Geomorphology 79 (2006) 488 – 506
www.elsevier.com/locate/geomorph
Geomorphic effects of rural-to-urban land use conversion on three
streams in the Central Redbed Plains of Oklahoma
Ranbir S. Kang a , Richard A. Marston b,⁎
b
a
Department of Geography, Oklahoma State University, Stillwater, OK 74078-4073, USA
Department of Geography, 118 Seaton Hall, Kansas State University, Manhattan, KS 66506-2904, USA
Received 2 February 2006; received in revised form 6 June 2006; accepted 6 June 2006
Available online 28 August 2006
Abstract
This research evaluates the impact of rural-to-urban land use conversion on channel morphology and riparian vegetation for
three streams in the Central Redbed Plains geomorphic province (central Great Plains ecoregion) of Oklahoma. The Deep Fork
Creek watershed is largely urbanized; the Skeleton Creek watershed is largely rural; and the Stillwater Creek watershed is
experiencing a rapid transition from rural to urban land cover. Each channel was divided into reaches based on tributary junctions,
sinuosity, and slope. Field surveys were conducted at transects in a total of 90 reaches, including measurements of channel units,
channel cross-section at bankfull stage, and riparian vegetation. Historical aerial photographs were available for only Stillwater
Creek watershed, which were used to document land cover in this watershed, especially changes in the extent of urban areas
(impervious cover).
The three streams have very low gradients (b0.001), width-to-depth ratios b10, and cohesive channel banks, but have incised
into red Permian shales and sandstone. The riparian vegetation is dominated by cottonwoods, ash, and elm trees that provide a
dense root mat on stream banks where the riparian vegetation is intact. Channels increased in width and depth in the downstream
direction as is normally expected, but the substrate materials and channel units remained unchanged. Statistical analyses
demonstrated that urbanization did not explain spatial patterns of changes in any variables. These three channels in the central
Redbed Plains are responding as flumes during peak flows, funneling runoff and the wash-load sediment downstream in major
runoff events without any effect on channel dimensions. Therefore, local geological conditions (similar bedrock, cohesive
substrates and similar riparian vegetation) are mitigating the effects of urbanization.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Channel morphology; Watershed; Riparian vegetation; Channelization; Urban; Urbanization
1. Introduction
Urbanization is a human activity that modifies (directly
and indirectly) components of the landscape that can alter
flow and sediment discharge into streams. In the 1956
⁎ Corresponding author. Tel.: +1 785 532 6727; fax: +1 785 532
7310.
E-mail address: rmarston@ksu.edu (R.A. Marston).
0169-555X/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomorph.2006.06.034
classic volume, Man's Role in Changing the Face of the
Earth, Strahler (1956) and Leopold (1956) both recognized the links between watershed scale changes and
stream response. Urbanization can affect river systems in
unexpected ways (Booth and Jackson, 1997). The primary measure of urbanization in a watershed is the area
under impervious cover (May et al., 2002). Impervious
cover refers to any surface that prevents the infiltration of
water into soil (Arnold and Gibbons, 1996) and can be
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
divided into two components: roof tops (non-transportation component) and the transport network composed of
roads, driveways, and parking lots (the transportation
component) (Schueler, 1994).
The increase in impervious cover, deforestation, soil
compaction, and decreased roughness that it often implies is the most obvious manifestation of urban development (May et al., 2002). Such surfaces decrease
infiltration capacity of land and lead to higher runoff by
adding more water to streams than pre-urbanization
periods. Because water runs faster on impervious surfaces (concrete, asphalt, roof tops, roads, and streets),
construction decreases the lag time of surface runoff
(from decreased infiltration) and increases flood peaks
that affect channel morphology in different ways, such
as alterations in channel cross-sections, types of bed
materials, types of channel units, and riparian vegetation
(Morisawa and Laflure, 1979; Nanson, 1981; Booth,
1990; Booth, 1991; Johnson, 2001; Jeje and Ikeazota,
2002; May et al., 2002; Avolio, 2003; Othitis et al.,
2004; Brierley and Fryirs, 2005). Therefore, a strong
association commonly exists between the degree of
urbanization, as measured by imperviousness in a drainage basin, and the morphology of its receiving stream
(Benfield et al., 1999).
The degree of association between urbanization and
channel morphology depends on various types of impervious surfaces (Schueler, 1994; May et al., 2002;
Avolio, 2003). The transportation component (road
networks) is a particularly pervasive type of urban development impacting stream morphology. The area
covered by roads usually exceeds the area under any
other impervious surfaces by a great margin (Schueler,
1994). Roads increase runoff by delivering large
amounts of storm water into stream channels during
heavy rains (Chin and Gregory, 2001). In addition, new
road crossings cause bank erosion and affect the presence or absence of pools, large woody debris (LWD),
and type of substrate materials that deteriorate geomorphic conditions of streams (Avolio, 2003). This impact,
however, varies locally with the degree of imperviousness (urbanization) and is determined by the watershed
and adjacent riparian conditions. This research examines
the spatial variations in such impacts on three streams
with different degrees of impervious surface cover.
Based upon the degree of imperviousness this project
considers three stages of urbanization in a watershed
(see regional settings): rural (Skeleton Creek), exurban
(Stillwater Creek), and urban (Deep Fork Creek). By
convention, the rural stage is a pre-urbanization period
with less than 3% of its area under impervious cover; the
exurban stage is the transition period from rural to urban
489
with 3–10% of its area under impervious cover (Neller,
1988); whereas the urban stage is characterized by
N 10% of the watershed area under impervious cover
(May et al., 2002). We hypothesized that in the case of
the exurban stream, Stillwater Creek, urbanization is the
main reason for increasing channel cross-sectional area
downstream of Boomer Creek in relation to upstream.
Boomer Creek delivers urban sediments and runoff
generated by the City of Stillwater. We also hypothesized that the exurban stream (Stillwater Creek) would
respond in a significantly different manner from the
other two streams because of a substantial generation of
sediment from construction activities.
