Catena 69 (2007) 197 – 205

www.elsevier.com/locate/catena

Temporal and spatial variability in root reinforcement of streambanks:

Accounting for soil shear strength and moisture
Natasha Pollen⁎
Channel and Watershed Processes Research Unit, USDA–ARS National Sedimentation Laboratory, PO Box 1157, Oxford, MS 38655, United States
Received 18 November 2005; received in revised form 8 May 2006; accepted 15 May 2006

Abstract

Riparian vegetation exerts a number of mechanical and hydrologic controls on bank stability, which can affect the delivery of sediment to
channels. Estimates of root reinforcement of soils have commonly been attained using perpendicular root models that simply sum root tensile
strengths and consider these as an add-on factor to soil strength. A major limitation of such perpendicular models is that tensile strength and
resistance is wrongly considered to be independent of soil type and moisture, and therefore variations according to these bank properties are
omitted in conventional models. In reality, during mass failure of a streambank, some roots break, and some roots are pulled out of the soil
intact; the relative proportions of roots that break or pull out are determined by a combination of soil moisture and shear strength. In this
paper an equation to predict the frictional resistance of root–soil bonds was tested against field data collected at Long Creek, MS, under two
soil moisture conditions. The root pullout equations were then included in the root-reinforcement model, RipRoot, and bank stability model
runs for Goodwin Creek, MS, were carried out in order to examine the effects of spatial and temporal variations in soil shear strength and
rooting density, on streambank factor of safety. Model results showed that at smaller root diameters breaking forces exceeded pullout forces,
but at larger root diameters pullout forces exceed breaking forces. The threshold diameter between root pullout and root breaking varied with
soil shear strength, with increasing soil shear strength leading to a greater proportion of roots failing by breaking instead of pullout. Root-
reinforcement estimates were shown to reflect changes in soil shear strength, for example, brought about by variations in soil matric suction.
Resulting Factor of safety (FS) values for the bank during the period modeled ranged from 1.36 to 1.74 with 1000 grass roots/m2, compared
to a range of 0.97 to 1.37 for the non-vegetated bank. Root reinforcement was shown to increase bank stability under the entire range of soil
moisture conditions modeled. However, the magnitude of root reinforcement varied in both space and time as determined by soil shear
strength and soil moisture.
Published by Elsevier B.V.

Keywords: Root reinforcement; Soil–root friction forces; Streambank stability; Soil shear strength; Soil moisture

