Sie sind auf Seite 1von 23

EARTH SURFACE PROCESSES AND LANDFORMS

Earth Surf. Process. Landforms 42, 191–213 (2017)


Copyright © 2016 John Wiley & Sons, Ltd.
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI: 10.1002/esp.4073

State of Science

Evaluating a process-based model for use in


streambank stabilization: insights on the Bank
Stability and Toe Erosion Model (BSTEM)
Kate Klavon,1 Garey Fox,1* Lucie Guertault,1 Eddy Langendoen,2 Holly Enlow,1 Ron Miller1 and Anish Khanal1
1
Biosystems and Agricultural Engineering, Oklahoma State University, Stillwater, OK USA
2
USDA-ARS, National Sedimentation Laboratory, Oxford, MS USA

Received 14 May 2016; Revised 20 October 2016; Accepted 21 October 2016

*Correspondence to: Garey Fox, Biosystems and Agricultural Engineering, Oklahoma State University, 120 Agricultural Hall, Stillwater, OK 74078-6016, USA. E-mail:
garey.fox@okstate.edu

ABSTRACT: Streambank retreat is a complex cyclical process involving subaerial processes, fluvial erosion, seepage erosion, and
geotechnical failures and is driven by several soil properties that themselves are temporally and spatially variable. Therefore, it can
be extremely challenging to predict and model the erosion and consequent retreat of streambanks. However, modeling streambank
retreat has many important applications, including the design and assessment of mitigation strategies for stream revitalization and
stabilization. In order to highlight the current complexities of modeling streambank retreat and to suggest future research areas, this
paper reviewed one of the most comprehensive streambank retreat models available, the Bank Stability and Toe Erosion Model
(BSTEM), which has recently been integrated with several popular hydrodynamic and sediment transport models including the Hy-
drologic Engineering Centerˈs River Analysis System (HEC-RAS). The objectives of this paper were to: (i) comprehensively review
studies that have utilized BSTEM and report their findings, (ii) address the limitations of the model so that it can be applied appropri-
ately in its current form, and (iii) suggest directions of research that will help make the model a more useful tool in future applications.
The paper includes an extensive overview of peer reviewed studies to guide future users of BSTEM. The review demonstrated that the
model needs further testing and evaluation outside of the central United States. Also, further development is needed in terms of ac-
counting for spatial and temporal variability in geotechnical and fluvial erodibility parameters, incorporating subaerial processes,
and accounting for the influence of riparian vegetation on streambank pore-water pressure dynamics, applied shear stress, and erod-
ibility parameters. Copyright © 2016 John Wiley & Sons, Ltd.
KEYWORDS: Bank Stability and Toe Erosion Model; borehole shear test; fluvial erosion; geotechnical parameters; jet erosion test; modeling;
streambank retreat

Introduction there are still important questions related to accurately model-


ing erosion processes and the practices designed to control
Numerous studies have highlighted the significant level of sed- them (Fox et al., 2016a). For example, stabilizing specific
iment contributed to surface waters through channel erosion reaches of streams may have little impact on total sediment
and failure (e.g. Walling et al., 1998; Bai and Lung, 2005; loads reaching downstream waters and in some cases stabiliza-
Simon and Rinaldi, 2006; Triplett et al., 2009; Miller et al., tion could exacerbate other problems downstream. Moreover,
2014; United States Environmental Protection Agency [US ecological and water quality benefits of stream stabilization
EPA], 2015). Sediment negatively impacts aquatic life, reduces may be limited by constraining rivers into a fixed planform.
recreational values of waters, increases the cost of water Numerical models have become increasingly applied for
treatment, and can impair navigation (Clark et al., 1985). simulating streambank retreat and river evolution at various
Streambank retreat rates have been reported to range over spatial and temporal scales, ranging from the century scale for
several orders of magnitude (Schumm and Lichty, 1963; Simon, meander migration to annual, monthly, or daily scales for river
1989; Simon et al., 2000; Miller et al., 2014) and can contribute engineering studies such as stream stabilization (Rinaldi and
as much as 90% of total sediment loads to surface waters Darby, 2007; Coulthard and Van De Wiel, 2012). Thus, bank
(Grissinger and Murphey, 1982; Fox et al., 2016b). One prom- erosion models fall along a range from empirical to process-
ising strategy for reducing sediment loads to surface waters is based approaches. Empirical approaches attempt to predict
through controlling streambank erosion and failure; however, channel instability from simple variables (Piégay et al., 2005).
192 KLAVON K. ET AL.

For example, the bank erosion hazard index (BEHI) is a com- of studies related to the factors influencing the processes of
monly applied example of an empirical approach, which at- streambank erosion and failure. A list of peer reviewed papers
tempts to relate short-term measurements of bank erosion and that used BSTEM as a part of their study and a summary of
estimates of ‘near-bank stress’ to create a risk estimate of future modeling objectives and findings are presented in Table I,
bank erosion (Rosgen, 1996, 2001; Heeren et al., 2012; Newton and detailed reviews of several of these studies are presented
and Drenten, 2015). The estimation of the ‘near-bank stress’ is by topic.
based on the calculation of vertical velocity profiles and shear Although the model has already proven highly useful in
stress for subsequent distribution of energy calculations in the certain areas of stream stability research and engineering, those
near-bank region (Rosgen, 1996). Factors in BEHI are assigned applications can be significantly expanded with additional re-
a rating based on both quantitative and qualitative assessments search and model development. The objectives of this paper
of the streambank. Relationships are then derived between were to: (i) comprehensively review studies that have utilized
bank erosion rate and near-bank stress, but separated into BSTEM and report their findings, (ii) identify the limitations of
applied stress categories called bank erosion risk ratings. How- the model so that it can be applied appropriately in its current
ever, considerable variability is typically reported for these form, and (iii) suggest research directions that will help make
relationships. An alternative empirical approach is the channel the model a more useful tool in future applications. The paper
stability index, CSI (Simon and Downs, 1995; Simon and is organized in the following sections:
Klimetz, 2008; Heeren et al., 2012). While BEHI is a local site
examination of a bankˈs condition, CSI is a reach-scale rapid • Section 2 – Description of BSTEM internal processes includ-
assessment incorporating form- and process-based sub-metrics ing geotechnical failure, fluvial erosion, and vegetation;
and the stage of channel evolution, as determined based on • Section 3 – Estimation and future research needs of model
morphological criteria, to indicate stability. inputs for BSTEM including geotechnical parameters from
Process-based models aim at simulating physical processes the borehole shear test (BST), fluvial erodibility parameters
responsible for bank retreat. Simpler process-based approaches from the jet erosion test (JET), and default parameters in
generally only consider one of the major driving processes BSTEM;
leading to streambank retreat, typically fluvial erosion pro- • Section 4 – Example application of BSTEM to estimate sedi-
cesses. Bank erosion models of this type, and in particular most ment loads from streambanks at the watershed scale;
meander migration models, commonly utilize a bank erodibil- • Section 5 – Suggested improvements to BSTEM including (1)
ity coefficient calibrated to historical retreat rates in a stream model run time and uncertainty/sensitivity analysis improve-
usually multiplied by a factor related to stream flow, power, ments, (2) complex near-bank flow and applied shear stress,
or shear stress (Posner and Duan, 2012). More complex and (3) subaerial processes and pore-water pressures;
process-based streambank models combine fluvial processes • Section 6 – Alternative models for stream-scale predictions
and geotechnical failure. Such spreadsheet based tools origi- and current developments with BSTEM, including integra-
nated with the work of Thorne and Tovey (1981), Osman and tion of BSTEM with the Hydrologic Engineering Centerˈs
Thorne (1988), and Thorne and Abt (1993) and modeled bank River Analysis System (HEC-RAS).
instability caused primarily by fluvial erosion of the bank toe.
This approach was further developed by Darby and Thorne This review paper focused on BSTEM instead of several other
(1996), Rinaldi and Casagli (1999), Darby et al. (2000), Simon integrated bank erosion and stability models because its algo-
et al. (2000), and Darby et al. (2002). All of these studies incor- rithms have recently been integrated, using modifications from
porated geotechnical failure processes and highlighted the im- the CONCEPTS model when appropriate, with several popular
portance of near-bank pore-water pressures on streambank hydrodynamic and sediment transport models, including
migration, leading to the current versions of more advanced HEC-RAS (Gibson et al., 2015), RVR Meander (Motta et al.,
process-based models used today (Rinaldi and Darby, 2007; 2012), SRH-2D (Lai et al., 2015), and TELEMAC (Langendoen
Rinaldi and Nardi, 2013). Examples of these more recent et al., 2016). The reader will please note that ‘the model’ in
models include numerous versions of one-dimensional or the following sections will be in reference to BSTEM.
two-dimensional models for hydrodynamics and/or river mean-
der migration with physically-based bank evolution (Rinaldi
et al., 2004; Rinaldi and Darby, 2007; Darby et al., 2010; Motta BSTEM Internal Process Modeling
et al., 2012; Nardi et al., 2013; El Kadi Abderrezzak et al.,
2014; Lai et al., 2015). Modeling geotechnical bank failure
The Bank Stability and Toe Erosion Model (BSTEM), devel-
oped by the National Sedimentation Laboratory in Oxford, Geotechnical failure occurs when the driving forces applied to
Mississippi, USA, has been described by several to be one of a bank exceed the resisting forces. In BSTEM, the process-based
the most advanced and commonly used tools for modeling calculation of bank stability takes the form of the computation
streambank retreat (Simon et al., 2000; Simon et al., 2011; of a factor of safety (FoS) ratio of driving forces to resisting
Curran and Hession, 2013; Rinaldi and Nardi, 2013; Lai forces for a given failure plane. The FoS is calculated using sev-
et al., 2015; Daly et al., 2015c). Since its creation, the model eral methods: horizontal layers, vertical slices, and cantilever
has transformed from a simple model able to analyze the stabil- shear failures. The model iteratively considers several failure
ity of a streambank at one particular time to a robust tool with planes and shear emergence elevations until the failure plane
the capability to dynamically predict fluvial erosion and geo- with the lowest FoS is determined. A detailed explanation of
technical failures over time for streams with various geometries, current BSTEM stability calculations is available in several pub-
soils, groundwater dynamics, and stream flow conditions lications (Langendoen and Simon, 2008; Simon et al., 2009;
(Simon et al., 2000; Simon et al., 2011) by adopting features Midgley et al. 2012; Daly et al., 2015c).
of its companion channel evolution computer model, the Con- The user input necessary to evaluate geotechnical stability in
servational Channel Evolution Pollutant and Transport System, BSTEM includes bank geometry and layering; effective cohe-
CONCEPTS (Langendoen and Simon, 2008). Consequently, sion, cˈ; effective internal angle of friction, ϕˈ; the angle that
the bank erosion simulation methodology of BSTEM is similar describes the relationship between matric suction and shear
to that of CONCEPTS. The model has been used in a variety strength, ϕ b; and parameters quantifying the dynamics of pore

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
Table I. Summary of peer reviewed studies using the Bank Stability and Toe Erosion Model (BSTEM) as a component of the research

Study Version of BSTEM Location Purpose Validation of results Major findings


a
Simon et al. (2000) BSM Goodwin Creek, MS Show that a dynamic bank stability Compared known bank failure times The bank stability algorithm sufficiently
algorithm can be used in combination to model predicted FoS represented the timing of bank failures.
with specific field data to accurately
represent the timing of bank failures
Simon and Collison (2002) BSM Goodwin Creek, MS Quantify mechanic and hydraulic Compared known bank failure timing The hydraulic and mechanical effects of
effects of various riparian vegetative under different riparian covers to riparian vegetation were both significant
species from BSM predictions predictions of FoS to bank stability.

Copyright © 2016 John Wiley & Sons, Ltd.


Simon and Thomas (2002) BSM Yalobusha River, MS Assess the impact of debris plug Not reported The rapid removal of the debris plug
removal on upstream bank stability caused all upstream banks to be at
risk of instability according to model
predictions.
Simon et al. (2002b) BSM Missouri River, MT Determine the effects of dam removal Not reported The key factors affecting bank stability
on stability, identify key factors were frequency and duration of flow
impacting stability, and compare events and both algorithms predicted
the BSM algorithm to another similar stability for the same hydraulic
stability model conditions.
Pollen and Simon (2005) BSM Fictitious Channel, MS Assess two root failure modelsˈ Compared both modelsˈ predicted The root bundle progressive model
(instantaneous and root bundle increased shear strengths to direct improved estimates of root shearing
progressive model) impact on shear tests completed on soils with resistance and therefore has the
bank FoS and without roots potential to improve estimation of
streambank stability in BSM.
Simon et al. (2006) 3.4 Upper Truckee River, CA Quantify mechanical and hydrologic Compared known bank failure timing BSTEM showed the effect of different
effects of two riparian vegetative under different riparian covers to riparian species on the frequency of
species on streambank stability model predicted FoS bank failures and delivery of sediment.
Lemmonˈs Willow withstood steeper
slopes and more saturated conditions.
Wilson et al. (2007) BSM Little Topashaw Creek, MS Evaluate model ability to account Compared laboratory measured to The model simulated increased instability
for streambank failure due to model predicted sediment loads of banks due to seepage erosion and
seepage erosion did not account for sediment loss
due to sapping.
Cancienne et al. (2008) 4.1 Little Topashaw & Determine importance of seepage BSTEM sensitivities matched previous BSTEM is most sensitive to pore water
Goodwin Creeks, MS undercutting relative to several studies for several parameters pressure distribution, small degrees
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM)

parameters (bank shear strength, of undercutting can counteract effects


bank angle, soil pore water pressure, of root reinforcement, and streambank
and root reinforcement) stability models need to account for
site-specific failure mechanisms.
Parker et al. (2008) — Goodwin Creek, MS Determine how variability in soil Compared known bank failures to Variability in soil parameters produced
parameters within and between sites range in predicted FoS a significant range in FoS, the current
impacts BSTEM predictions and how addressing of uncertainty in BSTEM is
to best incorporate uncertainty into not suitable, and a probabilistic
modeling approach is more appropriate than
deterministic approaches.

