&EPA
United States
Environmental Protection
Agency
Office of Environmental
Processes and Effects Research
Washington DC 20460
EPA 600/9-80-017
March 1980
Research and Development
Proceedings Stormwater
Management Model (SWMM)
Users Group Meeting
January 10-11,1980
Miscellaneous Reports Series
-------
EPA-600/9-80-017
March 1980
PROCEEDINGS
STORMWATER MANAGEMENT MODEL (SWMM)
USERS GROUP MEETING
10-11 January 1980
Project Officer
Harry C. Torno
Office of Environmental Processes and Effects Research
Office of Research and Development
U.S. Environmental Protection Agency
Washington, D.C. 20460
OFFICE OF ENVIRONMENTAL PROCESSES AND EFFECTS RESEARCH
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D. C. 20460
-------
CONTENTS
Page
Number
Foreword i v
Abstract v
STORMWATER MANAGEMENT PLANNING USING THE PENN STATE
RUNOFF MODEL; D.F. Lakatos 1
COMPUTER MODELING FOR WATERSHED MANAGEMENT IN NORTHERN
VIRGINIA; H.A. Bonuccelli, J.P. Hartlgan, D.F. Blggers 17
COMPARISON OF DESIGN PEAK FLOWS CALCULATED BY THE
RATIONAL METHOD AND THE EPA-SWMM MODEL; P. Wisner,
A. Kassem, P. Cheung 41
URBAN RUNOFF QUALITY IN METROPOLITAN TORONTO;
F.I. Lorant 75
AN EXAMINATION OF THE STORM WATER MANAGEMENT MODEL
(SWMM) SURFACE-RUNOFF-QUALITY ALGORITHMS; W.M. Alley 93
THE USE OF SWMM TO ECONOMICALLY MODEL SURCHARGED COMBINED
SEWER SYSTEMS; W.M. Parker, III 111
IMPROVEMENTS IN EXTRAN; L.A. Roesner, A.M. Kassem, P.E. Wisner 132
CSO IMPACT DETERMINATION BY LONG TERM SIMULATION; H.M. Shapiro,
J.B. Blenk, M.P. Allen 142
TOWARDS STANDARDS FOR COMPUTER-BASED MUNICIPAL DRAINAGE
STUDIES; W. James, M.A. Robinson 190
SIMULATION OF EFFECTS OF URBANIZATION ON STORMWATER
HYDROGRAPHS AND POLLUTOGRAPHS - A REGIONALIZED
PARAMETRIC APPROACH; D.E. Overton, W.L. Troxler, E.C. Crosby 207
STORMWATER MODELLING APPLICATIONS IN THE CITY OF
EDMONTON; M. Ahmad 223
STORMWATER RUNOFF MODELLING OF THE TAMPA PALMS
PROPERTY; K. Smolenyak 243
DRAINAGE SYSTEM DESIGN AND ANALYSIS OF THE TAMPA
PALMS PROPERTY; R.J. Motchkavltz 263
-------
Page
Number
WATER QUALITY IN THE FIRST, SECOND AND THIRD ORDER STREAMS
OF AN UPLAND AND FORESTED WETLAND WATERSHED; C.M. Courtney 286
IMPLEMENTATION OF STORM WATER MANAGEMENT MODELS: PROS AND
CONS OF STANDARDIZATION; P. Wisner, A. Kassem, P. Cheung 306
List of Attendees 326
-------
FOREWORD
A major function of the Research and Development programs of the
Environmental Protection Agency is to effectively and expeditiously transfer,
to the user community, technology developed by those programs. A corollary
function is to provide for the continuing exchange of information and ideas
between EPA and users, and between the users themselves. The Stormwater
Management Model (SWMM) users group, sponsored jointly with Environment
Canada/Ontario Ministry of the Environment, was established to provide such
a forum.
This report, a compendium of papers presented at the last Users Group
meeting, is published in the interest of disseminating to a wide audience
the work of Group members.
Allan Hirsch
Deputy Assistant Administrator
Office of Environmental Processes
and Effects Research
IV
-------
ABSTRACT
This report includes fifteen papers, on topics related to the development
and application of computer-based mathematical models for water quantity and
quality management, presented at the semi-annual meeting of the Joint U.S.-
Canadian Stormwater Management Model (SWMM) Users Group, held 10-11 January
1980 in Gainesville, Florida.
Topics covered include a description of two urban runoff models, an
examination of runoff quality algorithms in the SWMM, a discussion of improve-
ments to the Extended Transport (EXTRAN) portion of the SWMM, applications of
several urban drainage models in planning, analysis and design, and a compari-
son of the Rational Method and the SWMM. Also Included are a paper on sug-
gested methods for standardizing the process of acquiring modeling services,
and a paper on urban runoff quality data collection 1n the Toronto, Ontario
area.
-------
DISCLAIMER
This report has been reviewed by the Office of Environmental Processes
and Effects Research, Office of Research and Development, U.S. Environmental
Protection Agency, and approved for publication. Approval does not signify
that the contents necessarily reflect the views and policies of the U.S.
Environmental Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
-------
STORMWATER MANAGEMENT PLANNING
USING THE PENN STATE RUNOFF MODEL
by
David F. Lakatos, P.E,1
INTRODUCTION
Comprehensive planning for the control of stormwater runoff is a
very important part of the overall development objectives for establish-
ed communities, as well as those that are experiencing rapid growth at
the present time. The pattern of growth and development in most areas,
however, is such that many individual developments are built separately,
but together they eventually form the "overall community." The storm-
water management facilities for each individual development are in turn
designed and constructed separately, but they must eventually function
as a single hydraulic unit for the entire community. By the nature of
this pattern, it can be seen that some form of "comprehensive" analysis
of stormwater runoff control facilities should be performed for develop-
ments of all sizes in order to achieve optimum stormwater management on
a watershed basis.
At the present time there is a significant need for more up-to-date,
but readily usable, stormwater management technology for engineers who
design stormwater management systems for these individual developments.
These engineers are typically associated with small consulting organi-
zations and may not have the exposure to the state-of-the-art technology
that is being developed on a national level. Their applications do not
generally involve a high degree of complexity, and do not involve the
contract dollar amounts that are typically associated with stormwater
management projects in more heavily urbanized communities. Therefore,
the situation has evolved where the technological benefits of large-
scale research in stormwater management are restricted to only very
sophisticated applications; and the results of this research are not
accessible to a group with a very definite need.
One form of technology which has evolved from years of research in
the field of stormwater management is that of computer modelling. Com-
puter simulation models are very effective tools for analyzing the
effects of stormwater runoff in urban, as well as urbanizing, areas.
Consequently, a large number of simulation models have been developed.
Many of the existing models are very flexible in the sense that they
have the ability to analyze a broad range of urban stormwater problems,
1. Environmental Engineer, Satterthwaite Associates, Inc.
11 North Five Points Road, West Chester, PA 19380
-------
such as the very complicated aspect of storm runoff quality. This
degree of flexibility, while desirable in some urban applications, is
often wasteful in that large amounts of computer storage space and
compilation time are required for operations that may never be used.
In addition to this, most efforts to modify existing models have been
directed toward providing even more flexibility into the analysis rou-
tines, which typically increases overall model complexity.
In addition to this, the transfer of state-of-the-art technology
is limited for computer simulation models which can be used for
"smaller-scale" stormwater management analyses (i.e., for small com-
munities or individual developments). Many of the larger, more compre-
hensive models, toward which national attention as well as research and
development efforts have been directed, are not practical or directly
applicable for the analysis of stormwater runoff in small and mid-size
communities or individual developments. The benefits of major national
research efforts in stormwater management are therefore not directly
accessible to a great many engineers involved with stormwater manage-
ment.
This has led to the situation where stormwater management plans
for small communities, and individual large developments, are prepared
on a piece-meal basis. This is not a very effective technique for
stormwater management in that it is not coordinated, on a municipal
level, or certainly on a watershed level. The general lack of compre-
hensive, "watershed-level" stormwater management planning points direct-
ly to this need for a transfer of stormwater management technology on a
more local level. That is, there is a need for state-of-the-art tech-
niques and "tools" which can be used to evaluate individual stormwater
management designs in a comprehensive, watershed-level framework.
Many areas are identifying a real need to evaluate storm runoff,
and to provide stormwater management, on a watershed basis. This is
the case in Pennsylvania where recent legislation has been passed which
will require that comprehensive stormwater management plans be prepared
on a watershed basis. This action points even more toward the short-
term need for an analysis tool that can be used to evaluate the inter-
actions of many small, individual storm drainage systems on a watershed
basis. The Pennsyvlania Stormwater Management Act is an example of the
future direction and needs of stormwater management, i.e., comprehensive
and coordinated evaluations of many individual storm drainage designs in
order that the overall plan is responsive to the needs of the public, as
well as of the environment.
The Penn State Runoff Model was developed in response to the storm-
water management needs outlined above. The objectives that were adopted
for the development of the models were:
(1) To produce an urban runoff simulation model that will provide
acceptable hydraulic accuracy while remaining at a level of
sophistication compatible with minimum practice and data-
collection time, and therefore, minimum cost.
-------
(2) To keep the model as simple and concise as possible, and thus
convenient for use for small to medium-size communities, or
for individual developments.
(3) To provide a stormwater management tool for the analysis of
the timing of subarea flow contributions to peak rates at
various points in a watershed. This tool is known as the
Peak Flow Presentation Table, and will be described in a
later section of this paper.
The use of the Penn State Runoff Model is increasing for "smaller-
scale" applications because of its ability to provide a cost-effective
tool that can be used to evaluate state-of-the-art stormwater management
for any application, regardless of its size. In addition to this, it is
a model that can be cost-effectively applied on a watershed basis to
"tie together" several individual separate analyses into one coordinated
stormwater management plan. This capability satisfies the needs of com-
prehensive stormwater management planning.such as is being proposed, for
example, in Pennsylvania's recently-enacted Stormwater Management Act.
This technical paper presents a description of the Penn State Run-
off Model, along with a discussion of its
• general methods of analysis
t applications, with the aid of case study examples,
• capability to serve as a tool for comprehensive watershed
stormwater management planning
THE PENN STATE RUNOFF MODEL - A BRIEF DESCRIPTION
A visible flood flow is merely the result of the combination of
smaller flows from various subareas within a watershed. Subarea flow
combinations are a function of the travel times of runoff from these
subareas, particularly to junction points. The relative timing of peak
flows from different subareas determines the magnitude of aggregate flow
downstream, which in turn is directly related to the extent of flooding
that is experienced. This is illustrated conceptually in Figure 1.
It can be seen (Figure 1-A) that when the "time to peak flow"
(timing) for two subareas to a junction/combination point is the same,
in terms of time from the beginning of the storm event, a compounded
overall peak flow occurs at the junction point (Point X). This overall
peak flow is generated primarily by the simultaneous occurrence of peak
flow rates from the two contributing subareas, and the magnitude of the
total combined peak flow is equal to the sum of the two individual sub-
area peak flow rates. Conversely, if the "timing" of subarea peak flows
is such that the two subarea peak flow rates do not occur at the same
time (Figure 1-B), the overall peak flow rate at the junction point
(Point Y) will be significantly less even though the contributing area
is relatively the same.
-------
/Combined
% Hydrograph at
Point X
\
/ Subarea B %
* Hydrograph _
\
Time
Subarea Peak Flows Occuring Simultaneously at Point X
Combined
Hydrograph at
Point Y
Time
Subarea Peak Flows Occuring at Different Times Due to Different Lag Times
FIGURE 1
ILLUSTRATIVE EXAMPLE OF THE "TIMING" OF SUBAREA PEAK FLOWS
-------
Development of a watershed tends to increase the stormwater runoff
'peak flow rate and also decreases the time to the occurrence of the peak
flow rate at a given downstream point. Therefore, the development that
typically takes place in a watershed can cause the peak flows for two
streams, or "branches" of a watershed, to occur at the same time after
the start of a storm where this was not the case under natural conditions,
This situation typically causes storm runoff flooding, with its assoc-
iated damages, that stormwater management seeks to alleviate. The Penn
State Runoff Model was developed to provide the capability of evaluating
this situation in detail, and also of analyzing and optimizing engineer-
ing alternatives for cost-effective stormwater management.
The Penn State Runoff Model is basically an urban runoff timing
analysis model. A "watershed timing analysis," as it is used here,
refers to a computer study of the combinations of subarea runoff flows
and of their relationship to total watershed runoff for a particular
storm event. This model, developed in 1976 at the Pennsylvania State
University (1), was a response to the lack of an existing runoff simula-
tion model that could be used for the analysis of the timing of subarea
flow combinations. The Penn State Runoff Model can be used to analyze
the effectiveness of stormwater management facilities, such as the use
of stormwater detention structures, as a function of their location
within a watershed drainage system.
The model was designed to be as concise as possible. Simpler meth-
ods of analyzing infiltration, of generating runoff hydrographs, and of
routing flow through a drainage system were programmed into the model
to reduce computer execution time and cut overall operational costs.
The capability of analyzing subwatershed flow combinations through
easily-interpreted illustrations was also an important objective in the
development of this model.
GENERAL METHOD OF ANALYSIS
The Penn State Runoff Model simulates rainfall-runoff events on the
basis of the following information:
(1) Rainfall inputs:
t rainfall hyetographs, which can vary both temporally
and spatially
(2) Watershed representation:
t physical characteristics of the watershed
• conveyance system characteristics
• retention/detention basin storage charac-
teristics
Based on this input, the model predicts the outcome of the storm in the
form of runoff hydrographs, which represent:
-------
(1) Overland flow to a drainage point
(2) Pipe flow leaving a drainage poing
(3) Surcharge flow at a drainage point
The available documentation for the model describes the calculation
techniques in detail (2); the general outline of these techniques pre-
sented here provides an understanding of the basic processes being per-
formed.
Rainfall Analysis
To allow for the spatial as well as the temporal variation of a
rainfall event, the data from several recording and non-recording rain
gauges can be applied to any subwatershed, and can be used to account
for a system of rain gauges or for moving storm systems in a watershed.
Weighting factors can be applied to rain-gauge data to provide a more
accurate representation of the rainfall characteristics of particular
subwatersheds. There are two techniques that are available in the model
to specify, for a particular subarea, the volume and pattern of rainfall
from a single design storm being used for the entire watershed. These
techniques are:
(1) Thiessen-type manual weighting -- where the model user must
supply a weighting variable for each recording and non-record-
ing rain gauge being used. These variables are specified by
the user for each subarea and may be determined by a Thiessen
diagram or other methods. This weighting factor is then used
to adjust the volume of rainfall being specified for a given
time period in order that a representative value is used for
each individual subarea.
(2) Pattern conserving weighting -- for this method the user
must supply an x-y coordinate reading for each recording
rain gauge, and also for the centroid of each subarea.
The model then computes the rainfall weighting factors for
each subarea as values which are inversely proportional to
the square of the distances between the rain gauges and
the subarea. Also, in order to avoid the situation where
a distorted hyetograph is formed (excessively long dura-
tion and low intensity) by weighting several rain gauges
with differing temporal rainfall distributions, the model
computes the total storm depth and the temporal center of
gravity of the storm for each rain gauge. To develop a
design storm for each subarea, the pattern of the hyeto-
graph from the closest rain gauge (that with the largest
weighting factor) is adopted. The weighting factors are
then applied to adjust the total storm depth and the storm
center of gravity, thus resulting in an upward or downward
scaling and a time shift of the hyetograph. This is illus-
trated in Figure 2.
-------
1 -
1 -
&
1 -
1 -
1 -
Raingage 1
V\feight = 0.3
(a)
Raingage 2
Weight = 0.2
Raingage 3
Weight = 0.5
Conventionally
weighted hyetograph
Hyetograph weighted
by pattern conserving
method.
(b)
(c)
(d)
(e)
1 23456789 10
Time Units
FIGURE 2
EXAMPLE OF HYETOGRAPH WEIGHTING METHODS
-------
Watershed Character!zations
The physical system, for model purposes, consists of the water
conveyance and storage systems and the physical characteristics of the
watershed itself (e.g., proportions of pervious and impervious sur-
face areas). To facilitate calculation of actual runoff, the water-
shed can be divided into any desired number of subwatersheds or subareas,
These subareas are numbered in downstream sequence as shown in Figure 3.
Subarea
Subarea' -^^ Subarea 3 .,•
SOUTH
DRIVE -
SYSTEM
Aberdeen
Proving
Ground
JAMES
STREET
SYSTEM
Watershed Boundaries
— Subarea Boundaries
"•"•- Storm Drainage System
Railroad
FIGURE 3
TYPICAL STORM DRAINAGE SYSTEM SCHEMATIC
Only the main sewer (or natural drainage element) is considered,
and wherever a tributary joins a main drainage stem, the subarea number-
ing system jumps to the upper extreme of the tributary. The numbering
system then proceeds in a downstream order to the next junction, where
the process is repeated. Up to three incoming drainage elements are
allowed to combine at any one junction, but only one outgoing element is
accepted. Sewer overflow is assumed to proceed parallel with the desig-
nated drainage element to the next subarea outlet, with a travel time
equal to a specified multiple of the sewer travel time.
-------
Infiltration Calculations
Infiltration losses are estimated by a manipulation of the Soils
Conversation Service (3) runoff equation into the form
AF =
(P-IA+S)2
in which:
t AF andAP are infiltration and precipitation increments
in inches or millimeter per unit time interval.
• S is the soil water storage capacity in the same units
as F and P, as determined by the Soils Conservation Ser-
vice (SCS) method.
• P is cumulative precipitation since the beginning of the
storm.
• IA is the initial abstraction, assumed to be equal to
0.1S, in contrast to the SCS assumption that I A = 0.2S.
This approach for infiltration calculation was preferred over the
traditionally used Morton (4) equation, .principally because the Horton
equation depends on infiltration or permeability parameters which cannot
be quantified without specific field tests, whereas the SCS parameters
are obtained from data which are mapped with some degree of consistency.
Various possible alternatives of the SCS-based infiltration estimating
routine are described by Aron et al (5).
Overland Flow Calculations. Overland flow is computed by the
approximate kinematic wave routing method, which makes runoff a func-
tion of accumulated water depth on a subarea. The technique estimates
average depth by balancing the water budget, accounting for rainfall,
inflow and outflow as well as infiltration and initial losses. The
continuity equation and Manning flow equation are then solved simul-
taneously at each calculation time step by a simple iterative method to
determine depths of runoff for an area. An example of this technique
is presented in Figure 4.
Drainage System Flow Calculations
The model routes the runoff hydrographs through the storm drainage
system in a very simple, straightforward manner, The time that it takes
for water to move from one drainage point to another through the storm
drainage system is considered to be divided into a number of discrete
steps. The specific number of steps for a given drainage system ele-
ment is a function of the travel time in the element (e.g., a pipe or
-------
= f(y)bythe
Manning Equation
Flow Depth, y, in feet
RUNOFF DEPTH CALCULATION PROCEDURE:
1.
2.
3.
4.
Starting depth Y^ = Y0 + Precipitation excess (PE)/2 is
selected as a first approximation, from which a runoff
value (Qj) can be computed using the Manning Equation.
Qj is used in Y = YQ + (PE - Q)/2 to compute depth Y2>
A third approximation Y- is made by averaging Y- and Y?
and the cycle is repeated by computing Q
A final depth of runoff is converted into a runoff rate
in cfs.
FIGURE 4
EXAMPLE OF ITERATIVE DEPTH-AVERAGING TECHNIQUES FOR
DETERMINING RUNOFF DEPTH
10
-------
swale) and the time increment being used in the calculations. For each
time increment, flow moves through the pipe by one step, continuing
until it leaves the pipe and combines with either overland flow at a
downstream subarea or pipe flow from a tributary. This process is re-
peated until all flow leaves the watershed.
ANALYSIS OF SUBAREA PEAK FLOW TIMING
As was previously pointed out, the Penn State Runoff Model was de-
veloped to analyze and present in an easily-interpreted manner the char-
acteristics of the timing of subarea flow contributions. That is, a
total runoff volume at any particular point in the watershed drainage
system, and at any particular time, is merely the sum of individual flow
contributions from all the subareas draining to that point. The model
was developed to have the capability of illustrating the specific char-
acteristics of these subarea flow contributions, given that the program
itself had to calculate this information as part of its normal runoff
calculation routines. The particular model output which presents this
information is referred to as the Peak Flow Presentation Table.
The main function of the Peak Flow Presentation Table is to display
the individual runoff contributions from upstream subareas to a chosen
flood-prone location, including the timing of such peak flow contribu-
tions. The flow rates presented in the Peak Flow Presentation Table
reflect the travel time in the drainage system from an individual sub-
watershed to the particular point of ••'"tsrest downstream. This point
of interest can be a point of :!:__, ^ed stormwater flooding, or it can
be any point in the drainage system where the analysis of the effects
of stormwater runoff is desired. Presentation and review of this table
enables the model user to see which subwatersheds are contributing the
most critical flows to a downstream point, and to spot particularly
harmful combinations of subwatershed flow rates. Thus, it points the
planner to those locations chiefly responsible for the flood problem,
and allows the strategic placement of stormwater management facilities.
Figure 5 contains a sample Peak Flow Presentation Table taken
from model output for the analysis of a ten-subwatershed system, along
with a description of the major components of the Table.
MODEL INPUT AND OUTPUT
The major operations performed by the Penn State Runoff Model are
the generation of hydrographs and the routing of these hydrographs
through the storm drainage system. Therefore, the general input data
requirements are those which define the rainfall event, the area from
which runoff will take place, and the storm drainage system that will
transport the flow through the watershed. The specific input data re-
quired for the computer program includes sub-basin areas (acres), ap-
11
-------
SU[TF
7i
/;
s
7
e
9
ID
I
ITERSHEO >/
QTjX ,8
16. 5
11.2
12.6
12.1
11.6
3.1
C.C
11AI
OUTFLOW
n.o
a.L
r.
a.
n •
0.
0.
r, .9
IHE ABOVE;
*••»* PEAK
FLOW PRESENTATION TABLE •*••• 3
•*• SUtlUATERSHEO CCNT RIBU I ION S TO PEAK FLOWS AT INLET NO. |lo|»«*
21
D.
0.
0.
a'.
D.
D .
2.
b-
TABLt
\ FLOWS AKRIVINT AT SPECIFIED
1 .1 3.9 11 .5
l.S ^.C 19. 6
1.1 1.3 17.3
,'.i ic. e iJ.j
I'.. 2 18.8 21.1
's 2J.<1 56.3 133. F
5X
•••••DESCRIPTION OF
LISIS THE FLOIiRATES FROC
FOR INDIVIDUAL SUhU« TESSHLOS ARE TRANSLATED
TRAVEL FROH
TABLE. THE
THE 0
17.1 11. t
ll.C 9.3
13.1 It.b
17.1 16.6
15. C IS.1
22.2 23.3
| 1 1 76 .?] 161.8
/
37.7
'.I
11. S
11.9
16.0
21.1
IT HE:
66
72 78
32 II 50.6 29
61 S.« 3
111 8 10.) 7
.1
.3
.«
11D 13. & 10.8
16 6 1S.1 13
22 3
20.2 18
1«7.2 131.6 125. i 109
PEAK FLOW PRESENTATION
EACH INDIVIDUAL
IN TI»E 8< IHE
KAINAGE INLET OF THE SUBUAIER5HEC TO THE
.7
.S
.6
81 90
23.5 19
2.1 1
6.0 1
8.7 7
12.5 11 5
17.1 IS «
87.0 73.1
2.2
15,
1.0
1.C
5.8
10.
1«.6
62,
TABLE*****
SUeWAURSHEO AS 11 COMBINES WITH THE FLO*
AMOUNT
POINT OF
OF TINE THAT THE FLOW
INTEREST FOR THE PEAK
SPECIFIED T1HE INTERVALS REFER TO THE POINT OF INTEREST, AND THE FLOWRATES IN
\ SURCHARGE, OR OVERFLOW,
L>
3
1 S ICi
1
HAS OCCURRED AT INLETS NO
TAKES TO
FLOW
THE
PRESENTATION
ROW
MAJOR COMPONENTS OF "PEAK FLOW PRESENTATION" TABLE:
1. Subwatershed Identification Numbers.
2. Travel Time for a Particular Subarea - amount of time that it takes for flow to travel from the particular
subarea to the point of interest.
3. Point of Interest - point in the storm drainage system for which the analysis and table are being presented.
4. Time Steps in Minutes - time periods for the hydrograph at point of interest.
5. Calculated Total Flows for the Point of Interest - Includes overland flow (Total Outflow) for the subarea at
the point of interest, plus upstream flow contributions.
6. Calculated flow for the particular subarea (ID), for the time when this flow reaches trie point of interest.
7. Indication of those subareas where surcharging has occurred.
FIGURE 5 PEAK FLOW PRESENTATION TABLE
-------
proximate land slopes and overland flow widths, percentages of impervious
area, roughness coefficients, and SCS curve numbers for the pervious
areas, as well as pipe conveyance capacities and travel times between
points in the storm drainage system. Aside from these physical data,
rainfall increments per chosen time interval must be entered for at least
one and at most twelve rain gauges.
The program output consists of
(1) A tabular list of all rainfall-runoff information, i.e., rain-
fall in inches, runoff in cubic feet per second, storage in
acre-feet, for each printing time interval selected by the
user.
(2) Cumulative volumes of each element of the rainfall-runoff
process for the particular storm event.
(3) A listing of input data used by the program.
(4) Hydrograph plots when requested.
(5) Peak Flow Presentation Tables
PROGRAM SIZE AND REQUIREMENTS
The Penn State Runoff Model contains approximately 800 Fortran IV
statements and, once compiled, requires approximately 146 K-bytes of com-
puter memory to load and 134 K-bytos to execute. No off-line or scratch
files are required. A recent execution having the following character-
istics
o 67 subareas
o 365 acres (total watershed area)
\
o Three continuation runs of 28, 13, and 26 subareas each
o Two reservoirs
required 4.5 seconds of central processing unit time and cost approxi-
mately $6.70 at a commercial service bureau. Comparisons performed
during the initial development of the model indicated that run-time costs
for the Penn State Runoff Model were significantly less than those for
either the HEC-1 (6) program or the EPA Stormwater Management Model (7).
These comparison runs were for identical watersheds, and for output that
was as similar as possible.
13
-------
APPLICATIONS AND CASE STUDY USE
The Penn State Runoff Model is beginning to be applied and used by
consulting engineering firms for various practical stormwater manage-
ment and planning studies. Original verification of the computational
routines in the model was performed using the data for the Winohocking
Watershed in Philadelphia, and the Boneyard Creek Watershed in Illinois
(1). A complete description of the development of the model, including
its application and verification is given in a Pennsylvania State
University research publication (1). A detailed description of a case
study application has been prepared for several national stormwater
management conferences (8, 9). These references should be consulted for
specific use of the model for cost-effective stormwater management evalu-
ations. Descriptions of input data needs, as well as the potential
sources of input data, are presented in (8).
A recent application of the model for analysis of stormwater manage-
ment facilities highlights is practical use for developments as well as
small to medium-size communities. The application involved the analysis
of an existing storm runoff detention pond for a medium-size residential
development. The existing detention pool was designed using' the more
traditional approach of determining the pre- and post-development runoff
for the 100-year design event, and then conservatively sizing the pond
to be able to basically contain the difference. This resulted in a
basin that had a total available volume of approximately 40 acre-feet,
of which approximatey 20 acre-feet was used for a permanent wet pond.
The Penn State Runoff Model was then used to determine if there
were more optimum locations for smaller detention basins which would
still provide for a similar degree of stormwater management. In addi-
tion to this, the model was used to simulate the operation of the pro-
posed basin during a design storm event. The results of this simula-
tion illustrated the major problem associated with traditional ap-
proaches to storm runoff analysis, which do not consider either the
timing of runoff flows or the detailed operation of a detention struc-
ture. That is, the simulated operation of the proposed basin illustra-
ted that even during the 100-year design storm, only approximately 60%
of the available, useful storage would be used. This is primarily be-
cause the level of detail in the analysis provided by the model allows
for a "finer" evaluation of the inflow/storage utilization/outflow
characteristics of the basin. This is primarily due to the relatively
short calculation time steps which can be used with the modeling
approach. A "hand calculation" analysis of the basin does not provide
sufficient detail to optimally size the basin, and therefore a signi-
ficant degree of conservatism must be used.
In addition to this evaluation of the proposed detention pond, an
analysis of alternative basin locations was performed. The flexibility
provided by the model allowed for the analysis of nearly a dozen alter-
native locations within a very short period of time. An evaluation of
14
-------
the results of these alternative analyses quickly provided information
that was used to select the most cost-effective location for a combina-
tion of two very small basins. The total volume required was approxi-
mately seven acre-feet, and the overall reduction in total watershed
peak flow was similar to that provided by the much larger basin.
This application highlights the central theme of the approach to
stormwater management which utilizes the Penn State Runoff Model. This
theme is basically that, by evaluating the characteristics of the tim-
ing of subarea flows using the Peak Flow Presentation Table, an optimum
runoff control strategy can be adopted which is significantly more cost-
effective than would be possible using more traditional calculation
techniques. This theme, and the basic approach, has been tested and
proven, and is primarily applicable to individual developments, a com-
bination of developments, or a small- to mid-size community.
SUMMARY
The Penn State Runoff Model is an effective tool for analyzing
cost-effective stormwater management systems for developed and develop-
ing areas. The model was developed in response to a need for such a
tool for use in evaluating the state-of-the-art techniques for storm-
water runoff control. A thorough verification of the program was per-
formed as part of its development, and it has been used for several
practical stormwater management applications. This paper presents an
overall description of the capabilities of the Penn State Runoff Model
and highlights its: applicability for state-of-the-art stormwater manage-
ment.
REFERENCES
(1) Aron, G., D.A. Long, A.C. Miller, D.F. Lakatos, and M. J. O'Brien,
"Quantitative Implications of Urban Storm Runoff Abatement Measures,"
Research Publication No. 97, Institute for Research on Land and Water
Resources, the Pennsylvania State University.
(2) Aron, G. and David F. Lakatos, "Penn State Urban Runoff Model -
User's Manual," Research Publication No. 96, Institute for Research on
Land and Water Resources, the Pennsylvania State University, University
Park, Pennsyvlania, December 1976.
(3) Soil Conservation Service, "National Engineering Handbook, Sec-
tion 4, Hydrology," U.S. Department of Agriculture, Washington, D.C.,
1972.
(4) Horton, R.E., "An Approach Toward a Physical Interpretation of
Infiltration Capacity," Proceedings Soil Science Society of America,
Volume 5, pp 399-417, 1940.
15
-------
(5) Aron, G., A. C. Miller and David F. Lakatos, "An Infiltration For-
mula Based on SCS Curve Number," A.S.C.E., Journal of the Irrigation
and Drainage Division, Volume 104, No. IR1, February, 1978.
(6) Hydrologic Engineering Center, "Flood Hydrograph Package, HEC-1,"
U.S. Army Corps of Engineers, Davis, California.
(7) Environmental Protection Agency, "Stormwater Management Model"
(Four Volume Set), Water Pollution Control, 11024, DOC07-10 (with up-
dates, 1977), Washington, D.C.
(8) Lakatos, David F., "Analysis of Urban Storm Drainage Systems Using
the Penn State Runoff Model - A Case History," Satterthwaite Associates,
Inc., Publication for an Urban Hydrology short course at the Penn State
University, 1979.
(9) Lakatos, David F., ,and K. C. Wiswall, "Storm Sewer Systems Analysis
Using the Penn State Runoff Model: A Case Study," presented at the
American Society of Civil Engineers Annual Convention, Chicago, Illinios,
October 1978.
16
-------
COMPUTER MODELING FOR WATERSHED MANAGEMENT IN NORTHERN VIRGINIA
by
12
Hugo A. Bonuccelli , John P. Hartigan , and David J. Biggers
Introduction
The basinwide impacts of urbanization are seldom adequately analyzed
prior to actual development. As a result, corrective measures to address
such impacts are often reparative rather than preventive in nature. The
problem is even more difficult in watersheds which traverse several
jurisdictions, each of which might typically administer its own land use
planning and stormwater management programs with little regard for
impacts which cross political boundaries. Local jurisdictions are seldom
equipped with either the analytic tools or the institutional mechanisms
required to address runoff-related problems which originate outside of
their boundaries, even though the benefits of local stormwater management
programs may be jeopardized or negated by stormwater management
activities (or the lack therof) in neighboring jurisdictions.
As the regional planning agency for the Virginia portion of the
Washington, D.C. metropolitan area, the Northern Virginia Planning
District Commission (NVPDC) provides administrative and technical support
for a number of multijurisdictional watershed management programs. These
programs utilize both single-event and continuous simulation models as
impact assessment tools to evaluate alternative solutions/ to basinwide
problems.
This paper describes the development and application of the computer
models used for this purpose, as well as the institutional frameworks and
procedures under which ongoing impact assessment studies are performed.
Accomplishments of the management programs to date are also summarized.
The Four Mile Run Watershed Management Program
Background
As shown in Figure 1, the Four Mile Run Watershed encloses portions
of two counties and two cities that are located in the Virginia suburbs
of Washington, D.C. As a result of intensive suburban development which
occurred within the 19.5 sq mi watershed following World War II, much of
the basin's natural drainage system was replaced by an elaborate storm
sewer network, which was designed to transport storm flows downstream as
Chief, Environmental Systems Engineering Section, Northern Virginia
Planning District Commission, (NVPDC), 7309 Arlington Boulevard,
Falls Church, VA 22042
o
Director, Regional Resources Division, NVPDC
3Water Resources Engineer, NVPDC
17
-------
FIGURE 1
00
GENERALIZED MAP OF FOUR MILE RUN WATERSHED
f' \ •? ': FAIRFAX CO.
/ FALLS V j £_ -
( cmmtHST
LEGEND
— --•— Watershed Boundary
Jurfsdictiona! Boundary
_».. USAGE Flood Control Project
-O- NWS Recording Raingage at National Airport
® Main Stem USGS Stream Gage at Alexandria
WASHINGTON O.C.
0 2000 1(000
jaJ 1=
feet
FIGURE 2
WATERSHED DESIGN STORM
10
CO
z
bJ
<
u.
<
a:
peak 7.64 inch/hour
2 3
TIME (hours)
-------
quickly and efficiently as possible. Approximately 37% of the land
surface within the watershed is currently blanketed with impervious
cover.
Residential and commercial areas located near the mouth of Four Mile
Run sustained property damages totalling more than $40 million as a
result of seven floods between 1963 and 1975. The periodic flash floods
in this area were attributed to the cumulative impacts of sewered urban
development in the basin's four jurisdictions.
In March 1974, Congress authorized a $50 million U.S. Army Corps of
Engineers' (USAGE) flood control project that is designed to provide
protection from the 100-year streamflow event. The location of the flood
control project is shown in Figure 1. In order to qualify for the USAGE
channelization and bridge replacement project, the Four Mile Run
jurisdictions—Fairfax County, Arlington County and the cities of
Alexandria and Falls Church—have been required by Congress to develop
and implement a basinwide stormwater management program. This
prerequisite is intended to assure that runoff from future urban
development in the watershed does not produce streamflows which could
impair the effectiveness of the federal flood control improvements. It
is the first case in the history of USAGE flood control projects in which
a basinwide stormwater management stipulation has been attached to the
commitment of federal funding.
In April, 1974, the Four Mile Run jurisdictions agreed to develop
such a program. NVPDC was requested to provide coordination and
technical support. A nine-member Technical Advisory Committee (TAG)
composed of public works engineers and planners from the four
jurisdictions in the watershed was formed by NVPDC to guide program
development.
