.00 90.00 90.00 90.00 90.00
11.30 0.00 1.00 1.00 3.00 3.00 3.00 3.00 58.00 58.00 58.00 58.00
11.40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
11.50 0.00 100.00 100.00 105.00 105.00 105.00 105.00 108.00 248.00 248.00 258.00
11.60 0.00 125.00 131.00 133.00 133.00 133.00 290.00 290.00 290.00 290.00 290.00
11.70 0.00 132.00 132.00 135.00 180.00 132.00 132.00 140.00 132.00 135.00 135.00
11.80 0.00 132.00 132.00 127.00 134.00 134.00 134.00 134.00 136.00 132.00 132.00
12.10 0.00 0.00 0.00 -1.00 -1.00 -1.00 -5.00 -11.00 -5.00 4.00 4.00
12.20 0.00 0.00 0.00 0.00 0.00 0.00 92.00 90.00 90.00 113.00 113.00
12.30 0.00 160.00 160.00 165.00 165.00 165.00 285.00 295.00 300.00 307.00 307.00
12.40 0.00 150.00 155.00 160.00 160.00 160.00 215.00 220.00 225.00 225.00 250.00
12.50 0.00 212.00 209.00 235.00 235.00 235.00 302.00 300.00 355.00 365.00 364.00
12.60 0.00 245.00 245.00 235.00 235.00 235.00 313.00 313.00 370.00 370.00 380.00
13.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -8.00 -8.00 -8.00 -8.00
13.20 0.00 0.00 0.00 0.00 0.00 0.00 35.00 40.00 50.00 63.00 63.00
13.30 0.00 0.00 0.00 0.00 0.00 10.00 120.00 118.00 117.00 125.00 125.00
13.40 0.00 0.00 0.00 0.00 0.00 0.00 40.00 38.00 75.00 80.00 85.00
13.50 0.00 0.00 0.00 1.00 1.00 1.00 1.00 1.00 11.00 11.00 11.00
14.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
14.20 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
14.30 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
14.40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
15.10 0.00 115.00 115.00 125.00 125.00 125.00 125.00 125.00 125.00 135.00 140.00
15.20 0.00 80.00 90.00 90.00 90.00 90.00 105.00 110.00 110.00 110.00 115.00
15.30 0.00 70.00 90.00 95.00 95.00 95.00 390.00 390.00 385.00 385.00 395.00 block failure, tree cantilever
15.40 0.00 55.00 55.00 45.00 45.00 45.00 131.00 130.00 130.00 130.00 130.00
15.50 0.00 78.00 78.00 78.00 78.00 78.00 88.00 88.00 93.00 93.00 93.00
15.60 0.00 40.00 40.00 42.00 42.00 42.00 42.00 52.00 37.00 37.00 37.00
16.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
16.20 0.00 87.00 87.00 92.00 92.00 92.00 122.00 122.00 127.00 127.00 127.00
16.30 0.00 80.00 80.00 80.00 80.00 80.00 80.00 85.00 80.00 80.00 80.00
16.40 0.00 92.00 92.00 97.00 97.00 97.00 137.00 137.00 137.00 137.00
16.50 0.00 0.00 0.00 3.00 3.00 3.00 3.00 13.00 8.00 28.00 28.00
17.10 0.00 0.00 0.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
17.20 0.00 0.00 0.00 0.00 0.00 0.00 0.00 5.00 5.00 5.00 5.00
17.30 0.00 46.00 46.00 56.00 56.00 211.00 211.00 211.00 211.00 211.00 211.00
17.40 0.00 92.00 127.00 157.00 157.00 287.00 287.00 287.00 287.00 312.00 312.00
17.50 0.00 200.00 200.00 200.00 200.00 200.00 557.00 560.00 560.00 560.00 560.00 block failure, tree cantilever
17.60 0.00 215.00 240.00 230.00 235.00 435.00 555.00 555.00 560.00 575.00 580.00 block failure, tree cantilever
18.10 0.00 50.00 50.00 50.00 50.00 50.00 50.00 50.00 50.00 50.00 50.00
18.20 0.00 0.00 0.00 2.00 15.00 15.00 15.00 15.00 5.00 5.00 5.00
18.30 0.00 50.00 50.00 sliding
18.40 0.00 0.00 0.00 0.00 0.00 0.00 0.00
18.50 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
18.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
18.70 0.00 0.00 0.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00 20.00
19.10 0.00 0.00 5.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00 9.00
19.20 0.00 0.00 0.00 35.00 35.00 35.00 35.00 35.00 35.00 35.00 35.00
19.30 0.00 0.00 0.00 0.00 5.00 5.00 20.00 40.00 40.00
19.40 0.00 0.00 0.00 0.00 0.00 0.00 3.00 3.00 2.00 2.00 2.00
20.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
20.20 0.00 62.00 62.00 62.00 62.00 62.00 72.00 72.00 72.00 72.00 72.00
20.30 0.00 0.00 0.00 10.00 10.00 10.00 3.00 3.00 3.00 3.00 3.00

Table 4.3. Measured cumulative bank top change from November 2003 - September 2005. Shaded cells
indicate bank failures (>20 cm). Maximum change was 580 cm.
 Experimental setup 75

date
bend 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
1.1 0.00 -3.00 -2.80 -2.80 -2.80 -7.50 -6.00 -7.00 -9.00 -8.00 -9.00
1.2 0.00 2.00 1.00 0.00 -1.00 -2.00 -2.00 -2.00 -4.00 -1.50 -1.50
1.3 0.00 2.50 3.00 1.40 0.00 -0.50 -3.00 -4.00 -3.50 0.50 3.50
2.1 0.00 1.00 3.00 1.50 1.50 1.50 1.50 1.50 2.50 8.00 9.00
2.2 0.00 2.30 2.80 2.80 2.80 60.30 61.30 61.30 65.00 69.00 72.00
2.3 0.60 7.60 -0.40 -0.40 1.60 1.60 16.60 9.60 9.60
2.4 1.00 1.00 4.00 13.00 60.00 69.00 73.00 79.00 85.00
2.5 0.00 58.50 50.00 46.00 61.00 45.50 55.00 59.00 67.00
2.6 0.00 72.00 74.00 58.00 54.00 81.00 73.00 73.00 73.00 80.00
3.1 0.00 117.00 120.00 120.00 118.00 124.00 123.00 124.00 127.00 132.00
3.2 0.00 104.00 102.00 102.00 102.00 102.00 104.00 108.00 108.00 108.00
3.3 0.00 173.00 173.00 173.00 173.00 272.00 259.00 259.00 259.00 259.00
4.1 0.00 1.00 43.00 43.00 43.00 62.00 35.00 26.00 43.00 47.00 62.00
4.2 -1.50 -14.00 17.00 22.00 12.00 30.00 11.00 15.00 23.00 23.00
4.3 -76.00 -76.00 -76.00 -79.00 -73.00 -73.00 -73.00 -69.00
5.1 -0.50 109.50 105.50 114.50 114.50 140.50 138.50 127.50 153.50 154.00
5.2 0.00 2.50 2.50 2.50 -0.50 -0.50 -0.50 89.50 89.50 91.50 91.50
5.3 0.00 2.00 2.00 2.00 0.50 -3.00 1.50 3.00 3.00 20.50 -25.50
6.1 0.00 2.50 1.50 5.50 5.50 11.50 -0.50 9.50 29.50 35.50
6.2 0.00 8.00 51.00 10.00 10.00 10.00 17.00 4.00 35.00 30.00 30.00
6.3 0.00 90.00 93.00 93.00 99.00 95.00 126.00 129.00 153.00 160.00 178.00
6.4 0.00 127.00 139.00 149.00 149.00 149.00 163.00 172.00 200.00 212.00 221.00
6.5 0.00 81.00 57.00 75.00 80.00 80.00 106.00 113.00 132.00 132.00 140.00
6.6 0.00 157.00 159.00 159.00 159.00 153.00 325.00 209.00 209.00 224.00 224.00
6.7 0.00 160.00 189.00 189.00 176.00 163.00 116.00 116.00 covered covered covered
6.8 0.00 160.00 156.00 188.00 170.00 170.00 245.00 240.00 232.00 216.00 216.00
6.9 0.00 144.00 132.00 163.00 138.00 132.00 208.00 208.00 204.00 214.00 covered
6.9 0.00 91.00 80.00 80.00 80.00 80.00 184.00 184.00 153.00 121.00 158.00
6.9 0.00 68.00 65.00 65.00 65.00 59.00 146.00 117.00 107.00 112.00 150.00
7.1 0.00 104.00 217.00 217.00 269.00 269.00 269.00 282.00 282.00 269.00
7.2 40.00 230.00 312.00 356.00 418.00 427.00 420.00 428.00
7.3 0.00 129.00 130.50 202.50 207.50 205.50 206.50 206.50 206.50 209.50 209.50
7.4 0.00 79.00 130.50 137.00 137.00 130.00 130.00 140.00 157.00 163.00
7.5 0.00 232.00 232.00 232.00 232.00 229.00 224.00 224.00 235.50 244.00
7.6 200.00 230.00 230.00 230.00 230.00 231.00 231.00 231.00 237.00 237.00
8.1 0.00 0.00 0.00 0.00 41.00 41.00 46.00 46.00 46.00
8.2 0.00 8.00 8.00 13.00 10.00 10.00 69.00 77.00 77.00 74.00 74.00
8.3 0.00 181.00 176.00 172.50 171.50 171.00 170.00 167.50 175.50
9.1 0.00 142.00 135.00 140.00 140.00 140.00
9.2 -3.50 155.50 152.50 145.50 150.50 169.50 231.50 167.50 163.50 227.50
9.3 -1.00 65.00 65.00 70.00 61.00 68.00 68.00 68.00 57.00 73.00
10.1 0.00 4.00 6.00 7.50 13.50 31.50 34.50 35.50 35.50 39.00 42.00
10.2 0.00 0.60 0.60 1.60 2.10 3.10 3.60 7.10 7.10
10.3 -0.90 1.00 1.00 0.00 0.00 0.00 0.00 2.00 3.00 7.00 17.00
10.4 0.00 3.50 103.50 158.50 158.50 158.50 168.50 185.50 200.50 193.50 191.50
11.1 0.00 4.00 4.00 2.00 2.00 0.50 84.00 64.00 67.00 85.00 54.00
11.2 -0.50 28.00 -47.00 -47.00 -47.00 -47.00 -49.00 -49.00 -49.00 -42.00 -45.00
11.3 0.00 40.00 40.00 30.00 34.00 34.00 39.00 54.00 37.00 58.00 46.00
11.4 0.00 3.50 5.00 5.00 6.50 6.00 8.00 5.00 74.00 89.00 53.00
11.5 0.00 100.00 98.00 102.00 102.00 97.00 107.00 175.00 219.00 224.00
11.6 0.00 125.00 129.00 129.00 126.00 133.00 218.00 255.00 262.00 256.00 262.00
11.7 0.00 132.00 132.00 126.00 174.00 128.00 126.00 133.00 111.00 98.00 133.00
11.8 -1.50 121.50 121.50 118.50 122.50 116.50 115.50 110.50 113.50 113.50
12.1 0.00 0.00 1.50 2.70 10.00 14.00 17.00 30.00 35.00 36.50 41.50
12.2 0.00 35.00 38.00 38.00 36.00 38.00 161.00 164.00 170.50 168.00 173.50
12.3 0.00 180.00 183.00 166.00 176.00 137.00 291.00 304.00 296.00 301.00 318.00
12.4 0.00 138.00 125.00 130.00 130.00 235.00 235.00 235.00 235.00 246.00
12.5 0.00 205.00 205.00 187.00 202.00 323.00 303.00 331.00 340.00 340.00
12.6 0.00 245.00 224.00 224.00 218.00 224.00 307.00 307.00 325.00 316.00 349.50
13.1 0.00 0.00 1.00 1.00 1.00 7.00 23.00 27.00 27.00 27.50 27.50
13.2 0.00 1.00 1.00 2.00 2.00 3.00 68.00 62.00 74.00 75.00 78.00
13.3 0.00 3.00 29.50 44.50 50.00 52.50 11.00 103.00
13.4 -1.00 11.50 24.50 25.50 25.50 57.00 30.50 54.50 77.50 83.50 88.50
13.5 0.00 3.00 5.50 13.00 12.00 12.00 12.00 14.00 28.50 36.00 36.50
14.1 0.00 2.00 4.00 4.00 4.00 4.00 4.00 4.00 8.00 8.00 16.00
14.2 0.00 1.50 1.50 -0.50 1.00 1.00 1.50 1.50 1.50 1.50 6.00
14.3 0.00 -1.00 6.00 4.00 4.00 9.00 20.00 20.00 21.00 21.00 21.00
14.4 0.00 3.00 4.50 6.50 6.00 6.00 10.50 8.50 15.00 18.50 20.00
15.1 0.00 115.00 113.00 108.00 108.00 101.00 107.00 113.00 113.00 105.00 153.00
15.2 0.00 80.00 97.00 81.00 71.00 71.00 58.00 58.00 45.00 57.00 83.00
15.3 0.00 70.00 80.00 80.00 80.00 48.00
15.4 0.00 55.00 57.00 53.00 37.00 55.00 55.00 55.00 61.00 35.00
15.5 -1.00 77.00 61.00 67.00 61.00 68.00 98.00 112.00 132.00 135.00
15.6 0.00 11.00 1.00 -6.00 -7.00 34.00 34.00 36.00 46.00 46.00
16.1 -2.00 9.00 5.00 5.00 4.00 3.00 3.50 3.50 3.50 7.50 7.50
16.2 0.00 87.00 91.00 91.00 97.00 84.00 89.00 89.00 94.00 86.00 92.00
16.3 0.00 80.00 76.00 76.00 56.00 75.00 75.00 75.00 75.00 95.00
16.4 8.00 99.00 100.00 100.00 100.00 106.00 103.00 118.00 124.00
16.5 8.00 38.00 51.00 41.00 41.00 40.00 40.00 74.00 74.00 60.00
17.1 0.00 2.00 2.00 0.00 0.00 2.00 2.00 6.50 6.50 5.50
17.2 0.00 15.00 34.50 34.00 34.00 59.00 61.00 57.00 75.50 75.50 77.00
17.3 0.00 38.00 48.00 62.00 87.00 87.00
17.4 -3.00 89.00 238.00 245.00 206.00 208.00 213.00 226.00 214.00 235.00
17.5 0.00 200.00 200.00 212.00 210.00 444.00
17.6 0.00 248.00 249.00 250.00 459.00 489.00
18.1 0.00 5.00 4.00 5.00 -0.20 0.50 0.50 -2.50 -2.50 -2.50
18.2 0.00 12.00 11.50 9.00 9.00 9.00 7.50 7.50 5.00 5.00 5.00
18.3 -3.00 47.00
18.4 -1.00 -10.00 -9.50 12.30 19.80
18.5 0.50 9.00 9.20 -1.00 -4.20 7.50 9.00 11.50 12.00 14.00
18.6 0.00 9.00 12.50 12.00 11.50 11.00 12.50 12.50 19.00 26.00 28.00
18.7 0.00 18.00 -1.00 0.00 0.00 4.00 3.00 13.00 17.00 19.50
19.1 0.00 0.50 2.00 2.00 2.00 2.00 2.50 2.50 1.50 4.50 4.50
19.2 0.00 14.20 14.20 13.70 12.20 11.60 18.20 16.20 16.70 21.20 21.70
19.3 -1.00 20.00 -3.00 1.50 29.00 4.00 66.00 44.00 53.00
19.4 -1.00 -0.50 -4.50 -3.00 -3.50 -5.90 -5.90 -5.40 -6.40 -5.40 -6.40
20.1 0.00 -7.50 -7.50 -7.50 15.50 27.00 20.50 24.50 22.20 24.00 28.00
20.2 -3.50 12.00 13.50 16.00 14.00 12.50 12.00 11.00 11.50 13.50 14.50
20.3 0.00 2.50 3.00 8.00 19.50 25.50 27.00 27.50 26.00 23.50

Table 4.4. Measured cumulative bank bottom changes from November 2003 - September 2005.
Maximum bottom erosion was 444 cm.
76 Venoge River Data for Streambank Erosion Modeling

4.3 Historical and remotely sensed channel data

4.3.1 Available data

4.3.1.1 Historical maps

Lance and Consuegra (1998) investigated the list of available historical maps during the HYDRAM
Venoge Study. In the cantonal archives (1:50’000 scale) and those of the Federal Office of Topography
(Swisstopo) (1:25’000), maps from 1839, 1849, 1882, 1892, and 1945 are available. The current
1:25’000 maps of Swisstopo were first mapped in 1952 (map 1242 Morges) with a latest impression in
2000 (content 1998) and in 1955 (map 1222 Cossonay) with a latest impression in 1999 (content 1997)
(Swisstopo Status, 2005). The different revisions of these maps are of course available at Swisstopo. The
Vaud canton has a map at the 1:10’000 scale. The last revision of the digital maps was December 1990.

4.3.1.2 Remotely sensed data

An impressive quantity of remotely sensed data1 for the Lower Venoge River is available. Aerial photos
were taken in 1964 by the EPFL laboratory of photogrammetry as well as April 15, 1996. More photos
were taken by the Swiss railways (SBB) in March 1994. Besides these dates, photos were also taken in
1933, 1957, and 1974 (Lance and Consuegra, 1998), although their sources are unknown. Other photos
of the Venoge exist and are available at the Swiss Federal Office of Topography (Swisstopo). In their
recent campaign of orthophoto coverage of Switzerland in its entirety, called SwissImage, they produced
orthophotos of the entire Venoge catchment in 1998. These orthophotos were made available by the
Vaud Canton for this thesis.

Unfortunately, for the orthophotos of 1964, 1994, and 1996, no quality information was found
concerning the ortho-rectification of the photos. Concerning the orthophotos SwissImage, they have a
quality of 1 meter precision (under smooth relief conditions) with pixel sizes of 50 cm.

Besides the photogrammetric data available, recent advances in airborne laser scanning using LIght
Detection And Ranging (LIDAR) technology have permitted high resolution digital terrain models to be
developed. It was possible to buy from Swisstopo, the laser elevation points issued from their 2001-2002
flights over the Lower Venoge River. Two data sets were available: the MNS-MO - the surface model
points, as well as the MNT-MO, the actual terrain points. The surface model points are the original
LIDAR points. To take out points that represent objects (vegetation, buildings,...), and not the actual land
elevation, filters are employed resulting in the terrain model points of the MNT-MO. There are typically
1-2 points per square meter for the surface model points. MNS-MO precision is advertised as 1.5 meters

1. Remotely sensed data is defined by Burrough and McDonnell (1998) as: "the collection of data about
an object without coming into contact with it." In this broader sense, aerial photography is considered
in this research as remotely sensed data.
 Historical and remotely sensed channel data 77

standard deviation in wooded areas and 0.5 meters in open areas, and 0.5 meters for the MNT-MO
(Swisstopo MNS, 2005).

Processing the LIDAR point data proved difficult. Modern Geographical Information Systems (GIS) are
not necessarily equipped to handle the volume of point data supplied. This was true for the software,
Mapinfo® Professional®, in combination with the spatial analysis tool, Vertical Mapper®. LIDAR data
points are delivered by tile (size of 3 by 4.125 km) and this was approximately the size that was
reasonable for processing in the GIS. Points from adjacent tiles were added to the tile in question before
interpolation of the points. This was necessary to provide an overlap between tiles. A DTM and DSM
(Digital Surface Model) was then created with each set of extended tile points with the Inverse Distance
Weighting interpolation algorithm available in the Vertical Mapper® software (Northwood, 2001).
Although the algorithm is labeled as an interpolation program, it is used here more as an aggregation
algorithm (Metzger, 2003) due to the original point density of 1 - 2 points per square meter compared to
the one meter resolution DTM/DSM to be produced. For the DSM, the IDW was carried out using an
exponent of 2 for the exponential distance weighting and the nearest 10 points were used in a radius of
10 meters. For the DTM, an exponent of 2 was also used although the first 25 points encountered in a
radius of 250 meters were used. A lower radius and number of points for the DSM was used to avoid too
much smoothing. After the creation of each separate DTM/DSM, their edges were trimmed of 100
meters to cutoff any border effects and then the 6 tiles which make up the Lower Venoge stream corridor
were spliced together averaging any overlapping points. As a result, a DTM and DSM of the Lower
Venoge River stream corridor were obtained.

Besides the high resolution DTM and DSM that were developed, the MNT25 (DTM at 25 meter pixel
size) was also available from Swisstopo. This DTM is developed from the elevation contours of the
1:25’000 national topographic map and the advertised precision is 1.5 meters (Swisstopo MNT25,
2005).

4.3.2 Derived information

4.3.2.1 Lower Venoge River centerline

The first application for the historical and remotely sensed data was the development of the Lower
Venoge River centerline. The centerlines for the 1996 Lower Venoge as well as the 2004 Lower Venoge
were desired because of the cross sections achieved approximately during these periods. As simple as
the task might seem, it proved itself to be much more difficult. Digital river data is available from the
Vaud canton as well as from Swisstopo (Vecteur25), but this data was not considered because of the
precision necessary. Using the available maps to reference the river centerline was also avoided due to
the map scale or the year of revision of the maps. Quality issues of the 1996 ortho-rectification were
raised when planimetric differences of up to 4 meters were found in places compared to the Swissimage
1998 orthophotos. To produce the 1996 centerline, it was finally decided to base it on the 1998
Swissimage orthophotos and where significant channel change was obvious between the 1996 and 1998
photos, the 1998 line was changed in those places to produce the 1996 channel centerline. Producing the
78 Venoge River Data for Streambank Erosion Modeling

1998 centerline was not always easy though because of the vegetation that appears in the Swissimage
orthophotos (based on summer aerial photos).

The task was even more complicated for the 2004 centerline because there are no more recent
orthophotos than those of 1998. To approximate the channel centerline changes that needed to be
incorporated into the 1998 centerline to make the 2004 centerline, the LIDAR DTM and 2004 cross
section (cf. §4.4) were used. Figure 4.16 show the most important channel changes during the course of
1996-2004 for the study reach.

2004
2004 centerline
centerline
1996
1996 centerline
centerline

Reproduced with permission

from swisstopo (BA068334)

0
000
0 100
100
100
100
100 200
200
200
200
200 0
00
00
0 100
100
100
100
100
100 200
200
200
200
200
200

meters
meters meters
meters

Figure 4.16. Channel centerlines determined with 1998 orthophotos and 2001/2002 LIDAR data and
2004 cross sections. Background images: SWISSIMAGE © 1998, Swisstopo (DV012716).

4.3.2.2 Channel migration and erosion estimates

It is often cited in literature of the possibility to estimate erosion based on two sets of aerial photos
(preferably orthophotos). Lawler (1993) includes using historical sources for medium timescale
measurement of river bank erosion. He warns of the following problems that must be considered with
the use of historical information: "assumption of continuity and/or linearity in channel, change of time -
representativeness of temporal sampling, survey errors/plotting errors/map distortion, confusion over
map revision, change in channel definition criteria between two surveys, misleading information, and
possibility of contradictory evidence". Although the problems Lawler cites are important and could
completely negate the use of a historical source, it is still possible to glean significant information from
them.
 Channel topography 79

Lance and Consuegra (1998) were able to make general conclusions on the bank erosion, channel
changes and stream corridor width of the Lower Venoge River by analyzing the different historical maps
and orthophotos available to them. Concerning the study reach of this thesis, they concluded the
following:
• the meander near km 10 (see appendix 2 for locations), progressed very little according to the
historical maps (1850-1950), but seems to be progressing more rapidly in the last 30 years, most
likely due to the industrial zone implanted on the raised right bank,
• from the "Abbaye" weir (km 7300) to the federal stage recorder weir (km 4900), the river planform
has changed very little. The channel is deeper in this section.
• from the federal stage recorder weir (km 4900) to the football field (km 4100) the river planform has
changed little in the last century because of bank protection works although the river will probably
regain ground as the bank protection works will not be maintained. From the football field down to
the "Valeyre" weir (km 2150), bank failures are important and seem to be occurring more frequently
than in the last century. New meanders are developing.

The use of the aerial photos and the LIDAR DTM was already seen in section 3.7.1 to help classify the
Lower Venoge River type and to better understand the meander cutoffs that have occurred in the last
forty years.

It is of primary interest for this study to evaluate the use of the aerial photos for a quantitative assessment
of mean annual erosion. Due to the fact that Swissimage orthophotos are taken in the summer, nearly all
of the study reach banks in the 1998 orthophotos are covered by vegetation. At some places, vegetation
completely masks the river channel. As mentioned above, the 1964 and 1996 orthophotos have accuracy
problems. Thus only along river reaches where there is another fixed reference on the map, and the bank
is relatively visible, is it really possible to use the 1964 and 1996 orthophotos to evaluate relatively
precisely (±1 meter). Due to the very partial information that could be obtained with the aerial photos, it
was decided not to try to estimate annual erosion rates over the entire study reach.

4.4 Channel topography

Probably one of the greatest hurdles to overcome in streambank erosion modeling is that of the
topographic data of the stream channel. Banks shapes and slopes can vary significantly even over tens
of meters, due to soil and vegetation heterogeneities. Lane et al. (1994) estimated bedload transport rates
and aggradational and degradational zones of the stream bed by repeated tacheometric surveying. They
concluded that bed cut and fills could be estimated to 20% of the correct value with cross section spacing
of 2 meters. Although their study on bedload transport is a different context, it does reflect the spacing
necessary to define changes in the channel bottom accurately. This amount of cross section data is
impractical for a long stream reach.

High resolution LIDAR can provide a very accurate DTM. Its application in the river environment has
been investigated and it has been found though to be accurate only in exposed, unvegetated channels
(Charlton et al., 2003). Accuracy suffers also on steep slopes because the ground surface becomes greater
than the horizontal projected surface reducing the number of points per surface area. LIDAR surveying
80 Venoge River Data for Streambank Erosion Modeling

usually uses an infrared frequency for measurement. This frequency will not penetrate water. Thus, when
riparian vegetation is sparse, it is possible to pickup the channel water surface. Other LIDAR techniques,
such as the Scanning Hydrographic Operational Airborne Lidar Survey (SHOALS), have been
developed using other frequencies so bathymetry can be detected (Irish and Lillycrop, 1999). The
SHOALS system and other related remote sensing techniques show the future for obtaining high
definition channel topography. Until these methods become widespread, proper interpolation methods
need to be used to approximate the channel bottom to be combined with a high resolution DTM issued
from LIDAR.

Many cross sections of the Lower Venoge River were surveyed in 1995-1996 during the HYDRAM
Venoge study and during a fieldwork campaign of EPFL environmental engineering students. A
comparison of those cross sections with cross sections surveyed in 2004 during this thesis permits an
evaluation of the stability of the thalweg. Cross section density is insufficient in these two measuring
campaigns to provide accurately interpolated meander scour holes throughout the study reach. A new
interpolation method is presented to improve interpolated cross sections.

LIDAR data was available for this thesis. In areas of significant riparian vegetation and steep
streambanks, LIDAR terrain point density suffers and a poor topographic representation results. A linear
interpolation method is presented to improve river representation in a LIDAR derived DTM. This
method interpolates linearly between channel LIDAR points in the streamwise direction. This linear
interpolation could also be applied to integrate real cross section data into the DTM. The improved DTM
could be very significant for developing a regional erosion model as described in chapter 2.4.1.

4.4.1 Cross section data

During the 1996 HYDRAM Venoge study, 98 cross sections were surveyed in the Lower Venoge River
from kilometer 0.7 to 27. Cross sections locations were primarily at hydraulic structure locations and in
urban areas. From kilometers 0 to 22.5, it was decided to reprofile some of the 1996 cross sections and
add other cross sections in areas insufficiently covered in 1996, especially in the areas of erosion
monitoring. One hundred and fifty-five cross sections were surveyed in 2004. After the 2004 survey,
other cross sections from the EPFL environmental engineering students fieldwork campaign in 1996
were found in the archives. One hundred and five new cross sections were added to the 98 cross sections
to give a total of 203 for 1996. No x-y reference points were found for the 1996 fieldwork campaign cross
sections so their exact location is unknown. The majority of the locations of the cross sections could be
estimated though from paper maps found in archives.

The measured cross sections allowed for detailed longitudinal profiles to be plotted (figure 4.17). These
plots show that during the eight year period little bed elevation movement occurred. The bed of the
Lower Venoge River seems to be stable, except the reach from kilometer 7.3 to 7.7 seems to have
degraded. In a personal communication, Bujard (2005) related the history of the "Abbaye" weir. It failed
during the March 9, 2001 high flow. To combat the regressive erosion that was occurring after this
failure, a temporary reparation was done in October 2002. Large boulders were put in place to simulate
the effect of the old weir. Two other areas have possibly changed their bed elevations: downstream from
the Abbaye weir, and in the vicinity of kilometer 12. Due to insufficient cross sections and uncertainties
 Channel topography 81

in the exact location of some of the 1996 cross sections in these areas, it is impossible to confirm the bed
movement. It is possible in both locations that the low points from 1996 all come from scour holes and
the high points of 2004 come from riffles, although the thalweg values downstream from the "Abbaye"
weir seem too big to be explained simply as scour holes.

395
1996 2004
Change
390
elevation (m.a.s.l.)

385

380

Change?

375

370
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
distance from outlet (m)

425
1996 2004
420

415
elevation (m.a.s.l.)

410

405

400

395

390 Change?

385
8500 9500 10500 11500 12500 13500 14500 15500 16500 17500
distance from outlet (m)

455
1996 2004
450

445
elevation (m.a.s.l.)

440

435

430

425

420

415
17000 18000 19000 20000 21000 22000 23000 24000 25000 26000
distance from outlet (m)

Figure 4.17. Longitudinal profiles of the Lower Venoge River based on cross sections from 1996 and
2004. Points indicate cross section locations.
82 Venoge River Data for Streambank Erosion Modeling

1996 channel channel

Besides the evaluation of the stability of the
profile id width (m) km 2004 width (m) longitudinal profile, cross sections were analyzed
360 24 10.336 26 in the study reach to see if average channel width
6012 23 9.413 24 was changing. It was possible to compare 16 cross
6011 25 9.262 27
sections for channel width changes. Some of the
6010 24 9.158 24
6009 30 9.126 27 cross sections are not exactly at the same place so
370 22 8.228 21 it is normal to expect some difference. Table 4.5
381 21 8.002 16 shows that on average cross section width has not
400 21 7.602 19
changed. Some cross sections have changed
430 19 6.117 19
440 18 5.702 18 though because of the dynamic nature of the river.
450 20 5.515 15 Other differences are due most likely to slight
8007 26 3.5 26 location differences.
470 16 3.005 19
9001 19 2.841 25
The locations of the 2004 cross sections that will
9002 22 2.752 16
9003 19 2.682 26 be used for the erosion modeling are seen in figure
average 21.81 21.75 4.2. The CCHE1D model asks for cross section
data in a text file format. It is possible, though, to
Table 4.5. Comparison of 1996 and 2004 cross
avoid the text export with careful preparation of
section widths
the GIS cross section database. To help in the cross
section database preparation, a cross section tool was programmed. As will be seen in this section, the
cross section tool became much more than simply a viewer. The interface of the cross section tool is
shown in figure 4.18. Most all GIS’s will open and work with dbf files. The strategy for the 2004
surveying was to do it in 3D coordinates rather than by leveling. GPS points were implanted during the
spring of 2004. Tacheometric (Total Station) surveying was then performed using the GPS points as base
points. These points were then imported into Mapinfo Professional® and saved as a dbf file. These dbf
points are not possible to visualize rapidly without some programming in Mapinfo®. Rather than being
dependant on a GIS software, it was decided to program in Borland® Delphi®.

CCHE1D needs the cross sections to be ordered according to a cross section identifier and according to
its position from left to right. Once the dbf points have been identified by cross section via manipulation
in a GIS, and the leftmost point has been indicated as the first cross section point, it is possible with the
cross section tool to calculate the cross section transverse distance, cp_w, and its cross section position
index, cp_posidx. Once this has been done, it is easy with the viewer to visualize where the bank top and
bank toe are located and to check the appropriate boxes in the database spreadsheet to assign them.

Other capabilities were also added to the cross section tool. Unnecessary points can be removed from
the cross section by running the smoother and points can also be added. Parts of the cross section can be
labelled as the left overbank, left bank, main channel, right overbank, or right bank. This labelling will
be used in other routines in the cross section tool and is helpful for other GIS manipulations.

If cross sections are numerous enough and correctly spaced, a linear interpolation of cross section points
from one cross section to the next will give a sufficient approximation of the channel topography. As
mentioned in chapter 2.4.2, nodes for the CCHE1D calculations are selected in a GIS. If these nodes are
in dbf format, necessary for CCHE1D, they can be opened in the cross section viewer with the "Node
Points" tab. This part of the cross section tool allows the user firstly to recalculate the x-y coordinates of
 Channel topography 83

cross sections to straighten the points in a perpendicular fashion around the channel centerline if desired,
although it is not absolutely necessary. Secondly, cross sections can be linearly interpolated. This
interpolation is based on master chords that run from a definition point on one cross section, such as the
left bank top, to the corresponding definition point on the next cross section (figure 4.19). This
interpolation scheme is similar to the scheme used in HEC-RAS (USACE, 2002), although HEC-RAS
uses a thalweg master chord rather than two toe master chords.

Figure 4.18. Database file cross section tool for viewing cross sections in dbf format. The tool
interpolates cross sections linearly and with respect to bend curvature. The tool can also interpolate
and smooth LIDAR topographic points to estimate channel geometry.

The interpolation of the cross sections based on node locations ensures a cross section at each node
location for CCHE1D. The CCHE1D model needs two other information for the cross section points. A
roughness value, cp_rgh, that of Manning, must be specified for every point, as well as an effective flow
logical value, cp_block, which if set to "0" means that no flow will be calculated from that cross section
point to the next transverse point. In the case of the Venoge study reach, effective flow was only set to
"0" for bridge pillars. These values are updated in Mapinfo Professional® via Manning and ineffective
flow polygons as shown in figure 4.20.
84 Venoge River Data for Streambank Erosion Modeling

Right end chord

Left end chord

Left toe chord

Right toe chord

Right bank top chord Left bank top chord

Figure 4.19. Linear cross section interpolation based on master chords between cross section
definition points

0
00
00
0 10
10
10
10
10
10 20
20
20
20
20
20

meters
meters
meters
meters
meters
meters

Reproduced with permission

from swisstopo (BA068334)
Ineffective flow area

Homogeneous
Homogeneous
Manning
Manning zones
zones
Example Manning Values
ID: 3
LOB: 0.120
LB: 0.070
MC: 0.040
RB: 0.070
ROB: 0.120

Cross section points

0
00
00
0 100
100
100
100
100
100 200
200
200
200
200
200

meters
meters
meters
meters
meters
meters

Figure 4.20. Attribution of Manning values and ineffective flow areas to cross section points within a
GIS. Background images: SWISSIMAGE © 1998, Swisstopo (DV012716).
 Channel topography 85

4.4.2 Interpolation in meanders

4.4.2.1 Development
The number of cross sections necessary to roughly describe a channel bottom would be one for each
riffle and one for each scour pool. With Lower Venoge River wavelength roughly in the 100-160 meter
range, this would mean a cross section every 25-40 meters. Because of erosion and depositional
complexities that occur due to flow complexities, bed changes between a riffle and a scour pool will not
follow a linear interpolation between the riffle cross section and the bend cross section. This is the reason
for Lane et al.’s (1994) recommendation of 2 meter spacing for an 80% accuracy in calculating cuts and
fills. For the erosion modeling, a cross section for every riffle and scour pool is probably sufficient, but
finding the deepest scour hole can be time consuming as for many rivers it requires the use of a boat.

Cross section surveying of the Lower Venoge River in 2004 is sufficiently dense in the bends that are
monitored for erosion, but elsewhere is insufficient. For this reason, it was decided to explore the
possibilities of interpolating cross sections with respect to channel curvature. A theoretical "model of
equilibrium bed topography for meander bends with erodible banks" has been proposed by Darby and
Delbono (2002), but it was chosen here to follow a calibration approach that is much simpler to program.
The bend scour pool location is related to the zone of maximum outer bend shear stress. The location of
that maximum shear stress and maximum migration rate have been estimated via the use of weighting
upstream migration rates by the equation 2.15. It is proposed that by weighting upstream curvature, it
will be possible to estimate the radius of curvature that can be used with equation 2.5 to estimate bed
topography. The equation for bed topography then is:
h rw ⎞ K
------ = a ⎛ ------- Eq. 4.3
hm ⎝ r cw⎠

where h is the water depth, hm is the mean cross section water depth, rw is the weighted radius of
curvature at a cross section bottom point, rcw is the weighted radius of curvature at the channel
centerline, and a and K are calibration coefficients. The coefficient a is introduced because it is observed
that at the channel bottom center in bends the flow depth is often slightly lower than hm. An a value in
the 0.75 to 1 range permits this adjustment. Inspired by equation 2.15, the weighted curvature, c1, can
the be estimated by the following equation:
∞ ∞ –1

c1( s ) = ∫ c0 ( s – ζ )G ( ζ ) dζ ∫ G ( ζ ) dζ Eq. 4.4

0 0

where c0(s-ζ) is the channel centerline curvature at a distance of ζ upstream from point s and G(ζ) is the
weighting function. Curvature and radius of curvature are related by c = 1/rc. It is proposed to use the
same exponential decay weighting function as in equation 2.16 that uses the decay parameter α.

As the bankfull discharge is considered to be approximately the formative channel discharge, and the
bankfull discharge is often between a one and two year return period (Bravard and Petit, 1997), the mean
86 Venoge River Data for Streambank Erosion Modeling

flow depth, hm, can be based on the maximum water surface elevation of the simulation of the one year
return period flood (or the steady flow simulation of the one year return period maximum discharge).