(Continues)

Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)


193
Table 1. (Continued)
194

Study Version of BSTEM Location Purpose Validation of results Major findings

Lindow et al. (2009) 4.1 Fictitious Channel, NC Analyze and model laboratory Compared known bank failures to The model could not predict the observed
experiments to determine the predicted FoS failures due to lateral seepage forces,
effect of seepage, pore water therefore near bank groundwater flow
pressure, and bank geometry processes should be incorporated.
on stability
Pollen-Bankhead and BSM Fictitious Channels Evaluate bank stability with Not reported Combining RipRoot and BSTEM models
Simon (2009) variations of root density with allowed for analysis of spatial and
depth to correctly apply RipRoot temporal variations on bank stability

Copyright © 2016 John Wiley & Sons, Ltd.


estimates to BSTEM layering for various riparian vegetative species.
Simon et al. (2009) 4.1 Upper Trucke River & Evaluate effectiveness of three bank Not Reported Toe protection reduced sediment loads
Blackwood & Ward stability treatments (toe protection, by 69–100% and bank failures to a
Creeks, CA top of bank vegetation, and single episode, while other stability
decreased bed slope) by quantifying treatments reduced loads by 42–54%,
sediment load reductions on average. Results demonstrated
importance of toe protection.
Heinley (2010) 5.2 Osage River, Missouri Determine the influences of reservoir Validated with data from Rinaldi Most common failure mechanism is
releases on the rate and amount of et al. (2004) for Seine River in Italy mass wasting or cantilever failures,
streambank erosion which can be reduced through toe
protection of the banks to limit the
impact of releases
Pollen-Bankhead and 5.1 Ficticious Channels, MS Incorporate effect of root permeated Effects of vegetation appeared to The change in matric suction from
Simon (2010) bank and toe material on soil compare well to similar studies evapotranspiration showed the greatest
erodibility and hydrologic effect of potential benefit to FoS but only in
evapotranspiration into stability model the summer. Root reinforcement was
KLAVON K. ET AL.

the most important contributor to


FoS during the critical spring and
winter months.
Simon et al. (2011) 5.1 Goodwin Creek, MS Determine bank stability conditions under Simulations of existing conditions BSTEM was successfully used to design
a range of hydraulic and geotechnical matched 10 years worth of a stable bank with a constrained 1:1
conditions and erosion-control strategies, observations at the site, post slope requirement. The type of vegetation
design sustainable bank stabilization project monitoring required and rock toe protection were
measures tested until a stable design was
determined.
Simon et al. (2011) 5.1 Lower Tombigbee River, AL Use iterative modeling to quantify Known average 29 year retreat rate User should simulate at least a one year
sediment loads of existing conditions at the site for existing conditions “worst case” flow hydrograph (including
and mitigation strategies drawdown conditions) in order to quantify
sediment loads and load reductions.
BSTEM can be used to calculate
volumetric and retreat reductions from
various mitigation strategies.
Simon et al. (2011) 5.1 Upper Truckee River & Use iterative modeling to quantify Not reported RGAs can be used to assess the percent
Blackwood & Ward sediment loads of existing conditions of the banks considered highly or
Creeks, CA and mitigation strategies moderately erosive to estimate

(Continues)

Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)


Table 1. (Continued)

Study Version of BSTEM Location Purpose Validation of results Major findings

stream-scale loads. Load reductions and


unit cost per ton of soil for implementing
rock toe protection were calculated. Load
reductions to the lake and unit costs
ranged from 33 to 87% and $267 to
1
$2500 t respectively.
Simon et al. (2011) 5.1 Big Souix River, SD Determine average annual rates of Results of spatially extrapolated loads Loads were calculated for different
streambank erosion and the were compared to suspended sediment percentile annual flows. Bank failures

Copyright © 2016 John Wiley & Sons, Ltd.


effects of mitigation strategies loads from USGS gauge data were typically only predicted during
90th percentile years. Loads for lower
percentile flows were dominantly from
fluvial erosion. Rock toe protection
resulted in 85–100% reduction in load.
Loads from streambanks contributed
10–25% of suspended sediment.
b
Midgley et al. (2012) 5.2 Barren Fork Creek, OK Evaluate BSTEMˈs ability to predict Compared observed bank retreat and BSTEM was able to accurately predict
long term retreat and determine failure timing to model retreat the timing of bank failures, BSTEM
the importance of accurate predictions for both default and under predicted retreat for both default
geotechnical, fluvial, and near measured fluvial and geotechnical and measured soil parameters, and
bank pore water pressure properties BSTEM predicted a more accurate
parameters retreat with measured soil parameters
using the JET.
Motta et al. (2012) — Various Propose new approach that Application to several test cases for Improved physically-based method is
calculates rates of meander idealized and natural planform required to capture migration patterns
migration using physically-based geometry of natural channels, not predicted from
streambank erosion formulations hydrodynamics only
with RVR Meander and bank stability
routines in CONCEPTs (BSTEM)
Langendoen et al. (2014) 5.4 Debre Mawi, Ethiopia Design stable gully head and bank Compared observed stable and unstable Expansion of gullies up to six meters deep
profiles to halt gully expansion in gully banks to model predictions for can be halted by regrading gully head
the sub-humid northern Ethiopian various soil hydrologic conditions and banks to 45 degrees and protecting
highlands using both default and measured the surfaces with rock and vegetation
geotechnical properties
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM)

Polvi et al. (2014) — North Joe Wright and Use RipRoot within BSTEM to Not Reported Reported added root cohesion of four
Corral Creeks, CO determine the added cohesion major vegetation types (tree, forb,
due to roots for various sediment graminoid, and shrub) on 6 different
textures and plant species soil textures. The highest added
cohesion was for Salix geyeriana
due to a large number of large roots
for the species.
b
Daly et al. (2015c) — Barren Fork Creek, OK Assess model sensitivity to Compared observed bank retreat and Calibration of erodibility parameters
root cohesion and performance failure timing to model predictions, enabled the model to match both timing
in predicting bank retreat. and amount of retreat for sites with and

(Continues)

Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)


195
Table 1. (Continued)
196

Study Version of BSTEM Location Purpose Validation of results Major findings

Evaluate a simple method for calibrated erodibility parameters without vegetative cover. BSTEM was
incorporating riparian practices based on imagery not sensitive to additional soil cohesion
into the model due to roots.
Gibson et al. (2015) — Goodwin Creek, MS Integrate BSTEM with the sediment Compared model predictions to historical Integration included BSTEM using HEC-RAS
transport methods in HEC-RAS 5.0 bank migration times and magnitude hydraulics to determine water surface
of lateral retreat profiles, HEC-RAS updated cross-sections
when failure was predicted by BSTEM,
and HEC-RAS added failed bank material

Copyright © 2016 John Wiley & Sons, Ltd.


to the transport model. Location and
shape of the final cross-sections matched
observed.
Kheiralla and Siddeg (2015) — Demonstrate methodologies for Not Reported Most common failure mode was cantilever
streambank erosion control failure during drawdown conditions
due to toe erosion of a non-cohesive
sand layer
Lai et al. (2015) — Goodwin Creek, MS Develop a coupled flow, sediment Compared predicted retreat geometry to For five year modeling period the difference
transport, and bank stability model measured and previously modeled data between predicted and modeled retreat
that includes a geotechnical model, was 0.2–5.2%, developed model shows
appropriately distributes failed the benefit of multi-dimensional models
bank material to the toe, can in streambank erosion, model was robust
predict complex turbulent flow, and easy to apply, future need for
simulates fluvial erosion, and improving modeling of complex streams.
simulates sediment transport.
b
Khanal et al. (2016b) 5.4 Barren Fork Creek, OK Compare model performance Compared observed bank A non-linear detachment model predicted
KLAVON K. ET AL.

for predicting fluvial erosion retreat to model predictions for retreat closest to the observed retreat.
when a non-linear detachment different fluvial erosion models Linear models both over and under
model is used versus linear excess and solution techniques predicted retreat, depending on the
shear stress equation solution technique used to evaluate
the JET data.
Konsoer et al. (2016) — Wabash River, IL Evaluate the effect of lateral and Compared known retreat and failure Erodibility coefficients derived from JETs
vertical variability in mechanisms to BSTEM predictions, were up to four orders of magnitude
resistance parameters of streambanks RipRoot results were similar to higher than BSTEM default parameters.
on the rates and mechanisms previous studies BSTEM predicted erroneous bank profiles
of retreat with JET parameters. Model simulations
predicted mechanisms of failure. BSTEM
retreat and failure mechanisms are highly
sensitive to lateral and vertical variability
in erosion resistance, geometry, and
hydraulic conditions.
Thapa (2016) 5.4 Kodku River, Nepal Determine the causes of unstable Not reported Suitable bioengineering techniques are
streambanks to protect infrastructure, possible to limit bank toe erosion
agricultural land, and settlements and failure
a
BSM = Bank Stability Model, the original form of BSTEM including only the bank stability algorithm.
b
The device used to estimate fluvial erodibility parameters in situ was the ‘mini’ jet erosion test (JET) not the original JET.

Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)


INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 197

water pressure distributions including the saturated hydraulic the no-lag kinematic model (Crosato, 2007). Accordingly,
conductivity and van Genuchten soil moisture characteristic BSTEM users should be aware that the τ distribution methods
curve parameters. Geotechnical parameters, cˈ and ϕˈ, can used in BSTEM may be limited in being able to represent actual
be determined using four techniques: Iowa BST for in situ mea- field conditions.
surements (Lohnes and Handy, 1968; Lutenegger and Hallberg, When considering the quantification of fluvial erodibility pa-
1981), direct shear test (DST), triaxial shear test (TST), or the rameters for input into BSTEM, some studies suggest an empir-
BSTEM default parameters. The BST is the most commonly used ical relationship between τ c and the soil silt-clay content (Julian
technique for estimating geotechnical parameters in previous and Torres, 2006). However, Simon et al. (2011) suggested
BSTEM studies and will be discussed further in the next section. using the JET or default parameters for cohesive soils. For
non-cohesive soils, little information exists in the literature for
parameters of the excess shear stress equation (Criswell et al.,
Modeling fluvial erosion 2016). The particle size distribution can be used to estimate
the τ c and then empirical relationships can be used to estimate
Process-based fluvial erosion modeling was introduced to kd. For example, Hanson and Simon (2001) noted an inverse re-
BSTEM in version 3.3 (United States Department of Agriculture lationship between the erodibility parameters based on JETs
[USDA], 2016). Of the studies in Table I, seven studies used the from Midwestern US streams. Alternatively, default parameters
model to predict both fluvial erosion and geotechnical failures. in BSTEM can be used for non-cohesive soils. All of the seven
The model typically uses the excess shear stress equation studies in Table I that incorporated the fluvial erosion predic-
(Ariathurai and Arulanandan, 1978; Simon et al., 2000) to tion capabilities of BSTEM used JETs to quantify the erodibility
quantify the erosion rate, εr (in m s1): of the cohesive soils. Use of the JET and derivation of its erod-
ibility parameters will be discussed further later.
εr ¼ k d ðτ  τ c Þa (1) Additionally, fluvial erodibility parameters are suggested to
be temporally variable, dependent on subaerial processes, hy-
where τ is the average applied shear stress (in pascals), τ c is the drologic events, and vegetation cover (Daly et al., 2015b;
critical shear stress (in pascals) that must be exceeded for ero- Konsoer et al., 2016). Yet, no studies have suggested an ap-
sion to occur, kd is the coefficient of erodibility (in m3 N1 s1), proach for adjusting fluvial erodibility parameters with time.
and a is an exponent assumed to be unity. Therefore, one major research need is to study the temporal ef-
An accurate computation of the boundary τ applied on the fects on fluvial erodibility parameters and incorporate the find-
bank profile is necessary to predict fluvial erosion rates. The ings into the model.
boundary τ distribution across a channel cross-section is Khanal et al. (2016b) highlighted other issues with modeling
governed by a three-dimensional flow field and depends on fluvial erosion within BSTEM. Their research incorporated an
the channel form, roughness distribution and secondary cur- alternative to the excess shear stress equation, a non-linear,
rents. Generally, calculation of the τ distribution remains a mechanistic model (Wilson 1993a, 1993b), to quantify the de-
difficult task and ideally requires the computation of the tachment rate in BSTEM. The model, referred to as the Wilson
three-dimensional flow and turbulence. In BSTEM, an analyti- model, is based on the balance of all forces and moments driv-
cal method is used to estimate the reach-scale τ from the ing and resisting detachment of a two-dimensional particle or
cross-section geometry, stage, and channel slope assuming a aggregate:
uniform and steady flow. It is shown as τ = γRhS, where γ is
the specific weight of water, Rh is the hydraulic radius, and S pffiffi    
b0 τ b1
is the energy slope assumed to be the bed slope (S0) for uniform εr ¼ 1  exp  exp 3  (2)
steady flow. The hydraulic-radius separation method (Einstein, ρb τ
1942) is used to divide the cross-section into regions where
the flow is governed by a local hydraulic radius and a local where ρb is the bulk density of soil (in g m3) and b0 (in g
roughness coefficient. The division line between those regions m1 s1 N0.5) and b1 (in pascals) are mechanistically defined
is controlled by the bisector of the wetted bank and bed pro- parameters of the Wilson model. Retreat predicted with the
files. For each region, the hydraulic radius is estimated as the Wilson model was closer to the actual observed retreat in
ratio between the wetted area and the wetted perimeter. comparison to the retreat predicted with the excess shear
Khodashenas et al. (2008) performed a comparison among stress equation. The Wilson model parameters can be derived
several analytical methods for computing the τ distribution as easily as excess shear stress parameters using the JET
and laboratory data for several channel shapes (rectangular, (Al-Madhhachi et al., 2013b). Khanal et al. (2016b) also
trapezoidal, and compound) and roughness. They came to demonstrated that the solution methodology used to analyze
the conclusion that analytical methods similar to the one used JET results for excess shear stress parameters significantly
in BSTEM provide a suitable estimation of the τ distribution, altered BSTEM predicted profiles, highlighting a need to deter-
except near corners and at the edge of the cross-section. This mine which solution methodology should be used for deriving
weakness was attributed to the fact that they do not consider fluvial erodibility parameters. More studies on a variety of
the effect of secondary flows developing in cross-sections streambank conditions are needed to confirm the results of
with corners. Khanal et al. (2016b).
Kean and Smith (2006) and Darby et al. (2010) investigated
hydraulic erosion of bank toe materials by estimating boundary
shear stress for irregular bank topography. Bank topographic Modeling the effects of vegetation
features induced form roughness and significantly influenced
the total applied shear stress. Further, Papanicolaou et al. According to Millar (2000), bank vegetation has several im-
(2007) investigated the effects of secondary flows on cohesive pacts on bank stability as it reduces the near-bank velocity
bank erosion in straight channels, finding two-fold increases and effective bank τ, increases the bankˈs strength through
in average boundary shear stress. To take into account the ef- binding of the sediment by root masses, and enhances intersti-
fect of stream curvature on the τ distribution, a correction for tial deposition of fine washload sediment. Simon and Collison
near-bank τ has been implemented in BSTEM 5.4, based on (2002) found that the hydraulic and mechanical effects of