1. Introduction (USEPA, 2002). It is therefore critical to understand and

Streambank instability poses a number of economic and
ecological problems including land loss and destabilizing of
structures (e.g. bridges), while the delivery of excessive
sediment to channels can cause downstream aggradation and
impairment of water quality (Casagli et al., 1999). Sediment
is one of the principal pollutants of surface waters in the
United States, and recent studies have shown that stream
bank materials are a significant if not dominant source
⁎ Tel.: +1 662 281 5712.
E-mail address: npollen@ars.usda.gov.
0341-8162/$ - see front matter. Published by Elsevier B.V.
doi:10.1016/j.catena.2006.05.004
quantify the key controlling processes and more accurately
predict the role vegetation can play in stabilizing stream-
banks, thereby reducing sediment generation.
The presence of riparian vegetation on streambanks not
only provides ecological benefits, but is also widely believed
to increase the stability of streambanks (Abernethy and
Rutherfurd, 1998; Simon and Collison, 2002). A number of
studies have shown that lateral channel erosion is particularly
affected by vegetation type and density (e.g. Graf, 1978;
Andrews, 1984; Hey and Thorne, 1986), but the complexity
of natural, vegetated streams often makes it difficult to
establish a direct causal relation between riparian vegetation
198 N. Pollen / Catena 69 (2007) 197–205
and channel characteristics (Gran and Paola, 2001). As a
result of the potential benefits for stability and environmental
quality, interest in the use of riparian vegetation in stream
restoration and stabilizing schemes has grown in recent
years. However, quantification of the interactions between
vegetation and bank erosion processes has been unreliable
due to limited knowledge regarding the way in which roots
affect the geotechnical, hydrological and hydraulic processes
within a bank and its adjoining channel. Current stream-
restoration designs are, therefore, based on empirical
methods and standardized practices (Gregory and Gurnell,
1988; Wynn et al., 2004), with process-based models
examining vegetation largely unavailable or untested.
Riparian vegetation exerts a number of controls on bank
stability. These controls can be separated into mechanical
and hydrologic effects, some beneficial and some negative to
bank stability. The mechanical effects are for the most part
beneficial. Roots anchor themselves into the soil to support
above-ground biomass, producing a reinforced soil matrix. A
negative impact of vegetation on bank stability is the weight
of the vegetation, particularly mature trees. These forces add
additional load to the streambank, increasing the driving
forces acting on a bank, and thereby reducing bank stability.
The beneficial hydrological impacts of vegetation on
bank stability include vegetative canopy interception, which
decreases the amount of water available for infiltration, and
extraction of water from the root zone which lowers the local
phreatic surface (Selby, 1993), and increases matric suction
within streambanks (Simon and Collison, 2002). The
hydrologic disadvantages of vegetation on soil stability are
related to soil infiltration characteristics both at the soil
surface, and deeper within the soil profile. At the surface,
canopy interception and stem flow tend to concentrate
rainfall locally around the stems of plants, creating locally
higher pore-water pressures (Durocher, 1990). The presence
of stems and roots at the soil surface also act to increase
infiltration rate and capacity, at times accelerating the
delivery of water at depth into the slope or bank by creating
preferential flow paths (Simon and Collison, 2002). The
extent to which preferential flow affects bank stability
depends on properties of the bank materials and the
stratigraphy of the bank.
To quantify vegetation effects, several properties of the
root network need to be established. Possibly the most
challenging aspect of root investigations is the acquisition
of data pertaining to root architecture and root distributions
throughout the bank. The inherent complexity of root
networks and the spatial and temporal variability of the
soil in which they grow (Nielson et al., 1997), and their
opaque growing medium, creates numerous impediments
to data collection and sampling. Despite the difficulties
concerned with obtaining certain root data, a number of
papers have attempted to measure and quantify the
physical properties of riparian root networks and relate
these properties to streambank stability. These include the
studies conducted in Mississippi by Simon and Collison
(2002), Easson and Yarbrough (2002), Collins (2001), in
Virginia by Wynn et al. (2004), in California by Simon et
al. (2006), in various parts of the USA by Pollen et al.
(2004), and in Australia by Abernethy and Rutherfurd
(1998, 2000). In addition, the hydrologic effects of riparian
vegetation on bank stability have been explored by Simon
and Collison (2002), Simon and Pollen (2004) and Pollen
(2004).
1.1. Quantifying root reinforcement of soil
It is generally accepted that plant roots provide reinforce-
ment to a soil matrix due to the different material properties
of soil and roots (Greenway, 1987). Soil is strong in
compression but weak in tension. Conversely, roots are weak
in compression but strong in tension. The presence of roots
in the soil thus produces a reinforced matrix in which stress
in transferred to the roots during loading of the soil (Thorne,
1990) in a way that is similar to the reinforcement of concrete
structures by steel and fiberglass. Several factors can affect
the root-reinforcement of a soil, including root density, root
branching, Young's modulus and tensile strength of the roots
(Greenway, 1987), and root moisture content (Collins,
2001).
Estimates of root reinforcement of soils have commonly
been attained using simple perpendicular root models such as
those of Waldron (1977) and Wu et al. (1979), that calculate
root reinforcement as an add-on factor to soil strength. The
root reinforcement model of Waldron (1977) is based on the
Coulomb equation in which soil shearing resistance is
calculated from cohesive and frictional forces:
S ¼ c þ rN tan/ ð1Þ
where S is soil shearing resistance (kPa), σN is the normal
stress on the shear plane (kPa), ϕ is soil friction angle
(degrees), and c is the cohesion (kPa).
