JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
VOL. 37, NO. 2 AMERICAN WATER RESOURCES ASSOCIATION APRIL 2001
EFFECTS OF URBANIZATION ON CHANNEL INSTABILITY'
Brian P Bledsoe and Chester C. Watson2
ABSTRACT: Channel instability and aquatic ecosystem degradation
have been linked to watershed imperviousness in humid
regions of the U.S. In an effort to provide a more process-based
linkage between observed thresholds of aquatic ecosystem degradation
and urbanization, standard single event approaches (U.S. Geological
Survey Flood Regression Equations and rational) and
continuous hydrologic models (HSPF and CASC2D) were used to
examine potential changes in flow regime associated with varying
levels of watershed imperviousness. The predicted changes in flow
parameters were then interpreted in concert with risk-based models
of channel form and instability. Although low levels of imperviousness
(10 to 20 percent) clearly have the potential to destabilize
streams, changes in discharge, and thus stream power, associated
with increased impervious area are highly variable and dependent
upon watershed-specific conditions. In addition to the storage characteristics
of the pre-development watershed, the magnitude of
change is sensitive to the connectivity and conveyance of impervious
areas as well as the specific characteristics of the receiving
channels. Different stream types are likely to exhibit varying
degrees and types of instability, depending on entrenchment, relative
erodibility of bed and banks, riparian condition, mode of sediment
transport (bedload versus suspended load), and proximity to
geomorphic thresholds. Nonetheless, simple risk-based analyses of
the potential impacts of land use change on aquatic ecosystems
have the potential to redirect and improve the effectiveness of
watershed management strategies by facilitating the identification
of channels that may be most sensitive to changes in stream power.
(KEY TERMS: stormwater management; erosion; sedimentation;
land use planning; aquatic ecosystems; channel stability; model-
ing.)
INTRODUCTION
Urbanization can have many direct impacts on
aquatic ecosystems as streams broaden or deepen
to accommodate larger, more frequent, or longer
duration flows of water. In the initial construction
phase, a spate of sediment delivery to stream channels
is accompanied by an increase in flow (Wolman,
1967). After exposed soil has been stabilized or covered
with impervious surface, increased runoff coupled
with a decline in watershed sediment yield
typically results in channel enlargement through incision
(Booth, 1990), widening (Arnold et al., 1982) or
both (Hammer, 1972; Roberts, 1989). Channel erosion
due to urbanization can become the predominant
source of excess sediment to downstream reaches
(Trimble, 1995, 1997) and result in degradation of
biotic integrity (Waters, 1995).
Investigators from various humid regions in the US
have reported that channel instability and abrupt
declines in indices of aquatic ecosystem integrity are
observed at 10 to 20 percent watershed imperviousness
(Morisawa and LaFlure, 1979; Booth, 1990,
1991; Booth and Reinelt, 1993; Schueler, 1994). The
specific causative mechanisms behind this perceived
threshold have yet to be identified. Numerous studies
have linked watershed imperviousness with increased
peak discharge magnitudes (Urbonas and Roesner,
1993). In a synthesis of several previous studies, Holus
(1975) suggested that flows less than bankfull may
be increased by a factor of 10 or more depending on
the degree of urbanization. At 10 to 20 percent watershed
imperviousness, flows with recurrence intervals
of 1.5 to two years in the annual maximum series may
be doubled or tripled (Hammer, 1972; Hollis, 1975).
In general, observations suggest that stream channels
in smaller and more permeable watersheds exhibit a
greater and more abrupt erosive response to urban-
ization.
iPaper No. 00089 of the Journal of the American Water Resources Association. Discussions are open until December 1, 2001.
2Respectively, Research Assistant Professor and Associate Professor, Department of Civil Engineering, Colorado State University, Fort
Collins, Colorado 80523 (E-MaillBledsoe: bbledsoe@engr.colostate.edu).
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 255 JAWRA
Bledsoe and Watson
It is clear that increases in discharge and stream
power associated with urbanization have the potential
to severely destabilize streams in a variety of contexts.
It seems unlikely, however, that there is a
consistent relationship between imperviousness and
the potential for channel response across watershed
and stream types. The purpose of this paper is to
examine linkages between imperviousness, increases
in discharge and stream power, and the risk of channel
instability in urbanizing watersheds. Because the
disparate responses of stream channels are dependent
upon flow regime, the following sections explore the
variability in hydrologic response to urbanization
before specifically addressing the risk of channel
instability in subsequent sections.
RELATIONSHIPS BETWEEN IMPERVIOUSNESS
AND FLOW CHARACTERISTICS
Measured data on the effects of urbanization on
stream channel form are rare, and estimates are typically
extrapolated from watershed models. In this
analysis, the potential effects of watershed imperviousness
were examined using hydrologic models that
range from simple statistical relationships and the
rational method (McCuen, 1989) to complex processbased
models. Several USGS Flood Regression Equations,
other USGS data, and National Urban Runoff
Program (NURP) data were contrasted with continuous
simulations based on the CASC2D (Julien et al.,
1995) and Hydrologic Simulation Program-Fortran
(HSPF; Johanson et al., 1980; USEPA, 1984) hydrologic
models. Such a contrast provides insight into the
usefulness of simple, single event approaches in projecting
magnification of channel-forming flood events
as well as cumulative effects across a broad range of
flows occurring over an extended period of time.
Single Event Approaches
USGS Flood Regression Relationships. For
many years, the USGS has been involved in the development
of regional regression equations for estimating
flood magnitude and frequency at ungaged sites.
Regression equations for rural areas and, in some
instances, urban areas are periodically compiled in a
summary organized by state for the United States.
The most recent compilation is provided in Jennings
et al. (1993). These regression equations are used to
transfer flood characteristics from gaged to ungaged
sites through the use of watershed and climatic
characteristics as predictor variables. The USGSdeveloped
regression equations are generally unbiased,
reproducible, and easy to apply.
For the purposes of this analysis, we are most concerned
with some estimate of the dominant, channel
forming discharge, which is often approximated as
having a 1.5-year to two-year recurrence interval in
the annual maximum series (approximately an annual
flood). Estimates of the median annual flood event
(Q2) derived from the USGS flood regression equations
may be particularly useful in geomorphic assessment
and hydraulic design and often provide an
acceptable estimate of the dominant, channel-forming
discharge. Six sets of USGS urban regression equations
for Q2 provided in the 1993 summary were
selected for this analysis based on range of applicability,
data availability, and ease of application. The
urban equations, the relationships used to estimate
a rural Q2, and the sources for each are listed in
Table 1. The number of region-specific relationships
examined was limited to six since the other equations
required site-specific information that rendered a general
analysis infeasible. For example, several of the
USGS flood regression equations for locations outside
the six selected states utilize a "basin development
factor" (BDF) developed by Sauer et al. (1983) for use
in the USGS national flood regression equations for
urban watersheds. The BDF is a function of channel
improvements, channel linings, storm drains or sewers,
and curb-and-gutter streets. It is intended to represent
the efficiency of the drainage system.
Subsequent testing of the BDF approach has yielded
mixed results and the accuracy of the original data on
specific drainage characteristics used in the regression
analyses is uncertain (personal communication
with Ben Urbonas, Chief of Planning, Urban
Drainage and Flood Control District, Denver, Colorado,
1999).
