Stormwater management in coastal areas

[From the U.S. Government Printing Office, www.gpo.gov]

ATER
ANAGEMENT
IN COASTAL
AREAS-198Z

oe4

665
.M33
1982
STOR

f 11@111

STORMWATER MANAGEMENT IN COASTAL AREAS

Submitted by

Richard H. McCuen
Department of Civil Engineering
University of Maryland
College Park, Maryland 20742

COASTAL ZONE

INFORMATION CENTER

Submitted to;

Tidewater Administration
Department of Natural Resources
Annapolis, Maryland

September 1982

ACKNOWLEDGEMENTS

The authors appreciate the federal funding of this study under a Coastal

Zone Management Program grant from the Office of Coastal Zone Management, NOAA,

to the Tidewater Administration, Maryland Department of Natural Resources. The

efforts of Mr. Earl Bradley, Tidewater Administration, Maryland Department of

Natural Resources, who was contract administrator of this contract are especially

appreciated. Also appreciated are the efforts of the other persons who participated

on a,study review team with Mr. Bradley; Namely, Mr. Earl Shaver and Mr. Robert

Dannecker, Water Resources Administration, Maryland Department of Natural Resources;

Mr. Jeff Hutchins, Tidewater Administration, Maryland Department of Natural

Resources; and Mr. Edward Phillips, Coastal Zone Technical Coordinator for

Worchester and Wicomico Counties.

TABLE OF CONTENTS

Introduction :****'*****,**'"'*,***,**,******,**"*********''**I .... 1
A Literature Review ................................................... I
Stormwater Management Methods .......................................... 2
Stormwater Management Policy ........................................... 2
Test of the SCS Method ................................................. 3
The Use of SCS Methods for Design in Coastal Areas ..................... 4
Downstream Effects of Stormwater Detention Basins ...................... 6
Integrated Stormwater Runoff Design .................................... 6
Recommendations for Hydrologic Design .................................. 7
Channel Stability ...................................................... 9
Agricultural Drainage .................................................. 9
Hydrologic Effects of the Conversion of Forested
to Agricultural Uses ............................................ 13
Study Product I ........................................................ 16
Study Product 2 ........................................................ 18
Study Product 3 ........................................................ 20
Study Product 4 ........................................................ 21
Study Product 5 ........................................................ 22
Study Product 6 ......................................................... 21
Study Product 7 ........................................................ 23
Study Product 8 ........................................................ 24
Study Product 9 ........................................................ 25
Study Product 10 ....................................................... 26
Appendix A: Detention/Retention For the Control of the Quantity and
Quality of Stormwater Runoff: State-of-the-Art
Appendix B: Summary of Stormwater Management Methods and Their
Applicability in Coastal Wetlands
Appendix C: Elements of a Comprehensive Stormwater Management Policy
Appendix D: Evaluation of Methods of Determining Urban Runoff
Curve Numbers
Appendix E: Comparison of Urban Flood Frequency Procedures
Appendix F: Estimating the SCS Peak Rate Factor
Appendix G: An Automatic Fitting Version of the SCS TR-20 Hydrologic
Model
Appendix H: A Planning Method for Evaluating Downstream Effects of
Detention Basins
Appendix I: Integrated Stormwater Management Design
Appendix J: The Design of Vegetative Buffer Strips for Runoff and
Sediment Control
Appendix K: Design of Infiltration Trenches for Control of Stormwater
Runoff
Appendix L: The Effect of Land Use Change and Stormwater Management
on Flow-Duration Curves
Appendix M: Sensitivity of SCS Models to Curve Number Variation

INTRODUCTION

The following paragraphs outline the areas that were investigated for

this study. The specific results are summarized in reports that are included

as appendices. These reports should be read for a description of the methods

used and results obtained.

A LITERATURE REVIEW

one objective of this study was to perform a literature review. A

literature review was undertaken to identify past work on stormwater

management (SWM), with special emphasis on SWM in coastal areas. A detailed

literature review was undertaken and summarized in the following report

(see Appendix A):

"Detention/Retention for the Control of the Quantity and Quality
of Stormwater Runoff: State-of-the-Art" by R.H. McCuen, S.G. Walesh,
and W.J. Rawls

The report is currently being printed by the USDA-SEA-AR. Copies should be

available in the near future.

With respect to the objectives of this project, the literature contains

only a few articles. Research on stormwater management in coastal areas is

almost totally lacking, especially research that discusses the applicability

of existing SWM methods to coastal areas and the interaction between SWM and

agricultural drainage.

Because of the importance of policy and design and because the literature

review revealed very little information, the primary emphasis for the study was

placed on SWM policy and design. Because the SCS methods are widely used in

MD, a major emphasis was placed on the applicability of these methods in

coastal areas.

STORMWATER MANAGEMENT METHODS

There is a legitimate concern about the use of detention storage in low

relief areas such as coastal areas and wetlands. Numerous. summaries that

outline various methods have been compiled. The summary of methods provided

in Chapter 7 of TR-55 is fairly comprehensive. Another summary is provided

in the following report (see Appendix B):

"Summary of Stormwater Management Methods and Their Applicability
in Coastal Wetlands" by R.H. McCuen and S.L. Wong.

It is difficult to ma-ke universal statements about their applicability in

coastal areas. The effectiveness of any method is highly dependent on site

characteristics, as indicated in the report referenced above. It appears that

maximum efficiency can be obtained by selecting several methods that appear

to be effective for a given site and incorporating them into an integrated

design approach. From a hydraulic/hydrologic viewpoint, there is no reason

to reject any one method in coastal areas. Site characteristics, including

soil type, slope, elevation of the water table, and land use, will determine

the effectiveness of any method.

STORMWATER MANAGEMENT POLICY

While analysis of and research on stormwater management design methods is

important, stormwater management programs will not be effective if the design

methods are not a part of a comprehensive stormwater management policy. Inade-

quate stormwater management policies can lead to poor designs. Because most

governing units (i.e., state, county, and local governments ) have had to develop

policy statements on their own, there are wide differences in the extent and

content of most policies. A survey was undertaken as part of this study to

collect SWM policy statements and design standards. An analysis of these

documents identified numerous important elements. The following report out-

lines the aspects of SWM that should be considered in the development of a

new policy or modification of an existing policy (see Appendix C):

"Components of a Model Stormwater Management Policy" by
M.E. Hawley and R.H. McCuen

It is important to state that this report is intended only as a guide in

developing a policy. Because each locality is different, the individual

components should be considered in detail for each governmental unit that is

developing a policy. For example, agricultural areas may warrant a policy

that is different from a policy designed for urban areas. Similarly, a policy

for a coastal area may emphasize components that are not important to non-

coastal areas. A model SWM policy should not be adopted without proper

consideration of each component for the locality for which it is intended.

TEST OF THE SCS METHOD

Because of the widespread use of the SCS TR-55 methods in Maryland. two
aspects of design were tested. Firstly, methods of deriving a watershed runoff
curve number (CN) estimate were compared. While the data base was not solely
for Maryland, most of the wa tersheds were of low relief. The results are
summarized in the following report (see Appendix D):

"Comparison of Methods for Determining Urban Runoff Curve Numbers"
by W.J. Rawls, A. Shalaby and R.H. McCuen.

While the watersheds are urbanized, the results of the study are just as

applicable for agricultural watersheds.

secondly, the TR-55 methods were tested using a national data base.

While urban watersheds were used in the study (because the title of TR-55

specifies urban watersheds), the results should also be applicable to watersheds

with nonurban land uses. The results are summarized in the following

report (see Appendix E):

"Evaluation of the SCS Urban Flood Frequency Procedures" by
R.H. McCuen, W.J. Rawls, and S.L. Wong.

THE USE OF SCS METHODS FOR DESIGN IN COASTAL AREAS

It is current practice to use the SCS hydrologic design methods (TR-20

and TR-55) for design in Maryland. Because of the concern over the unit

hydrograph that is usually used with TR-20 and the tabular method of TR-55,

SCS analyzed hydrologic data from coastal areas in Maryland and Delaware.

The results of the study indicated that a different unit hydrograph should

be used in coastal areas. The commonly used unit hydrograph is characterized

by a peak rate factor (D f) of 484; the SCS stud y indicated that Df should be
about 284. SCS developed the unit hydrograph for a D f of 284; a version of
the SCS TR-55 tabular method was developed for a D f of 284; it is available

from the MD SCS.

If the SCS methods are to be used for hydrologic design, including storm-

water management computations, in coastal areas, then the unit hydrograph is

a very important element. Thus, one objective of this study was to examine

the peak rate factor, D fin more detail. The results of this part of the

investigation are summarized in the following report (see Appendix F):

"Estimating the SCS Peak Rate Factor" by R.H. McCuen and
T.R. Bondelid.

The results indicate that a constant peak rate factor of 284 is reasonable for

the coastal areas but it may be somewhat low. A low value of D f would produce

low estimates of the design peak discharge. However, a D f of 284 is better

than the D of 484 for coastal areas of MD. The paper indicates that more
f

accurate designs can be obtained by developing the time-storage curve for

a watershed and computing the Df from the time-storage curve. While some

additional effort is required, computation of the time-storage curve does not

require data beyond that normally compiled for a hydrologic study.

If storm event rainfall/runoff data are available, then the accuracy

of designs using the SCS methods can be improved by calibrating the unit

hydrograph directly from the rainfall/runoff data. A version of TR-20 that

can be used to calibrate the unit hydrograph was developed as part of this

study. It is described in detail, including a user's manaual, in the following

report (see Appendix G):

"An Automatic-Fitting Version of the SCS TR-20 Hydrologic Model"
by T.R. Bondelid and R.H. McCuen.

A copy of the computer program can be obtained from R. H. McCuen.

The TR-20 program can be modified for use in coastal areas by changing

the unit hydrographs. The ordinates of the unit hydrograph are contained in

the array DIMHYD, which is assigned values by reading the ordinates in

through lines 139-185 of subroutine READIN of TR-20. If one wants to permanently

modify the TR-20 program to include a dimensionless unit hydrograph other than

the standard SCS unit hydrograph, then lines 168-176 of the MAIN program and

lines 28-32 of the BLOCK DATA program need to be changed. The array DHX of

BLOCK DATA contains the standard dimensionless unit hydrograph. These could

be replaced by another dimensionless unit hydrograph, such as the one developed

by SCS for coastal areas (Df = 284).

DOWNSTREAM EFFECTS OF STORMWATER DETENTION BASINS

While stormwater detention basins are widely used for controlling the

hydrologic effects of development, there is some concern that these basins

may actually create downstream problems. Thus, one objective of this study

was to develop a technique that could be easily applied at a site to determine

whether or not a more detailed analysis is necessary. The method is presented

in the following report (see Appendix H):

"A Planning Method for Evaluating Downstream Effects of Detention
Basins" by M.E. Hawley, T.R. Bondelid, and R.H. McCuen.

The method requires very little computational effort; also it is not necessary

to collect additional data to use the method. If the method indicates that

there may be a downstream problem created by the detention basin, then a

comprehensive watershed analysis should be made using TR-20. The effect on

channel stability can be made using conventional hydraulic computations, such

as the critical velocity method.

INTEGRATED STORMWATER RUNOFF DESIGN

There is a legitimate concern about the use of detention storage in coastal

areas. Traditional design methods were often developed using data from non-

coastal areas. If these design methods-are used in coastal areas, they may

require an excessive amount of storage. This is especially critical because

of the land that would be required to satisfy the required storage specified

by the design method. An integrated approach to stormwater control appears

to offer the most efficient design for controlling stormwater runoff. The

following reports provide details on the incorporation of various design

methods into computation of detention storage requirements using TR-55 (see

Appendices I,J, and K):

1. "Integrated Stormwater Management Design" by S.L. Wong and
R.H. McCuen.

2. "The Effects of Vegetated Buffer Strips on Runoff Quantity and
Quality" by S.L. Wong and R.H. McCuen.

3. "Design of Infiltration Trenches for Control of Stormwater
Runoff" by S.L. Wong and R.H. McCuen.

These methods would be especially useful in coastal areas. Buffer strips would

be effective for both quantity and quality control; they are especially useful

for limiting sediment inflow into streams. If properly maintained, they can

reduce runoff volumes significantly. Even the volume of direct runoff from

a large storm can be reduced by 5 to 10 percent. Such a reduction can be

used to decrease the amount of detention storage required.

RECOMMENDATIONS FOR HYDROLOGIC DESIGN

I recommend the following general policy for hydrologic design:

1. Peak discharges for small drainage areas should be estimated using

the TR-55 methods, the chart and graphical methods. The graphical

method should be used when an accurate estimate of the time of

concentration is available. For low sloped coastal areas, the

graphical method based on a peak rate factor of 284 should be used.

If such an estimate is not available, then the chart method should

be used. When the chart method is used, every attempt should be

made to obtain the data necessary to include the corrections and

adjustments. Computation sheets for these methods are available in

the following text:

McCuen, R.H., A Guideto Hydrolo2ic Analysis Using SCS
Methods, Prentice-Hall, Inc., Englewood Cliffs, NJ, 1982.

2. If the drainage area has a significant channel system and very

accurate estimates of the travel times through the channel reaches

are not available, the USGS state equation should be used to

estimate peak discharges. The USGS equations are given in the

following report:

Walker, P.N., Flow Characteristicsof Maryland Streams,
Rept. of Investigators, No. 16, Maryland Geological Survey,
1971.

3. If a runoff hydrograph is needed for an inlet area, then the TR-55

tabular method should be used. The expected accuracy of the method

will be greatest when the estimates of the times of concentration

and travel times for the channel are accurate. The method is

especially useful forexaminingthe effects of projected watershed

modifications, such as when an area will be cleared of forest land

and put into agricultural uses. For low sloped coastal areas, the

tabular method based on a peak rate factor of 284 should be used.

4. If a runoff hydrograph is needed and the channel system is significant,

the SCS TR-20 program should be used. The TR-20 program should not

be used in place of hydraulic computations, such as for the calculation

of flow through bridge openings; the Corps of Engineers HEC-2

program should be used for such problems.

5. Where pipe systems represent a major component of the runoff

process, a computer model such as INDRA should be used. A Users

Manual for the INDRA model is available:

Donahue, J. and McCuen, R., "User's Manual: Integrated Direct
Runoff Algorithm: INDRA," Technical Report, Dept. of Civil
Engineering, University of Maryland.

CHANNEL STABILITY

The literature includes very little research concerning the interaction

between land use changes and the stability of channels downstream. The method

proposed in Appendix H can be used to evaluate the potential for SWM at a

site to cause downstream flooding problems. Research to date on the effect

of downstream flooding on channel stability is quite theoretical and does not

provide a universally accepted means of assessing the interaction. Site

specific data requirements are very significant, which prevents this research

from being widely used. However, preliminary efforts indicate that SWM does

not totally mitigate the effects of land use conversion, especially when

channel stability is in question. A method is outlined in report of

Appendix L:

"The Effect of Land Use Change and Stormwater Management on Flow-
Duration Curves," by S.L. Wong, R.H. McCuen, and M.E. Hawley.

While estimates made by the method given in the report may not be highly

accurate, there is no reason to believe that the relative effects are

inaccurate. The flow-duration curve is a better tool for analyzing the effects

of land use change and SWM when channel stability is a primary concern.

AGRICULTURAL DRAINAGE

The need for stormwater management can result from the conversion of

forested land to agricultural uses, as well as from urbanization. on the

Eastern Shore of Maryland the majority of changes in land use involve the

establishment of new agricultural areas; there is very little urban development.

This region is characterized by very low slopes and high water tables; most

of the soils are sandy. Conversion of forested areas to agriculturalland

can cause hydrologic problems due to both increases in the volume and

magnitude of peak runoff and increased erosion and sedimentation.

Cutting and clearing forest vegetation in order to develop agricultural

lands will result in increased volume of runoff for some types of storms.

The forest land cover intercepts a large percentage of rainfall, both in the

tree leaf canopy and in the dead leaf layer at the ground surface. When the

forest canopy is removed, this level of interception of rainfall will hot

occur; even if agricultural crops are planted, the amount of storage in the

land cover will generally be less than that of the forest cover. For low-

intensity storms, much (if not all) of this formerly-intercepted water will

infiltrate into the soil, especially in the sandy soils typical of the Eastern

Shore; for longer duration storms, the infiltration capacity may be exceeded

and surface runoff will occur.

Because of the generally high water tables and high permeability

rates found on the Eastern Shore, even moderate storms are likely to raise

the water table to the level of the ground surface in some areas. Usually

the low areas in agricultural fields are likely to be covered with standing

water after rainstorms. This is sometimes due to surface runoff, but it

can Also be caused by a rise in the water table.

Standing water on the surface and saturation of the root zone of the

soil can be harmful to agricultural crops. The usual approach to solving

these problems is to excavate artificial drainage channels that will remove

the excess water quickly. While this method will generally solve the problem

of increased volumes of water caused by conversion of forests to agricultural

land, these artificial drainage canals can adversely affect water quality.

Therefore, these drainage channels should be designed with the objectives of

solving both the quality and quantity problems.

The most important aspect of water quality in agricultural areas is

sediment. other pollutants, such as herbicides, pesticides, and fertilizers,

tend to be absorbed onto sediment particles; therefore, if the flow of sediment

through the drainage system is controlled, the movement of these other

pollutants can also be controlled to some extent.

There are two potential sources of sediment in these drainage channels.

First, erosion may result from the impact of rainfall and the resulting over-

land flow; thus, sediment may be washed into the channel. Secondly, sediment

may be entrained in the channel flow if the flow velocity becomes high enough

to scour the bed and sides of the channel. Both of these potential sources

of sediment can be controlled if proper design procedures are used in planning

the artificial drainage network. Also, provisions can be made for removing

sediment before it enters the channel.

The majority of the damage caused by sediment occurs when the sediment-

laden water flows out of the artificial drainage channel into tidal waters.

Because the velocity of the flow drops as the water leaves the channel and

enters the wetlands, the sediment tends to be deposited at the entrance of

the wetland that is adjacent to the end of the drainage channel. This excess

sediment can bury many types of organisms and the increased turbidity can

interfere with photosynthesis by submerged aquatic vegetation. Overall, the

effect is a reduction in the biological activity in the wetlands, which can

lead to lower productivity of fish and other food organisms. Therefore, it is

imperative to prevent the sediment from being carried into the wetland areas

or tidal waters into which the agricultural drainage channels usually drain.

Perhaps the easiest way to prevent sediment from entering the drainage

channels in surface runoff is to use vegetative buffer strips along the sides

of the channels. Design methods for these buffer strips are available (See

Appendix J);the strips should be designed for high trap efficiency in serve storms,

because the detrimental effects of overland flow are not likely to be significant
during small storms because of the high infiltration capacity of the soils.

Alternatives to vegetative buffers would be straw bales and/or filter cloth

barriers; either of these methods could be used until a buffer strip could be

developed. The other alternative would be to leave the forest cover standing

along the sides of the drainage channels.

The second potential source of sediment is erosion of the bed and sides of

the drainage channel. This will only occur if the velocity of the stream flow

exceeds the critical velocity, which is a function of the material lining the

channel. Channels can be designed so that scour velocities are only exceeded

in extreme storms, and it is usually possible to determine the critical scour

velocity from tables published in engineering handbooks. It should be noted

that if sediment is allowed to enter the channels from surface runoff, it is

likely to be deposited on the beds of the channels somewhere downstream.

Therefore, it might be best to assume the critical scour velocity to be the

velocity required to entrain this loose sediment, rather than the velocity

required to erode the channel itself.
As an alternative to limiting the velocity of flow in the channels, the
critical scour velocity of the bed can be raised by lining the channels at
critical points with gabion mattresses, filter cloth, or large particles. It
must be noted, though, that this will not prevent sediment that enters with overland
flow from being carried downstream. Also, these methods of armoring the beds

and sides of channels are likely to be expensive.

The final method of preventing sediment from damaging the tidal discharge

areas is to design the channel so that there is a sedimentation basin at some

point upstream from the tidal outlet. This basin can merely be an area where

the channel is considerably wider and deeper, although if the water table is

near the ground surface, deepening the channel may not help. The idea is to

reduce the velocity of the water so that much of the sediment will settle out.

Because a major part of the problem is likely to be caused by suspended

sediment, settling may take a considerable amount of time. Also, a sediment

trap of this type will require considerable maintenance in order to prevent

large storms from reentraining the sediment that has previously settled out.

HYDROLOGIC EFFECTS OF THE CONVERSION OF
FORESTED LAND TO AGRICULTURAL USES

Given the value of land that is placed into agricultural use, there are

economic incentives for converting forested lands to agricultural uses. Such

land use conversion, however, has significant hydrologic impacts. Forested

land has a very low runoff potential because of the significant amounts of

storage. The canopy intercepts significant amounts of rainfall; this amount

is especially significant for the more frequent events. The soil below the

canopy usually has a relatively high infiltration capacity; thus, the direct

runoff volumes and rates are relatively low, and the degree of groundwater

recharge is relatively high. While the infiltration and interception potential

of a forested area varies with soil type and the condition of the forest

canopy, forest lands have a lower runoff potential (i.e., lower SCS runoff

curve numbers) than most other land covers.

The conversion of forested land to agricultural use causes very significant

changes in the runoff potential of a land parcel. Agricultural lands have

significantly less canopy interception; just prior to planting,a canopy may

not exist. The loss of interception storage and the removal of camopy litter

reduces the potential for water storage; thus, the runoff potential increases

significantly. The SCS runoff curve numbers for agricultural land uses are

much higher than those of forested lands.

To examine the hydrologic impact of land use conversion, the SCS Chart

method was used. The peak discharges for various forested conditions were

computed; specifically, forested land in good, fair, and poor condition with

types A, B, and C soils were used as the base conditions. Agricultural use

was represented by two states: (1) row crops/straight row/poor hydrologic

condition and (2) small grain/contoured/good hydrologic condition; these

two land covers were selected becasue they represent relatively high and

relatively low runoff potentials, respectively. The peak discharges were

computed for each condition and for soil types A, B, and C. The results are

shown graphically in Figures 1-4 for soil types B and C. These four figures

indicate that land cover conversion from forested land to agricultural uses

have a very significant effect on the peak discharge. An increase in peak

discharge of 100 percent in not uncommon, even for cases where only one-third

of the drainage area is converted. For cases where the entire watershed is

converted., the change in peak discharge may be as high as 600 percent. The

changes for type A soils are not shown because the increases were as much as

5,000 percent, with changes of 1,000 percent not uncommon for type A soils.

These extreme changes for type A soils result because of the very low runoff

volumes for forested areas on type A soils.

The results of this simple analysis suggests that state and local

governments should give serious consideration to the hydrologic impacts of

land cover conversions. While stormwater management may be effective for

controlling the impact of some land cover conversions, there is reason to be

concerned about the conversion of forested land in areas where type A and B

soils are predominant.

1. Forest/good condition, type C soil
2. it of if type B soil
3. Forest/fair if 11 C it
350 4. 11 it it B
5. poor it it C
6. B

300

250

200

Cd 150
0

U 100

Cd
4J
r-

(0
50

0 25 so 75 100

Percentage of Forested Land Cover Converted to Small grain/
Contoured/Good Condition for a Low Sloped Watershed of 20
acres for a 10-year Recurrence Interval Storm Event

FIGURE 1

1. Forest/good condition, type C soil
2. it ti it type B soil
3. Forest/fair it 11 C
4. of it B
S. it poor C
6. B

600

Soo

400
cu

U
M

Cd 300

bO
r-
Cd
200

4J
r-
0
U

100

0 25 so 75 100

Percentage of Forested Land Cover Converted to Row Crops/
Straight Row/Poor Condition for a Low Sloped Watershed of
20 acres for a 10-year Recurrence Interval Storm Event.

2
FIGURE

1. Forest/good condition, type C soil
1350 2. It if it type B soil
3. Forest/fair it It C 1, 2-
4. if if it B It
5. it poor C
6. B

300

250

200

@4
Cd

tn
-4
IM
14 150
Cd
4)

Cd
100

tz
Cd

U
@4
(D
so 3)(0

0 25 50 75 100

Percentage of Forested Land Cover Converted to Small
Grain/Contoured/Good Condition for Low Sloped Watershed
of 50 acres for a 50-year Recurrence Interval Storm Event.

FIGURE 3

1. Forest/good condition, type C soil
2. 11 it If type B soil
3. Forest/fair if 11 C if
4. if it It it .B It
S. poor It of C
6. B

600

500
W
$-4
Cd

U
M

Cd
a) 400

bwo

Cd
300

Cd
4-)

200

100

0 25 so 75 100

Percentage of Forested Land Cover Converted to Row
Crops/Straight Row/ Poor Condition for a Low Sloped
Watershed of 50 acres for a 50-year Recurrence Interval
Storm Event
FIGURE 4

RESPONSES TO PART'ICULAR ISSUES
THAT WERE TO BE ADDRESSED AS PART OF THIS STUDY

Study Product 1: Identification of conditions under which stormwater runoff

has a significant effect in coastal areas including a determination of what

constitutes a significant effect.

What constitutes a significant effect is a legal question, rather than a

hydrologic consideration. From a hydrologic standpoint, any change must be

considered as'significant; however, at any given site, a large change may not

create hydrologic/hydraulic problems. But as a general policy, I believe that

conditions in the drainageways, streams, rivers, etc. should not be changed

above that which occurs naturally. This is especially important because neither

theoretical or empirical studies have adequately addressed the effects of changes.

The theoretical bases for the physical systems are not that well understood and

empirical studies have been too limited and inconclusive. While a "no in-stream

change" policy may appear too restrictive, it is not. Stormwater management

control methods have shown to be effective and are continuing to improve; for

example, the Soil Conservation Service has recently developed a two-step riser

design method. This method would enable stormwater detention basins to be easily

designed to control more than just a single point on the frequency curve. The

integrated stormwater management design method proposed as part of this study

offers the means for using various stormwater management methods as part of a

development plan; the use of alternative methods can reduce the need for detention

storage requirements. The modification of design methods such as TR-20 for

coastal areas should increase the accuracy of designs.

Those involved in establishing development and stormwater management

policies for coastal areas must give consideration 'not just to what constitutes

a significant effect, but the criterion variables that effects should be

measured on. Flood flow peak discharge is the most commonly used criterion

variable. For example, a policy may specify that the peak discharge for a

selected excedence probability should not be increased above that which

existed prior to the land cover conversion. The use of a single criterion

variable may be inadequate. The duration of peak flows is another important

criterion variable. The effect of land cover changes on the flow-duration

curve of a watershed should be considered in design and addressed in

stormwater management policies. While there is concern over sediment

loadings, as well as other water quality parameters, it is more difficult

to address these parameters because there is little basis for making

accurate evaluations of the effect of land cover changes on these parameters.

While methods exist for making estimates, it is generally recognized that

there is little basis for assessing the accuracy of the estimates in waterways

for which extensive testing results are not available. The problem is compounded

because there is little data available to indicate exactly what constitutes

the "natural" water quality state of all waterways. Thus, measuring changes

is difficult and unreliable.

A criterion variable is a quantity that varies from site to site and is used
as a criterion for comparison; in the discussion, criterion variables are
quantities that are used to judge the effectiveness of stormwater management
policies and design methods.

Study_Product 2: Analysis of the conditions under which stormwater management

methods, both structural and nonstructural, are likely to be effective in

coastal areas.

The fact that a watershed lies in a coastal area is not sufficient reason

to assume that the effectiveness of a stormwater management method will be

different from that for a noncoastal area. The same hydrologic and hydraulic

principles apply; however, the relative importance of the input variables

may differ somewhat. The effectiveness of stormwater management depends more

on the comprehensiveness of the policy and the adequacy of the design method.

This is not to imply that site conditions are unimportant. The slope,

.soil type, and depth to the water table will influence the relative effective-

ness of stormwater management alternatives. If such site variables are not

given adequate consideration, then a particular design may not be effective;

this is the result of poor design rather than the fault of the stormwater

management method. For example, stormwater management methods that rely on

An"ii-itrafibfi, sucii -ds an infiltration hed, will tiot be effective if a site

is characterized by a high water table. This is not a fault of the method

but the incompatibility of the method for the site conditions.

A major problem in the application of stormwater management in

coastal areas is the failure to properly account for the natural storage

in the watershed. Coastal areas with low slopes and sandy soils have the

potential for large volumes of natural storage, including noncontributing

areas. Failure to account for this storage will lead to overestimation of

the required volume of storage through stormwater management. Stormwater

management design methods are based on the computation of the volume of

runoff and the peak discharge. The time of concentration is an important

parameter in estimating the peak discharge, and, thus, the volume of storm-

water detention required. If proper care is taken in computing the time of

concentration and routing of flows through a watershed, the computed

peak discharge and detention volume will be more realistic. Both the peak

discharge and the volume of detention may be over-estimated if the time

of concentration is underestimated because the natural storage in the

watershed is not properly accounted for.

2 ()

Study Product 3: Analysis of the modifications needed to make TR-20 and TR-55

applicable in coastal areas.

On the basis of their conceptual framework, there is no reason not to use

TR-20 and TR-55 in coastal areas. However, because of the characteristics of

coastal watersheds several factors should be given special consideration.

First, the problem of natural storage discussed above should be accounted for.

Specifically, the time of concentration for overland flow paths should account

for depressions. Travel times in channels should reflect channel storage,

especially through man-made constrictions.

Second, empirical evidence, as well as conceptual rationality, indicates

that the unit hydrograph in coastal areas has different distributional

characteristics than unit hydrographs for non-coastal areas. The TR-20

program uses a peak rate factor to control the distributional characteristics

of the unit hydrographs. While a value of 484 is recommended for noncoastal

areas, the analysis of actual runoff hydrographs indicates a significantly

lower value should be used in coastal areas. While SCS recommends a value

of 284, this value is based on analyses using the HEC-1 program rather than

TR-20. Analyses using TR-20 suggest a value nearer to 350 is more appropriate

for the coastal areas of Maryland. A modified version of the TR-55 tabular

method is available for the peak rate factor of 284.

21.
Stua Products 4 and 6: Analysis of the relative importance of variables

that affect both stormwater runoff characteristics and the efficiency of

stormwater management methods, including a sensitivity analysis of runoff

and design characteristics to potential factors.

The dominant factors in predicting the magnitude and frequency of floods

are the drainage area, the slope, the land cover and condition, precipitation

characteristics, and soil type. Land cover and condition, and the soil type can

be represented by the runoff curve number (CN). The time of concentration (tc) can be

computed from the slope, the watershed length (which can be computed from the

watershed area), and the land cover and condition. The results in Appendix M

indicate the importance of error in estimating either tc or CN on both the

runoff volume and the peak discharge.

The relative importance and sensitivity of design factors of stormwater

management methods will depend on the policy used. A policy that requires

control of frequent events (e.g., the 2-year event) will result in SWM methods

that are relatively more sensitive to the characteristics of the inflow

hydrograph than the design characteristics of the SWM method. For SWM methods

that are designed to control the larger events, the design characteristics

are less sensitive to the characteristics of the inflow.hydrograph.

For a stormwater detention basin the magnitude of the outflow will be

especially sensitive to the characteristics of the outlet, specifically,

the diameter and height of the riser. The timing characteristics of the

outflow will be more sensitive to the volume of storage and the distribution

of inflow. Detention basins with two stage risers will not show the extreme

sensitivity of design characteristics that characterizes basins with 6ne stage

risers. The use of the two stage riser will result in outflow hydrographs

that more closely approximate the hydrographs of runoff prior to development.

StudX Product 5: A determination as to whether stormwater management in

coastal areas should address the 100-year event, the 10-year event, the

2-year event, or a combination.

Past research has shown that a SWM policy that specifies only a single

storm frequency as the probabilistic basis of design will not control

flood magnitudes for other storm frequencies. Therefore, design policies

should specify two or more storm frequencies that should be controlled. I

would suggest the 2-year and 100-year events. The coastal areas of Maryland

are characterized by flood frequency curves with large positive skew.

Thus, if the 100-year event is not controlled, stormwater management will

be very ineffective during the larger events. Control of the smaller

events is very important with respect to limiting channel degradation.

For unstable channels, which may result from modification for drainage

of agricultural lands, it is important to control the smaller events.
Thus, a "two frequency" policy is highly recommended.

It is not unreasonable to require the control of floods of two

frequencies, The Soil Conservation Service has just recently completed a

publication that outlines the design of two-stage control structures.

The method is easy to apply and should not increase the cost or complexity

of design. It will, however, improve the effectiveness of stormwater

management.

Study Product 7: In conditions under which stormwater management methods

are not practical, development of a methodology for determining the level of

development that can occur without having an adverse hydrologic impact.

It is difficult to define what is meant by "adverse". The problem is

compounded because many criteria can be used to assess the hydrologic impact.

For example, peak discharge, volume of runoff, sediment production, bed

load transport, or groundwater recharge rates. Even when considering peak

discharge, one must be concerned with changes in velocities and the duration

of the discharge. Recognizing that it is difficult to define what is meant

by "adverse hydrologic impact", I would follow the policy that if stormwater

management methods are not practical, then development should not be permitted.

The actual hydrologic impact of any proposed development could be assessed

using a hydrologic model such as TR-20. However, if development is permitted

where SWM is not parctical, then a precedent is set.

Study Product 8: Development of a methodology for assessing the interaction

between agricultural drainage practices and stormwater management measures,

and identification of the conditions under which such an interaction is

likely to have an adverse environmental impact.

If the stormwater management measures are properly designed and located,

they should not interact improperly with agricultural drainage practices.

However, improper maintenance may result in ineffective stormwater management.

Thus, the section of SWM policies concerned with maintenance must specifically

state the maintenance requirements and provide for inspection and enforcement.

Without such force SWM measures may be ineffective in controlling runoff
from agricultural areas. Certainly, SWM methods will not function as intended

by the designer if they are not properly maintained.

The interaction between agricultural drainage practices and SWM may

be assessed using hydraulic and hydrologic methods commonly used. Where
sedimentation and erosion are important elements, there is not widely accepted;

procedure this is especially true for evaluating watershed erosion and sedimenta-

tion on a storm event basis.

Study Product 9: Development of a methodology for determining whether

there isaneed for stormwater management when a forested area is cleared

and changed to cropland or when cropland is increased through agricultural

drainage and identification of the conditions under which such changes

are likely to have a significant effect on stormwater runoff.

Again, the determination what constitutes significant effect is not

possible without the specification of the criteria to be used for such an

assessment. However, the SCS hydrologic methods, TR-20 and TR-S5,can be

used to assess the hydrologic impact of changes in land cover, such as a

change from forested land use to agricultural uses. The effect of changes

in the agricultural drainage can be evaluated using the TR-20 model to

estimate the surface runoff hydrogriLphs and the Corps of Engineers HEC-2

model to evaluate the water surface profiles in the channels. The TR-20

RESVOR subroutine can be used to evaluate the effect of a stormwater

detention basin.

As noted in the discussion on pages 13-15 there is reason to be concerned
about the impacts that are likely to occur from the conversion of forested
land in areas where type A and B soils are predominant.

Study Product 10: Development of a methodology for assessing the impact of

stormwater management on water quality loadings in coastal areas and analysis

of whether it is more effective in coastal areas to modify land use practices

or install runoff controls to address water quality problems associated with

stormwater runoff.

In recent years, stormwater management has evolved into a discipline

that considers not only the dissipation, retention and/or elimination of peak

wet weather runoffs, but also the impact that these runoffs have on the quality

of receiving bodies of water. The growing awareness of the magnitude of the

impact that nonpoint source pollutants can have has led to an enormous activity

in trying to better understand what happens when precipitation falls upon the

earth. Sediment transport caused by rainfall induced erosion has been of con-

cern to those involved in stormwater management for years. Sediment flow into

receiving bodies of water has for years been recognized as the largest pollutant

in terms of quantity per year that is discharged into these bodies of water.

However, it has only been in the last ten or so years that people have been

concerned about what these soil particles actually consisted of in terms of

specific pollutants.

Sediment flow into bodies of water was initially of concern for two reasons:

the sediment could settle to the bottom thereby accumulating and filling in the

body of water; and, the sediment could become dispersed in the body of water

and inhibit the passage of sunlight thereby inhibiting photo'synthetig ac-

tivity of aquatic plants. As knowledge about sediment.-pollutant interaction

increased, concern was also expressed for the effects that the sediment assoc-

iated pollutants had on water quality. Sediments were no longer considered

as inert solids with only their physical size and shape of concern. It was

recognized that many pollutants could be associated with sediments. These

include various Organic substances which can lead to a depletion of

dissolved oxygen in bodies of water when these organics are oxidized by

microorganisms; plant nutrients such as nitrogen and phosphorus compounds;
Yarious heavy metals; coliform bacteria; and, a number of pesticides, herbi-

cides, and fungicides.

Nonpoint Source Modeling. Historically, the study of nonpoint source pollu-

tants has been conducted using extensive sampling programs. However, because

of the varying nature of nonpoint sources, the cost required to accurately

determine annual nonpoint source pollutant loadings has been typically beyond

the resources available to most planning agencies. Where data were available,

people often developed various pollutant loading factors whicb could hopefully

be related to some particular land use for various hydrologic events (e.g.,

kilograms of phosphorus per hectare of farm land). However, it was soon

realized that not only was it sometimes difficult to properly assess land use,

often differences in soil type, differences in application of chemicals to

agricultural land, and Qther differences precluded this method giving reliable

values of nonpoint source pollutants for areas other than that for which the

loading factor was developed.

A great deal of work has recently been undertaken to combine water

quality aspects of surface runoff with various storm runoff predictive models.

A number of predictive models have been developed and reported in the literature.

They range from complex computer-based models of rainfall/washoff to simple

statistical relationships between runoff and areal pollutant yield rates.

These mathematical models have been used to assess nonpoint source pollution

and evaluate various Best Management Practices. Some of the most widely used

models include:

1) the Pesticide Runoff Transport (PRT) model to estimate runoff,

erosion, and pesticide losses from field areas (1).

2) the Agricultural Runoff Model (ARM) to estimate runoff, erosion,

and pesticide and plant nutrient losses from field areas (2).

3) the Agricultural Chemical Transport Model (ACINO) to estimate losses

from field or basin size areas (3).

4) the ANSWERS models to estimate runoff and erosion and sedimentation

from basin sized areas (4).

5) CREAMS, a model for estimating chemicals, runoff, and erosion from

field-size agricultural lands (5).

6) the Hydrologic Simulation Program - Fortran (HPSF) model used to

simulate watershed hydrology and water quality; this model uses various

pollutant -runoff models such as the ARM model and the Nonpoint Source

(NPS) model (6) as inputs.

7) SPNN, designed to predict the export of sediment, phosphorus, and

nitrogen from agricultural basins during individual storm events (7).

8) CMRA, a chemical migration and risk assessment model for pesticides (8).

9) several statistical correlation studies; an example of which is one

by Zison (9) in which sediment-pollutant relationships in runoff from

selected agricultural, suburban, and urban watersheds are developed.

These models and studies are all a step forward in the quest to truly

understand nonpoint source pollution. However, their application in even a

planning function is often rather limited. All of the models except CREAMS

-require data for both calibration and verifteation. CREAMS requires various

parameter and coefficient estimations.

The basic problem with the models is that they are attempts to describe

phenomena that are not well understood. Pollutants are transported by storm

runoff in'two ways: as a soluble fraction in the runoff; or, as something

attached in some way to a soil particle. Various control practices can be

applied to various land uses (e.g., Best Management Practices) to reduce

both the chemicals applied to land and the potential for stormwater runoff

and erosion. However, there will usually still be some quantity of surface

runoff to either retain or perform some treatment operation if the adverse
effects of nonpoint source pollution are to be eliminated or diminished. Even

being able to predict how much nonpoint source pollution exists by the use

of various models will not always solve the problem of what to do about it.

Sediment Capture. One apparently rational nonpoint source pollution control

method might be to capture the sediment in surface runoff. By thus preventing

this sediment from reaching various bodies of water, nonpoint source pollution

would be reduced. A stormwater management strategy that required for the re-

moval or detention of stormwater induced sediment would be a step in the right

direction. However, although it is apparent that sediment capture would reduce

nonpoint source pollution due to various sediment-pollutant interactions, the

complexities of these interactions make it difficult to quantify the reduction

of nonpoint sourse pollution due to sediment capture.

Sediment - Pollutant Interaction. The process of pollutant-soil interaction is

quite complex. Soils can retain, modify, decompose, or absorb pollutants. Every

year enormous quantities of organic materials, atmospheric pollution, and other

liquid and solid wastes are deposted and incorporated into soils and safely

decomposed into their basic constituents: carbon dioxide, nitrogen, phosphorus,

and other residues of biological decomposition. Because of a high concentration

of soil bacteria, the decomposition processes are quite intensive and effective,

and represent one of the best natural recycling processes. The processes which

participate in the soil decomposition and removal of pollutants include adsorption,

filtration, ion exchange, biochemical action of soil microorganisms, particle

charge interaction, pH effect, and precipitation. On account of the adsorption

and retention processes of the soil, almost all of the phosphorus, heavy

metals, many pesticides, and organic chemicals remain near the point of

application and move only with eroded soil. The exception is in sandy or

peat soils that have little tendency to adsorb pollutants. The residual

pollution that passes the soil layers plus that elutriated from the soils

enters the groundwater aquifer.

Most sediments fall into one of two broad categories according to

origin:

1) organic matter: usually the products of varying degrees of

decomposition of animal and plant material ranging in size from
-6
colloidal humus (0. 6X10 m diameter) to very large pieces of

material.

2) Mineral particles: sediments ranging in size from clay particles

of colloidal or near colloidal size, through silts and sands to

large bolders.

Colloidal mineral or organic materials are extremely active from an

adsorption standpoint, but larger particles are practically inert. How-

ever, this is sometimes the result when, for example, sand particles become

coated with very active fine organic matter.

These colloids are a concern from a water quality viewpoint; the colloids

may then be divided into two groups: clay and organic. Clay particles are
-6 -6
usually colloidal or near colloidal in size (0.2XlO to 4.0 X10 m) and are

made up of laminated plates or rods of alumina and silica. Because of their

construction, the specific area of particles (ratio of internal and external

surface area to mass) is huge. Their structure is such that each surface

will attract and adsorb cations present in dry surrounding water.

The structure of organic colloids is more complex and more variable

than clays. The nuclei consist of various compounds composed of mostly

carbon and hydrogen, which can be oxidized to inorganic material. So,

unlike clays, they are not conservative substances. However, they are

similar to clays in that the nuclei have a strong negative charge and

attract and adsorb cations. In fact, the cation exchange capacity of

organic colloids far exceeds that of even the most active clays. Hence,

organic colloids will be more active carriers of pollutants.

Adsorbed chemicals are, therefore, usually associated with clay and

organic fractions of sediment. Many studies have shown that phosphorus,

most heavy metals, ammonia, and some pesticides will undergo interaction

with the soil particles and become somewhat immobilized,'remaining near the

layer of application. Rainfall enduced erosion can, therefore, transport

these pollutants away with surface runoff. Nitrate nitrogen is much more

mobile and, if excess is present, will either be carried off in soluble

form in surface runoff or penetrate downward and possibly contaminate the

groundwater. The same phenomena is observed with some pesticides.

Pollutant Capture by Sediment Capture. Although it is known that many

pollutants are attached to sediments, it is difficult to generalize concern-

ing the removal percentage of various pollutants even if the removal percentage

of sediment is known. Information about the specific soil type would have

to be correlated with information concerning the application amount of the

various chemicals, the time from chemical application until rainfall event,

the time between rainfall periods, various control practices, and other factors

to predict pollutant association with sediment without conducting extensive

sampling.

However, a stormwater management strategy that contaiiiod provisions for

sediment capture would result in reducing the nonpoint source pollutant loading to

receiving bodies of water. For most effective removal of sediment related

pollutants, attempts should be made to capture as many of the colloidal

size particles as possible. While this might be somewhat difficult to

accomplish in conventional sedimentation basins, the use of aquatic vege-

tation in sedimentation basins could markedly increase the removal of these

colloided particles.

RecouLmended Polic . At the present time (1982), The state-of-the-art is

not.sufficient to describe the general effects of stormwater management on

water quality. At a specific site, an extensive data collection program can

provide the basis for reliable statements about stormwater management and

water quality. But there is no evidence to indicate that the results are

transferable to adjacant watersheds. While this may cause a legitimate

concern by regulatory agencies, it does not mean that attempts should not

be made to control the water quality effects of land cover changes.

The water quality models mentioned above can at least be used to

measure relative effects of land cover changes. That is, a model could be

used to predict the water quality state for the land cover conditions both

before and after development. While the individual estimates may not be

highly accurate, there is little reason to believe that the relative effect

is not accurate. The difference in the water quality states for the two land

cover conditions could then be used to assess the expected impact of the land

cover conversion. If the change in the water quality state is considered

significant, then a stormwater quality management plan could be required.

For purposes of the evaluation of land cover changes on the Eastern

Shore, I would recommend the use of the CREAMS model. This model, which

was developed by The Agricultural Research Service, does not require

calibration. It is based on existing empirical evidence and the experi-

ences of the Department of Agricultural. It is constantly being modified

to incorporate new data and experiences into the model framework. Futhermore,

the hydrologic component is based on the SCS Methods, which are required

for many project types in Maryland. These advantages are significant and

make it the most practical choice.

REFERENCES

1. Crawford, N.H., and A.S. Donigan, Jr., "Pesticide Transport and Runoff
Model for Agricultural Lands," EPA-660/274-013, Office of Research and
Development, U.S. Environmental Protection Agency, Washington, D.C.
(1973).

2. Donigan, A.S.', Jr., and N.H. Crawford, "Modeling Pesticides and Nutrients
on Agricultural Lands," EPA-600/2-76-043, Office of Research and Develop-
ment, U.S. Environmental Protection Agency, Washington, D.C. (1976).

3. Frere, M.H., C.A. Onstad,, and H.N. Holton, 11ACTMO, An Agricultural Chem-
ical Transport Model," ARS-H-3, Agricultural Research Service, U.S. De-
partment of Agriculture, Washington, D.C. (1975).

4. Beasley, D.B., E.J. Monke, and L.F. Huggins, "ANSWERS, A Model for Water-
shed Planning," Agriculture Expt. Sta. J., Paper No. 7038, Purdue Uni-
versity, West Lafayette, Indiana (1977).

S. Knisel, W.G., ed., "CREAMS: A Field-Scale Model for Estimating Chemicals,
Runoff, and Erosion from Agricultural Management Systems," Conservation
Research Report No. 26, U.S. Department of Agriculture, Washington, D.C.
(1980).

6. Williams, J.R.V 11SPNN, A Model for Predicting Sediment, Phosphorus, and
Nitrogen Yields from Agricultural Basins," Water Resources Bull., 16,
843 (1980).

7. Onishi, Y., S.M. Brown, A.R. Olsen, M.A. Parkhurst, S.E. Wise, and W.H.
Walters, "Methodology for Overland and Instream, Migration and Risk As-
essment of Pesticides," Draft Report, Environmental Research Laboratory,
U.S. Environmental Protection Agency, Athens, Georgia (1979).

8. Zison,, S.W., "Sediment-Pollutant Relationships in Runoff from Selected
Agricultural, Suburban, and Urban Watersheads. A Statistical Correlation
Study.," EPA-600/3-80-022, Environmental Research Laboratory, U.S. En-
vironmental Protection Agency, Athens, Georgia (1980).

APPENDIX A

DETENTION/RETENTION FOR THE CONTROL
OF THE QUANTITY AND QUALITY OF
STORMWATER RUNOFF: STATE-OF-THE-ART

Richard H. McCuen

Stuart G. Walesh

Walter J. Rawls

DETFNTION/RETENTION FOR THE CONTROL OF THE QUA14TITY AND

QUALITY OF STOR%,51ATER RUNOFF: STATE-OF-THE-ART
Richard H. Mc@'uen St;-.:i,:t G. Walesh2, and Walter J. Rawls3

INTRODUCTION

During the 1960's professional journals contained numerous technical

papers concerning the hydrologic effects of urbanization. Many of these

papers showed the increase in peak discharges and volumes, as well as the

effect on unit hydrograph cll=acteristics, that resulted as both the percent

and the degree of channelization aild flood, plain enroachment

increased. Recognizing that urbanization had detrimental flooding impacts

hYdic-logi.sts in the 1970's turned their attention towards reducing 'the amount

of 'Ln-:-ease and managing the increased stormwater runoff. As part of the

environmental movement of the late 1960's and early 1970's the effect of the

urbanization process on water quality received much attention in the

p--)fo@ssional literature. Thus, the professional literature of the 1970's has

c-ontained numerous articles concerning the planning, management, and design of

stormwater control methods, with articles on the management.of both stormwate-

quantity and quality. During the 1970's some state and local governments

,2@,veloped policies aimed at providing order to the elements involved in the

control of stormwater runoff. The stormwater man--ement proces-5, which

includes legislation, policy, planning, design, and management, will coatinue

to evolve during the 1980's.

1Associate Dean, College of Engineering, University of Maryland, College

Park, MD
2Manager-Water Resources Services, Donahue & Associates, Inc., Milwaukee

Division, 600 Larry Court, Waukesha, WI 53186
3Hydrologist, Hydrology Laboratory, USDA-SEA-AR, Beltsville, MD

-Al-

Recognizing nat urbanization has an impact on water quantity and quality

characteristics it may be of interest to examine, at least superficially. the

urbanization process and the intent of stormw -ater management practices.

Watershed development is a process that involves a change in land use that

often involves removing n.itu@-al vegetation, removing or covering topsoil,

grading of the surface, and covering part of the watershed with impervious

naterial. The loss of natural vegetation represents a loss of interception

storage that exi@,ts with natural vegetation. Grading reduces both

infiltration and depression storage. Similarly, the depression storage --)n

impprvious surfaces is less than for pervious land uses, and covering a

watefshed with impervious materials will obviously eliminate infiltratio.-, in

these areas. The losses of natural storage, i.e., interception, subsurface,

and depression, results in the increases in peak discharge ratc@s and runoff

volumes that are apparent in urbanized watersheds. The transition from a

natural to an urbanized state is an especially troublesome period. The

transition period is often characterized by runoff that is especially high in

pollution loadings. Even after the land has been stabilized, urban areas are

r1nown to have higher pollutant loadings than most natural land uses; this

results from the activities that are characteristic of urban watersheds.

Another problem that surfaces as urbanizatL@n increases is the change in time

c'tiaracteristics of runoff; this change is partly responsible for increased

floodiiig problems.

To mitigate Ciese detrimental effects that occur as a watershed

experiences development, a number of different stormwater control alternativ'Is

have been proposed and adopted by state and local governments. While these

measures may differ in content, the underlying intent should be to offset the

-A2-

effects of urbanization on the hydrologic cycle. Specifically, the stormwater

management measures should be designed to minimize hydrologic effects, to

replace teh natural storage that is lost due to development, and to neutralize

the effects of changes In time characteristics. For moist parcels of land

stormwater management methods, if properly designed and managed, should be

able to eliminate the detrimental effects of urbanization.

During the past decade, detention /retention (D/R) facilities have gained

increased popularity as major elements in urban stormwater control systems.

The original motivation for this evolving use of storm-oriented facilities

seems have been cost savings that are sometimes achieved relative to the

traditional conveyance-oriented systems. An apparent second motivation for

the growing use of D/R facilities has been their aesthetic, recreational, and

other values.

Because of growing concern with control of non-point source pollution in

unsewered or separately sewered areas, a potential third reason for.

considering utilization of D/R facilities has arisen: enhancement of the

quality of urban stormwater runoff. However, relatively little performance,

planning, design, and operation and maintenance experience is available and

that which is available along with the research and development studies that

are under way or have been completed, is scattered about varying institutions

and literature.

-A3-

REPORT OBJECTIVES

Because of the diversity of the journals in which technical papers have

been published and the variation in technical backgrounds of those interested

in the stormwater management process, it should be of interest to all involved

in stormwAter management, including hydrologists, engineers, and planners as

well as legislator.3 and managers, to have a literature summary and a

state-of-the-art analysis available. Such a summary should provide a

convenient source for identifying a sampling of the literature that is

pertinent to the specific @.:;pect of stormwater management that i-s of intelcnst

to th-L; iz,,'"vidual. In addition to a state-of-the-art analysis, this report

provides abstracts of sc-.ie articles that are concerned witli stormwater

management and have appeared in professional publications. Table 1 identifics

papers according to selected key words; the references are abstracted in

Attachment 8. Also, a list of other references on stormwater management,

which were not abstracted, are included in Attachment A as another source of

information. While it is not always possible to separate the quantity and

quality aspects of stormwater management, this report provides a separate

state-of-the-art analy!@!s of the literature for the rundff quantity and

quality aspects. Part I is devoted to a summary of quantity aspects and Part

N provides a detailed analysis of the quality

-A4-

L11 ArICaCninent B.

4j Ij 1,4 4J 4-)
(Ij U) r- -4 < 'Tj
:@ () t7) C) 4J -4 0) ::j 'T; 4j
Ck -, -i E C 0 4J @4 (f)
f _-; 0 ILI I @ Ln I )
@4 0 0 0) C, CY) F 4J r-i U U) @4 @J
0) Z -14 -A @: fu -4 M M (D c -H I Q) o
Aut h o r s 4) 0 r-I W (Z C "i -P M a U) _Q @4 C @4
U 0 @D -4 W 'j 0 0 (a @4 M 0 Q) 0
W i:@ C@ a@ li@ W C) U Z U @:) < Z W a4
A! I x x x x X--7-. -
2. Bakcr x x x x
3. Bedient and Amendes x x x x x x x x
4- Varned, and CharaAlis x x x x x x x
'1",@r and retQi:son x x x x x
6. Brandt, Conyers, and Ettinger x x x x x x
7. Butler and Maher x x x x x x x
S. -,'nlabrese x x x x x x x x x x
9. Cordero x x x x x x
10. Curtis and McCuen x x x x x x x
11. Davis, "IcCuen, and KamedUlski x x x x x x x x
12. Day and Ho x x x x x
13. Dendion, Delleur, and Talavage x x x x x x x
14. Diniz (1979) x x x
15. Diniz (1980) x x x x x
16. Fruend and Johnson x x x X. x x x x
11. Gbruck and Urban x x x x x x x
18. Grigg, et al. x x x x x
19. Grigg, Duda, and Morris x x x x x x x x
20. Guy x x x x x x x x x
21. Hawkins, Meloy, and Pavon x x x
22. Henry and Ahern x x x x x x x
23. Jnckson and Ragan x x x
24. Kamedulski and McCuen (a) x x x x x x x
25. NIamedulski and McCuen (b) x x x x
16. Krishnamurthi and Balzer x x x x
-17. Krishna,.-.,.urthi and Lenocker x x x x
18. Lai x x x x x x
29. Lakatos and Wiswell x x x x x
30. Lockwood x x x x x
31. Looper x x x x x x x x
32. Manz x x x x x x x x
33. Marile!-, Bribies-ca, and Mora x x x x x x
34. Ma t r. ra,.-7 x x x x x x
35. McCue;i (1980) x x x x x x x
@'j. McCu2n (1979a) x x x x x x x
37. McCuen (1979b) x x x x x
38. Moodie, Sclioles, and Thompson x x x x x x
39. Myers and Ho x x x x x
Nawrocki and Pietrzak x x x x
+1. Novitski (1978) x x x x x
+2. Novitski (1977) x x x x x
43. Patrick x x x x x x x
+4. Pennell x x x x
6. Peterson, Bubenzer, and Madison x x x x x x x x

-A5-

a@ Cy'% 0% k-n t-n t-n"t-n nLn un 1-n I@n
%0 W-j (7% t-n w " )-- (D ko co -j a,
. . . . . . . . . . . . . .

::E: :E: CO cn W cn o-d
0
r) t) 0 in. C-T' m0 '4 @3 ri
0 0 Q 01 0. (4 (D a. r7 r--.- ct f-t
?-h" " :3 :;@ m " - (D cn
t-h CL 0w " ;j (a 0
"0 ::r go = %-@@ v
lb C6 :@ w0 rr
0 R; @ a. ru 7;- m

0
0 pi M v >
H. rT ).- 0
Z I--
(D o CL CL C)
:3 @-3 0
w :;E: (D

I-h CL Z3

CA 0
CA 0
0 PA

> >4 >4 X X >4 Wat(

>4 Ecor

X Pol

>4 X x >4 >4 11.4 >4 >11 Des

>4 X >4 >4 X X >4 Pla

> Ma n

X >4 >4 >4 X >4 x

>4 >4 X x >4 Dat
x >4 x >4 x .1, Coa
XXX>@x >: Mod

X x >4 X Cas

>4 x X x>4 >4 X x;'k. Urb

x X Agr

X >4 N Non

>1 Sew

Por

TERMINOLOGY

The term 'stormwater management' is a term that has evolved because it is

reCognized that dpvelopment most often increases the potential for damage from

storm runoff, but does not properly mitigate the effects of runoff. A number

of management :-echni.cloes have been used to mitigate the hydrologic effects of

deve'-pment, w@th both structural and non-structural methods

be'LT-Ig used. @)Lructural rr-ichniques are construction related wbile

non-structural include legal and administrative procedures. A structure that

creates a reservoir is the most popular method of controlling runoff. A

number of terms have been applied to these structures, including on-site

ponds, detention basins, retention storage, and stormwater management basins.

The term "detention" implies a relatively short time period for detaining

stormwater runoff while the term "retention" implies a relatively longer time

period; however, it is not always clear how these two differ. "on-site

ponding" refers to storage methods that control runoff on the site where the

runoff is generated; it is a term that can be applied to both detention and

retention storage. The term "stormwater management basin," which is the most

general, can refer to both surface or subsurface storage. It is frequently

used, as it is quite frequently in this report, because it avoids the

ambiguities of the other terms and much of the literature that uses the other

terms is applicable to all types of structural storage, inclu,ling rooftop and

parking lot storage.

The terms "settli-il, efficiency" and "trap effLic@-ncy" and similar terms

also have varied and inconsistent meanings and interpretations in

erosion-sedimentation literiture. Malcolm and New (1975) draw a useful

distinction between these terms. Settling efficiency is "the fraction of

particles of a certain size that will be trapped in the basin under design

-A7-

conditions at peak outflow." Trap efficiency is "the fraction of material

removed from all runoff passing through the reservoir during its life." The

definition of trap efficiency is broadened for use herein by basing it on all

runoff passing through the reservoir during a period of time. Trap efficiency

is expected to significantly exceed settling efficiency for the design

condlLions bec-,use most runoff events will be less severe than the design

event.

-A8-

PART T: WATER QUANTITY

Because of regional variation in hydrologic problems a wide variety of

legislation and policy have evolved during the 1970's, the decade of the

infancy of small-scale stormwatpr management. These laws and policies

produced a wide variety of design methods. The lack of consistency in the

application of design methods has led many, especially the non-engineer, to

ques'_@on the value of stormwater management on small watersheds; this doubt

was further intensified by some ambiguous research results, with some research

suggesting that stormwater management may actually create problems (McCuen,

1979; Malcolm, 1980). Thus, the area of design methods is a topic of specia!

interest, especially as it relates to the comparison of the different design

methods that are available.

In spite of the few questions concerning the value of stormwaLer

mana.gement, its use has increased exponentially because, in part, studies have

almost without exception suggested that there are many economic benefits of

storrrwater management. Case studies have shown that downstream flood damage

may be reduced. Added benefits result because lower peak discharges mean

smaller conveyance systems are required. Benefits that are not easily

translated into economic terms include positive environmental impacts. In

many instakices, the economic factors are influential in decisions concerning

@--:)rmwatter management legislation. For this reason it may be useful to

examine the ID,-crature that specifically addresses the suloject of economic

impacts of stormwater management.

In recent ycars developmental pressures have been very great in coastal

areas. Thus, there is now a special interest in examining the value of

-A9-

stormwater management in coastal areas. Because of the low slope-s and high

water tables that exist in coastal areas there is considerable concern that

traditional structural methods of control may not be applicable. Thus,

coastal area stoctuwater management is a topic that will receive considerable

attention in the 1980's, and a review of existing literature is especially

[email protected].

1hese three important areas of stormwater management, i.e., design

in,@.-Lhods, ecoi-iomics, and coastal area analysis, will be examined in more

detail. Table 1 provides a more complete list of other publications on these

three topics, as well as references to other subject areas.

Design for the Control of Flow Rates

Cordero (1972) presented a simplified method for estimating the volume of

storage and release rate from a detention structure, with both the volume and

release rate being a function of the drainage area. Such methods, while easy

to apply, reflected the state-of-the-art in both stormwater design and policy

at that time. However, as the policies improved, the design methods became

more complex in both the conceptual development and application. Bouthillier

and Peterson (1978) provide a method of computing storage requirements; it

uses a graphical relationship of the maximum allowable outlet rate/maximum

inflow rate vs the required reservoir voL,me/total inflow volume.

As suggested previously, the storage in detention and retention basins is

intended to offse@ the natural storage that has been lost due to development.
0

Therefore, the next generation of dosign mathods involved hydrograph

atinlysis. Wycoff a7-1 Singh (1976) presented a technique that was typical of

the simple hydrograpli model; these raethods reflect the effect of land use

changes on both 'the time and flow rate characteristics of

-AlO-

the hydrograph. This generation of design/planning rnetho'ds did not require

routing, which for some design problems may be considered as a limitation.
Baker (1979) presented a method that recognizes that the rate of outflow

is dependent upon the depth of impounded water; thus, it uses a linear routing

equation to route the flow through the detention basin. The solution

procedure attempts to increase the accuracy of estimates by using an iterative

p.:-ocedure in that rainfalls of various durations ar@, used to determine the.

critical rainfall duration that requires the largest storage volume.
In the design of a detention/retention basi@, the discharge rate is

dependent on the inflow, the outlet structure configuration, and the basin

storage characteristics, all of which are dependent on the discharge rate

itself. Thus, an iterative procedure that allows for this interdependent

relationship should improve the accuracy of estimates of the required v..)Iume

of storage and the flow rate. Bondelid and McCuen (1979) presented a method

that is representative of the latest generation of design methods; this method

has the added features of allowing for multi-stage riser designs and being

computerized.

While everyone is concerned with the accuracy of prediction, it is

difficult to assess the accuracy of any technique because an adequate data

base for comparison does not exist. Thus, at the present time it is only

possible to compare the various design/planning methods using synthetic data.

Donahue, et al. (1981) prov@ded a comparison of various levels of

desi.gn/planning models, with the results indicating a very significant range

of proposed storage volumes and release rates. Thus, in order to provide

consistent estimates it may be prudent to select one method for all designs

-All-

in a region. This policy would result in consistent estimates and eliminate

institutional conflicts that arise from policies that are overly flexible. It

must be recognized, however, that the true design cannot be known and that, if

the method selected is inherently biased in its conceptual development, then

all designs will produce inaccurate answers. The latest generation of models

are probably sufficiently accurate that in the trade-off between desi-n

accura cy and 1--or-,istency in -@stimction, it would be best to identify in

stormwa@@er management policy statements a single procedure to be used for the

design of stormwater management basins.