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
198 KLAVON K. ET AL.

riparian vegetation were both significant to bank stability. In performed per soil layer. The number of tests reported in some
BSTEM, extensive work has been done to model the effects of of the studies inconsistently ranged from one to 16. One study
vegetation on soil geotechnical parameters and apply findings stated that at a minimum duplicates were necessary for each
to integrate a user friendly sub-model called RipRoot (Simon soil layer (Daly et al., 2015c), but no explanation was given
and Collison, 2002; Pollen and Simon, 2005; Simon et al., for this requirement. The potential advantages of BSTs are
2006; Pollen-Bankhead and Simon, 2009, 2010; Polvi et al., that tests are conducted in situ and the short test duration,
2014). The sub-model predicts added cohesion caused by roots with test completion commonly within one hour (e.g. Handy
for 22 common riparian species and allows users to either esti- and Fox, 1967; Lutenegger, 1987); however, some studies
mate root distributions based on the age of the species or enter have highlighted key BST research needs.
measured root diameter class data. However, the RipRoot While the BST device is an alternative to a direct shear labo-
sub-model was only designed and tested for top of bank vege- ratory test, interpretation of BST results has been problematic in
tation, not vegetation covering an entire bank slope. Addition- many cases. One essential difference is that in a laboratory
ally, when applied to long-term modeling on highly erodible setting the pore pressure (μ) is controlled, while in the field it
streams, Daly et al. (2015c) found that BSTEM was not sensitive is unknown. Accordingly, interpreting the results of BSTs
to the minimal added cohesion due to roots in rapidly migrat- appears to have been most straightforward with fine-grained
ing composite banks with gravel toes, and that adjusting fluvial or stiff clay soils at or close to saturation (Handy et al., 1985;
erodibility parameters was a better approach when calibrating Handy, 1986). When soil is nearly or completely saturated, cˈ
BSTEM input parameters for vegetation. and ϕ 0 are functions of the normal stress. When testing soils be-
In channels with vegetated banks, additional drag forces due low the water table, it is necessary to include the weight of wa-
to the stems and branches of the vegetation increase the ter. However, it is possible that the application of normal stress
hydraulic resistance (Kean and Smith, 2004). Traditional one- at the shear head will create an unknown degree of excess pore
dimensional modeling approaches combine vegetation drag pressure in the soil. The BST methods specify consolidation
and boundary roughness into a single roughness coefficient times, which are intended to allow that excess pore-pressure
based on reference values (James et al., 2004; Wilson et al., to dissipate from around the shear plates. Demartinecourt and
2006; Schoneboom et al., 2010), such as given by Chow Bauer (1983) found that a consolidation time of 1.5 hours was
(1959). In BSTEM, a lumped α factor can be included in the required before excess pore pressure completely dissipated in
excess shear stress equation to indirectly correct the applied τ testing of sensitive, soft clay with a BST specifically fitted with
to account for vegetation and other effects (Langendoen and a pore pressure sensor. Lutenegger and Hallberg (1981) noted
Simon, 2008, 2009; Daly et al., 2015c): that excess pore-pressure can accumulate during testing creat-
ing a secondary, shallower friction angle evident when shear
τc
data are plotted. Lutenegger and Tierney (1986) noted a range
εr ¼ k d ðατ  τ c Þ ¼ αk d τ  (3)
α of pressures across the shear head during the application of
shear stress, with the highest pore pressures encountered at
This results in one additional calibration parameter. Signifi- the leading edge and lower pore pressures near the center of
cant research efforts are needed to mechanically estimate the the shear head.
reduction of applied τ and incorporate the findings into the In unsaturated soils, an isometric component of soil strength
model, as was done for the effect of vegetation on soil strength is provided by matric suction. The theoretical formulation by
parameters. For instance, more sophisticated two- and three- Fredlund et al. (1978) postulates that ϕ b will be constant and
dimensional models allow for representing resistance induced that increased matric suction will increase the apparent cohe-
by bed roughness and form drag on the vegetation and can pro- sion. Laboratory direct shear unsaturated soil strength testing
vide a more realistic τ distribution over the river section (Naot includes a range of soil moisture including saturation, so that
et al., 1996; López and García, 1998; Wilson et al., 2006). cˈ, ϕ 0 , and ϕ b can be established. Experimental evidence
The roughness of the riparian vegetation also affects the shows apparent cohesion increasing with matric suction and
secondary flow pattern and alters the distribution of scour and with the apparent angle of friction remaining constant or in-
deposition across the channel. Numerical studies have shown creasing slightly (e.g. Lu and Likos, 2004; Gallage and
that the local effects of vegetation can propagate downstream Uchimura, 2015). In contrast, unsaturated strength testing of
due to the interactions between the channel, flow, and bed soils with the BST shows a decrease in apparent cohesion and
response (Van de Wiel and Darby, 2004). a large increase in apparent angle of friction (Miller et al.,
1998; Miller and Khoury, 2012).
The soil shearing during a BST occurs along a shear plane
Estimating Model Input within the soil, and staged tests with increasing normal stress
act to move this shear zone farther from the shear head and into
Estimating geotechnical parameters using the BST undisturbed soil (Handy and Fox, 1967; Lohnes and Handy,
1968). Handy and Fox (1967) suggested that the sheared soil
A variety of techniques have been used to obtain geotechnical becomes compacted and adhered to the shear head, and that
parameters for use in a wide range of streambank stability and subsequent tests at increased normal stress move the shear sur-
erosion models. The majority of studies used only the BST, face away from this compacted zone and thus each test occurs
others used a combination of techniques for different soil layers in essentially undisturbed soil. However, the nature and geom-
(i.e. BST and DST, BST and TST, or BST and default parameters), etry of this zone of compaction remains unknown in BSTs. Im-
some used solely default parameters, and a few did not state portantly for interpreting BST results, the compacted zone at the
parameters or methods used for obtaining geotechnical param- shear head is also an essential difference from laboratory direct
eters (Table I). Because of the large number of studies that used shear testing, where no such unknown compaction exists.
BSTs to obtain geotechnical parameters in previous BSTEM Lutenegger and Tierney (1986) noted that the compacted zone
studies, the BST will be the focus on this review paper. The introduces an unknown with respect to the distribution of pore-
BST manual makes no recommendation for the number of test water pressure along the shear zone. This can affect interpreta-
replications (Handy Geotechnical Instruments, 2013). Of those tion of results for unsaturated soils (Miller et al., 1998) and can
that used the BST, most did not report the number of BSTs alter conditions from test to test (Bechtum, 2012).

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 199

In many fields of science and in geotechnical investigations concluded that results were comparable to each other and
there is a need for acquiring soil strength data, and in situ test- values obtained from flumes with an adjustment factor
ing offers the ability to evaluate soil secondary properties and (Al-Madhhachi et al., 2013a).
site variability that may be important in reaching conclusions The experimental data obtained from the JETs have been an-
about a locale. The BST promises to be a relatively rapid alyzed using three different solution routines to derive the
method to supply those data. However, as noted earlier, there erodibility parameters for Equation (1), which will be referred
are important unknowns about BST-acquired data that raise to as ‘Blaisdell’, ‘scour depth’, and ‘iterative’ in the following
questions about interpreting results as equivalent to laboratory discussion. The Blaisdell solution method is oldest, and is cur-
soils testing. The theory and analytical methods required to as- rently the most popular method of analysis (Blaisdell et al.,
sess the components of soil strength are based on the controlled 1981; Hanson and Cook, 2004). It first determines the τ c based
outputs of laboratory tests. However, many of the variables on the equilibrium depth of the scour hole by fitting a hyper-
which are carefully controlled in a laboratory setting are un- bolic function to the data to estimate the scour depth as time
known and/or uncontrolled in BSTs. So, despite the linear ap- approaches infinity. The kd is then iteratively solved to mini-
pearance of the BST failure envelope, it is uncertain what the mize the error between the measured and predicted dimen-
test is actually measuring. Some important research areas still sionless time based on an integrated solution of the excess
need to be addressed in order to reliably interpret BST soil shear stress equation.
strength test results including the effect of BSTs on soil moisture In contrast to the Blaisdell solution, the more recently devel-
surrounding the test zone, the size of the undisturbed soil zone oped scour depth and iterative solutions solve for both τ c and kd
in which shearing occurs, the nature of the consolidated zone simultaneously through iteration. The scour depth solution
to which normal stress is applied, and the number of tests re- minimizes the squared error between the observed JET erosion
quired to appropriately represent soil shear strength parameters rate and the predicted erosion rate computed by the excess
in process-based models. This paper specifically calls for future shear stress equation to calculate τ c and kd (Daly et al., 2013).
research that can indicate the extent to which geotechnical pa- The iterative solution solves for the parameters in a similar
rameter diverge and under what conditions with data derived manner, but places an upper bound on τ c by assuming the
from both the BST and laboratory methods. Research is also scour depth reaches an equilibrium depth at the end of the test
needed to explore the extent that these quantified uncertainties (Simon et al., 2010). Note that the parameters (b0, b1) of the
are then significant in forcing uncertainty within BSTEM Wilson model, shown in Equation (2), are also estimated from
simulations. the JET data using the scour depth technique as described in
Al-Madhhachi et al. (2013b).
The Blaisdell solution estimates lower τ c and kd than the scour
Estimating fluvial erodibility parameters depth and iterative solutions (Simon et al., 2010; Daly et al.,
2013). At higher applied τ, εr predicted by the scour depth solu-
As with the BST, several research areas still need to be ad- tion and the iterative solution are thus much higher due to their
dressed in order to appropriately determine fluvial erodibility higher estimated kd values (Khanal et al., 2016b). When used in
parameters. Several techniques have been employed to esti- stability models such as BSTEM, kd is frequently used as a cali-
mate the fluvial erodibility parameters for different streambank bration parameter, and it has been observed that kd estimated
erosion and failure models like BSTEM. Examples of these tech- from JETs using the scour depth and iterative solutions requires
niques include flumes, rotating cylinder apparatus (Moore and significant scaling down to match the predicted and observed
Masch, 1962), hole erosion test (HET) (Wan and Fell, 2004), co- bank retreat (Daly et al., 2015a). This necessity raises questions
hesive strength meters (CSM) (Tolhurst et al., 1999) and JETs. about the assumption of linearity between εr and τ and the ad-
Flume tests involve creating open-channel flow conditions with vantage of a non-linear detachment model as discussed by
soil at the bed of the flume and visually or graphically deter- Khanal et al. (2016b). A consensus on which solution technique
mining τ c (Clark and Wynn, 2007). Several studies utilize this should be used for deriving the parameters for streambank
more traditional method of estimating erodibility parameters erosion modeling has not been reached. In fact, deriving these
(Kamphuis and Hall, 1983; Hanson, 1990; Sutarto et al., parameters may be dependent on regional soil properties and
2014). Flumes have also been modified specifically to measure there may not be a single solution approach.
the fluvial erodibility of sediments as in the cases of SEDflume One of the other major issues concerning the JET has been
(McNeil et al., 1996) and erosion function apparatus (EFA) the wide variability in the erodibility parameters estimated from
(Briaud et al., 2001). Flume studies have the advantage of the JETs. The parameters varied by as much as three orders of
allowing the test to be conducted in a controlled environment, magnitude when tests are compared at the site scale (Daly
but it is difficult to collect and place a sample in the flume with- et al., 2015a) and by as much as four orders of magnitude at
out some disturbance to the sample. The majority of BSTEM the watershed scale (Wynn and Mostaghimi, 2006; Wynn
studies report deriving erodibility parameters using the JET, et al., 2008; Karmaker and Dutta, 2011; Daly et al., 2015b).
likely because in situ testing is preferred for streambank erosion Certain variability in the parameters is expected due to the in-
studies. Consequently, only JET testing and analysis will be herent spatial heterogeneity of the soil properties and other ex-
discussed further in the following section. trinsic factors influencing the erosion (Hanson and Simon,
During a JET, a submerged jet of water generated under a 2001); much less variability is typically observed under labora-
constant hydrostatic pressure impinges on a soil surface. The tory conditions across all the post-processing techniques
jet exerts a shear pressure on the soil causing the soil particle (Khanal et al., 2016a). However, the high degree of variability
to detach and soil to scour. The depth of the scour is measured is problematic, especially when simulation models like BSTEM
periodically and the data is used to estimate the erodibility pa- use single values of the estimated parameters (Karmaker and
rameters. Currently there are two forms of JETs in operation: Dutta, 2015). Simon et al. (2006) reported two JETs per bank
the original JET and mini-JET. The original JET is described in layer, Daly et al. (2015c) suggested at least triplicates, and Lai
detail by Hanson and Cook (2004), and the mini-JET, a minia- et al. (2015) completed 16 JETs on different soil layers and used
turized version of the original-JET, was first described by the average for the BSTEM simulation. No consensus has been
Simon et al. (2010). Al-Madhhachi et al. (2013b) compared reached on the number of JETs needed to adequately account
the results obtained from these two forms of the JET and for this variability.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
200 KLAVON K. ET AL.