The values of α, a, and K need to be evaluated for each node. It is possible to linearize equation 4.3
permitting a linear regression to be performed to find an optimal a and K compared to a measured cross
section and that for a given α. Upper and lower bounds for these three parameters should be set to avoid
unrealistic values. For the study reach, ranges of realistic values are 5-50 meters, 0.75-1, and 1-5 for
1/α, a, and K, respectively. The coefficient of determination can then be used to judge the fit of the
linearized form of equation 4.3 to the cross section bottom data. Toe points are removed from the linear
regression fit because toe geometry is variable depending on the status of basal sediments. To avoid bias
due to poorly distributed bottom points, the linear regression and coefficient of determination is based
on evenly spaced interpolated bottom points. The values of α, a, and K giving the best coefficient of
determination are then written to the node database. Node points in between real cross sections are given
estimations of the three variables according to a linear interpolation of 1/α, a, and K between the
upstream and downstream real cross sections. The interface for determining these values in the cross
section tool is shown in figure 4.21.

Figure 4.21. Calculation of α, a, and K values based on measured cross sections

The cross section tool also allows a text file to be saved containing calculated channel curvature values
at measured cross sections as a function of the value of α. This output is useful to determine if the step
of the 1/α value was sufficient. Several distributions of curvature as a function of the inverse of alpha
are shown in figure 4.22. It can be noticed that channel curvature is sometimes maximized at values of
the inverse of alpha greater than 5. This response is of course to be expected and allows for a scour hole
to be located downstream from the bend apex.
 Channel topography 87

0.030

0.020

0.010
Curvature (1/m)

0.000

-0.010

-0.020

-0.030
5 10 15 20 25 30 35 40 45 50
1/alpha

Figure 4.22. Ten examples of the influence of the parameter alpha on the upstream weighted curvature

4.4.2.2 Procedure
The procedure then to interpolate cross sections with respect to meander curvature using the model
CCHE1D and the cross section tool then is the following:

1. Prepare nodes for CCHE1D as indicated in chapter 2.4.2.

2. Linearly interpolate the cross sections with the cross section tool.
3. With the linearly interpolated cross sections, run a CCHE1D calculation with fixed bed and banks to
calculate the maximum Water Surface Elevation (WSE) at each node.
4. Open the linearly interpolated cross sections, the cche1d nodes, and a text file containing the WSE
results for each node.
5. Calculate optimum α, a, and K values using the cross section tool.
6. For poorly fitted cross sections or zones where cross sections are sufficient, check "fixed_bed" in the
node database file to keep the originally interpolated cross sections.
7. Interpolate with respect to curvature using the cross section tool.
8. Pack the cross section table to eliminate deleted bottom points.
9. Recalculate transverse distance, cp_w, and the position index, cp_posidx.
10.Compare real cross sections with fitted ones.
88 Venoge River Data for Streambank Erosion Modeling

The above procedure has been done in connection with the model CCHE1D, although any other 1D
hydraulic model could be used to calculate the WSE’s.

4.4.2.3 Results
The meander interpolation procedure was applied to the 82 cross sections that make up the erosion
modeling study reach. Four of those cross sections are related to the weir in the middle of the study reach.
They were excluded from bed topography fitting. The sum of the coefficient of determination (R2) is
output for each cross section linear regression. If the weighted channel curvature forces a bed slope
opposite of that observed, the fit is considered bad and the R2 is zero in this case. The cross section tool
allows two modes for calculating optimum α, a, and K values. Alpha can be fixed for the entire zone or
it can be variable. For the case when alpha must remain constant throughout the reach, an optimum at
α=0.05 was found with a sum of the R2 = 41.3 and 24 bad fits out of 78 cross sections (figure 4.23). The
cross section tool also outputs the sum of the coefficients of determination without respect to any
constraints to give an impression of how well the model of bed topography performs. This unconditioned
sum of the R2 = 58.5. When alpha is allowed to be variable, the sum of the R2 = 48.2 and there are 20
bad fits. Using a variable alpha does improve the result significantly. With a result of 48 out of the 59
possible for the 78 cross sections, the bed topography model used here is capable of reproducing the
majority of the bed slopes. Examples of different bed fits are shown in figure 4.24. The top two figures
show well the basal sediments at the toe, while the basal sediments seem to have been eroded away in
the bottom cross section. The top cross section bed fit is probably a good estimation of the cross section
when basal sediments have been eroded away.

50 60

50
40

40
Number of bad fits

30
sum of R2

30

20
20

10
10
Sum R2
Bad fits
0 0
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
alpha

Figure 4.23. Results of the search for optimum parameters for the bed topography model applied to the
Venoge study reach
 Channel topography 89

Figure 4.24. Examples of best fit calculated channel bottoms. R2 values are 0.99, 0.63, and 0.36 for the
top, middle, and bottom cross sections, respectively.

To avoid poor interpolations, areas where poor fits occurred were marked "fixed_bed" in the nodes
database and original linearly interpolated cross sections were maintained in these zones. Although, it is
sure that some of the cross sections interpolated with respect to channel curvature will overestimate or
underestimate some bend scour holes, it will in the majority of cases help to correct very poor linear
interpolations as shown in figure 4.25. More objectively, real cross sections that were not included in the
parameter estimation should be compared to linear and meander interpolated cross sections to quantify
the improvement brought by the new interpolation method. Real cross sections measured in 1996 are
90 Venoge River Data for Streambank Erosion Modeling

compared with the linear and meander interpolated cross sections of 2004 (figure 4.26). This comparison
is permitted in that the channel bed elevation seems relatively stable in this zone according to the
longitudinal profile comparison (figure 4.17). The positioning of some of the 1996 cross sections is
uncertain though and this can be very critical. Sometimes the improvement when compared to the linear
interpolation is small, but often the meander interpolation places the most appropriate cross section just
upstream or downstream because the weighted radius of curvature evidently did not respect a linear
interpolation of the meander interpolation parameters. This means that upstream influence is extremely
variable and/or the digitization of the channel centerline does not accurately reflect the flow curvature.
The fact that interpolated cross sections are displaced too much upstream or downstream is important. It
can be suggested for cross sections that have been meander interpolated, erosion results should be
considered to apply to the upstream and downstream nodes (possibly several nodes in each direction if
the displacement is long or if the node spacing is close).

Figure 4.25. The meander interpolation method can significantly improve interpolated cross sections in
bends such as this one.

To complete the verification of the meander interpolated cross sections, a simulation of the one year
return period discharge was done with the new cross sections. Average absolute WSE difference was 1
cm and the maximum difference was 9 cm. This amount of difference is accepted for this study, although
the source of the difference should be investigated. It could be linked to the lower limit of a being too
low or more likely to higher point bar elevations that reduced effective flow capacity.
 Channel topography 91

Station Meander Linear Improvement 1996 real 1996 id Comment

scour scour scour
(m) elevation, side elevation, side (cm) elevation, side
8937 391.2F
8930 390.9R 391.2L 30 390.65F 6007
8728 390.2R
8721 390.3R 390.2R 5 389.9L 6005 1996 profile most likely at 8707
8707 390.2L 390.1R 50
8588 390.0F 389.6R 40 389.1C 6004
8581 389.3R
8488 389.2R 389.2R 0 389.3R 6003
8498 389.3R
8247 387.7L 388.3C -60 389.6C 370
8238 388.6F 388.2C
8213 388.2L 388.7L 50 387.6L 371
8221 387.6L 388.5C 90
8070 388.7R 388.6L 50 388.2R 380

Figure 4.26. Comparison of some 1996 measured cross sections with linear and meander interpolated
cross sections. Improvement is based on the difference of outer bend elevation between linear and
meander cross sections. F=flat, R=right, L=left, and C=center.

4.4.3 LIDAR and river topography

Airborne laser scanning has the potential of providing significant river topographic information. A DTM
derived from a non-water penetrating laser scan has the potential to accurately describe the water surface,
exposed bars, and channel banks if the last return laser point density is sufficient. This is usually not the
case for most airborne laser scans due to the effect of significant riparian vegetation. For the Lower
Venoge stream corridor, the average point density is around 1 point per square meter while the density
in the radius of 30 meters around the channel centerline is on average 1 point per 2 square meters.
Although, 1 point/2 m² may seem quite good, there are patches where the density is much less than this
causing poor river representation for most interpolation methods. Typical laser river cross section
interpolation problems are: widening, constriction, and bed raising (figure 4.27 shows examples from
the Lower Venoge River). Of course, the LIDAR derived DTM river topography will depend on the
interpolation method employed. Most GIS packages do not offer a polyline interpolation method which
is assumed to be necessary to improve the channel topographic description.

For an accurate description of the river topography, accurate bathymetric measurements must be made,
by appropriate laser scanning, tacheometric surveying, or sonar sounding, to be combined with standard
airborne laser scanning. Such combined surveys are rare. The next best possibility is to combine
surveyed cross sections with LIDAR topographic points. For such a method to give good results, a linear
streamwise interpolation method must densify the channel points before integration with the non-
channel airborne laser scanned last return points. Such a streamwise linear interpolation method is
elaborated here and the procedure for the integration of the points with a LIDAR derived DTM is
explained.
92 Venoge River Data for Streambank Erosion Modeling

Figure 4.27. Examples from the Lower Venoge River comparing cross sections extracted from a LIDAR
derived DTM and surveyed cross sections showing the interpolation problems of widening (top), bed
raising (middle), and constriction (bottom).

4.4.3.1 Linear streamwise interpolation

Linear and meander based cross section interpolations have been explained in the previous sections.
These methods are, of course, appropriate methods of interpolation, but require a certain effort in
defining cross section geometry and performing hydraulic modeling before the interpolations can be run.
To avoid these preliminary tasks, it is proposed here to linearly interpolate cross section points based on
real upstream and downstream points that are at a given perpendicular distance to the channel centerline.
 Channel topography 93

Such a method can be applied to the original channel LIDAR last return points or based on surveyed
cross section data.

The cross section tool was extended to perform this linear streamwise interpolation. Cross section points
can be created according to channel centerline node points at regular transversal spacing to an equal
width on each side of the channel node point. These points (called laser cross section points here) are
easily mapped in a GIS (figure 4.28). Surveyed cross section points and/or original LIDAR river points
can then be attributed to the laser cross section points (figure 4.28). After this step, the points can then
be opened again by the cross section tool to perform the interpolation. This interpolation is performed in
two steps: first, cross section points are interpolated transversely such that any null value point that is
surrounded by non-null points is interpolated, and secondly, interpolation proceeds by cross section
position index in the streamwise direction (shown by the example interpolation line in figure 4.28).

Figure 4.28. Top - Created cross section points (black) around the channel centerline (light grey).
Dark grey circles show an interpolation line. Bottom - White circles show points that were attributed a
laser point that will be used to interpolate elevation values onto points having a null elevation.
94 Venoge River Data for Streambank Erosion Modeling

This streamwise interpolation causes poor original laser points to propagate their error downstream. This
is often seen by abnormal bumps in the cross section. A smoothing algorithm can be applied to remove
these bumps. The algorithm was programmed to use only local minimums for the channel bottom and
all intermediate points are interpolated between the local minimums. The channel bottom is defined as
the elevation halfway vertically between the channel minimum and the lower of the starting and ending
laser cross section points. After this bump removal, an additional smoothing is run based on a moving
average over three points except for the minimum point that is preserved.

4.4.3.2 Creation of a DTM with streamwise interpolated cross section points

The procedure for the creation of a DTM based upon the streamwise interpolated cross section points is
then the following:

1. Laser cross section points are created such that the perpendicular cross section length is two times
that of the maximum estimated river width.
2. Laser points (or surveyed cross section points) are attributed to laser cross section points. For the
Lower Venoge River, any original laser point within 1 meter of the laser cross section point was
attributed to that point. Points are averaged if several points are within the attribution radius.
3. Laser cross section points are then interpolated and smoothed in the cross section tool.
4. A DTM as well as a grid of the slopes of the linearly interpolated laser cross section points are made.
For the Lower Venoge River, an Inverse Distance Weighted (IDW) interpolation was performed
based on exponential weighting of 2 on the five closest points to create the DTM. A slope model can
then be derived from the DTM in a GIS (figure 4.29).
5. The channel area is digitized from the slope model.
6. All laser points are opened and the channel area points are deleted.
7. Laser points without the channel area are interpolated.
8. The DTM produced in step 4 is trimmed according to the digitized channel area of step 5 and this
river channel DTM is spliced onto the DTM of step 7 to produce the final DTM.

The above procedure was adapted slightly for the Lower Venoge stream corridor because the DTM had
already been developed for the Lower Venoge without any special river interpolation (c.f. section 4.3).
In this case, step 7 involves the following before performing step 8:

7. All laser points are interpolated together. For the Lower Venoge stream corridor, an IDW
interpolation based on exponential weighting of 2 on the 25 closest points was used to create the
DTM.
7a.A grid of the river banks without the channel is created. For the Venoge, this was done over a
distance of two times the estimated maximum river width.
7b.This grid is maximized onto the grid created in step 7.
 Channel topography 95

4.4.3.3 Results
Due to differences in dates of cross sections and the LIDAR points, and also due to an insufficient
spacing of cross sections in some areas, it was chosen to interpolate LIDAR channel points streamwise
instead of surveyed cross section points. A DTM was created upon the steps elaborated in the previous
subsection. Although some artifacts remain from the linear interpolation and smoothing algorithms, the
bank outline is significantly better in most areas. Comparison of the river integrated DTM and the non-
river DTM shows a difference in channel volume of 119’260 m³, giving an average change in depth of
0.2943 m. The channel surface area is 405’208 m² for the downstream 18’488 m of length giving an
average channel width of 21.92 m.

Figure 4.29. Comparison of the LIDAR derived non-river integrated DTM (top left) and corresponding
slopes (bottom left) and the river integrated DTM (top right) and corresponding slopes (bottom left).
LIDAR points come from DTM-AV ©2004, Swisstopo (DV043683).

4.4.3.4 Possible future developments

The improved channel definition obtained with the river integrated LIDAR DTM and the cross section
interpolation methods leads to the following possible future developments:

1. Using the LIDAR DTM based only on LIDAR points, a relatively good water surface elevation
indicative of the energy slope can be obtained. This was tested for the Lower Venoge and the results
96 Venoge River Data for Streambank Erosion Modeling

seem good although more detailed survey data would be necessary to validate it (figure 4.30). The
LIDAR DTM water surface longitudinal profile has significant noise. The profile was smoothed by
the following procedure: the upstream elevation was used as the starting elevation, and then going
downstream, the next downstream water surface point was not added to the longitudinal profile until
its vertical difference was at least 50 cm. This was continued until the outlet. The longitudinal
profile minimum was used as the outlet value. A good energy slope estimation could be very
important for a regional streambank erosion model (see chapter 2.4.1).
2. Bank tops are much more clearly defined (figure 4.29). Either by manual digitization or by
automatic detection, bank tops can be delimited. This would permit the bank height to be estimated
from the LIDAR DTM which is an important variable in bank failure probability. Bank slopes could
also be approximated. Again this bank information could be used in a regional streambank erosion
model.
3. Cross section data that is from a different date than that of the LIDAR data and in an insufficient
quantity is more frequent than the contrary and thus a procedure to properly integrate such survey
data with a LIDAR DTM should be developed. Special attention needs to be taken when channel
migration has occurred or when the bank heights and channel widths are significantly different than
the interpolated cross sections. Such an integration method could be significant for streambank
erosion modeling because it would improve the bank height definition of interpolated cross sections.

395
2004 survey 2002 lidar 2002 lidar smoothed

390
elevation (m.a.s.l.)

385

380

375

370
0 2000 4000 6000 8000 10000
distance from outlet (m)

Figure 4.30. Approximation of the longitudinal water surface elevation profile with LIDAR data and
comparison with the thalweg based on surveyed cross sections. LIDAR data from DTM-AV ©2004,
Swisstopo.
 Bed 97

4.5 Bed

4.5.1 Necessary input data

The bed data needed for use in a mobile bed hydraulic model are porosity, depth to bedrock, and the
Grain Size Distribution (GSD) of the subpavement. Besides these data, the CCHE1D model uses a
logical variable, a non-erodibility variable, that can be turned on if a river bottom should not be eroded.
No bedrock exposure was found along the study reaches, so the depth to bedrock can be assumed to be
great and a very large value will be used in the model to allow unlimited degradation. The non-erodibility
option will be used only in the vicinity of the "Abbaye" weir. Rather than try to estimate an appropriate
GSD for this improvised, large boulder weir, it was decided to consider the bed to be non-erodible at the
brink down to its base. Concerning bed porosity, the tests necessary to determine it regularly along the
reach would be too time consuming as important volumes of bed sediment would have to be analyzed.
Values in literature range from 0.25 to 0.45 with more typical values being in the 0.30-0.40 (Church et
al., 1987). It was decided to use the standard porosity value of 0.4 used in previous studies done with the
CCHE1D model.

4.5.2 Grain size distribution

The Grain Size Distribution (GSD) of the bed sediment is of course a very important input data for the
determination of the bedload transport. With the concave longitudinal profile of the Lower Venoge River
it is to be expected that the average grain size will diminish as one progresses downstream. Thus, an
appropriate sampling method will need to be employed to detect the significant changes in this
downstream fining. The sampling method must also take into account that the subsurface needs to be
sampled, which is not the same as the surface layer due to armoring.

4.5.2.1 Sampling techniques

There are numerous possibilities available for the determination of the GSD of a gravel bed river, which
were initially described by Kellerhals and Bray (1971) and more recently by Petrie (1998):

1. Volume sampling: This method involves extracting a volume from the subsurface layer for sieve
and gravelometric analysis. Church et al. (1987) calculated sample sizes based on the percentage of
the weight of the largest stone compared to the sample weight. They calculated for a 10 cm
maximum diameter (approximate Venoge study reach dmax) that the 0.1% criteria would necessitate
a sample size of 1400 kg. They show also that the 0.1% criteria is similar to the ISO low precision
standard 4364-1977(E). Fehr (1987) suggests that a volume analysis needs a size of 2.5 × d max
where dmax is in m and the estimated volume is in m³. To calculate the GSD, the proportion of
sediment in predefined classes is calculated by dividing the class weight by the total weight.
2. Grid sampling: This method involves placing a grid on the surface and sampling grains located on
98 Venoge River Data for Streambank Erosion Modeling

each grid point. The b-axis, the middle dimension of the ellipsoid, is measured. Various grid spacing
techniques exist: Wolman pacing (1954), using a wire grid, or using a survey tape at regular spacing.
The grid spacing can be variable within an isotropic reach. The main criteria to respect is to avoid
bias in selecting grains. It is easier to select larger grains that are more visible. The biggest problem
with grid sampling is that there will be a truncation limit at approximately 1 cm because of the grain
size one can reasonable pickup with his/her hand. Various authors (Hey and Thorne, 1983; Wolman,
1954; Kondolf, 1997; Fripp and Diplas, 1993; Rice and Church, 1996) have proposed sample sizes
in the range of 40 - 1200 stones based on their replicate sampling tests, and the parameter and
accuracy sought. Recently Petrie and Diplas (2000) made an important advancement in determining
sediment sampling accuracy. They applied binomial and multinomial distributions for the
quantification of the number of stones to be sampled for a given confidence interval for a given
diameter (binomial) or for a given number of classes (multinomial). For example a confidence
interval of 95% with a bounds of ±10 percent finer on an analysis using 10 classes would need a
sample of 190 stones.
3. Areal sampling: It involves picking up all exposed surface grains in a given area for the
determination of the GSD. The proportions of each class can then be determined by weighing or
counting the grains. Usually this is done by weighing because a small grain size limit can be
obtained according to the pickup procedure. The pickup procedures usually involve adhesives: clay,
wax, or tape; or use spray paint to help in hand removal. Clay is often used because it can be used
underwater. There is unfortunately an upper size limit to what the adhesives can pickup which is
approximately a diameter of 4 cm. Analysis of photos is also possible, but is not recommended due
to the b-axis being hidden at times.
4. Transect sampling: This method is similar to grid sampling in that it involves picking up stones from
the surface, but is done by laying down a line (usually a tape measure) and then removing every
surficial stone that lies under the line. The Fehr (1987) method is often used in Switzerland,
although the method is rarely cited in scientific bed sampling literature. It recommends the pickup of
a minimum of 150 stones.

Very little advance has been shown in the scientific literature concerning methods to combat the
excessively large volumes needed to accurately describe the subpavement. The recent work of Petrie and
Diplas (2000) needs to be mentioned again here because they extended the multinomial analysis to give
sampling accuracy estimates not only for grid sampling but also for volume and areal sampling. In this
sense, the sampling operator can determine the quantity to sample for the desired accuracy. Work still
needs to be done to attack the problem of determining the GSD for the subpavement of a gravel bed river.
The only real practical method developed in scientific literature seems to be the hybrid method suggested
by Fripp and Diplas (1993) and revisited by Rice and Haschenburger (2004). This method involves
determining the upper curve of the GSD by a truncated grid sample and the lower part of the curve by a
truncated volume sample. Rice and Haschenburger (2004) postulate that the grid-by-number and the
volume-by-weight distributions overlap in the middle of the GSD and this provides the necessary
information to combine these two informations to form the total GSD. The other practical method, and
much quicker than the hybrid method, is that of Fehr (1987). The Fehr method is quicker because no
subpavement measurements are made. The distribution of fines in the total GSD is assumed to follow a
Fuller curve in which the percent finer is equal to the value d ⁄ ( d max ) .
 Bed 99

4.5.2.2 Analysis techniques

To determine the GSD, it is necessary to determine the proportion of grains in each predefined size class.
The number of classes depend on the size range to be covered. Fehr (1987) suggests classes that increase
–φ
by a factor of 2 . Others use classes that increase by the phi, D = 2 , or half phi ranges. The
proportion of grains in each class can be determined by different analysis techniques: frequency-by-
number, frequency-by-volume, frequency-by-area. Frequency-by-number involves counting the number
of particles in a size class compared to the total population to determine the class proportion. Frequency-
by-volume compares the class volume to the total volume. Volume is usually substituted by weight
which is valid for particles with the same specific gravity. Frequency-by-area determines the class
proportion by comparing the projected area of the class sediments to the total projected area.

4.5.2.3 Equivalence of sampling methods

Depending on the sampling method, the most common are volume-by-weight and grid-by-number and
area-by-weight, a conversion may be necessary. The standard is to express sampling results in a volume-
by-weight equivalent. Kellerhals and Bray (1971) laid the foundation for this standard. Their analysis of
different sediments (represented by blocks) in a cube allowed them to determine the conversion formulas
between the different sampling methods. Their conversion formula can be expressed generally by the
formula (Petrie, 1998):
x
p ( S )i Di
p ( V – W ) i = ------------------------------
x
- Eq. 4.5
∑ ( p ( S ) i i)
D

where p(V-W)i is the proportion by volume-by-weight of the sediment in class i, p(S)i is the fraction of
class i to be converted, Di is the sieve diameter for size fraction i, and x is the conversion exponent. The
conversion exponents proposed by Kellerhals and Bray are 0 for grid-by-number, -3 for grid-by-weight,
2 for area-by-number, and -1 for area-by-weight. Diplas and Sutherland (1988) used a modified
Kellerhals and Bray cube which included blocks that represented porosity to show that the exponent for
area-by-weight (when using wax as a pickup adhesive) varies in the range of -0.4 to -0.5 depending on
the porosity. The value of -1 from Kellerhals and Bray should be used for area-by-weight conversions
when using tape as an adhesive. No conversion exponents for line-by-number were found except for that
proposed by Fehr which is 0.8.

4.5.2.4 Fehr conversion of pavement GSD to subpavement GSD

Fehr (1987) converts a line-by-number sample to a volume-by-weight sample using equation 4.5 with
the conversion exponent of 0.8. He suggests that particles less than 1 cm should not be picked up. To
estimate the fines that were not picked up, he proposes a correction of the GSD with the formula:
i

p i c = 0.25 + 0.75 × ∑ p ( S ) i Eq. 4.6

1
100 Venoge River Data for Streambank Erosion Modeling

where pic is the corrected cumulative proportion smaller than the class Di. This GSD still does not
represent correctly the fine distribution of the subpavement according to Fehr and he proposes that the
fine part of the GSD should be replaced by a Fuller curve. He introduces the Fuller curve in the GSD by
looking for the point on the curve that has a slope that coincides with the Fuller curve. The GSD that is
finer than this point is replaced by the Fuller curve.

4.5.2.5 GSD’s of the Lower Venoge River

The Fehr method was chosen for determining the GSD of the subpavement of the Venoge. Although,
time will not allow this method to be compared in more depth to the hybrid method of Rice and
Haschenburger (2004), this should be done to validate the Fehr method and an accuracy quantification
should be done following the recommendations of Petrie and Diplas (2000).

Before the exact length of the Lower Venoge

study reach was determined, line samples were
collected from the zone of Lussery (km 22.5)
down to km 1. A survey tape was laid on the
ground and all grains falling below the tape
edge and greater than 1 cm were measured for
the b-axis dimension with a folding ruler
(figure 4.31). The sampling was usually
stopped at 150 stones. All sizes were recorded,
and the values were assigned to classes
following the suggestion of Fehr, applying a
factor of 2 ; the first class starts at 1 cm,
followed by 1.41, 2, 2.83 and so on. Forty-eight
samples were collected on the Lower Venoge
and 8 were collected on the four tributaries:
Ruisseau de Molomba, Sorge, Vaube, and
Arena.

Figure 4.32 shows an example of the results of

a Fehr line analysis. The pic curve shows the
Figure 4.31. The author performing a line sample on
distribution after correction with equation 4.6. the Vaube tributary of the Venoge
The piFu curve shows the Fuller curve
calibrated according to the appropriate intersection point of the pic curve. The resulting GSD is that of
the piv curve.
 Bed 101

100%

90%

80%

70%

60%
Percent Finer

50%

40% pic
piFu
30%
piv
20%

10%

0%
0.00 5.00 10.00 15.00 20.00
Diam eter [cm ]

Figure 4.32. Example of a GSD determined with the Fehr line-by-number analysis on the Lower
Venoge River

The GSD’s (in terms of the class fraction) of the 56 samples are given in appendix 3. Figure 4.33 shows
the results of the GSD’s in terms of their characteristic diameters.

16.00
d30
14.00
d50
12.00 d90
grain diameter [cm]

dm
10.00

8.00

6.00

4.00

2.00

0.00
0 5 10 15 20 25
distance to outlet (km)

Figure 4.33. Evolution of the characteristic diameters of the bed sediment of the Lower Venoge
102 Venoge River Data for Streambank Erosion Modeling

4.6 Bank

4.6.1 Input data and methods

The input data required by the CCHE1D model to describe streambanks are: bulk density (N/m3), critical
shear stress, τc (kPa), porosity (-), shear stress coefficient (-), cohesion (Pa), friction angle (°), specific
gravity and the particle size distribution. The bulk density is interpreted to be the saturated unit soil
weight as this is the weight that will typically be involved in mass failures. Values for cohesion and
friction angle are the effective cohesion and effective friction angle. More or less attention can be given
to certain parameters depending on the importance and type of bank failures and the type of streambank
erosion simulations to be run.

Bank classification methods

At present, streambank erosion numerical modeling has been applied mostly to short reaches and
consequently authors have used time consuming point measurements for determining the above listed
bank parameters. It would seem necessary, though, to go towards a field inspection type of bank
classification for streambank erosion modeling to be more practical for long reaches. It is assumed that
each of the above parameters has a given distribution and that the mean values of those distributions can
be used to represent the parameters throughout a reach with the same classification. Optimally,
uncertainty analysis in streambank erosion modeling would employ Monte-Carlo analysis to sample the
entire distributions.

In the Venoge case study, bank classification is based on the Unified Soil Classification System (USCS)
as described in the Swiss standard SN 670 008a (VSS, 1997a). This classification involves the
determination of the particle size distribution, Atterberg limits (liquid limit and plastic limit to determine
the plasticity index), and organic content. Their determination are standard laboratory analyses and are
described in many books (e.g. Terzaghi et al., 1996 and Mandal and Divshikar, 1995) and standards, so
they will not be discussed here. Besides the USCS classification tests, porosity, natural unit weight, and
water content were also determined by laboratory analysis.

Three levels of bank description were carried out on the Lower Venoge River:

1. USCS laboratory classification along with classification criteria suggested by Terzaghi et al. (1996)
as well as pocket shear vane and penetrometer tests
2. USCS field classification, laboratory particle size distribution, and pocket shear vane and
penetrometer tests
3. USCS field classification and pocket shear vane and penetrometer tests

The initial values of the parameters to be used will be based on typical values used in other streambank
erosion studies (Darby, 2005 and Simon et al., 2002) and from index properties (Vss, 1999b) based on
soil classification. Those values are given in table 4.6.
 Bank 103

Description Friction angle typical friction Cohesion c ' typical Saturated unit Matric suction
(°) angle (°) (kPa) cohesion c' weight (N/m 3) friction angle (°)
Gravel 45-35 35 0 0 20000 10
Sand 40-30 37 0 0 18000 15
Sandy Silt 35-30 35 0-10 2.5 18000 15
Silt 35-25 30 0-10 5 18000 15
Soft clay 32-20 28 0-20 10 16000 15
Stiff clay 30-10 25 10-30 15 18000 15

Table 4.6. Cohesion and friction angle values as well as initial values used according to soil
classification

Another possible method would be to rank banks according to their stability and erodibility. In the
project SAFER, this was done for the stability of the streambanks of the Enrick River (table 4.7 and
appendix 4). Point measurements are made at various places along the different bank types to estimate
the parameters’ distributions.

Bank type Stability

Bedrock Very High
Tree lined bank High
Fine grained bank High
Rip-rapped bank High
Walled High
Point bar Moderate - High
Vegetated valley-side Moderate but
slope vulnerable to large
floods.
Composite bank Low- moderate
Cobble bank Low
Eroding valley-side Low –Frequent
slope failures during high
flows.
Table 4.7. Bank types and stability as evaluated in the SAFER project on the Enrick River (Jacobs
Babtie, 2005)

Although the specific point measurements for determining effective cohesion, effective friction angle,
and critical shear stress were not used for the Venoge, they will be presented here for completeness.
Optimally, such point measurements should be used to provide the parameter distributions for bank types
rather than deriving parameter distributions from index properties based on soil classification.

Point measurements for critical shear strength

While the critical shear stress of non-cohesive sediments is relatively well known by the work done by
2
Shields and others and can be approximated by τ c ( N ⁄ m ) ≅ D 50 ( mm ) (Briaud et al., 2001), the
estimation of the critical shear stress of cohesive sediments is more complicated. Numerous soil
properties have been shown to influence the erodibility of a cohesive soil such as antecedent moisture,
clay mineralogy and proportion, density, soil structure, organic content, as well as pore and water
chemistry (Grissinger, 1982). Various test procedures for determining critical shear stress are reported
by Grissinger (1982), but recent literature has given most of its attention to three in situ measuring
apparatus. They are the submerged jet apparatus (Hanson and Cook, 2004), the Erosion Function
104 Venoge River Data for Streambank Erosion Modeling

Apparatus (EFA) tested for scour rate predictions (Briaud et al., 2001), and the cohesive strength meter
(CSM) tested for measuring the in situ erosion shear stress of intertidal sediments (Tolhurst et al., 1999).
The first two apparatus will be briefly described because they have been used specifically in the context
of stream erosion.

The submerged jet test has been used in many studies (e.g. Hanson and Simon, 2001 and Simon and
Thomas, 2002) and is described in detail by Hanson (2004). The principle behind the test is to erode a
material with a jet of the eroding agent (water in our case). The critical shear stress is determined from
the equilibrium scour depth. For a given jet velocity, scour will continue until the equilibrium scour
depth is reached. The scour depth data versus time used with the Blaisdell et al. (1981 in Hanson, 2004)
formulas permit the calculation of the critical shear stress. Once the critical shear stress, τc (Pa) has been
determined, the erodibility coefficient, kd (m³/N-s), can then be determined by the excess shear stress
erosion rate formula:
εr = kd ( τe – τc ) Eq. 4.7

where τe is the effective hydraulic stress (Pa) and εr is the erosion rate (m/s) which is known from the
scour depth versus time data.

The EFA is not an in situ test, but rather uses an intact soil sample in the laboratory apparatus. The
principle behind this test is to push a cylindrical soil sample perpendicularly into a rectangular pipe of
flowing water. The flow velocity for the pipe is controlled so the eroding shear stress is known from
hydraulic pipe formulas. The erosion rate is determined by pushing 1 mm of the soil sample into the
flowing water and measuring the time needed to erode it. This is done at several flow velocities
permitting the plotting of the shear stress versus erosion rate curve. The critical shear stress is determined
by the first velocity that is able to erode the sediment and the erodibility coefficient is then determined
with equation 4.5 by plotting the excess shear stress versus the erosion rate (Briaud et al., 2001).

Point measurements for effective cohesion and friction angle

The determination of effective cohesion and friction angle by soil shear box tests and triaxial
compression tests is well known and presented in many textbooks (e.g. Mandal and Divshikar, 1995,
Terzaghi et al., 1996, and Selby, 1993). The triaxial test is preferred because of its versatility in
performing drained, undrained, or consolidated undrained tests. These test methods won’t be described
further because their cost essentially negates their use over a long stream reach.

The Iowa Borehole Shear Tester (BST) is the in situ measuring device that has been used frequently in
streambank failure studies (e.g. Simon et al., 2002 and Simon et al., 2000). The BST is inserted into a
borehole. The head consists of a piston that can expand to apply a normal stress to the shear plates.
Transducers in the shear plates measure the pore-water pressure. An upward shearing stress is applied to
the plates. Thus, the normal and shearing stresses are known as well as the pore-pressures so that the
apparent cohesion and the resulting effective cohesion and effective friction angle can be determined.
Many tests can be done in the same borehole at different depths in the borehole and at different normal
stresses. The BST is the preferred device for determining the soil strength parameters because it is
applied to the soil in its natural state and at natural pore-pressures (Selby, 1993). Saturated tests can also
be performed by filling the borehole with water.
 Bank 105

Pocket shear vane and penetrometer tests (figure 4.34) can be used to gain a rough idea of the soil
strength. A shear vane has four rectangular blades on a rod which are set at right angles to each other.
The shear vane test indicates total stress and is dependant on the moisture content at the time of the test
so that the test result can not be used directly to determine the effective cohesion and friction angle.
Pocket penetrometers have a rod that can be pushed into the soil. The force needed to insert the
penetrometer rod into a soil a preset distance is recorded. The penetrometer reading is a useful indicator
of bulk density, but is not a reliable indicator of shear strength. Shear vanes are most suitable in fine-
grained soils and care must be taken with both of the instruments not to use them in soils with roots,
clasts, or strongly developed soil structure (Selby, 1993).

Field shear boxes can also be used, but are much more time consuming than the BST measurements.
Notably, the field shear box test has been used by Wu et al. (1988) to determine the contribution of roots
to the shear strength of soils.

Figure 4.34. Pocket penetrometer (left) and shear vane (right) used during the Lower Venoge bank
investigations

Bank classification methods and soil strength parameters

Various authors (Briaud et al., 2001, and Hanson and Simon, 2002) have shown that no correlation exists
between USCS soil classes and soil strength properties. Thus, there is a real dilemma of how to
practically determine the soil strength parameters for streambank erosion hazard modeling. Point
measurements are too timely and costly to be performed over a large area, but bank classification based
on soil type has failed up to date. It is hypothesized that even with a large number of point measurements,
bank heterogeneity is such that soil strength parameter calibration will be needed regardless. Concerning
the bank toe critical shear stress, one value is certainly not enough as the value for the original bank toe
sediment is probably different than that of failed basal sediments. Therefore, it is hypothesized that more
time should be devoted to understanding and observing erosion and failure processes than for performing
point measurements. Model parameters can be adjusted or calibrated to try to mimic the observed erosion
and failure processes.
106 Venoge River Data for Streambank Erosion Modeling

4.6.2 Bank classification and results

As mentioned in the previous section, three levels of investigation were done on the Lower Venoge
River. Level 1 investigation was carried out on each erosion monitoring bend, while level 2
investigations were carried out at a 1 kilometer interval and level 3 investigations at a 500 meter interval.
Sampling was mostly done on bend exteriors to avoid sampling of sandy bend interior deposits. Figure
4.35 shows the locations of the sampling and the type of investigation performed.

