Are BMP Criteria Really Environmentally Friendly?

Abstract

In the 1990's, a number of best management practices (BMPs) manuals have been developed that address the control of urban runoff to protect receiving water quality. More recently, several papers have investigated the effectiveness of these BMPs in protecting small urban watercourses, and have concluded that they do not. Investigations of both design practices and effectiveness reveals that there is a lot of ignorance in the scientific and engineering community about what constitutes a properly designed BMP and what it really achieves, with respect to environmental protection. This paper discusses the state-of-practice in BMP design in the United States and points out its strengths and weaknesses with respect to real protection of the downstream receiving water environment. The paper recommends an approach to design criteria development that can be applied over a wide variety of climatologic, topologic, and geologic conditions to protect receiving waters systems.

Full text

150 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT / MAY/JUNE 2001
A
RE
B
EST
-M
ANAGEMENT
-P
RACTICE
C
RITERIA
R
EALLY
E
NVIRONMENTALLY
F
RIENDLY
?
By Larry A. Roesner,
1
P.E., Fellow, ASCE, Brian P. Bledsoe,
2
P.E.,
and Robert W. Brashear,
3
P.E., Members, ASCE
A
BSTRACT
:In the 1990’s, a number of best management practices (BMPs) manuals have been developed that
address the control of urban runoff to protect receiving water quality. More recently, several papers have inves-
tigated the effectiveness of these BMPs in protecting small urban watercourses, and have concluded that they
do not. Investigations of both design practices and effectiveness reveals that there is a lot of ignorance in the
scientific and engineering community about what constitutes a properly designed BMP and what it really
achieves, with respect to environmental protection. This paper discusses the state-of-practice in BMP design in
the United States and points out its strengths and weaknesses with respect to real protection of the downstream
receiving water environment. The paper recommends an approach to design criteria development that can be
applied over a wide variety of climatologic, topologic, and geologic conditions to protect receiving waters sys-
tems.
INTRODUCTION
In the 1990’s, so called Best Management Practices(BMPs),
have been used more and more to control the pollution of
urban runoff and ostensibly protect the receiving waters to
which the runoff drains. More recently some investigators [e.g.
Maxted and Shaver (1997) and Schueler (1999)] have offered
opinions that these BMPs do not protect the downstream
aquatic environment, and Schueler (1999) now argues that a
different approach to management of urban runoff is required
to protect urban streams. However, it is the writers’ opinion
that the problem is not the BMPs themselves, but rather that
the design guidance for BMP outlet flow control does not take
into account the geomorphologic character of the stream. This
paper examines these issues and provides comments for im-
proving the design of BMPs so that they are more friendly to
the environment. The paper is directed at urban headwater
streams, not larger streams or rivers that flow through an urban
area.
IMPACT OF URBANIZATION ON RUNOFF
The effect that urbanization has on a watershed (Fig. 1) has
been well documented in the literature, but we cover the high-
lights here again as background for the subsequent discussion.
Undeveloped land has very little surface runoff; most of the
rainfall soaks into the top soil and evapotranspirates or mi-
grates slowly through the soil mantle as interflow to the
stream, lake, or estuary. As a result of this process, rainfall
effects are averaged out over a long period of time (Fig. 1).
But as a watershed develops and the land is covered over with
impervious surfaces (roads, parking lots, roofs, driveway, and
sidewalks), most of the rainfall is transformed into surface
runoff.
The resulting effect on the hydrology of the receiving water
is dramatic, especially for streams. A given rainstorm now pro-
duces significantly more runoff volume than before, and flow
1
Prof., Dept. of Civ. Engrg., Colorado State Univ., Fort Collins, CO
80523-1372.
2
Res. Assoc., Dept. of Civ. Engrg., Colorado State Univ., Fort Collins,
CO 80523-1372.
3
Sr. Water Resour. Engr., Camp Dresser & McKee Inc., 8140 Walnut
Hill Lane, Ste. 1000, Dallas, TX 57231.
Note. Discussion open until November 1, 2001. To extend the closing
date one month, a written request must be filed with the ASCE Manager
of Journals. The manuscript for this paper was submitted for review and
possible publication on November 20, 2000. This paper is part of the
Journal of Water Resources Planning and Management, Vol. 127, No.