Numerous user-dependent factors, including head setting, Default BSTEM parameters


initial reading time interval, and termination reading time inter-
val have been found to influence the parameters of the erodibil- In addition to uncertainties on the method of deriving geotech-
ity models. Khanal et al. (2016a) recommended an initial time nical parameters, default parameters listed in Simon et al.
interval of no more than 30 seconds and a termination interval (2011) are available for a very limited selection of soil types:
of at least 300 seconds for highly erodible soils. These recom- gravel, sand and gravel, sand, loam, and clay. A sensitivity
mendations are a step forward in the direction to standardizing analysis of the stability algorithm in BSTEM by Parker et al.
the operation of JET and similar studies should be conducted to (2008) revealed that the FoS is highly sensitive to input geo-
include other soil types. technical parameters and most sensitive to cˈ. The sensitivity
Another major criticism of the JET apparatus is the confine- analysis was performed for nine separate hydrologic events
ment of the fluid field and the alterations in the τ distribution on a single bank profile. Model FoS predictions were compared
within the submerged jet due to the confinement (Ghaneeized to known information on whether or not bank failure actually
et al., 2015). The confined conditions of the JET apparatus raise occurred for the given event. When the means of default geo-
questions about the assumption of a free jet and unconfined technical parameters were used, BSTEM predicted bank fail-
flow used in deriving the equations that predict the τ distribu- ures for three of the five banks known to be stable, and stable
tion. Ghaneeized et al. (2015) recommended investigating banks for all of the banks that had known failures. When the
the effect of confinement in the JET and the mini-JET and least- and most-resistant geotechnical parameters within a rea-
redefining the apparatus constants and relationships to im- sonable range were used for each bank, the FoS ranged from
prove the erosion assessments. Another consequence of the stable to unstable for all banks. This study highlights the effect
confined conditions could be the effect on the suspension of of geotechnical parameter variability within a single bank layer
the detached soil particle due to recirculation and their even- on BSTEM FoS (Parker et al., 2008). While this research was
tual resettlement at low τ. The JET assumes that the scouring conducted specifically using BSTEM, the conclusions would
is a continuous process during the test and all the detached be valid for any integrated fluvial erosion and mass wasting
particles are ejected from the flow. While the smallest particles model utilizing a mechanistic approach.
indeed are transported out of confinement with the waste Geotechnical parameters can be highly variable and are also
water, the authors have observed that the heavier particles extremely important to determining bank stability, and users
can remain in the immersion canister for a longer time, and should be aware of the implications of choosing default param-
eventually detached larger particles may not leave the scour eters. Measured geotechnical parameters and information on
hole as the applied τ decreases. The influence of such effects soil texture were gathered from studies listed in Table I, and
on the erodibility parameters must be investigated further. then separated into BSTEM soil default categories (sand, silt,
Although the JET is a unique device for measuring erodibility and clay). In most cases, the BSTEM default value did not cor-
parameters in situ and compares favorably to flume tests under respond to the median of in situ values, and in some cases,
controlled conditions, there remains uncertainty when it is was among the most extreme values (Figure 1). Users should
applied in field conditions. measure geotechnical parameters when possible instead of re-
Finally, fluvial erodibility parameters are typically used as lying on BSTEM default parameters for the most accurate repre-
calibration factors, with the JET results being used as a starting sentation of their study site.
point. Several calibration methods exist, for instance Lai et al. BSTEM also gives default parameters for fluvial erodibility
(2015) recommend calibrating kd only, with the justification parameters. Figure 2 shows the range of JET results when cate-
that field measurements produce highly uncertain kd estimates. gorized by BSTEM default soil textures for the studies in Table I
Other studies suggested calibrating fluvial erodibility parame- that reported parameters used for fluvial erodibility. Konsoer
ters using a lumped α factor shown in Equation (3) to account et al. (2016) noted that erodibility parameters derived from JETs
for spatial and temporal changes in soil resistance to fluvial were several orders of magnitude different than BSTEM default
forces due to several factors: roots, moisture content, and sub- parameters; however, the default parameters were sufficient at
aerial processes (Langendoen and Simon, 2008, 2009; predicting the failure type over long time periods. In contrast,
Rousselot, 2009; Daly et al., 2015c). Midgley et al. (2012) found that solutions from the JET were

Figure 1. Box and whisker plots of measured geotechnical parameters of papers listed in Table I versus default BSTEM parameters (triangles). Infor-
mation on soil texture provided by the paper was used to lump geotechnical parameters into sand, silt, and clay default categories.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 201

Figure 2. Box and whisker plots of measured fluvial erodibility parameters of papers listed in Table I versus default BSTEM parameters (triangles).
Information on soil texture provided by the paper was used to lump parameters into sand, silt, and clay default categories. Note that BSTEM default
parameters for silt and clay are classified as resistant, moderately erodible, or erodible, and hence have multiple values.

better at predicting retreat than the default parameters for long- has been shown to simulate the effects of reducing bank slope
term simulations. Daly et al. (2015c, 2016) also warned of over by regrading, reducing bed slope (i.e. constructing a meander
reliance on relationships constructed between erodibility pa- bend), covering the bank top with a variety of riparian vegeta-
rameters and soil texture such as the Julian and Torres (2006) tion species, and protecting the bank with rock.
relationship: using empirical equations to predict erosion prop- Theoretically, everything seems to be in order for BSTEM to be
erties based on soil characteristics will likely contain high un- a practical tool for stream stabilization. However, BSTEM has yet
certainty and thus should be used with caution. From to reliably quantify long-term sediment loads without significant
research on three streambanks of varying erodibility, Daly calibration (Simon et al., 2009; Simon et al., 2011; Midgley
et al. (2016) reported no consistently significantly correlated et al., 2012; Daly et al., 2015c). For example, Simon et al.
variables in terms of erodibility and soil physical properties (2011) completed a study on the Lower Tombigbee River in
from multiple linear regression and principal components Alabama. Detailed site information was known including geo-
analysis. technical parameters, fluvial parameters, stage, and several
years of measured bank retreat. The stream was expected to
experience an average retreat of 1.2 m yr1. Stabilization
BSTEM Applications practices, including combinations of rock toe and vegetation,
were modeled and compared to simulations of no treatment
Applying BSTEM to stream stabilization over a one-year simulation period. However, the no-treatment
simulation predicted a retreat of over 40 m during the
In recent years, BSTEM has shown potential for being an alter- simulation, which was not even close to the known average of
native to methods such as classification systems for stream 1.2 m yr1. The model results were stated to not represent actual
stabilization design (Simon et al., 2009; Simon et al., 2011). volumes of failures, but only should be used to highlight relative
The appeal of BSTEM is that it is a process-based method, differences between the existing, no-treatment case and various
and therefore can potentially be used to predict actual bank stabilization strategies (Simon et al., 2011). If it is to be used as
response to various erosion mitigation strategies, and future a design tool by engineers and conservationists, the model
conditions. BSTEM is capable of aiding in the design of site should be able to predict more reliable loads and retreat rates.
specific bank stabilization practices by determining minimum
conditions for bank stability, and quantifying retreat rates and
sediment loads of various bank conditions including conditions Watershed-scale sediment load estimation
after stabilization.
Simon et al. (2011) discussed using BSTEM for streambank A potential use of BSTEM for stream stabilization includes com-
stabilization by presenting a case study of the model for such bining site-scale BSTEM bank erosion predictions based on
purposes. The following guidelines were recommended for bank stabilization strategies with wide-scale reach information
using the model as part of a stream stabilization project: (i) to determine watershed scale sediment load reductions. Re-
worst case conditions (e.g. no riparian protection, high water search is needed to understand when and how to extrapolate
table, and no toe protection) should be tested to first determine site-specific BSTEM simulations to calculate stream-scale sedi-
if stabilization is even necessary; (ii) if stabilization is necessary, ment loads and load reductions. An example based on a previ-
BSTEM simulations can be completed with various mitigation ous study and data from the Fort Cobb Reservoir watershed in
strategies for worst case conditions until a stable design is deter- southwest Oklahoma (Klavon, 2016) is presented to demon-
mined for the site specific characteristics; (iii) after a stable de- strate both the potential and the uncertainty in using this
sign is determined for worst case scenarios, long-term method to extrapolate the results of calibrated BSTEM simula-
simulations should be performed to quantify retreat and sedi- tions. Sediment from unstable streambanks in the watershed is
ment load reductions due to various mitigation strategies; and one of the primary contributors of loading to the reservoir
(iv) BSTEM reductions can be combined with reach information (Figure 3). Four monitoring sites in the watershed were selected
to determine reach scale sediment load reductions. The model along two tributaries, Fivemile and Willow Creeks, to the

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
202 KLAVON K. ET AL.

Figure 3. Unstable streambank in the Fort Cobb Reservoir watershed. [Colour figure can be viewed at wileyonlinelibrary.com]

reservoir from which the fluvial and geotechnical properties of sampling at a number of sites in the watersheds and therefore
the streambank were estimated. were generated using Monte Carlo methods. The ls and aver-
The range of stream-scale sediment loads and load reduc- age lr were assumed known, and the same values were used
tions for three different stabilization practices simulated in for all simulations. The second approach utilized calibrated
BSTEM was performed using a Monte Carlo simulation (Dunn BSTEM simulations to calculate L10yr:
and Shultis, 2011). The practices included rock toe protection,
vegetation and grading only, and a combination of rock toe
L10yr ¼ LBSTEM l s (5)
protection, grading, and vegetation. Further details on the
BSTEM study, including site information, data collection,
model setup and calibration, and stabilization practice model- where LBSTEM is the 10-year sediment load per meter of bank
ing is reported in Klavon (2016). Statistical distributions were fit predicted by calibrated BSTEM models on nine sites in the
to the independent parameters in Minitab 16 (Minitab, 2009), watershed (in kg m1). For each monitoring site of the two
and were considered an acceptable fit if the p-value was tributaries, 10-year simulations were performed to generate
greater than 0.05. When two or more distributions were accept- LBSTEM using the Monte Carlo methods. The ls was again
able, the one with the lowest Anderson–Darling (AD) statistic known. Table II shows Monte Carlo distributions for the three
was chosen. In addition, distributions were checked to ensure Monte Carlo parameters (BH, ρb, and LBSTEM for both Willow
that negative values were not reflected in the distribution for and Fivemile Creeks) used to calculate sediment loads for
non-negative parameters. Once distributions were determined, both methods. Note that both approaches did not account
probability density function (PDF) and cumulative density func- for bed sediment dynamics due to the method of bank retreat
tion (CDF) of the parameters were obtained. Simulation param- estimation and limitations of BSTEM.
eter data were then generated with a minimum of 1000 random Figure 4 shows the results of the Monte Carlo simulation for
numbers between zero and one (Driels and Shin, 2004), setting calculating sediment loads. The range of sediment loads was
the random number equal to the previously determined CDF, much smaller for the aerial imagery method (Equation (4)) for
and back calculating the associated simulation parameter. both creeks; however, average loads were within the same
A 10-year sediment load was calculated using two different order of magnitude. To potentially decrease the range of pre-
approaches along the two tributaries. The first approach dicted loads for the BSTEM method (Equation (4)), more sites
utilized aerial imagery to estimate the 10-year load, L10yr (in would need to be modeled. For Fivemile Creek, the BSTEM
kilograms): method (Equation (5)) predicted higher loads than the aerial im-
agery method (Equation (4)). For Willow Creek, the opposite
L10yr ¼ Laerial l s (4)
was true. For all but the Fivemile Creek BSTEM method
(Equation (5)), using average data nearly matched the Monte
where ls is the length of the stream (in meters) and Laerial is Carlo simulation averages.
the 10-year sediment load per meter of bank from aerial im- Three mitigation strategies (toe riprap, vegetation and grad-
ages (in kg m1) calculated as the product of BH, τ b, and lr, ing only, and a combination of the two) were modeled at sites
where BH is the bank height (in meters), ρb is the average experiencing retreat as part of another Monte Carlo analysis
bank bulk density (in kg m3), and lr is the average lateral for potential total load reduction. The following equation was
bank retreat along the stream determined from aerial imagery used to calculate the maximum 10-year load reduction, LR10yr
(in meters). The BH and ρb are unknown variables based on (in kilograms), for each stream:

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 203

Table II. Parameter distributions and Monte Carlo inputs for bank height, BH; bulk density, ρb; and BSTEM predicted 10-year sediment loads,
LBSTEM, based on predictions from Klavon (2016)

Variable Creek Distribution Inputs for Monte Carlo* p Value AD value

BH Fivemile Weibull αa = 4.67; βb = 4.33 > 0.25 0.368


Willow Gamma α = 57.88; β = 0.07 0.061 0.725
ρb Fivemile Loglogistic β = 0.04 ; μc = 0.37 > 0.25 0.368
Willow 3-Parameter Weibull α = 5468.55; β = 527.89; γd = 526.47 0.383 0.340
LBSTEM Fivemile Uniform ae = 128; bf = 111919 — —
Willow Uniform a = 350; b = 25528 — —
a
α = shape.
b
β = scale.
c
μ = location.
d
γ = threshold.
e
a = minimum value.
f
b = maximum value.