2. Background
2.1. Impact of urbanization on channel morphology
Impervious surfaces cause increased storm runoff,
flood frequencies, and peak discharges compared to predevelopment conditions (Booth, 1991). Streams adjust
to such increased regimes by altering stream morphology through the undercutting of banks and the deposition of sediment downstream. Debris from storm scour
blocks stream flow, straightens stream channels, and
causes stream channel enlargement. Therefore, the increased runoff and sediment supply (from impervious
surfaces) affect channel morphology by altering channel
cross-sections (Hammer, 1972). All of these alterations,
however, are variable at different locations and lead to
complex hydraulic geometry (Morisawa and Laflure,
1979).
The first and most important factor in explaining
spatiotemporal variability in such alterations in channel
cross-sections is length of time an impervious area has
been in existence. Because downstream channel enlargement requires time to take place, impervious areas
that have been in existence for 4–15 years have the
maximum impact on channel enlargements; and such
impacts decrease considerably after 30 years because of
a tendency for recovery (Hammer, 1972).
The second factor is nonequivalent channel enlargements caused by the same urban growth in different
streams. Hollis found a same increase in imperviousness
leading to dissimilar increase in cross-sections of two
streams in southeastern England (Hollis, 1976). The
main reason for such variable channel enlargements,
caused by the same degree of urbanization, is the difference in local conditions (bedrock geology, soil structure,
entrenchment ratio, and riparian vegetation). Drainage
basin area is another factor that can affect the impact of
urbanization. Even small changes in imperviousness can
490
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
have significant downstream consequences in small
drainage basins. Urbanization has also been found to
lead to a reduction in channel cross-sectional area
(Leopold, 1972; Nanson, 1981; Booth and Henshaw,
2001).
Nanson (1981) observed downstream reduction in
channel cross-section area in an urbanizing river on the
Illawara escarpment (New South Wales, Australia) and
attributed it to resistant sediments, vegetation, and a
sudden decline in channel slope and associated stream
power. Similarly, Leopold (1972) observed a slow reduction in channel cross-section area in Washington, DC,
during the first decade of urbanization. During a later
period of urbanization, channel area increased because of
increased sediment deposition caused by annual flooding.
After 20 years of observation (1953–1972), however, the
channel area showed a 20% decrease (as opposed to an
increase advocated by Hammer, 1972). Booth and
Henshaw (2001) also observed a decrease in channel
cross-sectional area in urban channels in western
Washington because of geologic conditions that limited
erosion. Also in urban streams, sinuosity is lower
(8% lower), pools are less deep (31% shallower) channel
gradients are steeper, and the substrate is more easily
erodable (Hession et al., 2002).
2.2. Impacts of urbanization on riparian vegetation
Riparian vegetation performs various functions for
streams, such as reducing sediment loads, reducing
nutrient loads, attenuating peak flow, and initiating
fluvial adjustments (Leavitt, 1998; Simon et al., 2004).
Presence of riparian vegetation enhances the stream
bank stability and increases flow resistance by disrupting flow paths. The absence or removal of riparian
vegetation, however, leads to higher rates of runoff,
erosion, and the extraction of channel morphology
(Simon et al., 2004). During urban sprawl, improper
construction and maintenance of roads leads to degradation of processes, structures, and functions of riparian
Fig. 1. The Skeleton Creek Watershed.
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
corridors. For example, channel diversions, alterations
in channel morphology, alterations in organic debris in
streams, hillslope drainage alterations, and base flow
changes are likely.
Avolio (2003) agrees that increasing road density has
an impact on channel morphology and argues that a
thick riparian corridor can help mitigate such impacts of
road crossings on channel morphology. Riparian vegetation and channel cross-sections affect each other
(Hession et al., 2002), and riparian vegetation interacts
with stream flow during urban-induced high flow periods affecting channel morphology (Leavitt, 1998).
A variety of interpretations have been made concerning the impact of grass cover versus the impact of
trees or LWD on stream banks (Trimble, 2004). Long
grass with small woody plants provides the best protection to banks, and the growth of trees and forests
increases the rate of erosion more in humid regions than
in arid regions (Trimble, 2004). Although Trimble argues
for the role and importance of riparian vegetation in
491
managing sediment budgets, his argument de-emphasizes
the role of tree roots in controlling erosion and stabilizing
streams.
Streams have a geomorphic tendency to recover from
the temporary disturbances caused by urbanization
(Booth, 1991). In the case of watersheds experiencing
different degrees of urbanization, channel morphological recovery occurs at variable rates. In urban watersheds, the increased magnitudes and frequencies of peak
flows may inhibit geomorphic recovery, so urban
streams may not have enough time to return to their
pre-urban morphology. Most previous studies have
concentrated on small watersheds (b 100 km2). The
present study focuses on three watersheds that are
somewhat larger than earlier studies. Also, these watersheds are experiencing different levels of development
but similar local conditions. Because the focus is on
relatively large watersheds, the unpredictability of
results associated with small watersheds is minimized
(Booth and Henshaw, 2001).
Fig. 2. The Stillwater Creek Watershed.
492
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
3. Regional settings
This project studies three different streams: Skeleton
Creek, Stillwater Creek, and Deep Fork Creek (Figs. 1–3),
each experiencing different degrees of urbanization, rural,
exurban/urbanizing, and urban, respectively (Tables 1 and
2). All the three watersheds selected for this study are
marked by a sub-humid climate (Cfa) with a slight decline
in moisture as one moves westward. The average annual
climatic data reveal slight variability of climate (Table 3).
Also, the study areas share similar bedrock geology of the
central Redbed Plains (Figs. 1–3). Red Permian shales
and sandstones are dominant bedrock types in this region,
forming gently rolling hills and broad flat plains. These
bedrock formations of the Pennsylvanian and Permian
Periods contain red iron oxides that are commonly seen in
such rocks (Johnson, 1996). All three watersheds lie
within the Central Great Plains Ecoregion. They are
similar in most respects, except for land cover, which is
why they were selected for this study. A brief description
of these watersheds follows.
The Skeleton Creek watershed (Fig. 1) is a rural
watershed with b 3% of its area under impervious cover. It
is the largest of all three watersheds with an area of
1.59× 103 km2 (419 mi2) in Garfield, Kingfisher, and
Logan Counties in Oklahoma. The city of Enid where
Skeleton Creek starts constitutes the major impervious
cover in the north of this watershed. Because of the rural
nature, Skeleton Creek watershed is dominated by agricultural and pasture land separated by riparian vegetation
bordering the stream. No reservoirs exist in this watershed.
The Stillwater Creek watershed (Fig. 2) is an exurban
watershed with 4.49% of its area under impervious
cover. This watershed supports agricultural land (pasture, woodland, and crops) and an expanding urban area
of Stillwater, OK. It is marked by dry Mollisols along
with bluestem grama prairies. Experiencing rural and
urban land uses, this watershed has a drainage area of
733 km2 (283 mi2). As an ungauged tributary of the
Cimarron River, Stillwater Creek flows through Payne
County, OK, with a small fraction of the watershed in
Noble and Logan Counties.
Fig. 3. The Deep Fork Creek Watershed.
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
Table 1
Land use in the three watersheds
Type of land use
Table 3
Average annual climate conditions in the three watersheds
% of total watershed area
County
Average
annual
temperature
(°F)
Average
maximum
temperature
(°F)
Average
minimum
temperature
(°F)
Average
annual
precipitation
(in.)
Payne
Noble
Logan
Garfield
Oklahoma
Kingfisher
60
60
60
60
61
60
72
72
72
72
72
72
48
48
49
48
49
48
37.34
36.56
35.24
33.60
36.21
32.61
Skeleton
Stillwater Deep Fork
Creek
Creek
Creek
watershed watershed watershed
Pasture
16.1
Cultivated
52.9
Open Water
1.10
Transitional
0.00
Woody wetlands
0.00
Bare rock/sand/clay
0.00
Urban/recreational grasses
0.06
Urban/residential/commercial
2.99
Quarries/strip mines/gravel pits
0.00
Emergent herbaceous wetlands
0.25
Forest (deciduous/evergreen/
5.40
mixed)
Shrubland and grasslands/
21.2
herbaceous
Total
100.00
10.5
11.3
3.39
0.04
0.01
0.01
0.03
4.27
0.01
0.07
21.4
10.7
3.63
0.91
0.05
0.00
0.07
1.34
49.5
0.00
0.02
14.9
48.9
18.9
100.00
100.00
Three reservoirs exist (Lake Carl Blackwell, Lake
McMurtry and Boomer Lake) in the Stillwater Creek
watershed, two of which are located above the urban area
of Stillwater (Fig. 2). Lake Carl Blackwell is the largest
reservoir with an area of 14.2 km2 (5.47 mi2) and
shoreline 93.3 km (58 mi). Built in 1932, this lake is
located 11.3 km (7 mi) west of the City of Stillwater and
owned by the Oklahoma State University. The primary
purpose of this lake is recreation but it can serve as a
secondary source of water for the Oklahoma State
University. Lake McMurtry, with an area of 5.26 km2
(2.03 mi2) and shoreline of 43.5 km (27 mi), is located
14.5 km (9 mi) north of the City of Stillwater. It was built
for recreation, fishing and flood control in Stillwater.
Boomer Lake is the smallest of three reservoirs with an
area of 1.05 km2 (0.4 mi2) and shoreline of 9.66 km
Table 2
Comparison of imperviousness in the three watersheds
Characteristic
Rural
Exurban/ Urban
watershed converting watershed
watershed
Name
Skeleton
Creek
1590
30
Stillwater
Creek
733
30
Deep
Fork Creek
175
19
0.72
0.82
11.1
0.28
0.01
NA
2.01
1.51
0.01
2.15
4.49
0.1
0.32
NA
11.6
Total area (km2)
Number of transects surveyed
(cross-section and riparian
vegetation)
% Impervious area
(urban roads)
% Impervious area (rural roads)
% Impervious area (buildings)
% Impervious area (other)
% Total impervious area
493
(6 mi). It is located within the city limits of Stillwater and
was built for recreation, fishing and as a supply of water
to cool a natural gas plant that generated electricity.
Deep Fork Creek (Fig. 3) near Arcadia, OK, is an
urban watershed covering an area of 175 km2 (67.6 mi2)
with 11.6% of its area under impervious cover, the
majority of which lies in Oklahoma City. As the smallest
among the three watersheds, Deep Fork Creek flows
through central, northern, and northeastern parts of
Oklahoma County. The potential natural vegetation includes cross-timbers, a mosaic of bluestem prairie (blue
stem, and Indian grass) and oak/hickory forest. As an
urban stream, the riparian vegetation along this stream is
bordered by industrial buildings, governmental facilities,
homes and other urban structures. This is an urban
stream with occasional presence of rip-rap along stream
banks in a few reaches (Figs. 4–6, Table 1).
4. Materials and methods
This research studies the impact of varying degree of
urbanization on the morphology of three streams sharing
similar local conditions (bedrock, vegetation and climate). A discussion on the various types of research
methods and materials used to understand the effects of
urbanization on these three streams follows.
4.1. Effects of urbanization on channel morphology
4.1.1. Field survey of channel morphology, unit types,
and other variables
To understand the effects of urbanization on channel
morphology, field measurements were conducted of
channel cross-sections. Survey sites for channel crosssections in the three watersheds were determined by
dividing each stream into various reaches using USGS
topographic maps (1:24,000). A reach is defined here as
the channel segment between any two adjacent tributaries
(with changing channel form, valley form, vegetation type,
and land use) (Harrelson et al., 1994; Moore et al., 2002).
494
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
case of the urban stream, Deep Fork Creek, 11 reaches have
rip-rap along the banks. Such controlled reaches fail to
respond to changing runoff and sediment supply. Therefore, only 19 uncontrolled reaches (out of 30 surveyed)
from Deep Fork Creek were included in the analysis.
Measurements of channel cross-sections were completed
by standard topographic survey methods (Harrelson et al.,
1994) with a laser level, stadia rod, rebar, flags, and
measuring tapes. Measurements of channel cross-sections
included identification of bankfull height, bankfull width,
flood prone height (double of bankfull height), flood prone
width (top width at double bankfull stage), depth of the
stream along the transect at regular intervals, land use type,
Rosgen channel type, and entrenchment ratio (Rosgen,
1996). Measurements were made by stretching a tape from
the location of a tripod and measuring horizontal and
vertical distances at regular intervals.
The identification of the types of channel units was
based on the following four categories (Harrelson et al.,
1994; Moore et al., 2002) as applied along each transect
in the three streams:
Fig. 4. The urban stream, Deep Fork Creek.
Therefore, changing sinuosity and channel gradient were
also used to divide the three streams into reaches.
Channel cross-sections were measured along a transect
perpendicular to the main axis of the stream at the start of
every reach. A total of 90 reaches were surveyed in the
three watersheds (30 reaches in every watershed). In the
(i)
(ii)
(iii)
(iv)
pool (slow and deep),
glide (slow and shallow),
riffle (fast and shallow), and
run (fast and deep).
Other stream variables were calculated from secondary sources, such as USGS (1:24,000) topographic
maps. These variables included actual stream lengths,
straight-line stream lengths, sinuosity, and gradient.
Fig. 5. The urban stream, Deep Fork Creek.
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
495
Fig. 6. The urban stream, Deep Fork Creek with occasional presence of rip-rap and trash.
4.1.2. Use of GIS in delineating the three watersheds
Boundaries of all three watersheds and the subwatersheds were delineated by using standard GIS methods. This involved the use of digital elevation models
(DEMs) with 30 × 30-m resolution for the different
counties that covered the three watersheds. DEMs were
downloaded from the USGS web page. The stream networks were downloaded from Census Tiger files on the
Environmental Science Research Institute (ESRI) web
page. The Geoprocessing Wizard in ArcView 3.3 was
used to merge the data from different counties to make one
shape file for every stream. The two data sets (DEMs and
stream networks) were used in the ArcView Soil and
Water Assessment Tool (AVSWAT) to delineate the
boundaries of the three watersheds, and boundaries of
sub-basins within the three watersheds. AVSWAT is an
ArcView extension and a graphical user interface for the
Soil and Water Assessment Tool (SWAT). SWAT is a
physically based and computationally efficient watershed-scale model used to predict the impact of management practices on water, sediment, agricultural chemical
yields, and more (Luzio et al., 2002).
4.1.3. Fluvial data processing
Channel morphologic data were entered into a specially designed MS Excel Reference Reach Tool, an
Excel programmed macro that was used to calculate the
following hydraulic variables:
(i) Entrenchment ratio: an index value that is used to
describe the degree of vertical containment of a
river channel (width of the flood prone area at an
elevation twice the maximum bankfull depth/
bankfull width) (Rosgen, 1996).
(ii) Threshold grain size: particle size predicted to be
at the threshold of motion at the calculated shear
stress, derived from the Shields curve that is a
plot of particle size against the shear stress
required to initiate movement (this hydraulic
variable was calculated by the Reference Reach
Tool).
(iii) Maximum bankfull depth: the maximum depth of
flow at bankfull stage.
(iv) Mean bankfull depth: the stream channel crosssection at bankfull stage.
(v) Wetted perimeter: perimeter of the channel crosssection formed by bed and banks.
(vi) Width of flood prone area: flooded width at a
stage twice the maximum depth in a riffle or
straight section.
(vii) Cross-sectional area: area of the stream channel
cross-section at bankfull stage.
(viii) Froude number: the dimensionless number
expressing the ratio of inertial to gravitational
forces.
Stillwater Creek was analyzed individually before
comparing the three streams. It was divided into upstream and downstream reaches at the confluence of
Boomer Creek which brings urban runoff from the
City of Stillwater (Fig. 2). Various hydraulic variables
were used as dependent variables in a series of
496
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
univariate regression analysis with drainage area
always used as the independent variable. Dependent
variables such as sinuosity, gradient, mean bankfull
depth, bankfull width, bankfull area and threshold
grain size were used to assess how drainage area explains each of these variables. Residuals calculated
from the univariate regression analysis were used to
perform Mann-Whitney nonparametric test on the upstream and downstream reaches of Stillwater Creek.
This test was conducted to explain any possible impacts of urbanization on the downstream reaches of
Stillwater Creek which receive urban runoff from the
city of Stillwater.
The next step was to compare the three watersheds
selected on the basis of geomorphic similarities. The
objective was to determine if the changing degree of
urbanization can explain any geomorphic differences
among the three streams. This analysis involved a
similar series of univariate regressions with drainage
area as the independent variable. A variety of dependent variables were used to determine how well drainage area predicts other dependent variables for every
watershed (Marston and Wick, 1994). The three
watersheds were compared with each other to determine if channel morphologies of the three streams
respond differently to changing degrees of urbanization. This comparison was made by conducting a covariance test for dependent variables (Johnston, 1972).
Results of this test were used to decide if regres-
sions are significantly different for three streams. The
following formula was used for this test:
F¼
½Q4 − ðQ1 þ Q2 þ Q3Þ=ðm−1Þ
ðQ1 þ Q2 þ Q3Þ=mðn−1Þ
Where
m
n
Q1
Q2
Q3
Q4
number of transects measured
average number of transects measured for individual dependent variables
residual sum of squares for Skeleton Creek
residual sum of squares for Stillwater Creek
residual sum of squares for Deep Fork Creek
residual sum of squares using all three streams
The F value derived from this formula was compared
with the critical value obtained from an f table. If the
derived F value is greater than the critical value from
f table, regressions are significantly different from one
another and vice versa.
4.2. Effects of urbanization on riparian vegetation
An inventory of riparian vegetation was prepared that
consisted of a type of belt transect extending along the
riparian zone perpendicular to the stream channel on one
side of the stream (Fig. 7). Vegetation transects started
near the upstream half of the reach (same as channel cross-
Fig. 7. An example of transect extending across the riparian zone perpendicular to the stream channel (Lehmert and Marston, 2005).
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
497
Fig. 8. Aerial view of slightly entrenched meanders and land use adjacent to Skeleton Creek.
sections), extended 5 m perpendicular to the main axis of
the stream (on either the left or right side), and extended
30 m in the longitudinal dimension. This 30-m-long
transect was divided into three zones of 10 m each to
record the percent canopy closure, grass and shrubs, tree
groups (based on size and species), and number of trees.
The final data from field surveys were used to calculate
descriptive statistics. Basal areas were calculated for trees
in different diameter categories (3–15, 16–30, 31–50,
51–90 and N91 cm).
Fig. 9. Aerial view of a reservoir and land use in the Stillwater Creek watershed.
498
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
The use of a small plane was another tool that was
used to get a perspective of the three watersheds. This
technique provided oblique photographs that could not
be used for any quantitative analysis. Nonetheless, the
photographs (Figs. 8–10) and videos of the three watersheds were useful tools to understand the general land
use in the three watersheds.