3, May/June, 2001. 䉷ASCE, ISSN 0733-9496/01/0003-0150– 0154/$8.00
⫹$.50 per page. Paper No. 22350.
peaks are increased by a factor of 2 to more than 10. The
overall hydrologic effect is that the flow frequency curve for
a developed area is significantly higher than for an undevel-
oped area as shown in Fig. 2. This change in the flow fre-
quency curve manifests itself in two ways. First, as just men-
tioned, the peak runoff rate for a given return period storm
(rainfall) increases (point A in Fig. 2). In the Metropolitan
Denver (Joint Task Force 1998) area these increases range
from a factor of 2 to 60 as illustrated in Table 1. The extremely
high increase in the two-year peak runoff is due to the fact
that predevelopment runoff was nearly zero. In other localities
the increase may not be as dramatic, but in general, the per-
centage increase in the peak flow is highest for small, frequent
storm events as Fig. 2 illustrates.
The second effect that urbanization has on runoff is to sig-
nificantly increase the frequency of the predevelopment peak
flows (point B in Fig. 2). Impacts reported by Schueler (1987),
in Maryland are illustrated in Table 2. The table shows that
the two-year peak runoff rate in the predeveloped state occurs
three times per year if the area is developed as residential, and
eight times per year if the area is developed as industrial prop-
erty. This is an increase in frequency of 6–18 times!
FLOW IMPACTS ON RECEIVING WATERS
The increased magnitude and frequency of these flow peaks
can cause severe stream channel erosion and increased flood-
ing downstream. The most commonly observed effects are the
physical degeneration of natural stream channels. The higher
frequency of the peak flows causes the stream to cut a deeper
and wider channel (Fig. 3), degrading or destroying the in-
stream aquatic habitat. The eroded sediments are deposited
downstream in slower moving reaches of the stream or at the
entrance to lakes or estuaries, harming the aquatic habitat in
these areas by smothering the benthos, filling wetlands with
sediment, and so forth. As a result of the erosion and sedi-
mentation, decreases in biodiversity and numbers of aquatic
stream biota are commonly observed in both areas.
The hydroperiod of the wetlands in these watercourses are
also drastically changed, experiencing high flows for short pe-
riods during and after rainfall, followed by a period of much
reduced or zero flow, due to the reduction of interflow. Fresh-
water wetlands can dry up or become unsightly bogs. Saltwa-
ter wetlands will deteriorate due to the increases in the fre-
quency of large freshwater flow into them, or they may convert
to freshwater wetlands if the rainfall frequency is high enough
to keep a supply of freshwater running through them. The
effect of these changes in the wetland causes significant stress

JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT / MAY/JUNE 2001 / 151
FIG. 1. Urban Impacts on Hydrology
FIG. 2. Effect of Urbanization on Flow Frequency Curve
TABL E 1 . Impact of Urbanization on Peak Flow Rate (Joint Task
Force 1998)
Storm Ratio of postdevelopment to
predevelopment flow peak
100-year ⫻2
15-year ⫻3
2-year ⫻57
TABL E 2 . Increase in Frequency of Two-Year Peak Runoff Rate due
to Development (Joint Task Force 1998)
Percent impervious Frequency (times/year)
30 (residential) 3
50 (strip comm) 6
80 (industrial) 8
FIG. 3. Typical Channel Erosion due to Urbanization
to the native biota, resulting in loss in biodiversity and often
changes of species.
WATER QUALITY IMPACTS OF URBANIZATION
The principal constituents of concern in urban runoff are
total suspended solids (TSSs), nutrients (P and N), heavy met-
als (Cu, Pb, and Zn) and fecal bacteria (E. coli). The first
comprehensive reporting of typical concentrations of these pa-
rameters in urban runoff was the USEPA National Urban Run-
off Program (NURP) conducted in the late 1970s and early
1980s (USEPA 1983). Since then many investigators have re-
ported results of newer testing; these later measurements show
only slight variations from the USEPA data.
It is the writer’s opinion that, while typical storm water can
have negative impacts on the health of an urban aquatic eco-
system due to the pollutants contained in the runoff, those
effects are generally masked by the negative habitat impacts
caused by uncontrolled runoff. Therefore, urban runoff man-
agement programs should target flow control first. If this is
properly done, the quality issue will for the most part resolve
itself, because the flow management practices that must be
employed will generally result in removal of pollutants from
the runoff.