Figure 4. Histogram results of the Monte Carlo simulation to estimate total sediment loads to the reservoir over the 10-year simulation. Note that MC
stands for Monte Carlo and No MC is the load if site average values were used to calculate loads.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
204 KLAVON K. ET AL.

constructed where established riparian vegetation was not


LR10yr ¼ BHρb l sf l r R % (6) already present; therefore, lsf was determined using aerial im-
agery and was considered a known. The other parameters
match those in Equation (4).
where lsf is the length of stream without a forested riparian area The results of the Monte Carlo analysis for determining max-
(in meters) and R% is the BSTEM-simulated percent load reduc- imum load reductions are shown in Figure 5. The average load
tions for the different mitigation techniques at the nine sites. for all the Monte Carlo simulations was almost identical to the
The R% was an independent variable and was generated using load reduction using solely averages. However, the ranges of
Monte Carlo methods. Distributions for R% are shown in load reduction were over an order of magnitude different,
Table III. Because the amount of R% data was limited, simple suggesting that stochastic methods should be considered when
distributions (triangular or uniform) were used. The assump- using BSTEM to predict sediment load reduction for stabiliza-
tion was made that mitigation practices would only be tion practices. It is important to note that these simulations
do not include the impact of stabilization practices on
sediment transport capacity due to the one-dimensional con-
Table III. Parameter distributions and Monte Carlo inputs for percent straint of BSTEM.
load reduction, R%, for each of the stabilization practices modeled in
Klavon (2016)

Stabilization Distribution Inputs for


Suggested improvements to BSTEM
practice Monte Carlo
General model improvements
Toe riprap only Triangular aa = 53.9; bb = 100;
mc = 100 In order to improve BSTEM as a tool for streambank stabiliza-
Toe riprap + vegetation Triangular a = 95.4; b = 100; tion design the following research and development are sug-
and grading m = 100 gested. First, from experience with BSTEM, the model runtime
Vegetation and grading Uniform a = 75.9; b = 99.9 was long and therefore the calibration process for long-term
only simulations was time intensive. Advancements have been re-
a
a = minimum value. cently made that significantly reduce simulation run time;
b
m = mode. therefore, users need to be ensure that they are using the very
c
b = maximum value. latest version of the model. With a reduced simulation time,

Figure 5. Histogram results of the Monte Carlo simulation to estimate the maximum sediment load reduction to the reservoir using the 10-year
BSTEM simulation. Note that MC stands for Monte Carlo and No MC is the load if averages were used to calculate loads.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 205

one potential improvement to simulation capability would be correction factor method is an interesting step forward, but
an option for stochastic analysis to account for uncertainty in has yet to be fully validated. Lai et al. (2015) combined BSTEM
input parameters. Second, most studies that have utilized with a two-dimensional model to more appropriately model
BSTEM routines are focused on streams in the central United sediment transport, failed material as bank protection, second-
States (Table I). While this is not surprising as this is where the ary flow, and turbulent flow. The study demonstrated the
model was developed, future research should evaluate the benefits of pairing BSTEM with a two-dimensional model by
model across a range of conditions nationwide and worldwide. reproducing the bank retreat process as well as the formation
Most previous research suggests that failed material from of the new bank shape with tension cracks on Goodwin Creek
previous bank collapses can protect the bank toe until that in northern Mississippi, one of the most extensively monitored
deposited sediment is removed during subsequent streamflow streams in the country. An important benefit of these integrated
events (Parker et al., 2011). However, Hackney et al. (2015), models may be the ability to model in-stream stabilization
while investigating the influence of submerged blocks from pre- structures such as J-hooks or bendway weirs. These integrated
vious mass failures, reported that such blocks may also deflect models show good promise for resolving BSTEMˈs weakness
flow onto the bank and increase the local applied shear stress. at inaccurately quantifying long-term sediment loads; however,
The model currently lacks a routine for estimating the protec- future studies are needed to validate the integrated models for a
tion and/or flow deflection onto the bank. With that said, in- variety of stream conditions.
cluding such capabilities could lead to being able to simulate
temporally-varying erodibility properties of various streambank
layers. Note, such capabilities have been incorporated in the Subaerial processes
BSTEM-like bank erosion algorithms of the CONCEPTS,
TELEMAC and RVR Meander models (Motta et al., 2014). Subaerial processes refer to freeze–thaw cycling, desiccation
More studies on a variety of streambank conditions are cracking, and seepage processes that weaken the bank material
needed to confirm the results of Khanal et al. (2016b) relative to fluvial erosion and geotechnical failure. Additional advance-
to the use of linear versus non-linear detachment models for ments are needed in BSTEM to account for these preparatory
cohesive sediment. Such confirmations will require updated processes to erosion and mass failure. For example, seepage
versions of BSTEM to include options to use multiple detach- erosion, or the erosion of unconsolidated sand above a more re-
ment models. As part of such an update, it may be valuable strictive bank layer, is known to be an important erosion process
for BSTEM to alert the user if simulated applied shear stress ex- in certain incised streams and can significantly contribute to
ceeds the maximum applied shear stress used to derive erod- bank failure (Fox et al., 2007; Wilson et al., 2007; Fox and Wil-
ibility parameters in field tests. son, 2010). Two studies were completed to evaluate BSTEMˈs
Also, guidance is needed on how to parameterize the model ability to account for erosion due to seepage which highlighted
to represent stabilization practices, especially combined ‘hard’ the fact that BSTEM was missing capabilities to fully predict
(e.g. riprap toe) and ‘soft’ (e.g. bioengineering stabilization) seepage erosion (Wilson et al., 2007; Lindow et al., 2009). They
practices. Being able to incorporate several practices at once, found the model could accurately predict failures of the typical
including practices both installed on the bank and also in- seepage geometries, but in order to fully incorporate the seep-
stream practices, would be a significant benefit to the user age erosion process, the model needed to consider near bank
community. groundwater flow (Wilson et al., 2007; Lindow et al., 2009).
Chu-Agor et al. (2009) proposed empirical techniques for incor-
porating seepage undercutting processes for bank failure predic-
Complex near-bank flow tions. Midgley et al. (2013) noted the importance of seepage and
fluvial processes acting in concert to create unstable
BSTEM currently represents an individual cross-section, and streambanks. Fluvial erosion was noted as a potential mecha-
calculations are based on uniform and steady flow assumptions nism to prevent permanent self-healing of seeps. Relationships
on a straight reach. Further development and validation of the for seepage undercutting can be incorporated into the bank sta-
analytical method is still needed to better simulate τ distribu- bility analysis, but will require integration with the groundwater
tions for various channel shapes, roughness, and the presence flow component of BSTEM. In order to fully integrate such
of small-scale topographic features in the streambed. Second- modeling capabilities, additional research is needed to identify
ary flows that can form due to the cross-section shape, stream appropriate methodologies for simulating bank pore-water
curvature, and common obstructions such as woody debris or pressures and groundwater flow dynamics. For instance, is there
failed bank material are not taken into account directly in an advantage to modeling Darcian flow versus simpler water
BSTEM without the use of the τ factor in Equation (3). Vegeta- balance approaches in BSTEM? In summary, BSTEM could be
tion is currently simulated in the model from only the perspec- further developed to include an option for accounting for
tive of geotechnical instability. Routines to estimate the impact subaerial processes, groundwater flow, and pore-water pres-
of vegetation and above-ground biomass on τ are needed. sures (Rinaldi et al., 2004). Such advancements will also require
Roots and sub-surface biomass also influence the detachment infiltration and evapotranspiration routines and begin to expand
of cohesive sediment and should be considered in the model. the model into considering rainfall/runoff processes. With that
Numerical studies have shown that the local effects of vegeta- said further general research will then be needed to consider
tion can propagate downstream due to the interactions be- the role of vegetation in controlling pore-water pressure distri-
tween the channel, flow, and bed response (Van de Wiel and butions in the bank.
Darby, 2004). A site analysis, as provided by BSTEM is limited
in determining the net impact (beneficial or adverse) of riparian
vegetation on channel morphology beyond the site-specific Alternatives to BSTEM for Stream-scale
critical bank being simulated. Predictions
Several recent studies have addressed some of these limita-
tions. Konsoer et al. (2016) applied BSTEM 5.4, which includes As estimating sediment transport and sediment load reductions
a correction for near-bank τ due to the effect of curvature, to due to streambank stabilization practices on a stream scale be-
two meander bends on the lower Wabash River, Illinois. This comes more and more valuable for several applications, such

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
206

Table IV. Summary of peer reviewed studies, Masterˈs theses, conference proceedings, and reports using the CONservational Channel Evolution and Pollutant Transport System, CONCEPTS) as a component of the
research or application

Study Publication type Location Purpose Major findings

Langendoen et al. (1999) Conference Proceedings Goodwin Creek, MS Evaluate the process-based bank CONCEPTS under predicted basal
failure algorithm in CONCEPTSa scour, but accurately predicted
and validate predictions using timing and dimensions of failure.
field data Slight differences in failure
dimensions were due to CONCEPTS
not modeling cantilever failure
Langendoen and Report West Papillion Simulate the effects of channel Depth of incision decreased with

Copyright © 2016 John Wiley & Sons, Ltd.


Simon (2000) Creek, NB roughness and grade control decreased roughness and grade
structures on channel incision control could reduce level of incision
and widening
Simon et al. (2002a, b) Conference Proceedings James Creek, MS Determine in-stream sediment Combined models can be used to
load, sediment sources, and predict actual and reference
“reference” condition for sediment transport rates
sediment loads using CONCEPTS
and AnnAGNPSb
Thomas et al. (2002) Conference Proceedings Yalobusha River, MS Determine stream channel Thalweg profile and channel top
response to channelization width were in close agreement
within the river system to observed data and there was
some discrepancy between
observed and simulated deposited
bed sediment.
Simon et al. (2004) Project Report Shades Creek, AL Evaluate the effects of urbanization CONCEPTS could simulate a
KLAVON K. ET AL.

on stream channel erosion with reduction in bank erosion with


AnnAGNPS and CONCEPTS added bank protection and the
combined models could provide
adequate predictions of suspended
sediment loading for planning
purposes
Thames (2005) Masterˈs Thesis Beaver Creek, TN Conduct sensitivity analysis of CONCEPTS was not sensitive to
CONCEPTS and evaluate the τ c or kd c of the bed or ϕˈd and only
effect of urbanization on stream sensitive to cˈe approaching zero
channel bed sediment
Staley (2006) Conference Proceedings Stroubles Creek, VA Compare channel erosion CONCEPTS and GWLF under predicted
predictions from SWATf, GWLFg, the sediment load from channel
and CONCEPTS to field data erosion, SWAT significantly over
predicted channel erosion, and
CONCEPTS under predicted erosion
due to subaerial processes dominating
in the watershed and the modelˈs
inability to simulate subaerial processes
Yuan et al. (2006) Conference Proceedings James Creek, MS Combine AnnAGNPS and CONCEPTS
to determine sources and magnitudes

(Continues)

Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)


Table 4. (Continued)

Study Publication type Location Purpose Major findings

of sediment loads in a watershed to 70–90% of sediment originated from


assist in the development of a TMDLh the erosion of unstable streambanks
with in the James Creek Watershed
Carter (2007) Masterˈs Thesis Abrams Creek, TN Compare predictions of stability of a Natural Channel Design and CONCEPTS
stream reach using empirical and had high degrees of uncertainty due
analytical approaches to professional judgement of field data
collection and interpretation of output
Langendoen (2007) Conference Proceedings Goodwin Creek, MS Integrate CONCEPTS with REMMi Combined model simulated effects of

Copyright © 2016 John Wiley & Sons, Ltd.


vegetation on pore water pressures
and there was a need for improved
groundwater model in CONCEPTS
Wells et al. (2007) Journal Article Kalamazoo River, MI Investigate impact of dam removal Dam removal scenario predicted
on sediment dynamics twelve times the amount of sediment
compared to no dam removal
Langendoen and Journal Article Flume study Validate CONCEPTS simulation CONCEPTS accurately simulated bed
Alonso (2008) of stream flow hydraulics and and water surface profiles and bed
evolution of graded streams grain size distributions
Nieber et al. (2008) Final Project Report Knife River, MN Determine sediment sources and CONCEPTS over predicted bed erosion
Submitted to Minnesota compare CONCEPTS and BEHIj when uncalibrated, CONCEPTS and
Pollution Control Agency BEHI were in reasonable agreement,
and both models had great uncertainty
with respect to input data
Peterson et al. (2010) Conference Proceedings Mower County, MN Coupled CONCEPTS CONCEPTS simulated greatest amount
with SWMMk to evaluate the of incision in the upstream reaches,
stability of a two stage drainage but did not simulate any bank
ditch that conveys agricultural instabilities within the drainage
drainage channel.
Schwartz and Drumm (2010) Report to US Department New River, TN Use CONCEPTS along with Combined models helped to identify
of Interior, Office of AnnAGNPS and fine sediment major sediment contributors and
Surface Mining sampling to develop a sediment understand the impact of BMPsm
budget on the watershed scale
l
and conduct a CHIA for surface
mining permitting
Ramirez-Avila et al. (2011) Conference Proceedings Yonaba Creek, MS Assess the performance of CONCEPTS accurately predicted top
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM)

CONCEPTS in simulating temporal bank retreat and the timing and


and spatial changes in streambanks magnitude of streambank failures
Daly et al. (2013) Conference Proceedings Barren Fork Creek, OK Evaluate CONCEPTS for use Erodibility parameters could be
in areas with composite streambanks modified to represent sinuosity, ϕ’
and perform a sensitivity analysis was the most sensitive parameter,
on input parameters and CONCEPTS could be used to
select stabilization sites and evaluate
various stabilization practices

(Continues)

Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)


207
208

Table 4. (Continued)

Copyright © 2016 John Wiley & Sons, Ltd.


Study Publication type Location Purpose Major findings

Sutarto et al. (2014) Journal Article Clear Creek, IA Investigate bank heterogeneity and Bulk density, cohesion, internal angle
estimate bank retreat as a result of of friction, erodibility, and porosity
fluvial erosion, mass wasting, or varied in the vertical direction;
the combination of both processes heterogeneity must be accounted
for to accurately simulate bank stability;
and fluvial erosion was an important
factor in bank stability analysis.
a
CONservational Channel Erosion and Pollutant Transport System (CONCEPTS).
b
Annualized Agricultural Non-Point Source Pollution Model (AnnAGNPS).
c
τ c and kd are erodibility parameters of the excess shear stress equation (Ariathurai and Arulanandan, 1978) of the soil.
d 0
ϕ is the effective angle of friction of the soil.
e
cˈ is the effective cohesion of the soil.
KLAVON K. ET AL.

f
Soil and Water Assessment Tool (SWAT).
g
Generalized Watershed Loading Function (GWLF).
h
Total Maximum Daily Load (TMDL).
i
Riparian Ecosystems Management Model (REMM).
j
Bank Erosion Hazard Index (BEHI).
k
Storm Water Management Model (SWMM).
l
Cumulative Hydrologic Impact Assessment (CHAI).
m
Best Management Practice (BMP).

Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)


INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 209

as developing total maximum daily loads, transitioning to a processes and may under predict erosion in parts of the water-
model developed for stream-scale use may be more appropri- shed where subaerial processes dominate (Langendoen,
ate than extrapolating BSTEM results. The USDA-ARS devel- 2000). The authors suggest keeping the limitations of single
oped the CONCEPTS model, with bank erosion algorithms cross-section models for uses in determining the sediment
similar to BSTEM, to be used as a tool to evaluate stream stabi- transport dynamics and bank erosion rates on a stream net-
lization techniques (Langendoen, 2000, 2011; Langendoen work scale.
and Alonso, 2008). CONCEPTS models bank stability and Recently, the BSTEM/CONCEPTS suite of bank erosion algo-
fluvial erosion processes using the same routines as BSTEM, rithms have also been implemented in the two-dimensional,
but considers the entire cross-section and includes sediment depth-averaged computer models of river morphodynamics:
transport of the bed, allowing for vertical bed adjustment. RVR Meander (Motta et al., 2012, 2014), SRH-2D (Lai et al.,
Additionally, CONCEPTS considers multiple cross-sections 2015), and TELEMAC (Langendoen et al., 2016). Two-
and incorporates sediment transport processes between cross- dimensional models can more accurately simulate the applied
sections and is therefore more appropriate for predictions at a τ acting on streambanks and at higher spatial resolution. Impor-
larger scale. While BSTEM relies only on the stage of the water tantly, unlike one-dimensional models, two-dimensional
in the channel, CONCEPTS simulates one-dimensional, un- models can simulate the interactions between laterally-varying
steady flow along the centerline of the channel using the Saint bed morphodynamics (such as bar dynamics) and bank ero-
Venant equations. A more detailed description of CONCEPTS is sion. Such improvements in quantifying the applied forces act-
available in previously published work (Langendoen, 2000; ing on streambanks necessitates improved measuring
Langendoen and Alonso, 2008). techniques to characterize the spatial distribution and magni-
CONCEPTS requires information on channel and floodplain tude of bank-material resistance-to-erosion properties.
geometry, soil properties, soil layering, and channel and flood-
plain roughness for each cross-section as well as water and
sediment discharge information at the upstream boundary.
Streambank soil parameter inputs are the same as those used Conclusions
in BSTEM and are also used to describe the bed sediment.
CONCEPTS can be integrated with a watershed model, such BSTEM in its current form is one of the most robust, compre-
as AnnAGNPS (Simon et al., 2002a; Thames, 2005; Yuan hensively studied, and widely used process-based models
et al., 2006; Schwartz and Drumm, 2010), SWMM (Peterson available for predicting streambank erosion and failure, but still
et al., 2010), or SWAT to determine the water and sediment has several areas that need to be addressed so that the model
discharge at the upstream boundary. CONCEPTS predicts can be even more broadly applicable and accessible to a wider
changes in cross-sectional geometry and bed elevation, range of users. Major technical issues that need to be addressed
sediment discharge and yield, velocity, stage, factor of safety, include accounting for spatial and temporal variability in geo-
average τ, and bed sediment size (Langendoen, 2000). technical failure and fluvial erodibility parameters, incorporat-
CONCEPTS has been used in a variety of different applica- ing subaerial and other site specific processes into the model,
tions (Table IV) including the evaluation of two-stage drainage and accounting for the effect of riparian vegetation and roots
ditch design (Peterson et al., 2010), investigation of the impact on bank shear stress distributions and erodibility parameters. Is-
of dam removal on sediment transport rates (Wells et al., 2007), sues to address that affect BSTEM usability and inclusion in
determining the response of a stream to channelization stream stability practice include decreasing run-time for long-
(Thomas et al., 2002), and evaluating the effectiveness of stabi- term modeling, incorporating and parametrizing a larger suite
lization practices (Langendoen and Simon, 2000; Langendoen, of both in-stream and streambank practices, and confirming
2007; Daly et al., 2013). the preliminary results of integrated multi-dimensional models.
Studies have also shown CONCEPTS to accurately predict Several streambank erosion and failure studies have been com-
streambank retreat and flow hydraulics. Langendoen et al. pleted using BSTEM, ranging from specifically bank stability
(1999) determined that CONCEPTS accurately predicted timing modeling to evaluating several factors that influence erosion
and geometry of failure. Ramirez-Avila et al. (2011) also and failure such as the vegetation, pore-water pressures, seep-
showed that CONCEPTS predicted top bank retreat and the age, and streambank stabilization practices. In recent years
time and magnitude of streambank failures that were observed the model has been integrated with multi-dimensional flow
from field data. Thomas et al. (2002) determined that CON- and sediment transport models to address some of the major is-
CEPTS accurately predicted hydraulics parameter and erosion sues associated with the one-dimensional nature of the model.
and deposition patterns without calibration. General rates and Two-dimensional hydraulic models combined with BSTEM
trends of morphological changes were also correctly simulated. show promise in addressing some of the one-dimensional re-
In some cases CONCEPTS does not agree with measured re- lated problems within BSTEM, such as determining shear stress
sults, which highlights an area where future work is needed distributions due to sinuosity and vegetation. The model in de-
(Staley, 2006). sign is limited to the site scale and extrapolating site-specific re-
As an alternative solution to CONCEPTS, Gibson et al. sults to larger scales introduces uncertainty. When larger scale
(2015) recently integrated BSTEM 5.4 with HEC-RAS 5.0, and predictions are desired, models that incorporate sediment
showed that the two models were able to appropriately predict transport mechanisms such as CONCEPTS may be preferred.
observed cross-sections over a long-term simulation. This code
offers several advantages over CONCEPTS, as it is a common Acknowledgements—The authors acknowledge the financial support
tool used in engineering with several online support forums of the Buchanan Family Trust through the Buchanan Endowed Chair
and includes greater modeling capabilities, such as multi- and the Oklahoma Agricultural Experiment Station at Oklahoma State
University. This project was also supported by National Integrated Wa-
reaches and hydraulic structures modeling.
ter Quality Program Project Grant No. 2013-51130-21484 from the
As with any hydraulic model, it is important to consider the USDA National Institute of Food and Agriculture. The authors acknowl-
limitations of the model. For instance, CONCEPTS and HEC- edge the USDA-ARS Grazingslands Research Laboratory for providing
RAS were designed to model straight channels or channels flow input data for the BSTEM simulations in the Fort Cobb reservoir.
with low sinuosity and should be used with caution in highly The authors acknowledge Amanda Fox for reviewing an earlier version
sinuous channels. CONCEPTS does not simulate subaerial of this manuscript and two anonymous reviewers.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
210 KLAVON K. ET AL.

References Darby SE, Alabyan AM, Van de Wiel MJ. 2002. Numerical simulation of
bank erosion and channel migration in meandering rivers. Water
Al-Madhhachi AT, Hanson GJ, Fox GA, Tyagi AK, Bulut R. 2013a. Mea- Resources Research 38(9): 1163. DOI:10.1029/2001WR000602.
suring soil erodibility using a laboratory “mini” JET. Transactions of Darby SE, Gessler D, Thorne CR. 2000. Computer program for stability
the ASABE 56(3): 901–910. analysis of steep, cohesive riverbanks. Earth Surface Processes and
Al-Madhhachi AT, Hanson GJ, Fox GA, Tyagi AK, Bulut R. 2013b. De- Landforms 25: 175–190.
riving parameters of a fundamental detachment model for cohesive Darby SE, Thorne CR. 1996. Stability analysis for steep, eroding, cohe-
soils from flume and jet erosion tests. Transactions of the ASABE sive riverbanks. Journal of Hydraulic Engineering 122: 443–454.
56(2): 489–504. Darby SE, Trieu HQ, Carling PA, Sarkkula J, Koponen J, Kummu M,
Ariathurai R, Arulanandan K. 1978. Erosion rates of cohesive soils. Jour- Conlan I, Leyland J. 2010. A physically based model to predict hy-
nal of the Hydraulics Division 104(HY2): 279–283. draulic erosion of fine-grained riverbanks: The role of form roughness
Bai S, Lung WS. 2005. Modeling sediment impact on the transport of in limiting erosion. Journal of Geophysical Research 115: F04003.
fecal bacteria. Water Research 39(20): 5232–5240. DOI:10.1029/2010JF001708.
Bechtum TM. 2012. Automation and further development of the bore- Demartinecourt JP, Bauer GE. 1983. The modified borehole shear de-
hole shear test. Masterˈs thesis. Iowa State University, Ames, IA. vice. Geotechnical Testing Journal 6(1): 24–29.
Blaisdell FW, Clayton LA, Hebaus CG. 1981. Ultimate dimension of lo- Driels MR, Shin YS. 2004. Determining the Number of Iterations for
cal scour. Journal of the Hydraulics Division – ASCE 107(3): 327–337. Monte Carlo Simulations of Weapon Effectiveness. No. NPS-MAE-
Briaud J, Ting F, Chen H, Cao Y, Han S, Kwak K. 2001. Erosion function 04-005. Naval Postgraduate School, Department of Mechanical
apparatus for scour rate predictions. Journal of Geotechnical and and Astronautical Engineering: Monterey, CA.
Geoenvironmental Engineering 127(2): 105–113. DOI:10.1061/ Dunn WL, Shultis JK. 2011. Exploring Monte Carlo Methods. Elsevier:
(ASCE)1090-0241(2001)127:2(105). Amsterdam.
Cancienne RM, Fox GA, Simon A. 2008. Influence of seepage under- Einstein HA. 1942. Formulas for the transportation of bed-load. Trans-
cutting on the stability of root-reinforced streambanks. Earth Surface actions of the American Society of Civil Engineers 107: 561–597.
Processes and Landforms 33(11): 1769–1786. DOI:10.1002/ El Kadi Abderrezzak K, Moran AD, Mosselman E, Bouchard J-P,
esp.1657. Haberscak H, Aelbrecht D. 2014. A physical, movable-bed model
Carter DL. 2007. Stream Restoration Assessment of Abrams Creek in the for non-uniform sediment transport, fluvial erosion, and bank failure
Great Smoky Mountains National Park: Management Implications in rivers. Journal of Hyro-Environment Research 8(2): 95–114.
and Comparison of Empirical and Analytical Physical Assessment Ap- Fox GA, Sheshukov A, Cruse R, Kolar RL, Guertault L, Gesch KR,
proaches, Masterˈs Thesis. University of Tennessee, Knoxville, TN. Dutnell RC. 2016a. Reservoir sedimentation and upstream sediment
Chow VT. 1959. Open Channel Hydraulics. McGraw-Hill: New York. sources: Perspectives and future research needs on streambank and
Chu-Agor ML, Fox GA, Wilson GV. 2009. Empirical sediment transport gully erosion. Environmental Management 57(5): 945–955.
function predicting seepage erosion undercutting for cohesive bank DOI:10.1007/s00267-016-0671-9.
failure prediction. Journal of Hydrology 377: 155–164. Fox GA, Purvis RA, Penn CJ. 2016b. Streambanks: a net source of sed-
DOI:10.1016/j.jhydrol.2009.08.020. iment and phosphorus to streams and rivers. Journal of Environmental
Clark LA, Wynn TM. 2007. Methods for determining streambank criti- Management 181: 602–614. DOI:10.1016/j.jenvman.2016.06.071.
cal shear stress and soil erodibility: implications for erosion rate pre- Fox GA, Wilson GV. 2010. The role of subsurface flow in hillslope and
dictions. Transactions of the ASABE 50(1): 95–106. stream bank erosion: a review. Soil Science Society of America Jour-
Clark EH, Haverkamp JA, Chapman W. 1985. Eroding Soils: The Off- nal 74(3): 717–733. DOI:10.2136/sssaj2009.0319.
farm Impacts. Conservation Foundation: Washington, DC; 252 pp. Fox GA, Wilson GV, Simon A, Langendoen E, Akay O, Fuchs JW. 2007.
Coulthard TJ, Van De Wiel MJ. 2012. Modelling river history and evo- Measuring streambank erosion due to ground water seepage: corre-
lution. Philosophical Transactions of the Royal Society of London A: lation to bank pore water pressure, precipitation, and stream stage.
Mathematical, Physical and Engineering Sciences 370(1966): Earth Surface Processes and Landforms 32(10): 1558–1573.
2123–2142. DOI:10.1002/esp.1490.
Criswell DT, Al-Madhhachi AT, Fox GA, Miller RB. 2016. Deriving Fredlund DG, Morgenstern NR, Widger RA. 1978. The shear strength of
erodibility parameters of a mechanistic detachment model for unsaturated soils. Canadian Geotechnical Journal 15(3): 313–321.
gravels. Transactions of the ASABE 59(1): 145–151. DOI:10.13031/ Gallage C, Uchimura T. 2015. Direst shear testing on unsaturated silty
trans.59.11490. soils to investigate the effects of drying and wetting on shear strength
Crosato A. 2007. Effects of smoothing and regridding in numerical me- parameters at low suction. Journal of Geotechnical and
ander migration models. Water Resources Research. DOI:10.1029/ Geoenvironmental Engineering 142(3) 04015081-1–04015081-9.
2006WR005087. DOI:10.1061/(ASCE)GT.1943-5606.000141.
Curran JC, Hession WC. 2013. Vegetative impacts on hydraulics and Ghaneeized SM, Atkinson JF, Benett SJ. 2015. Effect of flow confine-
sediment processes across the fluvial system. Journal of Hydrology ment on the hydrodynamics of circular impinging jets: implications
505: 364–376. DOI:10.1016/j.jhydrol.2013.10.013. for erosion assessment. Environmental Fluid Mechanics 15(1):
Daly ER, Fox GA, Enlow HK, Storm DE, Hunt SL. 2015a. Site-scale var- 1–25.
iability of streambank fluvial erodibility parameters as measured with Gibson S, Simon A, Langendoen E, Bankhead N, Shelley J. 2015. A
a jet erosion test. Hydrological Processes 29(26): 5451–5464. physically-based channel-modeling framework integrating HEC-
DOI:10.1002/hyp.10547. RAS sediment transport capabilities and the USDA-ARS bank-
Daly ER, Fox GA, Al-Madhhachi AST, Storm DE. 2015b. Variability of stability and toe-erosion model (BSTEM). Paper presented at the
fluvial erodibility parameters for streambanks on a watershed scale. Federal Interagency Sediment Conference, SedHyd Proceedings.
Geomorphology 231: 281–291. Grissinger EH, Murphey JB. 1982. Present ‘problem’ of stream channel
Daly ER, Miller RB, Fox GA. 2015c. Modeling streambank erosion and instability in the bluff area of northern Mississippi. Journal of the
failure along protected and unprotected composite streambanks. Ad- Mississippi Academy of Sciences 27: 117–128.
vances in Water Resources 81: 114–127. DOI:10.1016/j. Hackney C, Best J, Leyland J, Darby SE, Parsons D, Aalto R, Nicholas A.
advwatres.2015.01.004. 2015. Modulation of outer bank erosion by slump blocks:
Daly ER, Fox GA, Fox AK. 2016. Correlating erodibility parameters disentangling the protective and destructive role of failed material
from jet erosion tests to soil physical properties at a site scale. on the three-dimensional flow structure. Geophysical Research Let-
Transactions of the ASABE 59(1): 115–128. DOI:10.13031/ ters. DOI:10.1002/2015GL066481.
trans.59.11309. Handy Geotechnical Instruments. 2013. Borehole Shear™ Test Instruc-
Daly ER, Storm DE, Fox GA, Langendoen EJ. 2013. Evaluation of the tions: Revised January 2013. Handy Geotechnical Instruments:
Conservational Channel Evolution and Pollutant Transport System Madrid, IA.
(CONCEPTS) Applied to Composite Streambanks in the Ozark High- Handy RL. 1986. Borehole shear test and slope stability. Proceedings of
lands Ecoregion, Paper 131620552, 2013 ASABE Annual Interna- In Situ 1986: ASCE Specialty Conference on Use of In-Situ Tests
tional Meeting, St Joseph, MI. DOI. 10.13031/aim.20131620552 Geotechnical Engineering, June 23–25. Blacksburg, VA; 161–175.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 211