Hypothetical watershed areas of 20, 50, and 100
km2 were analyzed using the six USGS urban equations
at levels of imperviousness ranging from 5 to 30
percent. The corresponding rural equations were used
to approximate discharges for watersheds with minimal
imperviousness. Since the Missouri and Wisconsin
rural equations require channel slope as an
independent variable (Jennings et al. 1993), the magnitude
of rural Q2 in those states was estimated using
the appropriate urban equation at a low level of
imperviousness (Table 1). The Missouri data set
included several watersheds at 1 percent imperviousness,
and the estimated Q2 at 1 percent imperviousness
was assumed to represent rural conditions.
Similarly, the Wisconsin rural Q2 was assumed equal
to the urban result at 5 percent imperviousness which
was above the lower end of the range of data used in
developing the regression equation. These analyses
provide estimates of increased discharge for different
JAWRA 256 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
Effects of Urbanization on Channel Instability
R regions as a function of imperviousness (Figure 1).cC
Results from the urban regression equations for lessa ,- c.
than 5 percent imperviousness are not shown since
I I some statistical models were not well calibrated atC)
— - impervious levels of 0 to 5 percent..- -CC
C) At 10 percent imperviousness, the ratio of devel-C) CC)
0 ci)- oped Q2 to rural Q2 ranges from 1.5 to 2.2. At 20 percent
imperviousness, five out of six equations suggest
that Q2 has increased by at least twofold with three
- C'1
N states exceeding 2.5. Over the range of 8 to 30 percentcC C) C)
'? C) -i ,- C)
'-, C) imperviousness, the Alabama and Tennessee equations
provide the median result, with peak discharge
increasing by a factor of 1.9 and 2.5 at 10 percent and
F! ,. a 0 0
0 20 percent imperviousness, respectively. Increasing—
basin area from 20 km2 to 50 or 100 km2 intensified
a the estimated increase in Q2 for some states since the
C') C')
,'. estimated urban Q2s increase more rapidly than the
C') '1 C')
rural Q2s with increasing basin area. Changes in
' I I basin area had little effect on the median results.
To further explore the potential effects of water-0 CC
a C')
shed imperviousness, the data of Sauer et al. (1983)II II II
were reanalyzed without BDF as an independent
- variable. Sauer et al. (1983) assembled a database of
269 sites from 31 states and 56 metropolitan areas.
, . These data represent a broad spectrum of watershedCC C'l C
'S . conditions from several states outside those withN CC
d S
urban equations as described above. Although many
Sa Q
_ of the data were taken from the eastern U.S., several—
gages in semi-arid to arid regions of Arizona, California,
and Colorado were also represented. As in the
a
previous analysis, Sauer et al. (1983) used USGS
S regression equations developed for rural watershedsC)
C) ? to provide an estimate of predevelopment discharges
bD'II (N
on a region-by-region basis.II II II
A simple statistical analysis of the Sauer et al.O
N (1983) data using categories of imperviousness from
(N Ii ') C')
- ' 0 to 30 percent indicates that the effects of totalII II II
- - a imperviousness on hydrologic response are extremely
variable (Figure 2). The data also suggest that a—
CC o twofold or greater increase in Q2 is frequent, even at
a a - C-. 0 s - very low levels of imperviousness. Of these 269 water-CD
COa-a 9-_ t sheds, 55 were reported to have "significant deten-0 .' a
. 'Z tion." Furthermore, 33 of the watersheds had
,CC
,.. estimates of rural Q2 that exceeded the estimated
a urban Q2. Only five of the sites identified as havingC- C-C N
C<
C
- C) N - "significant detention" had urban Q2 to rural Q2
o r ' ratios of less than one. All sites with "significant stor-a 0C'
- , age" and urban Q2 to rural Q2 ratios of less than one
C)
a a were included in the analysis. Including sites where
a E urbanization has decreased Q2 in the absence of "sigaaua
nificant storage" may be justified since unintentional-..s
ly undersized culverts and storm sewers have been
(N.—
a ' reported to significantly reduce peak discharges in
a some instances (Malcom, 1980).a — a aa S a 0
- -S a o U)
-e c aCD 4-'
( ZE ZJOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 257 JAWRA
Bledsoe and Watson
Impervious Area
Figure 1. Ratio of Median Annual Flood in an Urbanizing Watershed to That of a Rural Watershed as a Function of
Impervious Area Based on the USGS Flood Regression Equations for Six States. Watershed area is assumed
to equal 20 km2. The dashed line represents the relationship developed using NURP data.
Figure 2. Box and Whisker Plots of Post-Development Q2 to Predevelopment Q2 Ratio
of Different Levels of Imperviousness. (Data are from Sauer et al., 1983.)
U.S. Nationwide Urban Runoff Program Data
and the Rational Method. Data from the U.S.
Nationwide Urban Runoff Program (NURP; USEPA,
1983) indicate that the runoff coefficient used in the
rational method increases with the percentage of
imperviousness in a watershed according to the following
empirical equation:
C = (8.58 x l0-)I - (7.8 x 10-5)12 + 0.007741 + 0.04
(1)
where I = percent watershed imperviousness. The
rational method is a linear approach that relates the
JAWRA 258 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
5.00
4.50
4.00
3.50
0 2.50
2.00
1.50
1.00
0
0
,..-..'0—
/0.-;/
0
0
0
0
0
0
,
0
0
0
0/0 —
WI
NC
/___—
— TN
GA —
— MO —
—
—
5% 10% 15% 20% 25% 30%
7
6
5
2
0
0-5% 6-10% 10-15% 15-20% 20-25% 25-30% >30%
Total Impereousness
5%
75%
25%
0 Median
Effects of Urbanization on Channel Instability
peak discharge, drainage area, rainfall intensity, and
a runoff coefficient:
Q=C1A (2)
where C = the runoff coefficient (typically 0.75-0.95
for impervious areas); A = the watershed area corresponding
to the runoff coefficient; and i = rainfall
intensity at a duration equal to the time of concentra-
tion.
The units employed for rainfall intensity and
watershed area are typically in inches/hr and acres,
respectively. This results in discharge units of acreinches
per hour which is approximately equivalent to
ft3/s. Although several limiting assumptions are
implicit in the rational method (McCuen, 1989), it
remains one of the most widely used approaches for
basic hydrologic design in the United States. The
rational method is generally applicable to basin areas
less than about 0.8 km2 (ASCE, 1992; Wanielista et
al., 1997).
Runoff coefficients represent the combined effects
of land use, soils, geology, infiltration, and slope on
runoff potential. The data used to develop Equation
(1) were collected over a two-year period and are
indicative of urban runoff from events less than or
equal to a two-year storm. Multiple sites with less
than 2 percent imperviousness were included in the
study and resulted in a regression constant of 0.04 for
zero imperviousness. Using Equation (1), an estimate
of the relative increase in peak discharge was developed
for varying levels of watershed imperviousness.
This estimate is contrasted with increases in discharge
predicted by the USGS regression relationships
in Figure 1 and depicts the greater impact of
imperviousness on events less than Q2. The comparison
is based on the assumption that the pre- and postdevelopment
times of concentration are equal. In
reality, time of concentration decreases as runoff is
conveyed more rapidly by connected impervious areas
and drainage facilities. As time of concentration
decreases, the applicable precipitation intensity (i)
increases. Therefore, differences in the NURP and
USGS relationships depicted in Figure 1 would be
magnified if the effects of reduced times of concentration
were included. Conversely, a watershed with a
greater predevelopment C would likely exhibit a
smaller increase in peak flows. The data used to
develop Equation (1) indicate that C can vary from
0.03-0.35 for levels of impervious of 10 to 20 percent.