17
17
Reproduced with permission
11 from swisstopo (BA068334) 18
18

19
19
2.2
2.2 5.2
5.2
20
20
22

7.1
7.1
8.1
8.1
6.1
6.1
12.2
12.2
10.3
10.3
13.2
13.2
11.2
11.2
33 21
21 22
22
23
23
24
24
44 25
25

55 26
26

66 27
27

77

28
28

88
99
29
29
10
10
11
11
30
30

31
31

32
32
12
12 15.2
15.2
19.2
19.2
20.1
20.1
13
13 33
33
16.1
16.1
17.1
17.1

34
34 18.3
18.3

14
14

Level 1 35
35
Level 2
15
15 Level 3
36
36
16
16

37
37
17
17 38
38

18
18

0 0.5 1 0 0.5 1
19
19
2.2
2.2 5.2
5.2
kilometers kilometers
20
20

Figure 4.35. Bank sampling locations and the level of investigation

(Background maps: CP100 ©1997, Swisstopo)
 Bank 107

Table 4.8 shows the bank soil properties that were collected in the level one investigations and how they
are determined and the possible values for qualitative testing. Table 4.9 shows the principal results for
the monitored bends. It can be noticed that complete level one investigation was not possible of course
for non-cohesive soils: shear vane, penetrometer, atterberg tests, and intact soil sampling are not
possible. Some layers were not completely tested because of their limited thickness.

bend/layer
thickness
representative of
photo number
1. Color Munsell chart
2. Odor strong, medium, weak, none
3. Texture: A. appearance after fracture granular, dull, smooth, glossy
B. feeling to fingers floury, smooth, gritty, sharp
angular, subangular, subrounded,
4. Grain properties: A. form of particles rounded, well rounded
B. mica content high, medium, low, none
C1. type of organic matter fibers, twigs, leaves, …
visible, slightly visible,
C2. State of organic matter decomposed
very high, high, medium, low, very
5. Dry crushing strength low
6. Dilatancy fast, slow, none
7. Toughness at plastic limit high, medium, low, none
8. Ribbon length short, medium, long
9. Stickiness strong, medium, weak, none
10. Porosity (-) laboratory analysis
11. Moisture content (%) laboratory analysis
12. Natural unit weight (g/cm3) laboratory analysis
13. Dry unit weight (g/cm3) laboratory analysis
14. Unconfined compressive strength, qup
(kN/m2): field test
Unconfined compressive strength after
remolding, qur (kN/m2): field test
Sensitivity, qup/qur calculation
15. Shear vane result (kN/m2) field test
16. Liquidity limit, wl (%) laboratory analysis
17. Plasticity limit, wp (%): laboratory analysis
Plasticity index (%): Ip=wl-wp calculation
18. Particle sizes (mm): Dmax
D90 laboratory analysis
D60 laboratory analysis
D50 laboratory analysis
D30 laboratory analysis
D10 laboratory analysis
dm calculation
% Gravel laboratory analysis
% Sand laboratory analysis
% Silt laboratory analysis
% Clay laboratory analysis
Cud (d60/d10) calculation
Ccd (d30^2/(d10*d60)) calculation
19. Carbonate content: Reaction to HCl strong, medium, weak, none
Comments:
Field classification USCS of the fines
Laboratory Classification USCS:
Classification name

Table 4.8. Field and laboratory tests performed in the detailed level 1 of bank investigation for the
classification of the bank soil. The right column indicates the method or possible values.
108 Venoge River Data for Streambank Erosion Modeling

Laboratory Classification USCS

12. Natural unit weight (g/cm³)

15. Shear Vane result (kN/m²)

14. Unconfined compressive
13. Dry unit weight (g/cm³)
11. Moisture content (%)

16. Liquidity limit, wl (%)

19. Carbonate content

strength, qup (kN/m²)

Classification name
Plasticity index (%)
10. Porosity (%)

Particle size distribution

bend/ thickness % % Reaction
id layer (cm) Gravel Sand % Silt % Clay HCl
83 2.1 0-30 63.3 28.0 1.2 1.0 56.7 42.3 27.4 7.2 0.5 53.0 29.2 17.2 strong SC clayey sand
84 2.2 30-50 48.3 13.5 1.6 1.4 92.1 56.2 19.8 3.2 0.6 60.5 27.3 11.6 strong SM silty sand
silty, clayey
85 2.3 50-80 52.0 14.6 1.5 1.3 80.0 74.0 26.0 5.0 0.7 50.7 35.1 13.5 strong SC-SM sand
silty, clayey
86 2.4 95-130 46.7 17.3 1.7 1.4 12.5 20.3 27.6 6.8 3.6 59.6 25.5 11.3 strong SC-SM sand
88 5.2 20-110 60.0 22.8 1.4 1.1 80.4 69.2 27.0 3.6 0.0 20.9 67.8 11.3 strong ML silt with sand
89 6 0-20 cm 13.8 23.5 strong silty sand
90 6.1 20-130 cm 51.3 15.9 1.7 1.4 37.9 35.0 21.2 2.2 0.0 59.6 20.4 20.1 strong SM silty sand
91 6.2 130-140 cm 36.0 18.0 0.0 9.1 71.1 14.4 5.3 strong SM silty sand
Poorly graded
gravel with silt
92 6.3 140-fond 0.0 61.7 29.0 7.0 2.4 strong GP-GM and sand
clayey, sandy
93 7.1 10-120 cm 56.0 34.6 1.7 1.2 46.7 43.0 30.0 11.8 0.0 49.5 39.9 10.7 strong CM(CL) silt
94 8 0-60 cm 45.0 40.3 strong silty sand
95 8.1 60-140 cm 48.7 12.0 1.7 1.4 53.8 43.0 20.3 2.4 0.0 54.7 32.3 13.0 strong SM silty sand
96 8.2 140-160 cm strong sand with gravel
Well graded
97 8.3 160-fond 47.7 17.2 1.8 1.4 35.4 27.7 0.2 87.1 9.5 3.1 strong SW-SM sand with silt
98 10.1 0-30 cm 10.4 11.0 29.7 1.4 strong silt with sand
99 10.2 30-50 cm 71.0 30.7 1.3 1.0 28.3 24.0 42.2 8.3 0.0 23.2 55.5 21.4 strong ML ou OL silt with sand
100 10.3 50-190 cm 56.0 19.4 1.8 1.4 41.3 51.2 24.7 3.5 0.0 38.8 58.0 3.2 strong ML sandy silt
101 11.1 0-30 cm 21.3 43.5 35.0 4.1 strong ML sandy silt
102 11.2 30-240 cm 54.3 19.2 1.7 1.3 42.5 37.7 41.6 18.2 0.0 34.7 52.5 12.8 strong CM sandy lean clay
103 12 0-20 cm 110.0 89.8 limon sableux
ML (limit
CM-CH-
104 12.1 20-60 cm 58.3 20.1 1.6 1.3 137.1 84.5 46.3 18.3 0.0 10.6 43.1 46.3 strong MH) silt
silty, clayey
105 12.2 60-190 cm 50.3 14.3 1.7 1.4 145.8 90.4 0.0 67.5 19.6 12.9 strong SM sand
silty, clayey
106 12.3 190-fond 54.0 8.6 1.5 1.4 165.8 0.0 31.7 51.2 8.2 8.8 strong SC-SM sand with gravel
107 13.1 0-20 cm 48.3 13.4 1.6 1.4 94.2 37.0 8.3 42.7 35.0 14.0 silty sand
108 13.2 20-fond 49.0 15.0 1.7 1.4 75.0 66.8 0.1 37.1 46.2 16.6 sandy silt
silty, clayey
110 15.1 20-60 cm 53.3 12.6 1.4 1.2 50.4 26.8 22.8 4.8 0.0 79.6 16.1 4.2 strong SC-SM sand with gravel
111 15.2 60-200 cm 56.7 18.8 1.5 1.2 69.6 54.7 31.2 5.0 0.0 36.4 46.7 16.9 strong ML sandy silt
112 16.1 30-fond 54.3 22.3 1.6 1.2 42.1 45.7 25.6 3.6 0.0 41.3 45.2 13.5 strong ML sandy silt
113 17.1 10-140 cm 60.0 22.0 1.4 1.1 45.0 36.7 19.8 1.5 0.0 52.5 36.0 11.6 strong SM silty sand
114 17.2 140-fond 5.0 11.0 19.6 1.1 strong SM silty sand
115 18.1 0-90 cm 21.3 24.7 30.9 9.9 none sandy silt
Silty gravel with
116 18.2 90-170 cm 0.0 strong sand
117 18.3 170-fond 58.3 28.4 1.7 1.3 54.2 98.8 43.2 13.9 0.0 0.2 87.2 12.6 none ML silt
118 19.1 0-50 cm 69.0 17.6 1.5 1.2 88.8 40.0 26.9 3.0 10.1 47.8 37.6 4.5 strong SM silty sand
well graded
119 19.2 50- fond 74.9 24.1 0.6 0.4 none GW gravel with sand
120 20.1 80- fond 83.0 79.4 0.7 1.1 95.0 21.3 31.8 4.7 1.3 35.1 49.1 14.4 strong ML sandy silt

Table 4.9. Results of level 1 soil analyses for the monitored erosion bends and their USCS classification

Results of the level 2 and 3 investigations are shown in tables 4.10. For a specific map location, two or
three classification tests were done at different levels on the bank. The first test was done on the face of
the bank, the second one on or near the top of the bank, and the third one on the top of the bank. Shear
vane and penetrometer results are the average of 6 readings conducted on soils after removal of the
vegetation layer (approximately 20 cm depth). The apparent cohesion value is the minimum of the shear
vane result or half of the penetrometer reading. Of course this apparent cohesion value is dependant on
the moisture content of the soil. The frequency of the particle size distributions for the level 2
 Bank 109

investigations is evident by the Particle Size Distribution (PSD) column. For the level 2 investigations,
three PSD analyses were averaged to obtain the final distribution.

USCS
Shear Apparent Moisture Natural unit field
Vane, Penetrom cohesion, content weight classificat
date id id map su eter, qup cu roots PSD (%) (g/cm³) ion
kN/m² kN/m² kN/m²
29.03.2004 1 1 38.5 74.2 37.1
29.03.2004 2 88.2 257.5 88.2
29.03.2004 3 46.0 146.7 46.0 yes 27.57 1.09 SL
29.03.2004 4 2 24.0 50.0 24.0
29.03.2004 5 51.0 98.3 49.2
29.03.2004 6 3 21.5 40.0 20.0 1 S
29.03.2004 7 20.0 118.3 20.0 1 SL
29.03.2004 8 4 21.5 51.7 21.5 1 SL
29.03.2004 9 51.5 126.7 51.5 1 yes 27.34 1.01 LS
29.03.2004 10 5 27.0 52.5 26.3 SL
29.03.2004 11 19.8 61.7 19.8 SL
29.03.2004 12 9.0 42.5 9.0 SL
29.03.2004 13 6 56.0 125.8 56.0 LS
29.03.2004 14 51.2 288.3 51.2 yes 16.40 1.06 SL
29.03.2004 15 7 30.3 73.3 30.3 SL
29.03.2004 16 31.3 95.8 31.3
29.03.2004 17 8 81.2 253.3 81.2 LS
29.03.2004 18 86.2 242.5 86.2 yes 18.76 1.19 SL
29.03.2004 19 9 37.3 79.2 37.3 LS
29.03.2004 20 68.7 100.8 50.4 LS
19.04.2004 21 10 25.3 56.7 25.3 SL
19.04.2004 22 36.0 101.7 36.0 SL
19.04.2004 23 11 23.7 59.2 23.7 LS
19.04.2004 24 63.3 135.8 63.3
19.04.2004 25 12 52.7 215.8 52.7 LS
19.04.2004 26 25.2 50.0 25.0 yes 21.35 0.91 SL
19.04.2004 27 19.5 46.7 19.5 SL
19.04.2004 28 13 33.5 52.5 26.3 LS
19.04.2004 29 50.3 111.7 50.3 LS
19.04.2004 30 14 64.0 135.0 64.0 LS
19.04.2004 31 28.8 80.0 28.8 1 SL
19.04.2004 32 15 48.3 75.0 37.5 LS
19.04.2004 33 66.0 114.2 57.1 yes 24.88 0.94 LS
19.04.2004 34 16 28.3 57.5 28.3 SL
19.04.2004 35 18.7 31.7 15.8 S
19.04.2004 36 17 31.2 90.8 31.2 LS
19.04.2004 37 25.5 55.0 25.5 1 LS
19.04.2004 38 18 22.0 35.0 17.5 1 LS
19.04.2004 39 45.8 40.0 20.0 1 yes 32.18 0.88 SL
26.04.2004 40 19 22.0 47.5 22.0 1 S
26.04.2004 41 44.2 63.3 31.7 2 SL

Table 4.10. Results of the level 2 and 3 bank investigations on the Lower Venoge River. Root density is
medium with a value of 1 and high with a value of 2. LS is a sandy-silt and SL is a silty-sand.
110 Venoge River Data for Streambank Erosion Modeling

USCS
Shear Apparent Moisture Natural unit field
Vane, Penetrom cohesion, content weight classificat
date id id map su eter, qup cu roots PSD (%) (g/cm³) ion
kN/m² kN/m² kN/m²
26.04.2004 42 20 2.5 0.0 0.0 S
26.04.2004 43 10.8 21.7 10.8 1 S
26.04.2004 44 21 66.3 241.7 66.3 LS
26.04.2004 45 22.7 55.8 22.7 yes 15.72 1.30 SL
26.04.2004 46 23.8 35.8 17.9 1 S
26.04.2004 47 22 26.5 45.8 22.9 1 LS
26.04.2004 48 62.7 102.5 51.3 1 LS
26.04.2004 49 23 12.3 29.2 12.3 1 SL
26.04.2004 50 30.2 62.5 30.2 2 SL
26.04.2004 51 24 10.7 29.2 10.7 2 SL
26.04.2004 52 16.0 38.3 16.0 2 SL
26.04.2004 53 25 59.5 96.7 48.3 1 LS
26.04.2004 54 75.7 167.5 75.7 1 yes 22.47 1.01 LS
26.04.2004 55 26 26.2 78.3 26.2 1 LS
26.04.2004 56 23.8 44.2 22.1 1 LS
03.05.2004 57 27 61.3 119.2 59.6 1 LS
03.05.2004 58 32.7 62.5 31.3 1 SL
03.05.2004 59 28 77.0 148.3 74.2 1 SL
03.05.2004 60 61.3 176.7 61.3 1 SL
03.05.2004 61 29 22.7 91.7 22.7 2 SL
03.05.2004 62 25.0 72.5 25.0 1 yes 10.30 0.83 SL
03.05.2004 63 20.0 45.8 20.0 SL
03.05.2004 64 30 21.7 48.3 21.7 1 S
03.05.2004 65 39.2 133.3 39.2 1 LS
10.05.2004 66 31 21.0 30.8 15.4 2 LS
10.05.2004 67 16.8 45.8 16.8 2 LS
10.05.2004 68 32 54.2 116.7 54.2 2 LS
10.05.2004 69 45.0 148.3 45.0 1 yes 15.78 1.12 SL
10.05.2004 70 32.2 104.2 32.2 2 LS
10.05.2004 71 33 41.5 115.0 41.5 1 SL
10.05.2004 72 28.7 104.2 28.7 2 SL
10.05.2004 73 34 43.8 52.5 26.3 1 LS
10.05.2004 74 41.0 95.0 41.0 1 LS
10.05.2004 75 35 24.5 57.5 24.5 1 LS
10.05.2004 76 21.5 65.0 21.5 2 yes 14.46 1.09 SL
10.05.2004 77 36 53.5 81.7 40.8 2 SL
10.05.2004 78 33.7 73.3 33.7 2 SL
10.05.2004 79 37 45.2 58.3 29.2 1 LS
10.05.2004 80 46.3 60.8 30.4 2 LS
10.05.2004 81 38 47.7 67.5 33.8 1 SL
10.05.2004 82 25.0 56.7 25.0 2 yes 28.32 1.02 SL

Table 4.10 (cont.). Results of the level 2 and 3 bank investigations on the Lower Venoge River

In the level 2 PSD’s, it was evident that there was little PSD variability (c.f. appendix 5) and that one
PSD is sufficient at a location for the level 1 investigations to get a representative bank PSD. All of the
determined particle size distributions are shown in figure 4.36. It can be seen in this graphic the majority
of distributions are typical of silty sand or sandy silt soils. Only the two left distributions and the two
right distributions depart from these typical alluvial soil distributions. The particle size distributions are
also shown on textural triangles in appendix 5 to allow a comparison of the Lower Venoge PSD’s with
those shown in appendix 1.
 Bank 111

100

90

80

70
Percent finer

60

50

40

30

20

10

0
0.0001 0.001 0.01 0.1 1 10 100
Particle size (cm)

Figure 4.36. Particle size distributions of the Lower Venoge banks

As has been mentioned earlier, it is not expected to find correlations between the USCS soil classes and
the soil properties. The initial soil strength parameters to be used in the streambank erosion modeling
will be based on the observed soil classes and the typical values given in table 4.6. The soil strength
parameters will then be calibrated according to erosion observations and correlations between soil
classes and/or shear vane results will be investigated.

The bank analyses served to prepare the bank input for the CCHE1D model. Figure 4.37 shows the soil
areas that were estimated according to the point sample results. Further point sampling would be
necessary to ensure these soil areas are correct, especially at distances far away from the original
sampling points. Figure 4.37 also shows bank protection works. At these locations, subdivisions of the
bank soil areas have been made so that non-erodible banks can be incorporated, if desired, in the bank
erosion simulations.
112 Venoge River Data for Streambank Erosion Modeling

10.5
10.5
Bank Soils 77
Clay
40
40 Gravel 55
55
Sandy silt
Sand
Silty sand
6.5
6.5
84
84 10
10 Soil area
57
57
102 Id of soil area
102
88 93
88 93 66
10 station km
10
90
90 9.5
9.5 Soil area subdivisions
95
95 Venoge River
100
100
105
105 Bank protection 59
59
102 108
102 108 0 0.2 0.4
99 kilometers
5.5
5.5

47
47
61
61
8.5
8.5 49
49
51
51 55
88
53
53
7.5
7.5
64
64

4.5
4.5

77

55
55
44 66
66
68
68
6.5
6.5
111
111 3.5
3.5
119
119
120
120

71
71
33
112
112

113
113
117
117
2.5
2.5
73
73

Figure 4.37. Soil areas along the Lower Venoge River according to field investigations to be used in the
streambank erosion modeling
 Boundary conditions 113

4.7 Boundary conditions

Streambank erosion modeling necessarily implies the specification of boundary conditions. The
CCHE1D model needs to know sediment and discharge input the length of the study reach as well as the
downstream water depth for the subcritical hydraulic calculations of the model. Approximations for
these boundary conditions are made for the Lower Venoge River study reach in the following manners:

1. Discharge input will be based on hydrologic modeling which is described in chapter 5. A historical
discharge series (Ecublens, Les Bois, figure 4.2) is available which will permit the calibration of the
model and the refinement of the simulated discharges.
2. Sediment input is based upon a sediment transport capacity equation. Only three sediment transport
curves need to be developed for the Venoge study reach: the upstream Venoge input, and the two
main tributaries, the Vaube and the Arena, input, as all other tributary input are small and thus
assumed to be clear water. The sediment transport curves (figure 4.38) are based on the Smart and
Jaeggi equation (equation 2.3). Channel slopes for the Vaube and the Arena are approximated based
on the LIDAR DTM and their average channel dimensions near their outlets were measured at the
same time that bed GSD samples were taken.
3. The downstream rating curve (figure 4.39) is approximated with the Manning/Strickler formula.

Sediment and water discharge are input into the CCHE1D model via a text file requiring a given format
(Vieira and Wu, 2002). A program was written to prepare this text file. The hydrograph and sediment
transport curve and corresponding GSD for each node as well as the starting and ending dates of the high
flow events are read by the program. Sediment input is divided into the grain size classes according to
the GSD of the bed at the input location.

4.8 Summary and conclusions

The goal of this chapter was to show the data that was collected and measured to be able to simulate
streambank erosion and to describe the measurement methods. Erosion scars were observed on the
Lower Venoge River from km 22.390 to km 0.789 as shown in figure 4.1. This allowed two detailed
zones, km 10.100 to km 9.075 and km 3.510 to km 2.650, to be identified for detailed erosion
monitoring. These zones were monitored for erosion from November 2003 through September 2005 and
for flow depth from March 2004 through August 2005. The streambank erosion modeling zone was
chosen to encompass these erosion monitoring zones going from km 10.516 to km 2.538.

Erosion and failure processes were better understood with the detailed erosion measurements. Bank
failures on the Venoge River are mostly soil falls and slab failures. It is very difficult to distinguish
between these two failure types because it is unknown to what extent the bank is being undercut before
the failure. Cantilever failures occur also, mainly due to vegetative effects, but also due to composite
bank stratigraphy. Often soil fall occurs in combination with fluvial toe erosion to undercut highly
vegetated banks before a cantilever fail occurs.
114 Venoge River Data for Streambank Erosion Modeling

0.014 Venoge upstream

Outlet Vaube
0.012 Outlet Arena
Sediment Input (m 3/s)

0.01

0.008

0.006

0.004

0.002

0
0.00 20.00 40.00 60.00 80.00 100.00 120.00 140.00
3
Water discharge (m /s)

Figure 4.38. Sediment transport curves for the estimation of sediment input into the Lower Venoge
River study reach

160.00
Manning bed = 0.04
Manning banks = 0.10
140.00

120.00
Discharge (m 3/s)

100.00

80.00

60.00

40.00

20.00

0.00
377.00 377.50 378.00 378.50 379.00 379.50 380.00 380.50 381.00
Water surface elevation (m.a.s.l.)

Figure 4.39. Estimation of the downstream rating curve for the Lower Venoge River study reach
 Summary and conclusions 115

Bank retreat rate is determined by basal endpoint control. This is particularly important for the Lower
Venoge River where non-cantilever bank failures are small (usually less than 1 meter) and failures can
be multiple during a high flow event.

Erosion was monitored with erosion pins, two PEEP sensors, and bank profiling. The PEEP sensors were
difficult to install because of the gravel and roots in the alluvial soils. Measurement values also
fluctuated too much to be able to correctly measure fluvial erosion. One of the PEEP sensors did record
a soil fall that was distinguishable with the measurements. This measurement confirmed that fluvial
erosion was not responsible for the retreat recorded at that point and that fluvial toe erosion can cause
these small soil falls below a vegetative cover on the bank top.

If erosion is active enough, bank profiling is to be preferred because of the toe measurement that is
important. Bank profiling and pin erosion measurements can be used together, though, and the
experimental system to do so was explained.

Historical maps, orthophotos, and topographic LIDAR data (surface and terrain) were obtained for the
Lower Venoge River. By comparison of this data between each other, it provides valuable information
on past channel migration. If two of these data sets provide an accurate description of the channel bank
tops, it can provide valuable information on bank retreat rate. The only historical data set that seems to
provide this accurate bank line description is the LIDAR data. The historical maps and orthophotos
provide piecemeal information, but have accuracy problems or problems with determining the bank lines
as is the case for the summer SwissImage orthophotos. The available information was used to determine
the Lower Venoge River 2004 centerline.

The channel topographic description is a big hurdle to overcome for streambank erosion hazard
modeling. Linear interpolation of cross sections is possible if there is a cross section for at least every
riffle and pool, although even more frequent cross sections are necessary if an accurate interpolated bank
and bed definition is desired in between surveyed cross sections. This amount of cross sections is never
available unless surveyed specifically with sediment transport calculations in mind. Cross sections
measured in 1996 on the Venoge River were recovered, but those in the detailed erosion monitoring
zones are from leveling without an accurate plan precision. Cross sections in 2004 are georeferenced.
Comparison of thalwegs and cross sections for the period 1996 and 2004 in the modeling reach shows
definite channel instability in the region of km 7.3 to km 7.7, although it has supposedly been stabilized
after the "Abbaye" weir reparation in 2002. A few other zones are possibly instable, but the cross section
data in those areas is insufficient to confirm instability. Otherwise, the majority of the 22 kilometers
surveyed in 2004 show that the Lower Venoge River is stable.

A cross section tool was programmed to work with GIS cross section database files. This facilitates the
transformation of surveyed cross section points into two dimensional (transverse and elevation) cross
sections for modeling and is coordinated for the model CCHE1D. This tool can be used to visualize cross
sections for calculating transverse cross section distance and position and for defining bank definition
points. The tool can interpolate cross sections linearly or based on channel curvature along river
centerline points. Interpolated cross sections with respect to channel curvature demands the calibration
of bed topography parameters based on surveyed cross sections. The cross section tool does this
calibration and uses the parameters to interpolate cross sections with respect to channel curvature. For
116 Venoge River Data for Streambank Erosion Modeling

reaches with insufficient cross section data (missing riffle or pool cross sections), this interpolation
method can be used to interpolate cross sections that are coherent with the channel curvature. Scour holes
were found to be displaced upstream or downstream from some surveyed cross sections.

The possibility of improving the channel description in a LIDAR derived DTM was explored. With
sufficient cross sections surveyed at the same time as the LIDAR measurements, linearly interpolated or
meander interpolated cross sections can replace the original channel LIDAR points. This is usually not
the case as is for the Venoge. Streamwise linear interpolation of channel points was programmed into
the cross section tool. If sufficient cross sections are available and aligned with the LIDAR data, they
can be interpolated with this module and replace the original LIDAR channel points foregoing the more
timely linear interpolation procedure. Otherwise, as was done for the Venoge, the original LIDAR
channel points can be interpolated with this module. This significantly improved the Venoge channel
bank line description, and it is assumed that the DTM channel bottom now reflects the approximate water
surface at the time of the LIDAR measurement. The interpolation and smoothing procedure produces
some artifacts in the DTM. A more sophisticated interpolation procedure, such as an IDW in the
streamwise direction, could help this.

The Fehr line sampling method was used to collect bed grain size data and estimate the subpavement
grain size distribution. This method depends on the assumption that the subpavement follows a Fuller
curve. The Fehr method should be compared against the hybrid method of Rice and Haschenburger
(2004) to verify that the Fuller curve assumption is verified for different types of rivers. More bed
samples should also be obtained for the Venoge River and in different locations in a riffle-pool sequence
to better understand the sediment facies.

Three levels of investigation of bank soil properties were conducted. The investigated soil properties will
be compared against calibrated bank values to determine if they can explain the critical shear stress,
cohesion, and friction angles without measuring these values with the specific equipment necessary.

The upstream and tributary sediment input into the river is estimated with the Smart and Jaeggi sediment
transport capacity formula and the downstream rating curve is based on uniform flow.
 Chapter 5

Hydrologic Modeling for

Distributed Discharge Input

5.1 Objectives of the hydrologic modeling

The goal of the hydrologic modeling shown in this chapter is to provide a correct spatial representation of
the discharge input (and indirectly an appropriate sediment discharge input) into the river. This is possible
with a continuous hydrologic model, HydroRoute, that takes into account the important hydrologic
processes. Spatial interpolation of meteorological data is performed to properly generate the model input
data. The model is calibrated and the historical period is simulated producing discharge inputs for the
Lower Venoge River tributaries. This historical period is used as a basis for preparing three series of 300
years of discharge. These long series will permit the streambank erosion model to predict erosion for
extreme hydrologic events.

5.2 Hydrologic model

A continuous hydrologic model is used for the generation of the tributary hydrographs. The model GSM-
SOCONT (Glacier and SnowMelt - SOil CONTribution model) developed at the HYDRAM laboratory
was available for use. The origin of this model is the model SOCONT developed by Consuegra and Vez
(1996) based largely on the GR models (Edijatno and Michel, 1989). The model was adapted for mountain
hydrology during several research projects (Schaefli et al., 2005, Hamdi et al., 2005). The GSM-SOCONT
model theory was used as the basis for the generation of an object-oriented program called HydroRoute
developed by the company SODELO. The HydroRoute model is under development and its use in this

117
118 Hydrologic Modeling for Distributed Discharge Input

thesis served to test the implementation of the GSM-SOCONT model in the user-friendly, convivial
interface proposed by HydroRoute.

HydroRoute can be classified as a semi-distributed conceptual reservoir model. Hydrographs are

generated in a lumped fashion for each subcatchment, but the possibility to use many subcatchments
within a catchment and then to route them provides a semi-distributed catchment response. The model
also incorporates snow storage units, also called bands, which can be multiple for a subcatchment. This
permits a refinement in the quantification of the equivalent rainfall that will be routed in the model.

5.2.1 Elements of the HydroRoute model

The hydrologic elements (objects) of HydroRoute (figure 5.1) and their brief description are as follows:
• Junction: adds two or more upstream hydrographs to form a single hydrograph.
• Hydrograph-input: allows the input of a discharge time series.
• SOCONT: transforms a rainfall time series and Potential EvapoTranspiration (PET) time series into
a discharge time series through the use of a linear slow reservoir and a nonlinear runoff reservoir.
• SOCONT-Alpine: incorporates snow and glacier effects together with the aspects of the SOCONT
element. Precipitation, PET, and temperature time series are needed as input. Snow storage units
handle the calculation of the equivalent rainfall and snow height, and glacier units can be used to
calculate the snowmelt and glacier melt hydrograph contribution. The non-glacier parts of the
subcatchment use the same rainfall-runoff transformation as the SOCONT element.
• Muskingum: provides the channel routing of hydrographs before addition with downstream
hydrographs using the classical formulation (see Chow et al., 1988 for an example).
• Derivation: allows for an input hydrograph to be separated into two output hydrographs according to
the Qin-Qouts relationship specified by the user.

A model topology can be constructed with the above elements, as shown in the example in figure 5.1. A
global parameters window allows global parameters (Tc1, Tc2, Tc3, Tc4, θr) to be specified. Through the
parameters’ window, the user can indicate the input files, the output variables to be stored, and the
parameters specific to the selected element. Another window allows the simulation time, time step, and
precision to be specified. It is worthwhile to note here that the model can use input data at variable time
steps, even variably spaced, through a new time series format that has evolved from the CODEAU® time
series format (De Souza et al., 1994) developed by the HYDRAM laboratory in the last decade. A
simulation window allows the user to choose calibration or simulation mode. In calibration mode, the
user can search for an optimal parameter set using Monte-Carlo (MC) parameter sets generated by
random sampling of uniform parameter distributions or using sweep parameter sets which sample
uniform distributions at regular intervals. Calibration parameter sets and the resulting objective function
criteria (Nash or Log-Nash of the entire or superior truncated (low flows) or inferior truncated (high
flows) hydrograph) can be output to a text file for further analysis. For a "normal" (non-calibration)
simulation, the user can store the output time series for visualization within HydroRoute.

The SOCONT-Alpine element is heavily used in the Venoge catchment modeling so the theory behind
the snow storage and the rainfall-runoff transformation of it will be discussed in the next two sections.
 Hydrologic model 119

Junction Hydrograph-Input SOCONT SOCONT-Alpine Muskingum

Connectivity

Derivation

Figure 5.1. Elements of the HydroRoute model and its graphical interface

5.2.2 Snow storage

The flowchart in figure 5.2 illustrates the formulas involved in the transformation of precipitation into
rain and snow1, and the subsequent calculation of the equivalent rainfall at a time t in one snow storage
unit of an element SOCONT-Alpine. A subcatchment can have as many snow storage units as the user
decides, but precipitation, P, and temperature, T, must be provided for each snow storage unit. Model
performance will increase by increasing the number of snow storage units, but only up to a threshold that
depends on the catchment elevations (Schaefli et al., 2005). The quantity of precipitation that falls as
snow, S, and as rain, Rn, is determined according to the two critical temperatures, Tc1, and Tc2. Below
Tc1 all precipitation is snow, while above Tc2 it is all rain. Between the critical temperatures, a linear
transition determines the amount of snow and rain. The quantity of snow is then added to the snow
height, Hs. If the temperature is greater than a third critical temperature, Tc3, indicative of the melting

1. Snow is used here to refer to all solid precipitation.

120 Hydrologic Modeling for Distributed Discharge Input

temperature, the quantity of melted snow, Ms, is calculated. Snowmelt is calculated by a temperature
index method using a degree-day factor (mm/h/°C) and a coefficient brn (h/mm) which increases the
degree-day factor in case of rain on snow. The melted snow and rain are added to the snow water, Ws.
Snow has a capacity to retain a certain quantity of water. This is represented by the variable, θr. If the
quantity of water in the snow is greater than this capacity, the quantity that goes over it will be considered
to be equivalent rainfall, Peq.

Precipitation P(t), Temperature T(t)

S(t) Rn(t)

Yes No

Figure 5.2. Snow storage unit used in the SOCONT-Alpine element in

the HydroRoute model (adapted from Hamdi et al., 2005)

5.2.3 SOCONT-Alpine
The flowchart in figure 5.3 illustrates the SOCONT-Alpine element of the HydroRoute model. The
element has two parts: glacier and non-glacier. Only the non-glacier part is described here since the
Venoge catchment does not have any glaciers.The SOCONT-Alpine element uses snow storage units.
The equivalent precipitation issued from them is divided into effective rainfall, Peff, for runoff and
infiltrated rainfall, Pinf. The quantity of rainfall that infiltrates depends on the degree of saturation
(fullness), represented by the height, ha (mm), compared to the capacity of the slow reservoir,
represented by the total height, A (mm), of the reservoir and is governed by the equation shown in figure
 Subcatchment delineation 121

5.3. The exponent y can be calibrated. The slow reservoir is drained by evapotranspiration, ET, and by
baseflow, qa. The ET depends on the input PET time series, which can be adjusted by a multiplication
factor, FPET, and also on the degree of saturation in the slow reservoir raised to the power x, another
calibration coefficient. The baseflow is also a linear reservoir with a storage coefficient, ka (1/h). The
effective rainfall flows through a nonlinear runoff reservoir. The runoff specific discharge, qr, depends
on the calibration coefficient, β (m4/3/s), the catchment mean slope, J (m/m), and the height in the runoff
reservoir, hr. The sum of the non-glacier specific discharges multiplied by the non-glacier area, Angl (ha),
and the appropriate unit conversion factors will produce the non-glacier discharge, Qngl (m³/s). The sum
of the Qg and the Qngl gives the total discharge, Qtot, for the SOCONT-Alpine element.

Glacier Non-glacier
Pj,nk,Tj,nk Pj,nl,Tj,nl
Pj,k,Tj,k Pj,l,Tj,l
Pj,1,Tj,1 Pj,1,Tj,1
snow storage

Peq,gl Peq,ngl

g l a c i e r s t o r a g e

hgm hr A
hsm ha
Snowmelt Glaciermelt reservoirs Runoff Slow

Qtot(t)

Figure 5.3. SOCONT-Alpine element of the HydroRoute model (adapted from Hamdi et al., 2005)

The snow storage and transfer reservoirs in the system need the continuity equation to calculate their
respective heights. The generalized continuity equation applied is:
dh
------ = I ( t ) – O ( t ) Eq. 5.1
dt

where h (mm) is the height in the snow storage unit or reservoir and the I (mm/h) is the input to and O
(mm/h) is the output from the reservoir. The continuity equations are solved implicitly by the Newton-
Raphson method.

5.3 Subcatchment delineation

The delineation of the subcatchments (c.f. figure 3.5) follows that of the GESREAU database. This
delimitation is based on the 1:25’000 maps in the database (Crausaz, 2000). This topographic delineation
122 Hydrologic Modeling for Distributed Discharge Input

possibly underestimates the real hydrographic basin (Prétorian, 1994), but the limited hydrogeologic
information available is not sufficient to delineate a real basin. Subcatchments are regrouped to form the
principal tributaries and contributing slopes (contributing areas around the river channel but with no
tributary). These subcatchments are shown in figure 5.4. Notice the elevation bands, based on the 100
meter elevation contours, that are used in the large Upper Venoge subcatchment. Table 5.1 shows the
mean elevation, area, and mean slope of each of the subcatchments, as well as the mean elevation and
area of each band.

Upper Venoge Bands (in color) - id shown

1'200
1'200

500
500
400
400

4'053
4'053
1'100
1'100 501
501 4'025
4'025
1'000
1'000 4'054
4'054 4'048
4'048
900
900
1'201
1'201 800
800 4'023
4'023
700
700 600
600 4'024
4'024
1'300
1'300
1'400
1'400 4'046 4'029
4'046 4'029
1'500
1'500
4'036
4'036
4'038
4'038
4'012
4'012
4'059
4'059 4'057
4'057
4'058
4'058
0 2.5 5 4'021
4'021
4'008
4'008
kilometres 4'003
4'003
model subcatchments
(id in bold) 4'002
4'002

4'001
4'001
Lower Venoge (1996)

Figure 5.4. Delineation of the subcatchments for the hydrologic modeling. ID’s of the subcatchments
and Upper Venoge snow storage units are shown.
 Data and modeling periods 123

Mean Mean
elevation Mean elevation
ID (m.a.s.l.) Area (ha) Slope (-) Band (m.a.s.l.) Area (ha)
4001 401 297 0.051 400 472 912
4002 407 286 0.059 500 559 219
4003 423 309 0.075 501 568 2618
4008 478 733 0.065 600 655 4235
4012 505 668 0.077 700 741 1881
4021 431 329 0.084 800 847 892
4023 814 15046 0.135 900 947 615
4024 485 360 0.122 1000 1052 630
4025 482 110 0.086 1100 1152 687
4029 474 277 0.084 1200 1239 159
4036 510 269 0.076 1201 1256 511
4038 470 283 0.095 1300 1354 757
4046 554 382 0.069 1400 1444 567
4048 556 1350 0.061 1500 1559 365
4053 527 204 0.091
4054 521 365 0.1
4057 428 233 0.094
4058 412 18 0.058
4059 524 2254 0.059

Table 5.1. Properties of the subcatchments and elevation bands as identified in figure 5.4

5.4 Data and modeling periods

5.4.1 Available data

Precipitation

The Venoge catchment is equipped with only one precipitation gage (nonrecording at Cossonay) (figure
5.5). Fortunately, the surrounding catchments are well equipped with the nonrecording gages of
MeteoSwiss. Four automatic federal climate stations (ANETZ network of MeteoSwiss) surround the
Venoge catchment, but are relatively far. Most ANETZ stations were implemented in 1978 providing a
network of 10 minute and hourly stepped data. The Vaud canton has tried to overcome this lack of fine
temporal resolution precipitation data by installing additional tipping bucket gages. Two of those gages
(figure 5.5) are indirectly used in this study.
124 Hydrologic Modeling for Distributed Discharge Input

La
La Fretaz
Fretaz
Federal nonrecording
Federal recording
Cantonal recording

VALLORBE(VILLE)
VALLORBE(VILLE)
CHAVORNAY
CHAVORNAY

CHARBONNIERES.L.
CHARBONNIERES.L.
Goumoens
Goumoens

ECHALLENS
ECHALLENS

LES
LES BIOUX
BIOUX
COSSONAY
COSSONAY
LE
LE SENTIER
SENTIER

BIERE
BIERE
MARCELIN/MORGES
MARCELIN/MORGES
Pully
Pully

Féchy
Féchy

0 5 10

La Dole
La Dole kilometres
Reproduced with permission
Changins
Changins from swisstopo (BA068334)

Figure 5.5. Precipitation gages used in the hydrologic modeling. Polygons show the zone of influence
attributed to each nonrecording gage. Background image is the DTM25 © 1995, Swisstopo.