Waldron (1977) extended Eq. (1) for root-permeated
soils, by assuming that all roots extended vertically across a
horizontal shearing zone, and the roots act like laterally
loaded piles, so tension is transferred to them as the soil is
sheared. The modified Coulomb equation becomes:
S ¼ c þ DS þ rN tan/ ð2Þ
where ΔS is increased shear strength due to roots (kPa).
In the Waldron (1977) model, the tension developed in the
root as the soil is sheared is resolved with a tangential
component resisting shear and a normal component
increasing the confining pressure on the shear plane. ΔS
can be represented by:
DS ¼ Tr ðsinh þ coshtan/ÞðAR =AÞ ð3Þ
where Tr is average tensile strength of roots per unit area of
soil (kPa), AR/A is the root area ratio (no units). The root area
 N. Pollen / Catena 69 (2007) 197–205
ratio is the cross sectional area of roots crossing a plane
within the soil.
Gray (1974) reported the angle of internal friction of the
soil appeared to be affected little by the presence of roots.
Sensitivity analyses carried out by Wu et al. (1979) showed
that the value of the bracketed term in (3) (sin θ + cos θ tan
ϕ), is fairly insensitive to normal variations in θ and ϕ
(40–90° and 25–40° respectively) with values ranging from
1.0 to 1.3. A value of 1.2 was therefore selected by Wu et al.
(1979) to replace the bracketed term and the simplified
equation becomes:
DS ¼ 1:2 Tr ðAR =AÞ ð4Þ
Thus, according to the simple perpendicular root model of
Wu et al. (1979), the magnitude of reinforcement simply
depends on the amount and strength of roots present in the
soil. However, Waldron and Dakessian (1981) suggested that
these perpendicular root models tended overestimate root
reinforcement due to the inherent assumption that the full
tensile strength of each root is mobilized during soil
shearing, and that the roots all break simultaneously. Field
and laboratory tests carried out by Pollen (2004) and Pollen
et al. (2004) confirmed the presence of this overestimation
effect when using perpendicular root models. This overes-
timation was largely corrected by Pollen and Simon (2005)
by constructing a fiber-bundle model (RipRoot) to account
for progressive breaking during mass failure.
Observations of incised streambanks suggest however,
that when a root-reinforced soil shears, two mechanisms of
root failure occur: (1) root breaking and (2) root pullout. The
anchorage of individual leek roots was studied by Ennos
(1990), who developed a function for pullout forces based on
the strength of the bonds between the roots and soil:
FP ¼ Sd Ld 2kr ð5Þ
where FP is the pullout force for an individual root (N), S is
soil shear strength (kPa), r is the radius of the root (m) and L
is the length of the root (m). L can be estimated in the
absence of field data using:
L ¼ R rg ð6Þ
(Waldron and Dakessian, 1981), where the constants g and R
have ranges: 0.5 < g < 1.0; 200 < R < 1000, with values
varying according to a number of factors, for example
plant species and growing conditions.
For the saturated part of a bank profile the Coulomb
equation can be used to calculate S (Eq. (1)). For the
unsaturated part of the bank profile the criterion as modified
by Fredlund et al. (1978) can be used:
S ¼ c Vþ ðr−la Þtan/ þ ðla −lw Þtan/ b
ð7Þ
where c′ is effective cohesion (kPa), σ is normal stress
(kPa), μa is pore air pressure (kPa), μw is pore-water
199
pressure (kPa), (μa − μw) is matric suction, or negative pore-
water pressure (kPa), and tan ϕb is the rate of increase in
shear strength with increasing matric suction.. The quantity
(μa − μw) tan ϕb represents the additional strength provided
by matric suction along the unsaturated part of the failure
plane and is reflected in the apparent or total cohesion (ca)
term (although this does not signify that matric suction is a
true form of cohesion) (Fredlund and Rahardjo, 1993):
ca ¼ c Vþ ðla −lw Þtan/b ð8Þ
In root-reinforcement calculations, root breaking is
generally considered to be independent of soil moisture.
Work by Collins (2001) has however shown that root tensile-
strength values may be dependent on soil moisture, but this
factor was not measured or considered in this study. Root
pullout forces are a function of soil shear strength, which is
determined by effective cohesion, soil friction angle, and soil
matric suction. The forces required for root pullout thus vary
spatially with material type, and temporally with variations
in soil moisture. The original version of RipRoot did not take
into account root pullout forces, and as such could not take
into account the effect of differing soil types and moistures
on estimates of root reinforcement. This was considered to
be a major deficiency of the model and the perpendicular
root models that preceded it.
This paper focuses on the processes involved in
mechanical root-reinforcement of streambanks. A major
limitation of attempts to quantify root reinforcement of
streambanks to date is that they fail to take into account some
of the most important temporal and spatial variability that
can occur within the root zone of a streambank. In particular,
the effect of variations in soil moisture and soil properties on
root reinforcement are often omitted because root tensile
strengths are considered to be independent of soil type and
moisture. Such a reductionist view of the way that roots
interact with the soil matrix leads to oversimplification of the
processes occurring in a root-reinforced bank during bank
erosion and mass-failure processes (Pollen, 2004). In this
paper, the forces required to pullout roots were investigated
in a field study, and the results were tested against Eq. (5)
(Ennos, 1990). Root pullout forces were then compared to
root breaking forces obtained from tensile strength testing,
and the RipRoot model was modified to account for both
root-failure mechanisms. Results from the newly constructed
model are presented to show the variability in root
reinforcement as soil matric suction changed through a
wetting and drying cycle following a storm event.
2. Methods
Root tensile strengths and root pullout forces were
measured using a Root Puller device (based on a design by
Abernethy, 1999). The puller is comprised of a metal frame
with a winch attached to a load cell and displacement
transducer, both connected to a datalogger. A trench was dug
200 N. Pollen / Catena 69 (2007) 197–205