Runoff coefficients for undeveloped forest or grassland
areas are typically less than 0.2 for moderate slopes
with loamy soils but may range from virtually zero for
small events to values approaching impervious areas
when soils are compacted, clayey, or saturated by
large events.
Continuous Simulations of Hydrology in Urbanizing
Areas
The preceding analyses do not directly address the
frequency characteristics of sediment transport
events and flows less than a median annual flood. To
improve interpretation of the effects of urbanization
on the full range of channel-forming discharges, continuous
simulation models were also used to examine
changes in the magnitude, frequency, and duration of
flow and sediment transport characteristics in three
watersheds: over a 40-year period in the Hylebos and
Des Moines Creek watersheds in King County, Washington,
and over a growing season in Goodwin Creek
watershed in Mississippi at imperviousness levels of
0 to 30 percent.
HSPF Simulations of Hylebos and Des Moines
Creek Watersheds in King County, Washington.
In a series of papers examining the impacts of urbanization
on lowland watersheds in King County, Washington,
Booth (1990, 1991) and Booth et al. (1993,
1997) have suggested that even low levels of imperviousness
yield demonstrable, and probably irreversible,
loss of aquatic ecosystem function. These
papers evolved out of the innovative work of King
County, Washington, in developing strategies to mitigate
the adverse effects of land use change through
watershed-scale planning. A key component of the
King County approach is continuous hydrologic modeling
of changing land use scenarios using detailed
HSPF models.
To facilitate examination of the effects of imperviousness
over the entire range of flows, HSPF continuous
simulation models of watershed hydrology for
varying levels of imperviousness in two watersheds
were obtained from King County. The two watersheds,
Hylebos Creek and Des Moines Creek, were
selected because field data describing particle size
distributions and discharges resulting in mobility of
bedload were available. The watersheds are also very
similar in size and both stream beds are composed of
well-graded gravel with a sand matrix. King County
developed continuous HSPF models that were calibrated
to existing land use conditions in each watershed.
Selected characteristics of the two watersheds
at representative cross-sections are listed in Table 2.
For a full description of watershed and stream characteristics,
as well as calibration and validation procedures,
see King County (1991, 1997, 1998a, 1998b).
The Hylebos and Des Moines Creek models were
used in conjunction with 40-year records of meteorological
data to simulate daily maximum flows in the
watersheds with predevelopment forest conditions
and existing land use over the entire 40-year period.
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 259 JAWRA
Bledsoe and Watson
The results of these simulations were examined to
determine: (1) the magnitude of an annual flood relative
to pre-development conditions at various levels of
imperviousness, and (2) the frequency of flows that
result in a "significant scouring event" at various levels
of imperviousness. In contrast to a simple incipient
motion analysis, we defined a significant scouring
event as pervasive and extensive movement of all particle
sizes along with measurable scour as evidenced
by field measurements. The discharge threshold for a
significant souring event was based on field measurements
of bedload movement using tagged rocks and
scour pins over several large flow events (King County,
1997, 1998b). Based on these field measurements
and values of "critical discharge" identified by King
County, the site-specific flow magnitudes required for
the occurrence of a significant scour event were set at
2.4 m3/s and 1.4 m3/s for Des Moines and Hylebos
Creeks, respectively.
TABLE 2. Summary of Watershed and Representative
Stream Characteristics for Forested and Existing
Land Use Conditions in Des Moines and Hylebos
Watersheds, King County, Washington.
Des Moines Hylebos
Watershed Area 14.7 km2 15.1 km2
Existing Percent Impervious 37% 18%
D50 29 mm 20 mm
D84 51mm 56mm
Slope 0.01 0.015
Forested Q2 1.1 cms 0.9 cms
Existing Q2 4.8 cms 2.2 cms
Average Number of Significant
Scouring Events Per Year — Forested
<1 <1
Average Number of Significant
Scouring Events Per Year — Existing
Percent Impervious
11 5
The results of the 40-year simulations clearly indicate
the drastic changes in flow regime that may arise
from watershed imperviousness (Table 2). One of the
most striking changes is the estimated fivefold
increase in the frequency of significant scouring
events at 18 percent impervious area. At 37 percent
imperviousness, the frequencies of midbank to bankfull
flows and significant scouring events have
increased dramatically (Figure 3). In Hylebos watershed,
the estimated Q2 at 18 percent impervious area
is over 2.4 times the estimated predevelopment condition.
At 37 percent imperviousness, the estimated Q2
in Des Moines Creek watershed is over four times
larger than the estimated predevelopment annual
flood.
Continuous Hydrologic Simulations of Goodwin
Creek Watershed, Mississippi. Goodwin Creek
is an experimental watershed located in the bluff hills
region of north-central Mississippi. The watershed
area is approximately 21.4 km2, with elevations ranging
between 66 and 129 m above sea level. The watershed
is largely free of land management activities
with only 13 percent of the total area being cultivated.
The remainder of the land is idle, pasture, or forest
land (Kuhnle et al., 1996). Streams in the watershed
have relatively cohesive banks and beds composed of
sand and gravel. The watershed is divided into 14
nested subwatersheds with a flow-measuring flume
constructed at each drainage outlet. Continuous
streamflow and sediment load measurements are
made at these gaging locations by an electronic data
acquisition system. Twenty-nine standard recording
rain gages are uniformly located within and just outside
the watershed. Topographic surveys of channels
and bed materials are conducted regularly to track
morphometric changes in the basin. For a comprehensive
description of Goodwin Creek Experimental
Watershed, see Blackmarr (1995).
As one of the most highly instrumented watersheds
in the world, Goodwin Creek has served as the primary
watershed for developing the CASC2D watershed
model (Julien et al., 1995; Ogden, 2000). CASC2D is a
fully-unsteady, physically-based, distributed-parameter,
raster (square-grid), two-dimensional, infiltrationexcess
hydrologic model for simulating the hydrologic
response of a watershed subject to an input rainfall
field. Major components of the model include: continuous
soil-moisture accounting, rainfall interception,
infiltration, surface and channel runoff routing, soil
erosion, and sediment transport. Continuous simulation
capabilities have recently been added to
CASC2D. Although the continuous version of
CASC2D had not been released for public distribution
at the time of the study, it had been exhaustively tested,
calibrated, and validated for the growing season of
1987 in the Goodwin Creek watershed (Senarath et
al., 2000). Algorithms allowing application of the continuous
version of CASC2D to winter months when
freeze-thaw processes are important were not complete
at the time of the study and only the 1987 growing
season was modeled.
JAWRA 260 JOURNAL OF TI-1E AMERICAN WATER RESOURCES ASSOCIATION
700
C
0
E
(I)
I-
w
>-
400
300
C,
C)
0xw
I.-
0
I-
C,
E
z
600
500
200
100
0
Bedsiope
Effects of Urbanization on Channel Instability
Daily Maximum Discharge (m3Is)
Figure 3. Frequency Distribution of High Flow Events Over 40-Year Simulation of
Des Moines Creek Watershed With Two Land Use Scenarios.