CONTROLLING FLOW FREQUENCY CURVE WITH
TRADITIONAL DRAINAGE PRACTICE
Design for Flood Control
Urban hydrologists and drainage engineers have long ac-
knowledged the fact that flows increase as a result of devel-
opment. Many municipalities now have ordinances requiring
that larger storms be controlled so the postdevelopment peak
flow for a given return interval storm (rainfall event) does not
exceed the predevelopment peak flow for that same storm. The
state of Florida requires that such control be provided for the
25-year storm. Other places in the United States require that

152 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT / MAY/JUNE 2001
TABL E 4 . Unit Storage Volume to Achieve 90% Capture of Annual
Runoff Volume
City
Rainfall/Runoff
Characteristics
Annual rainfall
(mm) Runoff
coefficient
Storage volume
required for
90% capture of
annual runoff
(mm)
Tucson, Ariz. 295 0.50 9
Butte, Mont. 371 0.44 5
San Francisco 490 0.65 23
Edinburgh, Scotland 700 0.43 12
Chattanooga, Tenn. 750 0.63 15
Detroit 889 0.47 7
Cincinnati 1,013 0.50 12
Orlando, Fla. 1,270 0.35 15
TABL E 3 . Comparison of 25-Year Rainstorm with 90th Percentile
Storm
City 25-year, 24-h rainfall
(mm) 90th percentile 24-h rainfall
(mm)
Cincinnati 74 20
Orlando, Fla. 127 36
FIG. 5. Cumulative Rainfall Distribution for Orlando, Fla., and Cin-
cinnati
FIG. 4. Effect of Peak Shaving on Detention Basin Outflow Hydro-
graph
the 10-year storm is controlled to predevelopment levels; and
a few municipalities require that postdevelopment discharges
not exceed the predevelopment 2-year storm for all storms. To
the writers’ knowledge, no municipality requires that storms
less than the two-year storm be controlled to predevelopment
levels.
To achieve this control, the most common practice is to use
detention basins to peak shave the postdevelopment flow so
that basin outflow does not exceed predevelopment flow for
the design storm. The basic effect of peak shaving on the out-
flow hydrograph is illustrated in Fig. 4. But experience with
these facilities shows that while they reduce downstream
flooding, they are not effective at reducing the erosion in
stream channels for two reasons. The first is due to the pro-
tracted time of flow at the lower rate, as illustrated in Fig. 4.
If the basin outlet flow is large enough to cause stream bank
erosion in the downstream channel, the detention basin actu-
ally subjects the stream channel to the erosive flows for a
longer period of time. This has led some geomorphologists to
suggest that flow attenuation should not be practiced for runoff
peaks except as absolutely required for flood protection down-
stream. But, this would be difficult to administer in a devel-
oping community, and thus is probably not practical.
The second problem with flow attenuation, as it is generally
practiced, is that the outlet works for these basins are designed
only to attenuate the flow of the design storm. Any flow at-
tenuation that occurs for other storms is purely coincidental.
For those storms that have postdevelopment runoff peaks
smaller than the basin outlet capacity, very little flow attenu-
ation takes place. Thus, given the increased frequency of peak
flows in general, and the fact that most detention basins do
not regulate flows smaller than the 10 year predevelopment
flow, most storms will pass through the structure unregulated,
subjecting the downstream channel to erosive velocities on a
more frequent basis. To illustrate this, Fig. 5 shows the cu-
mulative frequency plots of daily rainfall amounts for Orlando,
Fla., and Cincinnati. The curves show that 90% of the annual
rainfall comes in storms smaller than 20 mm in Cincinnati and
36 mm in Orlando. Table 3 compares these volumes to the 25-
year storm for both cities. Table 3 reveals that the 25-year
storm is nearly four times the volume of the 90th percentile
storm for both cities. This is why flood control facilities have
very little effect of the runoff hydrograph of most storms.
Examination by Roesner (1999) of five other U.S. cities plus
Edinburgh, Scotland, with widely varying climatic conditions
confirms that most of the annual rainfall occurs in small
storms. Table 4 identifies the cities and the capture volume
required for 90% control of the annual volume of runoff.