A two-year study was required to produce a basinwide stormwater
management program that satisfied the Congressional requirement. Initial
stages of the NVPDC study focused on the technical, rather than
institutional requirements of a multijurisdictional stormwater management
program. Following the completion of the necessary technical analyses
and the development of required impact assessment tools, an institutional
mechanism for implementing the stormwater management program was
identified.
Development of Four Mile Run Watershed Models
After some deliberation, it was decided that the use of a
computerized mathematical model of watershed hydrology would best allow
the four jurisdictions to quantify and minimize flooding impacts of
future development. The planning tools chosen for use in the Four Mile
Run Watershed Management Program were the continuous simulation model
STORM (1) and the single-event model WREM (2). Water Resources
Engineer's, Inc. (WRE) of Springfield, Virginia was retained by NVPDC for
a two-year period to calibrate the models (3) .
STORM, a relatively simple tool utilizing a modified version of the
19
-------
rational formula to translate hourly rainfall into runoff, was used to screen
the watershed's 52-year rainfall record for critical rainstorms. Six
floods that occurred during the period 1963-1972 were used for model
calibration. Although STORM does not have any flood routing capabilities,
simulated streamflow peaks at the main stem gaging station shown in
Figure 1 were within 12% of measured values for five of the six
calibration floods. The surprisingly good results of STOPsM applications
to such a large drainage area led to the conclusion that the Four Mile
Run watershed is ideally suited to STORM because its high degree of
imperviousness and extensive storm sewer network have resulted in a very
rapid streamflow response to rainfall and a travel time from its
headwaters to the main stem streamgage of approximately one hour.
Further, since STORM relied on some of the simple equations that were
being applied by local public works departments for drainage system
design, its successful calibration helped to alleviate anxieties about
computerized hydrologic models that were shared by local public works
staff.
Since the STORM analysis indicted that no historical rainfall event
produced main stem flows with a recurrence interval in excess of 40
years, it was necessary to synthesize a 100-year design rainfall event
for use in the Watershed Management Program. The method of Keifer and
Chu (4_) was used to develop the design storm shown in Figure 2. This is
a very intense, 4-hour thunderstorm slightly skewed toward the receding
limb, assumed to occur instantaneously over the basin. This storm is
used in conjunction with model WREM to project the streamflow impacts for
the Management Program.
Since the Four Mile Run versions of WREM have been described in
earlier papers (5_,5_), only a brief summary is provided here. WREM is a
second-generation version of the USEPA Stormwater Management Model. The
model consists of three major programs which are executed sequentially:
(a) Land Use Management (LUM) Program, which converts land use into
impervious ground cover; (b) RUNOFF, which converts rainfall into surface
runoff and utilizes kinematic wave routing to develop overland flow and
flow in minor conduits; and (c) TRANSPORT, which routes flows through
major conduits by means of a numerical methods solution to the equations
of motion and continuity.
An advantage of model WREM is its sophisticated hydraulic routing
capabilities, which enable it to represent open and closed conduits of
varying cross-section as well as elements such as orifices and weirs. A
number of runoff control measures can be simulated, including wet and dry
ponds, parking lot and rooftop ponding, seepage pits, and porous
pavement.
Following an extensive data collection effort, the watershed was
divided into 179 subcatchments, averaging approximately 70 acres in size,
drained by 97 idealized RUNOFF conduits and 101 idealized TRANSPORT
channels. Three of the six floods of record were chosen for model
calibration. Simulated streamflow peaks at the main stem gaging station
on Four Mile Run compared well with field estimates, agreeing within 6.2%
for the two most recent storms. Discrepancies noted in some of the
upstream watersheds were attributed to local variations in rainfall data.
20
-------
Since the physical and hydraulic parameters which describe the watershed
did not require adjustment or tuning to produce satisfactory results, the
analysis was changed from one of model calibration to one of model
verification.
Following this exercise, model WREM was designated as the principal
planning tool for the Four Mile Run Watershed Management Program. The
April 30, 1975 land use pattern was chosen as the initial "baseline" for
the assessment of future development impacts.
Institutional Structure of Watershed Management Program
Various alternatives were considered for the institutional framework
within which the calibrated model would be utilized as a watershed
management tool. After more traditional institutions such as districts
and authorities were rejected by the TAG because of various technical and
political shortcomings, the Four Mile Run Watershed Management Program
was implemented under the "joint exercise of powers" institution (Section
15.1-21 of the Code of Virginia), which had not previously been utilized
for water resources planning and management programs in Virginia (10). It
permits two or more jurisdictions to jointly exercise any power,
privilege, or authority which they are capable of exercising
individually.
The Program formally began on March 31, 1977 with the signing of an
interjurisdictional Memorandum of Agreement that provides for assessment
of all "drainage modification projects" (i.e., any change in the
watershed's land use or drainage system) with the model WREM, and
requires implementation of corrective measures to offset any peak
streamflow increase which might exceed the capacity of the USAGE flood
control works. The Memorandum also establishes a Technical Review
Committee (TRC) composed of local public works officials, to oversee
Program operations, and a Runoff Management Board (RMB), composed of the
chief administrative officers of the four jurisdictions, to address
policy matters. The TRC meets on a quarterly basis, while the RMB meets
annually, or more often as required. Technical and administrative staff
support for the program is provided by NVDPC. Local progress in the area
of stormwater management is summarized in quarterly and annual reports
which are forwarded to the USAGE'S Baltimore District Office for review.
Impact Assessment Procedures
Once per quarter, the local jurisdictions submit standardized forms
to NVPDC summarizing all drainage modification projects (DMP's) approved
during the previous quarter. Local site plan review processes have been
altered to insure that appropriate data on DMP' s can be generated for use
in the Program. If a jurisdiction suspects that a particular project may
cause negative impacts, it can request a separate assessment, funded by
the developer, prior to local site plan approval.
Upon receipt of the summaries, NVPDC assigns each DMP to one of two
categories for impact assessment. Projects which are less than two acres
21
-------
in size are assigned to a Parameter Adjustment File. These small
projects are allowed to accumulate within one of the idealized model
subcatchments until a sufficient number are present to warrant adjustment
of the parameters describing the subcatchment in the model. Larger
projects are assessed through the use of detailed representations known
as SITE models, which are formulated as follows: (1) The subcatchment(s)
enclosing the DMP (i.e., development project plus runoff control measure)
are extracted from the watershed model. (2) A detailed SITE model of the
DMP and the residual subcatchment is developed to permit an in-depth
analysis of local conditions. (3) The extracted subcatchment(s) is
replaced in the watershed model by hydrograph(s) generated by the
detailed SITE model, and the watershed model is executed with the design
storm to define projected impacts in the flood control channel. (4) If
negative impacts are noted, the SITE model is used to evaluate the
effectiveness of alternate control measures, and downstream impacts are
checked by repeating step 3 with modified SITE model hydrographs.
Program Accomplishments
During the first 30 months of the Watershed Management Program, 223
local DMP's were reviewed. In all, 144 local DMP's have been
incorporated into the watershed model. Six SITE models have been
developed. The DMP's incorporated into the watershed model represent the
addition of 54.7 acres of impervious cover and 37 runoff control measures
providing a total of 5.6 acre-feet of detention storage.
In addition to the required quarterly review activities of the
Program, a number of other impact assessment studies have been conducted
with funding from private developers to determine the probable
effectiveness of control measures at nine DMP's. In four cases, the
downstream benefits of detention storage were documented. In four other
cases, control measures were deemed unnecessary, a determination which
resulted in a cost savings for the respective developers. As a result of
the ninth study, two of the jurisdictions are currently designing a
detention pond which will straddle a mutual boundary. A tenth impact
assessment study documented the benefits of downsizing a proposed culvert,
resulting in significant cost reductions for one of the participating
j uri sd ict ions,
In addition to these studies, a pair of major SITE model studies for
outside agencies have been completed. The first study (7), funded by the
Virginia Department of Highways and Transportation, involved the
assessment of a 4.5 mile"segment of Interstate Highway 66 which traverses
the northern section of the watershed. The 1-66 project involves the
addition of approximately 100 acres of impervious surfaces and extensive
stream channel improvements to the watershed. Initial watershed model
assessments indicated that the original 1-66 stormwater management
scheme, which included one detention pond with approximately 9 acre-ft of
storage capacity, focused primarily on control of runoff impacts in the
vicinity of the highway, and would result in adverse downstream impacts.
Following a series of watershed model evaluations of alternative runoff
control levels, it was determined that a major diversion structure and
two detention ponds with a total storage capacity of approximately 38
acre-ft were required to adequately address projected downstream impacts.
22
-------
If the Four Mile Run Watershed Management Program had not been in
existence, these supplemental controls would not have been considered.
The second study (Q) , funded by the U.S. Department of the Navy,
involved the assessment of a major stormwater detention facility at the
Henderson Hall U.S. Marine Corps station in the eastern portion of the
basin. The facility, which is to be constructed with excess excavation
material from redevelopment activities on the site, will not only provide
approximately 17 acre-feet of detention storage, but will also result in
a net savings to the Federal government of approximately $280,000, due to
reduced offsite disposal costs. Model WREM assessments indicated that
sizeable downstream peak streamflow reductions would be achieved by the
stormwater management project. The documentation of the basinwide
impacts and benefits associated with this DMP would not have been
possible had the Management Program not been in operation.
Both of these studies were further noteworthy in that neither the
Highway Department nor the U.S. Navy were signatories of the Memorandum
of Agreement under which the Program is implemented, and thus they were
not legally bound to accept the results of the model assessments.
Another special study (9) was undertaken because applications of
the watershed model during the course of Management Program operations
had revealed wide variations in peak flow impacts resulting from drainage
modifications at various points in the basin. This watershed sensitivity
study was intended to investigate such locational differences in
streamflow response. A hypothetical -20-acre DMP and stormwater detention
basin were simulated at twenty locations throughout the watershed and the
resulting flows in the flood control channel compared.
The results of this analysis indicated that, for the Management
Program design storm, the middle and upper middle portions of the
watershed are most sensitive to the addition of impervious cover. The
effectiveness of stormwater detention measures was likewise projected to
be greatest for these locations. Additions of impervious cover to areas
in the headwaters of the basin and near its mouth were projected to
result in little or no adverse peak flow impacts in the flood control
channel, while the provision of detention storage at these locations
tended to be ineffective or counter-productive.
As a result of this analysis, the watershed was divided into
stormwater management zones based on the probable effectiveness of
detention storage (9_). These zones are shown in Figure 3. Although
larger projects will still require detailed analysis with the watershed
model, these zones can be utilized by local staffs during the development
plan review process to determine whether detention storage should be
required at smaller sites.
The model assessments to date indicate that, as a result of the
activities undertaken by the Watershed Management Program thus far, the
22,500-27,000 cfs capacity of the flood control channel is not exceeded
in any idealized reach. Peak flow in the uppermost portion of the flood
control channel is projected to increase by only 10 cfs in response to
development since 1975, while peak flows farther downstream are projected
23
-------
ro
STORMWATER MANAGEMENT ZONES
JCJi?J STORAGE EFFECTIVE
££vj STORAGE LESS EFFECTIVE
liiiiiiiiil STORAGE LEAST EFFECTIVE
FIGURE 3. GENERALIZED MAP OF STORMWATER MANAGEMENT ZONES
-------
to decrease by as much as 370 cfs. For the same development condition
without stormwater runoff controls, projected increases are on the order
of 400-500 cfs for every reach of the flood control channel. The
recommended controls are thus projected to decrease flood control
channel peak flows by 450-790 cfs (9).
Conclusion
Since more detailed discussions of both the modeling and
institutional aspects of the Program have been provided elsewhere
(§.'i2/ii) ' the advantages and disadvantages of the adopted methodology
will not be described in this paper. However, it should be noted that
the incorporation of computer modeling techniques into a
multijurisdictional management framework in Four Mile Run has enabled the
participating jurisdictions to successfully address the basinwide
flooding impacts of development in the sensitive watershed on a
preventive basis, thus assuring the continued viability of the federal
flood control project.
The Occoquan Basin Nonpoint Pollution Management Program
Background
The 580 sq. mi. Occoquan River Basin, which traverses four counties
and two cities in the Virginia suburbs of Washington, D.C., is shown in
Figure 4. The 9.8 billion gallon water supply reservoir at the mouth of
the watershed was impounded in 1957 and currently serves more than
600,000 customers. It is one of the few major water supply impoundments
in the Eastern U.S. that is located downstream from an urbanizing region.
In the late 1960's, classical symptoms of cultural eutrophication
were observed in the reservoir. Following a one-year study (12) of water
quality problems in the watershed, the Virginia State Water Control Board
(SWCB) in 1971 promulgated an "Occoquan Policy" (13) for regional
wastewater management which required that the jurisdictions in the basin
replace eleven major secondary sewage treatment plants with a regional
advanced wastewater treatment (AWT) plant situated
immediately upstream from the Occoquan Reservoir. The $82 million
Occoquan AWT plant began treatment operations in early summer, 1978.
The Occoquan Policy was founded on the assumption (12) that
secondary wastewater treatment plants and agricultural runoff represented
the major sources of plant nutrients that were degrading the quality of
Occoquan Reservoir waters. Consequently, at the time of policy
promulgation, it was assumed that construction of the regional AWT plant
would not only eliminate wastewater sources of the contaminants but that
it would also reduce nonpoint pollution loadings by accelerating the
conversion of agricultural lands to suburban development.
The Occoquan Watershed Monitoring Laboratory (OWML) was created in
1972 to establish basinwide surface water quality records which could be
used to gauge the effectiveness of water quality management activities.
As a result, an extended record of runoff pollution loads, dry weather
25
-------
ro
LOUDOUN CO. /
, '
LOCATION MAP
/ BASIN BOUNOARy
/ : JURISDICTION BOUNDARY
\ / « STREAM SAMPLING STATION
RESERVOIR SAMPLING STATION
0
FIGURE 4. GENERALIZED MAP OF OCCOQUAN RIVER BASIN
-------
flow loads, and, receiving water quality has been developed for the
watershed.
The development of a regional nonpoint pollution management program
for the Occoquan River Basin originated with the formation of the NVDPC
Occoquan Study Group in mid-1973. Composed of local elected officials
and senior staff representatives of the affected jurisdictions and water
management agencies, it was charged with responsibility for formulating a
balanced water resources management program for the basin, which would
address water resources problems that were not covered by the SWCB's 1971
Occoquan Policy. The Study Group proposed a scope of work to develop
such a program which included the development of computer-based planning
tools to project the water quality impacts of futuj;e land use patterns
and control strategies.
In August 1975, a two-year NVPDC study based on the Study Group's
recommendations was approved for federal funding under Section 208 of the
1972 Clean Water Act. In January 1978, based on the results of NVPDC1s
208 Study of the Occoquan Basin and a related nonpoint pollution field
study (14) conducted by NVPDC and Virginia Polytechnic Institute and
State University (VPISSU), the Occoquan Study Group formulated a
basinwide nonpoint pollution management policy for inclusion in the
metropolitan region's 208 plan.
The 208 Occoquan policy requires the basin's jurisdictions to
develop local programs to control nonpoint pollution from future
development and established a regional nonpoint pollution management
program administered by a Policy Board composed of the chief
administrative officers of participating jurisdictions. The program
relies upon a computer-based model of the Occoquan Basin which was used
to establish nonpoint pollution loading goals for the basin's
jurisdictions and to monitor the multijurisdictional impacts of local
land use changes and associated nonpoint pollution controls.
The NVPDC 208 study which developed the Occoquan Basin Computer
Model used for impact assessments under this program relied primarily on
the U.S. Environmental Protection Agency's Non-Point Source (NFS) model
(15) and the Hydrocomp Simulation Programming (HSP) model (16).
Hydrocomp, Inc., developer of the models, was retained by NVPDC to
assist with model set-up and calibration (17).
NVPDC-VPI&SU Field Study of Nonpoint Pollution loadings
In order to determine the nonpoint pollution loadings generated by
each urban and rural-agricultural land use category, an intensive field
study (14) was carried out by NVPDC and the Civil Engineering Department
of VPI&SU" from June 1976 through May 1977. This study was a logical
extension of earlier OWML monitoring efforts (18) which had documented
the significance of nonpoint pollution loadings in the basin. Runoff
from twenty-one homogeneous watersheds ranging from 6 to 71 acres were
monitored for the following pollutants: plant nutrients, BOD, COD, heavy
metals (e.g., lead, zinc), sediment, and fecal coliforms. The study
areas included four residential categories, two commercial categories,
three agricultural categories, one construction site, and an undeveloped
watershed which served as a control area. Thirteen of the watersheds
27
-------
were located in the Occoquan Basin at installations upstream from
existing OWML monitoring stations. To project the impacts of dense urban
development that is currently being proposed for the Occoquan Basin,
eight additional monitoring stations were located in the more intensively
developed Four Mile Pun Watershed which is situated approximately 13 mi.
east of the Occoquan Basin.
The monitoring stations relied upon automatic discrete sampling
equipment that was activated by a suitable flow-measuring device at
preselected flow rates. Either a natural (e.g., ephemeral stream) or
artificial (e.g., storm sewer, H-flume) drainage control was used to
establish stage-discharge relationships at the outlet of each watershed.
Continuous recording raingages were installed in or near each study area
to develop input data for computer modeling studies of each watershed and
to permit periodic analyses of pollutant loadings released by rainfall.
During the 12-month field study, more than 300 runoff events were
monitored and 1,300 samples were analyzed for 29 constituents. Standard
statistical analyses (19) of and computer model applications (20) to the
observed data produced similar relationships between land use
characteristics and nonpoint pollution loading rates. The major
conclusions of the NVPDC/VPI & SU field study are as follows (14):
o For plant nutrients and organics, annual unit area loading
rates from urban land uses were, in general, positively related
to impervious ground cover and higher than loadings from all
rural-agricultural land uses, with the exception of cropland
(in the case of nitrogen and phosphorus). Unoxidized nitrogen
forms, which can eventually exert an oxygen demand on
downstream receiving waters, represented 70%-80% of the
nitrogen loadings in runoff from all land use categories.
o For heavy metals such as lead and zinc, annual unit area
loading rates from urban land uses were positively related to
impervious ground cover and considerably higher than loadings
from all rural-agricultural land uses.
o For urban land uses, mean dissolved loadings in runoff ranged
from 57%-73% of the total load for nitrogen, from 42%-55% of
the total load for phosphorus, and from 6%-13% of the total
load for lead. The high quantities of dissolved N and P
loadings are significant from a nonpoint pollution management
standpoint since this fraction is not only readily available to
stimulate algal blooms in downstream receiving waters, but it
is also generally not removed by stormwater detention ponds
typically used to manage peak runoff.
o Air pollution is an important source of urban nonpoint
pollution in the metropolitan Washington region. Atmospheric
contributions of plant nutrients do not appear to be dependent
on land use or distance from urban centers of poorer air
quality, due to the high levels of atmospheric mixing that tend
to exist during rainstorms in the region. However, since
highly impervious land uses will convert larger amounts of
28
-------
rainfall to runoff than will land uses of lesser
imperviousness, the significance of atmospheric loadings was
found to vary from one urban land use to the next. Estimates
of atmospheric contributions to the average annual nonpoint
pollution load from urban land uses ranged from 30%-99% for
nitrogen, from 20%-50% for phosphorus, and from 5%-10% for
lead.
NFS Set-up and Calibration
The NFS model (15) was used to derive "land use-nonpoint pollution"
relationships from the field study loading data. Nonpoint pollution
loading processes simulated by the NPS model are: (a) accumulation of
pollutants on the land surface and in the atmosphere during non-storm
periods; (b) generation of surface sediment loads during rainfall events
through raindrop impact and detachment of soil particles; and (c)
overland flow transport of pollutants that have accumulated on the land
surface and/or have been washed out of the atmosphere. Like most
computer-based urban runoff models (!_,£, l&j21), the NPS model relies upon
"dry weather pollutant accumulation rates" (Ibs/ac/day) to represent the
pollutant loading potential of urban land use classifications. Separate
pollutant accumulation rates were defined for the pervious and impervious
fractions of each urban land use category (20). Pollutant accumulation
rates were also defined for the majority of the rural-agricultural land
uses: forest land, idle land and pasture land. However, loading
projections for cropland land uses, where soil loss rather than pollutant
accumulation rate is the principal determinant of nonpoint pollution
washoff, were based upon sediment "potency factors" (i.e., ratio of
pollutant mass to sediment mass) that are multiplied by simulated
sediment yield. With the long-term record of observed nonpoint pollution
loadings and rainfall intensities collected during the field study,
NVPDC was able to use an NPS model of each of the 21 watersheds to derive
either dry weather pollutant accumulation rates or sediment potency
factors for each land use category. Following calibration of hydrology
parameters, each watershed model was executed with a five-minute rainfall
record* for the twelve-month monitoring period and assumed values for
accumulation rates and potency factors were iteratively adjusted until
reasonable agreement between simulated and observed mean concentrations
and loading rates was obtained.
Since pollutant contributions from urban lawn surfaces can vary from
storm to storm depending upon antecedent soil moisture conditions, the
pervious fraction of each watershed was extracted from the hydrology
calibration data set to facilitate the characterization of pervious area
loading rates. Separate models were set up for each fraction, and the
pervious fraction model was executed with its five-minute rainfall record
to identify those storms that produced runoff from the watershed's
pervious surfaces. Calibration of the impervious fraction model
involved iteratively adjusting dry weather pollutant accumulation rates
* A continuous record of five-minute rainfall volumes was used for
model calibration since several of the urban watersheds were
characterized by times of concentration on the order of 5-10 minutes.
29
-------
until there was reasonable agreement between simulated and observed loads
for storms that did not generate any runoff from the pervious fraction.
Differences between observed pollutant loads and simulated impervious
fraction loadings yielded calibration data for determining dry weather
pollutant accumulation rates for the pervious fraction model.
Goodness-of-fit statistics for the final "land use-nonpoint
pollution" relationships developed by the model calibration study are
summarized elsewhere (££»22_) • In addition, several other technical
papers (23_,2£,2S_,26_,2T_,28_,29) have been prepared by NVPDC and VPI&SU
investigators to facilitate applications of the "land use-nonpoint
pollution" relationships within and outside the metropolitan Washington
region.
Characterizations of Best Management Practices
In order to project the water quality improvements which might
result from the implementation of urban Best Management Practices (BMP's)
for nonpoint pollution control, it was necessary to develop techniques to
simulate the operation of such measures (30). Urban BMP's can generally
be categorized as follows:
1. Source Controls; Measures that reduce the accumulation of
available pollutant loads on the land surface and in the
atmosphere between storm events. These include land use
planning techniques, fertilizer management activities, and
street/parking lot sweeping programs.
2. Discharge Controls; Measures that detain runoff for several
hours and release it at a controlled rate to the drainage
network.
a. Detention Ponds: Online detention facilities that rely
upon Type I sedimentation processes to remove sediment and
suspended pollutants from urban runoff. Traditional
flooding/erosion control designs produce "peak-shaving"
facilities with relatively low detention times (e.g., 30
minutes) and considerable deviations from ideal quiescent
and plug flow conditions. As a result, peak-shaving
facilities achieve relatively low removal of the fine
particles which are associated with the majority of the
plant nutrients, BOD, and heavy metals in urban runoff.
The recommended multipurpose design criteria relies upon a
24-hr detention time and quiescent conditions in a
slow-release pool (50%-70% of peak-shaving storage volume)
to achieve high trap efficiencies for silt particles. The
use of subsurface drains to maintain a slow release pool
can produce additional filtration and adsorption within
the overlying soil profile. Sketches of typical outlet
structures for detention basin BMP's are shown in Figures
5 and 6. To simultaneously achieve flooding/erosion
control performance standards, detention basin storage
capacity must be increased to approximately 1.2 times the
30
-------
TOP OF FLOODING/EROSION CONTROL STORA6
CMP RISER
-------
peak-shaving level.
b. Stormwater Treatment: The addition of chemicals to
stormwater detention basins can result in the removal of
dissolved and colloidal pollutant loadings which would not
otherwise settle out.
3. Volume Controls; Measures (e.g., Dutch drains, seepage pits,
porous pavement) that reduce runoff volumes and associated
loadings by diverting rainfall excess into the soil profile.
Pollutant removal in the soil occurs through natural physical,
chemical, and biological processes documented for "land
treatment" of wastewater.
Source controls can generally be represented directly by the NPS
model. For instance, street and parking lot sweeping measures can be
simulated by modifying the model's dry weather pollutant decay parameters
for impervious surfaces. However, in order to simulate the dynamic
operation of volume and discharge controls, it was necessary to develop a
STORAGE-TREATMENT submodel (17). This submodel was designed to operate
in conjunction with the continuous output files generated by NFS,
providing relatively simple representations of the physical, chemical,
and biological unit processes that occur within stormwater management
BMP's.
Since measured data documenting the pollutant removal efficiencies
of typical urban BMP's is generally unavailable, average annual pollutant
removal rates for BMP's serving each land use category were developed by
operating the NPS and STORAGE-TREATMENT models with meteorologic data for
an "average" year. The pollutant removal efficiencies derived by these
BMP model applications are summarized elsewhere (30).
Occoguan Basin Computer Model
The Occoquan Basin Computer Model consists of 15 sub-basins (39 sq.
mi. average) represented by the NPS model linked by 12 idealized HSP
stream channels and 3 idealized HSP reservoirs. The HSP continuous
simulation model was selected to serve as the receiving water submodel
bacause of its ability to simulate pollutant transport and transformation
processes and to project long-term impacts with the use of "pollutant
concentration vs. frequency" relationships.
Model Calibration. To calibrate the NPS submodel's hydrologic
parameters, simulated and measured runoff volumes and streamflow peaks at
the Bull Run, Broad Run, and Cedar Run gages were compared for the period
October 1970 - September 1975. Verification of the hydrology calibration
results was based on comparisons of simulated and measured volumes for the
period October 1967 through September 1970 at the three calibration gages
and a fourth gage on Occoquan Creek. The calibration and verification
results were quite good. With the exception of a single year at one of
the gages, average error for the calibration period was less than 10%,
with a maximum error of less than 25%. Similar results were obtained for
the verification period. The discrepancies noted between simulated and
recorded values approximated the expected measurement errors in rainfall
and streamflow data at the respective gages.
To calibrate the HSP submodel's stream transport/transformation
32
-------
parameters and to verify the transferability of the "land use-nonpoint
pollution" relationships produced by the NVPDC/VPI&SU field study, the
NFS submodel was executed for each of the 15 sub-basins in the watershed
and the resultant output time series served as input for HSP submodel
executions. Simulated and measured water quality data were compared at
OWML monitoring stations for the period January, 1974 - December, 1976.
A pair of auxiliary computer programs were developed to aid in this
calibration/verification study. The first program was used to generate
line printer plots of observed and simulated "pollutant concentration vs.
frequency" curves for each set of calibration parameters. As a further
check, a second program which generated line printer plots and simple
linear regression equations for observed and simulated nonpoint pollution
loads at two free-flowing stream monitoring stations was developed.
After some adjustment of instream process parameters, it was concluded
that the model adequately represented land surface loadings and instream
transformations of pollutants in the basin.
A detailed summary of calibration results is presented elsewhere
Impact Assessment Studies. Long-term water quality impacts of land
development patterns and BMP strategies could be projected by using the
Occoquan Basin Computer Model to simulate nonpoint pollution loadings and
receiving water responses produced by the 28-year (1949-1976) hourly
rainfall record available at the basin's recording raingages. However,
to minimize computer costs, a twelve-month period was identified which
was characterized by simulated hydrologic and water quality statistics
most typical of the statistics associated with the entire 28-year record
(17) . This "average" year (January 1976 - December 1976) was identified
by operating the model for each year in the 28-year meteorologic record
and comparing the annual statistics with the 28-year statistics.
Projected hydrologic and water quality responses produced by operating
the computer model with the average year meteorologic records are assumed
to have statistics (e.g. , mean pollutant concentrations, concentration
vs. frequency relationships) that approximate those associated with the
entire 28-year period.
Procedures for assessing the basinwide water, quality impacts of various
land use and water quality management decisions have been developed for
use in conjunction with the Occoquan Basin Computer Model. Land use
changes can be assessed by altering the NPS input files and re-executing
the NPS and HSP submodels. Evaluations of structural BMP's require the
development of individual land use submodels for each land use for which
the BMP's are analyzed, so that the STORAGE-TREATMENT and HSP submodels
can be executed with appropriate NPS washoff files. The effects of
alternative wastewater discharges or streamflow diversions can be
investigated by altering the appropriate point source or diversion files
and re-executing the HSP portion of the model.
To derive a benchmark for assessments of urban BMP strategies, the
Occoquan Basin Computer Model was used to compare the water quality
impacts of existing and future land use patterns, assuming a year of
average wetness and AWT discharges in accordance with the 1971 Occoquan
Policy. The modeling study indicated that an uncontrolled Year 2005 land
33
-------
use pattern can be expected to produce a 16.1% increase in annual total
phosphorus loadings delivered to the Occoquan Reservoir, a 15.0% increase
in BOD loadings, and a 187.0% increase in lead loadings (31). Nonpoint
pollution loadings of phosphorus were shown to be capable of producing
eutrophic conditions, even in the absence of wastewater treatment plant
loadings. The sizeable increase in lead loadings could produce higher
accumulations of lead within Occoquan Reservoir bottom sediments which
could have an adverse effect on aquatic life. In short, these modeling
studies have demonstrated that high levels of wastewater treatment alone
will not eliminate water quality problems in the Occoquan Basin and have
confirmed the need for a balanced approach to water quality management in
the watershed.
Separate model executions were carried out to characterize the
benefits of each nonpoint pollution control strategy (31). Since the
model runs assumed that these BMP's would only be applied to future urban
development, BMP benefits are best viewed in terms of reductions in
increased pollutant loadings associated with additional urbanization.
The conclusions which were drawn from the BMP modeling studies are as
follows (31): (a) traditional stormwater management techniques such as
volume controls and detention basin controls (with modified design
criteria) should be capable of maintaining Year 2005 nonpoint pollution
loadings and ambient water quality at levels equivalent to or in the
vicinity of existing conditions, in effect achieving nonpoint pollution
management benefits that are reasonably close to those associated with
land use controls which would minimize future development in the Occoquan
Basin; (b) the adoption of traditional urban stormwater controls for
nonpoint pollution management programs will involve only a 10%-20%
increase in the total cost of a detention basin facility and no change in
the total cost of a volume control facility; (c) although stormwater
treatment BMP's promise to achieve the greatest reductions in nonpoint
pollution loadings, the water quality benefits associated with this
control measure do not appear to be great enough to offset its extremely
high costs; (d) in light of (a), (b), and (c), traditional urban
stormwater management BMP's appear to represent a much more
cost-effective approach than urban stormwater treatment and a viable
alternative to land use controls for the Occoquan Basin; and (e) in
conjunction with the application of multipurpose stormwater management
BMP's to future urban development, adoption of urban source control BMP's
and rural-agricultural BMP's can be expected to produce nonpoint pollution
loadings and receiving water quality impacts which are even lower than
existing conditions.
Institutional Structure of Watershed Management Program
Based on the conclusions of NVPDC's 208 planning study, the 208 plan
for the Occoquan River Basin provides for the establishment of a
basinwide nonpoint pollution management program to supplement the
benefits of the basin's wastewater management program. In striving to
control water quality problems which are not addressed by wastewater
treatment plants, the basinwide nonpoint pollution management program has
as its goal: (a) the implementation of the most cost-effective nonpoint
pollution mitigation techniques during the early stages of urbanization,
so as to minimize the risk of irreversible water quality degradation
34
-------
and/or the need for costly remedial control measures at some later date;
and (b) the management of nonpoint pollution loadings from agricultural
lands within the basin. The management program is strictly advisory in
nature, and as such, it is primarily a vehicle for fostering
interjurisdictional cooperation, for providing continuing technical
assistance to local staffs, and for monitoring local progress in the area
of nonpoint pollution management.
The basinwide management program was established in November 1978.
It is administered by a Policy board which is composed of the chief
administrative officers of participating local governments and is advised
by a special technical committee. Technical and administrative staff
support for the management program is provided by NVPDC.
According to the provisions of the 208 plan, the Policy Board meets
regularly to review local nonpoint pollution management activities, to
monitor associated water quality changes with the Occoquan Basin Computer
Model, to comment on the adequacy of local nonpoint pollution management
efforts, to prepare quarterly reports summarizing local progress in the
area of nonpoint pollution management, to review water quality data
collected by monitoring agencies to determine if changes in basinwide
water quality targets are warranted, and to adopt an annual operating
budget for the areawide nonpoint pollution management program. The
quarterly reports on local progress are forwarded to the governing boards
of participating jurisdictions and agencies, the State Water Control
Board, and the U.S. Environmental Protection Agency for review. The
basinwide nonpoint pollution goal that has been established to gauge the
progress of local management programs is minimal deterioration in surface
water quality, as forecast by the Occoquan Basin Computer Model.
Program Accomplishments to Date
In the fourteen months that have passed since the basinwide nonpoint
pollution management program was begun, activities have focused on the
development of nonpoint pollution planning tools for local staff
applications. The NVPDC staff has formulated an Urban BMP Guidebook (30)
which outlines estimated BMP efficiencies and cost-effectiveness
relationships for alternative design criteria. In addition, NVPDC has
assisted local staff with the review of urban development proposals, the
formulation of urban BMP recommendations, the evaluation of alternative
local frameworks for nonpoint pollution management, and the
identification of the most appropriate agricultural BMP's for the
Occoquan River Basin.
Some jurisdictions have already made considerable progress in
implementing nonpoint pollution controls. The Fairfax County Board of
Supervisors has recently agreed to incorporate urban BMP requirements
into the County Public Facilities Manual so that BMP's can be required
for all new development that occurs in the Occoquan Basin. Fauquier
County has relied upon its subdivision regulations to require
comprehensive BMP plans for several major single family developments; in
addition, the County is constructing seven major flood control/water
supply impoundments, funded by the U.S. Department of Agriculture's
PL-566 program, that are projected to achieve substantial nonpoint
35
-------
pollution management benefits as well. The City of Manassas relies upon
a vacuum street-sweeping program which covers existing as well as new
development. By late Winter 1979, it is anticipated that all
participating jurisdictions will have selected an approach for
institutionalizing urban nonpoint pollution management programs.
Conclusion
The Occoquan Basin Nonpoint Pollution Management Program allows the
participating jurisdictions to determine cost-effective solutions to
severe water quality problems which traverse political boundaries. The
calibrated computer model is a state-of-the-art planning tool that can be
used to forecast the basinwide nonpoint pollution impacts associated with
various hydrologic conditions and to compare the benefits of alternative
water quality management approaches. As with the Four Mile Run Watershed
Management Program, the Occoquan Program enables the participating
jurisdictions to successfully address the basinwide impacts of
development on a preventive basis, thus fulfilling the goals of both the
local jurisdictions and Section 208- of the Clean Water Act.
Other Computer Modeling Activities
Three other water quality modeling studies undertaken by NVPDC are
worthy of note. All three studies rely upon the "soil texture-hydrologic
model parameter" relationships, "land use-nonpoint pollution"
relationships, and water quality models developed during NVPDC's Occoquan
Basin study.
One of these, undertaken under a 208 planning contract with the
Metropolitan Washington Council of Governments (MWCOG), utilized computer
simulation techniques to rank the metropolitan region's watersheds in
terms of nonpoint pollution contributions to the Potomac Estuary. NPS
was used to generate "land surface washoff" (LSWO) files for various land
use-soil type combinations (32). These LSWO files were then used to
calculate weighted runoff loading files for each of 20 watersheds
tributary to the free-flowing portion of the Potomac. HSP was then set
up on the Potomac River and used to route the nonpoint pollution loads to
the Potomac Estuary. By removing the washoff loads generated by each
watershed one at a time, an estimate of the relative nonpoint
contribution of each tributary area was obtained. Watershed
contributions were ranked and the tributary areas in the top quartile
were designated for intensive nonpoint pollution management
investigations.
