Channel Disturbance and Evolution:
Controls and Implications for Stream
Restoration
Andrew Simon
USDA-ARS National Sedimentation Laboratory, Oxford,
MS, USA
National Sedimentation Laboratory
Introductory Points
• Alluvial streams are open systems that dynamically
adjust to variations in flow energy and sediment supply.
• Streams adjust their morphology to imbalances between
available force and sediment supply as a function of the
resistance of the boundary sediments to hydraulic and
geotechnical forces.
• Thus, two channels of similar morphology disturbed by
an identical perturbation can attain different equilibrium
morphologies
• Also, diverse streams subject to diverse perturbations
can respond similarly.
National Sedimentation Laboratory
CW
Impetus for Restoration Efforts
• Major land-clearing activities between the mid-1850’s and early
1900’s to bring land into agricultural production.
• Soil-conservation techniques were not used/available.
• Massive erosion from fields and uplands in many areas,
particularly the mid continent.
• Channels filled with eroded sediment causing reduced conveyance
and increases in the frequency and duration of flooding.
• Large-scale programs to dredge and straighten many fluvial
systems to improve land drainage and reduce flooding.
• Resulting channel instabilities caused incision and massive
erosion of main stem and tributary streams (valley-fill deposits ).
• Erosion from channel systems has become the dominant source of
sediment in many (most) of these watersheds.
• Clean-Water Act, TMDLs, Rosgen
Gravity is A Constant!!
• The physics of erosion are the same wherever you
are…no matter what hydro-physiographic province you
are in…whatever the stream type may be.
• Channel adjustment is driven by the imbalance
between the driving and resisting forces
• Differences in rates and magnitudes of adjustment,
sediment transport rates and ultimate channel forms are
a matter of defining those forces…deterministically or
empirically
Incision enhances channel response by creating flows
with greater transporting power
Case Study: Coastal-Plain System
Obion Forked-Deer River Basin
Downstream, anthropogenic
disturbance in a sand-bed, cohesivebank system causing an increase in
transport capacity (gQS)
Modified from Lutenegger (1987)
Adjustment Processes
Mississippi
Tennessee
Mississippi
Nebraska
Case Study: Sub-Alpine System
Upstream “natural” disturbance in
a coarse-grained, non-cohesive bank
system causing an increase in
transport capacity (gQS) and a
decrease in resistance (d50)
Adjustment Processes
Trends of Bed-Level Change
Elk Rock Reach
Salmon B Reach
Mt. St Helens
W. Tennessee
Function for Degradation/Aggradation
z / zo = a + (1-a) e (-k t)
z = elevation of the channel bed at time t,
z0 = elevation of the channel bed at t0 = 0,
a = dimensionless coefficient determined by regression
equal to z/z0 when equation becomes asymptotic,
1-a = total change in dimensionless elevation,
k = coefficient, determined by regression and indicative of
the rate of change on the channel bed per unit time,
t = time in years since the onset of the adjustment process.
a > 1, aggradation; a < 1, degradation
Trends of Bed-Level Change
Trends of Bed-Level Change
Coarse-grained material
for aggradation derived
from bank sediment.
Widening
Incision creates the
conditions for bank
instability and widening
by creating higher,
steeper banks
But why are
they so
different ?
Resistance
2,500 km of streams
in W. Iowa (1993-4)
Data from Hadish (1994)
FREQUENCY OF OCCURRENCE,
IN PERCENT
Idealized Adjustment Trends: Mid Continent
70
60
6% stable
50
40
80% with
unstable
banks
30
20
10
I
II
III
IV
STAGE
V
VI
Bed-Level
Adjustment:
Arno River, Italy
LOWER VALDARNO
Cross-section: 284
River kilometer: 59.01
ELEVATION, IN METERS
12
Pisa Plain
Lower Valdarno
Florence
Plain
10
Upper
Valdarno
Phase I
10
9
Phase II
1
8
7
2
3
4
8
6
1840 1860 1880 1900 1920 1940 1960 1980 2000
1845-1978
YEARS
1952-1978
6
UPPER VALDARNO
Cross-section: 867
River kilometer:160.54
1845-1987
4
136
Phase I
135
ELEVATION, IN METERS
CHANNEL BED-LEVEL LOWERING,
IN METERS
12
11
2
1845-1952
1960-1987
0
1845-1960
-2
0
20
40
60
80
100
120
DISTANCE FROM MOUTH, IN KILOMETERS
140
160
134
133
Phase II
132
1
131
2
130
3
129
4
128
1840 1860 1880 1900 1920 1940 1960 1980 2000
YEARS
Phases of Degradation Since 1900
Phase I: Land use changes with a reduction in sediment supply
Phase II: Gravel mining and upstream dam construction
12
136
Phase I
10
9
Phase II
1
8
7
Phase I
135
11
2
3
4
6
1840 1860 1880 1900 1920 1940 1960 1980 2000
YEARS
ELEVATION, IN METERS
ELEVATION, IN METERS
UPPER VALDARNO
Cross-section: 867
River kilometer:160.54
LOWER VALDARNO
Cross-section: 284
River kilometer: 59.01
134
133
Phase II
132
1
131
2
130
3
129
4
128
1840 1860 1880 1900 1920 1940 1960 1980 2000
YEARS
Boundary Resistance and Channel Response
• General trends of channel response to disturbance (channelization
and reduction of sediment supply) provide only a semiquantitative view of how different disturbances can cause
similar responses.