Stormwater Management in Coastal Areas

As indicated earlierg developmental pressures are currently greater in

coast.n'. areas than most other arens. Because of both environmental and

economic reasons, governmental bodies are especially concerned about the

hydrologic impact of development. Unfortunately, the literature contains very

few publications that provide conclusions that can be transferred to other

areas. Also, most studies concentrate on the problems created by development

but fail to show conclusively that an intensive stormwater management program

will eliminate the problems.

Day and Ho (1978) conducted research to determine the ability of natural

wetlands in south Louisiana to remove nutrients from agricultural runoff.

Grigg, Duda, and Morris (1980) provided an assessment of the magnitude of the

storinwater runoff and flooding problem in North Carolina coastal zone,

&!scribe@ cxisting rn:- -.,emcnt programs, and provided reco-_iaendat ions for

improving stormwater managem@-nt programs. In another case study, Patrick

(1976) examined water quality problems in Louisiana's coastal zone that might

-A12-

be handled using management techniques. Flooding problems in coastal areas

are compounded by tidal surges; Myers and Ho (1980) examined an approach to

tide frequency assessment, using, the Delmarva coast as an example.

Economics of Stormwater Management

Some early studies demonstrated that detention/retention basins could

provide economic benefits, such as a reduce cost for downstream conveyance*

systems (Leach and Kittle, 1966). Cuy (1978) indicates that stormwater

management, especially when well coordinated with natural drainage, may result

in lower initial costs of storm drainage systems; however he points out that.

the cost of maintenance may increase. Reuter and Fox (1976) provided a

methodology for evaluating the economic costs of nonstructural pollution

control techniques; nonstructural alternatives should increase the

effectiveness of structural pollution controls while reducing the costs of

achievment, environmental quality objectives.

Civen that a comprehensive methodology for assessing the economic benefits

and costs of a storm drainage system that includes one or more stormwater

management techniques has not been developed, it is difficult to make

generalized statements about cost. Several studies have provided economic

comparisons of stormwater management alternatives; such comparisons may

provide some help in evaluating the economic impact of stormwater management

alternatives. Henry and Ahern (1976) calibrated formulated for estimating sewer

pipe cost; the results indicated that sewer system costs for subdivisions
were reduced when storage was incorporated. Rawls and McCuen (1978) also

proviided regression equations for estimating the costs of storm sewer systems;

the equations were calibrated from actual projects throughout the United

States. They also provided a simple relationship between detention basin cost

-A13-

and project drainage area, which was based on 34 stormwater detention projects

in the Washington, D.C. area. Bedient and Amandes (1978) compared costs of a

variety of management alternatives to evaluate the relative effectiveness.

Calabrese (1980) developed cost-eff,@@ctiveness data on various management

practices and usej the method for establishing the least cost combination of

managemil-at p--::ctices for a case study in Florida.

Brendt, er al. (1972) made a comprehensive cost study on the Seneca Oreek

watershed ",,ear Washington, D.C. In this study costs were compared to

effe-cciv-eness for many erosion and sediment control systems. Drainage costs

in stream@@ w@--re also asses-cad. The results indicated that sediment damages

from uncontrolled erosion on urban construction sites could potentially each

cost $1,500/acre.

Butler and Maher (1978) also examined the economic impacts for downstream

sites. They concluded that the external costs of upstream development require

that both flooding and water quality aspects of stormwater runoff be managed

basin-wide; site-by-site action will probably not provide adequate control at

a reasonable cost.

Grigg, e.t al. (1976) provided the most comprehensive analysis of the

impacts of urban drainage and flood control projects. The measurement of

tangible benefits is described; however, they concluded that a direct.

objective techniqu-@ for qitantifying intangibles was not available. They

provided guidelines for selecting the proper discount rate and for estimaLin.-

flood damages.

-A14-

PART IT: WATER QUALITY

USE OF D/R IN URBAN STORM WATER MANAGEMENT: HISTORIC OVERVIEW

Tli,,re has been a gradual evolution in the state-of-the-art of storm water

management and also in actual practices in recent years (Poertner, 1974).

That evolution is briefly described here to demonstrate that the time has

arrived to consider the use of D/R for enhancing the quality of storm water

runoff in urban and urbanizing areas.

Impact of Urbanization: Water Quantity and Quality

Urbanization usually increases the volume of storm water runoff while

d,,@reasing the runoff time. The net effect of these two processes usually is

a significant increase in peak runoff rates and stages. Although natural

factors, such as land slope and the underlying soil type, are important in

determining the hydrologic-hydraulic response, land use superimposed on the

soil by man's activities will markedly alter the response.

In addition to increasing the quantity of runoff, often with adverse

consequences, the urbanization process can also increase the variety and

amount of potential pollutants washed from the land surface and eventually

into the receiving waters. This is the result of two factors which tend to

have an additive effect. First, urbanization increases the variety and

availability of potential pollutants. Examples include soil and other

rnat-.-.,ials that are loosened and exposed as a result of demoiition and

construction, the introduction of human litter, careless material storage and

handling, and the use of 61-reet de-icing compounds and sand. Second, and as

already d-'3cussed, urbanization typically inc-eases the volume and rate of

storm water runoff thereby providing a more effective means of transporting

the newly exposed potential pollutants from the land surface into the surface

waters. Additionally, sti-,@,,- cliannel erosion was a principal reason that SWM

be.-an aal is legislated.
-A15-

Controlling the Quantity of Runoff

The state-of-the-art of storm water management has developed to the point

that there are two fundamentally different approaches to controlling the

quantity of storm water runoff. Selected characteristics of these two,

approaches to storm water management are illustrated in Figure 1.

Conveyance-Oriented Approach

The first of the two approaches is the more traditional

$'conveyance-oriented" storm water system. Systems designed in accordaoLe with

this appronch provide for the collection of storm water runoff followed by the

immediate and rapid conveyance of the storm water runoff from the area of

collection to the point of discharge so as to minimize damage nad disruption.

The principal components of conveyance-oriented storm water systems are

culverts, storm sewers, and channels that are supplemented with storm water

inlets and catch basins.

Storage-Oriented Approach

A potentially effective but less common approach to storm water control is

the "storage-oriented" system. The function of this type of system is to

provide for the temporary storage of storm water in or near the site with

subsequent slow release to downstream storm sewers or channels. This approach

minimizes damage and disruption both within and downstream of the site. One

or more D/R facilities are the principal elements in a storage-oriented

s y:., m. These p-incipil elements are often supplemented with cul,!erts, storm

sewers, inlets, and catch b-!_3,*..-I-,.

-A16-

"BEFORE"

DIVIDE

D.EPP, ESS ION

0 V E Rr-LOW
AREA

C R E EK

DISCHARGE
FROM BASIN
A F T E R" "AFTERP
C-ONVEYANCE APPROACH STORr,"-.GE APPROACH

LOCAL STORM
SEWER
0 El
00 FILL romo
00
00 DEVELOPMENT
P, El D Ci
OPEN
E R SPAG r rovtr4
S]Fwtlt

DETENTION
POND

% CONCRETE 7'
C I-I A N N'E L

T
1. Thi, ppron,,.lic

-A17-

Comparion of Features

Principal advantages of the traditional conveyance -oriented approach to

storm water control are: applicability to both existing and newly developing

areas, rapid removal of storm water from the service area, minimal operation

and maintanance requirements and costs, and accepted analysis and design

Principal advantages of the storage-oriented approach are:

possible cost reductions in newly-deve loping urban areas, the prevention of

downstream adverse flooding and pollution associated with storm water runoff,

and potential for multi-purpose uses.

Potential Economic Advantage of the Storage-Oriented Approach

The original motivation for use of the newer storage-oriented approach

over the traditional conveyance-oriented approach apparently was the

realization that the former may offer cost advantages. Poertner (1974) and

Donoue & Associates, Inc. (1978, 1979) present cost comparison data for

several engineering studies, each of which indicates significant reduction in

storm water control costs when D/R is used in lieu of conveyance-oriented

systems particularly when the latter utilize storm sewers and concrete-lined

channels.

A complete comparison of conveyance-oriented and storage-oriented system

must consider other "costs" and "benefits" such as the reduction in available

land with the storage-oriented system and increased land values for areas

contiguous to D/R facilies. Therefore, cost analyses must be conducted on a

case-by-case basis. Nevertheless, there are already enough documented cases

of economic advantage of storage-oriented systems to suggest that D/R

facilities should always be considered for controlling the quantity of storm

water runoff.

-A18-

Recreatin 'Ae6thetic Value of D/R Facilities

D/P facilities are being increasingly planned, designed and used as

multi-purpose developments. In addition to their primary storm water control

function they can provide or be part of a site for recreation activities such

as fishing, boating, tennis, jogging, ski-touring, field sports, and

sle1j!",ig. A i7,-:M planned, desi@nedf and operated D/R facility will also have

aesthetic value for contiguous and nearby residential areas.

Mandacory Facilities

Need for Control

In additica to the obvious erosion and sedimentnt_-;,)n pr3blems often

associated with urbanization, it is becoming increasingly apparent that urban

storm water runoff contributes a significant portion of som2 pollutanL3 to the

surface waters. For example, Colston (1974) compared the quality of urban

runoff to that of secondary municipal sewage treatment effluent on the basis

of weight per unit area per year for Durham, North Carolina. On an annual

basis, the urban runoff contributed 91 percent of the chemical oxygen demand,

89 pe.:cent of the ultimate biochemical oxygen demand, and 99 percent of the

suspended solids. Many Public Law 92-500 "20811 studies also concluded that

urban storm water runoff was a major contributor of pollutants to surface

waters. For example, in southeastern Wisconsin, urban storm water runoff

accounts for 54 percent of the suspended solids, 27 percent of the f*ve-day

b;ochemical. oxygen demand, and 32 percent of the total phosphorus carried from

the land surface to the surface waters (Bauer, 1978).

-A19-

It appears as though controlling the quality of runoff--a-_ least the first

and most significant increment of control--should focus on suspended solids

partly because erosion and sedimentation are problems in urbanizing areas.

Furthermore, many pollutants such as phosphorus, pesticides, heavy metals,

bacteria, and other pollutants are carried by soil particles (e.g., Betz

Environmental Engin-2ers, 1976; Whipple, 1979; and Davis, 1979). Therefore,

successful erosion and sediment control is also likely to lead to significant

control of pliosphorus,*pesticides, heavy metals, bacteria, and other

Potenv.Lal Effectiveness of D/R. Fac''.ities

M-in-..- measures have been suggested for controlling urban area non-point

source pollution in general and erosion-sedimentation in particular. The use

of D/R is one of these measures. The state-of-the-art of the use of D/R

facilities for control of the quality of urban storm water runoff is the

subject of the remainder of this paper. Inasmuch as D/R facilities are

finding increased use for controlling the quantity of runoff be cause of the

sometimes cost advantages and recreation/aesthetic values, they should be

.considered for an important third function: water quality control.

Historic Overview: Summary

The evolution of the use of D/R facilities in urban storm water management

is showa graphically in Figure 2. Beginning with the single use of quantity

control., D/R facilities have evolved so that they now could serve three

compat@ble functions:

1. Quantity control;

2. Recreation, aesthetic, and other supplement3l uses; and

3. Quality control.

-A20-

0001,

4.00

QUANTITY

CONTROL
Q U AN? T I T Y

CONTROL R E C IR E A T11(31 N
QUANTITY AESTH. EE T I C
R'E C R E A T 10 1',11 & OTHER
CONTROL AESTHETIC' SUPPLEMENTAL
& OTHER USES
SUPPLEMENTALI
S QUANTITY

CONTROL

PHASE 1- PHASE 2-J PHASE

7. Historic Development of the Use of D/R Facilities.

-A21-

CATEGORIZING THE AVAILABLE LITERATURE

Ideas, information, and data reported in the literature were for purposes

of this paper, cited and summarized in these five categories:

1. Water quality performance of D/R facilities based on field studies!

2. Warer quality performance Of D/R facilities based on laboratory

studies.

3. Water quality performance of D/R facilities based on computer

modeling studies.

4. Guidelines for planning and designing the water quality features of

D/R facilities.

5. Operation and maintenance aspects of D/R facilities.

Literature findings are summarized in the subsequent sections of this

paper in accordance with the preceding five categories.

WATER QUALITY PERFORMANCE OF D/R FACILITIES BASED ON FIELD STUDIES

SCS Empirical Trap Efficiency Relationship

The SCS (USDA) has developed graphs of percent sediment trapped versus the

ratio of impoundment volume and inflow volume. These trap efficiency curves

are based on a study of sedimentation data for relatively large retention

reservoirs (Brune, 1953 and Geiger 1963, as cited in Chen, 1975)

For detention, as opposed to retention facilities, five and ten percent

reductions in trap efficiency are suggested for incolating suspended sediments

that are, respectively, course and fine textured. These adjustments in the

trap efficiency curves are intended to account for the increased likelihood of

flow directly through a detention facilities as opposed to a retention facility.

-A22-

The trap efficiency curves that D/R facilities, like those finding

increased use in urban storm water management, could capture essentially all

of the sediment that enters it. There are two qualifications on use of the

trap efficiency curves. First, they apply to long-term sediment trapping as

opposed to individual runoff events. Second, the curves are based on sediment

trap data for impoundments that are generally larger than the typical D/R

facility and, therefore. may not actually reflect the behavior of the latter.

Sedimentation Basin - Retention Pond in Montgomery County, MD

Davis (1978) reports on a field monitoring study of the hydrologic and

sediment trap performance of the sedimentation basin-retention pond. During

the study, the mildly sloped, 45 acre tributary watershed was being developed

for a complex of government structures. Watershed soils were reported as

"deep, well-drained, and moderately erodible."

The sedimentation basin-retention pond had an area of 1.3 acres at

the permanent pool level. Eight acre-feet of water and sediment storage

volume--equivalent to 2.1 inches of runoff from the tributary watershed--

were available between the bottom of the basin and the crest of the

non-perforated circular riser of the riser and barrel service spillway. A

four inch diameter port in the vertical riser maintained the normal pool 18

inches below the crest of the riser. This provided up to 2.4 acre-feet of net

available storage--equivalent to 0.64 inches of runoff from the tributary

area-at the outset of a runoff event.

-A23-

Inflow and outflow data for ten storms occurring from J@_ln,_- through August,

1977 were analyzed. Storm durations ranged from 0.83 to 14.25 hours, railifall

volumes varied from 0.08 to 2.2 inches and average intensities ranged from

0.04 to 0.94 inches per hour. Construction and development proceeded in the

tributary watershed during the monitoring as indicated by a general ir:rease

;a impervious pav--ment (from 14 -@j 39 percent of the tribuL3ry area) aad

[email protected] (from 8 to 10 percent) and the decrease iii expos-d soil (from 44 to 3

percenr_).

Settl:Lng efficiency for the ten storms ranged 'from 88.1 to 99.7 percent.

About 92.6 percent of the total. sediment that entered the facility during the

cen storms was trapped. With the exception of the largest storm--a total of

2.2 inches of rain in 3.9 hours--the peak discharge at the outlet was about 9

percent of the peak discharge entering the facility. In summary, the sediment

basin-retention pond was very effective both in trapping sediment and in

attenuating peak flows.

The author emphasizes the importance of the amount of storage available

between the normal pool level and the crest of the service spillway. A hood

fitted to the top of the riser on the riser and barrel service spill@way was

found to be very effective in preventing floating debris, which is especially

likely to be present during construction activities, from escaping from the

pond. The author favors non-perforated risers to perforated risers for

sediment control based on observed performance of both. Perforated risers

permit suspended sediment to pass through the facility and, under soint?

rai,ifall.-runoff situations, rosults in a "negative" trap efficiency, that is,

sediment outflow may exceed sediment inflow, especially during periods of low

flow.

-A24-

According to the author, mitigation of erosion and sedimentation problems

in urbanizing areas requires a two-pronged approach--erosion control on

disturbed areas, and sedimentation control in the form of sediment

basin-retention pond facilities.

Multi-Purpose Retention Facility in Milwaukee

Cherkauer (1977) monitored the quantity and quality of runoff from two

adjacent approximately 2.9 square mile urban watersheds in the Milwaukee area

for a period of two years. The two watersheds were similar in area, drainage

density, total relief, channel gradient, percentage of main channel improved,

and percentage of watershed developed.

The most significant difference between the two watersheds is that one

contained a 48 acre retention facility normally containing 308 ac-ft. of water

and having an average depth (quotient of volume and depth) of 6.5 feet. The

retention facility received runoff from the upstream 52 percent of the

watershed and was multi-purpose in that it served storm water control,

recreational, and aesthetic purposes. Water quality parameters monitored were

chloride, sodium, calcium, magnesium and total dissolved solids, as well as

sediment.

The retention facility had the expected effect of significantly reducing

peak flows at the watershed outlet while significantly extending the time of

hydrograph recession. Chloride, sodium and total dissolved solids monitoring

were used as indicators of the effect of winter road salting which was

practiced in the watershed for de-icing purposes. The monitoring indicated

-A25-

that the retention facility reduced the transport and concentration of

chloride from tb@! watershed during winter runoff events. This was explained

as a combination of a dilution effect on the incoming salt water and the

tendency the denser incoming salt water to move to the bottom of the

retenti-
and foycia& out uL' the retention facility the

overlying stored waLer having a lesser concentration. However, the mean

-.i@.:eaLrations in runoff from the watershed containing the retention facility

were larger. . Therefore, the aplarenL effect of the retention facility was to

rc..!ucc arinual \ia@-i.'LiliLy in salt concentrations.

Calcium and inagnesiWa concenL -,-at ions were monitored to serve as an index

to the behavior of substances transported [email protected] the soil to th,@ surface water

system. The mass transport and concentration of these SUbstances from the

watershed containing the retention facility were found to be significantly

larger--factor of two or more--during runoff events that occurred during late

summer and fall when the quality of baseflow dominated lake quality. The

runoff events flushed the stored baseflow from the retention facility and

increased downstream concentrations. Although calcium and magnesium are of

relatively little practical importince in assessing surface water quality, the

behavior of these soil-derived constituents in a drainage system containing a

retention facility suggests that the retention facility may store soil-derived

pollutants during baseflow periods and cause abrupt and potentially adverse

increases in downstream concentrations duri-,,.,- an ensuing runoff event. An

example of such a pollutant would be certain creosote co Impounds that have

entered the shallow aquifer from a poorly managed creosote operation. in

summary, if the soil and, therefore, the baseflow is polluted, the effect of

that pollutant on the receiving surface waters ma be aggravated if a
y

retention facility is introduced.
-A26-

Unfortunately, the quantity and quality of lake inflow were not determined

coincident with lake overflow characteristics and, therefore, it is not

possible to directly demonstrate the effect of-the facility on water quality.

Also, other water quality parameters of interest such as sediment, dissolved

oxygen, fecal coliform, and phosphorus were not included in the analysis-

Under3i@o etenti-oa Facility in Milwaukee
@nd D

A recent study (Milwaukee and Consoer, Townsend & Associates, 1975)

evaluated an underground detention tank as a method for abatement of pollution

from combined sewer overflow. The tank received most of the combi-ned sawage

from the 570 acre tributary area--relief and int--rceptor sewers captured some

of the combined sewaae before it reached the detention tank and transported it

past the tank.

Because this facility received combined sewage (a combination Of Istorn,

water runoff and sanitary sewage) as opposed to only storm water runoff, the

results of the study are not directly transferrable to storm water runoff. A

study by Dalrymple et al. (1975) suggests that storm water runoff may have

better SLIspended solids settling characteristics than combined sewage.

Therefore, the results of this combined sewage detention stuOy for suspended

solids may provide a conservative estimate of what could be expected with

storm water runoff.

The tank @iad inside dimenE-'Dns of 420 x 75 x 16 feet ind a capacity of 3.9

@)n ga 11 -_is, or 1.1. 97 ac - f t, wh ich 4. s equivalent to 0. 2 5 inches o f r uno f f

Crow. the tributary irea. Combined sewage flowed by gravitv to @_Iae tank and

entered throug
.,h a bor screon with 1.5 inch openings. The tank contain-d

several rotary mixers operated only to resuspend solids for pumping out of the

tank after a rainfall-runoff event.

-A27-

Inflow-outflow data were insufficient to directly determine suspended

solids and biochemical oxygen demand removal efficiencies. 17herefore, a

computer model was developed and calibrated using the inflow-outflow data.

The sedimentation equation, as discussed by Fair and Geyer (1954), was used in

the model to represent suspended solids and biochamical oxygen demand.

Historic hourly rainfall data for three years (wet, normal and dry) were

input to the model to test the effect of V'arl'.ablc@ tank size. The size of the

storage tank had a significant effect on predicted susperled solids and BCD

removal. For example, the modeling indicated @hat incre:-isiag tank From

one million gallons per-square mile (3.06 ac-ft/squa-re mile or 0.058 inches of

runoff) to seven million gallons per square mile (21.42 ac-@7t/square mile or

0.406 inches of runoff) would increase suspended solids removal from about 35

percent to about 90 percent. This is based on the quantity of suspended

solids and BCD reaching the tank--not all of it did since soine was taken from

the drainage area by relief and interceptor sewers. However, interceptor

capacity was set at 8 mgd or 12.4 cfs, which was small compared to peak rianoff

rates during storms.

A model run for 1972 was used to test the effect of pump-out rates on

settling efficiency with the conclusion that settling efficiency was

relatively insensitive to pump-out time for the range of pump-out time

examined. Decreasing the pump-out time from 96 hours to 24 hours for the

facility increased overall suspended solids removal _--fficieacy from about 68

to 72 percent and increased overall BOD removal efficie@-_",- f-rom about 65 to 70

percent.

-A28-

Sediment Ponds in Idaho

Bondurant, Brockway, and Brown (1975) discuss Ilia sediment removal

efficiency and other characteristics of two sediment ponds in Idaho based on

measurements taked period of several years. The first and larger basin

measured 60 x 500 x 5 feet and served an area of 6,000 acres, or 9.4 square

miles. The total volume was 3.44 ac-ft, which is equivalent to only 0.007

inches of runoff from the tributary area. The second basin measured 300 x 40

feet with a depth ranging from 2 to 4 feet and served an area of only 70

acres. The total volume was about 0.8 ac-ft, which is equivalent to 0.15

inches of runoff from the tributary area.

Longitudinal bottom profiles and sediment characteristics (percent clay,

silt, sand and density) at the smaller basin revealed a pattern of more

sediment and more sandy sediment at inlet end. Periodically determined

settling efficiencies varied from 56% to 96% and-were greatest during higher

flow rates because heavier (more-settleable) sediment was carried in. Three

years of data collected for the larger basin showed annual trap efficiencies

of 65%, 68%, and 74%. High efficiencies did not necessarily mean that less

sediment was discharged through either of the ponds.

Ponds in the Woodlands Development, Texas

Davis (1979) monitor low flows and storm water runoff from portions of

the Woodlands, a new urban development near Houston, Texas, for two and

one-half years. The strong area contained two closely spaced artifical lakes

in series having a total drainage area of 820 acres and a total volume of 110

ac-ft. The investigations focused on pathogenic and indicator bacteria and

included some limited laboratory and field studies of the effect of

detention. A few suspended solids and turbidity analyses were also run.

-A29-

Data presented for five rainfall events which occurred in March (2

events), April, September and Oc@-)ber, 1975, indicate large reductions in peak

concentrations of suspended solids and turbidity from the upper reaches of the

upstream lake to a point in the downstream lake. For example, an

approximately four inch rainfall in April produced an upstream peak suspended

solids of 2660 mg/l, whereas, the downstream peak was only 356 mg/1--an 87%

reduction. Similarly, peak turbidity was reduced from 900 to 210 JTU (Jackson

Turbidity Units)--a 77% reduction.

Additional data presented [or t'ie above five rainfall events :,dicated

sLgnificant reductions in pathn@-enic bacteria and in fecal coli-Eorm and fecal

streptococci bacteria. Most reductions in peak bacterial concentrations were

in the rance of one to three orders of magnitude. Peak concenu-ations of

fecal coliform bacteria in the downstream lake were generally reduced to about-.

100 to 1000 colonies per 100 ml. The author hypothesized the reductions in

pathogen and.indicator bacteria were caused by a combination of the processes

of dieoff and settling.

Laboratory settling experiments were conducted with 5 gallon samples of

storm water runoff. One of a pair of the samples was continuously agitated

while the other was not agitated. Samples were periodically drawn from both

samples at a depth of 15 cm and analyzed for indicator and pathogen bacteria.

Although some high reductions--50% or more--were achieved for indicator

bacteria after 30 minutes, the results were inconsistent and, therefore,

inconclusive.

Other conclusions were that indicator and pathogen bacteria both exhibit a

first flush, phenomonen during storm water runoff events and the times of peak

bacterial concentration co-relative closely with the peaking times of

-A30-

solids and turbidity. Finally, fecal coliforms as opposed, for example, to

total coliforms, appear to be the best indicators of pathogens in storm water

runoff.

D/R Basins for Sediment Control at a Maryland Construction Site

Oscanyan (1975) describes the use of two lakes in series (retention

facilities) and two-sediment basins (detention facilities) into a 65 acre

multi-use recreation facility in Montgomery County, Maryland. All four D/R

facilities were intended to serve as sediment traps during the construction

period.

The upstream lake had a tributary area of 37 acres and a surface area of

0.9 acres. The downstream lake had a tributary area of 75 acres, which

included the 37 acres tributary to the upstream lake, and a surface area of

2.0 acres. The larger of the two sediment basins had a tributary area of 15

acres and a surface area of 0.3 acres and the smaller-basin had a tributary

area of 2.5 acres and a surface area of 0.2 acres. Volumes of the lakes and

basins were not given.

A modest sediment inflow-outflow sampling program carried out in the D/R

facilities during construction indicated, according to the author, that the

two lakes and one of the sediment basins removed over 99 percent of the

incoming suspended sediment. However, the second sediment basin trapped only

about 30 percent of the sediment during rainfall-runoff events and had a

negative efficiency during some rainfall-runoff events. Based on the widely

differing sediment trapping performance, the author concluded that factors

enhancing sediment trapping include: (1) a high ratio of D/R surface area to

stored volume of water at the design storage elevation; (2) a long, narrow

shape with flow through the long dimension to prevent short-circuiting; and

(3) minimizing turbulance.

-A31-

Lake in Urban Watershed in Maryland

McCuen (1980) summarizes input-output mass transport quantities and trap

,efficiencies for 11 parameters. The study focused on runoff from a 148 acre

urban watershed in Montgomery County, MD containing a shopping mall, apartment

complexes, highways and townhouses under development. Watershed runoff pass Ied

through a 5.9 acre lake with outflow controlled by a vertical 24" diameter

CMP. The eleven water quality parameters were sampled during "several storm

events" the number and distribution of sampling is not indicated. The

monitoring data were used to develop equations for inflow and outflow
transport as a function of flow rate. Design storms were then used to

calculate inflow and outflow hydrographs for 2, 10 and 100-year recurrence

intervals. The hydrographs were in turn combined with the equations to

compute mass inflow, mass outflow and settling efficiencies for each of" the 11

parameters.

The resulting theoretical settling efficiencies were very high for most

potential pollutants. For example, for design storm recurrence intervals of

2, 10, 100-years and storm durations of 0.5, 1.0, 2.0, and 6.0 hours, settling

efficiencies ranged from about 20-70 percent for orthophosphate, about 75-90

percent for 5 day BOD and over 015 percent for lead. However, the settling

efficiency for total phosphorus was low being between 2 and 22 percent. As

expected, settling efficiency tended to decrease with increasing severity of

the design storm with increasing duration will any recurrence interval.

Artificial Marsh/Pond System in Upton, N.Y.

Small (1976) presents input-output concentration data for over 30

parameters collected over a 13 month period (August, 1975 through August,

1976) for an artificial marsh/pond system. This outdoor system, which is

-A32-

located in Upton, NY, received sanitary sewage that had been gritted,

screened, and aerated. The marsh/pond system was part of an experiment to

develop natural systems for treating sanitary sewage prior to discharging to

the groundwater. Although the marsh/pond system received "weak raw sewage or

the equivalent of secondary or primary treated sewage", the study is of

interest in this paper because of its uniqueness, particularly the use of an

artificial marsh and a pond in series.

The upstream, marsh portion of the system covered 0.2 acres and was

constructed by excavation, lining with an impermeable membrane, backfilling

with 4 to 6 inches of muck, and planting with cattails. Water depth in the

artifical marsh was control, so as not to exceed 2 inches. The downstream

pond portion of the system-was 5 feet deep and was constructed by excavation,

placing a liner, and stocking with carp and other tolerant fish. Because of

the carp, algae were the only vegetation supported in the pond. During the 13

month sampling period, the application rate to the artificial marsh varied

from 50,000 to 100,000 gpd/acre (1.1 to 2.2 gpd/sq. ft.).

The quality of the flow was significantly enhanced by passage through the

marsh system. This may be illustrated by comparing average inflow-outflow

concentrations of selected parameters. For example the average concentration

of suspended solids in the inflow was 353 mg/l and the average concentration
in the outflow was 43 mg/1--a 88 percent reduction in average values. Inflow

concentrations of suspended solids ranged from 50 to 4300 mg/l whereas the

range in effluent concentrations was 14 to 100 mg/l. Similarily, the average

concentration of biochemical oxygen demand was reduced from 170 mg/l to 19

mg/1--an 89 percent reduction, the average concentration of total phosphorus

-A33-

was reduced from 7.2 mg/l to 2.1 mg/l--a 71 percent reduction, and the

geometric mean of fecal coliform bacteria was reduced from 1560 colonies per

100 ml to 50 colonies per 100 ml.

Natural Marsh at Brillion, WI

Inflow to and outflow from a 385 acre natural marsh located at Brillion,

WI and having a 19 square mile drainage area were monitored for 15 months

(Fetter, et al. 1973). The marsh received surface runoff from the primarily
urban watershed and, because of a municipal wastewater treatment plant,

secondary effluent accounted for from 2 to 50 percent of the estimated monthly

flows into the marsh. The average depth of the marsh was about 1.5 feet and

the estimated theoretical detention time, based on mean annual runoff, was 48

days.

A comparison of inflow and outflow concentrations averaged over the study

period for various potential pollutants indicates that significant reductions

occurred. For example, BOD concentrations were reduced 80 percent, coliform

bacteria 86 percent, nitrate 51 percent, COD 44 percent, turbidity 44 percent,

suspended solids 29 percent, total phosphorus 13 percent, and orthophosphate

about 6 percent. A mass balance analysis indicated a net phosphorus reduction

attributed to the marsh of 32 percent for the 15 month study period.

Wayzata Wetland in Minnesota

A study was-conducted on the Wayzata wetland in the Minnehaha Creek

Watershed District (Wenck, 1977). A goal of the study was to determine the

effectiveness of wetlands in removing potential pollutants From incoming

streams thus preventing those substances from entering and impairing the

quality of lakes. The wetland removed 77 percent of the total phosphorus

-A34-

and 94 percent of the total suspended solids that entered during the study

period. Based on a biological assessment, no negative impacts on wildlife or

vegetation were noted.

Natural Marsh at Green Lake, WI

This study (Donchue & Associates, 1978) culminated in a management plan

for Green Lake, a 7346 acre lake in a preglacial valley about 75 miles

northwest of Milwaukee, WI, Green Lake water quality was deteriorating as

indicated primarily by increases in nuisance algae and other aquatic plants.

The observed water quality degradation and the shift toward a autrophic

status were attributed to an increased influx of nutrients.

One of the recommended management measures is investigation of the

feasibility of an artificial marsh on Silver Creek at the point where it

enters the east end of Green Lake, to trap incoming suspended sediment and

phosphorus.

The recommendations were based, in part, on monitoring of the nutrient

trap effect of County Park Marsh through which flow enters the west end of the

lake. Grab samples were taken over a period of one year during both wet and

dry weather periods. During a monitoring period, 71 percent of the total

phosphorus entering the marsh was trapped and prevented from entering the lake.

Artificial Marshes, WI

Studies on several artificial marshes, and a natural marsh at Brillion,

WI, suggest significant seasonal patterns (Spangler, et al. 1977). Primary

and secondary effluent was passed through the artificial marshes composed of

bulrushes and, over an 18 month period approximately one-third of the total

phosphorus in the incoming flow was removed by the marsh.

-A35-

The variable seasonal phosphorus accumulation pattern was "accumulation

over long periods in spring and summer and discharge of phosphorus in sudden

spurts either in autumn or winter. ... accumulation in the artificial marshes

seems to be associated with high rates of biological activity. Discharge

occurs after cold weather kills the flora."

The authors suggest that the water quality enhancement effect of a marsh

could be improved by harvesting vegetation and by pumping water from or

diverting it around the marsh during expected flushing periods.

PERFORMANCE OF D/R FACILITIES BASED ON LABORATORY STUDIES

Runoff from Durham, North Carolina

Colston (1974) studied a 1.67 square mile (1,070 acre) mixed land-use

urban basin Durham, North Carolina. The study area did not have a storm

sewer system--drainage was provided by street gutters, swales, small pipes,

and culverts. Thirty-six storms were sampled. For most pollutants,

concentrations had a standard deviation of 70% to 80%.

Laboratory experiments were conducted on storm water samples using

chemical oxygen demand, suspended solids, and turbidity as water quality

indicators. The objective was to explore the factors influencing the

physical-chemical treatment of storm water runoff. Laboratory sedimentation

tests were conducted in 1,500 ml beakers some without pH adjustment and

coagulants and others with pH adjustment and various combinations of

coagulants.

Fifteen minutes of plain sedimentation under quiescent conditions produced

average chemical oxygen demand, suspended solids, and turbidity removals of,

respectively 61%, 77%, and 53%. Fifteen minutes of plain sedimentation under

quiescent conditions using alum as a coagulant produced higher average chemical

-A36-

oxygen demand, suspended solids and turbidity removals of, respectively, 84%,

97%, and 94%. The author states "within the cho ices of treatment

alternatives, plain sedimentation is a reasonable, relatively inexpensive

alternate to chemical treatment of urban land runoff".

After completing the above jar tests, larger batch scale coagulation,

flocculation, and sedimentation tests were run on stormwater runoff. The

test apparatis is consisted of a 10 foot tall plexiglass cylinder having a 6.25

inch inner diameter. The results of these tests produced depth versus time

relationships for difference percent removal of suspended solids. These

relationships were in turn used to construct curves of percent suspended

,solids removal versus overflow rate. These relationships are for ideal

quiescent settling and "would have to be adjusted depending on the relative

efficiency of a designed sedimentation basin.". Even at an overflow rate of up

to 6,000 gpd/sq. ft., over 90% of suspended solids can be removed by

sedimentation, according to the settling curves presented in the report.

Unfortunately, the above large scale tests were not run as plain

sedimentation, that is, without coagulants or coagulation aids. Based on the

results of the jar tests a smaller maximum allowable overflow rate for a

given desired percent removal would be expected.

Runoff from Toronto

Dalrymple and others (1965) reviewed literature dealing with the settling

properties of sanitary sewage, combined sewage, and storm water runoff and

then performed laboratory suspended solids tests on a few samples of Toronto

storm water -_runoff.

-A37-

Studies reported in the literature were used to develop a particle size

distribution graph for sanitary sewage, combined sewage, and storm water. The

graph suggests that particle sizes in potential storm water runoff are

significantly larger than in combined sewage. This assumes that storm water

runoff will transport all the solids washed off in the street washing

experiment. Thus, all other things being equal, better settling should be

achieved with stormwater runoff than with combined sewage.

Laboratory settling tests were run on storm water runoff collected at an

outfall in Toronto. The sample was thought to be higher in and silt

content relative to most storm water runoff. In one run, one hour of settling

reduced suspended solids by 80% from an initial concentration of 337 mg/l. In

another run, one hour of settling reduced suspended solids by 87% from an

initial concentration of 323 mg/1- Based on their laboratory tests, the

authors suggest that the settling characteristics of storm water runoff might

be improved by storage prior to settling to provide an opportunity for

agglomeration of small particles.

THE WATER UALTTY PERFORMANCE OF D/R FACILITIES

BASED ON COMPUTER MODELING STUDIES

Computer Modeling Studies of Hypothetical D/R Facilities

Curtis and McCuen (1977) address the factors that influence the

effectiveness of a D/R facility in controlling both flow and sediment. The

authors developed a computer model consisting of a soil detachment transport

submodel, a detention/retention reservoir sedimentation submodel, and a

reservoir mouting submodel. All analyses used a 2-year, 6-hour design storm.

Sensitivity studies were run leading to the observations that settling

efficiency increased with increased proportion of larger (heavier) soil

particles, decreased depth of D/R, decreased initial volume of water in D/R,

-A38-

and decrc-...,sed size of orifices in outlet control _,-,ructure (a perforated riser

pipe). ieak outflow was (]-ccreased by all those f,irtors that increased

settling efficiency (except for particle size distribution). Because of timing

effects, it is possible that multiple D/R in a watershed could increase flow

at the watershed outlet.

Kamedulski and McCuen (1979) used a calibrated hydrologic-erosion-

sedimentation model to evaluate the consequences of various detention

policies. All modeling was based on design storiiis as opposod, for example, to

a series of historic storms. Peak flow reduction through a D/K facility

decreased i-rithin increased recur--ence intervalp increased urbanization, and

increased storm durations--all because of the increased volume of inflow.

Settling efficiencies decreased in similar fashion.

Most D/R policies constrain one point on the discharge-probability

relationship which may result in significant over or under design for other

possibilities. The general trend suggested by this modeling study was that if

a D/R facility is designed for a given recurrence interval, it will be over

designed for more severe events and under designed for less severe events. Of

interest is the v,2rv high--generally 85% or greater--settling efficiencies

predicted for all situations. This was explained by a dominance of heavy,

large-s'zed soil particles and long flow lengths.

S@-@ttlin Efficiency Equatio

Haan and Ward (1978) develope-I a digital computer model --o simulate the

passage of a hydrograph and sedimentgraph through a D/R basin. A limited

amount of data were available to verify the model. Model input includes:

inHow hydrograph, inflow sedimentgraph or inflow sediment mass, a stage-

discharge-area relationship, sediment particle size distribution and specific

gC;LI!Ity. The initial modeling resulted in development of an equation for

-A39-

from a large number of small storms carry m ost of the pollutants.

Therefore, dual purpose D/R facilities should be sized and provided with

outlet works that detain as much runoff as possible from small storms and hold

it for a long time to provide ample time for settling. Dual purpose D/R

facilities should also be designed to safely detain as much runoff from a

large storm but release it from the D/R facility as soon as, possible to

provide room, if needed, for storage of runoff from a subsequent flood.

Maintenance will be even more important for dual purpose D/R facilities than

for single purpose facilities.

SCS Sediment Basin Concepts and Aids

The Soil Conservation Service (USDA, no date; USDA, July, 1975) present

design guidelines for the outlet control structure of a permanent or temporary

facility intended to trap sediment. The principal elements of this structure

are:

1. a turf-covered earthen embankment with a trapezoidal cross-section,

2. a pipe principal spillway passing beneath the structure provided with

a metal riser and trash rack at the upstream end, and

3. an emergency spillway.

This type of hydraulic control structure is frequently modified to serve

the dual purpose of controlling large volumes of runoff associated with large

infrequent rainfall events while effectively trapping sediment and adsorbed

pollutants during smaller but more frequent rainfall and snowmelt events.

-A40-

settling efficiency as a function of D/R capacity, inflow volume, average

detention time, storm duration, pcak inflow and outflow rates, and percent of

sediment particles smaller than 5 and 20 microns, respectively, at the peak

inflow rates.

Potential Adverse Downstream Flooding & Erosion Effects

Malcolm (1978) investigated the effect of a D/R facility receiving flow

from 360 acres,marshed on channel bank and bottom erosion downstream of the

facility. A hypotenthical D/R facility was designed to reduce peak watershed

outflow for the selected design storm from 500 to 150 cfs. A digital computer

model was used to develop inflow-outflow hydrographs.

In addition to before and after hydrographs the design storm

condition, before and after graphs of tractive force versus time were also

constructed. The tractive force graphs indicated that the effect of the

storage facility was to reduce the maximum tractive force by about one-half

but to double the time of persistence of tractive force above the scouring

threshold. This suggests the counter-intuitive conclusion that D/R will not

necessarily reduce downstream channel bottom and bank erosion but may actually

aggevate it by increasing the duration of tractive forces and by

concentrating those forces in and near the channel.

GUIDELINES FOR PLANNING AND DESIGNING

THE WATER QUALITY FEATURES OF D/R FACILITIES

Planning and Designing for Control of Quantity and Quality

Whipple (1979) notes that D/R facilities can be planned and designed to

control both the quality and quanity of urban runoff. The few large storms

that typically occur in a year cause the most flood damage, whereas, runoff

-A41-

Design Procedures for the Sedimentation Aspects of a D/R Facility

Malcolm and New (1975) present a procedure for designing the sedimentation

aspects of a D/R facility. Necessary design criteria include:

1. Selection of a recurrence interval. A 10-year recurrence interval is

suggested for the settling performance of the D/R facility with a more

severe condition being selected for the hydrologic-hydraulic and

structural design of the entire D/R facility.

2. Selection of the size of the smallest particle to the settled with a

given settling efficiency at peak outflow. A 40 to 80 micron particle

is suggested to be removed at 70 percent efficiency.

(Note: 1 micron = 0.001 mm).

The design process, which is described in detail and illustrated with

example, consists of:

1. Sizing a shallow inlet zone to spread the incoming flow across the

width of the basin.

2. Sizing the settlng zone based on the design flow and the design,

particle size and trap efficiency. Hazen's (Fair and Geyer, 1954)

theory of sedimentation tank design is used to determine the

minimum allowable plan area of the settling zone.

3. Sizing the sediment storage zone which lies beneith the settling

zone. The sediment storage zone is sized to contain the total volume

of settled sediment that will be trapped during the life of the

facility or between sediment removal operations.

4. Sizing the perforated metal riser and the piped spillway which

passes beneath the earthen or other dam forming the outlet structure.

The riser diameter is based on a weir flow analysis; the number, size

and location of perforations in the riser area based on an orifice flow

analysis; and the size of the main pipe is based on an inlet control

approach.

-A42-

Oscanyan (1975) also describes and illustrates with an example a procedure

for designing the sedimentation aspects of a D/R facility. The procedure is

more elaborate than that presented by Rooney and New (1975) and requires more

data and, accordingly, may not be as readily applicable. For example,

Oscanyan specifies removal of all sediment particles greater than 5 microns in

diameter and the corresponding steps in the d,@sign procedurc requires a grain

size distribution c,.,,rve for the incoming suspended sediment.

The author provides a schematic plan and section drawings of a

sedimentation basin. The scheniatic complements those presented by others

(USDA, no date; USDA, July, 1975). Important elerients of the basin are:

an earthen dam or embankment, a settling zone on top of a sediment storage

zone, a metal riser and a pipe spillway which passes beneath the earthern

embankment.

The author does not addrass '.,e design of a combination sedimentation

basin-storm water control facility. However, this sedimentation basin design

procedure could be integrated with a design procedure for determinir,g the

vulume of stoca&@ and the OULlet control works needed to control a large

qua;,.tity of runoff like that which would be generatcd by a large storm.

Chen (1975) also presents a design procedure that is similar to the two

p-:-ceding procedures in that ir- is based on settling efficiency as determiaed

by settling velocity of particles. An illustrative examp'(,:, i-@ not provided.

,Usef-,;l design aids included in this paper are: a table of dry ,.!eights of

settled sediment, a table of settling velocities for a range of suspended

sed-iment sizes from fine clay to course sand, and a graph of settling

efficieticy versus outflow rate (unit overflow rate based on plan area) for the

same range in suspended sediment sizes.

_A43-

OPERATION AND MAINTENANCE ASPECTS OF D/R FACILITIES

A systematic maintenance program is required for the safe operation and

proper functioning of a D/R["facility is intended to control the quantity or

quality of storm water runoff. In areas such as northeastern Illinois where

D/R of storm water has been practiced or required for quantity control for a

decade or more, there are indications that proper maintenance has not been

provided. Inadequate maintenance prevents the D/R facility from functioning

as designed and creates safety hazards, flood risks nuisances, and aesthetic

problems.

Madison, W1

Operation and maintenance suggestions based on experience in Madison,

Wisconsin are presented by Schonbeck (1979). The planning and design process

should consider operation and maintenance needs by providing ease of access

and movement for trucks, front-end loaders and mowers. Consideration should

be given to provision of protective &rates on outlet control structures to

minimize entrapment of debris and to prevent children from being washed into

the outlet control structure.

Madison crews check outlet control structures during and after every

runoff event and are experimenting with a weather forecasting service to

provide advance information as to the expected occurrence and magnitude of

runoff events. Madison has experimented with weirs and other outlet control

structures intended to trap sediment.

Highland Park, IL

The City of Highland Park, Illinois (Highland Park, 1977) allows use of a

D/R facility as a temporary sediment trap during the construction phase of

land development provided that the facility is restored to its design storage

capacity. The continued proper functioning of D/R facilities is encouraged by

-A44-

requiring a maintenance procedure for such facilities as part of the plan

submitted for land development. Furthermore, the city engineer is required to

inspect every D/R facility at least once every five years and to serve notice

if reparis or maintenance are required.

Montgomery County, MD
As a result of the Maryland Sediment Control Act of 19**, the Montgomery
County) Soil Conservation District must approve sediment control plans prior

be initiation of land development projects. The District's responsibility

Includes controlling off-site erosion and sedimentation and, therefore, use is

made of D/R facilities. Formal maintenance responsibility arrangements must

be made prior to construction. If facilities are constructed on public land,

the head of the governmental unit must provide a letter indicating

responsibility for maintenance. For facilities constructed on private land,

the owner must sign a maintenance statement.

DISCUSSION AND CONCLUSIONS

Little is Known

With the exception of a few field and laboratory studies of the effects of

D/R facilities on removal of suspended sediment from. storm water runoff, few

field or laboratory studies have- been conducted on the use of D/R for

enhancement of storm water runoff. Therefore, the effectiveness of D/R

facilities in removing non-point source pollutants--other than sediment--has

not been convincingly demonstrated in the field.

A Favorable Prognosis

D/R facilities have, in many instances, been shown to be an effective

means for controlling the quantity of storm water runoff while also offering

-A45-

economic, recreation and aesthetic benefits. Suspended solids trap

efficiencies in excess of 90 percent have been reported for Cie few available

field studies. Although the state-of-the-art is rudimentary, it appears that

with additional effort leading to additional knowledge, D/R facilities could

be planned, designed, constructed and operated to also provide for control of

the quality of storm water runoff.

Effect on Pollutants Other Than Suspended Solids

Based on the results of very few field and laboratory studies, there are

indications that D/R facilities may significantly reduce the concentration of

pollutants other than suspended solids. Examples of these pollutants or

pollution indicators are biochemical oxygen demand, chemical oxygen demand,

pathogenic and indicator bacteria, turbidity and chloride. The solids appear

to be a transporter of pollutants such as phosphorus, pesticides, heavy

metals, and bacteria. Therefore, suspended solids should be the primary

target. The explicit successful control of suspended solids in D/R facilities

is likely to result in the removal of other non-point source pollutants.

Factors that Influence Performance

Field, laboratory, and computer modeling studies suggest that D/R

facilities can easily achieve suspended solids trap efficiencies of 70% or

more. Trap efficiency is improved by or increases with:

1. Volume available for temporary storage of storm water

2. The presence of a restricting outlet control device such as a

metal riser

3. Reduction in emptying time between runoff events

4. Increased fraction of larger soil particles

-A46-

5. Minimizing short circuiting

6. Large surface area for a given storage volume

7. Introduction of a coagulant, such as alum, and coagulant aids

8. An outlet control configuration that minimizes velocity in the D/R

facility in the velocity of the outlet.

The Leveling Effect

Just as D/R facilities reduce the long-term variation in stream flows in

downstream reaches, they may also reduce the long-term variation in stream

water quality, that is, in the concentration and transport of potential

pollutants. Although a leveling of water quality may be generally desirable,

there may be adverse effects.

A Cautious Approach to Design

Some design guidelines are available for the sedimentation aspects of D/R

facilities for use by practicing engineers. These aids are based on a

combination of plain sedimentation basin design procedures drawn from the

field of sanitary engineering and empirical sediment trap efficiency studies

,for large reservoirs. Because incorporation of water quality enhancement

features in DR facilities is relatively new, engineers should proceed with

caution and make full use of the few aids that are available.

Integrating Quantity and Quality Control Concepts

Preliminary engineering, investigations, of D/R facilities tends to

achieve quanity and quality control with supplemental recreation and

aesthetic benefits should draw on experience already gained in the following
three areas: hydrologic-hydraulic design of flood control D/R facilities,

empirical and theoretical sedimentation design procedures, and use of

vegetative filters. One D/R facility concept that integrates these three areas

-A47-

of experience is a series configuration consisting of, in downstream order, a

sedimentation basin, a vegetative filter, and a retention reservoir is shown

in Figure 3.

Possibility of Increased Downstream Erosion

Based on preliminary studies, the construction of a D/R intended to reduce

downstream flooding and pollution may under certain conditions actually

increase downstream erosion and sedimentation by incrasing the duration of

erosive forces, by concentrating those forces in or near the channel and by

the reduced sediment load in the D/R discharge. Designers should carefully

assess this possibility and its consequences.

The "Little Additional Expense" Idea

In terms of land acquisition and construction costs, relatively little

additional expenditures are likely to be required to add effective water

quality control components to a planned D/R facility. Storm water D/R

facilities intended for control of storm water quantity with supplemental

recreation and aesthetic benefits have typically been of such size and

configuration that relatively little additional earth work and components

would be required.

Potential Inspection-and Maintenance Problems

The short experience with operating D/R facilities, most of which were

intended to control the quanity --not quality--of runoff, suggests that

periodic and storm-related inspection and maintenance is vital to effective

operation. One problem that has occurred is accumulation of sediment and

debris. Because D/R facilities intended for both quantity and quality control

will, by design, accumulate even more sediment and debris, it follows that

systematic inspection and maintenance will become even more important.

-A48-

QUANTITY QUANTITY
COMTROI CONTROL

SEDIMENT
BA'S I N
PERFORATED SfDIME@tT
RISER PIPE StTTLING
FAC16T'Y ZONE
-D RETENTION VEGETAYIVE
-ROL FILTER Q)

TEMPORARY FLOOD
SEDIMEWC
CONTROL STORAGE BER
A A AGE
Ml@ STOR
ERM W/P.IPE ZONE
L SPILLWAY

V%BERM
W/SPILLWAYS

,ure 3. Intergrating Quantity and Quality Control Concepts.

-A49-

Inspection and Maintence Approaches

Proper inspection and maintenance of D/R, facilities may be accomplished by

municipal staff or be carried out by he owner, with the latter being assured

by legal agreement or municipal ordinance. inspection and maintenance

responsibities should be formalized prior to beginning construction.

One of Several Measures

D/R facilities are one of many structural and non-structural measures

potentially available for managing the quantity and quality of storm water

if significant advances are made in the state-of-the-art of D/R
facility knowledge, other management measures may more appropriate in a

given situation.

"Go Slow" in Mandating D/R Facilities

Given the limited knowledge, local, state and federal governmental units

should "go slow" in mandating D/R for controlling the quality of storm water

runoff in urban and urbanizing areas. Premature rule-making and regulation is

likely to result in "action" but little progress. More is likely to be

accomplished in advancing the state-of-the-art and in achieving significant

control of non-point source pollutants by (a) funding additional research and

development projects and pilot studies and by "encouraging" the control of

non-point source pollutants but not dictating the means.

-A50-

CITED REFERENCES

Bauer, Kurt W. 1978. An overview of the sources of water pollution in
southeastern Wisconsin. Southeastern Wisconsin Regional Planning Commission,
Technical Record.

Betz Environmental Engineers. 1976. Planning methodologies for analysis of
land use/water quality relationships. Technical Appendix.

Bondelid. T.R., and R. H. McCuen. 1979. A computerized method for designing
stormwater management basins. Technical Report, Department of Civil
Engineering, University of Maryland, College Park.

Bondiment. J.A., C.E. Brockway, and M. J. Brown. 1975. Some aspects of
sedimentation pond design. National Symposium on Urban Hydrology and Sediment
Control Proceedings. University of Kentucky, Lexington.

Brune, G. M. 1953. Trap efficiency of reservoirs. Transactions of the
American Geophysical Union. 34(3):

Chen, Charng-Ning. 1975. Design of sediment retention basins. National

Symposium on Urban Hydrology and Sediment Control Proceedings, Mini Course No.
2. University of Kentucky, Lexington.

Cherkauer, Douglas S. 1977. Effects of urban lakes on surface runoff and
water quality. Water Resources Bulletin, 13(5): 1057-1067.

Colston, Jr., Newton, V. 1974. Characterization and treatment of urban land
runoff. U.S. Environmental Protection Agency, Environmental Protection
Technology Series, EPA-670/2-74-096.

Curtis, D. C. and R. H. McCuen. 1977. Design efficiency of storm water
detention basins. American Society of Civil Engineers, Journal of the Water
Resources Planning and Management Division, 103(WRI): 125-140.

Cordero, E. T. 1971. Management of runoff from minor storms in Montgomery
County, Maryland. American Society of Agricultural Engineers, St. Joseph,
Michigan.

Dalrymple, Robert J., Stephen L. Hodd, and David C. Morin. 1979.. Physical
and settling characteristics of particulates in storage and sanitary
wastewaters. U.S. Environmental Protection Agency, Environemental Protection
Technology Series, EPA-670/2-79-050.

Davis, Ernet M. 1979. Maximum utilization of water resources in a planned

community-Bacterial characteristics of stormwaters in developing areas.
U.S. Environmental Protection Agency, Environmental Protection Technology
Series, EPA-600/2-79-050f.

Davis, William J. 1978, Sediment trap efficiency study - Montgomery County,
Maryland. American Society of Agricultural Engineers Proceedings, Winter
Meeting, Chicago, Illinois.

-A51-

Donahu, J. R., R. H. McCue@I, and T. R. Bondelid. (In press). Comparison of

stormwater detention planning and design models. American Society of Civil
Engineers, Journal of the Water Resourct2.s Planning and Management Division.

Donohue & Associates, Inc. 1973. A plan for the protection of Green Lake.

Donohue & Associates, inc. 1979. Storm water management plan for the Cushing
Creek watershed, city of Delafield, Waukesha County, Wisconsin.

Donohue & Associates, Inc. 1978. Storm water management plan for the

southeat area of Mount Pleasant storm water drainage district No. 1, Racine
County, Wisconsin

Fair, Cordon MI., and Joha C. Geyer. 1954. In Wator supply and waste-water
disposal, Chapter 2_9' Sedimentation and flotation. Wilev and Sons, New York,

Fetter, jr., C. W., William E. Sloey, and Fred L. Spangler. 1978. Use of
natural marsh for wastewater polisliing. Journal of Water Pollution Control
Federation.

Geiger, A. F. 1963. Developing sediment storage requirements for upstream
retarding reservoirs. Proceedings of the Federal Inter-agency Sedimentation
Conference, U.S. Department of Agriculture, Miscellaneous Publication No. 970,
pp.-881-885.

Haan, C. T. and A. D. Ward. 1978. Evaluation of detention for
controlling urban runoff and sedimentation. Research Report No. 112, Water
Resources Research Institute, University of Kentucky, Lexington.

Highland Park, City of, Illinois. 1978. Storm water management. Chapter 154
of Ordinances.

Kamedulski, Gregory E. and Richard H. McCuen. 1979. Evaluation of
alternative storm water detention policies. American Society of Civil
Engineers, Journal of Water Resources Planning and Management Division,
105(WR2): 171-186.

Lager, John A., William G. Smith, William G. Lynard, Robert IR. Finn, and Jobin
E. Finnemore. 1977. Urban storm water management and technology: Update and
uses guide. U.S. Environmental Protection Agency Environmental Protection
Technology Series, EPA-600/8-77-014.

Leach, L. G., and B. L. Kittle. 010)4i6. Hydraulic design of the Fort Campb-211
storm drainage system, Highway Research Record No. 116, pp. 1-7.

Malcolm, H. Rooney and Vernon E. New. 1975. Design approaches for stormwater
management in urban area,3, Part 4: Sediment basin design. North Carolina
State University,.Raleigh.

-A52-

Malcolm, H. Rooney. 1978. Culverts, flooding, and erosion. Paper presented
at The Engineering Foundation Conference, Water Problems in Urban Areas.
Henniker, New Hampshire.

Malcolm, H. R. 1980.A study of" detention in urban stormwater management.
Water Resources Research Institute, Report No. 156.

McCuen, R. H. 1979. Downstream effects of stormwater management. American
Society of Civil Engineers, Journal of Water Resoures Planning and Management
Division, 105(WR11): 1343-1356.

McCuen, Richard H. 1980. Water quality trap efficiency of storm water
management basins. Water Resources Bulletin, 16(1): 15-21.

Mein, R., 1980. Analysis of detention basin systems. Water Resources
Bulletin, 16(5): 824-329.

Milwaulkee, City of, and Consoer, Townsend & Associates. 1975. Detention tank
for combined sewer overflow - Milwaukee Wisconsin demonstration project
U.S. Environmental Protection Agency, Evironmental Protection Technology
Series, EPA-600/2-:5-071.

Montgomery Soil Conservation District. 1975. On-site storm water management
policy. Rockville Maryland.

Oscanyan, Paul C. 1975. Design of sediment basins for construction sites.
National Symposium on Urban Hydrology and Sediment Control Proceedings,
University of Kentucky, Lexington-

Poertner, Herbert G. 1974. Practices in detention of urban storm water
runoff. American Public Works Association, Special Report No. 43.

Rawls, W. J. and R. H. McCuen. 1978. Economic assessment of -storm drainage
planning. American Society of Civil Engineers, Journal of Water Resources
Planning and Management Division, 104(WRI): 45-54.

Schoenbeck, Robert H. 1979. Maintenance and operation of detention in
systems. Paper presented at the Stormwater Detention Systems Institute,
University of Wisconsin - Extension, Madison.

Small, Maxwell M. 1976. Data report - marsh/pond system. Brookhaven
National Laboratory, Upton , New York.

Spangler, Fred L., C. W. Fet.-, Jr., and William E. Sloey. 1977. Phosphorus
accumulation - Discharge cycles in marshes. Water Resources Bulletin, 13(6):
1191-1201.

-A53-

USDA, Soil Conservation Service. (no date). Minimizing erosion in urbanizin-7
areas. Section titled, Storm Water Removal Systems, Madison, Wisconsin.

USDA, soil Conservation service. 1975. Standards and specifications for soil
erosion and sediment control in developing areas. College Park, Maryland.

-structural methods of storm
-2nck, Norman C. 1977. Use of wetlands for non
water treatment. Abstract of a paper presented at the 10th Annual Water
Resources Seminar, Minneapolis, Minnesota.

Whipple, 7;illiam. 1979. Dual ?urpose detantion basins. American Society of
Civil Engineers, Journal of Wat,@r Resources Planning and Managem-@nt Division,
105(WR2): 403-412.

-A54-

ATTACHMENT A

ADDITIONAL REFERENCES

Abt, S. R., and N. S. Grigg. 1978. An approximate method for sizing
detention reservoirs. Water Resources Bulletin, 14(4): 956-965

Alley, W. M. 1976. Guide for the collection, analysis, and use of urban
stormwater data. American Society of Civil Engineers, New York, N.Y., 115 pp.

Amandes, C. B., and P. B. Bedient. 1979. Stormwater detention in developing
watersheds. American Society of Civil Engineers, Journal of Environmental
Engineering Division, 105(EEI): 1-17.

Bedient, P. B., and P. G. Rowe. 1979. Urban watershed management: Flooding
and water quality. Rice University Studies, Houston, Texas.

Berwick, R., M. Shapiro, J. Kuhner, and D. Luecke. 1978. Evaluation of storm
water management practices in new residential areas: Methodological
developments. International Symposium on Urban Storm Water Management,
University of Kentucky, Lexington.

Brandsletter, A. 1974. Comparative analysis of urban stormwater models.
Battelle Memorial Instutute, Richland, Washington.

Chen, H. S. 1978. A mathematical model for water quality analysis. 26th
Annual Hydraulics Division Specialty Conference Proceedings, University of
Maryland, College Park.

Cherkauer, D. S. 1977. Effects of urban lakes on surface runoff and water
quality. Water Resources Bulletin, American Water Resources Association,
13(5): 1057-1067.

Chu, C. S ., and C. E. Bowers. 1978. An optimization technique for a
mathematical urban runoff model. 26th Annual Hydraulics Division Specialty
Conference Proceedings, University of Maryland, College Park.

Collins, P. G., and J. W. Ridgway. 1980. Urban storm runoff quality in
southeast Michigan. American Society of Civil Engineering, Journal of
Environmental Engineering Division, 106(EE1): 197-210.

Curtis, D. C. 1976. A deterministic urban storm water and sediment discharge
model. National Symposium on Urban Hydraulics and Sediment Control,
University of Kentucky, Lexington.

Dodge, R. A. 1978. Sediment model verification for small diversions. 26th
Annual Hydraulics Division Specialty Conference Proceedings, University of
Maryland, College Park.

-A55-

Ellis, S. R. 1978. Hydrologic data for urban storm runoff from three
localities in the Denver Metropolitan Area, Colorado. U.S. Geological Survey
Open-file Report 78-410, Lakewood, Colorado, 135pp.

Field, R. 1980. EPA research in urban stormwater pollution control.
American Society of Civil Engineers, Journal of Hydraulics Division, 106(HY5):
819-835.

Ciessner, W. R., R. W. Cockburn, F. H. Moss, Jr., and M. E. Noonan. 1974.
Planning and control of combined sewerage systems. American Society of Civil
Engineers Proceedings, Journal of Enviromental Engineering Division,
100(EE4): 1013-1032.

Giggev, Michael D., and William G. Smith. 1978. National needs for combined
sewer overflow control. American Society of Civil Engineers Proceedings,
Journal of Engironmental Engineering Division, 104(EE2): 351-366

Hartigan, J. P., A. M. Lumb, J. T. Smullen, T. J. Grizzard, and P. A.
Alcivar. 1978. Calibration of urban nonpoint pollution loading models. 26th
Annual Hydraulics Division Specialty Conference Proceedings, University of
Maryland, College Park.

Heaney, J. P., S. J. Nix, and M. P. Murphy. 1978. Storage-treatment mixes
for stormwater control. American Society of Civil Engineers Proceedings,
Journal of Environmental Engineering Division, 104(EE4): 581-592.