A second study, currently underway, is developing 208 Watershed
Management Programs patterned after the Occoquan Basin Nonpoint Pollution
Management Program for three other multijurisdictional watersheds—Goose
Creek, Broad Run, and Sugarland Run. Watershed models (33), relying upon
the NPS and HSP submodels, will serve as the principal planning tool for
management studies of the urbanizing 490 sq mi study area. This
watershed management study will be completed in July 1980.
A final modeling study (34), funded by the Virginia State Water
36
-------
Control Board (SWCB), will produce nonpoint pollution loading projections
for 11 watersheds (886 sq mi total) tributary to Potomac Estuary
embayments. NFS and HSP submodels will be set up for each watershed.
Following model calibration, the watershed models will be used by NVPDC
to simulate monthly, seasonal, and annual loadings on the respective
embayments, which serve as the interface between the region's
free-flowing streams and the Potomac Estuary. The results of the
watershed modeling study will be used by the SWCB to determine whether
local nonpoint pollution management programs within the 11 watersheds
might justify a relaxation of effluent standards at four advanced
wastewater treatment (AWT) plants which discharge to Potomac embayments.
Thus, the NVPDC study could potentially affect capital and O&M costs at
AWT plants in four member jurisdictions. This watershed modeling study
will be completed in December, 1980.
Summary and Conclusions
Computer modeling techniques have been successfully incorporated
into a number of regional water resources management programs in Northern
Virginia. The modeling results have not only been successful from a
technical viewpoint, but the models themselves have been accepted as
impact assessment tools in ongoing planning programs. This acceptance
has enabled the participating jurisdictions and agencies to view the
watersheds as integrated systems for the first time. This allows them to
quantify and address basinwide impacts rather than approach water
resources problems in the piecemeal, after-the-fact fashion which had
previously characterized their efforts. It further helps to insure that
cost-effective, balanced stormwater management programs can be
implemented before the problems become economically irreversible.
37
-------
References
1. U.S. Army Corps of Engineers, Hydrologic Engineering Center, "Urban
Storm Water Runoff: STORM", Generalized Computer Program 723-58-
L3520, Davis, California, August, 1975.
2. Water Resources Engineers, Inc., "San Francisco Stormwater Model:
User's Manual and Program Documentation," Dept. of Public Works, City
and County of San Francisco, Cal., 1972.
3. Water Resources Engineers, Inc., "Four Mile Run Watershed Runoff
Control Program," prepared for Northern Virginia Planning District
Commission, Falls Church, Va., Dec. 1976.
4. Kiefer, C.J. and Chu, H.H., "Synthetic Storm Pattern for Drainage
Design," Journal of Hydraulics Division, ASCE, Vol. 83, No. HY4,
August, 1957, pp. 1-25.
5. Hartigan, J.P., and Bonuccelli, H.A., "Management of Urban Runoff in
a Multi-Jurisdictional Watershed," Proceedings of the 1977 Interna-
tional Symposium on Urban Hudrology, Hydraulics, and Sediment Control,
Report UKYBU14, College of Engineering, University of Kentucky,
Lexington Ky., Dec., 1977, pp. 27-41.
6. Bonuccelli, H.A., and Hartigan, J.P., "Assessing Land Use Changes
with an Urban Runoff Model," Verification of Mathematical and Physical
Models in Hydraulic Engineering, American Society of Civil Engineers,
New York, N.Y., Aug., 1978, pp. 827-835.
7. Northern Virginia Planning District Commission, "Impacts of the 1-66
Highway Project on Streamflows in the Four Mile Run Watershed," pre-
pared for Virginia Dept. of Highways and Transportation, Richmond,
Va., December, 1978.
8. Northern Virginia Planning District Commission, "Impacts of Stormwater
Detention at Henderson Hall on Streamflows in the Four Mile Run Water-
shed," prepared for U.S. Navy Dept., Chesapeake Division, Naval
Facilities Command, Washington, D.C., June, 1979.
9. Northern Virginia Planning District Commission, "Four Mile Run Water-
shed Management Program: Annual Report," prepared for Annual Meeting
of Four Mile Run Runoff Management Board, October, 1979.
10. Hartigan, J.P. and Bonuccelli, H.A., "Joint Exercise of Powers: A
Tool for Stormwater Management," Legal, Institutional, and Social
Aspects of Irrigation and Drainage and Water Resources Planning and
Management, American Society of Civil Engineers, New York, N.Y.,
July 1978, pp. 442-456.
38
-------
11. Bonuccelli, H.A., Hartigan, J.P., and Biggers, D. J. , "Multijurisdic-
tional Stonnwater Management: The Four Mile Run Watershed Management
Program," Proceedings of National Conference on Urban Erosion and
Sediment Control; Institutions and Technology, held at St. Paul,
Minn., October 10-12, 1979, (IN PRESS).
12. Metcalf and Eddy, Inc., "1969 Occoquan Reservoir Study," prepared
for Virginia State Water Control Board, Richmond, Va., 1970.
13. Virginia State Water Control Board, "A Policy for Waste Treatment
and Water Quality Management in the Occoquan Watershed," Richmond,
Va., 1971.
14. Northern Virginia Planning District Commission and Virginia Polytechnic
Institute and State University, "Occoquan/Four Mile Run Nonpoint Source
Correlation Study," Final Report prepared for Metropolitan Washington
Council of Governments, Wash., D.C., July 1978.
15. Donigian, A.S. and Crawford, N.H., "Modeling Nonpoint Pollution
from the Land Surface," EPA-600/3-76-083, U.S. Environmental Protection
Agency, Washington, D.C., July, 1976.
16. Hydrocomp, Inc., "Hydrocomp Simulation Programming: Water Quality
Simulation Operations Manual," Palo Alto, Cal., 1977.
17. Hydrocomp, Inc., "The Occoquan Basin Computer Model: Calibration,
Verification, and User's Manual," prepared for NVPDC, Falls Church,
VA, May 1978.
18. Occoquan Watershed Monitoring Laboratory, "Summary Edition of the
Occoquan Quarterly: 1973-1976," Vol. IV, No. 4, Spring 1976.
19. Smullen, J.T., Hartigan, J.P. and Grizzard, T.J., "Assessment of
Runoff Pollution in Coastal Watersheds," Proceedings of Symposium
on the Technical, Environmental, Socio-Economic, and Regulatory
Aspects of Coastal Zone Management, American Society of Civil
Engineers, New York, N.Y., March, 1978, pp. 840-857.
20. Hartigan, J.P., et al., "Calibration of Urban Nonpoint Pollution
Loading Models," Verification of Mathematical and Physical Models
in Hydraulic Engineering, American Society of Civil Engineers,
New York, N.Y., August, 1978, pp. 363-372.
21. Metcalf & Eddy, Inc., Univ. of Florida, Water Resources Engineers,
Inc., "Stormwater Management Model," Vol. I, 11024DOC07/71, U.S.
Environmental Protection Agency, Wash., D.C.,'1971.
22. Northern Virginia Planning District Commission, "Occoquan Basin
Computer Model: Summary of Calibration Results," Falls Church,
VA, January, 1979.
39
-------
23. Randall, C.W., Helsel, D.R., and Grizzard, T.J., "The Impact of
Atmospheric Contaminants on Stormwater Quality in an Urban Area,"
Journal of Water Pollution Control Federation, (In Press).
24. Hartigan, J.P., et al., "Planning for Nonpoint Pollution Impacts,"
Dynamic Planning for Environmental Quality in the 1980' s, American
Society of Civil Engineers, New York, N.Y., 1978, pp. 152-191.
25. Griffin, D.M., et al. , "An Examination of Nonpoint Pollution Export
from Various Land Use Types," Proceedings of International Symposium
on Urban Stormwater Management, Report No. UKYBU116, University of
Kentucky, Lexington, Ky., July, 1978, pp. 119-127.
26. Hartigan, J.P., et al., "Watershed Planning for Flood Control and
Runoff Pollution Management," New Directions in Century Three;
Strategies for Land and Water Use, Soil Conservation Society of
America, Ankeny, Iowa, 1977, pp. 109-117.
27. Grizzard, T.J., and Hartigan, J.P., "Characterizing 'Runoff Pollution-
Land Use" Relationships in Northern Virginia's Occoquan Watershed,"
Proceedings of the AMSA-MSDGC Workshop on Water Quality Surveys
for 208; Data Acquisition and Interpretation of Nonpoint Runoff,
Metropolitan Sanitary District of Greater Chicago, Chicago, IL. 1977,
pp. F-l—F-54.
28. Grizzard, T.J., et al., "Assessing Runoff Pollution Loadings for
208 Planning Programs," paper presented at ASCE National Environmental
Engineering Conference, held at Nashville, Tennessee, July 13-15, 1977.
29. Hartigan, J.P., et al., "Urban Land Use Characteristics and Runoff
Pollution Loadings, Timing, and Solubility," paper presented at Annual
Spring Meeting of American Geophysical Union, held at Washington, D.C.,
May 30-June 3, 1977.
30. Northern Virginia Planning District Commission, "Guidebook for Screening
Urban Nonpoint Pollution Management Strategies," prepared for Metropoli-
tan Washington Council of Governments, Washington, D.C., November, 1979.
31. Northern Virginia Planning District Commission, "Assessments of Water
Quality Management Strategies with the Occoquan Basin Computer Model,"
prepared for Occoquan Technical Review Committee, January, 1979.
32. Northern Virginia Planning District Commission, "Nonpoint Pollution
Modeling Procedures for 208 Planning," prepared for Metropolitan Washing-
ton Council of Governments, Washington, D.C., January, 1979.
33. Northern Virginia Planning District Commission, "Watershed Model for 208
Planning Studies of Goose Creek, Broad Run, and Sugarland Run Watersheds,"
prepared for Goose-Broad-Sugarland Technical Advisory Committee, July, 1979.
34. Northern Virginia Planning District Commission, "Work Program for
Modeling Study of Nonpoint Pollution Loadings from Potomac Embayment
Watersheds," prepared for Virginia State Water Control Board,
Richmond, Va., May, 1979.
40
-------
COMPARISON OF DESIGN PEAK FLOWS CALCULATED 3Y
THE RATIONAL METHOD AND THE EPA-SWM MODEL
by
Dr. Paul Wisner*
Atef Kassem**
Philip Cheung**
"...us should not readily ridicule our.
tested knowledge and accumulated ax~
variance in face of any new illusion,
but we should seek its basic concept,
if there is one, and then adopt and
improve it to the best for our purpose.
7en Te Chow CD
1. INTRODUCTION
The limitations of the Rational Method (RM) have been
known since early in its development, and have been extensive-
ly reviewed in several recent studies (2, 3, 4). A review of
Canadian practice indicated a wide discrepancy in the selec-
tion of parameters C, and the inlet time tj_ used in conjunc-
tion with the Rational Method by different municipalities (2)
Despite the development and widespread application of
urban hydrology models, many drainage engineers continue to
use the RM. A review of storm drainage design methods, pub-
lished in 1976 by the Hydraulic Research Station in Walling-
ford, England (4), concluded that the use of the Rational
Method should continue in the U.K., at least for a limited
period of. time.
Several recent Canadian drainage manuals recommend the
use of hydrologic models and accept with or without limita-
tions the RM (5, 6, 7, 3).
The parallel use of the Rational Method and more sophis-
ticated models may lead, however, to discrepancies. From the
practitioner's point of view, it is of particular importance
to establish whether the RM would give results which are at
least consistent with a more complex model such as SWMM. A
* Professor, Department of Civil Engineering, University of
Ottawa, Ottawa, Ontario, Canada, KIN 9B4.
** Research Associates, Department of Civil Engineering,
University of Ottawa, Ottawa, Ontario, Canada, KIN 934.
41
-------
question which also often arises is whether the Rational
Method gives results which are conservative/ i.e., whether
the design of storm sewers by the Rational Method has a higher
safety factor. Another question is that of the maximum drainage
area for which the RM can be applied. Several comparisons of
the RM and hydrologic models including SWMM were based on simu-
lations of measured flows, most of which smaller than those con-
sidered for design conditions (2, 4) .
Computation of "design flows" by means of a. model requires
the input of "design storms" (real or hypothetical) . A compari-
son of the RM with SWMM used in conjunction with design storms
was therefore considered as being of interest and was included
in a research program conducted at the university of Ottawa.
This paper presents for discussion at the SWMM Users Meet-
ing the preliminary results.
2. PARAMETERS IN THE RATIONAL METHOD
The Rational Method for peak runoff determination is ex-
pressed by the formula: Q - CiA; where Q is the peak rate of
runoff (cfs), C is the runoff coefficient, i is the design rain-
fall intensity (in/hr) for a duration which equals the time of
concentration, tc, and A is the tributary area (acres).* The
time of concentration is given by the sum of travel time and
"inlet time", t.j_. For a given intensity-duration-frequency curve
and pipe system we may assume that i is an input and tj_ a para-
meter. This section summarizes various methods for the selec-
tion of the Rational Method parameters C and tj_. More detailed
reviews of the Rational Method can be found in references (2,
3, 4).
The runoff coefficient C lumps the effect on runoff of
various factors such as impervious ness ratio, infiltration los-
ses, depression storage, basin slope, ... etc. The simplest and
most common way of selecting C is by relating it to land use
characteristics of the basin. Tables which give C for various
surface types and "area descriptions" can be found in the
ASCE/WPCF design manual (9) . More recently, attempts have been
made to determine the runoff coefficient in terms of impervious-
ratio which is a typical parameter in many of the models (5, 19).
A simple formula of this kind is a weighed relation of the type
CX Aper * Ci X A
-»• A.
per imp
. ?erv per imp imp
A -»• A.
*In the metric system a conversion factor should be included.
For example, for Q in m3/s, i in mm/hour and A in ha in order
to maintain the familiar C values, the RM becomes Q = kCiA
with k = 0.29.
42
-------
where
Cperv is the C value for the pervious area,
Ciinp is the C value for the impervious area.
Attempts have also been made to use a more realistic represen-
tation of runoff processes.
One_direction (11, 12) considers the effect of rainfall
intensities on the magnitude of abstractions. It assumes that
for higher intensity rainfalls the effective rainfall will in-
crease, that is the runoff coefficient will increase.
A simple relation was also developed which adjusts the
runoff coefficient by multiplying it by a frequency factor
"Cf". For more frequent storms (2-10 years frequency) Cf = 1.0,
for 1/25 years storms, C^ = 1.1 and for 1/100 year storm,
Cf = 1.25 (13).
Another direction is related to the fact that rainfall
abstractions will decrease with time, resulting in an increase
in the runoff coefficient. Graphs expressing C in terms of the.
time from beginning of rainfall based on the rates of infil-
tration can be found in the ASCE/WPCF design manual (9) and
(14, 15, 16). An empirical relation, developed by Mitci and
applied in Montreal, is given in Figure 1.
In most applications, a flat allowance is made for the
"inlet time", tj_. In a review of U.S. and Canadian practice (4) ,
it was found that the RM has been applied using inlet times
varying between 5 minutes and 15 minutes (2, 17, 19). These
values are larger than those typical for U.K. practice which
range between 1 minute (19) and 4 minutes (20). Various for-
mulae for calculating the inlet time or overland flow time
by Izzard, Kirpich, Kerby, Seelye and others were compared with
a relation based on the kinematic wave (3) and the results in-
dicate significant discrepancies (Table 1).
3. DIFFICULTIES IN COMPARING THE RATIONAL
METHOD AND THE SWM MODEL - PREVIOUS
STUDIES
A major difficulty in comparing the performance of the
RM with other models is more or less due to the subjective
selection of the RM parameters. Another reason why a direct
comparison is not possible is because of the different type
of input.
The Rational formula does not take into account the time
distribution of rainfall, and assumes a uniform rainfall.*
"To account for variable rainfall intensities, the
applied in conjunction with a time-area or isochrc
*To account for variable rainfall intensities, the RM should be
rone procedure.
43
-------
This "block rainfall" can be obtained from intensity-frequency-
duration curves, which are readily available.
Models or hydrograph methods, however, require as an in-•
put a "storm profile". Storms may be of a synthetic type or
real events. Application of real storms with antecedent condi-
tions may lead to peak flow frequency relations. In the RM it
is assumed that the return period of a storm is the same as
that of the peak flow.
Another factor to be considered in the comparison is the
routing procedure. The Rational Method substitutes routing by
consideration of different rainfalls of decreasing intensifies.
It does not give the designer the flow history of a runoff event
It gives only peak values and not hydrographs.
In modelling procedures, the transformation of rainfall
into overland runoff and channel routing is usually conducted
in two distinct parts of the analysis. This separation, which
represents the normal framework of analysis for a hydroiogist,
is not so obvious to those used with the RM in which the two
phases: rainfall-runoff transformation, and routing are lumped
into a single procedure.
In comparing the RM peak flows with measurements or models,
some assumptions should be made to circumvent these difficul-
ties. For example, in the comparison with measurements, the pro-
cedure indicated in Figure 2 was used in several studies (2,
21). According to this method, an average rainfall intensity
used as an input to the RM is selected for a duration equal to
tc of the real storm comprising the peak intensity.
Watkins (21) used this method to calculate the peak dis-
charges of 283 storm events observed on 12 catchments and com-
pared them with the observed flows. The value of mean absolute
error was between 10% and 20% on seven catchments and between
20% and 26% on three. For the remaining two catchments, mean
errors of 47% and 100% were found.
The same procedure was used by Wisner and Clarke in a
study carried out for the Canadian Urban Drainage Committee (2).
The SWM model together with other models (including the Rational
Formula) were tasted against measurements for small watersheds.
It was found out that SWMM predicted the peak flows with an
error in order of +_ 20%. Flows predicted by the Rational
Method had a larger error (see Table 2 and Figures 3 and 4).
The study concluded that the RM is not appropriate for the
simulation of runoff for a real storm event. Since for some o~
the areas it underestimated while for others it overestimated
the flows, it was considered that additional studies are re-
quired to avoid inconsistencies in design. Formula (1) was con-
sidered more adequate than runoff coefficients which do not
44
-------
include the irnperviousness ratio as a parameter.
Comparisons with measured flows by other authors indicate
different errors. Jens and McPherson (22) reported applications
in Baltimore, St. Louis, Los Angeles and Oxhey which gave a mean
absolute error in the prediction of peak discharge of 31.5%,
Swinr.erton et al (23) tested the method on 12 storms recorded
on motorways and obtained an average absolute error of 83.3%.
Chow and Yen (24) used the Rational Method to compute the peak
discharge for four storms on the Oakdale Avenue catchment, and
obtained an average error of 16%.
4. METHODOLOGY FOR THE COMPARISON UNDER
DESIGN CONDITIONS
Many of the previous comparisons between flows predicted
by the Rational Method, other hydrograph methods such as SWMM,
and measured flows were conducted for rainfalls which have
much smaller intensities than it is usually considered under
design conditions. The variation of the infiltration losses,
associated with these low intensity storms, may be very sig-
nificant. The relatively wide scatter of the Rational Method
flows for measured storms does not give the practicing drain-
age engineer a clue regarding the safety of a drainage system
designed by means of the Rational Method.
Verification of a model under design conditions would
•require high intensity storms. Measurements for storms which
have, for example, 5 years recurrence intervals are, however,
very rare and perhaps inaccurate. Because of the good per-
formance of a model such as SWMM for measured storms, it is
however assumed that under design conditions one may also
expect reasonably good predictions for the present state of the
art. This does not mean'a priory, that if the RM has a poorer
simulation for frequent storms it may not be used with ap-
propriate parameters and limitations for the more severe de-
sign conditions.
It was, therefore, considered that an assessment of the
Rational Method as a design tool can be made by a systematic
comparison with flows determined by a model such as SWMM.
SWMM was selected because of its sophisticated routing routine
which also has the capability of accepting hydrographs as an
input. The methodology for comparison can be applied using
other models and appropriate routing techniques.
Some of the Rational Method deficiencies such as lack
of aipe routing are not significant for small watersheds.
'analysis was therefore conducted in two steps:
1 The Rational Method parameters were first selected in
order to obtain the "best fit" with SWMM for a small
watershed (less than 20 acres).
45
-------
2. After the selection of the Rational Method parameters
for the small watershed, flow computations are ex-
tended to larger watersheds where routing effect may
be more significant.
For the first step, the main parameters used in conjunc-
tion with the Rational" Method are the runoff coefficient "C"
and inlet time "ti". Since the imperviousness "I" is a_ main
parameter in SWMM, selection of a runoff coefficient C in
terms of imperviousness, using formula (1) was considered
appropriate. The values of Cperv and C. in this formula were
0.2 and 0.9, respectively. " p
Since the Rational Method flows are determined from in-
tensity-frequency curves, it was also felt that for consis-
tency of input the SWMM flows should be computed from syn-
thetic design storms derived from these curves. The use of storm
profiles of the Chicago type (5) with three levels of dis-
cretization (2, 5 and 10 minutes) are presented in Figure 5
and is discussed in the next section. The methodology des-
cribed in this paper, however, can be applied for any storm
profile.
5. CONSIDERATIONS REGARDING THE USE OF
CHICAGO STORM PROFILES
The design storm concept has been criticized by McPherson
(25) and Marsaiek (26) and others for a number of valid reasons,
namely:
(i) the attempt to summarize widely varying storm patterns
in a single hyetograph shape;
(ii) the exclusion of antecedent conditions which may vary
from storm to storm;
(iii) the return period associated with a design storm can
be misleading and technically imprecise because the
frequency curves which form the basis for such com-
putations are themselves derived from different storms,
in a time sequence other than the actual occurrence
and often contain non-existent dummy values;
(iv) the time of concentration varies from point to point
in a watershed and the use of "design conditions" ob-
tained on the basis of a single design storm is mis-
leading.
The alternatives to using a single design storm, on the
other hand, are:
(i) to perform a long-term simulation using a calibrated
46
-------
computer model and recorded rainfall data;
(ii) to use a set of recorded rainstorms as being represen-
tative of a desired recurrence interval (27);
(iii) to use a series of design storms of different durations
which are comparable to real storms observed from local
rainfall records (27).
Arnell (28) reported the results of simulation of peak
flows^with various types of synthetic rainfall data and" return
periods and compared them with measured peak flows for a small
catchment area. It was found that the theoretical Chicago dis-
tribution by Keifer and Chu (5) yielded higher flows com-
pared to the measured ones.
Unpublished studies conducted by J.F. MacLaren Ltd. (29,
30) reported that the "Chicago" design storms of 2, 5, 10 and
25-year return period having rainfall volumes similar to those
of real storms yield flow frequency curves which are comparable
with the use of real storms (Figure 6 and 7).
An analysis of reports (26) and (29, 30) revealed a dif-
ference in the choice in the time step used for the discreti-
zation of the design storm. The MacLaren studies use time steps
of 5 min. while the Marsalek study uses 2 min. As shown in
Figure 5, the choice of time step can significantly affect the
"peakiness" of a design storm. A "peakiness factor" (?F) can
be defined as the ratio of the peak intensity to the average
intensity of the storm.
An examination of several critical storms from the Toronto
Airport records reveals that for a duration of 1-hr., the
"peakiness factor" value was in the range of 2.2 and 2.3
whereas the 1-hr., 5 yr. "Chicago" yielded ?F values of 5.9,
4.2 and 2.9 when discretized at 1, 5 and 10-minute time steps.
The sensitivity of design flows determined by 3WMM to
the "peakiness" (directly affected by discretization) of the
input" hyetograph, was examined in a series of systematic simu-
lations". Tests"were conducted using 1, 2 and 5-year "Chicago"
storms of 1-hour duration applied to a 96-acre hypothetical
catchment shown in Figure 3. The design storms were discretized
to 1, 5 and 10-minute time steps. Table 3 presents the results
from these simulations and shows that design peak flows model-
led with SWMM are sensitive to the "peakiness factor" of the
design storms.*
*These tests were conducted by Mr. S. Gupta, graduate student
at the University of Ottawa and are discussed in detail in a
non-published report, "Review of Design Storms", August 1979.
47
-------
Based on these results, it seems that peakiness of a
design storm should be related to the "peakiness" characteris-
tics of real storms. In general, short duration storms have a
higher probability of obtaining a high peak.
Selection of a ?F value for a design storm should, there-
fore, be related to the time of concentration (tc) of the catch-
ment from which the flows are being simulated. For example, for
a very small catchment, with a small tc/ it may be advisable
to model a short duration, high PF value design storm. These
results confirm the need to "design a design storm" (27). They
also show that by appropriate selection of the time step, the _
peakiness factor of the Chicago hypothetical storm can be modi-
fied to generate realistic flow frequency curves.
6. COMPARISON OF DESIGN PEAK FLOWS
As a first step, detailed analysis using both the Rational
Method and SWMM was carried out on a small typical residential
area indicated in Figure 9. The watershed which has an area of
23.3 acres (9.43.ha) is divided into 13 subareas ranging in
size between 1.2 acres (0.5 ha) and 2.6 acres (1.05 ha). For
such a small area, the inlet time represents a substantial
portion of the time of concentration.
Comparisons between the RM flows and SWMM flows with
two inlet times (5 and 10 minutes) and three levels of storm
discretization (2, 5 and 10 minutes) are presented in Figure 10.
For a storm with a peak intensity of 4.3 in./hr (10 mins. dis-
cretization) , the RM flows are very close to SWMM flows if the
inlet time is 10 minutes. On the other hand, for the higher in-
tensity corresponding to the 2 minutes discretization, -he in-
let time has to be reduced to approximately 5 minutes in order
to obtain RM flows in agreement with the SWM model.
These results show that for a small area the Rational
Method can be calibrated to give flows which are similar to
SWMM peak flows, using a unique value of C in terms of the
imperviousness and an inlet time related to the peakiness
factor of the storm profile or the peak storm intensity"!
For the analysis of routing effects, flows were cal-
culated by both methods (the Rational Method with 10 minutes
inlet time and SWMM with 10 minutes storm profile) for two
test areas: TESTVILLE A and TSSTVILLE B (see Figures 11 and 12).
The drainage arrangement of Testville A is a rather
theoretical layout consisting of 20 typical subwatersheds,
each having an area of 20 acres; total area is 400 acres.
Each subwatershed is a typical residential area with imper-
viousness ratio of 30% and ground slope of 2%. Testville 3
represents a cypical new development: of approximately 165 acres
48
-------
It includes a park and two school areas. The ground slope is
2%. The total development is discretized into 21 subareas with
sizes ranging between 6 acres and 12 acres.
Pipe sizes were selected in order to avoid surcharges.
Routing was done by means of the WRE Transport Version of
SWMM. In these tests, the SWMM peak flows from the small
subwatersheds are very close to the RM flows and, consequently,
any difference in the results will reflect the effect of rout-
ing. Results presented in Figures 13 and 14 for TESTVILLE B
indicate that as the drainage area increases, the RM will
underestimate the flows. A similar trend can be observed for
TESTVILLE A. as shown in Figures 15 and 16.
It also seems that the configuration of the sewer sys-
tem has some bearing on the results. The relatively longer
and narrower system in TESTVILLE A, which means longer time
of concentration, shows a larger discrepancy.
An apparent difference between the above results and
the comparison with low intensity measurements (Figures 2
and 3) is that under design conditions (for a given type of
storm) the RM_flows are systematically biased as compared to
the SWMM flows.
The comparison between the RM and SWMM flows under de-
sign conditions show that for relatively small areas (less
than 100 acres) the differences are small and may not be
economically significant. This confirms that the RM, if
properly applied, is an adequate tool for determination of
peak design flows.
It may, however, be desirable to apply the RM in such a
way that flows are in better agreement with models such as
SWMM. Several possibilities will be briefly discussed:
1. an increase in runoff coefficient "C"; In fact; some
drainage manuals have recommended for residential areas
G values which are slightly higher than those computed
by formula (1). This would lead to overestimation of
flows for the upper small areas. However, the effect on
the selection of pipe sizes may not be significant.
2. runoff coefficient "C" variable with the time of concen-
tration :It is possible to consider an increase of the
C value in terms of the time of concentration. The trend
shown in Figure 17 would support this approach. Results
based on Figure 1 developed by Mitci (15) are shown in
Figures 18 and 19 for TESTVILLE A and TESTVILLS 3, res-
pectively. For TESTVILLE 3 the RM flows computed by Mitci
relation are smaller than those computed with a constant C.
This can be explained by the low time of concentration for
49
-------
TESTVILLE 3. On the other hand, for TSSTVILLE A, the RM flows
computed using Mitci relation are closer to SWMM flows, but
still remain systematically smaller. With more systematic
testing, a new relation could be developed.
7. CONCLUSIONS
Previous comparisons against measurements show not only
errors but also inconsistencies in predicting the peak flows
by the RM. Most measurements correspond to rainfalls with low
intensities, where antecedent conditions, time-variation of
infiltration and other losses may be very significant.
Through systematic comparisons between the RM and SWMM
under design conditions described in this study, it was found
that the Rational Method may give consistent results:
a) For a given design storm and for small areas (less than
20 acres), the RM parameters can be selected in order to
obtain flows which are in close agreement with those deter-
mined by SWMM. The runoff coefficient C can be expressed
in terms of imperviousness by a simple relation (I).-The
"equivalent inlet time" varies with the peakiness of the
design storm.
•
b) For larger watersheds, where requirements for routing can
be significant, flows determined by the RM show a sys-
tematic bias when compared to SWMM flows. For the examined
configurations and storm profiles, the RM flows were con-
sistently smaller with the difference increasing with the
area of the watershed or the time of concentration.
c) The difference between the flows determined by the RM and
SWMM varies with the shape of the watershed and the con-
figuration of pipe system,
d) For watersheds of 100 acres and less, a slight increase
in C value would bring the RM flows very close to SWMM
flows for the downstream part. The consequent overesti-
.mation of flows for the upper small area may not be eco-
nomically significant.
e) Refinement for larger areas would consist in varying C
in terms of the time of concentration. Application of re-
lations developed by Mitci (15) resulted in better agree-
ment of the RM and SWMM flows. However, the RM flows were
still smaller than the SWMM flows.
f) The selection of design storm profile is important not only
for studies with SWMM but also for applications of the RM
(selection of an appropriate "inlet time").
50
-------
These results seem to confirm that with careful selection
of parameters/ the RM is a good tool for design purposes of
conventional storm sewer systems.The limiting size"of the
watershed can be increased"if a variable runoff coefficient is
adopted (e.g. C increasing with the time of concentration).
Advantages in using SWMM for design of storm sewers re-
sult mainly if surcharge and storage are analyzed. An example
of a design problem which cannot be studied by the RM is the
dual storage system, which is based on the principles of dual
drainage (pipe and street flows), runoff control, and inlet
control. Such an application would require sizing of storage
elements (e.g., underground storage and surface storage), con-
sideration of restricted outlets, backwater effects and over-
flows. The economic advantages and design aspects of this solu-
tion have been presented in another report (31).
51
-------
REFERENCES
1. Ben Chia Yen, Editor, "Workshop Notes on Storm Sewer System
Design", Department of Civil Engineering, Water Resources
Centre, University of Illinois, Urbana, Illinois, 1973.
2. MacLaren, James P., Ltd., "Review of Canadian Storm Sewer
Design Practice and Comparison of Urban Hydrologic Models",
July 1973.
3. Mechler, W.A., "The Rational Method of Peak Storm Flow
Determination", Grea-ter Vancouver Sewerage and Drainage
District, November 1976.
4. Colyer, P.J., and Pethick, R.W., "Storm Drainage Design
Methods, A Literature Review", Hydraulic Research Station,
Wallingford, Oxfordshire, England, March 1976.
«
5. MacLaren, James F., Ltd., "Drainage Criteria Manual for the
City of Winnipeg", November 1974.
6. Dillon, M.M., Ltd., "Drainage Criteria Manual for the City
of Burlington", 1977.
7. MacLaren, James F., Ltd., "Drainage Criteria for the Town
of Vaugham", 1973.
8. Brodie, A.F., and Wisner, P., "Drainage Criteria for the
Town of Oakville", 1979.
9. ASCE and WPFC, "Design and Construction of Sanitary and
Storm Sewers", manual of practice No. 9, 1970.
10. Schaake JC, Jr, Geyer JC and Knapp JW, "Experimental
Evaluation of the Rational Method", ASCE, J. of Hydr.
Division, Vol. 93, No. HY6 (1967).
11. Gregory, R., and Arnold, C., Run-off-Rational Run-off
Formulas, ASCE Trans., Vol. 96, 1932, pp. 46-50.
12. Ordon, G.J., "Storm Water Runoff", ASCE, Meeting pre-
print 2032, July 1932.
13. Corrugated Steel Pipe Institute, "CPS Sewer Manual",
Mississauga, Ontario, February 1977.
14. Chien, J.S., and Saigal, K.K.,"Urban. Runoff by Linearized
Subhydrograph Method", Jnl. Hyd. Div., ASCE, Vol. 100,
April 1974, pp. 1141-1157.
52
-------
15. Mitci, C. , "Rationalization of the Rational Formula, A
Simplified Method for Determining Urban Runoff", Water
& Wastes Engineering, January 1974.
16. McKinney, H.M. , "St. Louis Flood Protection: Interior
Drainage", Journal of Hydraulics Division, ASCE, Vol. 93,
July 1967, pp.. 129-147.
17. McPherson, M.B., Some Notes on the Rational Method of Storm
Drainage Design, ASCE Urban Water Resources Research
Programme, Tech. Memo G, 1969.
18. Ardis, C.V., Dueker, K.J., and Lenz, A.T., "Storm Drainage
Practice of 32 Cities", Prcc. ASCE, HYi, January 1969,
p. 383.
19. Lloyd-Davies, D.E., 'The Elimination of Storm Water from
Sewerage Systems', Prac. ICE, Vol. 164(2), 1906, pp. 41-67.
20. Road Research Laboratory, "A Guide for Engineers for the
Design of Storm Sewer Systems", Road Note No. 35, DSIR,
1963.
21. Watkins, L.H., "The Design of Urban Storm Sewers", Road
Research Technical Paper No. 55, DSIR, 1962.
22. Jens, S.W., and McPherson, M.S., "Hydrology of Urban Areas",
Section 20 of Chow V.T. (ed) Handbook of Applied Hydrology,
McGraw-Hill, 1964.
23. Swinnerton, C.'J. , Hall, J.M. , and O'Donnel, T. , "A Dimen-
sionless Hydrograph Design Method for Motorway Stormwatar
Drainage Systems", Journal Instn Highway Sngrs, November
1972.
24. Chow, V.T., and Yen, B.C., "Urban Storm Water Runoff and
Flowrates, Environmental Protection Agency, Cincinnati",
Draft Report, 1975.
25. McPherson, M.B., "The Design Storm Concept", Addendum 2,
Urban Runoff Control Planning, Miscellaneous Report
Series, Report No. EPA-600/9-78-035, October 1978,
pp. 100-118.
26. Marsalek, J., "Research on the Design Storm Concept",
Urban Runoff Control Planning, Tech. Memo No. 33,
September 1978, pp. 153-187.
27. Urbonas, B., "Reliability of Design Storms in Modelling",
Proc. International Symposium on Urban Storm Runoff,
Univ. of Kentucky, Lexington, Ky., July 23-26, 1969,
OD. 27-36.
53
-------
28. Arnell, V., "Analysis of Rainfall Data for Use in Design
of Storm Sewer Systems", Proc. International Conference
on Urban Storm Drainage, Univ. of Southhampton, U.K.,
April 1978, pp. 71-86.