Handy RL, Schmertmann JH, Lutenegger AJ. 1985. Borehole shear tests No. 16. US Department of Agriculture, Agricultural Research
in a shallow marine environment. In Strength Testing of Marine Service, National Sedimentation Laboratory: Oxford, MS. https://
Sediments: Laboratory and In Situ Measurements. ASTM Special www.ars.usda.gov/SP2UserFiles/Place/60600505/TechnicalReports/
Technical Publication 883ASTM Publications: West Conshohocken, NSLTechnicalReport16.pdf [9 May 2016].
PA; 140–153. Langendoen EJ. 2007. CONCEPTS: A process-based computer model
Handy RL, Fox NS. 1967. A soil bore-hole direct-shear test device. of instream and riparian processes. Proceedings, World Environmen-
Highway Research News 27: 42–51. tal and Water Resource Congress 2006. DOI. 10.1061/
Hanson GJ. 1990. Surface erodibility of earthen channels at high 40856(200)328
stresses part I – open channel testing. Transactions of the ASAE Langendoen EJ. 2011. Application of the CONCEPTS channel
33(1): 127–131. evolution model in stream restoration strategies. In Stream
Hanson GJ, Cook K. 2004. Apparatus, test procedures, and analytical Restoration in Dynamic Fluvial Systems, Simon A, Bennett SJ,
methods to measure soil erodibility in situ. Applied Engineering in Castro JM (eds). American Geophysical Union: Washington, DC;
Agriculture 20(4): 455–462. 487–502.
Hanson GJ, Simon A. 2001. Erodibility of cohesive streambeds in the loess Langendoen EJ, Alonso CV. 2008. Modeling the evolution of incised
area of the midwestern USA. Hydrological Processes 15(1): 23–38. streams: I. Model formulation and validation of flow and streambed
Heeren DM, Mittelstet AR, Fox GA, Storm DE, Al-Madhhachi AT, evolution components. Journal of Hydraulic Engineering 134(6):
Midgley TL, Stringer AF, Stunkel KB, Tejral RD. 2012. Using rapid 749–762.
geomorphic assessments to assess streambank stability in Oklahoma Langendoen EJ, Mendoza A, Abad JD, Tassi P, Wang D, Ata R, El Kadi
Ozark streams. Transactions of the ASABE 55(3): 957–968. AK, Hervouet J-M. 2016. Improved numerical modeling of
Heinley KN. 2010. Stability of Streambanks Subjected to Highly Vari- morphodynamics of rivers with steep banks. Advances in Water Re-
able Streamflows: The Osage River Downstream of Bagnell Dam, sources 93(Part A): 4–14.
Masterˈs Thesis, Paper 5022. Missouri University of Science and Langendoen EJ, Simon A. 2000. Stream Channel Evolution of Little
Technology, Rolla, MO. Salt Creek and North Branch West Papillion Creek, Eastern Ne-
James CS, Birkhead AL, Jordanova AA, OˈSullivan JJ. 2004. Flow resis- braska, Report of USDA Agricultural Research Service. National
tance of emergent vegetation. Journal of Hydraulic Research 42(4): Sedimentation Laboratory: Oxford, MS. http://www.ars.usda.gov/
390–398. SP2UserFiles/Place/60600505/AGNPS/Concepts/Doc/Nebraska.pdf
Julian JP, Torres R. 2006. Hydraulic erosion of cohesive riverbanks. [9 May 2016].
Geomorphology 76(1–2): 193–206. DOI:10.1016/j. Langendoen EJ, Simon A. 2008. Modeling the evolution of incised
geomorph.2005.11.003. streams. II: Streambank erosion. Journal of Hydraulic Engineering –
Kamphuis JW, Hall KR. 1983. Cohesive material erosion by unidirec- ASCE 134(7): 905–915. DOI:10.1061/(ASCE)0733-9429(2008)
tional current. Journal of Hydraulic Engineering 109(1): 49–61. 134:7(905).
Karmaker T, Dutta S. 2011. Erodibility of fine soil from the composite Langendoen EJ, Simon A. 2009. Closure to ‘Modeling the Evolution of
river bank of Brahmaputra in India. Hydrological Processes 25: Incised Streams. II: Streambank Erosion’ by Eddy J. Langendoen
104–111. and Andrew Simon. Journal of Hydraulic Engineering – ASCE
Karmaker T, Dutta S. 2015. Stochastic erosion of composite banks in al- 135(12): 1107–1108.
luvial river bends. Hydrological Processes 29: 1324–1339. Langendoen EJ, Simon A, Curini A, Alonso CV. 1999. Field validation of
DOI:10.1002/hyp.10266. an improved process-based model for streambank stability analysis.
Kean JW, Smith JD. 2004. Flow and boundary shear stress in channels Proceedings of the 1999 International Water Resources Engineering
with woody bank vegetation. In Riparian Vegetation and Fluvial Geo- Conference, ASCE, Reston, VA.
morphology, Bennett SJ, Simon A (eds). American Geophysical Langendoen EJ, Zegeye AD, Tebebu TY, Steenhuis TS, Ayele GK,
Union: Washington, DC; 237–252. Tilahun SA, Ayana EK. 2014. Using computer models to design gully
Kean JW, Smith JD. 2006. Form drag in rivers due to small-scale natural erosion control structures for humid northern Ethiopia. Proceedings
topographic features: 1. Regular sequences. Journal of Geophysical of the 11th International Conference on Hydroscience Engineering,
Research 111: F04009. DOI:10.1029/2006JF000467. September 28–October 2, Hamburg; 1137–1145.
Khanal A, Fox GA, Al-Madhhachi AT. 2016a. Variability of erodibility Lindow N, Fox GA, Evans RO. 2009. Seepage erosion in layered stream
parameters from laboratory mini jet erosion tests. Journal of Hydro- bank material. Earth Surface Processes and Landforms 34(12):
logic Engineering 21(10): 04016030. DOI:10.1061/(ASCE)HE.1943- 1693–1701. DOI:10.1002/esp.1874.
5548.0001404. Lohnes RA, Handy RL. 1968. Slope angles in friable loess. Journal of
Khanal A, Klavon KR, Fox GA, Daly ER. 2016b. Comparison of linear Geology 76(3): 247–258.
and nonlinear models for cohesive sediment detachment: Rill ero- López F, García M. 1998. Open-channel flow through simulated vege-
sion, hole erosion test, and streambank erosion studies. Journal of tation: suspended sediment transport modeling. Water Resources Re-
Hydraulic Engineering 142(9): 04016026. DOI:10.1061/(ASCE) search 34(9): 2341–2352. DOI:10.1029/98WR01922.
HY.1943-7900.0001147. Lu N, Likos WJ. 2004. Unsaturated Soil Mechanics. John Wiley & Sons:
Kheiralla KM, Siddeg AS. 2015. Control over river bank erosion: a case Hoboken, NJ; 556 pp.
study of Ganetti Station, Northern States, Sudan. Journal of Earth Sci- Lutenegger AJ. 1987. Suggested method for performing the borehole
ence & Climatic Change 6: 287. DOI:10.4172/2157-7617.1000287. shear test. Geotechnical Testing Journal, ASTM 10(1): 19–25.
Khodashenas SR, El Kadi Abderrezzak K, Paquier A. 2008. Boundary Lutenegger AJ, Hallberg GR. 1981. Borehole shear test in geotechnical
shear stress in open channel flow: a comparison among six methods. investigations. Laboratory Shear Strength of Soil: ASTM Special Tech-
Journal of Hydraulic Research 46(5): 598–609. nical Publication 740. R. N. Young and F. C. Townsend, Eds.
Klavon KR. 2016. Advances in Modeling Cohesive Sediment Detach- 566–578.
ment and a Process-Based Method for Stream Restoration to Deter- Lutenegger AJ, Tierney KF. 1986. Pore pressure effects in borehole shear
mine Sediment Loads, Masterˈs Thesis. Oklahoma State University, testing. In-situ 1986. In Proceedings of ASCE Special Conference on
Stillwater, OK. the Use of In Situ Tests in Geotechnical Engineering, Clemence SP
Konsoer KM, Rhoads BL, Langendoen EJ, Best JL, Ursic ME, Abad JD, (ed) , Geotechnical Special Publication 6. ASCE: Blacksburg, VA;
Garcia MH. 2016. Spatial variability in bank resistance to erosion 752–764.
on a large meandering, mixed bedrock-alluvial river. Geomorphol- McNeil J, Taylor C, Lick W. 1996. Measurements of erosion of undis-
ogy 252: 80–97. DOI:10.1016/j.geomorph.2015.08.002. turbed bottom sediments with depth. Journal of Hydraulic Engineer-
Lai YG, Thomas RE, Ozeren Y, Simon A, Greimann BP, Wu KW. 2015. ing 122(6): 316–324. DOI:10.1061/(ASCE)0733-9429(1996)
Modeling of multilayer cohesive bank erosion with a coupled bank 122:6(316).
stability and mobile-bed model. Geomorphology 243: 116–129. Midgley TL, Fox GA, Heeren DM. 2012. Evaluation of the bank stability
DOI:10.1016/j.geomorph.2014.07.017. and toe erosion model (BSTEM) for predicting lateral retreat on com-
Langendoen EJ. 2000. CONCEPTS – CONservational Channel Evolu- posite streambanks. Geomorphology 145: 107–114. DOI:10.1016/j.
tion and Pollutant Transport System version 1.0, Research Report geomorph.2011.12.044.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
212 KLAVON K. ET AL.