These capture volumes were determined by hydrologic simu-
lations using STORM (Roesner et al. 1974; Storage 1976)
with multiyear, hourly rainfall records as input. The computed
runoff was routed through capture (detention) facilities of var-
ious sizes (volumes), using a constant facility release rate
equal to the storage volume divided by 24, i.e., the drawdown
time for a full basin is 24 h—this criteria is commonly used
for BMP detention facilities intended to settle out suspended
solids in urban runoff. Table 4 shows that a unit storage vol-
ume of roughly 5–15 mm (except for San Francisco, which
requires 23 mm) is sufficient to capture 90% of the annual
runoff in the BMP facility.
Design Storm for BMPs
The recommended design storm for sizing most BMPs is
the storm volume that is just greater than 70–90% of the rain-
storms (Joint Task Force 1998). Using the 90th percentile as
the guideline, the storage volumes shown in Table 4 would be
the BMP design volume for the eight cities listed. Table 5,
taken from Roesner (1999), shows these design storms and
their frequency of exceedance (or overflow frequency).
Swales, infiltration basins, and extended detention facilities
would be sized to accommodate these flows. Because the vol-

JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT / MAY/JUNE 2001 / 153
TABL E 5 . Design Storm for 90 Percent Capture of Runoff
City Overflow frequency
(times/year) Design storm
(return interval)
Tucson, Ariz. 3 4 month
Butte, Mont. 6 2 month
San Francisco 4 3 month
Edinburgh, Scotland 4 3 month
Chattanooga, Tenn. 10 1.2 month
Detroit 12 1 month
Cincinnati 8 1.5 month
Orlando, Fla. 4 3 month
FIG. 6. Effect of Detention Storage and BMPs on Postdevelopment
Peak Flow Frequency Curve
FIG. 7. Effect of Urbanization on Macroinvertebrate Community
(Maxted and Shaver 1997)
umes are small, it is often possible to retrofit existing regional
flood control detention basins with small, low-level outlets
thus providing extended detention basins for treatment of these
small storms.
Effect of BMP Design on Flow FrequencyCurve
The small runoff volume controlled by BMPs means that
for storms larger than the design storm, the BMP will capture
the first flush, thereby attenuating the initial portion of the
runoff hydrograph. But once that volume is exceeded, the re-
mainder of the flow is unregulated until the outflow control
for the drainage detention basin begins to take effect. If the
drainage outlet is designed for control of the 10-year storm,
and the BMP design volume is for a 2-month storm, then ap-
proximately six times per year the flow peak will pass through
the facility essentially unaffected. The magnitude of that peak
will be somewhere between the 2-month postdevelopment
flow peak and the 10-year predevelopment flow peak. The net
effect of the BMPs and drainage detention basins on the flow
frequency curve is illustrated in Fig. 6. As Fig. 6 shows, a
significant portion of the flow frequency curve is still un-
regulated with this design protocol. It is this uncontrolled sec-
tion of the flow frequency curve that the writers believe causes
stream reaches downstream from BMPs to continue to exhibit
habitat degradation and reduced biological indices.
Correcting the Problem
Many investigators in the scientific community have looked
for a relationship between urban development and the ecologic
health (or condition) of the receiving stream. The currently
popular relationship to explore is percent impervious versus
biological assessment indices as illustrated in Fig. 7 taken from
Maxted and Shaver (1997).
Investigators then use these plots to draw conclusions about
impacts of urbanization on ecologic integrity. This type of
analysis has lead to the popular myth that the gross percent
impervious of urban development must be limited to 12–15%
in order to preserve stream biological integrity. But this type
of graph does not take into account the moderating effects that
changes in drainage system design can have on the hydrologic
characteristics of the runoff. For example, the reduction of
directly connected impervious area and inclusion of infiltration
devices—such as infiltration basins that are widely used in
Central Florida—or underdrained sand filtration basins that
are used in Austin, Tex., and San Antonio, will have a signif-
icant effect on the flow frequency curve. By integrating the
design of storage and outlet controls for flood control and
BMPs it is possible to better control wet weather discharges
to receiving waters over the entire range of flows that the
facility experiences.
But it is not effective or realistic to seek one-size-fits-all
prescriptions of land use controls or watershed controls for
ecologic protection of urban streams. This approach fails to
recognize the fundamental causes of channel instability and
aquatic ecosystem degradation. Streams adjust to the water and
sediment supplied to them by widely varied land use scenarios,
each with unique runoff and sediment conveyance character-
istics. Accordingly, stream protection and rehabilitation goals
must be defined and based on general physical principles (Wil-
cock 1997). Managers must ask the following question: Given
the potential flows of sediment and water out of the watershed,
what are our options for achieving desired social and biolog-
ical endpoints?