Fig. 1. Map showing locations of the Goodwin Creek and Long Creek fieldsites in Mississippi.

using an excavator to expose the roots of a stand of mature
river birch (Betulus nigra) trees located on the banks of Long
Creek, MS (Fig. 1). The Root Puller was attached to the side
of the trench opposite each tree, so that the roots being tested
were no longer anchored by the tree. This was necessary
because if the roots were still anchored to the tree, the forces
being measured would be the tensile strength of the roots; by
pulling roots embedded in the soil, but not attached to the
tree, we were able to measure the force required to break the
frictional bond between the soil and roots. The excavation
process may have loosened the bond between the roots and
the soil at the trench interface, but any disturbance did not
appear to extend more than a few millimeters into the soil,
thus limiting the effect on the root–soil friction forces
measured. The device was attached to the side of the trench
by tying a rope around metal stakes hammered into the
ground along the top of the trench. The load cell measured
the force required to either pull the root out of the soil, or for
it to break. Root–soil friction experiments were carried out at
the same site, at different times of the year (April and July
2002) to test under varying soil moisture conditions. In each
experiment 70 to 80 roots were tested to have sufficient data
to distinguish any trends in the data from natural variability.
3. Results and discussion
Hypothetical pullout forces were estimated using Eq. (5)
and compared to root breaking forces, calculated from tensile
strength–diameter relations of the form:
y ¼ ax−b ð9Þ
previously published in Pollen et al., 2004 and Pollen and
Simon, 2005), where y is root tensile strength (MPa) and x is
root diameter (mm), and a and b are constants.

Fig. 2. Diagrammatic representation of hypothetical root pullout forces compared to root breaking forces.
 N. Pollen / Catena 69 (2007) 197–205 201

Comparison of the variation in pullout and breaking forces

with increasing root diameters (range of root diameters was
0.5 to 8 mm) showed that for roots of a given riparian species
growing in a soil of a given shear strength, at small root
diameters breaking forces exceeded pullout forces, but at
larger root diameters pullout forces exceed breaking forces
(Fig. 2). Therefore smaller diameter roots are more likely to
be pulled out of the soil, and larger diameter roots are more
likely to break as a soil shears. The threshold between the two
failure mechanisms is a function of the shear strength of the
soil and thus the strength of frictional bonds between the roots
and soil, and plant species-specific tensile strength. Care was
taken to only test river birch roots. These were identifiable
from other species due to color and texture differences
between roots of different species.