TABLE 3. Characteristics of Goodwin Creek at the Experimental Watershed Outlet.
(Sources: Blackmarr,1995; and Molnár and RamIrez, 1998.)
0.001
Depth-Discharge Relationship
D84
Sand Concentration (ppm)
h = 0.294Q°428
0.97 mm
0.41 mm
11.52 mm
Cppm = 52.4e221461'; h > 0.372 m
Q2
Cppm = 320.21 h; h < 0.372 m
78.2 cms
CASC2D models have been previously calibrated to
the extensive monitoring network in the Goodwin
Creek watershed by other investigators (e.g. Johnson,
1997; Senarath et al., 2000). Pertinnt hydraulic characteristics
for the stream at the basin outlet are presented
in Table 3. The equations for concentration of
sand (bedload pius suspended load) were developed
and used by Willis (1991) to calculate sand loads for
the time period of 1985-1988 and have been independently
verified. A calibrated model of the Goodwin
Creek watershed with a 125 m grid size was used as a
base case for examining the effects of increased
imperviousness on the flow regime of a rural watershed.
Existing watershed conditions were contrasted
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 261 JAWRA
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 >2
Bledsoe and Watson
with multiple land use scenarios with impervious
areas ranging between 5 and 30 percent. Eleven input
maps representing infiltration, surface and subsurface
water movement, and evapotranspiration parameters
were modified for each scenario.
Both disconnected and connected imperviousness
scenarios were modeled. In the disconnected imperviousness
scenarios, selection of watershed areas for
conversion to imperviousness was based on two criteria.
Areas within the floodplain (defined as cells
either containing a stream segment or directly adjacent
to a cell containing a stream segment) and soils
identified as having low suitability for development
by the soil survey were excluded from conversion.
Outside the excluded areas, imperviousness was randomly
located throughout the watershed. This
approach distributed the impervious areas in a largely
disconnected pattern throughout the watershed
(Figure 4a). In contrast, the connected imperviousness
scenarios were developed by converting all cells
in a central region of the watershed to impervious
area (Figure 4b). Preliminary model runs revealed
that the effect of increasing impervious area on peak
discharge is very sensitive to the value of Manning's
n selected for overland flow. Recommended values in
the literature vary by an order of magnitude (Bedient
and Huber, 1992; HEC, 1990). As a result, two scenarios
of overland flow roughness were examined for
each spatial configuration with Manning's n for overland
flow equal to 0.1 and 0.013.
During the summer of 1987, Goodwin Creek experienced
three runoff events that approached or exceeded
10 m3/s (Events 1, 6, and 14; Figure 5). All
CASC2D simulations resulted in increased discharge
relative to base conditions at all levels of imperviousness.
Connected impervious areas clearly increased
peak flow magnitude with the effect particularly pronounced
for the combination of high conveyance and
connectivity (Figure 6). Increases in the peak discharge
of Event 1 at 15 percent impervious area varied
from 32 percent to 109 percent, depending on
spatial configuration and overland Manning's n value.
None of these simulated flow events approached
the magnitude of an annual flood at the outlet of
Goodwin Creek watershed. Nonetheless, changes in
the distribution of subbankfull flows at varying levels
of imperviousness suggest some interesting implications
for sediment transport processes in urbanizing
watersheds. Incipient motion and sand transport computations
were conducted using the scenario portraying
disconnected impervious area with Manning's n
for impervious areas equal to 0.1 to examine changes
in the frequency and duration of disturbance as
imperviousness increased in Goodwin Creek. The
threshold for movement of coarse sand and gravel size
classes was estimated using values of critical shear
stress from Julien (1995). The duration of sediment
movement for 15 percent and 30 percent imperviousness
was compared to existing conditions (Figure 7).
These estimates suggest that critical shear stress for
transport of D84 was only exceeded during the large
August event (Event 14) prior to the addition of
impervious area. With the addition of 5 percent or
more impervious area, critical shear stress for mobilizing
D84 was exceeded during all three large events
during summer of 1987. In addition, the duration of
sand concentrations in excess of 500 ppm was
increased over threefold at 30 percent impervious-
ness.
DISCUSSION OF HYDROLOGIC RESULTS
The preceding hydrologic analyses generally support
the assertion of Hollis (1975) that a twofold or
greater increase in the two-year discharge often
results at 10 to 20 percent watershed imperviousness.
It is clear, however, that the magnitude of the
response is highly variable and sensitive to watershed
context and the connectivity and conveyance characteristics
of developed areas. Data from Sauer et al.
(1983) indicate that measures of total imperviousness
fail to reflect watershed-specific controls on runoff
delivery and the ultimate effectiveness of impervious
areas (Booth and Jackson, 1997). The wide ranges of
runoff coefficients representing pervious areas and
similar levels of imperviousness (USEPA, 1983;
Schueler, 1995) reflect the sensitivity of the hydrologic
response to pre-development conditions and the
post-development efficiency of runoff delivery. The
extremely variable effects of imperviousness on peak
flows are echoed in theoretical analyses that couple
kinematic wave theory and differing combinations of
flow resistance and runoff coefficients (Wong and Li,
1999).
By including flows less than the annual flood
magnitude, the NURP study and the continuous simulations
indicate that flow magnification may be significant
for flow events smaller than Q2. Increasing
the frequency and duration of subbankfull flows can
result in a marked increase in time-integrated sediment
transport capacity in conjunction with more
moderate changes in annual flood peaks. Thus, the
continuous simulations and monitoring (Urbonas and
Roesner, 1993) suggest that standard hydrologic
design practices are inadequate for characterizing the
cumulative effects of urbanization on flow events that
are more frequent than Q2 but still significant in
terms of sediment transport and channel disturbance
JAWRA 262 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
Effects of Urbanization on Channel Instability
Figure 4. Maps of Imperviousness Scenarios in Goodwin Creek Watershed With Two Spatial
Configurations Represented. Shades correspond to subsets of cells that were converted in
each scenario to achieve the cumulative percent impervious area.
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 263 JAWRA
45
40
35
30
I
0a)I-
0.c 20
0U,
C
15
10
Bledsoe and Watson
Event 14
Figure 5. Continuous Simulation of Discharges in Goodwin at 0 and 30 Percent Disconnected
Imperviousness During Summer and 1987 With Overland Manning's n Equal to 0.1.
Imperviousness
Figure 6. Effects of Connectivity and Conveyance of Impervious Areas
on Flow Increases in CASC2D Simulation of Event 1.
JAWRA 264 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
5
Event 1
30%
0%
Event 6
-.0
19-May 29-May 8-Jun 18-Jun 28-Jun 8-Jul 18-Jul 28-Jul 7-Aug
1987
3.2
J
3
2.8
2.6
2.4
1.6
1.4
I.:
.'"
Disconnected, n = 0.1
——Connected, n = 0.1 H
—h— Disconnected, n = 0.013 I
-94—Connected, n = 0.013
.2___.:__
- ______
0% 5% 10% 15% 20% 25% 30%
Effects of Urbanization on Channel Instability
potential. It also follows that the impacts of urbanization
are likely to be greatest for suspended load channels
where a large percentage of the bed material load
is transported by frequent subbankfull flows.