The writers believe that a two-step approach is required to
link biological indices to watershed parameters:
1. Develop a relationship between stream geomorphologic
characteristics and biologic indicators.
2. Develop a relationship between watershed hydrogeo-
metric parameters and stream geomorphologic param-
eters.
Once this is accomplished, it will be possible to design holistic
runoff control facilities into an urban development so that the
community is afforded both protection from flooding and
maintenance of a sustainable ecosystem in the urban stream.
TOWARD HOLISTIC DESIGN OF URBAN
DRAINAGE SYSTEMS
Having come to the aforementioned conclusion, the United
States currently finds itself in a dilemma with respect to pro-
tecting aquatic environments from the hydrologic changes that
occur as the result of urban development. A basic problem is

154 / JOURNAL OF WATER RESOURCES PLANNING AND MANAGEMENT / MAY/JUNE 2001
that local agencies are responsible for drainage and flood con-
trols; and their mandates, with few exceptions, do not include
protection of aquatic ecosystems. Thus, design criteria for ur-
ban drainage systems are centered around drainage and flood
protection for the urban area with little or no account being
taken of the impacts of that drainage on the ecologic environ-
ment to which that water is discharged.
Congress, in Section 401p of the Clean Water Act, tried to
address the ecologic problem by requiring that pollutants in
urban runoff be removed to the ‘‘maximum extent practica-
ble,’’but as the writers have illustrated in this paper, these best
management practices, which are designed as treatment de-
vices not as flow regulating devices, do not provide the degree
of flow control needed to protect the aquatic environment.
More recent policy proposals for restoring the water quality
of U.S. streams and rivers will also be largely ineffective and
unsustainable in the long term unless policy makers recognize
that biological health is related more to the geomorphic
changes associated with land use conversion than the quality
of the runoff. The Clinton Administration’s Clean Water Ini-
tiative includes proposals for expanding wetland and riparian
area restoration and controls on non–point-source pollution.
The ecologic effectiveness of these efforts hinges on our abil-
ity to proactively manage the cumulative effects of wet
weather flows on geomorphic processes and aquatic ecosys-
tems. As stated in the preceding paragraph, storm-water con-
trols may reduce pollutant delivery from source areas, but they
still allow severe destabilization of streams and destruction of
aquatic life. Riparian areas may be restored or protected, but
eventually erosive processes linked to urbanization or some
other land use change will undermine them. Streams may be
only temporarily ‘‘restored’’ by adjusting channel form to a
reference condition that is assumed to be stable under current
hydrologic conditions. As land uses inevitably change, efforts
such as riparian area and channel restoration will be sustain-
able only if they are outgrowths of a watershed strategy at the
temporal scale of decades for managing fluxes of sediment and
water for channel stability and aquatic ecosystem integrity.
Another issue is that the regulation of altered runoff and
sedimentation processes during land use conversion usually
occurs on an ad hoc basis, without regional planning and con-
sideration of geomorphic processes (McCuen 1989; Urbonas
and Stahre 1993). Storm-water management programs in most
states and municipalities require site-by-site implementation of
BMP facilities in developing areas using common design cri-
teria because approach is the easiest to enforce administra-
tively. However, such an approach does not account for in-
creases in the frequency and duration of erosive forces and
boundary material characteristics in the individual channels re-
ceiving the runoff. Thus, neither the flows that transport the
most sediment, nor the duration of any flow, are maintained
at predevelopment levels (Booth 1990). As a result, the sedi-
ment transport potential of post-development flows may be
several times that of the pre-development conditions, despite
the installation of upstream stormwater management facilities
(see Hollis 1975; McCuen and Moglen 1988; Booth 1990;
MacRae 1991, 1993, 1997). Effective mitigation of channel
erosion hazards and aquatic ecosystem impacts necessitates the
application of geomorphic principles in improving storm-water
controls (Booth 1990; MacRae 1997).