Heaney, J. P. 1978. Water problems of urbanizing areas. Civil Engineering,
48(12): 20.

Huang, V. H., R. J. Mase, and E. G. Fruh. 1973. Nutrient studies in Texas
impoundments. Journal of the Water Pollution Control Federation, pp. 105-118.

Huber, W. C. 1975. Modeling for storm water strategies. APWA Reporter, pp.
10-14.

Jalal, K. F. 1977.. Water quality impacts of urbanization - A methodology.
American Society of Civil Engineers, Journal of Environmental Engineering
Division, 103(EEl): 49-57.

Jennings, M. E., K. M. Waddell, and B. C. Massey. 1978. Water quality of
urban storm drainage from Houston, Texas. 26th Annual Hydraulics Division
Specialty Conference Proceedings, University of Maryland, College Park.
I

Jewell., T. K., and D. D. Adrian. 1978. SWMM stormwater pollutant washoff
function. American Society of Civil Engineers, Journal Environmental
Engineering Division, 104(EE5): 1036-1040.

Jones, D. Earl, Jr. 1967. Urban hydrology - A redirection. Civil
Engineering, 37(8): 58-62.

-A56-

Kamedulski, G. E., and R. H. McCuen. 1978. Comparison of stormwater
detention policies. 26th Annual Hydraulics Division Specialty Conference
Proceedings, University of Maryland, College Park.

Lager, John A., and William G. Smith. 1974. Urban stormwater management and
technology: An assessment. Environmental Protection Technology Series
EPA-670/2-74-040, U.S. EPA Environmental Research Laboratory, Cincinnati,
Ohio, 447 pp.

Lager, J. A., T. Didriksson, and G. B. Otte. 1976. Development and
application of a simplified stormwater management model. Environmental
Protection Technology Series WPA-600/2-76-218. U.S. EPA Environmental
Research Laboratory, Cincinnati, Ohio, 139 pp.

Landwehr, J. M. l979. A statistical view of a class of water quality
indices. Water Resources Research, 15(2): 460-468.

Leach, L. G., and B . L. Kittle. 1966. Hydraulic design of the Fort Campbell
storm drainage system. Highway Research Record, No. 116, pp. 1-7.

Lohani, B. N., and N. C. Thanh. 1979. Probabilistic water quality control
policies. American Society of Civil Engineers, Journal of Environmental
Engineering Division, 105(EE4): 713-725.

Lystrom, D. J. 1979. Data gathering for stormwater runoff and surveillance.
In Water Problems of Urbanizing Areas, W. Whipple, Editor, American Society of
Civil Engineers, New York, N.Y.

McPherson, M. B. 1975. Need for metropolitan water balance inventories.
American Society of Civil Engineers Proceedings, Journal of Hydraulics
Division, 101(HY4): 409.

McPherson, M. B. 1975. Urban mathematical-modeling and catchment research.
American Society of Civil Engineers Urban Water Resources Research Program,
Technical Memorandum No. IHP-1.

McPherson, M. B. 1975. Urban hydrological modeling and catchment research in
the U.S.A. American Society of Civil Engineers Urban Water Resources Research
Program, Technical Memorandum No. IHP-1, 49 pp.

McPherson M. B. 1978. Urban runoff control planning. U. S. Environmental
Protection Agency, Report EPA-600/9-78-035, Washington, D.C., 187 pp.

McPherson, M. B. 1978. Urban runoff control master planning, American
Society of Civil Engineers Proceedings, Journal of Water Resources Planning
and Management Division, 104(WR1): 223-234.

-A57-

Matthew, H. C. J. 1978. Quality and quantity of storm water runoff from
three land use areas, Broward County, Florida. International Symposium on
Urban Storm Water Management, C. T. Haan, Editor, University of Kentucky,
Lexington.

Oberts, Gary L. 1977. Water quality effects of potential urban best
management practices: A literature review. Department of Natural Resources,
Technical Bulletin No. 97, Madison, Wisconsin, 25 pp.

Omernik, J. M. 1977. Nutrient concentrations in streams from nonpoint
sources. Corvallis Environmental Research Laboratory, U.S. Environmental
Protection Agency, Corvallis, Oregon, 6 pp.

Padmanabhan, G., and J. W. Delleur. 1978. Statistical and stochastic
analysis of synthetically gen rated urban storm drainage quantity and quality
data. Water Resources Research Center, Technical Report No. 108, Purdue
University, Lafayette, Indiana.

Pitt, Robert, and Richard Field. 1977. Water-quality effects from urban
runoff. Journal of American Water Works Association, 69(8): 432-436.

Poertner, H. G., R. L. Anderson, and K. W. Wolf. 1966. Urban drainage
practices, procedures and needs. American Public Works Association, Special
Report No. 31.

Poertner, H. G. 1973. Better storm drainage facilities - at lower cost.
Civil Engineering, pp. 67-70.

Poertner, H. G. 1976. Urban stormwater detention and flow attenuation.
Public Works Journal, pp. 83-84.

Polls, I., and R. Lanyon. 1980. Pollutant concentrations from homogeneous
land use. American Society of Civil Engineers, Journal of Environmental
Engineering, 106(EEl): 69-80.

Ragan, R. M., and A. J. Dietemann. 1975. Impact of urban stormwater runoff
on stream quality. Urbanization and Water Quality Control, American Water
Resources Association, Proc. No. 20, pp. 55-61.

Rawls, W. J., V. A. Stricker, and K. V. Wilson. .1980. Review and Evaluation
of Urban Flood Frequency Procedures. U.S. Department of Agriculture Science
and Education Administration, Bibliographies and Literature of Agriculture
No. 9, 63pp.

Rausch, D. L., and J. D. Schoerber. 1977. Callahan reservoir: I sediment
and nutrient trap efficiency. American Society of Civil Engineers, 20(2):
281-284, 290.

Respond, F. J. 1976. Roof retention of rainfall to limit urban runoff.
National Symposium on Urban Hydrology, Hydraulics, and Sediment Control,
University of Kentucky, Lexington.

-A58-

Rogers, Peter. 1978. On the choice of the 'appropriate model' for water
resources planning and management. Water Resources Research, 14(6): 1003-1010.

Russell, S. 0., B. F. I. Keaning, and G. J. Sunnell. 1979. Estimating design
flows for urban drainage. American Society of Civil Engineers Proceedings,
Journal of the Hydraulics Division, 105(HYl): 43-52.

Sartor, J. D., and G. B. B oyd. 1972. Water pollution aspects of street
surface contaminents. U.S. Environmental Protection Agency, EPA-R2-72-081.

Soil Conservation Service. 1975. Urban hydrology for small watersheds. U.S.
Department of Agriculture, Soil Conservation Service, Technical Release No.
55, 91 pp.

Terstriep, M. L., and J. B. Stall. 1974. The Illinois urban drainage area
simulator, ILLUDAS. Illinois State Water Survey, Bulletin No. 58.

U.S. Army Corps of Engineers. 1976. Storage, treatment, overflow runoff
model (STORM), users' manual. Hydrologic Engineering Center, Corps of
Engineers, Computer Program 723-S8-L7520, Davis, California, 170 pp.

Verhoff, F. H., and D. A. Melfi. 1978. Total phosphorus transport during
storm events. American Society of Civil Engineers, Journal of Environmental
Engineering Division, 104(EE5): 1021-1025.

Walesh, Stuart G. 1978. Urban nonpoint source pollution: Monitoring,
modeling, and management. A paper presented at the American Water Resources
Association, Wisconsin Section, Second Annual Meeting, "Non-Point Source
Problems" Session, Milwaukee, Wisconsin, l6pp.

Watkins, L. H. 1962. The design of urban sewer systems. Department of
Scientific and Industrial Re-search, Road Research Technical Paper No. 55,
London, England.

Wenzel, H. G., Jr., J. W. Labadie, and N. S. Grigg. 1976. Detention storage
control strategy development. American Society of Civil Engineers, Journal of
Water Resources Planning and Management Division, 102(WRI): 117-136.

Williams, J. R. 1978. A sediment yield routing model. 26th Annual
Hydraulics Division Specialty Conference Proceedings, University of Maryland,
College Park.

Whipple, W., Jr., J. V. Hunter, and S. L. Yu. 1977. Effects of storm
frequency on pollution from urban runoff. Journal of Water Pollution Control
Federation, pp. 2243-2248.

Younkin, L. M. 1978. Sediment yield from highway construction areas. 26th
Annual Hydraulics Division Specialty Conference Proceedings, University of
Maryland, College Park.

-A59-

ATTACHMENT B - Papers identified by selected key words presented in Table 1.

1. Alley, W. M. 1980. A simple lumped-parameter runoff quality model.

National Symposium on Urban Stormwater Management in Coastal Areas) C. Y. Kuo,

Editor, American Society of Civil Engineers, New York, pp 236-243.

Abstract: The state-of-the-art of urban runoff modeling is far in advance of

urban runoff-quality modeling. Whereas, a physically-based distributed-

parameter approach to rainfall-run off modeling has been successfully performed

in many studies, such approaches toward runoff-quality modeling have rarely

been a proven success. Recognizing this fact, a single runoff-quality model

has been developed. The model is a lumped-parameter model in that no spati

variations in model parameters are accounted for. This model has been linked

with a more complex distribution routing rainfall- runoff model. The purpose

of this paper is to describe the structure of the runoff-quality model and its

application to a small urban drainage basin along the coast of south Florida.

A description of the structure and an application of a simple lumped-parameter

urban runoff-quality model has been presented. The model utilizes exponential

constituent accumulation and washoff equations for runoff-quality simulation

and a modified-Rosenbrock direct search algorithm to identify model parameters.

An option is included in the model to account for wetfall contributions. The

model is limited in its applicability to watersheds and constituents for which

runoff loads originate predominantly from effective impervious surfaces and

for which constituent concentrations tend to decrease with time from start of

storm. The model was calibrated and verified for storm-runoff loads of total

nitrogen from a small urbanized watershed in south Florida; standard errors of

-A60-

approximately 60 percent were found during model verification. The potential

importance of accounting for wetfall contribution when determining model

parameters for total nitrogen was demonstrated in this application.

2. Baker, W. R. 1977. Stormwater detention basin design for small drainage

areas. Public Works, pp 75-79

Abstract: Control and management of stormwater runoff has resulted in

associated problems causing it to become an increasing source of concern for

public works engineers. One of the most effective methods of reducing the

rate of runoff is through the use of a storage or detention basin. The author

uses hydrographs to develop a model for uses in determining detention basin

design for small drainage areas. A modified version of the rational method

was used to generate the inflow hydrograph. This method represents a simple

procedure which yields acceptable design results when applied to small

drainage areas (i.e., up to 200 acres in area).

3. Bedient, P. B., and C. B. Amandes. 1978. Monitoring, modeling and

management of non-point sources in Houston, Texas.. International Symposium

on Urban Storm Water Management, C. T. Haan, University of Kentucky,

Lexington, pp 151-157.

Abstract: The Department of Environmental Science and Engineering at Rice

University has undertaken a coordinated research effort with the City of

Houston Health Department on stormwater sampling and management. A

comprehensive stormwater quality and low flow sampling program has been

designed for the Brays Bayou watershed and tributaries in order to define

pol loads (TSS, TIP, NO 3, NH 3, BOD, DO) and potential impacts. The

main objective of the study is to organize the City of Houston sampling effort

in order to differentiate the impact of point sources vs. stormwater runoff.

-A61-

Load-runoff relationships have been developed as a function of land use type

and are used in a predictive model (HLOAD) for multiple or individual storm

events. Stormwater responses are predicted using a hydrolic formulation such

as STORM in concert with the HLOAD methodology. Results from four watersheds

have been quite satsifactory for TSS, TP, TKN, and COD. Total annual load

calculations indicate 85-90 percent is due to nonpoint source runoff. For a

variety of stormwater management practices, including channel modification,

detention storage and land use controls, STORM and HLOAD models were used to

characterize the effects of storm f1ows and receiving water quality. Both

present and projected land use patterns were imput to investigate the effects

of future development in the watershed. Cost data for a variety of management

alternatives will be estimated and compared to evaluate relative effectiveness.

4. Bedient, P. B., P. A. Harned, and W. G. Characklis. 1978. Stormwater

analysis and prediction in Houston. American Society of Civil Engineers,

Journal of Environmental Engineering Division, 104(EE6) 1087-1100.

Abstract: Analyses of available stormwater hydrologic and water quality data

for several Houston area watersheds yielded direct linear relationships

between pollutant mass loading rates and cumulative storm runoff volume. These

relationships were then used to rank the watersheds on a pollutant load per

unit runoff basis, and to calculate total annual pollutant loads for each area.

Further, a plot of dimensic less cumulative flow versus the dimensionless

c flux provides a quantifiable means to evaluate the first flush.

allowing comparison of this effect between watersheds. Finally, the developed

load-runoff relationships provide the basis for a simple, yet effective method

of predicting pollutant mass flow for individual or sequential storm events.

-A62-

5. Bouthiller, P.H., and A. W. Peterson. 1978. Storage requirements for

peak runoff control. International Symposium on Urban Storm Water Management

C. T. Haan, Editor, University of Kentucky, Lexington, pp 13-18.

Abstract: Storage of storm runoff is becoming a common part of the design of

Lana developments because most environmental control authorities have imposed

limits on the pentrate of runoff. The objective of such regulations is to

reduce erosion by flows which would exceed those which occurred before

development. A method of computing storage requirements for storm runoff is

presented. A plot of maximum allowable outlet rate/maximum inflow rate vs

required reservoir volume/storage. A bottom outlet is used, the rate of

outlet varying as the square root of the head on the outlet. Curves are

provided for triangular and for cusp shaped hydrographs. Computations are

based or, total runoff (coefficient of runoff x rainfall) and a chosen base

length of hydrograph. Allowable peak outlet rates may be assumed or set by

regulatory agencies. Examples are given using Edmonton rainfall data. The

range of options provided, viz; return storm period, base length of hydrograph,
and runoff coefficients should enable wide applicability of this method.

Considering the uncertainties of rainfall intensity and patterns it is

believed that the method is sufficiently accurate for most applications.

6. Brandt, G.H., E. S. Conyers, M. B. Ettinger, F. J. Lowes, J. W. Mighton,

and J.W. Pollack. 1972. An economic analysis of erosion and sediment

control methods for watersheds undergong urbanization. The Dow Chemical

Company. Midland,

Abstract: Economic benefits are expected from controlling erosion and

sediment during urban construction, but control costs for urban erosion have

not been previously related to benefits. In this study, costs are compared to

effectiveness for many erosion and sediment control systems. Potential soil

-A63-

losses were calculated with the universal soil loss equation. Damage costs in

streams were assigned to specific construction sites by equations incorporating

estimated cost constants and sediment delivery ratios for various "legs" along

a drainage path. Specific damage values were determined for numerous

downstream sediment problems. The Seneca Creek watershed, located 20 miles

northwest of Washington, D.C., used as a model watershed for the study.

Control practices presently being used in the watershed include sediment

basins, diversion berms, level spreaders, grade stabilization structure, sodded

ditches, seeding, and straw mulch tacked with asphalt or disked. This

conventional system is estimated to cost $1,125 per acre and to control 91% of

the potential erosion. Control systems incorporating large sediment basins can

boost control to 96% at less total cost. Multipurpose impoundments design with

sediment forebays for chemical flocculation can boost urban sediment control

to 99%. In addition, such structures contribute very significantly to control-

ling sediment from agricultural and idle land. Sediment from these non-urban

sources must be controlled to significantly reduce the sediment pollution

problem. Urban construction sites can contribute large quantities of sediment

per unit area, but constitute a minor portion of the total sediment pollution

problem because total acreage is generally small compared to other land that

contributes sediment. Sediment damages from uncontrolled erosion on urban

construction sites in the Seneca watershed could potentially reach

$1,500/acre. This estimated damange value applies to specific conditions

prevailing in the watershed where 300 tons of sediment could erode per acre of

uncontrolled development during an 18-month construction period. The

estimated maximum soil erosion rate approaches 200 ton/acre/year or 128,000
ton/mile 2/year. However, downstream damages would be caused by only a

-A64-

portion of this total potential sediment. Sediment delivery and transport

ratios determine damage distribution along a stream and are a major source of

uncertainty in this economic analysis.

7. Butler, R. V., and M. D. Maher. 1978. The economics of urban stormwater

management. Center for Economic Theory and Econometrics, Rice University,

Houston, Texas.

Abstract: Recent studies have shown that urban flooding and water quality

problems are due in part to the effects of upstream urbanization. Economic

analysis suggests that upstream development will continue and the problem will

worsen because upstream residents have no incentive to alter their behavior

that causes stormwater problems. The consequences of the increased runoff

they generate are borne by others - by affected downstream residents and by

the general taxpayer if traditional government protection programs are used to

lessen damage downstream. The major policy implication of this analysis is

that treatment and mitigation efforts should focus on the source of the

problem as well as the site of damage. The external costs of upstream

development requires that both flooding and water quality aspects of

stormwater runoff be managed basin-wide, since uncoordinated community-by-

community action cannot simultaneously implement controls upstream, upland and

downstream. in addition, the efficient management strategy must consider both

flooding and water quality simultaneously. Since the pollution and flooding

problems are produced jointly, it will typically make sense to plan for joint

treatment. Several policy instruments for basin-wide control are discussed.

8. Calabrese, M. M. 1980. Optimization of stormwater management practices.

National Symposium on Urban Stormwater Management in Coastal Areas, C. Y. Kuo,

Editor, American Society of Civil Engineers, New York, pp 183-194.

-A65-

Abstract: In rec~,~2nt year--, stormwater has been found to be a major source ~0~q@

pollution to receiving waters. Major research efforts h~a~ie been directed in

this area, prim~a~-~L~-il~ly as a ~-~c~esult of the water pollution acts and amemdments.

Yet, there r~e~qma~qi~n~g a need for more data in this field of stormwater management,

data, and the planning methodology to optimally

select ~wa~n~ag~em~u~n~@ practices based o~a the cost-performance data. The

objectives of t~q!~.~.~-~*~,~s ~-~@~-~,~_~s~earch were to develop cost~-effective data on vari~O~L~I~S

management practices, and use these data to find optimal combinations of

storm~-~,~,~@ater management practices for storm-sewer systems. The optimal

combinat~qi~1p~p~p~f structural and non-structural manage~qm~@~nt ~qp-actices for

w~~t-weather pollution control was determined using linear progra~t~a~nin~qg methods.

The selection is a most critical ~-~_~.~'~a~qment~.of sto~v~i~i~iwater management research

because stormwater removals must be related to lake productivity and the best

combination of management practices must be specified to achieve desired water

quality improvement.. Combinations of computer analyses and mathematical

programming were used to analyze the data. The research culminating in the

computer program "MANAGE" described in the body of t~qhis paper reveals a

methodology for the choice of an optimal combination of stormwater management

practices. However, more work is n~eede~,~q! to estimate localized cost--

~92;9316;40;60qL- Incor~)orating mathematical
~~-~q1ci~encie~s for stormwater management practices. ~r

pro~qgra~;~@~,~@~-~,ing m~,~-~-~-~*~,~,-~)ds to t~-~,~,~,~1p~p~work ~s ~i-~;es t~i~:~z~a and money sit-ice i~L eliminate~@~@~,

~c~g~ki~e.~q, ~q%~q,~qi~qor~ql~qk allows one to combination of pra~qct~0qi~qr~q:~q,~qs wi~2q@_~q!

r~e~r~qq~qn~qv~q@~q, ~qi ~qm ~q@i~qm~qum ~qa~qc~qnou~qn~q@ of Pollutants at least cost. The program was

beneficial in es~0qLab~6qlishi~q;~qi~8qg the least cost combin-~q@~qon of man~qa~2qg~qem~q,~q@~q?nt pract~q,~qr~q@es

in the Lake E~qola ~q@~q-~q,~qitershe~2qd, Orlando, Florida.

-A66-

Cordero, C.T. 1972. Management of runoff from minor steams in

Montgomery County, Maryland. American Society of Agricultural Engineers,

St. Joseph, Michigan. (preprint)

Abstract: The purpose if this paper is to present briefly the action taken by

and in the County to relieve some of the problems connected with storm water

runoff in a rapidly urbanizing area. First will be presented the procedure by

which the criteria was developed, and a synopsis of the criteria itself. Then

several examples of how the criteria was applied will be briefly described.

Finally, an assessment of how well the program is succeeding will be offered

to give others the benefit of the experience to date.

10. Curtis, P. C., and R. H. McCuen. 1977. Design efficiency of stormwater

detention basins. Journal of American Society of Civil Engineers, Water

Resources Planning and Management Division, 103(WRl): 125-140.

Abstract: In addition to increased flood runoff, urban development has caused

a significant increase in sediment loads in streams. While many means of

sediment and runoff control have been proposed, stormwater detention has been

shown to be one of the more cost efficient means. Because detention

facilities have not been used extensively in the past, a data base is not

available for determining the effect of design factors on sediment trap

efficiency and runoff control characteristics. A mathematical model, which

includes erosion, sedimentation, and detention facility components, was

developed from principles of hydraulics and classical settling mechanics. The

model was used to examine the effect of: (1) Detention basin location; (2)

soil particle size distribution; (3) basin depth; (4) initial storage, and (5)

orifice diameter. An understanding of the relative importance of these

factors may lead to better design of stormwater detention facilities.

-A67-

11. Davis, W. J., R. H. McCuen, and C. E. K:imedulski. 1978. The effect of

storm water detention on water quality. International Symposium on Urban

Storm Water Management, C. T. Haan, Editor, University of Kentucky, Lexington,

pp 211-218.

Abstract: Recent studies have indicated that non-point sources of pollution

are often a major contributor to the degradation of sLream quality. While

storm wat---r detention basins have proven to be an effective means of control-

ling both runoff rates and sediment loads, the effectiveness of detention

facilities has not been documented. Montgomery County, Maryland, has completed

a Section 20P research study of tl,-- effectiveness of a functional sedimentation

controL basin and a storm water manag-ement basin, as examples of local non-

point sour-.; pollution control structures. These structures are recommended

for sLrategic use-in a plan for area-wide waste treatment management developcd

by the Metropolitan Washington Council of Governments for the Washington, D.C,

area. The study included monitoring of rainfall and water quality parameters

in basin inflow and outflow, as well as detention basin and land use

characteristics. Analysis of measured data provided relationships of inflow

sed-'-,ent and pollutant loadings to land use and rainfall. Trap efficiencies

for sedi---,nt and pollutant loads were evaluated and related to rainfall, land

use, and ':-as:,@ design characteristics. Using the measured data, a hydraulic/

hydrologic moJ@-l was calibrated and used to examine the effect of storm watrir

d@tenLiun on and other pollutant lo,-,ds. The mod@! used to examine

tb,_- ef' t tinit variation )f basin design chnract-ristics would have on

e'Z,iciency of the structures. Monir--)ring installations anJ data collection

are described, data analysis techniques and results are presented, and the

model, its calibration, and operation are discussed.

-A68-

12. Day, J. W., and C. L. Ho. 1978. Assimilation of Acricultural runoff by

a swamp forest wetland. Agricultural and Mechanical College, Center for

Wetlands Resources, Louisiana State University, Baton Rouge.

Abstract: The general objective of this research is to determine the ability

of natural wetlands in south Louisiana to remove nutrients from agricultural

runoff. We are specifically interested in an area of swamp forest near

Lac des Allemands. This lake is now highly eutrophic, principally because of

agricultural runoff. Under natural hydrologic conditions, runoff from higher

areas would percolate through the swamp. Now, because of a extensive network

of drainage canals, nutrient laden runoff flows directly into waterbodies

causing the eutrophication problem. This problem affects both fresh and

estuarine waters and is beginning to threaten commercial fisheries. Earlier

studies in Louisiana have shown that when fishery wastes were applied to

wetlands, the wastes were assimilated and productivity of wetland vegetation

increased by as much as 50%. If the research proposed herein is successful,

it will show the utility of natural treatment for non-point source runoff.

Thus, a widespread cause of deterioration of water quality in south Louisiana

may have an economical solution.

13. Dendrou, S. A., J. W. Delleur, and J. J. Talavage. 1978. Systematic

planning of urban storm-drainage utilities. International Symposium on Urban

Strorm Water Management, C. T. Haan, Editor, University of Kentucky, Lexington,

pp. 229-234.

Abstract: A computer program package is developed that intergrates and inter-

faces an urban growth simulation model, LANDUSE, and an urban hydrology model,

a modified version of STORM. Alternate growth scenarios can thus be directly

related to the corresponding storm-drainage systems. If these systmes are

-A69-

designed to achieve specified standards of performance, then a useful

comparision among several possible urban growth patterns can be performed.

While an urban area encompsses several natural watersheds, the hydrologic

models simulated one watershed at a time. The different watersheds that

partition an urban agglomeration create a tree-like or dendriform

configuration. The planning of a global storm drainage system for such a

conglomerate of basins can be, efficiently accomplished by a coordination of

the interactions among the different basins. A model for these interactions

is developed. The planning variables are the drainage network capacity, the

placement and size of the storage facilities, and the size of a central

treatment facility. An example of application is shown for a medium size

community in Indiana.

14. Diniz, E. V. 1979. Water quality prediction for urban runoff, an

alternative approach. Stormwater Management Model Users' Group Meeting

Proceedings, Project Officer, H. C. Torno, Washington. D. C.

Abstract: The accurate prediction of storm water quality resulting from

non-point sources of pollution has recently become a significant area of

interest. In general, storm water quality is a function of the total

pollutant accumulation on all surfaces exposed to rainfall. However, the

rainfall intensity and wash-off capacity of the pollutants are also important

factors. After, pollutants have been dislodged from exposed surfaces, overland

flow and channel hydraulics determine the pollutant concentrations evidenced

at a downstream observation of sampling site. Recent studies have indicated

that pollution accumulation rate are affected by several factors including

land use, population density, impervious area, traffic intensity, condition of

surface area, total overland flow length, time from last rain, street sweeping

-A70-

frequency, climate, and season. A water quality prediction strategy, as

normally formulated, requires the compilation of compilation of a data base, calibration of a

predictive model, simulation of existing and future conditions, analysis of

these predictions, and use of the results in planning decisions. Consequently,

the data base provides a foundation on which both the analysis and subsequent

decisions will be carried out. Because of the importance of valid data to the

overall strategy, numerous investigators have attempted to compile data which

could be used to develop predictive correlations between land use, watershed

characteristics, runoff, and water quality. The result of these efforts is a

very comprehensive data base for areas throughout the United States.

Unfortunately, the data collection and analysis procedures utilized by each

investigator were rarely uniform and consequently different sets of data have

to be carefully processed to be comparable.

15. Diniz, E. V. 1980. Quantifying the effects of porous pavements on urban

runoff. National Symposium on Urban Hydrology, Hydraulics, and Sediment

Control, Lexington, KY., pp 63-70.

Abstract: Porous pavements have been suggested as a means to reduce volume of

runoff as well as peak flows resulting from urbanization. The use of porous

pavements allows for infiltration into the ground from paved areas which would

otherwise be impermeable. Porous pavements can also be used to reduce the

overload on existing storm sewers. The effects on runoff quantity and quality

from porous pavements have been quantified by a modeling scheme which considers

the pavement and subgrade as two hydraulically connected control volumes

forming a single system. Inflow to the system are direct rainfall and an

overland flow hydrograph from contributing impervious areas. Outflows from

the system include vertical seepage, horizontal drainage, evaporation, and

-A71-

surface flow in the case of a surcharged porous pavement. Model input require-

ments include physical dimensions of the porous pavement, rainfall intensities,

and impermeabilities of the pavement, subgrade, and natural ground. A depth-

storage function for the pavement and subgrade is also necessary. A

comprehensive temporal accounting of flow and storage in each element of the

system is output from the model.

16. Freund, A. P., and C. D. Johnson. 1980. Comparison and relationships of

stormwater quality and basin characteristics: Madison, Wisconsin.

International Symposium on Urban Storm Runoff, C. T. Haan, Editor, University

of Kentucky, Lexington.

Abstract: The physical and hydrologic characteristics of three large and

diverse urban catchments in Madison, Wisconsin are examined and relationships

are developed between rainfall, discharge, sediment and nutrients using

regression analysis. The developed relationships provide a basis for comparing

the behavior of each basin as related to its physical characteristics. The

results of a street debris sampling program undertaken in each of the monitored

urban basins are reported, and implications for stormwater management are

briefly explored.

17. Gburek, W. J., and J. B. Urban. 1980. Storm water detention and

groundwater recharge using porous asphalt. Initial Results, International

Symposium on Urban Storm Runoff, Lexington, KY.

Abstract: Data collected and observations made during the first 2 years of

operation of the Willow Grove facility allow us to assess the potential of

porous asphalt pavement for storm water detention and groundwater recharge.

To date, the porous asphalt plot has produced no surface runoff from either

high-intensity or long-duration rainstorms to which it has been subjected;

-A72-

7.0 in/hr for 6 min (25 yr return period), and 0.37 in/hr for about 8 hr (5-yr

return period), respectively. Generally, 70 to 90% of the rainfall appears as

percolate below the plot on both the monthly and the individual storm basis,

although commonly no percolate appears during individual events of up to about

0.3 in. Groundwater beneath the porous asphalt plot responds relatively

rapidly to rainfall, usually within about 6 hours, and at the center of the

plot it rises approximately 5 ft. per inch of rainfall. A very localized

groundwater mound is formed by every storm that causes percolate to occurr, but

this mound forms and dissipates rapidly. Concentrations of both the critical

inorganic and organic water quality parameters within the percolate leaving

the porous asphalt plot are well below acceptable drinking water standards;

the percolate seems to pose no groundwater contamination threat. Field testing

of the strength of the porous asphalt plot showed that the plot as constructed

is able to support light to moderate traffic. Observations during severe

weather conditions indicate that the porous asphalt layer does not seem to be

affected by freeze/thaw conditions, and remains relatively skid resistant

during both wet and freezing weather. Finally, the initial results clearly

show that groundwater is recharged under the porous asphalt plot throughout

the year, whereas that under the adjacent grass cover plot is recharged little

or not at all during the growing season.

18. Grigg, N. S., L. H. Botham, L. Rice, W. J. Shoemaker, and L. S. Tucker.

1976. Urban drainage and flood control projects: Economic, Legal and

financial aspects. Hydrology Paper No. 85, Colorado State University, Fort

Abstract: Techniques for evaluating minor and major Urban Drainage and Flood

Control (UDFC) Projects are described. Economic, political, engineering,

-A73-

financial and legal problems must be faced prior to implementation of proper

levels of these projects. The measurement of tangible benefits is described

while a literature review revealed no direct objective techniques for

quantifying intangibles. However, some methods for establishing the relative

rankings of intangibles contributions show promise for improvement of evaluation

techniques. The legal problem of establishing benefits is described and a copy

of recent enacted Colorado legislation is included. Information on the

estimation of flood damages and the selection of discount rates is presented

for use by the analyst. Careful coordintion of land use and drainage control

measures is stressed. Related recent legislation and regulations are included.

19. Grigg, N. S., A. M. Duda, and J. Morris. 1980. Stormwater management in

coastal North Carolina, National Symposium on Urban Stormwater Management in

Coastal Areas, C. Y. Kuo, Editor, American Society of Civil Engineers,

New York, pp 33-34.

Abstract: The objective of this paper is to provide an assessment of the

magnitude of the stormwater runoff and flooding problem in the North Carolina

coastal zone, to describe existing management programs, and to provide

recommendations for improving stormwater management programs. The discussion

will be based on the practical aspects of a fragmented and pluralistic

approach to stormwater management in the coastal zone, including: three levels

of government, several categories of problems, and the multiple objective

approach to, problem solutions. The State of North Carolina approaches the

stormwater management problem realistically using the approach of seeking

multiple objectives through multiple means. We cannot expect that new

comprehensive stormwater management leg islation or extensive new management

programs will be approved by the General Assembly in the near future. The only

-A74-

altenative, is to make mid-course corrections in our various exising programs

at the local, state, and federal level to ensure that they work together and

work more effectively.

20. Guy, H. P. 1978. Sediment management concepts in urban storm water

system design. International Conference on Urban Storm Drainage.

P. P. Helliwell, Editor, University of Southampton, England.

Abstract: Storm drainage systems can be designed which will greatly reduce

peak rates of runoff and the amount of sediment and pollutants, normally

transported from urbanizing and urban areas to receiving water bodies.

Reduction in peak flow rates reduces the potential for serious channel

enlargement and additional sediment problems downstream from the development.

Optimum design can be achieved through good land-use planning that is well

coordinated with natural drainage. This in turn, will make it possible to

minimize excavation and soil exposure during construction, and provide a

maximum of individual and (or) community onsite storm water detention storage.

The resulting storm drainage system would usually have a lower initial cost

and result in a more esthetically pleasing neighborhood than generally exists

with conventional designs; may cause loss of convenience and be more

costly to maintain.

21. Hawkins, R. A., B. F. Maloy, and J. L. Pavon. 1979. Procedure for the

establishment of statewide wasteland allocations. Stormwater Management Model

Users' Croup Meeting Proceedings, Project Officer, H. C. Torno,

Washington, D.C.

Abstract: State regulatory agencies are faced with the problem of developing

effluent limitations for industrial, semi-public and municipal dischargers

which assure that water quality standards are met and that the economic burden

-A75-

of environmental protection is shared as equitably as possible among all

dischargers to a stream segment. In the past, emphasis has been placed on the

development of sound technical water quality modeling procedures and the

calibration and verification of the water quality models. The purpose of this

paper is to present the minimum treatment requirements which have been

established for dischargers by the United States Environmental Protection

Agency and a procedure which allows regulatory agencies to decide where limited

resources; for water quality sampling and field investigation should be placed

in developing defensible Waste load allocations.

22. Henry, J. G., and P. A. Ahern. 1976. The effect of storage on storm and

combined sewers. Canada-Ontario Agreement Research Program, Ottawa,

Ontario, Canada.

Abstract: The effect of storage on storm and combined sewers has been

investigated for a residential subdivision of approximately 100 acres area,

under development in Southern Ontario. The storage methods examined were:

(1) ponding of rainfall on flat roofs; (2) on-site storage of roof runoff;

(3) storage of flows from street gutters at catch basins; and (4) holding

reservoirs as part of the sewer network. A hydrograph model for estimating

stormwater runoff from residential areas has been developed and incorporated

in a computer program. The model was used to determine the outflow hydrograph

at the st sewer outfall from the subdivision for a synthetic two-year

design storm. An alternate design, replacing the separate sanitary and storm

sewers by combined savers, was carried out and the outflow hydrograph from a

combined sewer network was determined. The effects of different storage

methods on the outflow hydrograph were tested for both the separate and

combined systems. Substantial reductions in peak flow were found to occur as

-A76-

the level of storage was increased, with holding reservoirs providing the

greatest reductions. Cost data from the original separate system design were

used to calibrate formulae for estimating sewer pipe costs. The total costs

of systems incorporAting various storage options were compared. A combined

sewer system was found to be slightly less expensive than the separate system

and small savings in subdivision sewer costs resulted when storage was

incorporated.

23. Jackson, Thomas J., and Robert M. Ragan. "Hyfrology of Porous Pavement

Parking Lots,", Journal of the Hydraulics Division, ASCE, Vol. 200, No HY12,

Proc. Paper 11010, December, 1974, pp. 1739-1752.

Abstract: Numerical solutions of the Boussinesq equation were used to examine

the hydrologic behavior or porous pavement systems incorporating subdrains in

open graded base courses placed on impermeable membranes. A series of

numerical experiments showed that substantial control of the runoff hydrograph

from parking lots could be obtained through the use of porous pavements. The

numerical experiments conducted with synthetic design storms were used to

develop equations and graphs for use by engineers designing porous pavement

systems for runoff control.

24. Kamedulski, G. E., and R. H. McCuen. 1978. The effect of maintenance on

storm water detention basin efficiency. Water Resources Bulletin,

14 (4) : 1146-1152.

Abstract: Storm water detention is an effective and popular method for

controlling the effects of increased urbanization and development. Detention

basins are used to control both increases in flow rates and sedimentation.

While numerous storm water management policies have been proposed, they most

often fail to give adequate consideration to maintenance of the basin.

-A77-

Sediment accumulation with time and the growth of grass and weeds in the

emergency spillway are two maintenance problems. A model that was calibrated

with data from a storm water detention basin in Montgomery County, Maryland,

is used to evaluate the effect of maintenance on the efficiency of the

detention basin. Sediment accumulation in the basin caused the peak reduction

factor to decrease while it increased as vegetation growth in the emergency

spillway increased. Thus, the detention basin will not function as intented

in the design when the basin is not properly maintained. Thus, maintenance of

detention basins should be one component of a comprehensive storm water

management policy.

25. Kamedulski, G. E., and R. H. McCuen. 1979. Evaluation of alternative

stormwater detention policies. American Society of Civil Engineers, Journal

of Water Resources Planning and Management Division, 105(WR2): 171-186.

Abstract: Mathematical modeling is a quick, in Densive and effective weans

of evaluating existing stormwater detention policies and examining alternative

policies. The model was then calibrated with data collected from a watershed

located in Montgomery County, Maryland. The data base included measured

precipitation, and both sediment and flow into and from the detention basin.

Model studies indicate that many existing SWM policies do not meet the intent

of SWM because they ignore the volume-duration frequency concept that has long

been used in hydrologic design. SWM policies often fail to specify a duration

for detention basin design and limit the design frequency to a single return

period. A detention that is designed using a policy based on a single return

period will not limit the entire flood frequency curve to that of the

before-development conditions, and thus the policies do not satisfy the intent

of a SWM. Also, less costly designs are possible if both basin volume and

outlet characteristics are considered simultaneously.

-A78-

26. Krishnamurthi, N., and J. L. Balzer. 1978. Design-storm for

sedimentation ponds. Environmental Quality Department Utah International

Inc., San Francisco, California.

Abstract: The office of Mining and Reclamation has adopted a 10-yr,

24-hour precipitation event as the design storm for sedimentation ponds to

limit sediment concentrations in effluents. The authors feel that the

design-storm approach using the frequency analysis is seldom warranted often

leading to overdesign and undue costs. This paper discussed the physical

process of precipitation that cause runoff and how it should be considered in

developing the criteria for the design of sedimentation ponds. The authors

recommend that the Office of Surface Mining revise the design principles of

sedimentation ponds both for environmental and economical reasons and that

they increase emphasis on limiting the sediment volume flowing as a mining

facility rather than limiting its concentration in the effluent. It is their

hope that these recommendations could influence EPA effluent regulations in

the future.

27. Krishnamurthi, N., and W. T. Lenocker. 1980. Comparison of HEC I and

TR 20 Programs. National Symposium on Urban Stormwater Management in Coastal

Areas, C. Y. Kuo, Editor, American Society of Civil Engineers, New York, I

N. Y., pp 92-98.

Abstract: Determination of peak flows in an ungaged watershed has been of

interest to thepracticing engineers due to the increased interest on flood

plain management programs at the local, state and federal levels. For the

past two decades, research institutions and academic universities have

developed many rainfall-runoff simulation models for the ungaged watersheds

-A79-

and some Of the notable and commonly used models are: (1) Stanford Watershed

Model, (2) HEC 1, (3) U.S.G.S.-Dawdy Model, (4) Tr 20, (5) USDAHL, and

(6) ILLUDAS. While these models are well documented relative to the purpose

of programs, identification of variables and explanation of input and output

procedures, very little work is done on the comparision of these models. The

objective of this paper is to compare two of the above mentioned models,

name'.- HEC I and TR 20, with respect to their accuracy and resources needed.

program was developed by the Hydrologic Engineering Center of U.S. Army

Corps of Engineers to handle flood hydrograph computations associated with

precipitation and runoff on a complex, multisub-basin, multichannel river

basin (1). TR 20 program was developed by the Soil Conservation of U.S.

Department of Agriculture to compute surface runoff resulting from any

synthetic or natural rainstorm (2). Detailed descriptions of these programs

and of imput parameters are explained in their user's manuals (1, 2).

Based on preliminary testing, the authors feel that if rainfall/runoff data

are available, HEC 1 program is best suited to calibrate the input parameters

of a rainfall/runoff simulation and to predict the peak flow of floods.

However, Snyder and Clarks' coefficients for HEC I program are not readily

available for many ungaged watersheds. For such watersheds TR 20 programs can

be best utilized through the use of soil survey maps.

28. Lai, C. 1980. Urban storm sewer flows in coastal areas. NationaL

Symposium on Urban Stormwater Management in Coastal Areas, C. Y. Kuo, Editor,

American Society of Civil Engineers, New York, pp 244-254.

Abstract: Analyses and management of urban stormwater in coastal areas have

become increasingly impportant subjects in recent years. Urban basins are

generally small in size and generally consist of a large percentage of

-A80-

impervious surface Many sewers coastal areas are characterized by small

condult slopes and by tide affeccted flows. Additional features, such as rapid

change of discharge due to subtropical rainstorms, are common to the sewers of

Southern coastal areas. Storm rivers in the Miami., Florida area exhibit a

numbcr of thcse flow, characteristics typical of urbanized southern coastal

areas. Accurate simulation of this class of flows remains a difficult subject,

and general, rational and reliable deterministic methods for solving such flow

problems are needed. The purpose of this papaer is to examine and explore

methods for analyzing the transient flows of urban storm sewers in southern

coastal areas. The study considers three general cases: a) full-pipe or

closed conduit flow, b) free-surface or open-channel flow, and c) two-phase or

mixed flow. For ease of formulating a workable approach, storm sewer

configurations in Southeast Florida are used to develop the numerical solution

technique. Numerical schemes for simulating these three types of low have been

investigated, and a computer model for full-pipe flow has been applied to a

storm sewer flow in South Miami, Florida. The simulated discharge hydrograph

generally agrees well with the measured one; however, some refinement in low

discharge analysis seems to be necessary. The use of rigid water column theory

foc full pipe flow is adequate for most sewers. "Minor losses" contribute
significantly to hydraulic lost in the flow simulation. A U-shaped flow

control device may be used advantageously to obtain a point discharge in a

sewer pipe for the open-channel flow regime. In fact, some U-shaped

constrictions have been installed for this purpose in south Florida sewers by

the U.S. Geological Survey.

-A81-

29. Labitos, D. F. , and K. C. Wsell. 1978. Computer simulation of.storm

drainage system: A case study based on the use of the Penn State runoff model.

international Symposium on Urban Storm Water Management, C. T. Haan, Editor,

University of Kentucky, Lexington, pp 1-12.

Abstract: Successful management of urban water resources depends on the

ability of urban planners and managers to predict accurately the effects that

increased urban development will have on storm water runoff. The- lack of

simple methods for prediction of watershed response to storm events is a major

factor contributing to increased urban flooding. The Penn State Runoff Model

is simple and concise storm water simulation program, developed for the

purpose of analyzing the timing of subarea flow combinations and their effect,

on downstream flow rates. Information on the manner of combination of subarea

flows provides the basis for evaluation of flood-control altenatives for the

source of the flooding problem rather than the point of actual flooding. The

Peak Flow Presentation Table, a major output of the model, lists the

subwatershed flows that combine to form a flood flow downstream. The timing

of peak flows from individual subareas determines the magnitude of aggregate

flow downstream, which in turn determines the extent of flooding. The results

of a storm drainage study illustrates the practicality of the Penn State Runoff

Model in developing realistic control alternatives for storm-related flooding

problems. This study involved development of detailed schematics of the storm

drainage system and accumulation of ."Low and rainfall data required for use of

the model. Evaluation of the various runoff -control alternatives revealed

some situations where conventional control measures are required to alleviate

the immediate problems. The planning and analysis had to be sufficiently

comprehensive to insure that the control program developed for the Town did

not place an unwarranted runoff burden on downstream properties.

-A82-

30. Lockwood, G. 1980. New approach to storm drainage pipe design. Civil

Engineering, pp. 59-61.

Abstract,: Using a recently developed computer program, drainage swales can be

designed to provide short-term storage of stormwater. Benefits of this

approach to stormwater management include reduced sewer pipe size requirements,

minimal pollution of streams and lakes from run-off, reduced flooding during

peak storm periods and, if desired, a recharged aquifer.

31. Looper, R. D. 1980. Calibration and application of a water,shed planning

model. National Symposium on Urban Stormwater Management in Coastal Areas.

C. Y. Kuo, Editor, American Society of Civil Engineers, New York, pp 152-16

Abstract: Accurate determinaton of the storm water runoff process is critical

to the overall effectiveness of a stormwater drainage masterplan. For a storm-

water drainage masterplan consisting of sixteen basins in Pineillas County,

Florida, The Soil Conservation's TR-20 computer model was selected. This model

was formulated to improve the quality of watershed project planning by pro-

viding a large degree of flexibility in developing input parameters and by

quickly evaluating alternative drainage systems. However, in order to apply

this model to a coastal area, calibration of various input parameters )ad to

be accomplished. The TR-20 model utilizes synthetic unit hydrograph theory

which use input parameters based on the physical characteristics of a

watershed. Several of these pararmeters, such as rainfall, infiltration, soil

moisture capacity and time of 2 concentration required special attention due to

the topography and-hydrology of coastal areas. Verification of TR-20

parameters was accomplished through use of the continuous event model, STORM,

and the single event model, HEC-1. This paper will present the calibration

and application of the TR-20 watershed planning model to Alligator Creek Basin

-A83-

in Pinel1as County, Florida. problems encountered with the HEC

pertain mainly to the loss rate parameter relationship with STORM and TR-20-

By reconstituting- observed runoff events, HEC-1 Must Optimize low rate

parameters given estimates of starting values for input variables. These loss

rate parameters are not directly comparable to those used in the SCS curve

number runoff analysis. The principal difficulty in using the TR-20 model is

encountered when trying to apply the model to develop basings. Although TR-20

provides a routing procedure for reservoirs and channels, closed conduit

systems are not provided for This limitation excludes T2-20 from being used

in the analysis of s ered areas.

32. Manz, P. E. 1980. Development of a dynamic stormwater management

system. National Symposium on Urban Stormwater Management in Coastal Areas.

C Y. Kuo, Editor, American Society of Civil Engineers, New York, pp 163-172.

Abstract: Conventional stormwater drainage masterplans are generally

developed to provide both a solution to existing flood problems and to provide

adequate drainage for anticipated future development in a watershed. The

evolution and selection of drainage
improvements ace based on Ultimate and

Use (ULS) conditions, as envisioned at the present time Any changes to the

ULU conditions of a drainage basin or revision of the recommended improvement

to, a channel in the basin can the stormwater masterplan obsolete or

inadequate. Extensive revisions and lengthy compatations are,required

update the masterplan and maintain it as an effective tool in solving drainage

problems. A systems approach was used to develop dynamic stormwater drainage

masterplans for Pinellas County, Florida. Computer programs were obtained or

developed to digitize the basin, calibrate and develop the hydrology and model

the hydraulic characteristics of the basin. The strength of a dynamic

-A84-

masterplan lies in its flexibility after the final masterplan is submitted.

The client is provided with program decks on the computer models used and

copies of all input files. As land use patterns change, the masterplan can be

updated and checked to see if the alternatives are still adequate. If the

alternatives prove inadequate or other changes are proposed, they can be

quickly" evaluated and incorporated into the masterplan. In this way the

client always a current masterplan that can be an effective tool for use

in urban planning. The advantage of dynamic stormwater masterplans include;

1. The ability to maintain a current masterplan by incorporating changes in

the land use pattern. 2. The ability to quickly evaluate proposed changes to

the masterplan at a specific 1ocation. 3. The ability to quickly evaluate

downstream responces to changes in the masterplan. 4. The ability to obtain

water quality indications of the proposed changes.

33. Mariles, 0. A., J. L. S. Bribiesca, and R. D. Mora. 1978. A numerical

procedure to design drainage networks based on the hydraulic and storage

capacities of the pipes. International Symposium on Urban Storm Water

Management, C. T. Haan, Editor, University of Kentucky, Lexington, pp 333-346.

Abstract: A numerical procedure based on finite difference computations has

been developed by the authors to calculate, at any time and at any site of a

given network, the hydraulic parameters, i. e., velocity and depth. By this

way and baking into account the storage at registers and manholes, it is

possible to simulate the whole work of the network and to know the volume, of

storm water that is not possible to manange inside the pipes. Since such a

volume will produce a flood and cause a certain damage, it is possible to

associate the damage for several storms with different return periods with the

expected damage for a certain desig. In such way the expected damage

-A85-

associated to each design can be known. If the benefit of a design is assumed

to be the expected avoided damage at actual conditions and with a certain

design, two graphs can be drawn; the first one will be the cost of the design

against its capacity and the second will be the benefit associated against its

capacity. The selection of the right design becomes obvious.

34. Nattraw, H.. C. Jr., J. Hardee, and R. A. Miller. 1978. Urban stormwater

runoff data for a residential area, Pompana Beach, Florida. U.S. Geological

Survey open-file Report 78-324, Tallahassee, Florida, 108 pp.

Abstract: The U.S. Geological Survey has measured rainfall, runoff, and

runoff quality for three urban sewered basins in Broward County, Florida.

Approximately 2 years of records have been collected for a 47-acre single-

family residential area; a 58-acre, 3,000 foot secondary divided-highway

segment; and a 28-acre commercial shopping center. The three homogeneous

areas were gaged with an automatic, integrated instrumentation package or

urban hydrology monitor. The urban hydrology monitor collects and

synchronously records 36-second interval rainfall and water-level information

The collection time of 24 water-quality samples is also recorded. Discharge

is computed from the difference in stage through a U-shaped construction

(Venturi flume) placed in a sewer pipe. Approximately 100 rainfall-runoff

periods per site were digitized from analog records and stored in an urban

data management system. Sets of nutrient and heavy-metal water-quality data

collected for 30 or more storms at each of the three areas and stored in

the system Loads computed for the three areas indicate the importance oF. the

hydraulic interconnection between impervious areas. Factors which affect

storm-water runoff loads include landuse, proportion of hydraulically

interconnected impervious area, seasonal distribution of rainfall, and the

-A86-

antecedent dry period. The variability within a basin is great, indicating

the need for systematic collection of runoff-quality information for a variety

of conditions prior to any satisfactory calibration of currently available

mode 1 s.

35, McCuen, R. H. 1980. Water quality trap efficiency of stormwater

management basins. Water Resources Bulletin, 16(l):15-21.

Abstract: While the quality of rivers has received much attention, the

degradation of small streams in upland areas of watersheds has only recently

been recognized as a major problem. A major cause of the problem is increases

in nonpiont source pollution that accompany urban expansion. A case study is

used to examine the potential for storm water detention as a means of

controlling water quality in streams frequently used to control increases in

discharge rates, can also be used to reduce the level of pollutants in inflow

to receiving s treams. Data collected on a 148-acre site in Maryland shows

that a detention basin can trap as much as 98 percent of the pollutant in the

inflow. For the 11 water quality parameters, most showed reductions of at

least 60 percent, depending on storm characteristics.

36. McCuen, R. H. 1979. Downstream effects of stormwater management

basins. American Society of Civil Engineers, Journal of Hydraulics Division,

105(HYII): pp 1343-1356.

Abstract: Urbanization decreases the natural storage of a watershed, which

changes the timing characteristics of the runoff. Stormwater management (SWM)

basins are an attempt to put the storage lost through development back into

the runoff process. While SWM basins provide the proper volume of storage,

they fail to return the timing characteristics to those that existed prior to

development. The changes in storage and timing characteristics may have

-A87-

adverse effects on flow rates and bedload transport in downstream channel

reaches. A study of a 2.12-sq mile watershed in Montgomery County, Md.,

showed that a SWM basin increased both peak flows and bedload transport rates

in the channel downstream from the facility. This results from both the

change in timing , characteristics and the increased duration of bankfull flow.

These results iindicate that SIAM policies must require evaluation of SWM plans

on a regional basis and not just using on-site control criteria. Also,

methods that can evluate storage and timing changes must be used in the

design of SWM facilities.

37. McCuen, R. H. 1979. Stonwater management policy and design. Journal.

of Civil Engineering Design, (1): 21-42.

Abstract: Stormwater management is recognized as a requisite to controlling

storm runoff from developing areas. To best meet societal needs, numerous

policies have been adopted to meet the intent of stormwater management.

Unfortunately, many policies are deficient in their failure both to lead to

designs that meet the intent of stormwater management and to provide the

proper guidelines for translating policy into a design that provides the

maximum benefit to society. Specific deficiencies of many policies include:

1) the use of a single frequency; 2) neglect of storm duration; 3) inadequate

consideration of maintenance; 4) insensitivity to the importance of soil

characteristics; 5) lack of recognition of differences between water quality

and quality control; and 6) lack of consideration of downstream effects of

detention storage. The effect of these policy deficiencies were evaluated.

Different hydrologic methods are being used, to translate policy into design,

including graphical, empirical, unit hydrograph, and conceptual models. These

methods provide different designs and differ on other important criteria,

-A88-

including accuracy, training requirements, design cost, total cost, and their

sensitivity to policy components. The different hydroloic methods are

38. Moodic A. R., J. D. Scholes, and D. G. Thompson. 1978. Design of a

stormwater quality and quantity monitoring system for an urban catchment - An

Australian case study. International Symposium or. Urban Stormwater Management,

C. T. Haan, Editor., University of Kentucky, Lexington, pp 135-142.

Abstract The resultof a prelimainary investigation into stormwater runoff

quality from an urban catchment in Melbourne, Australia, are presented and

discussed. Similarities with the results obtained by workers in the USA are

noted. The results of the preliminary investigation are shown to be useful in

designing and operating a stormwater quality and quantity monitoring system

for the catchment. Aspects of the design of this system, such as the

selection of sub-catchments for instrumentation and the design of a device for

constant proportional depth sampling in a large Parshall flume, are also

outlined.

39. Myers, V. A., and F. P. Ho. 1980. Coastal tide frequency, Virginia to

Delaware. National Symposium on Urban Stormwater Management in Coastal

Areas. C. Y. Kuo, Editor, American Society of Civil Engineers, New York,

pp 74-82.

Abstract: More than 2000 of United States coast are at a risk from

temporary inudations by storm-driven sea water. Judicious decision on the

part of the coastal zone that shall be occupied and on rational use of this

occupied part, on lessening risk to life and property by structural means, on

routes and refuges that will be needed for evacuation when a major storm

threatens, and on flood damage insurance rates, requires knowledge--in

probability terms--of the elevation and lateral extent of the space that

-A89-

storm-driven sea water- is likely to occupy. The behavior of coastal storms is

ultimately assessed from past experience. The longer the return period of

flood heights to be determined, the larger the experience sample required.

Local experience with hurricanes does not necessarily reveal the true risk.

In this study, "experience" is enhanced by applying all the data of a region

to each coastal point, by deriving data needed (storm tides) from data

available (meteorlogical parameters), by computation with tested models, and

by developing hypotcal storms that collectively represent the full, range

of possibilities . This summary P er illustrates this approach to coastal

ride frequency , using the coastfrom the Virgina- North Carolina

border to Delaware Bay ("Delmarva" coast) as the example.

40. Nawrocki, M., and J. Pietrzak. 1976. Methods to control fine-grained

sediments resulting from construction activity. Hittman Associates, In,..,

Columbia, Maryland.

Abstract: This publication is the third in a series issued under Section

304(e)(2)(C) of Public Law 92-500 concerning the control of water pollution

from construction activity. This document was prepared for use by planners,

engineers2 resource managers, and others who may become involved in program a

to effectively provide for sediment control. Standard erosion and sediment

control measures are usually effective for preventing the runoff of the total

sediment load. The effectivenss of these standard techniques, however, has

been found to be relatively poor with regard to preventing the runoff of the

fine-grained fractions, such as the silts and clays. The objective of this

study was to research practical, cost effecive methods that would help to

reduce specifically the fine-grained sediment pollution derived from

construction activities. The prime consideration during this study was the

-A90-

use or adaptation of exisiting technology, as described in the, current

literature or data, to the finegrained sediment pollution problem.

41. Novitzki, R. P. 1978. Hydrology of Wetlands in Wisconsin. U.S.

Department of the Interior Geological Survey, Water Resources Division,

Madison, Wisconsin.

Abstract: Interest in the role of wetlands in the hydrologic system has

intensified in response to increased environmental concern. However, little

information is available concerning those hydrologic factors that influence

wet lands and conversely, the effects of wetlands on the hydrologic system. To

manage wetlands as a significant element in the total cnvironment will require

a better under standing of their function in the hydrologic system and their

effect on the quality of water moving through them. The initial purpose will

be to classify wetlands according to their function in the hydrologic system.

In particular, the effect of the wetland on runoff, sedimentation, and

chemical characteristics of lakes or streams and the wetland's relation to

water levels and water quality in the ground-water system will be defined.

Several wetland areas, each representative of one wetland category, will be

studied in detail. The hydrologic system near each wetland will be defined to

determine direction of water movement, both vertical and horizontal, to

determing areas of recharge and discharge to determine the function of the

wetland in the local. system, and to define the influence of the wetland on

flood flows and low flows and sedimentation in associated streams in the local

system. The chemical nature of water entering, within, and leaving the

wetlands will be examined to determine the effect of the wetland on water

quality.

-A91-

42. Novitzki, R. P. 1977. Hydrology of the Nevin fresh-meadow wetlands.

U.S. Department of the Interior, Geological Survey, Madison, Wisconsin.

Abstract: Wetlands have been vanishing from the Wisconsin landscape at a

rapid late Approximately half of the original 5,000,000 acres have been

drained or affected by drainage. The fresh-meadow wetlands have been

vanishingh most rapidly because of agricultural and urban pressures. To

preserve wetlands in the future, the Wisconsin Department of Natural Resources

meds prohability to evaluate the tangible- values of wetland areas. With such

an evaluation, they (DNr), may be able to predict the type, and magnitude of

disturbance that will be caused by proposed wetland changes. This project

will attempt to define the hydrologic system that presently maintains the

nevin wetland and to monitor and explain changes which may occur. The water.

quantity and-quality entering and leaving the area will be evaluated with

respect to the influence of land-use changes, drainage nutrients, sediment,

municipal and local pumpage, and the effects of the hatchery. Data

collections will include spring flows into wetlands, streamflow leaving

wetlands, water level, precipitation and evaporation. Water samples for

analysis of sediment, pesticides, and total organics be collected from

water entering and leaving welands and from ground water in shallow aquifers

in the study area. Inventory of hydrology and geology influencing wetlands,

and analysis of quantity and quality of water moving through wetlands and its

related drainage basin will be made. Changes; in quantity and quality detected

by monitoring will be useful in evaluating those factors which may be

affecting the wetlands.

-A92-

43. Patrick, W.W 1981 Effect of man's alterations on wetland and

estuarine chemistry. Agricultural and Mechanical College, School of

Agriculture, Louisiana State University, Baton Rouge.,

Abstract: Objectives are (1) to determine if specific lakes of Barataria

Basin are a source or sink for nitrogen and phosphorus; (2) to characterize

water intefaces in shallow bays, lakes, and

channels; (3) tp determine the significance of sedimentation throughout

Bacteria Basin in relation to, nitrogen,phosphorus, and carbon chemistry; (4)

to determine, the importance of mineral components of organic marsh soils in

to continue studying
neutralizaing plant toxins such as sulfide: and (5)

nitrogen reactions in, wetland system nitrogen fixation in water-sediment and

plant-soil systems using labelled nitrogen, and denl iofication in natural
systems. The information obtained from this study will be used by public

agencies concerned with management of coastal lands and resources, in problems

dealing with eutrophication, agricultural runoff, wastewater management, water

quality, primary production, etc., of Louisiana's coastal zone. Identified

benefits, to date are (1) a trophic state index based on cluster analysis and

factor analysis of several aquatic parameters has been developed; (2) a study

has examined the relationship between chemical and physical properties of

marsh soils and the intriguing variations in plant growth; (3) demonstrated

the significance of sediment_ transport to marsh accretion and nutrient

cycling; and (4) develop a better -understanding of the nitrogen cycle ;in a

Sparti alterniflora salt marsh.

44." Pennel, A. B. 1980. Retention/detention basins in coastal areas.

National Symposium on Urban Stormwater Management in Coastal Areas, C. Y. Kuo,

Editor, American Society of Civil Engineers, New York, pp 299-302.

-A93-

Abstract: Stormwater managementt is an integral part of urban planning and

development. Urban growth translates directly to an increase in the precent

of impervious surfaces that generate higher runoff rates and columes over a

shorter time period than previously generated from the undeveloped condition.

The traditional role of stormwater management was to remove the additional

water from the developed site as, rapidly and efficiently as possible to

minimize the flood hazard and inconvenience. Continued growth and development,

comp1eting with a renowed awareness, of environment issues and concerns, has

led to a re-evaluation of stormwater management practices. Present stormwater

management practice:, are moving to multiple objective concepts. These

generally include the traditional flood protection along with the preservation

or maintenance of the existing flow quality and quantity. Use of Ole

retention/detention basin system serves as one of the basic tools in achieving

these objectives. The practicality of retention/detention basins in coastal

areas is a function of the optimization of design parameters relative to the

value served. The design parameters of particular influence in this regard

are groundwater table elevations and tailwater stages. Utilization of a model

that accurately accounts for these parameters is an economical and timely

manner is paramount to fulfilling the obligations and goals of stormwater

management in coastal areas.

45. Peterson J. 0., G. Bubenzer and F. Madison. 1977. Evaluation of the

implentation of the White Clay Lake protection and management plan. School

of Agriculture and Life Sciences, University of Wisconsin, Madison.

Abstract: Implementation of the plan to provide protection of the water

quality of White Clay Lake presents an opportunity to assess the effectiveness

of protecting water quality by applying simple sediment and nutrient control

-A94-

measures in an agricultural watershed. Techniques such as clear-water

diversion, terracing, contour stripping, better manure handling systems, and

stream bank protection are useful soil and water conservation practices, but

institution of regulations to require application of such techniques on a

broad basis should be preceded by an evaluation of their effectiveness for

water quality protection major chang in water quality resulting from

implementation of conservation practices will be most easily noted in the

incoming streams. Continuous monitoring of two streams which drain about 45%

of the watershed provides two years of background data on water flows,

nitrogen, phosphorus, chloride and sediment transport. Evaluation of the

management plan implementation involves a continuation of current lake and

watershed monitoring activities plus some additional work using sediment cores

to delineate the history of lake infilling. Additionally, evaluation of the

effect which marsh areas have on the quality of water entering the lake will

help guide judgments about-the preservation of these wetlands for protection

of lake water quality.