29. MacLaren, J.F., Ltd., "Report on Design Storm Selection",
City of Winnipeg, Waterworks, Waste and Disposal Division,
May 1978, p. 24.
30. MacLaren, J.F., Ltd., "A Comparison of Historical and
Theoretical Design Storms for the City of Edmonton",
November 1978, p. 31.
31. wisner, P., Kassem, A., and Cheung, P., "Application of
the SWMM Model to the Design of Dual Storage", SWMM Users
Meeting, Montreal, May 24-25, 1979.
54
-------
TAIII.I: l
ciiif>ui»i ion of nut of- cu-ic£'iiHAi ion (HINUIESI BY six
lit i nous (Reference 2> )
en
en
L* 100. ft.
1- 0.50 lll/tlH C" 0.02
S-O.OI
S-0.02
S-O.OJ
S-G.O4
i-0.05
S-O.Oi
S-U.OI
I- 1.00 Ifl/liil C' O.Oi
S-U.OI
5-0.02
S-O.Oi
S-O.O'.
S=0.0i
S-0.0&
S-O.OJ
l- 1.50 i. -1/11:1 c- 0.07
i'<».OI
s^o.o2
S-O.OJ
S-O.04
S-0.0*
S- 0.116
s-o.ai
\- 2.00 Ill/Ill! C» O.UU
S-O.OI
S^o.ii?
S-O.03
S-0.0'.
S-O.U'J
S-U.06
s-o.or
I- 2. SO ln/liu C' 0.09
i-O.QI
S-0-02
S-O.OJ
S - (1 . 0 <.
S-O.O'J
i-0.06
S-0.0/
ASSU'HllG;
HANNIIIGS n
U/AUUS CR
'PAVtO
UKHO
21.)
16.9
H. a
13.4
12.3
11. 7
11.1
I12AHU
11.1
1O.6
9.3
a.'.
7.0
7.J
7,0
IJIAHO
10.',
0.2
7.2
6.5
6.1
5.1
5.^
I2ZAIIU
a. 7
6.'l
6.O
5.5
5.1
<..u
4.5
I//AHO
7.6
li.O
5.1
4.U
4.5
'..Z
4.O
KEHBV
2. a
2.J
2.1
2.0
!.'»
i!o
1.7
Ktiiur
2.4
2.J
2.1
2.0
l.'»
t.a
i- 1
KEiinr
2.8
2.1
2. 1
z.o
1.0
i.a
1.7
KEIIUY
2.1)
2.1
2.1
2.0
I.')
i.a
1.7
KEHUV
2.0
2.1
2.1
2.0
1.')
l .a
I. r
KI »L.'IAI 1C
/.-.
t.o
».l
4.')
4 .6
4.3
4. I
KIHtHAHC
'..7
3.0
3.4
3.1
2.0
2. a
2.6
KtHEMftMC
3.7
3.0
2.7
2.4
2. J
2.2
2.1
KlIlCMlUC
3.1
•2.5
2.2
2.0
t.O
1.0
1.7
KINLMMIC
2.7
2.2
I.'J
1 .6
1.7
1.6
1.5
Klltl'ICII
l.J
1.0
0.0
o.a
0.7
0.7
0.7
KIIIT ICH
-l.J
1.0
o.->
0.11
O.I
0.7
0.7
K HIP ICH
1.3
1 .0
O.'l
O.tt
0.7
0.7
0. 1
KIIU'ICll
1.3
l.O
O.'J
0.0
0.7
0.7
0.7
Kl III' ICH
1.3
1.0
<>.•>
fl.U
o.l
O.I
o.;
PIVEO
0.01)
0.013
CRASSED-fEftVIUUS
0.350
0.060
GHASSED-PERVIOUS
*
lEELlf 1:17777
«.2
6.S
5.7
5.2
4. a
4.5
4.3
SttrLYE
(1.0
6.3
5.5
5.0
4.7
4.4
4.2
StELYE
7.')
6.3
5.5
S.O
4.7
4.4
4.2
SlitLYE
7.0
6.3
5.5
5.0
4.6
4.4
4.1
SfcELYe
7.0
6.2
5.5
5.0
4.6
4.1
4.1
5.0
4.0
3.5
1.2
2.0
2.e
2.6
777?
4.3
3.
-------
Table 2
Performance of Che RM and SVM Model
as Assessed by "he Canadian Review
of Urban Drainage Practices (2)
Catchment Area
(acres)
Oakdala
Avenue 12.95
(Chicago)
Northwood
(Baltimore) 47.4
Gray Haven
(3aiciaore) 23.3
Calvin Park
Kingston, 89.4
Ontario
7.
Imp
45.3
53
52
27
„ No. of
Storms
14-17
14
10-14
10-13
Rational Method
X as
1.4 0.58 46
0.97 0.32 21
0.91 0.22 20
1.63 0.28 '63
STflffli
\ 3 Z
1.C8 0.20 19
1.18 0.51 33
1.13 0.22 19
1.20 0.22 23
* These results were rejected because of inaccurate date
X Signifies the ratio computed/measured peak discharge
a Signifies Che standard deviation of the individual values -about
z Signifies Che mean absolute percentage error
56
-------
TABLE 3 - SENSITIVITY OF SWMM (RUNOFF) TO PEAKINESS OF A CHICAGO STORM
RETURN PERIOD TIME STEP OF PEAK FLOW (IN./HR.)
(YRS) DISCRETIZATION AREA (ACRES)
(MINUTES) 12 64 96
1 1.36 1.09 0.94
5 1.01 0.92 0.83
10 0.79 0.75 0.72
1 1.57 1.28 1.12
5 1.19 1.09 0.98
10 0.93 0.88 0.85
1 2.01 1.69 1.49
5 -1.53 1.44 1.32
10 1.17 1.13 1.11
57
-------
en
00
MO |6O
100
TIME FHOM BEGINNING OF RAINFALL (min)
Figure 1: C. as u Inaction ol: Time (from Uel:. 15).
-------
avq.
ACTUAL HYETOGRAPK
I. fe DETERMINED AT OUTU2T
2. MAX. AVG. INTENSITY FOUND rOH fc
FROM RECORDED 5TOSM
1 C..A FROM PHYSICAL DATA
4. Q MAX. a C i A
Figure 2: Estimation of Peak Rainfall
Intensity from Recorded Rainfall
for JIM Peak Flow Confutation, (s.er.2 )
59
-------
10 20 30 40 SO 60 10 80 bO
» SWMM
Q Rational Method
Figure 3 Comparison Of Measured And Calculated Peak Flows For The Gray Haven
District In Baltimore (Data from Uofi>i:ence 2).
-------
O
O
a>
« SWMM
O Ralional Method
01 2 3
12 13 M 15
y Obs. (cfs)
Figure 4 Comparison Of Measured And Calculated Peak Flows For The Oakdale
District In Chicago (Data from Reference 2 ).
-------
ro
8.0
7.0
~ 6.0
- 4.0
Z 3.0
UJ
h~
S 2.0
1.0
[~TJ
r
j
r
i
...*
i
n
*
F
O t|IM
c MIN.
5 MIN.
— JO MIN.
-
i
tl
u
L
*""»
0
10 20 30 40 50 60 70 80 90 tOO 110 120 130 140
TIME (MIN.)
Figure 5: 5 Years Intensity-Duration-Frequency Curve and the
Corresponding Chicago Design Storm (10 mins., 5
mins., and 2 mins. discretization).
-------
1.10 1.20 1.401.SO 2.0 3.0 4.0 5.O 10
RETURN PERIOD (YEARS)
20
30 4O 50
LEGEND
REAL STORM
A DESIGN STORM
I cfs/ocre « 0.070 mVs hectare
FREQUENCY OF £XCEEDENCE
VS
RUNOFF
Tc= 15mtn.
FIGURE 6 (Reference 29)
63
-------
50-
a-s-
2-0-1
UJ
I-
to
u.
o
1
^ i-oH
0-5-
• Peak flow for real siorms
A Peak flow for design storms
APPROXIMATE
UPPER 95%
CONFIDENCE LIMIT
'LEAST SQUARES
REGRESSION FIT
IGNORING TOP POINT
APPROXIMATE
LOWER 95%
CONFIDENCE LIMIT
•'01
i
l-l
T
1-5
i
2
3.
i
•5-
i
10
r
20
i
50
100
RETURN; .PERIOD
FIGURE 7 (Reference 30)
FLOW FREQUENCY ANALYSIS'FOR REAL STORMS, LOCATION 680
(14-2 ACRES)
64
-------
cr>
en
SUBCATCHMENT
DETAILS
AREA (AC.)
PERCENTIMPV
JUNCTION
NO.
PIPE NO.
DIA.dW.)
LENGTH/SLOPE (%|
Figure 8: A Portion of lestville B, 96 Acres
-------
OUTFLOW
en
en
~) E.6 Ac. I
Figure 9: Schematic ot a Typical Small Residential Watershed Used for Systematic
Couinarisoii of the RM ami SWMM i:i/«.i«
-------
4O
35-
~ 30
b.
0
a
o
i
H-
LU
25
20
l5
OS
a
10
RM INLET TI>E
(min.)
10
SWMM STORM DISCR.
(min.)
10
10 15 20 25
Q. SWMM (CFS)
30
35
40
Figure 10: Calibration of Inlet Time to Obtain Flows
in Agreement with SWMM.
67
-------
CTl
00
figure ll: Schematic tuiyout of Teslville A.
-------
VO
FIGURE IZ- Schematic layout of TESTV1I.LE I>
-------
UJ
0.
0 30 100 ISC 200 250 300
PEAK FLOW-SWMM WITH 10 MIN. STORM
DISCRETIZATION
Figure 13': Comparison of Peak Flows Calculated with
SWMM and RM for Testville B
250
200
in
u.
o
— ISO
O
"• 100
UJ
Q.
50
• RM (10 MIN. INLET TIME)
• SWMM (10 MIN. CHICAGO STORM DISCRETIZATION 1
0 20 40 60 80 100 120 140 160 ISO
AREA (ACRES)
Figure 14: Peak Flows vs. Area for Testville B
70
-------
300
400
C/J 300
O
200 -
UJ
Q.
100 -
240
200
u 160
120
ui
O. 30
2
IT
40
40 8O 120 I6O 200
SWMM PEAK FLOW (c?s)
240
Figure 15; Comparison Between Peak Flows- Calculated
with SWMM and RM for Testville A.
• RATIONAL METHOD (10 MIN. INLET TIME)
A SWMM (10 MIN. CHICAGO STORM DISC.)
40 00 120 160 200 240 280 320 360 400
AREA (ACRES)
Figure 16: Peak Flows vs. Area for Testville A.
71
-------
1.5
1.4
1.3
1.2
S "
4/>
O
1.0
0.9
0.8
0.7
0.6
0.5
ta TESTVtLLE "A"
Q TESTVILLE V
. Q
a
Q
Q
Q
Q
-B-
10
35
40
15 20 25 30
TIME OF CONCENTRATION ( M.N. )
l-igure J.7: Ratio of QSWMM/QRM in Terms of the Time of Concentration.
45
-------
240
TESTV1LLE "A
• "C" CONSTANT
A "C" FROM MITCI
40
80 120 160
Q SWMM (CFS)
200
240
Figure 18: Rational Method Flows with Runoff Coefficient
Variable with the Time of Concentration (Mitci
(15)) as Compared to Flows with Constant C
and SWMM Flows - Testville A.
73
-------
240
200
£ 160
Q
O
X
120
Z
O
P 90
tr
o
40
RM WITH "C" VARYING WITH Tc (MITCI)
• RM WITH CONSTANT "C"
40
80 120 160
Q SWMM (CFS)
200 240
Figure 19: Rational Method Flows with Runoff Coefficient
Variable with the Time of Concentration (Mitci
(15)) as Compared to Flows with Constant C
and SWNM Flows - Testville B.
74
-------
URBAN RUNOFF QUALITY IN
METROPOLITAN TORONTO
by
F. Ivan Lorant,2 P.Eng
1. BACKGROUND
Following the Agreement between the United States and Canada
on Great Lakes Water Quality in 1972, a number of research
projects were undertaken under the guidance of Canada-Ontario
Urban Drainage Subcommittee to carry out research programs on
the abatement of municipal pollution. One of the first studies
included in this program was the collection of field data on
the quantity and quality of urban runoff in the Borough of
East York, Toronto. The study included the setup of a data
acquisition system, collection of data on quantity and quality,
the calibration and verification of the Storm Water Management
Model Runoff and Transport Blocks.
A previous paper presented in 1977 (see reference 1 in the
Bibliography) described in detail the data acquisition system
and the calibration of the quantity aspects of the Storm Water
Management Model. The intent of this paper is to summarize
the water quality information observed over the period
December 1975 - September 1976.
The total test area of 383 acres shown on Figure 1 was developed
in the 1920's and is served by a combined sewer system. The
predominant land use of the urban catchment area is residential
(89.1%) made up by single family residential units, while the
remaining area is institutional (5.7%), parks and open spaces
1. Chief Water Resources Engineer,
M.M. Dillon Limited
Consulting Engineers and Planners
Box 219, Station K
Toronto, Ontario M4P 2G5
75
-------
m
X
I
0
>
0 i
CQc
m x
LEGEND:
_ __ _ _
SINGLE FAMILY
RESIDENTIAL UNITS
MULTI-FAMILY
RESIDENTIAL UNITS
COMMERCIAL
INSTITUTIONAL-SCHOOLS,
HOSPITALS,CHURCHES, ETC.
OPEN SPACE - PARKS
PLAYGROUNDS
; OPEN SPACE - RAVINE LAND
URBAN TEST AREA BOUNDARY
SOURCE:
BOROUGH OF EAST YORK'S
SECONDARY PLAN STUDIES
FOR THE CENTRAL COMMUNITY
-------
(4.2%) and commercial (1%), There is no accurate data
available on the population, the estimated population of the
area is 14,600, with an average of 3.5 persons per household,
and an average gross residential density of 38 persons per
acre.
A detailed analysis of photographs and site inspection
determined that 49% of the study area i.e. 187 acres can be
classified as impervious, made up as follows:
52% roads, sidewalks and driveways
38% roofs connected to the sewers
4% roofs not connected to the sewers; and
6% miscellaneous.
A special survey determined that there are 481 catch basins,
therefore, the catch basin density is 1.26 per acre. The
inspection of the sumps showed that the total storage volume
is approximately 13 cu. ft. in each sump.
Limited records indicate that all pipes up to 24 in. in
diameter are glazed vitrified clay, while the remaining
larger sizes are concrete, except two 60 and 66 in. diameter
sewers which are believed to be brick.
77
-------
2. MUNICIPAL PRACTICES
A number of the municipal practices such as street cleaning,
sewer cleaning, snow and ice control can affect the quantity
and quality of storm water runoff.
During the snow and ice free period between April and November,
the roads within the study area are generally swept by a pick-
up sweeper once a week. In adverse weather conditions, the
road gutters are cleaned by manual labour. The total quantity
of litter and debris removed was 504 cu. yd. during March -
November 1976, and 280 cu. yd. during April - October 1977.
There is no regular program for cleaning sewers. Catch
basins are generally cleaned once a year by a vacuum hose. No
records are available on the debris removed, however, estimates
show that in 1976 the total amount of debris removed was
approximately 92,000 Ibs.
During the winter period rock salt is used with a 6 ton salt
spreader, adding occasionally anti-skid material of sand mixed
with the rock salt. The rate of application depends on the
weather conditions. During the December 1975 - March 1976
winter, the amount of salt spread varied from 0 to 24 tons per
day, totalling 397 tons.
Road plowing begins when the snow accumulation reaches about
3 in., however sidewalks are plowed generally after a snowfall
of Ik in. The snow from the roads and sidewalks are plowed
into the road gutter and the snow is only removed if the
accumulation along the road becomes unmanageable.
78
-------
3. DATA ACQUISITION SYSTEM
The purpose of the data acquisition system was to provide
accurate and sufficient information in order to prepare pre-
cipitation hyetogrpahs, air temperature graphs, flow hydro-
graphs and pollutographs. This data collected was used for
calibration and verification of the Storm Water Management
Model.
The data acquisition system shown on Figure 2 included the
following:
Preci pitation Measurement
A tipping bucket rain gauge with a print-out recorder
which prints the actual date and time to the nearest
minute every time the precipitation gauge bucket
tips, indicating 0.01 of an inch of rainfall. This
gives print-out capability up to 36 in. per hour
rainfall intensity.
Flow Measurement
The flow measuring device was installed in a 66 in.
diameter outlet pipe and it consisted of a partial
trapezoidal weir which was calibrated by model
tests.
A bubbler type system measured the head of water and
recorded it on a two pen strip chart recorder
together with the air temperature measurement.
79
-------
0)
0
I
m
0
0
•q
O
0
0
H
0
Z
5,
(Oo
CURRENT ALARM RELAY activates
pump and automatic sampler
STEVENSON SCREEN houses
a resistance thermometer bulb
DIFFERENTIAL PRESSURE TRANSMITTER
/measures difference in pressure between
compressed air outlet and atmosphere
tipping bucket
PRECIPITATION GAUGE
AUTOMATIC SAMPLER
/takes samples at pre
-set time intervals
sampler suction line
PRINT OUT RECORDER stamps the
date and time for each tip of the
precipitation bucket
ROTARY VANE AIR
COMPRESSOR
supplies constant compressed air line
pump discharge pipe
PUMP pumps sewage to
within operating limits
of automatic sampler
TWO PEN STRIP CHART RECORDER records
information from the differential pressure transmitter
(head of water at compressed air outlet)
:::;- '.';.-•' pump suction pipe
' air temperature as measured by the resistance
thermometer bulb
compressedair outlet
-------
Water Quality Sampler
The automatic sampler capable of collecting 24
discrete 1,000 ml. samples was located above
ground. This arrangement in spite of the 38 foot
head difference between the sample jars and the
sewer, was the most appropriate, as regulations
would have required to explosion proof all electric
instrumentation in the manhole.
The water quality samples of dry and wet weather
flows, were analysed at the Ontario Ministry of the
Environment laboratories.
Air temperature measurements were taken by a resistance thermO'
meter bulb. Recording of the temperature was on a strip chart
recorder.
Time synchronization was achieved by recording all events
except the precipitation on a two strip chart recorder and all
recorders kept excellent time throughout the study.
81
-------
4. DRY WEATHER FLOWS
The average measured dry weather flow was 2.75 cfs. Daily and
hourly variations were insignificant, when compared to the
wet weather flows.
The two pollutants analysed were BOD, average 95.3 mg/1, and
suspended solids, average 96.9 mg/1. A detailed summary of
the daily and hourly dry weather flow and pollutant variations
are shown in Table 1.
5. MODELLING OF FLOWS
The SWMM computer program used in the study was based on the
U.S. - E.P.A. Release II, dated September 1970, updated
February 1975, with snowmelt program added in October 1975.
The quantity calibration was based on low intensity precipita-
tion events, therefore, none of the recorded events had
sufficiently high intensities to overcome the estimated
infiltration and depression storage losses of pervious areas.
Three events have been used to calibrate the model. Subse-
quently, 20 precipitation - runoff events were used to verify
the model. The results were encouraging as shown by the
following summary:
RUNOFF VERIFICATION
PER CENT OF CALCULATED DATA
ERROR
Calculated Data <:20% <10% <5%
Flow Volumes 100% 81% 44%
Peak Flow Rates 91% 59% 34%
Time to Peak 97% 88% 72%
82
-------
TABLE 1
DAILY AND HOURLY DRY WEATHER FLOW VARIATIONS
Dry Weather Flow
DAY (Average 2.75 cfs)
1 -
2 -
3 -
4 -
5 -
6 -
7 -
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
0.942
0.985
1.036
1.076
1.033
0.975
0.945
HOUR
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
0.806
0.710
0.660
0.631
0.617
0.639
0.763
0.947
1.129
1.194
1.224
1.204
1.191
1.205
1.188
1.130
1.101
1.134
1.184
1.165
1.093
1.064
1.048
0.971
RATIO OF AVERAGE
BOD of Dry Weather Flow
(Average 95.3 mg/1)
1.000
1.000
1.000
1.000
1.000
1.000
1.000
0.826
0.330
0.175
0.206
0.165
0.144
0.372
SS of Dry Weather Flow
(Average 96.9 mg/1)
321
672
548
1.662
362
176
Oil
,053
094
0.970
1.104
1.620
1.775
1.465
1.053
0.888
1.001
000
000
000
000
000
1.000
1.000
0.766
0.472
0.273
0.231
0.199
0.178
0.818
,784
,689
,448
.448
,542
1.427
1.175
0.913
0.787
0.735
1.007
1.689
511
,280
049
,766
0.797
83
-------
6. WATER QUALITY DATA
The runoff quality data analysed are listed in Table 2 with
approximate range of values.
TABLE 2
RANGE OF OBSERVED WATER QUALITY DATA
Pollutants
BOD
COD
Chloride Cl
Sulphate SO.
Conductivity UMHOS/cm
Lead Pb
Total phosphorus P
Dissolved reactive
phosphorus P
Phenols
Free ammonia
Total Kjeldahl
Ni tri te
Nitrate
Suspended solids
Dissolved solids
Total solids
Range
25-250 mg/1
100-1,000 mg/1
30-2,000 mg/1
0.5-70 mg/1
200-7,000
0.2-3 mg/1
0.5-5 mg/1
0.1-1.0 mg/1
3-31 part per
billion
0.2-7 mg/1
4-30 mg/1
0.1-0.7 mg/1
0.1-2.0 mg/1
(dried or ashed)
30-2,000 mg/1
(dried or ashed)
100-4,000 mg/1
(dried or ashed)
100-4,000 mg/1
Remarks
High initial values
High initial values
Higher during winter
Higher during winter
Higher during winter
Higher during winter
Higher during winter
84
-------
A sample sheet of the recorded data is shown in Table 3.
Due to the shortage of funds and time constraints the project
included only limited amount of calibration and verification
of the BOD and suspended solid pol1utographs. However, in
order to permit future calibration and verification studies of
any of the pollutants, all recorded data will be published in
early 1980. (Reference 2).
7. MODELLING OF WATER QUALITY
The automatic sampler was programmed to record and collect
samples from significant storm events covering a reasonable
time span. This was achieved by activating the sampler when
the flow reached 15 cfs and continued to take samples at 6
minute 40 second intervals. Unfortunately this meant that the
initial part of the recorded pollutograph was not available.
Before carrying out any calibration of the quality parameters
simulated hydrographs were adjusted to agree with recorded
hydrographs.
Biochemical Oxygen Demand (BOD)
The observed BOD values generally start in the order of 150
mg/1 but after one to two hours the readings drop to 25 mg/1.
Out of the 18 events monitored only two had initial BOD values
higher than 200 mg/1 .
B'OD values can drop rapidly, however, the simulation could not
duplicate this rapid drop. Generally the initial stage of
simulations are below the observed values, while the latter
part of the simulated pol1utographs are higher than the
85
-------
TABLE 3
Units ore mg./Ulre unless otherwise indicated
RUNOFF QUALITY DATA
oo
Ul
i
uY/k'O/YR
30/12/75
Ul
HR MM
14:46
14:52
14:5E
15:06
15:12
15:19
^5:26
15:32
!5:39
15.46
15:52
15:59
16:06
16:12
16:19
16:26
16:32
16:39
16:46
16:52
16:59
1 7 : 06
(a
u
^
IS.
19.2
20.8
20.8
20.0
19.0
18.8
16.7
15.8
15.0
14.2
13.3
12.8
12.3
11.5
10.7
10.6
10.6
10.4
10.0
9.8
9.7
9.5
o
0
CD
61
65
75
64
43
45
38
33
33
35
31
31
31
38
35
58
53
48
58
46
44
45
o
q
o
255
495
260
250
280
16C
185
140
145
140
185
150
145
160
235
160
160
190
245
185
155
210
o
CO
Ul
o
X
o
747
718
687
683
697
712
790
810
881
916
930
971
975
1072
1051
1025
1034
1091
1072
1080
1033
1007
^
o
CO
CO
UJ
X
a.
CO
37
37
34
33
33
33
36
'37
37
38
40
40
53
46
44
46
48
55
50
50
43
49
>-
i-
> s
^i?
o ^
3 CO
o o
2: ^
O 2
cj n
2600
2600
2500
2500
2500
2500
2900
2950
3150
3300
3300
3400
3550
3600
3800
3700
3800
4000
3900
3900
3300
3750
JD
a.
CO
o
Ul
0.51
0.48
0.53
0.49
0.55
0.53
0.53
0.46
0.43
0.54
0.19
0.32
0.44
0.54
C.57
0.42
0.26
0.26
0.27
0.21
0.21
0.22
CO
3
IE
O
X
cv
o
X
a.
_j
< a.
r-
O CO
p<
2.70
2.10
2.60
1.70
1.00
0.76
1.20
1.00
0.84
0.44
0.76
0.68
0.84
0.96
1.00
0.80
1.00
0.96
1.20
H.10
1.00
1.20
Ul
> 3.
< <
Ul CO
a. o
a fc.
ui o
> 1.
-J CL
in g
0 t
1.00
0.74
0.60
0.44
0.34
0.22
0.36
0.20
0. i4
0.18
0.34
0.20
0:18
0.30
0.42
0.26
0.30
0.38
0.34
C.34
0.32
0.30
j.
ca
a.
0.
z
CO
o
z
Ul
X
CL
24
26
23
23
24
23
24
25
26
27
25
25
24
26
25
24
28
27
28
27
25
23
NITROGEN AS N
z
o
X
<
K
1.8
2.1
1.3
1.9
1.7
1.9
3.3
2.6
1.8
2.7
3.0
3.1
2.7
3.6
4.3
2.7
3.9
4.3
4.1
3.9
4.9
4.6
!
_]
Ul
-i
ac
1
9.1
8.7
10.0
7.2
3.3
5.4
6.8
5.6
4.7
3.9
4.5
5.8
5.6
6.2
6.6
5.4
7.2
7.0
7.4
7.2
8.0
4.3
hi
t-
z
0.04
0.04
0.02
0.04
0.08
0.10
0.08
0.28
0.34
0.34
0.34
0.30
0.44
0.40
0.46
0.02
0.16
0.08
0.04
0.66
0.36
0.02
Ul
z
0.4
0.4
0.2
0.2
0.3
0.3
0.3
0.3
0.5
0.7
0.7
0.5
0.8
1.0
0.5
0.2
0.2
0.2
0.2
0.7
0.4
0.2
SUSPENDED
SOLIDS
o
<£
Q
180
140
160
120
120
no
120
105
100
60
160
140
110
150
140
140
100
140
140
130
130
160
0
X
CO
-------
observed values. During calibration varying the BOD concen-
tration of the catch basins had little effect on the accuracy.
Examples of simulated and observed BOD pollutographs are shown
on Figures 3 and 4.
Suspended Solids
For the simulation of suspended solid loads the empirical
method (1SS=1) was selected as it produced better results than
the experimental equation (1SS=0). In the computation of dust
and dirt accumulation, observed dates of street cleaning
activities were utilized. The following equation was developed
to estimate the number of equivalent days of pollution
accumulation.
ND = tj_ (l-E)""^ (l-E)n"2+... tn (l-E)n"n
To estimate the length of dry period, the time of the last
storm was selected when the total rainfall amounted to one
inch. Time intervals between street sweeping and storm
events in days are represented by t. E is the efficiency of
the street sweeping and n represents the number of time intervals
An illustration of the above system is shown in the following
table for the 31 July 1976 storm event.
Storm analysed = 31 July 1976
Last storm (minimum one inch) = 20 July 1976
Street sweeping events during the two storms = 21 and 27 July
Therefore, tj = 21 July - 20 July = 1 day
t2 = 27 July - 21 July = 6 days
t3 = 31 July - 27 July = 4 days
n = number of events = 3
E = street sweeping efficiency = 0.75
N = 1(1-0.75)2 + ed-0.75)1 + 4(1-0.75)° = 5.56 = 6 days
87
-------
1
JU mii u u
U
COMPARISON OF RECORDED
A NO SIMULATED BOD
STORM Of MARCH 27 1976
w
rum
uuu
COMPARISON OF HEC9RT3
AND SIMULATED BOO
STORM OF MARCH 31.1*7*
u u uu u Din
COMPARISON OF HECOROEO
AND SIMULATED BOD
STORM Of APRIL 25,1976 I
COMPARISON OP RECORDED
AMD SIMULATED iOD
STORU OF MAY II. 1»76
\
COMPARISON OF RECOmED
AND SIMULATED BOD
STORM OF JUNEE.1978
COMPARISON Of RECORDED
AND SIMULATED BOD
STORM OF JUNE 50.1*7*
FIGURE 3
COMPARISON OF RECORDED
AND SIMULATED BOD
88
-------
COMPARISON or Htconoeo
AND SIMULATED BOO
STORM OF JULY 1, I976(pm]
Fl«.30b
* ,4
TTTT
COMPARISON OF RECORDED
STORM Of JULY 2,1978
COMPARISON OF flECOHDEO
AND SIMULATED QOO
STORM OF JULY20.I97C
Fffl.SIb
COMPARISON Or RECORDED
AND SIMULATED BOO.
STORM OF JULY 29,1976,
FIg.Mb
COMPARISON OF RBCOROEO
AND SIMULATED BOD.
STORM OF AUfl.l3.i97e.
Ftg.B6b
BOD
FIGURE 4
COMPARISON OF RECORDED
AND SIMULATED BOD
89
-------
The simulated and observed suspended solids pol1utographs are
shown on Figures 5 and 6.
Cone 1 us ions
The verification of the BOD and suspended solid pollutographs
demonstrated the possibility of simulating urban runoff water
quality parameters by a calibrated Storm Water Management
Model. Although the results of the verification exercise were
promising, the accuracy of predicting BOD or SS pollutographs
cannot reach the accuracy attained in simulating flow hydro-
graphs by a calibrated SWMM. However, both the Suspended
Solid and BOD simulated pollutographs presented could be
improved after an extensive re-calibration program. It is
hoped that the publication of the entire observed quantity and
quality data will provide the necessary impetus and incentive
for research students or other non-academic modellers to
improve the capabilities of the SWM model.
BIBLIOGRAPHY
1. Larsen, E., Recorded and simulated runoff from an urban
catchment area in Metropolitan Toronto, International
Symposium on Urban Hydrology, Hydraulics and Sediment
Control, Kentucky 1977.
2. Storm Water Management Model Verification Study, Canada-
Ontario Agreement on Great Lakes Water Quality,
Research Report No. 97.
90
-------
COMPARISON 0? RECORDED
AND SIMULATED SS
STOWM OF MARCH 27,19T«
TJTT
U U U
COMPARISON U* RECORDED
AND SIMULATES S3
STORM OP MARCH 51,1979
U
COMrARISOM OF RECORDED
AND SIMULATED S3
STORM OF JUNE I. IS76
j..
COMPARISON OF HECCROS9
AND SIMULATED 93
STORM Of JULY), 197* (pm)
U
COMPtiHISON OP nCCCRDED
AND SIMULATED 89
STORM OP JULf 2,1970
FIGURE 5
COMPARISON OF RECORDED AND
SIMULATED SUSPENDED SOLIDS
91
-------
COMPARISON OF HCCOHOE3
AND SIMULATED SS
STORM OF HAY 11, 1978
COMPARISON OF RECORDED
AND SIMULATED SS
STORM OF JUNE 30,1976
COMPARISON OF RECORDED
AND SIMULATED SS.
STORM OF JULY 29,1976.
FIGURE 6
COMPARISON OF RECORDED AND
SIMULATED SUSPENDED SOLIDS
92
-------
AN EXAMINATION OF THE STORM
WATER MANAGEMENT MODEL (SWMM)
SURFACE-RUNOFF-QUALITY ALGORITHMS
By William M. Alley-''
INTRODUCTION
The Storm Water Management Model (SWMM) is a computer model for
simulation of storm- and combined-sewer systems (Metcalf and Eddy, Inc.,
and others, 1971). The surface-runoff-quality algorithms contained in
SWMM have formed the basis for similar algorithms contained in several
other urban-runoff models including STORM (U.S. Army Corps of Engineers,
1976) and a modified ILLUDAS model (Han and Delleur, 1979). These models,
in some instances, have been used extensively. For example, STORM
(Storage, Treatment, Overflow, and Runoff Model) was the most commonly
used model in studies conducted under Section 208 of the Federal Water
Pollution Control Act of 1972 (P. E. Shelley and E. D. Driscoll, written
commun., 1979).
The purpose of this paper is to examine the functional relationships
used in SWMM for surface-runoff-quality simulation and to consider some
areas for improvement.
SWMM SURFACE-RUNOFF-QUALITY ALGORITHMS
There are two main components of the surface-runoff-quality algorithms
contained in SWMM: Constituent accumulation and constituent washoff.
Constituent accumulation is estimated as a function of land use, number of
i/Hydrologist, U.S. Geological Survey, Reston, Virginia
93
-------
days prior to the storm during which the accumulated rainfall was less than
1.0 inch, and street-cleaning frequency and efficiency. The quantity P in
milligrams of a constituent I on areas of land use L at the beginning
of a storm is computed as:
P(I,L) = F(I) x DD(L) x DRYDAYS x GLEN(L) x 453.6 (1)
where
F(I) = milligram of constituent I per gram of "dust and dirt;"
DD(L) = "dust and dirt" accumulation rate for land use L, in pounds per
day per foot of curb;
DRYDAYS = number of days prior to storm during which the accumulated
rainfall was less than 1.0 inch (modified to account for
street sweeping);
GLEN(L) = curb length for land use L, in feet; and
453.6 » conversion factor (grams per pound)
Constituent washoff is simulated using an exponential decay equation:
POFF(I.L) = PO(I,L) [l-e(~aRAt)] (2)
where
POFF(I,L) = amount of constituent I removed from areas of land use
L during a time step, in milligrams;
PO(I,L) = amount of constituent I on areas of land use L at the
beginning of time step, in milligrams;
a = decay coefficient;
R = runoff rate, in inches per hour; and
At = time step, in hours
94
-------
The primary assumption for use of equation 2 is that the amount of
constituent washed off during each time step is proportional to the amount
remaining on the land surface at the start of the time step.
When equation 2 is used to simulate suspended solids, it is multi-
plied by an availability factor (ASUS) where
ASUS = 0.057 + 1.4R1'1 (3)
Similarly, when computing the washoff of settleable solids, equation 2 is
multiplied by an availability factor (ASET) where
ASET = 0.028 + R1'8 (4)
Both availability factors are limited to a maximum value of 1.0, and the
user has the option of setting either or both of them to 1.0 for all time
steps.
After computation of constituent washoff during a time step using
equations 2-4, SWMM increases the amount of 5-day biochemical oxygen
demand (BODc), chemical oxygen demand (COD), total nitrogen, and total
phosphate washed off using the relationship:
POFF(I,L)A = POFFd.Du + F1(I) x SUS + F2(I) x SET (5)
where F1(I) and F2(I) are correction factors for constituent I; SUS and
SET are the amounts of suspended and settleable solids washed off, in
milligrams; and the subscripts A and U denote adjusted and unadjusted
values, respectively. Equation 5 is utilized to account for the insoluble
portion of constituents associated with suspended solids and settleable
solids.
95
-------
AREAS FOR IMPROVEMENT
Modeling experience and data collected since the original development
of the runoff-quality portion of SWMM have provided indications of potential
areas for improvement of the model. These areas for improvement will be
discussed under the general headings of constituent accumulation, constituent
washoff, and source identification.
Constituent Accumulation
Several features of the constituent accumulation equation (equation 1) ar
worthy of note. First, the equation assumes a linear buildup of constituents
on the land surface with no upper limit on the amount accumulated.' Second,
antecedent conditions are defined by the rather arbitrary 1.0-inch DRYDAYS
criterion. Third, the assumption is made that the land surface is completely
void of constituents at the start of the DRYDAYS period.
Several studies have suggested that the rate of accumulation of
constituents on urban surfaces is not linear and that there is a limit to
the amount of constituents that can accumulate between storms, regardless
of the length of dry period (Sartor and Boyd, 1972; Jewell and others,
1978; and Smith and Jennings, 1979). Data collected by Sartor and Boyd
(1972) suggest that the accumulation rate is largest for several days after
a period of street cleaning or rainfall, and then the rate decreases and
approaches zero. Apparently, constituents are resuspended by wind and land-
use activities such as vehicles moving along a highway. Barkdoll, Overton,
and Betson (1977) have pointed out that a particle with a diameter of 246
microns could be resuspended by air masses with velocities of less than 5
miles per hour. Sartor and Boyd (1972) found that from 40 to 90 percent of
96
-------
constituent loads on street surfaces are associated with particle sizes
less than 246 microns.
Overton and Meadows (1976) suggest the following alternative to
equation 1:
f - Kl - K2P (6)
where
P = amount of a particular constituent on subareas of a given land
use, in pounds per acre;
K^ = a constant rate of constituent deposition, in pounds per acre
per day;
K2 = a rate constant for constituent removal, in day ; and
t = time, in days
Integration of equation (6) yields:
P = !i (l-e-K2T) <7>
K2
where T is the accumulation time in days. Use of equation 7 would limit
constituent accumulation to a maximum value of K^/t^.
The definition of antecedent dry period as the number of days prior
to the storm during which the accumulated rainfall was less than 1.0 inch
(DRYDAYS in equation 1) should be further examined. For example, Alley
and Ellis (1979) compared simulated total nitrogen loads from two residential
areas near Denver, Colorado, with measured loads. They found SWMM to over-
estimate runoff loads with the extent of overprediction increasing with the
value of the model parameter DRYDAYS. DRYDAYS exceeded 30 days in five
of the six events simulated, resulting in as much as +1000 percent errors
in simulated loads.
97
-------
Traditionally, equation 7 has been derived with the assumption that
urban land surfaces were completely washed by the last cleaning, either
mechanical (street sweepers) or by storm runoff. In order to eliminate
this assumption, T could be redefined as:
T = t + t0 (8)
where t is the time since last cleaning, and
to.-i ma-!**) <9>
K2 K-L
where P£ is the available land-surface load in pounds per unit area of land
surface at the end of the last period of street sweeping or storm runoff.
Figure 1 illustrates the two options for exponential constituent accumu-
lation, as well as the linear buildup option currently in SWMM. Use
of equations 8 and 9 is limited to continuous simulation. However, it
would eliminate the need for an arbitrary definition of antecedent conditions.
An offshoot of the pervasiveness of linear accumulation equations in
urban runoff models appears to be the reporting of accumulation rates as
linear values in pounds per unit area of land surface per day. Even
Sartor and Boyd (1972), who first suggested nonlinear accumulation, reported
accumulation rates in pounds per curb mile per day. Often the linear rates
are determined by dividing the total measured runoff load over a period of
time by the contributing area and by the time period in days. Use of
such a variable in a runoff-quality model might yield good estimates
of annual runoff loads but individual storm-runoff loads could be poorly
simulated.
98
-------
'0 t0
TIME SINCE LAST CLEANING, EITHER MECHANICAL OR BY STORM RUNOFF, IN DAYS
Figure 1.—Constituent accumulation.
99
-------
In summary, more information is needed on the nonlinear characteristics
of constituent accumulation rates on urban land surfaces. Implementation
of equations 7, 8, and 9 as an option in urban runoff-quality models
such as SWMM might encourage the gathering and reporting of such information.
Constituent Washoff
Several features of the washoff equations warrant further discussion.
First, the exponential washoff equation (equation 2) appears to be the
subject of some misinterpretation. It is often assumed that equation 2
accounts for the effects of both runoff intensity and duration on water
quality. In actuality, because equation 2 is a function of the product
of runoff intensity and duration, the amount of constituents washed off
during a storm, according to equation 2, is a direct function of the total
volume of storm runoff. Equations 3 and 4 are used to account for
the effects of runoff intensity.
Generally, the value of a in equation 2 is set at a constant value
of 4.6. The rationale for this value is usually stated as—"Assuming that
a uniform runoff of 0.5 inches per hour would wash off 90 percent of a
constituent in 1 hour." This is a misleading statement in that the
actual assumption is that a runoff volume of 0.5 inches will wash off 90
percent of a constituent from urban land surfaces regardless of duration
and whether or not the runoff was uniform. Even this statement is only
the true assumption if the availability factors ASUS and ASET in equations
3 and 4 are equal to 1.0 for all time steps.
100
-------
Generally, the value of a is set to 4.6 for all sites and all consti-
tuents. However, several studies have reported—the perhaps, not unexpected
result—that a varies for different constituents and for different watersheds
(Barkdoll, 1975; Smith and Jennings, 1979; and Ellis and Sutherland, 1979).
Both Barkdoll (1975) and Sonnen (1979) have found 4.6 to be an overestimate
of a in many cases.
The availability factor equations (equations 3 and 4) were derived
using constituent accumulation data from Chicago streets and runoff-quality
data from Cincinnati. Thus, the coefficients in these equations, if not
the form of the equations, could be very site specific. Dustfall data
reported for the Chicago and Cincinnati studies show considerable difference
(Metcalf and Eddy, Inc., and others, 1971). Thus, the applicability of
Chicago constituent accumulation information to a Cincinnati watershed is
questionable.
The original formulation of SWMM did not contain the correction
factors (F1(I) and F2(I) in equation 5) for insoluble constituents. A
value of F1(I) of 0.05 for BOD was somewhat arbitrarily assigned to improve
initial testing of SWMM. The rationale for this adjustment was that in-
soluble BOD was not included in the Chicago data used in developing the
model. As other constituents were added to the model, similar corrections
were made. When the capability for settleable solids simulation was added
to the model, values of the second correction factor F2(I) were assigned.
However, inclusion of a settleable solids correction was redundant, since
settleable solids are a fraction of suspended solids. Studies by Jewell
101
-------
and Adrian (1978) and by Alley and Ellis (1979) found that the correction
for insoluble constituents (equation 5) is the main source of simulated
constituent washoff for BOD5, COD, total nitrogen, and total phosphate.
For example, Alley and Ellis (1979) found that application of equation
5 resulted in an average 22-fold increase in simulated total nitrogen
washoff over that predicted by the exponential washoff equation (equation 2)
alone. Jewell and Adrian (1978) concluded that:
The original calibration problems appear to
have been caused by use of the Chicago data
to predict pollutant buildup rather than a
need for any insoluble pollutant correction.
A later study (Sartor and Boyd, 1972) indi-
cated that pollutant buildup rates for
other cities are one to two orders of
magnitude greater than those reported in
the Chicago study. This alone could explain
why initial predicted pollutant washoff was
low.
Inclusion of site specific correction factors, such as those embodied
in equations 3-5, complicates model calibration and may hinder achievement
of satisfactory results. However, equation 2 by itself will often be
insufficient for-predictive purposes, due to the limitations previously
discussed.
An alternate approach could be to eliminate equations 3-5 and
include an availability equation for each constituent such as:
A(I,L) = c-^I.L) + c2(I,L) x R (10)
where A(I,L) is the availability factor for constituent I on land use L,
and Cj^d.L) and c2(I,L) are coefficients which could be calibrated for each
constituent and land use based on local data. Alternately, a term related
to shear stress might be included in equation 10 in place of R. Limitations
102
-------
of this approach include difficulties in transferring the coefficients
C-L and c2 to unsampled watersheds and the fact that constituent availability
relationships would be expected to change from event to event and within
events due to changes in particle size distribution of constituents.
Another approach might be to replace equations 2-5 with a washoff
scheme based on sediment transport theory. An important consideration
would be simulation of constituent transport by particle size. Particle-
size distributions are not only important for simulating transport of consti-
tuents, but also for assessing the effectiveness of management strategies
such as street sweeping and detention storage, and for determining impacts
on receiving waters.
The U.S. Geological Survey is currently investigating these and other
approaches.
Source Identification
An important aspect of runoff-quality modeling is often identification
of sources. Two potentially important sources of runoff loads generally have
been neglected in urban runoff-quality modeling. These are pervious-area
runoff and atmospheric fallout.
Pervious-Area Runoff
The constituent accumulation and washoff equations discussed, thus
far, have pertained to impervio.us-area runoff only. However, the relative
contributions of pervious and impervious areas to runoff loads will
affect the utility of management strategies such as street sweeping, as
well as the outcome of modeling.
103
-------
A study by Barkdoll, Overton, and Betson (1977) suggests that "semi-
pervious and pervious areas are highly significant in their contribution
to urban water pollution." Theoretical studies by the originators of
SWMM (Metcalf and Eddy, Inc., and others) indicated that "very large
rates of runoff would be required to remove dust and dirt from grass plots,
and that unless erosion takes place from ungrassed areas, the contribution
of pervious surfaces to suspended solids content is minor." However, they
/
noted that runoff from pervious surfaces may contain significant amounts
of soluble constituents.
Several indicators of the relative contributions of pervious and
impervious areas to runoff loads can be used. These include empirical
knowledge of sources, results of regression analyses, distributions of
constituent concentrations and loads over storm hydrographs, results of
street-surface sampling, and the effectiveness of management strategies
such as street sweeping.
Empirical knowledge of sources can provide an indication of the
relative contributions of pervious and impervious areas to runoff loads
for certain constituents. For example, one might expect pesticides to
originate primarily from pervious areas and lead which is associated with
automobile exhaust to originate primarily from impervious areas. Other
constituents such as nitrogen species which are associated with organic
pollution, fertilizers, and automobile sources may originate primarily
from either pervious or impervious areas.
Regression analyses of storm-runoff loads with parameters such as
average daily traffic or pavement condition may provide an indication of
sources by identifying significant variables.
104
-------
The distribution of constituent concentrations and loads over storm
hydrographs may provide some indication of the relative contributions of
pervious and impervious areas to runoff loads. For example, the decay
coefficient («) in equation 2 would be expected to be smaller for con-
stituents with a larger pervious-area contribution to runoff loads. In
addition, the relative proportions of pervious- and impervious-area
contributions to runoff, as determined from a rainfall-runoff model, might
be compared to trends in constituent concentrations and loads.
Sampling of street-surface solids in conjunction with runoff-quality
studies may provide a useful indication of the potential impervious-area
contributions to storm-runoff loads. A limitation of street-surface sampling
is that it provides an estimate of the total accumulation of constituents
on the street surface, but not of the street-surface load "available" for
washoff by storm runoff. However, analysis of constituent partitioning
amongst street-surface solids by particle size can provide an indication
of constituent availability for washoff.
Finally, the effectiveness of management strategies such as street
sweeping may provide an indication of the relative importance of pervious
and impervious areas as sources of storm-runoff loads. For example,
effectiveness of street sweeping as measured by removal of constituents
from streets could be compared to its effectiveness in reducing constituent
loads in storm runoff.
Due to rapid response times of most urban watersheds, of primary
interest would be surface runoff and quick-return flow (i.e., interflow
with shallow penetration of the soil). Chemical constituents added to
solution would normally be those characteristics of surficial soils.
105
-------
Unfortunately, techniques for predicting the interactions between these
soils and the runoff are not far advanced. Given the state-of-the-art,
simple empirical equations for pervious-area runoff might be included
in the model. For example, Leonard and others (1979) report simple power
functions decribing soil-based herbicide transfer to runoff. They also
note that average storm herbicide concentrations in runoff were correlated
with herbicide concentrations at the 0- to 1-centimeter depth increment
of the watershed soils at the time of runoff.
Atmospheric Fallout
Most atmospheric fallout samples have been collected with devices that
remain open continuously. Generally, analyses of such samples are reported
as bulk precipitation. Bulk-precipitation data provide measures of the
dissolved material in wet precipitation (wetfall) plus water-soluble materials
that have been leached from the dry fallout (dryfall) as the sample awaited
processing. From the modeling standpoint, these data are more useful if
wetfall and dryfall are separated. The data are also more useful if the
water-insoluble components are included in the wetfall and dryfall analyses.
Various studies have suggested that both wetfall and dryfall may be important
sources for many constituents in runoff from urban areas (Betson, 1978; and
Barkdoll, Overton, and Betson, 1977).
Urban runoff can be viewed as chemically modified rainwater. Accounting
for the original chemical composition of this rainwater may be important
in terms of both model reliability and evaluation of management practices.
For example, wetfall contributions to runoff loads would be unaffected
by street sweeping practices. The simplest means of accounting for wetfall
106
-------
contributions to runoff loads in a model would be to add a concentration
representing wetfall to the concentrations of a constituent predicted
by the washoff equations. This adjustment could be constant or could
vary according to season. Although it is well known that the chemistry
of wetfall can change within a particular storm, few data exist to quantify
this effect. Thus, a simple adjustment might be all that is currently
justified. Certainly, more wetfall data in conjunction with runoff-quality
data are needed.
Dryfall data are generally collected without consideration of re-
suspension. Thus, dryfall data could provide an indicator of the value
of K, in equation 7. Ideally, dryfall data may provide a basis for
transferring constituent accumulation information from one urban water-
shed to another. As for wetfall, seasonal values for dryfall indicators
might be used.
SUMMARY AND CONCLUSIONS
An analysis of the SWMM surface-runoff-quality algorithms has been
presented. Limitations of the constituent accumulation equations include
the assumption of linear buildup of constituents on urban surfaces with
no upper limit on the amount accumulated, the arbitrary antecedent-conditions
criterion (DRYDAYS), and the assumption of completely clean urban surfaces
at the start of the DRYDAYS period. An alternative approach which eliminates
these assumptions has been presented.
107
-------
Several features of the washoff equations were also discussed. These
included the misinterpretations often resulting from use of the exponential
washoff equation (equation 2) and the limitations of the availability factor
equations (equations 3 and 4) and the insoluble correction factor equation
(equation 5). A different approach might include deletion of equation 5
and substitution of availability factor equations for each constituent in
place of equations 3 and 4. Either a runoff rate or shear stress term
might be included in these availability factor equations. Alternately, a
sediment transport approach might be used.
Finally, the need to address the contributions of pervious-area runoff
and atmospheric, fallout to runoff loads was discussed.
108
-------
REFERENCES
Alley, W. M. and Ellis, S. R., 1979, Rainfall-runoff modeling of flow and
total nitrogen from two localities in the Denver, Colorado, Metropolitan
Area, in Stormwater Management Model (SWMM) Users Group Meeting,
May 24-25, 1979, Proceedings: U.S. Environmental Protection Agency,
EPA-600/9-79-026, p. 362-403.
Barkdoll, M. P., 1975, An analysis of urban stormwater quality and the
effect of dustfall on urban stormwater quality, East Fork of Third
Creek, Knoxville, Tennessee: M.S. Thesis, Department of Civil Engi-
neering University of Tennessee, Knoxville.
Barkdoll, M. P., Overton, D. E., and Betson, R. P., 1977, Some effects of
dustfall on urban stormwater quality: Water Pollution Control
Federation Journal, V. 49, no. 9, p. 1976-1984.
Betson, R. P., 1978, Bulk precipitation and streamflow quality relationships
in an urban area: Water Resources Research, v. 14, no. 6, p. 1165-1169.
Ellis, F. W., and Sutherland, R. C., 1979, An approach to urban pollutant
washoff modeling, in International Symposium on Urban Storm Runoff,
July 23-26, 1979, Proceedings: University of Kentucky, Lexington,
p. 325-340.
Han( J. and Delleur, J. W., 1979, Development of an extension of ILLUDAS
model for continuous simulation of urban runoff quantity and discrete
simulation of runoff quality: Tech Rept no. 109, 136 p.
Jewell, T. K., and Adrian, D. D., 1978, SWMM stormwater pollutant washoff
functions: American Society of Civil Engineers Proc., Journal
Environmental Engineering Division, v. 104, no. EE5, p. 1036-1040.
Jewell, T. K., Nunno, T. J., and Adrian, D. D., 1978, Methodology for
calibrating stormwater models, in Stormwater Management Model (SWMM)
Users Group Meeting, May 4-5, 1978, Proceedings: U.S. Environmental
Protection Agency, EPA-600/9-78-019, p. 125-169.
Leonard, R. A., Langdale, G. W., and Fleming, W. G., 1979, Herbicide
runoff from upland piedmont watersheds-data and implications for
modeling pesticide transport: Journal Environmental Quality, v. 8,
no. 2, p. 223-228.
Metcalf and Eddy, Inc., University of Florida, and Water Resources Engineers,
Inc., 1971, Storm Water Management Model: U.S. Environmental Protection
Agency, EPA-11024 DOC 07/71, 4 volumes.
109
-------
Overton, D. E., and Meadows, M. E., 1976, Stormwater modeling: New York,
Academic Press, 358 p.
Sartor, J. D., and Boyd, G. B., 1972, Water pollution aspects of street
surface contaminants: U.S. Environmental Protection Agency, EPA-R2-
72-081, 236 p.
Smith, P. E., and Jennings, M. E., 1979, Accumulation and washoff of
pollutants on urban watersheds: EOS, Transactions of American
Geophysical Union, v. 60, no. 18, p. 259.
Sonnen, M. B., 1979, Information needs for the quality of urban runoff:
American Society of Civil Engineers, National Convention, Atlanta,
Georgia, Preprint no. 3789, p. 33-50.
U.S. Army Corps of Engineers, 1976, Storage, treatment, overflow, runoff
model (STORM): Hydrologic Engineering Center, Davis, California.
110
-------
THE USE OP SWMM TO ECONOMICALLY MODEL
SURCHARGED COMBINED SEWER SYSTEMS
by
WILLIAM M. PARKER HI-/
The Village of Skokie, Illinois (pop. 70,000) is a
near-northern suburb of Chicago approximately 10 square
miles in area. The Village owns and operates a combined
sewer collection system tributary to the Metropolitan
Sanitary District of Greater Chicago (MSDGC) interceptor and
treatment system. Intercepted flows are treated at the
MSDGC's Northside Sewage Treatment Works, located in Skokie.
Flows in excess of the interceptor system capacity are
discharged to the nearby North Shore Channel. An area map
relating the above-described features is shown in Figure 1.
The combined sewer and interceptor systems each have
limits which are exceeded during certain precipitation
events. For many sections of Skokie, the collection system
limit is exceeded several times each year. The result is
that combined sewage backs up into basements and onto
streets on a Village-wide basis and is not restricted merely
to downstream portions of the collection system near the North
Shore Channel. The MSDGC interceptor system capacity is
currently exceeded over 90 times each year and the resulting
water pollution problems have a major regional impact, as
there are 53 other suburban communities connected to the
MSDGC system.
The MSDGC has adopted and is now implementing the
Tunnel and Reservoir Plan (TARP), which will essentially
increase systemwide interceptor capacity and limit overflows
I/ Environmental Engineer, Harza Engineering Company
150 South Wacker Drive, Chicago, Illinois 60606
111
-------
I
NORTH
I
0 1 2
Scale In Thousands Of Feet
LEGEND
Subdistrict Boundary —— ——
Area Serviced By iiljilliliilil
Other' Sewer District $;888S:!88&^8
Figure 1
SUBDrSTRICT DELINEATION
COMBINED SEWER CONVEYANCE PROBLEMS
SKOKIE • ILLINOIS
112
-------
to local waterways. The system components of TARP are
designed to accomodate improvements in the local collection
systems, such as Skokie's, by providing a hydraulic outlet
for the 5-year rainfall event. Thus, while the MSDGC is
proceeding to improve the outlet conditions for the local
collection systems, the initiative to correct remaining
local system inadequacies remains with each of the tributary
communities.
A study —has recently been completed for the Village
which addresses local system inadequacies and has recommended
alternative corrective measures to upgrade the Skokie system
in order to obtain the full benefit of TARP. The RUNOFF and
EXTENDED TRANSPORT blocks of the U.S. EPA's Stormwater
Management Model (SWMM) were selected to conduct extensive
computer simulation of the existing collection system. The
EXTENDED TRANSPORT block of SWMM was selected to be used
rather than the conventional TRANSPORT block so that
simulated water surface vs. time data could be calibrated
and verified against similar data collected in the field.
Model Set-up and Development
The Skokie collection system consists of three major
sub-districts served by large diameter trunk sewers running
in an east-west alignment toward the North Shore Channel
and the MSDGC interceptor. The Mainstream Tunnel component
of TARP runs directly below this interceptor and will accept
flows from Skokie in excess of the interceptor capacity at
the points which the three major trunk sewers join the
interceptor. Three catchments were selected to be simulated
with SWMM, each corresponding to a major trunk sewer
sub-district, namely the Emerson, Main, and Howard Street
catchments.
The Howard Street catchment was selected to be
monitored with four, continuous water surface vs. time
113
-------
recorders placed in key manholes. The Howard Street
catchment was chosen in this regard principally because
it contained a 105 acre area which had previously
2/ 3/
been studied by the Village Engineering Department.— —
This area, known as Fairview South, had been the subject of
a Village initiated rooftop downspout disconnection program
and offered a comprehensively developed data base. Fairview
South, therefore, was selected as the area to develop the
most economically suitable fashion to apply SWMM on a
Village-wide basis and obtain reasonable simulation results.
Because 1) system surcharging is a Village-wide problem
in Skokie, and 2) it was not practical to simulate the entire
collection system due to model limitations and economics,
lateral and branch sewers were necessarily simulated as
gutter/pipes in the RUNOFF block for the overall sub-district
simulations. Accurate characterization of gutter/pipes to
best represent lateral and branch sewer performance under
surcharged conditions was therefore necessary to develop
realistic alternative correction actions. This need resulted
in the study of a two block length of Lunt Avenue within
Fairview South at two levels of detail to accurately
characterize gutter/pipe capacities for larger scale,
Village-wide assessment.
An additional concern was the loss of model resolution
and possible accuracy when model sub-catchments were
increased in size from Lunt Avenue (10 acres) to Fairview
South (105 acres). This concern prompted an additional
sensitivity analysis of key RUNOFF block parameters on two
levels of detail for the entire Fairview South area. A
description of each effort follows.
114
-------
Characterization of Gutter/Pipe Capacity
The Lunt Avenue area was modeled at two levels of
detail to determine the sensitivity of the results to
changes in gutter/pipe capacity. The gutter/pipe capacity,
which is implicitly stated in the RUNOFF block of SWMM, is
specified in terms of slope, diameter, and friction
coefficient. The gutter/pipe capacity is important because
it affects the shape and peak of the hydrograph that is
routed in the EXTENDED TRANSPORT block. Accurate simulation
of the hydrograph that is to be routed was fundamental to the
sizing of relief conveyance facilities for the Village trunk
sewers.
The Lunt Avenue area consists of two typical blocks in
the Fairview South area. The land use in this 10 acre area
is all single-family residential. As shown on Figure 2, the
area drains to catch basins which direct runoff to a street
lateral. This lateral is approximately 1200 feet long with
an average slope of 0.0026 and an outlet diameter of 18
inches. Based on a Manning's "n" value of 0.015 and a
slope of 0.0055 in the last downstream pipe section, the
lateral outlet capacity under full flow, gravity conditions
is 6.7 cfs. This outlet capacity is also based on the
assumption that the trunk sewer is not surcharged.
It appears that this street lateral has inadequate
capacity to convey the runoff from an event with a 5-year
recurrence interval. Based on the Chicago Hydrograph Method
by Kieffer and Tholir4/, the required outlet capacity would
be about 15 to 18 cfs for the 5-year event. The apparent
inadequacy of the lateral sewer capacity indicates that
surcharging during the 5-year event will occur even with no
surcharging in the downstream branch or trunk sewers. If
surcharging occurs in the lateral sewer, then the hydrograph
from the Luiit Avenue lateral will have a peak flow for the
115
-------
FAIRVIEW
LANE
3D
>
s
m
r-
1
r
i
4
i
r
I
L
2>
<
m
z
c:
rn
1
mt A
ibcat
>de
ransp
i i 2
3 1 4 1
« !
. . .. •*• — - •- •!'- • ..... . . _A
v^ X
P X ,
5 1 6
* S
L7 1 8
•
o
r-
>
M
m
i
•
-4-
•
1
w
m
*
i
i
-i-
•
i
•
9 i 10 ;
• i
; 11 i 12
i
A
F *
13 j 14
i t ^^
? w
15 1 16
•
VERGNE _
1 t
1
-1 !
1
2p
1
i
COYLE
r i
LEGEND
venue Study Area — — — —
chment Boundary — —
AVENUE
AVENUE
i i
Fi
LUNT AVENUE STUDY
i*
•<.
m
21
c:
m
1
gure
ARE
SKOKIE . ILLINOIS
-------
5-year event greater 6.7 cfs. To properly size downstream
relief facilities it appears that the lateral (gutter/pipe)
capacity should be specified as greater than 6.7 cfs.
The most detailed characterization of the Lunt Avenue
area was used to evaluate the problem of specifying the
gutter/pipe capacity. For this most detailed characteriza-
tion, the Lunt Avenue Area was delineated into 16
sub-catchments and the RUNOFF and EXTENDED TRANSPORT blocks
set up to model the two block long sewer system.
The detailed characterization provided an opportunity
to simulate the routing of flow through the street lateral.
Results of this analysis were then compared to the results
of the coarser Lunt Avenue characterization, for which the
area was modeled with the RUNOFF block as a single
sub-catchment and the lateral sewer was represented as a
gutter/pipe (with no routing). Because the cost and time
involved in characterizing the entire study area at the same
level as the more detailed Lunt Avenue characterization was
prohibitive, the coarser characterization required adjust-
ments so that its results would closely approximate the
more detailed characterization results. By this comparison,
a means of calculating gutter/pipe capcity for larger areas
was developed.
The more detailed level of modeling tended to exaggerate
the effect of downspout disconnection. To ensure that the
results of this analysis were reasonably representative, two
conditions were identified for the RUNOFF block. Condition
1 was set up assuming that all roof area was pervious
(downspouts disconnected), resulting in a total impervious
area of 17 percent. Condition 2 represented the situation
wherein the roof areas were impervious and connected directly
to the lateral sewer, resulting in a totaZ impervious area
of 36 percent.
117
-------
Four simulation runs were made for the more detailed
characterization of the Lunt Avenue area. Conditions 1 and
2 were each run for the 2-year and 5-year (both 2-hour)
rainfall events, and the four outlet hydrographs are shown
in Figure 3. The 5-year event, as simulated, caused
surcharging and indicated that for both Conditions 1 and 2
the peak flow to an unsurcharged downstream sewer would
exceed 6.7 cfs (full flow capacity of the 18-inch lateral).
Also shown on Figure 3 are two hydrographs from the coarser
characterization. These hydrographs are for the 2-year
recurrence interval event and indicate (from the flattened
peak) that a specified gutter/pipe capacity of 6.7 cfs
constrains the flow. By setting the outlet capacity on the
basis of having surcharged conditions (in effect, increasing
the gutter/pipe slope), it appears that a more
representative hydrograph is determined.
Neither Condition 1 nor Condition 2 quite accurately
represents roof areas being disconnected or connected,
respectively. The RUNOFF block simulates various abstrac-
tions (rainfall losses) from pervious areas. Considering
roof area to be pervious in Condition 1 allows the model
to abstract too much rainfall from an artificially
designated pervious area. Thus, the resultant modeled
outlet hydrograph peak flow and total runoff volume estimate
are lower than would be expected to actually occur. In
Condition 2, the model routes roof area runoff by gutters
to a catch basin rather than entering the runoff directly
into the lateral sewer. Consequently, the resultant
modeled outlet hydrograph peak flow may be inaccurate. The
actual hydrograph for Condition 1 is estimated to fall
between the two modeled hydrographs.
Inspection of the simulated hydraulic gradeline along
Lunt Avenue indicated that even with free outfall conditions
to the branch sewer along Lavergne Avenue, surcharging
118
-------
9.6
8.8
8.0
7.2
6.4
Modified Outlet Capacity
Based on 4' Surcharge on Most Upstream Node
-/y-;"'
/I \ ! \ \ *— 18" Dia. Full Flow Capacity =6.7 cfs
_ i \ / \ .^^ _ _
n \j - A- \
// r\ v^ \
5.6 -
01
o
3
s
4.8-
4.0
3.2
2.4
1.6
0.8
';/ •'» \
W\\
LEGEND
CODE
RECURRENCE LEVEL OF SYSTEM PERCENT
INTERVAL(yrs) CHARACTERIZATION IMPERVIOUS_
Coarse i
Coarse 2/
Fine
Fine
Fine
Fine
26
26
36
17
36
17
I/ Outlet Capacity =6.7 cfs
2/ Outlet Capacity =7.3 cfs
10 20 30 40 50 60
90
TIME (minutes)
120
150
180
Figure 3
OUTLET HYDROGRAPHS
LUNT AVENUE AREA CHARACTERIZATION
COMBINED SEWER CONVEYANCE PROBLEMS
SKOKIE • ILLINOIS
-------
would occur and could cause basement backups. For the 5-year
event, Condition 2 (rooftops directly connected), the hydraulic
gradeline rose to within three feet of the ground surface
midway along the two block length and was within only
several inches of the ground surface at the upstream end of
the lateral. The duration of the surcharged condition is
about 30 minutes for the 5-year event. The 2-year event,
Condition 2, also caused surcharging, but to a smaller
degree, as the most upstream manhole was surcharged to
within four feet of the ground surface for approximately
10-15 minutes.
Two conclusions were reached in studying the analysis
of the Lunt Avenue area. One conclusion is that the lateral
sewers are not adequate to convey the 5-year event without
surcharging. Thus, improvements to increase the trunk
sewer and branch sewer conveyance capacity would not
entirely eliminate surcharging in the laterals. The second
conclusion was that the gutter/pipe capacity should be
specified to reflect surcharged conditions in the laterals.
The extent of simulated lateral surcharging is
insignificant relative to actual conditions if the trunk
and branch sewer do not surcharge. When the trunk sewers
surcharge, the duration, magnitude, and extent of lateral
surcharging is greatly increased.
Evaluation of the Lunt Avenue Area indicated that the
gutter/pipe capacity should be calculated for the
Village-wide analysis as 0.9 cfs/acre tributary to EXTENDED
TRANSPORT block nodes. The Skokie sewer system was checked
to determine whether or not a parametric analysis was
applicable for specifying outlet capacity of each
gutter/pipe. Most laterals appeared to be sized similarly
to the Lunt Avenue lateral and were expected to have
similar capacity limitations. Thus, for 10 acre (two
block) sub-catchments, the outlet sewer capacity was
120
-------
calculated to provide 0.9 cfs/acre for each gutter/pipe.
Based on overall similarity of the various lateral systems,
this quantity was subsequently used for gutter/pipes in all
sub-catchments rather than individual analysis of the
existing outlet pipe for the sub-catchment.
Sensitivity to Other Runoff Parameter
Necessary data for the RUNOFF and EXTENDED TRANSPORT
blocks of SWMM were developed on two levels.for the Fairview
South Area. Initially, the area was broken down into 16
sub-catchments of approximately 10 acre size (fine). Each
sub-catchment represented a two block area approximately
similar to the Lunt Avenue Area. Laterals through these
sub-catchments were represented by the gutter/pipe routine
of the RUNOFF block, with the branch sewer (running north
and south down Lavergne Avenue) modeled in the EXTENDED
TRANSPORT block. The Fairview South Area was also
delineated as a single 105 acre sub-catchment (coarse) for
use as part of the entire Howard Street catchment modeling
effort. Figure 4 illustrates the two characterizations of
the Fairview South Area.
The Fairview South Area was as a calibrating link
between large scale, 100 acre sub-catchment modeling and the
smaller, 10 acre size approach, which was represented by the
Lunt Avenue area analysis. Investigations were made to
determine the sensivity of various model output parameters
to several varied input parameters in the RUNOFF block
necessarily associated with increased coarseness of drainage
area representation. Various outlet hydrographs resulting
from the application of the 10-year, 2-hour storm on the
Fairview South Area are shown in Figure 5. The
conditions producing each hydrograph are also shown in
Figure 5.
121
-------
J L
ro
ro
Li) 1
r < <
r ITCH
_J [
i
E3TES
( 1 I
' ~I 1 1
FAIRVIEW
L __l L_ .... J
! 1 1
LUNT
L IL _._ _l
[ 1 L I
COYLE
LI J
I I
MORSE
1 L _
r~ ki i
FARWELL < I
r £i [
1 Ujl , 1
r1 -• -\
\((
D '•X
Ayevl
L 1_
1
1
1
1
1
1
r
I
i
L
^
\
^
1 ^:
AVE
LANE
_ J
1
AtfC
1
|
AVP.
1
1
_AVt
AVg
D
LJ
COARSE CHARACTERIZATION
LEGEND
Catchment Boundary
Node
Transport System
Gutter/Pipe With
Identifying Number
OJJ 1
1 - —fflfrr.tr".'-"~~
FINE CHARACTERIZATION
Figure 4
CHARACTERIZATION OF FAIRVIEW SOUTH AREA
COMBINED SEWER CONVEYANCE PROBLEMS
SKOKIE • ILLINOIS
-------
i:>o I j i i r
110 -
ro
3
3
50
40
30
2O
10
1 1 1 1 1 T
LEGEND
CODE
MODELING
TECHNIQUE UTILIZED
No Gutter Pipe
Runoff only
Runoff only
Runoff only
Runoff only
Runoff only
Runoff + Transport
WIDTH
(ft)
4,120
4,120
7,257
20,000
26,922
26,922
LENGTH
(ft)
320
320
320
320
2,575
1O 20 30 40 50 60
90
TIME (minutes)
120
150
180
Figure 5
OUTLET HYDROGRAPHS • 10 YEAR « 2 HOUR STORM
FAIRVIEW SOUTH CHARACTERIZATION
COMBINED SEWER CONVEYANCE PROBLEMS
SKOKIE • ILLINOIS
-------
Through this sensitivity analysis, it was found that
the total volume of runoff was most sensitive to the width
of sub-catchment. The width of sub-catchment is an arbritary
measure of the total width of overland flow and is commonly
approximated by double the length of the longest (main)
drainage conduit through the sub-catchment, since two
plane catchments contribute flow along this length. It was
found that as the size of the sub-catchment increased, the
width of sub-catchment lost its significance as previously
described. Additionally, as the size of the sub-catchment
was increased, more rainfall abstraction was simulated than
had been anticipated. This is thought to be due to a
relatively large increase in the length of overland flow for
increased sub-catchment sizes without a corresponding
increase in the width of sub-catchment.
Model output, specifically the shape of the outlet
hydrograph, was found to be fairly sensitive to land slope,
as evidenced when default values for the parameter were
replaced with measured quantities. Model output was not
found to be excessively sensitive to the other input para-
meters tested, namely imperviousness, overland flow "n"
values, or infiltration rates. As such, input quantities
were determined from existing data where available. Where
existing data were not available, default values recommended
by the SWMM User's Manual — were used.
It was determined from efforts to match hydrographs for
the coarse and fine representations of the Pairview South
Area that the width of sub-catchment could reasonably be
expressed as a function of sub-catchment size. The
following relationship was therefore used to determine
sub-catchment widths for other sub-catchments in the study.
(1) Width of sub-catchment (feet) = 250 ft/acre x
sub-catchment acreage
124
-------
Calibration and Verification
The general purpose of the field studies conducted for
the Skokie study was to gather data concerning the existing
sewer system and factors that affect the quality, amount,
and rate of runoff to the sewer system. More specifically,
field data were obtained to permit comparison of actual and
predicted hydraulic responses of the sewer system to several
precipitation events. The comparison of field data and
model output showed that the model produces representative
results within reasonable accuracy and verified the
conclusions derived during the analysis of model results.
Monitoring and Data Collection Program
Field studies were limited to the Howard Street
catchment. The Howard Street catchment was chosen as
representative of a typical drainage area within the Village.
Additionally, included within this sub-district is the
Fairview South Area, unique in the respect that approximately
85 percent of the roof downspouts have been disconnected
from the sewer system.
Rainfall data for storms that occurred during the flow
monitoring program were necessary for calibration of SWMM
and verification of the program once calibration was
completed. Rainfall data were collected for the storms of
June 5, 8, 11,,17, 24, and 30, 1977 from the MSDGC. The
MSDGC maintains a rainfall gauge at their Northside Sewage
Treatment Works, located in southeastern Skokie at the
intersection of Howard Street and McCormick Boulevard.
Field data were also collected with regard to Manning's
"n" values at various points within the Howard Street sewer
system. These data were obtained to provide representative
input data to the EXTENDED TRANSPORT block. Roughness coef-
ficients were calculated based on several different field
125
-------
velocity and depth of flow measurements as well as other
hydraulic properties of the particular sewer section.
Continuous measurement of water surface elevations at
selected points within the Howard Street sewer system was
also included in the field studies program. These
measurements were conducted for about 45 days to obtain data
during the six before-mentioned precipitation events.
Information collected from four of these events was used
for calibration of SWMM. Two of these events were used for
verification. No quality measurements were taken, as only
the capacity of the system was investigated in the field
monitoring program.
Four recorders were used for continuous monitoring of
water surface elevations. One unit was located in the
Fairview South Area to check the response in the sewer
system to the control measures previously mentioned.
This unit was located in the manhole at the corner of Fitch
Street and Lavergne Avenue. Another re.corder was located
at the corner of Howard Street and Lavergne Avenue to
monitor the area downstream of Fairview South. The
third and fourth recorders were located in manholes along
Howard Street in the main trunk sewer itself. The locations
along Howard Street provided data representative of the
system response to runoff from both the total and
intermediate areas.
Presentation £f Results
The following section presents information on rainfall
and runoff for events observed in the field during the
course of the study. Six independent rainfall/runoff
events were observed during the field study program.
Pertinent data concerning these six events are summarized in
Table 1.
126
-------
Table 1
DATA FOR FIELD
MONITORED STORMS
2/
_ . _, Total Maximum i/ Duration Recorders
Date Rainfall Intensity-7 of Rainfall Operational
June 05, 1977 0.95 in. 0.44 in./hr. 4 l/2hrs. 1,3,4
June 08, 1977 0.32 in. 0.60 in./hr. l/2hr. 1,3,4
June 11, 1977 1.90 in. 0.44 in./hr. 3/4hrs. 1,3,4
June 17, 1977 0.77 in. 0.42 in./hr. 5 3/4hrs. 1,4
June 24, 1977 0.20 in. 0.12 in./hr. Ihr. 1,2,3,4
June 30, 1977 1.49 in. 0.70 in./hr. 7hrs. 1,2,3
I/ Intensities for a 15 minute duration.
2/ Key to Recorders: 1 - Howard and CNW RR (node 2)
2 - Howard and Kenneth (node 8)
3 - Howard and Lavergne (node 12)
4 - Fitch and Lavergne (node 17)
The events of June 05, 08, 11, and 17, 1977 were used
for calibration purposes. The events of June 24 and 30,
1977 were used to verify the model. Representative
calibration and verification plots are presented in Figures
6 and 7, respectively.
The calibration process pointed out the sensitivity of
the gutter/pipe capacity to the results of the simulation.
It appeared that surcharging in the trunk sewers caused the
sub-catchment outlet hydrograph to be affected. The
simulated outlet hydrograph is also affected by the speci-
fied characteristics of a gutter/pipe. During surcharging
of the trunk sewer the actual hydraulic gradelines on the
lateral and branch sewers will be less than the pipe slope
at certain times. Thus, the actual outlet peak flows might
127
-------
789
TIME (hours)
RAINFALL HISTORY
10 11 12
ffit
ffi
6739
TIME (hours)
NODE 2
10 11 13
^.L.. .r- ^ .- ---:L_) [-- j'i
10 11 12
NODE 12
NODE 17
LEGEND
Field Observations
Simulation Results
Sewer Crown Elevations
128
Figure 6
JUNE 5 • 1977
WATER SURFACE ELEVATIONS
COMBINED SEWER CONVEYANCE PROBLEMS
SKOKIE • ILLINOIS
-------
10 '.'. 12 an
TIME (hours)
RAINFALL HISTORY
NODE 2
--..
4_Z^ '~~" p'~'.-- :p""- .----. p-.
NODE 8
7 MI 8 9 10 11 12 pn 1 2 3
TIME (hourJI
NODE 12
LEGEND
Field Observations
Simulation Results
Sewer Crown Elevations - — •
Figure 7
JUNE 30 • 1977
WATER SURFACE ELEVATIONS
COMBINED SEWER CONVEYANCE PROBLEMS
SKOKIE • ILLINOIS
129
-------
be less than those simulated by the RUNOFF block of SWMM.
Because of program limitations, it is not possible to
adequately compensate for the varying gutter/pipe outlet
conditions. Thus, for each event a specific gutter/pipe
slope was selected with all other parameters held constant
to obtain a reasonable match between observed and simulated
results.
For this study, it was most important that the
simulated results match the observed depth measurements at
the most downstream recorder. The response simulated for
this location is less subject to errors caused by
limitations the model has in simulating runoff from all the
sub-catchments. Over the full length of the modeled system,
these limitations tend to balance out resulting in a
reasonable representation of routed flow. In contrast, the
comparison of simulated results and observed measurements at
the recorder located in Fairview South is affected to the
greatest degree by the modeling limitations.
Summary
Flows in the Village of Skokie combined sewer collection
system were modeled with the RUNOFF and EXTENDED TRANSPORT
blocks of SWMM for six observed precipitation events. Model
set-up and development indicated that 1) gutter/pipe
characterization, and 2) width of sub-catchment were key para-
meters in modeling the system because of Village-wide .
surcharging experienced during frequent rainfall events.
Accurate representation of system flows was possible at a
computer run time of about 0.06 CPU/acre modeled for a 2
hour storm.
130
-------
References
1. "Study of Combined Conveyance Problems." Harza
Engineering Company, Chicago, 111., 1978.
2. "Fairview South Sewer Study." Engineering Dept. Staff
Report, Village of Skokie, 111., 1974.
3. Pahlke, E. C., and Mark, S. P., "An Interim Flood
Control Program for a Combined Sewer Residential Area."
Proceedings from National Symposium on Urban Hydrology,
Hydraulics, and Sediment Control, July 26-29, 1976,
Univ. of Kentucky, Lexington, Kentucky, 1976.
4. Tholin, A. L., and Keifer, J. C., "Hydrology of Urban
Runoff." Transactions of the American Society of Civil
Engineers, Paper No. 3061.
5. Huber, W. C., et al., "Storm Water Management Model
User's Manual - Version II." USEPA Report No. EPA-670/2-
75-017, NTIS No. PB-203 291, Cincinnati, Ohio, 1975.
131
-------
IMPROVEMENTS IN EXTRAN
by
Larry A. Roesner, Atef M. Kassem, and Paul E. Wisner
WRE-TRANSPORT MODEL
The original WRE-TRANSPORT model was developed in 1974 by CDM/Water
Resources Engineers. The formulation of the model is contained in the
EXTRAN Documentation.
A conceptual presentation of the model is shown in Figure 1. The
model is based on a link-node description of the sewer system. Links
represent pipes and flow diversions. The nodes correspond to man-
holes or pipe junctions in the physical system.
The model is based on a complete solution of the gradually varied,
unsteady free surface flow equations (Saint-Venant equations). The
numerical integration is accomplished byamodified Euler method -
for details on the numerical solution refer to EXTRAN Documentation.
The model has the following capabilities:
- computation of surcharges and backwater effects;
- computation of flow diversions by orifices, weirs, pumps, etc.;
- it accepts inflow-hydrographs by direct input from cards or
hydrographs computed by the SWMM-RUNOFF routine.
The model has been extensively applied in Canada and the U.S. However,
many users have experienced stability problems under certain conditions,
particularly with weirs, orifices, large drops, or superpipes (usually
used for underground storage) with restricted release.
Subsequent to the release of the original WRE-TRANSPORT model, many
modifications were incorporated by various users, many times without
consulting the model developers. Consequently, a large number of
modified versions are currently used by different consultants and
organizations. Not surprisingly, comparative analysis carried out in
Associate, CDM/Water Resources Engineers
8001 Forbes Place, Suite 312, Springfield, VA 22151
IMPSWM Project Manager
University of Ottawa, Ottawa, Ontario KIN 9B4
Professor of Civil Engineering
University of Ottawa, Ottawa, Ontario KIN 984
132
-------
co
CO
LINK N-l
OI-E- °M™
NODE J
5 Q, = AS,U)
Q = flow
S = storage
Q»(N)
" Q|(N)
LINK N
FIGURE 1
CONCEPTUAL REPRESENTATION OF THE TRANSPORT MODEL
-------
the frame of IMPSWM project indicated differences between results
obtained from various versions when tested under the same conditions.
RECENT IMPROVEMENTS OF WRE-TRANSPORT MODEL
' The latest version of the WRE-TRANSPORT model represents a major
improvement of surcharge computation. It also has a better print-out.
Preliminary tests showed that the program is stable in most cases.
The following outlines the features of the new program:
(a) improved surcharge computation
(b) weir transfers are node to node, which improved stability;
(c) print-out summary for junctions, which shows:
' maximum computed depth and time of occurrence;
' maximum computed surcharge;
' minimum depth below ground elevation;
* duration of surcharge.
(d) print-out summary for conduits which shows:
' design flows;
" maximum computed flows and velocities and their time of
occurrence;
* ratio of maximum to design flows;
" maximum depth above inverts at conduit ends.
' However, the new version did not completely solve the instability
problems with orifices and weirs. Improvement of stability under
these conditions is currently undertaken jointly by WRE and IMPSWM.
PRESENT RESEARCH ON "IMPROVEMENT" OF WRE-TRANSPORT
' An agreement is reached between the CDM/Water Resources Engineers
and IMPSWM to work jointly on the improvements of the WRE-TRANSPORT
in order to eliminate the remaining problems with the model.
(a> Improvements of stability with orifices;
- two directions are followed for improving
the orifice simulation: (i) an approach
similar to the one presented in the current
release, and (ii) by cpnverting the orifice
into equivalent pipe, which is done inter-
nally in the model;
134
-------
- two options of the orifice are incorporated
in the model: (i) bottom orifice, and (ii)
side orifice;
- preliminary tests with the orifice simulated
as an equivalent pipe indicate good results.
(b) Improvements of stability with weirs:
Although the weirs are seldom unstable, model-
ling of weirs by means of equivalent conduits
is being investigated.
(c) Computation of surface area in order to eliminate any artificial
storage. The proposed scheme is explained in Figure 2. With
this scheme, one should avoid large drops (high pipe). However,
work continued to account for this condition;
Cd) Improvements of modelling on-line underground storage (example,
superpipe).
(.e) Some improvements of printout.
SURCHARGE COMPUTATION BY THE WRE-TRANSPORT MODEL
All the previous versions of the WRE-TRANSPORT simulate surcharge
based on an analogy with a surge tank (storage node).
Assumptions for the "artificial" storage node in the original model
(1974) are shown in Figure 3.
Subsequent modifications of "surcharge computation" applied the same
concept, assuming different shapes of the "artifical" storage node.
Examples are presented in Figure 4. It followed that various models
have predicted different surcharges when tested under the same
conditions.
The most recent version of the model (WRE-IMPSWM, 1979) no longer
assumes a surge tank analogy for surcharge computations, and a modi-
fied Hardy-Cross method is applied instead as described below.
During surcharge, the continuity equation for node j at time t is
ZQ (t) = 0
135
-------
///it,
A/or TO
Ji/ie Vton
Surfece
Hica
FIGURE 2
f's
///
136
-------
«>• ' r-p I f ?. «^f 5 f i? r .•<*'• ,\-I r:i:;"r* 11
I,.) L.:I. I'r^'i.-lsco 'li'^N::i'0,:
co
••4
CtJ .» folio".!
Tli.^ -i-tll'lclfll itoruv.o
(sec I'ijsure 7)t
h
o
h.
ad folloua
CRELEV - ZCROUH
0.75 h.
(I)
'A. • surface area of no4e ftt beginning
0 of surcharge - Computed from surface
width at 0.96 of the diameter of the
uppermost pipe connecting with sur-
charged node
A • Bin (5.0/at, 0.5 A /4t)
*1 &0
h. • water depth above crown elevation
A. * computed surface are* of surcharged node
2 corresponding to h.
wherei Y. • new nodal water depth at end of
Y. • nodal water depth at start of * t
£Q • junmatlon of flow* into junction
FIGURE 3
-------
FIGURE 4 / Fr*m
f/lf'Insert'aj
DEPTH
ABOVE
NODE
(FT.)
GRSLEV
B!$T 1 = 0,75 *
i^ DI$r 3 =0.25
(J)
D£UOM < ZOO
SURFACE ARZA OF SURCHARGED NODE
. fr. )
REDUCTION OF ASU) UNDER SUP.CHARGE
IN SWMM 77
138
-------
where lQ(t) is all inflows to and outflows from the node from inface
runoff, conduits, diversion structures, pump, and outfalls.
Since the flow and continuity are not solved simultaneously in
Transport, the flows computed in the links connected to node j will not
satisfy equation V-lla. However, computing 3Q for each link connected
3H3
to node j, a head adjustment can be computed such that the continuity equation
is satisfied. Rewriting equation V-lla in terms of the adjusted head gives:
z(Q(t) +|pi A Hj(t)) - 0 (V-llb)
which can be solved for AH. as
J
AHj(t) - -EQ(t)/E jjji (V-llc)
J
This adjustment is made by half-steps during surcharge so that the half-
step correction is given as:
Hj (t + ^|) = Hj(t) + kAHjU + ^f) (V-lld)
«
where AH.(t + Q] is given by equation V-llc, while the full-step head
is computed as:
H^(t + At) = H.(t + 4) * fe AHiCt) (
j j t j
where AH.(t) is described by equation V-llc. The value of the constant
k theoretically should be 1.0. However, it has been found that
equations V-lld and e tend to overcorrect the head; therefore, a value of
0.5 is used for k which gives much better results.
139
-------
For the various types of links connected to a node, sQ/sH is
computed as follows:
Conduits
where
K(t)
At
A(t)
L
n
R
v(t)
(V-llf)
time interval
flow cross sectional area in the conduit
conduit length
Manning n
hydraulic radius for the full conduit
velocity in the conduit
System Inflows
(V-llg)
Orifice, Weir, Pump, or Outfall Diversions
3M
il = (Q(t - At ) - Q(t - At))/(H.(t - At } - H-(t - At)) (V-llh)
. —n J ~~n J
140
-------
Because the head adjustments computed in equations V-lld and e
are approximations, the computed head has a tendency to "bounce1' up and
down when the conduit first surcharges. This bounding can cause the solu-
tion to go unstable in some cases, therefore, a transition function is
used to smooth the changeover from head computation by equations V-7 and 8
to equations V-lld and e. The transition function used is:
where
DENOM is given by
DEflOH . + (As.(t) .
and
AS = the nodal surface area at 0.96 full depth
DJ = pipe diamater
y. = water depth.
xJ
The exponential function causes equation (V-lli) to converge within 2 percent
of equation V-llc by the time the water depth is 1.25 times the full flow
depth.
141
-------
CSO IMPACT DETERMINATION
BY LONG TERM SIMULATION
HOWARD M. SHAPIRO, DIRECTOR OF ENGINEERING
JOHN B. BLENK, PROJECT MANAGER
MARK P. ALLEN, PROJECT ENGINEER
LOZIER ARCHITECTS ENGINEERS
ROCHESTER, NEW YORK
PRESENTED AT SWMM USERS GROUP MEETING
JAN. 1980
GAINESVILLE, FLORIDA
142
-------
STUDY SCOPE
GENERAL
The County of Chemung and City of Elmira, New York have to
upgrade the treatment of their wastewaters in order to further
preserve the Chemung River for its intended best usages.
For the normal wastewater flows (excluding overflows) the most
cost effective approach is the construction of a single upgraded
facility at the site of the existing City of Elmira Sewage
Treatment Plant.
The overflows, of which the city sewer system has two major
and three minor, will under present plans continue to operate,
pending further study as to their impact. Originally the
effect of the overflows was to be studied and analyzed as part
of the Southern Tier Central Regional Planning Board's 208
study. Due to financial constraints this aspect of the 208
study was cancelled.
Proper planning and design coordination, however, dictate that
the synergistic and complimentary aspects of solutions for wet
weather flows and solutions for dry weather flows be examined
before finalizing the design of the upgraded regional wastewater
treatment plant. Areas to be studied are:
143
-------
» What effect will increases in plant treatment efficiency
have on a needed reduction in overflows.
• What effect will increases in system storage of overflows,
with subsequent treatment at the dry weather facility have
on project costs and plant operation (dual use of
facilities).
• How often do combined discharges of dry weather and treated
or untreated overflows contravene stream standards, and
for what durations .
To account for the above, the present 201 facility planning
study incorporated elements of the 208 planning study and sought
to undertake the necessary analyses. These analyses would be of
a planning level, and would focus on those combined sewer overflow
abatement alternatives that could affect the design or operation
of the wastewater treatment plant. For example, what are the
consequences of split flow treatment, whereby, all flow beyond a
threshold level receives high rate settling in selected settling
tanks, with the flow below the threshold receiving normal
treatment. If such split flow could achieve significant wet
weather contravention reductions, then it may be cost effective
to include construction of the required facilities at what could
be a relatively minor cost.
144
-------
Level of Analysis
Urban water management analysis has been refined
by the U.S.E.P.A. R & D program as progressing in basically
four (4) levels--
/Level I - Desktop Analyses;
Planning ^ Level II - Simplified Continuous Models;
Level III - Refined Continuous Models; and
Design J
S Le
(^J_evel IV - Sophisticated single event and
continuous models
The level at which the subject study is being performed
is with.in Level III. This is at the upper end of the planning
analysis spectrum. A study of the City of Elmira's overflow
situation at this level was desired because, one, the results
could impact on the design of the treatment plant, and two,
the results of this study will be used to determine if a CSO
problem exists to what severity, and to formulate a continued
program.
To accomplish the above, knowledge of the overflows and
receiving waters during transient storm events is required.
The scope of the problem then has determined the need of account-
ing for the quantity and pollutional content of the overflows for
dry and wet weather conditions. Complete dependence on
empirical measurements is not desirable; the time requirements
and number of monitoring stations would be prohibitive. What
was needed was a basis of analysis which augmented available
data by incorporating empirical measurements, and generaliza-
tions derived from other systems and from theory, to calculate
historical overflow quantity and quality.
145
-------
NEEDS OF ANALYSIS
REQUIREMENTS OF ANALYSIS
The magnitude and comprehensiveness of the study required the
use of computerized mathematical modeling and associated
software to make it possible to process the necessary informa-
tion in a reasonable time period and with useful accuracy. The
basic requirements to be satisfied by the methodology were:
• To account for the time variations in precipitation,
runoff, dry periods, and receiving water conditions by a
continuous multiple event analysis.
• To have the model initialize each event.
• To provide statistical analysis of the output data.
0 To simulate the flows and pollutant characteristics
in sufficiently short time intervals to adequately
represent varying flow rates and receiving water reactions.
NEED FOR MULTIPLE EVENT ANALYSIS
Urban stormwater runoff and stream flows are determined by a
randomly distributed hydrological process wherein rainfall,
runoff, and overflow, and impacts on receiving waters each have
differing frequencies of occurrence. Meaningful and realistic
assessments concerning urban runoff pollution, intermittant
loadings of sewer rystem overflows and receiving water impacts
are not readily obtained from simulation of singular events.
The numerous, highly interdependent influences and especially
the random nature of storm events and stochastic nature of stream-
flows necessitate the investigation of many events, to gain the
proper perspective of the entire process.
146
-------
5.
A basic consideration is that the impact of overflows on river
water quality is highly event-specific. Analyses based on total
annual removal of overflow, or on a percent removal from a design
event, have little meaning. Overflow characteristics are clearly
dependent on storm intensity for surface wash-off and sewer
scour, quantity of dust and dirt on surfaces and deposition in
sewers, accumulated pollutant wash-off during the process of a
storm, and the filling of surface depressions during the progress
of a storm. The river's assimilative capacity is dependent upon
flow, temperature, photosynthetic activity, etc.
It is interesting to note that for a given storm event the
respective frequencies of rainfall, runoff, overflow quantity,
overflow quality, and, river assimilative capacity may all differ,
making it difficult to fabricate a single, or even a set of
representative "design" storms.
SINGLE EVENT SIMULATIONS
During a prior study the feasibility and practicality of
benchmarking single event storms with long-term analyses for
purposes of reducing the work effort on future analyses was
examined. The results of the analysis showed that such a
technique could be erroneous and misleading.
147
-------
6.
The benchmarking study was performed using long-term continuous
simulation output for the parameters of flow, BOO, and TSS. The
total quantity discharged of each parameter from storages of
several select sizes for several select real time single-events
were compared. The primary thesis employed in the benchmarking
attempts was that the frequency associated with each event on
the existing system, would not shift or would shift only
slightly, while the actual quantities of flow, 300, and TSS
would change for each size of storage improvement. The test was
performed by varying the size and outlet flow of the storage,
with each size being considered as a different level of
improvement. For each level of improvement a frequency curve
was developed (Figures A, B, C & D). The selected individual storm
results from each improved system were then benchmarked against the
frequency curve, developed from continuous simulation, for the exist-
ing system. The frequency curve generated through the benchmarking
procedure was then compared with the frequency curve based on
the long-term analysis.
As it developed, for the method of analysis employed, the
thes'is proved false. The parameter most closely approximating
the primary thesis was that of total overflow quantity. Even
here though, the deviations are significant, to the point that
confidence in the procedure is lacking.
148
-------
20
u.
6
'o
3 15-J
IU
2
10
U.
OC
Ul
h-
o
I
o
LEGEND
© Storm 1
m Storm 2
® Storm 3
• Storm 4
4> Storm 5
LONG TERM SIMULATION RESULTS
————— Existing System
— Upgraded System 1
Upgraded System 2
" Upgraded System 3
SINGLE EVENT STORM INTERFACING RESULTS
Upgraded System 1
Upgraded System 2
A - FREQUENCY DEVIATION
B - MAGNITUDE DEVIATION
2 3 4 5 6 7 8 9 10 II IZ
FREQUENCY
-------
en
O
(0
jQ
U.
cc
til
a
o
03
LEGEND
a
®
B
Storm 1
Storm 2
Storm 3
Storm 4
Storm 5
LONG TERM SIMULATION RESULTS
Existing System
— Upgraded System 1
Upgraded System 2
SINGLE EVENT STORM INTERFACING RESULTS
Upgraded System 1
Upgraded System 2
A - FREQUENCY DEVIATION
B - MAGNITUDE DEVIATION
5 10
FREQUENCY
-------
200
OT
CO
150
u.
DC
LU
O
I-
(0
V)
100
SO
LEGEND
Storm 1
Storm 2
Storm 3
Storm 4
Storm 5
LONG TERM SIMULATION RESULTS
Existing System
Upgraded System 1
Upgraded System 2
SINGLE EVENT STORM INTERFACING RESULTS
Upgraded System 1
Upgraded System 2
A - FREQUENCY DEVIATION
B - MAGNITUDE DEVIATION
1234
FREQUENCY
-------
tVJ
FREQUENCY OF OCCURRENCE
(times/year)
(As Determined From Long -Term Simulation)
STORM EVENT No. 3
Existing 1/5 Yr. Design 1/2 Yr. Design
Overflow Q (cf) 2.4 3.1 3.7
BOD (Ibs) 1.0 1.6 2.5
TSS (Ibs) 1.0 1.1 2.4
STORM EVENT No. 4
Existing 1/5 Yr. Design 1/2 Yr. Design
Overflow 5.0 7.0 No
BOD 7.7 7.0 Overflow
TSS 4.0 2.5
-------
n.
A further analysis was performed in a similar manner, using a
different set of single-event storms chosen from the same
modeled years of record; selected from events having
overflow/frequency relationships approximating the first set.
The interfaced frequency curve generated from the second set of
storms not only failed to agree with the long-term frequency
curve, but also failed to agree with the interfaced frequency
curve generated from the first set of storms.
The failure of the tested procedure to demonstrate adequate
generation of frequency relationships for the improved systems
is understandable. The model employed for the study considered,
among other items, input parameters such as storm intensity for
surface wash-off and sewer scour, antecedent dry period for
pollutant wash-off during the progress of a storm, and the filling
of surface depressions during the progress of a storm. The difficulty
in benchmarking individual storms further demonstrates the need
for an approach other than by "design storm".
NEED FOR CONTINUOUS SIMULATION
The build up of surface contaminations, the build up of settled
matter in the sewer network, the emptying of storage, and the
drying of surface areas are dependent upon the period between
storms. Each of these parameters affects the runoff, or the
pollution potential, or both. In turn each of these parameters
is affected by the intensity and duration of rainfall. For
153
-------
example, there might have been only partial washoff from a
previous storm. For reasonable assessments of overflow
quantity, duration, and quality, simulated for each
storm event, it is necessary that these parameters be
initialized. To initialize these parameters for all components
for each event is quite time consuming, and introduces another
source of possible error in the analysis. It is desirable to
have the means to trace the conditions through a continous
simulation.
NEED FOR DYNAMIC SIMULATION
The overflow process is closely related to the hydraulic
capacity and the actual flows within the sewer network which it
relieves. Similarly the receiving waters behavior is closely
related to its hydraulic capacities and actual flows. In
simulating overflow occurrences, it is then important to
represent the unsteady flow rates in the network and receiving
waters with reasonable accuracy. This requirement implies
sufficiently short time intervals of simulations to correspond
with intermittent conditions in the system.
In previous studies on the Rochester, New York system, a
computerized simple mass balance computation on long-term
rainfall data was accomplished to test the difference using
daily and hourly rainfall input into a model. The analysis
indicated that an hourly data input would produce a sixty (60)
percent increase in overflow quantities over that produced by
daily input data. Initial calibrations of the storm model on
154
-------
this study indicated that by using hourly increments reasonable
results in matching peak and total flows can be obtained. Given
the above and the nature of stream hydraulics, including initial
mixing diffusion, backmixing, and the relative speed of the
quality reactions, one (1) hour time intervals were selected.
NEED FOR STATISTICAL APPROACH
Rainfall, runoff, overflow and streamflow are all probabilistic
or stochastic parameters, and should be analyzed as such. With
the amount of data to be analyzed, the only practical approach
is a statistical one, incorporating records of many events into
a picture of the network as it operates under varying conditions.
A statistical approach offers two definite additional advantages.
One, the model is only an approximation of the actual network
and deviations from reality exist in all parameters and all
internal functions. By operating the model over many simulated
storm events under varying conditions and then statistically
analyzing the data, the effect of errors associated with an
individual event is minimized. Two, the receiving water
standards are open to interpretation from a frequency of contra-
vention standpoint. A statistical analysis fits nicely with
this interpretation.
155
-------
Model Selection
As a summary of the previous discussion, a model
package to be selected had to meet the following criteria:
• Be suitable for planning level purposes;
• Dynamically and continuously analyze the overflow
process;
• Dynamically and continuously analyze the effects
of the overflows on the receiving waters; and
• Statistically analyze the output.
The model package seen to best meet the above
criteria was the: U.S. Army Corps of Engineers Hydrologic
Engineering Center's (HEC) Storage, Treatment and Overflow
Reduction Model (STORM); the HEC Water Quality River Reservoir
Systems (WQRRS), and the HEC statistical post processor program.
156
-------
THE STORM MODELING CONCEPT
Storm is a continuous simulation model used for prediction
of the quantity and quality of storm water and dry weather
flow. The model provides analysis that satisfies two primary
study objectives. These are: (1) formulation of wet weather
hydrographs and pollutographs for use in receiving water impact
assessment and (2) provide statistical information to aid in the
selection of storage capacities and treatment rates to achieve
a given level of control of storm water runoff. The approach
used in the model to achieve these objectives is the recognition
of the properties of storm duration and intensity, the storage
capacity of the system and the affects of storm event spacing.
The overall model operation involves the interaction of 8 main
processes. These are: precipitation, runoff, pollutant
accumulation, pollutant washoff, dry weather flow, storage,
treatment and overflow. Figure 1 shows a schematic represent-
ation of the 8 storm water elements modelled. The model computes
runoff from rainfall or rainfall plus snow melt and associated
pollutant washoff for a lumped basin. In this process, rainfall
washes dust and dirt and associated pollutants off the watershed,
scour of debris in the conveyance system is not directly modelled
as such but can be accounted for by an appropriate calibration
technique. Runoff in excess of a specified treatment rate is
diverted into storage for subsequent treatment. Runoff in excess
of both the treatment rate and storage capacity is considered
overflow and is diverted directly into the receiving waters.
157
-------
en
00
RAINFALL / SNOWMELT
DRY WEATHER
FLOW
POLLUTANT
ACCUMULATION
POLLUTANT
WASHOFF AND
SOIL EROSION
TREATMENT,
MAJOR PROCESSES MODELLED BY STORM
Figure 1
-------
The quantity, quality and number of overflow events are thus
functions of hydrologic characteristics, land use, treatment
rate and storage capacity. In keeping with this there are
three major steps involved in estimating storm water runoff
and quantity and quality, these are: 1) the computation of
runoff quantity, 2) the computation of runoff quality, and
3) the computation of treatment, storage and overflow.
COMPUTATION OF RUNOFF QUANTITY
Runoff quantity can be computed by one of three methods, the
coefficient method, the U.S. Soil Conservation Service Curve
Number Technique or a combination of the two. For a predominately
impervious lumped urban catchment the coefficient method is
selected based on the relative ease of calibration and the
relative significance of the previous area. Storm employs the
Soil Conservation Service triangular unit hydrograph as a means
of routing basin excess to the point of treatment/storage/
overflow for both the coefficient method and the SCS curve number
technique.
The coefficient method defines runoff as the product of a runoff
coefficient and hourly rainfall excess. The runoff coefficient
is the weighted average of empirical runoff coefficients for
pervious and impervious areas, as such it is a composite runoff
coefficient which accounts for infiltration losses. The rainfall
excess is defined as the difference between hourly rainfall and
losses to depression storage. The composite runoff coefficient is
used for every rainfall in the rainfall/snow melt record regardless
159
-------
of rainfall characteristics or antecedent moisture conditions.
Antecendent depression storage is defined as the available
depression storage at the end of the pervious rainstorm plus a
linear recovery to account for evaporation during the period of
no 'precipitation. The evaporation rate can be verified monthly.
The coefficient method is a simplification of the hydrology,
but it can be expected to perform relatively well on an urban
catchment of relatively high percent imperviousness for which
losses due to infiltration are relatively small. The runoff
coefficients, available depression storage and depression storage
recovery factor are derived in the calibration procedure using
observed data.
COMPUTATION OF RUNOFF QUALITY
Computations for the stormwater runoff quality are based on
formulations first used in the EPA Stormwater Management Model.
Empirical equations considering land use, street sweeping
practices, and days between rainstorms define the amount of each
pollutant on the ground at the beginning of the rainstorm. An
exponential washoff equation relates the mass of pollutants washed
off during each hour to the current mass of pollutants on the water-
shed, the runoff rate and an exponent governing the rate of
pol1utant washoff.
160
-------
Two methods are available In STORM for the determination of
pollutant accumulation. These are the dust and dirt method
and the daily pollutant accumulation method, the latter is not
intended for use on an urban catchment. The dust and dust
method assumes that all pdllutants are associated with the dust
and dirt accumulation in the streets. This assumption is
acceptable for storm sewers and open channel conveyance system
but is not a proper characterization of a combined sanitary and
storm water drainage system. In a combined sewer system there
is typically a build-up in the sewers of grit and organic debris
between rainstorms. Thus, in addition to the surface accumulation
of dust and dirt there is the addition pollutant load flushed
from the conveyance system, i.e. the conveyance system scour
and washout.
Calibration of the dust and dirt method for separate stormwater
conveyance systems involves adjustment of the dust and dirt
accumulation rates, the pollutant fractions of dust and dirt
and the washoff coefficient so that the predicted pollutant
concentrations most nearly match those from measured data. When
calibrating the dust and dirt method for combined storm and
sanitary conveyance systems the system washout must be lumped
in with the surface dust and dirt. A combined sewer overflow
sampling program is required when calibrating STORM for a combined
Storm/Sanitary conveyance ^system. The qualtty constituents
predicted by the surface runoff portion of STORM are suspended
solids, settleable solids, biochemical oxygen demand, total
nitrogen, total orthophosphate and total coliform. Each constituent
161
-------
is determined for an hourly computation interval. The computation
interval affects the lower limit of size of catchment that may
be modelled. As normally used,STORM should not be expected to
produce properly shaped hydrographs or pol1utographs for
catchments having times of concentration less than one hour.
The quantity and quality of dry weather flows are input to
STORM as average values for domestic, commercial, industrial and
infiltration loads. They are modified on a daily and hourly
basis by a user specified coefficient.
The dry weather flow and quality hourly hydrograph and pollutograph
is added to those of the runoff and total hydrographs and
pol1utographs are generated.
STORAGE. TREATMENT AND OVERFLOW
Computation of treatment, storage, and overflow proceeds on
an hourly basis throughout the rainfall snowmelt record. Periods
of no rain are skipped, however, the number of dry hours is used
for various purposes including build-up of surface dust and dirt.
Each hour in which runoff occurs the treatment facilities are
utilized to treat as much runoff as possible. When runoff rates
exceed the treatment rate, storage is utilized to contain the
runoff. When runoff is less than the treatment rate, the excess
treatment rate is utilized to diminish the storage level. If
the storage level is exceeded, all excess runoff is considered
overflow and does not pass through the storage facility. This
162
-------
RUNOFF [c(P-f)]
o>
.20 -i
.18 •
.16 •
tf.
| .«-
-x
2 .12-
z
ui .10-
H
OL
^ •««•
b.
O
3 O6 •
a:
6
%
^
W
!
i
I
/A;
12
TREATMENT RATE
V////A RUNOFF AVAILABLE FOR
"f'fA STORAGE OR OVERFLOW
P
^y
^ -^w
n n
18 24 3O
TIME (HOURS )
TIME HISTORY OF RUNOFF USING THE
COEFFICIENT METHOD
Fi(;jre Z
-------
cr»
UJ
x
o
Id
13
IE
O
.20
IB
. 16 •
. 14
. 12
.10 •
.08
.06
.04
.02
0
DURATION OF
EVENT
Ln
12 18
TIME (HOURS)
STORAGE LEVEL
STORAGE
CAPACITY
OVERFLOW
26
32
TIME HISTORY OF STORAGE USING THE COEFFICIENT METHOD
Figure 3
-------
overflow from the system becomes the Input hydrograph and pollute-
graph for a receiving water quality model. Plug flow is assumed
for routing of pollutants through storage, water quality is
not modified during the storage routing interval.
Figures 2 and 3 show the time histories of run-off, storage
and overflow. From the time histories it can be seen that for
a given planning horizon, overflows to a receiving water could be
minimized by examining a matrix of land use alternatives and
treatment rate/storage alternatives.
Logically, the planning horizon should consider the impact of
overflows from the existing urban catchment on the receiving
water. Based on the overflow impact of the existing system a
matrix of alternatives land use and treatment/storage systems
can be designed to determine the best cost benefit/cost effective
solution to the CSO problems.
LOZIER ENHANCEMENTS TO STORM
Several program enhancements have been developed into the
original HEC version of STORM by Lozier. These program
changes were developed to enable the model to more realistically
represent the operational aspects of the storage/treatment
system. It was recognized by Lozier that at the end of a
rainfall/runoff event (the end of excess rainfall) wastewater
remaining in storage was lost to the analysis. The model was
altered to permit the event to continue until all storage was
bled back through the treatment works resulting in the time exten-
165
-------
sion of the pollutograph output. Additionally, the option was
provided to permit the user to specify the treatment efficiency
for reduction of the six modelled quality constituents. A
different treatment efficiency can be.assigned to each quality
constituent. This permits the user to examine the impact on the
receiving water of treated effluent and stormwater overflow as
the treatment works is pushed hydraulically during the period
of wet weather treatment. This option can also be used to
add a dimension to the matrix analysis of land use/storage/
treatment by permitting the comparison of different treatment
processes at various levels of hydraulic loading.
THE WORRS MODELLING CONCEPT
The Water Quality River-Reservoir modelling concept is based
on the fundamental principal of the Law of Conservation of
Mass and the Kinetic Principle, as they appl.y to studies of
the Continually Stirred Tank Reactors (CSTR). Laboratory
studies are largely responsible for the current level of under-
standing of each individual process that occurs in a receiving
water environment. The problem of receiving water modelling
is to assemble the body of theoretical and emperical knowledge
developed from CSTR experiments into a simulator that properly
predicts the phenomena observed in the field, i.e., the prototype.
Assembling such a simulator is not a simple task because CSTR
studies are normally conducted at relatively small scale usually
under well controlled conditions and often under different
circumstances than those existing in a prototype. The receiving
water is not a single simple reactor, to overcome the disparity
between model and prototype the receiving water is conceptualized
as a system of discrete hydraulic or volume elements.
166
-------
c,
c.
cn
\ X
r
A.
A CONTINUOUSLY STIRRED TANK REACTOR, CSTR
OUT u IN
B.
VOLUME
IDS
BOD
DO
TEMP
ALGAE
ZOO
FISH
AN IDEALIZED HYDRAULIC ELEMENT
Figure 4
167
-------
Figure 4 represents an idealized hydraulic element where the
state of the ecosystem can be defined and where the CSTR analogy
holds. The element has a volume, a surface area and one or
several cross-sectional areas between adjacent elements. Flows
are advected from adjacent elements and local inflows and outflows
are accounted for. The conceptualization is carried one step
further by inter-connecting the hydraulic elements in such a
way that they can be viewed as a series of CSTR's whose contents
can be transferred from one to another. This analogy improves
as the number of hydraulic elements is increased. Figure 5
illustrates the hydraulic element placed in series to represent
stream system geometry and mass transport mechanism.
The mechanics of receiving waters are usually highly dynamic.
The CSTR on the otherhand usually represents steady state
conditions. Consequently, the CSTR analogy can only be considered
valid when the prototype approximates steady state conditions.
To overcome this disparity the CSTR or hydraulic element must be
observed at a sufficiently frequent interval that will permit
the characterization of the dynamic process as a set of successive
steady state processes. As the observation interval or time
step is shortened the CSTR/hydraulic element analogy improves.
Many reactions are occurring simultaneously in the receiving
stream. There is no theoretical difficulty in dealing with this.
if in the mathematical simulation the reactions are considered
to properly take place in parallel (independent reactions) or
in series (inter-dependent reactions). Each process contributing
168
-------
Control Volume Element
Control Volume Nodes
hydrologic balance
(QxCx)i
Qi
3X/j+| material balance
DISCRETIZED STREAM SYSTEM
Rgure 5
169
-------
in part to the total water quality change is mathematically
represented by a partial differential according to kinetic
principles. Partial differentials can be added together to form
the total differential, allowing the computation of total response
in the water quality. Therefore, for each hydraulic element
a mass balance equation in differential form is solved for
each water quality constituent. The whole array of differential
equations representing the coupled nature of the water quality
constituent is entered for each time step and solved to determine
the dynamic response of the receiving water to wasteloads.
WQRRS Model Structure
The WQRRS model consists of three separate but
integrable modules; the reservoir module, the stream hydrau-
lics module and the stream quality module. WQSTAT, a statis-
tical and graphical package, is available for reduction of
the stream quality module output. The reservoir and system
hydraulic module are stand alone programs and may be executed
analyzed and interperted independently. The stream quality
module has no hydraulic computation capability and requires
a hydraulic data file generated by the stream hydraulics
module. The three computer programs can be integrated for
a complete river basin water quality analysis through the
automatic storage of result for input to downstream simula-
tions.
170
-------
• NODE POINT
REACH tt 1
REACH * 4
REACH « 5
REPRESENTATION MODEL
TYPICAL STREAM SYSTEM
Figure 6
171
-------
STREAM HYDR'AULICS MODULE
The stream system is represented conceptually as
a linear network of hydraulic elements. Several elements
make up a stream reach, several reaches comprise a stream
system. Figure 6 illustrates a typical stream system
representation, note that branch and looped flow can be
modeled. Each element is characterized by length, width,
cross section, and certain other parameters such as location
of tributaries and local inflows and outflows. Six methods
of hydraulic computation are incorporated into the hydraulic
module. They include:
1. Backwater hydraulic solution (steady flow)
2. Solution of the full St. Venant equations
3. Solution of the kinematic wave equations
4. Direct input of a stage-flow relationship
(steady flow)
5. Muskingum hydrologic routing
6. Modified puls hydrologic routing
The first three methods represent hydraulic behavior of the
prototype stream system under gradually varied flow and assumes
the system can be represented by the St. Venant equation of
motion. Additional assumption for the first three methods are:
172
-------
1.) the system is one dimensional in a mathmatical
sense, flow and velocity are uniform both laterally
and vertically
2.) linear interpolation between cross-sections pro-
vides adequate definition of the element systjem.
3.) The rate of energy loss for gradually varied
steady and unsteady flow is the same as that for
uniform flow having the same velocity.
4.) The slope of the channel bottom is small
(ie. , cos 9 3 1 )
The final three hydraulic computation methods are all indepen-
dent of the fundamental hydraulic relation-ships as represented
by the St. Venant equation. Each method requires specific
boundary conditions of flow, stage, or flow versus stage.
Input to the hydraulics module consist of a table
of elevation versus channel characteristics for each element.
The table of characteristics consist of cross-section area,
width, hydraulic radius, and friction factor. This input
can be generated by the use of the HEC program, Geometric
Element From Cross-Section Data (6EDA). GEDA processes
available cross section data to produce the elevation versus
channel characteristics table at desired locations. The
hydraulic module processes inflow and withdrawal recoras
input at user specified locations and time intervals, ie.,
hourly, daily, monthly. The module output consists of,
flow depth, velocity, cross-section area.surface area and
width for each element at a user specified computational
interval. This output is saved on an interface storage
devise, such as magnetic tape or disc, for subsequent use
173
-------
in the stream quality model. Thus the hydraulic module
output characterize the spatial and temporal relationships
of the stream system. This characterization includes base
stream flow and tributary flows, inflows such as waste water
discharge rate and rainstorm overflow rates and stream flow
withdrawals in addition to the hydraulic aspects relevant to
element and reach travel time and stream reaeration.
Quality Module
The receiving water quality module is based on the
assumption that the dynamics of each chemical and biological
water quality component can be expressed by the law of conservation
of mass and the kinetic principle. An important assumption is
that all chemical and biological rate process occur in an aerobic
envi ronment.
The ecological processes within the quality module
are centered around benthic alage where the tropic relationship
between benthic alage and aquatic insects form the base of the
food chain. Figure J_,s hows the interrelationship of the food
chain constituent and water quality constituents. Figure 7
also illustrates the inter-dependance or coupled nature of the
water quality co-nsti tuents and ecological constituents.
174
-------
MAN-INDUCED
WASTE LOADS
BENTHIC
ANIMAL
'1$
FOOD
FISH
DEFINITION OF AN AQUATIC ECOSYSTEM
Figure 7
175
-------
The quality module incorporates a water quality
constituent disconnect capability which allows the user to
restrict the simulation of constituents to those of interest
or importance in a specific study. The user may also hold a
constituent at a constant value during the simulation. This
permits the user to adapt the model to a particular level of
study, which could range from a simple first cut simulation
of BOD, dissolved oxygen and temperature or a very rigorous
simulation of the aquatic food chain. The importance of this
option should be fully considered by the user, as once the
modeling data base has been assembled for the stream system,
it can be updated as information becomes available and the level
of simulation can be upgraded accordingly.
A Gaussian reduction technique is used to solve the
set of simultaneous equations representing those constituents
which are passively transported with the movement of water.
For those constituents which are assumed affixed to the bottom
or are self mobile, such as fish, the equations are solved by
simply multiplying the time derivatives by the computation time
step increment. The differential equations representing the
constituents are processed sequentially from the least dynamic
to the most dynamic. The array of differential equations is struc-
tured such that the coupled nature of the constituent is recognized
in the sequential regression.
176
-------
Water quality constituent sources and sinks resulting fro.m coupling
are assumed constant over the computational time step. These
sources and sinks may include: sett!ing, first order decay,
reaeration, biological uptake and release, growth, respiration,
and mortality including predation. With the exception of
conservative constituents the differential equations representing
water quality relationships incorporate one or more physical,
chemical or biological rate coefficients. These rate coefficients
are based upon theoretical and empirical understanding of the processes
governing the sources and sink terms, tlany of these rate coefficients
are highly variable and depend upon such factors as regional
climatic variation, time of day, synoptic weather patterns, stream
system geometry and type and general levels of pollution.
In the aquatic environment the rates at wnich chemical
and biological processes take place are normally functions of
temperature, therefore rate coefficients describing these processes
must be adjusted to the ambient temperature. Two approaches
are used in the model to make the temperature adjustment. The
temperature limit method assumes a reaction takes place as a function
of two exponental curves constrained by temperature tolerance
limits. One curve considers growth, respiration, mortality and
decay, the other considers growth only. The temperature correction
method assumes the rate at which a reaction takes place, increases
exponentially with increase in temperature. This method applies to
stream reaeration coefficients and coliform die-off rates,
remaining constituents can be specified for temperature correction
by either method.
177
-------
The ambient temperature is arrived at in the model
by considering external heat sources and sinks. In a stream
system heat is-transferred at the air water interface and the
stream bottom sediments.
Two methods may be selected by the user to calculate
the water surface heat flux; the heat budget method and the
equilibrium temperature method. Determination of the rate of
heat transfer per unit surface area for both methods requires
the following five components; net rate of short-wave solar
radiation, net rate of atmospheric long-wave radiator, rate
of long wave radiation from the water surface, rate of heat
loss by evaporation and rate of convective heat exchange between
the water surface and the overlying air mass. This information
is generated in the model from standard meteorological input
consisting of wet and dry bulb temperature,cloud cover, barometric
pressure and wind speed in addition to invariant data such as atmos-
pheric turbidity, evaporation coefficients and latitude and longitude.
Variant meterological data is imput at a time interval (hours,
days, months) consistent with available information and the level
of study required.
Heat exchange with bottom sediments may be significant
in shallow streams where the heat exchange with the bottom will
have a moderating effect on water temperature fluctuations.
The maximum daily temperature will be reduced by conducting of
heat away from the water to the cooler bottoms sediments. The
reverse will occur at night, thus the total temperature fluctuation
is moderated.
178
-------
WQRRS
ANALYTIC PROCESSES
QUALITY MODULE
• THE USER HAS THE OPTION OF SELECTING FROM THE
FOLLOWING SEVEN METHODS FOR COMPUTING THE DISSOLVED
OXYGEN REAERATION COEFFICIENT:
.969
CHURCHILL K2 = 5.031 ' V
H ''673
V5
O'CONNOR 8 DOBBINS K.2 =3.951 '
67
OWENS K2 =5.346 ' V
LANGBIEN AND DURUM K- =5.133
2 -"•"•"' V 1.333
THACKSTON | KRENKEL K2 =24.95
TSIVOGLOU 2, WALLACE K- =3.78
DIRECT INPUT OF
• SHEAR VELOCITY = (H . S , 9 )'J
TABLE 1
179
-------
The response of water quality to external wasteloads
is determined by routing the wasteload through the stream system.
In the model this takes place after initial conditions have
been satisfied for the water quality constituents selected for
simulation. During the routing procedure, for each computational
interval, the temperature is determined, biological and chemical
rate coefficients are adjusted, the stream dissolved oxygen -
reaeration capacity is determined and the water quality constituent
response is predicted. This result becomes the initial condition
for the next time step and so on as the water moves downstream.
As the water moves through the stream system, the dissolved
oxygen concentration can be determined based on the hydraulic
characteristics of the stream flow. Table 1 lists the six
computational methods available to the user for determining the
dissolved oxygen reaeration coefficient. In addition, the user
may directly input the reaeration coefficient.
Quality module output consists of a water quality
constituent profile of the stream system in tabular format
for a user specified reporting interval. This interval will
most likely be much greater than the computation interval
and as such represents average conditions for the reporting
interval. This is a practical limitation with a long-term Sim-
ula tion,based on computer printing cost and the users ability
to synthesize the mass of data generated for a given computation
interval. The user can specify that the computation interval output
be placed on a magnetic storage device for later statistical
and graphical analysis by the HEC program WQSTAT.
180
-------
KQSTAT
Statistical Post-Processor
WQSTAT is a statistical and graphical package
consisting of three major components: two graphical plotting
programs and a statistical program. Table 2 lists the water
quality parameters that can be analyzed by the package. Each
of these programs can be executed independently of the other
two and each accepts input from a magnetic storage device
containing, output (at the computation interval) from the
Stream Quality Module.
The first plotting program plots the results of the
stream simulations as a function of time for any number of these
parameters at user specified locations throughout the study
reach. The user has the option to superimpose observed values
on these plots.
Water quality standards can be specified and plotted
as well. In the case of temperature and pH, both maximum and
minimum standards values can be plotted.
The second plotting program plots longitudinal profiles
of water quality parameters at all nodes throughout the study
reach. For each of the eleven parameters, plots of the maximum
and/or minimum simulated values over the study period at each node
can be generated. The user can specify that observed data and
water quality standards be plotted as well. The water quality
standards can be changed as a function of river mile to reflect
different stream segment quality designations. As a separate option,
the user can also input values of the water quality parameters
that are indicative of a particular critical quality condition. The
181
-------
GENERAL CAPABILITIES OF THE
STATISTICAL AND GRAPHICAL
POST-PROCESSOR
HOURLY STATISTICS GENERATED FOR UP TO ELEVEN
WATER QUALITY PARAMETERS
• STREAM FLOW
* WATER TEMPERATURE
• DISSOLVED OXYGEN
• AMMONIA NITROGEN
• NITRATE NITROGEN
• .'HOSPHATE PHOSPHORUS
• ALKALINITY
• COLIFORMS
• TOTAL DISSOLVED SOLIDS
• PH
• 5-DAY BIOCHEMICAL OXYGEN DEMAND
TABLE 2
182
-------
mean error and standard deviation associated with these
critical values can also be plotted. The input values for
this option generally come from the results of the statistical
program.
The Statistical Program has the capability of summarizing
by node the maximum, minimum, mean, and standard deviation of
the simulated values over the study period, or over a selected
sub-period, for each of the eleven water quality parameters. It
can also compare simulated parameter values against selected
water quality standards and determine the number and percent
of points exceeding the selected standards. Lastly, it can
accept observed data for various locations throughout the study
reach and then determine the mean and standard deviation of
the difference between the simulated and pbserved values.
Determination of CSO Impact on the Chemung River
The determination of CSO impact on the Chemung
River is based on a long term simulation of meterological,
and urban hydrological patterns of the study area. Twenty
miles of the Chemung River was selected for study, the study
area begins at River mile 23.0 (Fitch Bridge) in West Elmira.
This location is approximately 5.0 miles upstream of CSO point
source discharges. The study area ends at mile 3.0, approximately
2 miles downstream of a USGS flow gage and NYSOEC water quality
sampling station.
183
-------
The study period selected, dates from June 1966 to
October 1975 for 10 years of records. The five month modeling
period from June to October is considered to represent the
critical period in terms of the river water qua!ity. Daily river
flow records are available from the USGS stream gage at Chemung
(mile 5.0) for the period, in addition to sparse river quality
data. Hourly rainfall data for the period was taken from the
Binghamton airport. A comparison study of daily rainfall for
Elmira and Binghamton showed a favorable super position, from this
it was assumed the rainfall intensity/duration was also comparable
ahd the hourly Binghamton data was taken as representative of the
Elmira meterological history.
The drainage basin bordered by the study area
includes the Chemung River mainstem and several major tributaries
These include Newtown Creek and Seeley Creek and Sing Sing Creek,
just upstream of the study area boundary. Daily flow for the
tributaries was developed from USGS recorded flow for Newtown
Creek by simpl-e proportion based on drainage area.
The meteorological/hydro!ogical data base consists of
10 years of daily stream flows and hourly rainfall for the months
of June through Oct'ober.
Application of "Storm"
"Storm" determines surface run-off rate and pollutant
concentration for rainfall events based on a lumped run-off
coefficient aggregated by land usage, and a pollutant accumulation
function. The run-off quantity calibration must be accomplished
before proceeding with run-off quality calibration. Calibration
of Storm for run-off volume is accomplished by adjusting certain
coefficient which regulate the volume and timing of run-off. In
184
-------
the coefficient method, the first factor to adjust is the run-off
coefficient for the previous areas of the watershed so that the
observed and computed magnitudes of run-off volume show fair agree-
ment. Since the coefficient method is a very crude model of the
urban run-off process, it may not be meaningful to compare
individual event volumes. Other variables that can be adjusted,
although to a lesser extent, are the run-off coefficient for the
impervious areas and the initial abstraction in the form of
depression storage and infiltration.
Calibration of run-off quality can be approached
in serveral different ways. One method consists of a sampling
program of dust and dirt accumulation and associated pollutant
concentrations. Another method involves the determination
of daily pollutant accumulation rates and adjustment of wash-
off exponents. These two methods apply principally to pollutants
associated with surface wash-off, but do not consider the
accumulation of material in combined storm and sanitary sewers
where the type of deposited material differs significantly from
surface pol1utants.
The calibration technique employed for this study
consists of water quality sampling at CSO point discharge during
rainfall run-off events. From sampled data the dust and dirt
accumulation rates and wash-off exponent are somewhat arbitrarily
set to force an hourly correlation of pollutograph shape and
magnitude for several sampled storms. By calibrating on storms
with intensity duration relationships comparable to a majority
of rainfalls in the data base it can be assumed that Storm will
produce reasonable hourly pol1utographs for use in receiving
water quality studies.
185
-------
A sensitivity analysis can be performed on this
calibration to examine variation in total pollutant concentration
for the study period. This is done by varying the calibration
parameter that produces a change in the magnitude of predicted
pollutant concentration, with the superposition shape remaining
unchanged. The Storm model is run with the long term data base
with variations in the superposition magnitude, and the resulting
long term pollutant concentrations statistics are examined to
determine if the statistics are sensitive to the variation. If
long term statistics are observed to be insensitive to small
(+10% to 20%) variations in pollutant concentration magnitude
it can be assumed that the initial calibration produces hourly
pollutant concentration which on the average are reasonable.
From here the hourly pol 1 utograp'n output for the
10 year study period was written to magnetic tape storage for
later input to the receiving water quality model.
Calibration of storm for the City of Elmira was
based on overflow sampling at a point source to which approximately
1/3 of (1216 acres) the total, urban areas is tributary.
Application of "WQRRS"
WQRRS consists of a hydraulics module and a stream
quality module. First the hydraulics module is calibrated
and run, with the results placed on magnetic tape storage
for input to the receiving water quality module. Calibration,
6~f both modules is based on available data and engineering
judgement. It is unlikely that any receiving water system has
or will have been studied in the field to the extent that a
186
-------
dynamic hydraulic and water quality model can be rigorously
calibrated. Generally speaking the hydraulics module is more
easily calibrated as the hydraulic problem is less complicated,
field data is more easily obtained, and knowledge of a few basic
parameters may be sufficient for calibration. The stream quality
module is many times more complex, inherently involving many more
parameters and interrelationships. For this reason, a direct
calibration of the stream quality module is not a practical
pursuit when dealing with a study level project. If a CSO
impact is demonstrated a CSO abatement study should be undertaken
at which time in stream sampling during overflow events should
be attempted.
Hydraulics Module
The Chemung River was broken into 26 elements varying
in length from 0.5 miles to 2.5 miles, the shorter elements being
placed in the area of the river believed to be most sensitive
to CSO impact. A mainstream base flow daily hydrograph and three
tributary base flow daily hydrographs characterize the hydrology
of the study area. Two hourly hydrograph/pol1utograph inputs
are received from STORM. The hydraulics of the study area
are developed in the model by the stage flow method, from input
characterizing the river channel cross section, slope and
friction coefficients. The physical data describing the low flow
channel was seen to be lacking in required precision. Therefore,
steps were taken to obtain a better hydraulic characterization
of the low flow channel. Prior low flow travel time studies were
reviewed with NYSDEC to determine if they could be used to
187
-------
benchmark low flow hydraulics. With NYSDEC concurrence on the
usefulness of the travel tinre studies the hydraulic module low
flow output was adjusted by formulating a correlation function
and using a simple driver computing routine to update the
hydraulic characteristics as they are read into the stream
qua!i ty module.
Stream Quality Module
For a first cut at determining the Elmira CSO impact
the stream quality module was set-up for a study level analysis
looking at six basic pollutants, these are: suspended solids,
settleable solids, BOD, nitrogen, phosphorous, and coliform
bacteria. The primary water quality indicators being examined
are BOD, dissolved oxygen, temperature, and coliform bacteria.
Background levels of these water quality consituents were
determined from intermittent sampling data collected at Fitch
Bridge (mile 23.0) in West Elmira and Chemung (mile 5,0)
The predictions of changes in water quality
constituents as the pollutants are routed downstream are
largely uncalibrated. This is the result of insufficient
data to calibrate the complex processes occurring in the stream.
Therefore, the model predictions are based partly on theoretical
knowledge and empirical relationship. The primary empirical
function is that of stream reaeration. This is handled in
the model by direct input of reaeration coefficients (in a
long term analysis this results in essentially steady state,
because little data is available) or by selecting one of six
reaeration equations. The strategy developed for dynamically
determining reaeration coefficients is one of judgment in
188
-------
the comparison of a test run on each of several rea-eration
equations, the equation producing the most reasonable coefficients
over the 26 elements of the stream is selected for the long
term simulati-on.
The results of the stream quality module is a massive
listing of water quality constituents concentrations, hour by
hour, for each element of the stream for the entire study period.
It is obvious that the output in this form is of little practical
value in the evaluation of the CSO impact. Therefore, the
WQSTAT statistical program package is employed to reduce the output
to more usable form. Statistical comparison can be made of
modeled results to established stream standards or to the degree
and duration of violation of the stream standards. Therein will
be found the significance of the CSO impact on the receiving water
quality.
189
-------
TOWARDS STANDARDS FOR COMPUTER-BASED
MUNICIPAL DRAINAGE STUDIES
by
William James1 and Mark A. Robinson2
ABSTRACT
The main purposes of a model study are a reasonable understanding of the
physical processes involved in the study problem, and a careful evaluation of
reasonable alternative designs. The central concern is credibility.
Increasingly, government and municipal engineers control studies in which
sophisticated computer models are being used by specialists. Because of the
rapid evolution of some of these models, their complex structures and the very
large variety of programs available, the situation is becoming increasingly
complex. Guidelines for control of the studies may be useful, especially where
the local municipal engineer has limited experience with this type of computing.
The following activities could be specified in detail in the initial study
terms of reference, even before bids are accepted: problem review, study
objectives, performance criteria, requisite accuracy, review of available programs,
available data and study resources, program selection criteria, model verification,
model calibration, model validation, minimum level of discretization, sensitivity
analysis, data preparation and output interpretation, documentation of the modified
program actually used, and preparation of machine readable input and output files
for archiving.
Each of the above activities is discussed in general terms but the examples
used are appropriate to stormwater management modelling studies particularly
i Professor, 2Research Engineer, Department of Civil Engineering and
Engineering Mechanics, McMaster University, Hamilton, Ontario L8S 4L7
(416) 527-6944
190
-------
some verification and sensitivity analysis for the hydrological portion of the
Stormwater Management Model.
The delivery of computer-based engineering design services could be improved
by enlightened study terms of reference and attentive control by the responsible
municipal engineers. The suggestions in the paper should reduce the chances of
using an inappropriate model, inappropriate use of an otherwise suitable model,
such as erroneous data preparation or output interpretation, and ignorance of the
model even when the results are acceptable. Credibility would thus be improved.
INTRODUCTION
This paper attempts to set out some special requirements of computer-based
studies in municipal drainage. These requirements are additional to those
relating to the specifics of the drainage study. The purpose of these additional
requirements is to establish the credibility of the computer modelling and to
ensure that confidence in the results is transmitted from the engineer doing the
computing, or the study, through to those engineers and politicians ultimately
responsible for the implementation of the drainage improvements. It is suggested
that these special computer-modelling requirements could be set out in the initial
Study Terms of Reference. These suggested additional points in the Study Terms
of Reference are summarized again at the end of this paper.
Programs such as SWMM are capable of describing drainage hydraulics and hydrology
to a high level of system detail. Their strength lies in the fact that they take
into account the interaction of a reasonable number of processes and are able to
report on the status of certain important variables within these processes through-
out the simulation. Of course, there would be little point in using SWMM if the
system under study was simple enough to be modelled with the use of cheaper
calculating machines than the usual main-frames. The main benefit in using such a
191
-------
complex model therefore appears to be that many, if not all, of the relevant
processes and their interactions can be sufficiently accurately investigated and
ultimately sufficiently well understood by the engineers responsible for the
design. Thus such models requiring main-frame support provide a thorough evalu-
ation of reasonable alternative designs for complex drainage systems. A
difficulty that arises is that the person responsible for writing the original
program is usually not available for interpreting the performance of the program.
Also, the backroom engineers responsible for the computer simulation, often on
a considerably modified version of the original program, are frequently remote
from the interpretation of the design and ultimate presentation of the results
to the public. Whereas the original author of the program, and engineer
responsible for submitting the computer runs, may be highly confident of the
reliability of his particular computer program, the municipal engineers, and the
people responsible for budgeting and supervising the ultimate project may not have
the same confidence in the computer methods. Indeed, they may be wise to be
sceptical until the results are proven to be sound.
In our work we have tried to establish controls to ensure that the best model
is used correctly. This should ensure that-credibility and confidence in the
results is created and transmitted sequentially from the backroom engineers to the
project engineer, perhaps in the consultant's office, and thereafter to the
municipal engineers responsible for supervising the study, and ultimately through
the engineering committees and other political bodies to the beneficieries, namely
the public. In other words, it is not sufficient that the computer model represents
all relevant physical processes of the study accurately enough, or that the results
are correct or sufficiently'accurate. It is highly important that the study be
carried out in such a way that there is little chance of using a wrong model, or
using wrong data, or wrongly interpreting the results, or simply of not under-
standing the model or the design.
192
-------
Of course, there are obvious difficulties. For example, which of the large
variety of programs available are best suited to the particular study, or in
which order certain models should be used, may be uncertain. Similarly, if the
program adopted is very complex, it may not be clear which processes may be
safely ignored, or that the data set used is the best of many possibilities.
So we argue herein that many computer-based studies can be enhanced if certain
precautions are taken at the outset of the study. We suggest that the Study Terms
of Reference include among other things a minimum number of standard tests for
verification, sensitivity analysis and output interpretation. Evidently such
tests will not materially increase the workload for the consultant, nor will it
increase the cost of the study.
COMPLEXITY
Several factors have increased the complexity of computer-based drainage
studies over the years:
(1) more and more models are becoming available;
(2) the models are including more processes;
(3) the variety of computing hardware is increasing;
(4) the cost of computing is decreasing;
(5) computer communication between design offices and remote mainTframes is
becoming easier;
(6) the software capabilities of the computers is becoming more complex; and
(7) there is a professional drive within most engineers to improve their under-
standing of computer modelling and computer methodology, despite the
increasingly longer learning times associated with this methodology, and
higher salaries and related costs.
193
-------
These developments increase the pressure on municipal engineers to use
sophisticated design technology which they may not have had time to otherwise
use themselves.
Computer Models
Most tried-and-tested programs, such as ILLUDAS, SWMM and HSP have been
considerably enhanced recently. For example, the latest versions of each may
operate in continuous mode, including water quality processes. Since these programs
have been under development for ten years or more, it becomes very important that
the precise stage of the evolution of the program be correctly identified when the
program is used. Program documentation seldom matches the capabilities of the
program at any point in time; consequently it becomes important that the actual
program capabilities are accurately and carefully identified.
A recent study1 lists several hundred computer programs currently available for
solving problems associated with water resources development. In fact, it seems
that every university and consulting engineering office is bent on establishing
their own computer programs for the solution of stormwater management problems.
Moreover, while most of the popular models are being enhanced to better describe
constituent processes, or to include more processes, other programs describe some of
these constituent processes in greater detail. Hence the problem that now arises
is the selection of the correct sequence of models to be used in a study. It may be
better to use a sequence of process models than to use one of the system models
taking into account similar processes, but inaccurately. Evidently few guidelines
exist for this model selection process. On the other hand, as a municipality gains
experience with one model, there is a strong tendency to prefer studies based on
the same model. McPherson describes these problems in a recent paper2.
194
-------
Hardware
Readers will be aware of recent advances in programmable calculators3 and the
rapid evolution of micro-computers with their bewildering range of configurations
and capabilities. Similar situations exist within the range of mini-computers
and main-frames and the machines that lie between these three classes. Increasingly,
it is becoming necessary to distribute the computational effort involved in any one
study between local micro- or mini-computers and remote main-frames. Many programs
are being modified so that their solution methods are better suited to the large,
cheap memories and small bit words available on the small systems4'5. Again there
are few guidelines for this distribution of computational effort.
Systems Ware
New systems software products include local disc operating systems and sophisti-
cated word processors. These products could considerably aid both output
interpretation and input data editing. These in turn have benefitted from enhanced
communication speeds. In addition, there is an increase in the number of national
and international computer networks available, so that the smaller design offices
may easily access the largest computers. These trends help smaller consulting
engineering offices to gain remote job entry into the large programs maintained by
the vendors. With the skills of a recent graduate or postgraduate engineer and
an inexpensive terminal, any office has immediate access to the largest computers
and the most sophisticated programs on the continent without the burden of maintain-
ing or updating the packages.
STUDY RESOURCES
Individual studies often differ greatly in their objectives and available
resources, and consultants usually review the complete study problem and resources
at the outset in order to establish the mutually agreed scope and basis for the
195
-------
model study. The problem review should not be done in a general way, capable of
ambiguous interpretation. It should be directed so as to list all the hydraulic
and hydro!ogical characteristics of the study problem so as to aid in the selection
of a computer program. It should attempt to list specific hydraulic and hydro-
locial processes that ought to be included in any model study undertaken. For
example, if hydraulic jumps or energy dissipation at manholes are important, these
should be identified. Few, if any, programs adequately describe these phenomena.
The study objectives and related objective functions should be listed. The
purpose of this is to demonstrate how the numbers generated by the computer model
relate to the general objectives of the study. Computer models often produce only
hydrographs and pollutographs resulting from a certain hydrologic time-series. On
the other hand, the study objectives often require a least cost alternative to a
specific flooding or drainage problem, such as the production of toxic or other
harmful pollutants as a result of rainfall or snowmelt. The review of the objective
functions is a clear and logical explanation of how the model study results relate
directly to the objectives of the design.
To many observers, the model produces only a sequence of numbers. It is
advisable to describe early in the study how these numbers may be compared between
several computer runs so as to establish the validity of the computer program. In
other words, the performance criteria for assessing the various design proposals
must be carefully explained. In some cases it may be sufficient to estimate the
total annual loadings whereas in another case the peak flow is more important. The
criteria for the performance of one design against another may therefore appear to
vary, unless it has been settled initially.
Accuracy is of overriding importance. If the required accuracy can be established
early in the study, it would make the selection of the models and the level of
discretization for the model study an easier task. In a sense the accuracy level
196
-------
predetermines the model to be used. There seems to be a mismatch between the
accuracy of the results demanded from the computer and those obtainable from the
field observation program. There is little point in producing simulation results
that are much more accurate than the field observations which are used to validate
the model. Consider for example the accuracy of rain gauge sampling of storms,
and the whole question of design storms. In any case, high accuracy may not be
necessary in view of the often tenuous relationship between the study objectives
and the performance criteria.
Review of Models
There is a large amount of published literature and data on the performance of
various models1'2'6. A simple review of the documentation will indicate the
accuracy with which selected processes are modelled. The engineers conducting the
study should attempt to show why selected models are deficient or sufficiently
accurate for their purposes in the light of the problem review and special hydro-
logic and hydraulic processes that are required for their particular study. The
purpose here is to show how models may be used to support one another, or to
establish a reasonable sequence of model use in the study. No model selection
should be undertaken previous to this stage; first of all review all available data
and available study resources including the models, then establish the criteria for
selecting the models and finally select the model sequence.
It is much simpler to seek out and review the available data than to collect
data from scratch. In two recent Canadian studies several years of sewer water
level records as well as rainfall records were surprising finds. Such data could
form part of the model selection criteria; to make maximum use of available data
the study should not be constrained to use a model unsuited to these (unknown) data.
197
-------
The initial review should include a description of available manpower,
time, and money resources; the management of a study is an attempt to maximize
the level of detail subject to manpower, time and money constraints.
Model selection criteria should be clearly stated in the study report. It
is easy to criticize a study at a later date for not using a model which would
describe processes to a much higher level of system detail if it is not made
clear that the simpler model was used because of limited time and money. Other
criteria include availability of the model, availability of advice, experience
with similar or alternative models, availability of data, and costs.
MODEL CREDIBILITY
An additional series of verification, validation and sensitivity tests will
produce sufficient information to answer a wide range of questions about the
model's performance.
Verification Tests
Verification tests use some specific conditions for which the model response
can be exactly predicted to check if indeed the model is structured as intended.
Verification tests are not conducted by comparison of model responses with those
of the actual system to be modelled; rather, comparisons between model responses
and theoretically anticipated results are made in as many cases as possible. The
input data need not be physically reasonable.
It would be useful to create a standard input file for verification tests
for stormwater models. A hypothetical system comprising two simple, square sub-
catchments, of say one acre each would be suitable. The sub-catchments could be
joined by a pipe of standard diameter and simple form. The hydrograph from the
first sub-catchment is attenuated in the pipe. At the downstream end of this pipe
the hydrographs from both sub-catchments are superimposed. At the outlet of the
198
-------
pipe, the combined hydrograph could be routed through a simple standard storage
tank. The purpose of this verification data set would be simply to test the
algorithms for:
(1) the generation of the overland flow hydrograph;
(2) the addition of two hydrographs;
(3) routing in the pipes; and
(4) storage routing in storage tanks.
For example, we have used the following tests:
(1) zero rain to ensure that no runoff is generated;
(2) very steep catchments to ensure that the hydrographs generated are very
similar to the input hyetograph;
(3) light rain and high infiltration rates to ensure that no runoff hydrographs
are generated;
(4) completely impervious catchments to ensure that the total volume of runoff
is equal to the total volume of rainfall;
(5) low infiltration rates and high runoff to ensure that the correct amount of
infiltration is subtracted;
(6) very flat pipe gradients to check the surcharge calculation;
(7) similar tests with small diameter pipes and high pipe roughness;
(8) high and low initial abstractions; and
(9) tests on the storage routing parameters, viz. the outlet rating curve and
the storage curve.