The time is right to bring together the urban hydrology and
environmental effects communities to develop guidance for
improved storm-water management on urban watersheds. This
guidance includes receiving water ecosystem impacts as well
as drainage and flood control. USEPA, state water quality reg-
ulators, and an increasing number local storm drainage agen-
cies have become aware of the importance of flow manage-
ment for aquatic ecosystem protection, but they lack guidelines
for design of such systems. Heaney et al. (1999) recognized
this fact in a recent article on research needs in urban wet
weather flows. But it is extremely important that the research
be pragmatic and have its focus on developing pilot/demon-
stration studies that will lead to design guidance that munici-
palities can use to design new systems, or design improve-
ments to existing systems that will not only protect the health
and welfare of the citizenry that it serves, but will provide
protection of the aquatic ecosystems that receive the wet
weather discharges from these urbanized sites.
REFERENCES
Booth, D. B. (1990). ‘‘Stream channel incision following drainage basin
urbanization.’’ Water Resour. Bull., 26(3), 407–417.
Heaney, J. P., Wright, L., and Sample, D. (1999). ‘‘Research needs in
urban wet weather flows.’’ Water Envir. Res., 71(2), 241–250.
Hollis, G. E. (1975). ‘‘The effect of urbanization on floods of different
recurrence intervals.’’ Water Resour. Res., 11(3), 431–435.
Joint Task Force of the Water Environment Federation andASCE. (1998).
‘‘Urban runoff quality management.’’ ASCE Manual and Rep. on
Engrg. Pract. No. 87, ASCE, Reston, Va.
MacRae, C. R. (1991). ‘‘A procedure for design of storage facilities for
instream erosion control in urban streams.’’ PhD dissertation, Dept. of
Civ. Engrg., University of Ottawa, Canada.
MacRae, C. R. (1993). An alternative design approach for the control of
stream erosion potential in urbanizing watersheds.’’ Proc., 6th Int.
Conf. on Urban Storm Drain.,
MacRae, C. R. (1997). ‘‘Experience from morphological research on Ca-
nadian streams: Is the control of the two-year frequency runoff event
the best basis for stream channel protection?’’ Effects of Watershed
Devel. and Mgmt. of Aquatic Ecosys., L. A. Roesner, ed., ASCE, Res-
ton, Va., 144–162.
Maxted, J., and Shaver, E. (1997). ‘‘Use of retention basins to mitigate
stormwater impacts on aquatic life.’’ Effects of Watershed Devel. and
Mgmt. on Aquatic Ecosys., L. A. Roesner, ed., ASCE, Reston, Va., 494–
512.
McCuen, R. H., and Moglen, G. E. (1988). ‘‘Multicriterion stormwater
management methods.’’ J. Water Resour. Plng. and Mgmt., ASCE,
114(4), 414–431.
McCuen, R. H. (1989). Hydrologic analysis and design, Prentice-Hall,
Englewood Cliffs, N.J.
Roesner, L. A. et al. (1974). ‘‘A model for evaluating runoff-quantityand
quality in metropolitan master planning.’’ Tech. Memo. No. 23, ASCE
Urban Water Resources Research Program, ASCE, New York.
Roesner, L. A. (1999). ‘‘The hydrology of urban runoff quality manage-
ment.’’ Sustaining Urban Water Resour. in the 21st Century, ASCE,
Reston, Va., 229–241.
Schueler, T. R. (1987). Controlling urban runoff: A practical manual for
planning and designing urban BMPs, Metropolitan Washington Coun-
cil of Governments, Washington, D.C.
Schueler, T. R. (1999). ‘‘Keynote speech in proceedings of the conference
on diffuse pollution, Edinburgh, Scotland, August, 1998.’’ Water Sci.
and Technol., 39(12),
Storage, treatment, overflow, runoff model ‘‘STORM.’’ (1976). U.S. Army
Corps of Engineers, Hydrologic Engineering Center, Davis, Calif.
Urbonas, B., and Stahre, P. (1993). Stormwater: Best management prac-
tices and detention for water quality, drainage, and CSO management,
Prentice-Hall, Englewood Cliffs, N.J.
U.S. Environmental Protection Agency (USEPA). (1983). ‘‘USEPA, Re-
sults of the nationwide urban runoff program, final report.’’ NTIS No.
PB84-185545, Washington, D.C.
Wilcock, P. (1997). ‘‘Friction between science and practice: The case of
river restoration.’’ EOS, Trans., 78(41), 454.