Ele va tion sta rt- 1934- 1961- 1975- 1983-

ISM ID Sta tion name (m.a.s.l.) Pe riod Gage type 2003 2003 2003 2003 2003
6038 CHARBONNIERES.L. 1135 7.1982 - present nonrecording 1712 1712
6033 LES BIOUX 1025 7.1996 - present nonrecording
6030 LE SENTIER 1020 10.1900 - 6.1996 nonrecording 1551 1571 1639 1804 1739
6060 VALLORBE(VILLE) 762 1.1961 - present nonrecording 1370 1370 1483 1427
8245 BIERE 695 5.1974 - present nonrecording 1341 1389 1313
6100 ECHALLENS 618 1.1934 - present nonrecording 1061 1061 1030 1087 1023
8180 COSSONAY 570 10.1919 - present nonrecording 953 963 973 1034 965
6115 CHAVORNAY 439 1.1974 - present nonrecording 889 925 897
8230 MARCELIN/MORGES 425 1.1961 - present nonrecording 1051 1051 1149 1089
8290 Changins 430 1.1978 - present tipping bucket 1023
8280 La Dole 1670 8.1978 - present tipping bucket 1890
6185 La Fretaz 1202 1.1978 - present tipping bucket 1333
8100 Pully 461 1.1978 - present tipping bucket 1174
Féchy 470 11.1992 - present tipping bucket 1190
Goumoens 614 11.1992 - present tipping bucket 937

Table 5.2. Precipitation station data used in the hydrologic study. Mean annual precipitation is given
for different periods. Les Bioux and Le Sentier stations are summed together.
 Data and modeling periods 125

Temperature

The temperature data for the Venoge

Federal continuous stage recorder
La
La Fretaz
Fretaz
catchment is also located outside of Federal ANETZ climate station

the catchment. The nearest stations

are those of the ANETZ stations
(figure 5.6). These stations are
representative of the entire catchment
considering the Dole station is
located at 1670 meters, Fretaz at 1202
Lussery
Lussery
meters, and Pully at 461 meters.

Discharge
Ecublens
Ecublens

Pully
Pully
Discharge measurements have been
made by the Swiss National
Hydrologic Service (SHN). 0 5

kilometres
10

La
La Dole
Dole
Measurements exist for the station of Changins
Changins

Lussery and Ecublens. Daily

Figure 5.6. Climate and discharge stations used in the
measurements for Lussery exist from hydrologic model. The polygons show the zone of influence for
1944-1974 and are continuous during each climate station. Background is the DTM25 © 1995,
the period 1974-1978. The reproduced with permission from swisstopo (BA068334).
measurements at Ecublens are
continuous and started in 1979 and are still running. Without overlapping discharge series, the routing
aspects of the Lower Venoge River cannot be calibrated.

Potential evapotranspiration

Potential evapotranspiration can be obtained from MeteoSwiss. This data was examined but found to
seriously underestimate the PET (it is less than the average real evapotranspiration recorded in the Swiss
hydrological atlas (Spreafico et al., 1992)). Other European sources of PET were available, but data for
the streambank erosion modeling needed to be up-to-date. This requirement made it necessary to
calculate the potential evapotranspiration. Calculation of the PET by the Penman-Monteith equations
requires sunshine duration, relative humidity, wind, temperature, and global radiation data. This data,
without interruptions, is available from MeteoSwiss, from 1981 to present.

The available data is summarized in table 5.3.

5.4.2 Modeling strategy

The ANETZ station data begins in 1978 and the Dole station data is only available from August 1978,
which does not coincide well with the end of discharge measurements at Lussery in 1978. A 1978
simulation is possible for Lussery, but the use of calibrated model parameters based on only one year
would be perilous. Maximum annual discharge values are available though for Lussery so they will be
used to evaluate the simulated maximum annual discharges with the caution that is due.
126 Hydrologic Modeling for Distributed Discharge Input

A daily time step for the Venoge catchment is too long. Prétorian (1994) analyzed 19 runoff events from
1984-1991. The mean lag time of those events was 16 hours with values ranging from 6 to 27 hours.
Mean rising limb time was 32 hours, mean hydrograph duration was 100 hours and the mean time of
concentration was 66 hours. With a mean lag time of less than 24 hours, it is absolutely necessary to
study the Lower Venoge River at a smaller time step. As mentioned in chapter 3.4, the impermeable
urban area of the Venoge catchment is small, so that a small time step for simulating urban runoff is
unnecessary. The temporal precipitation structure of the farther hourly rain gages will be used to
disaggregrate the precipitation recorded at the daily nonrecording gages.

1944-1974 1974-1977 1978 1979-1980 1981-1992 1993-2.2005 3.2005

Lussery Q daily, max annual instantaneous
Ecublens Q instantaneous
Prec. - detailed
Prec. - less detailed
PET - Temp
PET - PM
Simulation Periods validation calibration

Table 5.3. Summary of the available hydrological data and simulation periods

Discharge information is available at Ecublens from January 1979 - March 2005. This period is long
enough for calibration and validation. From 1993 on, the cantonal hourly rain gages are also available
so the calibration period is from August 1993 - August 2004. The August starting point is chosen to avoid
water balance issues related to snow since all of the Jura snow typically melts by June. The validation
period is chosen from August 1979 - August 1993. This permits the use of the Charbonnieres data, but
it is assumed the associated error will be minimal due to the reasonable coverage of the surrounding Jura
gages.

To completely exploit the Ecublens discharge series for the streambank erosion modeling, it is necessary
to find a solution for the PET for 1979-1980. Recent advancements in this area (Oudin et al, 2005b) have
shown that a simple temperature based PET model is sufficient for continuous rainfall-runoff modeling.
Calibration will be done using Penman-Monteith calculated PET and simple temperature-based PET to
evaluate the performance of the two PET calculations. Holes in the ANETZ temperature data for the
1978-1980 period will need to be corrected.

5.4.3 Corrected and derived data

Precipitation

The daily precipitation data of the nonrecording gages shown in figure 5.5, must be transformed to
hourly data. This is done with the daily precipitation structures of the surrounding cantonal and federal
recording gages. The method consists of applying the daily recorded precipitation volume proportionally
onto the daily structure of a nearby recording gage. The optimal manner to do this procedure would be
to apply the precipitation event volume onto the event volume of the recording gage, but its
programming is much more complicated for a small gain. Occasionally, there are daily precipitation
volumes without any precipitation measured at the recording gage. In this case, the precipitation will be
uniform for that day. The other complication that occurs is that of the different processing of snow in the
 Data and modeling periods 127

precipitation gages by cantonal and federal measurement systems. Federal precipitation volumes take
into account snow that falls during the recording period. Snow in the cantonal gages is measured at the
time it melts. This proves to be a significant complication for using their data in combination with the
data of the federal stations. For this reason, only the cantonal stations of Goumoens and Féchy are used.
The modeling improvement that can be gained from the hourly recorded structures of these devices for
1993-2005 is greater than the few snow complications that occur at the elevations of these two stations.
For the period 1993-2005, the precipitation structures-volumes combinations are the following: Féchy
with Bière and Marcelin/Morges, Goumoens with Echallens, Cossonay, and Chavornay, La Dole with
Sentier-Bioux and Charbonnieres, and La Fretaz with Vallorbe. For the period 1979-1992, the structure
of the federal station of Pully was used in combination with the volumes of Echallens, Chavornay,
Cossonay, and Marcelin/Morges, Changins was used with Bière, La Dole with the Sentier, and La Fretaz
with Vallorbe. Prétorian (1994) analyzed the number of days a given precipitation intensity was
exceeded for the majority of the nonrecording stations shown in figure 5.5 and it clearly shows that
intense precipitation is more frequent at higher elevations. Ideally then, the rainfall structure should
come from a station that is at the same altitude or higher then the daily volume station. This was for the
most part respected, except for the Biere station. Further analysis would have been necessary to
determine more precisely if the Dole station or the Changins and Féchy station would be more
appropriate, but the Changins and Féchy stations were chosen because of the important elevation
difference between Biere and Dole.

Temperature

Correction of ANETZ temperature during the period 1978-1980 was necessary. This was done in one of
two ways. For holes that were greater than 6 hours, missing values were replaced from a nearby station.
Replacement values were taken from the closest ANETZ station if possible, but otherwise from climate
stations that measure temperature three times daily. The values were corrected for elevation differences
with a standard lapse rate of -0.005 °C/m. For holes of 6 hours or less, a cubic spline interpolation was
used to replace missing values.

Potential evapotranspiration

Potential evapotranspiration was calculated for the Venoge with two methods: Penman-Monteith (PM)
and a temperature-based model. The PM method is documented in many sources (ASCE, 1996 or Allen,
1998 for example) and will not be described in detail here. PM PET hourly time series were calculated
for the Pully, Dole, and Fretaz stations from January 1981 - August 2004.

Oudin et al. (2005a) demonstrated that lumped rainfall-runoff models only need long term mean PET
values. In a second companion article (Oudin et al., 2005b), they proposed a simple temperature based
model, which they called the "adjusted PE model with mean air temperature", and concluded that
rainfall-runoff model efficiency can be improved by using simple temperature based PET models. This
improvement is most likely due to adjustment factors in their PE formula that can compensate for model
128 Hydrologic Modeling for Distributed Discharge Input

structure deficiencies. Their model is issued from the Jensen-Haise and McGuinness models which take
the following generalized form:
Re Ta + K2
PET = ------ ------------------ for Ta + K2 > 0
λρ K 1 Eq. 5.2
PET = 0 for Ta + K2 ≤ 0

where PET is the potential evapotranspiration (m/day), Re is the extraterrestrial radiation (MJ/m2day)
which can be calculated with the latitude and Julian day, Ta is the mean daily air temperature (°C), λ is
the latent heat flux (standard value is 2.45 MJ/kg), ρ is the density of water (kg/m3) and K1 and K2 are
coefficients to calibrate. Oudin et al. (2005b) indicate that the chosen values of K1 and K2 could depend
on the model structure, but they assert that with values of 100 and 5 respectively, the model performance
due to the PET should be optimized. Temperature based PET was calculated for the Pully, Fretaz, and
Dole stations with the values of 100 and 0 for K1 and K2, respectively.

Muskingum parameters

No upstream and downstream hydrographs are available for the Lower Muskingum
Venoge River during the same period so it was decided to approach the id K (s)
hydrograph routing more as hydrograph translations (X=0.5) rather 4025 209
than attenuation to avoid a possible underestimation of the discharge 4054 586
4024 907
peaks due to an exaggerated attenuation. Translation times are based 4029 1144
on the K muskingum coefficient. This value can be estimated by 4038 1290
dividing the length of the reach by the mean reach velocity (Musy and 4057 1403
4058 146
Higy, 1998). The estimated mean reach velocities are based on the
4021 1434
HEC-RAS simulations used for the description of the Lower Venoge 4003 1116
hydraulic geometry (c.f. chapter 3.7.4). 4002 865
4001 1284

Table 5.4. Muskingum K values

for the Lower Venoge
5.5 Spatial interpolation of data
Estimates of hourly rainfall, temperature, and evapotranspiration must be provided to each HydroRoute
subcatchment. These estimates can be based on the recorded or derived values of nearby stations or an
areal estimate. The method used here follows the one presented by Hamdi et al. (2005). A brief
description is given hereafter.

5.5.1 Precipitation
The first step in the precipitation interpolation is to correct for elevation. A precipitation correction factor
is calculated according to:

P̂ an, i, k φi ( Zk – Zi )
r i, k = ---------------- = 1 + -------------------------- Eq. 5.3
Pan, i P an, i
 Spatial interpolation of data 129

where ri,k is the precipitation correction factor for the ith station and the kth band, P̂ an, i, k is the
estimated mean annual precipitation based on station i for the kth band, P an, i is the mean annual
precipitation for station i, φ is the lapse rate for station i as determined by a linear regression analysis,
and Zk and Zi are the elevations of the kth band and ith station, respectively. The estimated precipitation
for band k based on station i, P̂ i, k , then is:

P̂ i, k = r i, k P i Eq. 5.4

where Pi is the precipitation of station i. The estimated precipitation of the kth band, P̂ k , is calculated
according to:
m

P̂ k = ∑ α i, k r i, k P i Eq. 5.5
i=1

where αi,k is a weight expressing the influence of the ith station on the kth band according to the overlap
area of the Thiessen polygon of the ith station on the kth band.

The appropriate lapse rates

are determined by a linear 2000
regression analysis of
Mean annual precipitation (mm)

annual precipitation data 1800

y = 0.8398x + 808.8
(figure 5.7). A lapse rate of
1600
1.21 was found with the y = 1.2094x + 414.16
R2 = 0.8801
linear regression using all 1400
of the 1983-2003 federal
nonrecording stations. 1200 all 83-03

This lapse rate is too high Jura 83-03

1000 Linear (all 83-03)
for the Jura stations
Linear (Jura 83-03)
because for an elevation
800
equivalent to La Dole, the 400 500 600 700 800 900 1000 1100 1200
estimated precipitation Elevation (m .a.s.l.)
would be 2500 mm
Figure 5.7. Analysis of the effect of elevation on mean annual precipitation
compared to the recorded
to determine precipitation lapse rates
1900. A linear regression
was performed with only the Jura stations providing a lapse rate of 0.84 that gives a better estimate for
these high altitude stations.

The Thiessen weights necessary for the precipitation spatial interpolation are based on the polygons
shown in figure 5.5. The initial Thiessen polygons over the Jura mountains were adapted to prevent the
Jura stations from having any influence on the rainfall interpolation on the plateau part of the catchment.
Thiessen weights were generated based on their overlap with the snow storage units and subcatchments
of figure 5.4. Each Lower Venoge subcatchment of figure 5.4 has one snow storage unit. With the band
elevations, station elevations, station lapse rates, and thiessen weights matrix known, a spatial
interpolation program was run and the precipitation series of the 18 subcatchments and 14 bands of the
Jura subcatchment were produced for the period 1.1983-2.2005. The same procedure was run, but with
130 Hydrologic Modeling for Distributed Discharge Input

Thiessen weights adapted accordingly to replace the missing Charbonnieres station data, to produce the
precipitation series for the period 1.1978-2.2005.

5.5.2 Temperature
The spatial interpolation of temperature follows a procedure that is similar to that of the precipitation.
Rather than fixing the lapse rate for the entire interpolation period, it was decided to calculate a lapse
rate for every time step. This will allow periods of temperature inversion (higher temperatures at higher
elevations) to be taken into account. As for precipitation, the first step is to calculate an estimated
temperature, T̂ j, k , for the kth band with the jth temperature station according to:

T̂ j, k = T j + G j ( Z k – Z j ) Eq. 5.6

where Tj is the temperature of the jth station, Gj is the hourly lapse rate, and Zk and Zj are the elevations
of the kth band and the jth station, respectively. The estimated temperature for each band then is
determined as:
n

T̂ k = ∑ wj, k T̂j, k Eq. 5.7

j=1

where wj,k are the Thiessen determined weights. Here again, the initial Thiessen polygons were adapted
so that the Dole and Fretaz Jura stations only influenced the Jura region. Since the Dole and Fretaz
temperature series had been corrected to the beginning of 1978, only one temperature spatial
interpolation had to be run. This run calculated the temperature series for the 14 Jura subcatchments
bands. Due to the small elevation differences in the subcatchments of the Lower Venoge, the Pully
temperature data was used directly with no correction. Figure 5.8 shows the hourly lapse rate calculated
with the Dole, Fretaz, and Pully ANETZ stations for the year 2004. This example shows the inversion
that can occur during the fall and winter months.

Figure 5.8. Hourly temperature lapse rate (°C/m) calculated with the Dole, Fretaz, and Pully ANETZ
stations for the year 2004
 Spatial interpolation of data 131

5.5.3 Potential evapotranspiration

Potential evapotranspiration must be supplied to each subcatchment in the HydroRoute model. Since the
Upper Venoge is encompassed in one subcatchment, it needs only one potential evapotranspiration
series. The weights of 0.53, 0.21, and 0.26 for the three PET series: Pully, Fretaz, and Dole, respectively,
are to be used for the calculation of the Upper Venoge PET according to Thiessen polygons (same as
those used for temperature in figure 5.6). The Pully PET series is used for the Lower Venoge
subcatchments.

5.5.4 Accuracy of data

The accuracy of the utilized data must be assessed. A qualitative assessment is possible based on the
density of the gaging network and the spatial interpolation method.

The first question to ask is if the gaging network is sufficiently dense to accurately describe the spatial
rainfall. The Cossonay daily rainfall gage covers nearly half of the Venoge catchment (figure 5.5). This
indicates that the density is not sufficient. Huff (1970) measured areal precipitation in Illinois for areas
of 50-550 mi² and suggested that gage spacing should be 65 km2 per gage to ensure average sampling
errors of less than 6% for hourly rainfall, but also that the storm type affects sampling errors. Duncan et
al. (1993) also showed that the 5% accuracy gage spacing depends on the catchment and the storm types
and pattern and calculated 5% accuracy values (standard deviation of the error equals 5%) for gage
spacing in the range of 25-1000 km per gage. They also suggested though that for a 5% accuracy for peak
flow estimates, a much higher gage density is needed in the range of 1-41 km2 per gage. It can be
deduced then from this recommendation and the Venoge catchment area that the minimum number of
hourly recording gages necessary for the Venoge catchment is 6 and most likely more due to the
orographic effects of the Jura slopes.

Andréassian et al. (2001) also looked at the gage density question in light of its effect on hydrologic
modeling results. They showed that models have a capacity to adapt to poor rainfall input, but that the
performance and the variability of model efficiency improved with increasing gage density.

Concerning spatial interpolation, Beven (2001) noted that a precipitation correction for elevation is
justified in the case of orographic situations. Dirks et al. (1998) compared interpolation methods on a
high density network and showed that areal-mean, thiessen, inverse-distance, and kriging methods
performed similar. They concluded for high density networks that inverse-distance weighting should be
used for a better interpolation of the precipitation field, but also for its simplicity. For low to medium
density networks, Creutin and Obled (1982) showed that kriging outperformed Thiessen or areal-mean
methods. Goovaerts (2000) suggested that multivariate geostatistical algorithms can be used to integrate
DEM or radar information with ground gage information. These recommendations suggest that further
work can be done to improve the spatial interpolation of precipitation for the Venoge catchment.

Hamdi et al. (2005) noted that the spatial interpolation of temperatures seems to be as delicate as
precipitation interpolation. This remark is especially pertinent for the lack of temperature information
available for the Jura slopes of the Venoge catchment. For the rest of the catchment, this lack of
132 Hydrologic Modeling for Distributed Discharge Input

temperature information should be less important because of the small snowfall recorded on the plateau
part of the Venoge catchment.

5.6 Simulations and results

Two sets of calibration runs were done for the period August 1993 - August 2004, one with the PM PET
and the other with the temperature-based PET. Validation runs were done for August 1979 - August
1993. The parameter setup of the model is explained first, and then the results of the simulations are
presented.

5.6.1 Parameter estimation

It was necessary to estimate as many parameters as possible with standard values found in literature to
avoid an excessive number of parameters to calibrate. Based on the work done by Hamdi et al. (2005),
the critical temperature values for the snow/rainfall separation, Tc1 and Tc2, were taken to be 0°C and
2°C, respectively. They estimated the mean degree-day factor, as, to be 0.15 mm/h/°C for elevations in
the range of 500-1000 meters, which corresponds to suggested literature values (ASCE, 1996). The value
used for the Venoge was 0.16. The value of brn (h/mm) was estimated to be 0.3 h/mm. This value was
calculated so that a strong rain pushes the combined degree-day factor, as(1+brn*Rn), up to a value of
around 0.3 which is the value found in literature for snowmelt during windy and heavy rain (ASCE,
1996). The snow water retention capacity, θr, was set to 0.1 (Hamdi et al., 2005). The critical temperature
for snowmelt was determined from the Lussery discharge series, in which the Venoge spring snowmelt
is clearly visible. The diurnal temperature effect and the snowmelt peaks are clearly visible in figure 5.9.
The critical temperature was adjusted so that the snowmelt of the highest model snow storage unit
finished according to the end of the snowmelt in the 1978 Lussery discharge series. The critical
snowmelt temperature, Tc3, was estimated to be 0.25°C. The remaining parameters are calibrated.

40
35
30
Lussery discharge (m3/s)

25
Daily snowmelt peak
20 End of Jura snowmelt
15
10
5
0
800
700
Band 1500 snow equiv. (mm)

600
500
400
300
200
100
0
27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

May 1978 June 1978

Figure 5.9. Estimation of the snowmelt temperature based on the end of the simulated Jura snowmelt
and the snowmelt visible in the 1978 Lussery discharge series
 Simulations and results 133

5.6.2 Calibration
Adopted procedure

The HydroRoute model was adapted during this thesis to allow a global calibration. The parameters of
each of the subcatchments as well as the global parameters are available for calibration. Beyond two
subcatchments, the number of parameters to calibrate becomes unmanageable because no optimization
routine is available in HydroRoute. It was necessary to add four coefficients to allow for a global
calibration of the primary Venoge subcatchment parameters. These four coefficients are CA, CB, Cka,
CPET. They are multiplied against each subcatchment A, β, ka, and FPET, respectively. No subcatchment
differentiation of parameters was done except for the upstream β that was diminished because of the size
of the subcatchment. The parameter A was set to 250 mm, β to 7500 m4/3/s (6500 for the Upper Venoge
subcatchment), ka to 0.00022 h-1, and FPET to 1. These parameters (except for the Upper Venoge β)
correspond to those used in the Venoge study (Lance and Consuegra, 1997).

The adopted calibration procedure involved the simulation of 10’000 parameter sets generated via
Monte-Carlo random sampling of uniform distributions of each of the coefficients. The uniform
distributions for CA, CB, Cka were from 0.5 to 1.5 and for CPET the distribution was 1 to 1.75 for the PM
PET series and 2 to 3.5 for the temperature-based PET. The Nash or Log-Nash objective function criteria
is available for model performance evaluation. The Nash criterion was chosen to evaluate the model
performance over the entire period, with all values below the threshold of 4 m³/s (under-threshold), and
with all values above 40 m³/s (over-threshold). Using these three objective functions, should provide
parameter sets that are optimal for the entire range of discharges.

Results

The objective function results for 25 of the 10’000 simulations are shown in tables 5.5 and 5.6. Columns
2-5 show the randomly generated coefficients used in the simulations. The three Nash criteria are given
as well as their sum. The next three columns show the ranking of each of the Nash objective functions
(1 to 10000). The next two columns present two ways to combine the information of the ranks, one by
summing and one by choosing the maximum rank of the three. Combining the results of the three Nash
criteria helped to find a parameter set that was better for high flows. The best Nash results (for the entire
discharge series) for each of the two calibrations are 0.91 for the PM calibration and 0.89 for the
temperature-based PET one. This confirms the hypothesis of Oudin et al. (2005) that a simple
temperature-based PET is sufficient for hydrologic modeling. It should be noticed though, that the a
posteriori "good" parameter distributions might be different depending on the parameters used in the
temperature-based PET formula. As can be seen in the tables 5.5 and 5.6, the resulting "good" estimates
of β are significantly different for the two calibrations. The reason for this can be seen in figure 5.10.
The temperature-based PET (K1=65, and K2=0 of equation 5.2) produces lower winter and spring PET
than that of PM. Since the majority of high flows for the Venoge are in the winter and spring, the β of
the PM calibration must be higher due to the slow reservoir saturation that is lower. Figure 5.10 also
shows that using the temperature-based PET with K1=100 and K2=7 is roughly equivalent to the PM
equation for the Venoge catchment.
134 Hydrologic Modeling for Distributed Discharge Input

simula- rank rank rank maxi-

tion Nash Nash Nash 4 sum Nash Nash Nash 4 sum mum of
number C.ß C.A C.K C.PET entire 40 m³/s m³/s Nash entire 40 m³/s m³/s ranks ranks comments
11 0.64 1.00 0.52 1.44 0.91 0.37 0.28 1.55 1 1564 737 2302 1564 best entire
286 1.00 1.26 0.51 1.08 0.86 0.59 -1.02 0.43 5280 1 7282 12563 7282 best high flow
8982 0.71 1.44 0.53 1.74 0.84 -1.45 0.61 -0.01 6786 8793 1 15580 8793 best low flow
5655 1.26 1.11 0.58 1.73 0.88 0.49 0.42 1.79 2595 499 222 3316 2595 best sum Nashs
9780 0.95 1.06 0.58 1.56 0.90 0.49 0.36 1.75 238 487 416 1141 487 best sum ranks
7331 1.01 1.06 0.59 1.60 0.89 0.50 0.38 1.77 463 438 343 1244 463 lowest maximum rank
3679 0.93 0.98 0.60 1.57 0.90 0.51 0.32 1.73 269 352 554 1175 554
4953 0.84 0.94 0.57 1.55 0.90 0.50 0.30 1.70 181 409 641 1231 641
7126 0.77 0.95 0.54 1.56 0.90 0.47 0.34 1.71 85 679 487 1251 679
7720 0.91 1.08 0.57 1.51 0.90 0.48 0.32 1.70 108 604 556 1268 604
2292 0.84 1.00 0.55 1.61 0.90 0.46 0.38 1.75 232 719 321 1272 719
665 0.97 0.99 0.64 1.53 0.90 0.51 0.28 1.69 241 323 726 1290 726
5706 0.94 1.00 0.51 1.63 0.89 0.52 0.37 1.78 610 297 389 1296 610
1046 1.08 1.09 0.56 1.51 0.89 0.54 0.30 1.73 518 150 655 1323 655
8937 0.78 0.95 0.50 1.63 0.90 0.47 0.37 1.74 285 686 368 1339 686
9329 0.85 0.98 0.60 1.47 0.90 0.51 0.24 1.65 76 349 914 1339 914
5087 1.01 1.01 0.64 1.51 0.90 0.52 0.26 1.68 288 264 819 1371 819
9545 0.96 1.09 0.56 1.58 0.90 0.46 0.39 1.75 326 746 303 1375 746
5322 0.97 0.96 0.62 1.56 0.90 0.51 0.30 1.70 376 347 664 1387 664
5213 0.95 0.94 0.66 1.61 0.90 0.49 0.33 1.71 353 518 542 1413 542
8301 0.83 1.02 0.59 1.48 0.90 0.47 0.28 1.65 35 645 736 1416 736
2664 0.98 1.01 0.67 1.49 0.90 0.51 0.24 1.65 172 320 950 1442 950
4309 0.93 0.96 0.53 1.54 0.89 0.52 0.28 1.70 466 286 714 1466 714
612 0.84 0.93 0.56 1.50 0.90 0.51 0.24 1.65 219 366 900 1485 900
834 1.09 1.05 0.65 1.63 0.89 0.49 0.38 1.76 653 483 351 1487 653

Table 5.5. Calibration objective function results for 25 of the 10000 simulations using Penman-
Monteith PET

simula- rank rank rank maxi-

tion Nash Nash Nash 4 sum Nash Nash Nash 4 sum mum of
number C.ß C.A C.K C.PET entire 40 m³/s m³/s Nash entire 40 m³/s m³/s ranks ranks comments
1815 0.63 0.96 0.77 2.38 0.89 0.54 -0.10 1.33 1 101 3931 4033 3931 best entire
2197 0.85 0.84 1.09 2.55 0.89 0.61 -0.19 1.32 250 1 4886 5137 4886 best high flow
5090 0.55 1.47 0.52 3.23 0.80 -1.97 0.44 -0.73 8810 8810 1 17621 8810 best low flow
2178 0.81 0.94 0.68 3.04 0.88 0.58 0.19 1.65 1549 1558 909 4016 1558 best sum Nashs
5167 0.58 0.95 0.59 2.69 0.89 0.50 0.17 1.57 183 273 993 1449 993 best sum ranks
5924 0.62 1.02 0.70 2.84 0.88 0.29 0.21 1.39 754 785 753 2292 785 lowest maximum rank
8350 0.57 0.93 0.67 2.68 0.89 0.48 0.14 1.51 101 198 1266 1565 1266
1230 0.56 0.99 0.58 2.55 0.89 0.46 0.14 1.49 96 193 1294 1583 1294
6235 0.55 0.91 0.59 2.78 0.89 0.50 0.19 1.57 309 381 895 1585 895
250 0.54 0.93 0.62 2.59 0.89 0.49 0.12 1.51 53 152 1404 1609 1404
8442 0.54 0.95 0.62 2.76 0.89 0.39 0.20 1.48 352 421 845 1618 845
8386 0.64 0.93 0.75 2.77 0.89 0.52 0.14 1.55 148 241 1255 1644 1255
422 0.60 0.97 0.55 2.74 0.89 0.49 0.20 1.58 400 465 816 1681 816
7516 0.53 0.92 0.54 2.77 0.89 0.49 0.20 1.58 394 459 830 1683 830
7114 0.63 1.05 0.59 2.66 0.89 0.40 0.19 1.48 378 443 865 1686 865
8276 0.60 0.92 0.67 2.86 0.89 0.50 0.19 1.58 382 447 889 1718 889
5066 0.68 0.97 0.63 2.71 0.89 0.55 0.16 1.60 275 350 1100 1725 1100
368 0.58 0.90 0.57 2.77 0.89 0.55 0.18 1.62 359 427 970 1756 970
2480 0.60 0.92 0.63 2.60 0.89 0.56 0.11 1.56 71 168 1547 1786 1547
7027 0.59 1.00 0.70 2.61 0.89 0.39 0.11 1.40 111 207 1482 1800 1482
876 0.61 0.91 0.74 2.68 0.89 0.54 0.10 1.53 41 140 1623 1804 1623
6534 0.72 1.00 0.74 2.68 0.89 0.52 0.12 1.53 147 240 1423 1810 1423
1063 0.67 0.94 0.80 2.72 0.89 0.52 0.10 1.52 77 174 1618 1869 1618
8282 0.62 1.06 0.65 2.71 0.89 0.30 0.20 1.38 552 601 847 2000 847
2460 0.74 0.90 0.87 2.89 0.89 0.58 0.11 1.58 228 312 1499 2039 1499

Table 5.6. Calibration objective function results for 25 of the 10000 simulations using temperature-
based PET

All of the parameter sets of tables 5.5 and 5.6 had problems reproducing the extreme discharge events
(figure 5.12). The HydroRoute model doesn’t save output variables for multiple runs so it was unfeasible
to test a lot of simulations for their extreme value distributions. Parameters were changed from the initial
parameter set until an acceptable result in terms of the entire Nash, high flow Nash, and the extreme
value distribution was found. The parameter set found was: β = 7500 (except for the Upper Venoge β =
6500), A=250, ka=.00018, and FPET=2.5 for the temperature-based PET calibration. The entire Nash was
 Simulations and results 135

0.88, the high flow Nash 0.47, and the low flow Nash -0.08. The low flow Nash is not good, but in the
case of streambank erosion, it does not need to be better. It should be remembered that the high and low
flow Nash criteria are very strict so the reader should not be shocked by these values. Figure 5.13 shows
the results of the Gumbel frequency analysis. The fitted, simulated maximum annual discharges at the
Ecublens station correspond very well to the observed 1993-2004 trend, which is different from the
1979-2004 trend (figure 5.13). The Lussery 1993-2004 fitted maximum annual discharges doesn’t
correspond well though to the Lussery 1948-1978 gumbel fit (figure 5.14). This can be explained by the
two very important high flows that occurred during the 1993-2004 period. The 2004 high flow was one
that was particularly generated from the Upper Venoge catchment precipitation. Replacing those two
high flows by lower flows typical of the 1948-1978 period produced the fictive Lussery fitted line that
is shown in figure 5.14.

The high flows as well as the low

160
flows are relatively well PM
Temp,K1=65,K2=0
reproduced as illustrated in figure 140
Temp,K1=100,K2=7
5.11. The variability around the 120

regression line shows the error 100

PET (mm)

due to the model structure and 80

especially the error due to the 60

inaccuracies of the input data. The
40
regression line’s slope is 1.00
20
showing that there is no bias in the
0
calibrated result. 1 2 3 4 5 6 7 8 9 10 11 12
month

Figure 5.10. Comparison of mean monthly PET calculated with PM

and the temperature-based method

70
70
simulated
60 60
observed
Simulated Mean Daily Q (m³/s)

50 50
discharge (m³/s)

y = 1.0014x
40 R2 = 0.8916
40

30
30

20
20

10
10
0
0.1 1 10 100 1000 0
0 10 20 30 40 50 60 70
Days per year exceeded
Mean Daily Q Ecublens (m³/s)

Figure 5.11. Flow duration curve (left) and the observed vs. simulated mean daily flows (right) for the
period 1994-2004.
136 Hydrologic Modeling for Distributed Discharge Input

Qi obs Ecub79-04 fitted 140

Qi obs Ecub79-04
Rp = 2
Rp = 5
Rp = 10 120
Rp = 30
Rp = 100
Qi obs Ecub93-04 fitted 100
Qi sim Ecub93-04 fitted
Discharge (m /s)

80
3

60

40

20

0
-2 -1 0 1 2 3 4 5

Reduced Variate

Figure 5.12. Comparison of maximum annual discharge behavior of the best "sum of Nash" of the
temperature-based PET calibration versus observed discharges using a Gumbel (EVI) frequency
analysis

110
5 years

10 years

30 years

100 years
Qi obs Ecub79-04 fitted
Return period = 2 years

Qi obs Ecub79-04
Qi obs Ecub93-04 fitted 100
Qi sim Ecub93-04 fitted
Qi sim Ecub79-04 fitted
90

80
Discharge (m3 /s)

70

60

50

40

30

20
-2 -1 0 1 2 3 4 5
Reduced variate

Figure 5.13. Gumbel frequency analysis of maximum annual discharges for the station of Ecublens,
observed and simulated
 Distributed hydrographs for streambank erosion modeling 137

90

5 years

10 years

30 years

100 years
Qi obs lus48-78 fitted

Return period = 2 years

Qi obs lus48-78
Qi sim Lus93-04 fitted 80

Qi sim Lus fictive

70

60
Discharge (m /s)
3

50

40

30

20

10

0
-2 -1 0 1 2 3 4 5
Reduced variate

Figure 5.14. Gumbel frequency analysis of maximum annual discharges for the station of Lussery,
observed and simulated

5.6.3 Validation
A simulation for the period August 1978 - August 1993 was run to validate the calibration. It was run
only with the temperature-based PET. The resulting entire Nash is 0.83, the high flow Nash is 0.13, and
the low flow Nash is -0.47. These results are lower than the calibration, even considerably for the high
and low flow Nashs, but this is to be expected because the station Charbonnieres was not available, and
especially because of the lack of the two cantonal recording precipitation gages that provide a better
temporal structure for the majority of the catchment.

Although, the hourly discharge estimates of the simulation were degraded, the gumbel frequency
analysis of the 1979-2004 shows a very good fit with the observed series (figure 5.13).

5.7 Distributed hydrographs for streambank erosion modeling

The goal of the work presented in this chapter is to produce hydrographs for the upstream and tributary
inputs to the streambank erosion study reach. The calibration of the hydrologic model provided good
results and the validation is within the values expected. The simulated values for Jan. 1, 1979 - Aug. 22,
1993 and the calibration simulated values for Aug. 23,1993 - February 28, 2005 were used for generating
the hydrograph inputs for the streambank erosion study reach. With a historical discharge series
138 Hydrologic Modeling for Distributed Discharge Input

available though, its information should be capitalized on for the simulation of the historical period. It is
proposed then to correct the simulated discharges according to the observed discharge with the following
formula:
Q obs, i
Qest, j = --------------Q sim, j Eq. 5.8
Q sim, i

where Qest,j is the estimated discharge for catchment j, the Qobs,i is the observed discharge at station i,
Qsim,i is the simulated discharge at station i, and the Qsim,j is the simulated discharge for subcatchment
j. This correction of simulated discharges was done to produce the estimated discharges of the combined
subcatchments upstream and including subcatchment 4057, and the subcatchments 4012, 4058, 4008,
4021, 4003, and 4002 (see figure 5.4 for the subcatchment locations). Figure 5.15 illustrates the
correction of simulated discharges for the January 14, 2004 flood.

80 80
75 Ecublens obs. 75
Upstream est.
70 70
Ecublens sim.
65 Upstream sim. 65
60 60
55 55
Discharge (m3/s)

50 50
45 45
40 40
35 35
30 30
25 25
20 20
15 15
10 10
5 5
.12-12: .13-12: .14-12: .15-12: .16-12: .17-12:
<_2004.01.12-00:00:00 2004.01.18-00:00:00_>

Figure 5.15. Simulated and historically corrected discharges

5.8 Extreme hydrographs and scenarios

The historical period for the Lower Venoge River is long enough to provide an adequate picture of the
effect of high, low and typical precipitation years on the river flows. As a simpler alternative to the
generation of discharge with stochastically generated meteorological data, it is proposed here to generate
300 years of Lower Venoge River discharge based on the discharge series created in section 5.7. Twenty-
five years of discharge for the Lower Venoge River is available from 1980-2004. It is proposed to use
the high flows during these years to generate extreme hydrographs and to replace the historical floods
with the generated ones.
 Extreme hydrographs and scenarios 139

5.8.1 Flood volume and duration frequency analysis

Streambank erosion depends not only on the event peak discharge, but also on the event mean discharge.
Thus, when generating flood hydrographs it is necessary to take into consideration frequency analysis of
not only the peak discharges, but also of the flood volumes and durations. One possible approach is the
QDF (discharge-duration-frequency) approach (Musy and Higy, 1998). Frequency analysis is performed
on the mean discharge for various time steps and then can be graphed together (figure 5.16). A
hydrograph of a given frequency can then be constructed based on the QDF curve of that frequency and
typical rising limb times. This approach can lead to unrealistic hydrographs (single peak instead of multi-
peaks), though, so it is preferred to generate hydrographs based on historical ones. As all floods above
9 m³/s will be simulated for streambank erosion, volume and duration frequency is analyzed specifically
at that threshold and the results are given in table 5.7.

160

140

120
Mean Discharge (m /s)
3

100

80

60

40

20

0
0 10 20 30 40 50 60 70
Duration (days)
Return period 2 5 10 20 30 50 100 300 1000

Figure 5.16. Mean discharge-duration-frequency curves for the Ecublens Venoge River recording
station
140 Hydrologic Modeling for Distributed Discharge Input

Frequency analysis of instantaneous peak discharge, Return

volume and duration were performed with the period
HYFRAN© (HYdrological FRequency ANalysis) (years) Q (m³/s) Vol. (m³) Dur. (days)
software. The Gamma distribution is recommended 25 92.7 5.41E+07 32.95
27 94.1 5.51E+07 33.56
by the software for fitting the annual series of the
30 95.6 5.63E+07 34.22
Venoge River Ecublens data, although the Gumbel 33 97.3 5.75E+07 34.96
distribution gives a good fit to the data also. It is 38 99.2 5.89E+07 35.78
preferred to work with the Gumbel distribution as 43 101.4 6.05E+07 36.71
50 103.8 6.24E+07 37.77
there is not enough years of data to confirm either the
60 106.8 6.46E+07 39.04
Gamma or Gumbel distribution and the Gumbel 75 110.3 6.72E+07 40.58
distribution predicts higher rare flood peaks, volumes, 100 114.9 7.07E+07 42.57
and durations. 150 121.3 7.55E+07 45.36
300 132.3 8.38E+07 50.13
1000 151.4 9.81E+07 58.39
Table 5.7. Gumbel frequency analysis results for
5.8.2 Venoge River historically based instantaneous peak discharges, and flood
flood hydrographs volumes and durations of the Venoge River for
the Ecublens recording station
To estimate a 300 year discharge series, the 25 year
historical record will be repeated twelve times. Of course this historical record needs to be correctly
completed with higher return period floods. Twelve floods respecting the characteristics of table 5.7
must be generated and inserted properly into the 300 year series. Six historical hydrographs were
selected for the generation of these twelve hydrographs. The characteristics of those 6 historical
hydrographs are placed in the context of the annual floods in table 5.8.