• Similar channels may respond differently as a function of the
relative and absolute resistance of the boundary (bed and
banks) to hydraulic AND geotechnical forces
• Alluvial-channel response has been defined by many with nonlinear decay functions that become asymptotic and reach
minimum variance with time.
Minimization of Energy Dissipation
Channels adjust such that their geometry provides
for a minimum rate of energy dissipation given the
constraints of the upstream sediment load,
roughness and resistance of the boundary materials
If this holds true, then we should be able to track
this over time in disturbed, adjusting streams
Flow Energy and Energy Dissipation
E = z + y + v2/2g
hf = (z1 + y1 + v12/2g) - (z2 + y2 + v22/2g)
hf
v12/2g
v22/2g
y1
y2
z1
z2
1
L
Energy slope: Se = hf / L
2
Processes That Effect Components of
Total Mechanical Energy (E)
For each parameter comprising E,
what processes would result in a
reduction in those values?
• z:
degradation
• y:
widening, aggradation
• v2/2g: widening, increase in relative roughness,
growth of vegetation, aggradation,
Thus, different and often opposite processes can have
the same result
Adjustment by Different Processes
Aggradation and widening
Degradation and widening
Importance of Widening in Energy
Dissipation
• Salmon B reach: aggradation and widening
• Elk Rock reach: degradation and widening
Differences in
sediment supply
1. Reduces flow depth (pressure head) for a given flow;
2. Increases relative roughness, and therefore,
3. Reduces flow velocity (kinetic energy);
4. Combined with degradation (potential energy) is the most
efficient means of energy reduction because all components
of E are reduced;
5. Counteracts increase in potential energy from aggradation
Effect of Bank Materials on Incision
• Assume that gQS a Qsd50 is balanced
• How does a channel respond if disturbed?
• Will the channel incise?
• Will the channel fill?
• Will the channel widen?
• Will the channel narrow?
• Will it equilibrate to the same geometry?
Provides Only Limited Insight
gQS a Qsd50
g = unit weight of water
Q = water discharge
S = bed or energy slope
Qs = bed-material discharge
d50= median particle size of bed material
Where will erosion occur?
How will channel form change?
Simulated using a numerical model of bed deformation and
channel widening (Darby, 1994; Darby et al., 1996)
Disturbing a Sand-Bed Channel
• Assume that gQS a Qsd50 becomes un-balanced
• Qsd50 = 0.5 * capacity
• Slope = 0.005
• Initial width/depth ratio = 13.5
Bank material
Sand
Silt
Clay
Bed d50
(mm)
1.0
1.0
1.0
Bank cohesion
(kPa)
4.0
7.5
40.0
Friction angle
(o)
32.5
32.5
32.5
Sand content
(%)
100
20
10
Energy Dissipation for Different Boundary
Materials
0
100
0
200
300
400
0
100
200
300
400
DAYS FROM START OF SIMULATION
0.00005
0.0005
0.005
0
100
200
300
@ 0.90
@ 0.87
@ 0.80
Energy adjustment is similar at given slope because of an equal,
but excessive amount of flow energy relative to sediment supply.
Do each of these channels reach equilibrium similarly?
400
Styles of Adjustment for Different
Boundary Materials
DAYS FROM START OF SIMULATION
Disturbance: Upstream sediment supply cut in half
Adjustments for Different Boundary
Materials
Response to similar disturbance: Sediment supply = 0.5 * capacity
From Simon and Darby (1997)
Response with Different Bank Materials
• How does the channel respond?
It depends!
• How much will the channel incise*? 0.4 – 3.5 m
• How much will the channel widen*? 0 – 13 m
• What is the stable W/D ratio*? 5.6 – 16.4
* For a given initial slope of 0.005 m/m
Can A Form-Based Design System
Address these Issues?