E
a
iia0.
a
E
Figure 7. Duration of Transport for Various Bed Material
Fractions in Goodwin Creek for Two Levels of
Imperviousness Relative to Predevelopment Conditions.
Increases in the frequency and intensity of sediment
transport events not only affect channel morphology,
but in many instances may have adverse
effects on ecological processes (Newcombe and Jensen
1996; Waters 1995). The effects might include a shift
to small, vagile, and/or colonizing assemblages of
invertebrates and fish, selection for fish species possessing
a morphology that increases the likelihood of
surviving flood flows, and reduced influence of biotic
interactions on community structure (Poff and Ward,
1989). Altered stream hydraulics could also affect
zonation patterns in stream insects (Statzner and
Higler, 1986). Comprehensive analyses of the potential
impact of land use change should therefore carefully
consider effects on subbankfull flows and
substrate disturbance regime in addition to peak flow
magnitude.
FLOW ENERGY AND RISK OF
CHANNEL INSTABILITY
In a study of 270 streams and rivers, Bledsoe and
Watson (In Press) suggested that logistic regression
models (Tung, 1985; Menard, 1995; Christensen,
1997) can accurately predict unstable channel forms
with a "mobility index" based on slope (S; bedslope or
valley slope), an estimate of channel-forming discharge
(Q), and median bed material size (d50). The
mobility index is defined as:
(3)
The numerator of this ratio is a surrogate for specific
stream power (Bagnold, 1966; Schumm and Khan,
1972; Bull, 1979; Edgar, 1973, 1976; Brookes, 1988;
Nanson and Croke, 1992; Rhoads, 1995) or the rate of
potential energy expenditure per unit area. Specific
stream power is proportional to the third power of
depth-averaged velocity and is defined as (Leopold et
al., 1964):
(4)
where y is specific weight of the water and sediment
mixture, Q is discharge, S is slope, and w is channel
width. By assuming that channel width is proportional
to the square root of dominant discharge
(Knighton, 1998), a relation for approximating specific
stream power is (Ferguson, 1987):
YQ°5SoS..j (5)
where a is a regression coefficient computed for a particular
collection of streams. Equation (3) therefore
represents an expression of flow energy relative to
bed material size. Because we are often interested in
predicting unstable channel forms a priori, Equation
(3) does not require channel width or depth since
these are often a consequence of the adjustment process
and may not represent channel form prior to an
increase in stream power. The logistic analyses performed
by Bledsoe and Watson (in press) and a comparison
of channel stability indices by Doyle et al.
(2000) indicate that assessments of channel stability
based on a ratio of erosive to resisting forces or some
measure of excess power outperform absolute measures
of flow energy where bed materials are variable.
Using Equation (3), logistic regression models
depicting the risk of an incising or braided channel
form were developed (Figure 8). These models explicitly
provide the estimated risk of an unstable channel
form with greater than 80 percent accuracy for the
270 streams and rivers examined. Moreover, analysis
of the 270 channels suggested that sinuosity tends to
decline abruptly at S(Q/d50)05 values greater than
approximately 0.5 mIs05 and 0.2 mIs05 in sand and
gravel bed rivers, respectively (Bledsoe and Watson,
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 265 JAWRA
3.5
3
2.5
2
1.5
sand @500 ppm 4mm gravel 12mm gravel (084)
0.9
0
0.8
0C
0.7
C)
0)
.0
. 0.50)
C
0.4I-0
0.3
.0
0.2
.0
2
. 0.1
0
Bledsoe and Watson
Figure 8. Results of Logistic Regression Models Comparing Stable Meandering Channels With Braiding
and Incising Channels. Median particle sizes ranged from 0.075-1.69 mm and 10-190 mm
in the sand and gravel channels, respectively (after Bledsoe and Watson, in press).
2000). In gravel channels, the threshold associated
with a sharp decline in sinuosity may be more pertinent
than the transition to braiding. Logistic regression
models comparing meandering and braiding
streams in gravel and coarser material suggested that
an abrupt decline in sinuosity seems to occur at the
mobility index associated with a 25 to 40 percent risk
of braiding in gravel channels. Thus, S(Q/d50)°5 values
of approximately 0.18 to 0.3 mIs05 may bracket a
region of accelerated widening and straightening in
gravel bed channels where forms intermediate
between meandering and braiding would be expected
to predominate. To the extent that bank resistance to
erosion is independent of bed material size, the mobility
index described above is limited in its applicability
to lateral adjustment processes. In reality, the severity
of response to an increase in specific stream power
in gravel bed channels may be controlled more by
bank characteristics than bed material. As we obtain
more information on how vegetation, materials, and
stratigraphy affect the relative erodibility of various
bank types, simple indices of flow energy may be useful
in risk-based models of lateral adjustment poten-
tial.
INTERPRETING THE EFFECTS
OF URBANIZATION
Examining the urban hydrology analyses in light of
the logistic regression models provides a framework
for explaining channel instability in terms of excess
specific stream power. It appears that in many watersheds
in humid regions, the increases in stream
power associated with low levels of imperviousness
(10 to 20 percent) are sufficient to markedly increase
the likelihood of severe channel instability. This finding
reinforces the observations of Booth and Reinelt
(1993) and Morisawa and LaFlure (1979) that signifi.
cant channel instability can occur at 10 percent
watershed imperviousness in some contexts. Making
general statements about the ultimate effects of a
flow increase on channel stability, however, is hardly
straightforward. A final look at the case studies on
Hylebos, Des Moines and Goodwin Creeks illustrates
this point and suggests some interesting implications
for applying probabilistic models of channel pattern
stability in the management of flow increases associated
with urbanization.
Hylebos and Des Moines Creeks
Under existing land use conditions, Hylebos and
Des Moines Creeks have mobility indices of 0.16 and
0.13 mIs05, respectively. Although significantly higher
than the estimated pre-development values, these
existing values of the mobility index are still well
below a 50 percent risk of braiding and the apparent
limits of meandering in gravel bed channels. In reality,
both channels have experienced bank erosion as a
result of flow increases. For example, Hylebos Creek
is reported to have a width that is approximately
twice that of streams in undeveloped watersheds
of similar size (King County, 1997). These results
JAWRA 266 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
S(QId)°5(mis°5)
Effects of Urbanization on Channel Instability
indicate that gravel bed channels may experience
severe bank erosion at relatively low levels of risk for
braiding.
It is interesting that although Des Moines Creek
has experienced widening, it is apparently much less
severe than that observed in Hylebos Creek, despite
the fact that Des Moines Creek watershed contains
about twice the impervious area of Hylebos. The geologic
and land use histories of Hylebos and Des
Moines Creek watersheds partly explain this difference.
Both basins are underlain by a sheet of glacial
till deposited during the most recent continental
glacial advance (King County, 1991, 1997, 1998a,
1998b). In post-glacial time, Hylebos Creek has eroded
through the till sheet and exposed the underlying
advance outwash. The advance outwash is composed
primarily of sand and is much more erodible and permeable
than overlying till. In Hylebos watershed,
sandy advance outwash underlies essentially the
entire length of the stream bed. In contrast, the lower
reach of Des Moines Creek has developed a graded
profile into Puget Sound by incising through the
upper glacial and lower mixed deposits to reveal compact
and non-erosive deposits along most of the
stream valley sidewalls. The urban development in
Des Moines watershed is also relatively old compared
to Hylebos (one to four decades). As a result, Des
Moines Creek has had a longer period to adjust to the
post-development hydrologic regime. Riparian disturbance
is apparently greater along Hylebos Creek as
well.