The low probability floods are generated by associating one of the low probability floods of table 5.7
with a historical flood and then are generated in three steps: firstly, all discharges are increased to raise
the desired length of the hydrograph above the limit of 9 m³/s, secondly, discharges of one or multiple
peaks of the historical flood are increased so that the peak discharge is respected, and finally, all other
non-peak discharges are increased or decreased to produce the required flood volume. This procedure
assumes that the return period of the instantaneous peak discharge, volume, and duration of low
probability floods is equal. This assumption is not valid on an event basis as evident in table 5.8, but on
average the volume and duration of the flood does increase with increasing flood peak discharge. Putting
the low probability volumes and durations together with the same probability peak discharge will be a
conservative estimation of low probability hydrographs for streambank erosion modeling. The twelve
low probability floods as well as the 1000 year flood, which will be used as the extreme flood, are shown
in figure 5.17. These low probability floods are proportioned into the upstream input and tributary
hydrographs based on the subcatchment areas compared to the Ecublens measuring station catchment
area.
 Extreme hydrographs and scenarios 141

Return Return
Return period period
Q peak Duration period Q volume duration
Event start date (m³/s) Volume (m³) (days) (years) (years) (years)
01.01.1987 22:06 20.1 3.12E+06 2.30 1.0 1.1 1.0
09.12.1994 18:58 24.4 3.99E+06 2.82 1.1 1.1 1.0
30.11.1996 00:27 27.7 5.84E+06 4.39 1.1 1.1 1.1
03.02.2003 16:00 28.3 2.78E+06 2.09 1.2 1.0 1.0
29.06.1997 18:58 30.2 1.02E+07 8.22 1.2 1.2 1.3
02.03.1989 23:58 32.2 4.90E+06 3.31 1.3 1.1 1.0
18.01.1998 20:26 35.1 5.59E+06 4.48 1.3 1.1 1.1
22.12.1991 01:33 36.6 7.15E+06 4.08 1.4 1.1 1.1
23.01.1986 15:45 40.0 5.56E+06 3.10 1.5 1.1 1.0
10.10.1988 02:05 40.7 1.02E+07 5.12 1.6 1.2 1.1
01.05.1983 12:06 46.4 4.33E+07 31.88 1.7 10.4 21.5
25.12.1999 22:10 46.5 1.23E+07 6.83 1.8 1.3 1.2
22.01.1985 17:49 48.3 1.01E+07 5.59 1.9 1.2 1.1
31.01.1984 17:07 50.1 1.72E+07 11.51 2.1 1.7 1.7
29.02.2000 16:36 51.3 9.83E+06 5.75 2.3 1.2 1.1
20.01.1995 21:52 52.6 2.94E+07 15.16 2.5 3.6 2.4
28.09.1993 13:09 54.6 3.56E+07 22.81 2.7 5.7 6.1
15.11.1992 11:26 63.1 2.53E+07 14.89 3.1 2.7 2.3
31.12.1981 01:32 63.4 2.61E+07 15.28 3.5 2.8 2.4
31.01.1980 18:46 63.7 2.06E+07 11.32 4.0 2.0 1.6
26.01.1979 21:29 68.5 9.74E+06 5.11 4.7 1.2 1.1
14.02.1990 02:31 74.4 2.34E+07 9.83 5.8 2.4 1.4
12.01.2004 00:54 77.8 2.36E+07 10.69 7.4 2.4 1.5
03.03.2001 12:08 80.5 6.78E+07 33.02 10.4 78.8 25.3
07.12.1982 23:05 81.8 3.68E+07 18.58 17.3 6.2 3.5
09.11.2002 20:45 93.4 3.97E+07 23.23 52.0 7.8 6.5

Table 5.8. Annual peak floods recorded at the Ecublens Venoge River recording station. Events shaded
in grey are those selected for the generation of low probability floods.

5.8.3 Discharge scenarios

Three 300 year discharge time series (scenarios) are constructed based on the simulated historical
discharges from 1980-2004 and the low probability floods of figure 5.17. The base 300 years are
constructed by repeating the period 1980-2004 twelve times to form a period from 1980-2179. The
twelve events to be inserted into the 300 year period are chosen in a random order. This is done for each
of the three scenarios (table 5.9 and figure 5.18) and for each tributary and the upstream input within
each scenario.
142 Hydrologic Modeling for Distributed Discharge Input

120
observed
Rp=38 yr
100 Rp=43 yr

80
Q (m 3/s)

60

40

20

0
10.12.1981 20.12.1981 30.12.1981 09.01.1982 19.01.1982 29.01.1982 08.02.1982

160 observed
Rp=150 yr
140
Rp=300 yr
Rp=1000 yr
120

100
Q (m 3/s)

80

60

40

20

0
06.04.1983 16.04.1983 26.04.1983 06.05.1983 16.05.1983 26.05.1983 05.06.1983

120
observed
Rp=27 yr
100 Rp=33 yr

80
Q (m 3/s)

60

40

20

0
07.12.1993 17.12.1993 27.12.1993 06.01.1994 16.01.1994 26.01.1994 05.02.1994

Figure 5.17. Low probability floods generated from historical floods

 Extreme hydrographs and scenarios 143

120
observed
Rp=50 yr
100 Rp=60 yr

80
Q (m 3/s)

60

40

20

0
16.01.1995 26.01.1995 05.02.1995 15.02.1995 25.02.1995 07.03.1995 17.03.1995

120
observed
Rp=75 yr
100 Rp=100 yr

80
Q (m 3/s)

60

40

20

0
28.02.2001 10.03.2001 20.03.2001 30.03.2001 09.04.2001 19.04.2001 29.04.2001

120
observed
Rp=25 yr
100 Rp=30 yr

80
Q (m 3/s)

60

40

20

0
03.11.2002 13.11.2002 23.11.2002 03.12.2002 13.12.2002 23.12.2002 02.01.2003

Figure 5.17 (cont.). Low probability floods generated from historical floods
144 Hydrologic Modeling for Distributed Discharge Input

Scenario 1 Scenario 2 Scenario 3

To respect the historical frequency
Return Return Return analyses, it is necessary to take out
Period Year Period Year Period Year
the historical flood that corresponds
150 1983 30 2002 27 1993
300 2008 150 2008 60 2020 to the date of the low probability
50 2045 100 2051 50 2045 flood and to take out the floods that
100 2076 38 2056 33 2068
had return periods higher than 25
25 2102 43 2081 43 2081
33 2118 300 2108 100 2126 years. This is the case for the 2001
30 2152 60 2145 150 2133 (high volume and duration) and 2002
27 2168 75 2176 30 2177
75 2201 25 2202 38 2181 (high peak discharge) floods. These
43 2206 27 2218 75 2226 floods are eliminated from each of the
60 2245 50 2245 25 2252
scenarios.
38 2256 33 2268 300 2258
1000 2283 1000 2283 1000 2283
The resulting 300 year scenarios each
Table 5.9. Insertion of low probability floods into the 300 year
contain 2826 streamflow events with
record to form three 300 year scenarios. The 1000 year flood is
added to the end of each scenario. flow greater than 9 m³/s.

constructed
236 floods during the 25 year historical period low prob. floods

25 year
300 years 27
scenario 1 30
33
scenario 2 38
43
scenario 3 50
60
extreme flood

75
100
150
300
Figure 5.18. Example illustrating the insertion of the 25 and 27 year return period floods into the three
300 year series. One rectangle represents the 236 floods of the 25 year historical period.

Only three 300 year scenarios are constructed here to serve as an example. More should be constructed,
but to do so an automated method would be necessary. Rather than automating such a method, it would
be preferable to generate meteorological scenarios (Wojcik and Buishand, 2003) for input into the
hydrologic model to generate many discharge series.

5.9 Summary and conclusions

Hydrologic data for the Venoge catchment was collected. Although, the data measured within the basin
is largely insufficient, using the surrounding data allows for reasonable estimates of hydrologic input
 Summary and conclusions 145

data. Daily precipitation is disaggregated to hourly precipitation via the 10 minute recording gages.
Precipitation, temperature, and PET are spatially interpolated for each Venoge subcatchment. PET is
calculated by both the Penman-Monteith method as well as a temperature based method.

The model HydroRoute was used to do continuous hydrologic modeling. This model is able to take into
account snowfall and snowmelt, which is important for the Jura mountains of the Venoge catchment. The
Upper Venoge is modeled as one big subcatchment with many snow storage units, while the Lower
Venoge is divided into many subcatchments to provide a detailed evolution of the Lower Venoge River
flows.

Calibration coefficients were added to HydroRoute. These calibration coefficients allow a regional
calibration of subcatchment parameters. One of the calibration coefficients is multiplied against the
subcatchment PET. This calibration coefficient can be important to overcome evapotranspiration
deficiencies in the model structure.

Calibration of the four calibration coefficients was done for the Venoge catchment using Monte-Carlo
simulations. The best calibration yielding a Nash value of 0.91 is an excellent result for continuous
hydrologic modeling although the stricter criteria of using the Nash criteria on the high flows or the low
flows shows the difficulty of reproducing hydrograph peaks and low flows precisely. Validation gave
poorer results as expected due to the diminished quantity of input data for this period. The Nash criteria
was insufficient for finding parameter estimations to reproduce high flows adequately. Output of high
flow statistics by the HydroRoute model would permit an improved model calibration of high flows.

Temperature based PET produced an equally good calibration as that done with the Penman-Monteith
PET. This verified the hypothesis of Oudin et al. (2005a) that temperature based PET is sufficient for
continuous hydrologic modeling. Other model parameters can be affected though if the temperature
based PET gives an annual PET series that is significantly different than more sophisticated PET
calculations such as Penman-Monteith.

The simulated January 1979 - March 2005 discharges were proportioned according to the discharge
observed at the Ecublens recording station to improve the historical estimations for the Lower Venoge
tributary inputs.

Frequency analysis of the instantaneous peak flow, flow duration, and flow volume was done for the
Ecublens observed discharge series. This provides the criteria to respect in adapting historical high flow
events to generate low probability flood events. Twelve low probability flow events are generated and
combined in a 300 year flow series based on the 25 year historical period. This procedure was repeated
three times to produce three series of 300 years. An extreme flood is added onto each of the 300 year
discharge scenarios.

More than three discharge scenarios should be produced, although three discharge scenarios for the
Venoge case will be sufficient to serve as an example in the streambank erosion hazard mapping
methodology. To produce more scenarios, the scenario generation would need to be automated, or better
yet, discharge scenarios should be generated based on stochastically generated meteorological data.
146 Hydrologic Modeling for Distributed Discharge Input
 Chapter 6

Geofluvial Streambank Erosion

Hazard Modeling and Danger
Mapping

6.1 Introduction
As described in chapter 2.3.2, geofluvial modeling combines mechanistic fluvial and bank failure models
to be able to perform streambank erosion hazard modeling. To calculate bank failures and their
probabilities it is necessary to use this type of model. The methodology for calculating and mapping a
streambank erosion hazard was outlined in chapter 2.4. The topographic, hydraulic, geotechnical, and
hydrologic data that is used in connection with the geofluvial model, CCHE1D in this case, was presented
in chapters 4 and 5.

This chapter will firstly show how the model is setup and then calibrated. Bank parameters will be
spatialized to permit the simulation of streambank erosion throughout the reach. Finally, the streambank
erosion hazard simulations will permit their mapping.

6.2 Model setup

The modeling reach is from station kilometer 10.516 to station kilometer 2.538 (map shown in figure A.3
of the appendix). Although, the data has been presented in the previous chapters, the procedure to bring
that data into the required CCHE1D format is summarized here:

1. CCHE1D calculation nodes. After the channel centerline has been digitized and smoothed, nodes
from that polyline are extracted according to cross section and hydraulic structure placement and

147
148 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

according to the spatial resolution needed. The procedure is fully described in chapter 2.4.2.
2. Hydraulic structures. If structures are to be modeled, the appropriate CCHE1D dbf files must be
prepared. For the Venoge, only one inline weir is modeled as a drop structure, that of the recording
station at kilometer 4.908. Bridges are modeled as regular cross sections with ineffective flow areas
where flow obstruction occurs.
3. Channel and link CCHE1D files. These files are prepared based on the node and hydraulic
structures. Scripts for the GIS, Mapinfo Professional® in this study, were developed to prepare these
files.
4. Grain size classes. Bed and bank data require grain size distributions of the bed and bank material.
The size classes must be specified for the model. This is done in the CCHE1D *_sedclas.dbf file.
5. Grain size fractions. Each Grain Size Distribution (GSD) for the bed and bank material must be
divided into the grain size classes by indicating the fraction of material in each class. This grain size
fraction information is put into the *_grfrac.dbf file.
6. Bed data. According to the sampling locations, polygons in the GIS are made to show the area of
influence of each sample. These polygons become the basis for the CCHE1D *_sedbed.dbf file.
Each *_sedbed.dbf entry contains attributes defining if the bed is erodible or not, and if so to what
depth, and also the bed porosity, and identifier of the grain size fraction histogram. Nodes must be
updated with the identifier of the bed data that applies to that node.
7. Bank data. Polygons are drawn in the GIS to represent homogeneous bank areas. These polygons are
the graphical representations of each entry in the *_sedbank.dbf file. Each entry must define each of
the required bank soil properties as described in detail in chapter 4.6. Each entry must also contain
the identifier that links it to its grain size fractions. Nodes must be updated with the identifier of the
bank data that applies to that node.
8. Cross sections. This step requires several sub-steps using the cross section tool developed during
this thesis. They are described in detail in chapter 4.4 and are summarized here:
• Cross sections points are identified by their station number which is the downstream distance to
the outlet in meters for this study.
• Cross section positions and transverse distances are calculated using the cross section tool.
• Bank definition points are chosen.
• Cross sections are interpolated linearly onto each node.
• Manning values are based on the homogeneous roughness zones as shown in chapter 4.4.1 and
ineffective flow areas are attributed to the appropriate cross sections points.
• Cross section lines for the CCHE1D program are generated by a script in the GIS based on the
cross section points CCHE1D file.
9. Baseflow. A text file is prepared according to the CCHE1D user’s manual specification (Vieira and
Wu, 2002). The baseflow for the Venoge is set at 9 m³/s since the hydrographs to be simulated begin
at roughly this value.
10.Upstream boundary condition. The upstream boundary condition is prepared with a program that
takes a discharge time series and multiplies it against a sediment load curve to prepare the CCHE1D
*.bc file. This is fully described in chapter 4.7.
11.Downstream boundary condition. A text file is prepared for CCHE1D specifying the rating curve at
the downstream outlet. This is also described in chapter 4.7.
12.CCHE1D input file. This file is generated automatically if using the ArcView® interface of the
CCHE1D program. Otherwise, it must be prepared manually. This was done by running the
 Model setup 149

ArcView interface for an example and reusing it for the Venoge case. Many important parameters
are specified in this file. The various parameters used for the Venoge are:
• Sediment transport equation: code 2, Wu-Wang formula
• Bed material porosity: code 2, default method
• Roughness method: code 1, user specified
• Washload method: code 2, default method
• Suspended load adaptation length method: code 1, user specification, α = 0.5 for the Venoge
• Bed adaptation length: code 3, method based on alternate bar spacing
• Mixing layer method: code 2, related to the bed material diameter
• Timestep: 1 hour, no significant difference was found between 1 hour and 1 minute timesteps
• Dynamic or Diffusive Wave: Diffusive; significant differences occur only in the vicinity of the
weirs
13.CCHE1D output definition file, *_mp_df.txt. This file indicates to the CCHE1D program the
variables that need to be output and at what locations and what times. The variables that can be
output and the output formats are described in the user’s manual (Vieira and Wu, 2002).

The following files are used specifically for this thesis and are not required by the standard CCHE1D
program:
14.Second output definition file, "jb_output_def.txt." This is a second file used for output of other
CCHE1D variables that can not be exported via the standard CCHE1D file.
15.Simulation type file, "inicon.txt". The CCHE1D program can be run with or without the initial
conditions run. This can be helpful for repetitive simulations.
16.Centerline and erosion parameters file. This file contains the x and y coordinates of each node as
well as the variables k1, k2, b, and k5 of equation 2.22, and α of equation 2.16.

It is reminded here that the above steps are necessary when running CCHE1D without the ArcView®
interface. The manner of preparing data for CCHE1D using the ArcView® interface is fully described
in the user’s manual (Vieira and Wu, 2002).

Once the above steps have been completed, it is possible to run the CCHE1D program from a command
line. To facilitate the program execution, a program was written to allow CCHE1D to be run in batch
mode. This program asks for the location of the CCHE1D executable and the location of the input files
and will run CCHE1D repeatedly until all input files have been run.

In step 8 above, cross sections were linearly interpolated. To replace the cross sections with cross
sections that respect radius of curvature (see chapter 4.4 for details), the Manning roughness values of
the linearly interpolated cross sections are calibrated with historical water surface elevations. Water
depth values exported from CCHE1D then permit the cross section tool to calibrate bed topography
parameters and to interpolate cross sections respecting the centerline radius of curvature. These new
cross sections are once again prepared as in step 8.
150 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

6.3 Calibration

6.3.1 Water surface elevation

The first step in the calibration procedure is to calibrate the Water Surface Elevations (WSE’s). This is
possible if historical WSE’s are available. They were available or measured during this thesis and were
reported in chapter 4.2. The majority of WSE’s were measured during the October 26, 2004 event so it
is used as the basis for simulating WSE’s. For higher discharge values reported in table 4.2, those
observed WSE’s were used qualitatively only. For WSE’s from 1996, the observed WSE was adjusted
if a significant difference in the bottom thalweg exists between 1996 and 2004. The Manning values
were changed and updated in the cross section points file according to the homogenous roughness zones
as shown in chapter 4.4.1 until the differences in simulated and observed WSE’s were acceptable.
Typical Manning values for the are 0.04 for the bed, 0.07 for the banks, and 0.11 for the overbanks. The
results are shown in figure 6.1 and table 6.1.

395

390
elevation (m.a.s.l.)

385

380
Simulated WSE
Thalweg
Observed WSE

375
2500 3500 4500 5500 6500 7500 8500 9500 10500
station (m)

Figure 6.1. Comparison of observed and simulated water surface elevations for the October 26, 2004
high flow event

6.3.2 Bank erosion

The next step in the calibration procedure is to calibrate the bank erosion and according to calibrated
values estimate bank property values for other banks where erosion was not monitored. Initial
simulations to calibrate bank erosion showed the gamma toe shear stress correction function (figure 6.2)
to produce erosion in a less localized manner than that observed in the field. To try to localize erosion to
the very small areas as observed, the gamma function was changed as shown in figure 6.2.
 Calibration 151

The calibration period for bank distance observed

to outlet discharge maximum simulated difference
erosion is from January 12, 2004
(m) point id (m³/s) WSE WSE (m)
through March 29, 2005. Historically 10323 360 50 396.42 396.50 0.08
corrected simulated data are available 10282 361 55 396.61 396.59 -0.02
10104 limni1 57 396.37 396.41 0.04
through February 28, 2005. For the 10053 2.1 57 396.14 396.31 0.17
month of March 2005, input 9804 4.1 60 395.61 395.59 -0.02
9728 limni6 60 395.49 395.41 -0.09
hydrographs are determined according 9635 7.1 60 395.18 395.24 0.06
to the subcatchment area in relation to 9625 7.4 60 395.03 395.23 0.20
9549 8.3 60 395.21 394.96 -0.25
the area of the Ecublens stage recorder 9269 limni-eros11 60 394.31 394.20 -0.11
catchment and the measured 8213 371 55 392.24 391.90 -0.34
7771 390 50 390.49 390.44 -0.05
discharge. When running the 5241 461 50 385.22 385.26 0.04
simulations with sediment transport a 4910 limni Ecublens 60 384.47 384.41 -0.06
3510 15.2 60 381.00 380.89 -0.11
few cross sections significantly 3330 limni14 60 380.65 380.63 -0.02
changed their bottom elevation. To 3275 20.2 60 380.45 380.57 0.12
3005 470 55 379.89 380.20 0.30
avoid problems with changing bed 2963 471 60 380.35 380.27 -0.08
elevations, the bed elevations were 2859 16.4 60 379.93 380.02 0.09
2734 limni-eros17 60 379.74 379.72 -0.02
fixed for the bank erosion calibration.
Table 6.1. Comparison of historical and simulated WSE’s
With incorporation of tension
cracking in the CCHE1D bank failure algorithm, failure widths were typically in the 40-80 cm range,
which was observed in the field, so that specific calibration of failure width was not sought. Also,
without cross sections measured immediately before and after the erosion monitoring period at each
erosion monitoring point, it is impossible to try to simulate the bank failures at a precise time. Thus, it
was decided to compare erosion measurements with simulated bank erosion. An average erosion width
for each measuring vertical during the calibration period was calculated based on the measured values
given in table A.1 of the appendix. This eroded width is then compared with the simulated eroded width.
The simulated eroded width, in terms of bank properties, is dependant on the critical shear stress and
porosity. Assuming the measured bank porosity to be correct, bank erosion calibration was performed
by changing the critical shear stress until simulated erosion was equivalent to the observed erosion.

2.5
Correction factor, C b

1.5

0.5
0 1 2 3 4 5 6 7 8 9 10
Rc/w

Figure 6.2. Final toe shear stress correction factor (in grey) as a function of the ratio of radius of
curvature to width used in the streambank erosion modeling. Test case curve is the dashed line.
152 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

estimated critical The results of the bank

eroded simulated shear sum of erosion calibration are
bank cross erosion difference stress differences %
bend width (m) section id width (m) (cm) (N/m²) (cm) difference shown in table 6.2. The
1.1 -0.01 10080 0.293 -30 5.3 10.2 0.01 results vertical by
1.2 0.06 10080 0.293 -23
vertical (line by line in
1.3 0.01 10074 0.363 -35
2.1 0.12 10053 0.637 -52 table 6.2) are poor as
2.2 0.37 10045 0.723 -35 shown by the difference
2.3 0.40 10045 0.723 -32
2.4 0.46 10045 0.723 -26 in the eroded width and
2.5 0.74 10039 0.759 -2 the simulated eroded
2.6 0.94 10033 0.759 18
3.1 1.26 10022 0.921 34 width, although the
3.2 1.11 10022 0.921 19 average eroded width
3.3 2.63 10016 0.877 175
4.1 0.54 9804 0.651 -12 5.2 9.4 0.02
throughout a bend was
4.2 0.49 9804 0.651 -16 reproduced as shown by
4.3 0.55 9796 0.734 -18
the percentage
5.1 1.64 9788 0.864 78
5.2 0.95 9782 0.803 14 difference (table 6.2).
5.3 0.28 9776 0.657 -37
The important
6.1 0.30 9728 1.474 -118 4.45 -5.7 0.00
6.2 0.21 9728 1.474 -126 differences vertical by
6.3 1.73 9722 1.624 11 vertical are assumed to
6.4 1.76 9716 1.853 -9
6.5 1.47 9716 1.853 -38 come from the varying
6.6 2.39 9709 1.8 59 flow conditions from
6.7 1.63 9709 1.8 -17
6.8 2.30 9701 1.44 86 one vertical to another
6.9 2.22 9701 1.44 78 during one event to
6.91 1.40 9694 1.113 29
6.92 1.20 9686 0.799 40
another. A bank failure a
7.1 2.93 9635 1.902 102 3 -34.6 -0.02 few meters upstream
7.2 4.71 9635 1.902 281
will influence the flow
7.3 2.19 9625 2.843 -66
7.4 1.54 9625 2.843 -131 conditions for the next
7.5 2.39 9625 2.843 -45 vertical downstream. It
7.6 2.55 9620 4.31 -176
8.1 0.48 9557 1.036 -56 5.2 0.9 0.00 can also be assumed that
8.2 0.56 9557 1.036 -48 the bends will not
8.3 1.79 9549 0.745 105
9.1 1.81 9460 1.491 32 4.2 4.8 0.01 continue to progress in
9.2 1.49 9455 1.204 28 the same proportions as
9.3 0.48 9449 1.035 -55
observed during the
10.1 0.17 9348 0.346 -18 4.95 -2.2 -0.01
10.2 0.09 9339 0.716 -63 calibration period as this
10.3 0.09 9339 0.716 -63
would eventually
10.4 2.06 9330 0.658 141
produce an extremely
Table 6.2. Results of the bank erosion calibration. Calibration of the bank jagged bank line.
critical shear stress is based upon groups of measured verticals as shown by
areas outlined in solid black.
 Calibration 153

estimated critical
eroded simulated shear sum of
bank cross erosion difference stress differences %
bend width (m) section id width (m) (cm) (N/m²) (cm) difference
11.1 0.68 9262 0.92 -24 5.25 18.0 0.02
11.2 0.22 9255 1.172 -95
11.3 0.54 9248 1.053 -51
11.4 1.28 9241 1.054 22
11.5 2.02 9234 1.269 75
11.6 2.54 9224 1.389 115
11.7 1.09 9224 1.389 -30
11.8 1.27 9214 1.209 6
12.1 0.18 9176 0.94 -76 3.25 -16.2 -0.01
12.2 1.69 9167 1.418 27
12.3 3.35 9167 1.418 194
12.4 2.32 9158 3.206 -88
12.5 3.35 9158 3.206 15
12.6 3.37 9150 4.239 -87
13.1 0.17 9080 0.218 -5 3.75 -8.9 -0.03
13.2 0.69 9073 0.573 12
13.3 0.95 9073 0.573 38
13.4 0.74 9073 0.573 17
13.5 0.19 9067 0.89 -71
14.1 0.04 3357 0.113 -7 6.4 -6.2 -0.12
14.2 0.04 3348 0.153 -11
14.3 0.20 3339 0.152 4
14.4 0.23 3339 0.152 8
15.1 1.12 3519 1.095 3 4.35 17.6 0.03
15.2 0.75 3510 1.066 -32
15.3 1.85 3500 0.892 96
15.4 0.76 3492 0.883 -13
15.5 1.12 3476 0.856 27
15.6 0.32 3468 0.944 -63
16.1 0.04 2882 0.887 -85 4.8 -8.0 -0.02
16.2 0.93 2870 0.86 7
16.3 0.85 2864 0.747 11
16.4 1.14 2859 0.602 54
16.5 0.55 2853 0.499 5
17.1 0.11 2728 2.362 -225 3.75 9.8 0.01
17.2 0.46 2716 2.1 -164
17.3 1.15 2710 2.198 -105
17.4 2.71 2703 2.541 17
17.5 4.85 2693 2.852 199
17.6 4.79 2688 1.914 288
18.1 0.31 2670 0.149 16 4.3 -3.2 -0.02
18.2 0.03 2670 0.149 -12
18.3 0.40 2664 0.171 23
18.4 0.27 2664 0.171 10
18.5 0.12 2658 0.197 -8
18.6 0.18 2646 0.311 -13
18.7 0.10 2640 0.282 -18
19.1 0.02 3447 0.16 -14 5.7 0.0 0.00
19.2 0.15 3440 0.162 -1
19.3 0.39 3433 0.166 22
19.4 0.06 3420 0.131 -7
20.1 0.36 3284 0.206 15 6.05 0.3 0.01
20.2 -0.04 3275 0.178 -22
20.3 0.21 3266 0.141 7

Table 6.2 (cont.). Results of the bank erosion calibration. Calibration of the bank critical shear stress is
based upon groups of measured verticals as shown by areas outlined in solid black.
154 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

The calibration of critical shear stress is based on the cumulative bend erosion. It is worthwhile though
to consider the behavior of the model event by event. As explained above, the cumulative erosion results
for each bank profile are often poor so that event by event the results will be expected to be even poorer.
For each bank profile and for each event, the three or four measurements (top, mid top, mid bottom, and
bottom) were averaged to estimate a mean bank profile erosion. These mean values are only estimates
of the event bank profile erosion. This is not only due to the fact that the point measurements are limited,
but it is also largely due to the fact that the bank toe is not fully represented in the point measurements.
These mean values were compared to the simulated mean bank profile erosion event by event. The
absolute value of the difference of simulated and measured average profile erosion were summed over
the 100 profiles. This sum of the differences is compared to the total measured average profile erosion
to calculate a percentage difference. The results are shown graphically in figure 6.3. Firstly, the sums of
all of the measured and simulated erosion at the measured profile locations are given, and secondly the
percentage difference is shown with the right vertical axis. In terms of the global erosion, the model
respects relatively well the measured quantity although the graph suggests slightly too much erosion for
small discharge events and not enough for higher events. The percentage difference results are very poor.
This is especially true for the lower discharge/volume events. Without putting the exact starting bank
configuration into the model (the cross sections are from summer 2004 while the simulations start in
January 2004) and without taking into account the effect of bank soil and tree falls on the 3-D flow field,
it is inevitable to have poor results profile by profile. This should not be considered though to invalidate
the model, but rather qualify the results of the model as indicative of global bend and reach erosion rather
than specific location erosion.

As seen in figure 6.3, the lower discharge/volume event erosion appears to be overestimated and the
higher discharge/volume events underestimated. This behavior is most likely a result of the erodibility
coefficient equation used in the model. The CCHE1D model equation (c.f. equation 2.7), based on the
work of Arulanandan et al. (1980), is compared to the erodibility coefficient equation in the work of
– 0.5
Simon and Thomas (2002) (figure 6.4) in which the erodibility coefficient, k = 0.1τ c . It can be seen
that the low calibrated shear stress values result from the Arulanandan et al. (1980) formula. Using the
Simon and Thomas (2002) formula would produce a higher calibrated critical shear stress that would
permit proportionally more erosion to occur in the higher discharge/volume events. This result confirms
that the equation of Simon and Thomas (2002) is an improvement over the Arulanandan et al. (1980)
formula, although in the article by Hanson and Simon (2001), the equation is different by a factor of 2.
Such a difference shows the need to further research the erodibility coefficient equation.
 Calibration 155

Measuring date 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05
Erosion sim. (cm) 5227 147 359 380 360 2270 373 765 860
Erosion meas. (cm) 6082 478 342 73 431 2539 119 445 244
% difference by profile 74 548 188 129 240 114 180 122 101

7000.0 700
Erosion meas.
6000.0 Erosion sim. 600

% difference between simulated and

% difference
sum of profile erosion (cm)

observed profile erosion

5000.0 500

4000.0 400

3000.0 300

2000.0 200

1000.0 100

0.0 0
0.0E+00 5.0E+08 1.0E+09 1.5E+09 2.0E+09
Qmax*Volume (m 6/s)

Figure 6.3. Comparison of total measured and simulated erosion as well as the average absolute
percentage difference for measured and simulated profile erosion

0.7
Arulanandan et al. (1980)
Erodibility coefficient, k (cm N s )
-1 -1

0.6 Simon and Thomas (2002)

3

0.5

0.4

0.3

0.2

0.1

0
0 20 40 60 80 100
Critical shear stress (Pa)

Figure 6.4. Comparison of two equations for erodibility as a function of critical shear stress.
156 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

6.3.3 Critical shear stress estimation

With the calibrated bank critical shear stresses, it is now possible to investigate if there are relationships
between them and the various soil properties described in chapter 4.6 in view of developing an equation
to estimate critical shear stress for non-calibrated banks. Firstly, correlation coefficients are explored to
gain an appreciation of the correlations between the variables. Besides the variables given in table 4.9,
characteristic grain sizes (D10, D30, D50, D60, D90, Dm), derived values of %silt/%clay, % cohesive,
% large silt-fine sand (0.02-0.316 mm), % fine sand (0.05-0.316 mm), fraction bare observed, fraction
bare laser, and vegetation height laser values were also explored for correlation. The variable, fraction
bare observed, is a subjective judgement of the fraction of the bank observed to have no vegetation or
root coverage, while fraction bare laser is derived from the LIDAR topographic data. It is assumed that
the ratio of terrain LIDAR points to surface LIDAR points should be indicative of the vegetatative cover.
This ratio is expressed in the variable fraction bare laser. The difference between surface point elevations
and terrain point elevations can be considered to be representative of the vegetation height and is given
by the variable vegetation height laser. Little correlation exists between these variables and the calibrated
critical shear stress values. Only variables that have a correlation coefficient greater than the absolute
value of 0.4 with critical shear stress are shown in the correlation matrix in table 6.3.

τc n γ Cu Ip Cud Ccd fract.

bare
obs
Critical shear stress, τc 1.00
Porosity, n 0.50 1.00
Natural unit weight, γ -0.43 -0.84 1.00
Shear vane, Cu -0.43 -0.32 0.41 1.00
Plasticity Index, Ip -0.44 -0.16 0.24 0.61 1.00
Uniformity Coef, Cud -0.47 -0.24 0.03 0.32 0.72 1.00
Curvature Coef., Ccd -0.45 -0.01 -0.06 0.40 0.86 0.86 1.00
fract. bare observed -0.74 -0.49 0.50 0.71 0.61 0.40 0.44 1.00
Table 6.3. Correlation matrix of critical shear stress and soil properties

The poor correlation coefficients confirm the suggestion of other authors (Briaud et al., 2001, and
Hanson and Simon, 2002 for example) that critical shear stress is poorly correlated with other soil
variables and must be measured. This does not preclude an attempt at a multiple linear regression,
though, to see if a combination of variables can explain the calibrated shear stresses for a given study
reach. Stepwise regression was performed in Matlab® based on the same variables explored above in the
correlation analysis. A multiple regression using just two variables, % fine sand - large silt, and fraction
bare observed explained 75% of the variation (table 6.4). Although, this regression can not be used
outside of the reach from which it was obtained, it does indicate the importance of these two variables.
 Calibration 157

critical % fine sand

shear and large fract. bare
bend stress silt observed
1-3 5.3 63.09 0.25 RESIDUAL OUTPUT
4-5 5.2 48.32 0.25
6 4.45 64.22 0.40 Bend Predicted Residuals Stand. Residuals
7 3 75.27 0.50 1-3 5.06 0.24 0.54
8 5.2 54.84 0.50 4-5 5.38 -0.18 -0.42
10 4.95 65.71 0.25 6 4.40 0.05 0.10
11 5.25 59.84 0.30 7 3.74 -0.74 -1.67
12 3.25 47.38 0.80 8 4.19 1.01 2.27
13 3.75 56.65 0.40 10 5.00 -0.05 -0.11
15 4.35 63.05 0.50 11 4.92 0.33 0.74
16 4.8 63.50 0.30 12 3.10 0.15 0.34
17 3.75 62.80 0.50 13 4.57 -0.82 -1.85
18 4.3 1.79 0.75 15 4.01 0.34 0.76
19 5.7 14.37 0.30 16 4.84 -0.04 -0.09
20 6.05 66.59 0.05 17 4.02 -0.27 -0.60
18 4.32 -0.02 -0.04
SUMMARY OUTPUT 19 5.93 -0.23 -0.51
20 5.82 0.23 0.52
Regression Statistics
Multiple R 0.87
R Square 0.75
Adjusted R Square 0.71
Standard Error 0.48
Observations 15

ANOVA
df SS MS F p value
Regression 2 8.34 4.17 18.14 0.0002356
Residual 12 2.76 0.23
Total 14 11.10

Coeff. Stand. Error t Stat P-value Lower 95% Upper 95%

Intercept 7.500 0.563 13.312 0.0000 6.273 8.728
% fine sand - large silt -0.022 0.007 -3.183 0.0079 -0.037 -0.007
fract. bare obs -4.192 0.705 -5.944 0.0001 -5.729 -2.656

Table 6.4. Results of a multiple linear regression to explain calibrated critical shear stress

Without numerous particle size distributions and observations of vegetation density on bank faces in
areas of the study reach not monitored for erosion, it was decided to attribute a mean calibrated critical
shear stress value of 4.6 Pa to all other banks in the study reach. This value was then increased or
decreased based upon the location of erosion scars (see figure 4.1) and upon the approximate river
centerline movement from 1964 (from the 1996 Venoge study, Lance and Consuegra, 1998).

6.3.4 Sediment transport

Sediment transport can only be considered qualitatively because there are not enough cross sections
spanning a long enough period to quantify significant bed elevation changes in the Venoge River and no
bedload or suspended sediment measurements were taken.

Sediment transport including bank erosion was simulated for the historical period from January 1979
through March 2005. The areas from km 4.950 to km 5.527 and km 5.819 to km 6.102 seemed to have
excessive bed erosion. It seems to be excessive compared to a lack of bed lowering indicators based on
158 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

field observations. Bed lowering occurred upstream from the damaged "Abbaye" weir in the zone km
7.335-7.884 during 2001 and 2002, but this zone was supposedly stabilized by a temporary reparation to
the weir in 2002.

Bed sediment size was increased in the excessive bed lowering areas until a relative stability was ensured
for the 1979-2005 simulation. The change in the D50 seems excessive though, as the values from km
4.950 to km 5.527 were changed from approximately 1 cm to 4.5 cm, and from 1.1 to 4 cm for km 5.819
to km 6.102, and from 1.3 to 7 cm for km 7.335 to km 7.884. As the maximum sampled D50 for the last
20 kilometers of the Venoge River is 4 cm (see figure 4.33), there seems to be data problems in these
areas.

Besides the adjustments to bed sediment sizes, it was also noticed that significant deposition was
occurring near the upstream boundary condition. This was corrected by multiplying the sediment
transport curves of figure 4.38 by 0.75 such that a reasonable balance between deposition near the inlet
and erosion following the depositions was obtained (figure 6.5).