• To many, the Rosgen classification and associated “natural
channel design” have become synonomous with the terms
“stream restoration” and “fluvial geomorphology”
• Is required for many restoration proposals, job applications
etc.
• Empowerment of individuals, groups and agencies with
limited experience in watershed sciences to engineer wholesale
re-patterning of stream reaches using technology never
intended for engineering design
• Many of these projects are being implemented across the
country with varying degrees of success…
Uses and Misuses: Classification vs.
“Natural Channel Design”
Uses
• Characterize the SHAPE and the average composition of boundary
sediments of stream reaches
• As a communication tool for the above
• Classification is rapid and easy to perform
Misuses
• Predicting river behavior…processes
• Engineering design in disturbed systems
• Use of a single discharge (bankfull)
• Ignores temporal and spatial scales
• Ignores processes and the concept that rivers are dynamic and part
of open systems
Problems with Application of Classification
Definition of bankfull level, particularly in unstable
systems.
“Bankfull” discharge and the dimensions
represented by hydraulic geometry refer to
stable channels. In unstable channels, they are
changing with time.
Problems with Application of Classification
Inconsistent determination of stream type among
observers with no clear guidance for determining
stream type when more than one was possible
Entrenchment
“…the classification system … appears to do
little to improve communication among
practitioners beyond what the raw measures
of channel attributes would have done.”
Roper et al., (2008), in press, JAWRA
10
9
8
7
6
5
4
3
2
1
0
“Rosgen A stream type that 5 out of 8
times was misclassified by observer
measurements as a B channel type”
From Roper et al., (2008), in press
ey
isk
Wh k
Lic
WF
il
Tra
r
ke
Tin us
t am
Po
rtle
My
ian
Ind ish
f
aw
Cr
e
an
Cr s
mu
Ca
e
idg
Br
Big
Implications for a Form-Based System:
Bed or Channel Material?
(two different populations)
National Sedimentation Laboratory
CW
Channel/Bed Material
From Rosgen (1996); Fig. 5-3
CW
However…
From Rosgen, 1996; Fig. 5-2
CW
Avoiding the Problem??
Rosgen (2006)
“Streambanks generally make up five percent or less of the
channel boundary…This would avoid the problem…”
True if width/depth (W/D) ratio is about 40 or greater. However,
they comprise 29% for example if W/D = 5. For example, using
guidance to proportionately sample pools and riffles (p. 5-27,
Rosgen, 1996):
Why is this Important?
• Sites may not classified correctly: Example:
“C” channel shape: gravel bed, silt/clay banks
“C” channel shape: sand bed, sand banks
C5
These two C5’s represent very different transport regimes
• As we have seen differences in bank materials are critical to
predicting channel response and stable geometries
• Particle-size data cannot be used for incipient motion or
transport analysis
• Extensive data sets collected by various agencies cannot be
used for analysis of hydraulic erosion, geotechnical
stability, or channel response
CW
Perhaps Explains Why…
“The consequence of a wide range of
stream channel instability can be
described and quantified through an
evolution of stream types (Figure 1).”
Rosgen (2001)
•
•
•
•
From Rosgen (2001)
E to E
C to C; C to Bc; C to D
B to B
Eb to B
CW
Forcing a Form-Based System to Describe Process
“The consequence of a wide range of
stream channel instability can be
described and quantified through an
evolution of stream types (Figure 1).”
Rosgen (2001)
•
•
•
•
E to E
C to C; C to Bc; C to D
B to B
Eb to B
Can this be predicted a priori?
Can this truly be quantified?
CW
From Rosgen (2001)
If not, what does this mean for the
“Natural Channel Design”
approach since “…stream
classification does not attempt to
predict…stability…” Rosgen (2006)
And Aren’t Most of These Similar?
+
From Rosgen (2001)
Incision
Widening
Filling
From Rosgen (2006)
Implications for Sediment TMDLs
• What are background, natural, stable rates of
sediment transport/ bed-material characterisitics?
• We must be able to discriminate between stable
and unstable conditions to determine departure
from natural or background conditions
A Rapid Means of Evaluating
Thousands of Streams was Needed
The very popular Rosgen Classification offers one
such means of rapidly classifying streams
• easy to understand
• novices can perform
• excellent communication tool about channel form
We don’t have the time or the money to perform detailed
analyses at every site that needs to be evaluated and
that may require a TMDL
Still, a scientifically defensible procedure is required
Stream Types With No “Reference”
Condition for Sediment TMDLs
• D: Active lateral adjustment with abundant sediment
supply….aggradational processes…high bedload and
bank erosion (Rosgen, 1996).
• F: Entrenched…laterally unstable with high bankerosion rates (Rosgen, 1996).