The differences in response between Des Moines
and Hylebos Creeks underscore the fact that channel
adjustment is always context dependent and influenced
by historical processes. Des Moines Creek is
formed in relatively resistant material while the
banks of Hylebos Creek are highly susceptible to erosion
as a result of geologic history and vegetation
removal. Thus, descriptors of erosive power such as
Equation (3) must be applied to the limiting channel
boundary with some knowledge of its relative resistance
to erosion. When viewed in terms of a limiting
boundary, the significant lateral adjustment of Hylebos
at a low value of S(Q1d50)°5 becomes less anomalous.
If one assumes that channel width may be
estimated with a typical downstream hydraulic geometry
equation for gravel bed rivers (w = 4Q05;
Knighton 1998), it is likely that the actual specific
stream power in Hylebos has increased from less than
35 W/m to more than 50 W/m2. These conditions are
capable of mobilizing bank material that is noncohesive
and sandy, especially in the absence of dense veg-
etation.
Goodwin Creek
Like Hylebos and Des Moines Creeks in the Pacific
Northwest, Goodwin Creek has a history that will
undoubtedly continue to influence the course of channel
adjustments to watershed changes. During European
settlement in the 1830s, much of Goodwin Creek
watershed was deforested and converted to intensive
cotton agriculture. Severe erosion followed and incision
is still in progress in many reaches (Kuhnle et
al., 1996). At the watershed outlet, where the hydrologic
modeling was focused, the stream is relatively
stable when compared to several upstream reaches
(Molnár and RamIrez, 1998). The mobility index of
the stream at this location is 0.28 mIs05. This value
suggests that the stream may be poised near a geomorphic
threshold since the risk of severe incision
begins increasing very rapidly at this point and reaches
a risk of 50 percent at a mobility index of 0.44
mis05. If we assume that specific stream power varies
with the square root of discharge, a twofold increase
in flow would correspond to a 41 percent increase in
the mobility index. But such an approach is misleading
since Goodwin Creek is a highly entrenched channel
with very steep, cohesive banks. The channel has
the capacity to contain a flow event much larger than
Q2. As a result, shear stress and specific stream
power will increase in direct proportion to discharge.
A key determinant of channel response to
increased discharge is how the excess flow is conveyed,
for example, in an entrenched channel or on a
broad floodplain. If we assume that the increased discharge
is conveyed at the approximate width of the
pre-development channel, then stream power will
increase in direct proportion to discharge. Conversely,
if the excess flow is spread across a wide floodplain,
the resulting change in specific stream power will be
relatively small. It follows that entrenched channels
are particularly susceptible to flow increases. Thus,
even a modest increase in discharge might be adequate
to initiate severe incision throughout the Goodwin
Creek watershed, particularly in upstream
reaches that have a much higher stream power when
compared to the watershed outlet reach considered
here (Molnár and RamIrez, 1998). Given the localized
changes between sand and gravel bed materials that
occur in the Goodwin watershed, areas of relatively
high specific stream power are more difficult to interpret.
In this regard, the Goodwin Creek context points
to the importance of identifying spatial variability in
stream power relative to sedimentary characteristics
throughout the drainage network area potentially
affected by flow increases.
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 267 JAWRA
Bledsoe and Watson
CONCLUSIONS
The previous discussion indicates that probabilistic
models do not provide a substitute for field observations,
sound judgment, and an appreciation for the
importance of antecedent events. Nonetheless, simple
risk-based analyses of the potential impacts of land
use change on aquatic ecosystems have the potential
to redirect and improve the effectiveness of watershed
management strategies by facilitating the identification
of channels that may be most sensitive to
changes in stream power. Energy-based indices of
channel stability should, however, be applied with an
understanding of context and be referenced to the
erodibility of the limiting channel boundary. The variables
included in risk-based models of channel
response should be calibrated regionally and extended
beyond those described in this study. Decision-based
models of stream stability and ecological integrity
may be developed using descriptors of key flow regime
attributes, the condition of channel banks and riparian
zones, geologic or wood influences, floodplain connectivity,
and development style in addition to
channel stability indices.
Low levels of imperviousness (10 to 20 percent)
clearly have the potential to severely destabilize
streams, but changes in discharge, and thus stream
power associated with increased impervious area are
highly variable and dependent on watershed-specific
conditions. In addition to the storage characteristics
of the pre-development watershed, the magnitude of
alteration is sensitive to the connectivity and conveyance
of impervious areas as well as the specific
characteristics of receiving channels. Different stream
types are likely to exhibit different levels of resilience,
depending on entrenchment, relative erodibility of
bed and banks, riparian condition, mode of sediment
transport (bedload versus suspended load), and proximity
to geomorphic thresholds. Thus, the relationship
between channel instability and imperviousness
is complex and involves several factors and processes
(Bledsoe and Watson, 2000; Figure 9).
The potential impacts of urbanization on subbankfull
flow frequency, time-integrated sediment transport
capacity, and channel stability are difficult to
characterize with single event hydrologic models that
rely on measures of total impervious area. Since first
and second order streams comprise over 70 percent of
the stream length in the U.S. (Leopold et al., 1964)
and as much as 90 percent in some regions (Benda
and Dunne, 1997), engineered systems with designs
based on the rational method may directly affect a
large percentage of the drainage networks in practice.
Mitigating stream degradation in urban watersheds
will often necessitate consideration of flow regime as
opposed to a single design event in order to realistically
assess the cumulative impacts of urbanization
on fluvial systems.
IMPERVIOUSNESS I
I
Connectivity, Conveyance, Change in Storage, and
Mitigation Scheme
',
CHANGE IN
SPECIFIC STREAM POWER
1
Proximity to a Geomorphic Threshold
Entrenchment
Floodplain and Bank Condition
Time-integrated Sediment Transport
Resilience of Stream Type
RISK
OF STREAM INSTABILITY
Figure 9. Linkages Between Imperviousness
and Channel Stability.
Strategies intended to mitigate the degradation of
urban streams should be guided by an understanding
of the underlying processes. An incomplete understanding
of the mechanisms involved, however, does
not preclude immediate and meaningful improvements
in existing management programs. Most studies
associating urbanization with loss of stream
integrity do not address the particular styles of development
involved and their effects on specific physical
processes (e.g., Kennen, 1999). Ultimately, the diverse
responses of aquatic ecosystems to various types of
development provide important feedback and an
opportunity for adaptive management of urbanizing
watersheds. Because all imperviousness is not created
equal, there is a pressing need to link specific development
styles, flow regime, risk-based models of
stream integrity, and monitoring. Such a linkage
JAWRA 268 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION
[
Effects of Urbanization on Channel Instability
would provide an improved foundation for innovative
site planning, ecologically effective stormwater practices,
and meaningful estimates of watershed development
capacity.