6.4 Streambank erosion hazard simulations and danger mapping

6.4.1 Streambank erosion danger map

With the calibration of the Venoge River study reach completed, it is now possible to simulate bank
erosion and failures with the 300 year scenarios shown in chapter 5.8. The objective of this simulation
is to be able to analyze the empirical frequency of the bank failures to calculate the probability of a given
failure width. As explained in chapter 2.4.2 concerning the changes made to the CCHE1D model for this
thesis, to calculate the probability of failure for a current bank, flow conditions and bank geometry
should not change during the simulations. To do this, bank geometry is not changed after a bank failure,
but new bank failures are not allowed to occur until nearly all basal sediments in the virtual reservoir
have been emptied.

Besides the adjustments made to the sediment transport as shown in the previous section, 6.3.4, there are
still important bed fluctuations that are occurring during the 25 year period. These fluctuations will
change the flow conditions and bank geometry situations. Thus, it is decided for the 300 year simulations
to fix the river bed. This option is available in the CCHE1D sediment transport calculations.

To practically handle the 300 year simulations with the computer resources at hand, the 300 years are
divided into 100 year simulations. This means that the initial bank conditions are repeated 3 times rather
than once. This is deemed a reasonable compromise for banks that are initially stable, but become
unstable during the course of the simulations due to toe erosion. If the simulation period is too long, the
probability of failure of such banks will be overestimated. If the simulation period is too short, though,
the bank will never fail. Thus, the length of the simulations should be at least as long as the period of
validity designated for the maps, in other words, the length of time before the maps will be renewed. In
the case of the Venoge River, it would be suggested to renew the maps every 10 years or so. Of course,
 Streambank erosion hazard simulations and danger mapping 159

erosion in active meanders will most likely change during the 10 years as the erosion attack will migrate
downstream, but the bank failure probabilities can be roughly assumed to follow the bend migration path
during this period. Due to some poor cross section interpolations, it is preferred to go beyond this 10 year
simulation length. How much beyond this 10 year simulation length is reasonable is unknown. For
convenience, a 100 year simulation length was chosen. This simulation time takes roughly 13 hours on
a 2.8 Ghz desktop personal computer. This is convenient for running simulations during the evening on
several workstations and recovering the results the next work day. Thus, the three 300 year simulations
were divided among 9 workstations.

final thalweg
395
initial thalweg

393
elevation (m.a.s.l.)

391

389

387

385

383
6500 7500 8500 9500 10500
station (m)

386
final thalweg
initial thalweg
384

382
elevation (m.a.s.l.)

380

378

376

374
2500 3500 4500 5500 6500
station (m)

Figure 6.5. Bed elevation changes simulated with the 26 year discharge record
160 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

The bank failure results for the simulations are output in a text file. The three 100 year bank failure text
files are combined to form the 300 year bank failures. These bank failures are read by a program which
ranks the failures and calculates a frequency according to the number of events. This event frequency is
transformed into an annual frequency according to the Langbein-Takeuchi relation (Meylan and Musy,
1999):
F A ( x ) = exp { –λ [ 1 – F E ( x ) ] } Eq. 6.1

where FA(x) is the annual cumulative probability, λ is the number of events divided by the number of
years, and FE(x) is the event cumulative probability. Examples of bank failures and their probabilities
are given in figure 6.6. The top left graph shows a bank that fails infrequently while the bottom right
graph shows very frequent failures. The multiple steps in the bottom right graph are due to multiple bank
failures in a single event. For banks to have a more varied bank failure width response, the CCHE1D
program would need to include pore-water pressure modeling. This would force the bank geometry to
steepen when the bank has enough matric suction to resist bank failure causing a different failure
geometry the next time the bank is saturated enough to fail. Varied bank failure response is also most
likely due to heterogeneities in the bank which could be incorporated in the model if those
heterogeneities are quantified (with probability distributions of cohesion and friction angle in the bank
for example).

3.5000 3.5000

3.0000 3.0000

2.5000 2.5000
failure (m)

failure (m)

2.0000 2.0000

1.5000 30 m³/s, 13 m³/s, 30 m³/s 1.5000

1.0000 1.0000

0.5000 0.5000

0.0000 0.0000
0.001000 0.010000 0.100000 1.000000 0.001000 0.010000 0.100000 1.000000

annual exceedance probability annual exceedance probability

3.5000 3.5000

3.0000 3.0000

2.5000 2.5000
failure (m)

failure (m)

2.0000 2.0000

1.5000 1.5000

1.0000 1.0000

0.5000 0.5000

0.0000 0.0000
0.001000 0.010000 0.100000 1.000000 0.001000 0.010000 0.100000 1.000000

annual exceedance probability annual exceedance probability

Figure 6.6. Examples of bank failure widths and probabilities for the following cross sections: km
2.693 right (top left), km 10.080 left (top right), km 10.016 (bot. left), and km 2.949 (bot. right)

With the bank failure widths and their probabilities quantified, it is possible to quantify the danger level
of the streambank erosion hazard according to the recommendations of the Federal Office for Water and
Geology (figure 1.3). For example, the bank represented in the top left graph of figure 6.6 would have a
 Streambank erosion hazard simulations and danger mapping 161

weak danger level because its value is just less than 0.5 meters and a failure can be expected only every
one hundred years on average. It can also be noticed in this graph that the events causing the failures are
all below the one year return period maximum annual discharge showing that bank failures often are not
correlated with the maximum event discharge. The top right and bottom left graphs of figure 6.6 show
medium streambank erosion dangers, while the bottom right graph shows a high streambank erosion
danger. The danger levels are calculated for each bank and are mapped (figure 6.7 and appendix 7).
There are very few weak dangers on the map and they are hardly visible because of their size. This is due
to the stringent criteria of all 30 year return period failures being classified as a medium or high danger.
It seems reasonable though that nearly all of the bank failures should be classified as medium or high
dangers because of the dangerousness of a failure.

It should be noticed that danger levels have been calculated where bank protection exists. This was
possible in the simulation with a fixed bed calculation. With a non-fixed bed simulation, taking out all
of the bank protection would have caused too much sediment supply causing significant bed changes.
Danger levels at the bank protection in this map are not the final danger levels. These danger levels
would have to be combined with the probability of failure of the bank protection works to determine a
final danger level.

Besides the erosion danger level, the mean annual erosion is also represented in the map according to the
length of the bar. This information is provided to qualify the danger level. It is meant to show where the
erosion is active. It is not meant to show where the river will progress in the future. For actively eroding
banks, the cross section and planform geometry will progressively change, changing in turn the mean
annual erosion widths. Besides, the bars showing the danger level, a buffer is placed around the river
channel outline showing the possible river extension during an extreme flood. This value is set here as
1.5 times the maximum simulated bank failure of the river. The maximum extreme failure of 10.53
meters for a cross section is at km 9.614 so that the extreme danger buffer is set to 16 meters around the
existing bank tops. It seems reasonable to imagine, that obstructions in the river could deflect the river
flow to cause excessive shearing velocities beyond that which was simulated, thus permitting the use of
a multiplication factor. The average width of the Venoge River in the study reach is approximately 24
meters. It would be prudent to use this value as the security distance for a management strategy such as
the one shown in figure 1.4.
162 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

Streambank Erosion
Danger Level

Low

Medium

High

Length of danger bar shows

mean annual erosion
exaggerated by 20 times

Extreme danger limit

Venoge River
2004 centerline

Bank protection work

8 Erosion measuring profile

11 and bend number
33 22
0 100 200

44 meters
55
77

66

88

99

12
12
10
10

11
11
13
13

Intensity
(Failure 30 yr 100 yr 300 yr extreme
width) Streambank Erosion
strong Danger Level

2m high

medium
medium
low
0.5 m
residual
weak

high medium low

Occurrence Probability (Frequency)

Figure 6.7. Streambank erosion danger map based on CCHE1D simulations and the FOWG danger
matrix (bottom left inset). Bank protection failure probability is not taken into account.
 Streambank erosion hazard simulations and danger mapping 163

6.4.2 Streambank erosion hazard results

4.5 4

4 y = 0.0001x2 + 0.0061x y = 0.0011x 2 - 0.0059x

3.5
R2 = 0.6785 R2 = 0.681
3.5
3
Average bank erosion (m)

Average bank erosion (m)

3
2.5
2.5
2
2
1.5
1.5
1
1

0.5 0.5

0 0
0 20 40 60 80 100 120 140 160 0 10 20 30 40 50 60 70
Qmax (m³/s) Shear Max (N/m²)

4 4
y = 3E-18x 2 + 4E-08x
R2 = 0.9807
3.5 3.5

3 3
Average bank erosion (m)

Average bank erosion (m)

y = 0.0046x 2 - 0.0392x
2.5 2.5
R2 = 0.3179

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 5 10 15 20 25 30 0.0E+00 2.0E+07 4.0E+07 6.0E+07 8.0E+07 1.0E+08 1.2E+08
Qmean (m³/s) Flood volume (m³)

4 8
y = 5E-10x
y = -2E-19x2 + 2E-09x R2 = 0.6181
3.5 7
R2 = 0.9497

3 6
Average bank erosion (m)

Average bank erosion (m)

2.5 5

2 4

1.5 3

1 2

0.5 1

0 0
0.0E+00 5.0E+08 1.0E+09 1.5E+09 2.0E+09 0 2E+09 4E+09 6E+09 8E+09 1E+10 1.2E+10 1.4E+10 1.6E+10

Qmean*Volume (m 6/s) Qmax*Volume (m 6/s)

Figure 6.8. Results of the streambank erosion simulations for the right bank at kilometer 2.693

The correlation between measured streambank erosion and flood variables was shown in figure 4.12.
With the streambank erosion simulated for a period of 300 years, it is possible to examine the model
behavior to determine if it mimics the behavior shown in the graphs of figure 4.12. Maximum event
discharge (Qmax), event flood volume (event start and end at 9 m³/s), mean event discharge (Qmean),
maximum shear stress, Qmean*volume, and Qmax*volume are compared against the average bank
erosion. Graphs for the right bank of node 1123 (km 2.693) serve as an example in figure 6.8. It can be
seen in these graphs that the flood volume is the best indicator of the bank erosion that will occur while
the indicator Qmax*volume is not linear as seen in figure 4.12. The rare flood values of Qmax*volume
are much greater though in figure 6.8 so it is difficult to conclude if the difference in behavior is due to
the very few observations in figure 4.12 or due to the model behavior. Figure 6.3 would seem to indicate
164 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

though that the model does not erode enough at high shear stresses so that the erosion shown in figure
6.8 might be underestimated for the high discharge/volume events. The CCHE1D model should be tested
with the Simon and Thomas (2002) erodibility coefficient and erosion rate formulas to see what
difference it would make on the erosion values for high discharge/volume events.

6.5 Streambank erosion simulation tests

It can be noticed in the danger map (figure 6.7) that there are a lot of straight reaches with medium danger
levels. This result is most likely due to shear stresses that have not been diminished (or critical shear
stresses that are not high enough) due to the significant vegetation that lines these straight sections.
Thompson et al. (2004) have used shear stress partitioning theory to estimate the shear stress exerted on
the vegetation and non-vegetation surfaces. They have shown that with significant vegetation, the
particle shear stress is significantly reduced (down to 13% in their experiments). With this reasoning in
mind, it would seem appropriate to reduce the bottom end of the shear stress correction function (figure
6.2) to 0.5 rather than 1.0. This should significantly reduce the simulated erosion in straight river reaches.

To test this hypothesis, a test case is run. The test case consists of using a trapezoidal channel with the
average dimensions and roughness for the study reach. This same cross section is used for every node.
The average slope of the study reach is used and does not vary throughout the reach. The bank and bed
properties are also homogeneous throughout the test case study reach. With this test, variability in
erosion due to hydraulic, soil, and vegetation conditions is removed. The test case is run for the
calibration period of January 2004 through March 2005.

The result of the test case is shown visually for the upstream detailed study reach (figure 6.9). It can be
seen that with the shear stress correction of 1 in straight sections that erosion often occurs on both sides
of the stream compared to none or one-sided erosion when the correction is 0.5. The Venoge channel
width is stable which would suggest the use of the correction factor for straight sections of 0.5 compared
to 1 for vegetated banks.

A second test is performed to verify the necessity of smoothing the channel centerline (c.f. chapter 2.4.2).
This test case uses the same hypothetical test reach as in the first case. Four cases are tested: a smoothed
channel with no upstream weighting (α = 0.99), a smoothed channel with upstream weighting as for the
Venoge (α = 0.05), an unsmoothed channel with no upstream weighting (α = 0.99), and an unsmoothed
channel with typical Venoge upstream weighting (α = 0.05). It can be seen in figure 6.10 that for an
accurately digitized centerline, smoothing is unnecessary when the upstream weighting is significant,
but when no upstream weighting is used, the smoothing procedure is necessary.
 Streambank erosion simulation tests 165

Figure 6.9. Comparison of erosion results when the shear stress correction function does not reduce
the average shear stress in straight sections (left) or reduces it to half (right). The simulated eroded
distances are shown by the lines perpendicular to the channel centerline.
166 Geofluvial Streambank Erosion Hazard Modeling and Danger Mapping

smoothed, α = 0.05 unsmoothed, α = 0.05

smoothed, α = 0.99 unsmoothed, α = 0.99

Figure 6.10. Test case erosion to determine the importance of centerline smoothing. The simulated
eroded distances are shown by the lines perpendicular to the channel centerline.
 Chapter 7

Conclusions and Future

Research Recommendations

7.1 Conclusions
The streambank erosion hazard can be viewed from two aspects: land-use management and flood event
management, which can be restated in the questions: "Where will the river be in 30 or 60 years?" and "How
much will the stream erode its bank during a 100 year flood?" This research investigated the second
question. Knowing the event streambank erosion behavior can also be helpful for land-use management
by providing a streambank intervention distance that dictates when streambanks must be protected to
avoid damage to vulnerable objects.

This research proposes a methodology for streambank erosion hazard mapping. It must begin with
geomorphological mapping. This step is paramount for understanding the sediment sources of the river
reach in question. In this step, the fluvial geomorphologist reads the field signs that indicate the current
erosional and depositional behavior of the river.

Geomorphological mapping can be followed by streambank erosion hazard modeling and mapping. The
model CCHE1D was adapted in this research to be able to calculate the width of a streambank failure and
its probability, as needed for the calculation of the streambank erosion danger according to the directives
of the Swiss Federal Office for Water and Geology. A bank toe shear stress correction function was added
to the CCHE1D model to correct the 1-D toe shear stresses based on channel planform curvature, as well
as an upstream weighting function to permit the phase lag of maximum bend erosion compared to
maximum bend curvature. These shear stress correction functions properly placed the maximum erosion
at the outside of meander bends, but permitted too much erosion in straight river reaches.

167
168 Conclusions and Future Research Recommendations

Tension cracking was added to the bank failure algorithm of the CCHE1D model. Tension cracking
significantly affects the failure width of slab failures so it must be taken into account for this type of
failure mode. It needs to be further researched so that algorithms can be improved to better predict
tension crack depth and placement.

For the special case of determining bank failure probability, the CCHE1D model was modified so that
bank geometry does not change after a bank failure. Bank failure sediments are placed in the virtual basal
sediment reservoir of CCHE1D and a new bank failure can not occur until the basal sediments have been
eroded away. This strategy allows bank geometry and flow conditions to stay the same throughout
simulations so that a probability of a given bank failure width can be calculated.

The Lower Venoge River fieldwork showed the difficulties associated with the input data necessary for
streambank erosion hazard modeling. Monitoring erosion allows for an understanding of the erosion and
failure processes to be gained. This step is absolutely necessary so that the modeler can select a
streambank erosion hazard model that includes the observed processes. For the monitoring to be more
practical, it can be restricted to a few banks that are representative of the different bank types found on
the river.

Estimation of the mean bank critical shear stress is extremely important in the CCHE1D modeling.
Attempts have been made in this research and by other researchers to correlate critical shear stress with
other measured soil properties, but no equations have proven to be globally satisfactory. Critical shear
stress must be measured by an apparatus such as the submerged jet device (Hanson and Cook, 2004) or
calibrated as done in this research. A multiple linear regression equation involving the percentage of fine
sand - large silt and the fraction of the non-vegetated bank surface explained 75% of the variation for the
monitored bends. Such a regression equation can be used for the reach for which it was developed.

Calibration of critical shear stress was done by comparing simulated cumulative bend erosion (9 events)
to that which was measured. If the channel topography is known precisely at two different times, with
two or more sets of orthophotos or high resolution (1-2 m) DTM, then these remote sensing data can
replace erosion measurements.

The topographic needs for streambank erosion are enormous. Streambank erosion is usually very
localized so that the computational network must be refined. This requires an enormous amount of cross
sections which are impractical to survey. Thus, remotely sensed topographic data must be used or cross
sections must be interpolated respecting channel curvature. Standard LIDAR data, such as the national
Swisstopo data used in this research, is insufficient for describing the topography of vegetated channels
and is not water penetrating. Thus, a cross section interpolation tool was developed to interpolate cross
sections with respect to curvature. This tool was used to interpolate 78 measured cross sections to
provide cross sections for the 1149 computational nodes. Although this tool needs to be used in various
channel contexts before giving firm suggestions on its use, it would appear in a homogeneous zone
(similar bank and bed conditions, and river slope) that one cross section for every fourth bend pool would
be approximately sufficient. The bank toe definition point is very important in the interpolation program
and its influence on the interpolation and on the subsequent calculations should be further investigated.
 Conclusions 169

A qualification is warranted, though, concerning the topographic needs. For slowly migrating channels,
the initial topographic description will be important. Without a correct topographic description, the bank
might fail too soon or remain stable too long during the simulation. For fast migrating channels, a precise
description is unnecessary if the simulation time is long enough. Such banks will have the opportunity
to fail many times throughout the simulation.

Depending on the type of geotechnical failures observed, more or less attention should be given to
collecting data on the cohesion and friction angle of the banks. For the Venoge River, soil fall and
cantilever failures are the dominant process so that specific calibration of the cohesion and friction angle
parameters was not undertaken. In a case where planar and curved failures are wider, cohesion and
friction angle should be measured by laboratory analysis or by an in situ device such as the Iowa
Borehole Shear Tester.

Bed and bank particle size data are required by the CCHE1D model. These data can be rough if the
streambank erosion simulations are conducted with a fixed bed.

Flood volume was shown to be a very important indicator of reach erosion. This shows the necessity of
streambank erosion modeling to be unsteady. High flow hydrographs are necessary and usually need to
be produced by continuous hydrologic modeling. The spatial interpolation of data and the semi-
distributed approach used in the hydrologic modeling in this thesis serve as an example how to generate
spatially coherent input data for the streambank erosion modeling. Temperature based potential
evapotranspiration was shown to be sufficient for the hydrological modeling compared to more data
intensive formulas.

Three 300 year simulations were conducted with the CCHE1D model. Bank failures were output from
the model for the 1149 cross sections covering the 8 kilometers of study reach. The probability of a given
bank failure width was calculated based on its empirical frequency. For a given bank and for a given
probability, the median bank failure width of the three 300 year scenarios was used as the final bank
failure width. Danger levels were calculated and mapped for each bank based on the directives of the
Swiss Federal Office for Water and Geology.

Bank failure width for an extreme event was also calculated. The maximum simulated for the study reach
was 10.5 meters. This failure width was multiplied by a safety factor for the extreme failure width of 16
meters to be mapped on the danger map. It is assumed that such a multiplication should be made due to
extreme shear stress conditions that might occur during an extreme flood. Research should be done to
better determine this multiplication factor or the method by which a simulation should be conducted to
best reproduce extreme shear stress conditions.

Erosion hazards in straight reaches of the study reach are too high. This would suggest that the toe shear
stress correction should be less than one. This is a reasonable assumption because of vegetation which
most likely reduces particle shear stress in these zones.

The streambank erosion hazard mapping methodology succeeded in producing the streambank erosion
danger map. The methodology can be used for other alluvial, single threaded, subcritical flow rivers.
Curved failure surfaces are not implemented in CCHE1D, so reaches with such geotechnical problems
should not be modeled with it. The planar failure algorithm with tension cracking might not work
170 Conclusions and Future Research Recommendations

appropriately for steep, high (> 8 m) banks. Also in such a case, unsaturated conditions might begin to
have important stability effects. Unsaturated conditions are not taken into account by the CCHE1D bank
failure algorithm.

When large failure width (>10 m) geotechnical problems are present, the model could be used in
combination with a geotechnical model. CCHE1D can be used to estimate the toe erosion, while the
geotechnical failure should be explored by an appropriate geotechnical model.

It must be remembered that CCHE1D is a 1D model. Vegetation can only be taken into account by
adapting the critical shear stress and bank failure parameters. The effect of bank soil and vegetation falls
into the river, which in turn affect the bank shear stresses, can not be taken into account. This was quite
evident in the very poor correlation between measured and observed event bank profile erosion. Average
bend erosion was calibrated so that the average bend migration is expected to be realistic. Observed and
measured reach erosion were relatively well correlated. The tendency to overestimate lower discharge/
volume events and underestimate higher discharge/volume events should be further investigated.
Changing the erodibility coefficient equation might correct this problem.

It can be concluded from this research that streambank erosion hazard mapping is possible with the
results of geofluvial 1-D models. The amount of input data for this type of modeling is prohibitive though
and will restrict its use to cases in which the streambank erosion problem has significant impacts.
Research must be conducted to reduce data acquisition costs for this type of modeling to be used more
regularly.

7.2 Future research recommendations

The following recommendations can be given for improving streambank erosion modeling in the
CCHE1D model:
• Confining and pore pressure effects should be included in the bank failure algorithm to better
reproduce natural variability in bank failure width. Bank failures should not be restricted to failing
through the bank toe.
• Rather than using a virtual basal sediment tank for failed material, it would be better to slough the
material onto the toe area of the bank. This would entail improving cross section deposition.
• Channel migration modeling could be explored with CCHE1D, but the cross section deposition
would have to be improved. It is debatable whether a geofluvial 1-D model should be used for
channel migration modeling. It is most likely preferable to concentrate on including bank erosion
and bank failure algorithms in a 2-D sediment transport model.

The following recommendations can be given for completing the streambank erosion hazard modeling
research on the Lower Venoge River:
• Critical shear stress must be measured for the bends monitored for erosion. The in situ jet apparatus
described by Hanson and Cook (2004) could be used for this. Knowing these values will allow the
improvement and validation of the shear stress correction function used in this thesis. In this respect,
3-D hydraulic modeling could also be conducted to better understand the shear stress being applied
to the toe and its phase lag with maximum bend curvature, but it would be difficult to correctly
 Future research recommendations 171

quantify the actual particle shear stress due to vegetation effects.

• Stochastically generated meteorological scenarios should be generated for the Venoge catchment to
produce more discharge scenarios. The use of many discharge scenarios will give a higher
confidence in the calculated failure width probabilities.
• For more precision in failure width, more needs to be known about the soil fall and slab failures of
the Venoge River. This entails better knowledge of the bank topography just before and after the
failure. New monitoring tools must be researched and developed to do this.
• Event based modeling should be compared against continuous modeling once enough discharge
scenarios have been modeled. In the fluvial erosion dominated case of the Lower Venoge River,
event modeling will probably be sufficient. As in any event modeling, the proper initial conditions
must be chosen though, and this is especially true for the bank topography of slowly eroding banks.
Thus, research would need to be done to determine how to properly choose the initial conditions of
the banks.

Other diverse recommendations are:

• A sensitivity analysis should be done to quantify the influence of the CCHE1D shear stress
correction and bank parameters.
• Data acquisition methods must continually be pushed to be less costly and time consuming.
Important advances have been made with LIDAR topographic data, although this data is usually
insufficient for accurately describing the channel topography unless special measures are taken
(water penetrating LIDAR and sufficient resolution for vegetated banks). Water penetrating LIDAR
should be used in a stream such as the Venoge River to quantify its potential in that setting.
Otherwise, other rapid land-based laser techniques should be tested in combination with standard
LIDAR data (such as that used in this thesis) for accurately describing the channel topography.
Research also must be continued on maximizing the potential of the data on hand. One aspect of this
was started in this research by improving the channel description by proper interpolation methods.
The streamwise interpolation of channel points presented in this research should be improved with
more sophisticated interpolation algorithms. More rapid critical shear stress, cohesion, and friction
angle testing also need to be developed.
• The channel migration modeling of the 1996 HYDRAM Venoge study should be further researched.
This model is practical because it does not require hydraulic modeling. A relation of event erosion
compared to channel migration should be possible to find by taking into account bank erosion theory
and the hydrologic events that produced the channel migration.
172 Conclusions and Future Research Recommendations
 References

Abernethy, B. and I. D. Rutherfurd (1998). "Where along a river's length will vegetation most effectively
stabilise stream banks?" Geomorphology 23(1): 55-75.
Abramson, L. W., T. S. Lee, et al. (2002). Slope stability and stabilization methods. New York, Wiley.
Allen, R. G. (1998). Crop evapotranspiration guidelines for computing crop water requirements. Rome,
Fao.
Andreassian, V., C. Perrin, et al. (2001). "Impact of imperfect rainfall knowledge on the efficiency and the
parameters of watershed models." Journal of Hydrology 250(1-4): 206-223.
Arulanandan, K., E. Gillogley, et al. (1980). Development of a quantitative method to predict critical shear
stress and rate of erosion of natural undisturbed cohesive soils. Vicksburg, Mississippi, U.S. Army
Engineers Waterways Experiment Station.
ASCE (1996). Hydrology handbook. New York, NY, American Society of Civil Engineers.
ASCE Task Committee (1998a). "River width adjustment. I: Processes and mechanisms." Journal of
Hydraulic Engineering-Asce 124(9): 881-902.
ASCE Task Committee (1998b). "River width adjustment. II: modeling." Journal of Hydraulic
Engineering-Asce 124(9): 903-917.
Beven, K. J. (2001). Rainfall-runoff modelling the primer. New York, John Wiley.
Blackmar, W. (1995). Documentation of hydrologic, geomorphic, and sediment transport movements on
the Goodwin Creek Experimental Watershed, Northern Mississippi, for the period 1982-1993.
Research Report #3. Oxford, MS, USDA-ARS, National Sedimentation Laboratory.

173
174

Blaisdell, F. W., L. A. Clayton, et al. (1981). "Ultimate dimension of local scour." Journal of the
Hydraulics Division, ASCE 107(HY3): 327-337.
Blanckaert, K. and W. H. Graf (2001). "Mean Flow and Turbulence in Open-Channel Bend." Journal of
Hydraulic Engineering 127(10).
Blick (2005). Engelberg ist keine Insel mehr. Zurich, retrieved from http://www.blick.ch/news/schweiz/
hochwasser/artikel25138, October 22, 2005.
Borah, D. K. and P. K. Bordolai (1989). Stream bank erosion and bed evolution model. Sediment
Transport Modeling. Proceedings of the international symposium. S. Wang. New York, ASCE.
Bravard, J.-P. and F. Petit (1997). Les cours d'eau. Dynamique du système fluvial. Paris, Colin.
Briaud, J. L., F. C. K. Ting, et al. (2001). "Erosion function apparatus for scour rate predictions." Journal
of Geotechnical and Geoenvironmental Engineering 127(2): 105-113.
Bridge, J. S. (1976). "Bed topography and grain size in open channel bends." Sedimentology, Int.
Associatin of Sedimentologists 23: 407-414.
Bridge, J. S. (2003). Rivers and floodplains forms, processes, and sedimentary record. Oxford,
Blackwell.
Bujard, P. (2005). Failure of the Abbaye weir. Morges, personal communication.
Burrough, P. and R. McDonnell (1998). Principles of geographical information systems. Oxford [etc.],
Oxford University Press.
Charlton, M. E., A. R. G. Large, et al. (2003). "Application of airborne LiDAR in river environments:
The River Coquet, Northumberland, UK." Earth Surface Processes and Landforms 28(3): 299-306.
Chorley, R. J., S. A. Schumm, et al. (1984). Geomorphology. London New York, Methuen.
Chow, V. T., D. R. Maidment, et al. (1988). Applied hydrology. New York [etc.], McGraw-Hill.
Church, M. A., D. G. McLean, et al. (1987). River Bed Gravels: Sampling and Analysis. Sediment
Transport in Gravel-bed Rivers. C. R. Thorne, J. C. Bathurst and R. D. Hey. Chichester etc., John
Wiley and Sons.
Collet, L. W. (1955). Notice Explicative. Feuille 5. Carte géologique générale de la Suisse. 1:200'000.
Berne.
Consuegra, D., Beck, J., and Musy, A. (2000). Gestion quantitative des eaux de la Seymaz et de ses
affluents. Etude Hydrologique et Hydraulique. Diagnostic de l'état actuel des dangers liés aux
inondations et à l'érosion, EPFL, HYDRAM.
Consuegra, D., J.-M. Lance, et al. (1999). "Modélisation conceptuelle de la dynamique érosive d'un
cours d'eau." Ingénieurs et Architectes Suisses(9).
Consuegra, D. and E. Vez (1996). AMIE - Analysis et Modélisation Intégrées du cheminement des Eaux
en zones habitées, modélisation hydrologique, Application au bassin versant de la Haute Broye.
Lausanne, EPFL-HYDRAM.
Couper, P. (2003). "Effects of silt-clay content on the susceptibility of river banks to suaerial erosion."
Geomorphology 1338: 1-14.
Crausaz, P.-A. (2000). Du rôle intégrateur des systèmes d'information à référence spatiale dans la gestion
institutionnelle des eaux: analyse, méthodes, limites et perspectives. HYDRAM. Lausanne, EPFL.
 References 175

Creutin, J. D. and C. Obled (1982). "Objective Analyses and Mapping Techniques for Rainfall Fields -
an Objective Comparison." Water Resources Research 18(2): 413-431.
Cunge, J. A., F. M. Holly, et al. (1980). Practical aspects of computational river hydraulics. Boston -
Mass. a.o., Pitman.
Custer, W. (1935). Notice Explicative. Feuille 1222. Cossonay. Atlas géologique de la Suisse. 1:25'000.
Berne.
Daniels, M. D. and B. L. Rhoads (2004). Riparian Vegetation and Fluvial Geomorphology. Riparian
vegetation and fluvial geomorphology. S. J. Bennett and A. Simon. Washington, DC, American
Geophysical Union: 282 S.
Darby, S. E. (2005). "Refined hydraulic geometry data for British gravel-bed rivers." Journal of
Hydraulic Engineering-Asce 131(1): 60-64.
Darby, S. E., A. M. Alabyan, et al. (2002). "Numerical simulation of bank erosion and channel migration
in meandering rivers." Water Resources Research 38(9).
Darby, S. E. and I. Delbono (2002). "A model of equilibrium bed topography for meander bends with
erodible banks." Earth Surface Processes and Landforms 27(10): 1057-1085.
Darby, S. E. and C. R. Thorne (1994). "Prediction of Tension Crack Location and Riverbank Erosion
Hazards Along Destabilized Channels." Earth Surface Processes and Landforms 19(3): 233-245.
Darby, S. E. and C. R. Thorne (1996). "Numerical simulation of widening and bed deformation of
straight sand-bed rivers .1. Model development." Journal of Hydraulic Engineering-Asce 122(4):
184-193.
Darby, S. E., C. R. Thorne, et al. (1996). "Numerical simulation of widening and bed deformation of
straight sand-bed rivers .2. Model evaluation." Journal of Hydraulic Engineering-Asce 122(4): 194-
202.
De Souza, P. B., D. Consuegra, et al. (1994). CODEAU: A database package for the assessment and
analysis of hydrometeorological data. Hydroinformatics '94 proceedings of the First International
Conference on Hydroinformatics, Delft, Netherlands, 19-23 September 1994, Rotterdam [etc.],
Balkema.
Diplas, P. and A. J. Sutherland (1988). "Sampling Techniques for Gravel Sized Sediments." Journal of
Hydraulic Engineering 114(5): 484-501.
Dirks, K. N., J. E. Hay, et al. (1998). "High-resolution studies of rainfall on Norfolk Island Part II:
Interpolation of rainfall data." Journal of Hydrology 208(3-4): 187-193.
Duncan, M. R., B. Austin, et al. (1993). "The Effect of Gauge Sampling Density on the Accuracy of
Streamflow Prediction for Rural Catchments." Journal of Hydrology 142(1-4): 445-476.
Edijatno and C. Michel (1989). "Un modèle pluie-débit journalier à trois paramètres." La Houille
Blanche 2: 113-121.
Engelund, F. (1974). "Flow and bed topography in channel bends." Journal of Hydraulics Division-Asce
100(11): 1631-1648.
Fehr, R. (1987). "Einfache Bestimmung der Korngrössen-verteilung von Geschiebematerial mit Hilfe
der Linienzahlanalyse." Schweizer Ingenieur und Architekt 38.
176

FEMA (1999). Riverine Erosion Hazard Areas. Mapping Feasibility Study, Federal Emergency
Management Agency: 154.
Franca, M. (2005). A field study of turbulent flows in shallow gravel-bed rivers. Civil Engineering PhD
dissertation. Lausanne, EPFL.
Fripp, J. B. and P. Diplas (1993). "Surface Sampling in Gravel Streams." Journal of Hydraulic
Engineering-Asce 119(4): 473-490.
Furbish, D. J. (1991). "Spatial autoregressive structure in meander evolution." Geological Society of
America Bulletin 103: 1576-1589.
Goovaerts, P. (2000). "Geostatistical approaches for incorporating elevation into the spatial interpolation
of rainfall." Journal of Hydrology 228(1-2): 113-129.
Graf, W. H. and M. S. Altinakar (2002). Fluvial hydraulics flow and transport processes in channels of
simple geometry. Chichester, Wiley.
Graf, W. L. (1984). "A Probabilistic Approach to the Spatial Assessment of river Channel Instability."
Water Resources Research 20(7): 953-962.
Graf, W. L. (2001). "Damage control: Restoring the physical integrity of America's rivers." Ann. Assoc.
Am. Geog. 91: 1-27.
Grattier, M. (1980). "Les Sols du Plateau Vaudois." Mémoires de la Société vaudoise des Sciences
Naturelles 16(3): 188.
Grissinger, E. H. (1982). Bank erosion of cohesive materials. Gravel-bed rivers. Fluvial processes,
engineering and management ( First International workshop on gravel-bed rivers; Gregynog, 23-27
June 1980). R. D. Hey, J. C. Bathurst and C. R. Thorne. Chichester a.o., Wiley: XV, 875.
Hamblin, W. K. and E. H. Christiansen (1998). Earth's dynamic systems. Upper Saddle River, N.J.,
Prentice Hall.
Hamdi, Y., B. Hingray, et al. (2005). "Un modèle de prévision hydro-météorologique pour les crue du
Rhône supérieur en Suisse." Wasser Energie Luft(11/12 2005).
Hanson, G. J. (1990). "Surface Erodibility of Earthen Channels at High Stresses .1. Open Channel
Testing." Transactions of the Asae 33(1): 127-131.
Hanson, G. J. and K. R. Cook (2004). "Apparatus, test procedures, and analytical methods to measure
soil erodibility in situ." Applied Engineering in Agriculture 20(4): 455-462.
Hanson, G. J. and A. Simon (2001). "Erodibility of cohesive streambeds in the loess area of the
midwestern USA." Hydrological Processes 15(1): 23-38.
Hanson, G. J. and A. Simon (2002). "Discussion of "Erosion function apparatus for scour rate
predictions" by J. L. Briaud, F. C. K. Ting, H. C. Chen, Y. Cao, S. W. Han, and K. W. Kwak." Journal
of Geotechnical and Geoenvironmental Engineering 128(7): 627-628.
Hersberger, D. S. (2003). Wall roughness effects on flow and scouring in curved channels with gravel
bed. Lausanne, Laboratoire de Constructions Hydrauliques Ecole Polytechnique Fédérale de
Lausanne.
Hession, W. C., J. E. Pizzuto, et al. (2003). "Influence of bank vegetation on channel morphology in rural
and urban watersheds." Geology 31(2): 147-150.
 References 177

Hey, R. D. (1997). Stable River Morphology. Applied fluvial geomorphology for river engineering and
management. C. R. Thorne, R. D. Hey and M. D. Newson. Chichester [etc.], Wiley: VI, 376 S.
Hey, R. D., G. L. Heritage, et al. (1991). Streambank Protection in England and Wales. R&D Note 22.
London, National Rivers Authority: 75.
Hey, R. D. and C. R. Thorne (1983). "Accuracy of Surface Samples from Gravel Bed Materials." Journal
of Hydraulic Engineering-Asce 109(6): 842-851.
Hey, R. D. and C. R. Thorne (1986). "Stable Channels with Mobile Gravel Beds." Journal of Hydraulic
Engineering-Asce 112(8): 671-689.
Hickin, E. J. and G. C. Nanson (1975). "The Character of Channel Migration on the Beatton River,
Northeast British Columbia, Canada." Geological Society of America Bulletin 86: 487-494.
Howard, A. D., and Knutson, T.R. (1984). "Sufficient Conditions for River Meandering: A Simulation
Approach." Water Resources Research 20(11): 1659-1667.
Huff, F. A. (1970). "Sampling Errors in Measurement of Mean Precipitation." Journal of Applied
Meteorology 9: 35-44.
Hutchinson, J. N. (1995). Landslide hazard assessment. Landslides proceedings of the sixth international
symposium 10-14 February 1992, Christchurch. D. H. Bell. Rotterdam [etc.], Balkema: 3 vol.
HydroScientific (1999). The Photo-Electronic Erosion Pin (PEEP) System: User guide for models PEEP
110 and PEEP 200. Stratford-upon-Avon.
Irish, J. L. and W. J. Lillycrop (1999). "Scanning laser mapping of the coastal zone: the SHOALS
system." Isprs Journal of Photogrammetry and Remote Sensing 54(2-3): 123-129.
Jacobs Babtie (2005). SAFER - Flood risk assessment and survey on the River Enrick,
Geomorphological appraisal. Glasgow, Jacobs Babtie.
Johannesson, H., and Parker, G. (1989). Linear theory of river meanders. River meandering. S. Ikeda,
and Parker, G. Washington D.C., American Geophysical Union. Water Resources Monograph 12:
103-126.
Jowett, I. G. (1998). "Hydraulic geometry of New Zealand rivers and its use as a preliminary method of
habitat assessment." Regulated Rivers-Research & Management 14(5): 451-466.
Kellerhals, R. and D. I. Bray (1971). "Sampling procedures for coarse fluvial sediments." Journal of the
Hydraulics Division 97(HY8): 1165-1180.
Kienholz, H. and B. Krummenacher (1995). Légénde modulable pour la cartographie des phénomènes.
Bern, Office Fédéral de l'environnement, des forêts, et du paysage (OFEFP) and Office Fédérale des
eaux et de la géologie (OFEG) Suisse.
Knighton, D. (1998). Fluvial forms and processes a new perspective. London [etc.], Arnold [etc.].
Kondolf, G. M. (1997). "Application of the pebble count: Notes on purpose, method, and variants."
Journal of the American Water Resources Association 33(1): 79-87.
Lamouroux, N. and H. Capra (2002). "Simple predictions of instream habitat model outputs for target
fish populations." Freshwater Biology 47(8): 1543-1556.
Lancaster, S. T., and Bras, Rafael L. (2002). "A simple model of river meandering and its comparison to
natural channels." Hydrological Processes 16: 1-26.
178

Lance, J.-M. and D. Consuegra (1997). Bassin Versant de la Venoge. Etude des risques liés aux
inondations. Etude hydrologique, IATE/HYDRAM, EPF Lausanne.
Lance, J.-M. and D. Consuegra (1998). Bassin versant de la Venoge. Etude des dangers liés à l'érosion,
IATE/HYDRAM, EPF Lausanne.
Lance, J.-M., R. Thielen, et al. (1997). Bassin Versant de la Venoge. Etude des risques liés aux
inondations. Diagnostic de l'état actuel, IATE/HYDRAM, EPF Lausanne.
Lane, E. W. (1953). "Design of stable channels." Transactions, ASCE 120: 1234-1260.
Lane, S. N., J. H. Chandler, et al. (1994). "Developments in Monitoring and Modeling Small-Scale River
Bed Topography." Earth Surface Processes and Landforms 19(4): 349-368.
Langendoen, E. J. (2000). CONCEPTS-Conservational Channel Evolution and Pollutant Transport
System. Research Report No. 16. Oxford, USDA-ARS National Sedimentation Laboratory.
Lawler, D. M. (1993). "The Measurement of River Bank Erosion and Lateral Channel Change - a
Review." Earth Surface Processes and Landforms 18(9): 777-821.
Lawler, D. M., C. R. Thorne, et al. (1997). Bank Erosion and Instability. Applied fluvial geomorphology
for river engineering and management. C. R. Thorne, R. D. Hey and M. D. Newson. Chichester [etc.],
Wiley: VI, 376 S.
Leopold, L. B. and T. Maddock (1953). The hydraulic geometry of stream channels and some
physiographic implications. Washington, United States Government Printing Office.
Leopold, L. B. and M. G. Wolman (1957). River channel patterns. Washington, United States
Government Printing Office.
Loat, R. and A. Petrascheck (1997). Danger naturels : recommandation : Prise en compte des dangers
dus aux crues dans le cadre des activités de l'aménagement du territoire. Bienne, OFEG, OFAT,
OFEFP.
Mandal, J. N. and D. G. Divshikar (1995). Soil testing in civil engineering. Rotterdam [etc.], Balkema.
Maxey, G. B. (1964). Geology. Handbook of Applied Hydrology. V. T. Chow. New York, McGraw-
Hill.
Metzger, R. (2003). Modélisation des inondations par approches déterministe et stochastique avec prise
en compte des incertitudes topographiques pour la gestion des risques liés aux crues. HYDRAM.
Lausanne, EPFL.
Meylan, P. and A. Musy (1999). Hydrologie Fréquentielle. Bucarest, Editions *H*G*A.
Mosselman, E. (1992). Mathematical Modeling of Morphological Processes in Rivers with Erodible
Cohesive Banks. PhD thesis. Delft, The Netherlands, Delft University of Technology.
Mosselman, E. (1998). "Morphological modelling of rivers with erodible banks." Hydrological
Processes 12(8): 1357-1370.
Musy, A. and C. Higy (1998). Hydrologie Appliqué. Lausanne, Course book.
Musy, A. and C. Higy (2004). Hydrologie. Lausanne, Presses Polytechniques et Universitaires
Romandes.
Nanson, G. C. and J. C. Croke (1992). "A Genetic Classification of Floodplains." Geomorphology 4(6):
459-486.
 References 179

Northwood Technologies (2001). Vertical Mapper 3.0 User guide. Ontario.