Midgley TL, Fox GA, Wilson GV, Heeren DM, Langendoen E, Simon A. root-reinforcement the whole story? Geomorphology 116(3–4):
2013. Seepage-induced streambank erosion and instability: In situ 353–362. DOI:10.1016/j.geomorph.2009.11.013.
constant-head experiments. Journal of Hydrologic Engineering Polvi LE, Wohl E, Merritt DM. 2014. Modeling the functional influence
18(10): 1200–1210. DOI:10.1061/(ASCE)HE.1943-5584.0000685. of vegetation type on streambank cohesion. Earth Surface Processes
Millar RG. 2000. Influence of bank vegetation on alluvial channel pat- and Landforms 39(9): 1245–1258. DOI:10.1002/esp.3577.
terns. Water Resources Research 36(4): 1109–1118. Posner AJ, Duan JG. 2012. Simulating river meandering processes using
Miller GA, Azad S, Haskell CE. 1998. Iowa borehole shear testing in stochastic bank erosion coefficient. Geomorphology 163–164:
unsaturated soil. Geotechnical site characterization. Proceedings of 26–36.
the First International Conference on Site Characterization, Robertson Ramirez-Avila JJ, McAnally WH, Langendoen EJ, Achury SLO, Martin
PK, Wayne PW (eds), 19–22 April, Atlanta, GA; 1321–1326. JL. 2011. The role of streambank erosion contributions to sediment
Miller GA, Khoury CN. 2012. Observations from borehole shear testing loads in the Town Creek Watershed in Mississippi. In International
in unsaturated soil. Proceedings of the Second European Conference Symposium on Erosion and Landscape Evolution (ISELE), 18–21 Sep-
on Unsaturated Soils, Mancuso C, DˈOnza F, Jommi C (eds), June, tember 2011, Anchorage, Alaska. American Society of Agricultural
Naples; 113–122. and Biological Engineers: St Joseph, MI.
Miller RB, Fox GA, Penn C, Wilson S, Parnell A, Purvis RA, Criswell K. Rinaldi M, Casagli N. 1999. Stability of streambanks formed in partially
2014. Estimating sediment and phosphorus loads from streambanks saturated soils and effects of negative pore water pressures: the Sieve
with and without riparian protection. Agriculture, Ecosystems, and River (Italy). Geomorphology 26(4): 253–277.
Environment 189: 70–81. DOI:10.1016/j.agee.2014.03.016. Rinaldi M, Casagli N, Dapporto S, Gargini A. 2004. Monitoring and
Minitab. 2009. Minitab Statistical Software. Version 16. Minitab Inc.: modelling of pore water pressure changes and riverbank stability
State College, PA. during flow events. Earth Surface Processes and Landforms 29:
Moore WL, Masch FD, Jr. 1962. Experiments on the scour resistance of 237–254.
cohesive sediments. Journal of Geophysical Research 67(4): Rinaldi M, Darby SE. 2007. Modelling of river-bank-erosion processes
1437–1446. DOI:10.1029/jZ067i004p01437. and mass failure mechanisms: progress towards fully coupled simula-
Motta D, Abad JD, Langendoen EJ, Garcia MH. 2012. A simplified 2D tions. Developments in Earth Surface Processes 11: 213–239.
model for meander migration with physically-based bank evolution. Rinaldi M, Nardi L. 2013. Modeling interactions between riverbank hy-
Geomorphology 163–164: 10–25. drology and mass failures. Journal of Hydrologic Engineering 18(10):
Motta D, Langendoen EJ, Abad JD, García MH. 2014. Modification of 1231–1240. DOI:10.1061/(ASCE)he.1943-5584.0000716.
meander migration by bank failures. Journal of Geophysical Re- Rosgen DL. 1996. Applied River Morphology. Wildland Hydrology
search: Earth Surface 119(5): 1026–1042. Inc.: Pagosa Springs, CO.
Naot D, Nezu I, Nakagawa H. 1996. Hydrodynamic behavior of partly Rosgen, D.L. 2001. A practical method of computing streambank ero-
vegetated open channels. Journal of Hydraulic Engineering 122(11): sion rate. Proceedings of the Seventh Federal Interagency Sedimenta-
625–633. tion Conference, March 25–29, Reno, NV; Vol. 2: 9–15.
Nardi L, Campo L, Rinaldi M. 2013. Quantification of riverbank erosion Rousselot P. 2009. Discussion of ‘Modeling the evolution of incised
and application in risk analysis. Natural Hazards 69: 869–887. streams. II: Streambank erosion’ by Eddy J. Langendoen and Andrew
DOI:10.1007/s11069-013-0741-8. Simon. Journal of Hydraulic Engineering – ASCE 135(12):
Newton SE, Drenten DM. 2015. Modifying the Bank Erosion Hazard In- 1107–1107.
dex (BEHI) protocol for rapid assessment of streambank erosion in Schoneboom T, Aberle J, Dittrich A. 2010. Hydraulic resistance of veg-
northeaster Ohio. Journal of Visualized Experiments 96: e52330. etated flows: contribution of bed shear stress and vegetative drag to
DOI:10.3791/52330. total hydraulic resistance. Proceedings, International Conference on
Nieber JL, Wilson BN, Ulrich JS, Hansen BJ, Canelon DJ. 2008. Assess- Fluvial Hydraulics (River Flow 2010), 8–10 September,
ment of Streambank and Bluff Erosion in the Knife River Watershed, Braunschweig.
Final Report. Minnesota Pollution Control Agency, University of Min- Schumm SA, Lichty RW. 1963. Channel Widening and Flood-plain
nesota: St Paul, MN. Construction along Cimarron River in Southwestern Kansas, US Geo-
Osman MA, Thorne CR. 1988. Riverbank stability analysis: I. Theory. logical Survey Professional Paper 352–D. USGS: Reston, VA.
Journal of Hydraulic Engineering 114: 134–150. DOI:10.1061/ Schwartz JS, Drumm EC. 2010. Method Development for Watershed
(ASCE)0733-9429(1988)114:2(134). Sediment Budgets to Support the CHIA/PHC Process: A Focus on Sed-
Papanicolaou AN, Elhakeem M, Hilldale RC. 2007. Secondary current iment Modeling for Estimating Sediment Loads, Final Report. US De-
effects on cohesive river bank erosion. Water Resources Research partment of Interior, Office of Surface Mining Appalachian Region,
43(12): W12418. Knoxville Field Office: Knoxville, TN. http://www.osmre.gov/pro-
Parker C, Simon A, Thorne CR. 2008. The effects of variability in bank grams/TDT/appliedScience/2007UT-SchwartzWatershedFR.pdf [9
material properties on riverbank stability: Goodwin Creek, Missis- May 2016].
sippi. Geomorphology 101(4): 533–543. DOI:10.1016/j. Simon A. 1989. A model of channel response in disturbed alluvial
geomorph.2008.02.007. channels. Earth Surface Processes and Landforms 14(1): 11–26.
Parker G, Shimizu Y, Wilkerson GV, Eke EC, Abad JD, Lauer JW, Paola Simon A, Bingner RL, Langendoen EJ, Alonso CV. 2002a. Actual and
C, Dietrich WE, Voller VR. 2011. A new framework for modeling the Reference Sediment Yields for the James Creek Watershed, Missis-
migration of meandering rivers. Earth Surface Processes and Land- sippi, National Sedimentation Laboratory Research Report 31. Na-
forms 36(1): 70–86. tional Sedimentation Laboratory: Oxford, MS; 185 pp.
Peterson JR, Wilson BN, Kramer G. 2010. Two-stage ditch assessment Simon A, Collison AJC. 2002. Quantifying the mechanical and hydro-
using the CONCEPTS model. Proceedings, American Society of Agri- logic effects of riparian vegetation on streambank stability. Earth Sur-
cultural and Biological Engineers Annual International Meeting 2010. face Processes and Landforms 27(5): 527–546. DOI:10.1002/
ASABE: St Joseph, MI. esp.325.
Piégay H, Darby SE, Mosselman E, Surian N. 2005. A review of tech- Simon A, Curini A, Darby SE, Langendoen EJ. 2000. Bank and near-
niques available for delimiting the erodible river corridor: a sustain- bank processes in an incised channel. Geomorphology 35(3–4):
able approach to managing bank erosion. River Research and 193–217. DOI:10.1016/s0169-555x(00)00036-2.
Applications 21(7): 773–789. Simon A, Downs PW. 1995. An interdisciplinary approach to evalua-
Pollen N, Simon A. 2005. Estimating the mechanical effects of riparian tion of potential instability in alluvial channels. Geomorphology
vegetation on stream bank stability using a fiber bundle model. Water 12(3): 215–232.
Resources Research 41(7): 11. DOI:10.1029/2004wr003801. Simon A, Klimetz L. 2008. Magnitude, frequency, and duration rela-
Pollen-Bankhead N, Simon A. 2009. Enhanced application of root- tions for suspended sediment in stable (“reference”) southeastern
reinforcement algorithms for bank-stability modeling. Earth Surface streams. Journal of the American Water Resources Association
Processes and Landforms 34(4): 471–480. DOI:10.1002/esp.1690. 44(5): 1270–1283.
Pollen-Bankhead N, Simon A. 2010. Hydrologic and hydraulic effects Simon A, Langendoen EJ, Bingner R, Wells R, Yuan Y, Alonso C. 2004.
of riparian root networks on streambank stability: is mechanical Suspended-sediment transport and bed-material characteristics of

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)
INSIGHTS ON THE BANK STABILITY AND TOE EROSION MODEL (BSTEM) 213

Shades Creek, Alabama and Ecoregion 67: Developing water-quality Thorne CR, Tovey NK. 1981. Stability of composite river banks. Earth
criteria for suspended and bed-material sediment. U.S. Department Surface Processes and Landforms 6(5): 469–484.
of Agriculture Research Service, National Sedimentation Laboratory, Tolhurst TJ, Black KS, Shayler SA. 1999. Measuring the in situ erosion
Technical Report 43, Oxford, MS. shear strength of intertidal sediments with the Cohesive Strength Me-
Simon A, Pollen N, Langendoen E. 2006. Influence of two woody ripar- ter (CSM). Estuarine, Coastal and Shelf Science 49(2): 281–294.
ian species on critical conditions for streambank stability: Upper DOI:10.1006/ecss.1999.0512.
Truckee River, California. Journal of the American Water Resources Triplett LD, Engstrom DR, Edlund MB. 2009. A whole-basin strati-
Association 42: 99–113. DOI:10.1111/j.1752-1688.2006.tb03826.x. graphic record of sediment and phosphorus loading to the St. Croix
Simon A, Pollen-Bankhead N, Mahacek V, Langendoen E. 2009. Quan- River, USA. Journal of Paleolimnology 41(4): 659–677.
tifying reductions of mass-failure frequency and sediment loadings United States Department of Agriculture (USDA). 2016. Bank Stability
from streambanks using toe protection and other means: Lake Tahoe, and Toe Erosion Model, v5.4, Model documentation. USDA-ARS:
United States. Journal of the American Water Resources Association Oxford, MS.
45(1): 170–186. DOI:10.1111/j.1752-1688.2008.00268.x. United States Environmental Protection Agency (US EPA). 2015. Na-
Simon A, Pollen-Bankhead N, Thomas RE. 2011. Development and ap- tional Summary of State Information. https://ofmpup.epa.gov/wa-
plication of a deterministic bank stability and toe erosion model for ters10/attains_nation_cy.control#status_of_data [25 April 2016].
stream restoration. In Stream Restoration in Dynamic Fluvial Systems, Van De Wiel MJ, Darby SE. 2004. Numerical modeling of bed topogra-
Simon A, Bennett SJ, Castro JM (eds). American Geophysical Union: phy and bank erosion along tree-lined meandering. In Riparian Veg-
Washington, DC. DOI. 10.1029/2010GM001006 etation and Fluvial Geomorphology, Bennett SJ, Simon A (eds).
Simon A, Rinaldi M. 2006. Disturbance, stream incision, and channel American Geophysical Union: Washington, DC; 267–282. DOI.
evolution: the roles of excess transport capacity and boundary mate- 10.1029/008WSA19
rials in controlling channel response. Geomorphology 79(3): 361–383. Walling DE, Owens PN, Leeks GJ. 1998. The role of channel and flood-
Simon A, Thomas RE. 2002. Processes and forms of an unstable alluvial plain storage in the suspended sediment budget of the River Ouse,
system with resistant, cohesive streambeds. Earth Surface Processes Yorkshire, UK. Geomorphology 22(3): 225–242.
and Landforms 27(7): 699–718. DOI:10.1002/esp.347. Wan CF, Fell R. 2004. Laboratory test on the rate of piping erosion on
Simon A, Thomas RE, Curini A, Shields FD. 2002b. Case study: channel soils in embankment dams. Geotechnical Testing Journal 27(3):
stability of the Missouri River, eastern Montana. Journal of Hydraulic 295–303.
Engineering – ASCE 128(10): 880–890. DOI:10.1061/(ASCE)0733- Wells RR, Langendoen EJ, Simon A. 2007. Modeling pre-and post-dam
9429(2002)128:10(880). removal sediment dynamics: the Kalamazoo River, Michigan.
Simon A, Thomas RE, Klimetz L. 2010. Comparison and experiences JAWRA – Journal of the American Water Resources Association
with field techniques to measure critical shear stress and erodibility 43(3): 773–785.
of cohesive deposits. Proceedings, 2nd Joint Federal Interagency Wilson BN. 1993a. Development of a fundamentally based detach-
Conference, Las Vegas, NV. ment model. Transactions of the ASAE 36(4): 1105–1114.
Staley NA. 2006. Modeling Channel Erosion at the Watershed Scale: A Wilson BN. 1993b. Evaluation of a fundamentally based detachment
Comparison of GWLF, SWAT, and AGNPS/CONCEPTS, Paper num- model. Transactions of the ASAE 36(4): 1115–1122.
ber 062016, 2006 ASAE Annual Meeting. ASAE: St Joseph, MI. Wilson CA, Yagci O, Rauch HP, Stoesser T. 2006. Application of the
Sutarto T, Papanicolaou AN, Wilson CG, Langendoen EJ. 2014. Stabil- drag force approach to model the flow-interaction of natural vege-
ity analysis of semicohesive streambanks with CONCEPTS: coupling tation. International Journal of River Basin Management 4(2):
field and laboratory investigations to quantify the onset of fluvial ero- 137–146.
sion and mass failure. Journal of Hydraulic Engineering. Wilson GV, Perketi RK, Fox GA, Dabney SM, Shields FD, Cullum RF.
DOI:10.1061/(ASCE)HY.1943-7900.0000899. 2007. Soil properties controlling seepage erosion contributions to
Thames BA. 2005. The Effects of Urbanization on the Streambed Sedi- streambank failure. Earth Surface Processes and Landforms 32(3):
ment Characteristics in a Ridge and Valley Watershed, Masterˈs The- 447–459. DOI:10.1002/esp.1405.
sis. University of Tennessee, Knoxville, TN. Wynn TM, Mostaghimi S. 2006. The effects of vegetation and soil type
Thapa I. 2016. Study of Streambank Erosion Processes and Bank Stabil- on streambank erosion, southwestern Virginia, USA. Journal of the
ity Condition of the Kodku River, Kathmandu Valley, Central Nepal, American Water Resources Association 42(1): 69–82. DOI:10.1111/
MSc Thesis. Tribhuvan University, Institute of Science and Technol- j.1752-1688.2006.tb03824.x.
ogy, Nepal. DOI. 10.13140/RG.2.1.1639.2727 Wynn TM, Henderson MB, Vaughan DH. 2008. Changes in streambank
Thomas RE, Langendoen EJ, Bingner RL, Simon A. 2002. Channel re- erodibility and critical shear stress due to subaerial processes along a
sponses and management strategies in disturbed channels: a numerical headwater stream, southwestern Virginia, USA. Geomorphology 97:
simulation approach. Proceedings, 2002 Conference on Water 260–273. DOI:10.1016/j.geomorph.2007.08.010.
Resources Planning and Management, Roanoke, VA. ASCE: Reston, VA. Yuan Y, Bingner RL, Langendoen E, Wells R, Simon A, Alonso C. 2006.
Thorne CR, Abt SR. 1993. Analysis of riverbank stability due to toe Determination of Watershed Sediment Sources using USDA Water-
scour and lateral erosion. Earth Surface Processes and Landforms shed and Channel Models, Paper number 062048, 2006 ASAE An-
18(9): 835–843. nual Meeting. ASAE: St Joseph, MI.

Copyright © 2016 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms, Vol. 42, 191–213 (2017)

Das könnte Ihnen auch gefallen