3.1. Field data results

The field data collected were used to examine whether a

clear threshold existed between the two root-failure
mechanisms. Field data results from Long Creek, MS are
presented in Fig. 3. The data show that above a certain root
diameter, all of the roots broke. Below this threshold
diameter some of the roots broke and some were pulled
out of the soil intact. This suggests that the theoretical
threshold diameter between root pullout and root breaking
does exist, above which the friction between the soil and
roots exceeds the root tensile strength. Below the threshold,
if the force required to break the root–soil friction bond is
less than the force required to break the root, the roots were
pulled out of the soil. The breaking of roots below the
threshold diameter may have been due to the presence of root
Fig. 3. Field data for root breaking and root pullout at the Long Creek site,
branching, causing the friction forces to vary from the MS. In (a) April and (b) July. Also shown are the predicted breaking and
idealized situation of straight, unbranched roots as assumed pullout forces calculated for the apparent soil cohesion at the site (6 and
in Eq. (5). The presence of branches off the main root would 8 kPa for April and July respectively), and the constants for root length, R
have increased the surface area in contact with the soil, hence (200) and g (0.7) determined from the field data.
increasing the root–soil friction bond to an extent where the
force required to break the root was smaller than that to pull 2.9 mm respectively), and the maximum root diameters that
the root out of the soil. could be pulled out of the soil were different. During the April
Moisture contents of 21.1% and 11.3% were experienced tests, the threshold value was approximately 3.5 mm whereas
during the root–soil friction tests conducted in April and July in the July tests the threshold value was approximately
2002, respectively. Samples were taken from the mid-point 2.4 mm. This change in threshold diameter for root pullout
of the rooted layer being tested (0.5 m), and were weighed, suggests that as the soil dries out the frictional bonds between
oven-dried and reweighed to obtain percent moisture the roots and the soil become stronger, as apparent cohesion
content. Statistical analysis confirmed that there was not a of the soil increases with increasing soil matric suction. Root
significant difference between the median root breaking breaking is therefore likely to be the predominant mode of
forces for the April and July data (Mann–Whitney Rank root-failure in dry soils or those with higher shear strengths,
Sum Test, p = 0.220). This result shows that the tensile whereas soils that are moist or have lower shear strength will
strengths of the roots studied, and therefore their breaking exhibit greater root pullout than breaking up to a certain root
forces, were largely independent of soil moisture. diameter. These results are similar to those of Ennos (1990)
Change in soil moisture did, however, affect the threshold who tested pullout resistance of leek radicles.
diameter between root pullout and breaking. Although the
mean diameters of roots that pulled out of the soil during the 3.2. Inclusion of pullout forces in the RipRoot model
April and July tests were not significantly different (t-test,
p = 0.221), the range of root diameters that were pulled out Root-pullout forces were included in the progressive
of the soil was less in July than in April (range of 1.7 mm and breaking algorithm of the RipRoot model using Eqs. (5) and
202 N. Pollen / Catena 69 (2007) 197–205
(6). Root lengths for the six species included in the RipRoot
model (Pollen and Simon, 2005) were calculated using
values in the suitable range outlined by Waldron and
Dakessian (1981) (R = 200 and g = 0.7 are typical suggested
values). Root tensile strength curves used to calculate root
breaking forces for each species were obtained from Pollen
and Simon (2005). The first set of model runs adding pullout
forces, were conducted using a specified number of roots for
each species, using varying soil shear-strengths. The results
of these model runs are shown in Fig. 4.
Frictional bonds between roots and soil were smaller for
soils with low shear strength. Thus small diameter roots may
be more easily pulled from the soil than broken. As soil shear
strength increased, the threshold diameter between root
pullout and root breaking decreased. Therefore, at higher soil
shear strengths, more roots were predicted to break than be
pulled out of the soil, and above a certain soil shear strength
all roots were predicted to break. The location of this
threshold soil shear strength also varied according to the
species considered because of variations in root tensile
strengths between species. Using Fig. 4a and b it is possible
to determine the threshold diameter (indicated by leveling off
Fig. 4. Variations in (a) threshold diameter for pullout and (b) root
reinforcement for each riparian species, with changing soil shear strength.
Modeled species were River birch (Betula nigra), Eastern sycamore
(Plantanus occidentalis), Black willow (Salix nigra), Longleaf pine (Pinus
palustris Miller), Switchgrass (Panicum virgatum) and Western cottonwood
(Populus fremontii).
in Fig. 4a and b) above which all roots are predicted to break.
Thus, the effects of both soil type and soil moisture can now
be accounted for in the RipRoot model.
3.3. Effect of changing soil matric suction on root
reinforcement values
An additional set of model runs were performed using
data from the Goodwin Creek Experimental Bendway
(GCEB) in Mississippi (Fig. 1) to examine temporal
variability in root reinforcement estimates before and after
a storm event (December 7–23, 2004). The period of study
selected was chosen because tensiometer data exhibited a
drying curve followed by a large storm event during which
soil matric suction dropped rapidly. To estimate soil shear
strength (using Eq. (7)), the following variables were
required: effective soil cohesion, pore-water pressure, σ, ϕ′
and ϕb. The GCEB was used for these model runs because
geotechnical properties were available (c′ = 1.4 kPa,
ϕb = 17°, ϕ′ = 28.5° for the upper 1 m of the bank profile
that contains the root zone, Simon et al., 1999). In addition,
detailed matric-suction data were available from nests of
tensiometers installed in different vegetative treatments
(Simon and Collison, 2002). Normal stress (σ) was
calculated for a soil depth of 0.5 m (the assumed center of
the root-reinforced zone), and using the regression equation
established in Simon et al. (1999) that relates soil matric
suction to soil unit weight for the upper 1 m of the modeled
bank.
Mean daily soil shear strength was calculated using the
bank geotechnical properties and mean matric suction value
for each day in the top layer of the grass plot, during the
period December 7–23, 2004. The values of soil shear
strength were used in the RipRoot model to calculate daily
root-reinforcement values for root densities of 500 and 1000
roots/m2. The mean daily root-reinforcement estimate for
1000 roots/m2 and the soil matric suction record for the
period studied are shown in Fig. 5. Changes in root-
reinforcement values reflect changes in soil matric suction,
since c′, ϕ′ and ϕb remain constant for a given soil. In
theory, as matric suction decreases towards the point of soil
saturation, apparent cohesion approaches zero, as the value
of the term (μa − μw) tanϕb is equal to zero at the point of soil
saturation. Therefore, when the soil is saturated, the effect of
root reinforcement on bank stability is likely to be greatly
reduced, although the soil will maintain some shear strength
as effective friction and cohesion remain.
In the case of the upper 1 m of the Goodwin Creek bank
profile, c′ was small (1.4 kPa), and values of matric suction
recorded by the tensiometers, in combination with soil
friction angle produced soil shear-strength values ranging
from 7.3 to 10.7 kPa during the period. Soil shear strength
was therefore in the range of soil strengths in which most of
the roots were predicted to break during soil shearing, with
very few roots being pulled out from the soil (Fig. 4a and b).
Variations in this range of soil strengths produced almost
 N. Pollen / Catena 69 (2007) 197–205 203