Osman, A. M. (1985). Channel width response to changes in flow hydraulics and sediment load. PhD
thesis. Fort Collins, Colorada State University.
Osman, A. M. and C. R. Thorne (1988). "Riverbank Stability Analysis .1. Theory." Journal of Hydraulic
Engineering-Asce 114(2): 134-150.
Oudin, L., F. Hervieu, et al. (2005b). "Which potential evapotranspiration input for a lumped rainfall-
runoff model? Part 2 - Towards a simple and efficient potential evapotranspiration model for rainfall-
runoff modelling." Journal of Hydrology 303: 290-306.
Oudin, L., C. Michel, et al. (2005a). "Which potential evapotranspiration input for a lumped rainfall-
runoff model? Part 1 - Can rainfall-runoff models effectively handle detailed potential
evapotranspiration inputs?" Journal of Hydrology 303: 275-289.
Parker, G. and E. D. Andrews (1986). "On the time development of meander bends." Journal of Fluid
Mechanics 162: 139-156.
Petrie, J. (1998). The Accuracy of River Bed Sediment Samples. University of Virginia - Civil
Engineering. Blacksburg, Virginia.
Petrie, J. and P. Diplas (2000). "Statistical approach to sediment sampling accuracy." Water Resources
Research 36(2): 597-605.
Piégay, H. and S. A. Schumm (2003). System Approaches in Fluvial Geomorphology. Tools in fluvial
geomorphology. G. M. Kondolf and H. Piégay. Chichester, Wiley: 688 S.
Prétorian, R. (1994). Analyse hydro-météorologique et modélisation hydrologique synthétique du
transfert de de l'acheminement entre sous-bassins versant de la Venoge". Lausanne, HYDRAM.
Primault, B. (1972). Etude mésoclimatique du Canton de Vaud. Lausanne, Office Cantonal Vaudois de
l'Urbanisme.
Rakotondranaly, N. (2001). Développement et Validation d'une Méthodologie pour l'Identification et
l'Evaluatiion des Dangers liés à l'Erosion des Berges des cours d'eau. Lausanne, EPFL/HYDRAM.
Rice, S. and M. A. Church (1996). "Sampling surficial fluvial gravels: the precision of size distribution
percentile estimates." Journal of Sedimentary Research 66(3): 654-665.
Rice, S. P. and J. K. Haschenburger (2004). "A hybrid method for size characterization of coarse
subsurface fluvial sediments." Earth Surface Processes and Landforms 29(3): 373-389.
Richards, K. (1982). Rivers form and process in alluvial channels. London New York, Methuen.
Richardson, W. R. (2002). "Simplified Model for Assessing Meander Bend Migration Rates." Journal of
Hydraulic Engineering 128(12): 1094-1097.
Rodi, W. (1993). Engineering turbulence modelling and experiments 2 proceedings of the Second
International Symposium on Engineering Turbulence Modelling and Measurements, Florence, Italy,
31 May - 2 June, 1993. Amsterdam [etc.], Elsevier.
Rosgen, D. and H. L. Silvey (1996). Applied river morphology. Pagosa Springs, CO, Wildland
Hydrology.
Schaefli, B., B. Hingray, et al. (2005). "A conceptual glacio-hydrological model for high mountainous
catchments." Hydrology and Earth System Sciences 9.
180

Schneider, T. (2001). Caractérisation multicritère des formations géologiques du canton de Vaud et de

leur prédispositions face aux Dangers Naturels. "Géologie de Ingénieur et de l'Environnement"
Postgraduate Diploma Presentation. Lausanne.
Schumm, S. A. (1981). "Evolution and response of the fluvial system, sedimentologic implications."
Society of Economic Paleontologists and Mineralogists Special Publication 31: 19-29.
Schumm, S. A. (2005). River variability and complexity. New York, Cambridge Univeristy Press.
Selby, M. J. (1993). Hillslope materials and processes. Oxford, Oxford University Press.
Shields, A. (1936). Anwendung der Aehnlichkeitsmechanik und der Turbulenzforschung auf die
Geschiebebewegung. Berlin, Eigenverlag der Preussischen Versuchsanstalt für Wasserbau und
Schiffbau.
Simon, A., A. Curini, et al. (2000). "Bank and near-bank processes in an incised channel."
Geomorphology 35(3-4): 193-217.
Simon, A. and P. W. Downs (1995). "An Interdisciplinary Approach to Evaluation of Potential
Instability in Alluvial Channels." Geomorphology 12(3): 215-232.
Simon, A. and C. R. Hupp (1986). Channel evolution in modified Tennessee channels. Proceedings of
the 4th Federal Interagency Sedimentation Conference, Las Vegas, Nevada, US Government Printing
Office, Washington D.C.
Simon, A. and R. E. Thomas (2002). "Processes and forms of an unstable alluvial system with resistant,
cohesive streambeds." Earth Surface Processes and Landforms 27(7): 699-718.
Simon, A., R. E. Thomas, et al. (2002). "Case study: Channel stability of the Missouri River, eastern
Montana." Journal of Hydraulic Engineering-Asce 128(10): 880-890.
Smart, G. M. (1984). "Sediment Transport Formula for Steep Channels." Journal of Hydraulic
Engineering 110(3).
Spreafico, M., Schweiz Landeshydrologie und -Geologie, et al. (1992). Hydrologischer Atlas der
Schweiz. Bern, Edmz.
Stewardson, M. (2005). "Hydraulic geometry of stream reaches." Journal of Hydrology 306(1-4): 97-
111.
Swisstopo MNS/MNT-MO., retrieved from http://www.swisstopo.ch/fr/products/digital/height/
dom_dtmAV, October 21, 2005.
Swisstopo MNT25, retrieved from http://www.swisstopo.ch/fr/products/digital/height/dhm25, October
21, 2005.
Swisstopo Status of Updates, retrieved from http://www.swisstopo.ch/pub/down/products/analog/maps/
status_de-fr-en-it.pdf, October 22, 2005.
Taylor, D. W. (1948). Fundamentals of soil mechanics. New York, Wiley.
Terzaghi, K., R. B. Peck, et al. (1996). Soil mechanics in engineering practice. New York [etc.], Wiley.
Thoma, D. P., S. C. Gupta, et al. (2005). "Airborne laser scanning for riverbank erosion assessment."
Remote Sensing of Environment 95(4): 493-501.
Thompson, A. M., B. N. Wilson, et al. (2004). "Shear stress partitioning for idealized vegetated
surfaces." Transactions of the Asae 47(3): 701-709.
 References 181

Thorne, C. R. (1981). Field measurements of rates of bank erosion and bank material strength. Erosion
and Sediment Transport Measurement, Forence Symposium, International Association of
Hydrological Sciences Publication.
Thorne, C. R. (1982). Processes and Mechanisms of River Bank Erosion. Gravel-bed rivers fluvial
processes, engineering and management (record of the proceedings of the International workshop on
engineering problems in the management of gravel-bed rivers = First International workshop on
gravel-bed rivers; Gregynog, 23-27 June 1980). R. D. Hey. Chichester a.o., Wiley: XV, 875.
Thorne, C. R. (1997). Channel Types and Morphological Classification. Applied fluvial geomorphology
for river engineering and management. C. R. Thorne, R. D. Hey and M. D. Newson. Chichester [etc.],
Wiley: VI, 376 S.
Thorne, C. R. and N. K. Tovey (1981). "Stability of Composite River Banks." Earth Surface Processes
and Landforms 6(5): 469-484.
Tolhurst, T. J., K. S. Black, et al. (1999). "Measuring the in situ erosion shear stress of intertidal
sediments with the Cohesive Strength Meter (CSM)." Estuarine, Coastal, and Shelf Science 49: 281-
294.
USACE (2002). HEC-RAS River Analysis System. User's Manual. Version 3.1. Davis, CA, USA, US
Army Corp of Engineers.
Vereinigung Schweizerischer Strassenfachleute (1997a). Identifikation der Lockergesteine
Labormethode mit Klassifikation nach USCS méthode de laboratoire avec classification selon
l'USCS. Zürich, Vss.
Vereinigung Schweizerischer Strassenfachleute (1999b). Bodenkennziffern. Zürich, Vss.
Vieira, D. and W. Wu (2002). One-Dimensional Channel Network Model CCHE1D 3.0 - User's Manual.
Oxford, National Center for Computational Hydroscience and Engineering.
Winterbottom, S. J. and D. J. Gilvear (2000). "A GIS-based approach to mapping probabilities of river
bank erosion: Regulated River Tummel, Scotland." Regulated Rivers-Research & Management
16(2): 127-140.
Wojcik, R. and T. A. Buishand (2003). "Simulation of 6-hourly rainfall and temperature by two
resampling schemes." Journal of Hydrology 273(1-4): 69-80.
Wolman, M. G. (1954). "A Method of Sampling Coarse River-Bed Material." Transactions, American
Geophysical Union 35(6): 951-956.
Wood, A. L., A. Simon, et al. (2001). "Bank-toe processes in incised channels: the role of apparent
cohesion in the entrainment of failed bank materials." Hydrological Processes 15(1): 39-61.
Wu, T. H., P. E. Beal, et al. (1988). "In-situ shear test of soil-root systems." Journal of Geotechnical
Engineering 114(12): 1376-1394.
Wu, W. and D. Vieira (2002). One-Dimensional Channel Network Model CCHE1D 3.0 - Technical
Manual. Oxford, National Center for Computational Hydroscience and Engineering.
Wu, W. and S. Y. Wang (2004). "Depth-averaged 2-D calculation of flow and sediment transport in
curved channels." International Journal of Sediment Research 19(4): 241-257.
182
Appendix

183
184 Appendix 1

Appendix 1 Textural triangles of the Vaud plateau

Appendix 1.1 Molassic and Morainic particle distributions

MOLASSE L = Loam
Stampien LU = Silt loam
Aquitanien U = Silt
Burdigalien LA = Clay loam
LAU = Silty clay loam
AY

LS = Sandy loam
MORAINE
CL

SL = Loamy sand
Stampien domain
Aquitanien domain

SIL
Burdigalien domain

SAND

Figure A.1. Textural triangles (FAO) of the Vaud plateau morainic and molassic particle distributions
(adapted from Grattier, 1980)
 Textural triangles of the Vaud plateau 185

Appendix 1.2 Alluvial soil particle distributions

horizon A
Orbe plain L = Loam
horizon C
LU = Silt loam
horizon A U = Silt
Broye plain
horizon C LA = Clay loam
LAU = Silty clay loam
large facies
(near streams and old AY LS = Sandy loam
CL

meanders) SL = Loamy sand

fine facies (old marshes)

SIL
limit between

T
fine and large
facies
samples from
same profile

SAND

Figure A.2. Textural triangles (FAO) of the Vaud plateau alluvial soil particle distributions
(adapted from Grattier, 1980)
186 Appendix 2

Appendix 2 Map of the Lower Venoge Study Reach

Reproduced with permission

from swisstopo (BA068334)

Figure A.3. Upstream section of the study reach of the Lower Venoge River (background: CP25
©1997, Swisstopo). Distance in meters from outlet is shown by points. 2004 centerline in black.
 Map of the Lower Venoge Study Reach 187

Reproduced with permission

from swisstopo (BA068334)

Figure A.3 (cont.). Downstream section of the study reach of the Lower Venoge River (background:
CP25 ©1997, Swisstopo). Distance in meters from outlet is shown by points. 2004 centerline in black.
188 Appendix 3

Appendix 3 Lower Venoge River and tributary GSD’s

class number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
representative diameter (cm) 0.0001 0.0011 0.004 0.008 0.021 0.066 0.18 0.35 0.58 0.85 1.20 1.69 2.39 3.38 4.78 6.76 9.56 13.52 19.12 27.04
lower (cm) 0 0.0002 0.002 0.005 0.010 0.032 0.10 0.25 0.45 0.70 0.99 1.40 1.98 2.80 3.96 5.60 7.92 11.20 15.84 22.40
upper (cm) 0.0002 0.0020 0.005 0.010 0.032 0.100 0.25 0.45 0.70 0.99 1.40 1.98 2.80 3.96 5.60 7.92 11.20 15.84 22.40 31.68
distance
to outlet
river (m) field id
Venoge 22654 1 0.004 0.009 0.008 0.009 0.023 0.040 0.05 0.05 0.05 0.05 0.05 0.07 0.08 0.09 0.11 0.13 0.16 0.02 0.00 0.00
22091 2 0.004 0.008 0.007 0.008 0.020 0.036 0.05 0.04 0.04 0.04 0.05 0.06 0.07 0.08 0.10 0.12 0.14 0.10 0.03 0.02
20513 5 0.005 0.012 0.010 0.011 0.029 0.052 0.07 0.06 0.06 0.06 0.07 0.08 0.10 0.12 0.07 0.05 0.05 0.05 0.00 0.03
19687 6 0.004 0.009 0.007 0.008 0.022 0.039 0.05 0.05 0.05 0.04 0.05 0.06 0.07 0.09 0.11 0.13 0.11 0.02 0.06 0.02
18848 8 0.004 0.008 0.007 0.007 0.020 0.035 0.05 0.04 0.04 0.04 0.05 0.06 0.07 0.08 0.10 0.11 0.14 0.12 0.02 0.00
18602 9 0.004 0.009 0.008 0.009 0.024 0.043 0.06 0.05 0.05 0.05 0.06 0.07 0.08 0.10 0.12 0.14 0.12 0.01 0.00 0.00
18323 10 0.004 0.008 0.007 0.008 0.021 0.037 0.05 0.05 0.04 0.04 0.05 0.06 0.07 0.08 0.10 0.12 0.13 0.04 0.03 0.03
17706 11 0.004 0.010 0.008 0.009 0.025 0.044 0.06 0.05 0.05 0.05 0.06 0.07 0.08 0.10 0.12 0.14 0.11 0.00 0.00 0.00
17276 a1 0.006 0.012 0.011 0.012 0.032 0.056 0.07 0.07 0.07 0.06 0.08 0.09 0.11 0.13 0.15 0.04 0.00 0.00 0.00 0.00
17157 12 0.004 0.009 0.008 0.009 0.023 0.041 0.05 0.05 0.05 0.05 0.06 0.07 0.08 0.09 0.11 0.13 0.10 0.07 0.00 0.00
16733 13 0.005 0.011 0.010 0.011 0.029 0.052 0.07 0.06 0.06 0.06 0.07 0.08 0.10 0.12 0.14 0.08 0.04 0.00 0.00 0.00
16195 14 0.006 0.013 0.011 0.012 0.033 0.059 0.08 0.07 0.07 0.07 0.08 0.09 0.11 0.13 0.13 0.01 0.02 0.00 0.00 0.00
16041 b3 0.006 0.014 0.012 0.013 0.035 0.062 0.08 0.08 0.07 0.07 0.08 0.10 0.13 0.12 0.06 0.06 0.00 0.00 0.00 0.00
15468 15 0.005 0.011 0.010 0.011 0.029 0.052 0.07 0.06 0.06 0.06 0.07 0.08 0.10 0.12 0.14 0.10 0.01 0.00 0.00 0.00
15085 16 0.003 0.007 0.006 0.007 0.019 0.034 0.04 0.04 0.04 0.04 0.05 0.05 0.07 0.08 0.09 0.11 0.13 0.13 0.05 0.00
14569 17 0.004 0.008 0.007 0.008 0.021 0.037 0.05 0.05 0.04 0.04 0.05 0.06 0.07 0.09 0.10 0.12 0.14 0.04 0.03 0.02
13578 19 0.005 0.011 0.009 0.010 0.027 0.048 0.06 0.06 0.06 0.06 0.07 0.08 0.09 0.11 0.11 0.12 0.04 0.03 0.00 0.00
13499 20 0.004 0.009 0.008 0.009 0.024 0.043 0.06 0.05 0.05 0.05 0.06 0.07 0.08 0.10 0.12 0.14 0.12 0.01 0.00 0.00
12021 22 0.005 0.011 0.010 0.011 0.029 0.051 0.07 0.06 0.06 0.06 0.07 0.08 0.10 0.12 0.14 0.12 0.01 0.00 0.00 0.00
11728 d1 0.005 0.011 0.010 0.011 0.029 0.052 0.07 0.06 0.06 0.06 0.07 0.08 0.10 0.12 0.14 0.08 0.03 0.00 0.00 0.00
11200 d2 0.005 0.012 0.010 0.011 0.030 0.053 0.07 0.06 0.06 0.06 0.07 0.09 0.10 0.12 0.14 0.09 0.01 0.00 0.00 0.00
10624 23 0.005 0.010 0.009 0.010 0.026 0.047 0.06 0.06 0.06 0.05 0.06 0.08 0.09 0.11 0.13 0.11 0.07 0.02 0.00 0.00
10329 d3 0.005 0.012 0.010 0.011 0.030 0.054 0.07 0.07 0.06 0.06 0.07 0.09 0.10 0.12 0.15 0.07 0.01 0.00 0.00 0.00
10058 28 0.006 0.012 0.010 0.012 0.031 0.055 0.07 0.07 0.07 0.06 0.07 0.09 0.11 0.13 0.15 0.06 0.00 0.00 0.00 0.00
9867 e2 0.005 0.011 0.009 0.010 0.027 0.048 0.06 0.06 0.06 0.05 0.07 0.08 0.09 0.11 0.13 0.15 0.03 0.00 0.00 0.00
9726 e3 0.005 0.011 0.009 0.010 0.027 0.049 0.06 0.06 0.06 0.06 0.07 0.08 0.09 0.11 0.13 0.13 0.05 0.00 0.00 0.00
9600 29 0.006 0.012 0.010 0.012 0.031 0.055 0.07 0.07 0.07 0.06 0.07 0.09 0.11 0.13 0.08 0.10 0.04 0.00 0.00 0.00
9399 e4 0.005 0.010 0.008 0.009 0.025 0.045 0.06 0.06 0.05 0.05 0.06 0.07 0.09 0.10 0.12 0.14 0.09 0.00 0.00 0.00
9247 30 0.005 0.010 0.009 0.010 0.026 0.047 0.06 0.06 0.06 0.05 0.06 0.08 0.09 0.11 0.13 0.15 0.04 0.01 0.00 0.00
9100 31 0.005 0.011 0.010 0.011 0.029 0.051 0.07 0.06 0.06 0.06 0.07 0.08 0.10 0.12 0.14 0.09 0.04 0.00 0.00 0.00
8875 e5 0.005 0.011 0.009 0.011 0.028 0.050 0.07 0.06 0.06 0.06 0.07 0.08 0.10 0.11 0.14 0.10 0.04 0.00 0.00 0.00
8625 e6 0.005 0.011 0.009 0.010 0.027 0.048 0.06 0.06 0.06 0.05 0.06 0.08 0.09 0.11 0.13 0.12 0.06 0.00 0.00 0.00
8562 e7 0.006 0.013 0.011 0.012 0.032 0.058 0.08 0.07 0.07 0.07 0.08 0.09 0.11 0.13 0.16 0.01 0.00 0.00 0.00 0.00
8169 32 0.006 0.012 0.010 0.011 0.030 0.054 0.07 0.07 0.06 0.06 0.07 0.09 0.10 0.12 0.15 0.05 0.03 0.00 0.00 0.00
7127 33 0.006 0.012 0.010 0.012 0.031 0.056 0.07 0.07 0.07 0.06 0.08 0.09 0.11 0.13 0.15 0.03 0.02 0.00 0.00 0.00
6496 34 0.007 0.014 0.012 0.014 0.037 0.065 0.09 0.08 0.08 0.07 0.09 0.11 0.13 0.15 0.05 0.01 0.00 0.00 0.00 0.00
5714 35 0.007 0.015 0.013 0.014 0.038 0.068 0.09 0.08 0.08 0.08 0.09 0.11 0.13 0.15 0.03 0.00 0.00 0.00 0.00 0.00
5226 36 0.007 0.014 0.012 0.013 0.036 0.064 0.08 0.08 0.08 0.07 0.09 0.10 0.12 0.11 0.09 0.02 0.00 0.00 0.00 0.00
4729 g1 0.006 0.014 0.012 0.013 0.035 0.062 0.08 0.08 0.07 0.07 0.08 0.10 0.12 0.13 0.12 0.01 0.00 0.00 0.00 0.00
4528 g2 0.006 0.013 0.011 0.012 0.032 0.057 0.08 0.07 0.07 0.06 0.08 0.09 0.11 0.13 0.15 0.02 0.00 0.01 0.00 0.00
3451 g3 0.007 0.015 0.013 0.014 0.038 0.068 0.09 0.08 0.08 0.08 0.09 0.11 0.13 0.16 0.02 0.00 0.00 0.00 0.00 0.00
3283 g4 0.007 0.016 0.013 0.015 0.040 0.070 0.09 0.09 0.08 0.08 0.10 0.11 0.14 0.13 0.01 0.01 0.00 0.00 0.00 0.00
2889 37 0.007 0.015 0.013 0.014 0.038 0.068 0.09 0.08 0.08 0.08 0.09 0.11 0.13 0.13 0.05 0.00 0.00 0.00 0.00 0.00
2702 h1 0.007 0.014 0.012 0.014 0.036 0.065 0.09 0.08 0.08 0.07 0.09 0.10 0.12 0.15 0.05 0.02 0.00 0.00 0.00 0.00
2538 38 0.006 0.013 0.011 0.013 0.033 0.059 0.08 0.07 0.07 0.07 0.08 0.10 0.11 0.14 0.09 0.06 0.00 0.00 0.00 0.00
2083 41 0.008 0.016 0.014 0.016 0.042 0.074 0.10 0.09 0.09 0.09 0.10 0.12 0.14 0.09 0.02 0.00 0.00 0.00 0.00 0.00
1522 39 0.006 0.012 0.010 0.011 0.030 0.054 0.07 0.07 0.06 0.06 0.07 0.09 0.10 0.12 0.09 0.08 0.05 0.00 0.00 0.00
1101 40 0.006 0.013 0.011 0.012 0.033 0.059 0.08 0.07 0.07 0.07 0.08 0.10 0.11 0.13 0.11 0.04 0.00 0.00 0.00 0.00
Molomba 125 3 0.004 0.008 0.006 0.007 0.019 0.034 0.05 0.04 0.04 0.04 0.05 0.06 0.07 0.08 0.09 0.11 0.13 0.15 0.01 0.00
25 4 0.004 0.008 0.007 0.008 0.021 0.037 0.05 0.05 0.04 0.04 0.05 0.06 0.07 0.08 0.10 0.12 0.14 0.09 0.01 0.00
Sorge 340 21 0.004 0.009 0.007 0.008 0.022 0.039 0.05 0.05 0.05 0.04 0.05 0.06 0.07 0.09 0.11 0.13 0.15 0.07 0.00 0.00
85 18 0.005 0.011 0.009 0.010 0.028 0.049 0.07 0.06 0.06 0.06 0.07 0.08 0.09 0.11 0.13 0.13 0.03 0.00 0.00 0.00
Vaube 165 24 0.005 0.010 0.009 0.010 0.026 0.046 0.06 0.06 0.06 0.05 0.06 0.07 0.09 0.11 0.13 0.15 0.06 0.00 0.00 0.00
45 25 0.005 0.010 0.008 0.009 0.025 0.045 0.06 0.06 0.05 0.05 0.06 0.07 0.09 0.10 0.12 0.13 0.05 0.02 0.04 0.00
Arena 100 27 0.006 0.012 0.010 0.012 0.031 0.055 0.07 0.07 0.07 0.06 0.07 0.09 0.10 0.12 0.15 0.03 0.01 0.02 0.00 0.00
70 26 0.007 0.015 0.013 0.014 0.038 0.068 0.09 0.08 0.08 0.08 0.09 0.11 0.13 0.08 0.09 0.02 0.00 0.00 0.00 0.00

Figure A.4. Grain size fractions as determined by the Fehr method for the Lower Venoge, the Ruisseau
de Molomba, the Vaube, and the Arena
 River Enrick streambank classification 189

Appendix 4 River Enrick streambank classification

Type Photograph Description Stability
Tree lined bank Cobble or coarse High
gravel bank, lined
with trees. Cobbles
often covered by
moss indicating
high stability. The
stability is
encouraged by tree
roots.
Banks often low.
Cobble bank Cobble or coarse Low
gravel banks, often
with a fine gravel
and sand matrix.
Generally free from
vegetation. Banks
usually around 1 m
o
high sloping at 45 .
Vegetated valley- Valley side bluff Moderate
side slope consisting of but
reworked vulnerable
Quaternary to large
deposits. Covered floods.
by vegetation
including trees and
mixed scrub.
Eroding valley-side Valley side bluff Low –
slope consisting of Frequent
reworked failures
Quaternary during high
deposits. Free from flows.
vegetation often
with accumulation
of material at toe.
Mass failures
common.
Fine grained bank River bank High
composed entirely
of fine sand, silts
and clay. These
banks are highly
cohesive. Presence
of moss on bank
face indicative of
stability.

Figure A.5. Classification of streambanks used in the project SAFER on the River Enrick
(Jacobs Babtie, 2005)
190 Appendix 4

Composite bank Bank consists of Low-

two layers. A coarse moderate
cobble-gravel
horizon at base
over laid by fine
sandy sediments.
Thickness of
horizons varies.
Fluvially derived –
alluvium.
Point bar River bank very low Moderate -
or absent due to the High
presence of a
gravel bar. In most
cases the bar
surface is at the
same elevation as
the surrounding low
floodplain.
Rip-rapped bank Bank composed of High
large boulders
placed to protect
the bank from fluvial
entrainment. The
size of boulders
vary, the larger the
boulder the greater
the stability.
Walled Bank protected by a High
wall of interlocking
river cobbles. These
are not cemented.

Bedrock River bank formed Very High

entirely of bedrock.
With the exception
of the gorge
upstream of
Corrimony, these
tend to occur along
only one bank.

Figure A.5 (cont.). Classification of streambanks used in the project SAFER on the River Enrick
(Jacobs Babtie, 2005)
 Textural triangles of the Lower Venoge River 191

Appendix 5 Textural triangles of the Lower Venoge River

100
FAO Classes
C Clay
L Loam 90 10
CL Clay loam
Si Silt
SiC Silty clay
SiCL Silty clay loam 80 20

SiL Silt loam

SC Sandy clay
SCL Sandy clay loam 70 30
SL Sandy loam C
C
LS Loamy sand
S Sand, unspecified 60 40

%
%s
yy
ccllaa

siilltt
50 50
%
%

SiC
SiC
40 SC
SC 60

CL
CL SiCL
SiCL
30 4.2/2
4.2/2
4.2/2 70
4.2/2
4.2/2
4.2/2
SCL
SCL 4.2/3
4.2/3
4.2/3 4.2/1
4.2/1
4.2/1
4.2/1
4.2/1
4.2/1
4.2/3
4.2/3
4.2/3
L
L
SiL
SiL
20 1.3/1
1.3/1
1.3/1
1.3/1
1.3/1
1.3/1 1.3/2
1.3/2
1.3/2
1.3/2 25.2/2
1.3/2 25.2/2
25.2/2
25.2/2 80
29.2/1
29.2/1
29.2/1
29.2/1 8.2/1
29.2/1 8.2/1
8.2/1 1.3/3
1.3/3
1.3/3
1.3/3
1.3/3
1.3/3
8.2/1
8.2/1
8.2/3
8.2/3
8.2/3
8.2/3
8.2/3 18.2/3
18.2/3
18.2/3 15.2/1
SL
SL 21.2/2
21.2/2
21.2/2
21.2/2
21.2/2 15.2/1
15.2/1
15.2/1
15.2/1
8.2/2
8.2/2
8.2/2
21.2/2
21.2/2
21.2/2 18.2/1 8.2/2
18.2/1
18.2/1
18.2/1
18.2/1 8.2/2 25.2/3
25.2/3
25.2/3
25.2/3
25.2/3 25.2/1
25.2/1
25.2/1
25.2/1
25.2/1
25.2/1
21.2/2
21.2/2 12.2/2
12.2/2
12.2/2
12.2/2
12.2/2 15.2/3
15.2/3
15.2/3
21.2/3
21.2/3
21.2/3 15.2/3
15.2/3
10 21.2/3
21.2/3 12.2/1
12.2/1
12.2/1
12.2/1
12.2/1 32.2/2
32.2/2
32.2/2 90
LS
LS 6.2/1
6.2/1
6.2/1
32.2/2
32.2/2
15.2/2
15.2/2
15.2/2
6.2/1
6.2/1
6.2/1 6.2/3 35.1/1
35.1/1
35.1/1 29.2/1 32.2/1
29.2/1
29.2/1
29.2/1
29.2/1 32.2/1
32.2/1
32.2/1
32.2/1 15.2/2
15.2/2
6.2/3 35.1/1
6.2/3
6.2/3
6.2/3
6.2/3 35.1/1
38.2/1
38.2/1
6.2/2
6.2/2
6.2/2 38.2/1
38.2/1 29.2/3 32.2/3
29.2/3
29.2/3
38.2/1 29.2/3
29.2/3
29.2/3 32.2/3
32.2/3
32.2/3
32.2/3
32.2/3
6.2/2 18.2/2
18.2/2
18.2/2
S
S 35.1/3
35.1/3
35.1/3
38.2/2
38.2/2
38.2/2
38.2/2
38.2/2
38.2/2 29.2/2
29.2/2
29.2/2
29.2/2
29.2/2
29.2/2
18.2/2
18.2/2
18.2/2 Si
Si
35.1/2
35.1/2
35.1/2
35.1/2
35.1/2 35.1/3
35.1/3
35.1/3
12.2/3
12.2/3
12.2/3
12.2/3
12.2/3
12.2/3 100
100 90 80 70 60 50 40 30 20 10

% sand
% sand

Figure A.6. FAO textural triangle representation of Lower Venoge 2004 level 2 investigation PSD’s
192 Appendix 5

100
FAO Classes
C Clay
L Loam 90 10
CL Clay loam
Si Silt
SiC Silty clay
SiCL Silty clay loam 80 20

SiL Silt loam

SC Sandy clay
SCL Sandy clay loam 70 30
SL Sandy loam C
C
LS Loamy sand
S Sand, unspecified 60 40

%
%
yy
ccllaa

ssiilt
50 50

lt
%
%

12.1/2
12.1/2
12.1/2
12.1/2
12.1/2
12.1/2
SiC
SiC
40 SC
SC 60

CL
CL SiCL
SiCL
30 8.1/2
8.1/2
8.1/2
8.1/2
8.1/2
8.1/2 70
SCL
SCL
LL
6.1/2
6.1/2
6.1/2
SiL
SiL
20 6.1/2
6.1/2
6.1/2 10.2/2
10.2/2
10.2/2 80
10.2/2
10.2/2
10.2/2
2.5/1
2.5/1
2.5/1
2.5/1
2.5/1
2.5/1 15.2/3
15.2/3
15.2/3
15.2/3
15.2/3
15.2/3
13.1/1
13.1/1
13.1/1
!

SL
SL
SL 13.1/1 13.2/2
13.2/2
13.2/2
13.2/2
13.2/2
13.2/2
2.5/3
2.5/3
2.5/3
2.5/3
2.5/3 20/2
20/2
20/2
20/2
20/2
!

10 16/2
16/2
16/2
16/2
16/2
!

11.2/2
11.2/2
11.2/2 90
!

12.2/3
12.2/3
12.2/3
12.2/3
12.2/3
12.2/3 2.5/2
2.5/2
2.5/2
2.5/2
2.5/2
2.5/2 17.1/2
17.1/2
17.1/2
17.1/2
17.1/2
17.1/2 11.2/2
11.2/2
11.2/2
!