• G: Gullies…deeply incised…unstable with grade control
problems and high bank erosion rates (Rosgen, 1996).
• Thus, alluvial stream types D, F and G have no
REFERENCE condition for sediment transport and
sediment TMDLs
“…stream classification does not attempt to
predict…stability…
Rosgen (2006)
“A two-week course is required to teach professionals (including
individuals who have graduated from college with advanced
degrees in engineering, geology, hydrology, fisheries, etc.) how to
conduct a watershed and stream channel stability analysis”.
Wow…
Consider This…
“Concepts that have proved useful in ordering things
easily achieve such authority over us that we forget
their earthly origins and accept them as unalterable
givens…The path of scientific progress is often made
impassable for a long time by such errors.”
Einstein, 1916
So, What to Do? Potential Approaches
• Empirical (including “Natural Channel Design”):
regime equations; not cause and effect; time
independent
Morphology related to discharge (hydraulic geometry) etc.
Can address tractive force and bed-entrainment issues
It’s aIgnores
big toolbox!
Use what
is appropriate
for the scale
bank processes,
flow variability
and sediment
contribution
banks Approaches are NOT
and objective
of thefrom
project.
mutuallybased;
exclusive!
• Deterministic: physically
cause and effect
Quantifies driving forces and resistance of boundary sediments
to the appropriate processes and functionally linked to upland
delivery of flow and sediment.
This is Our World of Disturbed Systems
1953
Hotophia Creek, MS
System-wide disturbance to
channel system caused by
lowering of the water surface of
the trunk stream due to dam
closure
How do you analyze this
system? Do you need to
consider dynamic processes
with time? Will a “reference
reach” approach be
appropriate?
1977
Dynamic System?
Unstable reach
“Reference reach”
Unstable reach is 100 m from “reference reach
How would a practitioner address this situation for a potential mitigation
project?
A Tiered Approach
• Reconnaisance Level:
1. Use form to define dominant processes and relative
stability. Determine if the instability is localized or
systemwide (scope) from rapid geomorphic assessments
(RGAs), BEHI, gauging station records, dendrochronology, air photos. Identify the problem not just the
symptom.
2. Use regional, historical flow and sediment transport
data to define transport conditions (rates) for stable
streams in the region.
A Tiered Approach
If the problem is localized (ie. Bridge constriction; local
structure; livestock impacts; deflected flow) the practitioner
has more options, including a “reference-reach” approach.
But you can just as easily use a deterministic approach that is
based on implicitly analyzing the specific processes (ie. bank
instability).
Quantify the Variables that Control the
Processes
Applied (Driving) Forces
versus
Resisting Forces
•Hydraulic processes (bed, bank toe)
• Geotechnical processes (bank mass)
We have the tools.
It’s not hard or very time consuming!
Bank-Stability Model
• 2-D wedge-failure model
• Incorporates both positive
•
Tensiometers
(pore pressure)
Confining
pressure
1.6
1.5
85
Factor of safety
Effect of confining pressure
B
84
1.4
83
Bank failures
1.3
1.2
82
1.1
1.0
0.9
12/29/97
81
Stage
01/05/98
01/12/98
01/19/98
01/26/98
80
02/02/98
RIVER WATER
STAGE, INLEVEL,
METERSM
ABOVE
SEA LEVEL
•
FACTOR OF SAFETY
•
and negative pore-water
pressures
Simulates confining
pressures from stage
Incorporates layers of
different strength and
characteristics
Inputs:
gs, c’, f’, fb , h, uw,
k, tc
shear surface
Hydraulic Bank-Toe Erosion
6.00
‘Toe Erosion Step 2’ worksheet
5.00
Results
Base of layer 1
4.00
ELEVATION (M)
Base of layer 2
Base of layer 3
3.00
Base of layer 4
2.00
Base of layer 5
Eroded Profile
1.00
Water Surface
Initial Profile
0.00
-1.00
0.00
1.00
2.00
3.00
4.00
STATION (M)
5.00
6.00
7.00
Click this button to export eroded profile to Option
A in Input Geometry worksheet
Planar Failures
Factor of Safety
Conditionally
stable
Failure width
Failure volume
Sediment loading
Constituent load
2.35
487
814176
814
Partly controlled by failure plane angle
Based on reach length
m
m3
kg
kg
Based on constituent concentration
Select material types, vegetation cover and water table depth below bank top
(or select "own data" and add values in 'Bank Model Data' worksheet)
Layer 1
Layer 2
Gravel
Angular sand
Rounded sand
Silt
Stiff clay
Layer 3
Gravel
Angular sand
Rounded sand
Silt
Stiff clay
Layer 4
Gravel
Angular sand
Rounded sand
Silt
Stiff clay
Gravel
Angular sand