Fig. 5. Soil matric suction and root-reinforcement at 0.3 m depth for the grass plot at Goodwin Creek during December 2004, assuming 1000 roots per m2 and soil
shear strengths ranging from 7.32 to 10.71 kPa.

indistinguishable (approximately 1–2%) variations in root-
reinforcement, ranging from 5.55 to 5.61 kPa (500 root/m2)
and 11.10 to 11.22 kPa (1000 roots/m2) (Fig. 5). Slight
increases in root reinforcement were seen with increasing
soil matric suction until soil shear strength increased above
10 kPa, at which point root reinforcement values leveled off
because all roots were predicted to break (as seen in Fig. 4).
Further increases in matric suction above 10 kPa, therefore,
had no effect on root reinforcement.
To test the effect of changing matric suction on a soil with
weaker shear strength, RipRoot runs were repeated using
mean daily shear-strength values that were half those
calculated for the first set of runs. The results of this second
set of runs (Fig. 6) showed that because soil strength values
were in the range where both root breaking and root pullout
occurred (Fig. 4a and b), any variation in matric suction had
an effect on calculated root reinforcement values. In this
case, changes in matric suction were reflected in root-
reinforcement values with no plateau in root-reinforcement
values as soil shear strengths were not great enough to force
all roots to break. Root-reinforcement estimates for the lower
range of soil shear strengths (3.66 to 5.36 kPa) varied from
9.12 to 11.2 kPa at a root density of 1000 roots/m2.
Therefore, the effect of soil type and moisture on root
reinforcement depends on spatial and temporal variability
and relative contributions of effective and apparent cohesion,
matric suction and soil friction angle to soil shear strength.
3.4. Effect of dynamic root-reinforcement values on
Streambank Factor of Safety (FS)
The streambank stability model of Simon et al. (1999)
was run for the grass plot at Goodwin Creek for the period
December 7–23, 2004. Previously published model runs
using this bank stability model have included root
reinforcement as a static add-on factor, assuming that all