LS
LS
LS 2.5/4
2.5/4
2.5/4
2.5/4
2.5/4
2.5/4 5.2/2
5.2/2
5.2/2
5.2/2
5.2/2
5.2/2
12.3/4
12.3/4
12.3/4
12.3/4
12.3/4
12.3/4 7/2
7/2
7/2
7/2
7/2
7/2

SS Si
Si
Si
8.3/4
8.3/4
8.3/4
8.3/4
8.3/4
8.3/4 15.1/2
15.1/2
15.1/2 19.2/2
19.2/2
19.2/2 10.3/3
10.3/3
10.3/3
15.1/2
15.1/2
15.1/2 10.3/3
10.3/3
10.3/3 100
100 90 80 70 60 50 40 30 20 10

% sand
% sand

Figure A.7. FAO textural triangle representation of Lower Venoge 2004 level 1 investigation PSD’s.
Lines join PSD’s sampled from different depths of a same bank.
 Cumulative measured bank positions changes 193

Appendix 6 Cumulative measured bank positions changes

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
1.1 note:
bank top -20 0 0 0 0 0 0 0 0 0 0 0
top -30 0 1 1.5 1.5 1.5 2 2 3 3 4 4
mid top -108 0 2 2.5 1.8 0 0 1 1 1 1 1
bottom -147 0 -3 -2.8 -2.8 -2.8 -7.5 -6 -7 -9 -8 -9
1.2 note:
bank top -35 0 0 0 0 0 0 0 0 0 0 0
top -45 0 6 6 6 6 6 6 6 7 9 10.5
mid top -90 0 2.5 3 3.5 3.5 3.5 5 6 6 10 12
bottom -125 0 2 1 0 -1 -2 -2 -2 -4 -1.5 -1.5
1.3 note:
bank top -21 0 0 0 0 0 0 0 0 -6 -6 -6
top -30 0 1.5 1.5 1.5 1.5 1.5 1 1 1 1 1
mid top -96 1 1.5 1.5 2 1.5 1 1 1.5 1.5 2.5 1.5
bottom -151 0 2.5 3 1.4 0 -0.5 -3 -4 -3.5 0.5 3.5
2.1 note:
bank top -20 0 0 0 0 0 0 3 3 3 3 3
top -52 0 5.5 6 5.5 5.5 5.5 7 8.5 8.5 10.5 12.5
mid top -70 0.5 5.5 5.5 5.2 5.5 5.5 7 7 7 17 18.5
bottom -100 0 1 3 1.5 1.5 1.5 1.5 1.5 2.5 8 9
2.2 note:
bank top -23 0 0 0 0 -2 -20 -7 46 46 46 46
top -50 0 4 3.5 3.5 3.5 3.5 10 10 2.7 5.5 7
mid top
bottom -100 0 2.3 2.8 2.8 2.8 60.3 61.3 61.3 65 69 72
2.3 note:
bank top -61 0 0 0 0 -10 -10 -10 -10 -10 -10 -10
top -90 0 10 1 5 12 12 10 9 9 12 16
mid top -130 0 22 25 25 35 21 81 80 98 98 101
bottom -160 0.6 7.6 -0.4 -0.4 1.6 1.6 16.6 9.6 9.6
2.4 note:
bank top -34 0 31 31 36 36 36 36 38 41 46 46
top -40 0 38 39 39 39 47 50 48 37 37 22
mid top -80 0 7 7 7 7 7 7 7 16 22 22
bottom -120 1 1 4 13 60 69 73 79 85
2.5 note:
bank top -21 0 83 83 83 83 83 83 83 83 83 83
top -50 0 86 86 86 86 86 86 86 95 95
mid top -90 0 78 78 78 65 74 74 56 69 76
bottom -120 0 58.5 50 46 61 45.5 55 59 67
2.6 note:
bank top -28 0 83 83 83 83 83 88 88 88 93 103
top -40 0 92 92 92 92 89 94 94 117 122
mid top -80 0 83 83 83 66 82 92 92 92 78
bottom -110 0 72 74 58 54 81 73 73 73 80
3.1 note:
bank top -15 0 124 124 127 127 128 143 147 147 147 147
top -30 0 119.5 120.5 121.5 119.5 122.5 121.5 124.5 124.5 124.5
mid top -80 0.2 117.7 117.7 117.7 117.7 126.7 124.7 124.7 126.7 130.7
bottom -120 0 117 120 120 118 124 123 124 127 132
3.2 note:
bank top -11 0 132 137 142 142 142 142 142 137 137 137
top -30 0 109.5 106.5 106.5 106.5 110.5 110.5 110.5 110.5 60.5
mid top -70 0 104 104 104 112 110 110 110 115 115
bottom -120 0 104 102 102 102 102 104 108 108 108

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
194 Appendix 6

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
3.3 note:
bank top -10 0 175 175 175 175 175 278 313 313 313 313
top -40 0 173 173 173 173 253 265 259 259 259
mid top -80 0 170 170 170 170 255 260 270 270 280
bottom -110 0 173 173 173 173 272 259 259 259 259
4.1 note:
bank top -41 0 0 0 5 5 5 88 88 88 88 88
top -85 0 0 0 0 0 0 68 58 58 58 58
mid top -125 0 1.5 1.5 34 29.5 78.5 48.5 55.5 55.5 55.5 65.5
bottom -180 0 1 43 43 43 62 35 26 43 47 62
4.2 note:
bank top -6 0 0 65 65 65 65 110 110 110 110 110
top -48 0.8 8.5 66 63 63 63 68 89 89 89 89
mid top -93 0 9 44 30 51 37 36 36 36 36 36
bottom -138 -1.5 -14 17 22 12 30 11 15 23 23
4.3 note:
bank top -19 0 0 0 -10 -10 0 35 35 35 50 50
top -35 -2 14.5 17.5 20.5 11.5 16.5 91.5 83.5 85.5 87.5 95.5
mid top -60 0 3 95 82 86 162 163 173 165 155
mid bot -100 3 29 17 17 48 48 33 42 44
bottom -140 -76 -76 -76 -79 -73 -73 -73 -69
5.1 note:
bank top -4 0 96 111 106 106 106 134 136 131 173 177
top -15 0 111 122 113 113 158 160 160 160 160
mid top -75 0 114 114 117.5 117.5 162 163 176.7 179.7 183.7
bottom -135 -0.5 109.5 105.5 114.5 114.5 140.5 138.5 127.5 153.5 154
5.2 note:
bank top -22 0 9 9 9 9 9 9 9 9 9 9
top -75 -0.4 1 2 2 2 2 2 94 94 96 96
mid top -130 0 2 2 2 2 2 10 99 94 96 97
bottom -160 0 2.5 2.5 2.5 -0.5 -0.5 -0.5 89.5 89.5 91.5 91.5
5.3 note:
bank top -17 0 4 4 4 4 4 4 -1 -4 -4 -4
top -75 0 9 6 6 6 6 15 15 19 19 1.5
mid top -95 0 11 13 13 20 21 30 30 40.5 45.5 -3.5
bottom -150 0 2 2 2 0.5 -3 1.5 3 3 20.5 -25.5
6.1 note:
bank top -19 0 0 8 18 18 18 18 18 18 18 43
top -80 0 1 2 2 2 2 2 8 32 36 41
mid top -130 0 3.5 19.5 -4.5 24.5 -12.5 -2.5 -7.5 -7.5 23.5 30.5
bottom -160 0 2.5 1.5 5.5 5.5 11.5 -0.5 9.5 29.5 35.5
6.2 note:
bank top -39 0 0 0 15 15 15 15 0 35 35 70
top -60 0 2.5 2.5 2.5 1.5 -3.5 16.5 2.5 48.5 48.5 48.5
mid top -100 0 38 41 41 41 34 49 37 -20 -15 -3
bottom -150 0 8 51 10 10 10 17 4 35 30 30
6.3 note:
bank top -5 0 90 95 90 95 95 125 165 165 175 175
top -30 0 90 101 101 101 101 118 170.5 127.5 180 180
mid top -80 0 90 101.5 90.5 101.5 101.5 119.5 179.5 179.5 179.5 191.5
bottom -130 0 90 93 93 99 95 126 129 153 160 178
6.4 note:
bank top -23 0 127 135 130 130 130 130 133 135 140 200
top -50 0 127 136 136 136 136 140 140 140 140 194
mid top -110 0 127 132 132 132 132 132 149 175 176 198
bottom -150 0 127 139 149 149 149 163 172 200 212 221
6.5 note:
bank top -39 0 81 81 96 96 96 96 96 96 96 116
top -40 0 81 107 107 107 107 111 103 103 103 126
mid top -100 81 84 92 92 92 168 188 206 206 229
bottom -150 0 81 57 75 80 80 106 113 132 132 140

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
 Cumulative measured bank positions changes 195

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
6.6 note:
bank top -7 0 157 162 162 162 162 232 277 277 282 282
top -30 0 157 163 163 163 163 308 239 239 239 256
mid top -80 157 161 161 161 161 340 290 254 254 254
bottom -130 0 157 159 159 159 153 325 209 209 224 224
6.7 note:
bank top -5 0 160 160 160 160 160 380 380 380 380 380
top -40 0 160 163 163 163 167 338 338 328 336 336
mid top -110 160 156 156 156 156 165 159 179 154 154
bottom -150 0 160 189 189 176 163 116 116 covered covered covered
6.8 note:
bank top -11 0 160 165 175 175 175 247 247 252 270 270
top -30 0 160 177 165 165 180 248 248 248 259 266
mid top -80 160 158 168 168 176 219 219 242 214 214
bottom -120 0 160 156 188 170 170 245 240 232 216 216
6.9 note:
bank top -7 0 144 144 144 144 144 159 159 259 259 259
top -40 0 144 141 129 140 151 196 196 243 243 243
mid top -100 144 144 144 144 149 178 178 208 208 208
bottom -140 0 144 132 163 138 132 208 208 204 214 covered
6.91 note:
bank top -6 0 91 91 96 96 96 121 121 121 121 121
top -40 0 91 86 90 90 90 116 123 116 125 124
mid top -90 91 92 92 92 92 176 176 191 175 191
bottom -120 0 91 80 80 80 80 184 184 153 121 158
6.92 note:
bank top -5 0 67 72 82 82 82 92 127 132 132 132
top -20 0 67 67 72 72 72 138 122 114.5 126 123
mid top -80 68 71 71 71 71 142 122 113 123 127
bottom -120 0 68 65 65 65 59 146 117 107 112 150
7.1 note:
bank top -58 0 104 169 204 204 314 319 314 324 329 329
top -74 0 104 158 200 200 320 320 320 320 320 320
mid top -114 0 104 151 165 170 268 270 270 276 276 276
bottom -144 0 104 217 217 269 269 269 282 282 269
7.2 note: measurement error
bank top -26 0 190 231 281 281 501 506 516 521
top -69 26 216 256.5 322.5 322.5 532.5 540.5 535.5 533.5
mid top -109 40 230 299 334 378 464 478 457 464
bottom -129 40 230 312 356 418 427 420 428
7.3 note:
bank top -5 0 129 154 249 249 254 254 254 256 254 264
top -30 0 129 140 238 238 238 238 238 238 238 238
mid top -60 0 129 129 214 208 204 204 204 204 208 208
bottom -90 0 129 130.5 202.5 207.5 205.5 206.5 206.5 206.5 209.5 209.5
7.4 note:
bank top -4 0 79 99 141 141 134 134 144 154 154 154
top -30 0 79 109 132 137 150 142 142 151 151 151
mid top -70 0 79 120 142 146 146 142 142 142 153 161
bottom -100 0 79 130.5 137 137 130 130 140 157 163
7.5 note:
bank top -32 0 204 243 253 253 253 248 253 253 258 258
top -50 0 235 235 235 235 230 230 234 243 248
mid top -80 0 236.5 236.5 236.5 231.5 232.5 232.5 232.5 239.5 256.5
bottom -110 0 232 232 232 232 229 224 224 235.5 244
7.6 note:
bank top -40 200 275 281 281 281 281 290 290 290 290
top -60 200 270 270 270 260 266 266 266 266 266
mid top -90 200 257 251 251 258 252 261 261 261 261
bottom -110 200 230 230 230 230 231 231 231 237 237

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
196 Appendix 6

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
8.1 note:
bank top -32 0 10 10 15 15 15 70 70 70 85 85
top -80 0 8 8 11 11 11 5 12 12 12 18
mid top -125 0 32 32 28.5 32 32 79 79 79 85 85
bottom -160 0 0 0 0 41 41 46 46 46
8.2 note:
bank top -18 0 0 10 10 10 10 10 5 5 60 75
top -45 0 7 7 7 7 7 43 44 44 40 55
mid top -100 0 0 2 2 2 2 37 40 50 53 53
bottom -160 0 8 8 13 10 10 69 77 77 74 74
8.3 note:
bank top -14 0 183 183 183 183 183 183 183 183 183 209
top -70 0 183 183 183 183 183 183 183 183 190
mid top -105 0 185 185 185 185 185 185 191
bottom -220 0 181 176 172.5 171.5 171 170 167.5 175.5
9.1 note:
bank top -24 0 150 150 150 150 380 380 380 380 380
top -90 0 161 161 161 161 300 300 300 300 300
mid top -125 0 149 149 149 136 269 249 239 243 251
bottom -185 0 142 135 140 140 140
9.2 note:
bank top -7 0 150 148 148 148 148 150 150 145 145
top -50 0 151 151 151 151 153 153 153 153 153
mid top -100 0 153 141 134 126 157 157 152 129 163
bottom -120 -3.5 155.5 152.5 145.5 150.5 169.5 231.5 167.5 163.5 227.5
9.3 note:
bank top -16 0 150 152 152 152 152 152 152 152 152
top -40 0 53 53 53 62 61 61 57 57 57
mid top -100 0 56 56 56 56 52 97 49 31 31
bottom -140 -1 65 65 70 61 68 68 68 57 73
10.1 note:
bank top 0 0 0 0 0 0 0 3 3 3 3 3
top -60 0 1 2 2 2 2 2 2 3 4 4
mid top -110 0 4 4 4 4 4 4 4 8 8 12.5
bottom -160 0 4 6 7.5 13.5 31.5 34.5 35.5 35.5 39 42
10.2 note:
bank top 0 0 4 4 4 4 4 4 4 4 -11
top -65 0.4 1 2.5 2.5 2.5 2.5 2.5 2.5 1.5 6 8.5
mid top -110 0 2.8 0.8 -0.2 0.8 4.3 5.8 9.8 13.8 13.8
bottom -160 0 0.6 0.6 1.6 2.1 3.1 3.6 7.1 7.1
10.3 note:
bank top 0 0 13 13 13 17 17 1 1 1 1
top -70 0 0 0 0 0 0 0 0 6 6 6
mid top -110 0 0 0 0 0.5 12 12 12 12 14 14
bottom -150 -0.9 1 1 0 0 0 0 2 3 7 17
10.4 note:
bank top -19 0 23 123 123 138 163 262 263 263 263 263
top -50 0 0.3 100.3 125.3 126.8 173.3 207.3 207.3 205.3 218.3 218.3
mid top -90 0 0.5 100.5 146.5 144.5 175.5 201.5 203.5 205.5 207.5 210.5
bottom -120 0 3.5 103.5 158.5 158.5 158.5 168.5 185.5 200.5 193.5 191.5
11.1 note:
bank top -18 0 0 0 0 0 0 15 15 15 15 15
top -70 0 3 5 5 5 5 59 5 57 61 46
mid top -140 0 -2 -1 -1 -1 -1 64 64 60 57 57
bottom -180 0 4 4 2 2 0.5 84 64 67 85 54
11.2 note:
bank top -18 0 85 85 93 93 93 93 90 90 90 90
top -60 0 80 76 76 76 76 76 76 76 76 76
mid top -100 -1 39 33 33 33 33 33 33 33 33 33
bottom -150 -0.5 28 -47 -47 -47 -47 -49 -49 -49 -42 -45

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
 Cumulative measured bank positions changes 197

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
11.3 note:
bank top -22 0 1 1 3 3 3 3 58 58 58 58
top -60 0 4 5 5 5 5 23 34 34 34 50
mid top -130 0 40 53 49 49 49 74 37 77 70 70
bottom -180 0 40 40 30 34 34 39 54 37 58 46
11.4 note:
bank top -7 0 0 0 0 0 0 0 0 0 0 0
top -80 0 7 9 9 9 9 9 13 138 154 154
mid top -120 0 3.5 0.5 0.5 -1.5 -1.5 -1.5 1 121.5 140.5 140.5
bottom -175 0 3.5 5 5 6.5 6 8 5 74 89 53
11.5 note:
bank top -11 0 100 100 105 105 105 105 108 248 248 258
top -60 0 100 100 99 98 99 99 105 194 193 218
mid top -120 0 100 104 103 103 105 105 109 165 194 206
bottom -170 0 100 98 102 102 97 107 175 219 224
11.6 note:
bank top -14 0 125 131 133 133 133 290 290 290 290 290
top -56 0 125 139 139 139 139 230 236 236 259 259
mid top -116 0 125 124 137 137 137 232 232 227 246 250
bottom -156 0 125 129 129 126 133 218 255 262 256 262
11.7 note:
bank top -20 0 132 132 135 180 132 132 140 132 135 135
top -70 0 132 132 132 179 131 133 133 128 145 149
mid top -130 -1.5 130.5 106.5 106.5 154.5 105.5 106.5 112.5 129.5 84.5 107.5
bottom -150 0 132 132 126 174 128 126 133 111 98 133
11.8 note:
bank top 0 0 132 132 127 134 134 134 134 136 132 132
top -40 0 131 131 131 131 131 135 135 135 135
mid top -90 0 132 132 132 132 137 137 132 132 143
bottom -130 -1.5 121.5 121.5 118.5 122.5 116.5 115.5 110.5 113.5 113.5
12.1 note:
bank top 0 0 0 0 -1 -1 -1 -5 -11 -5 4 4
top -70 0 1 2 2 2 2 3 3 5 5 6
mid top -120 0 1 2 2 2 2 4 4 8.5 12 12
bottom -175 0 0 1.5 2.7 10 14 17 30 35 36.5 41.5
12.2 note:
bank top -7 0 0 0 0 0 0 92 90 90 113 113
top -55 0 33.5 33.5 33.5 33.5 33.5 163.5 163.5 169.5 170.5 171.5
mid top -95 0 35 35 36 36 36 160 160.5 167.5 167.5 169.5
bottom -190 0 35 38 38 36 38 161 164 170.5 168 173.5
12.3 note:
bank top -14 0 160 160 165 165 165 285 295 300 307 307
top -70 0 180 178 178 178 280 404 404 407 411 411
mid top -120 0 180 181 179 181 180 290 289 293 294 295
bottom -170 0 180 183 166 176 137 291 304 296 301 318
12.4 note:
bank top 0 0 150 155 160 160 160 215 220 225 225 250
top -55 0 151 151 151 151 216 219 232 238 240
mid top -115 -1 149 149 149 149 215 215 224 224 224
bottom -175 0 138 125 130 130 235 235 235 235 246
12.5 note:
bank top 0 0 212 209 235 235 235 302 300 355 365 364
top -60 0 216 216 226 218 319 319 345 345 345
mid top -140 0 213 213 213 213 306 306 321 321 321
bottom -190 0 205 205 187 202 323 303 331 340 340
12.6 note:
bank top -9 0 245 245 235 235 235 313 313 370 370 380
top -68 0 245 245 245 245 245 327 332 353 353 353
mid top -138 0 245 252 245 250 250 336 336 342 342 342
bottom -188 0 245 224 224 218 224 307 307 325 316 349.5

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
198 Appendix 6

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
13.1 note:
bank top -12 0 0 0 0 0 0 0 -8 -8 -8 -8
top -90 0 1.5 2 3 3 2 8 8 8 7 10
bottom -125 0 0 1 1 1 7 23 27 27 27.5 27.5
13.2 note:
bank top -8 0 0 0 0 0 0 35 40 50 63 63
top -30 0 3 4 4 4 4 55 55 68 68 68
mid top -80 0 1.5 4 4 4 4 30.5 54 64.5 64.5 64.5
bottom -135 0 1 1 2 2 3 68 62 74 75 78
13.3 note:
bank top -9 0 0 0 0 0 10 120 118 117 125 125
top -56 0 0 0 0 1 21 74 68 81 91 94
mid top -100 0 0 1.5 1.5 1.5 37.5 66 63 86 94 97
bottom -180 0 3 29.5 44.5 50 52.5 11 103
13.4 note:
bank top -4 0 0 0 0 0 0 40 38 75 80 85
top -40 0 4 7 7 7 7 62 56 71 74 76
mid top -95 0 2 4 4 4 4 43 43 61 64 64
bottom -160 -1 11.5 24.5 25.5 25.5 57 30.5 54.5 77.5 83.5 88.5
13.5 note:
bank top -23 0 0 0 1 1 1 1 1 11 11 11
top -65 0 5.5 4 4 4 4 3.5 3.5 3.5 3.5 3.5
mid top -95 0 6 7 7 8 8 10.5 11.5 14 16 16
bottom -150 0 3 5.5 13 12 12 12 14 28.5 36 36.5
14.1 note:
bank top -35 0 0 0 0 0 0 0 0 0 0 0
top -120 0 0 -4 -0.5 -0.5 -0.5 -0.5 -0.5 -1.5 1 0
mid top -210 0 2 2 2 3 3 3 2 2 3 3
bottom -260 0 2 4 4 4 4 4 4 8 8 16
14.2 note:
bank top -36 0 0 0 0 0 0 0 0 0 0 0
top -160 0 5 5 5 5 5 5 7 7 9 8
mid top -230 0 6.5 2.5 2.5 2.5 0.9 -7.5 2 0.5 2 2
bottom -255 0 1.5 1.5 -0.5 1 1 1.5 1.5 1.5 1.5 6
14.3 note:
bank top -50 0 0 0 0 0 0 0 0 0 0 0
top -175 0 5 4 8.2 6.5 3.5 18 18 16.5 18 18
mid top
bottom -220 0 -1 6 4 4 9 20 20 21 21 21
14.4 note:
bank top -63 0 0 0 0 0 0 0 0
top -100 0 7 6 8 10.5 10.5 21 21.5 20 21 21.5
mid top -130 0 19 18 18 18 19.5 21 24.5 26.5 30 39
bottom -175 0 3 4.5 6.5 6 6 10.5 8.5 15 18.5 20
15.1 note:
bank top -20 0 115 115 125 125 125 125 125 125 135 140
top -50 0 115 110 116 111 118 113 113 113 121 156
mid top -100 -1 114 92 89 93 94 96 91 97 110 149
bottom -160 0 115 113 108 108 101 107 113 113 105 153
15.2 note:
bank top -12 0 80 90 90 90 90 105 110 110 110 115
top -50 0 80 81 81 81 76 76 69 69 84 90
mid top -110 -0.8 79.2 87.7 65.2 87.2 87.2 84.2 72.2 89.2 83.2 105.2
bottom -160 0 80 97 81 71 71 58 58 45 57 83
15.3 note:
bank top -10 0 70 90 95 95 95 390 390 385 385 395
top -74 0 70 68 74 70 70 353 362 354 354 346
mid top -144 0 70 18 36 26 30 153
bottom -184 0 70 80 80 80 48

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
 Cumulative measured bank positions changes 199

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
15.4 note:
bank top -5 0 55 55 45 45 45 131 130 130 130 130
top -60 0 55 55 55 61 56 72 79 79 79 79
mid top -120 0 55 60 60 60 60 63 78 87 87 70
bottom -170 0 55 57 53 37 55 55 55 61 35
15.5 note:
bank top -24 0 78 78 78 78 78 88 88 93 93 93
top -50 1 79 89 89 89 89 90 90 90 97 100
mid top -100 0 78 87 81 78 78 87 77 100 108 108
bottom -160 -1 77 61 67 61 68 98 112 132 135
15.6 note:
bank top -28 0 40 40 42 42 42 42 52 37 37 37
top -70 0 20 22 22 22 22 25 25 29 29 29
mid top -120 0 11.5 4.5 5.5 4.5 0 9.5 9.5 19.5 19.5 24.5
bottom -170 0 11 1 -6 -7 34 34 36 46 46
16.1 note:
bank top -5 0 0 0 0 0 0 0 0 0 0 0
top -65 0 6 6 6 6 6 7 7 5 9.5 10.5
mid top -110 -8 -4 -4 -4 -4 -2.5 -2.5 -2.5 -4.5 -4.5 -2
bottom -155 -2 9 5 5 4 3 3.5 3.5 3.5 7.5 7.5
16.2 note:
bank top -7 0 87 87 92 92 92 122 122 127 127 127
top -40 0 87 85 85 85 94 111 111 111 111 111
mid top -100 0 87 90 90 90 90 92 92 82 82 82
bottom -140 0 87 91 91 97 84 89 89 94 86 92
16.3 note:
bank top -12 0 80 80 80 80 80 80 85 80 80 80
top -40 0 80 80 80 80 80 87 87 87 87 87
mid top -130 0 80 80 80 80 89 90 90 90 94 102
bottom -170 0 80 76 76 56 75 75 75 75 95
16.4 note: no stakes
bank top -17 0 92 92 97 97 97 137 137 137 137
top -40 0 90 90 90 96 92 106 106 106 106
mid top -110 0 92 92 92 92 92 97 97 110 113
bottom -150 8 99 100 100 100 106 103 118 124
16.5 note:
bank top -10 0 0 0 3 3 3 3 13 8 28 28
top -55 0 6 8.5 8.5 8.5 8.5 8.5 8.5 17.5 25.5 25.5
mid top -120 7.5 20 29.5 29.5 29.5 29.5 27.5 27.5 53.5 64.5 69.5
bottom -150 8 38 51 41 41 40 40 74 74 60
17.1 note:
bank top -52 0 0 0 1 1 1 1 1 1 1 1
top -105 0 18 17 17 17 17 15.5 17 15 15 15
mid top -150 0 1 2.5 2.5 2.5 2.5 3 3 9.5 12.5 12.5
bottom -180 0 2 2 0 0 2 2 6.5 6.5 5.5
17.2 note:
bank top -16 0 0 0 0 0 0 0 5 5 5 5
top -60 0 2 2.5 2.5 2.5 2.5 34.5 34 34 35 35
mid top -100 0 8 2 1 3 16 16.5 16.5 28 28 31
bottom -150 0 15 34.5 34 34 59 61 57 75.5 75.5 77
17.3 note:
bank top -7 0 46 46 56 56 211 211 211 211 211 211
top -70 0 48 106 106 106 167 204 206 206 206 206
mid top -130 0 25 52 52 52 116 80 83 83 88 88
bottom -180 0 38 48 62 87 87
17.4 note:
bank top -14 0 92 127 157 157 287 287 287 287 312 312
top -50 0 92 162 178 178 315.5 318 322 327 327 327
mid top -110 0 92 149 236 236 283 271 271 271 271 283
bottom -150 -3 89 238 245 206 208 213 226 214 235

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
200 Appendix 6

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
17.5 note:
bank top 0 0 200 200 200 200 200 557 560 560 560 560
top -60 0 203 203 203 203 526 549 549 549 556
mid top -110 0 202 202 202 207 485 492 492 505 505
bottom -160 0 200 200 212 210 444
17.6 note:
bank top -18 0 215 240 230 235 435 555 555 560 575 580
top -60 0 242 250 250 267 566 579 595 595 604
mid top -110 0 247 270 270 271.5 504 555 563 592 575
bottom -160 0 248 249 250 459 489
18.1 note:
bank top 0 50 50 50 50 50 50 50 50 50 50
top -5 45 44.5 44.5 44.5 46 44.5 41.8 44 43
mid top -2 48 47.5 47.5 47.5 51.5 51.5 50 51 50
mid bot -1 49
bottom 0 5 4 5 -0.2 0.5 0.5 -2.5 -2.5 -2.5
18.2 note:
bank top 0 0 0 2 15 15 15 15 5 5 5
top -0.2 3 4 2.5 4 5 5 5 7 6 6
mid top -1 1 -1 0.5 -0.5 2 0.5 -3.5 -3 -3 -3
mid bot -0.5 6 6 7 7 5.3 5 -2.5 1 2 3
bottom 0 12 11.5 9 9 9 7.5 7.5 5 5 5
18.3 note:
bank top 0 50 50
top -1.5 48.5
mid top 30 80
mid bot 30 80
bottom -3 47
18.4 note:
bank top 0 0 0 0 0 0 0
top 0 1.5 1.5 1.5 1.5 1.5 41.5
mid top -1 -1 -0.5 -4.5 2 19.5
mid bot 0 6 6
bottom -1 -10 -9.5 12.3 19.8
18.5 note:
bank top 0 0 0 0 0 0 0 0 0 0 0
top -1 2 -5 -10 -12 -10.5 -5 -3.5 -5 -5 -12
mid top -0.5 10 5 0 -7.5 0.3 12.3 12.8 12.8 12.8 10.8
mid bot -0.5 15 14.5 9 5 6.2 24 23 24 26.5 25
bottom 0.5 9 9.2 -1 -4.2 7.5 9 11.5 12 14
18.6 note:
bank top 0 0 0 0 0 0 0 0 0 0 0
top -1 5 7 5.5 7 7 7 7 4.5 6 6
mid top 0 8.5 9 9 9 9 10 10 6.5 13 13.5
mid bot 0 20 18.5 20.5 21 20 22 22 25.5 28 29.5
bottom 0 9 12.5 12 11.5 11 12.5 12.5 19 26 28
18.7 note:
bank top 0 0 0 20 20 20 20 20 20 20 20
top -0.5 11 0 0 0 0 4 4 2.5 3 5
mid top -1 8 -3 0.5 -2 -2 3 4.5 4.5 5.5 27.5
mid bot 0 13 -2 -2 -2 -2 4.5 3 11 13.5 20
bottom 0 18 -1 0 0 4 3 13 17 19.5
19.1 note:
bank top 0 0 0 5 9 9 9 9 9 9 9 9
top -40 0 3 1.5 1.5 1 1.4 1.5 1.5 0.5 1.5 1.5
mid top -85 0 0.5 0.5 0.5 0.5 0.5 0.5 -1 -2 -1 -0.5
bottom -150 0 0.5 2 2 2 2 2.5 2.5 1.5 4.5 4.5
19.2 note:
bank top 0 0 0 0 35 35 35 35 35 35 35 35
top -60 -4 7 6.5 6.5 6.5 16.5 24.5 23.5 19 21 23.5
mid top -110 -1 4 4 3 0.3 0.3 4.5 3 3 3 4

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
 Cumulative measured bank positions changes 201

date depth 04.12.03 28.01.04 18.03.04 24.05.04 26.07.04 30.08.04 09.11.04 17.01.05 28.02.05 11.04.05 26.09.05
bend
bottom -140 0 14.2 14.2 13.7 12.2 11.6 18.2 16.2 16.7 21.2 21.7
19.3 note:
bank top -5 0 0 0 0 5 5 20 40 40
top -50 -12 -2 -1 0 19 41 38 35 57
mid top -85 -14 22 -12 -7 40 25 59 38 42
bottom -125 -1 20 -3 1.5 29 4 66 44 53
19.4 note:
bank top 0 0 0 0 0 0 3 3 2 2 2
top 0 9.5 11.5 11.5 11.5 11 11 11.5 9.5 16.5 20.5
mid top 0 1 2.5 2.5 2.5 3 3 3.5 8.5 7.5 13
bottom -1 -0.5 -4.5 -3 -3.5 -5.9 -5.9 -5.4 -6.4 -5.4 -6.4
20.1 note:
bank top -19 0 0 0 0 0 0 0 0 0 0 0
top -95 0 -4 -4 0 0 48 43.5 46 41.5 42 42
mid top -120 0 3 4 4 18 46 41.5 44 41 42 42
bottom -180 0 -7.5 -7.5 -7.5 15.5 27 20.5 24.5 22.2 24 28
20.2 note:
bank top -21 0 62 62 62 62 62 72 72 72 72 72
top -80 -8 -9 -8 -7 -8.8 -9 -10 -13 -13 -18
mid top -150 -5.5 -5.5 -5.5 -9.5 -5.5 -9.5 -11 -14.5 -13.5 -13.5
bottom -205 -3.5 12 13.5 16 14 12.5 12 11 11.5 13.5 14.5
20.3 note:
bank top -50 0 0 0 10 10 10 3 3 3 3 3
top -90 0 7 8 7 8 9.5 10 11.5 16 16 19
mid top -120 0 4 5 4 4 4 6 8.5 15 21.5 24.5
bottom -150 0 2.5 3 8 19.5 25.5 27 27.5 26 23.5

Table A.1. Cumulative bank position change. Medium gray shaded cells indicate bank failures (>20
cm).
202 Appendix 7

Appendix 7 Streambank erosion danger maps of the Venoge River

Streambank Erosion
Danger Level

Low

Medium

High

Length of danger bar shows

mean annual erosion
exaggerated by 20 times

Extreme danger limit

Venoge River
2004 centerline

Bank protection work

0 100 200

meters

Figure A.8. Streambank erosion danger map of the Venoge River (bank protection failure probability
not taken into account)
 Streambank erosion danger maps of the Venoge River 203

Streambank Erosion
Danger Level

Low

Medium

High

Length of danger bar shows

mean annual erosion
exaggerated by 20 times

Extreme danger limit

Venoge River
2004 centerline

Bank protection work

0 100 200

meters

Figure A.8 (cont.). Streambank erosion danger map of the Venoge River (bank protection failure
probability not taken into account)
204 Appendix 7

Streambank Erosion
Danger Level

Low

Medium

High

Length of danger bar shows

mean annual erosion
exaggerated by 20 times

Extreme danger limit

Venoge River
2004 centerline

Bank protection work

0 100 200

meters

Figure A.8 (cont.). Streambank erosion danger map of the Venoge River (bank protection failure
probability not taken into account)
 Streambank erosion danger maps of the Venoge River 205

0 100 200

meters
Streambank Erosion
Danger Level

Low

Medium

High

Length of danger bar shows

mean annual erosion
exaggerated by 20 times

Extreme danger limit

Venoge River
2004 centerline

Bank protection work

Figure A.8 (cont.). Streambank erosion danger map of the Venoge River (bank protection failure
probability not taken into account)
206 Appendix 7

0 100 200

meters
Streambank Erosion
Danger Level

Low

Medium

High

Length of danger bar shows

mean annual erosion
exaggerated by 20 times

Extreme danger limit

Venoge River
2004 centerline

Bank protection work

Figure A.8 (cont.). Streambank erosion danger map of the Venoge River (bank protection failure
probability not taken into account)
 JOHN RAYMOND BECK

P E R SO N A L I N F O R M AT I ON
Marital status: Married, three children
Nationality: American
Date of birth: February 15, 1971 at Cedar Rapids, Iowa, USA

E D U C A T I ON
DOCTORAL THESIS 2002-2006
S WISS F EDERAL I NSTITUTE OF TECHNOLOGY, EPFL, L AUSANNE , S WITZERLAND
Dissertation: “Streambank Erosion Hazard Mapping: Concepts, Methodology and Application on the
Venoge River (Switzerland)”

POST-GRADUATE SPECIALIZATION IN HYDROLOGY 1998-1999

S WISS F EDERAL I NSTITUTE OF TECHNOLOGY, EPFL , L AUSANNE , S WITZERLAND
Master’s thesis: “Methodology for the Evaluation of Flood Hazards: Application on the Seymaz
Watershed.”

BS CIVIL ENGINEERING WITH HONORS AND HIGHEST DISTINCTION 1989-1993

U NIVERSITY OF I OWA , I OWA C ITY, I OWA
G PA 4 .0 0/ 4. 0 0
Honors Paper: “The Effects of Long Combination Vehicles on Rigid Pavement Damage”

HIGH SCHOOL DIPLOMA 1985-1989

I OWA M ENNONITE S CHOOL , K ALONA , I OWA
Valedictorian

COMPUTER EXPERIENCE:
Microsoft Office, Adobe Framemaker, Canvas, MapInfo, HEC-RAS, HEC-HMS, Visual OTTHYMO
Programming experience in Pascal, Fortran, and Delphi.

LANGUAGES:
English (mother tongue)
French (fluent)
Chadian Arabic (medium speaking abilities)
Spanish (medium written skills)

P R O F ES S IO N A L E XP E R IE N C E
HYDROLOGY AND LAND IMPROVEMENT LABORATORY 1/2000-6/2006
S WISS F EDERAL I NSTITUTE OF TECHNOLOGY, EPFL , L AUSANNE , S WITZERLAND
Assistant researcher on the following projects:
• Quantitative water management of the Seymaz and its tributaries
• Flood Risk Scenarios and Hydrological Mapping (FRHYMAP), IRMA-SPONGE
• Flood Analysis and Management of the Orlice River (FLAMOR), Czech Republic
• Strategies and Actions/Implementations for Flood Emergency Risk Management (SAFER)

MENNONITE CENTRAL COMMITTEE 6/1994-6/1998

W INNIPEG , M ANITOBA , C ANADA
Volunteer worker in Chad, Africa
Managed a team of 11 people doing community development work: building wells and micro-dams
and pumps, promoting proper water hygiene, managing cereal banks, and creating a public
library.
C O N N I E B E L I N C E N T E R,
SUMMER INSTITUTE FOR C REATIVE E NGINEERING AND INVENTIVENESS 5/93-9/93
U NIVERSITY OF I OWA , I OWA C ITY, I OWA , USA
Coordinated summer 3 week camp on hydropower for 30 gifted high school students

B A R R E N G I N E E R I N G C O M PA N Y 5/1992-8/1992
M INNEAPOLIS , M INNESOTA , USA
Assisted surveyors during summer internship

A C T I V IT I ES /H ON O R S
American Society of Civil Engineers University of Iowa chapter vice president (spring ’93) and
executive committee (Fall ’93)
Chi Epsilon University of Iowa chapter president (’93) and vice president (Fall ’92)
Inducted as a member of Tau Beta Pi (Fall ’92)
Student Body President of Iowa Mennonite School (‘88-89)

P U B L IC A T I ON S
Beck, J. (2005). Streambank erosion hazard mapping on the Venoge River, Switzerland. Increasing
the Awareness of Flooding to Improve the Risk Management of Flood Events, Heidelberg,
Germany.
Beck, J., R. Metzger, et al. (2001). FRHYMAP - WP3: Security Deficit Risk Analysis on the Upper
Alzette River - Luxembourg. Lausanne, EPFL: 17 pp. + appendices.
Beck, J., R. Metzger, et al. (2002). Flood risk analysis based on security deficits. Experimental
Hydro-climatological Atlas of the Alzette River Basin - Grand-duchy of Luxembourg. L.
Pfister and L. Hoffman. Luxembourg: pp. 43-55.
Beck, J., R. Metzger, et al. (2002). Flood risk assessment based on security deficit analysis.
Geophysical Research Abstracts. 27th General Assembly of the European Geophysical
Society, Nice, France.
Beck, J., R. Metzger, et al. (2002). Two Dimensional Modeling of Floods in an Urban Environment.
The International Conference Preventing and Fighting Hydrological Disasters, Timisoara,
Romania.
Beck, J., P. Valenta, et al. (2003). Flood Analysis and Mitigation on the Orlice River. Final Report.
Lausanne, EPFL, CTU, VUMOP, IH: 145.
Beck, J., P. Valenta, et al. (2003). "Flood Analysis and Mitigation on the Orlice River (FLAMOR)."
ERWG Letter 14: 8-11.
Consuegra, D., J. Beck, et al. (2000). Gestion quantitative des eaux de la Seymaz et de ses
affluents: Etablissement de scénarios de gestion. Lausanne, IATE/HYDRAM.
Consuegra, D., J. Beck, et al. (2000). Gestion quantitative des eaux de la Seymaz et de ses
affluents. Etude Hydrologique et Hydraulique. Diagnostic de l'état actuel des dangers liés
aux inondations et à l'érosion. Lausanne, IATE/HYDRAM.
Metzger, R., A. Musy, and J. Beck (2001). Flood Risk Procedures and Management: A
comprehensive approach based on the Swiss experience. 19th European Regional
Conference of ICID on Sustainable Use of Land and Water, Brno and Prague.