ACKNOWLEDGMENTS
We are grateful to two anonymous reviewers for their insightful
comments and to Neil Grigg, Ellen Wohl, and LeRoy Poff for
reviewing an earlier version of this work. We also wish to express
our sincere thanks to Sharika Senarath, Hatim Sharif, and Fred
Ogden of the University of Connecticut for assistance with the
CASC2D continuous simulations; and to David Hartley and Jeff
Burkey of the King County Basin Planning Program for providing
HSPF models and background information on Hylebos and Des
Moines Creeks. Darcy Molnar also provided invaluable assistance
with the CASC2D models. Robin Kelley and Lejo Flores provided
skillful assistance with statistical analyses and hydrologic modeling.
This article was developed as a product under joint sponsorship
of the USEPA and the U.S. Army Corps of Engineers
(USACE). Although it has been reviewed by the USACE, it has not
been formally reviewed by USEPA. The views expressed in this document
are solely those of the authors and neither USEPA nor the
USACE endorse any products or commercial services mentioned in
this publication.
LITERATURE CITED
ASCE, 1992. Design and Construction of Urban Stormwater Management
Systems. Manual of Practice No. 77, American Society
of Civil Engineers.
Arnold, C. L., P. J. Boison, and P. C. Patton, 1982. Sawmill Brook:
An Example of Rapid Geomorphic Change Related to Urbanization.
Jl. Geology 90: 155-166.
Bagnold, R. A., 1966. An Approach to the Sediment Transport Problem
From General Physics. U.S. Geological Survey Professional
Paper 4221.
Bedient, P. B. and W. C. Huber, 1992. Hydrology and Floodplain
Analysis. Addison-Wesley, Reading, Massachusetts.
Becker, L. D., 1986. Techniques for Estimating Flood Peak Discharges
From Urban Basins in Missouri. U.S. Geological Survey
Water-Resources Investigations Report 86-4322, 38 pp.
Benda, L. and T. Dunne, 1997. Stochastic Forcing of Sediment Supply
to Channel Networks From Landsliding and Debris Flow.
Water Resources Research 33(12):2849-2863.
Blackmarr, W. A. (Editor), 1995. Documentation of Hydrologic, Geomorphic,
and Sediment Transport Measurements on the Goodwin
Creek Experimental Watershed, Northern Mississippi, for
the Period 1982-1993 (Preliminary Release). Research Report
No. 3, USDA-ARS, National Sedimentation Laboratory; Oxford,
Mississippi.
Bledsoe, B. P. and C. C. Watson (in press). Logistic Analysis of
Channel Pattern Thresholds: Meandering, Braiding, and Incising.
Geomorphology.
Bledsoe, B. P. and C. C. Watson, 2000. Observed Thresholds of
Stream Ecosystem Degradation in Urbanizing Areas: A ProcessBased
Geomorphic Explanation. In: Watershed Management
2000: Science and Engineering Technology for the New Millennium,
M. Flug and D. Frevert (Editors). American Society of
Civil Engineers.
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. Northwest Environmental
Jl. 7(1):93-118.
Booth, D. B. and L. E. Remelt, 1993. Consequences of Urbanization
on Aquatic Systems — Measured Effects, Degradation Thresholds,
and Corrective Strategies. In: Proc. of Watershed '93. EPA
840-R-94-002, pp. 545-550.
Booth, D. B. and C. R. Jackson, 1997. Urbanization of Aquatic Systems
— Degradation Thresholds, Stormwater Detention, and the
Limits of Mitigation. Jl. American Water Resources Association
33(5):1077-1090.
Brookes, A., 1988. Channelized Rivers: Perspectives for Environmental
Management. John Wiley and Sons Ltd., Chichester,
England.
Bull, W. B., 1979. Threshold of Critical Power in Streams. Geological
Society of America Bulletin 90:453-464.
Christensen, R., 1997. Log-Linear Models and Logistic Regression.
Springer-Verlag, New York, New York.
Conger, D. H., 1986. Estimating Magnitude and Frequency of
Floods for Wisconsin Urban Streams. U.S. Geological Survey
Water-Resources Investigations Report 86-4005, 18 p.
Doyle, M. W., J. M. Harbor, C. F. Rich, and A. Spacie, 2000. Examining
the Effects of Urbanization on Streams Using Indicators of
Geomorphic Stability. Physical Geography 21(2): 155-181.
Edgar, D. E., 1973. Geomorphic and Hydraulic Properties of Laboratory
Rivers. M.S. Thesis, Colorado State University, Fort
Collins, Colorado.
Edgar, D. E., 1976. Geomorphology and Hydrology of Selected Midwestern
Streams. Ph.D Dissertation, Purdue University, West
Lafayette, Indiana.
Ferguson, R. I., 1987. Hydraulic and Sedimentary Controls on
Channel Pattern. In: River Channels: Environment and Process,
K. S. Richards (Editor). Blackwell, Oxford, pp. 129-158.
Gunter, H. C., R. R. Mason, and T. C. Stamey, 1987. Magnitude
and Frequency of Floods in Rural and Urban Basins of North
Carolina. U.S. Geological Survey Water-Resources Investigations
Report 87-4096, 52 pp.
Hammer, T. R., 1972. Stream Channel Enlargement Due to Urbanization.
Water Resources Research 8:139-167.
HEC, 1990. HEC-1 Flood Hydrograph Package Users Manual.
Hydrologic Engineering Center, U.S. Army Corps of Engineers,
Davis, California.
Hollis, G. E., 1975. The Effect of Urbanization on Floods of Different
Recurrence Intervals. Water Resources Research 11(3):431-
435.
Inman, E. J., 1988. Flood-Frequency Relations for Urban Streams
in Georgia. U.S. Geological Survey Water-Resources Investigations
Report 88-4085, 36 pp.
Jennings, M. E., W. 0. Thomas, Jr., and H. C. Riggs, 1993. Nationwide
Summary of U.S. Geological Survey Regional Regression
Equations for Estimating Magnitude and Frequency of Floods
for Ungaged Sites, 1993. U.S. Geological Survey WaterResources
Investigations Report 94-4002.
Johanson, R. C., J. C. Imhoff, and H. Dana, Jr., 1980. Users Manual
for Hydrological Simulation Program B Fortran (HSPF). U.S.
Environmental Protection Agency, EPA/9-80-015, Athens, Geor-
gia.
Johnson, Jr., B. E., 1997. Development of a Storm Event Based
Two-Dimensional Upland Erosion Model. Ph.D Dissertation,
Colorado State University, Fort Collins, Colorado.
Julien, P. Y., B. Saghafian, and F. L. Ogden, 1995. Raster-Based
Hydrologic Modeling of Spatially-Varied Surface Runoff. Water
Resources Bulletin 31(3):523-536.
Julien, P. Y., 1995. Erosion and Sedimentation. Cambridge University
Press, New York, New York.
JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 269 JAWRA
Bledsoe and Watson
Kennen, J. G., 1999. Relation of Macroinvertebrate Community
Impairment to Catchment Characteristics in New Jersey
Streams. Jl. American Water Resources Association 35(4):939-
954.
King County, 1991. Executive Proposed Basin Plan, Hylebos and
Lower Puget Sound. King County Public Works Department,
Surface Water Management Division.
King County, 1997. Conceptual Alternative Design Report, South
360th Street Regional R/D Facility, Hylebos and Lower Puget
Sound Basin Plan Project No. 0A1685. King County Public
Works Department, Surface Water Management Division.
King County, 1998a. Des Moines Creek Basin Plan. King County
Public Works Department, Surface Water Management Divi-
sion.