Fig. 6. Soil matric suction and root-reinforcement at 0.3 m depth for the grass plot at Goodwin Creek during December 2004, assuming 1000 roots per m2, and
soil shear strengths varying from 3.66 to 5.36 kPa.
204 N. Pollen / Catena 69 (2007) 197–205
Fig. 7. Factor of safety values for the grass plot at Goodwin Creek, MS, during December 2004, with no roots, 500 roots/m2 and 1000 roots/m2 at higher soil shear
strengths (7.3 to 10.7 kPa) and lower soil shear strengths (3.66 to 5.36 kPa).
roots broke and that root reinforcement was independent of
soil type and moisture (Simon and Collison, 2002; Pollen et
al., 2004). As such, the addition of vegetation to any bank
produces FS for bare and vegetated banks that show sets of
parallel lines, with vegetated banks having higher FS values.
Root-reinforcement estimates including root pullout as well
as root breaking, result in root-reinforcement forces that vary
with soil shear strength. The effect of temporal variability in
root-reinforcement values on streambank FS values is shown
in Fig. 7. The FS for vegetated and non-vegetated banks
diverges slightly as FS increases (in this case the difference
in FS was 0.01 to 0.05; Fig. 7). This trend occurs because
root reinforcement is greatest when matric suction and thus
FS are high. Root reinforcement is lowest when bank FS
conditions are critical, although some reinforcement is still
provided through soil–root friction forces even when pullout
is the dominant root failure mechanism. Critical conditions
generally occur during the receding limb of a hydrograph,
when the bank material is saturated, and the confining forces
acting on the banks during high flows are removed as flow
stage recedes.
FS values where soil shear strength was high and root
reinforcement thus varied only slightly, showed that FS for
the vegetated and non-vegetated bank diverged slowly as FS
increased. The difference in minimum FS was 0.35 and the
difference in maximum FS was 0.36 for a density of 1000
roots/m2 (Fig. 7). Fig. 4 also shows the FS runs repeated with
root reinforcement estimates for the soil of lower shear
strength. In this second set of model runs, at minimum FS the
difference in FS between the bare bank and the bank with
1000 roots/m2 was 0.31, but at maximum FS, the difference
was 0.36. At lower soil shear strengths the range of root
reinforcement values was thus reflected in a wider range of
FS values. Results of model runs show that for a specified
rooting density, the range of root-reinforcement values is
dependent on how soil shear strengths affect the predomi-
nance of species specific root pullout and root breaking
mechanisms. Fig. 7 shows that the variations in model runs
for soil of lower and higher shear strengths were greatest at
lower FS values. This suggests that the inclusion of pullout
forces is particularly important when bank stability condi-
tions are critical.
4. Conclusions
Field tests confirmed the presence of a threshold root-
diameter, above which all roots broke under applied stress,
and below which, both root breaking and pullout occurred.
The threshold diameter between root-failure mechanisms is
determined by the shear strength of the soil. Model results
have thus shown that the effect of soil shear strength and
moisture on root reinforcement is dependent on spatial and
temporal variability in several factors including soil matric
suction and geotechnical properties. Results showed that root
reinforcement was at its lowest when soil moisture was high
and soil shear strength was low. Therefore, root reinforce-
ment is likely to be at a minimum value when a bank is most
at risk of failure, for example during the receding limb of a
hydrograph, when the bank may be saturated, but the
receding flow-level removes the confining force acting on
the bank. The fact that root reinforcement is likely to be at its
lowest during critical bank-failure conditions, further high-
lights the overestimation of root-reinforcement calculated
using simple perpendicular models such as that of Wu et al.
(1979). The potential range of spatial and temporal
variability in root-reinforcement should be thus considered
 N. Pollen / Catena 69 (2007) 197–205
carefully when assessing risk with regards to streambank
failures.
When soil shear strengths were calculated for Goodwin
Creek, root reinforcement values during December 2004
ranged from 5.55 to 5.61 kPa (500 grass roots/m2) and 11.10
to 11.22 kPa (1000 grass roots/m2). The root-reinforcement
values varied very little throughout the period of time
modeled because root breaking was predominant at that
range of soil shear strengths; changes in matric suction
therefore had little or no effect on root-reinforcement
estimates. When soil shear strength was halved, root-
reinforcement estimates decreased correspondingly, but the
range of values was greater as both root pullout and root
breaking occurred in this range of soil shear strengths. Any
changes in soil matric suction therefore affected the
proportion of roots that would break or pullout of the soil,
thereby causing larger variations in root-reinforcement
estimates.
Quantification of spatial and temporal variability in root-
reinforcement can now be taken into account in RipRoot by
varying input variables such as root density, species
assemblage, and soil shear strength. In the first version of
RipRoot, the roots and soil were considered separately as
root breaking was assumed to be independent of soil
properties. The addition of root pullout forces in this new
version allows for interaction between the root–soil matrix,
thereby improving the representation of the processes