4.3. Degree of urbanization and imperviousness
Digital data on imperviousness were collected from
existing sources and subjected to spatial analyses. Shape
files for impervious surfaces in the three watersheds
were obtained from the following sources:
(i) Stillwater Creek watershed: City of Stillwater office in Stillwater, OK.
(ii) Skeleton Creek watershed: City of Enid and
Garfield County Assessor office in Enid, OK.
(iii) Deep Fork Creek watershed: Oklahoma City office in Oklahoma City, OK.
clipped along watershed boundaries to remove areas
lying outside of the three watersheds. Shape files for
roads in the three watersheds were line features, so
buffers were created to find areas by using ArcToolbox
9.0. Roads were divided into two categories for this
purpose: (i) urban roads (i.e., roads within the city limits
of Enid, Oklahoma City, and Stillwater) were given a
10-m buffer width, and (ii) rural roads (i.e., roads
outside the city limits of Enid, Oklahoma City, and
Stillwater) were given a 7.5-m buffer width. Areas of the
road buffers were calculated using the “Open Tool”
option in ArcMap 9.0.
Historical photos were available for only Stillwater
Creek watershed for the year 1979 and 2003. These
photos (1: 20,000) were used to calculate the change in
impervious surface area in this watershed during
1979–2003. This method involved tracing aerial
photographs on a mylar sheet. Then a grid proportional
to 1:20,000 was overlaid on the traced mylar sheet to
calculate the area of the impervious surface.
4.4. Land cover analysis
The respective city offices prepared these GIS shapefiles for various purposes, such as property management, code enforcement, emergency management and
infrastructural maintenance. These shapefiles are prepared from different sources and provide comprehensive
digital details of imperviousness. All GIS shape files
were re-projected using ArcToolbox 9.0 into the
NAD 1983 UTM Zone 14 projection system. The
shape files of roads and other impervious surfaces were
A detailed study of land cover in all three watersheds
was conducted by using 1993 National Land Cover Data
(NLCD) from the USGS. At a resolution of 30 m, this
grid form of data revealed that N 50% of the area in the
Skeleton Creek watershed is under cultivation, whereas,
the area under cultivation in Stillwater Creek watershed
and Deep Fork Creek watersheds is 11.4% and 3.63%,
respectively (Table 1).
Fig. 10. Aerial view of urban land use adjacent to Deep Fork Creek.
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
499
5. Results and discussion
5.1. Effects of urbanization on channel morphology
5.1.1. Field survey of channel morphology
Skeleton Creek, the rural stream, shows the expected
morphological changes in downstream hydraulic geometry. Bankfull width and depth increase in the downstream
direction, as do bankfull area and wetted perimeter. These
variables carry very low values in the origin of this stream
near Enid and a consistent increase occurs with increasing
drainage area downstream. The exurban stream of
Stillwater Creek also shows a similar increase in these
variables in a downstream direction. Although three reservoirs (Lake Carl Blackwell, Lake McMurtry and Lake
Boomer) exist in this watershed, only two are upstream of
the urban area of Stillwater. These reservoirs were built for
recreation, flood control and urban use (see Section 3).
The urban stream of Deep Fork Creek, however, shows a
slightly different trend in the variation of channel
morphology. Mean bankfull depth, bankfull width, bankfull area, and threshold grain size do not show an increasing trend in the downstream direction in Deep Fork
Creek. Supported by statistical analysis, these findings are
similar to personal observations made during field
Fig. 12. Riparian corridor in Stillwater Creek Watershed.
surveys in the three watersheds. These trends are analyzed
in Section 5.4 on statistical analysis.
5.1.2. Types of channel units
The three streams show similar types of channel
units. The rural stream of Skeleton Creek shows the
random presence of all four types of channel units: pool,
glide, riffle, and run. With the increasing degree of
urbanization, however, glides appear slightly more often
in the exurban (Stillwater Creek) than the urban stream
(Deep Fork Creek). Visual observations revealed that
substrate materials remain almost unchanged in all three
streams, which is predominantly silt and clay. In the case
of Deep Fork Creek, certain areas in this stream have
bedrock as the actual bed material. Only 19 sections
were used in the final analysis (those without any
engineering control). Channel gradients (calculated
from topographic maps) were very low (b 0.001), and
sinuosity was consistently low (b 2) in the three streams.
5.2. Effects of urbanization on riparian vegetation
Fig. 11. Riparian corridor in Skeleton Creek Watershed.
The riparian corridor along the rural stream is
bordered by agricultural fields that rarely adjoin the
stream. Personal discussions with farmers in the rural
500
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
watershed of Skeleton Creek revealed that the riparian
buffer has been unchanged since the 1950s. Similar,
riparian buffers exist along exurban and urban streams
that are commonly bordered by pastures and impervious
areas (see regional settings), respectively. All three
streams have barbed-wire fencing along the riparian
corridors. Riparian corridors included three types of
vegetation: trees, shrubs and grass. The dominant trees
are cottonwoods (Populus sp.), green ash (Fraxinus
pennsylvanica), and American elm (Ulmus americana)
in the three watersheds. Field surveys revealed substantially similar riparian corridors in all three watersheds
(Figs. 11–13), which included measurements of
30 m × 10 m riparian plots perpendicular to stream
reaches (Fig. 7). Riparian buffers rarely extend beyond
30 m of any stream. Differences in acreage of riparian
vegetation as well as percent change in riparian
vegetation during different time periods were not
dependent on location along the stream or the width,
depth, channel area, or degree of urbanization for the
streams. Therefore, many geomorphologic variables can
be ruled out as the cause of the width of the riparian
corridor. Human factors, such as land-use changes from
Fig. 13. Riparian corridor in Deep Fork Creek Watershed.
agriculture to residential or from grazing to recreation
and urban area, may be a key factor in the width and
quality of the riparian zones that appear almost intact in
all three study regions (Lehmert and Marston, 2005).
5.3. Degree of urbanization and imperviousness
GIS analysis of imperviousness (Table 2) revealed
that roads (rural and urban) alone constitute the major
impervious surface in the three watersheds. Skeleton
Creek, the rural watershed contains the least area under
impervious cover (b 2.01%) compared to Deep Fork
Creek, the urban watershed (N 11%). The percentage of
areas covered by roads in Skeleton, Stillwater, and Deep
Fork Creeks is 2%, 2.33%, and 11.2%, respectively.