Rounded sand
Silt
Stiff clay
Layer 5
Gravel
Angular sand
Rounded sand
Silt
Stiff clay
Bank top
vegetation cover (age)
None
Vegetation safety margin
50
Reach Length
(m)
100
Constituent
concentration (kg/kg)
0.001
Water table depth (m) below bank top
4.00
Use water table
6.00
Input own pore pressures (kPa)
5.00
bank profile
base of layer 1
4.00
ELEVATION (M)
1.09
base of layer 2
3.00
Own Pore
Pressures
kPa
Layer 1
Pore Pressure
From Water Table
-34.34
Layer 2
-24.53
Layer 3
-14.72
Layer 4
-4.91
Layer 5
4.91
base of layer 3
2.00
base of layer 4
failure plane
1.00
water surface
0.00
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
water table
-1.00
STATION (M)
57.5
Shear surface angle used
Export Coordinates back into model
Factor of Safety
1.09
Conditionally
stable
Failure width
Failure volume
Sediment loading
Constituent load
2.35
487
814176
814
m
3
m
kg
kg
Cantilever Failures
National Sedimentation Laboratory
A Tiered Approach
However, If the problem is systemwide instability, or in an urban
setting, the practitioner had better obtain a complete quantitative
understanding of hydrology, magnitudes and trends of adjustment
processes, as well as the absolute and relative resistance of the
boundary materials to erosion by hydraulic and geotechnical
forces.
If this is the case, then the practitioner needs to rely on
validated numerical models, populated with field data to
predict response and stable geometries.
A Tiered Approach
• Analytic Level: Static and Dynamic Numerical Modeling
1. Collect data to define the variables that control processes
(force and resistance)
2. Use the best available numerical models for prediction
We cannot ignore the watershed and its delivery of energy and
materials to the channel system. In fact, changes to the watershed
may be the cause (problem) of the channel instability. An upland
model that provides flow and sediment loadings as the upstream
boundary condition should be coupled with a deterministic
channel-process model (that also handles mass failures). This way
changes at the watershed level can be incorporated into potential
channel effects
NSL Channel-Model Capabilities
Process
Bank StabilityToe Erosion
CONCEPTS
Hydraulic bank erosion


Bank stability


Bed erosion

Sediment transport

Vegetation effects


‘Hard’ engineering



Channel evolution
Rapid Assessments

CONCEPTS
• Hydraulics
HEC 6
•
– Steady, one-dimensional,
i.e. steady-state only
– Unsteady, one-dimensional,
i.e. dynamic inputs
• Sediment transport and
bed adjustment
•
•
Streambank erosion
– No bank processes
simulated
– Fluvial erosion and mass
failure simulated
• Instream structures:
culverts, bridge crossings,
grade control structures
Sediment transport and
bed adjustment
– Equilibrium sediment
transport – excess dumped
– Non-equilibrium sediment
transport – excess settles
• Streambank erosion
Hydraulics
•
Instream structures:
culverts, bridge crossings,
grade control structures
Discretizing the Stream Corridor
Inputs - Cross Sections and Rating Curves
Variable Roughness Elements
Sediment Sources and Fate
Sediment can come
from the
surrounding area,
banks, bed or
upstream
When excess sediment is
entrained, the surplus settles
out based on particle size,
rather than being simply
instantaneously dumped
Sediment Transport
Streambank-Erosion Modeling
soil layer 2
• Hydraulic erosion of
cohesive soils is
expressed by an excess
shear stress relation
soil layer 1
• Combination of
hydraulic erosion and
mass failure
E  K t  t c 
FOS 
Resisting Forces
Driving Forces
Wi
assumed groundwater surface
actual groundwater surface
Si
Ni
soil layer 3
• Bank stability is
expressed by a Factor of
Safety
slice i
failure surface and
bank profile after failure

lateral erosion and
bank profile after erosion
Summary and Conclusions
• Whether disturbances are “natural” or anthropogenic, occur at slow rates
over long periods of time or are catastrophic and instantaneous, incision occurs
because of an imbalance between sediment supply and transporting power.
• Resistance of the boundary to hydraulic and geotechnical forces provide
partial control of adjustment processes and stable channel morphologies.
•Restoration has to be in the context of the condition of the watershed and the
channel system… spatial and temporal aspects of the instability
• Restoration of an unstable reach within an dynamic, unstable system will not
likely be successful
• An energy- or deterministically-based approach provides a reliable means of
analyzing adjustment processes and predicting stable-channel geometries in
unstable systems.