occurring during mass failure of a streambank. Future
work will use field data to validate the level of improvement
made by using varying root-reinforcement estimates instead
of static add-on values. More accurate estimates of root
reinforcement of streambanks will ultimately aid in future
studies of bank stability and may help explain and quantify
rates of channel widening and lateral migration of channels
with riparian corridors of various species compositions.
References
Abernethy, B., 1999. On the Role of Woody Vegetation in Riverbank
Stability. PhD Thesis. Monash University: Melbourne.
Abernethy, B., Rutherfurd, I.D., 1998. Where along a river's length will
vegetation most effectively stabilize stream-banks? Geomorphology 23,
55–75.
Abernethy, B., Rutherfurd, I.D., 2000. The effect of riparian tree roots on the
mass-stability of riverbanks. Earth Surface Processes and Landforms 25,
921–937.
Andrews, E.D., 1984. Bed-material entrainment and hydraulic geometry of
gravel-bed rivers in Colorado. Geological Society of America Bulletin
95, 371–378.
Casagli, N., Rinaldi, M., Gargini, A., Curinin, A., 1999. Pore-water pressure
and stream-bank stability: results from a monitoring site on the Sieve
River, Italy. Earth Surface Processes and Landforms 21, 1095–1114.
Collins, A.J., 2001. The role of willow root architecture and character in root
reinforcement potential. Unpublished doctoral thesis, University of
Nottingham.
Durocher, M.G., 1990. Monitoring spatial variability in forest interception.
Hydrological Processes 4, 215–229.
Easson, G., Yarbrough, L.D., 2002. The effects of riparian vegetation on
bank stability. Environmental and Engineering Geoscience 8 (4),
247–260.
205
Ennos, A.R., 1990. The anchorage of leek seedlings: the effect of root length
and soil strength. Annals of Botany 65, 409–416.
Fredlund, D.G., Rahardjo, H., 1993. Soil Mechanics of Unsaturated Soils.
John Wiley and Sons, New York.
Fredlund, D.G., Morgenstern, N.R., Widger, R.A., 1978. The shear strength
of unsaturated soils. Canadian Geotechnical Journal 15, 313–321.
Graf, W.L., 1978. Fluvial adjustments to the spread of tamarisk in the
Colorado Plateau region. Geological Society of America Bulletin 89,
1491–1501.
Gran, K., Paola, C., 2001. Riparian vegetation controls on braided stream
dynamics. Water Resources Research 37 (12), 3275–3283.
Gray, D.H., 1974. Reinforcement and Stabilization of Soil by Vegetation.
Journal of Geotechnical Engineering ASCE (100(GT6)) 695–699.
Greenway, D.R., 1987. Vegetation and Slope Stability. In: Anderson, M.G.,
Richards, K.S. (Eds.), Slope Stability. John Wiley and Sons Ltd, New
York, pp. 187–230.
Gregory, K.J., Gurnell, A.M., 1988. Vegetation and river channel form and
process. In: Viles, H. (Ed.), Biogeomorphology. Blackwell, Oxford, pp.
11–42.
Hey, R.D., Thorne, C.R., 1986. Stable channels with mobile gravel beds.
Journal of Hydraulic Engineering 112, 671–689.
Nielson, K.L., Lynch, J.P., Weiss, H.N., 1997. Fractal geometry of bean root
systems: correlations between spatial and fractal dimension. American
Journal of Botany 84 (1), 26–33.
Pollen, N., 2004. The Effects of Riparian Vegetation on streambank stability:
Mechanical and Hydrological interactions. Ph.D. Thesis. King's
College, University of London, 320 pp.
Pollen, N., Simon, A., 2005. Estimating the mechanical effects of riparian
vegetation on streambank stability using a fiber bundle model. Water
Resources Research 41, W07025, doi:10.1029/2004WR003801.
Pollen, N., Simon, A., Collision, A.J.C., 2004. Advances in Assessing the
Mechanical and Hydrologic Effects of Riparian Vegetation on
Streambank Stability. In: Bennett, S., Simon, A. (Eds.), Riparian
Vegetation and Fluvial Gomorphology: Water Science and Applications,
vol. 8, pp. 125–139.
Selby, M.J., 1993. Hillslope Materials and Processes. Oxford University
Press, Oxford.
Simon, A., Collison, A.J.C., 2002. Quantifying the mechanical and
hydrologic effects of riparian vegetation on stream-bank stability.
Earth Surface Processes and Landforms 27 (5), 527–546.
Simon, A., Pollen, N., 2004. Hydrologic controls of riparian vegetation on
the geotechnical stability of streambanks: experimental results. Proceed-
ings of the 6th Int. Conf. on Hydroscience and Engineering (ICHE-2004).
Simon, A., Curini, A., Darby, S., Langendoen, E., 1999. Stream-bank
Mechanics and the Role of Bank and Near-Bank Processes in Incised
Channels. In: Darby, S., Simon, A. (Eds.), Incised River Channels. John
Wiley and Sons, New York, pp. 123–152.
Simon, A., Pollen, N., Langendoen, E., 2006. Influence of two woody
riparian species on critical conditions for streambank stability: Upper
Truckee River, California. Journal of the American Water Resources
Association 42 (1), 99–113.
Thorne, C.R., 1990. Effects of vegetation on riverbank erosion and stability.
In: Thornes, J.B. (Ed.), Vegetation and Erosion. John Wiley and Sons
Inc, Chichester, pp. 125–143.
USEPA, 2002. http://www.epa.gov/OWOW/tmdl/sed_miles.html.
Waldron, L.J., 1977. The shear resistance of root-permeated homogeneous
and stratified soil. Journal of the Soil Science Society of America 41,
843–849.
Waldron, L.J., Dakessian, S., 1981. Soil reinforcement by roots: calculation
of increased soil shear resistance from root properties. Soil Science 132,
427–435.
Wu, T.H., McKinnell III, W.P., Swanston, D.N., 1979. Strength of tree roots
and landslides on Prince of Wales Island, Alaska. Canadian Geotech-
nical Journal 16, 19–33.
Wynn, T., Mostaghimi, S., Burger, J.A., Harpold, A.A., Henderson, M.B.,
Henry, L., 2004. Variation in Root Density along Stream Banks. Journal
of Environmental Quality 33, 2030–2039.