46. Pitt, R. 1978. The potential of street cleaning in reducing nonpoint

pollution. International Symposium on Urban Storm Water Management,

C. T. Haan, Editor, University of Kentucky, Lexington, pp 289-301.

Abstract: This paper briefly describes the conclusions available at this tim e

from an KPA-sponsored demonstration study of nonpoint pollution abatement

through improved street cleaning practices. An important aspect of the study

was development of sampling procedures to test screet cleaning equipment

performance, in real-world conditions. These procedures and the experiment

design are described in detail in the interim report. This paper summarizes

accumulation rate characreristics of street dirt in the area studied. The

-A95-

rsults of performance tests for street cleaning equipment and the factors

that are thought to affect this performance are also summarized. These data

are used to draw preliminary conclusions about elements that must be

considered in designing an effective street cleaning program. The study of

urban runoff yielded information on overall flow characteristics,

concentrations and total. mass yields of monitored pollutants in the runoff,

and street dire removal capabilities and effects on deposition in the sewerage

for various, kinds of storms. These data are summarized here, and urban runoff

water quality is compared with recommended water quality criteria and the

quality of sanitary wastewater effluent. Costs and labor effectiveness of

street cleaning, runoff treatment, and combined runoff and wastewater

treatment ace also compared, and preliminary results from a study of airborne

dust losses from street surfaces are summarized.

47. Ports, M. A. 1978.. Erosion and sediment control design. Watershed

Permits Section, Maryland Water Resources Administration, Annapolis, Maryland.

Abstract: Since soil erosion and sedimentation by water are complex processes,

a better understanding of them provides a sound basis for developing improved

prediction and control methods for urban areas. The Universal Soil Loss

Equation is an important technique for evaluating urban erosion rates, and

various conservation practices are effective for controlling urban erosion and

sedimentation. Design of control plans requires the selection of appropriate

control practices. These topics are discussed and illustrated as they relate

to solving urban problems in erosion and sedimentation.

48. Kitt, R. A., and R. Johnson. 1978. On-site stormwater retention.

International Symposium on Urban Scorm Water Management, C. T. Haan, Editor,

University of Kentucky, Lexington, pp 91-94.

-A96-

Abstract: Many urban areas have tried various stormwater detention and

controlled release schemes to decrease peak flow rates for many reasons,

including lower costs for smaller storm sewers and decreased downstream

flooding. However, detention schemes have some disadvantages,,which include

land requirements and safety problems. An alternate method for reducing storm

sewer requirements is to use retention, and reroute part ofthe flow that

would ordinarily be urban runoff into the soil for subsequent percolation into

the ground water supply. Onsite retention methods can provide the same,

advantages of peak flow reduction as detention without the extra land

requirements. Most stormwater runoff in developed areas is from

cover. Reduction of runoff, both volume and rate, can be accomplished by

retaining a portion of the runoff from roofs using porous media

drains." Modifications were made to the Runoff Block of the Storm Water

Management Model (SWMM) from the Environmental Protection Agency. The

computer program was changed to allow a fraction of the runoff from impervious

surface to be immediately directed to onside retention units of specified

volume. Overflow from the retention units during extreme storm periods was

onto adjacent pervious surfaces. Using a hypothetical residential drainage
area, retention volumes of 500 ft 3/acre and 1000 ft 3/acre were studied.

The fraction of impervious area was, set at 0.30 and one-half of this area

drained into the retention storage. The modified SWMM computer program showed

that for a 5-year, I-hour storm, the total runoff volume with no retention was
3 3 3
363 ft which was reduced to 256 ft and 218 ft for 500 ft /acre and
1000 ft3/acre of onisite retention respectively- For the same storm even

the sum of the peak runoff rates of 132 cfs with no retention was reduced to
95 cfs with 500 ft 3/acre and 79 cfs with 1000 ft 3/acre. Reduction of this

magnitude could have a large impact on stormwater management strategies.

-A97-

49. Reiter,F, and C. Fox, Jr. 1976. Evaluation of the cost-effectiveness

of non-structural pollution controls: A manual for water quality management

planning. CONSAD Research Corporation, Pittsburgh, Pennsylvania.

Abstract: The report presents a methodology for evaluating the economic costs

of nonstructural pollution control techniques such as land use control. Used

instead or as a complement to structural pollution controls, nonstructural

alternatives can increase the effectiveness and or decrease the costs of

achieving environnmental quality objectiveness. in the report are

chapters or legal social impacts, and alternative compensatory

mechanisms for nonstructural controls.

50. Schluchter, S., and M. Teubner. 1960. Coastal flooding of tidal

estuaries. National Symposium on Urban Stormwater Management in Coastal Areas,

C. Y. Kuo, Editor, American Society of Civil Engineers, New York, pp 275-278.

Abstract: At the present time, several different two-dimenisional surge models

are being used by Federal and private agencies to predict the magnitude of

hurricane surges along the open coast; however, surge levels in estuaries can

vary significantly from open coast values. Flood levels in these estuaries

can be accurately and economically determined using a one-dimensional unsteady

flow model. This paper considers such a model which includes both dynamic and

storage effects associated with the channel flow and the tidal flats, as well

including the effects of miltiple-connected river junctions. An area along

the Texas coast has been selected to demonstrate the model's capabiiities.

-A98-

In this paper, one-d imensional unsteady flow model has been briefly

described. Some of the features of the model include:

(1) its capability of being able to handle a large number of rivers,

tributaries and canals at one time;

(2) its adaptation to using more accurately determined input data

involving cross-sectional parameters;

(3) Its adaption to uising predetermined elevaluation hydrographs, either from

in associated two-dimensional model or from measured data, as input

elevations.

Unfortunately, both the original one-dimensional model and the combined model

suffer from one main problem: insufficient surge data for calibration and

verification. It is relatively easy to recreate existing tidal conditions,

but to accurately determine storm surge elevations, more detailed data is

required.

51. Simons, D. B., R. M. Li, and T. J. Ward. 1978. Estimation of sediment

yield for a proposed urban roadway design. International Symposium on Urban

Storm Water Management, C. T. Haan, Editor, University of Kentucky, Lexington,

pp 309-315.

Abstract: Roadways are one of the primary sources of sediment in urban

environments. Soil erosion and the resultant sediment yield are important

problem- during construction and post-construction periods associated with the

development of transportation road surfaces. Increased soil erosion and

sediment yield often result from scarification of the natural terrain and

changes in drainage patterns caused by urban road construction. The increased

sediment loading to streams and drainages from roadway construction affects

water quality, riparian and aquatic ecosystems, and flooding potential due to

-A99-

aggradation of channels. Techniques have been developed for estimating
erosion and sediment yield from roadways. One approach used complex computer

simulation of physical processes that determine erosion and sediment yield.

Although a useful technique, complex computer simulation is often difficult to

comprehend for many field users. To avoid problems associated with the

complex approach, a simpler and more readily understood physical process

computer simulation model has been developed. This simplified model is based

on the same physical processes as the complex model but integrates and

averages the processes to develop a physically realistic solution. In this

paper,the, simplified version of a roadway sediment yield model is presented

along with applications to selected roadway designs. Using this model and

available, roadway data, a user can quickly estimate sediment yield from a

proposed roadway design. Roadway components considered in the model include

the road surface, cut and fill slopes, drainage ditches and culverts. Impact

of design alternatives such as a roadway cross section, road gradient, road

surfacing, and size and spacing of cross drains can be determined by the

model. Use of the simplified model in road design can alert the urban

planners and roadway design engineers to potential erosion and sediment

problems and help the designers avoid such problems.

52. Slyfield, R. E. 1978. Management of urban runoff by remote sensing in

south Florida. International Symposium on Urban Storm Water Management

C. T. Haan, Lditor, Unversity of Kentucky, Lexinton, pp 235-244.

Abstract: A continuous urban area 100 Tiles long by 10 miles wide borders the

south east coast f Florida. Because of the unique character of this area -

very flat topography, average elevation of about 5 feet above sea level, and

very intensive tropical rainfall - immediate water management decisions are

-A100-

required. Consequently, the South Florida Water Management District which

operates the areas's primary water management facilities is constructing a

remote sensing and control system. This system transmits vital hydro-

meteorological data via a terrestial telemetry system to a control center

where management decisions are made. Operational directions are then

transmitted over the system to control facilities and executed remotely.

Simultaneously these actions are monitored at the control center. Because the

system's main function is to perform during periods of high stress, design

requirements have been adapted to ensure that it will be operational when

needed most. Various innovative features are described which increase

reliability. The system consists of a miscrowave backbone in a loop

configuration and capable of fully duplex transmission under computer control.

Remote acquisition and control units (RACU's) contain environmental sensors

and control elements which operate water control facilites. RACU's also

contain transceivers powered either by commercial or solar power. Many

sensors are of unique design, and high accuracy requirements have been

verified.

53. Smolenyak, K. J. 1980. Stormwater runoff modeling of a planned

community. National Symposium on Urban Stormwater Management in Coastal Areas,

C. Y. Kuo, Editor, American Society of Civil Engineers, New York, pp 378-387.

Abstract: The Tampa Palms development, which is located approximately eleven

miles northeast of downtown Tampa, is a residentially oriented master planned

community totaling 5400 acres in size with an ultimate density of 13,000

units. Stormwater discharges from this property will enter the Hillsborough

River, a pot-able source of water for the City of Tampa. In order to assess

the effects of the development of the Tampa Palms Property on the quantity and

-A101-

quality of stormwater runoff entering the receiving systems, the Storage

Treatment overflow Runoff Model (STORM) was employed. STORM, a widely used

and proven hydrologic model (continuous simulation), was initially applied to

the project in its current undeveloped condition, using water quantity and

quality data collected from several monitoring stations strateically Located

at surface runoff sites throughout the property for model calibration. The

proposed changes in land use and drainage characteristics due to development

Were than incorporated into the model to predict future stormwater runoff

quantity and quality conditions and to analyze potential impacts to receiving

systems. The net impacts of stormwater discharges from the Tampa Palms

property on the Hillsborough River should be minimal, because significant

treatment of stormwater runoff will be achieved in the Tampa Palms drainage

system and the pollutant loads (mass) from the developed property will only

constitute a small percentage of the loads carried by the river. For all the

pollutants modeled, total loads for the post-development basins represent a

minor to very minor percentage of the total loads for the Hillsborough River.

In addition, stormwater discharges from the developed drainage system will be

attenuated (via controlled release) which will increase the ability of the

Hillsborough River to assimilate the loads

54. Ward, A. J., C. T. Haan, and B. J. Barfield. 1977. Simulation of the

sedimentology of sediment detention-basins. University of Kentucky Water

Resources Reseach Institute, Lexington.

Abstract: Sediment detention basins are a widely used means of controlling

downstream sediment pollution resulting from stripmining and construction

activities. A mathematical model for decribing the sedimentation.

characteristics of detention basins has been developed. This model requires

-A102-

as inputs the inflow hydrograph, inflow sediment graph, sediment particle size

distribution, detention basin stage-area relationship and detention basin

stage-discharge relationship. Based on this information the model routes the

water and sediment through the basin. In this routing process the outflow

sediment concentration graph, the particle of sediment depostion in the basin

and the sediment trapping efficiency are estimated. Comparison of predicted

results - with measured sediment basin performance indicates the model accurately

represents the sedimentation process in detention basins. This report details

the model illustrates its use in design, explains how to process the model on

a digital computer and presents a program existing of the model.

55. W. G. Jr., and C. Wilson. 1976. Evaluation of sediment control

dams. Pennsylvania Department of Transportation, Bureau of Materials, Testing

and Reseach, Harrisburg, Pennsylvania.

Abstract: Small dams in waterways have been extensively used by PennDOT in
erosion control. The evaluation of the performance of these small dams was

conducted to determine how well they performed in removing sediment, from

flowing water. Also the various methods of constructing these dams were

compared. The results indicate that only minor amounts of suspended sediment

are removed from the flowing water. However, the bottom transport is trapped

by these dams. The rock dams generally were stable.

36. Widseth, R. A. 1978. Utilizing a storm sewer outlet. International

Symposiim on Urban StormWater Management, C. T. Haan, Editor, University of

Kentucky, Lexington, pp 317-322.

Abstract: Crookston is located in Northwestern Minnesota. The University of

Minnesota operates a college and an agricultural experiment station north of

the City. In 1970 the University installed an 11,000 foot long storm sewer

-A103-

outlet to drain the campus area. This outlet crossed undeveloped University-

owned farmland along its route, but was not sized to serve this property

because a need was not anticipated. Now 100 acres of this land has been sold

for development and requires storm sewer service. By the use of a storm water

detention basin and a controlled outlet this development can be drained by the

existing storm sewer for approximately one-seventh the cost of constructing a
outlet.

Wilison, al. 1977. Preventive aproaches to stormwater

management. U. S. Environmental Protection Agency, NPS Branch WH-554),

Washington, D. C.

Abstract: This document was prepared for use by local agency administrators,

and others who may be involved in programs to abate pollution from urban

runoff. The concept of source control (BMP) has been discussed in the past and

this report lists many techniques that would be included in a Best Management

Practice. The problems associated with implementing these practices, legal,

financial, and institutional are also discussed. The objective of this study

was to provide a basic understanding to local administrators on what BMP is

those techniques which would comprise a BMIP.

58. Woodward, D. E., P. I. Welle, and H. F.Moody- 1980. Coastal plains unit

Hydrograph studies. National Symposium on Urban Stormwater Management in

Coastal Areas, C. Y. Kuo, Editor, American Society of Civil. Engineers, New York

pp 99-107.

Abstract: A hydroglogic study initiated in 1975 indicated that the Soil-

Conservation Service (SCS) was not able to simulate recorded flood hydrographs

or regional peakflow frequency curves on the Delmarva Penisula in Delaware,

Maryland, and Virginia. The standard SCS unit hydrograph was used in these

-A104-

detailed hydrologic studes. The peak rate equation for this standard SCS
unit hydrograph can be expressed as q = 484QA/Tp where 484 is a shape and
units conversion factor, q p is the peak discharge in cfs, Q is the volume of
runoff in inches, A is the drainage area in square miles, and T is the time

to peak in house . The 484 factor is referred to in this paper as a

dimensionless unit hydrograph peak factor (DUPE). This paper describes the

studies made to develop a dimesionless unit hydrograph for the Delmarva
Peninsula. The standard SCS unit hydrograph was developed from small

agricultural watersheds primarily in the Midwest. These watersheds are

generally characterized by local relief of 50 to 100 feet with little or no

natural storage. Since physical characteristics in the Midwest differ from

those in the Delmarva Peninsula, it was felt necessary to develop a unit

hydrograph specifically for the study area. The basic shape of the new

hydrograph appears raticial because when compared to the standard SCS unit

hydrograph, the recession limb contains more volume and the peak rate is

lower. Analysis of the August 3, 1967 storm on Murderkill River was used as a

check on the reasonableness of the selected unit hydrograph.

59. Wu, J. S., and R. C. Ahlert. 1978. Prediction and analysis of

stormwater pollution. International Symposium on Urban Storm Water

Management, C. T. Haan, Editor, University of Kentucky, Lexington, pp 183-188.

Abstract: In recent years, pollution from nonpoint source has become an

increasingly important consideration in water quality planning and management.

Stormwater pollution can be characterized, in magnitude in concentration

of pollutants, as intermittent and impulse-type discharges into receiving
waters, causing shock-loading problems to the ecosystems of these water bodies.

The classical approach, using critical low flow as the design criterion for

-A105-

water quality management schemes, must include the effect of storm runoff.

This paper aims to summarize state-of-the-art methodologies for predicting

storm runoff loads. Four categories of prediction methods are presented;

these include zero-order, rational, statistical and descriptive methods, for

achieving different levels of prediction, i.e. (i) average annual storm load,

(ii) storm load per event and (iii) storm load distributions within events.

Also, it is intended to present the authors' ongoing efforts to develop an

analytical method for assessing stormwater impacts, with regard to biochemical

oxygen demand (BOD) and sediments on receiving water quality. The receiving
water body is represented by a completely mixed reactor and a settling basin,

during storm runoff periods. Surface runoff and pollutant loads enter the

receiving stream at an average uniform rate over each time interval, i.e. a

step-wise, steady-state approach to input variables. Deposition and/or

resuspension of sediment are limited by the sediment transport capacity of the

stream. Model equations and analytical solutions will be presented, with the

Assunpink Basin of New Jersey as the illustrative example.

60. WycnfE, R. L. 1978. Computer-aided hydraulic design of storm drains.

intreaational Symposium on Urban Storm Water Management, C. T. Haan, Editor,

University of Kentucky, Lexington, pp 325-321.

Abstract: Conventional hydraulic design of closed conduit storm drains

usually consists of selecting conduct sizes based on open channel flow

complications. Conventional design of highway culverts, on the other hand,

use of the available static energv head at the culvert site. A

computer-aided design procedure is presented whereby the hydraulic equations

of culvert flow are utilized to size branching, closed conduit, storm drainage

networks. By this method, the available static energy head is utilized as

-A106-

efficiently as possible. The design algorithm selects a pipe size array which

will carry the design flows without overtopping, while minimizing the largest

conduit size selected. The pipe size array obtained by application of the

computer-aided design procedure to an example network is compared to the pipe

size array obtained by open channel flow design procedures. In general, the

computer-selected sizes are smaller the channel size,resulting in

construction cost savings.

61. Wycoff, R. L., and V. P. Singh. 1976. Preliminary hydrologic design of

small flood detention reservoirs. Water Resources Bulletin, 12(2): 337-349.

Abstract Flood detention reservoir design is a common problem encountered by

engineers are others involved with water resources problems. The's paper

presents a method by which the volume of flood storage required for a single

reservoir or a series of reservoirs may be estimated without using numeric

flood routing techniques. An idealized model based on triangular hydrographs

was developed to define the major relationship among the variables. A

generalized equation was then obtained by multiple linear regression analysis

of computed flood-routing data. The effect of storage distribution on flood

peak reduction efficiency was then investigated by means of a computer model

study of equal-sized reservoirs located in series. This resulted in a

relationship between number of reservoirs and peak reduction efficiency and,

together with the generalized storage volume equation, form a proposed

procedure for temporary hydrologic design of small flood detention

rescue,

-A107-

62. Wyroff, , J.E., School, and S. Kissoon. 1979. 1978 needs survey

Cost mcthodology for control of combined sewer overflow and stormwater

discharges. U.S. Environmental. Protection Agency, Municipal Construction

Division, Office of Water Program Operations, Washington, D.C.

Abstract: The major objective of this project is to develop updated nation-

wide cost estimates for control of pollution from combined sewer overflow and

urban stormwater runoff on a State-by-State basis. A secondary objective is

to establish the National Combined Sewer System Data File. This file contains

information on every known combined sewer system in the nation, including

location, sewer system and receiving water characteristics, and the status of

current Combined Sewer Overflow (CSO)

-A108-

APPENDIX B

SUMMARY OF STORMWATER MANAGEMENT METHODS
AND THEIR APPLICABILITY IN COASTAL WETLANDS

Richard H. McCuen

Stanley L. Wong

SUMMARY OF STORMWATER MANAGEMENT METHODS
AND THEIR APPLICABILITY IN COASTAL WETLANDS

Engineering projects, if not properly designed, can have serious physical

impacts on wetlands. Changes in the characteristics of surface and subsurface

flows may cause major biological changes and changes in the characteristics of
the existing ecosystem. Projects may also interfer with the existing character-
istics of tidal flows, which would have physical, as well as biochemical effects.

In addition to projects located outside of wetlands,but affecting wetlands,

engineering projects within wetlands may change the physical, chemical, and

biological characteristics of wetlands. Physical impacts of specific interest

are:

� changes in the mean water level of the wetland

� changes in local water table levels

� changes in periodic variations of both surface flows and tidal flows

� changes in circulation and salinity patterns

Because of the potential damage to wetlands that may result from these changes

every engineering project should include an assessment of the physi6al impacts

of construction and land use change and the identification of methods for miti-

gating these changes.

Changes in Mean Water Level

Engineering projects may involve damming effects and increases in drainage-

ways. If water is retained outside of a wetland area, the mean water level with-

in the wetland may decrease. Damming within the wetland may increase the mean

water level. Increased drainage within the wetland area may decrease the mean

water level. Changes in the water level indicate changes in the storage of

the wetland.

Changes of the mean water level may alter the spatial arrangement of plant

species. Additionally, changes in the size of the wetland,and therefore the

-Bl-

st ora go , inn y a f f cc t t lie ab L I i t )- o f t lie we t I and t o 1) rov i de down,, t r cam f i ood 1.) ro -

tection. Also, the reproduction characteristics of aquatic species may change

as the water level and storage change.

Assessment of the impact of engineering projects would require water level

information both before, after, and during construction. Measurements should be

made over a sufficient time frame so that an accurate estimate of mean water

level can be made.

Changesin Local Water Table Levels

Wetlands are characterized by high water tables, which intercept surface

water levels where dictated by local topography. Wetland flora may be quite

sensitive to permanent changes in local water table levels. Therefore, the

piezometric surface should be identified prior to construction and should be

monitored both during and after construction. Changes in the water table can

be minimized through proper drainage.

changes in Periodic Variations in Water Levels

In addition to the mean water level of both surface and subsurface waters,

the variation in water levels will affect the flora and fauna of wetlands.

Periodic variations may result from both daily and seasonal causes. Tidal

patterns will require measurements throughout the day to measure differences in

flood and ebb tide stages, as well as measurements over the lunar month to meas-

ure spring and neap tide stages. In addition to tidal fluctuations in coastal

wetlands, seasonal periodicites must be assessed. These periodic variations

in water levels affect the primary productivity of the wetland and the repro-

duction cycles of flora and fauna. Many wetland plants depend on periods of

inundation as well as periods for drying out; these periods are instrumental

in maintaining the supply of necessary nutrients. Measurements may be taken

using either stage recorders or continuous measurement devices.

-B2-

Changes in Circulation and Salinity Patterns

Circulatory patterns within a wetland are an important factor in determining

the distribution of nutrients and dissolved gases. The salinity gradient in

coastal wetlands is affected by the circulatory patterns. It is important to

chart the circulatory patterns within a wetland because the flora and fauna have

different requirements for nutrients and tolerances to pollutants.

Changes in the existing mixing of fresh and salt water can cause significant

changes in flora and fauna in a coastal wetland. Nontoxic dyes and periodic

measurements can be used to identify circulatory patterns. Seasonal variations

in the patterns should be assessed before construction and monitoring should take

place during,construction. Deviation from normal patterns should be minimized

through stormwater management.

Compilation of Site Information

The first step in runoff and sediment control is to prepare a control plan,

which should include the identification of local requirements, the evaluation of

on-site and off-site information, and the development of a control strategy. In

developing a control strategy it may be necessary to identify various alterna-

tives and select the one, or combination, that is considered to be the most

effective. The development of a control strategy should be followed by a field

review to examine the technical feasibility of the control strategy. The field

review may identify necessary revisions in the finalization of the control plan.

Assuming that the engineer is familiar with the legal requirements, the

engineer should proceed with the evaluation of site information. It is important

to make a thorough study of the site because the site information may dictate

the control methods adopted as part of the control strategy; it will also help

identify the most cost-effective strategy. Site information includes topography,

hydrologic response characteristics, climatic characteristics, soil characteristics

land use, and geologic information.

-B3-

Topographic maps are available from the U.S. Geological Survey and local

planning agencies. These maps provide useful information about relief and slope

gradients, drainage basin divides, road locations, and stream systems.

Soil maps and surveys are available from the U.S. Soil Conservation Service

or the local office of the Soil Conservation District. The surveys provide in-

formation about the soil types and extents, the engineering properties of the

soils, the erodibility of soils, and general textural characteristics of the soils.

While the topographic maps and soil surveys often identify land use, it is

worthwhile obtaining a recent low level aerial photograph of the study site and

downstream locations. For large projects Landsat imagery are available and can

be of value. Local land use and zoning maps may provide information about pro-

jected land uses that may be useful in control plan formulation.

Geologic maps are available from the U.S. Geological Survey and local offices

of the state Geological Survey. These maps identify the geologic trends in the

area, types of strata, and the location of geologic hazards. It may also be

important to identify locations with high-water tables, especially in coastal

areas near wetlands.

The information obtained for the site is necessary in the selection of miti-

gating measures in the planning, construction, and completion stages of a project.

Methods for runoff and sediment control can be categorized into three classes:

off-site conveyance, nonstructural on-site control, and structural on-site con-

trol. Off-site conveyance techniques include storm drains, open channels and

ditches, and channelization; the purpose of these methods is to remove the runoff

from the site to a downstream point where the runoff can be controlled without

causing damage. Nonstructural on-site control methods include swale and vege-

tative strips, porous pavement, and infiltration beds; the intent of these methods

is to provide the maximum posgible opportunity for runoff to infiltrate and, there-

fore, use natural storage. Structural on-site methods include detention, rooftop,

-B4-

and parking lot storage; the purpose of these methods is to use on-site, man-made

storage to replace the natural storage lost during development. While it may

appear that all of these methods are intended primarily for runoff control, it

is important to recogn@ize that the control of -runoff quality can not be separated

from the control of runoff quantity. The design factors that are important in

controlling runoff rates are also important in quality control. Furthermore,

the methods should not be viewed as independent methods; quite often, a combina-

tion of the methods may lead to the most efficient control of stormwater runoff

quantity and quality. For example, an infiltration bed should be separated from

the runoff source by a vegetative strip; this will provide for some particle

removal, which will help prevent the infiltration bed from becoming clogged.

Methods for Mitigating the Physical Impacts of Engineering Projects

Before examining specific runoff,erosion,and sediment control measures, it

may be worthwhile identifying the general objectives of mitigating measures:

� control runoff volumes into the wetland to maintain water levels and

the extent of the wetland
� control peak runoff rates into wetlands to maintain periodicities
and variations in both the surface water level and the water table

level
� control the volume of storage on upstream (adjacent) lands so as to
maintain the extent of the wetland.
� control sediment volumes into the wetland to prevent deposition with-
in the wetland, which may have physical impacts on both the bottom
taxa and the circulation patterns
� control sediment concentrations of runoff into the wetland to prevent
increases in turbidity
� maintain natural drainageways within the wetland to prevent both
damming of flow paths or the creation of channelized flow where
none existed prior to construction.

While these general objectives concentrate on runoff and sediment problems there

are other pollutants that result from construction activities. Construction

_B5_

activities may result in the discharge of petroleum products and other chemicals

into wetlands. In some cases, the measures used to control runoff and sediment

will reduce the detrimental effects of these other pollutants. In some cases,

it may be necessary to include other control measures into the control strategy

to ensure the mitigation of pollutants other than sediment.

Strategies for controlling runoff, sediment, and erosion have been instru-

mental in protecting the environment from the adverse effects of urbanization and

development. Without proper controls the changes in the land use and physical

features of the landscape (e.g., drainage paths) cause the response of a water-

shed to differ significantly from the original predevelopment conditions. The

increased flow rates and volumes of runoff often overload the capabilities of

the natural drainage. The transition period when land is being cleared for
development often produce-4 unusually high volumes of sediment and other pollutants.

The following is a summary of the advantages, disadvantages, and design

factors for the more common techniques used in controlling the increased amount

of runoff and sediment from developing areas. The design factors should be con-

sidered when attempting to implement the proposed method into a control plan. The

summary will aid in the selection of the appropriate approach, or combination of

approaches, that are available for mitigating the effects of development in or

near wetlands. For small construction sites many other methods exist. The

selection of these other methods will depend on the information obtained in the

site survey. A brief outline of these other methods is given at the end of the

section.

-B6-

ROUTING OF FLOW

Storm Drain Construction

Storm drainage systems have a primary purpose of conveying surface runoff

from developing sites.- They are especially valuable for routing flow around

environmentally sensitive areas. An underground pipe systent transports the

increased volume of water to receiving streams, retention ponds, or other stor-

age structures that are capable of accommodating larger quantities of water with-

out degradation of the local environment. The construction of storm drainage

systems helps to alleviate existing runoff problems and can also serve as a

preventive measure against potentially damaging (i.e., intense) rainfall events.

Advantages. Storm drainage systems have a major advantage in the small

amount of surface land that is required. Since the extensive conduit layout is

underground, the aesthetic value is acceptable to,the public. Storm drain systems

have additional benefits. The flexibility of including other stormwater manage-

ment techniques as well as expanding the conduit system allows for further develop-

ment with little disruption to existing areas. The drainage system is relatively

maintenance free and may have a service life up to 50 years.

Disadvantages. A significant drawback of storm drainage systems is the

changes that occur in the hydrologic response of a watershed. The effect of

increasing the conveyance of runoff decreases the times of concentration, hence

the peak discharge may be greater and occur sooner. The overall consequence may

be a transference of the problem to another location that is downstream of the

development.

Because of their impact on the hydrologic response of a watershed, the use

of storm drain systems may decrease surface and subsurface water levels in wet-

lands when the flow is routed around the area. If pipe systems are directed

into a wetland, the variations in flow, with increased peak rates of flow and

-B7-

decreased base flow, will increase; furthermore, pipe systems will eliminate the

natural cleansing action of a watershed and thus increases in pollution loadings

may result.

DesignConsiderations. The designing of a storm drainage pipe network re-

quires hydrologic and hydraulic knowledge of the watershed and the manner in

which the flow travels through a conduct. The sizing of the components in the

pipe system will depend on the predicted runoff rates. The peak discharge is

usually of primary concern in the determination of pipe sizes. Also., the invert

slope of the pipe and the Manning's roughness coefficient, n, of the pipe are

important considerations that affect the velocity of flow.

in addition to design factors one must also consider site factors. Pipe

systems may not be affective in extremely low sloped areas. The construction

of a system should not interfere with other utilities or natural environmental

factors that can not withstand the construction disturbance. Outlets from pipe

systems represent a danger to the local environment unless they are protected

from scour that results from high, concentrated flow rates. Such scour in wet-

lands can upset the balance within the bottom areas.

Open Channels and Ditches

Open channels and ditches are typically found in areas of low density, which

have sufficient surface land that is not required for other purposes. Open

channels may have a continuous flow (i.e., base flow). Ditches located along

roadsides are not intended to carry the larger quantities of water that open

channels are capable of handling. Ditches adjacent to construction sites must

be properly designed to control erosion.

Advantages. Open channels and ditches may be designed with different linings

such as concrete, rip-rap, or grass. The infiltration of flow in unlined channels

and ditches is also a beneficial result because it serves to recharge groundwater.

-B8-

Concrete or rip-rall linings improve the stability of the channel bed in tile

event of high velocity flows. Ilie crodibility of the soils is, therefore, re-

duced. Erosion can also be prevented by using check dams*and drop box culverts.

Disadvantages. Open channels may have adverse consequences if not properly

designed. If the drainage area in the channel bed is not large enough, flooding
may occur. There should-be an overflow area for those occasions where aimajor

stom could fill the available drainage area. The amount of required land is

thereby increased. Maintenance is frequently needed should the channel become

clogged with debris or become filled with siltation.

Channels and ditches constructed in wetlands may increase drainage from

the wetland; this may result in decreased storage and water table levels and

thus a decrease in the extent of the wetland. Channels and ditches constructed

adjacent to wetlands may cause significant changes in the circulation patterns

within wetlands.

Design Considerations. The factors that affect the design of open channels

and ditches are similar to those that are involved with the design of storm

drainage systems. The size and shape of a channel is dependent on the quantity

and velocity of the runoff. The velocity of the flow determines the type of

lining that is needed in the channel bed. The velocity varies directly with

the square root of the bed slope and is inversely proportional to Manning's

roughness coefficient n. The values of n have been derived for various sur-

faces and are available in most hydraulic textbooks.

To control erosion of the bed, the tractive force must be balanced with the

oppositely acting force of gravity. The most important factors are the slope

of the bed and side slopes, and the shape, size, and weight of the soil particles.

Bed slopes can be reduced when check dams and drop inlet culverts are used.

_B9-

Channelization

Channelization, which is used to prevent flooding in susceptible areas by

increasing the conveyance of water, requires the straightening or dredging of a

stream and the clearing of the banks. Channelization allows a larger amount of

water to travel a shorter distance so that the flood flow passes rapidly.

Channelization of a stream may occur naturally when flood waters force the flow

to travel in a straight path.

Advantages. Channelization offers the greatest advantage in the protection

of life and property from dangers that result from flooding. Communities that

have already built in the flood plain may benefit from the flood relief.

Channelization also provides a proper conveyance for boating. The aesthetics

of urban streams may improve as the banks are cleared of debris.

Disadvantages. The process of channelizing a stream may have adverse effects

if not properly implemented. The soil should be sufficiently fertile to support

vegetative growth to prevent erosion. In the event of high velocities, artificial

surfaces should be used to line the stream bed. Concrete or rip-rap lining is

commonly used in these cases.

The effectiveness of channelization as a stormwater management alternative

is not always favorable. The increase in the flood passage may transfer the

problem to another point causing increased peak flows downstream. Scour may

occur behind the channelized sideslopes if proper maintenance is not provided.

In many cases, the public find a natural environment to be aesthetically favorable

when compared with a concrete lined channel.

Channelization within a wetland reduces the surface area available for

aquatic biota; the increased velocities may also be unfavorable to the biota.

Channelization in areas adjacent to wetlands can cause changes in circulation

patterns within wetlands.

-B10-

Design Considerations. The hydraulic principles that control the stability

of channel are well documented, In general, one must consider the slope of the

channel, the soil characteristics, and the cross-sectional properties in design-

ing a stable channel. The sediment discharge is proportional to the square of

the discharge and the square of the slope and inversely proportional to the 1.5

power of the mean particle size of the soil. If a channel is to be modifted,

the engineer should take the necessary procedures to ensure stability of the bed

and side slopes. If a vegetated surface is not possible, the channel may have

to be lined.

GROUNDWATER RECHARGE SYSTEMS

Natural Infiltration Technique

Engineering projects often increase the percentage of impervious area, thus,

decreasing the available natural storage. As the watershed is developed the

amount of surface runoff is increased by the excess water that is no longer in-

filtrating or stored in depressions. The timing of the runoff from the watershed

is also affected, causing peak flows to occur sooner. The increase in surface

runoff is accompanied by decreases in base flow.

Infiltration of the runoff will decrease the initial volume of the overland

flow. Areas that are covered with good vegetative growth have higher infiltration

rates than those that are bare. Water flows over vegetation at a lower velocity,

enhancing the opportunity to infiltrate into the groundwater system. Hence, by

routing overland flow over grassy areas, the runoff is detained, allowed to

infiltrate, and decrease in amount.

Advantages. The advantages of utilizing natural infiltration to decrease

runoff is evident. The natural drainage system is maintained, and the use of

construction methods to prevent erosion by keeping existing vegetation intact

is a good conservation practice. The recharge of aquifers is also accomplished

when maximizing infiltration.

Disadvantages. Areas that have little or no vegetative cover are excluded

from using this technique. Also, vegetation may require maintenance.

Design Considerations. The effectiveness of natural infiltration depends on

the hydraulic flow length, the density and type of vegetation, the infiltration

capacity of the soil, and the slope. The length of the hydraulic flow path should

be maximized. For paths characterized by high slopes, the vegetation should be

selected so that high velocity runoff does not cause erosion. An increase in

depression storage, when acceptable, will increase infiltration and decrease

runoff volumes.

Porous Pavement

Porous pavements reduce the total volume of direct runoff from parking lots,

highways, and paved areas. The development of porous pavements as a stormwater

management technique is a relatively new concept. While still in the experi-

mental stage, porous pavements are being evaluated as a method to limit runoff

by allowing water to percolate into the ground. The practicality of the method

depends on its ability to carry loads, withstand freeze-thaw cycles, and be

cleaned, should the pores become clogged.

Advantages. Porous pavements have many advantages that make it a desirable

technique. Because pavements are normally impervious to water, the effect of having

an absorbent pavement may reduce the peaks of local floods and reduce the need

for storm drainage systems. Moreover, it increases groundwater recharge by

utilizing the natural drainage system. While not believed to be applicable to

roadways, it may be used on shoulder areas that do not receive heavy traffic

loads.

-B12-

Disadvantages. The effectiveness of porous pavements has not been widely

assessed under actual circumstances. More testing should be conducted to eval.u-

ate the feasibility of usage. The problem of maintaining clogged pores has not

yet been resolved and applicability is subject to the characteristics of the soil

beneath the pavement.

Design Considerations. The effectiveness of porous pavements will depend on

the slope and roughness of the pavement, the hydraulic conductivity of both the

pavement and the underlying soil, and the degree to which the pavement is clogged.

Porous pavement can be used in areas subjected to heavy erosion.

Infiltration Beds

Infiltration beds are designed to collect runoff and store it until it

infiltrates into the ground. The quantity of runoff is thereby reduced allowing

the natural storage to be maintained. The bed is excavated and filled with a

highly permeable material such as gravel or rocks. These facilities are useful

around commercial buildings or parking lots.

Advantages. The properly designed infiltration bed will serve as a means to

maximize infiltration. The local flooding potential is reduced and the recharge

to groundwater supplies are enhanced. The required capacity of the downstream

storm drainage system is reduced because of the lower volumes of direct runoff.

Infiltration beds located on construction sites near wetlands can decrease peak

runoff rates and sediment coiicentrations in flows into wetlands.

Disadvantages. The primary disadvantage of infiltration beds is the fre-

quent maintenance that is required. Measures should be taken to screen leaves

and trash before runoff enters the bed. If surface runoff that is directed into

an infiltration bed contains significant volumes of sediment, the runoff should

be intercepted by a sediment trap or a vegetation strip prior to discharge into

-B13-

the infiltration bed; otherwise, the sediment will reduce the.infiltration and

storage capacity of the infiltration bed. The method also requires an overflow

area, which may increase the amount of land required for the facility. Scour

at the inlet must be prevented. Safety is also a necessary consideration.
I

DesignConsiderations. The effectiveness of the beds is highly dependent

on the soil characteristics of the bed, the volume of storage, the location of

the bed within the hydraulic flow path, and the location of the water table.

An infiltration bed should be sited within the flow path in order to have the

greatest effect on runoff rates near the time of the peak discharge. For areas

having a high water table, the effective storage will be highly dependent on the

surface area allocated to the bed.

STORAGE OF EXCESS RUNOFF

On-Site Detention Ponds

On-site detention ponds are effective in reducing the peak flow both at the

site of development and in downstream areas that are sensitive to frequent flood-

ing. The runoff is detained in these ponds and released at a rate that is similar

to the predevelopment runoff rate. The size of a detention pond depends on the

volume of water that must be stored to prevent increases in peak flow rates and

velocities. Construction of detention ponds requires excavation and fill opera-

tions, which must be properly controlled to prevent adverse environmental impacts.

Advantages. Detention ponds serve as a multipurpose facility. One of the

primary purposes, which initiated the usage of detention ponds, is the control of

sedimentation from the erosion of soil during construction. The larger ponds may

also be used for recreation activities such as boating or fishing. As a result of

controlling the maximum runoff, detention ponds also effectively reduce the size

of the storm drain system required. If properly sited and designed, ponds can be

_B14-

aesthetically pleasing. Large ponds can also be used for recreation. They will

also help to maintain base flow into wetlands. Detention ponds can help maintain

water levels in wetlands and reduce the changes in variations of flow rates that

result from construction activities.

Disadvantages. The implementation of detention ponds requires the allocation

of a significant amount of land; such land then may not be used for development.

Another problem of implementing the facility is in attempting to convince local

communities that such stormwater management measures are worthwhile when the

downstream neighbors reap the benefits. The responsibility of maintenance, pro-

viding safety measures, and the legal water rights should also be considered.

If not properly designed, detention ponds may create flooding problems in wet-

land areas downstream of the construction area.

Design Considerations. In order to maximize the efficiency of a pond for

controlling runoff rates, sediment rates, and other pollutants, the volume of

storage must be sufficient to provide adequate detention time. This will allow

for sufficient setting of pollutants and helping flow rates at a minimum during

periods when uncontrolled flow have maximum runoff rates. The loss of storage

due to inadequate maintenance should be considered in design. In wetland areas

with high water tables, the expected location of the water table must be con-

sidered when determining the spatial area/depth ratio of the pond. The capacity

of the outlet facility must be determined with consideration of both allowable

downstream flow rates and control of the time of detention within the basin.

The dead storage/total storage ratio must be considered. Baffles can be used

to increase flow lengths and detention times and thus settling rates of pollutants.

Rooftop Ponding

Rooftop ponding provides storage for stormwater runoff by limiting the.flow

to the down spout. The technique is used primarily in highly urbanized areas

that include a significant number of multi-family housing units; it will have a

significant effect on reducing the runoff if a large percentage of roofs contri-

bute to runoff. The structural integrity of each roof should be evaluated before

incorporating rooftop storage. The provision of an emergency overflow must be

made to prevent structural damage or leakage. Runoff should overflow before the

maximum load is attained.

Advantages. Rooftop storage facilities can provide many benefits. There

is no inconveniences to the public since the storage structure is not visible

from the ground. The technique utilizes existing structures in urban areas where

there is usually no available land for other types of controlled storage.

Disadvantages. This method of storage requires planning when the structure

is designed; that is, it is often not possible with existing buildings. The

location of the housing units within the tract of land can have a significant

impact on the effectiveness of rooftop storage. Rooftop ponding requires mainte-

nance including the repairing of leaks, removal of debris and ice, and maintaining

the downspout and emergency overflow facility.

Design Considerations. In addition to the design of the structure, it is

necessary to consider the design characteristics of the outlet facility; these

should be designed to provide optimum control of the timing of the runoff.

Parking Lot Ponding

Parking lot storage reduces the runoff rates and pollution loadings on storm

drainage systems. The storage is temporary and used in paved areas around com-

mercial buildings, offices, and shopping centers. The stormwater is collected

and released at a regulated rate.

-B16-

Advantages. Parking lot ponding is easily implemented in areas that are al-

ready developed. Water can be ponded around the least used portions of the

parking lot, perhaps channeling the runoff to infiltration beds. Maintenance

is easily performed using mechanical street cheaning machines.

Disadvantages. Ponding in parking areas may cause an inconvenience to the

4
public during storms. The problem is minimized by designating storage areas in

spaces least used.

Design Considerations. The slope and roughness of the area will control

the rate of inflow to these areas of the parking lot used for storage of storm-

water runoff. The capacity of the outlet will determine the effectiveness of

the storage. Because dead storage is not usually provided, unless the effluent

is directed to an infiltration bed, parking lot storage is usually not effective

for pollutant control.

Additional Control Measures

Engineering projects require measures to limit physical impacts on wetlands;

four general impacts were identified: (1) changes in the mean water level of

the wetland, (2) changes in the local water table level, (3) changes in periodic

variations of both surface flows and tidal flows, and (4) changes in circulation

and salinity patterns. In order to minimize these impacts on wetlands, runoff

and pollution characteristics (i.e., volumes, peaks) must be controlled. While

the control methods discussed in detail are the primary control measures, other

effective measures are available.

The following control measures are effective for controlling peak runoff

rates into a wetland:

(1) temporary barriers made of straw bales can impede surface runoff and
induce infiltration;

-B17-

(2) diversion ditches can increase flow paths and are especially useful
for diverting surface flow around cleared areas;
(3) the omission of grubbing increases the surface roughness and
impedes flow;
(4) level spreaders can be used to convert channel or pipe flow to
sheet flow, which results in reduced rates of runoff;
(5) a roughened surface, which is made by disking or ploughing,
reduces runoff rates and increases infiltration.

The following measures are effective for controlling increases in surface

runoff volumes:

(1) benches constructed on the steeper slopes serve to increase infiltrntion;
(2) spray irrigation of surface runoff can be used to spatially distribute
the water on pervious surfaces;
(3) level spreaders can be used to divert water to flow over pervious
surfaces;
(4) serrated cuts will increase infiltration and reduce erosion potential.

The following measures are effective for controlling increases in sediment

volumes:

(1) check dams prevent erosion and lengthen runoff times, thus increasing
settling of suspended solids;
(2) chemical stabilization reduces surface erosion potential;
(3) compaction of cleared areas can decrease erosion rates;
(4) fertilization can increase the growth rate of vegetated covers, thus
increasing erosion resistance;
(5) filter fences placed in drainageways and at slope toes can be effective
in limiting the passage of sediment;
(6) gabions can dissipate the energy of surface runoff in channels and
along ditches;
(7) jute and plastic matting is used as a surface and drainageway
protection;
(8) pipe outlet protection can eliminate scour at outlet;

-B18-

(9) rip-rap can dissipate energy and reduce channel erosion;
(10) sediment traps are especially useful where flows are concentrated
and contain significant volumes of sediment;
(11) seeding, followed by fertilization or chemical stabilization, can
increase the vegetative cover and reduce the exposed surface area;
(12) vegetative buffer strips, which are especially useful in low sloped
areas, serve as filters to limit sedimentation.

_B19_

APPENDIX C

ELEMENTS OF A COMPREHENSIVE
STORW,ATFR MANAGEMENT POLICY

Richard H. McCuen

Mark E. Hawley

ELEMENTS 01: A COMPREHENSIVE STORMWATLIZ AANAGEMENT POLICY

In developing a SWM policy it is important to keep the goals of SIVM in

mind. These goals will not be met if the policy is ambiguous or incomplete.

The elements discussed below should be considered for inclusion in any SW74

policy that is being formulated at the level of government that has the primary

SWM, responsibility. In most cases, this level of government is the municipal

or county level, although regional councils and planning agencies may have the

primary responsibility for SWM in some areas. In an case, the goals of SIVM

will be more easily achieved if the SIVM policy contains those elements because

the policy will be clearer, more specific, and more comprehensive.

Statement of Authority to Regulate

A comprehensive policy should always include a statement explaining the

source of the authority under which the policy is promulgated. This statement

is helpful to developers, regulators, and the general public because the limits

of the regulators' authority are defined. Regulators may not know the basis and

extent of their powers if this statement is not included; this may lead to

misunderstandings and less than optimum performance of their duties. Further-

more, an explicit statement of the authority to regulate may eliminate unnecessary

lawsuits and disagreements caused by misinterpretation on the part of the de-

veloper. Institutional conflicts may also be avoided because the limits of

each government authority are specified in the policy.

The statement of authority to regulate must not only contain the legal

basis for regulations, but must also assign responsibilities. In man), areas,

more than one agency may have jurisdiction over development; therefore, the

statement of authority should specify the responsibilities of each agency so

_C1_

that conflicts are eliminated. For instance, a techni,cal agency such as tile

State Soil Conservation Service might be given tile responsibi lity for developing

a suitable set of SWIM design criteria, while re:@ponsibility for issuing permits

and enforcing the SIN regulations might be given to another government body,

such as a Department of Public Works. This division of responsibilities is

dependent on the nature of the laws under which the authority to regulate is

established. All responsibilities should be specified in the regulations so

that developers and administrators will know the limits of the authority and

responsibilities of each government agency.

Goals and Purpose of the Policy

Every SWNI policy shOLIld include a statement of tile goals and purposes of

the policy. This is actually tile most important part of the entire policy

because all the other parts are merely the ways of onsurin- that these goals

are met.' The statement of goals and purpose should be comprehensive; both

water quality and water quantity goals should be specified. The primary pur-

pose of most policies is to ensure the public's health and safety; tile -specific

goals are the standards that must be met in order to achieve this primary pur-

pose.

Definitions of Terms

Much of the vocabulary used in SWNI policies and regulations is technical,

and many of the people that will have to interpret the regulations do not have

a technical background; therefore, it is important to include a section in

which all terms are defined. Inclusion of such a section in a set of SWM

regulations will make the regluations less ambiguous and may lessen the

possibility of legal difficulties caused by misinterpretation of the regulations.

-C2-

Previous Relevant Legislation

The enactment of specific S101 legislation and regulations may cause con-

fusion due to the existence of previously, enacted legislation concerning flo od

plain management, sediment control, and other related topics. Therefore, the

relationship between the previous regulations and the newly enacted SWM reg-

ulations should be described precisely in the SIN legislation. In many cases

it is simpler to enact a new, comprehensive set of regulations that concern

SWM and all-related topics at once; this new set of regulations can incorporate

or replace the old floodplain and sediment control laws. This approach is

preferred because there is no possibility of conflicts with existing regulations
C, C,

if the comprehensive set of rules is properly for1flUlated. Conflicts with

existing laws can lead to confusion on the part of the developers and regu-

lators, and should therefore be avoided if possible.

Sediment and Grading Control

Many governments that have no SWM regulations already have sediment control

laws in effect; because one of the goals of SWN1 is the control of erosion and

sedimentation, sediment control should not be considered as a separate subject.

The most reasonable and efficient way of achieving all the goals of floodplain

management, sediment and grading control, and SWM is to formulate one compre-

hensive set of regulations covering these and any other related topics such as

regional planning. This is the only feasible way of eliminating conflicts that

can be detrimental to the public's well-being.

Floodplain Management

The subjects of stormwater management and floodplain management cannot be

completely separated. SKM involves controlling the rates and volumes of flow

through stream channels, while floodplain management involves controlling the

land use in the channel areas. Because the rates and volumes of flow determine

.-C3-

the size of the floudplain, these subjects should be considered togcthcr rather

than as two separate problems. In writing SIN regulation,;, this connection

should be considered because one of the goals of SWM is also the goal of flood-

plain management, namely, to minimize flood damage. If possible, the flood-

plain management rCgL1l:1tiO11, should be incorporatcd into the SWNI regulations

so that potential conflicts are avoided.

Cooperation with Other Local Governments

In many cases, storm runoff from one county or municipality drains into

streams and channels that pass through downstream jurisdictions. The upstream

jurisdictions have a responsibility, morally if not legally, to control their

runoff to avoid adverse effects in the downstream reaches. Many waterways also

serve as political boundaries; the banks of a stream mav lie in two or more

separate jurisdictions. In order to avoid conflicts between upstream and down-

stream counties and between counties on opposite banks of a waterway, all local

SIM regulations should require cooperation between the various local governments.

Each government should be required to consider the effects of its actions on

downstream areas and to regulate the activity of developers so that downstream

areas will be protected. This type of cooperation is probably best ensured by

the creation of regional planning bodies.

Elimination of Conflicts Between Local, State, and Federal Laws

The Federal government and state governments have jurisdiction over many

of the larger rivers, lakes and streams in the U.S. Confusion can arise as

to which government actually controls the use of a particular body of water.

Because the Federal, state, and local government regulations often differ, it

is important for the developer to know which set of regulations applies to the

area being developed. In most cases, state authority supersedes that of local

government and Federal authority supersedes that of the state; but because the

-C4-

limits of state and Federal authority are often ambiguous, it is not always

obvious where the authority to regulate lies. Therefore, a comprehensive set

of local regulations should include a statement describing the limits of the

local authority and requiring compliance with other authorities where they

have jurisdiction. In this way, conflicts between the various levels of

government can be avoided, and the design process can be expedited.

Exemptions for Government Projects

The SWIM regulations of many jurisdictions contain exemptions for government

projects, although the reasons for these exemptions are seldom explained. in

cases where the exemption is from the application and permit process only, these

exemptions make sense because it is often important to the public welfare

that government projects be completed as quickly as possible. The project

is usually subject to the same design requirements as a private project would

be, but the approval of the design is expedited by having the government engi-

neers design the facilities themselves. Some SWIM regulations contain an exemp-

tion from the inspection clause for government projects, but because many govern-

ment projects are actually constructed by private contractors, exemptions from

inspection requirements may not always be reasonable. The critical factor in

determining whether government projects should be exempt from any particular

regulation should be the impact of the exemption on the public welfare. If

an exemption is not in the public's interest, it should not be included in

the regulations. Any exemptions for government projects that are deemed to

be in the public's interest should be specified in the SWM regulations to

avoid ambiguity.

-C5-

Design Criteria

The heart of any S11M policy is the set of do,,-,ign criteria with which the

developer must comply. These criteria should be developed at the local level

if possible because different environments require different SWNI procedures.

In most parts of the U.S., each county has a local branch of the state Soil

Conservation Service (SCS); these local branches are familiar with the terrain

and soils in their areas and are probably the people best qualified to formu-

late the design criteria for SWM. However, in most areas tile local SCS branches

do not have the authority to institute ShIM regulations; that authority is

cither not exercised b@- WIN' government, exercised bY the state government, or

delegated by the state government to the local governments. Because of the

diversity of human and hydrologic environments encountered in many states,

the most sensible approach is usually for the state government to delegate

SWM authority to local governments and regional commissions and then play an

advisory role. In this case, the local Department of Public Works (DPW) is

usually charged with developing and enforcing tile SWM design criteria. Alter-

nately, tile local SCS can develop the criteria and the DPW can enforce tile policy.

Usually the agency that is charged with dvveloping tile design criteria call

receive help and advice from the state and Federal governments.

Regardless of the means of administrating a SWM policy, a comprehensive

set of design criteria is necessary if the goals of the policy are to be met.

Incomplete policies are ambiguous and lead to misunderstandings. The following

is a list of design elements that should be in a complete set of criteria;

under some circumstances, other elements should also be included.

Computational Methods. Many different hydrologic models are available for

evaluating the hydrologic effects of land use changes. Because different models

will often give different results when applied to the same situation, it is

-C6-

important for the reviewing agency to adopt one computational method (model)

for use in evaluating the developer's SWM plan. This standardization of methods

will help to eliminate technical disagreements because the design engineer and

the personnel with the authority to approve the plans will both be using the

same computational methods.

Design Limits on Slopes. Steep slopes are much more susceptible to erosion

than flatlands and, therefore, special limitations should be specified for areas

containing steeply sloping land. This is not much of a problem in areas where

the topography is generally very flat, but the problem of erosion on slopes

may still occur along the banks of channels or in areas where the local topog-

raphy is modified by man.

Design Storm Characteristics. Most SIVM regulations require that certain

design storms be controlled. The question of which return frequency to select

should be determined locally, based on economics (there is no point in spending

thousands of dollars to prevent flooding that will only cause hundreds of dollars

of damage). Many jurisdictions only specify the return period of the design

storm; this is a major failure on the part of the policymakers, because the

hydrologic effects of a particular design storm are not only a function of

total rainfall but also of storm duration, storm pattern, and antecedent con-

ditions. Therefore, the characteristics of the design storm that must be speci-

fied are: 1) the return period, 2) the storm duration, 3) the antecedent conditions,

and 4) the pattern of rainfall intensity. If references for determining the

return period and all the listed characteristics of the dosign storm are speci-

fied, then designers and evaluatui-s should have no disagreements over the

characteristics of the design storm.

-C7-

Emergency Spillways. Many SWM plans include structures designed to im-

pound water behind them; these structu res are designed to control the rate

of flow under a certain range of conditions. When the runoff from a SO-yoar

storm encounters a structure designed to control a 10-year storm, the top of

the structure will usually be inundated. This inundation may dainage or destroy

the structure unless an emergency spillway is provided to ensure that the excess

water can pass through the structure. Therefore, all impounding structures

should be required to include a spillway capable of handling runoff from ex-

treme storm events. Otherwise, sudden failures of SWIM structures mav cause

more damage than would have occurred with no StYLlCtUrC in plaCUP also, tile

costs of maintaining the structures will usually be hiolier if no spillway is

provided.

Fencing of Ponds. Stormwater management ponds are often considered to be

a hazard to public health due to their attractiveness to children. IMany juris-

dictions require that all ponds be fenced; some others require fences but issue

waivers of this requirement in rural areas. In some cases, nearby homeowners

will insist on fencing to protect their children, while others will think

that fencing is an unnecessary eyesore. Actually, fencing is usually only a

deterrent to small children-, only a very high barbed-wire topped fence is likely

to stop teenagers. Some states do not recognize liability cases based on

"attractive nuisance," so theoretically the government is not liable for damages

due to pond-connected accidents regardless of whether or not the pond i s fenced.

If fences are required, it should be because it is the will of the local populace.

Location of SM Facilities. One of the most important elements of a set

of SWIM design criteria is a regulation that controls the location of SIN facil-

ities. Many jurisdictions require that all SWM structures be located on the

site of the proposed development and that the postdovelopment. ronoff rates do

_C8-

not exceed predevelopment rates for specific storms. Alternately, off-site and

multi-site control facilities may be allowed.

When on-site control of runoff is required, each developer will plan his

own system independently. No coordination of the discharges from various develop-

ments will occur. If the postdevelopment runoff rate is regulated to be no

greater than the corresponding predevelopment rate, the postdevelopment duration

of peak flow must be larger than the predevelopment duration because the volume

of runoff is greater in the postdevelopment condition. Thus, each development

in the watershed will be releasing storm runoff at the predevelopment peak rate

for a longer period of time than in the predevelopment situation; these hydro-

graphs with increased durations of peak flows may combine in the channel and

cause increased flooding downstream due to the constructive interference of

the runoff hydrographs. This phenomenon is often referred to as peak overlap.

Peak overlap can usually be avoided by planning on a larger scale. If each

developer makes his own plans, there is no reason for him to consider the pos-

sibility of peak overlap; but if a regional authority participates in the plan-

ning process, SWM systems can be designed to eliminate the possibility of.peak

overlap.

Watershed-wide planning is a much more sensible approach to SWM than simply

requiring each developer to provide on-site controls. Besides eliminating peak

overlap, planning can also lead to the use of off-site and multi-site structures

that may provide greater benefits to the public at lower cost to the developers.

Economies of scale may be realized in both construction and maintenance if one

large structure serving a number of development sites can be substituted for a

series of smaller structures. In some jurisdictions a developer may be issued

a waiver of on-site control requirements if he cooperates with other developers

in planning multi-site facilities. In other areas the local government or

regional planning body may require the developer to contribute funds or land

_C9-

to be used for an off-site control structure in lieu of on-site control facilities.

One of the most important problems associated with changes in land use is

the increase in erosion that results from the increases in runoff velocities

and durations that accompany development. After development, most areas erode

more quickly than they did in the undeveloped state; therefore, the amount of

sediment delivered to the drainage channels is increased. Detention basins,

either on-site or off-site, are commonly used to control the increa.sed rates

of runoff from developed areas; as the silt-laden runoff from a developed area

flows into a detention basin, it loses velocity and consequently the sediment

carrying capacity of the flow is lessened. This results in sedimentation in

the detention basin. As the impounded water is released into the downstream

channel, the velocity increases, and creates the potential for erosion of the

channel immed.iatel), downstream. Also, because of the increased volume of run-

off, water flows through the downstream channel at near-peak rates for a much

longer time than in the predevelopment state; this can lead to enlargement of

the channel by erosion of the stream banks. Somc of thi.s erosion can he con-

trolled by use of outfall and channel protection methods, but a typical de-

tention structure is often collecting sediment on the upstream side and causing

increased channel erosion on the downstream side. Because the presence or

absence of sediment in the stream is one of the most important aspects of water

quality, these erosion problems should be considered in locating detention basins.

In some cases, on-site structures may be the only practical means of S101, while

in other cases off-site basins may be necessary if both the quantity and quality

of storm runoff are to be controlled. Requirements should be developed on a

watershed-by-watershed basis, rather than mandating on-site control for all

situations.

_C10-

Maintenance Considerations. Maintenance is an important factor in the

development of a SWIM policy due to the erosion and siltation problems associated

with development. Usually either the local government or the landowners are

assigned the responsibility for maintaining the SWNI.system; in many cases,

maintenance consists primarily of mowing grass), areas and removing the col-

lected sediment from detention structures. Regardless of whether the local

government or the landowners have maintenance responsibilitv, SW,%1 systems should

be designed so as to require as little maintenance as possible and to facilitate

the maintenance that is required. Consideration of maintenance in the design

of the SWM system will result in lower costs for the responsible parties and

will also lessen the amount of damage caused by neglect in most cases. There-

fore, consideration of maintenance in design of S101 systems is beneficial to

the general public as well as to the party responsible for maintenance.

Methods of SWN1. A complete policy should include a list of the types of

SWM facilities that a developer may use because some types of facilities are

not appropriate under certain conditions. For instance, in areas where water

tables are high or soils are very shallow, infiltration facilities may not

perform very well. If the allowable types of facilities are listed, the design

engineer will know what options are available. A statement should also be in-

cluded that allows the use of types of facilities that are not included in the

list if their effectiveness in a particular situation can be satisfactorily

demonstrated. The decision as to the types of facilities that are permitted

should be based on the computational methods used in evaluating proposed SIVINI

site plans; if the model will not account for the effects of a particular type

of facility, there is no means of judging the facility's effectiveness. The

model should be chosen with this problem in mind so that an engineer designing

a SWM system will be free to propose innovative solutions to SWM problems if

he desires to do so.

_C11-

Reference List. A list of rq@ferences to be used in evaluating SIN plans

should be included in the policy so that the design engineer knows where to

look for any required information. This will eliminate ambiguity and dis-

agreements over the methods used to evaluate the plan, lowering the cost to

both the developer and the government.

Sensitive Areas. In many localities, the government may find that there

are areas within its jurisdiction that require special consideration. These

are often called sensitive areas;'designation as a sensitive area may be due

to land use (historic districts or landfills), geomorphological features

(highly erodible soils or slopes, scenic areas), or biological considerations

(water supply reservoirs and trout streams). In an), case, it may be necessary

to impose SWM requirements of a more stringent nature in some localities in

order to protect these sensitive areas. The localities subject to these special

regulations should be described in the SWN1 regulations; in this way, the public's

health and welfare will be protected.

Inspection-s-

In order to ensure that the goals of the SWM policy are met, approved

projects must be inspected both during and after construction. Many of the

dimensions of detention basins are critical; if these structures are to per-

form as expected, the elevations of the top of the riser'and the emergency

spillway must be as specified in the design. Any part of a SWM system that

is located underground should be inspected before it is covered over to ensure

that it has been constructed as designed. Inspections are necessary after con-

struction has been completed to ensure adequate maintenance and proper function-

ing of the facilities. Inspections are usually the responsibility of the DPW

or a separate government inspection agency. The following list of elements

should be considered in developing inspection clauses1for SWM policies.

C12-

Access to Site. Free access to the site for government inspectors must

be guaranteed both during and after construction. This will help to prevent

an unscrupulous developer from deviating from the approved design in order

to cut costs and will also help to ensure adequate maintenance. Only by

having guaranteed access can the government inspectors protect the interests

of the public.

Inspection During Construction. An inspector should be on hand at all

critical times during construction of the system to ensure that SWM facilities

are built as specified. It is common practice to require the developer to give

notification at least 24 hours in advance of performing such critical operations

as setting the riser and spillway crest elevations in the construction of de-

tentibn basins. Also, any construction that will be covered when the system

is completed should be inspected before backfilling.

Final Certification. A final inspection of the completed SWN1 system should

be required, specially in cases where the government will assume the responsibility

for operation and maintenance. If the system passes the final inspection, the

local government should certify that the system has been built as designed.

This will limit the contractor's liability for failure of the system or its

components.

File of Inspection Reports Many jurisdictions require that the agency

responsible for inspections keep an up to date file of the inspection reports

for each project in a place accessible to the general public as well as to the

contractor. This allows concerned citizens to ensure that the inspections are

being carried out as required and protects the integrity of the inspection de-

partment. It is important that the contractor have access to the inspection

reports so that they may be reviewed for accuracy; many S101 policies require

-C13-

the inspection department to provide the contractor with a copy of each report.

Maintenance Inspections. Inspections to ensure proper maintenance of all

SWM facilities are necessary if the system is to function as expected. Usually,

inspections are made to ensure that pipes and channels arenot clogged with

natural or man-made debris or sediment and that vegetation, mulch, and other

sediment-controlling land covers are in good condition. Inspections for main-

tenance may be required either at set time intervals (every spring, for in-

stance) or after certain flood events (such as any, time the emergency spillway

is inundated). Underground facilities can present special problems for in-

spectors because not all of them are easily accessible; where possible, access

structures such as manholes should be incorporated into the design. If this is

not possible, it may be necessary to observe the operation of such facilities

during actual storm events to gage their performance.

Notification of Violations. It is wise to specify the method of notifi-

cation of violations in the inspection clause. Also, the time period allowed

for correction of violations should be stated. In this waY, both the developers

and the inspection agency will know how to proceed in cases where violations

occur.

Maintenance

Maintenance is one of the most important factors in a SIVM policy because

even a properly designed and constructed facility will not perform

as intended by the designer if it is not properly, maintained. The usual main-

tenance problems involve man-made and natural-debris or sediment clogging pipes

and infiltration surfaces. Another problem is the performance of erosion -res i stant

land covers and mulches. Good maintenance consists of regular cleaning of pipes

and basins and the mowing and repair of land covers.

-C14-

and to collect the funds required for this purpose from the private owners.

Bond.,; for Nlaintcnanct,. It is common practice to requirc the developer and

subsequent landowners to bear the financial burden of maintaining the SWN1

system that serves their land. Many governments require that a maintenance

bond be posted to C11SUre that the landowners will muct this obligation; the

amount of bond is usually the estimated cost of maintaining the system for a

specified time period, such as five years. This bond will ensure that the

general public does not have to pay to maintain a system that benefits some

far more than others. Bonds may be required for publicly owned, as well as

privately owned, systems if the SWN1 policy requires that the developer and

subsequent landowners pay for maintaining the system.

Responsibility for Maintenance. The responsibility, for maintaining a SIN

system can be subdivided into two components; these are the financial responsi-

bility and the administrative responsibility. Some jurisdictions require that

the landowners in the developed area assume both parts of this responsibility.

Some require only that the landowners assume the financial burden, and some local

governments assume all of the responsibility themselves. Assignment of the

administrative responsibility to the landowners is often avoided because of the

complications that arise if the landowners fail to perform the necessary main-

tenance. In most cases, the most equitable policy seems to be one requiring

that the landowners pay for maintaining the system because they derive the

primary benefit from it (that is, the use of their land). This question of

benefits should be considered in deciding who shall bear the responsibility for

maintenance. It is important to note that any policy that assigns any main- .

tenance responsibility to the landowners must include a provision stating that

this responsibility is transferable with the title to the land.

-C15-

There are two distinct approaches to assigning maintenance responsibilities

that are commonly used. In the first approach, all coml)leted SWM systems are

dedicated to a government agency and the government assumes all maintenance

responsibilities. The other approach is to assign the maintenance responsi-

bility to the landowners and require that maintenance be performed whenever

inspection reveals that the system is not functioning properly. Variations

and combinations of these two approaches are also used. Some jurisdictions

assume control of systems in residential areas but require that systems in

commercial and industrial areas be maintained by the owners. Other policies

accept the system for government maintenance but levy special taxes on the

property owners in the developed area to pay for maintenance of the system.

Any policy that assigns maintenance responsibility to the landowners must in-

clude a statement making that responsibility transferable with ownership.

Factors to be considered in developing a maintenance policy for SIN systems

are listed below.

Acceptance of Systems by the Local Government. Often the local government

feels that the only way of assuring proper maintenance of SIN systems is to

assume that responsibility itself. Standards for acceptance should be developed

in order to avoid the construction of systems that require excessive maintenance.

As noted above, maintenance should be considered in the design of SWM systems,

and designs that minimize maintenance requirements should be chosen whenever

possible when the public is responsible for maintenance.

Authority to Perform. A statement giving the local government the authority

to perform needed maintenance on privately owned systems will allow the government

to protect the public interest. Sometimes the owners of a system will fail to

perform necessary maintenance, even after being notified of violations. In such

cases, the local government must have the authority to perform the maintenance

-C16-

Stormwater N@.' nt Taxcs. A stormwater t:Ix is of'ten includcd

in policies that assign financial responsibility for maintenance to the land-

owners. This tax is an additional property tax and is usually collected in

the same manner with the same consequences for failure to pay. Such fees may

be charged regardless of whether the administrative maintenance responsibility

is assigned to the landowners, as in private systems, or assumed by, the local

government. The inclusion of a stormwater management tax is the easiest way

for many jurisdictions to force the landowners to bear the financial burden of

maintenance.

Multi-Site SWM Systems

Regional planning is essential if multi-site facilities are to be utilized

extensively and effectively. in many situations, multi-site systems are

advantageous for both technical and economic reasons. These facilities can be

designed to correct existing stormwater problems, as well as those that may be

caused by proposed dcvelopment, if tile planning agency has the authority to

participate in the design and financing of SWN1 systems. To maximize the ef-

ficiency of SIN systems, the agency responsible for SWM should have the powers

described in the following paragraphs. Unless a comprehensive regional planning

effort is to be made, multi-site systems are often not feasible because of the

time and effort required to evaluate plans for these systems.

Government Participation in SWM Systems. If the government agency respon-

sible for SM has the authority to participate in the financing and planning

of SWNI systems, then existing runoff problems can often be corrected and potential

problems can be avoided. When a developer submits a preliminary SWM plan for

a project, he may be required to construct a system capable of controlling runoff

from other areas as well; the government usually finances the increased cost of

the larger system. With comprehensive planning and government participation,

-C17-

the number and cost of SWM facilities can be, minimizcd; this increased efficiency

is expected to result in lower maintenance costs.

Runoff from areas that were developed before SWM systems were required causes

problems in many localities; such problems can often be corrected by requiring

a downstream developer to construct a system that will correct the existing

problem while also controlling the runoff from the new site. If tile comprc-

hensive plan allows for development of additional upstream areas, the downstream

developer may be required to build a system large enough to control anticipated

future conditions; this is usually economically feasible only if the upstream

areas are to be developed within a few years. If a downstream developer is

required to build an oversize S101 system in order to control runoff from other

areas, the government should provide financing and desion assistance. In tile

case where the oversize system is needed to control anticipated future upstrean'

development, the government can recover the costs of tile larger system from

the upstream landowners when they apply for permission to develop their land.

Where the oversize system is needed to control ;in existing runoff problem,

the government generally absorbs Lhe cost because tile correction benefits the

general public. By participation in the financing and planning of facilities

the government can develop an efficient SWN1 system for its entire jurisdiction.

Planning Responsibility for Multi-site Systems. The government is often the

only party with the knowledge, authority, and staff required to design multi-

site systems, but in some cases multi-site systems are proposed and designed

by developers. A single developer may have two or more sites in one locality

and his engineers may propose a multi-site system; alternately, two or more

developers with adjoining property may decide to cooperate in building a multi-

site system. Therefore, while the government should bear the primary responsi-

bility for design of multi-site systems, developers should be encouraged to

_C18-

initiate multi-site planning where they find it to be feasible. However, the

government must review all of the developers' 111,111S to C11SUre that tile goals

of SWM are met before giving construction authorization.

Waivers of On-Sit6 SWNI Requirements. In areas where multi-site systems

are used, some developers are not required to provide complete on-site SWM

systems because the runoff from their sites is controlled elsewhere. In these

cases, on-site requirements are waived and donations of money or land are

usually required. The funds collected from these developers are used to

finance the construction of multi-site systems. Donations of land may be re-

quired in order to accommodate larger SWM systems or for other public uses.

In multi-site areas, the government must have the authority to waive on-site

requirements in favor of donatjOTIS.

Permits and SWM Plans

Most local jurisdictions have the authority and responsibility to require

permits for construction and development. Permits are usually issued only after

a review of the design plans submitted by the developer. Once the reviewing

agency has evaluated the plans and determined that the goals of the SWM regu-

lations will be achieved, building and land use modification permits can be

issued. Some jurisdictions require permits for any land-diStUrbing activity

while others only require permits for construction; in most populated areas,

any change in land use can result in storm runoff problems, so permits should

be required for all such activities. Specific elements of a comprehensive SWM

policy that concern the permit process are discussed below.

Evaluation of Plans. In order to obtain a permit, the developer must pro-

vide a set of plans of the SWM facilities he intends to use to control any

potential runoff problem. A government technical agency such as the DPW,planning

_C19-

board, or local soil conservation district must be assigned the responsibility

for evaluating SWM plans. A specific set of design criteria must be developed

and specified in the SIVN regulations so that there is.no ambiguity concerning

the requirements. Permits for development should only be issued after an

evaluation to ensure that the system will function as designed.

Exemptions from Permit Requirements. Projects of less than a specified

size should be exempt from the permit requirements because the effects of

land use changes on a small scale is usually negligible. The exact circum-

stances that warrant an exemption should be a function of the degree of im-

perviousness proposed for the future land use a,, well as the size of the area.

The construction of a single house does not usually require a S101 plan,but a

small parking lot may have the hydrologic impact of inany detached houses.

Expiration and Renewal of Permits. Permits should have an expiration

date to ensure speedy completion of projects; this feature will decrease the

potential for erosion because the area will be permanently stabilized faster.

ln addition, temporary stabilization with straw, mulch, or vegetation should

be used whenever possible during the construction period. Renewal of permits

should not be automatic. Renewals should only be approved when they will bene-

fit the public, but appeals of denials to an authority specifiedin the permit

regulations should be allowed.

Fees for Review of Plans. Many governments charge the developer a fee

for reviewing and evaluating SWM plans. Usually this fee is intended to off-

set the actual costs to the government. These fees are imposed to ensure that

.the general public is not forced to subsidize the costs of development. The

amount of the fee may be a function of the size of the area being developed

and of the proposed land use, but some jurisdictions have one set fee for all

plans.

-C20-

Liability Insurance. Developers are sometimes required to obtain liability

insurance before a permit can be issued; this requirement is intended to reduce

the possibility of the developer failing to complete development due to financial

problems; thus, this requirement protects the public. Construction sites are

often hazardous, and there have been cases where developers have been bankrupted

by suits for damages. In these cases, the project is not likely to be completed

in a timely manner and stormwater erosion on the disturbed area can become a

problem. Therefore, liability insurance may be required to ensure the financial

well-being of the developer, which will in turn help to ensure speedy completion

of the project.

Modifications to Plans. Either the developer or the government may find it

desirable to modify development plans at any time during the review and con-

struction period. At the review and evaluation stage of the permit application

process, the reviewing authority may demand changes in order to meet the goals

of on-site SWM or to provide for multi-site systems. During the construction

period, conditions are often found to be different from those assumed in pre-

paring and reviewing the plans; depths of soil layers, availability of materials,

and water table levels may cause problems with the construction based on approved

plans. In cases where conditions are not as expected, either the developer or

the government may modify the plans. The developer should be required to secure

permission for any necessary modifications; permission should be obtained before

construction of the proposed change is made.

in many jurisdictions the permit application process requires submission

of both preliminary and final design plans. The preliminary plan is reviewed

by the government authorities to ensure compliance with all pertinent regulations;

then the plans are returned to the developer with the comments from the reviewing
C.

agency. Only then can the developer prepare his final plans, which must be sub-

-C21-

nd rted for a re-eva ILI It i011 I) el'ore permits may be i F's I i0d.

Performance Bonds. To ensure that projects that. are started are also

completed, a performance bond is required before permits may be issued in

many jurisdictions. The amount of this bond is usually the estimated cost

of completing the constrLlCtiOll of the required SWNI facilities. Such hunds

are forfeited to the government if the system is not completed within the

period of time specified on the permit unless an extension or renewal is

granted. The money collected in such cases of forfeiture is used to complete

the system. Performance bonds thus help to ensure that the public will be

protected from the adverse effects of the failure of a developer to complete

his project. In man)- jurisdictions, provisions are made for the release of

portions of the performance bond upon partial coiiipletion of the required SWINI

system.

Responsibility for Compliance with Permit Conditions. The responsibility

for ensuring that the project is constructed as specified in the design plans

should be assigned to the developer. While the government may inspect the con-

struction and contractors may perform the actual work, the periiiit should be

issued to the developer and compliance must be his responsibility. In cases

where a performance bund is required, this assigiiiiient of responsibility is

implicit; in other cases, a clause specifying the developer's responsibility

should be included in the SWM policy.

Prevocation and Suspension of Permits. The government may find it necessary

to revoke or suspend development permits if the construction is not proceeding

in accordance with the approved plans. Regular inspections should be performed

during the construction period, and if the conditions of the permit are not

being met, then the developer should be notified. If violations are not cor-

rected within a reasonable period of time, stop-work orders may be used to pre-

-C22-

vent the construction from proceeding. These stop-work orders are equivalent

to temporary permit suspensions; if the developer fails to demonstrate his

intent to correct the violations and complete the project as designed, his

permit should be revoked. The power to suspend and revoke permits is one

of the necessary tools for enforcement of SM11 regulations.

Situations That Require Permits. A complete SWIM policy must state exactly

which situations require that permits be obtained. This statement will result

in a less ambiguous policy and minimize the number of misunderstandings. A

common way of explaining which situations require permits is to state that

permits are necessary for all land use alterations affecting areas greater

than a certain minimum size. Exemptions to these guidelines can then be listed

separately. Exemptions are commonly given for changes in agricultural use or

for projects involving a single private residence.

Specifications for Plans. In order for the reviewing agency to accurately

evaluate the effects of a given design, the plans must include certain features.

Complete descriptions of facilities, including material specifications, are

required; the use of a standardized format will also expedite the review.

Many jurisdictions require that plans be of a certain size, with blocks for

the approval authorities to indicate their findings and comments on the plans.

Also, many jurisdictions require that the plans be certified by a registered

professional engineer before being submitted. The reason for these requirements

is to facilitate the plan evaluation and thus hell) to ensure that the develop-

ment will not adversely affect the general public.

Waivers and Appeals. In cases where preliminary investigation shows that

a proposed change in land use will not have detrimental results, there is no

reason to require a permit. Waivers from the permit requirements should be

-C23-

issued in such cases in order to ease the burden on both the developer and

the permit agency. The most common way of providing for these waivers is to

require that the developer demonstrate to the permit agency that the effects

of his proposed project will either not be detrimental or will be insignifi-

cant. If the developer can demonstrate this to the satisfaction of the govern-

ment, a waiver may be granted. It is wise to specify the procedure for appeal

when waiver requests are denied in the SWM regulations, because the situation

in which a developer is denied a waiver that he feels he deserves is almost

certain to arise.

Zoning A2proval. Before issuing permits for development, a consultation

with the local zoning authority should be mandatory. This consultation should

ensure that no zoning regulations will be violated by the proposed development

so that the goals of the zoning statutes are met.

Grandfather Clause

MO'S't SWM policies include a so-called grandfather clause in order to avoid

legal complications. A grandfather clause merely provides exemptions to the

provisions of the policy for projects that have already reached a certain stage

of completion by the date the SWM regulations become effective. A developer

who has already received approval for a project under the old regulations is

likely to protest if informed that the project must be reevaluated in order

to comply with a new �et of regulations. In addition, a grandfather clause

is often included because the reviewing agency is likely to be overloaded with

work if required to review all the previously approved projects that have not

yet been completed.

Severability of Provisions

A clause stating that each element of the SWM legislation is an inde-

pendent statute is often included in such legislation. These clauses state

-C24-

the principle of severability of the provisions, meaning that if one provision

is found to be unenforceable (due to lack of authority, unconstitutionality,

or other reasons) the other provisions remain in force. This clause strengthens

the legislation by ensurin@-- that minor faults will not result in the entire

ordinance being invalidated.

-C25-

APPENDIX D

EVALUATION OF METHODS OF
DETERMING URBAN RUNOFF CURVE NUMBERS

Walter J. Rawls

Ahlam I. Shalaby

Richard It. McCuen

Evaluation of Methods for Determining
Urban Runoff Curve Numbers
7,

Walter J. Rawls, Ahlam Shalaby, Richard H. McCuen _7
MEMBER
ASAE

ABSTRACT tion 4 (SCS, 1972) lists the many soil types and their
M
HE runoff curve number, which integrates land corresponding hydrologic soil group. The soil type can be
over, soil type, and antecedent soil moisture condi- identified from soil surveys which are available from
tions, is an important input parameter to the Soil Con- local SCS offices. Land cover is separated into the
,@ Z
versation Service (SCS) hydrologic procedures. The categories of Table 1, which also shows the CNs for C;
average soil moisture conditions (antecedent soil
determination of curve numbers is a time consuming
moisture condition 11).
procedure requiring detailed data on land cover and
soils. Data from 175 urban watersheds in various sec- A watershed is rarely composed of both homogeneous
7
tions of the United States were used to evaluate whether land cover and soils of the same hydrologic soil group.
runoff curve numbers developed by measuring the frac- Therefore, it is necessary to integrate the soils data with
tions of land use and soils separately and integrating the land use information to derive a value of the runoff
them using a weighted average scheme (the contingency CN for the watershed or subarea. The determination of'
table approach) or using land use maps developed by the CNs is often a time consuming procedure because of the
U.S. Geological Survey (USGS) were comparable to the way that land cover and soils data are integrated.
curve numbers developed by the conventional SCS Several investigators (Ragan and Jackson, 1980; Slack F
methods. The results indicated that estimates of runoff and Welch, 1980; Bondelid et aL, 1980) have investi-
curve numbers are sensitive to the land use classification gated the use of Landsat data for estimating curve
system and not very sensitive to the method of integrating numbers. Jackson and Ragan (1977) showed that land
soils and land cover data for watersheds of 0.1 to 472 cover data can be obtained in a cost effective manner us- 41
kml. Also, from the data set, a set ofcurve numbers was ing Landsat. The past studies using Landsat were per-
developed for the four hydrologic soil groups for level II formed with watersheds in Georgia, Maryland and Penw.
USGS land use categories. sylvania and included a range of land covers such as Z:
forest, agriculture, urban, strip mine and wetlands. The
INTRODUCTION investigators found that Landsat data could not be used
Many of the hydrologic models used in the design and to identify land cover at the level of detail described in
planning of urban water resources projects require input Table 1. However, Ragan and Jackson (1980) found that
parameters that are defined in terms of land cover so that by using a less detailed land cover classification system
alternative forms of development can be planned and developed for Landsat data, the curve number could be
future changes assessed. The series of hydrologic models estimated with an accuracy of two curve numbers.
developed by the Soil Conservation Service (SCS, 1969, Bondelid et al. (1980) investigated the use of the land
1972, 1975) are among the most widely used models in cover maps developed by the USGS for three rural water-
water resources planning and design. Originally, they sheds in Pennsylvania. They studied the USGS land use
maps because these maps are available for about 80 per-
were developed for agricultural areas; however, they have cent of the United States. Bondelid et al. (1980)
been extended for use in urban areas (SCS, 1975). These estimated curve numbers for the USGS land use classifi-
models require a value of the runoff curve number (CN), cation system and found that for the rural study area the
which is an indication of the flood runoff potential of a ,
watershed. The CN is a function of the land cover, the Qata produced curve numbers comparable to those pro-
soil type and the antecedent soil moisture conditions, duced by conventional methods.
Four hydrologic soil groups (A, B, C, D) are used to The objectives of our study were to evaluate whether
reflect variation in runoff potential for different soil urban runoff curve numbers developed by integrating
types. The National Engineering Handbook (NEH) Sec- land use and soils separately using a weighted average
scheme (contingency table approach) or using USGS
land use maps were comparable to the runoff curve
numbers developed using the conventional SCS ap-
Article was submitted for publication in November 1980: reviewed proaches-
and approved for publication by the Soil and Water Division of ASAE
in May 1981. Presented as ASAE Paper No. 80-2565. METHODS FOR INTEGRATING LAND USE
Contribution of' the USDA-SEA-AR Hvdrolog@ Laboratory.
AND SOILS DATA
Beltsville, MD and the Department ot'Civil Engineering, Unkersity of
Maryland, College Park, MD. The SCS hydrologic methods are very sensitive to the
The authors are: WALTER J. RAWLS. Hydrologist. USDA-SEA- CN (Hawkins', 1975). Therefore, it is important to obtain
AR, Hydrology Laboratory, Beltsville, MD: AHLAM SHALABY, estimates of the CN that accu ately reflect the runoff
Graduate Research Assistant, and RICHARD H. McCUEN. Pro- r
fessor, Department of Civil Engineering. University of Maryland. Col- potential of a watershed. Three critical aspects of
lege Park, MD. estimating the CN are the determination of soils, the

-D1 -
1562 TRANSACTIONS ofthe ASAE-1981

TABLE 1. RUNOFF CURVE NUMBERS FOR SELECTED AGRICULTURAL, SUBuRBAN. AND URBAN
LAND USE. (ANTECEDENT MOISTURE CONDITION 11, and 1, -.2S)
(FROM SCS TR-55 1975)

Hydrologic soil group

Land use description A B C D

Cultivated land*: Without conservation treatment 72 81 88 91
With conservation treatment 62 71 78 81
Pasture or range land: poor condition 68 79 86 89
good condition 39 61 74 80
Meadow: good condition 30 58 71 78
Wood or forest land: thin sstand, poor cover, no mulch 45 66 77 83
good covert 25 55 70 77
Open space (lawns, parks, golf courses, cemeteries, etc)
good condition: grass cover on 75% or more of the area 39 61 74 80
fair condition: grass cover on 50% to 75% of the area 49 69 79 84
Commercial and business area (85% impervious) 89 92 94 95
Industrial districts (72% impervious) 81 88 91 93
Residentialt:
Average lot size Average % impervious �

1 /8 acre or less 65 77 85 90 92
1/4 acre 38 61 75 83 87
113 acre 30 57 72 81 86
1 /2 acre 25 54 70 80 85
1 acre 20 51 68 79 84
Paved parking lots, roofs, driveways, etc.11 98 98 98 98
Streets and roads:
paved with curbs and storm sewersii 98 98 98 98
gravel 76 85 89 91
dirt 72 82 87 89

For a more detailed description of agricultural land use curve numbers refer to National Engineering Handbook,
Section 4, Hydrology, Chapter 9. Aug. 1972.
t Good cover is protected from grazing and litter brush cover soU.
Curve numbers are computed assuming the runoff from the house and driveway is directed towards the street with
a minimum of roof water directed to lawns where additional infiltration could occur.
� The remaining pervious areas (lawn) are considered to be in good pasture condition for these curve numbers.
11 In some warmer climates of the country, a curve number of 95 may be used.

determination of land cover and the method used for in- Li = the fraction of land use i, Sj = the fraction of soil
tegrating the land cover data with the soils data. group j (i.e., A, B, C or D) and CNii = the curve number
associated with Li and Sj. This method will be identified
The Conventional Approach as the contingency table approach.
The conventional method of determining curve
numbers involves low attitude aerial photographs, land ESTIMATING LAND USE WITH THE USGS
use maps and onsite investigations. To integrate land LAND USE SYSTEM
cover and soils data, the hydrologic soils maps are The USGS is currently producing land use and land
overlaid on aerial photographs or quad sheets that cover maps for the United States with a base of
delineate the land cover. The area for each land cover- 1:250,000. The data are being digitized and will be made
soil type complex is planimetered, and a curve number available in both graphic and digital form. This
for the entire watershed or a subwatershed is estimated Geographic Information Retrieval Analysis System
by weighting the CNS for each unique land cover-soil type (GIRAS) is designed to manipulate, analyze and output
complex. the land use and land cover data (Mitchell et al., 1977).
The Contingency Table Approach The maps currently are available for level 11 classifica-
We hypothesized that the effort required to estimate tion. A minimum mapping unit of about 4 ha (10 acres)
watershed CNs could be reduced, without an appreciable is used for all urban and water covers and a few other
loss of accuracy, by replacing the overlaying of soils and categories. For all other cover types a minimum unit of
land use delineation with a separate analysis of soils and 16 ha (40 acres) is used. Each spatial mapping unit must
land use data; this would make it possible to use com- be assigned to a single category. Because these maps are
puterized data bases of land use and soils data. readily available for a major portion of the United States
Specifically, the percentage of each soil group would be and are frequently updated, they provide a convenient
determined independently from the percentages of the alternative to the task of obtaining land use data from in-
different land covers. A weighted mean CN would then dividual sites by collection and interpretation of low level
be determined using the product of three values: the aerial photographs.
percentage of land use, the percentage of a soil group DATA BASE
and the CN associated with the soil and land use:
Land cover and soils data from 175 urban watersheds
n 4 were used to investigate the study objectives. The
CNw = Z Y Li Sj CNij . . . . . . . . . . . . . . . . . . . . . distribution of watersheds over the United States is given
j=1 in Fig. 1. The size of the watersheds ranged from 0. 1 to
472 km' with a mean size of 41.4 km'. The land cover,
in which CN. = the weighted CN for a watershed, n = soils data and conventionally determined curve numbers
the number of different land uses within the watershed, were compiled by SCS state hydrologists for the particular

1981 -TRANSACTIONS of the ASAE 1563

TABLE 2. ritKQtJKNCV DISTRIII(JTION OF SCS IIVDROI,O(;IC
'S SOIL GROUPS FOR THE WATERSHEUN

V llawdoit 175 %vatersheds Bast-d Im 120 waterOwt1%
Hydrologic Standard Standard
.7 soil groiip Mean, de@iation, % M-1, %, deviation, 'I,,

1.46 5.52
A 1.91 6.93
B 36.66 33.151 32.04 31.72
C 39.84 35.04 42.43 34.06
D 21.59 32.55 24.07 36.08

HAWAII 40NOLULV (5),
PZ RL CITY' AA ".11 of the 175 watersheds is given in Table 3. Approximately
KAANEHOC. HILO
44 percent of the land use is classified as residential.
Other significant land use types are pasture, forest, open
4ymbol d-0116 .1 LSOS L&hd C11#F M-1, space and commercial.
The USGS developed a land use classification system
FIG. I Metropolitan areas included In the land use study. that includes several levels of detail (Anderson et al.,
1976). Table 4 shows the level 11 land use categories used
in the study. The data base included 120 watersheds for
state in which the watershed was located. The SCS nor- which USGS land use data were available. Table 5 shows
mail), uses the curve numbers shown in Table 1; however, the frequency di 'stribution of land use in the 120 water-
local CN values that differed from those in Table I were sheds on the basis of USGS land use classification. Ap-
used for ten ofthe 175 watersheds. Five of the ten water- proximately 51 percent of the land use is classified as
sheds are located in the state ofWashington. Differences residential. Other significant land use types are commer-
in CNs covered the spectrum of land cover types, in- cial, urban and forest. I
cluding pasture, forest. residential and commercial. The EVALUATION OF CURVE NUMBERS ESTIMATED
use of different curve numbers appears to be location USING THE CONTINGENCY TABLE APPROACH
dependent because of differences in runoff potential for
similar land covers in different parts of the country. The conventional SCS approach for estimating curve
USGS land cover data were available for 120 of the 175 numbers was used as the true value for determining the
watersheds. These 120 watersheds are identified accuracy of CNs estimated using the contingency table
separately in Fig. 1. The USGS data were compiled by approach, although CN values obtained using the con-
hydrologists of the USGS through a cooperative project ventional approach are also estimates and, thus, are sub-
with the Federal Highway Administration (Sauer et al., ject to error. Use of the CNs estimated by the conven-
1980). The size of the watersheds ranged from 0.8 to 472 tional approach as a standard was the most feasible
km' with a mean size of 47.0 kml. The frequency criterion for estimating the accuracy of CNs derived us-
distribution of the SCS hydrologic soil groups is given in ing the contingency table approach. The residuals
Table 2 for both the 175 and 120 watershed groupings. (difference between the conventional and contingency
table approach curve numbers) were computed for each
Land Use Distributions of the 175 watersheds; the distribution of the residuals is
The frequency distribution for the SCS land use data given in Fig. 2. The residuals had a mean of -0.064,

TABLE 3. FREQUENCY DISTRIBUTION AND CORRELATION COEFFICIENTS OF THE RESIDUALS
OF THE SCS LAND USE FOR THE 175 WATERSHEDS.

Standard Correlation
Land use description Mean. % deviation, % Coefficient

Cultivated land: Without conservation treatment 1.50 6.64 0.00
Cultivated la-nd: With conservation treatment 1.51 5.68 0.14
Pasture or range land: poor condition 2.94 11.28 0.00
good condition 6.29 14.01 0.09
Meadow: good condition 2.14 6.69 -0.01
Wood or forest land: thin stand, poor cover, no mulch 5.42 12.93 -0.02
Wood or forest land: good cover 10.47 18.99 -0.08
Open space (lawns, parks, golf courses, cemeteries, etc)
good condition: grass cover on 75% or more of the area 6.23 9.93 -0.18
fair condition: grass cover on 50% to 75% of the area 3.10 8.03 -0.03
Commercial and business area (85% impervious) 6.84 9.03 0.03
Industrial districts (72% impervious) 4.91 9.77 0.02
Residential:
Average lot size Average % impervious
1/8 acre or less 65 10.17 17.79 0.03
1/4 acre 38 16.71 23.53 0.02
1/3 acre 30 5.66 12.82 0.02
1/2 acre 25 6.43 13.71 0.00
1 acre 20 5.06 13.59 -0.01
Paved parking lots, roofs, driveways, etc. 1.71 3.29 0.03
Streets and roads: 2.89 4.29 0.02
paved with curbs and storm sewers
gravel
dirt

1564 -D3- TRANSACTIONS of the ASAE-1981

TABLE 4. USGS LEVEL 11 LAND COVER CURVE NUMBERS. TABLE 5. FRKQUFNCY DISTRIBUTION FOR USGS LAND
USK DATA t'011. Tilt 120 WATERSHEDS
Hydrologic soil group
Sto:iflanl
Land use description* A B C D Land use description ti"n""..
Residential 61 70 8:1 87 Residential MMM 22.75
Commercial and services 87 89 91 93 Commercial and services 10,63 7.87
Industrial 81 88 91 93 Industrial 1.74 3.23
Transportation, communications, 84 87 90 92 Transportation, communications, utilities 2.32 3.82
utilities Industrial and commercial 0.89 2.50
Industrial and commercial 83 85 89 91
Mixed urban or built-up land 55 71 83 88 Mixed urban or built-up land 0.93 2.93
Other urban or built-up land 72 78 84 88 Other urban or built-up land 5.34 5.72
Crop or pasture 63 69 77 82 Crop or pasture 12.83 20.18
Orchards, groves, nurseries, and 23 52 69 75 Orchards, groves, nurseries, arid 0.18 1.17
ornamental horticultural areas ornamental horticultural areas
Deciduous forest land 55 63 71 76 Deciduous forest land 5.81 12.84
Evergreen forest land 35 60 73 79 Evergreen forest land 1. 15 4.54
Mixed forest land 43 64 7 r) 81 Mixed forest land 4.88 12.14
Lakes 100 100 100 100 a es 0.04 0.26
Reservoirs 100 100 100 100 Reservoirs 0.09 0.25
Forested wetland 45 66 77 83 Forested wetland 0.04 0.45
Strip mines, quarries. and gravel pits 74 82 87 89 Strip mines, quarries, and gravel pits 0.29 1.38
Transitional areas 63 74 81 84 Transitional areas 1.02 2.52
Shrubs and brush 0.87 4.83
Shrubs and brush 49 69 79 84 Other agricultural 0.0075 0.04
Other agricultural 59 74 82 86

*Ordy the level 11 categories used in the study are included.
-0.062, which is not significant. However, Fig. 2, a
which indicated the absence of a significant bias, and a graph of the residuals versus drainage area size, shows
standard deviation of 1.50, which indicated a small level an apparent trend toward wider variation for the smaller
of imprecision. drainage areas, especially those smaller than about 26
kml. The correlation coefficients were also computed be-
In an attempt to identify the cause of the imprecision, meen the residuals and the percentages of each of the
we first analyzed the 12 most extreme residuals to deter- SCS land covers; the resulting values are given in Table
mine if the error variation was due to a bias in drainage 3. The two largest correlation coefficients are 0.18 and
area size, soil type or land cover. No bias was found due 0.14, with 14 of the 18 values less than or equal to 0.03
to drainage area size, soil type or land use. indicating also land cover did not appear to cause the im-
The second analysis involved correlation analyses be- precision..
tween the residuals and drainage area and land use. The
correlation coefficient for the 175 residuals and the EVALUATION OF CURVE NUMBERS ESTIMATED
drainage areas of the corresponding watersheds was USING USGS LAND COVER
The data for the 120 watersheds were used to define
the CNs for the level 11 land use categories shown in
Table 4. Using the description of the SCS and USGS
land use categories, the SCS categories were grouped to
correspond to the USGS categories and curve numbers
were derived maintaining internal consistency between
soil groups. When these curve numbers were used on the
120 watersheds, they were found to have a positive bias of
approximately one curve number and an imprecision of'
approximately 3.5 curve numbers when compared with
4 the curve numbers that were estimated using the conven-
tional approach. Thus, the initial estimates of the curve
numbers using USGS land use data were adjusted until
the bias was reduced to near zero and the imprecision
was made as small as possible. The best-fit curve
numbers are shown in Table 4.
Because the same data base that was used to fit the
0 (1 41 1@ !100
AREA CNs for the USGS land use classification system was
used to estimate the bias and precision, an unbiased
measure of the accuracy of the curve numbers is not
available. However, because the data base included such
-4 r diversity ot'geographic location, degree of urbanization
and soil type, the CN values of Table 4 may be con-
sidered reasonablv accurate for making runoff computa-
-6 tions in the future; that is, the estimates of bias and
precision are good estimates ofthe accuracy obtainable
-7 when using the values of Table 4. In comparing the CNs
_8@ derived using the USGS land use system with the CNs
FIG. 2 Plot of residual curve numbers vs. the drainage area size for the derived conventionally, the bias was near zero and the
175 watersheds. imprecision was approximately 3 curve numbers. which

1981 -TRANSACTIONS ofthe ASAE -D4- 1565

use and soils data in the estimation of SCS runoff curve
numbers and simplifying the method of obtaining land
use data.
The conventional approach to CN estimation involves
overlaying land use and soils data. Because planimeter-
ing areas having homogeneous land use and soils is very
time consuming, we investigated the accuracy of CNs
estimated by integrating independently the estimates of
the percentages of each soil group and the land use. The
resulting estimates of CNs were accurate, indicating that
4 ..1 the detailed conventional approach may not* be neces-
sary. However, the results suggested that the conven-
2 ;J.. tional approach should be used for watersheds of less
than about 26 km'. This new method of integrating the
land use and soils data is important because the indepen-
0
-0 200 300 @00 @00 dent estimation of the two inputs would make possible
-2
AR:A (I-) the use of computerized data bases in determining curve
numbers.
-4 Land use maps developed by the USGS are readily
available and frequently updated, thus providing a cor@
venient alternative to the task of obtaining land use data
-6
from individual site collection and/or low level aerial
photograph interpretation. Reasonable estimates of the
watershed CN, especially for watersheds 'larger than
about 26 km', were made using a set of curve numbers
developed for the level 11 USGS land use categories.
While the CNs were not in total agreement with those
FIG. 3 Plot of residual curve numbers vs. the drainage area size for the estimated from conventional field inspection, the level of
120 watersheds. imprecision was satisfactory for most types of hydrologic
analysis.
must be considered significant, since it can cause up to
about a 15 percent error in runoff predictions. The dis- e erences
tribution of the residuals is shown in Fig. 3 for the 120 1 Anderson, J. R., E. E. Hardy, J. T. Roach, and R. E. Witmer.
watersheds. Approximately 91 percent of the residuals 1976. A land use and land cover classification system for use with
were within 6 curve numbers of the conventionally de- remote sensor data. USGS Professional Paper 964. Washington, DC.
rived values. pp. 10-22.
2 Bondelid, T. R., T. J. Jackson, and R. H. McCuen. 1980. Com-
Analysis of the 12 most extreme residuals indicated a parison of conventional and remotely sensed estimates of runoff curve
tendency for the watersheds corresponding to the largest numbers in Southeastern Pennsylvania. Proceedings of the American
residuals to contain less residential and industrial land Society of Photogrammetry Annual meeting. St. Louis, MO.
3 Hawkins, R. H. 1975. The importance of accurate curve
and corresponding more than average of forest land. An numbers in the estimation of storm runoff. Water Resources Bulletin.
attempt at decreasing the level of imprecision by further American Water Resources Association. pp. 887-891.
adjusting the CN values of Table 4 was made with no suc- 4 Jackson, T. J.. and R. M. Ragan. 1977. State-of-the-art ap-
cess. Although the cause of the imprecision could not be plic.ations of Landsat data in urban hydrology. International Sym-
posium on Urban Hydrology, Hydraulics and Sediment Control.
identified conclusively, the limited separation of residen- University of Kentucky, Lexington, KY. pp. 275-285.
tial land use seemed to be a major factor. Separation of 5 Mitchell, W. B., S. C. Guptill, K. E. Anderson, R. G. Feageas,
the USGS residential land use into categories according and C. A. Hallam. 1977. GIRAS: A geographical information retrieval
to lot size in a manner similar to the SCS residential land analysis system for handling land use and land cover data. USGS Pro-
use separation would, perhaps, reduce the imprecision to fessional Paper 1059.
6 Ragan, R. M. and T. J. Jackson. 1980. Runoff synthesis using
the USGS land use CNs. Landsat and the SCS model. Journal of the Hydraulics Division.
A correlation analysis between the 120 residuals and ASCE.
either drainage area or the USGS land use categories was 7 Sauer, V. B., W. 0. Thomas, Jr., V. A. Stricker, and K. V.
also performed in an effort to reduce the imprecision. Wilson. 1980, A nationAide study of urban flood frequency: a
preliminary report. Proceedings ot'International Symposium on Urban
This analysis indicated that the size of the drainage area Storm Runoff. University of Kentucky. Bulletin UKYBUT 121. Lex-
or the USGS land use categories did not cause the im- ington, KY. pp. 37-42.
precision. However, Fig. 3, a graph of the residuals ver- 8 Slack, R. B. and R. Welch. 1980. Soil Conservation Service
sus drainage area size, shows an apparent trend toward runoff curve number estimates from Landsat data, submitted to Water
Resources Bulletin.
wider variation for smaller drainage areas, especially 9 Soil Conservation Service. 1969. Computer program for project
those smaller than about 26 km'. formulation hydrology. Technical Release No. 20. USDA-SCS,
Washington, DC.
SUMMARY AND CONCLUSIONS 10 Soil Conservation Service. 1972. National engineering hand-
book section 4: Hydrology. USDA-SCS, Washington, DC.
The objectives of this study were centered around two 11 Soil Conservation Service. 1975. Urban hydrology for small
problems: reducing the effort required to integrate land watersheds. Technical Release No. 55. USDA-SCS, Washington. DC.

1566 -D5- TRANSACTIONS ofthe ASAE-1981

APPENDIX E

COMPARISON OF URBAN FLOOD
FREQUENCY PROCEDURES

Walter J. Rawls

Stanley L. Wong

Richard H. McCuen

EVALUATION OF THE SCS URBAN PEAK FLOW METHODS
Richard H. McCuell 2/' WaLter J. Rawls V, and Stanley L. Wong 4/

M. ASCE M. ASCE

ABSTRACT

The peak flow hydrologic methods in the SCS TR-55 are widely used for

planning and design on small urban watersheds. The urbanization factors of

the TR-55 methods (i.e., the lag factors for the graphical method and the peak

factors for the chart method) were apparently not directly based on analyses of

measured data and have not previously been systematically tested. Therefore,

the objective of this study was to test the TR-55 methods and recalibrate the

urbanization factors if the methods were found to be biased. A data base was

compi.led for 51 small urban watersheds under 4,000 acres (1,600 ha) in the

United States. The annual maximum flood series were used to develop Log

Pearson Type III frequency curves, from which peak discharge estimates were

estimated for the following return periods: 2-year, 5-year, 10-year, 5-year,

50-year, And 100-year. The bias and accuracy of the five methods were

evaluated using the Log Pearson Type III estimate as the expected true value.

The results indicate that no urbanization correction factors were needed for

the graphical method when time of concentration was determined using the

velocity method. Modifications to the urban adjustment factors used in the

chart method were proposed.

Key Words: Hydrology, urban floods, frequency analysis, runoff, drainage,

stormwater, peak flow.

V Contributions from the USDA-ARS, Hydrology Laboratory, Beltsville,
Maryland, and the University of Maryland, Department of Civil Engineering,
College Park, Maryland
2/ Professor, Department of Civil Engineering, University of Maryland,
College Park, Maryland 20742
3/ Hydrologist, USDA-ARS, Hydrology Laboratory, Beltsville, Maryland 20705.
4/ Graduate Research Assistant, Department of Civil Engineering, University
of Maryland, College Park, Maryland 20742

-El-

EVALUATION OF THE SCS URBAN PEAK FLOW METHODS I/
2/ 1/ 4/
Richard H. McCuerr ,Walter J. Rawls and Stanley L. Wong@

M. ASCE M. ASCE

INTRODUCTION

Man's impact through urbanization dramatically affects the flood response

of a watershed. The conversion from pervious to impervious surfaces typically

inhibits infiltration and groundwater recharge and reduces surface roughness,

surface retention, and depression storage. Thus, rainfall losses are reduced

and direct storm runoff is increased. The alterations and improvements made

to the existing drainage networks cause a more rapid runoff response because

of increased flow velocities over smooth surfaces to drainage inlets and then

by pipe to improved natural channels. Thus, the impervious surfaces and

improved drainage systems increase the amount of runoff and reduce flow travel

times, which produces a flood hydrograph of increased magnitude with a

shortened time-to-peak and recession limb.

With urban growth and development, there is an ever-increasing need for

flood information and estimating techniques for use in areas where little or

no data exist (6). Therefore, design flood estimation procedures capable of

Contributions from the USDA-ARS, Hydrology Laboratory, Beltsville,
Maryland, and the University of Maryland, Department of Civil Engineering,
College Park, Maryland
2/ Professor, Department of Civil Engineering, University of Maryland,
College Park, Maryland 20742
3/ Hydrologist, USDA-ARS, Hydrology Laboratory, Beltsville, Maryland 20705.
4/ Graduate Research Assistant, Department of Civil Engineering, University
of Maryland, College Park, Maryland 20742

-E2-

accounting for urbanization are required at two levels (1). First, simple

procedures are needed both to design drainage facilities on small basins and

to evaluate alternative developmemt schemes. Second, more detailed models are

needed to design major drainage facilities. The simple procedures should

produce a flood discharge of a given frequency, while the complex models

should compute and route the design flood hydrograph through. the drainage

system. A recent survey (6) of Federal agencies, State highway departments,

and the private sector indicated that about 80 percent of hydrologic design

problems are currently evaluated using "simplified" hydrologic methods.

The Soil Conservation Service TR-55 (4) presents two methods for estimating

peak discharge rates in urbanizing areas. One method, identified as the

graphical method, is described in Chapter 5 of TR-55. The second method,

identified as the chart method, is described in Chapter 4 and appendices D and

E of TR-55. These methods are widely used because of their national

applicability and computational simplicity (1). The SCS TR-55 procedures are

based on well known hydrologic techniques (4), with the addition of ad justment

factors to account for increased imperviousness and modifications to the

hydraulic length of channels resulting from urbanization. Although the SCS

hydrologic procedures have been extensively used, the validity of the

urbanization adjustments have not been verified. Hence, the objective of this

study was to evaluate the SCS TR-55 urban peak discharge methods using data

from urban locations throughout the United States, and if the methods are

found to be inaccurate or blased, then they will be recallbrated.

-E3-

SCS TR-55 PROCEDURES

Graphical Method

Input requirements for the Craphical method are the time of concentration

(hours), the drainage area (sq. mi.), and the depth of precipitation (inches)

for the 24-hour SCS type II storm distribution. A graph is provided (Fig. 5.2

of TR-55) relating the unit peak discharge (cubic feet per second per square

mile per inch of runoff) to the time of concentration for the SCS type II,

24-hour storm distribution. The volume of runoff is computed using the SCS

rainfall runoff relationship.

The SCS recommends a velocity method for estimating the time of

concentration. This method is based on the velocity of flow and the travel

length and requires as input the hydraulic length, slope, and land use type.

The average flow velocity should be determined for each segment of the flow

path using the average slope and land use. The time of concentration for each

flow segment equals the ratio of the hydraulic flow length to the velocity.

The watershed time of concentration equals the sum of times of the

concentration of the flow segments. TR 55 does not specify any limita tions on

the drainage area when using the velocity method to calculate the time of

concentration for use with the graphical method.

The volume of direct runoff required for the graphical method is computed

using the depth of the 24 hour precipitation for a given recurrence interval

and the runoff curve number (CN). The curve number is a function of the land

use, the hydrologic soil group, and the antecedent soil moisture condition.

An estimated peak discharge is computed as the product of the unit peak

discharge, the drainage area, and the depth of runoff. The graphical method

is applicable for watersheds where: 1) valley routing is not required, and 2)

land use, soil group, and cover are reasonably uniformly distributed

throughout th2 watershed (4).

-E4-

Chart Method

The input requirements for the SCS Chart method are the land use, the SCS

hydrologic soil groups, the hydraulic length (feet), the drainage area (acres),

the percentage and general watershed location of pond and swampy areas (%), the

24-hour precipitation for the given recurrence interval (inches), the percent

of imperviousness M, the percent of the hydraulic length modified (%), and

the watershed slope M. Charts are provided in appendix D of TR-55 for

determining the unit peak discharge (cfs per inch of runoff) for nonurbanized

conditions. This discharge is based on the drainage area, the watershed slope,

and the curve number. Adjustments are provided in appendix E of TR-55 for

conditions where pond and swampy areas exist and for conditions where either

the average watershed slope or the watershed shape vary significantly from

those used in the initial calculations.

The depth of direct runoff is computed using the depth of the 24-hour

precipitation for the given recurrence interval and a runoff curve number.

With the adjusted unit peak discharge and runoff depth, a nonurbanized peak

flow can be computed. The peak flow is then adjusted for the impact of

urbanization on travel time using adjustment factors (Figure 2) given in

Chapter 4 of TR-55. The adjustment factors are based on the percent

imperviousness and the percent modification of the hydraulic length. This

method is applicable for watersheds where: 1) the drainage area is less than

2,000 acres; 2) valley routing is not required; 3) the land use, soil group,

and cover are reasonably uniformly distributed throughout the watershed; and

4) where accurate estimates of the time of concentration are not available.

-E5-

ACCURACY ASSESSMENT

in testing a method for predicting the peak discharge, it is necessary to

establish criteria for assessing the quality of prediction. The criteria

should reflect both systematic and random variation about the true, or

accepted, value of the peak discharge. In general, one wants a prediction

method that, on the average, closely approximates the true value; this is a

measure of the bias. As a general rule, the bias is the average difference

between the predicted and true values and, therefore, is a measure of the

systematic variation of the computed estimates from the true value. When

comparing peak discharge estimates, it is necessary to modify the general rule

for computing bias because the true value is different on each watershed. In

terms of a plot of the predicted and true peak discharges, the bias can be

measured by the difference between the slope of the line that provides the

"best" fit to the data points (i.e., in a least squares sense) and the slope

of the line equaling the predicted and true values. Therefore, the bias of

the model is:

n
Bias J1 Q - Q - 1.0
-P p
n Q 2
i p

in which Qp is the predicted peak discharge, Q p is the true peak discharge,
and n is the number of watersheds. A positive bias value indicates over-

prediction, while a negative bias value indicates underprediction.

In addition to the systematic variation, one also expects nonsystematic,

or random, variation of estimated peak discharges about the true value; this

random variation results from factors affecting runoff rates that are not

accounted for by the model. The accuracy is a measure of the closeness of a

-E6-

predicted peak flow to the true peak flow. Representing the accuracy in the

form of a standard error, it is computed by:

n 2 0.5
Accuracy E Q Q (2)
n-I i=l

The accuracy, as given by Eq. 2, measures the deviation of the predicted
discharge from the true discharge; it is standardized by dividing by Q P to
reduce the effect of variation in the size of the events. That is, if Eq. 2

were not standardized, then errors for a few of the larger events might

dominate the computed value of the accuracy.

DATA BASE

Data used in this study were obtained from the U.S. Geological Survey

National Urban Frequency Study, which was reported on by Sauer, et al. (3).

This data base includes a comprehensive list of topographic, climatic, land

use variables, including indices of urbanization, and urban and rural flood

frequency estimates for 269 watersheds located throughout the United States.

The watersheds range in size from 76 acres to 73,000 acres (30 to 29,0'00 ha)

and have exhibited a relatively constant degree of urbanization over the period

of record. Data from a total of 56 metropolitan areas located in 31 states are

included in this data base. For this study, 68 watersheds were used with

drainage areas less than 4,000 acres (1,600 ha) and no channel storage. The

location and distribution of the watersheds are shown in Figure 1. Watershed

characteristics are summarized in Table 1. The impervious area averaged

24 percent of the total watershed area, and ranged from 3 to 98 percent. The

percentage of the hydraulic length modified averaged 31 percent and ranged

from 0 to 100 percent. Of the 68 watersheds, 51 had measured times of

concentration. These time of concentration values were determined using the

-E7-

velocity method. Watersheds in California, and Oregon were excluded from this

testing because the SCS procedures in TR-55 are primarily for areas where a

SCS type II storm is applicable. Also, since it has been shown (2, 7) that

the procedures do not perform well on watersheds exhibiting low slope

(one-half percent and less), the Texas watersheds were excluded, reducing the

data base to 40 watersheds having detailed channel information and measured

time of concentration values.

Flood frequency estimates for peak flows obtained from the data base were

derived from analyses of annual peak flow series. Log Pearson Type III

procedures, as recommended by the Water Resources Council Bulletin 17A, (5),

were used to fit each frequency curve to the data. A peak discharge was

estimated for each watershed for recurrence intervals of 2-, 5-, 10-, 25-,

50-, and 100-years. While these are actually predicted values themselves, for

these analyses they were assumed to be the true values. The record length of

record for the 68 watersheds of concern ranged from 9 to 18 years, averaging

about 13 years.

RESULTS

Graphical Method

The Graphical method was applied to the 40 watersheds. The resulting

values of the bias (Eq. 1), accuracy (Eq. 2), and the mean and standard

deviation of the errors are shown in Table 2. Standardized bias values

indicate that the graphical method is essentially unbiased over all return

periods (bias values are not statistically different from zero at the I percent

level). The mean of the errors shows a positive bias ranging from 210 cfs
(5.9 m 3Is) for the 2-year to 440 cfs (12.5 m 3/s) for the 100-year return

periods. The residual errors were examined to determine if an adjustment for

-E8-

the time of concentration measurements could be developed, based on watershed

characteristics such as area, percent impervious area, etc., that would more

accurately reflect urban conditions and, in turn, improve the bias and accuracy

of the procedure. It was determined that the errors were not correlated with

watershed characteristics and adjustments could not be developed that would

significantly improve the accuracy of the model.

Chart Method

Because the TR-55 Chart method is based on the SCS lag formula, its use is

limited to 2,000 acres (800 ha). This limitation was believed to be arbitrary

and, therefore, was tested using a straight-line extrapolation of the TR-55

Appendix D charts. Because the data base used in this analysis contained a

s.Lgn.Lficant break in watershed size at about 4,000 acres (1,600 ha), it was

decided to use those watersheds for which the other limitations of the method

were satisfied, to test the applicability of the method for watersheds up to

4,000 acres (1,600 ha). Results of this analysis for the six return periods

are summarized in Table 3. Comparing the statistics resulting from the

analyses of the 45 watersheds less than 2,000 acres with those resulting from

the analyses of the 68 watersheds less than 4,000 acres (1,600 ha), suggests

that the drainage basin size limit can be extended to drainage areas up to at

least 4,000 acres (1,600 ha) without a change in the bias or accuracy. Based

on these results, all further analyses were performed using watersheds having

drainage areas up to 4,000 acres (1,600 ha).

The Chart method of TR-55 provides for a correction of the percentage of

ponds and swamps in the watershed; the adjustment depends on the location of

the ponded areas, with different correction factors provided for ponding in

the lower, middle, and upper portions of the watershed. Because the data base

-E9-

used for testing did not indicate the location of the ponds, the bias and

accuracy statistics were evaluated for each of the three options to determine

the effect of the location of the ponding. Statistical testing of the

resulting values showed that there was no significant differe nce; therefore,

for our analysis we assumed aLl ponding to be located in the middle of the

watersheds. Lack of a significant difference resulted because there was very

little ponding (i.e., an average of less than I percent) in the watersheds

included in the test.

The Chart method was applied to the 40 urban watersheds. Resulting values

of the bias (Eq. 1), accuracy (Eq. 2) and the mean and standard deviation of

the errors are shown in Table 4. The statistics for the Chart method (Table 4)

are similar to those for the Graphical method (Table 2). The Chart method

incorporates urban factors based on the percentage of imperviousness and the

percentage of the hydraulic length modified to adjust the rural peak discharge

for urban conditions. Because these urban factors are essentially an addition

to the rural SCS procedures to account for urbanization and are apparently

based on limited data, our efforts to improve the Chart method were directed

to these urban correction factors. The data base was subdivided according to

whether or not channel modifications were made. Analysis of the residual

errors for the 9 watersheds with no channel modifications indicated that the

adjustment factor for the percentage of imperviousness (Figure 2) could not be

improved significantly. Most of the watersheds in the data base had impervious

areas from 0 to 30 percent; thus, the results were considered to be a verifica-

tion of the impervious area adjustment factor for values less than 30 percent.

In testing the adjustment factor for hydraulic length modified relationship

(Figure 2) we thought it would be more reproducible and hydrologically rational

if this factor reflected only channel modifications rather than modifications

-El 0-

to the total time of concentration hydraulic length path, which is the present

SCS interpretation. Statistics reflecting the change in the hydraulic length

definition are summarized in Table 4. Comparison of these statistics for the

two definitions of hydraulic length indicates that there is no significant

difference. Analysis of the residual errors using the new definition of

hydraulic length indicated that the hydraulic length modification factor need

not be a function of curve number and that adjustment factor could be best

represented by the curve for a curve number of 90 (Figure 2). A comparison of

the statistics in Table 4 indicates that the one curve is slightly better

primarily because of the significant decrease in the mean error.

CONCLUSIONS

Evaluation of the SCS urban peak flow methods on 40 urban watersheds

located around the United States indicated:

(1) A time of concentration adjustment to reflect urban conditions for use

in the SCS Graphical method is not necessary when time of concentration is

determined by the velocity method.

(2) The SCS Chart method can be used for areas up to 4,000 acres (1,600

ha) without a 1,oss of accuracy.

(3) The adjustment factor for the percentage of imperviousness (Figure 2)

that is used with the Chart method was found to be adequate.

(4) The adjustment factor for the hydraulic length modified that is used

with the Chart method should be limited to modification of the channel only,

and the hydraulic length modification adjustment relationship is not a function

of curve number. The curve for a curve number of 90 (Figure 2) provides the

most accurate prediction.

-E11-

REFERENCE

1. Rawls, W. J.,. Stricker, V. A., and Wilson, K. V., "Review and Evaluation

of Urban Flood Frequency Procedures," USDA Bibliographies and Licerature

of Agriculture, No. 9, 1980, 63 pp.

2. Rawls, W. J., Wong, S. L. , and McCuen, R. H., "Comparison of Urban Peak

Flow Methods," Journal of Water Resources Planning and Management, ASCE,

(in press).

3. Sauer, V. B., Thomas, W. 0., Stricker, V. A., and Wilson, K. V., "A

Nationwide Study of Urban Flood Frequency-A Preliminary Study," National

Symposium on Urban Hydrology, Hydraulics and Sediment Control, University

of Kentucky, 1980.

4. U.S. Soil Conservation Service, "Urban Hydrology for Small Watersheds,"

Technical Release No. 55, 1975.

5. U.S. Water Resources Council, "Cuidelines for Determining Flood Flow

Frequency," Water Resources Bulletin 17A, 1977.

6. U.S. Water Resources Council, "Estimating Peak Flow Frequencies for

Natural Ungaged Watersheds-A Proposed Nationwide Test," 1981.

7. Woodward, D. E., Welle, P. I., and Moody, H. F., "Coastal Plains Unit

Hydrograph Studies," Proceedings of National Symposium on Urban Stormwater

Management in Coastal Areas, ASCE, Blacksburg, Virginia, 1980, pp. 91-108.

-E12-

Table I.--Watershed Characteristics

..........
Standard Range
-Mean deviation Minimum --Maximum

Drainage area (mi2) 2.6 1.5 0.12 6.17

Watershed length (Mi2) 2.8 1.1 0.5 5.3

Watershed slope M 4.2 3.5 0.5 20.0

Pond and swampy area M 0.8 1.6 0.0 8.0

SCS runoff curve nUMber 80. 5.7 65. 91.

impervious area W 24. 16.4 3. 98.

SCS Hydrologic soil type C A D

Hydraulic length modified 31. 19. 0. 100.

-E13-

Table 2.--Results of Testing the SCS Graphical Method (n 40)

Statistic Return p-eriod years)
2 5 10 25 50 100

Bias 0.09 0.06 0.5 -0.02 -0.4 -0.07

Accuracy 0.96 0.81 0.97 1.02 1.11 1.17
Mean Error (cfs)l/ 210 312 383 380 438 440

Standard deviation 640 992 1,259 1,616 1,903 2,267
-of errors (cfs)

-1/ To convert cfs to m3/s multiply by 0.0283

-E14-

Table 3.--Evaluation of the Effects of Watershed Size on the Bias and Accuracy of the SCS Chart Method

Limit Return period-
4etho on area 2 years - -5 years 10 years 25 years __50 years 100 years
(acres) /
Bias Accuracy Bias Accuracy Bias Accuracy Bias Accuracy Bias Accuracy -Bias Accurau
,'hart 2,000 @-O. 46 1.11 -0.42 1.23 -0.38 1.26 -0.37 J. 59 -0.36 1.55 -0.33 2.48
(800 ha)

0.28 1.08 -0.29 1.16 -0.28 1.2 1. -0.32 1.46 -0.34
',hart 4,000 1.47 -0.34 2.17
i(1,600 ha) ip

'-/There were 45 watersheds less than 2,000 acres (800 ha) and 68 less than 4,000 acres (1,600 ha).

U-1
I

Table 4.--Results of Testing the SCS Chart MethodY

Modifications
to Chart Method Statistics --Return period
2 years 5 years 10 years 25 years 50 years_ 100 year@_

None Bias 0.03 0.02 0.01 --@O. 04 -0.05 -0.08
Accuracy 1.09 0.92 0.96 0.93 1.01 1.05
2/3/ Mean error (cfs)4/ 165 220 267 227 261 240
Standard deviation 551 769 945 1,155 1,351 1 1,587
of errors (cfs)

Recalibrated Bias -0.02 -0.03 -0.04 -0.09 -0.10 -0.13
impervious area and' Accuracy 1.13 1.02 0.99 0.95 1.02 1.06
143
hydraulic length Mean error (cfs) 133 174 208 157 174
adjustment factors Standard deviation 554 790 974 1,212 1,418 1,679
of errors (cfs)

Hydraulic length Bias 0.05 0.04 0.04 --.02 -0.03 -0.06
defined as Accuracy 1.09 0.91 0.93 0.89 0.95 0.99
channel2/3/ mean error (cfs) 166 219 264 222 254 232
Standard deviation 576 799 975 1,181 1,376 1,606
of errors (cfs)

Hydraulic length Bias -0.11 -0.12 -0.12 -0.17 -0.17 -0.20
defined as Accuracy 1.02 0.81 0.82 0.79 0.85 0.89
channel (use CN 90 mean error (cfs) 57 60 63 -21 -34 -96
curveT Standard deviation 537 759 932 1,159 1,355 1,601
of errors (cfs)

1/ 40 watersheds
Use Figure 2 hydraulic length adjustments
Use Figure 2 impervious area adjustments
To convert cfs to u@/s multiply by 0.0283

LOCATION OF WATERSHEDS

-*SIT: AT TLE

PAT
YOWA CITY IA@
0
C HIC A O,IL
CO'LUIABUS CAi
SAN@PRANCISa@' CA dOULD:7-:1. CO (2
-----------

DENVER, CO MDIANAPOL@' IN @f AS
% ST Louis. M 0. it*

%J
OKLAHOMA C14, Y. OK NASHVILLE. TN

A-T L A
NTA,
TUCSON'AZ. (1)
@CANTON, MS
DALLAS, TX IFT. WO TH. 40)
(2)
ATCHE@, MS

AUSTIN, TX
(2) 13ATON RQVGE LA
AkASKA SAN 44TON 0', TX Hou'sl`O-N, T K
\\ (21)
ONOLULU, Hi
c:>

LEGEND
lp I;> Number of watersheds

rigure

1.8

k ct)
z z
0 0

1.6

a:
0

U
<
LL 1.4

LU
Q-

1.2
C)0

1.0
0 25 50 75- 100
PERCENT IMPERVIOUS-
AREA OR HYDRAULIC
LENGTH MODIFIED

jure

-E18-

-------

APPENDIX F

ESTIMATING THE SCS PEAK RATE FACTOR

Richard H. McCuen

Timothy R. Bondeled

ESTIMATING THE SCS PEAK RATE FACTOR

ABSTRACT

The Soil Conservation Service (SCS) dimensionless unit hydrograph model

TR-20 is widely used for hydrologic design. The model uses a dimensionless unit

hydrograph based on a peak rate parameter of 484. While vague guidelines have

been provided for modifying the value of the parameter, almost all designs are

made using a value of 484. A method is presented that provides a systematic means

of obtaining a value of the peak ra te factor that is unique to a watershed. The

method only requires a runoff curve number map, which should be developed to use

the TR-20 model anyway. Application of the method to six watersheds, including

three coastal basins, indicates that the method increases the accuracy of pre-

diction; the peak rate factors computed using the procedure outlined compared

very favorably with the values obtained by optimizing on measured rainfall hyeto-

graphs and runoff hydrographs.

ESTIMATING Till'. SCS PEAK RATE FACTOR

Richard 1-1. McCuen and Timothy R. Bondelid

INTRODUCTION

A recent nationwide study (WRC, 1981) indicated that the Soil Conservation

Service (SCS) hydrologic methods were some of the most widely used methods in

hydrologic analyses. The SCS TR-20 program (SCS, 1969) provides the means for

developing runoff hydrographs, routing hydrographs through both channel reaches

and structures, and combining hydrographs. Manual hydrologic computations can

be performed using the methods detailed in TR-55 (SCS, 197S); these methods can

be used to obtin cither a peak discharge or a runoff hydrograph. The methods

that are presented in TR-55 are based on the concepts behind the development of

the TR-20 program.

A unit hydrograph is used in the TR-20 program for developing runoff hydro-
graphs (SCS, 1972). The unit hydrograph is dimensionless with axes of q/q p and
t/tp. in which q is the discharge at any time t and q P is the peak discharge at
time tP; thus, the peak of the dimensionless unit hydrograph equals 1.0 and t/t P
equals 1.0 at time t P The dimensionless unit hydrograph is dimensionalized by
multiplying the time values by the estimated time-to-peak and the discharge

ordinates by the peak discharge, which is given by:

DfAQ
P tP

in which qp is the peak discharge in cubic feet per second, A is the drainage
area in square miles, Q is the runoff volume in area-inches, and D f is called

IProfessor and Graduate Research Assistant, Department of Civil Engineering,
University of Maryland, College Park, MD 20742.

-Fl-

the peak rate factor. For the dimensions specified for A, Q, t p and q p and
the assumption that 37.5 pe reent of the total volume of runoff occurs prior to
the tp, Df will equal 484; most application of TR-20 assume this value of D f'
SCS (1972) indicates that D f can vary from 300 to 600, with a value of 300

in very flat swampy country and a value of 600 in steep terrain. However,
an accurate, systematic method for selecting the appropriate value of D f for

a watershed has not been proposed.

Hydrologists who must use either TR-20 or TR-55 in coastal areas recognize

that the standard value of 484 for D f does not reflect the flood runoff charac-

teristics of coastal watersheds. Since policies in Maryland require the use

of SCS techniques for various hydrologic computations there was considerable

interest in an examination of the accuracy of the standard peak rate factor in

coastal areas. A study by Woodward, et al., (1980) suggested a value of 284 for

the Delmarva peninsula. This created some question on the applicability of the

284 value for low sloped areas on the western shores of Maryland. This concern

emphasizes the need for an accurate, systematic method for determining the peak

rate factor and dimensionless unit hydrograph for ungaged watersheds where the

design engineer believes the standard SCS dimensionless unit hydrograph to be

inappropriate.

Because runoff-hydrographs computed using TR-20 and TR-55 are very sensitive
to the value selected for DfV it is important to have an accurate estimate of
its value. While regional estimates could be derived from an analysis of measured

storm hydrographs, this may not be either practical or possible for most hydro-

logic designs. Furthermore, if the watershed under investigation is not similar
to those used to obtain the regional estimate, then the estimate of D f may be

inaccurate.

Thle objective of this study is to present a method of estimating the dimension-

less unit hydrograph and the peak rate factor for ungaged watersheds.

-F2-

THE PEAK RATE FACTOR AND HYDROGRAPH CHARACTERISTICS

The value of 484 for the peak rate factor reflects both the units in which
qp A, Q, and tq) are given and the area under the rising limb of the hydrograph.
The shape and magnitude of the rising limb of a hydrograph is a function of the

storage characteristics of a watershed and tile precipitation characteristics

(Linsley, et al., 1958). If the distribution of precipitation is assumed con-

stant for the incremental duration of the unit hydrograph, then one can assume

that the rising limb of the hydrograph depends primarily on the watershed storage

characteristics. For a watershed in which storage characteristics are relatively

homogeneous, the time-are a curve can be used to represent the volume of detention

storage within the watershed. Tile time-area curve shows the distribution of the

drainage area that contributes to runoff during various time increments. Using

the hydraulic features of the "bank-full" channel system,the travel times to

the watershed outlet are estimated for a number of points in the drainage basin

and time contour lines with equal time intervals are drawn (Kraijenhoff van de

Leur, 1966). If the storage characteristics are not relatively homogeneous through-

out the watershed, then a time-storage curve could be developed instead of the

time-area curve. If a time-storage curve defines the volume under the rising

limb of a hydrograph, then the SCS peak rate factor (i.e., Df of Eq. 1) can be

estimated from the time-storage curve analysis.

PROCEDURE FOR ESTIMATING THE PEAK RATE FACTOR

The following procedure is recommended for obtaining the SCS peak rate

factor:

1. Develop a runoff curve number (CN) map by integrating the soils data

and the land use data;

2. develop a slope map from a topographic map;

-F3-

3. computer runoff velocities (using the CN and slope maps) for a large

number of points throughout the watershed and form a velocity map;

4. use the runoff velocity map to get iso-travel time lines, with approxi-

mately 15 to 20 intervals recommended-

5. find the fraction of the, total area of the watershed contributing during

each time interval, and form a plot of the fraction of the watershed

within each iso-travel time lines versus time;

6. form a time-storage plot by dimensionalizing the plot by dividing the

ordinates by the largest fraction, and find the proportion of the area

under the rising limb of the time-storage plot; and

compute, the peak rate factor by:

Df = 1290.67 p (2)

in which p is the proportion of the area under the rising limb of the time-

storage curve.

It is important to note that the development of the time-storage curve,and

computation of the peak rate factor,does not require any information beyond that

required to develop the input required for a typical TR-20 evaluation. Obtaining

the soils, land use , and topographic information is usually a first step in the

watershed analysis process. Computing the velocity map and time storage curve

is a simple process and uses only the curve number and topographic maps, in

addition to a formula for estimating the velocities as a function of CN and

slope.

APPLICATION OF THE TIME-STORAGE PROCEDURE

The procedure outlined previously was applied to six watersheds, including

two of the coastal watersheds used by Woodward, et al. (1980). The Powells Creel,.

watershed W-1 is a small upland watershed near Blacksburg, VA; it is an undeveloped

-F4-

woodcd and agricultural area of moderate sloj)cs Obout 2 percent). 'I'lic ra i n t'o i I

runoff. and basin characterist ics were obtained fi-om SEA-AR experimental witcr-

sheds as reported in ARS publications (e.g., ARS, 1972). The Powells Creek

watershed has a drainage area of 182 acres and a length of approximately 4,300

feet. Soils include Enon, Wi Ikes, Bremo, and LloYd lo.,jm. The watershed is

comprised of 16 percent hardwood forest, 60 percent native grass pasture, 6 per-

cent row crops (mostly,corn and tobacco), 7 percent small grain crops, 5 percent

alfalfa and other hay crops, 4 percent idle land, and 2 percent roads. These

data were used to develop a runoff curve number map, which was necessary to

develop the time-storage curve. The dimensionless unit hydrograph (DUYI) re-

sulting from the time-storage curve is shown in Fig. 1. Since approximately

38 percent of the Volume of tile DUII lies under tile rising limb, Eq. 2 provides

a value of 490 for the peak rate f@ictor; this is in very close agreement with

the standard value of 484 suggested by the SCS.

The three coastal watersheds lie on the Delmarva peninsula. The Nturderkill

River watershed near Felton, MD, has an area of 8,704 acres and is predominately

cultivated and forested, with over 60 percent of the watershed in crops and

approximately one-fourth wooded; the remaining land includes marshland, farm-

steads, and roadways. The main drainage channels lie within narrow wooded flood-

plains, which provide for increased watershed storage. Fig. I shows the time-

storage curve for the watershed. A numerical integration of the volume under

the rising limb indicates that p equals 0.307, which yields a va lue for D f of

396 (Eq. 2).

The Faulkner Branch watershed near Federalsburg, MD, drains approximately

4,544 acres. Except for farmsteads and roads, the drainage basin is predominately

cropland and woodland. The main drainage channels lie within narrow wooded flood-

plains. The soils consist of sassafras (SCS group B) and Woodstown (SCS group Q

sandy loams. The time-storage curve is shown in Fig. 1. Approximately 27.3 per-

-F5-

cent of the volume lies under the rising limb, which yields a value of 354 for

the peak rate factor.

The Morgan Creek watershed is located near Kennedyville, MD, and has a

drainage area of 12.7 square miles. The watershed is primarily cropland, although

there are woodlands, marshy areas, and farmsteads throughout the watershed. The

main channels lie within narrow wooded floodplains or open marshy areas. Channel

slopes are less than one percent, while land slopes are mostly less than two per-

cent. The dimensionless unit hydrograph was obtained from the time-storage curve;

a peak rate factor of 435 was computed.

Pony Mountain Branch Watershed W-1 is located in Culpeper County, VA. The

192 acre watershed is roughly trapezoidal with an- average elevation of 2,400 feet.

Approximately 66 percent of the basin has a slope range less than 4 percent, but

the remaining portion of the watershed has slopes from 12 to 25 percent. The

soils are of the Penn (SCS Soil Group C) and Bucks (Soil Group C) series. The

watershed has a mixed cover, with about 50 percent in farm woods, predominantly

hardwood,and 30 percent native grass;'the remaining portion of the watershed con-

sists of a mixture of orchard grass, clover, alfalfa, small grain, and road sur-

faces. The dimensionless unit hydrograph, which was obtained from a time-area

curve, yielded a peak rate factor of 360. ,

The Coshoction W-177 Watershed is an ARS experimental watershed in Coshocton,

Ohio. The 75.6 acre watershed is predominantly cropland, with small portions in

woodland, farmsteads, and roads. While the average slope is about 9 percent, the

slopes range from 5 to 20 percent. The dimensionless unit hydrograph was com-
puted from a time-area curve, and the peak rate factor of 495 was obtained from

Eq. 2.

Data on the Miaryland watersheds were provided by Dr. Helen F. Moody, College Park,
MD, and f1r. Donald Woodward, USDA-SCs, Broomall, PA; the authors are especially
grateful for their effort F6-

TEST OF THE TIME-STORAGE METHOD

The proposed procedure for estimating a value of the peak rate factor for

an ungaged watershed was tested using measured rainfall and runoff data. Data

for twenty-two storm events were available for the six watersheds. The rainfall

and runoff data were measured at two minute intervals for the Powells Creek,

Pony Mountain, and Coshocton watersheds;a one-hour interval was used for the

data from the three coastal watersheds, Morgan Creek, Faulkner Branch, and the

Murderkill River.

A version of the SCS TR-20 program that provides for automatic fitting

to observed hydrographs (Bondelid and McCuen, 1981) was used to identify the
best-fit relationship between Df and tc for each of the 22 storm events. The
program was used to develop a response surface that shows the value of the

objective function (i.e., a weighted least squares fit-of the hydrograph) as

a function of tc and Df* The response surface for the May 31, 1962, storm event

on the Powells Creek watershed is shown in Fig. 2. Each response surface was

characterized by a linear valley having a positive slope. While the fitting
precision changes rapidly as the combination of t c and Df moves perpendicular
to the valleysthe response surface indicates that there are an array of combina-
tions of values of tC and Df that provide almost an identical fit of the computed
runoff hydrograph to the measured hydrograph.

A straight line relationship was obtained for each storm event. A linear

relationship would be expected, as evident from a transformation of Eq. 1:

t AQ )D (3)
p q p f

Using the SCS relationship between tP and tc yields:

-F7-

t = "'SA%D (4 a)
C p f

= KDf (4b)

in which K is the peak rate slope coefficient. Table I gives the values of K
derived from the measured values of A, Q, and qp and the corresponding values
obtained from fitting the h),etographs and hydrographs. For storm events on

which the volume of runoff exceeded one inch, the two estimates of K showed

good agreement; for the remaining events the degree of agreement between the

two estimates varied. For storm events based on small volumes of runoff, the

value of K from Eq. 4 would depend on the representativeness of the peak dis-

charge.

The accuracy of the procedure described herein for estimating the peak rate

factor was tested using the twenty-two storm events. The peak rate factor
(D f1min ) and time of concentration (t c1min ) that provided the best fit between
the measured and predicted hydrographs are shown in Table 1 for each storm event.

The mean value of Df for the storm events were computed for each watershed and
are given in Table 2; the average error for the six watersheds was 4.8 percent.

The mean peak rate factor computed from the storm event data showed good agree-
ment with the values of Df obtained from time-storage analyses for four of the
six watersheds. The two values of D f differed by 39 and 50 on the Murderkill
River and Powells Creek watersheds, respectively. Five of the seven storm events

from Powells Creek had runoff volumes less than 0.3 inches and are less than the

I-year event. For this reason, the storm response may not be indicative of

the runoff potential of the watershed as indicated by both the peak rate factor

and the dimensionless unit hydrograph. For the two larger storm-events, with

runoff volumes of 0.9 and 2.0 inches, the computed peak rate factor showed per-

fect agreement with the value obtained from the time-storage curve. The two

-F8-

storm events for the Murderkill River watershed were the two largest runoff

volume producing events included in the study. The difference of 39 in the

hydrograph computed and watershed derived peak rate factors.is not considered

significant because there was little difference in the value of the objective

function between Dfvalues of 3SO and 400; that is, the valleys on the response

surfaces were very long and not steeply sloped along the valley. The fit of

the observed hydrograph for a Df of 396 was almost identical to that obtained

at the optimun values of 355 and 360-for the two storm events.

The mean values of the best-fit and watershed times of concentration are

also given in Table 2. The percentage error ranged from 5 percent for the

Morgan Creek and Murderkill River watersheds to 29 percent for the Pony Moun-

tain Branch watershed. The mean value for Pony Mountain Branch watershed was

based on three storm events all having runoff volumes less than O.OS inches.

Thus, the fitted times of concentration for the six watersheds are in good

agreement with the times of concentration obtained by conventional means. The

weighted average error was 10.7 percent.
CONCLUSIONS

The SCS methods (i.e., TR-20 and TR-SS) are some of the most widely used

hydrologic design methods. They are generally applied using the standard SCS

dimensionless unit hydrograph with a peak rate factor of 484. While these are

probably valid for many applications, it is recognized that the design accuracy

can be improved if the dimensionless unit hydrograph is calibrated. The regional-

ization provided by Woodward, et al., for the Delmarva peninsula resulted in a

more rational dimensionless unit hydrograph for those coastal areas. However,

past experience has suggested that unit hydrographs are not constant for all

watersheds in a region.

The hydrologic effects of land cover conversion from forested to agriculture

land use is of special concern to those involved in hydrologic analysis on the
Delmarva peninsula. If channel modifications are to be made for the purpose of

improving the drainage efficiency within the watershed, it is important to ac-
-F9-

curately evaluate the hydrologic response of the direct runoff. Furthermore,

in coastal areas where stormwater detention must be provided to control the ef-

fects of land use conversion, the accuracy of the shape of the hydrograph is an

important determinant of the volume of required storage, Given the economic

value of land, farmers in coastal areas are very concerned about the surface

area requirements of detention facilities. Hydrologic computations of detention

storage requirements could be highly inflated if the unit hydrograph is more

peaked than the true hydrologic response of the watershed. Thus, agricultural

drainage in coastal areas represents a case where standard unit hydrograph shapes

are of special concern.

A procedure was outlined herein that can be used to derive a dimensionless

unit hydrograph for use with the SCS hydrologic design methods. The method does

not require any data that is not normally used in the application of the SCS

design methods. The method was tested on six watersheds, including three coastal

watersheds; the method provided good agreement with the analysis of the measured

data. Use of the procedure should increase the accuracy of designs.

-FIO-

1. Agricultural Research Service, Hydrologic Data for Experimental Agricultural
Watersheds in the United States, 1966, U.S. Dept. of Agriculture, Miscel-
laneous Publication No. 1226, Washington, D.C., 1972.

2. Bondelid, T.R., and McCuen, R.H., "An Automatic Optimization Version of
the SCS TR-20 Computer Program," Technical Report, Dept. of Civil Engineering,
University of Maryland, 1981.

3. Kraijenhoff van de Leur, D.A., "Runoff Models with Linear Elements,"
Chapter 2 of Recent Trends in Hydrograph Synthesis, Proceedings of
Technical Meeting 21, Versl. Meded. Comm. Hydrol. Onderz. T.N.O. 13,
The Hague, 1966.

4. Linsley, R.K., Jr., Kohler, M.A., and Paulhans, J.L.H., Hydrology for Engineers,
McGraw- I I i I I Book Co., I iic. , New York, 19S8.

S . Soil Coliscl,vzltioll sci-vicc' "Coiiijinter Program foi, Pi,o'iecl Formulatioj)-
Hydrology," Washington, D.C., Tcclinical Release No. 20, Supplement No. 1,
Central Technical Unit, 1969.

6. Soil Conservation Service, "National Engineering Handbook, Section 4,
Hydrology," Washington, D.C., U.S. Department of Agriculture, 1972.

7. Soil Conservation Service, "Urban Hydrology for Small Watersheds,"
Washington, D.C., Technical Release No. S5, 197S.

8. Water Resources Council, Estimating Peak Flow Frequencies for Natural
Ungaged Watersheds, Hydrology Committee, Washington, D.C., 1981.

9. Woodward, D.E., Welle, P.I., and Moody, H.F., "Coastal Plains Unit Hydro-
graph Studies," Urban Stormwater Management in Coastal Areas, American
Society of Civil Engineers, New York, pp. 99-107, 1980.

-F11-

TABLL I . Compai,i.son ol' Best Fit and Stoilii Lvent VAJAICS

Date of l.SAQ atc Storm Storm
Watershed Storm- q1) 3D f Imin Event CN Event Q Df Lin tC@in tcIDf

Powells 7/10/59 0.00132 0.01440 68.1 0.047 350 0.305 0.496
Creek 10/ 8/S9 0.00132 0.00114 71.3 0.223 550 0.802 0.601
5/31/62 0.00155 0.01728 84.2 0.892 490 0.605 0.605
7/29/63 0.01324 0,00140 83.5 0.290 350 0.360 0.571
7/11/65 0.00198 0.00221 83.4 1.980 485 0.729 0.725
9/20/66 0.00215 0.00093 79.6 0.106 425 0.488 0.554
8/23/67 0.00264 0.005SS 83.5 0.102 430 1.070 1.390

Coshocton 6/18/40 0.00104 0.00147 90.6 0.342 49S 0.482 0.490
W-177 6/12/57 0.00103 0.00133 79.1 1.374 490 0.318 0.335
4/25/61 0.00238 0.00123 93.6 1.03S 42S 0.391 0.484
5/13/64 0.00099 0.01620 78.2 0.127 495 0.5S1 O.SS4

Pony Mtn. 5/26/62 0.00263 0.00667 64.0 0.028 3SO 0.180 0.184
Branch 6/14/67 0.00276 0.00174 72.3 O.OS4 300 0.239 0.370
6/17/68 0.00368 0.00455 81.8 0.049 425 1.384 1.085

Morgan 1955 0.0371 0.0360 52.1 1.211 350 10.40 12.61
Creek 1958 0.0311 0.0260 75.0 1.341 490 6.35 5.02
1960 0.0287 0.0342 79.5 2.317 Soo 11.89 11.05

Murderkill 1960 0.0670 0.0835 59.0 2.642 360 17.7 20.5
River 1967 0.0660 0.0551 85.7 6.726 355 10.1 12.7

Faulkner 1955 0.0470 0.0280 51.1 1.901 320 7.20 8.20
Branch 1958 0.0580 0.0128 66.5 2.291 355 4.86 4.85
1960 0.0416 0.0400 61.7 2.538 370 11.75 11.05

-F12-

TABLE 2. Summary of Means for Best-Fit and Storm Event Values

Mean Peak
Rate Slope Peak Rate Factor Time of Concentration
Watershed n Storm Event Fitted Watershed Fitted Watershed Fitted

Powells 7 0.00346 0.00613 490 440 0.670 0.623
Creek

Coshocton 4 0.00136 0.00506 495 476 0.500 0.436

Pony Mtn. 3 0.00302 0.00432 360 3S8 0.467 0.601
Branch

Morgan 3 0.0323 0.0321 435 449 9.12 9.55
Creek

Murderkill 2 0.0665 0.0693 396 357 13.3 13.9
River

Faulkner 3 0.0489 0.0269 352 348 8.70 7.94
Branch

-FI 3-

Powells Creek W-1 near Blacksburg, VA
Murderkill River near Felton, MD
Faulkner Branch near Federalsburg, MD

0.8

Cd
P4 0.6

0.41

t4-4 o.2..
0
.0
4j '.

0
0 1 2 3 4

Ratio of time (t) to time-to-peak (t p

FIGURE 1. Dimensionless Unit Hydrographs Derived from Time-Storage Curves for Three Watersheds

0.9

0.8 Q5

X:
0.7
0

4J
Cd
P
0 Q)
r-
Q) V Q@ *
U
9:
0 0.6
U

44
0

0.5

0.41

400 Soo 600

Peak Rate Factor

,FIGURE 2. Peak Rate Factor Response Surface for 5/31/62 Storm Event on

Powells Creek Watershed

-Fl 5-

APPENDIX G

AN AUTOMATIC FITTING VERSION
OF THE SCS TR-20 HYDROLOGIC MODEL

Timothy R. Bondeled

Richard H. McCuen

An Automatic Fitting Version of the SCS TR-20 Hydrologic model

Timothy R. Bondelid and Richard H. McCuen

INTRODUCTION

While peak discharge formulas, such as the rational method, are still widely

used, hydrologic methods that produce an entire runoff hydrograph are being used

more frequently in hydrologic design. This has resulted from the need to con-

sider storage in design, such as detention storage in urban design problems.

The hydrologic methods developed by the Soil Conservation Service (SCS) are

some of the most widely used methods for hydrologic analyses (WRC, 1981). The

SCS TR-20 computer program (SCS, 1969) can be used for hydrologic analyses where

it is necessary to generate flood hydrographs, route them through either channels

or structures, or both, and combine hydrographs. The SCS TR-55 (1975) tabular

method, which is the result of numerous TR-20 runs, also can be used to develop

flood hydrographs.

As with most flood hydrograph generating methods, the flood hydrograph algo-

rithm of TR-20 is based on the concepts of a unit hydrograph and convolution. The

SCS dimensionless unit hydrograph (DUH) is synthetic. A unit hydrograph is

generated from the DUH using the depth of runoff, which is a function of the

24-hour precipitation depth and the runoff curve number (CN), the drainage area,

the time of concentration, and a peak rate factor, which is a parameter that

indicates the proportion of the volume that is under the rising limb of the

unit hydrograph. A value of 484 is most frequently used for the peak rate factor.

1Graduate Research Assistant and Professor, Department of Civil Engineering,
University of Maryland, College Park, MD 20742.

-Gl-

It is believed that the value of this parameter will change for different regimes,

with a value of 300 recommended for coastal areas and a value of 600 for moun-

tainous watersheds (SCS, 1972). Woodward, et al. (1980) used measured rain-

fall and runoff to develop unit hydrographs for watersheds on the Delmarva penin-

sula; from these unit hydrographs they estimated the peak rate factor to be 284.

Other studies have shown that urbanization also changes the shape and distribu-

tion of a unit hydrograph. McCuen and Bondelid (1981) provided a procedure for

obtaining the dimensionless unit hydrograph and peak rate factor for ungaged

watersheds. The value of the parameter is supposed to reflect the design ac-

curacy. Therefore, it is deemed important to be able to estimate the correct

value of the parameter for ungaged watersheds where the,value of 484 is not

considered to be the optimal value.

In most regions hydrologic data are available that could be used to obtain

a regional estimate of the unit hydrograph parameter. However, this rainfall/

runoff data cannot be used with the TR-20 program because the current program

form does not provide for data analysis. Thus, a version of TR-20 that could

calibrate the dimensionless unit hydrograph and the peak rate factor using gaged

data would be of value in obt aining a regional estimate ; this should increase

the accuracy of designs made with the SCS hydrologic methods.

The purpose of this study was to develop a computerized automatic cali-

bration version of TR-20 that could be used in analyzing storm event rainfall

and runoff data to estimate the best-fit dimensionless unit hydrograph. While

the method proposed by McCuen and Bondelid (1981) should increase the design

accuracy over that obtained by using a peak rate factor of 484, the accuracy

could be further improved by calibrating the parameters of the TR-20 model using

measured rainfall and runoff data.

-G2-

AN AUTOMATIC OPTIMIZATION STRATEGY

Optimization, or fitting, is a term used to indicate the process in which a

model is fitted to data and "best" estimates of model coefficients are obtained.

Optimization is most efficient when a systematic strategy is used. The actual

fitting requires four components: (1) a model, (2) an objective function,

which is the criteria used to define "best" fit; (3) a data set; and (4)

a set of constraints, if applicable. For the case of fitting the SCS hydrologic

method, the dimensionless unit hydrograph was assumed to be the model. The

strategy provides several options for obtaining a best fit, with the options

depending on the available input data. Two objective functions are available

for fitting, with one based on the accuracy of the peak discharge and the other

based on a weighted least squares fitting of the entire runoff hydrograph. The

optimization program recognizes that the available input data and problem objec-

tive will vary; thus, several options are available for specifying the input

data structure.

Objective Function

A common technique in curve fitting is to minimize the sum of the squares

of differences between the observed and computed data points. As usually applied,

the least squares principle assumes the observations to be independent and, there-

fore, gives each observation equal weight; this is not the case in hydrograph

analyses. In fitting flood hydrographs, there is, in general, more concern in

being closer on the larger discharges, especially the peak of the hydrograph;

that is, it is more important for flood hydrograph prediction to be accurate

in predicting the high flows than in predicting the lower flows, which occur

at the beginning and end of the runoff event.

An objective function that combines the least squares concept and weighting

of the higher discharges is given by:

_G3-

- Qi 2
min F (Qi 3 ) Qi]
lQi

in which the summations are taken over the observed discharge values, Q i is
A
the observed discharge, Qi is the computed discharge occurring at the same time
as Qi, and F is the value of the objective function. The numerator in Eq. I

represents a weighted least squares and the denominator makes the F value

dimensionless. The weighting by Q, ensures that the larger flows, especially

the peak discharge, will be given a greater influence on the values of the

.fitting parameters. 'rhe goai of the optimization procedures for this objec-

tive function will be to minimize the right hand side of Eq.-I.

.As an alternative to Eq. 1, the program includes a second objective func-

tion. Specifically, if the peak discharge is of principle interest and the timing

of minor importance, then values for the fitting parameters can be obtained such

that the absolute value of the difference between the computed and measured peak

discharges is a minimum:

min F Q'p - Qp (2)

in which lp and 1P are the computed and measured peak discharge, respectively.

Dimensionless Hydrograph Generation

The structure of the SCS dimensionless unit hydrograph is not specified

by a known probability density function. While it seemed desirable to have a.

unit hydrograph that would closely approximate the shape of the SCS unit hydro-

graph, it also seemed reasonable to select a known probability density function

so that its characteristics would be known. In addition to selecting a known

probability density function, there are several constraints on the.shape of

the dimensionless unit hydrograph. The dimensionless unit hydrograph must peak

at a time of 1.0 and have a peak discharge value of one; also, the DUH must be

-G4-

unimodal and have the general shape of a typical runoff hydrograph. Additionally,

the unit hydrograph must be able to be varied in a consistent and stable manner

by varying specific parameters.

The gamma density function satisfies these requirements. Because it has

a flexible shape, the gamma function can take the extreme forms of a decaying

exponential function or almost an equilateral triangle; in general, it takes

the form of a typical unit hydrograph. The gamma function has been widely

used in hydrologic analysis; Dooge (1973) described its use for predicting

agricultural runoff, and Sarma, et al., (1969) used it for evaluating the ef-

fect. of urbanization on small watersheds. The gamma density function is given

by:
f(X) (X) C-1 exp(-X/b) (3)
b bF(c)

in which X is the random variable defined on the abscissa, f(X) is the ordinate,

b is a scale parameter, c is a shape parameter, and r(c) is the gamma function

with argument c. For c>1.0, f(X) has a shape similar to typical runoff hydro-

graphs. The mode of f(X) is given by b(c-1). The gamma density function can,

thus, be easily scaled so that the peak occurs at X=l, with f(X)=l. Also, b

and c can be readily varied to produce a wide range of unit hydrograph shapes.

Sensitivities of the Gamma Parameters. The parameters of Eq. 3 are para-

meters that define the dimensionless unit hydrograph ordinates. However, the

parameter of real interest in TR-20 is the dimensionless hydrograph factor, or
peak rate factor, Df* Therefore, a logical strategy is to first evaluate the
parameters of the gamma function (c and b) in relation to D f and then evaluate
the sensitivity of TR-20 to Df*
Initial analyses indicated that Df is much less sensitive to ihe scale
parameter (b) than to the shape parameter (c).- that is, Df is almost entirely
determined by the shape parameter of the gamma function. The lack of s.ensi-

-G5-

tivity of the b parameter was verified on all of the test watershed events;

the gradient of the b parameter with respect to the objective function F of

Eq. I was always at least an order of magnitude less than the gradient of the

scale parameter with respect to F. In all subsequent analyses, the scale

parameter was set so that the mode of the gamma distribution occurred at a

value of 1.0. The mode can be set at 1.0 by using the equation:

b = c11 (4)

The accuracy of fitting would not significantly improve if both parameters

were optimized simultaneously because there is high correlation between the

sensitivity functions of the two parameters (McCuen, 1973).

A Limitation of the Gamma Function. Use of the gamma distribution function

for generating dimensionless unit hydrographs has one slight*limitation. Specifi-

cally, the maximum value of Df that can be produced using the gamma distribution

is 645. The value of 645 corresponds to one-half of the area of the dimension-

less unit hydrograph being under the rising limb; the gamma distribution can

not have more than half of the area before the mode of the distribution function.

As Df approaches 645, the dimensionless hydrograph approaches a "spike', shape.

In spite of this constraint, the gamma function is still a very reasonable

dimensionless unit hydrograph. For most natural watersheds, one would not expect

more than 50% of the area under the unit hydrograph to occur before the peak.

Even on urban watersheds this might only occur if the drainage patterns were

altered significantly and the storage characteristics were drastically modified.

In fact, the gamma function is still more flexible than the standard SCS dimension-

less unit hydrograph; this is necessary if the peak rate factor is to be optimized.

-G6-

Opt imum CN

A desirable goal in hydrograph prediction is to have the simulated runoff

volume equal to the observed runoff volume. The observed runoff volume can be

readily computed by integration. Because the total rainfall depth P is also

known, the SCS rainfall-runoff formula can be used to estimate the retention:
(P-0.2S)2 (5)
(P+0.8S)

in which P is the rainfall depth (in.), S is the watershed retention (in.), and

Q is the runoff volume (in.). The retention S is related to the CN by:

S =(1000/CN) -10 (6)

By selecting the program option that uses the CN computed from Eq. 4, the

simulated and ovserved runoff volumes can be made equal. The program also

provides the option that the CN obtained by conventional means can be part of

the data base for input.

Estimation of the Time of Concentration

Most formulas for estimating the time of concentration are based on very

limited data bases; however, most hydrologic design methods, including the SCS

methods, are very sensitive to the time of concentration. Therefore, an option

was included in the automatic fitting program to use measured rainfall and run-

off data to find the optimum storm event time of concentration. The precision

in the estimated tC is controlled by an input parameter. The time of concen-,
tration can be fitted using a two-phased search procedure, which is part of

the optimization strategy.

-G7-

The Optimization Strategy
If the user elects to find the optimum value of either tC or Df@ a two-

phased strategy is used to ensure that the optimum is obtained. The first

phase of the search is a systematic search of the response surface (i.e., the

value of the objective function versus the unknown, Df or tc); this phase of

the search requires the extent of the search, (i.e., the upper and lower values

of tc or Df),and the number of intervals into which the search region is divided.

The second phase of the fitting process uses the point obtained in the first

phase that has the minimum value of the objective function as the initial point.

A pattern -search algorithm is used to find the minimum value of the objective

function. The degree of precision is controlled by the number of iterations,

which is specified by the user.

-G8-

PROGRMI DESCRIPTION

The objective of the automatic fitting version of the SCS TR-20 hydrologic

model is twofold. The first objective is to provide a straightforward proce-

dure for calibrating the TR-20 model parameters to observed rainfall and runoff

data. The second objective is to provide a flexible tool that can be used for

future research. Therefore, a wide variety of user-selectable options have been

incorporated into the program.

The computer program consists of a main routine and sixteen (16) subroutines.

The program structure is modular in that most of the primary operations are con-

tained in separate subroutines; this allows for flexibility in updating and re-

using the program. The program has been written in FORTRAN IV and is,therefore,

essentially machine independent.

The program analyses are rainfall/runoff event per execution. The TR-20

reservoir and channel routing routines have not been incorporated into the pro-

gram. Inclusion of the routing options would greatly increase.the complexity

of the calibration procedures. Therefore, the program utilizes only the TR-20

RUNOFF subroutine for hydrograph generation.

Numerous program options are available to the user. These options include:

(1) either specifying the CN or having the program compute the optimum CN value;

(2) optimize an either tc or Df; (3) inputting a SCS dimensionless hydrograph

or have the program generate dimensionless hydrographs using the gamma function;

(4) utilize either of the two objective functions specified by Eqs. I and 2;

(5) input an observed hydrograph or have the program generate hypothetical

observed hydrographs; (6) separate baseflow from the observed hydrograph; and

(7) produce graphical output in two plot sizes.

The program also performs a hydrograph analysis including computation of run-

off volume and centers of mass of runoff, rainfall, and rainfall excess. The

graphical output consists of three plots. The first plot shows the observed

-G9-

and optimum hydrographs. The second plot shows both the optimum dimensionless

hydrograph and the standard SCS dimensionless hydrograph with D 484. The

third plot shows the rainfall and runoff hyetographs. These plots are very use-

ful for judging the adequacy of the optimization procedure.

Input Description

The input data consists of essentially three (3) groups of data: (1) control

data for specifying the program options to be used; (2) the observed rainfall

in the form of a TR-20 format rainfall table; and, (3) the observed discharge

in cfs. Table I presents the formats, variable names, and variable descriptions

for the input data. A maximum of nine (9) card types are required. A descrip-

tion of each of the input data elements is described below.

Title. This is an 80-column title used for watershed and event description.

AREA. This is the measured watershed area in sq. mi.

CN. The user can specify the CN to be used in the optimization by spec-

ifying a non-zero CN. If the input CN is 0.0, then the CN computed from the

observed rainfall and runoff volumes is used.

TC. This is the time of concentration in hours to be used when the peak rate

factor., Dfy is the optimization parameter. If t C is the optimization parameter,

then a value of zero should be input.

DMHF AC. The peak rate factor, D., to be used when the optimization parameter
is t C An input value less than or equal to 0.0 specifies that a TR-20 format
dimensionless hydrograph table is to be read. If the optimization parameter is

D... then a value of zero is input.

VOLMIN. The gamma function is unbounded in the right tail of the distri-

bution; therefore, it is necessary to specify a bound in order for the gamma

function to be used as a unit hydrograph. The parameter VOLMIN specifies

_G10-

TABLE 1. Input Data for the Automatic Fitting Version of TR-20

Card Type Variable Name Columns Variable De'scription

Title card 1-80 Title card

Primary Control AREA 1-7 Area (sq. miles)
CN 8-14 Runoff curve number (if zero, then the CN is computed
from observed rainfall and runoff)
TC 15-21 Time of concentration, hours (if zero, then the value
is optimized)
DMHFAC 22-28 Peak rate factor (if zero, then the value is optimized)
VOLMIN 29-35 Minimum fraction of the gamma function unit hydrograph;
used to set limit on right tail
TZ 36-42 Time of start of rainfall distribution (hours)
RDEPTH 50-56 TR-20 Rainfall depth (set equal to 1.0 when an actual
event is used)
RAINDU 57-63 TR-20 Rainfall table duration (set equal to 1.0 when an
actual event is used)
DELTM 64-70 TIZ-20 Computation interval (hours)
NTER 71-74 Maximum number of iterations in a step for a phase Il
search of t or D f
NSTEP 75-78 Number of steps in a phase Il search
ITELL 79 Graphical output option (0, 1, or 2)
IOPT 80 Specifies objective function and type of hydrograph input
(1, 2, 3, or 4)

Card T)_Te Variable Name Columns Variable Description
Parameter NINCTC 1-7 Number of increments in the phase I search for t C
Optimization (NINCTC = 0; if optimizing DO
NINDMH 8-14 Number of increments in the phase I search for D f
(NINDMH = 0, if optimizing t C)
TCMIN 15-21 Lower value of t C for phase I search 0 for D optimization
TCMAX 22-28 Upper value of tC for phase I search I f
DMMIN 29-35 Lower value of Df for phase I search 0 for t optimization
DMMAX 36-45 Upper value of Df for phase I search) = c
Dimensionless - Some format as the SCS dimensionless hydrograph table
Hydrograph Table (DIMHYD and ENDTBL cards required);
input only if DMIFAC <, 0 on card 2

Rainfall Table - Same format as the SCS rainfall table (RAINFL and
ENDTBL cards required);
Baseflow TBI 1-7 Times (hrs) for first baseflow point (relative to input
Specification hydrograph)
QB1 8-14 Discharge (cfs) at TB1
TB2 15-21 Time (hrs) for end of baseflow point
QB2 22-28 Discharge (cfs) at TB2

Format FORM 1-80 Format for observed hydrograph (used if IOPT I or 2)
Hydrograph NPT 1-5 Number of points on the observed hydrograph
Length (used if IOPT = I or 2)
Hydrograph Data HYDR FORM Observations on observed hydrograph

the minimum volume allowed under the gamma unit hydrograph generating function.

VOLMIN must be less than 1.0 and is used to avoid excessive "chopping off" of

the tail of the unit hydrograph. A value of 0.99 is recommended.

TZ. Specifies the time in hours of the first entry in the TR-20 rainfall

table.

RDEPTH. This is the TR-20 rainfall depth in inches. If the rainfall table

contains actual depths,then RDEPTH should be set at 1.0. If the rainfall table

is scaled from 0.0 to 1.0, as in the 24-hour type II rainfall distribution, then

RDEPTH should be set equal to the actual rainfall depth.

RAINDN. This parameter specifies the TR-20 rainfall table duration and

is normally set at a value of 1.0 for the analysis of actual storm events.

DELTM. DELTM is the TR-20 computation interval, in hours. This value

should be selected on the basis of the general watershed runoff characteristics.

For instance, if the event being analyzed is on a small watershed and the event

lasts for only a few hours, then a DELTM of 0.1 hour should be used; if the

event is for a large watershed lasting on the order of 24 hours or more, then

a DELTM of 1.0 may be appropriate.

NTER. The maximum number of iterations per step in a phase II search.

This will depend on the uncertainty in the bounds of phase I search. A value

of 25 may be adequate for small watersheds; a value of 50 to 100 would be more

appropriate for a large watershed in which there is uncertainty in t C or D f,

NSTEP. NSTEP is the maximum number of steps in a phase II search. The

maximum possible number of iterations is NSTEP times NTER.

ITEL. The variable ITEL specifies the plot size for the graphical output.

If ITEL = 0, then no graphical output is produced. If ITEL = 1, then plots of

80 columns by 40 lines are produced; this size is suitable for 83-2 x 11 in. paper.

-G1 3-

If ITFL = 2. then plots of 100 columns by 50 lines are produced; this size pro-

vides the greatest detail and uses a full page of computer paper.

IOPT. The variable IOPT specifies both the type of objective function to

be used and the type of hydrograph to be used for calibration. Four values of

IOPT are possible:

IOPT Value Program Options

1 Use observed runoff hydrograph and the
weighted least squares objective function
(Eq. 1).

2 Use observed runoff hydrograph and the
difference in peak discharges as the
objective function (Eq. 2).

3 Use a TR-20 generated hydrograph as the input
TC and D 484 as the "observed" hydro-
graph ans the weighted least squares
objective function (Eq. 1).

4 Same as IOPT = 3, except use the difference
in peak discharges as the objective function (Eq. 2).

The 1OPT values of 3 and 4 will not normally be used in actual calibration; the

options are provided for possible research into TR-20 parameter interactions.

DIMYD Table. This is a complete TR-20 dimensionless hydrograph table

in TR-20 format and must include the DIMHYD and ENDTBL cards. This table is

read only if DMHFAC is less than or equal to 0.0. If DMHFAC is greater than

0.0, then do not include the DINIHYD table in the input dock. This option

provides an alternative to the gamma function; it could be used, for example,.

where the gamma function may not be appropriate, such as for watersheds with

a D. greater than 645.

RAINFL Table. This is a complete TR-20 rainfall table in TR-20 format and

must include the RAINFL and ENDTBL cards. This table specifies thb observed

rainfall distribution for calibration.

-G14-

NINCTC. The variable NINCTC is the number of increments in t c in a phase I
search when optimizing tC. NINCTC must be set at zero when the optimization
is on Df*

NINDMH. This is the number of increments in D f in a phase I search when
optimizing on Df* NINDMH must be set at zero when the optimization is on t C.

Also, NINCTC and NINDMH cannot both be set at zero in any one program execution.

TCMIN and TCMAX. These variables are the lower and upper bounds, respectively,
of tC in a phase I search. If NINCTC = 0, then these values are ignored.

DMMIN and DMMAX. Tnese variables are the lower and upper bounds, respectively,

of Df in a phase I search. If NINDMH = 0, then these values are ignored.

TBI and QB1. TB1 and QB1 are the time (hrs.) and discharge (cfs), respec-

tively, of the first baseflow separation point.

TB2 and QB2. These variables are the time (hrs.) and discharge (cfs),

respectively, of the last baseflow separation point. TB1, QB1, TB2, and QB2

specify two points on the runoff hydrograph; during program execution the runoff

is understood to be the hydrograph above a straight line connecting these points.

When baseflow is not present, these values should all be set to 0.0.

FORM. The FORTRAN format specifications, up to 80 columns long,of the

observed hydrograph cards. The format should include the left and right paren-

theses and specify reading of REAL variables(F or E format), e.g., (10F8.2).

The left parenthesis should be placed in column 1.

NPT. This variable specifies the number of points on the observed input

hydrograph.

-G15-

Input Hydrograph. Each point on the input hydrograph requires two (2)

values, the time (hrs.) and discharge (cfs). The values are read in pairs so

that the time and discharge for the first point is read, then the time and

discharge for the second point, etc. The format specified by FORM is

used so that, for instance, if the format is (10F8.0), then each card can con-

tain five (5) points on the hydrograph.

TESTING OF THE FITTING METHOD

Data consisting of observed rainfall/runoff data for six watersheds were

selected to test the automatic fitting procedure. The watersheds were selected

to provide a wide range of drainage areas and slopes, including coastal water-

sheds. Three coastal watersheds were included in the study: the Faulkner,

Morgan, and Murderkill basins; these watersheds are rural and are located on

the eastern shore of Maryland. The watersheds range in area from 3,072 to 8,704

acres and the slopes range form 0.5 to 5 percent. Two or three rainfall events

for each watershed were used. The approximate return periods of the rainfall

events ranged from less than 5 years to greater than 100 years.

The three remaining watersheds that were used for testing the program were

the Coshocton, Powells Creek, and Pony Mountain Branch experimental agricultural

watersheds. The Powells Creek and Pony Mountain Branch Watersheds are located

in southeastern Virginia near Blacksburg. The Coshocton watershed is located

in central Ohio. The watershed areas range from 75.6 to 192 acres, and the aver-

age slopes range from 1.9 to 10.0 percent. Four to seven rainfall events on

each watershed were selected. The approximate return periods of the rainfall

events ranged from less than 2 years to about 100 years.

The automatic fitting program was used to compute the optimum.time of con-

centration for each of the twenty-two storm events. The mean value of the

optimum times of concentration, which are shown in Table 2, were computed for

-G16-

TABLE 2. Watershed and Fitted Times of Concentration (hours)

Number of Mean Storm Percent
Watershed Storm Events Watershed t c Event tC Error

Morgan Creek 3 9.12 9.55 4.5
Murderkill River 2 13.3 13.9 4.3

Faulkner Branch 3 8.70 7.94 9.6

Powells Creek 7 0.670 0.623 7.5

Coshocton W-177 4 0.500 0.436 14.7
Pony Mountain 3 0.467 0.601 22.3

-G17-

each of the six watersheds. The time of concentration was also computed by

conventional means (i.e., the velocity method) using the land use, slope, and

hydraulic length as input; these watershed t Cvalues are shown in Table 2.

The percentage of error for four of the 6 watersheds was less than 10 percent,

with the largest error being 22.3 percent. The largest percentage error, which

was for the Pony Mountain Branch, resulted from three storm events, each have

less than 0.06 inches of runoff. Thus, there appears to be good agreement be-

tween the time of concentration optimized by the automatic fitting version of

TR-20 and times of concentration computed conventionally from the land use,

slope, and hydraulic lengths of the watershed.

CONCLUSIONS

There has been increased use of unit hydrograph methods in hydrologic de-

sign because of the need to incorporate storage computations in the design.

The SCS TR-20 program and the SCS TR-55 tabular method, which is based on the

TR-20 program, are unit hydrograph based methods that have seen increased usage.

It is recognized that the accuracy of a unit hydrograph method can be improved

significantly if it is calibrated to data measured either on the watershed where

the design point is located or within the region. McCuen and Bondelid (1981)

provided a method for obtaining the SCS peak rate factor and the dimensionless

hydrograph using the time-storage curve of the watershed. While this should

increase the accuracy of designs based on the TR-20 program, the accuracy could

be improved further if the paramaters of the TR-20 program were calibrated to

measured rainfall and runoff data.

An automatic fitting version of the TR-20 program was developed and described

herein. The program can use measured rainfall hyetographs and runoff hydro-

graphs to compute optimum values of various parameters. Depending on the fitting

options selected, the calibration paramaters include the runoff curve number (CN),

-G1 8-

the time of concentration (tc), and the dimensionless unit hydrograph ordinates.
The calibration procedure uses automatic optimization to obtain a predicted flood

hydrograph that most closely agrees with the observed flood hydrograph. The

method was tested on 22 storm events for six watersheds. The time of concen-

tration was used as the fitting parameter. The mean optimum time of concentration

showed reasonable agreement with the watershed time of concentration; the weighted

average error was 10.4 percent, and four of the 6 watersheds showed errors less

than 10 percent. Simultaneously, the calibration process provided dimensionless

unit hydrographs for each watershed; the optimum peak rate factors varied from

352 to 495, with four of the six watersheds having values that were significantly

different from the 484 value that is used with the standard SCS dimensionless unit

hydrograph. Runoff hydrographs computed with a peak rate factor of 484 were not

as close to the observed runoff hydrograph as the hydrographs computed with the

correct time of concentration and peak rate factor. These results support the

hypothesis that the automatic fitting program can increase the accuracy of flood

hydrograph predictions.

-G19-

REFERENCES

1. Dooge, J.C.I., Linear Theory of Hydrologic Systems, Tech. Bull. No. 1468,
Agricultural Research Service, USDA, Washington, D.C., October 1973.

2. McCuen, R.H., "The Role of Sensitivity Analysis in Hydrologic Modeling,"
Journal qf Hydrology, Vol. 18, pp. 37-53, 1973.

3. McCuen, R.H., and Bondelid, T.R., "Estimating the SCS Peak Rate Factor,"
Technical Report, Dept. of Civil Engineering, University of Maryland,
College Park, 1981.

4. Sarma, P.B.S., Delleur, J.W., and Rao, A.R., A Program in Urban Hydrology,
Part Il. An Evaluation of Rainfall-Runoff Models for Small Urbanized
Watersheds and the Effect of Urbanization on Runoff, Purdue University,
Water Resources Research Center, Lafayette, Indiana, Tech. Report No. 9,
October 1969.

5. Soil Conservation Service, Computer Programfor Project.Formulation -
Hydrology, Washington, D.C., Technical Release No. 20, Supplement No. 1,
Central Technical Unit, 1969.

6. Soil Conservation Service, National Engineering Handbook, Section 4,
Hydrology, Washington, D.C., U.S. Department of Agriculture, 1972.

7. Soil Conservation Service, Urban Hydrology for Small Watersheds,
Technical Release No. 55, USDA-SCS, January 1975.

8. Water Resources Council, E-stimating Peak Flow Frequencies for Natural
Ungaged Watersheds, Hydrology Committee, Washington, D.C., 1981.

9. Woodward, D.E., Welle, P.I., and Moody, H.F., "Coastal Plains Unit
Hydrograph Studies," Proceedings of the National Symposium on Urban
Stormwater Management in Coastal Areas, American Society of Civil Kngi-
neers, New York, pp. 99-107, 1980.

-G20-

APPENDIX H

A PLANNING METHOD FOR EVALUATING
DOWNSTREAM EFFECTS OF DETENTION BASINS

Mark E. Hawley

Timothy R. Bondeled

Richard H. McCuen

I I i

American Water Resources Association

Reprinted from:
Water -
Resources
0
u eti n
M
6666 B 11 .
OCTOBER 1981

WATER RESOURCES BULLETIN
VOL. 17, NO. 5 AMERICAN WATER RESOURCES ASSOCIATION OCTOBER1981

A PLANNING METHOD FOR EVALUATING DOWNSTREAM
EFFECTS OF DETENTION BASINS'

Mark E. Hawley, Timothy R. Bondelid, and Richard H. McCuen 2

ABSTRACT: While storm water detention basins are widely used for the timing of the runoff hydrograph. While SWM policies are
controlling increases in peak discharges that result from urbanization, designed to limit the effect of increases in peak discharge rate,
recent research has indicated that under certain circumstances detention they tend to ignore the effects of urbanization on the time
storage can actually cause increases in peak discharge rates. Because of characteristics of both direct runoff and flow through the SWM
the potential for detrimental downstream effects, storm water manage-
ment policies often require downstream effects to be evaluated. Such basin.
evaluation requires the design engineer to collect additional topographic McCuen (1979) has shown that in some circumstances the
and land use data and make costly hydrologic analyses. Thus, a method, change in timing caused by a SWM basin can result in increased
which is easy to apply and which would indicate whether or not a de- downstream flooding. This type of situation is illustrated in
tailed hydrologic analysis of downstream impacts is necessary, should Figure 1. In the predevelopment condition, the peak discharge
decrease the average cost of storm water management designs. A plan-
ning method that does not require either a large data base or a com- rate from area 2 passes point A before the peak from area I
puter is presented. The time coordinates of runoff hydrographs are esti- arrives. After development, the peaks arrive at point A at
mated using the time-of-concentration and the SCS runoff curve num- nearly the same time; this causes constructive interference and
ber; the discharge coordinates are estimated using a simple peak dis- results in a larger total peak discharge rate at point A, even
charge equation. While the planning method does not require a detailed though the peak flow from area 2 has not been increased over
design of the detention basin, it does provide a reasonably accurate
procedure for evaluating whether or not the installation of a detention that which occurred prior to urbanization. Because of this
basin will cause adverse downstream flooding. possibility, the timing of the SWM basin outflow hydrograph
(KEY TERMS: detention; flood control; hydrology; planning; reser- should be considered when a SWM basin is to be used for con-
voirs; storm water management.) trolling runoff from an area that is to be developed.
If SWM policies required all SWM basin analyses to include
INTRODUCTION measuring the downstream effects, the cost of SWM design
would increase substantially. In addition to the computer
Storm water management (SWM) basins are the most widely costs, it would be necessary to compile hydrologic, land use,
used means of controlling the rate of runoff from developed and soil data bases for the downstream area. Because SWM will
areas. In many jurisdictions, SWM policies have been estab- not always have a detrimental downstream effect, the average
"Ished with the intent of limiting the detrimental effects of
I project costs would be decreased if a simplified method that
urbanization on downstream areas. Usually, the policy requires would indicate whether or not a SW basin would have a
that the peak runoff rate from the urbanized area be controlled detrimental downstream effect were available; that is, a method
so that the peak discharge rate after development does not ex- is needed that can be used in the planning stage of site develop-
ceed the predevelopment peak discharge rate. Unfortunately, ment to indicate whether or not downstream effects should be
Rmiting the peak discharge rate in this manner does not neces- incorporated into the design of a SWM basin. The intent of
sarfly ensure that the detrimental effects at downstream points this paper is to present such a method and illustrate its applica-
a:m minimized (McCuen, 1974, 1979). tion.
Urbanization decreases the natural storage capacity of a
w2tershed; therefore, the volume of direct runoff from a given
depth of rainfall increases with development. Urbanization DEVELOPMENT OF THE PLANNING METHOD
also has the effect of decreasing the time required for water to
Cow from remote points to the watershed outlet; that is, the Analysis of Planning Requirements
time of concentration is inversely proportional to the degree of The best way to determine the downstream effects of a SWM
urbanization. Thus, urbanization affects both the volume and basin is to compare the predevelopment and postdevelopment

I Paper No. 81036 of the Water Resources Bulletin. Discussions are open until June 1. 1982.
2 Respectively, Faculty Research Assistant and Graduate Research Assistant, Department of Civil Engineering; and Associate Dean, College of Engineer-
ing; University of Maryland, College Park, Maryland 20742.

WATER RESOURCES BULLETIN

Hawley, Bondelid, and McCuen

hydrographs at downstream points. In the planning stage it is SWNI basins. Examination of a few SWM basin hydrographs
not necessary to have exact hydrographs; reasonably accurate showed that the characteristic shape of a SWM basin outflow
estimates of the hydrographs should be sufficient. The hydro- hydrograph is considerably different from the characteristic
graph at a downstream point, such as point A in Figure 1, is shape of a direct runoff hydrograph. Examples of these two
the sum of the hydrographs from the upstream areas I and 2. hydrograph shapes are shown in Figure 1. A SWM basin out-
The predevelopment hydrograph at A is the sum of the pre- flow hydrograph generally has a smaller slope on the risinglimb
development direct runoff hydrographs from areas I and 2, and an extended period during which flows are at Or near the
while the postdevelopment hydrograph at A is the sum of the peak raw the shape of the basin outflow hydrograph is more
direct runoff hydrograph from area I and the SWM basin out- nearly trapezoidal than the triangular shape of a direct runoff
flow hydrograph from area 2. Thus, the planning method con- hydrograph. For this reason, separate models are required for
sists of nothing more than a means of quickly estimating three estimating the SWM basin outflow and direct runoff hydro-
hydrographs, the predevelopment direct runoff hydrograph graphs.
from area 2, the direct runoff hydrograph from area I , and the Storm water management policies often require SWM basins
SWM basin outflow hydrograph representing the postdevelop- to control the postdevelopment runoff so that the peak dis-
ment hydrograph from area 2. charge is no greater than the pTedevelopment peak discharge.
Therefore, the predevelopment and postdevelopment peaks can
be considered equaL the TR-55 graph method is a fast and
A- 0 (1. be dovelOped) A- 1 0. .-d widely used method of estimating peak discharge, although it
does not provi 'de any estimates of the timing characteristics
"00106:0 of the runoff. To be consistent, it is best to use the same tech-
.WM Is. niques for estimating both direct runoff and SWM basin out-
flow hydrographs-, therefore, the logical method is to use the
PREDEVELOPMENIT HYDROGIIAPHS TR-55 graph method for estimating the peak discharge rates
and to develop empirical equations for estimating the time
coordinates of all the hydrographs. Because of the funda-
mental differences between the two types of hydrographs, it
010CHARGE Composite hV010016ph was not possible to develop one set of timing equations that
would work for both types; therefore, separate sets of timing
A,;@ I 01foct --tt hydrogroph equations were developed for the runoff and SWM basin out-
A- 9 01-1 -aft flow hydrographs.

TIME 0-0 Data Generation
POSIDEVELOPMENT HYDROGRAPHS In order to develop equations for predicting time coor-
dinates of hydrographs, a data base encompassing a wide range
of watershed conditions was required. Collection of a set of
measured hydrologic data that included the proper mix of
watershed conditions was not possible-, therefore, a synthetic
A- .1 011.@I I... 11 data base was generated. Because of the fundamental dif-
A,** 2 *WM beat. ferences between the characteristics of the direct runoffhydro-
...... graphs and the SWM basin outflow hydrographs, different
TIME (he.,.) methods were used for synthesizing these two types of hydro-
graphs.
Synthesizing the Direct Runoff Hydrographs. The data
Figure 1. Illustration of the Possibility of Increased base for the direct runoff hydrographs was generated using the
Peak Discharges Due to SWM Basins. SCS TR-20 runoff model with a variety of watershed condi-
tions. The usual data requirements for TR-20 include the area
In order to simulate these hydrographs quickly, estimates of the watershed, the average slope, the curve number, the hy-
of both the timing Coordinates and the discharge coordinates draulic length, the time of concentration, and the depth of
are required. Two models that provide these estimates are the rainfall. The type 11 storm distribution with half-hour rainfall
SCS TR-20 model and the TR-55 tabular model. The TR-20 increments was used for all analyses in this study. The average
model requires a considerable data base and a computer, and slope and the hydraulic length are used only in computing the
is therefore not practical for preliminary planning purposes. time of concentration; therefore, if the time of concentration
is specified, then these data need not be included. The area
Tbe TR-55 tabular method is less complex than TR-20 and can does not. affect the timing of the runoff hydrograph if the time
be performed by hand, but it is not capable of estiniating SWM of concentration is specified. Because the TR-55 graph method
basin outflow hydrographs. Ile SCS methods were designed results in peak discharge in efs per unit area rather than in cfs,
to simulate natural storage conditions rather than man-made it was not necessary to vary the watershed area in generating

-H2- WATER RESOURCES BULLETIN

A Planning Method for Evaluating Downstream Effects of Detention Basins

the data base. The remaining watershed variables are the curve Because the outflow hydrograph is not sensitive to the basin
number, the rainfall depth, and the time of concentration. parameters, the Bondelid program for generating feasible SWM
Values of the curve number were 60, 70, 80, 90, and 95;values basin designs was modified to give only one feasible solution
of the rainfall depth were 2.8, 4.3, and 7.0 inches, which cor- for each combination of watershed conditions. The five water-
respond to the 2-year, 10-year, and 100-year storms in the shed conditions that were varied in creating this data set are
state of Maryland', values of the time of concentration ranged the watershed area, average slope, rainfall depth, predevelop-
from 0.25 to 2.0 hours in quarter-hour increments. Thus, 120 ment curve number, and postdevelopment curve number. The
combinations of values were used to generate 120 direct run- predevelopment and postdevelopment times of concentration
off hydrographs; these hydrographs form the data base used in were calculated using the SCS lag formula:
developing the timing equations for the direct runoff hydro-
graphs. tc = 5 x LO.8(S+1)0.7
Generating the SWM Basin Outflow Hydrographs. The SWM 3 1900 YO.5
basin outflow hydrographs were generated using the Bondelid
method (Bondelid and McCuen, 1980). Tests showed that the in which tc is the time of concentration (hrs), L is the com-
timing of the SWM basin outflow hydrograph is determined puted hydraulic length (ft), Y is the average slope (percent),
almost entirely by the depth of rainfall and the watershed and S is related to the curve number by the function:
conditions. The peak outflow rate at any one site may be
limited to the predevelopment peak by using many different S 1000 - 10 (2)
SWM basin configurations, but all of these configurations will (curve number)
result in nearly identical outflow hydrographs. This is rational
because the peak flow rate and the volume of basin storage The computed hydraulic length, L, is derived from the water-
required are not functions of the SWM basin design parameters shed area using the function:
(i.e., size and number of riser pipes, depth, side slope); rather,
the SWM basin parameters are dictated by the peak flow rate L = 209 x (area)0.6 (3)
and the volume of basin storage that is required. The peak
flow rate is determined by the predevelopment watershed con- All of these equations are designed for use with SCS hydro-
ditions, and the volume of basin storage is determined by the graph models (SCS, 1972) and are compatible with the Bonde-
differences between the predevelopment and postdevelopment lid method of designing SWM basins. The values of the water-
watershed conditions. Therefore, the SWM basin design shed area that were used are 20, 60, and 100 acres; values of
parameters have very little effect on the basin outflow hydro- the predevelopment curve number were 60, 70, and 80; post-
graph as long as the set of basin design parameters constitutes development curve numbers ranged from 65 to 90 in incre-
a feasible solution. Table I illustrates this point by showing ments of 5 and were always greater than the predevelopment
the timing of the outflow hydrograph for each of several dif- curve number; values of the average slope were 1, 3, 5, and 7
ferent feasible SWM basin designs. The outflow hydrographs percent; the rainfall depths were 2.8, 4.3, and 7.0 inches. Sub-
of the various basin designs exhibit almost no variation in tim- stituting the various combinations of these values into the
ing.

TABLE 1. Timing of SWM Basin Outflow Hydrographs From Various Basin Designs.
Riser Diameter Volume of Storage Surface Area Depth* Time (hrs) From Start of Precipitation
Design (ft) (acre-feet) (acres) (ft) T50R T75R T P T75F T50F

1 3.50 3.048 9.599 0.758 12.0 12.2 12.8 14.4 15.7
2 3.25 3.167 9.077 0.802 12.0 12.2 12.8 14.5 15.8
3 3.00 3.187 8.305 0.867 12.0 12.2 12.8 14.5 15.8
4 2.75 3.199 7.437 0.952 12.0 12.2 12.8 14.5 15.8
5 2.50 3.149 6.353 1.085 12.0 12.2 12.8 14.5 15.7
6 2.25 3.080 4.971 1.294 12.0 12.2 12.8 14.6 15.7

Watershed Conditions:

Watershed Area = 100 acres.
Predevelopment Curve Number = 70.
Postdevelopment Curve Number = 80.
Predevelopment Time of Concentration = 1.0 hours.
Postdevelopment Time of Concentration = 0.5 hours.
Rainfall Depth = 2.8 inches.
*Depth is maximum height of water surface above top of riser.

-H3- WATER RESOURCES BULLETIN

Hawley, Bondelid, and McCuen

Dondelid program resulted in a data set of 324 SWM basin variables were calculated separately for each of the 2-year,
outflow hydrographs. I 0-year, and I 00-year design storms.
Analysis of the resWts showed that the correlations between
Developing the 77ming Equations predictor and criterion variables were very similar for all three
An examination of the direct runoff hydrographs showed design storms. In each regression, the first variable to enter the
that they could be approximated with the required degree of equation was the time of concentration; the curve number was
accuracy by a triangular hydrograph; three points on the hy- always the second variable to enter. Because the differences
drograph had to be located, then joined by straight lines. The in the rainfall depths of the three design storms did not make a
SW' M basin outflow hydrographs are much flatter than the significant difference in either the correlation between the
direct runoff hydrographs; therefore, five points were used in variables or the regression coefficients, the data set was recom-
estimating the basin outflow hydrographs. Typical estimated bined and all 120 observations were used with the stepwise re-
hydrographs of both types are shown in Figure 2. gression program. The resulting timing equations for the direct
runoff hydrographs are shown in Table 2, along with goodness-
of-fit statistics; these equations are applicable to 24-hour type
11 design storms of any return period. The very high correla-
tion coefficients and low standard errors of estimate indicate
that these three points (TP, T50R, and T500 can be predicted
very accurately within the range of the calibration data.
7; Poo - Timing Equations for the SWM Basin Outflow Hydro-
... graphs. Equations were developed for estimating the times to
five discharge stages of the basin outflow hydrographs. These
loo -
five equations estimate the times to reach 50 and 75 percent
loo - of peak discharge on the rising limb (T50R and T75R, respec-
tively), and time to peak (Tp), and the times to reach 75 and
50 percent of the peak discharge on the falling limb (T75F and
T50F, respectively). The equations were developed by using
stepwise regression techniques with the data set generated by
"1 11.5 12 12.6 '3 13.6 .6 .6 the Bondelid method. The predictor variables that were avafl-
able were: 1) the area, the average slope, and the computed
hydraulic length of the watershed; 2) the predevelopment
Figure 2. Estimated Hydrographs for Example curve number, time of concentration, runoff depth, and peak
of Planning Method. discharge rate; 3) the postclevelopment curve number, time of
concentration, runoff depth, and peak discharge rate that
would occur if no SWM basin were installed; and 4) the change
Peak discharge rates for the estimated hydrographs were in the curve number, the rainfall depth, the volume of basin
estimated using the TR-55 method; equations that can be used storage required, and the ratios of the predevelopment to post-
to estimate the timing coordinates of the hydrographs were development runoff depths and the predevelopment to post-
developed. These equations were developed by using regres- development discharge rates. All 324 SWM basin outflow hy-
sion techniques with the generated hydrograph data base and drographs were used as the data base; no distinction was made
are referred to as the timing equations. on the basis of rainfall depths because one set of timing equa-
Timing Equations for Direct Runoff Hydrographs. The first tions was desired that would be applicable to any type 11 design
step in deriving the timing equations for the direct runoff hy- storm depth. The timing equations that resulted from this
drographs was to separate the synthesized data into three analysis are shown in Table 3, along with goodness-of-fit sta-
groups on the basis of the depth of rainfall. The hydrographs tistics and the standardized partial regression coefficients.
were synthesized for design storms of 2.8, 4.3, and 7.0 inches The standardized partial regression coefficients in Table 3
because these are the depths of the 2-year, I 0-year, and 100- are indicative of the relative importance of the predictor vari-
year design storms in Maryland. Separate equations were ables. On the rising limb of the hydrograph, the postdevelop-
developed for each rainfall depth in order to determine whether ment time of concentration is very important, while the runoff
one set of equations could be used for all return periods. A depth is considerably less important; both of these variables
stepwise regression was then performed with each data set. reflect the watershed conditions rather than the SWM basin
The criterion variables were the time to peak (Tp), the time to design, so it seems that The timing of the outflow hydTograph
50 percent of peak on the rising limb (T50R), and the time to on the rising side is almost entirely controlled by watershed
50 percent of peak on the falling limb (T50F). The predictor characteristics. At the peak discharge and on the falling side
variables were the curve number (CN), the time of concentra- of the hydrograph, the ratio of the postdevelopment peak dis-
tion (tc), the depth of runoff (RO), and the peak flow rate charge rate that would occur in the absence of a SWM basin
(Qp). Correlation matrices of the predictor and criterion to the predevelopment peak discharge rate is more important
as a predictor variable than the depth of runoff. This ratio of

-H4- WATER RESOURCES BULLETIN

A Planning Method for Evaluating Downstream Effects of Detention Basins

TABLE 2. Tinting Equations and Goodness-of-Fit Statistics for the Direct Runoff Hydrographs.

Correlation Standard Error of Standard Deviation of
Criterion Equation Coefficient Estimate (hours) Criterion Variable (hours)

TSOR 12.20278 + 0.30759 * tc - 0.00651 * CN 0.9559 0.06 0.20
(.8641) (-0.4089)

Tp 12.32SIS + 0.65588 * tc - 0.00622 * CN 0.9702 0.10 0.40
(.9491) (-0.2012)

T50F 13.16962 + 1.36070 * tC - 0.01613 * CN 0,9452 0.28 0.86
(.9136) (-0.2421)

NOTES:
tc = time of concentration (hours).
CN = curve number.
T5OR time to reach 50 percent of peak flow on rising limb (hours).
Tp time to reach peak flow (hours).
T50F time to reach 50 percent of peak flow on falling limb (hours).
Numbers inside parentheses are standardized partial regression coefficients.

TABLE 3. Timing Equations for SWM Basin Outflow Hydrograph Estimation.

Correlation Standard Error of Standard Deviation
Criterion Equation Coefficient Estimate (hours) of Criterion (hours)

T50R. 11.79903 + 0.53128 * tcpost - 0.04274 * Q post 0.9640 0.06 0.23
(0.8548) (-0.3063)

T75R 11.90993 + 0.70941 * tcpost - 0,04727 * Q 0.9533 0.09 0.31
(0,8679) (-0.2575) post

Tp 11.39838 + 1,51487 * t + 0.18329 * q t/q pre 0.9387 0.29 0.83
(0.6848) cpost (0.7337) pos

T75F = 9.95355 + 3.12555 * t + 0-91341 * qpost /q pre 0.9591 0.92 3.24
(0.3613) cpost (0.9349)

T50F = 9.1829 + 4.18501 * t + I'35179 * qpost /q pre 0.9632 1.28 4.74
(0.3306) cpost (0.9453)

NOTES:
tcpost = postdevelopment time of concentration (hours).
Qpost = postdevelopment depth of runoff (inches).
qpost = postdevelopment peak discharge rate without SWM basin (cfs).
qpre = predevelopment peak discharge rate (cfs).
Numbers inside parentheses are standardized partial regression coefficients.

peak flow rates is indicative of the degree to which the basin in rainfall depth is included in each equation. In the first two
must control the runoff hydrograph. Thus, the importance of equations the effect of rainfall depth is reflected in the post-
this ratio can be considered as an indicator of the importance development depth of runoff. In the last three equations, the
of the SWM basin characteristics in determining the timing of effect of rainfall depth is inherent in the ratio of the peak dis-
the SWM basin outflow hydrograph. The standardized partial charge rates, because the depth of rainfall is very important in
regression coefficients show that as the rate of discharge de- determining the ratio. The change in the ratio with rainfall
creases the SWM basin design becomes more important and the depth can be explained by reference to Figure 10-1 of Sec-
postdevelopment time of concentration becomes less impor- tion 4 of the SCS National Engineering Handbook (SCS, 1972).
tant to the timing of the hydrograph. For any pair of predevelopment and postdevelopment curve
The equations in Table 3 are suitable for use for any 24- numbers, the ratio of the postdevelopment to the predevelop-
hour, type I] design storm depth because the effect of variation ment runoff depth is much greater for small rainfall depths
- H5- WATER RESOURCES BULLETIN

Hawley, Bondelid, and McCuen

than for larger depths. This effect is due to the greater relative depth. These values are used with TR-55 to estimate the post-
-riportance of the initial abstraction in smaller storms. In the development peak discharge rate from area 2 that would occur
data base used in this study, the average value of the ratio of in the absence of SWM basin (qpost). Once qpost has been
peak discharge rates was about 5 for a 2.8 inch design storm determined, the ratio qpost/qpre can be calculated. This
and about 2 for a 7.0 inch design storm. This indicates that ratio, the postdevelopment depth of runoff (Qpost), and the
smaller storms require more control if the SWM basin outflow postdevelopment time of concentration (tcpost) are substituted
peak discharge rate is to be no greater than the predevelop- into the equations in Table 3 to estimate the time of coor-
ment direct runoff peak rate. dinates of the SWM basin outflow hydrograph. Because the
S*`M basin peak discharge rate is to be controlled so that it is
no greater than the predevelopment peak discharge from area 2
PLANNING PROCEDURE (qpre), qpre is used for the flow rate coordinates. The five
The planning procedure consists of estimation and cornpari- points on the SWM basin outflow hydrograph have coor-
son of the predevelopment and postdevelopment hydrographs dinates (T50R, qpre/2), (T75R, 3qpre/4), (Tp, qpre), (t75F,
at a critical downstream point. The critical point is usually 3qpre/4), and (T50F, qpre/2); the lines connecting these
the point at which the runoff from the area to be urbanized points represent the estimated SWM basin outflow hydrograph.
(area 2 in Figure 1) joins the runoff from the undeveloped Estimation of the Downstream Hydrographs
area (area I in Figure 1). This junction point (point A in
Figure 1) is critical because channel flow from this junction The important hydrographs are not those for the individual
tends to smooth out the hydrograph and decrease the peak areas, but rather the downstream hydrographs formed by the
discharge rate-, therefore, if constructive interference between addition of the individual area hydrographs. Figure 2 shows
the area I direct runoff hydrograph and the area 2 SWM basin how the downstream hydrographs are generated. The pre-
outflow hydrograph causes the postdevelopment peak discharge development downstream hydrograph is simply the sum of the
rate to be greater than the predevelopment peak, the effect predevelopment area 2 direct runoff hydrograph and the area I
will be most apparent at this junction point. . direct runoff hydrograph. The postdevelopment downstream
In estimating the predevelopment and postdevelopment hydrograph is the sum of the SWM basin outflow hydrograph
hydrographs at the junction point the discharge rates are esti- and the area I direct runoff hydrograph. In Figure 2 the in-
mated using TR-55, while the timing coordinates are estimated stallation of a SWM basin appears to increase the peak flow at
using the equations in Tables 2 and 3. An important part of downstream points significantly.
the planning process is the selection of a design storm; in many
jurisdictions, legislation or established policy dictates this Example of the Planning Method
choice. Selection of a design storm is specific to each situation To illustrate its use, the planning method was applied to
and will not be discussed in this paper. Once the design storm the Crabbs Branch watershed in Montgomery County, Mary-
has been selected, which establishes the depth of rainfall, the land. Figure I contains a schematic diagram of the watershed,
planning procedure proceeds as described in the following the data and calculations appear in Table 4, and the estimated
paragraphs. hydrographs are shown in Figure 2. The hydrographs in Fig-
Estimation of Direct Runoff Hydrographs ure 2 indicate that the proposed development of area 2 and the
installation of a SWM basin will result in an increased peak dis-
In using the TR-55 graph method for estimating peak dis- charge rate at point A. The predevelopment peak rate is about
charge rates, the required data are the watershed area, the rain- 230 cfs and the postdevelopment peak is about 265 cfs, in-
fall depth, the curve number, and the time of concentration. dicating an increase of roughly 15 percent. In order to sub-
A number of methods for estimating time of concentration stantiate these results, the example was recomputed using
are described in the TR-55 manual; all of these methods are TR-20 to generate the direct runoff hydrographs and the
compatible with the tiniing equations. Once the data have Bondelid method (Bondelid and McCuen, 1980) to generate
been collected, TR-55 can be used to estimate the peak flow the SWM basin outflow hydrograph. The results of this simula-
rates. Then the equations in Table 2 can be used to estimate tion indicated an increase in peak discharge only slightly lower
the time coordinates. The three points on the hydrograph may than that indicated by the planning method.
be located by the coordinate pairs (T50R, Qp/2), (T Q P ), and
7 50F, Qp /2). Connecting these points with straight Wnes, as in
Figure 2, completes the estimation of the direct runoff hydro- CONCLUSIONS
grap'hs for the undeveloped area (area I in Figure 1) and for On some watersheds the use of SWM basins to control run-
the predevelopment conditions on the area being developed off from developing areas can increase discharge rates at down-
(area 2 in Figure 1). stream points rather than limiting discharges to those which
Estimation of SWM Basin Outflow Hydrograph occurred prior to development. In this paper, a quick planning
method for estimating the potential for adverse downstream
The estimation of the SWM basin outflow hydrograph re- effects of a SWM basin has been developed. The planning
quires knowledge of the postdevelopment curve number and method involves estimation of direct runoff hydrographs and
time of concentration as well as the watershed area and rainfall SWM basin outflow hydrographs. The TR-5S graph method
-H6- WATER RESOURCES BULLETIN

TABLE 4. Worksheet for Planning Method.

Design Storm: Return Period = /0 years P (depth of rainfall) 3 inches

Area 2** Area 2**
Variable Area I* Predevelopment Postdevelopment Eq. No. Timing Equations for Area I

A = watershed area (square miles) A 170 A 0-07 A= 0. 0'7 1. T = 12.20278 + 0.30759 t - 0.00651 CN
1 2 3 50R cl 1 0
C.
CN = curve number CNI 65 CN2 CN3 2. Tp = 12.32515 + 0.65588 t cl - 0.00622 CN I E
tv,
.C
t time of concentration (hours) t=As t 0, us" t O.Ao 3. T = 13.16962 + 1.36070 t - 0.01613 CN
c cl c2 c3 Sol., cl

S
=1 S = retention = (1000/CN)-to S1 S2 = 3
Eq. No. Timing Eqs. for Area 2: Predevclopmen(
Q = runoff volume = (P-0.2S)2/(P+0.8S) (inches) .7. 3? 0
QI /-2/ Q2 = 0' r66 Q3 I
qu, '73 .5 q 4. T 12.20278 + 0.30759 t 0.00651 CN
q, = unit peak discharge from Fig. 5-2, TR-55 manual (CSW/inch) .23S qu2 u3 1'0'06 50R= c2 2
El
q = peak discharge = A * qu * Q (cfs) qQ36 q q S. Tp = 12.32515 + 0.65588 t 0.00622 CN rM
p PI p2 p3 -7 c2 2
a = qp3 /q p2 a= 41. ICA 6. T 50F= 13.16962 + 1.36070 t c2 0.01613 CN2 0
T = time-to-rise to 50 percent of q (hours) Fq. I /,?. aq Eq. 4//.90 Eq. 7 /1. 175*
50R p
Eq. No. Timing Eqs. for Area 3: Postdevelopment
T time-to-risc to 75 percent of q (hours) Eq. 8
75R = p w
= time-to-rise to qp (hours) Eq. 2 1.1.91 Eq. 5/Q./-3 Eq. 9 9 7. T 11.79903 + 0.5 3128 t - 0.04274
Tp SOR c3 Q3
-4 T75F= time-to-fall to 75 percent of q p (hours) - Eq. 10 13, 9 4/ 8. T 75R= 11.90993 + 0.70941 tc3- 0.04727 Q3
M
T time-to-fall to 50 percent of q (hours) Eq. 3 AJA Eq. 6 5@7 Eq. I t 16-0 J/ 9. Tp = 11.39838 + 1.5 1487 t c3 + 0.18329 a
50F p
M
8 10. T751-'= 9.95355 + 3.12555 tc3 + 0.91341 a
C
w It. T 9.18290 + 4.18501 t + 1.35179 c
0 sor. c3
M
CO Area I is the area upstream from the area being developed.
C "Area 2 is the area being developed.
r-
r
M
--I
z

Hawley, Bondetid, and McCuen

und for esfirr;;ting discharge rates, and empirical timing ACKNOWLEDGMENTS
@@quations are used for estimating the time coordinates of the The research on which this report is based was supported, in Part,
',Ydrographs. The method was calibrated using simulated hy- by the Tidewater Administration, Department of Natural Resources,
irographs developed with the TR-20 model. The calibration Annapolis, Maryland.
data included times of concentration from 0.25 to 2.0 hours,
znd testing indicates that the method is reasonably accurate
@,cr times of concentration as low as 0.10 hours. Once the al- LITERATURE CITED
lowable peak discharge rate from a SWM basin has been set,
Bondelid, T. R. and R. H. McCuen, 1980. A Computerized Method for
.rariations in the SWM basin design parameters do not signifi- Designing Stotmwater Management Basins. Tech. Report, Depart-
czntly affect the SWM basin outflow hydrograph-, therefore, ment of Civil Engineering, University of Maryland.
e planning method does not require that a SWM basin be McCuen, R. H., 1974. A Regional Approach to Urban Storm Water De-
@esigned. The planning method is a fast and reasonably ac- tention. Geophysical Research Letters 1(7):321-322.
curate way to determine whether or not installation of a SWM McCuen, R. H., 1979. Downstream Effects of Stormwater Management.
Journal of Water Resources Planning and Management, ASCE 105
basin will increase flooding at downstream points. (HY3):1343-1356.
Soil Conservation Service, 1972. National Engineering Handbook, Sec-
tion 4, Hydrology. U.S. Dept. of Agriculture, Washington, D.C.

WATER RESOURCES BULLETIN

APPENDIX I

INTEGRATED STORMWATER
MANAGEMENT DESIGN

Stanley L. Wong

Richard H. McCuen

INTEGRATED STORMWATER MANAGEMENT DESIGN

Stanley L. Wong and Richard H. McCuen

INTRODUCTION

The changes that occur during the development of a watershed have a direct

influence on the runoff response of a watershed to a precipitation event. The

impact of land development is often assessed by the extent of land use changes

that accompany land development. Land use changes, especially in urban areas,

usually cause increases in both the peak runoff rate and the runoff volume. The

degree of pervious cover is reduced, thereby changing the timing characteristics

of a watershed and decreasing the travel time. Less natural storage is avail-

able during the initial period of runoff, which causes the volume of the runoff

hydrograph to increase.

When the extent of land cover change is significant, the detrimental conse-

quences of such change, including downstream flooding, erosion, and poor water

quality, must be prevented with appropriate stormwater management (SWM) tech-

niques. Many of the adverse consequences may occur concurrently at different

locations and may create additional problems if not properly mitigated. For

example, the runoff response of construction sites that are left devoid of vege-

tative cover will result in larger direct runoff volumes within a shorter time

span, increases in the velocity of overland flow, and a significant increase in

erosion and sediment transport. Therefore, a variety of criteria must be examined

in measuring the effectiveness of a SWM program. These efficiency criteria may

include: 1) flood peak control; 2) flood volume control; 3) erosion control;

IGraduate Research Assistant and Professor, Department of Civil Engineering,
University of Maryland, College Park, MD 20742.

4) sedimentation capacity; 5) reduction in overland flow velocity; 6) induced

infiltration and groundwater maintenance; and 7) various water quality enhance-

ments. The selection of appropriate SWM control measures should reflect the effect

of the program on each of these efficiency criteria.

Various SWM measures have different efficiencies with respect to each of

the above criteria. Thus, single method control plans do not achieve maximum

efficiency. The use of more than one SWM method should take advantage of the

strength provided by each measure with respect to all of the criteria listed.

Hence, an integrated approach to S101 control can maximize efficiency by accounting

for the interaction and interdependent effect of the individual Sh1v1 methods. The

implementation of a combination of SWM controls depends on policy requirements,

as well as hydrologic and hydraulic considerations. The objectives are to assess

the efficiency criteria obtainable by various SWM control techniques and to pro-

pose an integrated SWM design method that will maximize the efficiency of a SWM

program.

EFFICIENCY CRITERIA FOR STORMWATER MANAGEMENT METHODS

The goal of a SWM program for controlling storm runoff volumes and rates,

erosion, and sediment transport is to provide the controls that are necessary

to ensure that there is little change in the response of a watershed when com-

pared to that resulting from the original predevelopment conditions. The changes

in land cover and drainage patterns may cause the natural drainage capabilities

to be exceeded, and thus, require implementation of SWM controls.

The following is a summa ry of various efficiency criteria that should be

used to assess the effectiveness of SWM control programs. The adaptability and

feasibility of a SIVM control technique is reflected by the extent of fulfillment

to each of these criteria. The summary should aid in the selection of the appro-

priate approach or combination of approaches to mitigate the adverse effects of

development.
-r2-

Peak Runoff Rate

One of the primary effects of land cover changes is the significant increase

in peak runoff rates when compared with the predevelopment peak rates. The amount

of increase will depend upon the extent of development and the distribution of

impervious cover. Reductions in the peak discharge are often the only criteria

that is required in the design of SWM facilities. Detention basins, for example,

are commonly accepted as being a sufficient SWM control because the structure is

capable of limiting the peak outflow release rate to predevelopment rates. How-

ever, the time-to-peak is often changed significantly and the duration of the peak

rate is extended (McCuen, 1979). This results in an extended period of downstream

bankfull. flows.

SWM controls that reduce peak runoff rates are effective in limiting downstream

flow rates. However, detention basin designs most often are limited to the control

of a single return period; this may lead to inadequate control of storms for other

return periods (Kamedulski and McCuen, 1979). Evaluation of risks associated

with peak flows often constitutes the major concerns of SWM strategies when a

watershed undergoes extensive urbanization.

Volume of Runoff

The decrease in the natural available storage of a watershed causes a cor-

responding increase in the volume of direct runoff. Hence, various SWM measures

that provide for either man-made storage or natural storage attempt to replace

the natural storage that was lost in development. The downstream flow capacity

requirements may, therefore, be alleviated. Man-made storage methods (e.g.,

holding ponds, rooftop ponding, and parking lot storage) are most often imple-

mented to substitute for losses in natural storage. The effectiveness in con-

trolling runoff volumes depends on the available storage within the structure and

the location of the storage control facility within the watershed.

-13-

Erosion Control

Construction activities are generally known to generate substantial amounts

of sediment. Prevention of erosion from exposed soil surfaces is a necessary

conservation practice that should be included in planning phases. Methods to

stabilize the surface to resist erosion are more effective in controlling sediment

than measures that attempt to mitigate effects on flows that are heavily ladened

with sediment. Hence, efforts to prevent erosion should be investigated, including

the lining of ditches.

Sedimentation Controls

Erosion occurs despite the careful implementation of preventive measures.

Hence, additional attempts should be made to avoid sediment transport to receiving

streams. Both structural and nonstructural methods are available, including:

1) check dams that increase the settling of solids; 2) vegetative covers that

act as filters to "strain" sediment; 3) filter fences; 4) energy dissipating

devices to reduce both scour at outfalls and channel erosion; and 5) sediment

traps. Site conditions will often dictate the overall effectiveness of the

alternative methods.

Reduction in Overland Flow Velocities

Increases in runoff rates are associated with increases in overland flow

velocities. High flow rates can cause serious damage to surface conditions,

drainage patterns, and the integrity of channels. Urbanization results in

velocities of sufficient magnitude such that extensive storm drain construction

is required to convey flows around or through environmentally sensitive areas.

Delaying of flows in storage structures, increasing surface retardance, and

temporary barriers are alternate methods of reducing flow velocities.

-14-

Induced Infiltration

Increases in the percentage of impervious area will decrease the natural

infiltration and depression storage causing changes in the timing of runoff.

Inducement of infiltration enables the maintenance of the natural drainage sys-

tem and replenishment of groundwater supplies. Volumes of runoff may be effec-

tively removed from runoff hydrographs if sufficient vegetative buffer zones

are maintained. Overland flow is detained and decreased by that amount which

is infiltrated. Porous pavements and infiltration beds also utilize natural

infiltration to decrease runoff.

Water Quality Enhancement

The runoff quality criteria is not independent of water quantity control

criteria. The factors that are important in controlling runoff rates are not

separate from those that are important in water quality control. For example,

the control of sedimentation is considered to be water quality criteria because

many pollutants are adsorbed on sediment surfaces and transported along with sus-

pended solids. Hence, many water quality parameters are linked to water quantity

criteria.

AN INTEGRATED APPROACH TO SWM DESIGN

The hydrologic model used as the basic element of an integrated approach is

the Soil Conservation Service (SCS) tabular method (SCS, 1975). The tabular

method provides a means of estimating the runoff hydrograph from small watersheds

undergoing development. The tabular method is adapted because of the versatility

in evaluating SWM designs. The method is designed to develop composite hydro-

graphs at any point within the watershed for one or more subwatershed areas.

Additionally, the tabular method permits the assessment of effects that a com-

bination of SWM controls have on the composite hydrograph.

-15-

Tabular hydrograph discharge values are computed for subareas using a time
of co ncentration (tc) and a total travel time (tt) to the design point as input.
The peak discharge is determined from the equation:

q = q P A Q (1)

where q is the estimated peak discharge in cubic feet per second, q p is the
tabular unit peak discharge rate in cubic feet per second per square mile per

inch of runoff, A is the drainage area in square miles, and Q is the direct run-

off in inches. The hydrograph coordinates at the design point due to each sub-
area is determined using the values of t C, ttP A, Q, and q p for each subarea.
The total composite hydrograph is computed by summing the subarea hydrographs.

The effects of development can be determined by computing the discharge

hydrographs for present and future conditions. Land cover changes alter both

the peak discharge and timing characteristics of the runoff. Additionally, the

direct runoff volume, Q, is often greater after development due to increases in

the percentage of impervious area.

To offset the effects of development, SWM control measures are implemented.

The effectiveness of a SWM method will vary depending on both the hydrologic and

hydraulic conditions of the site, the importance of the efficiency criteria dis-

cussed previously, and the extent of development. Control of the peak discharge

rate and volume of runoff is managed through reductions in the direct runoff, Q.

The values of direct runoff may be sufficiently reduced depending on the type

and spatial configuration of the SWM controls employed. When SWM controls are

employed extensively, greater reductions in direct runoff are achieved. The

composite hydrograph resulting from the addition of alternate SWM controls is

computed by:

q = q p A Q1 (2)

-16-

in which Q1 is an adjusted direct runoff depth that represents the future condi-

tions with the addition of alternate SWM control methods. The value of Q1 will

be assessed in the same way as the value of Q except that the value will be re-

duced by the amount of storage allocated to the storage losses for SWM control

in the subarea; that is, Q1 is the difference between Q and the losses due to

SWM methods within the subarea.

A method of estimating the required volume of storage within detention basins

and holding ponds is outlined by SCS (1975). These methods provide design pro-

cedures to estimate the storage required in order to limit the outflow to a pre-

determined flow rate. The methods are based on average storage and routing effects

for many single-stage structures, specifically for weir flow and pipe flow

structures.

Recognizing that detention facilities do not provide the most efficient

solution to all of the problems created by development, a SWM solution may pro-

vide for incorporating various SWM measures in an integrated program. Also, the

implementation of many alternate SWM measures may effectively delay and reduce

the runoff volume and discharge rates that enter detention basins. Thus, the

total amount of storage volume required in detention basins can be reduced when

including the use of alternate SWM methods. The performance efficiency of the

SWM methods can be evaluated in terms of the criteria previously discussed.

-17-

STORMWATER MANAGEMENT CONTROL METHODS

The control of stormwater runoff may be managed through a variety of alter-

native methods. The selection of particular control methods will vary depending

on the conditions of development, physical land features, and hydrogeologic

conditions. Regardless of the SWM method selected, proper planning and design

procedures must be maintained to ensure an efficient control strategy.

The design procedures of SWM practices are important because efficiency

criteria are not applicable for all development conditions. Limitations exist

for some SWM methods where conditions are not favorable. For example, the use

of infiltration trenches may not be practical in areas with high water tables.

Therefore, design parameters should be investigated to determine the feasibility

of implementing alternate SWM controls. A brief description of the design con-

siderations and efficiency criteria applicable to selected SWM methods are dis-

cussed.

Vegetative Buffer Strips

Vegetative buffer strips provide for infiltration and natural storage of

surface runoff. Buffer strips are also effective for controlling sediment load-

ings. The vegetation should be a deep rooted, dense grass that is placed perpen-

dicular to surface flows. Design parameters include: 1) soil type; 2) surface

slopes; 3) type of vegetation; and 4) overland flow velocity (Wong and McCuen,

1981a). Placement of vegetative buffer strips will depend on the area available

and funds allotted for installation.

Among the efficiency criteria satisfied are: reductions in the runoff volume,

erosion control, sediment removal, reductions in flow velocities, and water quality

enhancement. Provisions for infiltration are made available by delaying surface

flows to promote natural storage and thus, replenishing of groundwater supplies.

The vegetative cover serves as a filter to remove both suspended solids and

pollutants that are attached to the surface of sediment particles. Also, vege-

-18-

tative covers stabilize the soil surface, reducing the erosive forces of pre-

cipitation and overland flow. While vegetative buffer strips satisfy the

majority of efficiency criteria, they are seldom effective when applied without

a series of other SWM alternatives because of their limited effect on flow volumes.

Infiltration Trenches

Infiltration trenches decrease the volume of runoff by providing subsurface

storage. Trenches are excavated and filled with crushed stone where runoff is

stored within available pore space. Filter clothes and topsoil backfill are

placed over the stones to filter runoff, prevent the clogging of voids, and in-

crease the design life (Wong and McCuen, 1981b).

Design considerations are important to achieve maximum efficiency. Topsoil

backfill and filter cloths overlaying stones need to have high permeability

rates to allow surface flows to be transmitted into storage. High groundwater

elevations are to be avoided so that depletion of storage is permitted. Additional

vegetative buffer strips placed upstream will further filter sediment particles.

Maintenance is required when void spaces are clogged.

Rooftop Ponding

Rooftop ponding can be very effective for controlling runoff volumes,

especially in highly urbanized commercial, industrial., and multi-family resi-

dential developments. Controlled flow roof drains are commercially available

and provide the means of reducing both flow rates and the volume of immediate

runoff. Since rooftops must be designed to withstand snow loadings, the roofs

can also be used for detention storage and provide as much as five area inches

of detention without structural modifications. Scuppers should be installed to

prevent the depth of ponding from exceeding five inches.

Design requires estimation of the depth-discharge relationship for the roof-

top outlet structure (McCuen and Piper, 1975). The volume of precipitation re-

_19-

tained (as opposed to detained) can be used as the volume for reducing Q in
Eq. 1; it is especially important to convert this volume to an area-depth amount

that is compatible with the subwatershed subarea. The volume retained will be
reflected in the lower portion of the depth-discharge relationship; it is a

function of the outflow characteristics of the drain.

Parking Lot Storage

Parking lot storage can be implemented to control runoff from large acres

of paved areas, including shopping centers and industrial districts. Because

paved areas have verX high runoff potentials in terms of both volumes and peak

rates, parking lot storage may reduce runoff rates by providing temporary storage

where flows are collected and released at a regulated rate. Water can be ponded

around portions of the parking lot that are least used to minimize-inconveniences

@o the public.

The volume of storage available is dependent on site conditions. The slope

and roughness of the area will govern the rate of inflow, and the depth of ponding

should not exceed eight inches for public safety reasons. The rate of release

may be regulated at the outlet with control devices, such as constricting pipes.

Proper design should prevent flooding with provisions for overflow during extreme

storm occurrences. Permanent water storage is not usually provided and, therefore,

parking lot storage is not effective in pollution control. Maintenance is easily

managed using mechanical street cleaning machines.

The depth of runoff (i.e., Q in Eq. 1) can be reduced by the volume stored

only if the outlet facility is designed as a retention storage basiTj@ That is,

the outlet facility should consist of one pipe or opening..,that releases water

for a duration of approximately 24 hours and a second larger.pipe or opening to

handle runoff that is in excess of the available storage. Only the storage volume

released through the smaller pipe should be considered in adjusting the value of

Q in Eq. 1.

_I10-

Porous Pavement

Porous pavements are a relatively new concept in SWM. Development of porous
pavements to infiltrate runoff from traditionally impervious surfaces is still in
the evaluation phase; however, many porous pavement sites do exist today. Porous

pavements will become more widely accepted as a viable SWM control measure as

performance criteria are further assessed.

Porous pavements are advantageous in reducing runoff volumes by allowing

water to absorb into the pavement, thus utilizing the natural storage in the

ground below the pavement. Hence, groundwater recharge is enhanced. The ef-

fects of creating a pervious area that would normally be impervious to water

(e.g., parking lots and roadways not receiving heavy loads) can reduce the flow

capacity required for storm drainage systems and reduce peak discharge rates.

Many design considerations need to be addressed to ensure the proper per-

formance of porous pavement systems during operation. The physical feasibility

is often limited to areas where the hydraulic conductivity of the underlying

soil is sufficiently high to transmit the downward movement of water; sub-

drains may be installed at regular intervals to expediate the drainage of

saturated layers. The drainability requirement of the porous pavement surface

and base courses are satisfied with open graded aggregate mixtures. Both the

durability and the load-bearing strength of the pavement are also important

design components. Additional design requirements include problems of with-

standing freeze-thaw cycles and the maintenance of clogged pores.

_I11-

INTEGRATED DESIGN IN COASTAL AREAS

Currently, there is special concern over the stresses placed on coastal en-

vironments, including wetlands, because of increased development, both urban and

agricultural, in coastal areas. Many states, county, and local governments are

developing stormwater management policies to control the adverse impact of land

development. Because of the proven effectiveness of on-site detention storage,

many policies that require or favor on-site detention exist. While on-site may

be a very effective control measure, there is some concern about its applicability

in coastal areas because of the large surface area that would be necessary to meet

the volume requirements.

The integrated design approach has significant potential for use in coastal

areas because the use of a variety of stormwater management methods will reduce

the volume of direct runoff and, thus, the volume of storage required. For urban

development in coastal areas, surface storage methods, such as rooftop and parking

lot detention, should be encouraged. For areas subject to agricultural develop-

ment pressures, vegetative buffer strips and infiltration trenches amy be used to

decrease the volume of surface runoff and, thus, the required on-site detention

storage. In-areas where water tables are high, every attempt should be made to

increase the length of drainage paths and other surface storage,

-112-

CONCLUSIONS

In summary, the interaction of the various efficiency criteria are apparent.

The physical impact of land development may create a variety of problems that

need to be identified with respect to each of the many criteria outlined herein.

Once these problems are recognized, then adequate S*1 controls may be implemented,

with each SWM control measure exhibiting its strengths.

The selection of specific SWM control measures to be implemented into an

integrated design approach should be based on the effect of the control measure

on the efficiency criteria outlined. Such an integrated approach can maximize

performance efficiency with respect to all of the criteria by accounting for the

interdependent effects of the SWM control measure. A composite of SWM controls

becomes effective because individual controls are inadequate in satisfying all

the efficiency criteria individually.

The integrated design approach recommended herein uses the SCS methods to

estimate the runoff hydrographs and the storage-volume estimation techniques for

estimating the volume of detention storage. -The effects of combining SWM controls

into a comprehensive design scheme can be evaluated using these procedures. Such

hydrologic modeling of watersheds will permit the evaluation of design techniques

to ensure that the intended SWM goals are achieved.

Watersheds that undergo development are subject to many adverse consequences

that, if not corrected, may cause both structural and environmental damage,

Unplanned growth should be avoided because. potential problems are not determined.

Proper planning should stem from policies that specify the intended goals of SWM

and adequate design techniques to translate policy into realized benefits.

LITERATURE CITED

McCuen, R.H., "Downstream Effects of Stormwater Management," Journal of Water
Resources Planning and Management Division, ASCE, Vol. 105(WIJ-1): 1343-1356,
1979.

Kamedulski, G.E., and McCuen, R.H., "Evaluation of Alternative Stormwater
Management Policies," J. Water Resources Planning and Management Division,
ASCE, Vol. 105(WR2): 171-186, 1979.

Soil Conservation Service, Urban Hydrology for Small Watersheds, Technical
Release No. 55, USDA, 1975.

McCuen, R.H., and Piper, H.W., 11 Hydrologic Impact of Planned Unit Develop-
ments," Journal of Urban Planning and Development, ASCE, Vol. 101(UPl):
93-102, 09-75.

Wong, S.L., and McCuen, R.H., "The Effects of Vegetated Buffer Strips on
Runoff Quantity and Quality," Technical Report, Department of Civil Engi-.
neering, University of Maryland, College Park, 1981a.

Wong, S.L., and McCuen, R.H., "Design of Infiltration Trenches for Control
of Stormwater Runoff," Technical Report, Department of Civil Engineering,
University of Maryland, College Park, 1981b.

-114-

APPENDIX J

THE DESIGN OF VEGETATIVE BUFFER
STRIPS FOR RUNOFF AND SEDIMENT CONTROL

Stanley L. Wong

Richard H. McCuen

THE DESIGN OF VEGETATIVE BUFFER STRIPS FOR RUNOFF AND SEDIMENT CONTROL

Stanley L. Wong and Richard H. McCuen

INTRODUCTION

Structural measures, such as detention basins, are often used for the manage-

ment of stormwater -runoff; they are used to reduce both the volume and the Tate

of runoff that enters receiving streams. While detention basins provide storage

of runoff, they are costly, require special design techniques, and land must be

designated for the single-purpose usage of ponding water. A more economical

method of managing runoff is with vegetative controls, which usually consists

of dense turf judiciously placed across the path of surface runoff. Vegetative

buffer strips have other inherently favorable aspects; in addition to being

ecominical, they are a multi-purpose control method. When properly used, they

serve as a protective cover in reducing the potential of erosion from exposed

soils while promoting conservation practices. This multi-purpose control method

has aesthetic and recreational value. The savings in both cost and time during

both the design and construction phases are additional incentives.

Buffer strips have not received the attention they deserve as a means of

controlling stormwater runoff because they have not been viewed as a viable

alternative to the detention basin. Design methods have not been developed that

can account for the effects of buffer strips on runoff volumes and sediment load-

ings. However, it seems reasonable that the total volume of detention storage

that is required to mitigate the effects of development could be reduced if

properly designed and located buffer strips were used to reduce runoff volumes

and sediment loadings into the detention facility.

IGraduate Research Assistant and Professor, Dept. of Civil Engineering,
41
University of Maryland, College Park, MD 20742.

_J1_

The design of a buffer strip for controlling runoff and sediment will be

a function of both stormwater management policy considerations and physical

characteristics of the site, For example, the required length of the buffer

strip will depend on the desired sediment trap efficiency, which is usually

established by policy. The soil-cover complex at the site will control the

infiltration and thus the ability of the buffer strip to control runoff rates.

The objective of this study was to develop a method for sizing buffer strips

and computing the reduction in the volume of direct runoff.

LITERATURE REVIEW

A number of studies have been undertaken to examine the effectiveness of

buffer strips in removing sediment and other pollutants. However, those methods

that presented results of field studies often did not attempt to develop design

criteria.

The size distribution of soil particles is important in determining the

amount of sediment that is deposited within the buffer strip and the portion

of material that remains in the runoff (Meyer, 1976). Neibling and Alberts

(1979) measured the amount and the particle size distribution of naturally

eroded sediment retained by various length of vegetative filters under broad

sheet flow conditions. The sod buffer strips achieved reductions of 56, 70,

94, and 95 percent for lengths of 2, 4, 8, and 16 feet, respectively, for

particle sizes that are greater than 0.02 mm. The results suggest an effective

removal of medium silt sized sediment when the ratio of the slope length of the

contributing area to the buffer strip length is 10 percent.

Wilson (1967) studied sediment removal from runoff using grass filtration.

The development of an economical and efficient method of improving the quality

of water required for artificial recharge stimulated the research. Several field

experiments were conducted using various lengths of Bermuda grasses. A maximum

percentage of sand, silt, and clays were trapped at buffer lengths of 10, SO,

-J2-

and 400 feet, respectively. The Bermuda grasses (coastal and common) were found

to be the more efficient variety of vegetative filter as a result of a higher

roughness coefficient. A sediment trap efficiency of 95 percent for a 700 foot

buffer strip with very flat slopes and a flow rate of 0.011 cfs per foot of

width of buffer strip. The suggested criteria for selecting the grass filter

media are: 1) a deep root system, 2) dense, well branched top growth,

3) resistance to flooding and droughts, and 4) an ability to recover growth

after sediment inundation.

Vegetative filters may also enhance the water quality of streams with the

treatment of runoff from livestock feedlots, which are characteristically high

in both sediment concentrations and nutrients. Vanderholm and Dickey (1978)

observed reductions of nutrients, solids, and oxygen-demanding materials on the

order of 80 percent on a concentration basis and 95 percent on a mass-balance

basis. The results were obtained by monitoring four field systems over a period

of two years. Based on the recommendation for a minimum contact time of 2 hours,

the design criteria for overland-flow filter lengths ranged from 300 feet for

slope of 0.5 percent to 860 feet for a slope of 4 percent. The contact time is

a function of the flow length, velocity of flow, slope, and type of vegetation.

Young et al. (1979) reported an average reduction of 92 percent of the

sediment and 80 percent of the volume of runoff from vegetative buffer strips

80 feet in length. The experimentation was conducted using a rainulator to

simulate rainfall over an active feedlot and to induce runoff and erosion. The

authors concluded that vegetative filters provide sufficient treatment of feed-

lot runoff.

Considerable research has also been conducted in the area of controlling

erosion from highway construction activities. Many structural and vegetative

control methods are available to limit erosion and sedimentation during and

after construction. Vegetative controls are used quite extensively as a soil

-J3-

stabilizer and as a sediment filter. Ohlander (1976) used field data to develop

a regression equation to estimate the amount of sediment trapped by a vegetative

buffer located below a road drainage outlet. The standard buffer length cap-

able of trapping I ton of sediment per year was given by:

L = (16.16 + 17.69 K + 1.34 S) ( 31 CN (1)
6900 - 69 CO

in which L is the standard buffer length as a slope distance in meters, K is

the soil erodibility index from the Universal Soil Loss Equation, S is the

slope in percent, and CN is the SCS runoff curve number. The ratio of the

actual buffer length to the standard buffer length times I ton yields the an-

nual amount of sediment trapped. Eq. 1 has not been experimentally verified;

however, it does provide a means of assessing alternative planning and manage-

ment schemes.

Recognizing that field studies could not be totally controlled experiments,

Tollner, et al. (1976) conducted experiments using a laboratory flume to simu-

late the flow of sediment ladened water through an artificial rigid grass

media. Under laboratory conditions, a range of values for the independent

variables were obtained by changing experimental conditions. The dependent

variable of interest was the outflow sediment concentration. A model was

developed to estimate the fraction of sediment trapped using transformations

and performing a regression analysis. The probability of a particle being

trapped was related to both the level of turbulence and the potential number

of times in which the particle came into contact with the bottom while being

transported. The fraction of sediment trapped was given by:
TE = exp 1.05 x 10- V MRs\ 0.82. V SLT 0.911 (2)
V VMdf

-J4-

in which TE is the trap efficiency, V m is the mean flow velocity in feet per
second, Rs is the spacing hydraulic radius in feet, v is the kinematic viscosity
in square feet per second, v s is the setting velocity in feet per second, L T
is the section length in feet, and d f is the depth of flow in feet. A cor-

relation coefficient of 0.87 was reported for the data set. The authors found

the mean velocity to be the most influential parameter that effects the sediment

trapped. Eq. 2 relates the important physical parameters in a quantitative manner

and can serve as a useful design tool when all of the variables can be quantified.

DESIGN OF VEGETATIVE BUFFER STRIPS FOR SEDIMENT CONTROL

The sediment yield from a watershed is a nonpoint source pollution problem

that is not easily remedied. Improper management of highway construction or

land development, cultivation of agricultural fields, and strip mining, for

example, contribute to potential erosion problems by exposing the soil to the

weathering processes of nature. Eroded soil that result from these disturbances

often disrupt the delicate ecological balances of receiving waters through the

deposition of sediment.

The mathematical model of Eq. 2 includes variablesthat reflect the important

design factors for vegetative buffer strips, including soil characteristics,

cover complex characteristics, and runoff characteristics. The velocity of run-

off depends on the roughness of the vegetal cover and can be computed using a

modified Manning's equation:

V 1.49 R 2/3 S 112 (3)
m n s

in which V mis the mean flow velocity in feet per second, n is Manning's roughness
coefficient, R S is the spacing hydraulic radius in feet, and S is the buffer
strip slope in feet/feet. Recommended values of n are 0.20, 0.35, and 0.80 for

light turf, dense turf, and conifer/deciduous forest with dense grass understory

respectively; the values assume the grasses are in good hydrologic condition.
-is_

The characteristics of the gr;iss media may be described by the spzicing

hydraulic radius, R Hiis parameter reflects the arrangement of the grass

media spacing in combination with the flow depth and is given as (Tollner, et al.

1976):

SSdf
Rs 2-df-+ S (4)

in which RS is the spacing hydraulic radius in feet, S s is the grass media spacing
in feet, and d fis the fl ow depth in feet. The flow depth of runoff should be

restricted to a height,less than the top of the grass media since the develop-

ment of Eq. 2 applies to the case of nonsubmerged flow. This restriction does

not limit the applicability of Eq. 2 since little sediment is trapped if the

buffer strip were allowed to become flooded.

The trap efficiency is related to the mean settling velocity of the suspended

sediment particles. The sediment particles settle at a rate such that the weight

of the particle exceeds or equals the resistance of its downward movement. The

settling velocity of spherical grains under laminar conditions may be computed

from Stokes law:

vs = (s-l)@ g D
18,v

in which vs is the mean settling velocity in feet per second, s is the specific
gravity of the sediment particle with respect to the fluid, g is the acceleration
of gravity or 32.2 feet per sec 2, v is the kinematic viscosity of water in feet 2

per second, and D is the particle size in feet.

The solution of Eq. 2 for computing the sediment trap efficiency of a vege-

tative buffer strip can be represented graphically. Figure 1 shows the re-

lationship between trap efficiency and the length and slope of the buffer strip,

as well as the roughness coefficient of the vegetation. The required length of

a buffer strip is very sensitive to variation in the trap efficiency as it approaches

-J6-

10-
9- n:0.80 - 1500
Lu 8- n:0.35 - 1400
7- n:0.20

6-
C0 -1300
5-

4-
3- -1200

-M
-1100 In
TR: 9 9 % a
M
0-- TR:9 5 % C)
- looo j.
<
M

- 900 W

In
In
-800 in

- 700 f

-600 r-
M
TR: 9 0 % z

400

TS:85%
300
TR:80%

200

T;I: 7 5 % - 100

0.313 0.67 1.00 1.33
RUNOFF VE.LOCITY Ot/sec)
FIGLME 1. Effective Buffer Lenth Determination For Trap Efficiencies (Tp)
of 75, 95, and 01 percent

100 percent, indicating that a small incremental increase in the trap efficiency

rC(jUires a considerable addition in the buffer length. The curves also suggest

that a significant trap efficiency (up to 75 percent) may be achieved at rel-

atively short buffer lengths. Figure 1 assumes a coarse silt with a mean

settling velocity of 0.002 feet per second through a buffer strip with an

average spacing hydraulic radius of 0.010 feet. Curves similar to Fig. I are

easily derived for other soil textures.

The trap efficiency for other soil textures may also be determined using

Fig. 1. The settling velocity of sediment particles manifests the appropriate

trap efficiencies that are attainable using buffer strips for a particular

particle size. In general, the greater the settling velocity, the higher the

trap efficiency per length of buffer strip. For example, the ratio of the

settling velocities for a coarse silt and a fine silt is 4.9, the buffer strip

length obtained from Fig. 1 should be multiplied by this ratio to obtain the

buffer strip length for a fine silt. This would provide the same trap efficiency

indicated on Fig. 1. The settling velocity ratio of coarse silt to medium silt,

fine sands, and medium sands are 1.3, 0.02, and 0.005, respectively.

THE EFFECT OF VEGETATIVE BUFFER STRIPS ON RUNOFF QUANTITY

While vegetative buffer strips are designed primarily to control sediment,

they also reduce the volume of runoff that results from excess rainfall. A

reduction in the runoff volume occurs as the vegetation impedes and retards

the flow of water, allowing a portion of it to infiltrate into the soil. The

rate of infiltration is a function of: 1) the condition of the vegetative cover,

2) the properties of the underlying soil, 3) the rainfall intensity, and

4) antecedent soil conditions. These factors act in an interrelated manner to

influence the amount of water that infiltrates into a buffer strip.

_j8_

The infiltration of suz,face water Js a complex process that is time variant.

The rate at which a soil may absorb water will vary with time, decreasing in

an exponential manner until a final infiltration rate is achieved. Most methods

of estimating infiltration are based on empirical formulas that represent the

results of field observation. Other methods are based on theoretical solutions

of equations for porous media flow; however, many approximations have been

developed.

In addition to soil characteristics, the volume of water that infiltrates

is also a function of the physical characteristics of the buffer strip. The

ability of the vegetation to retard flow has the effect of decreasing the veloc-

ity of the runoff, which increases the detention time of the overland flow.

The increased detention time increases the opportunity for infiltration. Dif-

ferences in flow velocities can be significant for different types of vegetative

cover (i.e., different values of n).

Recognizing that the infiltration losses may be small, the estimated re-

duction of the surface runoff from a vegetative buffer strip may be computed

using the simplified model:

VL = fctcL (6)

in which V Lis the volume of infiltrated water per width of buffer strip in
cubic feet per foot, fc is the equilibrium or minimum infiltration rate in feet
per hour, tc is the overland flow travel time over buffer strip in hours, and

L is the length of buffer strip in feet. Values of the minimum infiltration

rate are given in Table 1 for various texture classes and soil types, where the

soil types are classified using the SCS hydr6logic soil groups A, B, C, and D;
the part of Table I used to estimate fc will depend on the type of soil data

that is available. The solution to Eq. 6 is provided in Figs. 2, 3, and 4 for

the vegetative covers of light grasses, dense grasses, and conifer/deciduous

forest with dense grass understory respectively. The reduction in

-J9_

TA B 1, 1:1 1NU n imum lilt" i I t vat ioll Rates (.f foi@ Nine Soi I Texture Classes
and Four SCS Soil Groups

Soil Texture f c(in/hr) SCS Soil Group f c(in/hr)

Sand 4.64 A 0.43

Loamy Sand 1.18 B 0.26
Sandy Loam 0.43 C 0.13

Silt Loam 0.26 D 0.03

Loam 0.13

Sandy Clay 0.06
Loam

Clay Loam 0.04
Silty Clay Loam 0.04
Sandy Clay 0.03
Silty Clay 0.02
Clay 0.01

_J10-

600
to 1w
co co
500-
n:0.20

(11oht turf)
400- SLOPE
z
w
-j 300-
Ix
w 0
U. r-
u- 200- c
D x
In m
100----- 0
SOIL GROUP In
- 0.18 M
0- A c
z
- 0.16 0
In
In

- 0.14
B In

- 0.12
>
--4
m
0.10

- 0.08

c
In
- 0.06 'n
M
M

0.04

0.02
D

10 20 30 40
TRAVEL TIME (minutes)

FIGURE 2. Determination of Volume of Infiltrated Runoff From
Vegetative Buffer Strip (Light Turf)

-J11-

Goo

a 4b WO CIV
9 % C@. C;. C@*
500-

400- n:0.35

z (dense turf)
w
-j 300-
0
r-

U. 200- x
U. m

0
SOIL GROUP n
0.18 M
0- A z
0
- 0.16 "n

- 0.14 n
P

- 0.12 M
>
-4
m
- 0.1 a

- 0.08

c
n
- 0.06 'n
m

0 . 04 co

.02
D

0.0
10 20 30 40
TRAVEL TIME (minutes)

FIGURE 3. Determination of Volume of Infiltrated Runoff From
Vegetation Buffer Strip (Dense Turf)

-J12- k

600
ale af
CO 10 Cill
C-)
500- C@ co
CO

F- 400-
0
z
W
300-

cc
W n:0.80
u- 200-
U-
D <
c o n i f e rd e c 1 d u 0 u s0
100- r
forest with dense grass)

M
SOIL GROUP 0
In
A - 0.18 M
c
z
- 0.16 0
,n
B - 0.14 i;
,n
p
- 0.12 -4

- 0.10 M
a

ca
- 0.08
c

- 0.06

M
M
- 0.04 o
-4
:0

- 0.02
D

1 40
10 20 30
TRAVEL TIME (minutes)

FIGURE 4

-J13-

the Vol I I 111c of, 1.11 n o f t' is ex I) ress ed a s inch(-.,s of in t' i It i-a t ed wat ci- I) ei- tit ii t. a rea

of huffci- st t-ip.

The use of rigs. 2, 3, and 4 for computing losses from vegetative buffer
strips provides a means of assessing its effectiveness as a stormwater management
control measure in reducing the runoff volume. The procedure is a conservative
approach because the condition of minimum infiltration rate is assumed.

APPLICATION OF THE METHOD

The area of study in which the design procedures are illustrated is a 47.7

acre watershed in Montgomery County, Maryland. The watershed area has five

distinct subwatersheds; a detention basin is located at the lower end of the

watershed. Fig. 5 depicts a schematic of the watershed and summarizes the

characteristics of the subareas. The land use consists of grass, roadways,

rooftops, and exposed soils. The soils are predominantly Chester silt loams

that are well drained and moderately erodible.

The average annual watershed erosion is estimated using the Universal Soil

Loss Equation. The average watershed erosion may be determined from the empiri-

cal formula:

A = R-K-LS-VM (7)

in which A is the soil loss in tons per acre per year, R is the rainfall factor,

K is the soil erodibility factor, LS is the slope-length factor, and VM is the

vegetatibe-erosion control factor. The computation of the average annual water-

shed erosion is shown in Table 2. The total estimated annual erosion is 49.14

tons/year.

The area adjacent to the upstream segment of the detention basin may be

utilized to establish a buffer strip, which under other circumstances might be

either bare soil or grass in poor condition. Fig. 3 was used to obtain the

-J14-

SECTION
SECTION 3
1
AREA = 4.13acros
AREA = $.$$acres
SLOPE = 2.6%
SLOPE = 3.0%
LENGTH = 730feet
LENGTH = 670feet IMPERVIOUS AREA =49

IMPERVIOUS AREA

37%

SECTION SECTION1 I JISECTIJON
2 4 11 1 5
AREA = 13.04 AREA = I I AREA
a c r e s 10.20 7.71
SLOPE = 4.0% acres acres
LENGTH = 9 7 5 1 t SLOPE SLOPE
3.0% 1.9%
IMPERVIOUS LENGTH 9 LENGTH
w AREA = 0% = a 7 0 = 860
I :at e e I
IMP R- IMPER-
VIOUS VIOUS
AREA=i AREA
27%

BUFFER STRIP
AREA

DETENTION
BASIN
(390 x 235ft)

FIGURE S. Conceptualization of Study Watershed and Subarea Characteristics

-Jl 5-

TABLE 2. Estimation of the Average Annual Erosion From the Study Watershed*

Percent of
Previous Area VM
Section Area Slope Length Covered Vegetative LS A Total Soil Loss
No. (acres) (ft/ft) (ft) With Grass Factor Factor (tons/acre/yr) (tons/y-r)

1 3.33 0.030 570 85 0.03 0.48 0.81 2.68

2 13.04 0.040 975 75 0.04 0.97 2.17 28.33

3 4.13 0.026 730 80 0.04 0.45 1.01 4.16

4 10.20 0.030 870 85 0.03 0.55 0.92 9.43
L S 7.71 0.019 860 85 0.03 0.35 0.59 4.53
CA
I
Average Annual Soil Loss = 49-14 tons/yr
= 1.03 ton/acre/yr

*R 175; K 0.32

ti'ap efficiency foi, variotis ICngthS Of I)Uffel' StI-ij)S tIVII could be located im-

mediately upstream from the detention facility. 'Fable 3 lists the potential

trap efficiencies of various vegetative buffer length when implemented within

the available area. The corresponding reduction in the estimated average annual

sediment loading is also given in 'Fable 3. Table 4 is a summary of observed

sediment volumes at the inlet of the existing detention basin. The 90 percent

sediment trap efficiency from a 150 foot length of buffer strip will effectively

reduce the observed volume of sediment entering the detention basin by the amount

recorded in Table 4.

The disparity between the total estimated annual and the observed sediment

volumes is due to the degree of development that the watershed was undergoing

when the observed sediment volumes were measured. During the period of the ten

storm events, the SCS curve numbers ranged from 77 to 90, 70 to 81, 79 to 88,

79 to 89, and 8S to 94 for sections I through 5, respectively. The computed

soil loss, using Eq. 7, does not reflect the transition period of the watershed

during the development stages. The computations in Table 2 represent the erosion

in the post-development stage.

A reduction in the volume of runoff depends upon the overland flow travel

time, the minimum infiltration rate, and the length of available buffer strip.

The volume of runoff infiltrated by a buffer strip may be determined from Figs.

2 and 3, which is given in inches of losses per area of buffer strip. The

available buffer strip located upstream of the detention basin has dimensions

of a 150 foot length and a 400 foot width. The infiltration losses over a

buffer strip having 3 percent slope and B type soil will be 0.041 inches,

from Fig. 2. This reduction in runoff results as the flow travels the distance

of 150 feet within a 9 minute travel time. Thus, the total volume of infiltrated

flow for the duration of the runoff is equal to the product of the area of buffer

strip and the duration divided by the travel time over the buffer strip. A

volume of 20S cubic feet of runoff is infiltrated over the buffer strip in 9

-J17-

TABLE 3. Potential Trap Fff @ciency Attainable and Reduction in
E'stimated Sedinient Using Vegetative BLiffor Strip.,

Available Reduction in Sediment
Buffer Strip Potential Trap Using Buffer Strip*
Length (ft) Efficiency (tons/year)

so 7S 36.8S

100 S2 40.219

150 9 44.221

200 93 45.69

2SO 94 46.18

300 95 46.67

*a slope of 3.0%, light turf (n 0.20)

-J18-

TABLE 4. Summary of Measured Sediment Volumes for Ten Storm Events and the
Potential Sediment Reduction from a Vegetative Buffer Strip

Potential Sediment
Measured Watershed Reduction from
Sediment Volume Sediment Loading 150 foot Buffer Strip
-Date (tons) (tons/acre) (tons)

6/28/77 24.83 0.52 22.35

7/12/77 8.09 0.17 7.28

7/17/77 79.24 1.66 71.32

7/20/77 0.20 0.01 0.18

8/1/77 21.23 0.45 19.11

8/5/77 0.26 0.01 0.23

8/8/77 2.85 0.06 2.57

8/10/77 1.74 0.04 1.57

8/14/77 31.45 0.66 28.31

_J19-

minutes. Table 5 summarizes the reductions in volumes that would have occurred

for the ten actual storm events for which data were measured. The percent re-

duction of runoff into the detention basin is greatest for the smallest storm

event and, in general, the total volume loss will increase when the duration

of the runoff is longer.

CONCLUSIONS

Vegetative buffer strips can be used to control both the quality and quantity

of surface runoff. Because the quantity of runoff intercepted by the buffer strips

will be minimal,the design is most often based on quality control criteria,

specifically the sediment trap efficiency. In actual practice, the sediment

trap efficiency would be set by a stormwater management policy or regulation.

Thus, the designer would determine the buffer strip length that is required to

trap the required fraction of the sediment in.the inflow. Once the length of

the buffer strip has been determined for quality control, the effect on the run-

off volume can be determined. The depth given in area-inches by Figs. 2 and 3

can be converted to a total volume by multiplying by the area of the buffer strip.

For most design problems, buffer strips will not be capable of controlling

sediment and runoff volumes by themselves. Thus, they will have to be used as

one component of an integrated stormwater control system; that is, they may be

used in either series or parallel with other control methods to achieve a desired

level of control at a site. In order to be effective, the characteristics of

buffer strips must be considered in the site location plan. Thedesign curves

provided herein assume nonsubmerged flow. Therefore, flow rates may have to be

controlled by other means to ensure that acceptable flow rates are maintained.

Additionally, if sediment volumes in the inflow are excessive, considerable

volumes of sediment may be deposited at the upper end of the buffer strip. With

time, a berm may form, and the borm may be washed@out when an extreme rainfall

-J20-

TABLE S. Summary of Observed Volumes and the Potential Reduction
of Flow Entering Detention Basiii

Total
Observed Flow Duration of Volume of Flow Percent
Date Volume (ft3) Runoff (min) Infiltrated (ft3 Reduction

6/28/77 33,890 240 S,460 16.1

7/12/77 24,4921 180 4,100 16.7

7/17/77 77,441 47.5 1,082 1.4

7/20/77 75896 240 5,467 69.2

8/1/77 42,373 117.5 2,676 6.3

8/5/77 320 ISO 320 100.0

8/8/77 13@485 180 4,100 30.4

8/10/77 6,831 120 2,733 40.0

8/14/77 60,847 117.5 2,676 4.4

*for a Buffer Strip having dimensions of 150 ft x 400 ft, 3% slope, n = 0.2,
Hydrologic Soil Group B.

-J21-

event occurs. Thus, it is important either to prevent the berm from forming

or to provide proper maintenance. It is possible to prevent the berm from

forming by trapping extremely large volumes of sediment by other methods.

The processes of filtration and deposition within a grass buffer strip

will also lessen the loadings of any nutrients and pesticides that are carried

with the runoff by trapping those sediment particles which transport them.

Additional quality control will result if significant volumes of water are

encouraged to infiltrate. Infiltration of water containing soluable pollutants

will reduce the loadings of such pollutants that reach the stream with the direct

runoff. Infiltration can be encouraged by reducing the slopes of the area al-

located to buffer strips or increasing the surface roughness. If the natural

soil is not highly porous, the buffer strip may be placed over an infiltration

bed that uses a soil that has a high infiltration rate. Significantly larger

volumes of runoff, and thus soluable pollutants, will be removed from the direct

runoff.

The design of effective vegetative buffer strips to properly manage runoff

requires a strategy of evaluating performance criteria. A design method was

provided herein which can be used to evaluate both the sediment trap efficiency

and the reduction of runoff volumes due to the infiltration of overland flow

from vegetative buffer strips. The design method uses soil characteristics,

site characteristics, and policy information, all of which are important in

design. Vegetative buffer strips are recommended for those areas that would

otherwise be. exposed soil.

-J22-

REFFRENCES

I .Meyer, L.D. and Ports, M.A., "Prediction and Coiarol of Orhan E'rosion and
Sedimentation,'' National Symposium on Urban Hydrology, Hydraulics, and
Sediment Control, University of Kentucky, 1976.

2. Neibling, W.H. and Alberts, F.E., "Composition and Yield of Soil Particles
Transported Through Sod Strips," A.S.A.E. Paper No. 79-2065. Presented
at the A.S.A.E. and C.S.A.E. 1979 Summer Meeting, Winnipeg, Canada, 1979.

3. Ohlander, C.A., "Defining the Sediment Trapping Characteristics of a
Vegetative Buffer, Special Case: Road Erosion," Proceedings of the Third
Federal Inter-Agency Sedimentation Conference, Water Resources Council,
Washington, D.C., 1976.

4. Tollner, E.W., Barfield, B.J., Haan, C.T., and Kao, T.Y., "Suspended
Sediment Filtration Capacity of Simulated Vegetation," Transactions of
the A.S.A.E., 19(11), pp. 678-682, 1976.

S. Vanderholm, D.H. and Dickey, E.C., "Design of Vegetative Filters for
Feedlot Runoff Treatment in Humid Areas," A.S.A.E. Paper No. 78-2S70.
Presented at the A.S.A.E. 1978 Winter Meeting, Chicago, Illinois, 1978.

6. Wilson, L.G., "Sediment Removal from Flood Water by Grass Filtration,"
Transactions A.S.A.E. 10(11), pp. 35-37, 1967.

7. Young, R.A., Huntrods, T., Anderson, W., "Effectiveness of Nonstructural
Feedlot Discharge Control Practices," A.S.A.E. Paper No. 78-2572.
Presented at the A.S.A.E. 1978 Winter Meeting, Chicago, Illinois, 1978.

-J23-

APPENDIX K

DESIGN OF INFILTRATION TRENCHES

FOR CONTROL OF STORMWATER RUNOFF

Stanley L. Wong

Richard H. McCuen

DESIGN OF INFILTRATION TRENCHES FOR CONTROL OF STORMWATER RUNOFF

Stanley L. Wong and Richard H. McCuen

INTRODUCTION

In the past, urbanization has usually begun with the installation of storm

sewers for controlling the hydrologic effects of increased imperviousness in

developing communities. After considerable development has taken place, often

with little or no advanced planning, it was recognized that storm sewer sys-

tems are inadequate by themselves. Thus, it became necessary to incorporate

other structural measures, such as detention ponds, to control the increased

volume and rate of runoff. While the stormwater detention provided control of

runoff rates, its effect on the volume and duration of storm runoff has been

shown to be less than acceptable in many cases (McCuen, 1979). Additionally,

these structural control methods fail to reverse the trend towards reduced

groundwater recharge that accompanies urbanization and they require expensive land.

Thus, there is considerable interest in stormwater management methods that pro-

vide for more "natural" control, especially vegetative and nonstructural control

methods. While structural controls can probably not be eliminated entirely, the

use of these measures mav often be improved by combining them with methods that

utilize the process of induced infiltration.

Infiltration trenches are also applicable in agricultural areas. Although

infiltration rates of cultivated land are often thought to be high, a combination

of various agricultural practices may result in large quantities of runoff.

For example, the seasonal transitions of the crop growth cycle effects the volume

Graduate Research Assistant and Professor, Department of Civil Engineering,
University of Maryland, College Park, MD 20742.

-K1-

of flows, with Jarger volumes of runoff occurring during periods of fallow land-

use. Poor conservation practices coupled with soil textures having a high run-

off potential frequently create runoff problems that require control measures.

The control measures that reduce the volume of-runoff from agricultural fields

have an additional benefit of reducing pollutant loadings that often accompany

flows to receiving streams.

An infiltration trench,is a control method that enhances the infiltration of

stormwater into groundwater flow systems. Past research has not provided a

procedure for designing infiltration trenches. This method of inducing infil-

tration is analyzed herein, with an emphasis upon the design criteria for esti-

mating the dimensions of a trench that could control a portion of the volume of

excess direct runoff.

CONSIDERATIONS IN DESIGN AND CONSTRUCTION

An infiltration trench may be described as a structural device for inducing

infiltration into the subsurface soils, thus replenishing groundwater supplies.

However, the primary goal of such a stormwater management device is more often

used to aid in the control of surface flows, rather than to serve principally

as an artificial recharge device. Trenches are excavated and filled with crushed

stone, which stabilizes the side walls and provides a significant increase in

the storage capacity of the subsurface area; the trench can also serve to filter

sediment from the runoff. The volume of void spaces tends to clog with time;

hence, a filter cloth can be placed over the stone and beneath a topsoil back-

fill. The filter cloth should prevent clogging and, thus, increase the design

life of the infiltration trench. The infiltration characteristics of either the

filter cloth or the topsoil backfill can be the limiting design factor; thus, it

is important to consider their characteristics in design. The topsoil over-

burden will prevent damage to the filter cloth and will also serve to filter

-K2-

pollutants included in the stormwater runoff; such pollutants could clog the

void spaces in the bottom and sidewalls of the trench. The configuration of

an infiltration trench is shown in Fig. 1.

Sizing of Trench Dimensions

Runoff from both impervious surfaces and surfaces with low infiltration

capacities may be intercepted by infiltration trenches, which are preferably

located in low sloped areas. The temporary storage and eventual percolation

of storm runoff into the soil is the primary purpose of an infiltration trench.

The desired dimensions of an infiltration trench will depend upon the volume

of direct runoff that requires control and the characteristics of the water-

shed and soils. In general, the design of a trench must consider the limiting

effects of the storage volume and the infiltration characteristics of both

the topsoil backfill and the soil surrounding the infiltration trench. Quite

often, the design assumes that the infiltration capacity of the surrounding

soil is very small; this is a reasonable assumption because it would not be

practical to install infiltration trenches in areas with highly porous soils.

The volume of available storage within an infiltration trench, which is

the primary design variable, is a direct function of the porosity of the crushed

stone or gravel fill material, The porosity, defined as the ratio of the vol-

ume of voids to the total volume of the infiltration trench, may be expressed

as a percentage of the total volume; it is also called the percent voids. The

void spaces are filled with water where available storage exists to detain

surface flows. The maximum volume of storage occurs when the crushed stone is

saturated, so that all available voids are occupied.

Various combinations of depths and widths of infiltration trenches may be

used to achieve the necessary cross-sectional area of storage. The total volume

of storage is computed by:

-K3-

BERM VEGETATIVE COVER (sod) BUFFER 41
AREA
L

TOPSOIL DIRECTION OF FLOW
BACKFILL

FILTER CLOTH

C:@

CRUSHED STONE
AGGREGATE
(D.C
Dx
C) UNDISTURBED
SURROUNDING SOIL

FIGURE 1. Infiltration Trench Configuration for Controlling Storm Runoff

- K4-

V ndwL

3
in which Vs is the volume of available storage in ft , d is the depth of gravel
fill in ft., w and L are the width and length of the trench in ft., respectively,

and n is the porosity of gravel fill in dimensionless units. The dimensions

of the trench are shown in Fig. 1.

The gravel backfill material is typically coarse aggregate from 3/8-inch

to 1-12-inch in size. The percent voids range in value from about 30 to 40 per-

cent at dense compaction and loose compaction, respectively, for a gravel mix-

ture from the Mid-Atlantic coastal plain (Wills, 1967). The void content of

coarse aggregates may differ between gravels obtained from various regions

throughout the country. Particular specifications of gravels are available

upon request when purchasing fill material.

Storage of the direct surface runoff within the infiltration trench is

accomplished by sizing the dimensions to accommodate the volume of water infil-

trated from the surface flows. By simply equating the volume of storage with

the volume of available surface runoff, the dimensions of the trench may be

determined. The product of the depth of direct runoff and the upland area are

equated with Eq. I to determine the ratio of the upland area and the length

of the infiltration trench:

A ndw (2)
L Q

in which A is the upland drainage area in sq. ft., and is the depth of the

direct surface runoff in feet. The required length of the infiltration trench

is computed from either Eq. 2 or Fig. 2 for a given upland area. Fig. 2 is a

graphical solution of Eq. 2, with curves that represent different trench cross-

sectional areas and a void content of 35 percent for the gravel fill. The

procedure of computing the volume of direct runoff is often dictated by the
regulatory policy for the state of local jurisdiction.

- K5-

250

4J
200

1501

Cd r-
Q)

100--
0
Cd -4
-4 +j Cross Section Area of Trench
$a4 M

4-) 4.4
r-
4-4
0 4-4 SO.-
0 0
.H
4J
M

0.1
0 0.2 0.4 0.6 0.8 1.0 1.2 11-1. 41-, 1.6 1.8 2.0
Direct Runoff Depth from bpland@ -Area (in.)

FIGURE 2. Sizing of Infiltration Trenches to Control Direct Runoff from Upland Areas
(coarse aggregate fill poroslity equal to 0.35)
\J,

-K6-

Infiltration of Surface Flow as a Design Constraint

While the volume of void space within the infiltration trench is a limiting

factor, the infiltration rate through the topsoil overburden and filter cloth

must also be considered in the design of the infiltration trench dimensions.

The stormwater must infiltrate thruugh the topsoil overburden and into the

trench prior to percolation into the surrounding soils. The total volume that

enters an infiltration trench will depend on the infiltration capacity of the

overlying topsoil backfill. The topsoil material should be selected such that

the infiltration rate is high so storm runoff is more easily infiltrated into

the trench. A sandy textured soil is recommended to fulfill the requirement

of a high infiltration rate; a sandy soil is also capable of sustaining vege-

tative growth. A dense turf cover over the'infiltration trench provides a leaf

canopy that protects against the sealing of pores by sediments created upon

rainfall impact. Also, vegetative root systems penetrate the soil and increase

the infiltrating capacities by furnishing small channels between soil pores.

The volume of stormwater entering the infiltration trench may be estimated

if the minimum infiltration capacity of the topsoil overburden is known. The

volume of infiltrated water, VI (inches),may be estimated by:

VI = f LwtD (3)

in which fc is the minimum infiltration rate of the topsoil overburden in inches

per hour, L is the length of the infiltration trench in ft. (determined from

Eq. 2), w is the width of the infiltration trench in ft., and tD is the duration

of the direct runoff in hours. The duration of direct runoff may be approximated

as twice the time of concentration of the flow from the upland contributing area.

Solutions to Eq. 3 are given graphically in Fig. 3. The design curve chosen will

depend upon the type of soil information that is available; curves are provided

in Fig. 3 for different soil textures and for the four SCS soil groups,

-K7-

2.2

2.0.-

1.8- Soil Texture: Sand

4J
-4

P, 4.4
- Loamy Sand
4-4 =
4-4 -4
0
z

Cd
"0 4-4
W $-, SCS Hydrologic
-W :@ 1.0
Soil Group
4-)
U
.'-q r-
4-4 Q) 0.8 Loam
FF-, 1-1@ A
4-4 Silt Loam
0 0
Q) 0.6 Sandy Loam
0
> 0.4

0.2
C

0 10 20 30 40 so 60 70 80 90

Duration of Direct Runoff, tD (min)

FIGURE 3. Volume of Infiltrated Runoff into Infiltration Trench for
Various Soil Textures and SCS Hydrologic Soil Groups

-K8-

The minimum, or equilibrium, infiltration capacity was selected for Eq. 3;
this will result in a conservative design. Estimates of f c are based upon the
final constant rate of infiltration within the upper soil horizon after pro-
longed wetting of the soil. The value of f c will depend on the soil type,

antecedent soil moisture, and the duration of the rainfall event. Hence, the

minimum infiltration rate represents a long-term steady-state condition that

neglects the early decay portion of infiltration capacity curve and should

correspond to the saturated hydraulic conductivity of the soil. Additionally,

the saturation of the topsoil overburden is assumed to have occurred prior to

the interception of direct runoff by the infiltration trench.

Strategies for Increasing the Infiltration Potential

The total volume of surface runoff intercepted by an infiltration trench

is constrained by the volume of voids within the trench, the infiltration rate

of the topsoil overburden, and the surface area above the trench. The total

volume of infiltrated runoff also depends on the duration of the direct runoff;

the duration of the direct runoff from impervious surfaces is frequently short

due to the low resistance associated with the relatively smooth surfaces that

are associated with developed areas. Because of all of these potential limit-

ing factors, the fraction of the total surface runoff volume that is intercepted

by a trench is often small. Therefore, a means of inducing the infiltration

would enhance the efficiency of the trench.

A number of different strategies may be used to effectively increase the

volume of runoff entering the infiltration trench. To be effective, a strategy

must overcome one of the constraints identified, For example, a vegetative buf-

fer strip that is placed between the runoff generating surface and the site of

the infiltration trench would reduce the flow velocity and thus increase the

duration of flow across the surface area above the trench. Similarly, grading

-K9-

of the site so that the surface slope of the trench contact area is minimized

should increase the contact time, and thus, increase the volume of water in-

filtrated. Also, a similar effect will be attained by increasing the surface

roughness of the trench contact area. An additional advantage of installing

a vegetative buffer strip, increasing the surface roughness, or decreasing the

slope is the filtration and removal of sediment particles from the surface

flow (Wong and McCuen, 1981).

The temporary ponding of water with various structural devices may be

employed to prolong the contact time of the runoff. Retaining a portion of

the surface runoff on the surface above the trench will reduce the Volume of

runoff to downstream areas, while allowing for increased infiltration of water

into the trench. A berm may be placed along the edge of the trench when the

overland flow is in the direction across the width of the trench, as shown in

Fig. 1. Berms should be constructed with soils that are not easily eroded and

seeded to prevent washing back into the trench. The required height of the berm

will vary with slope. The volume of storage retained by the berm (V b) will de-
pend on the height of the berm, the surface slope, and the surface area behind

the berm. The volume of infiltration given by Eq. 3 can be modified to include

the volume retained by the berm:

V f LwtD +Vb (4)

Check dams may be installed to detain the water when an infiltration trench is

placed beneath a swale and the flow is directed along the length of the trench.

Check dams may consist of either graded stones, sheet pilings, or staked straw

bales. Both the berm and the check dam serve to reduce the flow velocity, lengthen

the detention time, and thus, increase the volume of infiltrated water.

-K1 0-

Siting of an Infiltration Trench

In addition to the factors previously identified., the effectiveness of an

infiltration trench will depend on siting factors. The selection of a satis-

factory location to implement an infiltration trench as a control measure will

necessitate a site investigation; thus, information obtained pertaining to the

inherent soil characteristics may reveal the suitability of the environment to

accommodate groundwater storage and recharge. Hydrologic, topographic, and

geologic investigations will provide the basis for a decision.

Much of the necessary information may be obtained from existing sources of

data. County soil surveys that identify the soils of the particular county are

available from the Soil Conservation Service. The permeability rate is of

particular interest because it reflects the rate at which the water moves in

the soil. The permeability rate may be measured on-site by various methods

(e.g., percolation test and well pumping) and should be used whenever possible.

The individual soil horizons between the ground surface and the aquifer should

not have any impermeable layers that restrict the downward movement of water.

The depth between the water level in the saturated aquifer and the soil

surface should be determined to ensure that the trench will not become rapidly

inundated due to a rise in the water table elevation. Areas with water table

elevations tnat are seasonally high may require consideration of additional

design requirements. Coastal areas, for example, are frequently faced with

groundwater tables that rise rapidly, which limit the ability of the soil to

adequately drain the stormwater runoff. Circumstances such as this may cause

infiltration trenches to become ineffective in both the storage of stormwater

and the recharge of groundwater. Furthermore, the inundation of infiltration

trenches by a rapidly rising water table may lead to the intrusion of soil

particles into the trench from the surrounding soil, which clog available

void space. Placement of filter cloths along the bottom and sidewalls, may be

necessary to avert soil particle intrusion. However, high costs may prohibit

such a requisite. Water table elevations should be several feet below the

trench bottom to allow for adequate clearance as water levels rise. The

actual distance will depend on soil characteristics.

Considerations in Selecting the Filter Cloth

The implementation of filter cloths as a separating layer between the

topsoil backfill and coarse aggregate fill should not be slighted. These

fabrics are critical for the continued effective functioning of the infiltra-

tion trench. If the installation of a filter cloth is neglected, the topsoil

overburden has no means of being held in place. Hence, the void space in

coarse aggregate fill would quickly clog, with the particles rendering the

trench useless. Filter cloths are available with various specifications for

different criteria. The selection of filter cloths should be based upon:

1) permeability (i.e., size of pore openings), 2) strength of fabric, and

3) percent of open area. Filter cloths are being used more frequently to

stabilize drain systems and care should be taken to ensure that it is not the

limiting permeable material.

Recognizing that filter cloth restricts the movement of the topsoil mate-

rial into the coarse aggregate fill, particles will tend to clog the pore open-

ings within the filter cloth and reduce the permeability rate. This causes

soil particles t6 form a mud-caked layer on top of the filter cloth, which re-

quires periodic maintenance. Depending upon the depth of topsoil backfill,

the accessibility of the filter cloth will vary. Maintenance schedules should

be considered in design when considerations are made for the implementation

of infiltration trenches. Furthermore, the removal of the aggregate fill to

be washed or replaced with new aggregate material may be required; however,

proper maintenance of the filter cloth should make this an infrequent main-
tenance requirement.

- K1 2-

DISCUSSION AND CONCLUSIONS

While surface detention storage appears to be an effective measure for

controlling the increased runoff rates and volumes that accompany development

there is some concern about the need for groundwater recharge and the value

of land used for detention storage. Infiltration trenches provide a means

of simultaneously decreasing surface runoff volumes and increasing groundwater

recharge rates. The effectiveness of infiltration trenches was shown to be

a function of the volume of storage within the trench, the infiltration capacity

of the topsoil overburden, the surface detention time, the proper selection of

filter cloth, and various siting characteristics. In addition to a method for

determining the appropriate dimensions of a trench, various strategies were

identified herein for increasing the effectiveness of the trench.

Integration of Stormwater Management Methods

In most applications, infiltration trenches will not be capable by them-

selves of controlling increases in runoff volumes that accompany urbanization

and land cover conversions in agricultural areas. Therefore, they will probably

be most effective when combined with other stormwater control methods, such as

detention ponds, parking lot and rooftop storage, and vegetative buffer strips.

The design procedure outlined herein can be used to incorporate infiltration

trenches into an integrated stormwater control program.

The use of methods such as infiltration trenches and buffer strips can

reduce the total storage volume required for detention basins. The usual

design procedure involves estimating the required volume of detention storage.

However, with an integrated design program, design curves such as those provided

herein for infiltration trenches could be used to reduce the volume of required

storage. The reduced volume of detention storage that would be required should

increase the flexibility of the stormwater management plan; that is, a smaller

-K1 3-

required -,torage volume may make it possible to locate the detention basin at

a site having characteristics less desirable for either urban or agricultural

land uses. This could increase the return on the development of the land.

Policy Considerations

Stormwater management policies and regulations play a critical role in the

effectiveness of control methods, including infiltration trenches. Policies

and regulations must provide control over the design, installation, and main-

tenance of the trenches. Failure at any of these aspects may lead to an

ineffective infiltration trench. In order for a policy to be effective it

must provide for assurrances that the design will properly coordinate the

storage volume with both soil and site characteristics, as well as the char-

acteristics of the filter cloth. Regulations must be sufficiently specific

to limit the installation of infiltration trench to a period during which up-

land areas are not exposed; otherwise, eroded soil may clog the storage space

within the infiltration trench. For maximum effectiveness, the infiltration

trench should be installed after most of the land development has been completed.

It should also be recognized that a poorly maintained infiltration trench will

not function as the designer intended. Therefore, policies must provide con-

trol of maintenance, and regulations must provide for inspection and enforcement

of the policies. Only when properly designed, installed, and maintained will

an infiltration trench contribute to the control of stormwater runoff.

-K14-

REFERENCES

1. McCuen, R.H., "Downstream Effects of Stormwater Management Basins,"
J. Hydraulics Division, Proc., Amer. Soc. of Civil Engineers, Vol. 105,
No. HYII, Nov., 1979, pp. 1343-1356.

2. Wills, M.-H., "How Aggregate Particle Shape Influences Concrete Mixing
Water Requirement and Strength," Journal of Materials, Vol. 2, No. 4,
Dec. 1967, pp. 843-865.

3. Wong, S.L., and McCuen, R.H., "The Design of Vegetative Buffer Strips
for Runoff and Sediment Control," Technical Report, University of
Maryland, College Park, MD, 1981.

K1 5-

APPENDIX L

THE EFFECT OF LAND USE CHANGE AND

STORMWATER MANAGEMENT

ON FLOW-DURATION CURVES

Stanley L. Wong
Richard H. McCuen
Mark E. Hawley

ITIE EFFECT 01, LAND USE CHANGE AND STORMWATER

MANAGEMENT ON FLOW-DURATTON CURVES

Stanley L. Wong, Richard 11. McCiicn, and Mark E. Hawley

INTRODUCTION

Hydrologists have recognized the effect of land use changes on peak

discharge rates for many years, but the effects on other runoff characteristics

have received much less attention. While the effect on peak discharge is

important, urbanization can also cause significant changes in the velocity

of runoff, the duration of near-peak flows, and the sediment transport capa-

bilities of the runoff. Changes in these characteristics usually lead to

changes in stream morphology due to increased erosion of the stream banks and

deposition of eroded.material at points downstream.

While studies have shown the effects of urbanization and stormwater

management on the magnitude and duration of flooding independently, the inter-

action of the two watershed modifications has not been examined. The inter-

us, -'low can b represented by a
action between the. magnitude . _I dIlration oJ7 J'j )d Is

flow-duration curve. Flow-duration curves for urbanized and non-urbanized water-

sheds were computed for Long [sland, New York (USGS, 1981). The effects of urban-

ization on the runoff characteristics of a watershed can be evaluated by developing

flow-duration curves for various design storm events. These curves are graphs of

the duration of the period during which the discharge rate equals or exceeds a

certain magnitude versus the magnitude of flow. Flow-duration curves Are an

alternative to flood frequency curves; however, they are preferable to the flood

frequency curve beca@ise they provide information about both the duration and

frequency of runoff.

Department of Civil Engineering, University of Maryland, College Park, MD 20742

The objective of' tliis stildY IS to cxalllinc 01c effects of' ul'banization

and stormwater management on flow-duration curves and to discuss the method of

incorporating flow-duration curves into the stormwater management design process.

Alterations in the time characteristics of direct runoff may be readily assessed
by the analysis of flow-duration curves for various watershed conditions. com- 6

parison of flow-duration curves before development and after development may serve

to illustrate the changes in the hydrologic response of the watershed. Flow-

duration curves may also be employed to evaluate the effectiveness of SWM measures

in controlling the increased runoff volume and peak discharge rates. The extent

of channel degradation and changes in stream morphology due to increased duration

of bankfull flows,is also investigated where detention basins and other SWM controls

are to be constructed.

FLOW-DURATION CURVES AND STORMWATER

MANAGEMENT DECISIONMAKING

Flow-duration curves are not commonly used in evaluating the effects of

urbanization due to deficiencies in the existing policies for stormwater manage-

ment (SWM). Most current SWM policies require only control of the peak discharge

rate for a selected recurrence interval, such as the 10-year event. These policies

resulted from the belief that most problems are the result of peak flow rate and

that the peak discharge rate was related to the other runoff characteristics;

from these beliefs it then followed that, if the peak discharge were controlled,

then the problems associated with the other stormwater runoff characteristics would

also be solved. Therefore, most drainage and stormwater management policies

-L2-

developed in the last ten yeat's have reqUil-ed that the peak discharge rates

be controlled so that the post development peak is no greater than the pre-

development peak for a specified Mceedence probability, or return period.

Peak discharge rates are commonly estimated using the rational or SCS 'design

methods, and detention basins are designed to control, the discharge rate.

Although detention basins are qijite effective in controlling peak drainage,

they also affect the duration characteristics of the runoff. Detention basins

provide artificial storage Of runoff to compensate for the loss of natural storage

and increased volume and duration of runoff caused by urbanization. Previous

studies (Kamedulski and McCuen, 1979) have demonstrated that a basin designed to

control the peak discharge rate from a certain design storm may not adequately

control the runoff from other storms. Other studies have shown that current SWM

policies that are concerned only with controlling peak discharge can resillt in

significant increases in the duration of peak and near-peak flows (McCuen, 1979).

Field inspections have demonstrated that these policies are not effective in

controlling sediment problems and can, therefore, lead to significant changes in

stream morphology.

Concern over the detrimental effects of urbanization on infiltration and

stream characteristics has led to development of alternative methods of SWM.

Porous and modular pavements, infiltration trenches, and vegetative buffer strips

are examples of alternative measures that encourage the infiltration of stormwater

on site. These methods provide a more natural type of control because they cause

some of the stormwater volume to enter the soil , rather than flowing directly into

the -stream. Tlitis, Hic inci-cii,;C ill VOILIHIC 01' I'M101T dUC LO Lli-banizatiun is lessened.

-L3-

On some'watersheds, soil types and SUbSM'fICC conditt,iolls, such as im-

permeable layers or high water tables, may limit the potential rate of infiltration,
while on intensively developed wat;rsheds the value of the land may be so great as

to make those methods economically unsuitable. In those cases, alternAtive means

of providing temporary storage may be useful; commonly used methods of providing

this storage are rooftop storage, detention basins, and parking lot ponding

structures. The volume of storage provided by each of these methods serves to

regulate the rate of discharge, but these methods do not lessen the total volume

of runoff to any significant extent. Thus, they do not give adequate consideration

to the duration characteristics of the runoff.

Although these structural methods of controlling discharge rates are useful

in meeting the requirements of SWM policies, they often fail to control the adverse

downstream effects of urbanization. SWM policies that are concerned only with peak

flow rates are deficient because they ignore the effects of urbanization on the

time characteristics of both direct runoff and discharge from SWM structures

(Hawley, et al., 1981). The erosive capacity of streams can increase tremendously

due to changes in the timing characteristics of runoff, leading to rapid changes in

downstream morphology.

Stormwater management decisionmaking should not be limited to the control of

a peak discharge rate for a single exceedence probability. While the control of

two or more points of the before development flood frequency curve is certainly

an improvement, stormwater management decisionmaking will be most effective when

the policy requires control of the flow duration curve for two or more exceedence

probabilities. Such a policy is not unreasonable given tho'CLIrrent state-of-the-art

-L4-

of stormwater management design. Recent advances have provided the design basis
for integrated stormwater management design (Wong and McCuen, 1982) and two stage

riser design (Soil Conservation Service,1982.). The integrated design approach

permits the use of control methods other than the detention basin; the integrated

approach will also decrease the required storage volume of"a detention basin.- The,

following analysis will demonstrate how the integrated design approach and the

flow duration curvecan be used to properly control the effects of urbanization.

STUDY STIE DESCRIPTION

The study site if a 47-acre subwatershed of Crabbs Branch, a tributary

of Upper Rock Creek in Montgomery County, MD. The subwatershed includes part
of the County Service Park, a complex of County government warehouses and main-

tenance depots. Land uses on the Study site include portions of a railroad,

highway, and a Department of Liquor Control warehouse with associated offices and

parking lots. A Department of Transportation materials and equipment storage

and maintenance depot and a main access road were under development during the

study. The percentage of cleared area ranged from 34 to 44 percent. Slopes of

less than 5 percent existed before total grading in preparation for the development

of the largely impervious complex of large, flat-roofed buildings and parking lots.

The predominant native soils are Chester silt loams, which are characterized as

deep, well-drained, and moderately erodible. These soils are typical of the upland

soils in Montgomery County.

At the foot of the slope, adjacent to the valley stream channel, is a

broad rectangular sediment basin, measuring 400 ft x 140 ft and having a

-L5-

permanent'@ pool surface area of 1.29 acres. The storage volume and riser

characteristics were designed using a policy based on a 10-yr.return period.

The intent of the SWM basin is, therefore, to limit the peak runoff rate from

a 10-yr storm to that which would have occurred prior to development. During

thedevelopment period, the primary basin outlet was a nonperforated corrugated

metal pipe (CMP) riser and barrel. The 6-ft high, 48-in diameter riser was

topped with a closed-lit, circumferential CMP antivortex box, or hood. The

riser also had a shielded 4-inch port, which maintained a normal pool level

18-in below the crest of the riser. A 32-ft wide grassed overflow channel was
provided as an emergency sp@[email protected]. The basin was altered and r@tained for

permanent use as a SKM basin following completion of the upslope development

site.

FLOW-DURATION CURVES AS INDICATORS OF THE EFFECTS OF URBANIZATION

The effects of urban development on the magnitude and duration of

flooding are reflected in the hydrograph ordinates at downstream points.

The downstream direct runofff hydrographs may be synthesized for various

watershed states, such as before development, after development, and after

development with SWM control measures. To examine the effects of each

condition, it is necessary to estimate the runoff hydrographs using a hydrologic

design model. The SCS TR-20 hydrologic model (SCS, 1965) is used to.account

for the changes in land use and the storage provided by SWNL The hydrologic

model is based on Soil Conservation Service techniques (SCS, 1972)... Flow-

druation curves may be completed for the hydrgraph discharge rates of each

watershed condition. A flow-duration curve is constructed, by plotting the

magnitude of flow versus the duration of the magnitude.

-L6-

The data from Crabbs Branch subwatershed were used in CVa1U,1ting the

changes in the storage and timing characteristics brought about by urbanization.

The data were used as input parameters for the TR-20 computer program to

examine the hydrologic response for selected return periods using the SCS

Type II*rainfall distribution. The direct runoff hydrographs were computed at

the downstrown outlet for the 2-, 10-, and 100-yr storm events for various

watershed conditions. The watershed conditions include: (1) before develop-

ment, (2) after development without SWM, (3) after development with a detention

basin, (4) after development with rooftop ponding and infiltration trenches, and

(5) after development with a detention basin, rooftop detention, and

infiltration trenches; the latter watershed condition will be referred as

the integrated SWM control system.

The computed hydrographs for each storm event were used in developing

the flow-duration curves at the down stream outlet for all five watershed

conditions, shown in Figs. 1, 2 and 3. The impact of urbanization for all

storm events are readily apparent by the increased magnitude of the peak

discharge rates, when comparing the after development conditions with that of

the before development conditions. It is evident from the figures that

urbanization causes a more substantial increase for the 2-yr peak rate

(286 percent) than for both the 10-yr (126 percent) or the 100-yr (SO percent)

storms. Additionally, the effect of urbanization on the runoff volume is also

evident from Figs. 1, 2, and 3; the increases in the runoff volume are

represented by the portion of the curve above the predevelopment flow-duration

curve.

Thus, the impact of future development on the magnitude of peak discharge
rates for all storm frequencies is recognized. Clearly, urbanization will affect

-L7-

Before Development
After Development
With SWM Basin
Rooftop Ponding and Infiltration Trenches
Composite SWM Control

-L8-

30 4&D 7 77--

DvJAOPMe-'Jk
1'0

Ovkpo 40rkrv

T!-

4 T

[I -rj

ud IA @AJIO.44A 4@1
Pic

4 -1+

Squares to the Inch -L9-

i2_260

A6
-TT

1;0

uws
12b

A v
r
o--
Ito
_!j

lt
Wit 1,A!
tit
J
T
i
t +

go
A_:A
L J J L
_1 _J_ A-1
So A 4

T-

t
41
I J
A

77-

L
It
114-
xL 4
+

J
F44_
-j-t

T.
t

I f _LL I I
T-7
@ 17
T F --$I"
1_4
LLE

V) Squai*@% to the Iiich L 10

both the magnitude and frequency of flooding as well as causing a change

in the timing characteristics of the runoff hydrograph. The development of

flow-duration curves for urbainizing areas serves to illustrate the changes in

the original durations of flow. Hence, remedial SIVM control methods.may

be further evaluated on additional criteria, other than that of controlling

peak rates, with the analysis of flow-duration curves.

The flow-duration curves of the study watershed are simulated for the

implementation of SWM measures. Fig. 1 reveals that,when the single SWM

detention basin is used, a reduction is acheived in the peak discharge rate

after development to levels that existed prior to development; however, the

duration of the peak rate has been extended from an instantaneous occurrence

to a flow of 1.21S hours. Similarly, the 10-yr and 100-yr predevelopment

peak flow rates have increased in duration by 0.45 hours and 0.50 hours,

respectively. In addition to the peak rates, the flow rates throughout the

runoff hydrograph also increase in duration. Thus, the manner in which

SWM basins manage urban flows is to detain the runoff volumes to be released

at a regulated rate with the intent of limiting the peak, while increasing the

duration of the residual flows.

In comparison with detention basins, the combination of rooftop ponding

and infiltration trenches have a lesser effect in reducing both the magnitude

and the duration of the after development flows, especially nearer to the peak

rate'. This may be attributed to the greater available storage within the

detention basin. Figs. 1, 2,and 3 show that rooftop,ponding and infiltration

trenches are more beneficial in reducing the flow-duration of runoff rates that

are considerably lower than the peak rate; although, the Teductions do not

differ significantly from the uncontrolled, developed state of the study area.

The integrated SWM control system presents the most efficient means of

controlling the after development peak to the predevelopment level. The

reductions in runoff volume due to rooftop detention and infiltration trenches

during the initial period of the storm event provide an additional amount of

storage within the detention basin; subsequently a greater reduction in the peak

runoff rate results. While the integration SWM control system provides further

reductions in both the magnitude of discharge rates and flow-durations, the

increased volume of direct runoff due to urbanization will continue to cause

the effects of prolonging the flow-durations for the predevelopment watershed

discharge rates.

FLOW DURATION CURVES AND DOWNSTREAM BEDLOAD TRANSPORT

Urbanization causes increases in the volume of direct runoff. While SWM

controls are effective in limiting peak flow rates, their effect on volumes

of flow are minimal. The flow-duration curves show that the SWIM controls do

little to limit the volume of flow. In streams with the potential for bedload

transport, the significant increases in the duration of bankfull flows can

cause additional bedload movement. Consequently, changes in the stream

morphology of unstable, undeveloped streams are likely to*occur more rapidly..

Deposited sediments will be scoured and washed from stream bottoms to be

transported to larger water courses downstream.

The transport of sediment occurs as a result of scour from banks. The

actual bedload discharge is difficult to determine due to the complex nature

of the many factors involved. Accurate field measurements are difficult to

obtain, because most existing equipment tends to disturb the natural movement

of sediment which leads to discrepancies in data. Thus, many bedload formulas

-L12-

are based on the statistical theory of random bedload movement or on the

analysis of the equilibrium of forces acting on a sediment particle.

The predicition of bedload movement has been empirically developed by

Schoklitsch (Shulits and Hill, 1976). This method is used herein as a means

of assessing the impact of urbanization and SWM controls on the stream

morphology. The bedload formula gives the peak sediment discharge per unit
width of channelGl, in lbs/sec/ft, as a function of the grain diameter, D,
in ft, the slope, S, the critical discharge, Q 01, and the stream discharge,
Ql, in cfs/ft:

3/
G1- 25 SI/ (Ql - Q01) (1)
D 2

The critical discharge is computed for a sediment specific gravity of

2.65 as:

Q 0.0638-D (2)
01 4/ 3

While the quantities of sediment computed with Eqs. I and 2 may not agree

closely with-actual field results, Schoklitsch's equations are useful in

evaluating the various watershed conditions that reflect urbanizat.ion and

@SWM.

The bedload discharge rates were computed for the 2-, 10- and 100-yr

storm events and each of the watershed conditions. The weighted average

value of 0.0015 ft was computed as the effective grain particle si ze using

the particle size diameters given by McCuen (1979). The mean slope at the

downstream outlet is 0.006 ft/ft. Thus, the effects of the watershed conditions

on bedload movement and total sediment volume may be computed; the results

-L13-

are shown in Table 1. The peak-sediment discharge rates after development

increase by factors of 3.9, 1.5, and 0.4 for the 2-, 10-, and 100-yr storm

events, respectively. However, the total volume of sediment movement after

development increases by factors of 6.5, 3.1, and 1.S for the 2-, 10-, and

100-yr storms, respectively, during the major portion of the event. This

illustrates that, while the peak sediment discharge rate will increase as

a function of the peak flow rate, the total volume of sediment movement will

increase by a greater percentage than the peak sediment discharge when the

stream flow rates increase in duration and magnitude.

Schoklitsch's bedload equations are particularly useful in comparing

the effects of various SWM methods on stream morphology. The concern of

se di-ment transport in natural streams is receiving more attention as the

imapct of SWM methods on stream erosion becomes more evident. The results

of Table 1 reveal that detention basins create changes in the pattern of the

stream stability, which is reflected by increases in both the sediment peak

discharge rates and total volumes of bedload movement. The installation

of a detention basin does, however, reduce the impact of the development

flow rates on stream morphology. Additionally, rooftop ponding and infiltration

trenches are somewhat effective, although a significant reduction in stream

erosion is not expected when examining the flow-duration curves.

The integrated SM control system limits the peak sediment discharge

rate as well as the peak runoff rate to predevelopment levels. The parallel

between the sediment and runoff peaks is related directly by the bedload

formula. Similarly, the flow-duration may be correlated with the total

volume of sediment movement. That is, as the effects of flow-duration increases

the stream discharge, the total volume of eroded sediment increases. Thus,

-L14-

TABLE 1 Effect of Watershed Conditions on Bedload Movement and Total Sediment Volume

I
Watershed Return Period, 2-yr Return Period, 10-yr Return Period 100-y9
Condition
G V F G V F G V F
I T I I T 1 1 T
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Before Development 2.8 1.9 --- 8.6 5.4 --- 26.3 17.7
After Development 13.8 14.1 6.5 21.8 21.9 3.1 36.8 44.0 1.5 1
Without SWM

With Detention Basin 4.8 9.3 4.0 12.7 18.0 2.3 32.7 41.7 1.4
with Rooftop Ponding 10.3 13.3 6.1 20.6 22.6 3.2 38.4 41.8 1.4
and Infiltration Trenches

with Detention Basin, 2.5 6.3 2.4 8.0 14.1 1.6 27.1 37.3 1 .
Rooftop Ponding,
and Infiltration Trenches

Note: G 1 = peak bedload, in lbs per second; V T = total sediment
volume for storm, in cubic yards; and F, = factor of increase
in VT over before development volume.

the total volume of bedload increases by factors of 2, 4, 1.6, and 1,1 during

the 2-, 10-, and 100-yr storm events, respectively, for the integrated SWM

system. The percentage of increase is greatest for the 2-yr event and becomes

less significant as the probability of exceedence decreases, due to the greater

relative consequence of the flow-duration on the smaller storm of greater

frequency.

CONCLUSIONS

The effectiveness of SWM metliods in controlling the impact of development

is often assessed by the degree in which the peak flow rate is controlled.

The goal is often to limit the downstream peak discharge to that of a pre-

development rate in order to lessen the adverse impact of urbanization.

However, when the extent of the landcover change is significant, the increases

in the volume of runoff over the duration of the storm is,more difficult to

manage. The development and analysis of flow-duration curves illustrates that

various SWM methods handle the increased volume of direct runoff in different

ways, with detention basins having the single most significant effect. The

integration of infiltration trenches and rooftop ponding controls provide

an additional degree of control; the integrated SWM control system will limit

the peak flow rate more effectively. However, the increased volume of runoff

after development prolongs the duration of the residual flows, which is eviden t

by the flow-duration curves.

The analysis of the flow-duration curves also addresses the importance

of the timing characteristics on the stream morphology. The prolonged

bankfull flow of streams is obsorvcd to cause the total volume of bedload

movement to increase by More th@m two fold for the 2-, 10-, and 100-yr

-L16-

storm events. Therefore, the impact of stream stability is an important

consideration in evaluating the effects of development and SWM methods.

The changes in the upstream,watershed characteristics can significantly alter

the flow-duration curves and change the streamflow characteristics.

-L17-

REFERENCES

1. "Computer Program for Project Formulation Hydrology," Technical Release
No. 20, Soil Conservation ServIce, Central Technical Unit, USDA-9CS,
May, 1965.

2. Hawley, M.E., Bondelid, T.R., and McCuen, R.H., "A Planning Method for
Evaluating Downstream Effects of Detention Basins," Water Resources
Bulletin, AWRA, Vol. 17, No. 5, 1980, pp. 806-813.

3. Kamedulski, G.E., and McCuen, R.H., "Evaluation of Alternative Stormwater
Management Policies," J. Water Resources Planning and Management
Division, ASCE, Vol. 05 (WR2): 171-186, 1979.

4. McCuen, R.H., "Downstream Effects of Stormwater Management Basins,"
Journal of the Hydraulics Division, ASCE, Vol. 105, No. HY11, 1979,
pp. 1343-1356.

5. National Engineering Handbook, Section 4, Soil Conservation Service,
Hydrology, USDA-SCS, August, 1972.

6. Shulits, S., and Hill, R.D., "Bedload Formulas," for Soil and Water
Conservation Research Division, USDA-ARS, Dec., 1968.

7. Soil Conservation Service, Design of Two-Stage Outlets by Short-Cut
Method, USDA-SCS, National Engineering Staff, Lanham, MD., January, 1982.

8. U. S. Geological Survey, Use of Flow-Duration Curves to Evaluate Effects
of Urbanization on StreairCflow Patterns on Long Island, New York, USGS
Water Resources Investigation 80-114, Albany, N.Y., 1981.

9. Wong, S.L., and McCuen, R.H., "Integrated Stormwater Management Design,"
Technical Report, Dept. of Civil Engineering, University of Maryland, 1982.

-L18-

APPENDIX M

SENSITIVITY OF SCS MODELS TO
CURVE NUMBER VARIATION

Timothy R. Bondeled

Richard H. McCuen

Thomas J. Jackson

SFNSITIVITY OF SCS IMODELS TO CURVE NUMBER VARIATION

1 2 3
Timothy R. Bondelid Richard H. McCuen and Thomas J. Jackson

INTRODUCTION

The Soil Conservation Servi.ce (SCS) models, including the TR-20 computer

program (SCS, 1969) and the simplifi.ed methods in TR-55 (SCS, 1975), are widely

used in hydrologic design. Applications include small urban watershed analyses

and stormwater detention basin design, as well as designs that require either

a peak discharge estimate or an entire flood hydrograph. In fact, numerous

government agencies recommend that the SCS models be used for hydrologic design.

Examples include the states of Maryland and Michigan.

The SCS methods are replacing methods such as the rational formula, which

has been and still is one of the most widely used methods. Application of SCS

methods is increasing rapidly for numerous reasons. First, the inherent defi-

ciencies of methods such as the rational formula are becoming more important as

the cost of inaccurate designs increases. Second, recent studies (Ragan and

Jackson, 1979; Bondelid, et al., 1980) have shown that land use or vegetation

data for the SCS methods can be obtained using remotely sensed data, which can

greatly reduce the cost of data collection. Third, the SCS methods provide a

1Graduate Research Assistant, Dcpartment of Civil Engineering, University of
Maryland, College Park, MD 20742.
2Professor, Department of Civil Engineering, University of Maryland, College
Park, MD 20742.

3
Hydrologist, Hydrology Laboratory, USDA-SEA-AR, Beltsville, MD 20705.

-Mi-

a high degree of flexibility in the type of hydrologic analyses that can be per-

formed using the same type of input data, varying from peak discharge estimates

for small urban inlet areas to analyses on large watersheds that involve structures.

Existing methods such as the rational formula are valid only on small watersheds

where a peak discharge is required; for other types of problems it is necessary

to collect other types of input data.

One of the reasons for the popularity of the SCS models is that the primary

input parameter, the runoff curve number (CN), is defined in terms of land use

treatments, hydrologic condition, antecedent soil moisture, and soil type. The

models can, therefore, be readily applied to ungaged watersheds. The accuracy

of models that use the CN for estimating design peak discharge and runoff volume

are especially sensitive to the estimated CN values.

The objective of this study is to evaluate the sensitivity of the SCS

models to errors in CN estimates. Specifically, the objective is to quantify

in as straightforward a manner as possible the changes in runoff volume and

peak discharge computations as a function of changes in CN estimates.

OVERVIEW OF SCS MODELS

The SCS hydrologic models were developed from field data on a moderate num-

ber of small watersheds (Rallison, 1980). Several versions of the SCS models

are described in the National Engineering Handbook, Section 4: Hydrology (NEH-4)

(Soil Conservation Service, 1972). These versions range from simple to quite

complex and have been adapted for urban areas (SCS, 197S). All versions use

the CN to estimate runoff volume, with the basic expression:

2
Q (P-0.2s) for P>0.2S
(P+0.8S)

in which Q is the runoff volume, P is the precipitation, and S is a retention

parameter. All of these variables are expressed in inches. In practice, the

-M2-

retention parameter, S, is related to the runoff CN through the following trans-

formation:

S = (1000/CN)- 10 (2)

The SCS curve numbers are based on the assumption that the initial abstraction

equals 0.2S.

The runoff curve number, which integrates land cover, soil type, and ante-

cedent soil moisture conditions, is an important input parameter to the SCS

procedures. The determination of curve numbers is a time consuming procedure

requiring detailed data on land cover and soils. While there are numerous methods

for obtaining the necessary input data, Rawls, et al., (1980) showed that esti-

inates of runoff curve numbcrs are. not very sensitive to the method of integrating

soils and land cover data; however, the land use classification system signifi-

cantly affects the accuracy of runoff curve numbers.

Two of the most commonly used SCS models are TR-20 (SCS, 1972) and TR-55

(SCS, 1975). TR-20 is a computerized single event model that incorporates the

surface runoff hydrograph development and includes streamflow and reservoir rout-

ing procedures.

TR-S5 is a distillation of' the results of a large number of TR-20 runs.

The results are expressed in TR-55 as a series of tables and graphs. TR-55

utilizes the standard SCS 24-hour type 11 rainfall distribution. This design

storm is applicable for use in all of the continental United States except for

locations on the Western coast. It has been shown that the TR-55 methods pro-

vide peak discharge estimates that are not significantly different from estimates

obtained from TR-20 as long as the initial abstraction is not greater than 25%

of the total rainfall (Bondelid, 1980; McCuen, 1981). The advantage of using

TR-55 in the analyses of peak discharge estimates instead of using TR-20 is that

the analyses can be performod analytically. The results from analyses with TR-55

-M3-

methods are valid to TR-20 as long as the conditions under which TR-55 applies

are not violated.

TR-55 (SCS, 1975) includes several methods for peak discharge and runoff

hydrograph estimation. The graphical methods provides a relationship between

the unit peak discharge and the time of concentration on a log-log plot. The

unit peak discharge (q ) is expressed in cubic feet p er second per sq. mi.
U
(CSM) per inch of runoff (Q); the time of concentration is in hours. Because

the graphiceLl relationship does not plot as a straight line on log-log paper

the curve can be expressed as a piecewise linear curve.

0.
476.51 t- 0.32192 for t < 0.2 hours (3a)
c c
379.14 t- 0.46394 for 0.2< t < 0.4 hours (3b)
qu c O.S7678 - c
341.90 t c for 0.4< t c < 0.7 hours (3c)
321-48 t- 0.74946 for 0.7< t < 5,0 hours (3d)
c - c-

The peak discharge (q p) in cfs equals the product of the unit peak discharge,'the
drainage area (A) in square miles, and the runoff volume (Q) in inches (i.e.,
qP = quAQ). Eqs. 1, 2, and 3 will be used for determining the magnitude of
errors in Q and q p as functions of variation in estimated CN's.

SENSITIVITY ANALYSIS OF THE TR-5S GRAPHICAL METHOD

Hawkins (1975) used equations 1 and 2 to perform a sensitivity analysis of

runoff volume to error in precipitation and the curve number; a figure showing

the errors in runoff resulting fromlO percent errors in either CN or P was pro-

vided, and a figure that showed the domain of greatest absolute error and

error contours incalculatingQ assuming a 10 percent in either P or CN. Hawkins

concluded: (1) for a considerable range of P, accurate values of CN are more

important than accurate estimates of P; (2) errors in estimating Q are especially

dangerous near the throshold of runoff, Hawkins did not examine the sensitivity

of peak discharges to orroys in P and CN.
-M4-

Sensitivity Definitions
The variation inQ and q pas functions of variation in estimated CN's can be
expressed in terms of the sensitivity of Q and q p to change in CN. In general,
-the absolute sensitivity of a variable y to changes in a variable x is given by:

S = dy (4)
dx

where S is the sensitivity of y to changes in x. Another useful measure is the
relative sensitivity (R s of y to changes in x (McCuen, 1973); Rs is defined by:

R = dy/y = @-y x
s dx/x dx y

Rs measures the proportional change in x for a proportional change in y. A

principal advantage of using the relative versus absolute sensitivity is that
Rs is non-dimensional, so that sensitivities of different variables can be

compared.

A modification of Eq. 5 is desirable for expressing the sensitivities of
Q and qP to changes in CN. The modification is to-express proportional changes
in Q and qp per unit change in CN. These proportions can be expressed by:
R dQ/Q = dQ 1 (6)
Q dCN dCN

and
dqp/qP dq p
(7)
q dCN dCN q

where RQ and R qare the proportional changes in Q and q p, respectively, per
unit change in CN.

The use of Eqs. 6 and 7 rather than Eq. S is desirable because proportional

changes in CN are not as meaninoful as unit changes in CN for several reasons.

The CN is nondimensional, so there is no dimensional advantage to CN proportions.

Also, variations in CN result from misclassifying either land cover, treatment,

-M5-

hydrologic conditions, and/or soil type, so that the magnitude of the CN devia-

tion depends on both the size of the area misclassified and the type of misclass-

ification; the proportional change in CN is not determined only by the proportion

of measurement variation.

Analysis of Derivatives
Eqs. 6 and 7 can be expressed in terms of P, CN, and t c by application of

Eqs. 1, 2, and 3. The design rainfall depth, P, is a constant in terms of CN,
but the t C varies with the CN. A commonly used formula developed by the SCS
for computing t C is:
5 X0.8 (S+I) 0.7
t = -
c 3 1900 Y 0.5 (8)

where Z is the hydraulic length in feet, Y is the average watershed slope in

percent, S is given by Eq. 2, and the t C is in hours. Eq. 8 will be used in
developing an expression for R q as a function of P and CN. A time of co ncentra-
tion should be modified using the adjustment coefficients provided in Chapter 3

of TR-55 when there is a significant percentage of impervious area or the hydrau-

lic length has been modified.

Analysis of R Eq. 6 can be rewritten as:

R = 3Q dS /Q (9)
Q 3S dCN

By application of Eqs. 1 and 2, and rearranging terms yields:

R 0 @4 + 0.8 1000 (io)
P_ .2S) (P+0.8S) CN2

Eq. 10 can be directly applied to determine the proportional change in Q that

results from a unit change in CN for a particular design rainfall depth.

-M6-

Analysis of R Eq. 7 can be rewritten as:

Dq @d 1, t
T /q
q DQ dCN e) t C] p

Differentiation of Eq. 3 with respect to Q yields:
;q P = a t -b (12)
DQ C

in which a and b are the coefficients of Eq. 3. Differentiation of Eq. 3 with

respect to t Cyields:

Ri -(b+l)
T -Qabt (13)
at C

Application of Eqs. 1 and 2 yields:

0
dQ 0.4(P+0.8S)(P-0.2S)-0.8(P-0.2S) -11 00 (14)
dCN (P+O-8S) 2 CN2

By Eq. 8 and rearranging terms:

dt C at C dS -700 t C
= - . - = (15)
dCN DS dCN (S+.I)(CN2

After rearranging, the combining of Eqs. 3, 11, 12, 13, 14, and 15 yields:

R 0.4 0.8 1000 700b (16)
q (P-0.2S) (P+0.8@) I CN 2 ' (S+I)(CN 2)

It is important to notice that R q is not a function of t C The coefficient b
is dependent on tC , but b is constant over various ranges of t c Therefore,
for tC between 0.7 and 5.0 hours, R q is dependent only on P and CN.
Separation of Error in --R-(I Eqs. 10 and 16 can be combined to yield:

-M7-

R R 7001) (17)
(S+1) (CN

The second term on the right-hand side of Eq. 17 represents the proportional
change in qp due to the variation in estimating tc that may result from
deviation in CN, Z, and Y. Therefore, the variation in q p is equal to the
sum of the deviation in runoff volume plus the deviation associated with esti-

mating tC*

FVAIAJATION 01: SENSITIVITIES

RQ is a function of P and CN as observed in ELI. 10. Fig. I presents curves
of R versus P for CN's ranging from 50 to 95. Fig. I shows that R decreases
Q Q

as P and CN increase. This is reasonable because the CN determines the initial

abstraction and the infiltration rate. As P and CN increase, the proportion

of the rainfall that goes into the initial abstraction and infiltration decreases,

so the proportional error also decreases.

Fig. 1 can be used directly to determine the percent error in Q associated

with a unit change in CN for a particular design rainfall depth. Fig. 1 can also

be used for CN variation other than unity. For instance, for a design rainfall
depth of 6 inches and a CN of 70, the value or R Q is approximately 0.033. A
CN error of +5 will, therefore, produce an error of about 0.003 x 5, or 16.5

percent.
Fig. 2 presents curves of R q versus P for CN ranging from 50 to 95. A
value for b of 0.75 is used, which is associated with t c between 0.7 and 5.0
hours. Fig. 3 is analogous to Fig. I and can be used for estimating proportional
errors in qp in the same manner as Fig. 1 can be used.
Fig. 2 shows that R q decreases as P increases for a particular CN. However,
Rq does not necessarily decrease as the CN increases. This behavior is in con-
trast to that shown in rig. 1, in which R Q always decreases as the CN increases.

-M8-

The difference in the behavior of R q as a function of CN as compared to
the behavior of R Q can be explained by examination of Eq. 17. The second term
on the right-hand side is the error associated with the error in estimating
tc This term is independent of P, so for a given value of CN, R q is equal to
RQ plus a constant. As P increases, R Q decreases, so the effects o f the t c

term on R become more dominant. The effects of the t term on R cause the
q c q
difflerence in the behavior of R q as compared to R Q*
The effects of the t c -term in Eq. 17 are illustrated in Fig. 3, which shows
the proportion of the xro r Ln R (I that is caused by the error in tc, The effects
of the error in t c increase as hoth 1) and CN increase. The effects can be

significant; at P=8 in. and CN=95, the error in t accounts for 75% of the total
C
error in qP*

APPLICATIONS OF THE SENSITIVITY ANALYSIS

The sensitivity analysis as summarized in Figs. I and 2 can be used directly

for evaluating the hydrologic effects of differences in estimated CN's. Differ-

ences in CN estimates can occur for several reasons. One reason can be the use

of different methods for estimating CN's- Another reason is random variation;

ra ndom variation can be caused by both human judgment and data base errors.

Finally, CN's are at best approximations of the true runoff potential; the sensi-

tivity analysis can serve as a bridge between the degree of inaccuracy in.the CN

tables and the hydrologic effects of the inaccuracy.

Effects of Different CN Estimation Methods

An example of direct use of the sensitivity analysis is to evaluate the

hydrologic effects due to the use of two different methods.for estimating CN's.

For instance, a study of three watersheds in southeastern Pennsylvania (Bondelid,

et al., 1980) compared CN's estimated by conventional methods vs. using Landsat

_M9-

data. The sensitivity analysis can be readily applied to this study to determine

the hydrologic effects of using Landsat vs. conventionally derived CN's.

The study by Bondelid et al., (1980) was conducted by estimating the con-

ventional and Landsat CN's on three watersheds. These watersheds are the

Quitiapahilla, Chickies Creek, and Little Mahanoy basins. Tables 1, 2, and 3
summarize the percent differences in Q and q p for each subwatershed in the
Quittapahilla, Chickies Creek, and Little Mahanoy watersheds, respectively, for

the 100-year, 24-hour rainfall of 6.3 inches. The values were determined by
computing RQ and Rq from Eqs. 10 and 17, multiplying by the residual, and then
multiplying by 100 to convert to percent. The residuals were computed by sub-

tracting the Landsat CN from the conventional CN. If the conventional CN's are

taken as the standard by which other CN estimation procedures are evaluated, then
Table 1, 2, and 3 give the percent error in computed val ues of Q and q p that
can occur as a result of using Landsat.

Tables 1, 2, and 3 show that the differences in q are approximately twice
p
as large as the differences in Q for the 100-yr. storm. The average differences
for the 53 subwatersheds in computed values of Q and qP are -4.5 and -6.3 per-
cent, respectively. Given the accuracy of the models and the random error

associated with CN estimation, the overall differences are probably not signifi-

cant. This example illustrates how the sensitivity analysis can be applied in

evaluating alternative CN estimation methods. The differences in the methods

can be readily evaluated in terms of the resulting hydrologic differences.

Effects of Input Variation

A recent study, which is not published at the present (1981), involved the

estimation of peak d1scharges on approximately 70 watersheds, with estimates

made independently by S hydrologists on each watershed. The study involved the

comparison of nine hydrologic methods, including the TR-55 graphical method.

Thus, the data base included a large number of curve number estimates, including

_M10-

independent estimates for the same watershed. Curve number estimation requires

the synthesizing of land use,hydrologic condition and treatment, and soils data.

Thus, any variation in the delineation of areas of homogeneous land use, hydro-

logic conditions, or soil type will be reflected in the variation of CN estimates.

For watersheds in the midwestern United States, where the 100-year, 24-hour

precipitation is on the order of 6.5 inches, the five CN estimates often varied

by 3 curve numbers. For a CN of 70, Figs. 1 and 2 show that the relative error

in runoff volume and peak discharge would be 0.035 and O.OS, respectively. Thus,

for a variation of 3 curve numbers one could expect 10 percent and 15 percent

differences in runoff volume and peak discharge, respectively, as a result of

variation solely in the delineation of land use and soil type. For watersheds

that are characterized by wide variation in land use and soil type, one can ex-

pect larger variations. This is not the fault of the CN concept, but results

from failure of the hydrologist to take the proper care in gathering the input

data for CN estimation.

CONCLUSIONS

A straightforward technique for evaluating the hydrologic effects of CN

variation has been developed in this study. These effects are evaluated in

terms of the proportional changes in total runoff and peak discharge for a unit

change in CN. The results are summarized in a series of curves in which the

proportional changes in total runoff and peak discharge are shown as functions

of design rainfall depth. The curves show that the effects of variation in CN

decrease as design rainfalt deptli increases. This supports Hawkins (1975) con-

clusion that errors are especially dangerous near the threshold of runoff.

Examples of the applicability of the analysis are presented. These examples

include the evaluation of the hydrologic effects of Landsat vs. conventionally

derived CN's. Another example is evaluation of the effects of human judgment in

_M11-

estimating CNIs. The curves of Figs. 1, 2, and 3 should be especially useful

for those who do not apply SCS methods on a regular basis.

Figs. 1, 2, and 3 were developed using the TR-55 graphical method. The

graphical method was developed from numerous TR-20 computer runs. Therefore,

Figs. 1, 2, and 3 should provide good estimates of the sensitivity of volumes

and peak discharges estimated with TR-20 when the conditions specified in TR-55

for using the graphical method are not violated.

-M12-

REPERENCES

1. Bondelid, T.R., 1980, Test of TIZ-55 Graphical Method, Technical Report,
University of Maryland, Department of Civil Engineering.

2. Bondelid, T.R., Jackson, T.J., and McCuen, R.H., 1980, Comparison of
Conventional and Remotely Sensed Estimates of Runoff Curve Numbers in
Southeastern Pennsylvania, Proceedings of the Annual Meeting of the
ASP, St. Louis, Mo.

3. Hawkins, R.H., 1975, The Importance of Accurate Curve Numbers in the
Estimation of Storm Runoff, Water Resources Bulletin, Vol. 11(5):887-891.

4. McCuen, R.H., 1973, The Role of Sensitivity Analysis in Hydrologic Modeling,
Journal of Hydrology, Vol. 18, pp. 37-S3.

S. McCuen, R.H., 1981, A Guide to Hydrologic Analysis Using SCS Methods,
Prentice-Hall, Inc., Englewood Cliffs, N.J.

6. Ragan, R.M. and Jackson, T.J., 1980, Runoff Synthesis Using Landsat and
the SCS Model, Journalof tho Hydraulics Division, ASCE, Vol. 106,
No. 11Y5, pp. 667-678.

7. Rallison, R.E., 1980, Origin and Evolution of the SCS Runoff Equation,
Proceedings, Symposium on Watershed Management, ASCE, Irrigation and
Drainage Division, Now York.

8. Rawls, W.J., Shalaby, A., and McCuen, R.H., 1980, Comparison of Methods
for Determining Urban Runoff Curve Numbers, preprint for American Society
of Agricultural Engineers, paper No. 80-2565, Chicago.

9. Soil Conservation Service, 1969, Computer Program for Project Formulation-
Hydrology; Technical Release No. 20.

10. Soil Conservation Sci-vice, 1972, National Enginecring 111ndbook Section 4:
Hydrology.

11. Soil Conservation Service, 1975, Urban Hydrology for Small Watersheds;
Technical Release No. 55.

-M13-

TABLE 1. HYDROLOGIC EFFECTS AT LANDSAT CNIS IN QUITTAPAHILLA

Subarea Conv. CN Landsat CN Residual Whange in Q Whange in QP

1 84. 81. -3. -7.410 -14.824
2 84. 81. -3. -7.410 -14.824
3 83. 80. -3. -7.S42 -14.794
4 70. 72. 2. 6.212 10.308
5 77. 7S. -2. -5.582 -9.954
6 72. 72. 0. .000 .000
7 81. 79. -2. -5.167 -9.855
8 85. 82. -3. -7.283 -14.873
9 82. 80. -2. -5.074 -9.8S7
10 72. 70. -2. -6.212 -10.308
11 85. 83. -2. -4.814 -9.937
12 82. 80. -2. -S.074 -9.857
13 72. 73. 1. 3.004 5.088
14 85. 84. -1. -2.387 -4.981
is 81. 78. -3. -7.823 -14.787
16 84. 76. -8. -20.669 -39.419
17 78. 78. 0. .000 .000
18 81. 79. -2. -S.167 -9.95S
19 70. 69. -1. -3.215 -5.232
20 7S. 74. .-1. -2.878 -5.017
21 81. 75. -6. -16.098 -29.650
22 77. 76. -1. -2.763 -4.966
23 81. 77. -4. -10.S29 -19.727
24 73. 74. 1. 2.940 S.050
25 84. 82. -2, -4.897 -9.897
26 87. 86. -1. -2.309 -5.049

Average: -4.929 -9.508

TABLE 2. HYDROLOGIC EFFECTS OF LANDSAT CN's IN CHIMES CREEK

Subarea Conv. CN Landsat CN Residual %Change in Q %Change in QP

1 68. 69. 1. 3.293 5.291
2 S9. 58. -1. -4.347 -6.242
3 S9. 59. 0. .000 .000
4 74. 74. 0. .000 .000
S 70. 72. 2. 6.212 10.308
6 78. 74. -4. -11.164 -19.908
7 79. 81. 2. 5.167 9.85S
8 79. 79. 0. .000 .000
9 79. 79. 0. .000 .000
10 7S. 74. -1. 2.878 -5.017
11 76. 77. 1. -2.763 4.966
12 78. 77. -1. -2.709 4.949

13 79. 79. 0. .000 .000

14 79. 80. 1 . -2.608 4.929

is 77. 76. -1.. -2.763 -4.966

16 77. 76. -1. -2.763 4.966

Average: -4.11 -.669

TABLE 3. HYDROLOGIC EFFECTS OF LANDSAT CNIS IN LITTLE MAHANOY

Subarea Conv. CN Landsat CN Residual %Change in Q %Change in QP

1 62. 60. -2. -8.049 -11.865

2 63. 64. 1. 3.747 S.676

3 67. 67. 0. .000 .000

4 74. 72. -2. -5.943 -10.136

5 74. 71. -3. -9.011 -15.263

6 71. 70. -1. -3.142 -S.179

7 73. 70. -3. -9.213 -15.392

8 73. 73. 0. .000 .000

9 7S. 73. -2. -S.817 -10.065

10 84. 82. -2. -4.897 -9.897

11 73. 72. -1. -3.004 -5.088

Average: -4.121 -7.019

0.20- C N
so -
60 -

To
so
90
95

0.15-

0.10

0.05-,.

0
3 4 6 8
RAINFALL (in)

FIGURE RELATIVE SENSITIVITY OF TO CN
(0. 7 < tc< 5. 0 hours)

-Ml 7-

0.20 CN
50
60
70
80
90

0.16-

cl
Cr

00.10-

0.05

01
3 4 6 7 8
24-hr. RAINFALL (in)

FIGURE 2. R-FLATIVE SENSITIVITY OF q p TO CN-
(0. 7 < tc< 5.0 hours)

_M18-

PROPORTION OF ERROR IN qp DUE TO tc

I NIN,

@Tj
0
l7i

A
@:i

A tZ--'
Ln
C
@Xll

I - - - -, ----
.. r
.I. i lllullllWllm@
3 6668 14102 3673