5.4. Statistical analysis of downstream trends
Stillwater Creek, the exurban stream, was analyzed
individually before comparing the three streams with
each other. One of its tributaries, Boomer Creek, was
used to divide this stream into upstream and downstream
reaches. According to a univariate regression analysis,
dependent variables (mean bankfull depth, bankfull
width, bankfull area, and threshold grain size) exhibited
an increase in the downstream direction. Residuals from
univariate regression analyses were calculated to see
what additional trends could be attributed to urbanization. These residuals refer to the difference between the
estimated and actual values calculated from the regression equation.
The Mann-Whitney nonparametric statistical test was
performed on these residuals, grouping reaches upstream and downstream from the confluence with
Boomer Creek, which delivers urban runoff and sediment from Stillwater. The p values reported from this
test are greater than 0.05 for sinuosity, gradient, bankfull
depth, bankfull width, bankfull area, and threshold grain
size. Therefore, after accounting for the effects of greater drainage basin area as one moves downstream by
analyzing regression residuals, the remaining unexplained variation is not significantly related to results of
urbanization. As one moves downstream of Boomer
Creek (the tributary that delivers runoff and sediment),
the river attains a lower sinuosity, lower gradient, greater
bankfull depth, width and area, and the threshold grain
size gets larger. These trends would be expected normally, however, none of these downstream changes show a
statistically significant change that can be attributed to
urbanization.
The one variable that does exhibit statistically
significant changes downstream of Boomer Creek is
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
the number of trees along the banks (significant at
p = 0.02). This increase in tree density downstream helps
explain the other results that show no response in channel morphology downstream of Stillwater. The greater
density of trees helps to stabilize the banks against
increasing flows. The channel bed and bank materials do
not change over the entire length of Stillwater Creek.
They consist of 95–100% silt–clay. Bedrock does not
change. The only variables that change are contributing
area, riparian vegetation, and urbanization. Stillwater
Creek is classified as an E6b channel (Rosgen, 1996),
which is a very stable channel type. Although the impervious surface area increased by 65% in 24 years
(1979–2003) in the Stillwater Creek watershed, no
statistically significant impact of urban runoff and sediment can be discerned on the lower reaches of Stillwater
Creek. Field observations in this watershed also
revealed the entrenched nature of this stream along
with occasional presence of woody debris jams. Therefore, the riparian vegetation, type of substrate materials
(silt and clay), and presence of woody debris jams
provide possible answers to why Stillwater Creek is not
exhibiting significant changes in morphology (downstream of Boomer Creek) as expected in a watershed
converting from rural to urban. The hypothesis is rejected that urbanization is the major reason for increase
in the channel cross-sectional area downstream in
relation to upstream of Stillwater Creek.
Skeleton Creek, the rural stream, shows a similar
trend of increasing mean bankfull depth, bankfull width,
501
and bankfull area in the downstream direction. As stated
earlier, Skeleton Creek is a rural stream with less than
3% of its area under impervious cover. Field observations revealed a very rural nature of this stream with an
absence of substantially industrial or urban related impervious surfaces adjoining the stream corridor. This
stream shares the same bedrock geology as the Stillwater Creek watershed. Riparian vegetation, substrate
materials (silt–clay) are also similar to Stillwater Creek
and these do not change in the downstream direction.
Field observations revealed an entrenched character
similar to Stillwater Creek. Informal talks with local
residents revealed that the majority of the riparian
corridor has been unchanged along the stream since the
1950s.
In the case of Deep Fork Creek, the urban stream,
bankfull mean depth, bankfull width, bankfull area, and
threshold grain size show a different trend. These
variables do not increase in the downstream direction
based on field observations for the uncontrolled reaches
(19 of the 30 surveyed). The analysis revealed, however,
that Deep Fork Creek exhibits a downstream hydraulic
response similar to Skeleton and Stillwater Creeks with
regards to sinuosity and gradient. Deep Fork Creek has
bed materials (silt–clay) similar to the other two
streams, although bedrock was observed in few sections
of this stream.
The three watersheds exhibit similar geomorphic
characteristics. The sizes (drainage area) are slightly
different but in the same order of magnitude. The biggest
Fig. 14. Regression plot of Log10 sinuosity as a function of Log10 drainage area.
502
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
Fig. 15. Regression plot of Log10 gradient as a function of Log10 drainage area.
difference is the degree of urbanization (Table 2). The
univariate regression analysis (based on log-transformed
values) was performed to compare the regression lines of
the three streams. Drainage area upstream of each transect was used as the independent variable. The dependent variables were sinuosity, gradient, mean bankfull
depth, bankfull width, bankfull area and threshold grain
size. Regression lines (Figs. 14–19) for the three streams
are plotted on the same graph for each dependent variable to determine if urbanization can explain any
changes in the channel morphology with increasing
drainage area. The values of r2 are low (Table 4) indicating a large proportion of unexplained variance. The
regression lines of the three watersheds fail to explain
differences in channel morphology between the three
streams. Now the question arises: do the three streams
exhibit any significant differences in terms of morphology because of the changing degree of urbanization?
This was addressed by performing a covariance test for
each dependent variable to test for significant difference
between those regression line (Y) intercepts (Johnston,
1972; Marston and Wick, 1994). This analysis provided
Fig. 16. Regression plot of Log10 mean bankfull depth as a function of Log10 drainage area.
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
503
Fig. 17. Regression plot of Log10 bankfull width as a function of Log10 drainage area.
F values for every variable that was compared with the
critical F value from the F table at the 1% significance
level.
The results of this test revealed that no significant
difference exists between regression lines of the three
streams for gradient, sinuosity or total number of trees.
At the same time, the three streams share the same
ecoregion (geology, climate, soils and potential natural
vegetation) with low local relief. The only differences
exhibited by statistical analysis are in case of bankfull
mean depth, bankfull width, bankfull area and threshold
grain size of the three streams.
Now the question arises: are the three streams really
different? If the three watersheds share same ecoregion,
why do statistical differences exist in the mean bankfull
depth, bankfull width, bankfull area and threshold grain
size? The changing degree of urbanization fails to
explain any differences in the three streams. The second
hypothesis that the exurban stream, Stillwater Creek
would respond significantly different from the other two
Fig. 18. Regression plot of Log10 bankfull area as a function of Log10 drainage area.
504
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
Fig. 19. Regression plot of Log10 threshold grain size as a function of Log10 drainage area.
streams (because of the generation of a substantial
amount of sediment from construction activities) is
rejected. Therefore, other factors are affecting these
streams. Any possible geologic variable is ruled out
because the watersheds are situated in the same
ecoregion. Although Stillwater Creek has three reservoirs two major reservoirs (Lake Carl Blackwell and
Lake McMurtry) are upstream of the urban area. The
presence of reservoirs in Stillwater Creek watershed is,
therefore, ruled out as a substantial reason for differences in bankfull mean depth, bankfull width, bankfull
area and threshold grain size of the three streams.
One other possible reason for the unexplained variance can be the presence of Central Oklahoma aquifer
under the sandstone bedrock of these entrenched
streams. It is possible that this aquifer is contributing
to the discharge of water in the three streams (personal communication, Dr. Paxton, School of Geology,
Table 4
Results of univariate regressions with drainage area (km2)
Dependent variables
Stream
N
r2
Intercept
Slope
p value
Sinuosity
Skeleton Creek
Stillwater Creek
Deep Fork Creek
Skeleton Creek
Stillwater Creek
Deep Fork Creek
Skeleton Creek
Stillwater Creek
Deep Fork Creek
Skeleton Creek
Stillwater Creek
Deep Fork Creek
Skeleton Creek
Stillwater Creek
Deep Fork Creek
Skeleton Creek
Stillwater Creek
Deep Fork Creek
Skeleton Creek
Stillwater Creek
Deep Fork Creek
30
30
19
30
30
19
30
30
19
30
30
19
30
30
19
30
30
19
30
30
19
0.03
0.11
0.20
0.31
0.17
0.15
0.07
0.17
0.50
0.52
0.05
0.49
0.42
0.13
0.33
0.03
0.20
0.00
0.02
0.39
0.80
0.07
0.47
− 0.19
0.32
1.01
1.88
− 0.48
− 2.28
2.69
0.25
0.36
2.97
− 0.24
− 1.93
4.80
− 0.66
94.3
20.9
1.27
− 1.58
7.81
0.02
− 0.14
0.13
− 0.64
− 0.90
− 1.59
0.11
1.16
− 1.16
0.35
0.40
− 0.84
0.47
1.56
− 1.58
5.04
− 30.7
− 4.24
− 0.12
1.09
− 3.27
0.36
0.80
0.06
0.00
0.02
0.10
0.15
0.02
0.00
0.00
0.22
0.00
0.00
0.05
0.01
0.37
0.01
0.82
0.54
0.01
0.10
Gradient
Bankfull depth mean
Bankfull width
Bankfull area
Total trees
Threshold grain size
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
Oklahoma State University). Overlaid on the sandstone
bedrock with some shale, the three streams respond like
flumes for the runoff. This would explain the entrenched
response of the three streams. Stable riparian buffers and
cohesive silt-clay as bed materials also support this
argument.
505
Environmental Science Doctoral student at Oklahoma
State University also helped in data analysis. Marissa
Raglin, Susan Basta, and Jack Vitek provided critical
editing assistance. The manuscript was improved
through by addressing the comments of anonymous
reviewers and the guest editors, Allan James and Andrew
Marcus.
6. Conclusions
Local conditions mitigated the impact of urbanization on channel cross-sections in all three study streams.
The process of urbanization has not significantly affected channel morphologies of the three streams that
are experiencing different stages of urbanization (rural,
exurban, and urban), except where rip-rap or channelization directly altered Deep Fork Creek. The statistical
analysis of Stillwater Creek, the exurban stream,
revealed that urbanization does not explain any morphological changes in the downstream direction for rapidly
urbanizing portions of the watershed.
Some unexplained variation remains among the three
streams in terms of downstream trends in mean bankfull
depth, bankfull width, bankfull area, and threshold grain
size. This natural variability is not explained by the
changing degree of urbanization. Field observations also
revealed the entrenched nature of the three streams.
Sandstone along with shale combined with silt and clay
and dense riparian vegetation render the bed and banks
as cohesive. These streams respond as flumes for runoff
and sediment without any significant impact on
morphology. Depositional features are largely absent.
Ties between the stream channel and underlying Central
Oklahoma Aquifer are suspected as one possible reason
for such natural variability in these streams, which
otherwise share the same low-relief geomorphic province and ecoregion. It is possible that the interaction
between the aquifer and streams has diluted the impact
of urbanization. Because urbanization does not explain
the natural variability among these fluvial systems, further research should be conducted on the interaction of
aquifer and streams in the same ecoregion to provide
more understanding of such natural variability among
these streams.
Acknowledgements
This project involved extensive fieldwork in three
watersheds that was completed with the help of John
Stapleton and Kim Stapleton who deserve special words
of appreciation. Dr. John Comer in the Department of
Geography at Oklahoma State University provided help
in regression analysis by using SPSS. Dale Splinter, an
References
Arnold, C.L.J., Gibbons, C.J., 1996. Impervious surface coverage: the
emergence of a key environmental indicator. Journal of American
Planning Association 62, 243–258.
Avolio, C.M., 2003. Local impacts of road crossings on Puget Lowland
Creeks: M.S. Thesis, Department of Civil and Environmental
Engineering, University of Washington, Seattle, WA, 119 pp.
Benfield, F.K., Raimi, M.D., Chen, D.D.T., 1999. Once There Were
Green Fields: How Urban Sprawl is Undermining America's
Environment. Environment, Economy, and Social Fabric. Natural
Resources Defense Council, New York, NY.
Booth, D.B., 1990. Stream channel incision following drainage basin
urbanization. Water Resources Bulletin 26 (3), 407–417.
Booth, D.B., 1991. Urbanization and the natural drainage system:
impacts, solutions, and prognoses. The Northwest Environmental
Journal 7, 93–118.
Booth, D.B., Henshaw, P.C., 2001. Rates of channel erosion in small
urban streams. In: Mark, S.W., Stephen, J.B. (Eds.), Land use and
Watersheds: Human Influence on Hydrology and Geomorphology
in Urban and Forest Areas. American Geophysical Union,
Washington, D.C., pp. 17–38.
Booth, D.B., Jackson, C.R., 1997. Urbanization of aquatic systems:
degradation thresholds, stormwater detention, and the limits of
mitigation. Journal of the American Water Resources Association
33, 1077–1090.
Brierley, G.J., Fryirs, K.A., 2005. Geomorphology and River
Management. Blackwell, Malden, MA. 398 pp.
Chin, A., Gregory, K.J., 2001. Urbanization and adjustment of
ephemeral stream channels. Annals of the Association of American
Geographers 91 (4), 595–608.
Hammer, T.A., 1972. Stream channel enlargement due to urbanization.
Water Resources Research 8 (6), 1530–1540.
Harrelson, C.C., Rawlins, C.L., Potyondy, J.P., 1994. Stream Channel
Reference Sites: An Illustrated Guide to Field Technique. General
Technical Report RM-245, USDA Forest Service. Rocky Mountain Forest and Range Experiment Station, Fort Collins, CO. 61 pp.
Hession, W.C., Pizzuto, J.E., Johnson, T.E., Horwitz, R.J., 2002.
Influence of bank vegetation on channel morphology in rural and
urban watersheds. Geology 31 (2), 147–150.
Hollis, G.E., 1976. The response of natural river channels to
urbanization; two case studies in southeast England. Journal of
Hydrology 30 (4), 351–363.
Jeje, L.K., Ikeazota, S.I., 2002. Effects of Urbanization on Channel
Morphology: The Case of Ekulu River in Enugu, Southeastern
Nigeria. Singapore Journal of Tropical Geography 23 (1), 37–51.
Johnson, K.S., 1996. Geology of Oklahoma. In: Johnson, K.S., Suneson,
N.H. (Eds.), Rockhounding and Earth-Science Activities in
Oklahoma, 1995 Workshop. Oklahoma Geological Society, pp. 1–9.
Johnson, M.P., 2001. Environmental impacts of urban sprawl: a survey
of the literature and proposed research agenda. Environment &
Planning A 33 (4), 717–735.
506
R.S. Kang, R.A. Marston / Geomorphology 79 (2006) 488–506
Johnston, J., 1972. Econometric Methods, 2nd ed. McGraw Hill,
New York.
Leavitt, J., 1998. The Functions of Riparian Buffers in Urban
Watersheds. M.S. Thesis, Department of Civil Engineering,
University of Washington, Seattle, WA, 38 pp.
Lehmert, K., Marston, R.A., 2005. Spatial and temporal trends in riparian
vegetation patterns on Stillwater Creek: Stillwater, Oklahoma.
Abstracts, Association of American Geographers Annual Meeting,
Denver, CO.
Leopold, L.B., 1956. Land use and sediment yield. In: Thomas Jr.,
W.L. (Ed.), Man's Role in Changing the Face of the Earth.
University of Chicago Press, Chicago, IL, pp. 639–647.
Leopold, L.B., 1972. River channel change with time: an example.
Geological Society of America Bulletin 84, 845–860.
Luzio, M.D., Srinivasan, R., Arnold, J.D., Neitsch, S.L., 2002.
ArcView Interface for SWAT 2000: User's Guide. Texas Water
Resources Institute, College Station, TX.
Marston, R.A., Wick, D.A., 1994. Natural variability and forestry
impacts in mountain streams of southeastern Wyoming. In:
Marston, R.A., Hasfurther, V.R. (Eds.), Effects of Human Induced
Changes on Hydrologic Systems. AWRA 1994 Annual Summer
Symposium of the American Water Resources Association,
Jackson Hole, Wyoming, pp. 225–238.
May, C.W., Horner, R.R., Karr, J.R., Mar, B.W., Welch, E.B., 2002.
Effects of urbanization on small streams in the Puget Sound
Ecoregion. In: Schueler, T.R., Holland, H.K. (Eds.), The Practice
of Watershed Protection. Center for Watershed Protection,
Ellicott City, WA, pp. 87–98.
Moore, K., Jones, K., Dambancher, J., 2002. Aquatic Inventories Project.
12.2. Oregon Department of Fish and Wildlife, Corvallis, OR. 36 pp.
Morisawa, M., Laflure, E., 1979. Hydraulic geometry, stream
equilibrium and urbanization. Adjustments of the Fluvial System:
A Proceedings Volume of The Tenth Annual Geomorphology
Symposia Series. Kendall/Hunt, Dubuque, IA, pp. 333–350.
Nanson, G.C., 1981. Downstream reduction of rural channel size with
contrasting urban effects in small coastal streams southeastern
Australia. Journal of Hydrology 52 (3-4), 230–255.
Neller, R., 1988. A comparison of channel erosion in small urban and
rural catchments, Armidale, New South Wales. Earth Surface
Processes and Landforms 13 (1), 1–7.
Othitis, M., Kang, R.S., Marston R.A., R.A., 2004. Effects of Urbanization on Stream Channel Morphology: Stillwater Creek,
Oklahoma. Abstracts, Southwestern Division. Association of
American Geographers, Nacogdoches, TX.
Rosgen, D., 1996. Applied River Morphology. Wildland Hydrology,
Pagosa Springs, CO.
Schueler, T., 1994. The importance of imperviousness. Watershed
Protection 1 (3), 100–111.
Simon, A., Bennett, S.J., Neary, V.S., 2004. Riparian vegetation and
fluvial geomorphology: problems and opportunities. In: Bennett, S.J.,
Simon, A. (Eds.), Riparian Vegetation and Fluvial Geomorphology.
American Geophysical Union, Washington, D.C., pp. 1–10.
Strahler, A.N., 1956. The nature of induced erosion and aggradation.
In: Thomas Jr., W.L. (Ed.), Man's Role in Changing the Face of the
Earth. University of Chicago Press, Chicago, IL, pp. 621–639.
Trimble, S.W., 2004. Effects of riparian vegetation on stream channel
stability and sediment budgets. In: Sean, B.J., Simon, A. (Eds.),
Riparian Vegetation and Fluvial Geomorphology. American
Geophysical Union, Washington, D.C., pp. 153–170.