King County, 1998b. Des Moines Creek Regional CIP Preliminary
Design Report. King County Public Works Department, Surface
Water Management Division.
Knighton, A. D., 1998. Fluvial Forms and Processes: A New Perspective.
John Wiley and Sons Ltd., New York, New York.
Kuhnle, R. A., R. L. Binger, G. R. Foster, and E. H. Grissinger,
1996. Effect of Land Use Changes on Sediment Transport in
Goodwin Creek. Water Resources Research 32:3189-3196.
Leopold, L. B., M. G. Wolman, and J. P. Miller, 1964. Fluvial Processes
in Geomorphology. W. H. Freeman and Co., San Francisco,
California, 522 pp.
Malcom, H. R., 1980. A Study of Detention in Urban Storm-Water
Management. University of North Carolina Water Resources
Research Institute Report No. 156, 78 pp.
McCuen, R. H., 1989. Hydrologic Analysis and Design. Prentice
Hall, Englewood Cliffs, New Jersey.
Menard, S. W., 1995. Applied Logistic Regression Analysis. Sage
Publications, Thousand Oaks, California.
Molnár, P., and J. A. RamIrez, 1998. An Analysis of Energy Expenditure
in Goodwin Creek. Water Resources Research 34(7):1819-
1829.
Morisawa, M. and E. LaFlure, 1979. Hydraulic Geometry, Stream
Equilibrium and Urbanization. In: Adjustments of the Fluvial
System, D. D. Rhodes and G. P. Williams (Editors). KendallHunt,
Dubuque, Iowa, pp. 333-350.
Nanson, G. C. and J. C. Croke, 1992. A Genetic Classification of
Floodplains. Geomorphology 4:459-486.
Newcombe, C. P. and J. 0. Jensen, 1996. Channel Suspended Sediment
and Fisheries: A Synthesis for Quantitative Assessment of
Risk and Impact. North American Jl. Fisheries Management
16(4):693-727.
Ogden, F. L., 2000. CASC2D Reference Manual. Version 2.0,
Department of Civil and Environmental Engineering, University
of Connecticut.
Olin, D. A., 1984. Magnitude and Frequency of Floods in Alabama.
U.S. Geological Survey Water-Resources Investigations Report
84-4191, 105 pp.
Olin, D. A. and R. H. Bingham, 1982. Synthesized Flood Frequency
of Urban Streams in Alabama. U.S. Geological Survey WaterResources
Investigations Report 82-683, 23 pp.
Poll, N. L. and J. V. Ward, 1989. Implications of Strearnflow Variability
and Predictability for Lotic Community Structure: A
Regional Analysis of Streamfiow Patterns. Canadian Jl. Fisheries
and Aquatic Sciences 46:1805-1818.
Price, M., 1979. Floods in Georgia, Magnitude and Frequency. U.S.
Geological Survey Water-Resources Investigations Report 78-
137, 269 pp.
Putnam, A. L., 1972. Effect of Urban Development on Floods in the
Piedmont Province of North Carolina. U.S. Geological Survey
Open-File Report, 87 pp.
Rhoads, B. L., 1995. Stream Power: A Unifying Theme for Urban
Fluvial Geomorphology. In: Stormwater Runoff and Receiving
Systems: Impact, Monitoring, and Assessment, E. E. Herricks
(Editor). Lewis Publishers, Boca Raton, Florida, pp. 65-75.
Robbins, C. H., 1984. Synthesized Flood Frequency for Small
Urban Streams in Tennessee. U.S. Geological Survey WaterResources
Investigations Report 84-4182, 24 pp.
Roberts, C. R., 1989. Flood Frequency and Urban-Induced Channel
Change: Some British Examples. In: Floods: Hydrological, Sedimentological
and Geomorphological Implications, K. J. Beven
and P. A. Carling, (Editors). John Wiley and Sons Ltd., Chichester,
England, pp. 57-82.
Sauer, V. B., W. 0. Thomas, V. A. Stricker, and K. V. Wilson, 1983.
Flood Characteristics of Urban Watersheds in the United
States. U.S. Geological Survey Water Supply Paper 2207.
Schueler, T., 1994. The Importance of Imperviousness. Watershed
Protection Techniques 1(3):100-111.
Schumm, S. A. and H. R. Khan, 1972. Experimental Study of Channel
Patterns. Geological Society of America Bulletin 83:1755-
1770.
Senarath, S. U. S., F. L. Ogden, C. W. Downer, and H. 0. Sharif,
2000. On the Calibration and Verification of Two-Dimensional,
Distributed, Hortonian, Continuous Catchment Models. Water
Resources Research (in press).
Stamey, T. C. and G. W. Hess, 1993. Techniques for Estimating
Magnitude and Frequency of Floods in Rural Basins of Georgia.
U.S. Geological Survey Water-Resources Investigations Report
93-4016, 75 pp.
Statzner, B. and B. Higler, 1986. Stream Hydraulics as a Major
Determinant of Benthic Invertebrate Zonation Patterns. Freshwater
Biology 16: 127-139.
Trimble, S. W., 1995. Catchment Sediment Budgets and Change.
In: Changing River Channels, A. M. Gurnell and G. E. Petts
(Editors). John Wiley and Sons Ltd., Chichester, England, pp.
201-215.
Trimble, S. W., 1997. Contribution of Stream Channel Erosion to
Sediment Yield From an Urbanizing Watershed. Science
278:1442-1444.
Tung, Y., 1985. Channel Scouring Potential Using Logistic Analysis.
Jl. Hydraulic Engineering 111(2):194-205.
USEPA, 1983. Results of the Nationwide Urban Runoff Program.
Final Report, NTIS Access No. PB84-18552, Washington, D.C.
USEPA, 1984. Hydrologic Simulation Program — Fortran (HSPF).
EPA-600-R93-174 and EPA-600-3-84-065.
Urbonas, B. R. and L. A. Roesner, 1993. Hydrologic Design for
Urban Drainage and Flood Control. In: Handbook of Hydrology,
D. R. Maidment (Editor). McGraw-Hill, Inc., Chapter 28, p. 28.1.
Wanielista, M., R. Kersten, and R. Eaglin, 1997. Hydrology Water
Quantity and Quality Control. John Wiley and Sons Ltd, New
York, New York.
Waters, T. F., 1995. Sediment in Streams: Sources, Biological
Effects, and Control. American Fisheries Monograph 7, American
Fisheries Society, Bethesda, Maryland, 251 pp.
Weaver, J. D. and C. R. Gamble, 1993. Flood Frequency of Streams
in Rural Basins of Tennessee. U.S. Geological Survey, WaterResources
Investigations Report 92-4165, 38 pp.
Willis, J. C., 1991. Sand Transport on Goodwin Creek. In: Proc. 5th
Fed. Interagency Sedimentation Conference, Las Vegas, Nevada,
pp. 4-131 to 4-138.
Wolman, M. G., 1967. A Cycle of Sedimentation and Erosion in
Urban River Channels. Geografiska Annaler 49A:385-395.
Wong, T. S. W. and Y. Li, 1999. Theoretical Assessment of Changes
in Design Flood Peak of an Overland Flow Plane for Two Opposing
Urbanization Sequences. Hydrological Processes 13:1629-
1647.
JAWRA 270 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION