PLANNING METHODOLOGIES
FOR ANALYSIS
OF
LAND USE / WATER QUALITY
RELATIONSHIPS
U. S. Environmental Protection Agency
Washington, D. C. 20460
October, 1976
-------
EPA REVIEW NOTICE
This report has been reviewed by the Environmental
Protection Agency and approved as satisfying the
terras of the subject contract. Approval does not
signify that the contents necessarily reflect the
views and policies of the Environmental Protection
Agency, nor does mention of trademarks or commercial
products constitute endorsement or recommendation
for use.
-------
PLANNING METHODOLOGIES FOR ANALYSIS OF
LAND USE/WATER QUALITY RELATIONSHIPS
by
Thomas R. Hammer, Ph.D.
In partial fulfillment of
EPA Contract No. 68-01-3551
for the
U.S. ENVIRONMENTAL PROTECTION AGENCY
Water Planning Division
EPA Project Officer: William C. Lienesch
October 1976
-------
PROJECT PARTICIPANTS
Betz Environmental Engineers, Inc.
Project Director
William K. Davis, AIP
Asst. Vice President B.E.E.
Principal Investigator
Thomas R. Hammer, Ph.D.
Principal Socio-Economic Planner
Major Contributors
Francis X. Browne, Ph.D., P.E.
Victor J. DePallo
William H. Gammerdinger
James V. Husted
Thomas G. May, P.E.
D. Kelly O'Day, P.E.
Jacquelyn G. White
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
DATE: DEC 3 1976
SUBJECT: Planning Methodologies for Analysis of Land Use/Water
Quality Relationships
uhd Notzon, AttTng Director
Water Planning tW vision (WH-554)
TO: All Regional Water Division Directors
ATTN: Regional 208 Coordinators
Technical Guidance Memorandum: TECH- 24
Purpose
This memorandum transmits the recently completed report, "Planning
Methodologies for Analysis of Land Use/Water Quality Relationships."
It is intended for use by state and areawide agencies in the develop-
ment of their water quality management programs.
Guidance
This report evaluates the potential usefulness and practicality of
various planning methodologies which can be used to quantitatively
determine the relationship between land use and water quality. It also
evaluates various land use and land management controls which can be
used to reduce pollutant loadings. In carrying out these evaluations,
the report reviews much of the current literature on the relationship
between land use and water quality.
While the report examines various land uses, it is intended to focus on
land uses commonly found in developed and developing areas. It is also
intended to focus on stormwater related pollution sources in such areas.
As a result municipal and industrial point sources as well as nonurban
nonpoint sources are treated peripherally.
After evaluating a range of planning methodologies and control measures,
the report examines the analysis and control of pollutants resulting
from hydrologic modifications, on-lot disposal systems, and construction
activity. These pollutant sources were chosen for in-depth examination
because it is felt that they are major sources of pollution which can be
prevented and which occur in many areas of the country.
The Office of Research and Development in conjunction with the Water
Planning Division has recently published related guidance, the Areawide
Assessment Procedures Manual. The manual differs from this report on
EPA Form 1320 6 fRcv. 3-76)
-------
planning methodologies in that it is more comprehensive, covering a
range of point and nonpoint sources. The material presented in this
report is intended to provide more detail on certain issues and in
general to supplement the more comprehensive guidance presented in the
manual.
If you would like further information on this report, please contact
Bill Lienesch of the Program Development Branch (426-2522).
Enclosure
cc: State and Areawide Agencies
-------
TABLE OF CONTENTS
Section Page
1 INTRODUCTION i
2 SUMMARY 4
3 CONCLUSION H
4 RECOMMENDATIONS 17
5 POLLUTANT GENERATION 23
6 REVIEW OF PLANNING METHODOLOGIES 59
7 ANALYSIS AND CONTROL OF EXISTING PROBLEMS
DUE TO UNRECORDED POLLUTANT SOURCES 84
8 IMPLEMENTATION OF CONTROLS FOR NEW
URBAN DEVELOPMENT 121
9 CONTROL OF HYDROGRAPHIC MODIFICATION 133
10 ASSESSMENT AND CONTROL OF ON-LOT
DISPOSAL SYSTEM PROBLEMS 172
11 EVALUATION AND CONTROL OF EROSION
FROM CONSTRUCTION SITES 191
BIBLIOGRAPHY
206
vii
-------
LIST OF FIGURES
Figure Title Page
1 Scope of the Present Study 2
2 Elements of Urban Water Quality Problems 5
3 Sediment Concentration in Runoff from
an Urbanizing Basin in Maryland 1957-1962 39
4 Sediment Yields from Construction Sites
and Non-construction Lands in Maryland
and Virginia 39
5 Response of SWMM Program to Input
Parameters 69
6 Technical Planning Activities Relevant
to Existing Unrecorded Sources 94
7 Estimated Relationships between Impervi-
ousness and Population Density 118
8 Hypothetical Flood-Frequency Relation-
ships 153
9 Hypotehtical Watershed Used in Hydro-
graph Computations 155
10 Relationship for Determining Required
Storage Capacity of Detention Basin 156
11 Discharge Hydrographs at Point "A" 157
12 Discharge Hydrographs at Point "B" 159
13 Summary of OLDS Impact on Water Quality 174
14 Schematic Diagram of OLDS Water Quality
Impact Evaluation 181
15 Dilution Ratio Analysis 186
16 Sediment Delivery Ratio for Relatively
Homogeneous Basins 198
17 Nomograph for On-site Erosion Control
Planning 199
IX
-------
LIST OF TABLES
Table Title Page
1 Organic Pollutant Loadings from Urban
Areas with Separate Sanitary Sewers 26
2 Comparison of Average Heavy Metal
Concentrations in Urban Storm Runoff 29
3 Nutrient Loadings in Pounds/Acre/Year 33
4 Summary of Heavy Metal Data from
Philadelphia Watersheds 36
5 Comparison of Surface Pollutant Accumu-
lation Rates with Instream Loading Rates 45
6 Significance of Runoff from Traffic Related
Roadway Deposits to Urban Water Pollution 47
7 Comparison of Hypothetical Storm and
Nonstorm Conditions 49
8 Number of Observations Required to Esti-
mate Average Concentrations of Water
Constituents, Based on Oklahoma Data 108
9 Sample Computation of Precipitation
Weighted Average Lead Concentrations for
Basins in Lodi, New Jersey 112
10 Characteristics of Pollution Controls 129
11 Descriptive Data for Hypothetical
Watershed 155
12 Sample Computation of Storage Capacity
in Infiltration Devices for New
Development 167
13 Comparison of Characteristics of Raw
Sewage and Septic Tank Effluent 176
-------
List of Tables (Continued)
Table Title
14 Analysis of Annual Sediment Yield for
Small Rural Watersheds in Chester
County, Pennsylvania 194
15 Values of Sediment Delivery Ratio for
Small Agricultural Basins in Texas 194
16 Computed Soil Loss from 5-acre Grid
Cells in the Wissahickon Watershed,
Philadelphia 204
XI
-------
SECTION 1
INTRODUCTION
Past water quality control activity in the United States
dealt primarily with treatment and disposal of wastewater
from industry and municipal sewer systems. In contrast, a
basic element of the current Federally-sponsored program of
water quality management planning is the recognition that
many other pollutant sources must also be considered if
water quality objectives are to be achieved. It is now ac-
knowledged .that an extremely wide variety of human activi-
ties which'affect the land surface can constitute sources of
water pollution. In order to overcome the generally low
level of experience which exists in dealing with these
sources, the U.S. Environmental Protection Agency has spon-
sored a series of studies to provide guidance to state and
areawide water quality management agencies. This study is
one such effort. The present objective has been to develop
a planning methodology for analyzing land/water relation-
ships and evaluating possible control measures.
«!.•
The scope of the present study of pollutant sources is il-
lustrated in Figure 1. The study focuses strictly upon
urban influences on water quality. The definition of urban
land would include both developed and developing areas.
Pollutant loadings involving urban land are categorized as
either recorded or unrecorded effluent discharges. For
present purposes, recorded sources can be defined as efflu-
ent discharges covered by National Pollutant Discharge
Elimination System (NPDES) permits as of 1974. The. main
recorded sources are industrial and municipal treatment
plant effluents. The present study is concerned with unre-
corded pollutant discharges, which include all other influ-
ences on water quality. The principal mechanisms whereby
unrecorded pollution occurs in urban areas are the follow-
ing: washoff and erosion of materials from land surfaces;
unauthorized disposal of wastes in surface waters and storm
sewers; outflow of contaminated groundwater (from on-site
septic systems, landfills, and sewer leaks); overflow of
municipal sewer systems; and hydrographic modification. The
definition of recorded and unrecorded sources in terms of
present coverage by NPDES permits is purely a practical
distinction, which does not necessarily relate to the manner
in which pollutants are conveyed to receiving waters.
-------
HUMAN INFLUENCES
ON WATER QUALITY
I
NON-URBAN
LAND AREAS
1
1
URBAN LAND
I
RECORDED
EFFLUENT
DISCHARGES
UNRECORDED
POLLUTANT
DISCHARGES
I
COMBINED
SEWER
OVERFLOWS
OTHER INFLUENCES ON
WATER QUALITY IN
URBANIZED AREAS
I
EXISTING
URBAN
LAND
FUTURE
LAND DEVELOPMENT
Source: Betz Environmental Engineers, Inc.
Figure 1 SCOPE OF THE PRESENT STUDY
-------
An extremely important type of unrecorded pollutant source
in many urban areas is combined sewer overflows. During
storms, the inflow of surface runoff to combined sewer
systems may cause the total wastewater flow to exceed
system capacity; the resulting overflows deliver to re-
ceiving waters a mixture of untreated sewage, storm water
pollutants,,and materials which .have accumulated in sewer
pipes over time. Due to the attention which tjlis problem
has receive^ in other studies, and the fact thft it gen-
erally does not involve current types of urban development,
the explicit focus of the present study is limited to areas
without combined sewers. However, the material presented is
relevant to pollutant loadings from combined sewer areas
which originate from sources other than sanitary sewage
overflow pet" se.
The final distinction noted in Figure 1 is between existing
urban land and future urban development. A ma^or emphasis
of water quality management planning is the use of preventive
measures for urban water quality control, which can be inte-
grated in the design of new land development projects. It
is felt that preventive measures are likely to be much more
cost-effective, and perhaps easier to implement, than reme-
dial contro£ measures which can be applied to urban land
after development. .The pjcimaj^t^pbje.Cjtive pf> tbe present
study has bf|en to provide assistance in designing and eval-
uating these preventive controls. ^
A fairly extensive review of unrecorded pollutant loadings
and problems has been conducted as part of this study in
order to address several critical issues which are discussed
in the next^section. The findings of this review are sum-
marized here in Section 5, and are presented in detail in
the Technical Appendix., A number, .of planning methodologies
which have f>een developed by others are then evaluated in
terms of their potential usefulness for water quality
management planning. Based on this and other information,
an overall strategy for dealing with the water quality
impacts of new urban development is suggested, along with a
general approach for analysis and control of existing prob-
lems. The specific methodologies which are recommended as
planning tools are then discussed in detail in the final
sections.
-------
SECTION 2
SUMMARY
The present study has been undertaken with the goal of de-
veloping a planning methodology for use in urban water
quality studies. To this end, four general classes of ex-
isting methodologies have been reviewed: (1) stormwater
modeling, (2) environmental synthesis techniques, (3) sta-
tistical methods, and (4) other predictive tools. The plan-
ning approach for design of preventive controls which is
ultimately recommended in this report stresses the use of a
variety of single-purpose techniques rather than the first
three of these classes of methodologies.
A viewpoint which developed in the course of the study is
that the choice of a planning approach is critically de-
pendent upon a number of underlying issues which do not
appear to have received adequate attention in the litera-
ture. Much of the discussion in this report is therefore
oriented toward these general issues, as opposed to the
mechanics of applying specific planning methodologies. In
order to describe these issues briefly here, a categoriza-
tion of relevant water quality problems is presented in
Figure 2.
As shown in Figure 2, three categories of urban land influ-
ence on water quality are considered, excluding recorded
discharges and combined sewer overflows. Surface runoff
activated pollutant sources (abbreviated as "SRA sources")
include all cases in which materials are washed or eroded
from land surfaces by stormwater or snowmelt. Certain other
mechanisms such as sanitary sewer bypasses and scouring of
catch basins and storm sewers are also included. Hydro-
graphic modification refers to the influence of urban devel-
opment on the magnitude and timing of water flows through
the hydrologic system. This type of effect, which relates
to water quality in several ways, is noted separately due to
its strategic role in the planning approach suggested here
for new development. Non-SRA pollutant sources include all
other influences on water quality in urban areas, such as:
sewer system leakage; unauthorized waste discharges to
surface waters and storm sewers; leachate from landfills and
on-site waste disposal systems; and other forms of ground-
water pollution.
-------
URBAN LAND INFLUENCES
ON WATER QUALITY
(Excluding Recorded Pollutant
Discharges and Combined Sewers)
SURFACE RUNOFF
ACTIVATED
POLLUTANT SOURCES
("SRA Sources"-)
HYDROGRAPHIC
MODIFICATION
DUE TO URBAN
DEVELOPMENT
NON-SRA
POLLUTANT
SOURCES
POLLUTANT
WASHOFF FROM
STREETS AND
HIGHWAYS
OTHER SRA
POLLUTANT
SOURCES
CHANNEL AND
BENTHIC CONDITIONS
IN RECEIVING
WATER BODIES
TRANSIENT
STORM-RELATED
WATER QUALITY
PROBLEMS
LONG-TERM WATER
QUALITY PROBLEMS
(Including Problems at
Extreme Low Flow)
Source: Betz Environmental Engineers, Inc.
Figure 2 ELEMENTS OF URBAN WATER QUALITY PROBLEMS
-------
Pollutant washoff from street and highway surfaces is dis-
tinguished in Figure 2 from other SRA pollutant sources, due
to the emphasis which has been placed upon the former in the
literature and in discussions of urban problems generally.
Channel and benthic conditions are noted because they
represent a potentially critical element of water quality
which may not be sufficiently acknowledged. Stream channels
can be important both as a source and as a sink for storm-
water pollutants. Benthic accumulations of materials also
tend to form in estuaries and standing water bodies. The
impacts of these accumulated pollutants, and the mechanisms
whereby they are released to the overlying water, are
generally not well known.
Figure 2 distinguishes between two classes of water quality
problems: transient problems which occur during and imme-
diately after storm events, and long-term problems which
tend to be persistent. The former usually involve temporary
dissolved oxygen depletion, and/or temporarily high con-
centrations of toxicants or pathogens. The seriousness of
these transient problems is critically dependent upon the
nature of the receiving waters affected. Long-term problems
can either be the result of continuous (non-SRA) pollutant
discharges, or the delayed effects of SRA loadings. Three
prominent examples of long-term problems are: aquatic plant
overgrowths due to nutrient enrichment; sedimentation of
stream channels, rivers and harbors; and buildup of toxic
materials in aquatic food chains. All of these problems can
be caused by transport of particulate materials and associ-
ated pollutants during storms, as well as by continuous dis-
charges. Two important reasons for distinguishing between
transient and long-term problems in the design of management
studies are that: (1) it is usually easier to evaluate the
existence and seriousness of long-term problems than is true
in the case of transient problems (although the linkages to
unrecorded pollutant loadings are often complex and poorly
understood in both cases); and (2) different types of load-
ing estimates are required for analysis of transient and
long-term problems.
An issue of immense importance in urban water quality plan-
ning is that it is often unclear just where unrecorded pol-
lutant loadings are coming from. Commonly, washoff of
materials from street surfaces is assumed to be the pre-
dominant source. This is almost certainly true in center
-------
city areas, where the only other eligible sources are side-
walks and parking lot runoff, sewer system overflows, pol-
lutants from the atmosphere, and improper waste disposal in
storm sewers. However, for the preponderance of urban land,
the relative importance of specific pollutant sources tends
to be uncertain. This fact has obvious implications for the
ability of investigators to evaluate the benefits of selec-
tive land management controls such as streetsweeping.
Studies which seek to develop loading coefficients or rela-
tionships for specific classes of urban land use may provide
valuable information, but will not necessarily resolve this
dilemma. For example, average loading coefficients for com-
mercial or industrial districts may not indicate the rela-
tive pollutant contributions of streets, loading areas,
sewer system overflows, unauthorized discharges, etc.; and
may not distinguish between cases in which pollutant genera-
tion is widespread, versus cases in which a small proportion
of establishments are largely responsible. This issue is
particularly important since the customary focus of atten-
tion—street surface contaminants—may be one of the least
amenable sources to control.
The second major issue has to do with the magnitude of un-
recorded pollutant loadings from urban land. An extremely
wide variety of loading magnitudes and stormwater pollutant
concentrations have been reported in the literature. Al-
though this question can be answered readily for a study
area, once monitoring activities have been conducted, it is
an important issue in the design of management studies since
the anticipated loading magnitudes have an important bearing
on the choice of analytical methodologies and the allocation
of planning resources.
A third issue involves the general difficulty of analyzing
transient water quality problems. Concentrations of toxi-
cants and pathogens during storms may not be amenable to
accurate prediction, especially when the objective is to
estimate values associated with "critical conditions" having
specific recurrence characteristics. Dissolved oxygen
levels during dynamic flow conditions are notoriously dif-
ficult to simulate accurately, due to short-term variation
in reaeration and reaction rates as well as pollutant in-
puts. An equally important fact is that, even if transient
water chemistry is well known, it is generally hard to es-
tablish the impacts of short-term chemical phenomena on
aquatic biota—which are commonly the basis for design of
-------
controls. The available literature provides relatively
little assistance in conducting such an assessment, par-
ticularly with regard to the synergistic effects of multiple
pollutants (e.g., high sediment, low DO, and high metals
concentrations during storms). Existing water quality
standards are likely not to provide adequate guidelines for
design of controls, since the standards may not be relevant
for transient conditions and/or may fail to cover many im-
portant water constituents. Finally, defining the appropri-
ate critical conditions for design purposes is itself a
complex problem, since stormwater impacts are related to the
spatial extent of precipitation, antecedent land conditions
and the intensity and duration of rainfall at a given point.
The fourth issue has to do with the overall philosophy of
water quality management planning. Due to the comprehen-
siveness of current studies and the consideration which is
given to unrecorded pollutant sources, these studies differ
in a number of fundamental respects from traditional fa-
cility planning. In facility planning, the water quality
problems addressed tend to be fairly well known (although
this is becoming less true as a wider variety of water
constituents are considered). The loadings produced by the
sources under consideration can be measured directly and/or
forecast with reasonable accuracy. The effectiveness of
control options in limiting pollutant discharges can usually
be quantified with some precision. And finally, the auth-
ority and responsibility for implementing the recommended
plan usually rest with a limited number of actors who are
identifiable throughout the planning process. Given these
conditions, the development and selection of plans can often
be based upon straightforward application of cost-effective-
ness criteria, subject to environmental impact constraints.
However, in the case of comprehensive areawide water quality
management planning, none of the above conditions may hold
to a high degree. It is therefore possible that funda-
mentally different criteria for development and selection of
plans will be appropriate. For example, detailed compari-
sons of alternatives in terms of cost-effectiveness may have
limited value if the effectiveness of controls is imper-
fectly known, and costs are to be incurred by widely dif-
ferent actors.
All of these issues have important implications for the
selection of technical planning methodologies. In particu-
lar, they affect the planning resources which can profitably
8
-------
be allocated to: (1) the development of general pollutant
generation relationships for urban land (with or without
formal stormwater modeling); (2) the analysis of transient
water quality problems, as opposed to long-term problems;
and (3) the use of procedures which allow the investigator
to "trade off" different control alternatives.
In order to gain some perspective on these issues, the re-
view of urban unrecorded pollution problems has attempted to
address the following questions:
1. What are the typical magnitudes of unrecorded
loadings from urban areas without combined sewers?
2. What are the important sources of variation in
loading rates?
3. Do loadings from new development differ from
average urban pollutant yields?
4. To what extent is urban development without com-
bined sewers likely to result in transient water
quality problems?
5. What are the major factors affecting the serious-
ness of SRA pollutant loadings?
The findings of the review have been suggestive rather than
definitive, but appear to have several important implica-
tions for water quality planning. These findings are sum-
marized in the next section, and discussed in greater detail
in Section 5 and the Technical Appendix. As a consequence
of this review, an overall planning strategy has been de-
veloped, incorporating simplistic analytical procedures,
which might be favorable in cases where planning resources
are limited. The evaluation of planning methodologies pre-
sented in Section 6 reflects this approach, in that it
points out various liabilities of methodolgies which attempt
to deal in a comprehensive fashion with unrecorded loadings.
The recommended strategy distinguishes between controls
applying to existing urban development and controls for new
development. In the latter case, controls would be an in-
tegral part of development design, and would be specified on
the basis of non-degradation principles rather than close
linkages to present or projected water quality. A favorable
approach might be to focus upon the water quantity effects
-------
of urban development (see Section 4). In planning acti-
vities which deal with existing water quality conditions,
the highest priority would be placed on direct monitoring
and analysis of problems, rather than development of gen-
eralized loading relationships. Second highest priority
would be given to identification and analysis of specific
sources responsible for high pollutant loadings. In both
cases, analysis would focus upon long-term loadings and
problems unless there is a clear indication that water
quality is dominated by transient effects. Land management
controls which apply broadly to both existing and new urban
development would be designed on the basis of general infor-
mation, plus careful analysis of implementation feasibility,
rather than formal cost-effectiveness analysis. Further
details of this approach are presented in the next section.
10
-------
SECTION 3
CONCLUSIONS
Only fragmentary data are currently available describing the
contributions of unrecorded sources to water quality prob-
lems in U.S. urban areas. The present review of these data
has been limited to measurements of waterborne pollutant
yields from urban land areas which do not contain combined
sewers or recorded effluent discharges. The most striking
characteristic observed is the variability of pollutant
concentrations and loadings among areas studied. The impli-
cation is that literature values of these quantities should
be interpreted with extreme caution when related to any
particular area.
The high loading values frequently cited in the literature
may have created the general impression that urban runoff
inevitably causes serious water quality problems. This
conclusion could in fact turn out to be largely correct;
but, the problems which are critical may not be those which
have traditionally received most attention in the water
quality literature. Specifically, the present review indi-
cates that organic loadings from urban land without combined
sewers are not necessarily problematic—i.e., are not neces-
sarily sufficient to cause transient dissolved oxygen depletion,
Nutrient loadings from urban and suburban areas are often no
higher on a per-acre basis than loadings from agricultural
land. The importance of organic and nutrient loadings from
urban land must therefore be judged on a case-by-case basis.
A major question mark is the role of heavy metals loadings.
Although the present review suggests that there is very
great variation among urban areas with regard to these
loadings also, it is possible that even the cases where
loadings are low could involve serious water quality prob-
lems on a long-term basis.
An important possibility, which is strongly suggested but
not proven by the existing data, is that small urban basins
yield greater pollutant loadings, per acre of land, than
larger basins containing similar land uses. The implication
is that there is a tendency for pollutants in stormwater to
settle out in stream channels. On the one hand, this pos-
sibility means that pollutant loading estimates based upon
small-catchment data, or upon rates of surface pollutant
11
-------
accumulation, may overstate the contribution of urban land
to water quality problems at downstream points. On the
other hand, there may be a strong need to consider potential
problems created by deposited material, such as benthic
oxygen demand and accumulation of toxicants in aquatic food
chains. The destination of pollutants may be a very cri-
tical issue with respect to stormwater.
Relationships between pollutant loadings and watershed
characteristics have been estimated statistically in the
present study utilizing data for multiple urban watersheds
in two geographic areas. The strongest explanatory vari-
ables for most of the water constituents studied were em-
ployment density and the percent of watershed land rendered
impervious. Population density was generally not found to
be an important explanatory variable, unless older housing
was differentiated from new housing. A relatively low
influence was attributed to the latter. This indicates that
average loading values may seriously overstate the water
quality impacts of new urban development. Finally, indus-
trial land was found in two cases to constitute an influence
on pollutant loadings over and above the general effects
attributed to employment and impervious surfaces. The
general impression created by the analysis was that un-
recorded pollutant loadings tend to be more closely related
to economic activity than to residential population. Ex-
ceptions would be areas in which residential neighborhoods
are relatively old and/or are not served by separate sani-
tary sewers.
The problem of soil erosion from construction sites, and
resulting sedimentation of stream channels and other water
bodies, has been discussed extensively elsewhere and thus is
not reviewed here in detail. A related urban problem which
deserves emphasis is hydrographic modification. Construc-
tion of impervious surfaces and land drainage alterations
increases the quantity of storm runoff, which in turn causes
sediment production through the phenomenon of stream channel
enlargement. Hydrographic modification also involves direct
alteration of stream channels as part of the land develop-
ment process, which disrupts aquatic habitats and increases
discharge-related problems downstream. The net effects of
these factors can in some instances be more serious for
water use than the changes in water chemistry which accom-
pany urban development.
12
-------
With regard to unrecorded pollutant loadings generally,
existing data provide little or no direct evidence regarding
the typical importance of different source classes, i.e.,
the relative contributions of roadways, other impervious
surfaces, and urban sources which do not involve impervious
surfaces. There is reason to believe, however, that washoff
of diffuse materials from impervious surfaces is not always
the predominant source mechanism, and may not be responsible
for some of the higher loadings observed. One indication is
the finding, observed in the present study and elsewhere,
that pollutant loading rates during storms may bear little
relationship to time since the previous storm—as would be
expected if progressive accumulation of materials on imper-
vious surfaces were the primary source. (Nonlinearity of
accumulation rates would explain this finding partially, but
not entirely.) Another indication is that pollutant load-
ings tend to differ by greater amounts than would be ex-
pected if the major sources are factors which are widespread
in urban areas. Groups of watersheds can be observed in
which sources of street dirt such as vehicular traffic,
litter, atmospheric fallout, decaying vegetation, and pet
wastes should be present in roughly equal degrees, yet
pollutant loadings are strikingly different.
Some of the other pollutant sources which have been found
very important in particular basins are: (1) dumping of
liquid and solid waste on land surfaces; (2) sanitary sewer
leaks and bypasses; and (3) unauthorized discharge of liquid
waste to storm sewers and receiving waters. Two critical
aspects of these sources are that, first, they involve
localized, site-specific conditions rather than extensive
land surfaces; and second, they tend to be difficult to
identify on the basis of land use or other general land
data. The potential importance of pollutant sources with
these characteristics is commonly recognized in the case of
non-SRA sources (e.g., leachate from landfills, waste
lagoons, and on—site septic systems) but frequently does not
receive adequate consideration in stormwater analysis.
The available literature contains only fragmentary informa-
tion for assessing the impacts of stormwater pollutant load-
ings on receiving water quality, particularly impacts on
biologic communities. Some of the major issues are: short-
term toxic effects, significance of short-term oxygen de-
pletion, availability of nutrients in urban runoff to sup-
port algal growth, transport and deposition of particulate
13
-------
materials from urban land, benthic oxygen demand, and re-
lease mechanisms for materials stored in bottom sediments.
The findings from a review of relevant literature are dis-
cussed in the Technical Appendix.
These observations have led to the following general conclu-
sions regarding analysis and control of urban unrecorded
pollution.
1. It appears likely that, for a large proportion of
existing urbanized land in the U.S., the measures
utilized for abatement of unrecorded pollution
will consist only of land, management controls.
Implementation of remedial structural measures—
e.g., runoff storage and treatment options—may be
limited primarily to areas with sewer overflow
problems, and commercial and industrial areas
which produce especially high pollutant loadings.
2. In current studies, the estimates of loading re-
ductions achievable by land management controls
will generally be quite rough due to uncertainty
regarding the share of existing loadings that can
be addressed by each control* as well as uncer-
tainty about the practical efficiency of controls.
Thus, the use of formal simulation procedures to
evaluate stormwater controls may be unwarranted
except in cases where the use of structural con-
trol measures is anticipated.
3. Development of general pollutant-generation rela-
tionships for urban land uses can be useful for a
variety of planning tasks, including estimation of
pollutant inputs to major water bodies, and prepa-
ration of wasteload allotments. However, such
relationships can easily be misleading, and often
fail to convey sufficient information about source
mechanisms to facilitate the actual design of
•controls. In cases where planning resources are
limited, this activity should probably be assigned
lower priority than analysis of existing water
quality problems and identification of specific
pollutant sources in the study.area.
4. There is reason to believe that unrecorded pol-
lutant generation by urban land^-which -includes
14
-------
sources such as unauthorized discharges and sewer
leakage as well as surface runof£-.—is not at all
uniformly distributed across urban areas, even
when variation in land use is taken into account.
This suggests that considerable improvement could
be brought about by implementing controls on a
site-specific basis, once high-yield pollutant
sources are identified. A reasonable strategy in
many study areas may be to emphasize investigative
activities that would contribute to this goal.
Such activities would consist largely of field
reconnaissance and chemical monitoring rather than
analysis of land use patterns, since imagery and
published data typically fail to capture many of
the most important aspects of pollutant genera-
tion.
5. The choice of critical flow conditions for the
design of controls is a significant issue in water
quality management planning. For water bodies
affected by SRA pollutant sources, critical con-
ditions may be defined in terms of selected storms
as well as dry-weather flows. Given the general
difficulty of predicting and interpreting short-
term water quality phenomena, the primary emphasis
of planning studies should probably be placed upon
steady-state dnd long-term conditions unless there
are clear indications that transient problems are
of major importance. In any case, the target
water quality criteria for each set of critical
conditions should be established directly on the
basis of desired water use, and should not be
dependent upon the expected levels of water qual-
ity during other critical conditions. If this
principle is followed, the most stringent condi-
tions for design of municipal and industrial
treatment facilities will tend to consist of
extreme low flow, in a large proportion of cases.
Opportunities for trade-off between control of
recorded discharges and control of unrecorded
effluents will exist in these cases only to the
extent that unrecorded loadings affect water
quality at low flow.
6. New urban development can be handled somewhat
differently than existing development in water
planning studies due to the opportunities which
15
-------
exist for incorporating control measures in de-
velopment design, and due to the relative ease of
establishing private responsibility for pollutant
reduction. A feasible planning approach in many
instances is to treat new development as a wholly
or partly independent problem, and to design con-
trols on the basis of uniform non-degradation
principles rather than on predicted levels of
water quality. As discussed in the next section,
a helpful strategy may be to focus largely on the
problem of hydrographic modification.
Various aspects of this general approach are discussed fur-
ther in Section 7, following the examination of empirical
data in Section 5 and the evaluation of specific planning
methodologies in Section 6.
16
-------
SECTION 4
RECOMMENDATIONS
The specific planning methodologies recommended here deal
primarily with new urban development, but are also relevant
in some instances to existing development. Perhaps the most
significant element of these methodologies is the selection
of problems to be addressed. The position taken here, which
has been based in part on review of empirical data, is that
the water quality impacts of completed urban development
with sanitary sewerage can be controlled adequately in most
instances by focusing upon the problem of hydrographic
modification. That is, preventive measures dealing with
water quantity will ordinarily prove to be adequate controls
for water quality, assuming that reasonably high standards
of public cleanliness can be maintained. The types of
problems and associated control measures which have been
selected for emphasis are thus the following.
1. Control of erosion from construction sites. Con-
siderable experience already exists in the design
and application of erosion/sedimentation control
measures. These measures should deal with hydro-
graphic modification—i.e., increased runoff—
caused by construction activity as well as with
soil loss per se. Thus, some form of runoff
detention should be included, along with other
physical controls. An overall approach for quan-
titative evaluation of erosion/sedimentation con-
trols is outlined here in Section 11.
2. Control of the location, design, and operation
of on-site sewage disposal systems. On-lot waste
disposal systems, particularly domestic systems,
constitute a very serious water quality problem in
many areas. Control of these facilities has been
oriented primarily toward prevention of nuisances
and health hazards, rather than protection of
water quality. Implementation of measures needed
to achieve adequate groundwater and surface water
quality may therefore require redefinition of
existing regulatory functions, with or without
additional enabling legislation. The controls
17
-------
implemented should include the following: (1)
prevention of new on-site disposal systems in
areas where soil characteristics, land slope, or
proximity to receiving waters will preclude sat-
isfactory operation; (2) design of new systems in
order to assure adequate performance under the
given land conditions; and (3) maintenance of
performance standards for septic system operation.
Prohibitive soil characteristics for new on-site
systems could include overly rapid percolation,
which would result in groundwater pollution, as
well as overly slow percolation. Maintenance of
performance standards may require significant
intensification of monitoring activities relative
to present practices, which typically are limited
to investigation of nuisances reported by resi-
dents.
Construction of leakproof, accessible sanitary
sewers. Sanitary sewer leakage and bypasses
contribute significantly to water problems in many
areas. Therefore, in the construction of new
systems, best available technology should be
utilized wherever possible. Leakproof sewers are
now technically feasible, if manufacturers' speci-
fications are followed closely. Although extra
construction and monitoring costs are likely to be
involved, these costs would appear to represent a
very wise investment on the part of a community.
For some types of development, the most cost-
effective strategy for water quality control could
conceivably involve combined sewers; but separate
storm and sanitary sewers will be assumed here.
An important objective in such cases is to keep
the systems as separate as possible. Roof drains,
foundation drains, and other sources of water for
which treatment is unnecessary should never be
connected with sanitary sewers, since this in-
flates waste treatment costs and may lead to
overload of the system and creation of bypasses.
Conversely, new development should be designed so
as to discourage the use of storm sewers for
disposal of wastes requiring treatment. Finally,
a very important objective is to design new de-
velopment in such a way that sanitary sewers are
readily accessible for repairs. A favorable
design may involve location of sewers beneath
18
-------
sidewalks or grass strips bordering the roadway,
rather than under the roadway itself.
4. Control of hydrographic modification. The control
of hydrographic modification due to urban develop-
ment involves three aspects: (1) prevention of
increase in peak discharge; (2) prevention of
decrease in the base flow of streams and the rates
of aquifer recharge; and (3) protection of water
courses from encroachment and alteration. Con-
trols dealing with hydrographic modification
should be designed to maintain existing conditions
as closely as possible; zero impact is in most
cases a feasible goal. With regard to the first
two objectives just listed, the most cost-effec-
tive design for a given development project is
likely to involve a combination of control
measures. Runoff detention devices are generally
needed to prevent increase in peak discharge,
whereas base flow maintenance and aquifer recharge
may require infiltration devices (with careful
attention paid to possible groundwater pollution
due to these devices). Overall modifications of
development design, such as restriction of imper-
vious coverage, can be extremely useful, but
rarely are sufficient per se to attain zero
impact.
A critical aspect of focusing upon hydrographic modification
is that the controls utilized will have very substantial
water quality benefits. Temporary storage of storm runoff
in detention basins brings about significant reductions in
most pollutant loadings, due to settling out of particulate
materials. Infiltration devices tend to achieve much
greater pollutant reductions. For most types of new urban
development, it is felt that the decrease in pollutant
loadings thus achieved by control of hydrographic modifica-
tion is likely to be sufficient for water quality protec-
tion—assuming that other measures as specified below are
also implemented.
The use of water quantity as an explicit basis for design
and defense of water resources protection measures has a
number of distinct advantages, given the present level of
knowledge concerning unrecorded pollution. The problems
created by hydrographic modification, which include in-
creased flooding as well as channel disturbance and sediment
19
-------
yield, are directly observable and intuitively understand-
able to the layman. The feasibility of zero impact with
regard to water quantity effects means that appropriate
design parameters for control devices can be established
unambiguously, through reference to existing conditions.
Implementation of controls can be accomplished by adoption
of performance standards, which allow some freedom of ad-
justment, or alternatively by requiring specific control
measures on an areawide or site-specific basis. The per-
formance of controls once in place is relatively easy to
monitor. Finally, considerable experience already exists in
many communities regarding various aspects of runoff con-
trol.
A fact which has considerable practical significance is that
control of hydrographic modification does not necessarily
require land use control, since mitigative measures can be
utilized at almost any buidable site. The direct and indi-
rect costs of providing these measures, which are usually
borne by private builders, tend to be small relative to
total project cost. Thus, prevention of hydrographic modi-
fication will not prohibit most types of development at
most locations. Regardless of the need for land use con-
trol in U.S. communities, it is felt that most water re-
sources protection measures developed in current planning
programs should not be closely identified with land use
control, since such an identification might limit imple-
mentation to a relatively few areas. The major exceptions
would be measures dealing with location and construction
of new on-site septic systems, which might involve a sub-
stantial degree of de facto land use control.
It is anticipated that runoff control for a large proportion
of new development projects will involve the use of storm-
water detention facilities. This fact is important to the
overall strategy recommended here. As knowledge is gained
regarding the water quality impacts of urban runoff, it
could eventually be established that chemical treatment of
stormwater is generally needed. If so, the availability of
runoff detention facilities will place a community in a
favorable position for implementation of additional con-
trols, since the very high costs of providing runoff storage
capacity in existing developed areas will be at least par-
tially avoided.
Some other comments regarding stormwater detention facili-
ties are the following. First, impoundments constructed for
20
-------
runoff detention should be designed so that there is virtu-
ally zero chance of structural failure. Second, care must
be taken to see that nuisances are not created by these
facilities, such as insect problems and safety hazards.
Third, some degree of maintenance is required for all runoff
control devices. In the case of stormwater detention fa-
cilities, it is essential that outlets be kept clear and
that accumulated sediment be removed periodically. The
material which is removed can be utilized as fill in con-
struction projects, preferably at locations where it is not
in contact with surface waters or groundwater. Various
aspects of the design of detention facilities are discussed
in detail in Section 9.
Protection of stream channels and other water bodies from
direct physical alteration is considered a very important
element of the approach suggested here. The objectives
include not only preservation of aquatic habitats, and
retention of the natural capacity of stream channels to
dissipate flooding effects (i.e., to reduce flooding down-
stream through temporary storage of stormwaters), but also
protection of the role of headwater stream channels as
"sinks" for sediment and other pollutants. As indicated
elsewhere, this role of stream channels may be very signifi-
cant. Although accumulations of pollutants are generally
undesirable at any location, it is probably better for such
materials to be deposited in headwater alluvial sediments
than to affect ambient water quality at downstream points
where water use is generally most intensive. A number of
issues involving protection of watercourses (including the
question of how to define the surface water system subject
to protection) are discussed in Section 9. An important
point is that efforts in this regard should be coordinated,
where possible, with ongoing or prospective flood plain
management programs.
The planning methodologies and control alternatives empha-
sized here have been chosen because of their relevance
specifically to new urban development. The present discus-
sion should not be interpreted to mean that other measures,
which could be applied to both existing and new development,
are not necessary. The success of the approach suggested
here is in fact dependent upon the assumption that land
conditions and waste management practices can be maintained
at levels which are presently above average relative to U.S.
urban areas generally. Thus, adequate control of the water
21
-------
quality impacts of newly-developed land will require at
least three classes of actions other than those discussed
extensively here, namely: (1) "housekeeping" measures, such
as routine cleaning of streets, parking lots, catch basins,
and other areas where pollutants accumulate; (2) public
information programs to create awareness of water quality
problems and their causes; and (3) actions to prevent the
occurrence of site-specific pollutant sources. The last
element could include monitoring of a wide variety of waste
management practices and facilities. A potentially promis-
ing area of action, which could be included as a fifth
control category in the list presented above, is waste
management planning for new development projects which are
expected to constitute especially high-yield pollutant
sources. Possible examples are establishments engaged in
petroleum distribution and sales, and certain types of
manufacturing operations. An objective would be to make
waste generation and waste management practices subject to
review by local planning agencies in a fashion similar to
other aspects of development design. The outcome could
consist of design modifications, operating agreements, or
special water quality control devices, as needed to prevent
impacts on a site-specific basis.
A general review of planning methodologies for water quality
control is presented in Section 6. Sections 7 and 8 outline
overall strategies for dealing with existing and new urban
development in current planning efforts, with emphasis in
the latter case upon the mechanisms which can be utilized to
implement control measures. The specific planning methodol-
ogies selected for emphasis are presented in Sections 9, 10,
and 11. Extensive discussion of underlying technical issues
in water planning, particularly the response of receiving
waters to pollutant loads, is contained in the Technical
Appendix to this volume.
22
-------
SECTION 5
POLLUTANT GENERATION
Introduction
The present study has included a rather extensive review and
analysis of existing data pertaining to unrecorded pollutant
sources in urban areas. The scope of this review has been
limited to urban land influences on water quality other than
recorded effluent discharges (i.e., municipal and industrial
effluents) and combined sewer overflows. Agriculture, which
is occasionally found within urbanized areas, is not dis-
cussed. These restrictions have been considered necessary
in order to concentrate upon the factors most relevant to
the impacts of new urban development. The objective of the
review and analysis has been to consider the following
questions, as posed in the summary section:
1. What are the typical magnitudes of unrecorded
pollutant loadings from urban areas without
combined sewers?
2. To what extent does variation exist in these
loadings; and what are the important sources of
variation?
3. Is washoff of diffuse materials from streets and
other impervious surfaces usually the predominant
source of unrecorded pollution?
4. Do loadings from new urban development differ
systematically from average urban loadings?
5. To what extent is urban development, with separate
sewers, likely to result in transient water qual-
ity problems?
For a variety of reasons, it has been impossible to obtain
definitive answers to any of these questions; and some have
been addressed only by inference. The present section con-
tains a somewhat abbreviated discussion of the findings,
emphasizing the materials which have greatest relevance for
design of water planning studies. Additional description is
23
-------
contained in the Technical Appendix to this volume, which
deals at length with the empirical land-water relationships
estimated as part of this study.
Literature discussion of unrecorded pollution problems in
urban areas has focused primarily upon stormwater pollution,
i.e., upon surface runoff activated (SRA) pollutant sources.
Aside from combined sewer overflows, two classes of SRA
pollutant sources have typically received primary attention:
soil erosion from construction sites, and washoff of dirt
and dust from impervious surfaces, particularly streets.
The potential importance of soil erosion caused by construc-
tion activity, and the need for control of this problem, has
been well established in the literature (although sediment
loading magnitudes and water quality impacts tend to be
difficult to predict in individual cases).
With regard to urban sources other than construction acti-
vity, perhaps the most extensive body of empirical research
available deals with rates of pollutant deposition and
accumulation on street surfaces. The present review has not
dealt at length with this information, however, but instead
has focused upon observed loadings:of waterborne pollutants
at in-stream locations. The reasons for this approach,
other than a desire to avoid redundancy with existing pub-
lications, are that: (1) major questions exist regarding
the manner in which dirt and dust accumulation on land
surfaces relates to in-stream pollutant loadings and water
quality conditions; and (2) accumulation of materials on
impervious surfaces is not the only urban SRA pollutant
source (notwithstanding combined sewers).
The loading data under discussion pertain only to stream
points and storm sewer outfalls which are unaffected by
recorded effluent discharges, combined sewers, or agricul-
tural land. Much of the information utilized has been
derived from the following sources, each of which contains
data for multiple urban watersheds.
1. An ongoing program conducted jointly by the U.S.
Geological Survey and the City of Philadelphia has
involved monthly sampling of numerous watersheds
in and near Philadelphia since 1970 (see Radziul,
et al, 1975). Many of these watersheds were
chosen explicitly on the basis of land use; thus.
24
-------
the network provides a good representation of
suburban and urban land in the Philadelphia area
(excluding heavy industrial land and urban core
areas). Data for ten of these basins, ranging in
size from 1 to 21 square miles, have been utilized
in the present study.
2. Storm runoff from 15 small basins in Tulsa, Okla-
homa, was analyzed intensively in a study by AVCO
Corporation (AVCO, 1970). The basins ranged from
64 acres to 938 acres in size, and provided very
good coverage of the residential, commercial, and
industrial land uses typically found in a medium-
sized urban area.
3. Numerous suburban watersheds in Montgomery County,
Maryland, were monitored in a study which was
concerned primarily with biologic effects of urban
runoff (Ragan and Dietemann, 1975). This study
yielded only limited chemical data but contained
very important implications for urban land impact.
Further discussion of these and other data sources is con-
tained in the Technical Appendix.
Variability of Pollutant Loadings
The most striking overall feature of observed pollutant
concentrations and loadings is the variability of values
among study areas. This characteristic is illustrated here
for organic pollutants in Table 1 and for heavy metals in
Table 2.
Average BOD concentrations in storm runoff from urban basins
range from about 3 mg/1 to upwards of 30 mg/1, as shown in
the second column of Table 1. Annual loadings, in pounds
per acre of watershed area, may vary to a lesser extent than
concentrations, although Table 1 is somewhat deceptive in
this regard since annual loadings are unavailable for sev-
eral of the basins with high average concentrations. A
similar situation prevails in the case of COD, for which
average concentrations range from about 20 to well over 100.
Literature discussions of organic loadings in urban storm
runoff have referred primarily to the higher BOD and COD
concentrations shown in Table 1, such as the values for Des
25
-------
TABLE 1
ORGANIC POLLUTANT LOADINGS FROM URBAN AREAS WITH SEPARATE SANITARY SEWERS
BOD
COD
Population
Density
(persona/
acre)
Basins Less Than
2 Square Miles
Des Moines, Iowa —
(4 areas)
Washington, DC2 38
Cincinnati, Ohio3'4 8.9
Durham. NC5 9.5
Durham, KC6 6.0
Tulsa, OX7 (8 resi-
dential basins) 9.3
Tulsa, OK7 (7 non-
residential basins) 2.6
Philadelphia. PA, :
Area (3 basins) 15
Rev Jersey8 (2 basins) 4.8
Basins 2 to 10 Sg. Hi.
Castro Valley, CA9 11
Ann Arbor, MI10
Philadelphia, PA,
Area (4 basins) 7.0
Ball Run, VA12 7.5
Basins Here than 10 Sg. Mi.
Philadelphia, PA,
Area (2 basins) 8.0
Nontgonery Co., HD11
(19 stations in 6 bsns.) 2-12
* Estimated
References: 1. Henningson, 1973
. 2. Meston. 1970
3. tfeibel, 1969
4. Weibel, 1964
Mean Concentration Mean Concentration
(mg/1) Annual (mg/1) Annual
Loading Loading
Net Dry (pounds/ Wet Dry (pounds/
Conditions Conditions acre/yr.) Conditions Conditions acre/yr.)
36
19 — 335
19 53 99 347
84
75-90* 15 170 29
12 23 87 171
12 — - 31 101 234
5.3 2.3 22 24 10 115
3.1 , 0.8 15
u ' _ _
*< • «— «— ___ ___ •.—
28 ._
5.0 3.8 24 25 14 127
8.5 2.5 21
4.4 2.8 19 24 11 112
1.8 15* :
5. Bryan, 1970 • 9. Bydrologic Engineering, 1972
6. Colston. 1974 10. Burn, 1968
7. AVCO, 1970 11. Ragan, 1975
8. Nhipple,- et al, 1974 12. Randall, 1975
26
-------
Moines, Washington, Cincinnati, Durham, and Ann Arbor. As a
result, urban runoff has frequently been compared unfavor-
ably to secondary sewage effluent in terms of strength.
However, the data included in Table 1 for basins in Pennsyl-
vania, New Jersey, and Maryland indicate that concentrations
of organic material in urban runoff are not necessarily high
relative to sewage effluent or other standards of compari-
son. Available data also suggest that these organic load-
ings do not necessarily create dissolved oxygen problems.
For the 10 Pennsylvania basins, in which dissolved oxygen
has been sampled along with BOD and other water constit-
uents, only three instances have been observed in which DO
was below 7.0 mg/1 during a storm period. (The dissolved
oxygen standard applying to these streams is 4.0 mg/1.)
Similarly, for the Montgomery County, Maryland basins, the
mean of minimum DO values observed at the 19 stations
draining urbanized areas was 7.1 (Ragan and Dietemann, 1975,
p. 58). The comparable figure for basins with no urban
development was 6.9 mg/1. On the other hand, organic load-
ings have been shown to affect dissolved oxygen in the
Castro Valley, California, basin (Lager and Smith, 1974, p.
81) and would be expected to have this result in cases where
BOD concentrations are as high as in Des Moines, Durham, and
Ann Arbor.
The loading variation shown in Table 1 can be explained in
part by several circumstances. As indicated in the first
column of the table, the Washington, D.C., data pertain to a
densely urbanized basin (containing 38 persons per acre, or
24,000 persons per square mile), which would not be gen-
erally representative of urban land outside the core areas
of major cities. The Durham, North Carolina, data obtained
by Colston (1974)—which have been much publicized in the
literature—pertain largely to a slum area which is charac-
terized by dilapidated housing, a total lack of storm
sewerage, numerous unpaved streets, and extremely poor
environmental conditions in terms of trash and garbage
accumulation. The Durham data thus may be important in
demonstrating extreme conditions, but do not appear to be
generally relevant for urban areas in the U.S.
A factor which could be important for a wide variety of
pollutants is the possibility that loadings are system-
atically higher in small, totally-sewered catchments than at
downstream points in natural channels. This possibility was
raised by Ragan and Dietmann in attempting to explain the
relatively low BOD concentrations observed in Montgomery
County (1975, p. 61):
27
-------
"Many of the studies reporting high BOD's in urban
storm water runoff have been conducted in very small
watersheds. Most of the watersheds in the present
study were in excess of ten square miles. It is
probable that the mechanism of BOD transport is similar
to that of sediment transport and, therefore, the role
of watershed size must be considered ..."
Constituent losses during transport would involve settling
of particulate material and associated pollutants in natural
channels. This factor would explain somewhat the figures
shown in Table 1, which have been arranged according to
watershed size. It is relevant that the 5-square-mile Ann
Arbor basin, in which relatively high BOD concentrations
were observed, was totally storm-sewered, so that losses of
material during transport would presumably be minimal.
Variation in pollutant loadings from urban land is also
related to the types of land development and activity
present, as might be expected. The relationships which have
been estimated as part of the present study are summarized
in the next sub-section.
Relatively less information is available for other pollu-
tants besides organics for urban basins such as considered
here. Table 2 presents a comparison of average heavy metal
concentrations observed in four areas, in this case, the
most striking variation is between the concentrations ob-
served in the ten Pennsylvania basins and the much higher
concentrations observed in other areas. For each of the
metals considered, the average Pennsylvania.concentration is
lower than all other concentrations by a factor of between 3
and 25. These differences could be explained partially by
the special characteristics of the Durham basin mentioned
earlier, and by the phenomenon of pollutant losses during
transport (which would affect the Pennsylvania basins but
not the New York or Lodi basins). Land use intensity is
probably also a factor. In any case, the differences are
significant in view of the fact that the Pennsylvania
basins are considered to be broadly representative of much
of the urban and suburban land presently found in the U.S.
(Further discussion of heavy metals is presented below and
in the Technical Appendix.)
The purpose of this discussion has been simply to indicate
that data from the literature should be interpreted with
considerable caution, and generally cannot be used to infer
the seriousness of problems in a particular area.
28
-------
TABLE 2
COMPARISON OP AVERAGE HEAVY METAL CONCENTRATIONS IN URBAN STORM RUNOFF (mg/1)
Location
Metal
Lead
Zinc
Copper
Chromium
Nickel
Durham ,
North Carolina
0.46
0.36
0.15
0.23
0.15
New York,
New York
1.6
0.46
0.16
0.15
Lodi -
New Jersy
0.90
0.62
0.15
0.03
0.08
Philadelphia4
Pennsylvania
0.035
0.12
0.003
0.01
0.019
Colston, 1974.
2Klein, et al, 1974.
3Wilber & Hunter, 1975.
^Average for 10 basins (wet days); see text and Technical Appendix.
-------
Relationships between Pollutant Loadings and Watershed
Characteristics
Use of formal statistical methods to estimate relationships
between pollutant loadings and watershed characteristics has
been limited in the present study by the fragmentary data
available and the need to control influences due to geo-
graphic location. The sample cases chosen for analysis
consisted of the 10 Pennsylvania basins and 13 of the 15
Oklahoma basins mentioned previously in this Section, plus 3
watersheds in New Jersey (Whipple, et al^ 1974). The quan-
tities analyzed were the annual loadings of various water
constituents in pounds per acre per year. The specific
constituents were: BOD, COD, total organic carbon, sus-
pended solids, organic Kjeldahl nitrogen, ammonia, nitrate,
soluble orthophosphate, and total phosphate. The watershed
characteristics utilized as explanatory variables in the
analysis were the following:
PI - Population density in persons per acre
P2 - Density of population in dwellings constructed
before 1940, in persons per acre
P3 - Density of population in dwellings constructed
after 1940, in persons per acre
i
M - Median family income (as reported in 1970 Census
for the year 1969) :•
:E - Employment density in persons per acre
I - Impervious surface as percent of watershed area
All of the density measures were gross rather than net
density, i.e., consisted of population or employment divided
by the acreage of the entire watershed. Due to problems of
data availability, areal measurements of land use were not
utilized in the analysis, except as a basis for forming sub-
samples of watersheds. Given the constraints of the analy-
sis and the results obtained, it is felt that further con-
sideration of land use variables wquld not have added sig-
nificantly to the explanation of loadings. (Further, discus-
sion of the form of the analysis and the derivation of
variables is contained in the Technical Appendix.) The
analysis involved simple and multiple regressions in which
dependent variables expressing pollutant loadings were
related to the above factors as independent variables.
30
-------
The findings of the analysis can be summarized as follows..
Strong relationships with watershed characteristics were «
identified'for almost all of the chemical loadings analyzed.
The best explanatory variable in almost all cases was em- <
ployment density. Percent imperviousness, which includes
the effects of both residential and non-residential land
uses, was also statistically significant in explaining the
observed loadings of a majority of constituents. Population
density was generally not found to be a good predictor of
pollutant loadings, except when residential basins were
segregated and separate consideration was given to popula-
tion in pre-1940 housing. The effects then attributed to
pre-1940 housing were much greater than the loadings attri-
buted to population in post-1940 housing. In the case of
nitrate, loadings were shown to be highly sensitive to the
presence of dwellings with on-site sewage disposal (utiliz-
ing data from another study by the present author). Fin-
ally, in two cases, the existence of industrial development
was found to be highly significant, over and above the
general importance attributed to employment and impervious-
ness.
The relationships obtained, which pertain largely to employ-
ment density and percent of impervious land, are presented
numerically and graphically in the Technical Appendix.
Overall, the results for different water constituents are
found to be remarkably consistent. On the basis of various
conversion factors, the equations appear to indicate that
each additional employee in an urban watershed increases
pollutant yields by roughly four times as much as each
additional resident. The predominant cause of unrecorded
pollutants in urban areas would thus be economic activity
rather than population (although these two factors are, of
course, inseparable for an urban area as a whole). This
finding may have been influenced, however, by the fact that
the sample watersheds studied here contained a somewhat
higher proportion of new residential development than would
be generally true for the U.S.
The importance of age of residential development as an
explanatory factor is considered highly significant. Dif-
ferences between the pollutant yields from new and old
development may be due to' associations between age of hous-
ing and various socioeconomic characteristics, which are 'in
turn correlated with factors directly affecting pollutant
31
-------
generation. Some other possibilities are: physical deteri-
oration of streets and buildings; correlations with air
quality; and especially, the condition of sanitary sewer
systems. In any case, an important implication is that
loading estimates obtained from the literature or from
monitoring of existing urban land may systematically over-
state the water quality impacts of new urban development.
A significant unanswered question is whether the loading
differences between new and old development are due strictly
to former construction practices, or whether the loadings
from new development can be expected to increase progres-
sively over time.
Although the estimated equations were reasonably accurate in
predicting pollutant loadings for the sample basins studied,
the extent of variation among land areas which is implied by
these relationships is not sufficient to explain the dif-
ferences which exist between loading values reported in the
literature. This circumstance is thought to reflect the
extent to which pollutant generation is not systematically
related to overall characteristics of urban land. As is
discussed below, a significant proportion of pollutant
loadings produced during both storm and nonstorm periods may
involve localized site-specific sources which cannot be
identified readily through land use analysis.
Nutrients
Urban land is frequently cited as an important source of
nutrients, due to factors such as lawn fertilizer applica-
tion, atmospheric fallout, domestic animals, erosion from
construction sites, and sewer leakage. In order to provide
some perspective on this problem. Table 3 presents areal
loadings of nitrogen and phosphorus, as N and P, obtained
from the literature and from other sources cited earlier.
The available data for urban basins with separate sanitary
sewer systems are fragmentary but fairly consistent. Load-
ings of total Kjeldahl nitrogen (ammonia plus organic nitro-
gen) range between perhaps 1 and 5 pounds per acre per year.
For both organic nitrogen and ammonia, loadings from com-
mercial and industrial areas appear to be higher on the
average than loadings from residential areas. Nitrate
yields, as N, typically range between 5 and 15 pounds per
acre per year. An important fact not illustrated in Table 3
is that nitrate yields may be extremely sensitive to the
presence of on-site septic systems (Howard and Hammer, 1973;
see also the Technical Appendix).
32
-------
TABLE 3
NUTRIENT IXJADINGS IN POUNDS/ACRE/YEAR
Phosphorus
U>
U)
URBAN LAND
Cincinnati, Ohio1
Ann Arbor, MI1
.Washington, DC2
Roanoke Basin, VA
Rock Creek, DC/Maryland
Philadelphia Area
(6 residential basina)
Philadelphia Area
(3 non-residential basins)
Tulsa, OK
(8 residential basins)
Tulsa, OK.
(7 non-residential basins)
Typical Urban Loadings*
SECONDARY MUNICIPAL EFFLUENT**
AGRICULTURAL LAND
Brandywine Creek, PA
Agricultural Basins
Cropland, U.S.1
Peedlot Runoff1
Total Soluble Total
OKN NH3-N TKN KOj-N „ OP04-P p
9.9
1.3 0.3 1.1
16.7 1.8
4.8
3.2 13.5 1.8
2.2 11.8 1.1
2.8 7.0 1.3
1.5 0.6
2.4 1-0
10-20 1-2
25-35 5 35-40 10-17
5 0.6
0.4 3 0.15-0.35
0.1-14 0.07-3.3
100-1800 10-3.3
Notes: 'Figures apply to medium-density urban development with sanitary sewers in
reasonably good condition.
"Loading rates due to municipal discharge assume 7 persons per acre and 100 gpd
wastewater generation.
References: (1) Loehr, 1974 (2) Jaworski, 1970 (3) Gizzard & Henelle, 1972
(4) Coughlin t Hammer, 1973
-------
Loadings of soluble orthophosphate, as P, from completed
urban development are typically between 0 and 1.5 lb/acre/
year; and total phosphorus is usually less than 2 lb/acre/
year unless there is construction activity or sanitary sewer
leakage. More than perhaps any other water quality param-
eter r phosphorus loadings are sensitive to the type and
condition of sewerage facilities (believed to be the factor
responsible for the high loading observed in the Roanoke
Basin), If sewers are tight and no major industrial sources
are present, phosphorus concentrations in streams draining
urban land can be less than 0.3 mg/1 during both wet and dry
weather. The other critical factor is the existence of
construction activity, which can result in very high phos-
phorus loadings. In contrast to nitrate, on-site sewage
disposal may not be a major source of phosphate in areas
where soils are favorable (Howard and Hammer, 1973).
Loadings of 10 to 20 Ib/acre/year for nitrogen and 1 to 2
Ib/acre/year for phosphorus are cited in Table 3 as typical
for medium-density urban development with sanitary sewers in
reasonably good condition. These figures are contrasted
with nutrient loadings due to secondary municipal effluent,
which assume a gross population density of 7 persons per
acre (substantially lower than the density for many of the
urban basins considered). A significant fact is that load-
ings of phosphorus and ammonia nitrogen due to municipal
effluent can exceed the loadings due to land drainage by a
factor of 10. The purpose of this comparison is simply to
indicate that sewer system maintenance and strict control of
construction activity may be adequate for control of nutri-
ents from urban land, even in cases where problems of nutri-
ent enrichment require advanced treatment of recorded ef-
fluents .
Comparisons with nutrient yields from agricultural land,
shown in the lower portion of Table 3, are also instructive.
As an example, rural land in the Brandywine Basin—of which
only about 50% is intensively used for agriculture—yields
about 5 pounds per acre per year of nitrate nitrogen, and
0..6 Ib/acre/year of phosphorus. Thus, the nutrient yields
from agricultural land per acre can be on the same order of
magnitude as loadings from urban land. These figures sug-
gest that in watersheds containing large agricultural
regions, the nutrient yields from completed urban develop-
ment with separate sanitary sewerage may not be a major
issue.
34
-------
Heavy Metals
Over the past few years there has been an increasing inter-
est in the contamination of receiving waters with heavy
metals such as mercury, lead, cadmium, chromium, and zinc,
due to their potentially high toxicity. A particularly
significant fact is that heavy metals are non-degradable and
hence persist in the environment for extended periods of
time. In addition, heavy metals tend to precipitate out of
solutions with relatively neutral pH values and some alka-
linity; and they may be adsorbed on clay particles or bound
by such compounds as the hydrous oxides of iron and man-
ganese. As a result, these materials are concentrated in
the solid phases of water systems. Even though the water
itself may contain only small amounts of these materials,
the particulate matter in the water, and especially the
benthic deposits, may contain considerable quantities.
Because metals are conservative they may undergo biological
magnification in the food chain, reaching concentrations in
the upper trophic levels several orders of magnitude greater
than those which originally existed in the water. Finally,
depending on specific environmental conditions, certain
metals such as mercury may undergo microbiological trans-
formations to forms exhibiting significantly more toxicity
than original forms.
Heavy metals are known to enter receiving waters from a
Variety of sources; but little information is available
regarding the typical loadings of these materials in urban
stormwater. The variability of heavy metals concentrations
among urban areas has already been suggested in Table 2.
The heavy metals data for the abovementioned Philadelphia
area watersheds (which are presented in detail in the Tech-
nical Appendix) are considered particularly significant,
since they are among the few existing examples of informa-
tion describing in-stream metals loadings due entirely to
urban unrecorded sources—in this case largely non-indus-
trial sources. As noted earlier, the metals concentrations
observed in the ten Philadelphia area basins are relatively
low. For example, the average lead concentrations for all
basins during both wet and dry conditions are below 0.05
mg/1; and less than 10% of observed values exceed 0.1 mg/1.
In contrast, the lead concentrations reported in other
studies and computed by indirect methods may exceed 0.5 mg/1
(see Table 2). This discrepancy could be indicative of a
strong tendency for heavy metals to settle out in stream
channels rather than passing through.
35
-------
TABLE 4
SUMMARY OF HEAVY METAL DATA FROM PHILADELPHIA WATERSHEDS
Number of Metals for Which
Criteria were Violated (11
Dry Weather '• Wet Weather
Station
1
2
3
4
5
6
u>
7
8
9
10
Max
6
6
5
6
5
7
6
6
6
4
Avg
4
3
3
3
2
2 .
3
2
1
4
Max
5
4
5
3
8
6
5
5
6
4
Avg
' 4 ':
4
4
3
4
3
3
3
3
3
Metal
«__•» .
Cadmium (Cd)
Chromium (Cr)
Copper (Cu)
, Iron (Fe)
Lead (Pb) (2)
Manganese (Mn)
(21
Nickel (Ni) v*'
Silver (Ag)
Zinc
-------
In any case, it appears possible that the contributions of
ubiquitous urban factors such as traffic and litter to in-
stream heavy metal concentrations may actually be fairly
low, and that high concentrations tend to reflect site-
specific factors, to perhaps a greater extent than is true
for other water pollutants. As an example, the Philadelphia
area watersheds as a whole contain more than 50 square miles
of land and over 300,000 people; yet well over a third of
the total lead yield appears to be coming from a single 5-
square-mile basin containing less than 20,000 people.
Another interesting aspect of the Philadelphia data is that
average wet-weather concentrations fail to exceed average
dry-weather concentrations for a large proportion of the
metals monitored. A similar finding was observed in Durham
(Colston, 1974). Non-SRA pollutant sources are usually not
mentioned in connection with heavy metals, but may be re-
sponsible for a significant proportion of total in-stream
loadings.
Relatively little definitive knowledge exists concerning the
long-term impacts of heavy metals on aquatic biota; and even
less is known about the possible shockloading effects of
temporarily high metals concentrations. A very important
point, however, is that even ambient concentrations on the
order observed in the Philadelphia basins may be serious.
In Table 4, these concentrations are related to the criteria
proposed by EPA for protection of agricultural water use,
aquatic life, and public water supply. The most stringent
criterion for each metal, expressed as a maximum concen-
tration, has been selected; and violation of this criterion
by either the maximum or the average concentration for a
given basin (station) has been noted. Table 4 tabulates the
number of metals for which each station is in violation, and
the number of stations in violation for each metal. It is
found that maximum concentrations violate the most stringent
criteria in about 60% of possible cases, and that average
concentrations violate criteria in 30% to 40% of possible
cases, for both wet and dry conditions. As shown in the
right-hand side of the table, the preponderance of viola-
tions occur for iron, manganese, and zinc, plus copper in
the case of violations by maximum concentrations. The most
stringent criteria for iron and manganese are based on
public water supply, whereas the zinc and copper violations
are related to aquatic life.
37
-------
Finally, the significance of heavy metal buildup in stream
channels and other benthic deposits is not well known, due
in part to lack of knowledge regarding the recycling pro-
cesses which affect bottom sediments generally. This aspect
of the heavy metals problem could well prove to be the most
serious, especially in cases such as the Philadelphia
basins where ambient concentrations are generally low.
Sediment and Hydrographic Modification
Sediment loadings in stormwater occur due to detachment and
transport of soil particles from earth surfaces, including
gullies and stream channels, and washing of particulate
matter from impervious surfaces. The rate at which sediment
is yielded from an earth surface is highly dependent upon
transient hydrologic conditions and upon the specific nature
of the surface. In addition, the loading produced at any
downstream point is governed by complex transport processes.
The problems caused by sediment loadings involve three
factors: (1) the role of particulate matter as a medium for
transport of chemical pollutants in storm runoff; (2) the
direct impacts of suspended solids per se upon water use;
and (3) the impacts produced by settling of particulate
material in stream channels, impoundments, and other water
bodies. The first of these factors tends to be of greatest
importance for particulate materials yielded by impervious
surfaces and other stable urban land, since the quantity of
these solids is generally less critical than their chemical
composition. The most significant sources of bulk sediment
in many urban areas are construction activity and erosion of
watercourses due to hydrographic modification.
Sediment from construction sites is a widely recognized
problem which has already been addressed by legislation in a
number of states. The magnitude of erosion which can occur
as a result of exposure and disturbance of the land in
construction sites is extremely great. Without control
measures, the loading rate of suspended solids during con-
struction can be from 10 to 100 times as great on a per-acre
basis as the loading rate from the land before development.
Figure 3 shows the pattern of sediment yields which occurred
during development of a 58-acre basin in suburban Maryland.
Figure 4 compares the sediment yields observed in a number
of urbanizing watersheds in Maryland and Virginia with data
for nearby rural basins. Typically, much of the material
38
-------
a 100,000
a
i
J, 60,000
LI
w 40,000
i
g
S
z
O
Of
t-
Ul
U
O
U
20,000
10,000
6000
4000
2000
1000
BEGININGOF
CONST
O Observed
Computec
MAX
oj'
MUM
CONSTRUCTION
ACTIVITY
END C
CONST RU
0
ITIONo
1957
1958
1959
YEAR
I960
1961
1962
Source: Guy, H.P. U.S. Department of Agriculture
Misc. Pub. 970, 1963
Figure 3 SEDIMENT CONCENTRATION IN
RUNOFF FROM AN URBANIZING BASIN IN MARYLAND 1957-1962
tc
1
O
vt
DO
00
UU
00
00
00
Ho
60
40
20
10
8
6
4
2
1
A
•
•
O
Urb
Nat
mil
iral
ng V
Wa
C
•
K
:"•-
c
h
ISTRl
t i«d
i
CTK
t
N
!
-
_.
ES
K1
--••.
'
ION
- (^
r
•CON
_^_
i
>IKU
O
1 "'" "
:Tlo^
>:-
LA
U
O
^PJ
^ MM
02 .005.01 .02.0456.08.1 .2 .4.6.8 2 4 6 810 20 50 100200400
DRAINAGE AREAfSquart Miles)
Source;
American Society of Civil Engineers,
"Urban Runoff Quantity & Quality,"
Wra. Whipple, ed., August 1974
Figure 4 SEDIMENT YIELDS FROM CONSTRUCTION SITES
AND NON-CONSTRUCTION LANDS IN MARYLAND AND VIRGINIA
39
-------
which is eroded from construction sites settles out on the
bottoms and banks of nearby stream channels, thus destroying
aquatic habitats and affecting water use in other ways.
A number of related effects are caused by hydrographic
modification. As associated with typical urban development,
hydrographic modification involves three aspects: (1)
covering of pervious soil with surfaces such as pavement and
structures which are impervious to rainwater; (2) alteration
of land drainage (e.g., construction of storm sewers in
conjunction with streets); and (3) direct alteration of
stream channels, in order to deal with flooding problems and
make land available for development. Creation of impervious
surfaces increases the quantity of direct runoff produced by
storm events; and all three aspects tend to speed the move-
ment of storm runoff through the surface water system. As a
result, the peak rates of stream discharge are increased,
often by a factor of 2 or 3 relative to pre-urbanization
conditions. One consequence, which is discussed at greater
length in Section 9, is a general process of erosion and
enlargement of watercourses, ranging from gully erosion at
headwater locations to gradual increase in the size of major
stream channels. This process involves significant yields
of sediment, which add to channel disruption and other
problems at downstream points.
The net effects of hydrographic modification are thus:
direct disruption of stream channels; indirect disruption
due to changes in the streamflow regime; and sediment
yields resulting from channel enlargement. These condi-
tions, along with sediment yield from construction sites,
are likely to have very important impacts on aquatic life.
In the study of Montgomery County, Maryland, streams by
Ragan and Dietemann (1975), changes in water chemistry due
to urbanization were found insufficient to explain the
marked shifts in fish species distribution which had occur-
red. Primary responsibility was assigned instead to stream-
flow and sediment as follows (p. 60):
"Changes in flow regime and sediment load have had a
dramatic impact on the stream system. A number of
studies reported in the literature have shown that
urbanization results in a substantial enlargement of
the stream cross-section. A survey showed that Paint
Branch, a tributary of the Anacostia, averaged nine
feet in 1948, but now has an average width of 37 feet.
Much of the gravel-bottomed channel delineated in 1948
40
-------
is now covered with silt, and bank erosion has des-
troyed much of the shelter for fish remaining in the
'unimproved1 sections of the streams.
"...A 'river walk' survey conducted as part of the
study revealed that approximately 25 percent of the
stream length in the Anacostia watershed was either
channelized or included substantial construction aimed
at bank stabilization. It is believed that the shifts
in fish species in the urbanized areas reflect these
changes in habitat. Because of the impact that in-
creased rates of runoff have on the streams, on-site
detention in storage is a major need in the study area
and probably in most other urbanizing watersheds."
Prevention of hydrographic modification, which involves both
runoff control and prevention of direct channel alteration,
should thus be considered an essential element of stormwater
planning, along with control of sediment from construction
sites. As implied elsewhere, an important issue is that
stream channel phenomena are relevant to loadings of storm-
water pollutants generally, due to,the role of natural
channels as a "sink" for pollutants as well as a source of
sediment. In spite of the fact that a large proportion of
urban unrecorded pollutant yields are known to involve
particulate matter (including chemical pollutants adsorbed
onto suspended particles during transport), stormwater
planning and research efforts do not appear to be well
integrated with ongoing work by hydrologists in sediment
transport and river mechanics. Despite the complexity of
these subjects, agencies should at least recognize the
general importance of channel-related effects, in the design
of analytical studies and especially in the formulation of
recommended plans for new urban development.
A final comment regarding hydrographic modification is that
construction of impervious surfaces tends to reduce stream-
flow during dry weather, as well as to increase wet-weather
flows (although this effect can be offset by imported
water, as discussed in Chapter 10). Effects upon base flow
may be important to the level of treatment required for
recorded effluents, since treatment levels may be. sensitive
to the amount of dilution received during extreme dry con-
ditions. Hydrographic modification should thus be con-
sidered in all aspects of water quality planning; and
41
-------
controls for new urban development should include measures
to assure that infiltration and dry-weather flows will be
sustained.
Application of Dirt and Dust Accumulation Data
Perhaps the richest available source of information per-
taining to urban unrecorded pollution is the literature
dealing with dirt and dust accumulation on street surfaces
(APWA, 1969; Sartor and Boyd, 1972; Shaheen, 1975). Ex-
tensive measurements of contaminant accumulations have been
conducted; and methodologies have been developed for utiliz-
ing these data to estimate stormwater pollutant loadings
(Amy and Pitt, 1974; McElroy, et aly 1975). Several impor-
tant issues regarding the use of these data are discussed in
the following paragraphs.
1. Nonlinearity of Relationships Between Pollutant
Accumulation and Time. Methodologies for estima-
tion of stormwater pollutant loadings on the basis
of dirt and dust accumulation commonly assume that
the amount of material present on impervious
surfaces increases at a constant daily rate over
time. (Examples are the STORM model, as utilized
up until the present writing, and the Midwest
Research Institute methodology discussed below.)
However, the recent study by Shaheen (1975) of
pollutant accumulation on roadways due to vehic-
ular traffic demonstrates that non-linear rela-
tionships may prevail, i.e., that the amount of
material may approach a limiting value rather than
increasing indefinitely between runoff and street
sweeping events. Shaheen thus distinguishes
between deposition and accumulation of materials
on street surfaces, as follows (page 49):
"Note that, although the deposition of traffic-
related materials occurs at a constant rate, the
accumulation of materials along the roadway tends
to level off after some period of time due, in
part, to traffic-related removal mechanisms—
However, all of the deposited pollutants are
available for transport to receiving waters during
storms and the deposition rates are valid esti-
mates of the contributions of motor vehicles to
water pollution."
42
-------
.The second statement indicates that, since pol-
lutants removed from street surfaces may still be
available for transport by stormwater, it may be
appropriate to utilize deposition rates rather
than accumulation rates for water planning pur-
poses. This conclusion would not appear valid in
all cases, however, since materials transferred to
pervious surfaces might reach receiving waters
only after long intervals (which would allow time
for chemical stabilization), or not at all. The
issue of linear versus nonlinear accumulation
rates tjhus remains critical, especially when dirt
and dust data are utilized to estimate the pollu-
tant loadings produced by storms which follow long
periods of dry weather.
2. Delivery of Washoff Pollutants to Downstream
Points. The possibility has been suggested that a
major proportion of pollutant loadings in storm-
water are deposited in stream channels, rather
than transported to downstream receiving waters.
This possibility would not necessarily detract
from the seriousness of pollutant loadings; in
fact, the accumulation of materials in stream
channels might result in more critical problems
than the conditions created by stormwater pollu-
tants while in transit. However, it is important
to note that this factor may cause loading esti-
mates based on dirt and dust data (or on pollutant
yields from small catchments) to overstate the
seriousness of problems at downstream points, such
as problems involving transient dissolved oxygen
depletion. M..
3. Omission of Other Stormwater Pollutant Sources.
As discussed elsewhere, washoff of materials from
roadways and other impervious surfaces is not the
only source of stormwater pollutant yields from
.urban land, and in at least some cases is not the
!predominant source. This fact should be kept in
mind when utilizing dirt and dust data.
i
The potential importance of the first two of these issues
can be illustrated by comparing, for a number of sample
areas, the estimated pollutant accumulations on watershed
surfaces with the pollutant loadings observed in-stream.
Pollutant accumulations on impervious surfaces have been
43
-------
estimated using a methodology developed by the Midwest
Research Institute (McElroy, et al, 1975, pages 186-198),
which is based upon information from the URS studies (Sartor
and Boyd, 1972; Amy and Pitt, 1974). The urban watersheds
considered are the Pennsylvania, New Jersey, and Oklahoma
basins discussed earlier (which range in size from 0.1 to 21
square miles, and contain no recorded discharges or combined
sewers). Curb length density, required for computation of
solids loading rates, has been estimated on the basis of
population density for the Pennsylvania and New Jersey
basins, using a methodology developed by the American Public
Works Association (McElroy, et al, 1975, p. 194).
The first column in Table 5 pertains to the estimated annual
deposition of various pollutants on impervious watershed
surfaces. Although the Midwest methodology may be intended
primarily for analysis of short-term situations, the compu-
tation of annual deposition rates should be no less valid
than computation of short-term buildup of materials, since
the Midwest methodology assumes that buildup occurs at a
constant daily rate. (That is, accumulation is not dis-
tinguished from deposition in this methodology.) The second
column in Table 5 pertains to the pollutant loadings ob-
served in-stream, during all periods in a typical year when
surface runoff is present. In each column of the table,
only the range of values obtained for individual watersheds
is listed.
For the Pennsylvania and New Jersey basins, the estimated
rates of pollutant deposition on impervious surfaces tend to
be strikingly higher than the in-stream loadings observed
during storm periods. In each of four cases—BOD, COD,
total phosphate (PO4T) , and ammonia (NH3)—the range of
deposition rates does not overlap with the range of in-
stream loading rates, and runs higher than the latter by a
factor of about 8. The ratios of deposition rates to in-
stream loading rates for individual watersheds range from
about 4 to 25. On the other hand, the estimated deposition
rates for nitrate (NO3) are significantly lower than the in-
stream loading rates. For the Oklahoma basins, the ranges
are much more similar. However, the deposition rates for
individual basins still tend to be consistently higher than
the in-stream loading rates. The former typically exceed
the latter by a factor of 2 or greater for BOD and soluble
orthophosphate (PO4O), and 1.5 or greater for COD.
44
-------
TABLE 5
COMPARISON OF SURFACE POLLUTANT ACCUMULATION RATES
WITH IN-STREAM LOADING RATES
Loadings in Pounds/Acre/Year
Pennsylvania/New Jerse
(12 Basins)
BOD5
COD
P04T
N03
NH0
Estimated Accumulation on
Watershed Surfaces*
y
115
851
16.8
4.5
15.2
- 162
- 1139
- 23.8
- 6.4
- 21.5
Observed In-Stream Loading
(Wet Conditions)
9
42
0.7
12
1.1
- 32
- 159
- 4.9
- 39
- 4.4
Tulsa, Oklahoma
(15 Basins)
BOD5
COD
P040
17.3 - 138
83 - 663
1.3 - 10.6
12 - 48
60 - 470
1.1 - 8.0
* Estimated using methodology developed by Midwest Research Institute
(McElroy, et al» 1975)
-------
These discrepancies could be due to the linearity assumption
in the Midwest methodology, to settling of materials in
stream channels, or.to other factors. The figures are cited
only to suggest that simplistic application of dirt and dust
data to predict in-stream pollutant loadings can potentially
result in very large errors, for either short-term or long-
term loadings.
Another important point is that the availability of pol-
lutant deposition data should not draw attention away from
the need to consider hydrologic conditions when evaluating
transient phenomena. In order to illustrate this issue, it
is worthwhile to examine in some detail a table contained in
the Conclusions section of the Shaheen report (1975, page
5). This table is reproduced here in its entirety as Table
6. For each of a variety of pollutants, Shaheen has esti-
mated the daily mass flow rate per person due to secondary
sewage effluent, and the daily rate of deposition on street
surfaces due to vehicular traffic, also on a per capita
basis. These figures are presented in the third and fourth
columns of Table 6, respectively. In the fifth column,
Shaheen considers a situation in which three days' accumu-
lation of street surface materials are delivered to surface
waters by a two-hour storm runoff event. The relative
contributions of sewage effluent and street surface material
during this two-hour period are compared by multiplying the
latter by 36 (equal to 72 hours divided by 2 hours) and
forming a ratio to the sewage flow rate. Traffic is shown
to be a much more important source of pollutant loadings
during the two-hour period than sewage effluent, for all
constituents except BOD, phosphorus, and Kjeldahl nitrogen.
Shaheen thus concludes (page 4):
"Traffic-related deposits by themselves would... con-
stitute a significant source of pollution on a shock-
load basis for each parameter listed; thus the im-
portance of traffic contributions to urban water pol-
lution is established."
Although Shaheen*s computations are valid, the above con-
clusion may be misleading, in that the importance of "shock-
loading" per se is not established. A two-hour storm
period in which all materials on street surfaces are de-
livered to receiving waters would involve a substantial
volume of runoff; thus, stream discharge could be many times
46
-------
TABLE 6
SIGNIFICANCE OF RUNOFF.FROM TRAFFIC-RELATED ROADWAY DEPOSITS TO
URBAN WATER POLLUTION
(Comparison with Secondary Sewage Treatment Plant Effluent)
Parameter
Suspended SiIds
BOD
COD
Kjeldahl-N
Phosphate-P
Lead
Zinc
Copper
Nickel
Chromium
Sewage
Raw
Tig/T)
235
140
200
30
10
-
_
-
-
Composition1 '
Final
Effluent
(mg/1)
24
14
20
3
7
0.03
0.08
0.03
0.01
Average Per Capita Haas Plow Rates
Final Effluent(b)
(g/cap-day)~
0.01
9.08
5.30
7.57
1.14
2.64
0.011
0.030
0.011
0.004
0.004
Traffic-Related
Depositions(c)
(g/cap-day)
26.3
0.06
1.41
0.004
0.016
0.31
0.039
0.003
0.005
0.002
Traffic{d)
Impact Ratio _
(Traffic/Effluent)
104
0.41
6.7
0.13
0.22
1015
47
9.8
45
18
(a) Estimates of raw sewage and final effluent concentrations are for separate domestic sewage and
have been derived from Fair and Geyer (4), EPA's manual on phosphorus removal (5) and a recent
publication on elemental analysis of waatewater sludges (6).
(b) Average per capita flow rates of pollutants in, final effluent have been calculated assuming a
per capita flow of 100 gallons of sewage per day.
(c)
Average per capita depositions of traffic-related pollutants available in urban stormwater run-
Off have been calculated assuming a per capita driving distance of 24.3 axle-miles per day and
deposition rates of traffic-related pollutants given in Table 1. The per capita driving distance
was derived from 1968 figures of 66 x 10 axle-miles per day from a population of 2,714,000 in
the Washington, D.C. Metropolitan area (7). For example:
S.43 x 10"6 Ibe. BOD
axle-mi.
24.3 axle-mi.
cap.-day
• *?/? g - 0.060 grams/capita-day
d) Runoff, during a two-hour storm event, of traffic-related materials deposited on roadways Over a
three-day period has been compared with sewage final effluent discharged to receiving waters dur-
• ing this same two-hour storm.
Source: Shaheen, Donald G. "Contributions of Urban Roadway Usage to Water Pollution."
Prepared for the U.S. Environmental Protection Agency. EPA-report No. 600/2-75-004.
1975.
-------
higher than during typical dry weather. The greater dilu-
tion of effluents could offset the addition of street sur-
face materials, with the result that increases in pollutant
concentrations might not occur.
/
As an example, consider a typical urban watershed in south-
eastern Pennsylvania with a population density of 10 persons
per acre (6,400 persons per square mile). It is assumed
that water for domestic supply is imported into the basin
from elsewhere; that treated municipal effluent is released
to surface waters within the basin; and that sewage effluent
and street surface washoff are the only two sources of water
pollutants. Under these conditions, streamflow during
nonstorm conditions would average about 1.71 cfs per square
mile (cfsm), of which 0.99 cfsra would consist of sewage
effluent. Based on Sheehan's data, typical pollutant con-
centrations during dry weather would be as shown in the
left-hand column of Table 7. For the two-hour storm period,
it is estimated conservatively that a rainfall sufficient to
wash off nearly all street surface pollutants would produce
at least 0.1 inch of runoff (from the watershed as a whole).
Assuming that this runoff is present in the stream system
for only two hours—which is also implied by Shaheen—the
average discharge during this period would be 34 cfsm. The
resulting average pollutant concentrations, produced by both
sewage effluent and washoff loads, are shown in the middle
column of Table 7. The ratio of each of these figures to
the corresponding dry weather concentration is shown in the
right-hand column.
The ratios in Table 7 convey a very different impression
from that created by the ratios in the right-hand column of
Table 6. Given the fact that 2-hour pollutant concentra-
tions must be much higher than long-term concentrations in
order to be limiting for aquatic biota, the figures in Table
7 would not necessarily lead to the conclusion that shock-
loading would be critical to water quality under the assumed
conditions. The purpose of this hypothetical example is by
no means to imply that control of street surface contami-
nants is unimportant. Rather, it is simply to suggest that
transient problems due to this source may be less important
than simplistic computations would indicate; and that in any
case such problems should be analyzed with a high degree of
sensitivity to hydrologic conditions.
Measurements of pollutant accumulation on roadway surfaces
have clearly performed an invaluable service in raising the
48
-------
ID
TABLE 7
COMPARISON OF HYPOTHETICAL STORM AND NONSTORM CONDITIONS
Hypothetical Pollutant Concentrations
Suspended Solids
BOD
COD
Kjeldahl-N
Phosphate-P
Lead
Zinc
Copper
Nickel
Chromium
Typical
Dry Weather
(mg/1)
13.9
8.1
11.6
1.7
4.0
.017
.046
.017
.006
.006
2 -Hour Storm
(mg/1)
73.5
0.6
4.5
0.1
0.2
.859
.110
.009
.014
.006
Ratio of
Concentrations :
Storm/Dry Weather
5.3
.07
.39
.06
.05
50.
2.4
.53
2.3
1.0
Source: Betz Environmental Engineers, Inc., based on Table 6
-------
level of knowledge regarding urban runoff problems. There
are a number of ways in which such data could be utilized in
current water planning studies; and their potential value
for this purpose is, of course, a matter of subjective
judgment. The opinion here is that use of such data for
analytical purposes should generally receive much less
emphasis than collection and analysis of in-stream water
quality data. As is discussed in Section 7, the latter data
would include: (1) chemical and biological information for
surface water locations at which unrecorded pollutant
sources actually appear to be causing problems; and (2)
measurements of pollutant contributions from selected high-
yield land areas.
The basis for this opinion is, first, that present pre-
dictive methodologies based on deposition and/or accumula-
tion rates do not appear to yield reliable estimates of in-
stream pollutant loadings (unless the rates are calibrated
using in-stream data), and thus have only limited value for
establishing wasteload allotments and determining the pol-
lutant reductions needed to meet water quality criteria.
Second, street dirt and dust is only one aspect of the urban
runoff problem, albeit perhaps the predominant aspect in a
major proportion of areas. This means that the range of
control alternatives which can be evaluated using the above-
mentioned methodologies is seriously limited. (Admittedly,
the effectiveness of many of these potential controls cannot
be quantified accurately using any known data sources. The
primary concern here is that controls for sources other than
streets will be overlooked.) Third, washoff of diffuse
materials from streets and other impervious surfaces may be
the aspect of urban unrecorded pollution which is least
amenable to control. Thus, it may be wise in many cases to
focus upon other aspects of the urban runoff problem.
A final comment regarding intensive street sweeping pro-
grams, and other broad-scale control measures which entail
substantial municipal expenditures, is that implementation
of such controls in the short run is likely to depend pri-
marily upon local goodwill and general awareness of water
quality issues. Technical inputs such as cost-effectiveness
comparisons of control alternatives are not likely to be
critical to the implementation of these measures, except
perhaps in metropolitan areas with long histories of water
planning and currently very well-funded programs. Thus, it
can be argued that technical planning efforts be directed
50
-------
primarily toward: (1) gaining a better understanding of
existing water quality problems; (2) detecting and docu-
menting the worst source areas, in such a fashion that con-
trol can feasibly be achieved even if opposition occurs; and
(3) developing highly selective control measures which do
not entail major public or private cost. These issues are
discussed further in Section 7.
Localized Pollutant Sources
Discussion of pollutant loadings due to urban land drainage
has tended to emphasize pollutant-generating factors which
are relatively ubiquitous in urban areas. A typical example
is the following description by Loehr (1974):
"Street litter, gas combustion products, ice control
chemicals, rubber and metals lost from vehicles, de-
caying vegetation, domestic pet wastes, fallout from
residential and industrial combustion products, and
chemicals applied to lawns and parks may be sources of
contaminants in urban runoff."
A concern here is that such discussion may have created the
impression in some minds that the relevant control measures
for urban runoff problems are limited to actions which
affect pollutant generation from extensive land areas.
Although such measures may in fact be necessary, it is
important to note the potential role of localized, or "site-
specific," pollutant sources which can be addressed by
fundamentally different types of controls. Site-specific
sources can be defined loosley as pollutant sources which:
(1) yield relatively large quantities of pollutants relative
to the land area involved; and (2) are not necessarily
associated with general types of land use or economic ac-
tivity.
The importance of continuous sources (i.e., non-SRA sources)
fitting this description is commonly recognized; two ex-
amples are malfunctioning on-site septic systems, and land-
fills which produce significant quantities of leachate. The
existence of site-specific SRA sources appears to have
received relatively little attention, however, due in part
to the fact that very few studies of urban runoff pollution
have attempted to determine the specific origin of pollutant
loadings in a given watershed. One case in which a detailed
investigative process was carried out is considered highly
instructive, and thus is described here in detail. In 1969,
51
-------
the Water Resources Research Institute at Rutgers University
began an intensive study of BOD loadings from unrecorded
pollutant sources in urban areas of New Jersey. Several
small, predominantly residential basins with no known ef-
fluent discharges were selected for study, one of which was
the Mile Run watershed in New Brunswick. The findings were
reported as follows (Whipple, et alr 1974, pp. 26-31).
"The upstream portion of Mile Run above its Livingston
Avenue crossing was chosen for study since it drained
an area which was predominantly residential in char-
acter and seemed to contain no gross pollution...This
drainage area selected is approximately one mile
square, of which 38.5% is devoted to residences, and
only 19.2% to industrial and commercial uses. Street
surface area accounts for some 14.0% of total land...
"During the early sampling of Mile Run, certain pecu-
liarities were noticed. Heavy oil slicks were observed
to occur during rainstorms with strong odors of fuel
oil at the gaging station. In addition, BODs were
running approximately three times what was expected. A
COD test run on a sample on November 7, 1969, showed a
relatively high result. This was the first indication
of organic loading other than street runoff. On
November 10, 1969, samples were taken at Georges Road
and Livingston Avenue on a non-rain day. The results
were revealing (see Table III-l). The BOD at Georges
Road was some six times the BOD at Livingston Avenue.
In addition, the pH at Georges Road was abnormally
high. On November 14, 1969, for a relatively small
rainfall a BOD above 25 mg/1 was recorded, again con-
siderably high. It was concluded that there was some
additional source of pollution affecting data collected
on Mile Run.
"The source of the organic load imparted to the stream
was investigated to locate the area and the possible
source...On March 10, 1970, the BOD concentration was
above 100 rag/1 at Georges Road and the temperature
difference was 11 degrees C from the Livingston Avenue
sample...Obviously, some heated organic load was en-
tering above Georges Road crossing...
"An on-site investigation was conducted to determine
the possible source along the stream banks on the
morning of April 28, 1970. The investigation revealed
52
-------
a small hot water source, possibly accounting for the
temperature rise at Georges Road, and fuel oil pollu-
tion, obviously entering in the Ward Street-Fulton
Street area due to fuel oil commerce in the area.
Drainage ditches in that area were heavily coated with
sludge fuel oil which emptied into Mile Run during
times of runoff and appeared to be dumped into ditches
as waste. The north bank of Mile Run in this area
appeared to be leaking oil from its bank and there was
evidence of waste oil being dumped from atop the bank.
What was difficult to determine was the cause of a
heavy brown scum buildup at two locations in the stream
just below Squibb and just below Georges Road.
"Obviously the pollution in Mile Run is far greater
than can be attributed to the housing. The mean 5-day
BOD during a two-year period of observation was about 9
mg/1 in dry weather and 17 in wet weather. The average
BOD loading varied to a much greater extent, being
about 26 Ibs/day/sq. mi. in dry weather and about 800
Ibs/day/sq. mi. in wet weather, giving a weighted mean
annual BOD loading of about 277 Ibs/day/sq. mi. It is
apparent that the storm runoff is of controlling
importance. The changes in suspended solids and COD
during rainfall are even more dramatic than those of
BOD...The influence of industrial wastes was evidenced
not only by the high BOD, but also by direct observa-
tions of heated and colored discharges, banks darkened
with oil, oil slicks and rapid changes in BOD at
certain times of the day.
"It is apparent that the pollution levels in such areas
are mainly dependent upon the degree and kind of com-
mercial and industrial development, and the effluent
controls employed. No methodology can be visualized to
forecast such loadings from commonly available planning
parameters."
The Mile Run case is clearly an extreme rather than a
typical situation. The important point, however, is that
site-specific sources such as these could be operative on a
smaller scale in many areas. Two critical characteristics
of these sources are the following. First, as indicated by
the last sentence of the above quote, their effects are
difficult to predict and identify on the basis of gener-
alized relationships or modeling programs. (Note that,
53
-------
unless unusually detailed calibration procedures and moni-
toring were utilized, a stormwater model would attribute the
pollutant loadings in Mile Run to ordinary washoff or ero-
sion processes, or else would leave the high loadings un-
explained. ) Second, the controls which are appropriate for
these sources ordinarily consist of very selective measures.
Broad-scale controls tend to be either irrelevant, as would
be true for street sweeping in the Mile Run case, or unnec-
essarily inefficient, as would be true in this case for
runoff detention and treatment. Several classes of site-
specific sources with these characteristics can be identi-
fied as follows.
Dumping of liquid and solid waste on land surfaces.
Waste disposal or storage of pollutant-generating
materials on land surfaces can occur either on-site or
off-site relative to points of waste generation (e.g.,
homes or businesses). Whether or not such activities
involve impervious surfaces, an important distinction
from other pollutant sources is that localized waste
accumulations tend to be less costly to remove than
diffuse materials such as street dirt; and there is
generally a higher probability that further accumu-
lations can be prevented.
Major sanitary sewer leaks and bypasses. Pollutant
generation by sanitary sewer systems, other than com-
bined sewer overflow and discharge of treated effluent,
is important in many areas. Potential problems include
both sewer leakage during dry weather, and major over-
flows during wet conditions due to inflow of storm-
water. Lager and Smith (1974, p. 67) have described
separate sewer systems in the following way:
"Most sanitary sewers in the United States are de facto
combined sewers. Stormwater enters these sewers through
cracks, unauthorized (and sometimes authorized) roof
and area drains, submerged manhole covers, improperly
formed or deteriorated joints, eroded mortar in brick
sewers, basement and foundation drains, and poorly
constructed house connections."
Inflow of surface and groundwater during wet conditions
may cause total wastewater flow to exceed the capacity
of pipes, pump stations or treatment plants downstream,
with the result that bypasses are necessary. In many
54
-------
communities, bypasses have been created on a casual
basis, without records being kept as to location and
design characteristics.* The extent of these problems,
which are important primarily for older urban areas,
may be difficult to predict on the basis of generalized
relationships, although empirical information for this
purpose has been developed by some private consulting
firms.
Direct discharge of liquid waste to storm sewers and
receiving waters. Many of the pollutant-generating
activities included in this category are illegal
according to federal, state, or municipal statutes, but
nevertheless occur. Unauthorized discharges to surface
waters may be either intermittent or continuous, and
may or may not involve fixed conveyance facilities. An
example of intermittent discharge without conveyance
facilities is dumping of septic system waste into
surface waters by professional scavenging operations.
An important problem in some areas is illegal con-
nections to storm sewers, which allow untreated wastes
to pass more or less directly to receiving waters.
Improper use of storm sewers and underdrains for
disposal of wastes such as crankcase oil is also not
uncommon.
Discharges in excess of permits. Even effluent dis-
chargers presently covered by permits can constitute
unrecorded pollutant sources if the discharge levels
exceed permitted amounts. This can occur for a variety
of reasons, including accidental spills and treatment
system malfunctions. Although it is not possible to
generalize regarding violation of NPDES permits, a
relevant observation is that the estimates of "nonpoint
source" pollutant loadings which are inferred in mass-
balance and modeling studies are often significantly
higher than loadings observed in comparable basins
without recorded effluent dischargers.
* In Allegheny County, Pennsylvania, for example, over 600
known sewer bypasses have been identified, and the actual
number of bypasses is estimated to be much greater.
(Personal communication with Dennis Burke, P.E., of The
Chester Engineers, Coraopolis, Pennsylvania.)
55
-------
The importance of these localized sources relative to other
unrecorded pollutant sources is difficult to judge except on
a case-by-case basis. However, the present study has
yielded one item of evidence which suggests indirectly that
sources other than washoff of diffuse materials may be
generally significant. Chemical data for the 15 urbanized
basins in Tulsa, Oklahoma, cited earlier have been subjected
to a pooled regression analysis in which individual pollu-
tant concentrations were related to precipitation variables
such as: time since start of rainfall, average intensity of
rainfall, and time since the previous storm. The most
notable finding was a failure to observe strong positive
associations with time since the previous storm, even when
the other factors (which did bear strong relationships to
pollutant concentrations) were controlled. Association with
time since the previous rainfall would be expected if water
quality is dominated by sources for which the available
supply of pollutants fluctuates markedly with the occurrence
of storm events—as tends to be true for street dirt and
similar materials. This finding, which has been noted in a
few other studies, could be due partly to nonlinearity of
pollutant accumulation rates, but is nevertheless considered
significant. (Further discussion is contained in the
Technical Appendix.) The conclusion is that the nature of
stormwater pollutant sources should be considered carefully
in each case, rather than simply assumed.
Conclusions
The present summary review has dealt primarily with pollu-
tant generation; an extensive discussion of water quality
issues is contained in the Technical Appendix. The overall
conclusion is that loadings and problems due to unrecorded
sources are highly variable, and may be due to site-specific
factors as well as to factors such as traffic and litter
which are generally ubiquitous in urban areas.
Agencies with limited resources may be well-advised to avoid
focusing upon transient water quality problems which occur
during and immediately after surface runoff events, unless
such problems are directly demonstrated to be important.
Dissolved oxygen has historically been over-emphasized in
water planning; a carryover of this pattern to analysis of
unrecorded pollution problems could be unfortunate since
transient DO phenomena are especially difficult to predict
and evaluate. With regard to other "shockload" effects of
urban runoff, the importance of these effects may not be
56
-------
demonstrable even if temporarily high pollutant concentra-
tions are established, due to the extreme lack of knowledge
regarding response of aquatic biota to conditions lasting
less than 24 hours. (Note that high concentrations due to
"first flush" effects are usually very short in duration,
when receiving waters consist of free-flowing streams.) An
important point is that transient impacts are highly sensi-
tive to receiving-water characteristics, and therefore
should not be a major focus of analysis unless an agency is
prepared to consider these characteristics in detail.
Given the fragmentary data reviewed here, it appears reason-
able to hypothesize that analysis of transient problems per
se is usually not essential in urban areas which do not
contain combined sewers or notably poor environmental con-
ditions, and for which the receiving waters do not consist
of estuaries or standing water bodies.* The alternative is
to focus upon the long-term problems created by SRA and non-
SRA sources, utilizing annual and/or seasonal pollutant
loading estimates. Preparation of such estimates on the
basis of observed data does not necessarily require the use
of loading simulation models. Similarly, analysis of the
effectiveness of control measures for both SRA and non-SRA
sources can be conducted on a long-term basis, using tools
such as the Hydroscience methodology cited in Section 9.
It is true that existing water quality criteria in some
areas are limited largely to dissolved oxygen and pathogens.
However, current water quality management planning studies
are not compelled to perpetuate the biases of the past. An
extremely important output of these studies could in fact be
the recommendation of additional criteria which will provide
more adequate protection of water quality.
An important finding of the present review is that the water
quality impacts of new urban development, once in place, may
be substantially less than the average effects of existing
* A major exception to this generalization may be urban
areas located in climatic regions where rainfall is not
well-distributed throughout the year. Extended pollutant
buildup on impervious surfaces may cause transient DO
problems to occur widely in these areas even if there is
reasonable public cleanliness.
57
-------
development. Given this finding, it is considered feasible
to adopt a strategy for new development in which, many of the
major control measures would be based explicitly upon water
quantity considerations rather than water quality per se.
This strategy is outlined in detail in Section 9. With
regard to existing water quality problems, the recommended
strategy when technical planning resources are limited is to
devote these resources primarily to analysis of: (1) in-
stream water quality problems; and (2) selected high-yield
pollutant sources. This approach and its underlying ra-
tionale are discussed in Section 7.
58
-------
SECTION 6
REVIEW OF PLANNING METHODOLOGIES
Introduction
The ultimate goal of the present project has been to develop
a "planning methodology" for use in water quality management
planning studies. Such a methodology is defined broadly as
any systematized approach which provides assistance in
quantifying present and future water quality problems, and
in evaluating possible control strategies. As noted earli-
er, the present emphasis is upon urban land, particularly
the impacts of future urban development. A broad objective
of the planning methodology is to promote and facilitate the
use of preventive water quality control measures.
This section discusses the planning methodologies which have
been reviewed. The four general types of methodologies con-
sidered are the following: (1) environmental synthesis
techniques (e.g., suitability mapping); (2) stormwater
modeling; (3) development and use of statistical relation-
ships; and (4) other predictive tools. As a result of this
review, and the review of urban stormwater problems dis-
cussed in Section 5 and in the Technical Appendix, the
present report will recommend the use of a variety of pre-
dictive tools, rather than a single unified methodology.
Suitability mapping and similar techniques are found not to
be favorable for general use in water quality management
planning studies, because they do not directly address the
technical issues which are felt to be most important. The
value of statistical studies is limited by the data require-
ments involved and the fact that the relationships obtained
may not be useful for evaluation of controls. One reason
for not focusing on stormwater modeling in the present
report is simply that the subject has been treated exten-
sively elsewhere (e.g., Meta Systems, 1975). Modeling is
also found to have a number of liabilities, particularly
when planning resources are limited. The discussion in the
latter portion of this document will therefore stress the
use of other methodologies.
59
-------
Stormwater Modeling
As introduction to some of the basic issues involved in
stormwater analysis, it is worthwhile to consider briefly
the distinctions between deterministic modeling—of which
stormwater modeling is representative—and stochastic model-
ing. The essence of deterministic modeling is an attempt to
provide direct mathematical representation of causal link-
ages. Model equations are formulated on the basis of a
priori knowledge and intuitive judgment; an important cri-
terion is the existence of direct parallelism with physical
phenomena. In contrast, stochastic modeling involves devel-
opment of equations on the basis of empirical data, utiliz-
ing statistical techniques. A stormwater model which
attempts to predict pollutant loadings from a particular
type of source area, for example, might provide separate
equations describing the washoff process and the transport
of pollutants through each link of the drainage system;
whereas a stochastic model might involve only one equation,
estimated statistically on the basis of repeated observa-
tions of source area characteristics and pollutant loadings.
The distinction between deterministic models and stochastic
models (notwithstanding the use of statistical techniques in
the latter case) is not always pronounced, since calibration
to observed data is a common characteristic of deterministic
modeling, whereas a strong theoretical basis is sometimes
available for development of stochastic models. Neverthe-
less, the difference in emphasis is very important.
The appropriateness of deterministic versus stochastic
modeling depends upon the complexity of the phenomena under
study and the level of knowledge which exists. Stochastic
modeling is clearly appropriate in economics, for example,
since it would be impossible to trace all of the individual
dollar flows and corresponding activities which comprise an
economy. Thus, in an econometric model, a quantity such as
personal consumption expenditures would be related to dis-
posable income and other aggregate variables (many of which
could be highly artificial constructs) which are demon-
strated statistically to bear some association with personal
consumption. The potential weaknesses of stochastic model-
ing are that: (1) large amounts of data are necessary for
model development; (2) the meaning of the relationships
obtained may be obscure; and (3) the predictive accuracy of
stochastic models may be poor—particularly in cases where
certain real-world quantities are to be manipulated for the
60
-------
purpose of system control. On the other hand, the potential
weakness of deterministic modeling in dealing with complex
situations is that a lack of empirical investigation prior
to model development may lead to oversimplification and
over-structuring of the phenomena under study, with the
result again that predictive accuracy and value for analysis
of controls may be limited. The latter limitation, when it
exists, tends to be more subtle than in the case of sto-
chastic modeling, and often goes unrecognized, due to the
fact that the relationships in deterministic models are
typically designed so that input of control actions per se
does not present difficulties.
With regard to the study and control of urban unrecorded
pollution problems, it is felt that the usefulness of both
deterministic modeling and stochastic modeling is seriously
restricted, due to the issues just discussed. The basic
problem in the case of deterministic modeling is a lack of a
priori knowledge—referring primarily to documentation of
conditions relevant in specific instances, rather than
knowledge at a theoretical level. The problems in the case
of stochastic modeling (discussed in a later sub-section)
are the difficulty of obtaining an adequate base of em-
pirical data, and the inherent limitations of this approach
for evaluation of control measures. On the whole, it may be
premature to rely heavily upon formalized analytical pro-
cedures in dealing with unrecorded pollution problems other
than, known sewer system overflows.
The present review of stormwater modeling will focus upon
only two computerized models: SWMM and STORM. The dis-
cussion is also relevant, however, to the use of simplified
techniques which incorporate similar assumptions. The
Storm Water Management Model (SWMM) has been developed under
the auspices of EPA and is expected to play a major role in
water quality management planning studies. STORM was de-
veloped as a planning tool by the U.S. Army Corps of En-
gineers. SWMM is highly comprehensive in that it includes
sophisticated routines for flow routing and simulation of
receiving water response (e.g., dissolved oxygen depletion).
In contrast, STORM is relatively simplistic and is concerned
primarily with pollutant generation. An important feature
of STORM is that it permits long-term simulation of runoff
quantity and pollutant loadings, whereas SWMM must be ap-
plied to pre-selected design storms. Both models are
61
-------
designed to permit evaluation of control alternatives, such
as street sweeping, runoff storage and treatment, and other
structural controls (in the case of SWMM).
An aspect of stormwater models which is particularly im-
portant for the present discussion is the manner in which
the models treat acquisition of pollutants from land sur-
faces (as opposed to sewer system overflow, the other major
category of unrecorded pollutant generation considered by
SWMM). SWMM and STORM consider two basic-mechanisms:
washoff of accumulated dirt and dust from impervious sur-
faces, and erosion of material from pervious land. Washoff
of material from impervious surfaces tends to receive pri-
mary attention when the models are applied to urban water-
sheds. The design storms utilized for planning purposes may
in fact be defined specifically as rainfall events which
produce direct runoff from impervious surfaces but not from
pervious land.
In both SWMM and STORM, pollutant yields due to washoff are
represented as the sum of three components: dissolved
material, suspended material, and settleable material. The
models assume that the rate of removal of dissolved material
from a surface at any given time during a storm is posi-
tively related to the amount of material present on the
surface at that time, and the rate of stormwater runoff at
that time. For suspended and settleable material, the rate
of removal is also positively related to the availability of
material for transport, which is considered to be a positive
function of tl\e runoff rate. A characteristic of this
formulation is that, when only the suspended fraction is
considered, the implied pollutant concentrations are nega-
tively associated with time since start of storm, and nega-
tively related to the rates of precipitation and runoff
(except at the very beginning of rainfall). The model thus
describes a "first flush" effect. When suspended and
settleable materials are considered, however, it is mathe-
matically possible to reproduce almost any pollutograph
shape.
The most critical determinant of pollutant loadings and
concentrations due to washoff is the amount of material
present on impervious surfaces at the start of a storm. The
STORM model simulates changes in the amount of this material
over time. Dirt and dust are assumed to build up continu-
ally on impervious surfaces, and be removed at discrete
intervals by storm events and by street sweeping. In the
62
-------
case of SWMM, it is necessary to supply dirt and dust infor-
mation for each individual storm event considered (unless
SWMM is linked to STORM in the manner suggested by Meta
Systems). An obviously important issue in applying STORM is
the possible nonlinearity of dirt and dust accumulation over
time. As noted in the previous section, Sheehan (1975) has
shown that pollutant buildup on roadway surfaces is likely
to bear a nonlinear relationship with time, approaching an
upper limit rather than increasing indefinitely. On the
other hand, he suggests that the deposition rate (a constant
daily rate) may nevertheless be more relevant than the'
accumulation rate, since pollutants removed from roadways
may still be available for transport by stormwater. In any
case, STORM incorporates a constant daily dirt and dust
accumulation rate, as of early 1976. The emphasis of STORM
and SWMM on washoff of diffuse materials from impervious
surfaces makes these models potentially very sensitive'to
the timing of storms. The pollutant loadings and concen-
trations predicted for isolated storm events can be an order
of magnitude higher than the values for storms which occur
in rapid succession.
Erosion of soil and associated pollutants from pervious
land, when this factor is considered, is modeled in both
SWMM and STORM by the Universal Soil Loss Equation (USLE),
which is discussed elsewhere in this report. The only
transient factor in the USLE relating to hydrologic condi-
tions is the rainfall intensity factor, which expresses the
ability of rainfall to detach and transport materials from
the soil surface. A significant fact is that, since pollu-
tant accumulation processes are not involved in this case,
the pollutant loadings predicted by the USLE are not sensi-
tive to the timing of storms. ;
A point which is obviously relevant to the use of SWMM and
STORM is the fact, discussed in the previous section, that
washoff of diffuse materials from impervous surfaces, gen-?
eralized land erosion, and combined sewers are not the only
unrecorded pollutant sources affecting urban water quality
during storm periods. Sanitary sewer bypasses, outflow of
contaminated groundwater, accidential spillage, overflow of
on-site septic systems, unauthorized and excessive waste-
water discharges, and other factors can be important, t
Even when washoff and erosion from land surfaces are dom-
inant, the 'relative importance of individual source areas
and source types may be a critical issue. An industrial
63
-------
docking area (impervious) or an unregulated dumping area
(pervious) may yield pollutant loadings that are an order of
magnitude higher per unit area than the loadings from
ordinary streets, lawns, and parking lots. Theoretically,
all of these factors can be represented in stormwater models
(although STORM is much more limited than SWMM in this
regard). Model structure per se is not necessarily con-
straining. The important issue is that source identification
is rarely an integral feature of modeling studies, except
perhaps for analysis of sewer system characteristics. Model
calibration and source identification are by no means the
same thing, since calibration data usually reflect a variety
of pollutant sources. As a result, the typical situation in
modeling studies is that all pollutant loadings which are
not accounted for by combined sewers, recorded discharges,
and known sanitary sewer bypasses are attributed to general-
ized washoff processes (and possibly general land erosion if
the USLE is utilized). This situation may or may not affect
the ability of the models to predict in-stream pollutant
loadings and problems. However, the failure to isolate
specific sources and source types has serious implications
for the evaluation of control measures.
The cases in which conventional modeling studies are likely
to be most successful are thus: (1) areas where sewer
system outflow accounts for a large proportion of pollutant
loadings, and is well-documented; and (2) urban core areas
where washoff of diffuse materials from street surfaces can
be safely assumed to account for most pollutant loadings
(due to the high level of street activity, and the general
absence of pervious land and exposed watercourses). These
issues are considered further below, after discussion of
model calibration.
Other aspects of stormwater models besides pollutant genera-
tion are, of course, important. The major elements of SWMM
consist of a highly sophisticated hydraulic transport,
storage, and treatment model, and routines for simulation of
dissolved oxygen dynamics. There are a number of critical
issues regarding simulation of receiving water response
which could be discussed, such as the ability of the model
to predict transport and deposition of particulate materials
in natural channels, and the appropriate values of para-
meters such as deoxygenation and reaeration rates. The
present discussion will be limited, however, to the "front
end" aspects of the models, i.e., the manner in which they
deal with pollutant generation.
64
-------
Need for Model Calibration and Verification
Calibration of stormwater models—i.e., the use of empirical
water quantity and .quality data to infer appropriate values
of the model parameters—tends to constitute a very large
proportion of total modeling costs. Given that default
values are supplied for most parameters, calibration is not
strictly necessary to obtain outputs. Thus, it is worth-
while to consider the need for calibration in some detail.
Since the objective of calibration and verification is to
increase the accuracy of a model in describing real-world
situations, a major issue is the extent to which accuracy is
actually required for planning purposes.
The potential functions of stormwater modeling as a tech-
nical planning tool could be categorized as follows: (1)
analysis of the magnitude and timing of pollutant inputs
from various sources; (2) prediction of pollutant loadings
and concentrations at in-stream points where problems
occur; (3) simulation of water quality problems (e.g.,
oxygen depletion) which result from complex interactions of
pollutant loadings and other factors; and (4) analysis of
system response to control measures, and comparison of
different controls in terms of effectiveness. The first
three of these functions involve description of water qual-
ity under existing conditions. The use of modeling for such
descriptive purposes is limited primarily to cases in which
model outputs are cheaper or otherwise more feasible to ob-
tain than direct water quality data (for example, cases in
which great spatial detail or coverage is required, or when
the necessary information pertains to infrequent events).
Predictive accuracy is clearly important when these des-
• criptive functions are emphasized.
The fourth function listed above, the use of modeling to
evaluate potential control measures, commonly receives pri-
mary attention in the design of modeling programs. The need
for predictive accuracy in this case is somewhat debatable.
It is argued that stormwater modeling can be useful for
comparative evaluation of controls, even if the model is not
successful as an absolute: predictor. Three comments can be
made regarding this position. First, models which do not
replicate real-world conditions reasonably well cannot
establish the overall levels of control needed. That is,
comparative evaluation of control measures will not neces-
sarily indicate the ability of a set of controls to achieve
65
-------
absolute water quality standards. Second, tests of pre-
dictive accuracy are often the best means of demonstrating
that underlying causal relationships are specified correctly
in the model, including relationships that affect the rela-
tive performance of control measures. Third, the extent to
which predictive accuracy can be claimed for analytical
methodologies utilized in planning could have an important
bearing on the implementation of controls, in cases where
recommended plans are controversial. The arguments which
discount the need for predictive accuracy in modeling may be
based on an implicit assumption that implementation of
recommended plans is not an issue—that the major question
is simply the choice from among available control altern-
atives. Such a description may not apply to current water
quality planning programs (as is discussed in the next
section). Given these factors, it is felt here that pre-
dictive accuracy should be an explicit objective of storm-
water modeling, if modeling is to be undertaken at all.
However, this is admittedly a matter of opinion.
SWMM and STORM clearly require calibration in order to simu-
late water quality phenomena accurately, although SWMM may
be a good hydraulic predictor with little calibration.
Possible sources of error in the pollutant loading estimates
yielded by these models are the following: (1) failure of
dirt and dust accumulation data from the literature to
describe local conditions; (2) inaccurate specification of
dirt and dust buildup processes; (3) omission of unrecorded
pollutant sources other than street dirt (and general soil
erosion, if the USLE option is utilized); and (4) failure to
consider the potential significance of.pollutant losses
during transport. Both STORM and SWMM incorporate default
values of dirt and dust accumulation rates based on the
original Chicago study (APWA, 1969), which could be replaced
by more recent, regional data. This might not reduce the
need for model calibration, however, A consequence of the
discussion in Section 5 is that direct measurements of
surface dirt and dust may not relate well to in-stream water
quality.
In general, the usefulness of uncalibrated models is felt to
be limited to the following tasks: (1) demonstrating to a
receptive audience that certain problems do exist; (2)
selecting control measures when there is widespread agree-
ment regarding both the seriousness of the problems ad-
dressed and the sources responsible; and (3) dealing with
water quantity problems. An example of the second case is
66
-------
use of the USLE to develop controls for erosion from con-
struction sites (see Section 11 of this report). Here,
neither the pollutant source nor the resulting problems are
likely to be a major subject of controversy; the principal
planning task is to design and implement appropriate con-
trols. It is relevant that erosion/sedimentation controls
can be successfully implemented without any quantitative
analysis of water quality impact, as has been demonstrated
in a number of communities.
A final point which should be noted in this context is that,
if accurate stormwater modeling is judged to be infeasible
in any particular case, the alternative is ordinarily not to
substitute another methodology that yields similar outputs
(which may not exist). Rather, the alternative is to re-
phrase the questions that are being asked, so that funda-
mentally different analytical techniques can be employed.
This point is discussed further below.
Model Calibration Procedures
Model calibration consists of a process wherein the model
parameters, and occasionally the input data describing the
hydrologic system, are adjusted so that water quantity and
quality predictions are consistent with observed data.
Verification consists of comparing predictions with reality
for cases which have not been considered in the calibration
process. In regional water quality planning studies, model
calibration programs can be oriented toward two somewhat
different objectives: (1) developing an accurate model for
a given basin or water body; and (2) developing general
values of model parameters, which can be widely applied in
modeling applications throughout the region. The model
parameters which are principally referred to here are dirt
and dust accumulation rates. In either case, the typical
situation is that adequate water quality and quantity data
for model calibration do not exist prior to a model study.
Thus, the first step is commonly to design and execute an
intensive stormwater monitoring program.
A systematic procedure for SWMM calibration and verification
has been developed by Dr. James Hagarman of the University
City Science Center, Philadelphia. This procedure illus-
trates the steps necessary to obtain generalized model
parameters, and is generally indicative of the important
issues involved in stormwater modeling.
67
-------
The areas used for model calibration in this procedure are
small catchments (less than 100 acres) which contain uniform
or nearly-uniform land uses. The overall set of areas
chosen is intended to be representative of land uses found
within the study area. Data are collected for about a half
a dozen runoff events, for the duration of each event. The
data include both quantity and quality of runoff, as well as
the rainfall hyetograph. For each area, detailed measure-
ments are made of imperviousness, curb length, slopes, and
other characteristics relevant to hydrologic response. If
an area is serviced by a combined sewer system, dry weather
flow data are collected for calibration of sanitary flows.
The calibration procedure starts with water quantity; after
quantity, quality is calibrated. Although there are no set
rules as to which program parameters to adjust during flow
calibration, there are some guidelines. Sensitivity analy-
sis has shown that the percent of land impervious has a very
significant effect on quantity. This factor is sometimes
treated as an adjustable parameter, even though it is phys-
ically measurable, since the adjustments can be considered
to express interactions between pervious and impervious
areas. Infiltration rates can also be adjusted. Such
parameters as pervious area resistance factors and surface
storage depth have almost no effect on the outfall hydro-
graph. Figure 5 gives some examples of program response to
input parameters. Generally, measurable data besides imper-
viousness should not be varied. For quality calibration,
the dust and dirt accumulation rates, and the pollutant
composition of dust and dirt, are the major adjustable
parameters. (If default values of these parameters are to
be utilized, it is extremely important to assure that the
conditions on which the values are based resemble conditions
encountered in the study area.)
Following the calibration of the model is a two stage veri-
fication process. The first stage is the application of the
model to a small, multiple land use basin using detailed
information. If the results of this stage are not satis-
factory, it is necessary to examine carefully the assump-
tions made during the model calibration stage. If the
results are satisfactory, the second stage of verification
can be implemented. This involves application of the model
to a larger basin with multiple land uses, using less detail
than before in representing basin characteristics. The
description of this basin may be facilitated by correlations
developed during the calibration stage (such as percent
68
-------
Predicted
Measured
^y—Predicted
Measured
(d) Minimum Rate of Infiltration Too Low
(a) Percent Impervious Area Too High
(b) Initial Rate of Infiltration Too High
9
Predicted
Measured
Measured
(e) Inadequate Routing Delay Modeled
(e) Decay Rate of Infiltration Too High
Source: Jewell, T.K. and P.A. Mangarella, "Applications of Stormwater
Management Models," EPA Short Course Study Guide, University
of Massachusetts, p. 45, 1975.
Figure 5 RESPONSE OF SWMM PROGRAM TO INPUT PARAMETERS
69
-------
imperviousness as a function of land use). Again, if the
results of this test are unsatisfactory, the initial assump-
tions must be closely examined.
After calibration and verification, the model can be applied
to the entire study area, using the less detailed methods
developed in the second stage of verification. For a se-
lected design storm or storms, the model can generate inputs
to river models and aid in developing abatement plans. The
calibration and verification process tends to be costly. As
an example, this methodology is being utilized in a major
Eastern metropolitan area at a cost of about $70,000 for 12
basins. Of this total, about 50% is for data collection,
and the other half is for data reduction and computer
charges (i.e., actual modeling costs). This figure does not
include the cost of monitoring water quality and quantity
for the test basins—which is expected to exceed the sum of
land measurement and modeling costs. Thus, the total cost
of model calibration is in the vicinity of $150,000.
This procedure represents a state-of-the-art approach for
obtaining model parameters that are specific to land uses.
In other modeling applications, calibration is conducted at
a grosser level, with the result that pollutant loads cannot
be assigned to specific land uses and source areas. How-
ever, as suggested earlier, even a very detailed calibration
program such as just described does not necessarily consti-
tute a process of source identification. It does not es-
tablish the relative pollutant contributions of streets,
other impervious surfaces, and unrecorded pollutant sources
besides washoff. This issue is especially important when
dealing with industrial areas, and medium-density.urban
development in which the overall imperviousness of land is
less than 50%.
Consider, for example, the situation when a model is cali-
brated for a basin in which water quality during storms is
affected importantly by a source such as an unregulated
dumping area or an unidentified sanitary sewer overflow.
The pollutant loadings from this source are likely to be
attributed to dirt and dust accumulation. If the calibra-
tion process is especially sensitive, the loadings might be
attributed to erosion by way of the USLE option, to account
for certain aspects of their behavior (e.g., the fact that
loadings are negatively rather than positively correlated
with time since previous rainfall). In either case, the
nature of the source, and the potential effectiveness of
70
-------
selective control measures, would not be identified by the
calibration process per se. To consider a less extreme
example, suppose that accumulation of materials on impervi-
ous surfaces is in fact the predominant pollutant-generation
mechanism in a basin, but that a few areas are responsible
for a disproportionate share of the loadings. The dirt and
dust accumulation rates obtained through model calibration
would be general rates, applying to the basin as a whole,
and might not be accurate in describing land conditions at
any particular point. The two aspects of modeling which are
illustrated by these examples are the following:
1. Stormwater modeling does not necessarily deal with
pollutant generation at the level of detail which
is relevant for evaluation of selective land
management controls.
2. The dirt and dust accumulation rates which are
obtained through model calibration must be viewed
largely as artificial constructs, whose corres-
pondence with land surface conditions is unknown.
The relevant level of detail for evaluation of controls is
the specific land surface or activity to which a control
will apply. Evaluation of streetsweeping programs, for ex-
ample, requires that pollutant loadings from street surfaces
be isolated from all other influences on stormwater. Con-
ventional modeling activities will not accomplish this; but
neither will any other methodology which is not based upon
extremely source-specific data. The point is that the
structural capability of a model to evaluate streetsweeping
programs might not justify its use if a required step in
performing such evaluations is to make a bald assumption as
to whether street surfaces account for, say, 40% or 80% of
the total pollutant load in any given case. The issue here
is not the relative merits of modeling versus other methods
of data manipulation, but simply the fact that model cali-
bration is not necessarily the same as source identifica-
tion.
Results of Model Testing
Model verification studies have indicated that SWMM and
STORM may be reasonably accurate in predicting runoff quan-
tity. Relatively little can be said about their ability to
predict runoff quality, however. Remarkably little testing
has actually been conducted using observed water quality
71
-------
data. The SWMM manual reports testing of water quality
predictions in four areas, with good results obtained in two
cases. On the whole, the accuracy of SWMM in predicting
water quality can be described as "questionable" (McElroy et
al, 1975, p. 5-29). The same would apply to STORM. A
reasonable hypothesis with regard to the ability of SWMM to
predict pollutant generation is that performance is rela-
tively good for combined sewer overflows, due to the fact
that discharge quantities are the most critical issue in
this case, but that predictive accuracy is more questionable
for other unrecorded pollutant sources.
An aspect of stormwater modeling which is relevant to cur-
rent planning efforts is that water quality modeling in
general is a specialized skill. Considerable experience is
required in order to be able to calibrate and operate models
effectively. This is true for steady-state oxygen modeling,
and therefore should hold to an even greater extent for
attempts to simulate dynamic receiving-water conditions
during and after storm events.
Overall Evaluation of Modeling
The decision as to whether stormwater modeling should be
utilized in a given situation is ordinarily not a straight-
forward choice between modeling and some other comparable
planning technique. No technique may exist which can sub-
stitute directly for modeling, in the sense that no other
technique may be capable of generating the same types of
outputs or of evaluating the same variety of control options
within a single framework. The alternatives to stormwater
modeling therefore tend to involve fundamentally different
planning approaches, in which different aspects of urban
water quality problems are emphasized. The choice which
confronts planners in this regard is explored in Section 7.
The following are some general comments.
1. The fact that modeling theoretically allows crisp
evaluation of control alternatives may have
limited significance if the underlying pollutant-
generation mechanisms are not well understood, as
is typically the case for urban stormwater. Un-
less a detailed process of source identification
is carried out along with model calibration, the
reliability of models in evaluating spatially-
selective land managment controls may be highly
suspect.
72
-------
2. The fact that modeling allows comprehensive eval-
uation of different forms of pollution control may
be less important than is commonly supposed, in
part because land management controls constitute a
weak link in the evaluation process, and in part
because of the nature of current planning pro-
grams. The ability to analyze many different
options within a single framework has greatest
value in cases where planning objectives, con-
straints, and criteria are clearly defined—as
when a "best" plan is to be selected and imple-
mented by a single actor. In present water qual-
ity management planning, however, a large number
of actors are typically involved; and issues in-
volving implementation feasibility are likely to
assume overriding importance.
3. Stormwater modeling is not necessary for evalua-
tion of long-term problems due to unrecorded pol-
lutant sources, as opposed to transient problems
which occur during and immediately after storm
events. Although some models can be utilized to
generate annual and seasonal loadings, they tend
to be relatively inefficient for this purpose.
The modeling question, as confronted in regional water
quality planning studies, largely boils down to the alloca-
tion of planning resources among various types of activi-
ties. Modeling is not inexpensive. The resources devoted
to model calibration and operation could be used instead for
collection of additional field data, design of pollution
control alternatives, investigation of pollutant sources on
an individual basis, management and institutional analysis,
or many other possible tasks. Given this fact, and the
issues mentioned above, it is felt that Stormwater modeling
should probably be limited to cases in which end-of-the-pipe
control measures for unrecorded pollutant sources are con-
sidered a realistic possibility in the near future. A
hypothesis offered here is that such cases are limited for
the most part to areas containing combined sewers and/or
very high-intensity urban development (see the next section).
The use of STORM is considered especially dubious, since
this formulation emphasizes the aspects of urban unrecorded
pollution which are least amenable to modeling, given
present knowledge. The model itself (as of early 1976)
contains a number of unrealistic features, such as Hortonian
73
-------
hydrology, linear dirt and dust accumulation rates, and an
absence of streamflow routing routines. The use of STORM to
establish "critical conditions" for pollutant loadings
through long-term simulation is considered an inherently
suspect procedure. Determination of critical transient
conditions on an a priori basis would require simultaneous
simulation of a wide variety of factors (e.g., antecedent
moisture and flow conditions, spatial pattern and timing of
rainfall) over the entire drainage area of the water body in
which problems occur—which is well beyond the scope of
STORM.
Environmental Synthesis Techniques
Environmentally-based community and regional planning has
received increasing attention in the U.S. in recent years.
It is therefore logical to consider ways in which water
quality management planning might be linked to such activi-
ties, and specifically, to investigate the present relevance
of methodologies which have been developed for purposes of
comprehensive environmental planning. The common charac-
teristic of these methodologies is that they attempt to
synthesize multiple environmental objectives in order to
develop an overall land use and public facility plan.
Primary attention will be paid here to the "suitability
mapping" approach, associated with the work of G. Angus
Hills, Philip Lewis, and Ian McHarg. A brief historical
account will illustrate the important aspects of this
approach.
The early work of Hills, a soil scientist and physical
geographer, was concerned primarily with biological produc-
tivity of land and water. A major objective of his land
planning approach was to provide criteria for subdividing
large regions into homogeneous units based on biologic
capabilities. Utilizing several different classes of
mapping units. Hills' approach involved evaluating the
capability, suitability, and feasibility of land areas for
agricultural production, forest production, wildlife pro-
duction and management, and recreation. These three levels
of evaluation were defined as follows:
Capability: the inherent potential of the combined
physiographic features (landform, ground and surface
water, soil and climate) of an area for biologic
production. '
74
-------
Suitability; the capacity of the site in its present
condition to respond to specific management practices
(kind and degree of effort) for a particular use.
Feasibility; the relative advantage of mapping, i.e.,
designating an area for specific uses having regard to
its suitability for these uses under existing fore-
casted socioeconomic conditions.
These concepts were refined and brought to bear more spe-
cifically on urban growth by Philip Lewis, in his develop-
ment of the "environmental corridor concept." The major
objective was to define and evaluate the ability of resource
patterns to be determinants of urban form. Planning would
involve inventory, mapping, and overlay analysis of resource
patterns involving above-surface, surface, and below-surface
land conditions. Primary emphasis would be placed upon
identification of areas which should be protected, par-
ticularly for purposes of recreation. A numerical rating
system would be utilized in the identification process;
prioritization of areas would be achieved by summation of
the ratings based upon individual resource factors.
The concepts of suitability and compatibility have been
developed most fully by Ian McHarg, in his book Design
with Nature and later work. The two leading characteristics
of McHarg's approach are: (1) a detailed description of
existing ecological processes; and (2) a systematic analysis
of land uses vis-a-vis natural land features using a com-
patibility matrix. Human and non-human life processes are
evaluated and subsequently presented as limiting or liberat-
ing criteria for land development. The compatibility matrix
then relates potential land uses to natural land categories,
where the latter involve factors such as climate, geology,
physiography, hydrology, soils, plant associations, wild-
life, and unique sites. Each combination is rated, utili-
zing a broadly-based value system; and the findings are
presented graphically in the form of composite suitability
maps. ,
Recent environmental planning activities have tended to
emphasize the data collection aspects of these approaches.
Digitization of land characteristics, interpretation of
satellite data, random point sampling, computer mapping, and
other modes of data collection and manipulation have re-
ceived increasing attention, particularly in studies which
deal with large regions. Analysis of this information
75
-------
usually does not include full implementation of the suit-
ability concept as outlined above. Typical activities are
delineation of environmentally sensitive areas, according to
accepted standards and criteria, and estimation of the
aggregate environmental impacts of land use and land de-
velopment, utilizing various assumed relationships.
Certain factors relating to the historical development of
these environmental planning techniques may limit their
appropriateness for current water planning studies. One
issue is that suitability mapping and related approaches
tend to be oriented toward a fundamentally different set of
problems than are relevant in water quality management.
Typically, comprehensive environmental planning must arrive
at a set of land use recommendations on the basis of a wide
variety of factors which are incommensurable and/or non-
quantifiable. Thus, the structure of planning methodologies
is strongly influenced by the need to provide a rational
basis for definition and weighting of objectives. In con-
trast, the objectives of current water planning studies are
relatively unambiguous. The constraints which must be met,
and the degree of success which is attainable through vari-
ous controls, are much more amenable to quantification than
many other types of environmental impact (e.g., impacts on
terrestrial ecosystems). This difference might not appear
important, given the deterministic tone of many discussions
of environmental planning, such as the following passage
from Design with Nature (p. 56):
"The formation of ... environmental protection
regulations requires no new sciences; we need move no
nearer to the threshold of knowledge than the late 19th
century. We can initially describe the major natural
processes and their interactions and thereafter estab-
lish the degree to which these are permissive or pro-
hibitive to certain land uses. This done, it will
remain with the government and the courts to ensure our
protection through the proper exercise of police
power."
However, a wide variety of assumptions are clearly required
to move from an analysis of natural processes-—often very
imperfectly known—to a specific land use plan. Many of the
aspects of planning methodologies which are oriented toward
making (or obscuring) these assumptions are essentially
irrelevant to current water planning programs.
76
-------
Another factor is that environmental planning methodologies
have been influenced by the situation which exists for many
terrestrial systems, in which land development is necessarily
antithetical to environmental values. Preservation of
woodland ecosystems, for example, can only be achieved
through exclusion of development. Until recently, this has
caused very heavy emphasis to be placed on the land use
aspect of environmental protection, as opposed to the pos-
sibility of applying mitigative measures. Suitability
mapping and related approaches can encompass the use of
mitigative measures, but generally are not well-developed in
this regard. In the case of water planning, however, miti-
gative measures are likely to be a dominant feature of
recommended plans. Measures are available which can offset
or prevent most water-related impacts of most types of land
development, at most locations. Major attention must there-
fore be paid to design and evaluation of these mitigative
controls, and to related activities such as establishment of
appropriate performance standards. Comprehensive environ-
mental planning methodologies are often not highly relevant
to these tasks (with some exceptions such as the Christina
Basin approach noted below).
It is felt that, with three major exceptions, land use
control tends to be a secondary aspect of urban water
resources protection. The three exceptions are: (1) con-
trol of land development with on-lot sewage disposal sys-
tems; (2) prevention of direct encroachment upon water-
courses and wetlands; and (3) protection of important
groundwater aquifers. Except for these cases, direct
control of land use is a relatively inefficient means of
achieving water-related objectives, due to the availability
of other mitigative measures. By the same token, water-
related objectives per se may not form a sufficient basis
for establishing and defending land use controls. Thus,
planning methodologies which are implicitly oriented toward
land use control are appropriate primarily in communities
where there is strong support for environmentally-based land
planning, and where it is possible to link land use controls
to a variety of objectives besides water. The value of such
methodologies for regional water planning studies, encom-
passing many governmental units, tends to be limited.
(These issues are discussed in greater detail in Section 8.)
A general comment regarding large-scale programs of land
data collection is that techniques for data collection,
manipulation, and presentation do not necessarily constitute
77
-------
a water planning methodology. Explicit linkages to the
hydrologic system must be established in order to utilize
such data for design and evaluation of water resources
protection measures. A major problem is that the ability to
collect and process land data tends to exceed the ability to
establish reliable land/water relationships. In the design
of planning programs, the existence of technical land/water
relationships is sometimes treated as an afterthought;
whereas this issue is crucial to the selection of variables,
the appropriate level of spatial detail, and other aspects
of land data collection. A second point is that implemen-
tation of at least some water resources protection measures
for new development does not necessarily require compre-
hensive land data of any kind. For example, erosion/sedi-
mentation controls can be enacted without detailed knowledge
of the amount and spatial location of land in the study area
having various erodibility characteristics—even if these
characteristics are treated explicitly in the regulations.
Analysis of particular areas of land, other than as neces-
sary to demonstrate the general need for controls, could be
deferred until the time at which specific development pro-
posals are reviewed. This approach would not allow sys-
tematic prediction of erosion and sediment loadings, with
and without controls, on a watershed basis; but such pre-
dictions are not always necessary. The purpose of this
discussion is simply to suggest that: (1) land data re-
quirements should be reviewed very carefully; and (2) col-
lection of land information on a comprehensive basis need
not be a dominant element of water planning studies. Such
data can be extremely useful for a wide variety of planning
purposes in addition to water resources protection; but
these other purposes should be made explicit if they are
important in justifying a given data collection, program.
A somewhat unusual planning methodology which should be
mentioned here was the approach utilized, by the Water Re-
sources Center at the University of Delaware, in preparing a
plan for the Christina River watershed (Tourbier, 1973).
This approach involved a very fine-grained analysis of the
mitigative measures necessary to prevent deterioration of
water resources due to land development at each point in the
watersheds The cost of prevention of impacts was then
computed for each prospective land use in each small area;
and this was considered to be the critical determinant of
feasible land use patterns. One potential limitation of
this approach is that it does not deal with the problem of
establishing appropriate performance standards for new
78
-------
development in cases where the "zero impact" concept is not
applicable. (This may not be a serious difficulty when
water quantity considerations are emphasized in the design
of controls, as is recommended here.) A second issue is the
extent to which it is actually possible to establish the
cost of preventive measures on a generalized basis for small
land areas. The least-cost control solution in a given case
is likely to depend upon very detailed aspects of develop-
ment design. The Christina Basin approach is considered to
represent an important contribution to environmental plan-
ning, however, due to its emphasis upon mitigative measures
and their relationships to land characteristics.
Statistical Techniques
I
Statistical analysis of empirical data can be utilized to
estimate relationships between land characteristics and
water quality for a given study area. Perhaps the most
important aspect of multivariate statistical analysis is
that it potentially allows the investigator to control for
variables whose effects cannot be isolated in the data
collection process. For example, suppose that the loading
of some pollutant, Y, is related to urban land uses, X and Z
(e.g., residential and commercial development). It might
theoretically be possible to establish these relationships
just by measuring pollutant loadings for basins containing
only land use X, and basins containing only Z. However,
when "pure" cases of X and Z are unavailable, the relation-
ships can, instead be estimated through multivariate analy-
sis of loadings (Y) for basins containing both X and Z in
varying degrees. The inclusion of both as explanatory
factors means that Z is controlled in estimating the rela-
tionship between Y and X; and X is controlled in estimating
the relationship between Y and Z.
Statistical analysis can be useful for estimating relation-
ships between water quality and hydrologic variables (e.g.,
relationships between pollutant concentrations and dis-
charge, for various phases of the hydrograph). However, the
present discussion will be limited to analysis of the
influence of watershed characteristics. A general comment
regarding the relative value of statistical analysis, versus
other means of obtaining predictive relationships, is that
the existence of- error in equations obtained statistically
is not a basic issue. All predictive relationships contain
error; the distinguishing feature of statistical analysis is
79
-------
that error is treated formally, and is utilized as a cri-
terion for accepting or rejecting relationships. Some of
the important general aspects of statistical approaches have
already been mentioned in the earlier discussion of sto-
chastic versus deterministic modeling. The present discus-
sion will pertain to: (1) the stringency of data require-
ments for statistical analysis; (2) the need for finesse and
intuitive judgment; and (3) the usefulness of the predictive
relationships obtained. The explicit focus is upon regres-
sion analysis, although most comments apply also to analyses
of variance and covariance.
Data requirements for statistical evaluation of land/water
relationships are severe, in that it is necessary to obtain
comparable data for a relatively large number of catchment
areas. These can consist either of individual basins, or of
stream segments for which the incremental effects of land
drainage have been isolated by upstream and downstream
sampling (the latter data being less desirable due to the
large percentage errors which are possible). The need for a
substantial number of catchments relates to the statistical
role of "degrees of freedom," a quantity which is equal to
the number of observations minus the number of explanatory
factors considered in a given equation (including the
constant term). Statistically significant results can some-
times be obtained with very few degrees of freedom; examples
exist in the water quality literature of regression analyses
with as few as one degree of freedom (Radziul, et al,
1973). However, very rarely can equations be considered
substantively significant, or usable for prediction, unless
there are many more observations than variables. Ordi-
narily, the number of degrees of freedom necessary to obtain
favorable results increases rapidly with the number of
explanatory factors to be retained in an equation. The
practice of artificially inflating degrees of freedom by
including multiple water quality observations for a given
catchment (so that values of the explanatory factors are
repeated) may be legitimate in a statistical sense, but
encourages overstatement of the reliability of the land/
water relationships obtained.
The variables utilized in statistical analysis must all be
measured in a similar fashion for all catchments considered,
a requirement that is likely to involve substantial effort
and may discourage investigation of any but the most general
explanatory variables. Typically, much of the existing
water quality data in an area will be found unusable for
80
-------
statistical analysis, due either to inadequate description
of pollutant loadings (e.g., lack of discharge data, or lack
of repeated sampling during storms) or the fact that water
quality in the given catchment is influenced by factors that
cannot be controlled adequately. Thus, collection of addi-
tional water quality data is likely to be necessary, as is
typically true in the case of stormwater modeling.
The application of statistical tools to obtain realistic,
usable relationships on the basis of observed data is a
subtle process requiring considerable judgment. Beyond the
need to establish statistical significance, and to avoid
problems such as collinearity and heteroscedasticity,
formal statistical theory provides relatively little indi-
cation of how to conduct a reasonable analysis. In fact,
the most common abuse of statistical tools in non-experi-
mental fields is over-reliance upon formal significance
testing in the choice and specification of variables. This
frequently leads to very unrealistic relationships. Sta-
tistical significance should be viewed as a necessary but
not sufficient condition for inclusion of a given variable
in a regression equation; many other criteria should be con-
sidered as well. In investigations dealing with land use
and urban form, there is generally a high probability that
statistical significance will be due to spurious associa-
tions or indirect linkages, rather than direct relation-
ships. For example, personal income tends to be a good
predictor of pollutant loadings from residential neighbor-
hoods; but this is obviously not due to any direct physical
linkage. Often it is best to view statistical analysis as a
learning process, rather than a search for definitive rela-
tionships. The major outcome may in fact be .to suggest the
importance of factors which have not been explicitly meas-
ured. A skilled investigator may be able to achieve many of
the same objectives through review of empirical data without
statistical tools. These tools may be important primarily
as a means of providing discipline when quantifying various
effects.
As noted earlier, statistically-estimated land/water rela-
tionships tend not to be usable for direct evaluation of
water quality controls, since they do not deal with pol-
lutant sources at the level of specificity necessary for
this purpose. Therefore, the greatest potential value of
statistical analysis in water planning studies is to: (1)
establish the relative importance of various aspects of
pollutant generation; and (2) predict in-stream pollutant
81
-------
loadings at points for which direct information is unavail-
able. Achievement of the first of these objectives requires
considerable sensitivity on the part of the investigator.
It is unfortunately true that most statistical investiga-
tions of land-water relationships do not add greatly to
existing knowledge. A significant aspect of statistical
analysis is that it is implicitly an averaging process, as
necessary to yield general relationships; but the most
critical factors in water planning are often those influ-
ences which are relatively unique (or are otherwise un-
suitable for consideration in statistical studies). An
investigator may therefore be faced with a choice between:
(1) focusing upon catchments in which the influence of site-
specific factors is likely to be minimal, thus enhancing
statistical explanation but obtaining equations that may not
be widely usable for predictive purposes; or (2) analyzing a
typical selection of catchments, and allowing the effects of
site-specific factors to be attributed to general watershed
variables such as imperviousness, population, and major
types of land use. Such a dilemma tends to be involved in
all methodologies for obtaining general land/water relation-
ships.
The value of statistically-estimated relationships for pre-
dictive purposes depends upon the care with which they have
been developed, and the extent to which prediction involves
extrapolation beyond the conditions represented in the study
sample. Empirically-based land/water relationships tend to
be conservative, in that they usually do not predict extreme
high and low values of water quality variables. This
central tendency is considered a favorable characteristic,
along with the fact that the relationships are a direct
reflection of observed data.
On balance, it is felt here that the only urban water qual-
ity problem which is likely to be highly favorable for
statistical studies is surface water pollution due to do-
mestic on-site sewage disposal. Several relevant factors in
this case are that: (1) in areas where on-site waste dis-
posal is common, its influence on water quality can often be
highlighted effectively through careful selection of sample
catchments; (2) the water quality effects of on-site waste
disposal are somewhat easier to characterize than is true
for other urban pollutant-generation mechanisms; (3) the
important explanatory factors are unambiguous and are not
overly difficult to measure; and (4) the relevant control
measures are well-known, so that the major needs are to
82
-------
estimate loading magnitudes, identify problem areas, and
establish the general importance of controls. (Further
discussion of such studies is presented in Section 10; one
example is contained in Howard and Hammer, 1973.)
With regard to other aspects of urban pollutant generation,
it is recommended that the use of statistical analysis be
limited to estimation of very simple relationships. The
relationships obtained would be used as predictive tools
only in cases where it was clearly not feasible to obtain
direct measurements of pollutant loadings. The primary uses
would be instead to: (1) assist in partitioning observed
pollutant loadings among source types; and (2) generate
"baseline" pollutant loading values, as an assistance in
identifying relatively high-yield areas within the study
region. The latter function is related to an overall
strategy for analysis of existing unrecorded pollution
problems which is discussed in the next section.
Other Planning Methodologies
The other methodologies reviewed as part of the present
study consist primarily of water planning tools which
address highly specific problems. Those which have been
selected for emphasis are discussed in later sections of
this report. The overall viewpoint developed here is that,
in dealing with unrecorded pollution problems other than
combined sewers, agencies should generally avoid heavy com-
mitment to specific planning methodologies until such time
as the types and magnitudes of existing problems are at
least roughly understood. The use of very straightforward
analytical procedures may be adequate, especially in cases
where planning resources are limited. When dealing with
future urban development, the most important issue may be
the selection of problems for emphasis, rather than the
choice of planning methodologies. Given the overall ap-
proach recommended here in Section 7, there are a number of
relatively simple methodologies which could fulfill the
needs of current water planning studies.
83
-------
SECTION 7
ANALYSIS AND CONTROL OF EXISTING PROBLEMS
DUE TO UNRECORDED POLLUTANT SOURCES
Introduction
Water quality problems vary tremendously among urban areas
of the U.S.; thus, it is difficult to generalize regarding
the appropriateness of any given planning approach. The
present section nevertheless attempts to develop an overall
strategy for analysis and control of existing unrecorded
pollutant sources, which may be potentially useful in areas
where planning resources are limited and existing water
quality problems are not well understood. The underlying
rationale for this approach is outlined in the next few
pages, followed by discussion of various technical issues.
As before, the emphasis is upon urban unrecorded pollutant
sources other than combined sewers. It should be emphasized
that "recorded" versus "unrecorded" pollution is a practical
distinction based upon existing documentation and regulation
of pollutant yields, not a physical distinction among source
types (as is true for SRA versus non-SRA sources). A con-
tinuous point discharge, for example, can constitute an
"unrecorded" source if the effluent is not covered by a
discharge permit, or exceeds permitted amounts in terms of
wastewater quantity or strength. A danger in planning
studies is that such pollutant loadings can be attributed to
"nonpoint" sources (e.g., on-site septic systems and street
surface runoff) when nonpoint source loadings are computed
as residuals. Another point to be mentioned is that the
scope of permitting activity is likely to increase in the
future, so that the balance of recorded and unrecorded
pollution may shift.
The present context for control of unrecorded pollution in
most U.S. urban areas could be described as follows. Al-
though many of the governmental mechanisms needed for con-
trol already exist, and although there are some corrective
measures which would entail little cost, the implementation
of controls will generally not be an easy task. In the case
of measures which require public expenditure, the most
critical fact is that Federal funding is usually not avail-
able—at present—as a lever for implementation. This
84
-------
represents a fundamental difference from municipal waste-
water facility planning. The intent here is not to dis-
parage the goodwill of municipalities with respect to pollu-
tion control, but simply to point out that the funding
issue, which is always important, is especially so at the
present time, due to the unfortunate coincidence of water
planning efforts with general stress upon municipal finance.
Except for industrial enclaves, the municipalities in which
unrecorded source control is most needed tend to be those
which are financially most hard-pressed. In the case of
controls which involve major private expenditure (or oppor-
tunity cost), the parties affected could well subject pol-
lution control regulations to legal testing as well as other
forms of challenge.
A second point is that, for a large proportion of urban
unrecorded pollutant sources, the possibility of over-
control is not a serious issue at present. The potential
importance of pollutants from watershed surfaces and waste
conveyance facilities has been established in the litera-
ture. Within the range of control which can feasibly be
achieved in the near future, it is usually safe to assume
that implementation of corrective measures will be desir-
able, if not actually required to meet specific water
quality criteria. For example, removal of pollutants from
streets and other impervious surfaces might not be necessary
in a given case for prevention of organic pollution or
nutrient enrichment; but the reduction thus achieved in
heavy metals loadings would be desirable and potentially
very important. Although little is known about the ultimate
effects of heavy metals and other toxic materials in urban
runoff, their presence in receiving waters must certainly be
regarded as a liability. In addition, "housekeeping"
measures involving watershed surfaces are often justified on
grounds other than water quality alone. Similar arguments
apply to measures such as sewer system maintenance. Thus,
although planners must avoid control measures which are
obviously ineffective and/or discriminatory, there is much
more danger that too little will be done than that too much
will be done.
In view of this set of circumstances, it is legitimate to
question the relevance of traditional planning approaches
when dealing with unrecorded pollution problems. One ele-
ment of these approaches is the emphasis placed upon com-
parison of control alternatives in terms of cost-effective-
ness. The discussion in the previous paragraph suggests
85
-------
that such comparisons may not be generally necessary, given
the range of possibilities which presently exist. A related
point is that formal comparisons and trade-offs of control
alternatives are often not sufficiently accurate to be
worthwhile. As discussed elsewhere, the urban stormwater
models which have been designed for this purpose are weak at
the "front end." Even"if calibrated carefully, they do not
actually establish where unrecorded pollutants are coming
from—e.g., streets, versus parking lots, versus the great
variety of other source areas and mechanisms which exist.
Thus, although the model structure may theoretically allow
simulation of a wide variety of pollutant-generation mechan-
isms, the models may in practice have substantial value only
in dealing with end-of-the-pipe control measures, such as
runoff storage and treatment options. Evaluation of other
types of controls is likely to involve such questionable
assumptions that the advantages of a complex methodological
format are largely nullified. Partly for this reason, it is
felt that trade-off analysis of control alternatives should
be a major focus of attention only in cases where runoff
storage and/or treatment for existing urban development is
considered to be a serious possibility. An hypothesis
offered here is that such cases presently do not include the
bulk of urban land in the U.S. with separate sanitary
sewers.
A general concern is that accepted water planning approaches
tend to be oriented toward criteria which are essentially
irrelevant to the task of establishing what is possible in
terms of unrecorded pollution control. In contrast, the
strategy suggested here is essentially a "bottom up" ap-
proach in which implementation feasibility is recognized as
a dominant issue. Technical planning efforts would not be
seen as attempting to develop.a "best" plan according to any
a priori criteria, but rather as a process of establishing a
favorable set of controls and providing specific documenta-
tion for defense of these controls. The difference in
emphasis is significant in that, due to the complexity of
unrecorded pollution problems, activities conducted in
pursuit of a "best" plan may not be particularly helpful
either in pointing out the control possibilities which
exist, or in defending the control plan which is ultimately
recommended. The alternative is to stress direct examina-
tion of water quality problems, and documentation of pol-
lutant sources at the same level of detail that is relevant
for analysis of controls. Allocation of technical planning
efforts among pollutant sources would be geared closely to
86
-------
the probability that a given item of information would
affect control implementation. Additional aspects of this
approach are discussed in the next sub-section.
An important issue regarding trade-off of controls has to do
with the planning implications of transient water quality
conditions versus conditions at low flow. Arguments have
been made in the literature that transient problems may
often be the more critical in urban areas, and that there-
fore direct substitutability exists between control of
recorded effluent discharges and control of SRA pollutant
sources, (Non-SRA unrecorded sources are ignored here for
simplicity.) One comment is that, if_ water quality is to be
maintained at desired levels during low flow, substitut-
ability should not be a highly important issue in most
instances. The hydrologic conditions requiring the most
stringent control of municipal and industrial wastewater
effluents are likely to consist of extreme low flow, re-
gardless of the relative significance attributed to tran-
sient versus long-term conditions. (For a number of rea-
sons, it appears unlikely that concern with transient prob-
lems will bring about higher levels of treatment for re-
corded effluents than are necessary to meet water quality
criteria at extreme low flow.)
On the other hand, such arguments may imply that transient
conditions place a limit on the level of water quality which
can reasonably be required at low flow. Substitutability
would then exist in the sense that, for any given degree of
SRA source control, the treatment levels for recorded efflu-
ents could be adjusted so that transient water quality
conditions and extreme low flow conditions were somehow
equivalent in terms of constraint on water use. Since this
would require adjustment of water quality criteria, the
implication is that the criteria applied at low flow should
be dependent upon the achievable level of SRA source con-
trol. This would mean that consideration of transient water
quality conditions could erode the basis for advanced treat-
ment of recorded effluents. Such a possibility appears to
be precluded by current EPA regulations, but is nevertheless
noted here as a potential danger. Given the present lack of
knowledge concerning the significance of transient condi-
tions, and the substantial probability that SRA pollutant
sources will continue to be somewhat under-controlled in the
near future, agencies should be skeptical of actions which
could result in under-control of recorded effluents as well.
87
-------
Trade-off between municipal and industrial discharge con-
trol, and control of unrecorded pollution, should not
ordinarily be emphasized as an issue.
Selection of Problems for Emphasis
Given that the present discussion is concerned primarily
with cases in which planning resources are relatively
limited, the issue of allocation of resources is important.
It is convenient when considering this question to divide
planning activities conceptually into two categories:
technical activities and promotional activities. "Promo-
tional" refers in a very general sense to all activities
oriented toward implementation of water quality controls,
including the drafting and promulgation of recommended
plans. It is clear that promotional efforts should be well-
balanced, in that specific problems and solution measures
should be given emphasis roughly in proportion to their
perceived importance. Considerable latitude exists in the
allocation of technical resources, however. The present
discussion considers briefly the question of allocating
technical resources among: (1) types of unrecorded pollutant
sources; and (2) types of in-stream water quality problems.
Perhaps the most important issue with regard to pollutant
sources is the extent to which attention should be focused
upon washoff of pollutants from street surfaces. When
considering any pollutant source in the present context, the
relevant criteria for allocation of technical effort would
appear to be the following: (1) the relative importance of
the source in terras of loadings; (2) the amenability of the
source to control; and (3) the likelihood that technical
planning efforts will make a decisive difference to imple-
mentation of controls. Street surface washoff .clearly ranks
high in terms of the first criterion, although the very
heavy emphasis upon street surfaces in the literature as the
origin of urban runoff pollutants may be somewhat unjus-
tified. It is felt, however, that street surfaces rank
relatively low in terms of the other two criteria. Control
of the problem through street sweeping is hampered by the
relatively low efficiency of most sweeping equipment, the
problem of parked cars, the generally poor state of munici-
pal finance, and the fact that this measure is most needed
88
-------
in communities which can least afford it.* End-of-the-pipe
measures tend to be limited by the great expense of instal-
ling runoff storage facilities in existing neighborhoods.
With regard to the role of technical research in encouraging
implementation, it is likely that communities will intensify
their street sweeping programs and other housekeeping
activities voluntarily, or not at all. For a variety of
reasons, technical inputs other than general descriptions of
the problem are not expected to exert a great deal of
leverage.
Thus, given any reasonable weighting of the abovementioned
criteria to form an overall index, it is felt that street
surface pollutants would rank relatively far down the list
of unrecorded pollutant sources, as a focus of technical
planning activity. The resources allocated to this subject
would then depend upon an agency's perception of its im-
portance relative to the overall level of pollutant reduc-
tion which can or must be achieved. If street surfaces are
thought to account for 75% of unrecorded pollutant loadings,
and if the minimum acceptable reduction in total loadings is
50%, then street surface runoff obviously should receive a
great deal of attention; but it is not clear that such a
case would be typical. The recommended approach in most
instances is therefore to encourage municipal housekeeping
measures strongly in promotional activities, but to minimize
the investment in technical backup for these measures (many
of which, such as improved garbage collection practices and
general cleanup operations, cannot be evaluated quanti-
tatively in any case). With regard to street surface con-
taminants per se, it is suspected that, although this
problem is not uncontrollable, the most important gains in
the long run will be achieved by change in the pollutant-
generating characteristics of motor vehicles, improvement in
air quality, and change in personal habits, rather than by
removal of pollutants once present on street surfaces.
* Given the "gap" in municipal services which is often felt
to exist in central cities, it is interesting to speculate
upon the reactions of ghetto residents when their streets
are forcibly cleared of cars and swept for the fourth or
fifth time in a given month.
89
-------
Finally, one overall problem created by emphasizing street
surface pollutants is that it reinforces the view that
"everyone is responsible" for pollution problems. This view
is partially valid, although some are more responsible than
others. The difficulty is that this view could have a
generally debilitating effect upon implementation of con-
trols. There are a number of arguments, some perhaps
cynical, for focusing at the present time upon the "bad
guys"; this could be viewed explicitly as an interim tac-
tical approach. It is felt that these localized influences
can be shown to constitute a large proportion of unrecorded
pollution in many areas, if agencies take the steps neces-
sary to document their existence.
i
The major issue regarding selection of water quality prob-
lems for analysis is the extent to-which attention should be
focused upon transient water quality problems, which occur
during and immediately after surface runoff events, as
opposed to long-term problems. The latter are defined as
water quality conditions, produced by SRA sources, non-SRA
sources, and/or recorded effluent discharges, which can be
observed during some or all periods of steady-state flow.
Transient water quality problems due to urban runoff are
generally limited to the following: (1) the direct effects
on aquatic biota of temporarily high concentrations of toxic
materials; (2) the effects of temporary dissolved oxygen
depletion on aquatic biota; and (3) limitation of water uses
involving human contact due to high pathogen concentrations
following storms. Obviously, the extent to which transient
problems must be considered depends upon their scope and
magnitude in a given study area. Several general comments
can be offered, however.
1. Except possibly in the case of pathogens, an
integral aspect of water quality planning on the
basis of transient conditions is to recommend
appropriate standards for receiving water bodies,
as well as to indicate how these standards can be
met. The reasons are that: (a) the water con-
stituents other than dissolved oxygen which are
relevant to transient biologic impact are general-
ly not covered by existing stream standards; and
(b) the standards which exist have virtually
always been established through reference to
steady-state conditions, and may not be appropri-
ate for control of dynamic conditions. Therefore,
if transient chemical phenomena•are considered, it
90
-------
is necessary to deal at a fundamental level with
their importance—specifically, their impacts on
aquatic biota—as well as to establish the exis-
tence and magnitudes of such phenomena.
2. A very slender base of information exists upon
which to establish the biologic importance of
transient pollutant loadings. Bioassay research
has dealt almost exclusively with time intervals
of several days or more (e.g., 96 hours). Few, if
any, attempts have been made to simulate directly
the biologic stress produced by stormwater, which
may involve important interactive effects among
pollutants. (These issues are dealt with in
considerable detail in the Technical Appendix to
this volume.) Given the critical relationship of
water use to aquatic life, the general lack of
biologic research in support of urban water plan-
ning studies is considered remarkable.
3. It goes without saying that organic loadings in
stormwater, expressed as oxygen demand, are
important only to the extent that transient
oxygen stress is in fact produced. If oxygen is
not affected significantly, for whatever reason,
these loadings are not an appropriate planning
parameter. Evidence suggests that organic load-
ings from urban land without combined sewers are
not necessarily problematic; and that transient
oxygen problems in general are extremely dependent
on receiving water characteristics. The optimal
conditions for transient oxygen depletion involve
a light, localized rainfall which delivers a
"first flush" of pollutants to a receiving water
body which is already in marginal condition and
for which the reaeration rate is low—followed by
a period of one or more days in which little
additional dilution is received. These conditions
are most likely to occur in an estuary or standing
water body during a period of generally dry
weather. They are considerably less likely to
occur in a free-flowing stream.
Due to these and other factors, it is felt that agencies
should generally attempt to minimize the resources devoted
to direct analysis of transient water quality problems.
That is, (1) agencies should not assume that such problems
91
-------
exist, especially in the absence of combined sewers; and (2)
major efforts should not be devoted to dynamic receiving-
water analysis unless the importance of transient conditions
can be established through direct observation (as opposed to
simulation). A determination that transient problems need
not be a major focus of attention, for part or all of a
given study region, has potentially far-reaching implica-
tions for technical planning efforts, since it means among
other things that stormwater modeling is probably not
necessary.
This view applies to transient water quality problems, not
transient pollutant generation. Clearly, any quantitative
analysis of SRA pollutant sources must deal in some fashion
with short-term loading rates, and their relationships to
discharge and other factors during storm events, if only for
the purpose of computing long-term loadings accurately. The
above suggestion applies only to the direct effects of
transient loadings on water use, when analyzed for indi-
vidual storm events. A second comment is that all influ-
ences on water quality exerted by benthic material (other
than benthic material which is resuspended during periods of
high discharge) are considered here as "long-term" problems,
due to the fact that benthic influences persist after the
direct runoff from a given storm has been flushed from the
surface water system. The distinction between transient and
long-term problems is admittedly arbitrary in some instan-
ces, but is very significant in terms of the manner in which
problems are described and analyzed.
Even when the critical conditions for a given problem
clearly occur during storm events, it may be adequate in
some instances to avoid dynamic receiving-water analysis by
employing long-term descriptions of the problem. For
example, suppose that monitoring of a given water body
indicates that the concentration of some toxicant exceeds a
critical value in one day out of every five, corresponding
to-days in which rainfall occurs. This information would be
adequate to establish that a problem exists, and that SRA
pollutant sources are at least partially responsible.
Suppose also that: (1) the total loading of the given toxi-
cant at the problem area in question can be computed for a
typical year or season; (2) the portion due to unrecorded
sources can be estimated (as a residual); and (3) additional
monitoring activities establish the existence of one or more
specific SRA sources of the toxicant which account for
disproportionately large shares of the total unrecorded
92
-------
loading (i.e., produce very high yield on a per acre basis).
It is believed that this could represent an adequate base of
information on which to design, defend, and implement con-
trol measures dealing with the specific sources in question.
This information would not, however, be sufficient to esti-
mate reliably the percentage of the time that the toxicant
concentration would exceed the critical level after imple-
mentation of the given controls. Even the reduction in
loadings, at the point where the problem has been observed,
might be difficult to estimate with accuracy, due to the
uncertain efficiency of controls and perhaps questions
involving transport/delivery of pollutants. The appropri-
ateness of this approach thus depends upon: (1) the rela-
tive importance of localized pollutant sources, and (2) the
issues discussed earlier in this section regarding develop-
ment of a "best" plan and the need to trade off different
forms of control.
A^Possible Approach for Analysis and Control of Existing
Problems
The present sub-section outlines at a conceptual level an
approach to remedial water quality planning which incor-
porates the ideas discussed thus far. This approach repre-
sents what might be done in a "worst case" planning situa-
tion, in which funds are limited, water quality problems are
diverse, the nature of unrecorded sources is largely un-
known, and the recommended plan is expected to face serious
challenge. The essential elements of the technical planning
problem are represented schematically in Figure 6. The
focus is upon urban unrecorded pollutant sources generally;
SRA and non-SRA sources are not distinguished except where
noted in the discussion.
Two overall classes of technical activities are distin-
guished: problem analysis, and source analysis. A more
balanced representation would include at least one addi-
tional class of activities, involving identification and
evaluation of potential control measures as a third class.
Problem analysis, based upon observed water quality data, is
considered to be perhaps the most important technical ele-
ment of current water planning studies. The outputs of this
element include the following: (1) quantitative description
of key water quality problems, by type, location, and rele-
vant water quality parameters; (2) total loadings of rele-
vant pollutants at problem locations; and (3) net loadings
93
-------
Problem Analysis
KEY WATER
QUALITY
PROBLEMS, BY
TYPE AND
LOCATION
RECORDED
EFFLUENT
DISCHARGES
TOTAL LOADINGS
OF RELEVANT
POLLUTANTS AT
PROBLEM LOCATIONS
1
±
URBAN UNRECORDED
POLLUTANT LOADINGS
AT PROBLEM LOCATIONS
REQUIRED
REDUCTION
IN RECORDED
EFFLUENT
DISCHARGES
1
REQUIRED
REDUCTION IN
UNRECORDED
POLLUTANT
LOADINGS
i
1
Source Analysis
1
i
IDENTIFICATION AND
DOCUMENTATION OF
HIGH-YIELD, SITE-
SPECIFIC POLLUTANT
SOURCES
'
k
CONTROL MEASURES
DEALING WITH
SITE-SPECIFIC
POLLUTANT SOURCES
l__
mt
mj
MIM
mm
mm
LAND DATA
BASE, FOR
HYDROLOGIC
SUBDIVISIONS
*
GENERAL
POLLUTANT
GENERATION
RELATIONSHIPS
QUANTITATIVE
EVALUATION
OF POTENTIAL
CONTROLS
1
CONTROL MEASURES
DEALING WITH
WIDESPREAD
POLLUTANT SOURCES
I
RECOMMENDED TECHNICAL PLAN
Source: Betz Environmental Engineers, Inc.
Figure 6 TECHNICAL PLANNING ACTIVITIES RELEVANT TO EXISTING UNRECORDED SOURCES
-------
at problem locations due to urban unrecorded sources (ob-
tained as residuals). Reference is made to "key" water
quality problems because of the fact that it may be advan-
tageous and/or necessary to be highly aelec* ive wnen choos-
ing in-stream problems for intensive analysis. Selectivity
applies to both the receiving water locations and the types
of problems which are considered. In the case of location,
the objective is to focus upon those points for which con-
trol of existing water quality problems would necessitate
£he greatest reduction in pollutant loadings. The selection
process is analogous to finding the.oxygen sag point when
analyzing recorded discharges under steady-state conditions;
but for a variety of reasons it is potentially much more
complex. (One general difficulty is that the most critical
problems tend to occur in those water bodies for which
analysis is most complicated, due to multiplicity of pol-
lutant sources.) In the case of water quality problems, and
associated chemical constituents, selectivity involves de-
emphasis of problems which are: (1) relatively unimportant;
or (2) secondary in terms of the required stringency of
control. As an example of the latter criterion, a water
body might be affected both by sedimentation and nutrient
enrichment; but analytical attention might be focused only
upon the latter problem, if it is judged that control of
nutrient loadings will necessarily entail adequate control
of suspended solids.
Estimation of pollutant loadings and concentrations at
problem locations is distinguished in Figure 6 from analysis
of the problem per se, in order to emphasize that the two
activities are not necessarily the same (unless a given
pollutant is defined to be a problem by existing water
quality criteria). Two important points in this regard are
that: (1) the investigation of problems can properly in-
clude a much wider variety of evidence than chemical analy-
sis of water quality (examples are analysis of benthic
materials, aquatic communities, and the tissues of aquatic
organisms); and (2) the descriptions of water chemistry
which are most relevant for problem assessment tend to
consist of pollutant concentrations, whereas linkage to
sources requires description in terms of loading rates. The
need to analyze loading rates as well as concentrations is
discussed later in this section, along with various issues
involving measurement of loadings. In general, it is teit
that the loading rates used in problem analysis should be
descriptive of long-term conditions unless it is very clear
that transient events are the dominant influence on water
quality.
95
-------
Source analysis, depicted in the right-hand portion of
Figure 6, also involves measurement of pollutant loadings.
However, the points at which such measurements are made
typically do not coincide with the key locations for problem
assessment, due to the emphasis in source analysis upon
isolating the pollutant yields of particular unrecorded
sources. It is emphasized again, with regard to analysis of
unrecorded pollutant sources, that such analysis must deal
with all influences on water quality other than effluent
discharges currently specified in discharge permits. Any
source mechanisms can be involved; and the relative im-
portance of different source mechanisms should not be
prejudged.
As indicated in Figure 6, source analysis involves three
types of activities which logically precede the design and
evaluation of controls. These are: (1) preparation of land
data relevant to unrecorded pollution; (2) estimation of
general pollutant-generation relationships; and (3) identi-
fication and documentation of specific pollutant sources.
These activities involve a categorization of unrecorded
pollutant sources along the lines suggested by earlier
discussion of site-specific versus ubiquitous pollutant
generation factors. It is hypothesized that, except for
core districts of major cities, and other areas with com-
bined sewers, the highest per-acre pollutant yields tend to
be associated with site-specific conditions (e.g., dumping,
sewer leaks, unauthorized discharges, handling of unusual
materials) rather than simply reflecting variation in
ubiquitous factors such as traffic and littering. Even if
the high-yield sources are associated with land uses that
are widespread in an area—e.g., manufacturing, petroleum
wholesaling and retailing, trucking operations--they tend to
be site-specific in that only a minority of establishments
of a given type are actually involved. Site-specific
sources are thought to be generally propitious for technical
analysis since these sources tend to be relatively amenable
to control once identified, due to the moderate public
expenditure required in many cases, and the frequent exis-
tence of precedents and linkages to ongoing programs. Thus,
detection and documentation of specific sources is dis-
tinguished from investigation of the overall pollutant-
generation characteristics of urban land. Although the
former activity may lead to controls which apply_to more
than one location, the important characteristic is that the
design and defense of these controls is based upon analysis
of highly specific cases.
96
-------
The development of pollutant-generation relationships
includes any activities which relate pollutant loadings to
land use, imperviousness, population, employment, or other
general characteristics of urban land. These activities can
be oriented toward three overall objectives: (1) prediction
of pollutant loadings at problem locations; (2) evaluation
of the potential effectiveness of control measures; and (3)
determination of baseline levels of pollutant generation for
use in reconnaissance studies (discussed below). A general
problem is that pollutant-generation relationships tend
either to be unreliable, or to omit important site-specific
pollutant sources (due to the characteristics of study areas
upon which the relationships are based). Also, as noted
earlier, relationships based upon monitoring of small catch-
ments may not be directly applicable for prediction of
pollutant loadings in major watercourses, due to factors
involving pollutant delivery. These are reasons why the use
of pollutant-generation relationships for prediction is
considered inferior to direct measurement of pollutant
loadings at problem locations, even though unrecorded load-
ings must typically be obtained in the latter case as a
residual. With regard to evaluation of controls, a charac-
teristic of pollutant-generation relationships noted earlier
is their limited usefulness for analysis of land management
measures (as opposed to end-of-the-pipe measures) due to
failure to deal with land use at the required level of
specificity.
Thus, agencies might not attempt to characterize pollutant
generation on a comprehensive basis, except in a very simple
fashion. The objective of this strategy in terms of data
collection would be to minimize expenditure of resources in
collecting "intermediate level" data—i.e., information
which does not directly pertain either to key receiving
water problems, or to pollutant sources at the level of
detail needed to evaluate controls. Referring again to
Figure 6, technical planning efforts would be devoted
primarily to the items situated to the left of the dashed
line. Control measures dealing with ubiquitous factors such
as street dirt would be formulated and actively promoted as
part of the plan; but only limited efforts would be devoted
to quantifying pollutant generation due to these factors,
and estimating the effectiveness of proposed controls. On
the other hand, the effectiveness of controls for site-
specific sources could often be estimated with considerable
accuracy, due to the nature of these sources and the docu-
mentation which would be provided.
97
-------
As indicated by the left-hand side of Figure 6, it is
assumed that required treatment levels for municipal and
industrial effluents can be determined more or less inde-
pendently, relative to the process of establishing controls
for unrecorded sources. Two relevant points from previous
discussion are that: (1) extreme low flow will usually
constitute the limiting condition for design of continuous
point source controls; and (2) consideration of transient,
storm-related problems should ordinarily not affect the
stream standards that apply at low flow. An additional
factor is that industrial treatment levels may be largely
predetermined. Thus, the major issue is the extent to which
unrecorded sources contribute to problems at low flow which
serve as the basis for design of. municipal effluent con-
trols. This issue could involve pollutant loadings from SRA
sources as well as non-SRA sources, insofar as the former
are retained in local surface waters through deposition in
benthic deposits and other mechanisms. The cases which
would appear most relevant in terms of trade-off with muni-
cipal treatment levels would be: benthic oxygen demand and
benthic BOD loads, due to pollutants from SRA sources; and
nutrient loadings, involving both SRA and non-SRA sources.
A more complete version of Figure 6 would thus perhaps show
a mutual interaction between unrecorded source control and
recorded effluent control in these instances. However, only
in the case of nutrients is it felt that such interaction
might justify major efforts to develop comprehensive pre-
dictive equations for unrecorded loadings (still assuming
that combined sewers are not present). These equations
would apply primarily to nutrient loadings from agriculture
and domestic on-site sewage disposal.
To complete the steps shown in Figure 6, the total allowable
pollutant loadings at each problem location would be deter-
mined on the basis of problem assessment and the magnitude
of existing loadings and concentrations. Given the recom-
mended treatment plan for recorded effluents, the required
reductions in unrecorded pollutant loadings could be estab-
lished (typically, using descriptions of loadings which
acknowledge variation due to hydrologic conditions). The
extent to which these reductions exceed the expected abate-
ment of site-specific sources would indicate the need for
more general control of urban unrecorded pollution. As
suggested above, major efforts would not be expended in
attempting to gear the control plan for general unrecorded
pollution sources to the required loading reductions, al-
though estimates of control effectiveness would be prepared
in some fashion.
98
-------
An unavoidable problem with this approach is the difficulty
of tracking down specific pollutant sources within a large
study area. It is clear that some form of land data base
and general loading relationships would be very helpful for
this task (as well as for estimating loadings at problem
locations which are not monitored directly) . The objective
would be to work back as efficiently as possible from
problem locations to high-yield source areas. In order to
execute this and other tasks, major watersheds would be
divided into sub-basins, and ultimately into hydrologic
subdivisions of perhaps one to five square miles each. The
role of pollutant-generation relationships would be to
indicate, on the basis of land data for a given catchment
area, the pollutant loadings and/or concentrations which
would be expected to occur under normal circumstances.
These "baseline" predictions would provide a means of evalu-
ating field data at each step, and selecting relatively
high-yield areas for further attention. Very simple rela-
tionships could be utilized for this purpose. (An extreme
case would be simply to select some variable such as popula-
tion or impervious surface and use this as a divisor to
"standardize" the data observed in the field.) An extended
example of such a reconnaissance process, utilizing em-
pirical relationships developed as part of the present
study, is presented in the Technical Appendix.
In addition to activities such as just described, agencies
may be able to select certain areas for monitoring on an a
priori basis. Likely candidates are industrial districts
and communities known to have deteriorating sewer systems.
Regardless of the manner in which high-yield areas are
identified, it becomes necessary at some point to conduct
on- site inspection of streams and potential unrecorded
pollutant sources, followed by highly selective monitoring
of the latter (which can potentially include permitted
discharges, if these are not already monitored). Although
one can argue that such activities are not appropriate for
current water quality management studies, the viewpoint
adopted here is that there will probably be^no better time
(or agency) for this task, and that unrecorded pollution
often cannot be well understood without direct observation--
even if only on a sample basis. It is well to remember that
current hypotheses regarding unrecorded pollution in urban
accumulation, and theoretical considerations. Little
99
-------
actually known about precisely where the bulk of pollutants
are coming from, in either typical or untypical situations.
Successful application of this approach requires (1) a
convincing demonstration that a water quality problem
exists, and (2) a demonstration that a specific source
contributes a disproportionately large share of the pollu-
tant loadings responsible for that problem. Notice that the
latter demonstration may be very imprecise in the case of
SRA sources, since a major proportion of loadings from a
given source may not reach a given problem location. Thus,
a convincing demonstration that a source is especially
important may require that the long-term, at-source loading
rate be at least 5 or 10 times as great, on a per acre
basis, as the total loading received from urban land at the
problem location. Such a differential is particularly
necessary when source documentation involves direct monitor-
ing of runoff from an impervious surface. Nevertheless, it
is felt that such sources commonly exist in urban areas, and
may account for very large proportions of unrecorded load-
ings, particularly in the case of water constituents other
than organics, nitrate, and suspended solids. (This dis-
cussion does not apply to areas with combined sewers, for
which fundamentally different approaches are likely to be
necessary.)
A final comment regarding land data and pollutant-generation
relationships is that, regardless of the planning approach
utilized, there is ordinarily not a real need for collection
of highly detailed urban land use data. One reason is
simply that not enough is known about pollutant generation
to utilize detailed land use information efficiently for
predictive purposes. As already indicated, many of the most
important effects of urban land are likely not to be cap-
tured by such data even if the relevant relationships are
understood. Various empirical studies suggest that the
maximum usable base of urban land information, other than
data obtained from field inspection and specialized sources
such as sewer plans, might consist of the following vari-
ables, measured for hydrologic subdivisions: population,
impervious surface, employment, land use (acreage of land in
5 to 10 categories); and the age and sewerage of dwelling
units. Socio-economic variables such as income could have
value as predictors, but are generally difficult to inter-
pret. As noted earlier, the case in which estimation of
detailed predictive relationships is likely to be most
useful is analysis of domestic on-site sewage disposal. In
100
-------
this case, however, the need for detail relates primarily to
natural land characteristics such as soils, rather than to
land use.
Estimation of Pollutant Loadings
Pollutant "loadings" refer here either to rates of mass flux
per unit time, or to the total mass load over a given time
period (e.g., a storm period or a season). Commonly, inves-
tigations of water quality problems focus upon pollutant
concentrations rather than loadings, since concentrations
are more directly relevant to water use and are the measure
in which water quality criteria are normally expressed.
However, loadings must be computed for purposes of source
analysis, whenever it is necessary to net out the influence
of one or more specific factors (e.g., municipal or indus-
trial discharges) or to deal with SRA pollutant sources.
The need to consider loadings as well as concentrations in
current water quality management planning studies is di-
rectly implied by the requirement that pollutant allotments
be developed.
The present discussion focuses upon the estimation of pol-
lutant loadings for water bodies affected by SRA sources.
The need to deal with non-steady-state conditions, highly
variable pollutant concentrations, and extended time inter-
vals makes this situation fundamentally different from
analysis of pollutant loadings at base flow. It is assumed
at present that the loading analysis will not involve extra-
polation to "critical conditions" involving a specific
design storm. In such situations, stormwater modeling would
ordinarily be utilized (although it is possible that equal
or greater accuracy could be attained through simple hand
computations, based upon ad hoc assumptions). The objective
instead would be to estimate total loadings over an interval
such as a season or a year. This is appropriate v/hen the
direct linkages between storm period loading rates and water
quality either cannot be quantified accurately, or can be
expressed on a long-term basis. (As an example of the
latter, it might be possible to determine that an average
of, say, 25% of phosphorus inputs to a lake during summer
storms are retained in the water after the storm period, and
thus are available to promote overgrowths of aquatic flora.)
Some form of extrapolation is nevertheless required for
development of long-term loading estimates, due to the fact
that it is rarely possible to observe all of the relevant
conditions directly. Loading estimation can therefore be
101
-------
viewed as a problem of statistical inference, in which it is
necessary to consider the variability of the factors in-
volved. In order to avoid confusion in the present dis-
cussion, the important variables are defined below. The
assumed objective is to compute the total load of some
pollutant (P) over a time interval which is T days in
length. Instantaneous points in time within this interval
are denoted by the subscript "i". It is assumed that the
behavior of pollutant concentrations and stream discharge
during the interval is completely characterized by the
values of these variables at "n" equally-spaced points in
time (where i = 1, ...» n, and n is an indefinitely large
number). Since it is impossible to observe pollutant con-
centrations and discharge at all n points, the missing
values must be implicitly assumed. Thus, the objective in
choosing a computational procedure is to utilize a form
which incorporates reasonable assumptions regarding the
missing values.
Quantities Pertaining to Instantaneous Points in Time ("i")
Qi = stream discharge in cubic feet per second at time i
Ci = pollutant concentration in mg/1 at time i
Li = pollutant loading rate in pounds per day at time i
Li = 5.4 QiCi
Quantities Pertaining to a Finite Time Interval
T = length of interval in days
n = number of points in time necessary for complete charac-
terization of pollutant loading behavior (i=l,...,n)
D = average discharge during interval (cfs)
1^
D = n (Ql + Q2 + ... + Qn)
102
-------
A = discharge-weighted average pollutant concentration
(mg/1)
Qlcl + Q2C2 + ... + QnCn
A = Ql + Q2 + ... + Qn
R = average pollutant loading rate (pounds/day)
.L
R = n (LI + L2 + ... + Ln)
P = total pollutant load during interval, in pounds
P = RT
or, P = 5.4 DAT
As indicated by the last two formulas (which are equiva-
lent) , there are two basic approaches to the problem of
estimating a pollutant load for a given time interval. One
is to estimate the average loading rate (R, in pounds per
day) and multiply by the number of days in the interval.
The alternative is to compute an overall loading rate for
the interval, using average discharge and the discharge-
weighted average pollutant concentration (A); and then to
multiply by the number of days involved. The significant
difference is that in the latter case it is usually possible
to utilize an estimate of average discharge, D, which is
based upon more comprehensive information than obtained in
the water quality sampling program per sje. For example, if
the interval in question is a summer season, average summer
discharge could be computed from long-term gaging station
records for either the given stream or a comparable nearby
basin. The estimation of D in this fashion usually involves
small errors relative to the overall error variance in
loading estimates. Thus, the major issue in choosing be-
tween the two formulas for P is the relative accuracy with
which R and A can be estimated.
For virtually all stream locations affected by SRA pollutant
sources, instantaneous loading rates are much more variable
over time than pollutant concentrations. Thus, when utiliz-
ing simple averaging methods to compute the total pollutant
103
-------
load, it is generally better to deal with A than R. That
isr it is safer to assume that observed values of pollutant
concentrations are representative of all values, than to
assume that observed loading rates are representative. This
is true whether or not storm periods are isolated from non-
storm periods in the computations.
The most accurate methods of computing pollutant loadings
take explicit account of association between pollutant con-
centrations and discharge. Observed values of C or L are
related statistically, usually through logarithmic regres-
sion, to corresponding values of Q. The resulting relation-
ship is then linked to a flow-duration curve (which indi-
cates the proportion of the time that discharge is within
given ranges, for the entire interval under consideration).
This makes it possible to establish average conditions for
the interval even if the observations at hand are known to
be untypical in terms of discharge. Since the choice of C
or L is immaterial in this approach, and since comprehensive
discharge data are required in any case, further discussion
will focus only upon the use of concentrations rather than
loading rates.* It is important to recognize that the
behavior of pollutant concentrations during storm periods is
likely to reflect a number of factors, as follows.
1. Dilution. Continuous sources which discharge at a
constant rate will affect in-stream pollutant
concentrations by an amount which is inversely
proportional to discharge. A less-than-full
dilution effect may prevail for sources involving
groundwater outflow.
2. Washoff. In the case of washoff of materials from
impervious surfaces, the rate of pollutant trans-
port to surface waters is largely dependent upon
* When L is related by logarithmic regression to Q, the
regression coefficient obtained for log Q is equal to
unity plus the value which would have been obtained if
log C rather than log L had been the dependent variable.
R-square is almost always higher in the former case, but
the standard error of the regression coefficient for log
Q is the same in both cases.
104
-------
the amount of material available for transport.
The resulting pollutant concentrations should
therefore be positively related to the time since
previous rainfall (although the nature of this
relationship may be uncertain), and should decline
throughout the course of a storm as material is
removed. Except for the initial "first flush"
period, pollutant concentrations should be nega-
tively related to the rate of runoff.
Erosion. The important characteristic of erosion
in the^ present context is that pollutant loading
rates are limited by the ability of rainfall to
detach and transport particulate matter, rather
than by the amount of material present. Two
classes of erosion processes can be distinguished,
corresponding to the two basic types of overland
flow. Erosion associated with Hortonian overland
flow (which occurs when the rainfall rate exceeds
the soil infiltration rate) would affect pollutant
concentrations throughout a storm, by amounts
which are positively related to the rates of
rainfall and runoff. Washoff of pollutants from
impervious surfaces can behave in a similar
fashion, when pollutants have accumulated to a
considerable depth, or consist of coarse-grained
particulate matter. The other category of over-
land flow, namely surface runoff due to progres-
sive saturation of the soil, would tend to yield
increasing quantities of eroded material as a
storm proceeds. (Pollutant yields from sanitary
sewer bypasses which are activated by inflow of
groundwater might resemble loadings due to satura-
tion overland flow.) On the whole, erosion
processes are distinguished from washoff processes
in that the resultant pollutant concentrations
tend to be positively related to the rate of
runoff, and unrelated or negatively related to
time since the previous storm. Also, in cases
where both washoff and erosion are operative
(which include a majority of watersheds in urban
areas), pollutant loadings due to erosion proces-
ses tend to occur later in a storm period than
loadings due to washoff.
105
-------
The purpose of the above discussion is simply to emphasize
the complexity of the mechanisms whereby in-stream pollutant
concentrations are determined. Except for very small urban
basins which are clearly dominated by washoff, the behavior
of pollutant concentrations over time should never be pre-
judged, either in conducting empirical studies or in apply-
ing stormwater models.
Returning to the loading estimation problem, if a pollutant
concentration is linked only to discharge, the relationship
obtained can be applied directly to compute expected con-
centrations for various ranges of discharges Although a
discharge-weighted average concentration could then be de-
veloped, the simplest way to obtain the total loading for
the time interval in question is to compute a loading rate
for each range of discharge, and then to multiply this by
the percent of the time that discharge is within the given
range, and sum over all discharge categories considered.
Greatest accuracy is potentially obtainable when this pro-
cedure is carried out separately for different hydrologic
conditions, e.g., steady-state conditions, rising stage, and
falling stage. However, it may not be possible to partition
the long-term discharge records accordingly in order to
obtain the necessary flow-duration information. When pol-
lutant concentrations are highly variable, meaning that
great accuracy is not obtainable in any case, it may be
adequate to utilize only a few discharge categories for
these computations. Section 11 depicts a case in which an
annual sediment loading was computed using only four cate-
gories. Further detail was not warranted, even though a
large number of observations were available.
An alternative procedure which should be noted is that of
estimating total pollutant loads for several discrete storm
periods, on the basis of numerous samples in each period,
and then relating these loads to precipitation variables in
order to extrapolate to long-term loadings. (This is basi-
cally what is accomplished by STORM, when calibrated for a
particular basin.) Although there may:be some difficulty in
defining storm periods for large watersheds, this procedure
is basically valid. However, the number of storms monitored
is rarely sufficient to allow critical examination of the
influence of precipitation variables on loadings. The
extrapolation process must therefore be either very simple,
or based upon a priori assumptions. Given the potential
dangers of the latter, it is felt here that procedures based
106
-------
on intensive monitoring of individual storms are not par-
ticularly advantageous for estimating long-term loadings.
It may be better to spread the sampling program over as many
storms as possible, in order to maximize the probability of
capturing the full range of conditions which exist.
A general concern is the extent to which loading estimates
are likely to involve random error and/or systematic bias.
An important aspect of pollutant concentrations is that they
tend to involve a substantial degree of random variation—
i.e., variation over time which cannot be explained by
variables such as discharge, time since start of storm, time
since previous storm, and average intensity of precipita-
tion. The variability, of pollutant concentrations has been
examined in the current study by analyzing the stormwater
data obtained by AVCO for 15 urbanized basins in Tulsa,
Oklahoma (AVCO, 1970; see Section 5). Table 8 presents a
tabulation, based on these data, of the number of observa-
tions required to estimate the average concentration of a
given constituent to a desired degree of accuracy and con-
fidence. Accuracy is expressed in terms of a multiplicative
factor, listed in the left-hand column. For example, sup-
pose that 8 independent observations of storm runoff from a
given basin are available, and that the mean BOD concentra-
tion observed is 9.0 mg/1. Table 8 indicates that there is
a 75% probability that the true average BOD concentration is
between 6.9 mg/1 (= 9/1.3) and 11.7 mg/1 (= 9(1.3)). Simi-
lar tables have been prepared using pollutant concentrations
which have been adjusted to eliminate variation due to
precipitation factors. However, the required numbers of
observations are reduced by only 10% to 15% relative to the
figures shown in Table 8.
Table 8 thus provides a general indication that large num-
bers of observations may be necessary, due to the extent of
random variation, in order to characterize pollutant con-
centrations accurately. The Oklahoma basins were all small
catchments (between 0.1 and 1.5 square miles); somewhat less
variability would be expected for larger watersheds. Also,
the required number of observations might be significantly
reduced if composite sampling were employed. On the other
hand, discharge-weighted average concentrations, which are
more relevant for loading analysis, are somewhat more liable
to error than unweighted averages. Another ifactor which
should be noted is that single chemical samples do not
necessarily represent accurately the total cross-section of
107
-------
TABLE 8
NUMBER OF OBSERVATIONS REQUIRED TO ESTIMATE AVERAGE
CONCENTRATIONS OF WATER CONSTITUENTS,
BASED ON OKLAHOMA DATA
BOD:
COD:
TOC:
OKN:
(Organic
Kjeldahl
Nitrogen)
OPR4:
Tsbluble
Ortho-
phosphate)
SS:
(Suspended
Solids)
Accuracy
Factor
2.0
1.7
1.5
1.3
1.2
1.1
2.0
1.7
1.5
1.3
1.2
1.1
2.0
1.7
1.5
1.3
1.2
1.1
2.0
1.7
1.5
1.3
1.2
1.1
2.0
1.7
1.5
1.3
1.2
1.1
2.0
1.7
1.5
1.3
1.2
1.1
Confidence Level
50%
1
1
2
3
6
20
1
1
1
2
5
15
1
1
1
2
4
13
1
1
1
2
3
8
1
1
1
1
2
5
2
2
3
7
15
53
75%
2
2
4
8
16
56
1
2
3
6
12
43
1
2
3
5
11
37
1
1
2
4
7
24
1
1
1
2
4
13
3
5
9
21
43
155
90%
3
4
7
16
32
114
2
3
5
12
24
88
2
3
5
10
21
76
1
2
3
7
13
48
1
1
2
4
8 -
27
6
11
18
42
87
316
95%
4
6
9
22
45
162
3
4
7
17
35
125
3
4
6
15
30
107
2
3
4
9
19
68
1
2
3
5
11
38
9
15
25
60
123
448
Source: Computed from statistical analysis of data for 15
urbanized basins in Tulsa, Oklahoma (AVCO, 1970).
See Technical Appendix.
108
-------
flowing water in a stream. For example, Colston (1974, pp.
55-56) found that pollutant concentrations near a stream
bottom may be systematically higher than surface concen-
trations.
A circumstance which is important to consider, since it
occurs frequently in research studies, is the use of un-
weighted average concentrations instead of discharge-weight-
ed average concentrations when computing total loadings.
This procedure clearly tends to produce unfortunate results
when storm periods are not isolated from nonstorm condi-
tions; but it is sometimes adequate when storm periods are
treated separately.* Considering just storm conditions,
three types of situations theoretically can occur: (1)
pollutant concentrations are uncorrelated with discharge
(over all points "i" in a typical series of storm periods);
(2) pollutant concentrations are significantly related to
discharge, and the observations available for computation of
loadings are randomly distributed over time; or (3) con-
centrations are significantly related to discharge, and the
available observations are non-random. In the first case, a
simple average of pollutant concentrations represents an
unbiased estimate of the discharge-weighted average, so that
there is no problem in using an unweighted average as an
estimate of "A" (see above). In the second case, use of an
unweighted average will impart a bias to the loading esti-
mate obtained; but the bias may not always be serious.
Storm-period pollutant concentrations in urban watersheds
frequently bear very mild overall relationships to stream
* Isolation of storm and nonstorm periods can be accom-
plished by a variety of methods, ranging from formal
base flow separation techniques to simplistic assump-
tions. Pollutant loading estimates do not appear to be
highly sensitive to the methods used, as long as both
storm and nonstorm loadings are ultimately considered,
and as long as storm periods include nearly all non-
steady-state conditions. (The definition and signifi-
cance of snowmelt conditions have not been considered
here.)
109
-------
discharge, due perhaps to the offsetting influence of wash-
off and erosion processes. For example, in Colston's study
of a 1.67-square-mile watershed in Durham, exponential re-
lationships between concentration and discharge were esti-
mated for 19 constituents (Colston, 1974, p. 59). The
discharge exponents obtained were between 0.15 and -0.15 for
all constituents except suspended solids, cobalt, iron and
calcium. (Similar findings were observed for the Pennsyl-
vania watersheds analyzed in the current study; see the
Technical Appendix.) As observed elsewhere, concentrations
were negatively related to time since start of storm, mean-
ing that a higher concentration would be associated with a
given discharge during the rising stage of the hydrograph
than during the falling stage. However, it is apparent
that, if water-quality samples are representative of the
full range of storm-period conditions, unweighted average
pollutant concentrations may be fairly close to the dis-
charge-weighted averages for most constituents. If the
discharge exponent for a constituent concentration is 0.15
in a watershed exceeding one square mile, the unweighted
average concentration would be biased downward by at most
about 20% relative to the discharge-weighted average; and
the upward bias should similarly not exceed about 20% if the
exponent is -0.15. It is risky to assume that bias will
always be within this range, however, particularly when
dealing with water constituents other than organics and
nutrients. ..;.•..
Very serious errors can result in the third case mentioned
above, in which pollutant concentrations are strongly re-
lated to discharge, and the observations available do not
provide a balanced characterization of storm periods. A
matter of particular concern here is that sampling programs
frequently under-represent major storms, and streamflow
recession periods generally, ,due either to preoccupation
with first-flush effects or to logistical considerations.
The result can be serious overestimation of pollutant load-
ings on a long-term basis. -
A good example of this problem is the computation of annual
lead loadings from Lodi, New Jersey, presented by Wilber and
Hunter (1975). The average lead concentration observed in
storm runoff from two small basins in Lodi was 0.90 mg/1.
This figure was multiplied by runoff from Lodi as a whole to
yield an annual loading. An interesting aspect of the
observed lead concentrations was that they pertained en-
tirely to light rainfall events. More than half of the
110
-------
storms sampled involved less than 0.1 inch of rain, and the
maximum was 0.26 inches. In order to examine the signifi-
cance of this circumstance, the published data have been
subjected here to a regression analysis in which the average
lead concentration for each storm event is related to the
amount of precipitation (and a dummy variable distinguishing
the two basins). The estimated equation is the following:
L = 0.237 KP~*35
where: L is average lead concentration for a given storm
event, in mg/1
P is precipitation, in inches; and
K is a constant equal to 2.614 for basin 1 and
1.00 for basin 2.
Both independent variables are significant at 5%; R-square
is 0.88. In this case, lead apparently bears a very strong
negative relationship to precipitation, and presumably also
to discharge. As an indication of the extent to which the
0.90 mg/1 concentration may be inappropriate for estimating
annual loadings, Table 9 shows the computation of a pre-
cipitation-weighted average concentration, using the above
equation and a realistic rainfall distribution (based on
southeastern Pennsylvania). The weighted average lead con-
centration is 0.767 mg/1 for basin 1, and 0.294 for basin 2,
yielding an overall average of 0.53. The discharge-weighted
concentrations would presumably be somewhat lower, since
runoff volume tends to be an increasing proportion of rain-
fall volume.
The purpose of this discussion is simply to indicate that,
when preparing long-term pollutant loading estimates, it is
necessary to consider the full range of hydrologic condi-
tions which exist.
Ill
-------
TABLE 9
SAMPLE COMPUTATION OF PRECIPITATION-WEIGHTED
AVERAGE LEAD CONCENTRATIONS FOR BASINS
IN LODI, NEW JERSEY
Storm
Magnitude
(inches)
0 - 0.09
0.1 - 0.29
0.3-0.59
0.6 - 0.99
1.00+
Distribution
of Rainfall
by Volume
5.5%
13.5%
20.0%
24.0%
37.0%
Estimated Average Lead
Concentration (mg/1)
Basin 1
1.770
1.087
0.818
0.669
0.537
Basin 2
0.677
0.416
0.313
0.256
0.206
Weighted average: 0.767 0.294
Source: Based on data from Wilber and Hunter, 1975.
Preparation of Land Data for Loading Analysis
The watershed characteristics which are relevant for unre-
corded pollution loadings can be categorized roughly as
follows:
1. Areal measures of "land use" (e.g., residential
land, industrial land, etc.)
2 . Descriptive measures of land surface character-
istics (e.g., impervious area, curb length)
3. Indices of human presence and activity (e.g.,
population, employment, traffic)
4. Direct measures of land management practice,
(e.g., fertilizer application, pesticide use)
5. Indices of environmental conditions and waste
management effectiveness (e.g., cleanliness^ sewer
system condition)
6. Measures of natural land features (e.g., slope,
soil characteristics, drainage density)
112
-------
The first two of these categories are likely to be over-
lapping, in practice, since "land use" tends to imply a
specific physical condition of the land, and vice versa.
Also, indices of human presence and activity, such as popu-
lation, are often considered as measures of land use. With
regard to the last category, natural land characteristics
are relevant primarily because of interactions between land
use and natural features, i.e., cases in which the pollutant
loadings yielded by a land use are dependent upon the char-
acteristics of the land on which it is located. With regard
to urban land use, such interactions are important for: (1)
pollutant generation due to on-site sewage disposal; (2)
soil erosion from construction sites; and (3) the effects of
hydrographic modification, as produced by urban areas with
less than complete storm sewerage. Interactions are rela-
tively unimportant in other cases due to the fact that
conventional urban development tends to homogenize the
landscape, thus eliminating the influence of natural features
A very serious problem in analysis of loading relationships
is the fact that comprehensive data can rarely be obtained
for the fourth and fifth categories of variables listed
above, namely measures of land management practice, waste
management, and environmental conditions. If measured
effectively, such variables would capture most of the im-
portant site-specific pollutant sources; but systematic
collection of such data is infeasible for large areas. The
necessary omission of these variables in, cross-sectional
loading studies is a major reason why such studies are not
emphasized here.
A general comment regarding variables that express land use,
physical land condition, and human presence and activity is
that no one type of measure is logically superior to another
as a predictor of urban pollutant loadings. Most of these
measures are, in effect, surrogates for actual pollutant-
generation processes. For example, the water quality im-
pacts of street litter, or disposal of motor oil in storm
sewers, could be attributed variously to the existence of
streets, residential land use, population, motor vehicle
use, etc. Thus, the selection of variables as predictors of
unrecorded pollution should be based largely on practical
considerations.
Areal measures of land use—e.g., acres of land in residen-
tial, commercial, and industrial use—have two major draw-
backs for pollutant-generation analysis: (1) existing land
113
-------
use measurements vary widely in terms of the delineation of
uses (especially the distinction between "developed" land
and other land owned by a given establishment or resident),
and the classification systems employed; and (2) areal
measurements typically fail to provide a sufficient des-
cription of land use intensity. The first of these problems
may not be relevant for a planning agency which intends to
conduct a new data collection program. The second drawback
can theoretically be overcome by creating sub-categories of
land use. For example, residential land use involving a
given housing type could be subdivided according to popu-
lation density. This solution is regarded as relatively
inefficient for present purposes. The recommended approach
is instead to focus upon activity measures, plus impervious
surface, and to utilize land use data primarily in a modi-
fying role. Measurements of employment or impervious sur-
face, for example, could be categorized according to the
general type of land use involved.
It is usually convenient in water quality planning studies
to partition the study region into hydrologic subdivisions,
or sub-watersheds, for purposes of wastewater facility
planning as well as analysis of unrecorded pollution. If
these hydrologic subdivisions are no more than about 5
square miles in size (in developed areas) they can ordin-
arily serve as the basic data collection units for land
information. However, areas within hydrologic subdivisions
which differ in terms of sewerage should be distinguished
where possible. Land served by combined sewers, land with
separate sanitary sewerage, and land without sanitary
sewerage facilities should be segregated and treated as
separate data collection units. If such a partitioning is
not possible for a given hydrologic subdivision (for ex-
ample, if there are residential neighborhoods containing
both sewered dwellings and houses with on-site disposal),
all variables measured for that subdivision should be
broken down by sewerage class using assumed percentage
distributions.
Measurements of variables such as land use, population,
employment, and impervious surface can consist of totals or
averages for these land areas. Information pertaining to
smaller areas may be needed in some instances to assist
field investigations in isolating high-yield pollutant
sources; but preparation of such data can be postponed until
specific needs are established. On the whole, when dealing
with unrecorded pollution from existing urban areas, it is
114
-------
felt that numerical summaries of information are generally
sufficient for technical planning purposes. Although much
of the information may be derived originally from maps and
photographs, there may not be a strong need to prepare new
maps as part of the study, other than overlays showing the
location of streams, sewers, hydrologic subdivision boun-
daries, and specific high-yield pollutant sources. This
point is mentioned because detailed mapping tends to be very
time-consuming, and thus can absorb a major proportion of
planning resources. The principal tasks for which it may be
necessary to map numerous land characteristics over large
areas are the following: (1) analysis of the environmental
impacts of proposed wastewater management alternatives,
especially secondary impacts in developing areas (i.e.,
growth stimulation due to sewer construction); (2) analysis
of existing and future water quality problems due to on-site
sewage disposal; and (3) analysis of problems due to agri-
culture. Mapping needs with regard to other tasks should be
examined very critically, in order to minimize unnecessary
expenditure of resources.
In the empirical investigations described in Section 5 and
in the Technical Appendix, employment was found to be the
best single predictor of urban unrecorded pollutant load-
ings. The reasons were presumably that economic activities
tend to be the most critical sources of unrecorded pollu-
tants, at least in areas resembling the basins studied, and
that employment is the best overall measure of economic
activity. Although employment is not ordinarily considered
in this context, the use of this variable in generalized
loading relationships has certain advantages. Employment is
a relatively unambiguous measure which is defined similarly
by most existing agencies; thus, data from a variety of
sources can ordinarily be utilized. Although comprehensive
employment information is ordinarily not available for areas
smaller than municipalities, reasonably accurate estimates
can be prepared for hydrologic subdivisions, with moderate
effort, by utilizing fragmentary sources such as industrial
directories, tax records, and telephone surveys of major
employers. Existing data for municipalities, counties, and
metropolitan areas can provide control totals.
The manner in which pollutant-generation relationships
should be developed, if at all, depends upon the character-
istics of the region studied, the availability of existing
data, and the feasibility of conducting sampling programs
for this purpose. In cases where such relationships are
115
-------
desired but resources are very limited, the following
approach might be utilized. Long-term pollutant loading
estimates would be developed for five to ten basins in the
study area which contain primarily urban development with
separate sanitary sewerage. If pollution due to on-site
septic systems is a major concern, an independent study
should be conducted to deal specifically with this factor.
The basins would not contain major construction sites or
recorded discharges, unless it was very clear that the
loadings due to these sources could be subtracted out
accurately. Various intensities of development would be
represented, with at least one basin containing very low-
density development (but not agriculture). Basin size could
be variable, although all basins would preferably exceed one
square mile and would contain at least some natural drainage
channels (unless complete drainage alteration is typical
throughout the urban area to be characterized). The pos-
sible inverse relationship between areal pollutant loadings
and basin size would be kept in mind. Since the number of
basins assumed here would be too small to permit multi-
variate analysis, the constituent loadings would be analyzed
simply by plotting observed values, expressed as mass per
unit area, against an overall urbanization index. The
results obtained in the statistical studies mentioned
earlier, which were highly consistent, suggest two versions
of such an index which might be appropriate. These are:
Ul = I + 5E
U2 = P + 4E
where: Ul and U2 are urbanization indices;
I is the percentage of basin land covered by
impervious surface;
E is employment per acre; and
P is resident population per acre.
The relationship between observed loadings and either of
these indices would then be usable for preparing very rough
estimates of long-term pollutant loadings at other points in
the surface water system. Such estimates could serve as a
116
-------
standard of comparison for evaluating field observations, or
as a highly approximate means of assessing water quality
problems when direct information is unavailable.
Some final comments concern the measurement of impervious
surface, which is required in stormwater modeling as well as
in other forms of loading analysis. Surprising differences
tend to exist among estimates of impervious coverage, which
may reflect differences in definition. For example, the
Poquessing Creek watershed in Philadelphia was estimated to
be 48% impervious in one study and 25% impervious in another
study conducted at roughly the same time (Radziul, et al,
1973; Coughlin, et al, 1976). As understood here, imper-
vious surface would consist simply of land covered by pave-
ment and structures. Semi-impervious surfaces such as
unpaved streets, railroad tracks, junkyards and unpaved
parking lots might be included if prevalent; but all areas
of uncovered soil would be excluded, even small areas such
as highway medians and grass strips between curbs and side-
walks. However, it appears in many instances that either
very different definitions are being employed, or that
impervious coverage is significantly overestimated.
Impervious coverage generally represents a difficult mea-
surement problem, since pervious surfaces and impervious
surfaces tend to be interspersed. When aerial photographs
are utilized, it is extremely difficult to identify imper-
vious areas accurately, unless the scale is greater than
1:10000. The accuracy of methods utilizing satellite data
(infared spectroscopy) is not known. In order to provide
assistance in dealing with this problem, various investi-
gators have estimated impervious coverage on the basis of
population density and other characteristics of urban
development. Several based on population density are pre-
sented in Figure 7.
The relationship by Graham (1974) was estimated by asympto-
tic regression using census tract data for the Washington,
D.C., metropolitan area. Similar relationships were esti-
mated linking imperviousness to employment density and
household density. It appears that the Graham relationship
tends to overestimate imperviousness for low and medium
density development. A notable feature is that impervi-
ousness is estimated at 22 percent for zero population
density. (The corresponding intercept values for the em-
ployment and household relationships are 35 percent and 26
117
-------
wo
UJ
a.
5
o
z
—I
5
>-
GRAHAM
RELATIONSHIP
STANK OWSKI
RELATIONSHIP
10 15 20 25
POPULATION PER ACRE
Source: Graham (1974); Stankowski (1974); Betz Environmental
Engineers, Inc.
figure 1 ESTIMATED RELATIONSHIPS BETWEEN IMPERVIOUSNESS AND POPULATION DENSITY
118
-------
percent, respectively.) However, the 95% confidence inter-
val estimated for the Graham relationship would include the
other curves shown in Figure 7.
The Stankowski relationship was developed in a study of
flood magnitudes in New Jersey (Stankowski, 1974). Esti-
mates of imperviousness were obtained on the basis of land
use data for 567 municipalities. These values were related
to population density by polynomial logarithmic regression.
The relationship obtained was the following:
I = exp (2.265 + 0.573 in P - 0.01695 (In P) (InP))
where: I = manmade impervious cover, as a percent of
total land area;
P = population density in persons per acre
An additional relationship, based on both population density
and employment density, has been prepared as part of the
present study. In this case, nonlinearity is expressed by a
simple upper limit on imperviousness at 90 percent, rather
than by an elegant functional form.
I = 8.0 + 2.0 P + 3.2 E, or 90, whichever is lower,
where I and P are as defined above, and E is employment
density in persons per acre.
This relationship is illustrated in Figure 7 for two cases:
a residential area with no employment, and a situation in
which employment is equal to 0.4 times population. Use of
this formula is recommended, either as an estimating tech-
nique or as a check on direct measurements, for urban land
with at least 1 person per acre on the average (preferably
2). The accuracy is reasonably good when the areas con-
sidered are greater than 1 square mile in size.
For many purposes, especially the analysis of hydrographic
modification, the most relevant variable is hydraulically-
connected impervious area, rather than total impervious
119
-------
coverage. Hydraulically-connected surfaces are those which
are directly linked to storm sewers or receiving waters by
impervious channels. Determination of which surfaces are
hydraulically connected can only be accomplished reliably by
field inspection, although reasonably accurate guesses can
be made from aerial photographs by a person familiar with
development characteristics.
120
-------
SECTION 8
IMPLEMENTATION OF CONTROLS
FOR NEW URBAN DEVELOPMENT
Introduction
The present section describes briefly the various ways in
which preventive water quality controls for new urban devel-
opment can be implemented. For more detailed discussion,
the reader is referred to the excellent report on this sub-
ject which has recently been prepared by Abt Associates
(1976) . The major focus here will be upon implementation of
the measures which have been outlined in Section 4. The
design of these controls will then be discussed in Sections
9, 10, and 11.
The term "control measure" is commonly used to refer to any
actions which benefit water quality, ranging from construc-
tion of physical facilities, to legislative and administra-
tive actions. In order to avoid confusion, control measures
will refer here only to actions which directly affect the
physical environment. The means whereby these controls are
achieved are termed implementation mechanisms. This dis-
tinction is not entirely clear in some cases, but is gen-
erally useful in view of the fact that a wide variety of
implementation mechanisms can often be utilized to achieve a
given physical control.
Implementation Mechanisms for Preventive Controls
The following are the major categories of actions which are
most frequently mentioned as implementation mechanisms for
water quality controls affecting new urban development.
1. State Laws
Two types of state legislation are potentially
relevant: enabling legislation, and laws which
directly govern the behavior of individuals and
organizations. Enabling legislation tends to be
most important, since a majority of the actions
needed for control of unrecorded pollution are
likely to be taken at the local level; and many
121
-------
of these actions may require amendments to zoning,
subdivision, and construction codes which are not
within the scope of existing enabling legislation.
Direct action at the state level is most likely to
involve problems which are important in rural as
well as urban areas, such as water pollution due
to on-site sewage disposal.
Local Ordinances
Numerous types of ordinances can be utilized at
the local level for water quality protection.
Some examples are erosion controls, special zoning
ordinances^ .environmental impact statement re-
quirements, -and environmental performance stan-
dards. An important class of ordinances might
deal with the management of potential pollutants
such as fertilizer, pesticides, and residual
petroleum products.. A reasonable procedure in
current water quality planning studies might be to
examine first the extent to which the needed
physical measures can be achieved through existing
mechanisms (i.e., local government operations,
subdivision review, construction codes), and to
recommend separate ordinances only as necessary to
supplement these mechanisms.
Municipal Operations
A significant proportion of municipal operations
involving sanitation and public health are rele-
vant or potentially relevant to pollutant genera-
tion. In many cases it may be possible to obtain
major water quality benefits by changing or aug-
menting these operations without legislative
action. Obvious examples would be intensification
of street sweeping programs, catch" basin cleaning,
and general cleanup operations. (An issue which
may be important is the extent to which munici-
palities can influence waste management practices
on private property through enforcement of exis.t-
ing ordinances dealing with nuisances and vector
control.) Significant benefits may also be
achievable by expansion of licensing and regu-
latory functions to include sewer scavengers and
others engaged in the handling of potential
122
-------
pollutants. On the whole, municipal operations
could prove to be critically important in de-
tecting and correcting site-specific pollutant
sources.
4. Construction Codes
Amendments to construction codes may be utilized
to minimize the impacts of new development on both
quality and quantity of runoff. An extremely
important action may be to upgrade construction
codes dealing with waste conveyance facilities.
Increased monitoring activities may be necessary
to assure compliance with sewer construction codes
once amended. Regarding impacts of new develop-
ment on runoff quantity, there are a number of
aspects of construction codes which affect the
magnitude of this problem significantly, such as
requirements involving roof drains, storm sew-
erage, elevated curbing, $aved sidewalks, street
width, and parking lot capacity for commercial
establishments. Important benefits may be gained
by reviewing existing codes carefully to determine
cases in which they may be producing unnecessary
impervious coverage or underutilization of the
ability of pervious soil to absorb stormwater.
5. Operating Agreements f
A serious problem confronting water quality con-
trol efforts at the local level is detecting and
preventing surface pollutant accumulations and
improper waste disposal practices on private
property when there is not a clear nuisance situ-
ation or public health threat. In the absence of
environmental ordinances dealing explicitly with
these problems, it may be difficult to affect
materials management in existing developed areas?
but greater opportunities exist for establishing
sound practices in new development. Whenever the
granting of approvals for new development involves
negotiation between developers and public agen-
cies, it may be possible to include waste manage-
ment as one of the issues under consideration.
Measures such as lot sweeping and maintenance of
storm drainage facilities could be included as
123
-------
part of operating agreements for shopping centers
and similar facilities. Operating agreements may
be especially important for establishments engaged
in the handling of materials which constitute
potential pollutants. Examples are gasoline
service stations, petroleum wholesalers, trash
collection services, and numerous types of manu-
facturing plants.
Subdivision Review
Runoff and drainage controls for new development
are most often implemented through amendments to
subdivision ordinances. Subdivision regulations
dealing with these and other water quality objec-
tives can be expressed either in terms of specific
control measures or as environmental performance
standards. Performance standards are favorable in
that they allow the developer some degree of
choice in the design of control measures. How-
ever, the success of this approach tends to be
highly dependent upon the technical expertise of
the reviewing agency. Subdivision review may also
be an appropriate point at which to require sub-
mission of environmental impact statements by
developers. An important issue regarding sub-
division review is that, if major reliance is to
be placed on this mechanism for water quality
control, the regulations should be sufficiently
broad to deal with the major development projects
which occur. Enabling legislation may therefore
be needed in order to cover various classes of
single-ownership developments as well as projects
involving actual subdivision of land parcels.
Zoning
Zoning still represents the primary mechanism of
land use control in most communities, to the
extent that control exists. The use of zoning as
a water quality control instrument typically
involves amendment of the zoning code to create
special protection districts for critical land
areas or water bodies. In view of the fact that
zoning has been only moderately successful in
guiding land use in most communities, it may be
124
-------
unrealistic to assign a major role to zoning as a
water resources protection measure unless: (1)
special zoning districts are based on detailed
assessments of land-water relationships in the
given area, and (2) a strong base of popular
support exists for this use of zoning power.
8. Capital Improvements Planning
The location of public facilities such as highways
and sewer interceptors is extremely important to
the spatial distribution of new development.
Thus, public facility planning is potentially a
very powerful lever for land use control. This
aspect of municipal wastewater facilities should
be considered in detail in current water quality
planning studies; and the development of waste
treatment alternatives should be coordinated with
plans for preventing unrecorded pollution, insofar
as the latter involve land use control. Heavy
reliance upon public facility planning as a means
of implementing land use controls tends to involve
a large number of legal and administrative issues,
however, and should be utilized only as part of a
carefully conceived comprehensive planning effort.
9. Acquisition of Land and Conservation Easements
Control of land use through land acquisition or
less-than-fee purchase by the public can be an
important water resources protection measure, to
the extent that the net effect is to shift new
development away from areas which are sensitive in
terms of either pollutant generation or vulner-
ability of receiving waters. Similar benefits can
be obtained through the action of private groups
such as conservation trusts. An alternative
approach to land use control, which is ordinarily
oriented toward preservation of land in agri-
cultural use, is preferential tax assessment.
Such programs have been implemented in a number of
states, with somewhat mixed results.
10. PUD/PRD Ordinances
Planned Unit Development and Planned Residential
Development ordinances allow much greater flexi-
bility in development design than would be
125
-------
possible under conventional zoning at a given
density. Such ordinances can thus be very im-
portant to water resources protection by encour-
aging design modifications which affect runoff and
infiltration of rainwater. The most important
single aspect of PUD/PRD ordinances is often the
reduction in coverage of land by streets and
driveways. PUD/ PRD ordinances are most effective
for water resources protection when combined with
specific runoff control requirements and con-
struction standards; these can either be included
as part of the ordinance or implemented through
other mechanisms.
A large proportion of these proposed mechanisms for water
resources protection are oriented toward land use control.
It is important to consider briefly, at a conceptual level,
the potential regional water quality benefits that can
be attained by implementing land use controls at the local
level.
Two general consequences of land use regulation can be
identified: (1) change in the spatial allocation of new
development within an urban region; and (2) net shift in the
types of new development constructed. The first consequence
involves the fact that exclusion or restriction of devel-
opment in one area may simply cause development to go else-
where within the region. The extent to which spatial real-
location occurs, as opposed to change in development type,
depends upon the comprehensiveness of controls within the
region and the existence of localized demand factors. For
example, suppose that a municipality institutes land use
controls which limit construction of conventional single-
family housing. .If alternative sites are available nearby,
and if developers feel that market demand is more signifi-
cantly related to housing type than to the amenities avail-
able in the given municipality, the overall effect could
simply be a shift of single-family housing construction to
other areas. On the other hand, if constraints upon single-
family housing are widespread, developers will be motivated
to construct and promote higher-density housing types. (The
latter situation is also likely to involve some degree of
spatial reallocation, since the relative advantages of
different locations within a region tend to be dependent
upon the type of development under consideration. For
example, high-density housing may tend to gravitate toward
existing population centers.)
126
-------
These rather obvious points are stressed because it is im-
portant to avoid the myopic view that exclusion or restric-
tion of new development in a particular area will neces-
sarily benefit water quality in the region as a whole. To
the extent that land use control results in spatial real-
location of development, net water quality benefits will
result if the land which is protected has at least one of
the following characteristics: (1) development of the land
would result in higher pollutant yields than equivalent
development at other sites; (2) development of the land
would constitute encroachment upon- surface water bodies,
flood plains, or wetlands; (3) the land constitutes an
especially important aquifer recharge area; or (4) the land
drains to surface waters which are especially sensitive to
impact. The first of these characteristics relates pri-
marily to land suitability for on-site sewage disposal and
susceptibility to erosion during construction. The im-
portant point is that, unless land use control measures
specifically direct development away from areas having the
above characteristics, there may be little water quality
justification for these controls, even though they may be
highly desirable for other objectives such as open space
preservation, ecosystem protection, efficient energy utili-
zation, transportation planning, protection of agricultural
land, and provision of public utilities. If these other
objectives are sought, they should be included explicitly as
a basis for land use controls.
In many cases, the aspect of land use control which is most
important for unrecorded pollution is the resulting shift
among development types, particularly the substitution of
clustered housing and multi-unit structures for traditional
house-and-lot development. The mechanisms which appear most
likely to achieve such a shift (with assistance from present
market trends) are PUD/PRD ordinances, and capital improve-
ments planning which encourages nodal development in se-
lected districts. However, the" water quality benefits thus
achieved can in most cases be achieved equally well through
the use of mitigative measures.
Choice of Controls
The physical control measures which are felt to be most
needed for new urban development have been described in some
detail in Section 4. The following areas have been empha-
sized: (1) control of erosion/sedimentation problems due to
127
-------
construction activity; (2) control of the location, design,
and operation of on-site sewage disposal systems; (3) con-
trol of hydrographic modification; and (4) construction of
leakproof sanitary sewer systems. The present discussion is
limited to control measures that are primarily suitable for
new development, either because they must be physically
incorporated in new facilities, or because they are tac-
tically easier to implement at the time of land use conver-
sion than after a land use has become established.
Table 10 presents a summary of physical controls by type,
scope, level of implementation, and nature of problem
addressed. The first four controls relate to hydrographic
modification (which is the subject of the next section of
this report). Runoff detention facilities, which include
measures such as storage of rainwater on rooftops and park-
ing lots, are expected to play a major role in water quality
control plans. On-site runoff detention devices must or-
dinarily be included in the original design of a development
project. However, detention facilities can be constructed
at off-site locations to serve existing development if
appropriately located land can be made available for this
purpose. In the case of infiltration devices, an important
characteristic is that unless soils are extremely permeable,
infiltration facilities must ordinarily be dispersed through-
out a developed area rather than centralized. The opportuni-
ties for use of infiltration devices in existing developed
areas thus tend to be very limited.
As discussed in the next section, detention facilities deal
with the problem of excessive rates of discharge during
storm periods, whereas infiltration devices are most rele-
vant to aspects of hydrographic modification which involve
groundwater recharge. Both types of measures reduce the
loadings of pollutants released to surface waters. For most
types of urban development, the use of properly designed
retention and infiltration facilities may provide adequate
control of stormwater quality, when combined with sound land
management practices. Implementation of these measures can
be accomplished at the local level by amendments to sub-
division ordinances or by adoption of separate environmental
ordinances, the latter of which may deal with a variety of
environmental impacts. In some areas of the U.S., voluntary
inclusion of runoff controls in major construction projects
is becoming common, particularly in cases where a zoning
variance is requested.
128
-------
TABLE 10
CHARACTERISTICS OF POLLUTION CONTROLS
10
Physical Control
Runoff detention
facilities
Runoff infiltration
facilities
Development design
modifications
to limit runoff
impact
Protection of
natural water-
courses
Erosion/sedimenta-
tion controls
for construc-
tion sites
Location and de-
sign of on-site
sewage disposal
systems
Construction of
leakproof sani-
tary sewer
systems
Private property
maintenance and
management of
hazardous ma-
terials
Maintenance of
on-site sewage
disposal sys-
tems
Public cleanliness
and facility
maintenance
Scope of Control
New and existing
development
New development
New development
New development
New development
New development
New development
New and existing
development
New and existing
development
New and existing
development
Level of
Implementation
Local
Local
Local
Local
State,
local
State,
local
Local
Regional
or local
Local
Regional
or local
Nature of
Problem Addressed
Peak discharge magnitudes,
pollutants
Base flow, aquifer
recharge, pollutants
Streamflow regimen,
pollutants
Aquatic habitats, peak
discharge, pollutant
transport
Sediment, runoff quantity
Groundwater quality,
pollutants
Infiltration/inflow,
pollutants
Pollutants
Pollutants
Pollutants
Source: Betz Environmental Engineers, Inc.
-------
Development design modifications, listed in the third row of
Table 10, refer to any integral features of development
projects which are intended to reduce or retard storm run-
off. These may range from simple measures such as limita-
tion of impervious coverage, to carefully-conceived drainage
plans which make use of grass swales and buffer strips to
promote infiltration. A potentially very beneficial measure
is the use of porous pavement rather than conventional pave-
ment, which could involve little or no additional expense in
some cases. Protection of natural watercourses from en-
croachment is important in order to maintain the natural
capacity of stream channels to absorb flooding, as well as
to preserve aquatic habitats. Ideally, protection measures
should extend to small headwater streams as well as major
watercourses, and should be linked to flood plain management
programs where possible. Policies regarding alteration of
watercourses should be included explicitly in subdivision
regulations and environmental ordinances.
Erosion/sedimentation controls for construction projects can
be implemented directly by state agencies, or more commonly,
by counties or municipalities under state enabling legisla-
tion. As discussed in Section 11, control of land erosion
usually involves the use of mitigative measures rather than
restriction of development in areas with high erosion po-
tential. Control of the location and design of on-site
septic systems represents largely an extension of existing
regulatory functions, which have traditionally been oriented
toward health rather than water quality. In regions where
large areas of land are intrinsically unsuitable for on-site
waste disposal, stringent regulation of new systems may have
major implications for land use, since development may not
be feasible in such areas without municipal sewerage or use
of package treatment plants. The impacts of on-site sewage
disposal regulations should therefore be considered care-
fully in population forecasting and wastewater facility
planning.
The last three items listed in Table 10 are controls which
can be applied similarly to new and existing development,
but which should ideally be established prior to land use
conversion. The potential importance of operating agree-
ments for commercial and industrial establishments has
already been mentioned. In cases where surface pollutant
accumulations are expected to be especially serious, such
agreements could include stormwater treatment. One approach
130
-------
is to require establishments engaged in the handling of bulk
materials to submit a mass balance accounting of all ma-
terials which will enter and leave the premises, given
various levels of operation. Satisfactory arrangements for
handling and disposal of residuals would then be established
as a precondition of the granting of approvals needed for
construction. An additional requirement which might be
imposed in the case of gasoline service stations is that
they be obligated to accept and properly dispose of residual
petroleum products (e.g., crank-case oil) when submitted by
residents in small quantities.
Maintenance of new on-site septic systems will not differ
technically from maintenance of existing systems; but
greater latitude may exist for enforcement of standards in
the case of new systems. All permits issued after a certain
date could possibly be made contingent upon observance of an
appropriate maintenance schedule (pertaining primarily to
sludge removal), to be verified by a licensed sewer scaven-
ger. The permitting system could also provide for testing
at long intervals to assure that a system is continuing to
function properly. Finally, with regard to public cleanli-
ness measures and facility maintenance, various steps might
be taken in the design of new development to facilitate
these activities. Two examples are location of sanitary
sewers so that they will be readily accessible for repairs,
and design of catch basins to facilitate cleaning opera-
tions.
There are numerous issues and opportunities involving the
use of stormwater controls which are not discussed here.
Some examples are: re-use of stormwater; reduction in re-
quired storm drainage when porous pavement is utilized;
linkage of the erosion/sedimentation controls provided
during construction with permanent stormwater control
measures; and design of runoff detention basins as multi-
purpose facilities. Some useful references in this regard
are Lager and Smith (1974), and Tourbier and Westmacott
(1974), as well as the Abt Associates report. The present
discussion also has not attempted to relate preventive con-
trol of unrecorded pollution to municipal wastewater plan-
ning, although these activities are linked by a number of
critical issues, which involve water supply, inter-basin
water transfers, land disposal versus stream disposal of
effluents, and regional versus localized treatment systems.
For example, one disadvantage of regional sewer systems
131
-------
may be the exportation of water from the areas served, which
reduces base flow of streams; but this could possibly be
offset by utilization of infiltration devices for ground-
water recharge, as described in the next section. Planners
should be aware of the existence of these issues, particu-
larly when the use of water quantity controls is under
consideration.
A final comment is that the list of preventive controls
recommended here for new urban development is intended to
represent a feasible strategy for widespread application in
U.S. metropolitan areas. It is clear that much more can be
done in certain counties and municipalities where interest
in environmental issues is high and ample resources exist
for implementation of controls. However, regional planning
agencies must consider what is achievable in all or a large
part of the region. The major preventive controls suggested
here—control of on-site sewage disposal, construction of
tight sanitary sewer systems, control of hydrographic modi-
fication, and control of erosion during construction—are
felt to be generally feasible, even though they represent a
quantum leap relative to existing practices in a large
proportion of communities. When combined with maintenance
of high cleanliness standards, these measures should, in
most instances, provide a reasonable level of protection
against water quality deterioration.
132
-------
SECTION 9
CONTROL OF HYDROGRAPHIC MODIFICATION
Introduction
Hydrographic modification refers to actions which affect the
flow regime of surface waters. The term is most commonly
used in describing the effects of major in-stream projects
such as dams/ diversions, channelization, and so on. The
major emphasis of the present section, however, is upon the
cumulative effects of widespread actions associated with
urban development, including changes in land surface charac-
teristics as well as direct manipulation of receiving water
bodies. The discussion will relate generally to all effects
of urban development on water quantity (defined as the
volume and rate of movement of water at each point in the
hydrologic system) and will refer to problems such as aqui-
fer recharge which are not typically associated with hydro-
graphic modification.
As discussed earlier, urban development involves construc-
tion of impervious surfaces and alterations to the natural
drainage network, which increase the volume of surface
runoff and accelerate its movement through the surface water
system. These impacts tend to be progressive in nature and
may require a long time to occur; the effects produced by
each individual development project may be imperceptible.
Significant differences in impact are associated with
different types of development. For example, with regard to
increase in peak discharge, the smallest effect per unit of
impervious surface is produced by development with no storm
sewerage; next is the effect of urban land with street
drains but no alteration of natural streams and swales; and
the greatest effect is produced by development in which all
watercourses have been replaced by artificial conduits.
Similar variation exists in the effects of urban development
on low flow magnitudes, although these effects tend to be
somewhat more difficult to predict.
It is extremely important to recognize the linkages which
exist between hydrologic effects and water quality. In the
case of effects on base flow, the linkages involve the
quantity of water available for dilution of wastewater
133
-------
effluents and support of aquatic life (as well as for water
supply). In the case of surface runoff during storm events,
the linkages involve the production of pollutants per se.
The same factors which increase runoff volume and rates of
discharge are critical to production and transport of storm-
water pollutants; and measures which control the former
effects also exert a substantial degree of control over pol-
lutant loadings. As noted earlier, the latter fact is
central to the control strategy suggested here for new urban
development.
The next three sub-sections discuss the hydrologic impacts
of urban development, and a number of related issues which
are important for water resources protection. The remainder
of this section deals specifically with the use of runoff
detention and infiltration devices as controls for water
quantity and quality.
Effects of Urban Development on Peak Discharge and Stream
Channel Conditions
Two quantity-related effects of urbanization on storm runoff
should be distinguished, namely the increase in total volume
of direct runoff, and increase in the peak rates of dis-
charge produced. The latter effect is typically much
greater than the former in percentage terms. (See for
example Anderson, 1968; and Putnam, 1972.) Urban develop-
ment in a small basin may increase the total direct runoff
resulting from a given rainfall by 50%, but increase peak
discharge at the basin mouth by 150% or 200%. This dis-
crepancy, which is due to the fact that runoff is accel-
erated as well as increased in volume, is attributable both
to drainage alterations such as storm sewerage and to the
presence of impervious surfaces per se.
The effects of urban development on peak discharge can be
summarized in either of two ways: CD the discharge asso-
ciated with any given frequency (e.g., once in 5 years) is
increased; or alternatively, (2) any given discharge level
is achieved more often. These effects tend to be most
important for small watersheds, due in part to the fact that
urban development usually accounts for a minor proportion of
land area in large basins (e.g., watersheds of several
hundred square miles or more). The impact of urbanization
on peak discharge also tends to be relatively greater for
134
-------
moderate stcr m events, which occur frequently, than for
extreme events. The 2-year flood discharge may be increased
by only 50%.
A major focus of attention here is the impact of changes in
peak discharge on stream channel conditions. The following
characteristics of stream channels are relevant to this
phenomenon/ as well as to a number of other water quality
issues.
Under static land use and land cover conditions, stream
channels tend to maintain a dynamic equilibrium (or "quasi-
equilibrium") condition with stream flow. A high level of
interaction tends to exist between the stream channel and
the stream water, in the form of erosion and deposition of
particulate matter. As described by the classic meander
model, a stream is likely to erode the outer bank of its
channel in a bend, and to deposit material on the inner
bank. Thus, the channel may shift laterally over time. The
channel is likely to remain fairly constant, however, in
terms of configuration, capacity, and elevation, even while
its location is shifting; and the amounts of material de-
posited and eroded are likely to balance roughly in the long
run. Even if the latter is not true, so that the amounts of
earth material entering and leaving a stream reach are
consistently different, the magnitude of the imbalance tends
to be small relative to the gross interchange of material
between the.stream channel and the stream water.
It has been observed (Leopold, Wolman, and Miller, 1964)
that stream channel equilibrium under natural conditions
tends to involve a situation in which channel capacity
remains roughly equal to the 1.5-year flood discharge. That
is, despite lateral movement of the channel, channel size
remains just sufficient to accommodate all flows up to the
discharge which is equaled or exceeded in two out of three
years on the average. Subsequent studies have suggested
that, when the streamflow regimen is changed due to urban
development, a process of channel enlargement ensues, in-
volving imbalance of erosion and deposition (Leopold, 1968;
Hammer, 1973a). The end result appears to be a new equi-
librium in which channel capacity is again roughly equal to
the 1.5-year discharge. The percentage increase in channel
cross-section area due to urbanization is thus comparable to
the corresponding increase in the 1.5-year flood. Logi-
cally, channel enlargement would not necessarily require a
135
-------
net removal of earth material equal to the change in channel
size. The size of a channel can be increased by deposition
of sediment on bank tops as well as by scouring. However,
surveys of channel behavior over time have suggested that
channel enlargement in a stream reach tends to involve a net
erosion effect which is very roughly equal to the increase
in channel volume (Hammer/ 1973a; pages 124-126).
Stream channel enlargement can therefore be an extremely
important source of suspended solids and bedload material in
urbanizing areas. This effect is a direct consequence of
change in streamflow regimen, and is independent of sediment
yields from construction sites, and of effects produced by
human alteration of stream channels (except that the latter
tend to increase peak discharge and channel enlargement
downstream). The channel enlargement process not only
yields sediment to downstream areas, but also involves
channel disturbance over an extended period of time, which
can have severe biologic and aesthetic impacts.
The above description, in which channel enlargement is said
to involve a temporary imbalance between erosion and depo-
sition processes, should be modified somewhat in the case of
rills, gullies, small headwater streams, and perhaps major
streams in mountainous and arid areas. The meander model
may have little relevance in these cases, since the streams
are persistently downcutting rather than interacting with
flood plains through selective erosion and deposition.
Change in streamflow regimen due to urbanization simply ac-
celerates the downcutting process. There may be no natural
limit to the net amount of material thus removed, unlike the
case when a stream is prone to lateral movement within a
flood plain.
Empirical studies of stream channel enlargement indicate
that the ultimate increase in channel cross-section area due
to urbanization ranges from 25%-50% for streams draining low
density residential development, to upwards of 200% in small
basins which are 50% impervious (Hammer, 1973a). The rela-
tionships governing channel enlargement are essentially the
same as those governing increase in peak discharge, and can
be summarized as follows:
1. The hydrologic impact of a given intensity of
urban development is less in large watersheds than
in small watersheds. Channel enlargement is a
136
-------
problem primarily in stream reaches draining less
than 100 square miles, and tends to be most dra-
matic in basins of 10 square miles or less.
2. Major differences in impact exist between hy-
draulically-connected and non-hydraulically-
connected impervious surfaces. The effects of the
latter are highly dependent upon several factors
which determine the probability that runoff will
reach the receiving stream, namely, the size of
the impervious parcel, the slope of the land on
which it is located, and the density of streets in
the vicinity.
3. The greatest land use impacts on flood discharge
and channel condition are produced by streets with
storm sewers, due to the role of street gutters in
collecting runoff from adjacent land as well as
from street surfaces. The effects of streets vary
somewhat according to the efficiency of storm
sewerage, as measured by variables such as the
density of inlets.
Equations estimated in Pennsylvania studies indicate that
the yield of sediment from an urbanizing basin, due strictly
to enlargement of stream channels draining more than 100
acres, could be substantially greater on an annual basis
than the total sediment yield from the basin while in agri-
cultural use (Hammer, 1973a, p. 140). For a given develop-
ment project, there are various circumstances under which
the total sediment yield due to channel enlargement could be
comparable to the yield due to erosion during construction,
with or without control measures applying to the latter.
Thus, increases in runoff quantity and peak discharge due to
urban development are per se an important factor in storm-
water pollution.
Prevention of Direct Alternation of Stream Channels and
other Water Bodies
Water resources protection measures should address at least
four objectives regarding stream and river channels, as
follows:
1. Prevention of disruption of aquatic habitats
2. Prevention of sediment yield due to channel
enlargement
137
-------
3. Preservation of the natural capacity of stream
channels to dissipate flood discharges
4. Preservation of the role of stream channels as a
sink for stormwater pollutants.
The discussion in the previous sub-section is relevant to
the last of these objectives. Given the interchange of
particulate material between stream channels and the stream
water which typically exists, and given the extent to which
urban stormwater pollutants are associated with suspended
material, stream channels would be expected to play an im-
portant role as a sink for pollutants. As suggested earli-
er, this role may be very important in reducing the pollu-
tant loads which reach major water bodies. Loss of pollu-
tants from the stream water does not necessarily require
that sediment loads per se diminish downstream; the depo-
sition of particulate matter which is high in chemical
pollutants could be balanced by erosion of relatively clean
material from the channel. Preservation of this role of
natural stream channels by preventing alterations such as
channelization and replacement by storm sewers is considered
generally desirable, especially for headwater streams,
although it is clearly not an adequate substitute for elimi-
nating stormwater pollutants altogether.
The capacity of stream channels to dissipate flooding has
not been discussed here but is generally well known. Peri-
odic flooding is a natural function of stream channels.
Channel alterations and encroachments which prevent this
occurrence have the effect of increasing peak discharges and
flood-related problems downstream.
Prevention of stream channel enlargement, and associated
channel disruption, will be accomplished largely through the
use of'runoff controls for new urban development, which are
the focus of the latter portion of this section. With re-
gard to direct alteration of streams and other water bodies,
planning agencies are urged to recommend a general policy of
non-intervention. New development should be designed so
that streams are left in their natural state, rather than
being straightened, channelized, or replaced by artificial
conduits. Attempts to improve the appearance of streams by
cosmetic measures, such as elimination of trash and grooming
of vegetation, should be encouraged; but such measures
should stop at the bank top, and should not interfere with
138
-------
natural erosion, deposition, and overflow processes. The
serious negative effects of channel alterations on aquatic
habitats have been mentioned in Section 5. Due to these
effects, and the other factors mentioned here, it is felt
that non-interference with stream channels can be recom-
mended as a water quality protection measure.
Clearly, alteration of watercourses is sometimes necessary
in order to prevent property damage and safety hazards due
to flooding of existing developed land. However, it should
be possible to avoid such situations in the future by pre-
venting construction of buildings and other facilities at
points where they could be vulnerable to flood damage (tak-
ing into account possible increases in flood levels due to
future urbanization). Efforts in this regard can be linked
to ongoing flood plain planning activities where they exist.
A significant issue is the definition of watercourses to be
protected, specifically the choice of an upstream limit of
protection. Typically, surface water drainage involves a
network of gullies, swales, ephemeral streams, and perennial
streams, within which there is no clear point at which
controls become appropriate. Channels draining small areas
of land can be a liability when affected by urbanization,
due to their erosion potential. On the other hand, many
small channels may be useful, when retained in natural
condition, for dissipation of flooding and entrapment of
pollutants. A suggested criterion is that non-intervention
policies should extend to all stream channels, no matter how
small, for which there exists a flood plain at least as wide
as the channel. (An unambiguous definition of flood plains
could be developed; see Hammer, 1973a, p. 112.) Alteration
or replacement by storm sewers would be permitted for other
watercourses draining into these channel reaches; this might
even be encouraged in cases where urban development does not
include runoff control. Protection of very small stream
channels is not without precedent. An ordinance applying to
the Wissahickon Creek Watershed in Philadelphia, for ex-
ample, contains minimum building setback requirements for
swales draining as little as 20 acres (Coughlin and Hammer,
1973).
Another potential aspect of hydrographic modification is the
creation of impoundments. Impoundments often have bene-
ficial effects on downstream water quality; but the im-
poundment itself is likely to be highly sensitive to water
139
-------
quality impacts, particularly impacts due to urban runoff.
If water quality in the impoundment is to be maintained,
more stringent controls on urban runoff quality are likely
to be necessary than would be the case without the impound-
ment. Thus, it is felt here that impoundments generally
should not be created in developing or developed areas,
except under three conditions:
1. Land development is essentially complete (or land
use is expected to remain stable for other rea-
sons) and the water quality of influent streams is
demonstrated to be suitable.
2. The unique value of the impoundment to the public
(e.g., for municipal water supply) clearly justi-
fies the additional public and private costs of
maintaining water quality through land use con-
trols and other measures.
3. The purpose of the impoundment is specifically to
mitigate the impacts of urban development on down-
stream receiving waters.
This is a complex issue, however, and must be resolved
largely on a case-by-case basis.
Effect of Urban Development on Groundwater Replenishment
Discussion thus far has focused upon the effects of urban
development which involve surface runoff, rather than the
consequences of decreased infiltration per se. In the
absence of urban development, infiltration typically is
equal to a large proportion of annual rainfall. (Water
which infiltrates only into the top layer of the soil, and
then re-emerges as storm runoff, is ignored throughout the
present discussion.) Part of the infiltrated water is
extracted by vegetation; part percolates downward into deep
groundwater reservoirs; and part travels through the soil to
emerge eventually in surface water bodies. The last of
these components is the major source of streamflow during
nonstorm periods. Urban development tends to reduce infil-
tration, due to construction of impervious surfaces and
temporary disturbance of the remaining pervious soil. The
expected effect is a reduction in all of the subsurface
water components just mentioned. Less water would therefore
be available for human use in groundwater aquifers; and
140
-------
streamflow during dry weather would tend to be reduced. As
mentioned earlier, reduction in the base flow of streams
could be relevant for water supply, maintenance of aquatic
life, and dilution of wastewater effluents. This quantity-
related effect is therefore considered to be an important
water quality issue.
The effects of urban development on groundwater and base
flow are somewhat difficult to predict, due to a number of
complicating factors. The issues involved can be illus-
trated by way of a simple example. Consider a small land
area which is being converted to medium-density urban use.
Assume that water for domestic and other uses will be
imported to the area by a municipal authority, and that
sewage will be exported via a separate sewer system. Due to
construction of impervious surfaces, surface runoff during
storms will increase relative to present conditions; and the
volume of water infiltrating into the soil will decrease.
If the area is to be 30% impervious, one might expect reduc-
tions of approximately 30% in the volume of infiltration,
the rate of aquifer recharge, and the contribution of the
area to base flow in local streams.
However, these factors will all change by considerably less
than 30%, for a number of reasons. First, the land develop-
ment process will involve removal of most or all of the
existing vegetation. The 70% of the land kept in soil cover
will eventually be replanted with grass, shrubbery, and
small trees. The net effect may be a large reduction in
evapotranspiration. Second, some of the impervious surface
may not be hydraulically connected to storm sewers, par-
ticularly if the area is to contain single-family housing.
Much of the storm water draining onto pervious soil from
these surfaces may infiltrate. Third, the new, carefully-
tended vegetation (e.g., grass) may maintain greater soil
porosity at the surface than existing vegetation, thus
increasing infiltration. And fourth, significant amounts of
imported water may be released from the municipal system—
i.e., may not be re-exported as sewage. Watering of lawns,
and leakage from water and sewer pipes, would be the two
primary mechanisms. A very large proportion of this water
may infiltrate, if not originally released into the soil.
These factors tend to be particularly important for residen-
tial land. In some cases they may cause urban development
to affect the net recharge of groundwater by only a small
141
-------
amount, or not at all. Limited evidence in the Philadelphia
area .has suggested that base flow of streams is just as
well-sustained in medium-density urban watersheds, with
imported water and no continuous wastewater effluents, as in
comparable rural watersheds. Extreme low flows may in fact
be somewhat higher. The only exception observed in the
Philadelphia case was an industrial basin containing an
airport (Hammer, 1973a, p. 229). It appears likely that
imported water is the factor most responsible for these
findings.
The situation becomes more complex when groundwater with-
drawals and sewage disposal are included. On-site sewage
disposal constitutes a direct addition to infiltration.
Pumpage of water from wells represents a subtraction from
the supply of water available in a given aquifer; but the
relevant recharge area may be located some distance away
from the wellhead. A fairly common case in urban areas is
one in which water is imported for domestic use, but sewage
is disposed on-site. It is quite clear that this form of
urban development involves no net reduction in groundwater
recharge, although groundwater pollution may be an issue.
Consideration of water supply and waste disposal opens a
number of issues regarding inter-areal water transfers and
the value of self-sufficiency in water resources. Without
assuming that self-sufficiency will be adopted as a goal, it
is recommended here that land development projects not
receive credit for the effects of imported water, when
impacts on groundwater are computed. This would apply
whether or not on-site sewage disposal is involved. Given
this position, it would follow that measures to promote
infiltration are generally needed for new urban development.
Control of Runoff and Infiltration
. .- l •
The effects of new urban development on runoff and infiltra-
tion can potentially be controlled in four ways:
1. Construction of stormwater detention devices
2. Construction of infiltration devices
3. Modification of development design
4. Control of development location
142
-------
Stormwater detention devices and infiltration devices are
offsetting measures which do not necessarily require altera-
tion in the design of other functional elements of a de-
velopment plan. Such measures can be integrated with other
facilities if desired, as is the case when rooftops and
parking areas are used for stormwater storage. Modification
of development design, as a hydrologic control, includes all
measures to reduce runoff and promote infiltration which do
not involve construction of specific physical facilities for
this purpose. Control of development location involves
change in overall land use patterns, relative to what would
otherwise occur, in order to minimize hydrologic impact.
Detention devices are considered here as a generic class of
facilities which operate automatically to store stormwater
for the purpose of slowing its rate of movement through the
surface water system. Water storage capacity is obviously a
necessary feature of detention devices, since reduction in
the rate of surface water movement can only be achieved by
creating a situation in which the rate of inflow to an area
is greater than the rate of outflow, during some critical
period. The simplest type of detention facility, which also
tends to be most cost-effective for development projects
involving large traces of land, is the detention basin.
This is simply a surface impoundment in which the outlet is
designed and operated so that excess storage capacity exists
prior to a storm. - Stormwater from areas served by the basin
is conducted to the basin by pipes or natural channels, and
then released at a controlled rate which is less than the
peak rate of inflow.
As applied to new development, a detention basin can be
visualized simply as a water storage area with a big pipe
coming in and a small pipe going out. A variant of this is
a paved area or rooftop upon which ponding occurs, so that
the rate of runoff from the surface is less than the rate of
precipitation. If a detention basin is not lined with
impermeable material, and is located on highly pervious
soil, a significant amount of infiltration may be achieved
during periods of storage, so that the total volume of water
released through the outlet is substantially less than the
volume of inflow. This does not hold for a majority of
detention facilities, however. Detention structures and
infiltration devices will thus be kept conceptually distinct
here for purposes of simplicity.
143
-------
Infiltration devices are facilities designed specifically to
promote infiltration of stormwater into the soil. Infiltra-
tion devices can be located either above or below ground,
although below ground facilities tend to be more common.
Examples are French drains, seepage pits, and Dutch drains,
A critical feature of infiltration devices is that they are
almost always designed to induce infiltration into a soil
surface which is much smaller than the catchment area served
by the facility. Thus, the rate of infiltration is general-
ly less than the peak rate of inflow, which means that
storage capacity must be provided. For example, an infil-
tration device serving a rooftop area could not consist
simply of an extension of the gutter downspout into the
soil. A subsurface area filled with porous material such as
gravel or stone would be required, in order to provide space
for temporary storage of water while infiltration is taking
place. The present discussion will assume that a surface or
subsurface storage area is always needed, and is likely to
constitute the major cost factor involved in constructing an
infiltration facility. Although an infiltration device may
include an outlet, as in the case of a Dutch drain, it will
be assumed here that there is no release of water from
storage other than through infiltration. When storage
capacity is fully utilized during a storm event, additional
stormwater simply bypasses the facility.
Detention devices and infiltration devices differ in that
the former affect only the timing of runoff release during a
storm period, not the volume, whereas the stormwater which
enters infiltration devices tends to make little or no
contribution to stream discharge during the high-flow period
associated with a given storm. Despite this advantage of
infiltration facilities, detention devices tend to be more
feasible for control of peak discharge, due to the generally
lower cost of providing large amounts of storage capacity,
and the fact that the operation of detention devices is
independent of soil characteristics.
Development design modifications for the purpose of mini-
mizing runoff and promoting infiltration fall primarily into
three categories: (1) limitation of impervious coverage;
(2) use of porous paving materials; and (3) drainage plans
which expose runoff from impervious surfaces to pervious
soil. Limitation of impervious coverage is a potentially
favorable approach which can often be accomplished without
reducing land use intensity. For example, clustered hous-
ing, as accomplished through PUD ordinances and other
144
-------
mechanisms, tends to involve less street area and other
impervious surface per dwelling unit than conventional
single-family housing. Impervious coverage can be reduced
by selectively eliminating paved areas, or substituting
other forms of surface, in cases where pavement is not
highly needed. Examples are paved sidewalks in low-density
residential areas, and overflow parking lots for shopping
centers. Often this can be accomplshed simply by amending
the requirements specified in existing municipal ordinances.
Porous paving material, which contains interstices that
allow stormwater to pass through to the underlying soil, has
not yet been widely tested, but is regarded as extremely
promising (Tourbier and Westmacott, 1974). The ability of
porous pavement to withstand heavy traffic and to remain
unclogged over long periods of time has not been proven; but
this material should be suitable for walks, drives, and
parking areas that are not intensively utilized. As is true
also in the case of infiltration devices, the value of
porous pavement for water quality control may be limited by
the fact that the surfaces containing the greatest pollutant
accumulations are likely to be those which are least suit-
able for use of this measure.
It is possible to induce infiltration and to moderate the
volume and speed of surface runoff by providing drainage
plans which allow runoff from impervious surfaces to cross
pervious soil. Runoff from rooftops can be directed onto
lawns, for example, rather than channeled to streets or
storm sewers. The use of grass swales rather than imper-
vious drainage conduits can also be beneficial in some
circumstances. For very -low-density development, it may be
possible to minimize the use of storm sewerage even for
street surfaces. Residential developments in southeastern
Pennsylvania, for example, which are less than 20% impervi-
ous overall and are located on moderately permeable soils,
typically show no ill effects from lack of elevated street
curbs and storm sewers; good grass cover is maintained to
the margin of the roadway. The potential importance of this
is indicated by the finding mentioned earlier that the
hydrologic impact of urban development, per unit area of
impervious surface, is very strongly related to the extent
to which impervious surfaces are hydraulically connected.
On the other hand, the danger of providing only minimal
drainage facilities, and/or utilizing pervious drainage
channels, is that erosion or ponding will occur, plus pos-
sible safety hazards in the case of street surfaces. Thus,
land slope and soil conditions must be considered carefully.
145
-------
Control of the location of new development, as a means of
minimizing runoff and infiltration effects, could involve
two aspects of location: proximity of development sites to
watercourses (or to specific stream segments considered
important for protection); and on-site characteristics of
the land, such as soils, slope, and existing vegetation.
Emphasis upon the first aspect might lead to plans in which
stream valleys were designated as corridors of protection,
whereas emphasis upon the latter might lead to protection of
dispersed "sensitive areas." Empirical evidence suggests,
however, that the hydrologic impact of completed urban
development is much less sensitive to location factors than
to development design (Hammer, 1973a). This finding applies
to conventional types of development, in which a large pro-
portion of impervious surfaces are hydraulically-connected.
An implication is that control of development location is
beneficial primarily when combined with design modifica-
tions, and/or the use of runoff control devices. This
conclusion is believed to be generally valid also with
regard to the impacts of development on water quality caused
by washing of pollutants from impervious surfaces.
In the approach recommended here, the use of runoff deten-
tion and infiltration devices is generally favored over
reliance upon development design modifications or land use
controls. The design modifications mentioned are worthy of
encouragement, but have the following disadvantages for use
as primary water resources protection measures. First,
these are essentially partial measures; thus, preventive
devices such as detention basins must also be utilized if
full protection against hydrologic impacts is to be achieved,
Second, the effectiveness of design modifications (other
than limitation of impervious surface) are difficult to
establish quantitatively. This is a very important issue
since many municipalities do not possess adequate technical
resources for detailed review of runoff control plans sub-
mitted by developers. In contrast, the appropriate design
parameters for detention and infiltration devices can be
established in a fairly unambiguous manner; and guidelines
regarding these parameters can be developed by planning
agencies at the regional level (see below). Third, develop-
ment design modifications tend to be most feasible in cases
where they are least needed in terms of water quality. With
regard to control of development location, such controls are
difficult to implement unless imbedded within an overall
environmental plan, which deals with land use issues at a
much more detailed level than is possible in current re-
gional water planning efforts.
146
-------
The strategy recommended here would involve stringent con-
trol of the hydrologic impacts of new development, as a
means of achieving both water quantity and water quality
objectives. Primary emphasis would be placed on use of
detention and infiltration devices (although performance
standards allowing partial substitution of other controls
might be appropriate in communities possessing strong
capability for technical review). The desired situation
would involve "zero impact" regulations, wherein new de-
velopment projects—public or private—would not be allowed
to increase peak discharge at any point in the surface water
system, and would not be allowed to decrease the net rate at
which groundwater is replenished. Further details regarding
these objectives are discussed below.
It is clear that achievement of these objectives through
stormwater detention and infiltration would produce signifi-
cant water quality benefits, over and above the benefits
involving stream channel conditions. Detention devices are
moderately effective in reducing stormwater pollutant load-
ings, due to the tendency of suspended materials to settle
out while stormwater is in storage. The degree of pollutant
removal is highly dependent upon the storage capacity pro-
vided and the time of detention. A detention basin which is
adequate for peak discharge control will reduce pollutant
loadings in stormwater by roughly the following amounts:
40% to 80% for suspended and settleable material; very
little reduction for dissolved material; and 25% to 50% for
constituents such as organics which are transported partly
in dissolved form and partly in association with particulate
matter. Greater pollutant reductions are potentially
achievable through the use of infiltration devices, since
runoff entering the device is prevented from contributing to
pollutant loadings during the given storm (although subse-
quent groundwater pollution could be a problem, as mentioned
below).
Given the review of empirical data which has been discussed
earlier, it is felt that the degree of water quality control
which will be achieved by strict regulation of runoff quan-
tity can probably be considered a reasonable goal, in cur-
rent water planning efforts, for most types of new urban
development. This assumes that other controls dealing_with
construction sites and on-site waste disposal will be imple-
mented; that tight sanitary sewer systems will be construct-
ed; that detention and infiltration devices will be main-
tained adequately; and that ongoing programs will be con-
ducted to maintain high cleanliness standards and proper
147
-------
management of potential pollutants. Also, it is recognized
that.additional control devices could be needed for specific
land uses with high pollutant-generation potential (e.g.,
certain manufacturing operations and other establishments
engaged in the handling of bulk materials).
This strategy of focusing upon water quantity effects as a
basis for water resources protection is geared to the rela-
tively low level of knowledge which presently exists regard-
ing the effects of stormwater pollution in urban areas.
Until more is known about the seriousness of transient
phenomena, and about the ultimate impacts produced by con-
servative materials contained in stormwater, it will be
difficult to establish the appropriate levels of control
with any certainty. The measures proposed here have the
advantage that they would involve construction of runoff
storage capacity in conjunction with new development. If
later determinations are made that greater pollutant removal
is needed, this storage capacity (primarily in detention
basins) could be utilized in chemical treatment programs.
Emphasis upon runoff storage capacity thus has an important
aspect of flexibility in allowing for future upgrading of
control practices.
The costs of runoff quantity control are generally moderate.
Since detention and infiltration devices operate automat-
ically and do not require a high level of maintenance, the
major expense is the initial investment in physical facili-
ties. Thus, most of the cost of runoff control can be
assigned to the builder (although the public may assume
responsibility for maintenance of facilities). As an
example, the direct expense involved in providing runoff
detention facilities adequate to deal with the 10-year flood
is estimated by Betz Environmental Engineers (1975) to range
from 0.1% of total development cost, up to an absolute
maximum of 5% of development cost. Although detention and
infiltration devices may create opportunity costs, by
constraining somewhat the utilization of land in development
sites, these measures are felt to be generally feaszble for
virtually all types of new urban development.
The use of water quantity rather than water quality as an
explicit basis for design of controls has a number of
tactical advantages. Quantity-related problems such as
flooding and erosion are well-known to the layman, and are
already a subject of concern in many communities. The
148
-------
performance of control facilities in terms of water quantity
is relatively easy to monitor. The goal of "zero impact"
(which could not be feasibly employed in dealing directly
with stormwater quality) provides an unambiguous benchmark
for design of controls. Such a benchmark is often very
important in allaying criticism that regulations are arbi-
trary. The ability of detention and infiltration devices to
achieve these standards is a function of relatively few
design parameters in each case. The appropriate values of
these parameters may be difficult to determine with cer-
tainty; but their logical basis is intuitively understand-
able. Finally, considerable experience already exists in
the use of detention and infiltration devices. The feasi-
bility of regulations requiring these measures has been
demonstrated, as well as the effectiveness of the measures
themselves. (See for example: ABT Associates, 1976.)
It is recommended that regional water planning agencies
attempt to provide as much technical guidance as possible,
within the scope of current studies, for implementation of
water quantity controls. Since the appropriate design
parameters for detention basins and infiltration devices are
related to relatively few variables, it might be feasible to
develop suggested values of these parameters for the whole
study region. (The values might be presented in tabular
form; and the subareas to which specific tables apply would
be mapped.) A particular concern here is that sufficient
runoff storage capacity be provided, in both detention
devices and infiltration devices. It is felt that regional
planning agencies may be in a somewhat better position than
local governments to determine the capacities needed to meet
zero impact standards.
The various types of runoff control devices available, and
the manner in which they should be constructed, have been
discussed elsewhere in the literature; an excellent source
is the compendium of controls prepared as part of the
Christina Basin Study in Delaware (Tourbier and Westmacott,
1974). The remainder of the present section will deal with
conceptual issues involving design of detention and infil-
tration facilities to meet specific performance standards.
149
-------
Design of Detention Devices
As indicated previously, the essential features of a deten-
tion device are a water storage area, an outlet structure,
and one or more inlet structures. The outlet is assumed to
operate automatically whenever there is water in the storage
area. Generally, the rate of release is positively related
to the surface elevation of the stored water. An overflow
structure (e.g., a spillway) is also provided in most cases.
This operates only when the storage area is full, and the
rate of inflow continues to exceed the capacity of the
outlet.
A detention device can only reduce discharge significantly
while excess capacity exists in the storage area. When the
storage area is full, and overflow is occurring, the total
rate of outflow approximately equals the rate of inflow, so
that the structure has little effect. The critical factors
in the design of a detention device are thus the volume of
storage provided, and the flow capacity of the outlet.*
These are the factors which determine the probability that
excess storage capacity will exist at any given time. A
detention device is said to provide adequate control of dis-
charge during a storm event if (1) the rate of release
through the outlet is not unacceptably high, and (2) over-
flow does not occur during the storm. Design of a detention
device usually involves selecting an outlet capacity and a
storage volume which will allow these conditions to hold
during a single "design storm," defined in terms of a given
intensity and duration of rainfall.
The "zero impact" standard for detention facilities is that
peak discharge should not be increased relative to condi-
tions which prevailed before development. Estimates of
discharge prior to development are thus neccessary for
* Since the rate of flow through the outlet at any time
depends upon the amount of water in storage, it is con-
ventional to define outlet capacity as the maximum re-
lease rate which occurs when the storage area is full
but not overflowing.
150
-------
design of these facilities. Although many estimating tech-
niques for small catchments are available, it is recommended
hetfe that hydrologic analysis be linked to materials pre-
pared by the U.S. Department of Agriculture, Soil Conserva-
tion Service. The specific reference utilized here is the
excellent S.C.S. manual entitled "Urban Hydrology for Small
Watersheds" (.S.C.S., 1973). This document contains simpli-
fied methods for routing streamflow, as well as methods for
selecting detention basin design parameters. The present
discussion will consider some of the issues which arise when
such a methodology is employed.
A major decision in runoff planning is the choice of a
design storm, which expresses the level of protection that a
detention facility is intended to provide. Use of a 10-year
storm, for example, represents a decision that there should
be no increase in the peak discharge produced by a storm
which is equaled or exceeded only once in every 10 years on
the average. It is commonly assumed that a detention fa-
cility which controls a given design storm will also prevent
increase in discharge during all smaller storms (although
this is not necessarily true, as shown below). Detention
facilities which are constructed for flood control purposes
usually are based on very low-frequency storms, such as a
50-year or 100-year storm.
When detention facilities are recommended as part of current
water planning studies the explicit justification for these
controls may be limited to prevention of stream channel
enlargement, channel disruption, and the resulting sediment
yield. These factors can only justify detention facilities
which deal with relatively frequent storm events, since
channel characteristics relate primarily to high-frequency
floods. As noted earlier, the channel-forming discharge
tends to be the 1.5-year flood.* However, the objective of
preventing channel-related effects can justify control of
* This means that (1) channel characteristics tend to be
correlated more highly with the 1.5-year flood discharge
(estimated using an annual series) than with the dis-
charges associated with other recurrence intervals? and
(2) channel capacity tends on the average to equal the
1.5-year flood. These characteristics are illustrated
in Hammer (1973a).
151
-------
somewhat rarer events than this. The general situation is
depicted in Figure 8. Relationships between discharge and
recurrence interval, based on a hypothetical small catchment
in Pennsylvania before and after urbanization, are shown by
solid lines. Without preventive measures, the effect of
urban development is to increase the discharge associated
with each recurrence interval. (The percentage increase is
inversely related to recurrence interval.) The upper dashed
line indicates the discharge relationship that might prevail
if detention facilities were utilized, which were capable of
controlling a 1.5-year storm event. As suggested earlier,
detention facilities have relatively little effect on dis-
charge during events which exceed the design storm, i.e.,
which cause overflow to occur. Thus, the discharge rela-
tionship under discussion rapidly approaches the relation-
ship denoting no preventive measures, when recurrence in-
tervals greater than 1.5 years are considered. It is
reasonable to assume that stream channel enlargement and
disruption would occur in such a situation, even though, the
1.5-year flood per se is controlled. The lower dashed line
denotes a situation in which detention facilities have been
designed to deal with the 5-year flood. Here, control
should be achieved for all of the recurrence intervals
relevant to channel characteristics. This is felt to be a
more realistic design, in cases where channel protection is
the major objective.
Control of events greater than the 5-year storm is neverthe-
less highly desirable, and should be required whenever it is
feasible to link water quality objectives with flood control
objectives. In addition to providing greater flood pro-
tection, the use of an infrequent event as the design storm
has important structural implications for detention facili-
ties. These facilities, as applied to new urban develop-
ment, typically consist of an impoundment formed by an
earthfill dam. If the basin is designed to deal with an
infrequent event such as the 50-year storm, overflow of the
basin will occur with a correspondingly low frequency.
Under these circumstances, it is normally adequate to pro-
vide an unpaved spillway, consisting of an undisturbed soil
surface located at either side of the dam. However, in
cases where overflow occurs every 5 or 10 years, it may be
152
-------
o
CK
<
u
Relationship After Urbanization,
Without Preventive Measures
Relationship
Before Urbanization
I
Possible Relationships
After Urbanization,
With Detention Basins
I
1.1
1.5 2.33 5 10
RECURRENCE INTERVAL IN YEARS
20
50
Source: Betz Environmental Engineers, Inc.
Figure 8 HYPOTHETICAL FLOOD-FREQUENCY RELATIONSHIPS
153
-------
necessary to provide a concrete spillway. This adds sub-
stantially to the cost of the dam, and may offset the
savings of providing less storage capacity.*
Conventionally, the event selected as the design storm for a
detention facility is assumed to involve a spatially uniform
pattern of rainfall. The duration of rainfall is subject to
discretion. (In applying the "rational method," for ex-
ample, the rainfall duration is a function of basin charac-
teristics.) The S.C.S. methodology cited here is based on a
24-hour storm, with a "Type II" distribution of rainfall
over time. The methodology is applied by consulting rain-
fall records or generalized maps to determine the 24-hour
rainfall magnitude corresponding to the chosen recurrence
interval (e.g., the 5-year, 24-hour rainfall in inches).
This quantity is used to compute runoff volume, which is
then routed through the watershed under study using various
simplified techniques. It should be noted that the S.C.S.
methodology is inherently conservative when;applied in many
regions of the U.S., since the Type II rainfall distribu-
tion—which is highly concentrated in time--represents a
worst-case condition rather than an average'condition.
Several issues involving design of detention facilities can
be illustrated by considering the hypothetical example shown
in Figure 9. An area of roughly 100 acres within a water-
shed is to be developed; this is depicted in Figure 9 as
"zone 3.M Since the area consists of a single catchment,
the developer has chosen to provide runoff control by con-
structing a single detention basin at the mouth of the
catchment (point A). The detention basin is to involve an
earthfill dam with a single outlet pipe through its base.
The watershed data which are utilized in evaluating the
performance of this basin are shown in Table 11; the reader
* A recent environmental ordinance in the City of Phila-
delphia requires runoff controls for new development in
the Wissahickon Creek watershed. Preservation of natural
stream channels was the primary justification for these
controls; and original specifications called for the use
of a 5-year design storm (Coughlin and Hammer, 1973).
The Water Department subsequently requested use of a 25-
year storm, however, in order to limit legal responsi-
bility for possible problems involving overflow.
154
-------
Source: Betz Environmental Engineers, Inc.
Figure 9 HYPOTHETICAL WATERSHED USED IN HYDROGRAPH COMPUTATIONS
TABLE 11
DESCRIPTIVE DATA FOR HYPOTHETICAL WATERSHED
Time of Time of
Drainage Concentration Travel
Zone No.
1
2
3-present
3-Puture
Area (Mi )
1.50
0.35
0.15
0.15
(Hours)
1.25
0.75
0.50
0.25
(Hours)
0.25
0.00
0.25
0.25
Runoff
Curve No.
65
65
65
80
Storm Runoff (inches)
5" Rainfall
1.65
1.65
1.65
2.89
3" Rainfall
0.51
1.65
Source: Betz Environmental Engineers, Inc.
155
-------
VS = VOLUME OF STORAGE (Acre - Feet)
VR=VOLUMEOF RUNOFF (Acre-Feet)
0.0= PEAK RATE OFOUTFLOW(Cfs)
QI = PEAK RATE OF INFLOW(Cfs)
0.1 0.15 0.2 0.3 0.4 0.5 0.6 0.8
RATIO QO/QI
Source: S.C.S., 1973
Figure 10 RELATIONSHIP FOR DETERMINING
REQUIRED STORAGE CAPACITY OF DETENTION BASINS
-------
250 i- 5-INCH RAINFALL
11=00
After Develpment
*V .With Detention Basin
'•
11:30 12:00 12:30
TIME OF DAY IN HOURS
1:00
r
!_•
c
u
5
K)O i- 3-INCH RAINFALL
After Development
Before Development
I I
•.With Detention Basin
0
11:00 J1:30 12:00 12:30
TIME OF DAY IN HOURS
Source: Betz Environmental Engineers, Inc.
Figure 11 DISCHARGE HYDROGRAPHS AT POINT "A"
157
-------
is referred to the S.C.S. manual for discussion of these
variables.
Assume that the county or township in which the hypothetical
development is located has an ordinance which specifies that
the maximum rate of discharge from a development site during
a 10-year storm shall not exceed the corresponding rate of
runoff before development. The 10-year, 24-hour rainfall in
the given region is a 5-inch rainfall. The S.C.S. method-
ology indicates that, before development, the peak discharge
produced by this storm at point A is 118 cubic feet per
second. Following conventional practice, the first step in
designing the detention basin is to specify that the flow
capacity of the outlet pipe—i.e., the rate of outflow when
the basin is full—will be equal to 118 cfs. It is then
determined that a storage volume of 5 acre-feet (215,000
cubic feet) will be sufficient. Given this volume of
storage, the basin will fill to capacity but not overflow in
the 10-year design storm. The S.C.S. handbook presents
graphical methods for determining storage volume very
quickly. One of these graphs is presented here as Figure
10? the curve shown is utilized to determine "Vs" on the
basis of the variables, Vr, Qo, and Qi.
The performance of the basin in a 5-inch storm is shown in
the upper portion of Figure 11. The solid lines are hydro-
graphs (plots of discharge versus time) representing dis-
charge from the catchment before and after development,
without the detention basin. (The time scale shown in the
diagram represents an arbitrary time-of-day assignment used
in the S.C.S. handbook.) The dashed line in Figure 11 shows
discharge from the catchment when the detention basin is
provided. This hydrograph represents flow through the basin
outlet.* Since the upper sqlid line represents inflow to
the basin, the rate of inflow exceeds the rate of outflow
from about 11:30 o'clock until 12:15. The basin is filling
* Unlike the other hydrographs, detention basin outflow in
this example was not computed using the simplified S.C.S.
methods. It was necessary to assume specific storage/
elevation and elevation/discharge relationships in order
to illustrate various points discussed below.
158
-------
900 i- TOTAL DISCHARGE AT POINT B
800
700
600
500
~ DISCHARGE AT POINT B FROM ZONE 3
c
CASES CONSIDERED
I— Before Development
II—After Development
III - After Development
with Retention Basin
N-.30 12=00 12=30 1-00
TIME OF DAY IN HOURS
Source: Betz Environmental Engineers, Inc.
1:30
Figure 12 DISCHARGE HYDROGRAPHS AT POINT "B"
159
-------
during this time. At 12:15, where the curves cross, the
basin is very nearly full, and the rate of flow through the
outlet is equal to the peak discharge which would have
occurred before development (118 cfs). After this point,
the rate of outflow exceeds the rate of inflow, and the
volume of water in storage decreases. Thus, the basin
effectively spreads out over time the release of runoff from
the catchment. This example illustrates two very important
points which are generally relevant for the design of ;
detention basins.
First, a detention facility which controls a given design
storm wxll not necessarily prevent increase in the dis-
charges produced by smaller storms. The lower portion of
Figure 11 illustrates what happens in the present example
during a 3-inch rainfall. Peak discharge increases by 100%,
even with the detention basin, since water entering the
basin runs directly through within a short period of time.
The outlet pipe, which has been sized to deal with a 10-year
storm, is too large to provide adequate control in this
case. Given that the 3-inch rainfall represents perhaps a
1-year event, it is very likely that this basin would not
prevent channel enlargement and disruption downstream.
Second, a detention facility which controls peak discharge
at one point may not provide adequate control at other
stream points. This fact is illustrated by Figure 12* The
hydrographs in Figure 12 pertain to stream discharge at
point B, which subtends the entire watershed shown in Figure
9. The lower portion refers just to the discharge at point
B which is contributed by zone 3 (the development area).
The upper portion of the diagram, which utilizes a different
vertical scale, shows the overall discharge at point B. An
important fact is that, without the, detention basin, runoff
from zone 3 tends to arrive at point B well before the
overall peak discharge at B, due to the circumstance that
zone 3 is located near the mouth of the given watershed.
One effect of development in zone 3 without the detention
basin is to advance the arrival of runoff at point B. As a
result, the overall peak discharge at B is increased very
little. (Compare curves I and II in the upper portion of
Figure 12.)
: • 'i ' '
The dashed curves in Figure 12 depict the conditions which
prevail if the detention basin is constructed as described.
An interesting fact revealed by the lower curve is that the
160
-------
peak rate at which discharge is delivered to point B from
zone 3 is greater with the detention basin that before de-
velopment, by about 22 cfs. This occurs even though the
basin effectively prevents increase in peak discharge at
point A. The reason is that, under most conditions, runoff
from a given land area tends to be dispersed during flow in
natural channels, due to overbank storage and other factors.
As a result, the peak rate of discharge contribution is
reduced downstream. In the present example, zone 3 before
development yields a peak discharge of 118 cfs at point A,
but only 92 cfs at point B. However, this dispersion effect
is reduced when detention basins are provided, since de-
tention basin outflow produces a relatively "flat" hydro-
graph. In the present case, discharge from the detention
basin at point A remains near its peak long enough for
equilibrium flow conditions to develop in downstream channel
reaches, with the result that the peak is effectively trans-
mitted to point B.
A second effect is that, because of the runoff delay accom-
plished by the detention basin, the arrival of this flow at
point B now coincides with the arrival of runoff from the
major land area in the watershed (zone 1) . The dashed line
in the upper portion of Figure 12 shows that the net result
of these two effects is an increase in overall peak dis-
charge, relative to both the pre-development condition, and
the post-development condition without the detention basin.
The magnitudes of these increases are 50 cfs and 38 cfs,
respectively. The detention basin thus makes matters worse
at point B.
Although this particular example is somewhat artificial, in
that it reflects the rigidities of the design storm concept
used by S.C.S. and others, it demonstrates a critical issue
in detention basin design, which can be summarized briefly
as follows. For any given watershed and recurrence inter-
val, peak discharges at points within the watershed tend to
vary less than proportionally with drainage area. (On the
East Coast, peak discharge typically varies as the 0.8 power
of drainage area.) This is due to: the dispersion effect
just discussed; variation among subareas in terms of travel
time to specific stream points; and the fact that real-life
conditions often involve non-uniform rainfall. Detention
basins, by creating relatively flat hydrographs for sub-
areas, tend to minimize the importance of the latter fac-
tors. The result is that peak discharge tends to be more
161
-------
nearly; proportional to drainage area. Situations can then
develop in which there is no increase in peak discharge from
each subarea of an urbanized watershed, and yet there are
serious flooding and channel enlargement problems on the
ma instem.
Given these issues, it is obviously desirable to develop
design criteria for detention basins which will guarantee
uniform protection against peak discharge increase, for all
recurrence intervals up to the design recurrence interval,
and for all relevant stream points.
The need to provide protection against a variety of storms
has special implications for basin design. Specifically, a
design in which the stored water drains directly into a
single outlet pipe or culvert may not be adequate. It may
not be possible to achieve an appropriate rate of outflow
when the basin is nearly full, without releasing too much
water when the basin is nearly empty. One way to increase
the variation in outflow rate is to have water enter the
outlet by way of a perforated riser pipe. The perforations
would be arranged so that the riser has a constraining
effect on the rate of release until a large proportion of
storage capacity is utilized. An alternative solution is to
provide two or more outlets at different elevations.
Determination of the optimal storage capacity and outflow
rates to provide uniform protection against discharge
increase is potentially very complex. Evaluation of deten-
tion basin performance, assuming a given design, may require
direct computations based on the storage-discharge rela-
tionship, rather than the use of simplified methods. In
determining design parameters, it is theoretically necessary
to consider the effects produced at numerous downstream
points, in order to determine the point which is most
constraining in terms of permissible outflow. (Also, the
most constraining point may shift over time, due to various
effects of urbanization.) Although such computations can be
carried out on a selective basis, to serve as demonstrations
or to defend runoff control requirements when challenged, a
much simpler procedure is suggested here.
Flood frequency relationships (such as depicted in Figure 8)
would be established on the basis of empirical data for a
number of watersheds about 5 square miles in size. The
162
-------
flood-frequency relationships would represent pre-urbaniza-
tion conditions. If peak flow characteristics were known to
vary systematically within the study region (due to factors
such as soil associations and topography) hydrologic zones
could be designated, each of which would be represented by
one or more of the sample basins. Such zones might already
have been designated in regional flood-frequency studies
published by the U.S. Geological Survey. Appropriate out-
flow rates for detention basins would then be obtained
directly by converting these discharge rates to an areal
basis. For example, suppose that a 25-acre development site
is located in a hydrologic zone for which the 5-year flood
discharge in a 5-square-mile watershed is 750 cfs, or 150
cfs per square mile. Assuming that a 5-year design storm is
relevant, the maximum outflow rate for the detention basin
serving this development would be: (25/640) 150 = 6 cfs.
Maximum outflow rates for storms less than the design storm
would also be computed in this fashion. For example, if the
1.5-year flood in the abovementioned watershed were 375 cfs,
or 75 cfs per square mile, the maximum outflow rate from the
detention basin during a 1.5-year rainfall should be about 3
cfs.
The use of a 5-square-mile watershed as the standard for
these computations represents a compromise among several
factors, and has no theoretical basis; but the outflow rates
thus yielded are thought to be appropriate. Regardless of
the methodology utilized for this purpose, the important
points are the following. First, reasonable rates of out-
flow for detention facilities (per unit area of land served
by the facility) can be established on a regional basis.
The storage capacity required to prevent overflow in the
design storm can then be determined for a given development
site, on the basis of land characteristics and the (tenta-
tive) storage-discharge relationship for the detention
basin. A regional planning agency could provide tabulated
values of storage capacity, for a series of assumed cases in
each hydrologic zone, as a type of planning assistance to
local communities. Second, whatever the fashion in which
outflow rates are derived, they should be very much lower
than the rates obtained by considering only discharge from a
development site per se. The methodology suggested here
yields maximum outflow rates which are only between 20% and
50% as great as the rates obtained in the fashion just
noted, for development sites ranging from 1 to 100 acres.
Third, the use of low maximum outflow rates, and additional
163
-------
constraints on outflow during partially-full conditions,
will have major effects on the volume of storage capacity
required.
Design of detention facilities with excessive outflow rates
is believed to be a major flaw in present runoff control
practices, which in some cases practically nullifies the
value of detention facilities. Although low outflow rates
are fully justifiable on the basis of runoff quantity con-
siderations, their recommendation here is importantly
related to the strategy of using detention facilities as
water quality controls. Low outflow rates greatly increase
stormwater detention time, and thus significantly improve
the efficiency of detention devices in reducing stormwater
pollutant loads.
Design of Infiltration Devices
In concept, the use of infiltration devices is perhaps the
most appealing form of water quantity control, since it
parallels most closely the natural processes which are lost
when land is covered by impervious surface. The present
discussion will be limited to the typical situations in
which soil permeability is not so great that infiltration
devices can be used feasibly to control peak discharge.
Thus, the following objectives of infiltration devices are
of primary concern:
1. Aquifer recharge
2. Maintenance of base flow in surface waters
3. Reduction in pollutant yields to surface waters
The design of infiltration as described here will relate
only to the first two of these objectives, although a
methodology for estimating the reduction in pollutant load-
ings achieved by these devices is mentioned. The "zero
impact" standard that infiltration devices would be designed
to meet would be that new development should not result in a
decrease in, net infiltration—defined as infiltration minus
evapotranspiration (minus infiltrated water which re-emerges
as surface runoff during storm periods).
Ideally, the design parameters for an infiltration device
would be determined through a process such as the following.
Factors which affect infiltration such as vegetation, soil
164
-------
type, land gradient, and hillslope position would be estab-
lished for the development site. On the basis of these
variables, and the detailed development specifications,
annual infiltration and evapotranspiration would be esti-
mated, for conditions before and after development. Effects
involving imported water, on-site sewage disposal, and
pumpage from wells would be ignored in the latter case.
Infiltration devices would then be designed to compensate
for the computed annual reduction in net infiltration, using
techniques such as the Hydroscience methodology discussed
below. The performance of infiltration devices involves a
trade-off between storage capacity and the rate at which
infiltration is achieved, which is analogous to the trade-
off between storage capacity and outflow rate for detention
basins. Since the rate of infiltration within a device
involves all of the variables listed above, these would have
to be consulted again in determining the appropriate design
for a specific facility.
Such a procedure is obviously infeasible in most instances,
and is described only to illustrate the factors involved.
The actual approach recommended here would involve an ex-
treme level of generality. The first task would be to
obtain a regional estimate of net infiltration under rural
conditions. A gaged watershed in a rural area would be
selected, for which it could be assumed that net groundwater
flow into or out of the basin was negligible relative to
surface and subsurface flows within the basin. Long-term
streamflow records would be analyzed; and base flow separa-
tion techniques would be applied to obtain an estimate of
annual base flow in inches. Annual infiltration minus
evapotranspiration would then be computed as the sum of the
following (all in watershed inches per year): base flow,
plus exported water, minus imported water. This estimate of
infiltration minus evapotranspiration would be considered to
hold uniformly for all areas of land before development.
The error involved in this assumption should not be overly
serious, since underprediction or overprediction of net
infiltration in any particular case should be followed by a
corresponding underprediction or overprediction of the
performance of infiltration devices after development.
In regions of the U.S. where natural vegetation is rela-
tively lush, it may be adequate to assume that infiltration
and evapotranspiration per unit area of pervious soil are
the same both before and after land development. The effect
165
-------
of development would then be simply to reduce net infiltra-
tion by.the same proportion that pervious soil is reduced.
A different formula could be used when natural vegetation is
sparse, although the above assumption might still be found
reasonable due to various offsetting factors. Only two
general design parameters for infiltration facilities would
be considered: storage capacity, and the percentage of
impervious surface in a development which drains into these
facilities. (It is assumed that no pervious land is drained
into infiltration devices.) Based on an assumed value of
the latter parameter, an estimate of required storage capa-
city would be obtained using a methodology developed for EPA
by Hydroscience (forthcoming report).
The Hydroscience methodology deals with three generic types
of treatment devices: (1) flow removal devices, (2) volume
removal devices, and (3) flow proportional removal devices.
Conceptually, an infiltration facility could be considered
as either a flow removal device or a volume removal device.
The latter is thought to be more appropriate here under
present assumptions. A volume removal device is a facility
which retains runoff up to a volume "V," and bypasses all
remaining flow. The Hydroscience materials permit computa-
tion of the proportion of annual runoff which is captured by
such a facility, as a function of storage capacity.
The graph developed by Hydroscience is based on the assump-
tion that the storage area is always empty at the start of a
storm event. In order to compensate for this assumption, as
well as to simplify the design problem for infiltration
facilities, the procedure adopted might make no allowance
for infiltration achieved during storm events (which would
allow re-use of storage capacity within a given storm). An
infiltration device would thus be represented as a facility
which captures runoff up to a volume V, holds this water for
the duration of the storm, and transfers it completely to
the soil at the end of rainfall. This assumption makes it
unnecessary to consider explicitly the areal extent of the
interface between the water storage area and the surrounding
soil. The only design parameter considered explicitly would
thus be the storage capacity of the device. (This would be
equal to the volume of the storage chamber, if underground,
times the porosity of the material occupying the chamber.)
The suggested procedure is demonstrated in Table 12, using
the example of a 30% impervious development with half of all
impervious surface draining into infiltration devices. This
166
-------
TABLE 12
SAMPLE COMPUTATION OF STORAGE CAPACITY IN
INFILTRATION DEVICES FOR NEW DEVELOPMENT
Land Characteristics
A. Annual rainfall:
B. Number of storm events per year (days
with rainfallSt 0.10 inch):
C. Annual net infiltration (infiltration
minus evapotranspiration) before
development:
Development Characteristics
D. Proportion of land area covered by
impervious surface:
E. Percent of impervious surface draining
into infiltration devices:
F. Proportion of land area consisting of
impervious surface draining into
infiltration devices (= D times E):
G. Runoff coefficient for impervious surfaces:
Derived Variables
H. Required infiltration to be achieved by
devices (= C times D):
I. Required infiltration, relative to land
draining into devices {= H/F):
J. Annual runoff from land draining into
devices (= A times G):
K. Required proportion of runoff to be
retained by devices (= I/J):
L. Required volume of storage relative to
runoff volume in average storm (from
Hydroscience, based on item K):
M. Required storage volume per acre of
land draining into infiltration devices
(= L times J/B):
N. Required storage volume, per acre of
land in whole development site (= M
times F):
3.33 acre-
feet/acre
80
0.83 acre-
feet/acre
0.30
50%
0.15
0.9
0.25 acre-
feet/acre
1.67 acre-
feet/acre
3.00 acre-
feet/acre
0.56
1.25
0.047 acre-
feet/acre
0.007 acre-
feet/acre
Source: Betz Environmental Engineers
167
-------
could be a single-family housing development with infiltra-
tion devices designed to serve only rooftops, patios and
parts of driveways. Annual rainfall is assumed equal to 40
inches (3.33 acre-feet per acre), with one-fourth of this
quantity contributing to net infiltration under natural con-
ditions. Construction of the development is assumed to re-
duce net infiltration by an amount proportional to impervi-
ous coverage. This reduction works out to 0.25 acre-feet
per acre of land in the development. Determination of the
storage capacity in infiltration devices needed to retain
this volume of runoff on an annual basis is illustrated in
steps I through M of Table 12. The key quantity is the
ratio of required infiltration (annual retention of runoff
in the devices) to annual runoff from land draining into the
devices. This quantity is applied to the Hydroscience graph
to obtain required storage capacity, as a proportion of the
average runoff per storm event.
In the example shown, the required storage capacity of in-
filtration devices works out to 0.047 acre-feet per acre of
land draining into the devices, or 0.007 acre-feet per acre
of land in the development as a whole. It is apparent that
this requirement is fairly modest. For example, if the de-
velopment involves 1/4-acre house lots, adequate infiltra-
tion facilities could consist of a gravel-filled trench 3
feet deep (excluding earth cover), 2 feet wide, and 26 feet
long on each lot. (This assumes a gravel porosity of 0.5.)
Construction of these devices and associated drainage fa-
cilities would increase total development cost by only a
very small percentage.
The overall procedure which is recommended here for promot-
ing the use of infiltration measures would be to develop
general guidelines for an entire study area or watershed.
Various values would be assumed for imperviousness and
percent of impervious surface draining into infiltration
devices. Storage capacity requirements would then be com-
puted as a function of these variables, using the Hydro-
science methodology and regional values of annual rainfall,
rainfall distribution over storm events, and infiltration
minus evapotranspiration. The resulting tabulations would
be made available as reference materials to local govern-
ments and other enforcement agencies. Review of development
proposals by the latter would then involve: (1) comparison
of storage capacity with the recommended amount indicated by
the tables; and (2) verification that the design of each
168
-------
device will allow the water storage area to empty within a
reasonable length of time, and that bypass water is handled
adequately.
It is recognized that, due to the role of imported water,
implementation of infiltration controls in this fashion
could produce situations in which groundwater recharge is
increased by new urban development. Nevertheless, the
controls are still considered justifiable, in view of the
greater demands for water use which will result from new
development. An impact of urbanization which is extremely
important is the increased demand for waste assimilative
capacity in receiving waters. Increased groundwater re-
charge, resulting in higher base flow of streams, could be
justified on these grounds alone.
It is clear that infiltration devices can achieve major
reductions in stormwater pollutant loadings. For storms
which do not exceed the design storage capacity, infiltra-
tion devices are potentially 100% effective in retaining
pollutants from the areas served; and the percentage removal
should be high even for larger storms, due to capture of the
"first-flush" effect. If reduction in stormwater pollutant
loadings is the primary objective of infiltration devices,
the facilities would presumably be designed to serve all
impervious surfaces in a developed area, or at least those
surfaces most likely to contain pollutants. (One problem is
that it is relatively difficult to serve street surfaces,
which are thought to contain the largest quantities of pol-
lutants, due to the safety factors involved in street drain-
age.) The effectiveness of infiltration devices in reducing
stormwater pollutant loadings can be evaluated using another
Hydroscience methodology, which takes account of variation
in pollutant concentrations over time. Values of two param-
eters describing this variation are assumed; the percentage
removal of pollutant loadings is then obtained by reading a
value off a curve as before. (The relevant figure is en-
titled: "Fraction of Runoff Loading Rate not Retained".)
The reduction in pollutant loadings is generally greater on
a percentage basis than the reduction in stormwater volume,
due to the inverse correlation between pollutant concentra-
tions and time since start of storm.
A major unresolved issue is whether the use of infiltration
devices for control of stormwater pollution runs the risk of
contaminating the groundwater. Some proportion of the
169
-------
stormwater pollutants which pass through infiltration de-
vices may reach the water table/ and eventually reach sur-
face waters* rather than being filtered out by the soil.
This would not appear to be a major problem for sediment,
phosphorus, or oxygen-demanding material in stormwater.
Nitrate is rather likely to reach the water table, but may
not be highly important since impervious surfaces do not
tend to be dominant sources of nitrate. Salinity, resulting
from the use of de-icing compounds, could definitely be a
factor. The major question marks are heavy metals and
perhaps pathogens. Delivery of the former to groundwater
supplies could have serious long-term effects, although the
probability of this occurrence due to infiltration devices
is unknown.
*-..'-
On the other hand, agencies might prefer to face the possir-
bility that contaminants are present in groundwater within a
limited area, than to know for sure that they are in the
stream. Infiltration devices are thus considered here to be
a legitimate and perhaps highly favorable means of reducing
stormwater pollutant loadings, but should be designed for
this purpose only if: (1) the bottom surface of each device
is situated at least several feet above the seasonal high
water table, and (2) there is no withdrawal of groundwater
in the vicinity for domestic or other use. These conditions
could well be met in many medium-density urban areas.
When infiltration devices are designed to serve impervious
surfaces with large contaminant loadings, a potential
problem is that particulate material could clog the device,
thus necessitating frequent rehabilitation. A necessary
step in such situations may be to provide a polishing pond
in addition to the infiltration device itself. Most of the
particulate material present would hopefully settle out
before water is presented for infiltration. This step might
allow reduction or even elimination of storage-capacity in
the infiltration device per se, but would complicate the
overall facility, and might tend to offset the advantages of
infiltration devices relative to stormwater retention and/or
treatment.
In cases where reduction of pollutant loadings is not the
primary objective, infiltration devices may be best applied
to runoff from rooftops and paved areas which are not in-
tensively used. Even so, rehabilitation of facilities may
be necessary at long intervals. This would involve exposure
of the storage area, if underground, and washing or sorting
of the porous material used to provide storage capacity.
170
-------
A major advantage of infiltration devices is that they can
be invisible, as well as automatic in operation. If prop-
erly designed, infiltration facilities need not affect the
land contour or vegetative cover (i.e., grass), although
maintenance of vegetation may require placement of an im-
permeable shield over the storage area in order to prevent
overly rapid drainage of the overlying soil. The use of
these devices may place considerable constraints on develop-
ment design, however. Each individual facility can or-
dinarily serve only a fairly small impervious surface, and
should be located some distance from building foundations
and other areas where soil wetness may be a problem.
The infiltration capacity of soils may also be a limiting
factor. Infiltration devices should ordinarily be located
above the seasonal high water table, in soils which are at
least moderately permeable. Soil requirements should per-
haps not be interpreted too stringently, given the facts
that: (1) infiltration devices tend to work best when they
are needed most, i.e., during dry weather; and (2) poor
performance of infiltration devices during wet periods need
not create special runoff problems if bypass is handled
adequately. Although it clearly makes little sense to
provide these facilities in cases where they do not produce
significant benefits on a long-term basis, the possibility
of their use should at least be considered for all new
development, especially residential development.
171
-------
SECTION 10
ASSESSMENT AND CONTROL OF ON-LOT
DISPOSAL SYSTEM PROBLEMS
Introduction
On-lot disposal systems (OLDS) receive the wastewater from
approximately 29% of the housing units in the United States.
Nearly 20 million year-round housing units, occupied by
roughly 58 million people, are served by OLDS. Of those
people served by OLDS, 34 million live outside standard
metropolitan statistical areas (SMSAs) and 25 million live
inside SMSAs. Of the latter, 2.5 million live in central
cities (Salvato, 1975). Thus, the use of OLDS is not
limited to rural areas. OLDS also serve vacation homes,
hotels, motels, camps, tent and travel trailer facilities,
restaurants, gas stations, and other commercial properties.
An OLDS, for purpose of this discussion, includes any system
or combination of systems which: (1) is currently receiving
or is designed to receive wastewater from individual dwel-
lings or establishments; (2) is designed to proyide some
degree of treatment for these wastewaters; and (3) dis-
charges an effluent to the soil where further renovation of
the effluent is expected to occur. Examples of OLDS in-
clude, but are not limited to:
septic tank - drain field
septic tank - seepage pit
cesspool
aerobic system - drain field
septic tank - alternate disposal system
Not included in the OLDS category are package treatment
plants or lagoon systems (unless these systems use the soil
for effluent renovation) and spray irrigation systems.
Because OLDS are being utilized for wastewater treatment in
many rapidly developing areas, a generalized methodology may
be needed to evaluate their existing and potential impact on
water quality. The present section discusses various
approaches which could be utilized.
172
-------
The Problem - OLDS Impact on Water Quality
Many people believe that only those OLDS which are malfunc-
tioning cause water pollution. While malfunctioning OLDS
have significantly greater impact on water quality than
properly operating systems, the latter can contribute to
groundwater quality degradation.* Figure 13 schematically
illustrates some of the water quality problems associated
with OLDS. The major causes of OLDS malfunctions are: (1)
improper system design or installation; (2) improper use;
and (3) inadequate maintenance. With regard to the first
item, a vast majority of system failures are caused by
installing an OLDS in soils which are not physically or
chemically suited for renovation of wastewater. Soils may
be unsuitable for OLDS due to the following factors.
1. Insufficient or excessive drainage
2. Insufficient thickness to limiting layer
3. Poor gradation (i.e., rocky)
4. Presence of a seasonally or permanently high water
table
5. Excessive slope
Even under ideal soil conditions for an OLDS, a system may
be doomed to failure because of lax construction and inspec-
tion practices, or the use of inferior construction materi-
als. Improper use of OLDS can cause sudden or gradual
malfunctioning of the system. The two major categories of
improper use are: (1) admission of extraneous water to the
system (i.e., downspouts, sump pump, etc.); and (2) use of
incompatible chemicals which harm or destroy the biological
balance in the system (petroleum, oils, solvents, and harsh
household chemicals such as sodium sulfate). Finally,
* This generally occurs when septic tank concentrations are
too dense and proper dilution of tile field drainage is
not achieved.
173
-------
OLDS
Normally Functioning
System
Malfunctioning
System
Leachate Renovated
by Soil
Direct Hydraulic
Connection to
Groundwater
Drain Field Installed
Too Far Below
Soil Surface
Soil Clogged
From Improper
Use or Maintenance
Ammonia and organic
nitrogen is converted
to nitrate form under
aerobic conditions as
soil and nitrate moves
downward with effluent
to groundwater
Little or no
renovation of
effluent by
soil before
mixing with
groundwater
Anaerobic conditions
in soil produce
organic intermediate
products which can
contaminate ground-
water
Soil becomes
saturated, which
causes anaerobic
conditionsf odor
and runoff problems
Sourcei B«tz Environmental Engineer*, Inc.
Fifon 13 SUMMARY OF OLDS IMPACT ON WATER QUALITY
-------
problems due to lack of maintenance generally involve
failure to remove periodically the accumulated sludge from
the system.
OLDS Effect on Surface Water Quality
Impact of OLDS on surface water quality can involve either
direct surface discharge to receiving waters, or outflow of
groundwater which has been contaminated by on-lot systems.
Direct discharge of OLDS effluent to surface watercourses
usually results from a hydraulic failure of the soil absorp-
tion portion of the system (seepage bed, tile field, etc.).
Wastewater applied to the seepage bed or tile field must go
somewhere; when it can no longer percolate through the soil,
it surfaces and flows to the nearest drainageway. Other,
less common causes of system failure to the surface include
broken or crushed distribution lines, and blockage of lines
by accumulated solids or root intrusions. Overflow of an
OLDS to a surface watercourse is, from a water quality
standpoint, similar to raw sewage discharge to that water-
course. Table 13 compares average concentrations of several
parameters in a medium-strength raw sewage to an average
septic tank effluent.
Effects on surface water quality due to outflow of ground-
water can be produced by normally functioning OLDS as well
as malfunctioning systems. The mechanisms involved are the
same as those discussed below for groundwater contamination.
For normally functioning systems which are well-located with
regard to water quality, the surface water effects tend to
be limited to nitrate and various dissolved solids of lesser
importance. In other cases, the pollutants delivered to
surface waters by way of groundwater can also include phos-
phate, pathogens, and oxygen-demanding material.
In regions where OLDS impact on surface waters is a major
issue, it may be helpful to develop generalized relation-
ships which can be used to predict the impact of this factor
over large areas. Evidence suggests that such relationships
can be successfully obtained from empirical studies, under
at least some circumstances, if steps are taken to control
for soil characteristics and land uses other than develop-
ment served by OLDS. The procedure is to monitor surface
water quality in basins (usually small) in which OLDS are
clearly the predominant influence on water quality other
than "background" pollutant loadings. Careful design of the
175
-------
TABLE 13
COMPARISON OF CHARACTERISTICS OF
RAW SEWAGE AND SEPTIC TANK EFFLUENT
Parameter
Suspended Solids
COD
Organic N
Total P
Bacteria
Concentration in mg/1
Medium Strength
Raw Sewage1
200
200
500
25
15
10
Septic Tank
Effluent
140 - 150(2)
100 - 120(3)
360 <3>
35<4)
10<4>
25<4>
12,000/ml
(5)
Sources:
(1) Metealf and Eddy, 1972
(2) Harkin, 1976
(3) Kreissel, 1976
(4) Dudley and Stephenson, 1973
(5) Harkin, 1976
176
-------
sampling program may allow estimation of relationships
between OLDS pollutant loadings and hydrologic conditions,
as well as relationships involving the density, location,
and characteristics of on-lot systems.
One example of such a study is a recent investigation of
OLDS impact on surface water quality in Chester County,
Pennsylvania (Howard and Hammer, 1973). The basins studied
in this case contained primarily housing units less than 30
years old, at various densities. Soils were moderately-
permeable and generally considered suitable for on-lot
disposal. The observed pattern of relationships between
pollutant concentrations and density of on-lot disposal
systems was the following.
Significant Relationships
with OLDS Observed
Nitrate
Total nitrogen
Chloride
Calcium
Magnesium
Sodium
Total dissolved solids
(3 measures)
Total coliform bacteria
Fecal streptococcus
bacteria
No Significant
Relationships Observed
Dissolved oxygen
COD
Organic nitrogen
Ammonia
Ortho phosphate
Total phosphate
Potassium
Bicarbonate
Iron
Fecal coliform bacteria
pH
The predictive equations obtained, which attributed very
large effects to OLDS in some cases, were generally consis-
tent with expectations. It is likely that greater effects
would have been observed for some constituents, such as
phosphate, if soil conditions in the study area had been
less favorable for OLDS. An important finding of the study
was that bacterial contamination of surface waters was
highly sensitive to the existence of OLDS within 100 feet of
streams and swales (the latter of which could largely be
considered perennial watercourses). Thus, in analytical and
planning activities it may be essential to consider the
location of OLDS with respect to surface waters, as well as
with respect to land features such as soils, geology, and
slope. Development of empirical relationships in this
fashion requires considerable effort, in terms of water
quality sampling and measurement of basin characteristics.
177
-------
(The latter must frequently be expressed in terms of inter-
action variables, i.e., number of OLDS located on land in
various categories.) Nevertheless, estimation of such
relationships may be worthwhile, given that the multiplicity
of OLDS tends to limit treatment of this factor on a site-
specific basis.
OLDS Effect on Groundwater Quality
The influence of OLDS on groundwater quality is dependent
primarily upon the following factors.
1. OLDS density in an area
2. Type, number, and degree of malfunctioning OLDS
3. Amount of rainfall and groundwater available for
dilution of effluent (i.e., groundwater flow,
quantity, and direction)
4. Hydraulic and chemical characteristics of soil,
subsoil, and aquifers
Figure 13 indicates that the greatest potential for ground-
water contamination from OLDS results from direct hydraulic
connections (gravel or solution channels) between- the drain
field or cesspool and groundwater. This type of problem
results in direct recharge of effluent, with little or no
renovation, to the groundwater. In an aquifer containing
numerous fractures, faults or solution channels, the efflu-
ent can travel with the groundwater for distances ranging
from less than one hundred feet to several miles. Wells
located a considerable distance away from the malfunctioning
OLDS, in the direction of groundwater flow, can therefore be
contaminated. This situation probably represents the most
severe groundwater contamination problem associated with
OLDS. Another potential problem is that drain fields which
have been installed too deep in the soil develop anaerobic
conditions and can contribute products of anaerobic decora-
position. These products include bacteria and viruses,
undegraded detergents, and partially decomposed organic
compounds which can contaminate groundwater and surface
water.
Perhaps the most difficult impacts to assess are those
produced by properly operating OLDS on groundwater quality.
A well-drained soil with an adequately designed and properly
178
-------
installed effluent disposal field is capable of providing an
adequate oxygen supply for reduction of BOD and COD (Miller
and Wolf, 1975). An important aspect of OLDS operation is
mineralization of organic nitrogen and ammonia nitrogen
(NH3) to the nitrate form (NO3) in the presence of oxygen.
If septic tank systems are situated in areas having well-
drained soils and relatively deep water tables, NO3 is the
nitrogen product of primary concern after renovation of
septic tank effluent by the soil (Holzer, 1975). Numerous
authors have addressed in detail the nitrogen transforma-
tions which take place in various components of OLDS (Hall,
1975; Holzer, 1975; Miller and Wolf, 1975; and Walker, et
al* 1973). —
Because NO3 is an anion and is extremely soluble in water,
the cation exchange capacity (CEC) of soil is not effective
in adsorbing NO3. Consequently, the NO3 moves with perco-
lating OLDS effluent to the groundwater. Indications are
that, although some denitrification (conversion of N03 to
N02 or nitrogen gas under anaerobic conditions) may occur in
fine-textured, well-structured subsoils, the NO3 concen-
tration (as N) of the percolate from a seepage bed generally
approximates the total N concentration of a septic tank
effluent (Sikora and Keeney, 1975).
Disposal of human wastes in a soil system can present a
potential public health hazard due to the admission of
pathogenic organisms into the groundwater supply. The
magnitude of the hazard depends on the presence, quantity
and accessibility of the pathogens. Prediction of this
problem is difficult since various physical, chemical and
microbiological properties of the soil have the ability to
render pathogenic organisms nontoxic. In general, the soil
presents a hostile environment to most pathogenic organisms.
(Pathogen survival times in soils have been summarized by
Miller and Wolf, 1975.)
Pathogen movement in soil systems by percolation has been
studied extensively and appears to be related to soil type
and hydraulic considerations (Miller and Wolf, 1975). Two
important factors are the size of the pathogen relative to
the pore sizes within the soil, and the adsorption processes
which are operative. Under favorable conditions, pathogen
movement through soils tends to be sufficiently restricted
that a minimal public health hazard is presented. However,
179
-------
as previously indicated, severe contamination can be pro-
duced by OLDS which have direct hydraulic connections (via
faults, fractures or solution channels) with the ground-
water.
Delineating OLDS Impact on Water Quality
The following discussion describes a mapping technique for
delineation of present and potential OLDS problems which
does not rely heavily upon water quality data. The pro-
cedure is outlined schematically in Figure 14. Since map-
ping is involved, a good quality base map of the study area
is required. U.S. Geological Survey quadrangle sheets (7-
1/2-minute series) are ordinarily suitable for this purpose.
Areas presently served by OLDS should be delineated on an
overlay of the base map. This information is generally
available from: municipal wastewater facilities planning
studies, municipal sewer system records, county health
department records, or state regulatory agency files and
records.
An important task is to delineate land areas where OLDS mal-
functions are likely to occur. Following an approach sug-
gested by Miller (1975), areas of probable OLDS malfunction
are designated "potential surface water contamination
areas;" and other land areas are designated "potential
groundwater contamination areas." (It is understood that
malfunctioning OLDS can also cause groundwater pollution,
and that properly operating systems can affect surface water
quality to some degree, as noted earlier.) Delineation of
these two categories of land is based upon soil character-
istics. The following characteristics are relevant:
drainage
depth to bedrock
slope
presence or absence of seasonally high groundwater
table
texture
presence of faults, fractures, or solution channels
in underlying geologic formations
The best sources of detailed soil information are the U.S.
Department of Agriculture (USDA) Soil Survey Reports. These
reports have been prepared for many counties throughout the
country. In many of the USDA reports, the above character-
istics have been considered for each soil series in order to
180
-------
Prepare Base Map of
Study Area
Delineate Areas Presently
Served by OLDS
Define Areas Having Probable Malfunctions
Using the Approach Described By Miller (1975)
Potential Surface
Water Contamination Areas
(Max. Water Table Depth <5 Feet)
Potential Groundwater
Contamination Areas
(Max. Water Table Depth >5 Feet)
Delineate Known or
Documented
Malfunctioning OLDS
Screen Potential
Surface Water
Contamination Areas
Screen Potential
Groundwater
Contamination Areas
Documented
Surface
Malfunctions
Documented
Water Well
Contamination
Delineate Soils
Having Severe
Limitations
For OLDS
(Use Soil Survey
if Available)
Delineate Areas
Having Probable
Groundwater
Contamination
(Use Dilution
Ratio Approach)
Rank Suspected
Problem Areas
Conducted Detailed Evaluation
And Monitoring in Priority Areas
Source: Betz Environmental Engineers, Inc.
Filire 14 SCHEMATIC DIAGRAM OF OLDS WATER QUALITY IMPACT [VALUATION
181
-------
estimate the degree of limitation for OLDS. This informa-
tion is usually presented in a table entitled "Major Soil
Properties and Estimated Degree of Limitation that Influence
Use of Soils for Community Developments" (USDA, 1967). Soil
limitations for OLDS are classified as "slight," "moderate,"
or "severe." When detailed Soil Survey data are available,
it is possible to proceed directly through several of the
steps shown in Figure 14 by mapping the soil series desig-
nated as having severe limitations for OLDS. (It may be
convenient simply to use the USDA soil maps as base maps,
rather than transferring soils information to other maps.)
When detailed USDA surveys are not available, other steps
are necessary to determine potential surface water con-
tamination areas. The hydrogeologic approach suggested by
Miller (1975) involves a distinction between areas having a
maximum groundwater table less than five feet below the land
surface, and areas in which the water table is always more
than five feet below the surface. Water table depth is a
good predictor of hydraulic failure of OLDS; areas in which
the maximum depth is less than five feet are therefore
selected as potential surface water contamination areas.
Groundwater contamination, primarily involving nitrate, is
most likely to occur where soils are permeable and/or well-
drained, in which case the water table is normally well
below the land surface. '
A number of sources of hydrogeologic information are avail-
able which would permit delineation of the five-foot ground-
water depth contour. These include:
U.S. Geologic Survey Hydrologic Investigations Atlas
Series
U.S. Geological Survey Professional Papers
State Geologist's records and reports
Having established regions of potential surface water and
groundwater contamination, further screening of problem
areas is conducted utilizing different approaches in the two
cases. Surface water problem areas are isolated through
further examination of soils information, including all of
the characteristics listed above. An additional factor
which is likely to be important, as noted elsewhere, is
hi11slope position. The distance of land to a stream or
swale is correlated with water table depth, and frequently
is associated with soil characteristics that are relevant to
182
-------
OLDS performance (due to the presence of alluvial soils near
streams). Thus, OLDS which are situated near the bottom of
hillslopes are relatively likely to affect surface waters
through both surface and subsurface flows. When attempting
to isolate surface water problem areas, the only available
soils information may pertain to soil associations rather
than soil series. Care must be taken in interpreting soil
association descriptions since these data are usually very
general, and in practice may not add significantly to the
information already conveyed by the five-foot groundwater
depth contour.
Analysis of groundwater problems due to OLDS in areas where
water table depth is not an issue should consider three pos-
sible causal factors: (1) overly rapid percolation, which
would allow unrenovated OLDS effluent to reach the saturated
zone; (2) improper location of wells; and (3) excessive
density of OLDS. The first two factors must often be ex-
amined on a site- specific basis, using geologic data and/ or
field inspection. The density question, involving either
existing or potential OLDS, can be examined using methods
based on the dilution ratio concept. The question is
whether renovated OLDS effluent is receiving adequate di-
lution in the groundwater. Dilution of the effluent by
uncontaminated groundwater is the principal means of de-
creasing the concentration of solutes, primarily NO3, in the
effluent.
The dilution ratio may be defined as the ratio of the
quantity of percolating effluent per unit area to the net
groundwater recharge per unit area. In equation form:
„ . . Q (effluent) _
Dilution Ratio =
(net recharge)
Net recharge is water reaching the surficial aquifer (the
same aquifer receiving OLDS effluent) which is available for
dilution of OLDS effluent. Net recharge is equal to infil-
tration minus groundwater evapotranspiration losses and
groundwater withdrawals via wells tapping the surficial
aquifer.
On the assumption that nitrate is the primary anion of
concern in properly renovated OLDS effluent, Holzer (1973)
183
-------
has concluded that net groundwater recharge must achieve at
least a one-to-one dilution of OLDS effluent, in order for
groundwater conditions to be acceptable.
"Based on the concentrations of nitrate measured by
Bouma et aJL, 1972, and Miller, 1972, a dilution of
renovated effluent of at least one-to-one may be re-
quired to reduce nitrate concentrations to the public
health standard of 45 mg/1 NO3 in permeable soils with
deep water tables." (Holzer, 1973).
The dilution ratio can be applied in two ways: (1) the net
aquifer recharge rate approach; and (2) the aquifer cross-
sectional approach. Only in the first case is it necessary
to determine the net aquifer recharge rate for an area of
land. This can be done using any technique which provides
an accurate water balance of the surficial aquifer. Factors
to consider include:
1. Total annual recharge from precipitation (usually
expressed in inches)
2. Total annual withdrawals via wells from the sur-
ficial aquifer
3. Groundwater evapotranspiration
An alternative is to determine net recharge for large areas
using surface water gaging records. (See the discussion of
infiltration devices in the previous section.) On an aver-
age basis, net groundwater recharge equals stream base flow,
plus imported water, minus exported water.
The second method of applying the dilution ratio—the aqui-
fer cross-sectional approach—is applied on a more detailed
basis than the net aquifer recharge rate apprpach. The
cross-sectional method assumes that any mixing of OLDS
effluent with groundwater takes place in the upper five feet
of the saturated zone (Dudley and Stephenson, 1973).
Groundwater flow through a selected cross-section of aquifer
may be computed using Darcy's Law:
Q = KIA
184
-------
where:
Q = flow through upper 5 feet of aquifer (gal/day)
K = aquifer permeability (gal/day/ft, sq.)
I = hydraulic gradient (ft/ft)
A = cross-sectional area of contaminated aquifer (ft.
sq.—based upon 5-foot depth and selected section
width)
The aquifer cross-sectional approach may be utilized to
estimate the impact of individual OLDS or subdivision OLDS
on groundwater quality at reasonably close distances (up to
1/4 mile).
Figure 15 relates the dilution ratio to both the net aquifer
recharge rate, and the flow rate through a cross-section of
surficial aquifer having a depth of five feet. Dilution
ratios greater than 1.0 indicate the potential of ground-
water nitrate contamination from OLDS. Derivation of the
population density curves in Figure 15 was based upon a
daily wastewater production of 50 gallons per capita-day.
Similar curves could be readily derived for any desired
wastewater production value.
From Figure 15, a population density of 3 persons per acre
would require a minimum of 150 gallons/day/acre of net
aquifer recharge, or 150 gallons per day of flow across the
section of aquifer which receives percolating OLDS effluent.
A net aquifer recharge rate of 150 gallons per day per acre
is equivalent to 2.0 inches of net annual aquifer recharge
from precipitation. Thus, if net recharge in a particular
portion of a study region is equal to 2 inches, 3 persons
per acre would be the upper limit population density which
is acceptable in the absence of sanitary sewerage, assuming
that groundwater is used for domestic water supply.
It is possible that the one-to-one dilution ratio suggested
by Holzman may not be sufficiently conservative. As indi-
cated in Table 13, the total nitrogen concentration of
septic tank effluent is approximately 45 mg/1. In general,
the total nitrate-nitrogen concentration of the percolate
from a seepage bed approaches the total N concentration of
the septic tank effluent (Sikora and Keeney, 1975). This
185
-------
00
(Tl
10
K>2
NET AQUIFER RECHARGE RATE^—— }
\Day-Acr*/
103 104
105
10«
0.01
10
102
FLOW RATE ACROSS UPPER 5 FEET OF AQUIFER-Gal/Day
Source: Betz Environmental Engineers, Inc.
Figure 15 DILUTION RATIO ANALYSIS
-------
translates to approximately 200 mg/1 as NO3. Thus, in order
to attain a nitrate concentration in groundwater of 45 mg/1,
as NO3, a dilution ratio as low as 0.29 could be required
(equal to 45/(200-45)) . Referring again to Figure 15, this
criterion would indicate that a population density of 3
persons per acre would require slightly more'than 500
gallons/day/acre of net aquifer recharge, or about 7 inches
on an annual basis.
A final step in mapping OLDS impact on water quality is to
delineate areas where OLDS malfunctions have been docu-
mented. Sources of information about malfunctioning OLDS
include:
1. State health or regulatory agency files
2. County health department records
3. Municipal health department or engineer's records
Nearly all documented malfunctions will probably be surface
malfunctions, which have been recorded because of complaints
about odor and unsightly conditions. In some cases, docu-
mentation of well water contamination from OLDS may be found
and should provide insight into groundwater flow patterns in
the area. Care must be taken to determine the cause of
malfunction before the defective OLDS is delineated on the
study area map. System failure due to misuse or lack of
maintenance does not necessarily reflect soil or geologic
conditions in the area. •
Control of OLDS Problems
The existing problem areas thus identified would be ranked
in importance, and subjected to detailed study if possible
through detailed monitoring activities. Remedial measures
would then be designed on the basis of the findings obtained.
These measures might include:
1. Conventional repairs to existing systems (e.g.,
moving absorption field).
2. Install alternate or experimental systems (e.g.,
aerobic treatment systems, elevated sand mound,
evapotranspiral beds).
187
-------
3. Renovate and maintain absorption area permeability
via peroxide treatments (see: Harkin, Jawson, and
Baker, 1975).
It is important to acknowledge that existing OLDS problems
are often legal and institutional in origin rather than
strictly technical. Government authority tends to be frag-
mented, with various roles assigned to city, county, re-
gional, and state agencies. Often there are no specific
criteria or standards which clearly define system failure,
and no specific programs to assure that systems are being
properly maintained. Agencies generally do not possess
adequate manpower to conduct the monitoring and enforcement
activities implied by such a program. Thus, perhaps the
most important step in correcting existing OLDS problems,
once identified, is to effect the necessary institutional
changes.
With regard to the location and design of new OLDS, it is
likely that ultimate determinations will continue to be made
on a site-specific basis, through the use of percolation
tests. Delineation of potential problem areas on a regional
basis as part of current studies is nevertheless valuable,
in order to integrate OLDS permitting activities with public
facility planning and land use planning. The important
question is how to deal with areas in which soils are gen-
erally unsuitable for OLDS use. The following options are
available in planning for such districts.
1. Discourage development through public improvements
planning. Highways and other facilities which
encourage urban growth can perhaps be planned on a
regional basis so that new development involving
OLDS is directed away from inherently unsuitable
areas.
2. Establish protection districts. Development can
be limited in undesirable areas by establishing
protection districts through such mechanisms as
zoning, conservation easements, and land purchase.
Areas unsuitable for OLDS might be given high
priority for open space uses.
3. Provide centralized sewage disposal. Within the
area which is unsuitable for OLDS use, public
sewerage can be provided to selected development
districts (and package treatment plants might be
188
-------
allowed in these districts). When public sewerage
is provided, a general reversal of strategy be-
comes appropriate: it is desirable to promote
high-density development in order to absorb as
much urban growth within the development district
as possible; whereas low density is generally de-
sirable when OLDS are utilized. One liability of
establishing development districts within areas
unsuitable for OLDS is that urban growth may tend
to sprawl beyond the boundaries of the district.
Thus, strict control of new OLDS permits becomes
especially crucial.
On the whole, the situation that must be avoided is the
circumstance which was typical in the past, wherein dwel-
lings with OLDS were allowed to sprawl haphazardly across
areas where soils were basically unsuitable, so that public
sewerage was eventually required as a remedial measure.
Sewer systems constructed in response to OLDS problems tend
to be much more expensive and inefficient than systems which
have been planned in an orderly fashion in anticipation of
growth.
In the granting of individual OLDS permits it may be im-
portant to augment standard procedures in a number of re-
spects. Percolation testing at multiple locations on a site
and/or times of year may be advisable. Overly rapid perco-
lation and overly slow percolation should be grounds for
denial of permits. Local geologic information may also be
consulted, for example, to avoid location of OLDS in areas
underlain by limestone where problems would be created by
rapid movement of unrenovated effluent through solution
channels. Steps should be taken to avoid location of OLDS
close to surface waters (e.g., within 100 feet), even if
percolation tests are positive. This would apply to small
headwater streams as well as major water bodies. Establish-
ment of protection districts along all perennial or nearly
perennial watercourses would be ideal; but important bene-
fits can be gained by considering this factor on an ad hoc
basis. In designing the type of system to be specified in a
given OLDS permit, the full range of alternatives should be
considered, including expensive designs such as elevated
sand mounds and dual absorption fields. The overall ob-
jective of OLDS permitting activities should be to shift as
much of the burden of water quality control to the design
and construction stage, as opposed to later monitoring,
surveillance, and remedial activities.
189
-------
Establishment and implementation of appropriate institu-
tional arrangements and regulatory programs for OLDS use
tends to be a difficult task, requiring a careful balancing
of objectives. On the one hand, in terms of water quality,
every OLDS may be a liability. In most urban regions there
are substantial water quality and health incentives to
minimize OLDS use as much as possible. On the other hand,
on-site sewage disposal may be essential for land develop-
ment of any kind in many instances. Denial of permits can
cause economic and social hardships for landowners and po-
tential residents. The overall posture adopted within a
municipality or other enforcement district should therefore
reflect a weighting of the following factors: (1) existing
problems due to OLDS; (2) variability of soils limitations
for OLDS use, i.e., opportunities for development at loca-
tions where limitations are not severe; (3) existing use of
groundwater and surface water; ;(4) probability that OLDS use
will create increasing problems in the future; (5) ability
of residents and potential residents to pay for elaborate
disposal systems; and (6) extent to which .OLDS use, as
opposed to joint sewage disposal, is actually necessary to
meet demands for housing and other facilities. Permitting
agencies should attempt to establish their role in positive
terms, assisting communities and individuals in developing
adequate solutions to waste disposal problems, rather than
appearing solely as an inhibiting influence oh community
development.
190
-------
SECTION 11
EVALUATION AND CONTROL OF EROSION
FROM CONSTRUCTION SITES
Introduction
Sediment produced by soil erosion is often considered to be
the most important single pollutant from nonpoint sources.
Sediment impairs water quality, chokes stream channels and
reservoirs with deposits, and adversely affects aquatic life
and the recreational value of water resources. In addition,
sediment eroded from topsoil often carries large amounts of
pollutants, such as nutrients, organic matter, pesticides,
and pathogens. Erosion of soil by water can be categorized
as follows:
Surface Erosion
sheet erosion
rill erosion*
Channel erosion
gully erosion
streambank erosion
The present section deals only with surface erosion, par-
ticularly soil loss from construction sites. The formula-
tion which is most commonly used for evaluating sediment
problems is the Universal Soil Loss Equation (USLE). As
shown below, the USLE is a very simple relationship involv-
ing a series of multiplicative factors which denote influ-
ences upon soil loss. Over the past two decades, consider-
able experience has been gained in the use of this formula-
tion, although the primary focus until recently has been
* Rill erosion is actually a form of minor channel ero-
sion. For purposes of analysis, rill erosion is usually
treated as sheet erosion.
191
-------
erosion from agricultural land rather than construction
sites. The advantages of the USLE as an analytical tool are
its extreme flexibility, the familiarity of many persons
with its use, and the fact that it relates directly to
control measures.
The development and application of the USLE have been dis-
cussed at length in numerous references, most notably the
comprehensive report on loading functions by Midwest Re-
search Institute (McElroy et al, 1975). The present dis-
cussion will therefore be limited to a brief treatment of
several conceptual issues, and presentation of a methodology
incorporating the USLE which has been developed by Chen
(1974).
The form of the Universal Soil Loss Equation is the follow-
ing:
SEDIMENT LOADING DUE
TO SURFACE EROSION = R K LS C P Sd
(TONS/ACRE/YEAR)
where:
R = the rainfall factor, expressing the erosion po-
tential of average annual rainfall.- (R is a sum-
mation across storms of the kinetic energy of
rainfall, in hundreds of foot-tons per acre, times
the maximum 30 minute rainfall intensity in inches
per hour);
K = the soil-erodibility factor, commonly expressed in
tons per acre per R unit;
L = the slope-length factor (dimensionless);
S = the slope-steepness factor (dimensionless);
C = the cover factor (dimensionless);
P = the erosion control practice factor (dimension-
less) ; and
Sd = the sediment delivery ratio.
192
-------
The rainfall factor, R, is expressed here in terms of
average annual conditions, but could relate to short-term
conditions or even a single design storm. Annual values of
R can also be selected which represent worst-case conditions
rather than average conditions (e.g., the worst year in ten
for erosion potential). Tabulations of R values applying to
specific localities are available in numerous publications
(McElroy, et al, 1975). K, the soil erodibility factor, is
a property of specific soil types and has been tabulated in
detail by the U.S. Soil Conservation Service. The factors L
and S, pertaining to slope length and gradient, are commonly
considered together. Tabulations are available for the
quantity LS as a function of hillslope characteristics.
The sediment delivery ratio, Sd, represents the proportion
of eroded soil which actually reaches a given point in the
surface water system. For example, a sediment delivery
ratio of 0.4 would mean that 40% of the soil lost from the
land area under study would reach the given stream point,
whereas the other 60% would settle out in stream channels
and other locations. In cases where the USLE is used to
predict sediment loadings from an entire basin, the value of
Sd is intended to describe sediment delivery from the catch-
ment as a whole. (Loadings are usually predicted in such
cases by measuring the other terms of the USLE for subdi-
visions of the catchment, summing the products, and then
applying the sediment delivery ratio.) Values of the sedi-
ment delivery ratio are highly dependent upon local condi-
tions, and therefore tend to be difficult to predict ac-
curately.
C and P, the cover factor and the erosion control practice
factor, are the two elements of the USLE which express human
influence on sediment yields. The appropriate values of C
and P for agricultural land uses have been a subject of
study for many years; but only recently have values been
developed which apply to construction sites under various
degrees of control. On the whole, the exercise of subjec-
tive judgment in applying the USLE centers largely around
the parameters Sd, C, and P, although considerable informa-
tion is now available for assistance (see below) .
Behavior of Sediment Loadings
Before discussing the use of the USLE for planning purposes,
some comments are offered regarding the general behavior of
sediment loadings. It should be recognized that sediment
193
-------
TA£LE 14
ANALYSIS OP ANNUAL SEDIMENT YIELD FOR
SMALL RURAL WATERSHEDS IN CHESTER COUNTY, PENNSYLVANIA
Stream Discharge, in
cfs Per Square Mile
of Watershed
Range Ave. Value
1 - 1.99 1.4
2 - 9.99 4
10 - 19.99 14
20+ 43
Annual Duration
of Discharge
within the Given
Range, in Days
104
37
1.3
0.9
Suspended Solids Loading
Ave . Concen-
tration in
mg/i
20
87
227
757
Total Load
in Tons/
Sq. Mi.
8
34
11
79
Total Annual Load:
132
Source: Hammer, 1973, p. 139 (based on data from Miller, Troxell and
Leopold, 1973)
. TABLE 15
VALUES OF SEDIMENT DELIVERY RATIO FOR
SMALL AGRICULTURAL BASINS IN TEXAS
Computed Value of the Sediment Delivery Ratio
(Dimensionless)
Year
1962
1964
1965
1966
1967
1968
1969
1970
Overall
SDR for
8 Years
Basin; Y2
0.23
0.09
1.26
0.85
0.06
1.24
0.16
0.24
0.67
HI
0.24
0.19
0.80
1.26
0.13
1.06
0.12
0.64
0.63
Y
0.24
0.06
0.86
1.09
0.07
1.22
0.33
0.54
0.66
D
0.16
0.11
0.84
1.20 -
0.09
0.34
0.38
0.43
0.48
G
0.18
0.06
0.45
0.65
0.12
0.73
0.38
0.44
0.42
Source: Williams fi Bernadt, 1972
194
-------
yields are very sensitive to hydrologic conditions and tend
to be extremely concentrated in time. The latter fact is
illustrated in Table 14, which is based upon suspended
solids data for a number of rural watersheds in Chester
County, Pennsylvania, ranging between 3 and 6 square miles
in area. The time in a typical year is subdivided into five
categories on the basis of stream discharge. (One category
is not shown since it involves negligible suspended solids
loadings.) For example, the second row indicates that
stream discharge is between 2 and 10 cfs per square mile for
a total elapsed time of 37 days per year; the average dis-
charge under these conditions is 4 cfs per square mile. As
shown in the fourth column, suspended solids concentrations
bear an extremely strong positive relationship with dis-
charge. The average concentration when discharge exceeds 20
cfs per square mile is nearly 40 times as great as the
average when discharge is between 1 and 2 cfs per square
mile.
The fifth column of Table 14 shows the total suspended
solids yield associated with each range of discharge during
a typical year. The total annual loading is 132 tons per
square mile. The most striking fact is that well over half
of this loading occurs during conditions which occupy only
about one day's time out of the year. (These conditions
account for slightly less than 10% of total annual runoff.)
In spite of the fact that rainfall in southeastern Pennsyl-
vania tends to be well-distributed throughout the year, it
is apparent that half a year's loadings of suspended solids
could occur during a single storm. This dominance of
extreme events, and the fact that suspended solids concen-
trations are imperfectly related to discharge, often makes
it difficult to compute suspended solids loadings accurately
even when field data are available.
Some additional data relating more directly to the USLE are
presented in Table 15. Five small agricultural basins in
Texas were studied by Williams and Berndt (1972) during the
1962-1970 period. Values of the sediment delivery ratio
were estimated for each basin in each year by dividing the
observed sediment load by the estimated soil loss, where the
latter was computed as the product of the factors R, K, LS,
C, and P. This procedure treats the sediment delivery ratio
as a "residual" term in the USLE—i.e., attributes any
errors in the equation to the sediment delivery ratio.
195
-------
The resulting estimates of the sediment delivery ratio for
the five basins are presented in Table 15. The variability
among the figures in each column is notable. For every
basin, the highest computed value of the sediment delivery
ratio over the eight years of record is at least ten times
as great as the lowest value. (No storm runoff occurred in
1963.) These data do not necessarily demonstrate that
sediment delivery per se varies over time; but indicate that
there may be a substantial degree of error in the USLE as a
whole. This circumstance is attributed by Williams and
Berndt to three factors: (1) the small number of storms in
each year; (2) variation in the grouping of storms (which
affects antecedent soil moisture, and hence relates to
runoff and erosion); and (3) a tendency of the USLE to
overpredict sediment for years when the rainfall index R is
low, and underpredict sediment for years with high values.
Such evidence suggests that sediment loadings are generally
difficult to predict accurately, regardless of the method-
ology utilized. An even more difficult problem may be to
estimate the effects of sediment on water use—for example,
the impacts on aquatic biota. These points are not intended
to discourage quantitative analysis of erosion and sedimen-
tation problems, but simply to suggest the limitations of
such analysis.
Quantitative Evaluation of Erosion Control Measures
The appropriate role of the USLE in planning studies depends
somewhat upon the manner in which control measures are
expected to apply—specifically, the extent to which con-
trols will be tailored to the circumstances encountered at
individual construction sites. At one extreme, the controls
which are implemented may be relatively uniform across the
study area, in that the physical measures required at con-
struction sites do not vary greatly with the characteristics
of the land or the receiving waters affected. At the other
extreme, the required control measures could be geared to
attainment of desired conditions in specific receiving water
bodies, and/or might be strongly related to the erosion
potential of land at the construction site. In the former
case, use of the USLE might be limited to sample demonstra-
tions of erosion magnitudes and the effectiveness of con-
trols; whereas in the latter case the USLE would be integral
to the planning process.
196
-------
A convenient methodology for linking erosion/sedimentation
controls to in-stream conditions and land characteristics
has been developed by Chen (1974) . This approach involves
partitioning the total sediment load into two components:
the loading from construction land, and the loading from
non-construction land. Each of these loadings is equal to
the per-acre soil loss from the given land type/ times the
acreage involved, times the overall sediment delivery ratio
for the catchment under consideration.
Chen presents a "macro model" which could be used to deter-
mine the allowable rate of soil loss from construction
lands, on the basis of allowable in-stream sediment concen-
trations. The macro model basically consists of a straight-
forward set of computations utilizing assumed values of soil
loss from non-construction lands and the sediment delivery
ratio. The latter could be estimated from sources such as
Figure 16, which relates the sediment delivery ratio to
drainage density and predominant soil type.
Chen's "micro model" for evaluation of erosion controls is
based upon the following equation:
Qc = R K LS C
where Qc is the soil loss from construction lands; and
C is the "control practice."
This is basically the USLE without the sediment delivery
ratio, and with the C and P terms combined into a single
control practice index. The nomograph developed by Chen for
evaluating this equation is reproduced here as Figure 17.
Working from left to right, the values of R, K, and LS are
utilized to determine the soil loss from the construction
site without control (axis 5). Relating this to the maximum
allowable rate of soil loss (from the macro model) yields
the value of the control practice index C which must be
achieved. The most important aspect of this approach is the
desaggregation of C into control "factors" "^.^associ-
ated with individual physical measures. The following
relationship is specified:
C = Cs Cr Ct Ce Co
197
-------
IO
CO
0.02
I/Drainage Density, Kilometers
1.0 10
Silty Cloy
Predominantly Silt
1.0 10
I/Drainage Density. Miles*7
Source: Midwest Research Institute, "Interim Report on Loading Functions for
Assessment of Water Pollution From Nonpolnt Sources," EPA, Washington, D.C., November 1975.
Drainage density equals total channel length (accumulated for all orders within a basin)
divided by the basin area.
Figiri 16 SEDIMENT DELIVERY RATIO FOR RELATIVELY HOMOGENEOUS BASINS
-------
<2)
(3)
(4)
(5)
(6)
(7)
4O-
90-
eo-
TO-j
eo-
IOO-
190-
.
.
200-
.
250-
-
"
300-
"
390 -I
4OO-
R
10 h
If
JS-
A-
x
x;v
\
N
-4
_ 5
-£
-7
~ •
-a
-10
-IS
-to
-so
-40
-to
- GO
-70
-SO
-SO
-100
^110^
7200 -
-25O
-306
-ISO
-4OO
-SCO
_
1c
i|
-'j
"•*
-.2
-.1
-.O$
-.04
"°1*CsO
-.02 "'"^-.^
-.01
-9O
-BO
-7O
-eo
-so
-40
-30
-23
-2O
•IS
-10
• 9
-T
-6
-S-r
-4
-3
- -s-r-Soil- Loss Limit
NOMENCLATURE
R - Roinfoll Erosion Factor p«r
Period (El,or Erotion
!nd«x Voluss
K - Soil Erodibility Factor
(Ton/Acre/EI Volu«)
LS- Slop*-Length Factor
C - Control Practice Factor
ae- Erosion Rate without Control
PracticetOI ) Ton/Acre /Period
5c- Erosion Rat* with Control
Practice (Ton/Acre /Period)
Source: Chen, C.N., 1974
Figure 17 NOMOGRAPH FOR ON-SITE EROSION CONTROL PLANNING
199
-------
where:
Cs = the control factor due to surface stabilizing or
protective treatments such as seeding, mulching,
and netting;
Cr = the control factor due to runoff-reduction prac-
tices such as diversion berms, interceptor dikes,
benches and terraces, sodded ditches, and level
spreaders;
Ct = the control factor due to sediment trapping
measures such as sediment basins or basins with
chemical flocculants;
Ce = the control factor due to restricting the spatial
and/or temporal exposure of the denuded site to
rainfall and runoff erosion; and
Co = control factors associated with other additional
practices.
Quantitative information on C factors has been developed
from available research and field data. Charts and graphs
presenting this information are contained in the reference
cited (Chen, 1974). A good example of the manner in which
various control practices can be combined to produce a
desired C value is provided by Chen, based on the situation
depicted in Figure 17/ for which C = 0.031. The measures
considered include surface stabilizing treatments, runoff
control measures, sediment traps, and limitations on the
extent of ground exposure.
"The denuded ground surface can be temporarily pro-
tected by properly tacked mulching. For instance, a
control factor, Cs, of 0.2 can be achieved by applying
roughly 1 ton per acre of straw, 7 tons per acre of
woodchips, or 120 tons per acre of crushed stone or
gravels (see Figure 7 in Chen, 1974). The erosion
process can be further checked by runoff-control
measures such as diversions across the slope. A single
diversion ditch cutting across the 300-foot length at
mid-slope, for example, provides a control factor, Cr,
of 0.7 (see Figure 8 in Chen, 1974). Additional pro-
tection can be achieved by trapping the sediments
200
-------
eroded from the slope. If a sediment basin of 20 acre-
inch capacity is available at the downstream end of the
50-acre site, the basin provides a control factor, Ct,
of 0.36 (see Figure 9 in Chen, 1974). The overall
control factor due to the combined practices of mulch-
ing, diversion and basin trapping is the product of Cs,
Cr and Ct, or 0.050. Since this overall C factor is
greater than the required value of 0.031, the system of
control measures does not constitute an acceptable
practice under the given control criterion.
"Further reduction of the erosion can be achieved by
establishing vegetative growth (such as annual rye-
grasses) by increasing the mulch rate, by increasing
the number of diversions, or by limiting the extent of
grading operations within the development site. For
example, a control factor, Ce, of 0.35 can be obtained
by proper scheduling of the grading operation such that
the overall exposed area is limited to 60 percent of
the site and the site is denuded from October to June,
avoiding the highly erosive months of July and August
(see Figure 10 in Chen, 1974). This additional control
practice with Ce = 0.35 brings the overall control
factor down to 0.018. This is considerably lower than
the required value of 0.031. If the scheduling prac-
tice is workable from the .management point of view,
then the overall control factor requirement of 0.031
can be satisfied without even considering the runoff-
diversion measure. This demonstrates the flexibility
of the tool presented herein in planning on-site
conservation practices."
This methodology might also be used in developing construc-
tion site controls based on pollutants other than sediment.
The most important case would be phosphorus, which is
closely associated with sediment loadings. For example,
suppose that study of eutrophication problems in a reservoir
indicates that phosphorus loadings must be controlled to 0.5
Ibs per acre per year from the catchment area of the reser-
voir. This area is undergoing extensive development, with
30% of the land in construction sites. Available data
indicate that the phosphorus yield due to soil loss from
non-construction land amounts to 0.2 pounds per acre per
year, which is not considered controllable in this particu-
lar case.
Suppose it is determined that the phosphorus-to-sediment
ratio is approximately 0.00005:1. To meet the phosphorus
201
-------
limit, the total sediment loading to the reservoir must then
be restricted to 5 tons per acre per year (equal to 0.5/2000
/O.00005). Assuming a sediment delivery ratio of 0.4, the
total allowable soil loss from land surfaces in the catch-
ment would be 12.5 tons per acre per year. Of this amount,
1.4 tons per acre per year would be accounted for by non-'
construction land (computed using the same phosphorus-to-
sediment ratio). Thus, the total allowable soil loss from
construction sites would be equal to: 11.1/0.3 = 37 tons
per acre per year. This maximum soil loss would be compared
with the expected soil loss without control, as determined
from Figure 17, in order to establish the control practice
C. Analysis of the controls necessary to achieve the given
value of C would then proceed as indicated above.
An alternative approach in applying the USLE might be to
consider soil loss only, not the sediment loadings delivered
to specific points in surface waters. The following argu-
ments would favor such an approach.
1. Soil loss may be a good overall statement of
erosion impact since all material eroded from
construction sites is potentially problematic,
whether or not it reaches a given receiving water
body.
2. Selection of a stream point as a basis for com-
puting allowable loadings may represent a somewhat
arbitrary decision; and quantitative description
of allowable in-stream conditions may be diffi-
cult, in view of the extreme variability of sedi-
ment concentrations.
3. Linkage of allowable soil loss to receiving
waters will increase the variability among areas
in the degree of control required.(C), and thus
may add to enforcement difficulties and the possi-
bility of challenge on equity grounds.
A reasonable approach may therefore be to select one or more
regional values of allowable soil loss (Qc) and to use these
as a basis for design of control measures. Sample computa-
tions of in-stream loadings would perhaps be conducted to
demonstrate that these values of Qc would provide adequate
overall protection of water quality. The control practice
index C would then be computed for individual construction
sites using the formula:
202
-------
c = QC/R
K LS
Within the region to which Qc applies, the variable factors
affecting control practice would be the terms in the denom-
inator, K and LS, pertaining to soil erodibility and slope
length and gradient.
A general characteristic of the USLE is that soil loss is
typically most sensitive to slope gradient. (The LS factor
varies roughly as the 0.5 power of slope length, and as the
1.5 power of slope gradient.) Differences in steepness of
slope can produce order-of-magnitude differences in computed
soil loss from small land areas. For example, a recent
watershed study in Philadelphia by Coughlin and Hammer
(1973) utilized the USLE to predict soil loss due to con-
struction, without control measures, for a large number of
5-acre grid cells. The computed values of soil loss are
tabulated here in Table 16. The degree of variability is
striking, in view of the fact that the soil erodibility
factor K ranged only between 0.28 and 0.43. The predicted
annual soil loss was less than 100 tons per acre in more
than 10% of the cases, but greater than 2,000 tons per acre
in 5% of the cases. If the approach mentioned above had
been utilized, and if the allowable soil loss had been set
at 30 tons per acre per year, the appropriate control prac-
tice index would have varied from 1.0 to less than 0.01.
Although this example is somewhat extreme, in that much of
the steep land with high soil loss potential would not be
considered suitable for building in any case, it is impor-
tant to note that direct application of the USLE can lead to
widely different levels of control for construction sites
within a given vicinity.
Control of Hydrographic Modification due to Construction
An impact of construction activity which is often not given
explicit consideration is the increase in runoff quantity
which occurs during the construction phase. Elimination of
vegetation, compaction of the soil, and exposure of subsoil
during construction may increase runoff very substantially,
in some cases by amounts which are as great or greater than
the effects of completed development. Thus, erosion/sedi-
mentation control plans should include measures which reduce
peak rates of runoff as well as sediment loads. This may
203
-------
involve design of sedimentation basins and other structures
so that detention storage capacity is adequate to control
peak discharge in moderately severe storms. The S.C.S.
materials referenced in Section 9 can be utilized for this
purpose.
TABLE 16
COMPUTED SOIL LOSS FROM 5-ACRE GRID CELLS
IN THE WISSAHICKON WATERSHED, PHILADELPHIA
Range of Soil Loss Number of Percent of
in tons/acre/year Grid Cells All Cells
Less than 50 14 1.4%
50 - 99 92 9.3
100 - 199 176 17.8
200 - 499 284 28.6
500 - 999 215 21.7
1000 - 1999 162 16.3
2000 or more 49 4.9
992 100.0%
Implementation of Erosion/Sedimentation Controls
Control of erosion problems due to construction is con-
sidered an extremely important aspect of water quality
planning; and there are a number of favorable circumstances
which should assist in implementation of these measures.
The problem is widely acknowledged, and is already subject
to control in a number of U.S. areas. The expense incurred
by builders in meeting erosion control regulations is usu-
ally small relative to total project cost. Although the
magnitude of the problem and the effectiveness of control
measures are often difficult to quantify accurately, regula-
tions are not highly vulnerable to challenge on technical
grounds, since there is not a question of "threshold"
effects. Even small sediment loadings are potentially im-
portant; thus it is difficult to argue that control in a
particular case is unjustifiable.
A major issue which has been touched upon earlier is the
extent to which the degree of control should vary among
204
-------
construction sites. If controls are based upon predeter-
mined values of allowable soil loss or stream impact, the
requirements are likely to vary widely in terms of the
physical control measures which builders must provide. This
situation could arise even if the factors which distinguish
between sites are not treated explicitly in official regu-
lations, since approval of specific erosion/sedimentation
plans by enforcement agencies might be based upon computa-
tions using the USLE. The potential problem is that control
programs which discriminate strongly among sites might be
considered inequitable.
An alternative approach is to apply erosion/sedimentation
controls more or less uniformly (e.g., to specify require-
ments in terms of the control index "C," rather than in
terms of soil loss) and to deal with the most problematic
areas through land use control. Restriction of development
on steep slopes is the most prominent example. An implicit
basis for this approach may be the view that, in cases where
highly elaborate countermeasures would be required to pre-
vent erosion during construction, adequate performance by
the developer would be very difficult to assure through
normal monitoring and enforcement procedures. Limitation of
development per se might be a safer approach, and could
yield a variety of benefits such as preservation of scenic
hillslopes. One problem is that, since known methods can
potentially prevent erosion from almost any buildable site,
land use restrictions might be difficult to justify on the
basis of erosion alone if challenged. Thus, land use con-
trols should be linked explicitly to other objectives in
addition to erosion control, and should preferably be sup-
ported by a comprehensive environmental inventory and
analysis.
205
-------
BIBLIOGRAPHY
Abt Associates, Inc. "Preventive Approaches to Urban Storm-
water Management" (Report forthcoming). Prepared for
U.S. Environmental Protection Agency by Abt Associates,
Cambridge, Massachusetts, 1976.
Ahl, T. "Effects of Man-induced and Natural Loading of
Phosphorus and Nitrogen on the Large Swedish Lakes."
Verhandlungen Internationale Vereinigung fuer Theo-
retische und Angewandte Limnologie, 19:1125-1132, 1975.
American Public Works Association. "National Characteriza-
tion, Impacts and Critical Evaluation of Stormwater
Discharges, Nonsewered Urban Runoff and Combined Sewer
Outflows, (Final Report Draft)." Prepared for the U.S.
EPA, Washington, D.C., August 1975.
American Public Works Association. "Water Pollution Aspects
of Urban Runoff." Prepared for Federal Water Quality
Control Administration, 1969.
Amy, G., et al. "Water Quality Management. Planning for
Urban Runoff." Prepared for U.S. Environmental Pro-
tection Agency by URS Research Company, EPA 440/9-75-
004, NTIS PB 241 689, 1974.
Anderson, D. G. "Effects of Urban Development on Floods in
Northern Virginia." U.S. Geological Survey Open File
Report, 1968.
Andersen, D. R. "Water Quality Models for Urban and
Suburban Areas." Nebraska Water Resources Research
Institute, University of Nebraska, Lincoln, Nebraska,
1974.
"Applications of Stormwater Management Models." Handout
at EPA seminar at University of Massachusetts, Amherst,
July 28-August 1, 1975.
Aron, G., et al. "A Method for Integrating Surface and
Ground Water Use in Humid Regions." Pennsylvania State
University, Institute for Research on Land and Water
Resources, Research Publication No. 76, University
Park, Pennsylvania, 1975.
206
-------
AVCO Economic Systems, Inc. "Storm Water Pollution from
Urban Land Activity." Prepared for U.S. Department of
the Interior by AVCO Economic Systems, Inc., Washing-
ton, D.C., 1970.
Bansal, M. K. "Deoxygenation in Natural Streams." Water
Resource Bulletin, Vol. 11, No. 3, pp. 491-501, 1975.
Battelle Columbus Labs. "Development of the Arizona
Environmental and Economic Trade-off Model." Prepared
for the state of Arizona Department of Economic Plan-
ning and Development, Columbus, Ohio, March 31, 1973.
Beck, Alan M. "The Ecology of Stray Dogs, a Study of Free
Ranging Urban Animals." York Press, Baltimore, Mary-
land, 1973.
Benjes, H. H. , et al. "Storm-Water Overflows from Cpmbined
Sewers." Journal of the Water Pollution Control
Federation, Vol. 33, No. 12, pp. 1251-1259, 1961.
Benzie, W. J., and Courphaine, R. J. "Discharges from
Separate Storm Sewers and Combined Sewers." Journal of
the Water Pollution Control Federation, Vol. 38, No. 3,
pp. 410-421, 1966.
Berger, Lewis and Associates. "Section 303(e) Water Quality
Management Basin Plan, Northeast New Jersey Urban
Area." Prepared for New Jersey Department of Environ-
mental Protection, 1975.
Betson, R. P., and McMaster, W. M. "Non-point Source
Mineral Water Quality Model." Journal of the Water
Pollution Control Federation, Vol. 47, No. 10, pp.
2461-2473, 1975.
Biesecker, James E., and Liefeste, D. K. "Water Quality of
Hydrologic Benchmarks: An Indicator of Water Quality
in the Natural Environment." USGS Circular 460-E,
1975.
Biggar, J. W. , and Corey, R. B. "Agricultural Drainage
of Eutrophication." Eutrophication; Causes, Conse-
quences, Correctives.National Academy of Sciences,
Washington, D.C., 1969.
207
-------
Black, Crow and Eidsness, Inc. "Storm and Combined Sewer
Pollution Sources and Abatement." U. S. Environmental
Protection Agency, NTIS PB 201 725, 1971.
Blackman, W. C., Jr., et al. "Mineral Pollution in the
Colorado River BasinT^ Journal of the Water Pollution
Control Federation, Vol. 45, No. 7, pp. 1517-1557,
1973.
Blackwood, K. R. "Runoff Water Quality of Three Tucson
Watersheds." U.S. Environmental Protection Agency,
NTIS PB 240 287, 1974.
Bowman, H. R., Conway, J. G. and Asaro, F. "Atmospheric
Lead and Bromine Concentrations in Berkeley, California,
1963-1970." Environmental Science and Technology, Vol.
6, No. 6, pp. 556-560, 1972.
Branch, Melville C., City Planning and Aerial Information.
Harvard University Press, Cambridge, Massachusetts,
1971.
Brandes, Charles E. "Methods of Synthesis for Ecological
Planning." Master's Thesis, University of Pennsylvania,
Philadelphia, 1973.
Brandstetter, A. "Comparative Analysis of Urban Stormwater
Models." Pacific Northwest Laboratories, Battelle
Memorial Institute, Richland, Washington, 1974.
Brown, H. E. "A System for Measuring Total Sediment Yield
from Small Watersheds." Water Resources Research, Vol.
6, pp. 818-826, 1970.
Brown, H. J., et al. "Empirical Models of Urban Land Uses:
Suggestions on Research Objectives and Organization."
Columbia University Press, New York, New York, 1972.
Brown, J. C., Shaw, C. M, and Read, N. P. "Nutrients and
Suspended Sediments for Forested Watersheds in the
East-Central Sierra Nevada." University of Nevada,
Reno, Nevada, n.d.
208
-------
Brown, J. C., Skau, C. M. and Howe, W. R. "Nutrient and
Sediment Production from Forested Watersheds." Paper
No. 73-201, Presented at the Annual Meeting of the
American Society of Agricultural Engineers at Lexing-
ton, Kentucky, June 17-20, 1973.
Brown, R. , et al. "Empirical Models of Urban Land Use:
Suggestions on Research Objectives and Organization."
Columbia University Press, New York, New York, 1972.
Brusven, M. A. and Phathen, K. V. "Influence of Stream
Sediments on Distribution of Macrobenthos." Journal of
Entomological Society of British Columbia (Canada),
Vol. 71, pp. 25-32, October 1974.
Bryan, E.H. "Concentrations of Lead in Urban Storm Water."
Journal of the Water Pollution Control Federation, Vol.
46, No. 11, pp. 2419-2421, 1974.
Bryan, E. H., "Quality of Stormwater Drainage from Urban
Land Areas in North Carolina." Water Resources Re-
search Institute of North Carolina, Raleigh, North
Carolina, 1970.
Burm, R. J., Krawczyk, D. F. and Harlow, G. L. "Chemical
and Physical Comparison of Combined and Separate Sewer
Discharges." Journal of the Water Pollution Control
Federation, Vol. 40, No. 1, pp. 112-126, 1968.
Burm, R. J. and Vaughan, R. D. "Bacteriological Comparison
Between Combined and Separate Sewer Discharges in
Southeastern Michigan." Journal of the Water Pollution
Control Federation, Vol. 38, No. 3, pp. 400-409, 1966.
Cahill, T. H., Imperato, P. and Verhoff, F. H. "Evaluation
of Phosphorus Dynamics in a Watershed." Journal of the
Environmental Engineering Division, Proceedings of the
American Society of Civil Engineers, Vol. 100, EE2, pp.
439-458, 1974.
Cairns, J., and Dickson, K. L., eds. "Biological Methods
for the Assessment of Water Quality." American Society
for Testing and Materials, Philadelphia, Pennsylvania,
1973.
209
-------
Carey, G. H., e_t a^. "Urbanization, Water Pollution,
and Public Policy." Center for Urban Policy Research,
Rutgers University, New Brunswick, New Jersey, 1972.
Catanese, A. J. "Scientific Methods of Urban Analysis."
University of Illinois Press, Chicago, Illinois. 1972.
Cherkauer, D. S. "Urbanization Impact on Water Quality
During a Flood in Small Watersheds." Water Resources
Bulletin, Vol. 11, No. 5, pp. 987-998, 1975.
Chen, C. N. "Evaluation and Control of Soil Erosion in
Urbanizing Watersheds." Proceedings of the National
Symposium on Urban Rainfall and Runoff and Sediment
Control, University of Kentucky, Lexington, Kentucky,
1974.
Chow, T. J. and Earl, J. L. "Lead Aerosols in the Atmos-
phere, Incremental Concentrations" (Report). Science,
Vol. 169, p. 577, 1970.
Chun, M. J., Young, R. H. F. and Anderson, G. K. "Waste-
water Effluents and Surface Runoff Quality." Water
Resources Research Center, Technical Report No. 63,
Honolulu, Hawaii, 1972.
Clark, L. J., Guide, V. and Pheiffer, T. H. "Nutrient
Transport and Accountability in the Lower Susquehanna
River Basin - Summary and Conclusions." U.S. Environ-
mental Protection Agency, Region III, Annapolis Field
Office Technical Report No. 60, EPA 903/9-74-014, 1974.
Cleveland, J. G., et al. "Storm Water Pollution from Urban
Land Activity T11" Combined Sewer Overflow Abatement
Technology, Water Pollution Control Research Series,
U.S. Environmental Protection Agency, 1970.
Cleveland, J. A., Reid, G. W. and Harp, J. F. "Evaluation
of Dispersed Pollutional Loads from Urban Areas." NTIS
PB 263 746, n.d.
Colston, N. V., Jr. "Characterization and Treatment of
Urban Land Runoff." Prepared for U.S. Environmental
Protection Agency, EPA 670/2-74-096, 1974.
210
-------
Colston, N. V., Jr. "Pollution from Urban Land Runoff."
University of North Carolina Water Resources Research
Institute at North Carolina state University, Durham,
North Carolina, 1974.
Colston, Newton V., Jr. and Tafuni, Anthony N. "Urban Land
Runoff Considerations." Urbanization and Water Quality
Control, W. Whipple, Jr., ed., American Water Resources
Association, Minneapolis, Minnesota, 1975.
Commonwealth of Pennsylvania, Department of Environmental
Resources. "Technical Manual for Sewage Enforcement
Officers." Harrisburg, Pennsylvania, 1974.
Corey, G. H. , e_t al. "Urbanization, Water Pollution, and
Public Policy." Center for Urban Policy Research,
Rutgers University, New Brunswick, New Jersey, 1972.
Coughlin, Robert E., Berry, David and Hammer, Thomas R.
"Environmental Study of the Poquessing Watershed."
Regional Science Research Institute, Philadelphia,
Pennsylvania, 1976.
Coughlin, R. E. and Hammer, T. R. "Environmental Study of
the Wissahickon Watershed within the City of Phila-
delphia." Regional Science Research Institute,
Philadelphia, Pennsylvania, 1973.
Coughlin, Robert E. and Hammer, Thomas R. "Stream Quality
Preservation Through Urban Development." Prepared for
U.S. Environmental Protection Agency by the Regional
Science Research Institute, Philadelphia, Pennsylvania,
EPA-R5-73-019, 1973.
Cowan, W. F. and Lee, G. F. "Leaves as a Source of Phos-
phorus." Environmental Science and Technology, Vol.
7, No. 9, p. 853, 1973.
Crecelius, E. A. and Piper, D. Z. "Particulate Lead Con-
tamination Recorded in Sedimentary Cores from Lake
Washington, Seattle." Environmental Science and
Technology, Vol. 7, pp. 1053-1067, 1973.
Crim, R. L. and Lovelace, N. L. "Auto-Qual Modeling Sys-
tem." U.S. Environmental Protection Agency, Region
III, Annapolis Field Office Technical Report #54, 1973.
211
-------
Dairies, R. H., Motto, H., and Chitko, D. M. "Atmospheric
Lead: Its Relationship to Traffic Volume and Proximity
to Highways." Environmental Science and Technology,
Vol. 4, p. 318, 1970.
de Gueare, T. V. and Ongerth, J. E. "Empirical Analysis
of Commercial Solid Waste Generation." Journal of the
Sanitary Engineering Division, Proceedings of the
American Society of Civil Engineers, Vol. 97, SA6, pp.
843-850, 1971.
Digiano, F. A. and Coler, R. A. "Definition of Procedures
for Study of River Pollution by Non-point Urban Sources,
U.S. Environmental Protection Agency, NTIS PB 237 972,
1974.
Dillion, P. J. and Kirchner, W. B. "The Effects of Geology
and Land Use on the Export of Phosphorus from Water-
sheds." Water Research (Great Britain), Vol. 9, pp.
135-148, 1975.
Dow Chemical Company. "An Economic Analysis of Erosion and
Sediment Control Methods for Watersheds Undergoing
Urbanization." Prepared for the U.S. Department of
the Interior, 1972.
Dudley, John G. and Stephenson, D. A. "Nutrient Enrichment
of Ground Water from Septic Tank Disposal Systems."
Upper Great Lakes Regional Commission, 1973.
Dugan, G. L. and McGaughey, P. H. "Enrichment of Surface
Waters." Journal of the Water Pollution Control
Federation, Vol. 46, No. 10, pp. 2261-2280, 1974.
Dunbar, D. D. and Henry, J. G. F. "Pollution Control
Measures for Stormwaters and Combined Sewer Overflows."
Journal of the Water Pollution Control Federation, Vol.
38, No. 1, pp. 9-26, 1966.
Durbin, Timothy, Jr. "Digial Simulation of the Effects of
Urbanization on Runoff in the Upper Santa Ana Valley,
California." U.S. Department of the Interior, Geo-
logical Survey Water-Resources Investigations 41-73,
1974.
212
-------
Ecology and the Economy...A Concept for Balancing Long
Range Goals." Urban and Rural Lands Committee. Pacific
Northwest River Basin Commission, November, 1973.
Edwards, D. "Some Effects of Siltation Upon Aquatic
Macrophyte Vegetation in Rivers." Hydrobiologia, Vol.
34, No. 1, pp. 29-37, 1969.
Elfers, K. and Hufachmidt, M. M. "Open Space and Urban
Water Management Phase 1: Goals and Criteria."
University of North Carolina, Water Resources Research
Institute, Report No. 104, Chapel Hill, North Carolina,
1975.
Elgmork, K., et al. "Polluted Snows in Southern Norway
During the Period 1968-1971." Environmental Pollution,
Vol. 4, No. 1, p. 41, 1973.
Emery, R. M. , Moon, C. E. and Welch, E. B. "Enriching
Effects of Urban Runoff on the Productivity of a
Mesotrophic Lake." Water Research (Great Britain),
Vol. 7, pp. 1506-1516, 1973.
Engineering-Science, Inc. "Comparative Costs of Erosion
and Sediment Control Construction Activities." Pre-
pared for U.S. Environmental Protection Agency, EPA
430/9-73-016, 1973.
Engman, E. T. "Partial Area Hydrology and its Application
to Water Resources." Water Resources Bulletin, Vol.
10, No. 3. pp. 512-521, 1974.
Engman, E. T. and Ragowski, A. S. "A Partial Area Model
for Storm Flow Synthesis." Water Resources Research,
Vol. 10, No. 3, pp. 464-472, 1974.
"Environmental Management for the Metropolitan Area -
Part II: Urban Drainage." U.S. Army Corps. Seattle
District, 1974.
"EPA Prepares Effluent Guidance for 21 Industries for
Permit Program." Environment Reporter, Vol. 3, No. 37,
pp. 1053-1057, 1973.
213
-------
Espey, W. H. Jr. , and Winslow, D. E. "Urban Flood Fre-
quency Characteristics." Journal of the Hydraulics
Division, American Society of Civil Engineers, Vol.
100, HY2, pp. 279-293, 1974.
"Eutrophication: Causes, Consequences, Correctives."
Proceedings of a Symposium, National Academy of Sci-
ences, Washington, D. C., 1969.
"Evaluation of the Effects of Urbanization on Aquatic
Ecology and Hydrologic Regimes." Hydrocomp, Inc., PB-
247 095/3WP, Palo Alto, California, 1975.
Field, R. "Coping with Urban Runoff in the United States."
Water Research (Great Britain), Vol. 9, pp. 499-511,
1975.
Field, R. "Urban Pollution and Associated Effects of
Street Salting." Edison Water Quality Research Lab,
National Environmental Research Center, Cincinnati,
Ohio, Environmental Protection Agency, Edison, New
Jersey, 1972.
Field, R. and Knowles, D. "Urban Runoff and Combined
Sewer Overflow." Journal of the Water Pollution
Control Federation, Vol. 47, No. 6, pp. 1352-1369,
1975.
Field, R. and Lager, J. A. "Urban Runoff Pollution Control,
State-of-the-Art." Journal of the Environmental
Engineering Division, Proceedings of the American
Society of Civil Engineers, Vol. 101, EEl, pp. 107-125,
1975.
Field, R. and Wiezel, P. "Urban Runoff and Combined Sewer
Overflow." Journal of the Water Pollution Control
Federation, Vol. 45, No. 6, pp. 1108-1115, 1973.
Fellos, John and Molof, Alan H. "Effect of Benthal Deposits
on Oxygen and Nutrient Economy of Flowing Waters."
Journal of the Water Pollution Control Federation,
Vol. 44, No. 4, pp. 644-662, 1972.
214
-------
Fillos, John and Swanson, William R. "The Release Rate of
Nutrients from River and Lake Sediments." Journal of
the Water Pollution Control Federation, Vol. 47, No. 5,
pp. 1032-1042, 1975.
Floyd, C. F. and Rowan, M. J. "Implications of Zoning as
an Urban Water Management Measure." Department of Real
Estate, University of Georgia and Environmental Re-
sources Center, Georgia Institute of Technology, 1976.
Foehrenbach, J. "Eutrophication." Journal of the Water
Pollution Control Federation, Vol. 45, No. 6, pp. 1237-
1244, 1973.
Franklin Institute Research Laboratories. "Selected Urban
Storm Water Runoff Abstracts." U.S. Environmental
Protection Agency, 11024 EJC, 1970.
Frey, J. C., Gamble, H. B. and Sauerlender, O. H. "Eco-
nomics of Water Supply Planning and Management."
Institute for Research on Land and Water Resources
Publication No. 90, Penn State University, University
Park, Pennsylvania, 1975.
Fruh, G. E. "The Overall Picture of Eutrophication."
Journal of the Water Pollution Control Federation, Vol.
39, No. 9, pp. 1449-1463, 1967.
Gambell, A. W. "Sulfate and Nitrate Content of Precipita-
tion Over Parts of North Carolina and Virginia."
U.S. Geological Survey Professional Paper 475C, C209,
1963.
Gannett, Fleming, Corddry and Carpenter, Inc. "Storm Water
Management Alternatives." From final report on Watts
Branch Storm Water Management Study to Montgomery
County, Maryland Department of Environmental Planning,
1975.
Gaufin, A. R. "Water Quality Requirements of Aquatic
Insects." U.S. Environmental Protection Agency, EPA
660/3-74-004, 1973.
Gburek, W. J. and Brogan, J. G. "A Natural Non-Point
Phosphate Input to Small Streams." Northeast Watershed
Research Center, University Park, Pennsylvania, n.d.
215
-------
Gizzard, T. J. and Jennelle, E. M. "Will Wastewater
Treatment Stop Eutrophication of Impoundments?"
Presented at the 27th Purdue Industrial Waste Con-
ference, West Lafayette, Indiana, 1972.
Graham, F. H., Costello, L. S. and Mallon, H. J. "Estima-
tion of Imperviousness and Specific Curb Length for
Forecasting Stormwater Quality and Quantity." Journal
of the Water Pollution Control Federation, Vol. 46, No.
4, pp. 717-725, 1974.
Grossman, D., Hudson, J. F. and Marks, D. H. "Waste
Generation Models for Solid Waste Collection." Journal
of the Environmental Engineering Division, Proceedings
of the American Society of Civil Engineers, Vol. 100,
EE6, pp. 1219-1230, 1974.
Grava, Sigurd. Urban Planning Aspects of Water Pollution
Control. Columbia University Press, New York, New
York, 1969.
Guy, H. P. "Research Needs Regarding Sediment and Urbani-
zation." Journal of the Hydraulics Division, Proceed-
ings of the American Society of Civil Engineers, Vol.
93, HY6, pp, 247-254, 1967.
Guy, H. P. "Residential Construction and Sedimentation at
Kensington, Maryland." Federal Inter-Agency Sedimenta-
tion Conference, AR5 Miscellaneous Publication 970,
.1965.
Guy,.H. P. and Ferguson, G. E. "Sediment in Small Reser-
voirs Due to Urbanization." Journal of the Hydraulics
Division, Proceedings of the American Society of Civil
Engineers, Vol. 88, No. HY2, pp. 27-37, March 2962.
Hagarman, James. Personal Conversation in Philadelphia,
Pennsylvania, April 29, 1976. ,
Haith, D. A. "Land Use and Water Quality in New York
Rivers." Journal of the Environmental Engineering
Division, Proceedings of the American Society of Civil
Engineers, Vol. 102, EEl, pp. 1-15, 1976.
216
-------
Hammer, Thomas R. "Stream Channel Enlargement due to
Urbanization." Water Resources Research, Vol. 9, No.
6, 1973.
Hammer, T. R. "Effects of Urbanization on Stream Channels
and Stream Flow." Regional Science Research Institute,
Philadelphia, Pennsylvania, 1973a.
Hammer, T. R. "Water Quality Determination in a Suburbaniz-
ing Basin: Brandywine Creek, Pennsylvania." Regional
Science Research Institute Discussion Paper No. 78,
Philadelphia, 1974.
Hanes, R. E., Zelazny, L. W., and Blaser, R. E. "Effects
of De-icing Sal,ts on Roadside Plants and Water Sup-
plies." Department of Agronomy, Virginia Polytechnic
Institute, Blackburg, Virginia, 1967.
Harbridge House. "Key Land Use Issues Facing EPA." Pre-
pared for U.S. Environmental Protection Agency, NTIS PB,
pp. 235-345, 1974.
Harkin, John M., Jawson, M. D., and Baker, F. G. "Cause
and Remedy of Failure of Septic Tank Seepage Systems."
Proceedings, Second National Conference on Individual
Onsite Wastewater Systems, National Sanitation Founda-
tion, Ann Arbor, Michigan, pp. 119-124, 1976.
Harms, L. L. and Southerland, E. V. "A Case Study in Non-
Point Source Pollution in Virginia." Virginia Water
Resources Research Center Bulletin No. 88, Virginia
Polytechnic Institute, Blacksburg, Virginia, 1975.
Hartt, J. P. "A Study of Pollution Loadings from Urban
Runoff." Water Pollution Research in Canada, Vol. 8,
pp. 16-25, 1973.
Hawkes, H. A. "Water Quality: Biological Considerations."
Chemistry and Industry, pp. 990-1000, December 21,
1974.
Hawkins, R. H. and Judd, J. H. "Water Pollution as Affected
by Street Salting." Water Resources Bulletin, Vol. 8,
No. 6, pp. 1246-1252, 1972.
217
-------
Heaney, J. P. and Sullivan, R. H. "Source Control of
Urban Water Pollution." Journal Water Pollution
Control Federation, Vol. 43, No. 4, pp. 571-579, 1971.
» ' :
Heaney, J. P., e_t al. "Urban Stormwater Management Model-
ing and Decision-Making." Prepared for U.S. Environ-
mental Protection Agency, National Environmental
Research Center, NTIS PB 242 290, 1975.
Heeps, D. P. and Mein, R. G. "Independent Comparison of
3 Urban Runoff Models." Journal of the Hydraulics
Division, Proceedings of the American Society of Civil
Engineers, Vol. 100, HY7, pp. 995-1009, 1974.
Heerdegen, R. G. and Reich, B. M. "Unit Hydrographs for
Catchments of Different Sizes and Dissimilar Regions."
Institute for Research on Land and Water Resources,
Reprint Series No. 44, Pennsylvania State University,
University Park, Pennsylvania, 1974.
Helly, Walter. Urban Systems Models. Academic Press,
New York, New York, 1975.
Henningson, Durham and Richardson, Inc. "Combined Sewer
Overflow Abatement Plan." Draft Report for the U.S.
Environmental Protection Agency, 11024 PEG, Des Moines,
Iowa, 1973.
Hergert, S. L. "Urban Runoff Quality and Modeling Study."
Prepared for U.S. Environmental Protection Agency, NTIS
PB, 237 141, 1972.
Hiemstra, L. A. V. "Joint Probabilities in the Rainfall-
Runoff Relation." Institute for Research on Land and
Water Resources, Reprint Series No. 1.4, Pennsylvania
State University, University Park, -Pennsylvania, 1969.
Hill, D. E. and Thomas, H. F. "Use of Natural Resources
Data in Land and Water Planning." The Connecticut
Agricultural Experiment Station, Bulletin 733, New
Haven, Connecticut, 1972.
Hoak, R. D. "Physical and Chemical Behavior of Suspended
Solids." Sewage and Industrial Wastes, Vol. 31, No.
12, pp. 1401-1408, 1959.
218
-------
Hobble, J. E. and Likens, G. E. "The Output of Phosphorus
Dissolved Organic Carbon, and Fine Particulate Carbon
from Hubbard Brook Watershed." Limnology and Ocean-
ography, Vol. 18, No. 5, pp. 734-742, 1973.
Halsworth, E. G. and Adams, W. A. "The Heavy Metals Content
of Rainfall in the East Midlands." Environmental
Pollution (Great Britain), Vol. 4, p. 231, 1973.
Holzer, Thomas L. "Limits to Growth and Septic Tanks."
Paper Presented at Conference on Rural Environmental
Engineering, Warren, Vermont, 1973.
Horbeck, J. W. "Storm Flow from Hardwood-Forested and
Cleared Watersheds in New Hampshire." Water Resources
Research, Vol. 9, No. 2, 1973.
Horton, J. P. "Street Cleaning Effectiveness: Vacuum
Sweepers." The APWA Reporter, pp. 20-22, April 1976.
Hossain, A., Delleur, J. W. and Rao, R. A. "Evaporation
Infiltration and Rainfall-Runoff Processes in Urban
Watersheds." Water Resources Research Center, Tech-
nical Report No. 41, Purdue University, West Lafayette,
Indiana, 1974.
Howard, W. T. and Hammer, T. R. "Water Quality Impacts of
Unsewered Housing." Regional Science Research Insti-
tute Discussion Paper No. 66, Philadelphia, Pennsyl-
vania, 1973.
Howells, D. H. "Water Quality Dimensions of Water Resources
Planning." Journal of the Hydraulics Division, Pro-
ceedings of the American Society of Civil Engineers,
Vol. 101, HY2, pp. 277-284, 1975.
Huber, Wayne C., et al. "Storm Water Management Model
User's Manual"- Version II." Prepared for U.S. Environ-
mental Protection Agency, 1975.
Huff D D., et al. "Simulation of Urban Runoff Nutrient
Loading,~ahd~Biotic Response of a Shallow Eutrophic
Lake." Institute for Environmental Studies, University
of Wisconsin, Madison, Wisconsin, 1974.
219
-------
Hwang, C. P., Huang, P. M. and Lackie, T. H. "Phosphorus
Distribution in Blackstrip Lake Sediments." Journal of
the Water Pollution Control Federation, Vol. 47, No. 5,
pp. 1081-1085, 1975.
Hydrologic Engineering Center, U.S. Army Corps of Engineers.
Urban Storm Water Runoff Model Storm Computer Program
Users Guide, 1975.
Hydrologic Engineering Center, U. S. Army Corps of Engineers,
"FY 1972 Annual Report on the Quality of Urban Storm
Runoff Entering the San Francisco Bay." 1972.
Hydroscience in U.S. EPA. Areawide Assessment Procedures
Manual, report forthcoming.
Hynes, H. B. N. The Ecology of^ Running Waters. University
of Toronto Press, Toronto, Ontario, 1970.
International Business Machines, Inc. "IBM Scientific
Computing Symposium, Land and Air Resource Management."
White Plains, New York, 1968.
Interstate Sanitation Commission. "Combined Sewer Overflow
Study for the Hudson River Conference." New York, New
York, 1972.
Jaworski, N. A. and Hetling, L. J. "Relative Contributions
of Nutrients to the Potomac River Basin from Various
Sources." U.S. Department of the Interior, Federal
Water Pollution Control Administration, Chesapeake
Technical Support Laboratory, Technical Report No. 31,
1970.
Johnson, R. E., Rossano, A. T. Jr. and Sylvester, R. O.
"Dustfall as a Source of Water Quality Impairment."
Journal of the Sanitary Engineering Division, Pro-
ceedings of the American Society of Civil Engineers,
Vol. 92, SA1, pp. 245-268, 1966.
Jordan, R. A. and Bender, M. E. "An in situ Evaluation of
Nutrient Effects in Lakes." Prepared for U.S. Environ-
mental Protection Agency by Virginia Institute of
Marine Science, Gloucester Point, Virginia, EPA-R3-73-
018, 1973.
Judd, J. H. "Lake Stratification Caused by Runoff from
Street De-icing." Water Research (Great Britain), Vol.
4, pp. 521-532, 1970.
220
-------
Judd, John A. "Effect of Salts from Street Runoff on
Benthic Organisms." University of Wisconsin, Great
Lakes Center, Milwaukee, Wisconsin, 1967.
Kaufman, W. J. "Chemical Pollution of Ground Waters."
Journal of the American Water Works, Vol. 66, pp. 152-
159, 1974.
Kerr, R. L. , e_t al. "Analysis of Rainfall - Duration -
Frequency for Pennsylvania." Institute for Research on
Land and Water Resources Research, Publication No. 70,
Penn State University, University Park, Pennsylvania,
1970.
Keup, L. E. "Biology of Water Pollution." W. M. Ingram and
K. M. Mackenthum, Eds., U.S. Department of the Interior,
Federal Water Pollution Control Administration, 1967.
Khanna, S. D. "Effects of Highways on Surface and Sub-
surface Waters." Public Works, Vol. 104, pp. 1171-
1182, 1973.
King, D. L. and Ball, R. C. "Comparative Energetics of a
Polluted Stream." Limnology and Oceanography, Vol. 12,
No. 1, pp. 27-33, 1967.
Klein, L. A., e_t al. "Source of Metals in New York City
Wastewater." Journal of Water Pollution Control
Federation, Vol. 46, p. 2653, 1974.
Kluesener, J. W. and Lee, G. F. "Nutrient Loading from a
Separate Storm Sewer in Madison, Wisconsin." Journal
of the Water Pollution Control Federation, Vol. 46, pp.
920-936, 1974.
Knauer, D. R. "The Effect of Urban Runoff on Phytophankton
Ecology." Verhandlungen, Internationale Vereinigung
fuer Theoretische und Angewandte Limnologie. Vol. 19,
pp. 893-903, 1975.
Kothand Araman V. "Water Quality Characteristics of Storm
Sewer Discharges and Combined Sewer Overflows.
Illinois State Water Survey, Illinois Department of
Registration and Education, Circular 109, Urbana,
Illinois, 1972.
221
-------
Kramer, J. R.f Herbes, S. E. and Allen, H. E. "Phosphorus:
An Analysis of Water, Biomass, and Sediment." Nutri-
ents in Natural Waters, Wiley-Interscience, New York,
New York, 1972.
Kreisal, James F. "Rural Wastewater Research." Proceedings,
Second National Conference on Individual Onsite Waste-
water Systems, National Sanitation Foundation, Ann
Arbor, Michigan, pp. 145-157, 1976.
Krenkel, P. A., Cawley, W. A. and Minch, V. A. "The
Effect of Impounding Reservoirs on River Waste Assimi-
lative Capacity." Journal of the Water Pollution
Control Federation, Vol. 37, pp. 1203-1217, 1965.
Kuhner, J. and Shapiro, M. "Discussion of 'Urban Runoff
Pollution Control - State-of-the-Art1, by R. Field and
J. A. Lager." Journal of the Environmental Engineering
Division, Proceeding of the American Society of Civil
Engineers, Vol. 102, EEI, pp. 220-223, 1976.
Kuo, Chin Y. "Evaluation of Sediment Yield Due to Housing
Construction: A Case Study." Department of Civil
Engineering, Old Dominion University, Norfolk, Virginia.
Lager, J. A. and Smith, W. G. "Urban Stormwater Management
and Technology: An Assessment»" U.S. Environmental
Protection Agency, National Environmental Research
Center, EPA 670/2-74-040, 1974.
Lamonds, A. G. "Chemical and Biological Quality of Lake
Dicie at Eustis, Flordia, with Emphasis on the Effects
of Storm Runoff." U.S. Geological Survey, NTIS PB 239
014, Tallahassee, Florida, 1974.
La Valle, P. D. "Domestic Sources of Stream Phosphates in
Urban Streams." Water Research (Great Britain), Vol.
9, pp. 915-927, 1975.
Lazrus, A. L., Lorange, F. and Lodge, J. R. Jr. "Lead and
other Metal ions in United States Precipitation."
Environmental Science and Technology, Vol. 4, p. 55,
1970.
222
-------
Leclerc, G. "Methodology for Assessing the Potential Impact
of Urban Development on Urban Runoff and the Relative
Efficiency of Runoff Control Alternatives." PhD Thesis,
Massachusetts Institute of Technology, 1973.
Lee, G. F. "Role of Phosphorus in Eutrophication and
Diffuse Source Control." Water Research (Great Britain),
Vol. 7, pp. 111-128, 1973.
Leopold, L. B. , Wolman, M. G. and Miller, J. P. "Fluvial
Processes in Geomorphology." W. H. Freeman and Company,
San Francisco, California, 1964.
Leopold, L. B. "Hydrology for Urban Land Planning - A
Guidebook on the Hydrologic Effects of Urban Land Use."
U.S. Geological Survey Circular 554, 1968.
Likens, G. E., ed. "Nutrients and Eutrophication: The
Limiting Nutrient Controversy." Special Symposia (Vol.
1), American Society of Limnology and Oceanography,
Allen Press, Lawrence, Kansas, 1972.
Likens, G. E. "The Runoff of Water and Nutrients from
Watersheds Tributary to Cayuga Lake, New York."
Cornell University Water Resources and Marine Sciences
Center, Technical Report No. 81, Ithaca, New York,
1974.
Likens, G. E. "The Chemistry of Precipitation in the
Central Finger Lakes Region." Cornell University Water
Resources and Marine Sciences Center, Technical Report
No. 50, Ithaca, New York, 1972.
Loehr, R. C. "Characteristics and Comparative Magnitude
of Nonpoint Sources." Journal of the Water Pollution
Control Federation, Vol. 46, No. 8, pp. 1849-1872,
1974.
Mallory, C. W. "The Beneficial Use of Storm Water." U.S.
Environmental Protection Agency, EPA-R2-73-139, 1973.
Man-Made Lakes: Their Problems and Environmental Effects.
W. C. Ackermann, G. F. White and E. B. Worthington,
eds., American Geophysical Union, Washington, D. C. ,
1973.
223
-------
Mansue, L. J. and Coramings, A. B. "Sediment Transport by
Streams Draining into the Delaware Estuary." Water-
Supply Paper 1532-H, U.S. Government Printing Office,
Washington, D.C., 1974.
Mantri, V. and Kaushik, K. "A Model of Time-Varying,
Non-Uniform Flow in Open Channels." Part II, 1975.
Manuel, A. D., Gustafson, R. H. and Welch, R. B. "Three
Land Research Studies." National Commission on Urban
Problems, Report No. 12, 1968.
Marsalek, J., et al. "Comparative Evaluation of Three
Urban Runoff Models." Water Resources Bulletin, Vol.
11, No. 2, pp. 306-328, 1975.
Martin, D. M. and Gaff, D. R. "The Role of Nitrogen in the
Aquatic Environment." Academy of Natural Sciences,
NTIS PB 213 496, Philadelphia, Pennsylvania, 1972.
Maryland Department of Water Resources, Burton C. Becker
and Thomas R. Mills. "Guidelines for Erosion and
Sediment Control Planning and Implementation." Pre-
pared for U.S. Environmental Protection Agency, EPA
R2-72-015, 1072.
McBean, E. A. and Loucks, D. P. "Planning and Analysis of
Metropolitan Water Resources System." Cornell Uni-
versity Water Resources and Marine Science Center,
Technical Report No. 84, NTIS PB 235 257, Ithaca, New
York, 1974.
McCuen, R. H. "Flood Runoff from Urban Areas." Water
Resources Research Center, Technical Report No. 33,
University of Maryland, College Park, Maryland, 1975.
McElroy, A. D., et al. "Interim Report on Loading Functions
for Assessment of Water Pollution from Nonpoint Sources.
Prepared for U.S. Environmental Protection Agency by
Midwest Research Institute, Kansas City, Missouri,
1975.
McElroy, A. D., Chiu, S. Y. and Aleti, A. "Analysis of
Nonpoint Source Pollutants in the Missouri Basin
Region." U.S. Environmental Protection Agency, EPA
600/5-75-004, 1975.
224
-------
McElroy, A. D. , et al. "Water Pollution from Non-Point
Sources." Water Research (Great Britain), Vol. 9. pp.
675-681, 1975.
McHarg, Ian L. "Design with Nature." Doubleday/Natural
History Press, Garden City, New York, 1969.
McPherson, M. B., Orlob, G. T., Kibler, D. F. and Chen,
C. W. "Management of Urban Storm Runoff." NTIS PB
234 316, May 1974.
Meta Systems, Inc. "Land Use Environmental Quality Rela-
tionship." Prepared for U.S. Environmental Protection
Agency under contract 68-01-2622, 1975.
Metcalf and Eddy, Inc., University of Florida and Water
Resources Engineers. "Storm Water Management Model."
(4 volumes). Prepared for U.S. Environmental Protec-
tion Agency, 11024DOC, 1971.
Metcalf and Eddy, Inc. "Storm Water Problems and Control
in Sanitary Sewers." Prepared for the U.S. Environ-
mental Protection Agency, 11024 EQG, 1971.
Metcalf and Eddy, Inc. "Wastewater Engineering - Collection
- Treatment - Disposal." McGraw-Hill Inc., New York,
New York, 1972.
Middlebrooks, E. J. "Modeling the Eutrophication Process."
D. H. Falkenborg and T. E. Moloney, eds., Ann Arbor
Science Publishers, Inc., Ann Arbor, Michigan, 1974.
Miller, Fred P. and Wolf, D. C. "Renovation of Sewage
Effluents by the Soil." Proceedings of the Second
National Conference on Individual Wastewater Systems,
National Sanitation Foundation, Ann Arbor, Michigan,
pp. 87-101, 1976.
Miller, John C. "Nitrate Contamination of the Water-Table
Aquifer by Septic Tank Systems in the Coastal Plain
of Delaware." Water Pollution Control in Low Density
Areas, University_Press ..of New England, Hanover, New
Hampshire, pp. 121-133, 1975.
225
-------
Miller, R. Adam, Troxell, J. and Lopold, L. B. "Hydrology
of Two Small River Basins in Pennsylvania before
Urbanization." U.S. Geological Survey Professional
Paper 701-A, 1971.
Miller, W. L. and Erickson, S. P. "Systematic Development
of Methodogies in Planning Urban Water Resources for
Medium Size Communities." Purdue University Water
Resources Research Center, Report No. 39, West Lafay-
ette, Indiana, 1973.
Mills, D. M. and Watson, P. S. "Regional Environmental
Assessment Procedure." University of Pennsylvania,
Philadelphia, Pennsylvania, 1974.
Minneapolis-St. Paul Sanitary District. "Dispatching
System for Control of Combined Sewer Losses." Prepared
for the U.S. Environmental Protection Agency, 11020
FAQ, 1971.
"Models for Managing Regional Water Quality." R. Dorfman,
H. Jacoby and H. A. Thomas, eds. Harvard University
Press, Cambridge, Massachusetts, 1972.
Moore, Charles A. and Silver, Marshall L. "Nutrient Trans-
port by Sediment-Water Interaction." .Water Resources
Center Research Report, Illinois University, Urbana,
Illinois, 1973.
Morrow, N. L. and Brief, R. S. "Elemental Composition of
Suspended Matter in Metropolitan New York." Environ-
mental Science and Technology, Vol. 5, No. 9, 1971.
Murray, T., et al. "Honey Hill: A Systems Approach for
Planning Multiple Use of Controlled Water Areas."
Department of Los Angeles Research Office, Harvard
University, Cambridge, Massachusetts, 1971.
"National Conference on Managing the Environment." Spon-
sored by the U.S. Environmental Protection Agency,
1973.
•
National Water Monitoring Panel. "Model State Water Moni-
toring Program." Environmental Protection Agency, EPA
440/9-74-002, 1975.
226
-------
Newton, C. D., et al. "Street Runoff as a Source of Lead
Pollution." Journal of the Water Pollution Control
Federation, Vol. 46, No. 5, pp. 999-1000, 1974.
"Non-Point Sources of Water Pollution." Proceedings of a
Southeastern Regional Conference at Virginia Poly-
technic Institute, Virginia Water Resources Research
Center, Blacksburg, Virginia, 1975.
Norton, J. L. "The Identification and Measurement of
Chlorinated Hydrocarbon Pesticides Accumulated from
Urban Runoff." Prepared for U.S. Environmental Pro-
tection Agency by the Oklahoma Water Resources Re-
search Institute, NTIS PB 226 307, 1973.
Norvell, W. A. and Frink, C. R. "Water Chemistry and
Fertility of TwentyThree Connecticut Lakes." Con-
necticut Agricultural Experiment Station, New Haven,
Connecticut, 1975.
"NOX Emissions from Stationary Combustion Sources."
Journal of Environmental Engineering and Design, p.
641, June 1974.
"Nutrients in Natural Waters." H. E. Allen and J. R.
Kramer, eds. John Wiley and Sons, New York, New York,
1972.
Ogumrombi, Joseph A. and Dobins, William E. "The Effects
of Benthal Deposits on the Oxygen Resources of Natural
Streams." Journal of the Water Pollution Control
Federation, Vol. 42, No. 4, pp. 538-552, 1970.
Ohio-Kentucky-Indiana Regional Council of Governments. "A
Method for Assessing Rural Non-Point Sources and its
Application in Water Quality Management." Cincinnati,
Ohio, 1975.
Oliver, B. G. , Milne, J. B. and La Barne, N. "Chloride
and Lead in Urban Snow." Journal of the Water Pollu-
tion Control Federation, Vol. 46, No. 4, pp. 766-771,
1974.
"Organisms and Biological Communities as Indicators of
Environmental Quality - A Symposium." Sponsored by
Ohio Biological Survey, Ohio Environmental Protection
Agency and U.S. Environmental Protection Agency at Ohio
State University, 1974.
227
-------
O'Shaughnessy, J. C. and McDonnell, A. J. "Criteria for
Estimating Limiting Nutrients in Natural Streams."
Pennsylvania State University Institute for Research on
Land and Water Resources Research, Publication No. 75,
University Park, Pennsylvania, 1973.
Palmer, C. L. "Feasibility of Combined Sewer System."
Journal of the Water Pollution Control Federation, Vol.
35, NO. 2, pp. 162-167, 1963.
Palmer, C. M. "A Composite Rating of Algae Tolerating
Organic Pollution." Journal of Phycology, Vol. 5, No.
1, pp. 78-82, 1969.
Palmer, C. L. "The Pollutional Effects of Storm-Water
Overflows from Combined Sewers." Sewage and Industrial
Wastes, Vol. 22, No. 2, pp. 154-165, 1950.
Papadakis, C. N. and Preul, H. C. "Testing of Methods for
Determination of Urban Runoff." Journal of the Hy-
draulics Division, Proceedings of the American Society
of Civil Engineers, Vol. 99, HY9, pp. 1319-1335, 1973.
Papadakis, C. N. and Preul, H. C. "Urban Runoff Model."
Journal of the Hydraulics Division, Proceedings of the
American Society of Civil Engineers, Vol. 98, HY10, pp.
1789-1804, 1972.
Parmele, L. H. and McGuinness, J. L. "Comparisons of
Measured and Estimated Daily Potential Evapo-transpira-
tion in a Humid Region." Journal of Hydroloay (Nether-
lands), Vol. 22, pp. 239-251, 1974.
Patri, T., e_t al_. "Early Warning System: The Santa Cruz
Mountains Regional Pilot Study." Department of Land-
scape Architecture, College of Environmental Design,
University of California, Berkeley, California, 1970.
Patrick, Ruth. "A Proposed Biological Measure of Stream
Conditions Based on a Survey of the Conestoga Basin,
Lancaster County, Pennsylvania." Proceedings of the
Academy of Natural Sciences of Philadelphia, Vol. 101,
Philadelphia, Pennsylvania, December 17, 1949.
Pheiffer, T. H. and Lovelace, N. L. "Application of Auto-
Qual Modeling System to the Patuxent River Basin."
U.S. Environmental Protection Agency, Annapolis Field
Office Technical Report No. 58, EPA-903/9-74-013, 1973.
228
-------
Pitt, R. E. and Amy, G. "Toxic Materials of Street Surface
Contaminants." NTIS PB 224-677, August 1973.
Pitt, R. E. and Amy, G. "Toxic Surface Analysis of Street
Surface Contaminants." Prepared for the U.S. Environ-
mental Protection Agency, 11034 FUJ, EPA R2-73-283,
1973.
Plews, Gary D. "The Adequacy and Uniformity of Regulations
for Onsite Wastewater Disposal - A State Viewpoint."
Proceedings, Second National Conference on Individual
Onsite Wastewater Systems, National Sanitation Founda-
tion, Ann Arbor, Michigan, pp. 139-144, 1976.
Plymouth Architectural and Planning Associates, Inc., and
Betz Environmental Engineers, Inc. "Workshop on Storm
Water Management." Prepared for Pennsylvania Depart-
ment of Community Affairs, 1975.
Poertner, H. G. "Practices in Detention of Urban Stormwater
Runoff." American Public Works Association, Chicago,
Illinois, 1974.
Pollution Ecology of Freshwater Invertebrates. C. W. Hart
and S. L. H. Fuller, eds. Academic Press, Inc., New
York, New York, 1974.
Pravoshinsky, N. A. "Description of the Drainage of Street
Flushing." Soviet Hydrology, Selected Papers, Issue
No. 2, pp. 168-170, 168, 1968.
Preul, H. C. "Contaminants in Ground Water near Waste
Stabilization Ponds." Journal of the Water Pollution
Control Federation, Vol. 40, No. 4, pp. 659-669, 1968.
Processes, Procedures and Methods to Control Pollution
Resulting from all Construction." U.S. Environmental
Protection Agency, Office of Air and Water Programs,
EPA 430/9-73-00 7, 1973.
Putnam, A. L. "Effects of Urban Development of Floods in
the Piedmont Province of North Caroline." U.S. Geo-
logical Survey Open File Report, 1972.
Putnam, D. and Olson, T. A. "An Investigation of Nutrients
in Western Lake Superior." School of Public Health,
University of Minnesota, 1960.
229
-------
Quan, Edison L., Young, R. H. F., Burbank, N. C. Jr. and
Lau, L. S. "Effects of Surface Runoff into the Southern
Sector of Kaneoke Bay." Water Resources Research
Center, University of Hawaii, January 1970.
Radziul, J. V., Cairo, P. R. and Smoot, G. S. "Does
Stormwater Damage?" Water Pollution Control Associa-
tion of Pennsylvania Magazine, pp. 26-36, September-
October 1975.
Radziul, J. V., Cairo, P. R. and Smoot, G. S. "Does Storm-
water Pollute?" Water Pollution Control Association
of Pennsylvania, 45th Annual Conference, Penn State
University, University Park, Pennsylvania, August 1973.
Ragan, R. M. and Dietemann, A. J. "Impact of Urban Storm-
water Runoff on Stream Quality." Urbanization and
Water Quality Control, American Water Resources Associ-
ation, Minneapolis, Minnesota, 1975.
Randall, Clifford W., e_t al. "Characterization of Urban
Runoff in the Oceogran Watershed of Virginia." Urbani-
zation and Water Quality Control, American Water Re-
sources Association, Minneapolis, Minnesota, 1975.
Rao, R. A. and Chenchagya, B. T. "Probabilistic Analysis
and Simulation of the Short Time Increment Rainfall
Process." Purdue University Water Resources Research
Center, Technical Report No. 55, West Lafayette,
Indiana, 1974.
Rao, R. A. and Rao, R. G. S. "Analysis of the Effect of
Urbanization on Rainfall Characteristics - I." Purdue
University Water Resources Research Center, Technical
Report No. 50, West Lafayette, Indiana, 1974.
Rao, R. A. and Rao, R. G. S. "Comparative Analysis of
Estimation Method in Non-Linear Functional Models of
the Rainfall-Runoff Process." Purdue University Water
Resources Research Center, Technical Report No. 56,
West Lafayette, Indiana, 1974.
Reed, L. A. "Sediment Characteristics of Five Streams
Near Harrisburg, Pennsylvania, before Highway Con-
struction." Geological Survey, Open File Report 74-
410, Harrisburg, Pennsylvania, 1974; Government
Printing Office, Washington, D.C., 1976.
230
-------
Reeves, Mark and Miller, Edward E. "Estimating Infiltration
for Erratic Rainfall." Water Resources Research, Vol.
11, No. 1, pp. 102-110, 1975.
Remson, I., Fungarolc, A. A. and Lawrence, A. W. "Water
Movement in an Unsaturated Sanitary Landfill." Journal
of the Sanitary Engineering Division, Proceedings of
the American Society of Civil Engineers, Vol. 94, SA2,
pp. 307-317, 1968.
Responses of Fish to Environmental Changes. W. Chavin, ed.
Charles C. Thomas, Inc., Springfield, Illinois, 1973.
Rho, J. and Gunner, H. B. "Micro Floral Response to
Aquatic Weed Decomposition." University of Massa-
chusetts, Department of Environmental Sciences, Am-
herst, Massachusetts, n.d.
Rickert, D. A., Hines, W. G. and McKenzie, S. W. "Methods
and Data Requirements for River-Quality Assessments."
Water Resources Bulletin, Vol. 11, No. 5, pp. 1013-
1039, 1975.
Roesner, L. A. "A Storage, Treatment Overflow and Runoff
Model for Metropolitan Master Planning." Applications
of Stormwater Management Models - 1975, EPA manual,
1975.
Roesner, L. A. "Quality Aspects of Urban Runoff." Water
Resources Engineers, Walnut Creek, California, n.d.
Roesner, L. A. , e_t al. "A Model for Evaluating Runoff-
Quality in Metropolitan Master Planning." American
Society of Civil Engineers, Urban Water Resources Re-
search Program, Technical Memo. No. 23, 1974.
Rogowski, A. S. "Variability of the Soil Water Flow
Parameters and their Effect on the Computation of
Rainfall Excess and Runoff." International Symposium
on Uncertainties in Hydrologic and Water Resources
Systems, Pennsylvania State University, University
Park, Pennsylvania, n.d.
Rovers, F. A. and Farquhai, A. "Infiltration and Landfill
Behavior." Journal of the Environmental Engineering
Division, Proceedings of the American Society of Civil
Engineers, Vol. 99, EE5, pp. 671-690, 1973.
231
-------
Ross, Hardies, O'Keefe, Babcock and Parsons, Inc. "EPA
Authority Affecting Land Use." Prepared for U.S.
Environmental Protection Agency, NTIS PB 235 331, 1974.
Ruane, R. J. and Fruh, E. G. "Effects of Watershed Develop-
ment on Water Quality." Journal of the American Water
Works Association, Vol. 65, No. 5, pp. 358-363, 1973.
Ruskin, A. J., ed. "Aqueous Environmental Chemistry of
Metals." Ann Arbor Science Publishers, Inc., Ann
Arbor, Michigan, 1974.
Ryden, J. C., Syers, J. K. and Harris, R. F. "Nutrient
Enrichment of Runoff Waters by Soils, Phase 1: Phos-
phorus Enrichment Potential of Urban Soils in the City
of Madison." University of Wisconsin Water Resources
Center, Madison, Wisconsin, 1972.
Salvato, Joseph A., Jr. Environmental Engineering and
Sanitation. John Wiley and Sons, Inc., New York, New
York, 1972.
Salvato, Joseph A., Jr. "Problems and Solutions of Onlot
Sewage Disposal." Proceedings of the Second National
Conference on Individual Onsite Wastewater Systems,
National Sanitation Foundation, Ann Arbor Michigan,
pp. 39-46, 1976.
Sankowski, Stephen J. "Magnitude of Frequency of Floods
in New Jersey with Effects of Urbanization." U.S.
Geological Survey, Special Report 38, 1974.
Sarme, P. B. S., Delleur, J. W. and Rao, A. R. "A Program
in Urban Hydrology, Part II: An Evaluation of Rainfall
Runoff for Small Urbanized Watersheds and the Effect of
Urbanization on Runoff." Prepared for U.S. Environ-
mental Protection Agency, NTIS PB 198-043, 1969.
Sartor, J. D. and Boyd, G. B. "Water Pollution Aspects of
Street Surface Contaminants." .A study by the URS Re-
search Company for the U.S. EPA (EPA-R2-72-081),
Washington, D.C., November 1972.
Sartor, J. D., Boyd, G. B. and Agandy, F. J. "Water Pol-
lution Aspects of Street Surface Contaminants."
Journal of Water Pollution Control Federation, Vol.
46, No. 3, pp. 458-467, 1974.
232
-------
Schultz, J. M. "Pollutional Characteristics of Stormwater
Runoff from Urban, Semi-Urban and Rural Watersheds in
the West Lafayette, Indiana Area." Purdue University
Department of Civil Engineering, M.S. Thesis, West
Lafayette, Indiana, 1969.
Seattle, Municipality of Metropolitan. "Maximizing
Storage in Combined Sewer Systems." Prepared for the
U.S. Environmental Protection Agency, 11022 ELK, 1971.
"Sediment Sources and Sediment Yields." Journal of the
Hydraulic Division, Proceedings of the American Society
of Civil Engineers, Vol. 96, HY6, pp. 1283-1329, 1970.
"Sediment Transportation Mechanics: Erosion of Sediment."
Journal of the Hydraulics Division, Proceedings of the
American Society of Civil Engineers, Vol. 88, HY4, pp.
109-27, 1962.
Selected Urban Storm Water Runoff Abstracts. U.S. Environ-
mental Protection Agency, 1968-1970.
Shaheen, Donald G. "Contributions of Urban Roadway Usage to
Water Pollution." Prepared for the U.S. Environmental
Protection Agency, EPA report No. 600/2-75-004, 1975.
Shaheen, D. G. "Passenger Cars are Big Water Polluters,
Biospherics Finds.: Chemical and Engineering News,
Vol. 51, No. 27, p. 10, July 1973.
Shakla, S. S. and Leland, H. V. "Heavy Metals: Review of
Lead." Journal of the Water Pollution Control Federa-
tion, Vol. 45, No. 6, pp. 1319-1331, 1973.
Shannon, E. E. and Brezonik, P. L. "Relationships between
Lake Trophic State and Nitrogen and Phosphorus Loading
Rates." Environmental Science and Technology, Vol. 6,
No. 8, pp. 719-725, 1972.
Sikard, L. J. and Keeney, D. R. "Laboratory Studies on
Stimulation of Biological Denitrification." Proceed-
ings of the National Home Sewage Disposal Symposium,
American Society of Agricultural Engineers, St. Joseph,
Missouri, pp. 64-73, 1975.
Sinqer P C. "Trace Metals and Metal Organic Interactions
in Natural Waters." Ann Arbor Science Publishers, Ann
Arbor, Michigan, 1973.
233
-------
Snodgrass, William J. and O'Melia, Charles R. "Predictive
Model for Phosphorus-in Lakes." Environmental Science
and Technology, Vol. 9, No. 10r pp. 937-944, 1975.
Soltero, R. A., Wright, J. C. and Horpestad, A. A. "Effects
of Impoundment on the Water Quality of the Bighorn
River." Water Research (Great Britain), Vol. 7, pp.
343-354, 1973.
Spiegelman, Robert. "Review of Techniques of Regional
Analysis, with Particular Emphasis on Applicability to
Regional Problems." Stanford Research.Center, Palo
Alto, California, 1962.
Spooner, C. S., Promise, J. and Graham, P. H. "A Demonstra-
tion of Areawide Water Resources Planning for Metro-
politan Washington, (Draft)." EPA, Washington, D.C.,
n.d.
Sridharan, N. and Lee, G. F. "Phosphorus Studies in Lower
Green Bay, Lake Michigan." Journal of the Water
Pollution Control Federation, Vol. 46, No. 4, pp. 684-
696, 1974.
Stankowski, Stephen J. "Population Density as an Indirect
Indicator of Urban and Suburban Land-Surface Modifica-
tions." U.S. Geological Survey, Geological Survey
Research Professional Paper 800-B, pp. B219-B224, 1972.
Steinitz, C., et al. "A Comparative Study of Resource
Analysis Methods." Department of LA Research Office,
GSD, Harvard University, Cambridge, Massachusetts,
July 1969.
Sutherland, R. and McCuen, R., R. "A Mathematical Model for
Estimating Pollution Loadings in Runoff from Urban
Streets." Preprint from Proceedings of the Interna-
tional Conference on Mathematical Models of Environ-
mental Problems, Southampton U.K., 1975. >
Sutterlin, A. M. j "Pollutants and the Chemical Senses of
Aquatic Animals - Perspective and Review." Chemical
Senses and Flavor, Vol. 1, pp. 167-178, 1974.
Sylvester, R. O. and DeWalle, F. B. "Character and Sig-
nificance of Highway Runoff Waters, A Preliminary
Appraisal." Washington State Highway Commission, Y-
1441, 1972; NTIS PB 220-083, December 1972.
234
-------
Tao, P. C. and Delleur, J. w. "Models of the Stochastic
and Chronologic Structure, Prediction and Simulation of
Runoff Sequences - Application to the Lower Ohio
Basin." Purdue Water Resources Research Center, West
Lafayette, Indiana, 1975.
Tarzwell, Clarence M., ed. "Biological Problems in Water
Pollution." Third Seminar, Robert A. Taft Sanitary
Engineering Center, Cincinnati, Ohio, 1962.
Task Group Report. "Sources of Nitrogen and Phosphorus
in Water Supplies." Journal of the American Water
Works Association, Vol. 59, pp. 344-366, 1967.
Terstries, M. L. and Stall, J. P. "Urban Runoff by Road
Research Lab Method." Journal of the Hydraulics
Division, Proceedings of the American Society of Civil
Engineers, Vol. 95, HY6, pp. 1809-1834, 1969.
Tholin, A. L. and Keiber, C. J. "The Hydrology of Urban
Runoff." Journal of the Sanitary Engineering Division,
Proceedings of the American Society of Civil Engineers,
Vol. 85, SA2, pp. 47-106, 1959.
Thomann, R. J. Systems Anaj-ysis and Water QualityManage-
ment. McGraw-Hill, New York, New York, 1972.
Thompson, G. B., et al. "Variations of Urban Runoff
Quality and Quantity with Duration and Intensity of
Storms - Phase III." Texas Tech University Water
Resources Center, Lubbock, Texas, 1974.
Toebes, G. H. and Chang, T. P. "Simulation Model for the
Upper Wabash Surface Water System." Purdue University
Water Resources Research Center, West Lafayette,
Indiana, 1973.
Torno, H. C. "A Model for Assessing Impact of Stormwater
Runoff and Combined Sewer Overflows and Evaluating
Pollution Abatement Alternatives." Water Research
(Great Britain), Vol. 9, pp. 849-852, 1975.
Tourbier, Joachim. "Water Resources as a Basis for Compre-
hensive Planning and Development of the Christina
River Basin." Prepared for U.S. Department of the
Interior by Water Resources Center, University of
Delaware, Newark, Delaware, 1973.
235
-------
Tourbier, J. and Westmacott, R. "Water Resources Protection
Measures in Land Development" - A Handbook, Water
Resources Center, University of Delaware, Newark, Dela-
ware, pp. 14-16, April 1974.
Tuffey, T. J., Hunter, J. V. and Matulewich, V. A. "Zones
of Nitrification." Water Resources Bulletin, Vol. 10,
No. 3, pp. 555-564, 1974.
Turner, Collie and Braden, Inc. "Stormwater Management
Report." (Draft) Prepared for New Castle County,
Delaware by Turner, Collie and Braden, Inc., Houston,
Texas, 1975.
ORS Research Company. "Water Quality Management Planning
for Urban Runoff." (Draft) Prepared for U.S. En-
vironmental Protection Agency, Contract No. 68-01-
1846, August 1964.
Urban Land Institute. "Residential Stormwater Model."
Washington, D. C.
"Urban Hydrology for Small Watersheds." U.S. Department
of Agriculture, Soil Conservation Service, Central
Technical Unit, Hydrology Technical Note 1, 1973.
m
Urban Stormwater Management Modeling and Decision-Making.
Prepared for National Environmental Research Center by
Florida University, PB 242-290, 1975.
Urban Systems Research and Engineering, Inc. "Evaluation
of the Use of Existing and Modified Land Use Instru-
ments to Achieve Environmental Quality." Urban Systems
Research and Engineering, Inc., Cambridge, Massachu-
setts, 1975.
U.S. Department of Agriculture. "Predicting Rainfall-
Erosion Losses from Cropland East of the Rocky Moun-
tains." Agricultural Research Service, Agriculture
Handbook No. 282, 1965.
U.S. Department of Agriculture, Soil Conservation Service.
"Soil Survey - Montgomery County, Pennsylvania."
Government Printing Office, Washington, D.C., 1967.
U.S. Environmental Protection Agency. "Characterization and
Treatment of Urban Land Runoff." EPA 670/2-74-096,
December 1975.
236
-------
U.S. Environmental Protection Agency. "Methods for Identi-
fying and Evaluating the Nature and Extent of Nonpoint
Sources of Pollutants." EPA-430-73-014, Washington,
D.C., 1973.
U.S. Department of Housing and Urban Development. "Urban
and Regional Informations Systems." Government Print-
ing Office, Washington, D.C.
Uttormark, Paul D., Chapin, John D. and Green, Kenneth M.
"Estimating Nutrient Loadings of Lakes." Water Re-
sources Center, EPA 660/3-74-020, Madison, Wisconsin,
1974.
Vice, R. B., Guy, H. P. and Ferguson, G. E. "Sediment
Movement in an Area of Suburban Highway Construction,
Scott Run Basin, Fairfax County, Virginia." U.S.
Geological Survey, Water Supply Paper 1591-E, 1969.
Viessmar, W. "Assessing the Quality of Urban Drainage."
Public Works, Vol. 100, No. 10, pp. 89-92, 1969.
Vitale, A. M. and Sprey, P. M. "Total Urban Water Pollution
Loads: The Impact of Stormwater." Prepared for U.S.
Environmental Protection Agency by Enviro Control,
Inc., Rockville, Maryland, NTIS PB 231 730, 1974.
Walker, W. G., et al. "Nitrogen Transformations During
Subsurface Disposal of Septic Tank Effluent in Sands I:
Soil Transformations." Journal of Environmental Quality,
Vol. 2, pp. 475-480, 1973.
Walker, W. G., et al. "Nitrogen Transformations During
Subsurface Disposal of Septic Tank Effluent in Sands
II: Ground Water Quality." Journal of Environmental
Quality, Vol. 2, pp. 521-525, 1973.
Walker, William H. "Groundwater Nitrate Pollution in Rural
Areas." Ground Water, Vol. 11, No. 5, pp. 19-22, 1973.
Wall, J. P., elt al. "Wisconsin Lakes Receiving Sewage
Effluent." Wisconsin Water Research Center, Technical
Report 73-1, EPA R-801-863, 1973.
Wallace, Douglas A. and Dague, Richard R. "Modeling of
Land Runoff Effects on Dissolved Oxygen." Journal of
the Water Pollution Control Federation, Vol. 45, No. 8,
pp. 1795-1809, 1973.
237
-------
Wallis,'!. G. "Options for Improving Water Quality."
International Journal of Environment Studies, Vol. 6,
pp. 107-120, 1974.
Warner, Maurice L. and Preston, Edward H. "A Review of
Environmental Impact Assessment Methodologies." Pre-
pared for U.S. Environmental Protection Agency, Office
of Research, EPA 600/5-74-002, April 1974.
"Waste Lube Oils Pose Disposal Dilemma." Environmental
Science and Technology, Vol. 6, No. 1, p. 25, 1972.
"Water Pollution Aspects of Urban Runoff." Prepared for
the Federal Water Pollution Control Administration,
U.S. Department of Interior, by the American Public
Works Association, Government Printing Office, Wash-
ington, D.C., 1969.
"Water Quality Criteria 1972." Ecological Research Series,
R3.73.033, Washington, D.C., March.1973.
"Water Quality Management for Urban Runoff." U.S. Environ-
mental Protection Agency. NTIS PB 241 689.
I
"Water Quality Management Planning for Urban Runoff." U.S.
Environmental Protection Agency, EPA 440/9-75-004,
1975. ;
"Water Quality Models for Urban and Suburban Areas."
Prepared for U.S. Environmental Protection Agency, NTIS
PB 238 622, University of Nebraska, Lincoln, Nebraska,
1974.
Water Resources Center, University of Delaware. Water
Resources Protection Measures in Land Development - A
Handbook. University of Delaware Water Resources
Center, 1974.
Water Resources Council. "A Summary Analysis of 19 Tests
of Proposed Evaluation Procedures on Selected Water and
Land Resources Projects." 1970.
Weibel, S. R., Anderson, R. J. and Woodward, R. L. "Urban
Land Runoff as a Factor in Stream Pollution." Journal
of the Water Pollution Control Federation, Vol. 36, No.
7, pp. 914-924, 1964.
238
-------
Weibel, S. R. "Urban Drainage as a Factor in Eutrophica-
tion." Eutrophication: Causes, Consequences, Correc-
tives. Proceedings of a Symposium, National Academy
of Sciences, Washington, D.C., 1969.
Weibel, S. R., e_t al. "Pesticides and Other Contaminants
in Rainfall and Runoff." Journal of the American Water
Works Association, Vol. 58, No. 8, pp. 1075-1084, 1966.
Weibel, S. R., et al. "Treatment of Urban Stormwater
Runoff." Journal of the Water Pollution Control Federa-
tion, Research Supplement, Vol. 40, No. 5, Part 2, R
162-R170, 1968.
Welb, D. M., et al. "Variation of Urban Runoff Quality with
Duration and Intensity of Storms - Phase II." NTIS
No. PB-223 930, 1973.
Werner, R. G. "Water Quality-Limnological Concerns about
Forest Fertilization." Forest Fertilization Symposium
Proceedings, College of Engineering Science and Forestry,
S.UvN.Y."Warrensburg, New York Campus, 1973.
Werschmeir, W. H. and Smith, D. D. "Predicting Rainfall
Erosion Losses from Cropland East of the Rocky Moun-
tains. " Agricultural Handbook 282, U.S. Government
Printing Office, Washington/ D.C., 1965.
Weston, Roy F., Inc. "Combined Sewer Overflow Abatement
Alternatives." Prepared for U.S. Environmental Pro-
tection Agency by Roy F. Weston, Inc., West Chester,
Pennsylvania, 11024 EXF, 1970.
Weston, Roy F., Inc. "Lancaster County Planning Commission
Storm Drainage Study." Roy F. Weston, Inc., West
Chester, Pennsylvania, 1970.
Whipple, William Jr. "Urban Runoff: Quantity and Quality."
Proceedings of a Research Conference at Rindge, New
Hampshire, American Society of Civil Engineers, New
York, New York, 1974.
Whipple, William Jr., ed. "Urbanization and Water Quality
Control." 'American Water Resources Association, Min-
neapolis, Minnesota, 1975.
239
-------
Whipple, William Jr., et al. "Unrecorded Pollution and
Dynamics of Biochemical Oxygen Demand." Rutgers
University, Water Resources Research Institute, New
Brunswick, New Jersey, 1974.
Whipple, William Jr. and Hafschmidt, M. M. "Reorientation
of Urban Water Resources Research." Rutgers University
Water Resources Research Institute, New Brunswick, New
Jersey, 1976.
Whipple, W. Jr. and Hunter, J. V. "Non-Point Sources and
Planning for Water Pollution Control." Presented at
the 48th Annual Water Pollution Control Federation
Convention, Miami Beach, Florida, 1975.
Whipple, W., Hunter, J. V. and Yu, S. L. "Unrecorded
Pollution from Urban Runoff." Journal of Water Pol-
lution Control Federation, Vol. 46, No. 3, pp. 873-885,
1974.
Wilber, William G. and Hunter, Joseph V. "Contributions of
Metals Resulting from Stormwater Runoff and Precipita-
tion in Lodi, New Jersey." American Water Resources
Association, pp. 45-58, June 1975.f
Wilber, William A. and Hunter, Joseph V. "Heavy Metals
in Urban Runoff." Rutgers University Department of
Environmental Science, New Brunswick, New Jersey, 1975.
Wiley, Morris A. The Petroleum Industry and Cost Effective
Water Quality Planning: I: Assessments of PL 92-500
and II: Improvement of Cost Effectiveness. Presented
at a Symposium on Urbanization and Water Quality
Control at Rutgers University, New Brunswick, New
Jersey, 1975.
Williams, J. R. "Sediment Yield Prediction with Universal
Equation Using Runoff Energy Factor." USDA Resource
Service, Oxford, Mississippi, November 28-30, 1972.
Williams, J. R. and Berndt, H. D. "Sediment Yield Computed
with Universal Equation." Journal of the Hydraulics
Division, Proceedings of the American Society of Civil
Engineers, Vol. 98, HY12, pp. 2087-2098, 1972.
240
-------
Williams, L. G. , Joyce, J. C. and Monk, j. T. Jr "
-lYof thCt* °n-the H6aVy MetilS Concentration"
Jf pi?? ££££ f^i: W°rks Association, Vol.
J» R ''Se*iment Routing for Agricultural Water-
965-974, 197
w
ReSOUrces Bulletin, Vol. 11, No. 5,
pp
Wischmeier, W. H., Johnson and Cross. "A Soil Erodibilitv
Nomograph for Farmland and Construction Sites "
Journal of Soil Water Conservation, Vol. 26, pp. 189-
193, 1971.
Wischmeier, W. H. and Smith, D. D. "Rainfall Energy and
its Relationship to Soil Loss." Transactions of the
American Geophysical Union, Vol. 39, No. 2, 1958.
Wolman, G. Gordan. "Stream Standards: Dead or Hiding?"
Journal of the Water Pollution Control Federation *, Vol
46, No. 3, 1974.
Wolman, M. G. and Schick, A. D. "Effects of Construction
on Fluvial Sediment: Urban and Suburban Areas of Mary-
land." Water Resources Research, Vol. 3, No. 2. pp
451-462, 1967.
Wulkowicz, G. M. and Saleem, Z. A. "Chloride Balance of
an Urban Basin in the Chicago Area." Water Resources
Research, Vol. 10, No. 5, pp. 974-982, 1974.
Yen, Ben Chie. "Methodologies for Flow Prediction in
Urban Storm Drainage Systems." Prepared for U.S.
Environmental Protection Agency, NTIS PB 225-480, 1973.
Young, R. A. and Wiersma, J. L. "The Role of Rainfall
Impact on Soil Detachment and Transport." Water Re-
sources Research, Vol. 9, No. 6, pp. 1629-1636, 1973.
Young, C. E. Current Research on Land Application of
Waste Water and Sludge. Penn State University, Insti-
tute for Research on Land and Water Resources, Uni-
versity Park, Pennsylvania, 1975.
Yu, S. L., Whipple, W. and Hunter^ J. V. "Assessing
Unrecorded Organic Pollution from Agricultural, Urban
and Wooded Lands." Water Research (Great Britain),
Vol. 9, pp. 849-852, 1975.
241
•US. GOVERNMENT PRINTING OFFICE: 1976 720-117/8791
------- |