In all of these tests particular attention is paid to the summary output:
total precipitation, total gutter flow, total snowmelt, total infiltration, etc.
This is an essential check that the numbers generated by the computer are.
sensible.
199
-------
Perhaps special algorithms and data files should be built into the model.
When the appropriate button is pushed the model automatically carries out a
series of verification tests. In our undergraduate classes the following
verification tests are normally carried out prior to any analysis being done:
1 Low imperviousness 5%
2 High imperviousness 95%
3 High detention 0.62 - 1.84 (in)
4 Low detention 0.01 - 0.1 (in)
5 PCTZER (High) 95%
6 PCTZER (Low) 555
7 High continuous rain 10 inch/hour
8 Low continuous rain 0.1 inch/hour
9 High continuous
infiltration 2 inch/hour
10 Low continuous
infiltration 0.1 inch/hour
Validation and Calibration Tests
Validation implies the. comparison of results to field measurements, to another
model known to be accurate, or to some other adequate criteria to ensure that the
model is producing accurate data. If these comparisons indicate that the model
results are not sufficiently accurate, the model is altered and the procedure is
repeated. This validation process generally involves several iterations before a
satisfactory confidence level is achieved. A recent paper describes the process
for combined quantity and quality modelling7. Techniques used in validation include:
(1) validation of parameters and results against field observations;
(2) cross correlation of model results with those of another proved, usually
discrete event model; and
(3) some combination of field observations and field modelling.
The most accurate method of validation is the comparison of output from the
verified model against corresponding field measurements.
The validation process should be limited to the range of parameter values
applicable to the normal operating conditions of the system. First, acceptable
tolerances must be established. They Should be related to achieveable field
200
-------
observations and the accuracy of field equipment. Then, emphasis should be placed
upon critical parameters -- those that have the greatest effect on the performance
of the system. Reasonable assumptions may be satisfactory for less critical
parameters.
The procedure, which may be necessary, of improving empirical constants in the
model to achieve proper correlations is also called calibration. Once verified
and calibrated, the model can be applied with confidence in the evaluation of the
real system. Important parameters are:
(1) percentage imperviousness
(2) width of sub-catchments
(3<) initial infiltration rate
(4) final infiltration rate
(5) infiltration decay rate
(6) ground slope
(7) Manning's roughness for overland flow
(8) depression storage for both pervious and impervious areas.
Validation tests using a full input data set may provide a useful test of the
logic underlying the data set.
Level of Discretization
The procedure for systematic disaggregation has been described in an earlier
paper by the author8. Disaggregation implies more sub-catchments of smaller area
and a finer time step. For coupling the time increment to the size of the sub-
catchment elements, the concept of an impulse response function is useful, i.e. an
instantaneous unit hydrograph. For example, it may be accurate enough for our
purposes to represent this response function by a time vector of 20 elements.
Sensitivity tests in which the time step is systematically changed may also be
appropriate. In practice, disaggregation and sensitivity analysis proceed
201
-------
simultaneously8. Often, careful disaggregation and appropriate selection of the
time step will produce better results than extensive optimization of empirical
factors such as Manning's roughness for pipes and overland flow. A recent paper9
by Alley and Veenhuis provides some guidelines.
There is an obvious conflict in the selection of the level of discretization.
Consultants' costs increase rather rapidly with an increasing number of sub-catch-
ments. More data has to be abstracted and prepared, and more expensive computer
runs will result. On the other hand, the client will usually prefer to have a high
level of discretization for the agreed fixed study price. It is useful to have, at
the outset of a study, an indication of the desirable level. As a guide, it is
important to identify all hydro!ogically significant elements in the system, where
it is necessary for hydrographs to be generated. If these elements can be identified
at the time the terms of reference are set out, consultants will have a better idea
of the scope of the work. It makes sense to face this possible conflict squarely
at the beginning of the study.
Sensitivity Analysis
Sensitivity analysis proceeds by holding all parameters but one constant at
their expected values, and perturbing that parameter within reasonable expected
limits such that the variation of the objective function can be examined. If what
appear to be small perturbations of the parameter produce large changes in the
objective function, the system is said to be sensitive to that parameter. The user
must obtain a measure of how accurate that parameter must be represented in his
model. If the objective function is not sensitive to the pertubated parameter, then
the parameter need not be accurately represented. If the system is insensitive to
the pertubated parameter, the parameter and its associated process is redundant and
the process should be deleted. It must be stressed that the actual values of the
constant parameters may affect the sensitivity analysis and so their values should
202
-------
be typical of the conditions being modelled.
Here again algorithms should be available that permit the user to easily
conduct a sensitivity analysis. When another button is pushed, the computer
requests the user to identify the parameter whose sensitivity is to be tested,
and the range of perturbed values. The data file will be automatically rebuilt
and the tests carried out. In addition, all output hydrographs should be plotted
on the same family of curves in order to present the impact immediately to the user.
Essentially there are two components of the hydrograph: peak flow and runoff
volume. The parameters which affect each component are as follows:
A: Peak Flow: slopes (pipe and land) B: Volume: infiltration
Manning's n surface retention
overland flow width PCTZER
pipe length % impervious
Control of Errors
Of course, it is not sufficient merely to carry out the required verification,
validation, and sensitivity analyses to various levels of discretization. These
results have to be presented in the report in a way which ensures that the
purpose of the tests has been achieved. The author of the report must explain the
results, in other words, interpret the results in a satisfactory way. The
verification tests must be shown to produce expected results. The calibration
and validation results must be shown to be reasonable. The trends resulting from
the sensitivity analysis must be shown to make sense. The purpose of this is to
provide evidence that the model is indeed performing in a reasonable way: (a) the
verification tests demonstrate that there are no serious errors in the coding of
the model, (b) the validation tests at a simple level of discretization, say one
or two sub-catchments and pipes, demonstrate that the model is being used in a
203
-------
in a reasonable way, (c) the validation tests on the full data set, i.e. for all
of the sub-catchments acting together as a hydrological system, indicate whether
serious blunders have been made in preparing the input data set, and (d) the
sensitivity analyses will indicate whether the level of effort put into estimating
the individual parameters is appropriate based on their significance in affecting
model results.
The required output interpretation, arguing that the results in fact make
sense, helps to ensure that the authors have not misinterpreted the model results.
CONCLUSIONS
In this paper we have suggested that the following topics should be included
in an initial terms of reference:
Problem Review - The problem review should identify all hydraulically and
hydrologically significant elements in the study area so that the model selected
can be shown to include all relevant processes.
Study Objectives - The study objectives should be reviewed to show how the objective
functions, viz. pollutographs and hydrographs, relate to the design alternatives.
Performance Criteria - The performance criteria for the comparison of one design
alternative against another must be correctly identified so that the simplest
possible model can be justifiably selected.
Requisite Accuracy - Accuracy of field measurements for validation should be
carefully reviewed in order to ensure that the model runs are not inordinately
expensive.
Review of Available Programs - Several programs should be suggested or selected for
review. The review should consider process models as well as system models, and a
sequence of models.
Study Resources - Study resources include time, manpower and money and these, in
turn, will determine which of the models may be selected.
204
-------
Model Verification - Verification tests should be required on a simple data set
consisting of two small catchments connected by a simple pipe and feeding into a
simple storage tank. The verification tests should demonstrate that the coding is
performing as intended.
Model Calibration and Validation - Validation tests should be carried out on one
of the sub-catchment data sets to demonstrate that the model is being correctly used.
Validation tests should be carried out on the full data set to demonstrate that the
input data are reasonable.
Minimum Level of Discretization - The smallest number of sub-catchments required
for modelling the system should be selected commensurate with the objectives of
achieving the best design at a reasonable cost. These minimum levels should correspond
to the disaggregation necessary to identify hydrographs at all hydrologically
significant elements in the drainage system, which could be listed.
Sensitivity Analysis - Sensitivity analyses should be carried out on a minimum
number of parameters, for example, infiltration parameters, roughness values, widths
of sub-catchments, etc., to identify which are of most significance and hence to
justify the effort put into their estimation.
Data Preparation and Output Interpretation - All output should be interpreted to
demonstrate that the model is performing in a logical way.
Pocumentation - The version of the program actually used in the study should be
identified and appropriate documentation sources listed in the report. In addition,
the machine readable input and output files should be archived for future use.
205
-------
REFERENCES
1. Chung San Chu and Bowess, C.E., "Computer Programs in Water Resources",
WRRC Bulletin 97, University of Minnesota, 1977, (263 pp.).
2. McPherson, Murray B., "Overview of Urban Runoff Tools of Analysis",
EP79-R-20, GREMU-79/01, Ecole Polytechnique de Montreal, 1979, (31 pp.).
3. Croley, Thomas E., "Hydrologic and Hydraulic Computations on Small
Programmable Calculators", Iowa Institute of Hydraulic Research, 1977,
(837pp.).
4. Patry, 6., et at, "Description and Application of an Interactive Mini-
Computer Version of the ILLUDAS Model", Proc. of SWMM Users' Group
Meeting, May 1979, p 242-274.
5. Thompson, L. and Sykes, J., "Development and Implementation of an Urban-
Rural Sub-Catchment Hydrologic Model (SUBHYD) for Discrete and
Continuous Simulation on a "Micro Computer", Proc. SWMM Users'
Group Meeting, May 1979, pp. 320-361.
6. Colyer, P.O. and Pethick, R.W., "Storm Drainage Design Methods", INT 54,
Hydraulics Research Station, Wallingford, 4th Ed., September 1977,
(110 pp.).
7. Jewell, T.K., et al, "Methodology for Calibrating Stormwater Models", Proc.
SWMM Users' Meeting, May 1978, pp. 125-173.
8. James, W., "Developing Simulation Models", Journal of Water Resources
Research, Vol. 8, No. 6, December 1972, pp. 1590-1592.
9. Alley, W.M. and Veenhuis, J.E., "Determination of Basin Characteristics for
an Urban Distributed Routing Rainfall - Runoff Model", Proc. SWMM
Users' Meeting, May 1979, pp. 1-27.
206
-------
SIMULATION OF EFFECTS OF URBANIZATION
ON STORMWATER HYDROGRAPHS AND POLLUTOGRAPHS—
A REGIONALIZED PARAMETRIC APPROACH
by
10 O
Donald E. Overton , William L. Troxler and Ernest C. Crosby
The University of Tennessee Runoff Model (TENN-I) for simulating
storm hydrographs and pollutant yields was developed in the course of
performing three separate but complimentary studies. All three studies
had the same fundamental objective, i.e. to evaluate the effects of
specialized land use on stormwater runoff and its associated quality.
The three studies were funded by the following three Federal Agencies
and dealt with the land use indicated:
Study
Federal Agency Contract-Grant No. Period Land Use
(1) U.S. Department of Energy EY-76S-05-4946 1975-79 Coal Strip
Mining
(2) U.S. Department of Interior TENN-A-046 1976-78 Urbanization
Office of Water Resources
Technology
(3) U.S. Air Force AF FO 8635C 77 1975-79 Air Force
Bases
TENN-I was developed in the course of analyzing 410 storms observed
on 36 watersheds. The storm sample included watersheds in agricultural,
urban and 100% forested land use conditions as well as watersheds under-
going coal strip mining and watersheds at a U.S. Air Force Base.
1. Assoc. Prof., Dept. of Civil Engr., the Univ. Of Tennessee,
Knoxville.
2. Environmental Engr., Hydroscience, Inc., Knoxville, TN
3. Assistant Prof., Dept. of Civil Engr., Univ. of Texas at Arlington.
207
-------
2. Objective of TENN-I
TENN-I was developed for the purpose of simulating stormwater hydro-
graphs from a real time or design storm rainfall time distribution and
land use and soil type of the watershed of interest. The watershed is
considered to be a lumped system, and the required basin characteristics
are: percent of watershed that is forested (PF), percent that is impervi-
ous (PI), percent in strip mining or denuded (PS), and surface drainage
area in square miles (SQMI).
Runoff volume and the associated rainfall excess time distribution
are simulated from the input rainfall using the U.S. Soil Conservation
Service Curve Number model (CN) [1]. If a runoff hydrograph is read into
the program associated with a rainfall hyetograph, the program will compute
a CN from this information. Otherwise, a CN must be read into the computer,
and would be simulated using the procedures specified in reference [1].
TENN-I makes provision for simulating a unit hydrograph or unit
response function, URF, which is convoluted with the rainfall excess
hyetograph simulated using the CN model. This computer program is an
adaptation of the simulation phase of the TVA double triangle model reported
by Ardis [2] and later modified by Betson [3]. A more detailed description
of the developement of TENN-I has been reported by Overton, Troxler and
Crosby [4] and Overton and Crosby [5],
TENN-I also simulates pollutant loads for the associated storm. This
simulation is a function of the above specified watershed and storm char-
acteristics. Hence, additional parameters need not to be read in.
208
-------
3. Simulation of Storm Hydrograph
a. Normalized Unit Response Function (NURF)
The URF in TENN-I is based upon Ardis1 [2] quadrilateral function.
The URF was coupled with the CN model to form the TVA double triangle
model. The shape of the URFs and associated CN have been optimized on a
total of 410 storms in studies performed by Ardis [2], Betson [3], Overton,
Troxler and Crosby [4] and Overton and Crosby [5]. Optimizations were
performed by a pattern search routine.
The quadrilateral URF was based on the concept of partial-area run-
off which assumes that the initial or quick response from a watershed
comes from the riparian areas. As other areas of the watershed become
saturated, they too begin to contribute to runoff in the form of a de-
layed response.
Ardis [2] assumed that these two responses could be simulated by
two separate triangle response functions as shown in Figure 1. When
added together, these two triangles form a quadrilateral unit response
function for the storm as shown in Figure 2.
In deriving the URF, it was assumed that (1) the peak of the delayed
response (UR) occurs at the end of the initial response (T2), and (2) the
time bases of both responses and the time to peak of the initial response
(Tl) must be integer multiples of DT. No assumption was made concerning
the relative volumes contained in the initial and delayed responses or
concerning the relative magnitudes of the peaks of the invidivual
responses.
The double triangle URF is defined by the five parameters UP, UR, Tl,
T2, and T3. T3 is determined by:
T3 = (NOBS - NRAIN +1) *DT (1)
209
-------
TIME (HR)
UNIT
INPUT
INITIAL RESPONSE
CO
z
o
a.
C/3
LU
BC
a
UJ
CO
ac
uu
I
DELAYED RESPONSE
T1 T2
TIME |HR)
T3
Figure 1. Partial Area Runoff Concept Represented by an Initial and
Delayed Response (After Ardis (2))
210
-------
TIME (HR)
1
i • 1/OT
I
•OT
T1 T2
TIME (HR)
T3
Figure 2. Double Triangle Model for Unit Response Function
(After Ardis (2))
211
-------
where NOBS = number of storm hydrograph ordinates in multiples of DT and
NRAIN = number of rainfall increments in multiples of DT. By maintaining
a unit volume, UR is calculated from:
UR = (2 - (UP*T2))/(T3-T1) (2)
Defining a storm URF therefore involves determining values of UP, Tl, and
T2.
The parameters UP, Tl, and T2 were optimized using the pattern searcii
technique. The objective function was the minimization of the sum-of
squares of errors between observed and simulated discharges. Since all five
parameters describing the model were allowed to vary from storm to storm,
the model is, considered nonlinear. Rainfall excess was optimized using the
SCS-Curve Number model after setting it equal to the observed direct
runoff volume.
The variability of the URF from storm to storm within a watershed
was expalined by normalizing the time and discharge scale by the associated
URF lag time, TL, where TL = time lapse between occurrence of 50% of the
rainfall excess block and 50% (or 1/2 inch) of the URF volume, as shown
in Figure 3. These normalized URFs are hereafter referred to as NURFs.
A NURF for each major land use category was identified and they are
shown in Figure 4. The categories are, (1) strip mined, (2) 100% forest,
(3) urban without extensive storm sewers, (4) urban with extensive storm
sewers, and (5) agricultural. As a matter of providing a reference or
bench mark, the NURF observed for sheet surface runoff from a plane,
reported by Overton and Meadows [6], and the NURF derived for the V-shaped
watershed of Overton and Brakensiek [7] (see Figure 5) from the kinematic
wave equations are also shown. The stripmined watersheds were in
212
-------
TIME (HfO
TL-
(T1.UP)
(0.0)
T50
TIME (HR)
T23 (T3.0)
Figure 3. Evaluating Lag Time of the Double Triangle Unit Response
Function
213
-------
ro
-IS.
100% FOREST
URBAN
AGRICULTURE
SHEET SURFACE RUNOFF
,_ URBAf! WITH EXTENSIVE STORM SEWERS
Figure A- Normalized Unit Response Functions for Different Land Uses.
-------
ro
en
to
c
m
en
i
in
zr
Qi
T3
ro
a.
z:
Q>
n>
Q.
ft-
tt>
-s
o
rt>
ct
o
Cu
3
Q.
CD
OJ
7
3
CO
IO
•^4
CO
-------
previously 100% forested areas. Hence, a pattern is shown where the
forested and strip mined watersheds have a small initial response whereas
the urban and agricultural watersheds have a larger initial response.
This implies that imperviousness and crop land produce much more surface
runoff. The storm sewered NURF has an even higher initial response (80%)
which is more likely generated by the runoff collection and rapid transport
to the watershed outlet. Sheet surface runoff has no delayed response as
does the V-shaped NURF. Hence, the variation between the two NURFs is
attributed to the geometry of the flow path of the V-shaped watershed.
b. Lag Time
Lag time for a storm is simulated in TENN-I using the concept that
it varies inversely with the generating rainfall excess intensity.
TL~ie"b (3)
This variation is well known for sheet surface runoff of overland flow
[6] and documentation of this variation for watersheds is growing [5,6],
The first such documentation of lag time variation on a watershed was
reported by Minshall [8] and is represented by the five URFs he obtained
from each of the five storms (see Figure 6). Lag time is increasing with
decreasing rainfall excess rate. These URFs were derived from storms of
10 minute duration as opposed to the long duration storms from which the
URFs were optimized in the noted studies [2, 3, 4, 5].
For sheet surface runoff, lag time is related to input intensity as
TL (min) = 0.58 (nL/v^)°-6/ie°-4 (4)
Eq. (4) can be derived from the kinematic wave equations [6], and lag
modulus, y, is defined as
216
-------
I 1 1
MAY 27,1948 X * 4.75 in/hr
SEPT. 2. 1941 X * 2.65 in/hr
APRIL 17. 1941 X* 1.95 in/hr
OCT. 22, 1941 X* 1.52 in/hr
JULY 20, 1948 X*0.95 in/hr
20 40 60 80 100 120
TIME FROM BEGINNING OF EXCESS RAINFALL,
-------
y = 0.58
(5)
where n = Manning roughness coefficient, L - length of plane in feet and
S = slope of plane.
Lag modulus is solely a function of the geometry and hydraulic
roughness of the plane.
The lag modulus model is used in TENN-I as an analogy for watershed
runoff and standardized to turbulent sheet surface runoff by fixing the
exponent on input intensity at 0.4. Lag modulus was optimized for all
36 watersheds [4, 5] using a weighted storm rainfall intensity, WRE,
WRE = Z ie2(J)/.I ie(J) (6)
where N is the number of time intervals equal to DT
For watersheds, Eq. (1) becomes
A
TL (min) = y/WRE°'4 (7)
Lag modulus was related to watershed characteristics in the following
manner:
Rural Watersheds
y (hrs) = 0.060 * SQMI + 0.0203*PF + 1.16 (8)
where PF = % forest
Urban Watersheds
y (hrs) = 3.24 [SQMI/PI]0'6 (9)
where impervious
218
-------
c. Convolution of the URF with Rainfall Excess Time Distribution
The final step in simulating a stormwater hydrograph is to convolute
the storm URF with the rainfall excess time distribution. The URF is
defined by simulating storm lag time from Eq. (7) using either Eqs. (8)
or (9) depending on the watershed land use. WRE is calculated by Eq. (6).
4. Simulation of Pollutant Yield
a. Load Modulus
TENN-I simulates storm pollutant yield using a load modulus (Ibs/acre
-in of storm runoff) as a function of percent stripped or denuded, PS, laq
modulus and percent forest or trees in the following form:
flw = C1 * PS - C2 *u* PF + C3 (10)
The coefficients were optimized using stormwater quality data on a total
of eleven watersheds, six undergoing coal strip mining and five urbanized
[5].
Eq. (10) was derived from a mass balance, and each of its terms
represents a component of pollutant yield.
Cl * PS = source of pollutant or soil loss
C2 * y * PF = deposition between source and outfall, and
C, = storage in watershed picked up and redeposited.
O
Eq. (10) is implicitly based upon a plug flow or first in first out invent-
ory concept where the processes are in equilibrium.
The coefficients optimized for the coal strip mined and urbanized
watersheds are shown in Table 1.
b. Storm Load
Once load modulus for the watershed pollutant has been simulated, the
219
-------
storm pollutant yield, SPY, is simulated by
SPY = u, * AREA * 640 * SRO (11)
w
where SRO is the total storm runoff in surface inches, and AREA is in
acres.
220
-------
Table 1
Coefficients in Pollutant Yield Model
(Equation 10)
URBAN [Ref. 4]
Coefficients
Pollutant
Suspended Solids
FE
MN
CA
MG
Sulfate
Total Alkalinity
Suspended Solids
FE
MN
CA
MG
Sulfate
Total Alkalinity
Source Deposition
Cl C2
16.7
0.442
0.0072
0.147
0.0597
0.0719
0.319
COAL STRIP MINED (Active)
18.2
0.323
0.0161
0.131
0.133
1.39
0.19
21.5
0.568
0.0092
0.189
0.113
0.0216
0.128
[Ref. 5]
1.28
0
0
0
0
0
0
Storage
C3
62.3
1.54
0.20
2.45
0.0323
1.24
6.71
576.7
0.12
0.002
0.38
0.40
2.45
2.00
221
-------
5. References
1. U.S. Soil Conservation Service, National Engineering Handbook
Section 4, Hydrology, Washington, D.C., 1972.
2. Ardis, C.V., Jr. "Storm Hydrographs Using a Double Triangle Model,"
TVA, Div. of Water Control Planning, January 1973.
3. Betson, R.P., Urban Hydrology - A Systems Study in Knoxville,
Tennessee, TVA, Div. of Water Management, 1976.
4. Overton, D.E., W.L. Troxler and E.G. Crosby, "Simulation of Effects
of Urbanization on Stormwater Runoff and Quality," Univ. of Tennessee,
Water Resources Center, Report No.74, Die. 1979.
5. Overton, D.E. and E.C. Crosby, "Effects of Contour Coal Strip Mining
on Stormwater Runoff and Quality - New River Basin, Tennessee,"
Department of Civil Engineering, The University of Tennessee,
Knoxville, 1979.
6. Overton, D.E. and M.E. Meadows, Stormwater Modeling, Academic Press
Inc., New York, 1976.
7. Overton, D.E. and D.L. Brakensiek, "A Kinematic Model of Surface
Runoff Response," Proc. of Intern. Symposium oh Results from Repre-
sentative and Experimental Watersheds. Wellington, N.Z., Dec. 1970/
IASH-UNESCO, Paris, 1973.
8. Minshall, N.E., "Predicting Storm Runoff on Small Experimental
Watersheds," J. of Hydr. Div., A.S.C.E., (HY8), pp. 17-38, 1960.
222
-------
STORMWATER MODELLING APPLICATIONS
IN THE CITY OF EDMONTON
BY
M. AHMAD
Drainage Systems Upgrading Engineer
Edmonton Water and Sanitation
12220 Stony Plain Road
Edmonton, Alberta
Canada
Presented at:
SWMM Users Group Meeting
10-11 January, 1980
Gainesville, Florida
U.S.A.
223
-------
STORMWATER MODELLING APPLICATIONS
IN THE CITY OF EDMONTON
by
M. AHMAD1
INTRODUCTION
The purpose of this paper is to present an overview of the various
modelling applications recently carried out by Edmonton Water and
Sanitation (Drainage Engineering Section).
The City of Edmonton has played a leading role in the implementation of
new modelling technology, and is among the first cities in Canada to use
sophisticated hydrologic simulation models for planning and design
purposes. Edmonton Water and Sanitation has been using the SWMM/WRE [1,2]
models since 1975 to upgrade the level of flooding protection in the
existing combined sewer areas.
Initially, the use of computer models was restricted to relief sewer
studies. At present, however, the City is employing a set of models
including SWMM/WRE [1,2], STORM [3], HYMO [4], and ILLUDAS [5] in a
variety of applications such as: (1) preparation of master drainage
plans for the combined and separated sewer areas, (2) development of a
long term program to provide relief to the existing overloaded combined
sewer systems, (3) evaluation of combined sewer pollution abatement
strategies, (4) development of a stormwater management criteria for new
•
areas, (5) analyzing small watersheds for drainage planning purposes, and
(6) verifying new system designs in newly developing areas prior to
construction.
Drainage Systems Upgrading Engineer, Edmonton Water and Sanitation
12220 Stony Plain Road, Edmonton, Alberta, T5N 3M9
224
-------
EDMONTON SEWER SYSTEM
The City of Edmonton is the capital of oil rich Canadian Province of
Alberta. It has a population of about 500,000 and is presently growing at
a rate of 2.0% per year. The City is located along the banks of the North
Saskatchewan River which divides the City into two parts - Northside and
Southside - as shown in Figure 1. At present, approximately 73% of the
total City area (123.0 sq. mi.) is fully developed. The remaining area is
developing at an increasing rate of about 1300 acres/year. Over 2,200
miles of sewers (including sanitary, storm and combined) service the
developed areas of the City. About 20% of the total developed area in the
City is drained by the combined sewers while the remainder by the
separated storm sewer systems. Figure 1 also shows the location of the
combined sewer area in the City- Most of the presently undeveloped areas
located within the City Limits are drained either by the existing storm
sewers or by the tributaries of the North Saskatchewan River such as Mill
Creek, Fulton Creek, Whitemud Creek, and Blackmud Creek.
The existing combined trunk sewer system drains an area of approximately
12,000 acres which is located in the older central core of the City, which
was developed prior to the early 1950's. During storm events this system
overflows to the river at 12 locations and at one location to Mill Creek.
The combined sewer system is characterized by deep tunnels (up to 125 feet
below ground) ranging in size from 48" to 126" and connected to local
lateral and collector sewers through drop structures. The existing
combined trunk sewer system is very complex and consists of a large number
of overflow weirs and interconnecting conduits. Portions of this system
are over 60 years old. Continual development and redevelopment in the
combined sewer areas gradually increased the imperviousness ratio
resulting in the creation of runoff greatly .beyond the design
considerations of the original sewer system. These increases in runoff
due to increased impervious areas have caused overloading of the existing
combined sewers with the resultant surcharging causing frequent basement
and surface flooding problems.
The existing combined sewer system is intercepted by two major
interceptors, one serving the combined system on the north of the river,
225
-------
and the other, the system lying south of the river. These interceptors
converge upstream of the Gold Bar Waste Treatment Plant.
The Gold Bar treatment plant provides treatment to approximately 92% of
the City1s sanitary sewage flow plus intercepted storm flow from the
combined sewerage system. This plant has a design capacity of 43.5 MIGD.
The average daily dry weather flow to plant is approximately 48.8 MIGD.
An extensive expansion program is currently underway which will expand the
design capacity to approximately 69 MIGD during the coming year which will
provide adequate capacity to accomodate the City's needs until
approximately 1989.
MODELLING APPLICATIONS
1. Combined Area Relief Studies
Combined Drainage Area Study [6]:
The first modelling study was commissioned in 1975 to investigate
storm drainage conditions and to determine relief requirements in the
combined sewer districts of the City. Prior to that time it was
considered that total separation of the existing combined system
would be required to provide effective relief. However, it was
recognized that the unusual and complex combined sewer system in the
City could not be analyzed with traditional methods. It was,
therefore, necessary to use the then new computer modelling methods
for the analysis and design of relief sewers in this area. The
original version of the WRE model [2] which was the most appropriate
model then available was used in this study. This study concluded
that neither major trunk sewers nor sewer separation were required
for hydraulic relief and the majority of the flooding problems in the
combined sewer area were found to be largely due to inadequate local
sewers. The study also concluded that considerable in-system storage
capacity was available for possible control of combined sewer
overflows. However, shortly after the completion of this study, it
was discovered that the WRE model overestimated the pipe system
storage in surcharged sewers. Subsequently, modelling results were
updated using the corrected version of the WRE model in an in-house
226
-------
study [7] completed last year. Revised modelling results indicated
that the original study underestimated the trunk relief requirements
and a more extensive relief program than previously determined would
be needed. To finalize the combined trunk relief requirements an
update study will be conducted this year.
Combined Area Relief Program:
Two earlier studies completed in 1976 [6] and 1979 [7] indicated that
the relief requirements for the existing combined sewer area are
two-fold: firstly, the major trunks and secondly, the lateral and
collector sewers. A 22-year long range program for relief and
upgrading of the existing overloaded combined lateral and collector
sewers was adopted by City Council and initiated in 1978 after
determining the relief requirements for individual subareas within
the combined sewer area. The entire 12,000 acre combined area was
divided into 32 drainage subareas which were then analyzed and ranked
according to the assessed priority for relief. The priority ranking
analysis was based on an evaluation of a number of factors for each
subarea including sewer density, length of the existing relief
sewers, reported basement flooding frequency, potential for
redevelopment and need for environmental improvements. The entire
combined sewer area has been divided into three groups, namely, high,
moderate and low, depending on the need and urgency of the required
relief. On the basis of this priority ranking analysis, an annual
construction program has been established for each year until year
2000, with specific areas identified for each year's construction.
This relief construction schedule has been adopted by City Council,
and hopefully will not be significantly altered to satisfy
independent political needs in the future.
Relief sewer construction in the combined sewer subareas is
tentatively scheduled for completion by the end of year 2000 and the
detailed relief analysis and design will be completed by 1986.
Relief construction schedule for the combined trunks will be
finalized upon completion of the update study which is to be
completed by mid 1980. The present relief program is based on a
relief sewer construction rate of 500 acres per year. Relief
227
-------
construction has already been completed in more than 1500 acres of
the combined area.
Financing for total cost of this relief program is being derived from
a special re-development assessment levy charged against all
re-developing properties. It is not anticipated that any portion of
the cost of this program will be financed through the general tax
levy.
The total construction cost for providing relief to the existing
combined lateral and collector sewers is estimated at approximately
$40 million (1977 dollars). An additional cost of about $1.3 million
was estimated in the 1976 study [6] in order to relieve the existing
combined trunk sewers. A cost comparison for two relief alternatives
namely, complete sewer separation and combined relief is presented in
Table 1 [8].
Table 1
COST COMPARISON FOR
RELIEF ALTERNATIVES
RELIEF ALTERNATIVE
SEWER SEPARATION ($) COMBINED ($)
Laterals & Collectors 104,000,000 40,000,000
Trunks & Interceptors 60,500,000 1,300,000
Overflow Control Structures - 7,700,000
Total 164,500,000 49,000,000
NOTE: Costs in 1977 dollars
From Table 1 it is evident that alternative 2 (combined) is the most
cost-effective solution. However, results of the update study (to be
completed in 1980) may indicate that trunk relief costs are
228
-------
considerably higher than those originally estimated. Relief costs
for the combined lateral and collector sewers were estimated using a
simple relationship between "impervious storage factor (ISF)" and
relief cost in dollars per acre for a subarea. A relationship as
shown in Figure 2 was developed using the actual cost data for three
subareas, Calder, Norwood and Alberta Ave/Eastwood. Relief
construction in these areas was completed prior to 1978. Impervious
Storage Factor for a drainage area can be calculated using the
following equation:
I.S.F.=
where
1. = total length in feet of sewers of diameter d.
d. = diameter of a sewer in feet
I = imperviousness ratio of the area
A = drainage area in acres
Impervious Storage Factor, in general, reflects the drainage
characteristics of a catchment by relating the pipe-full storage
available within the sewer system to the upstream contributing
impervious area. This concept was initially used in the City of
Winnipeg Relief Study [9].
Estimated relief cost for the recently analyzed Garneau area is found
to be in close agreement with that originally estimated using ISF
curve. However, actual relief construction cost in the Northcote
area in which construction was completed in 1979 was much lower than
originally estimated mainly due to overestimation of the
imperviousness ratio used in computating ISF value which resulted in
higher estimated cost for relief.
2. Combined Sewer Pollution Abatement Study [10]
As a follow-up to the first study in 1976 [6] an extension study was
229
-------
carried out to determine the feasibility of controlling surcharge and
storage in the deep tunnel system in order to reduce the number of
overflows from the combined sewage overflowing to the North
Saskatchewan River and to partly equalize flows to the treatment
plant. The number of overflows from the combined system was
simulated over 1 year period using the STORM model which was
calibrated against overflow measurements made by the City in 1975.
The modelling results indicated that relatively small volumes of
storage would be quite effective in reducing overflows and overflow
pollution. Subsequently, more detailed models were employed to
assess the effectiveness of employing a variable height regulator
gate to control storage in the sewer system, and for an accurate
assessment of first flush pollutant loadings. Automatic regulators
in the other cities were investigated and a programme for the
implementation of computer controlled variable regulator gates at
locations throughout the combined sewer system was developed. The
estimated cost of the real-time control system was $7.0 million. The
conclusions and the conceptual design of the real-time control system
will be unaffected by an increase in system capacity for relief
purposes.
3. Pollutant Loadings from Combined and Storm Sewer Discharges Study [11]
This study was carried out in order to determine the significance of
the proposed real-time control system in the context of a
comprehensive pollution abatement programme for the City. The scope
of the study included an assessment of the effects of combined sewer
overflows (controlled and uncontrolled), separated storm sewer flows,
and storm flows from undeveloped areas within the City of Edmonton on
the water quality of the North Saskatchewan River.
The STORM model was used to simulate flows and quality from the
combined and separated sewer systems using 1969 hourly precipitation
and temperature data.
Pollutant loadings from separated storm sewers and combined sewer
overflows were estimated for both present and future conditions with
230
-------
various pollution abatement measures employed. Contributions from
rural areas were considered only briefly due to lack of adequate
data. The STORM model was calibrated for BOD and suspended solids
using flow and quality measurements collected by the City. A simple
manual model of the river water quality was used to determine the
response of the river to pollutant loadings discharged from the major
sources.
A comparison of the pollutant loadings discharged annually to the
river under the present conditions by the sewage treatment plant
(both treated effluent and overflow by-pass), combined sewer
overflows, and separated storm outfalls is presented in Table 2. It
is evident that the Sewage Treatment Plant (STP) is the major source
of pollution in terms of BOD and coliforms. Combined sewer overflows
and STP bypasses contribute about 58% of the total coliforms load
discharged to the river annually. Separated areas account for 68% of
the total suspended solids loads, although they account for only 20%
of the total effluent.
A comparison of the effects of pollutant loads from different sources
was also made for the future conditions. Five different
alternatives, listed in Table 3, were compared by estimating the
pollutant loads from three major sources, i.e. STP, combined and
storm sewer discharges. For comparison purposes, it was assumed that
bypasses at the treatment plant would not occur after completion of
the ongoing expansion program in 1989- The relative effluent
volumes, and the estimates of BOD, SS, and total coliforms loadings
discharged annually to the river from three major sources for the
five alternatives considered in this study are compared in Figures 3
to 7. Table 4 compares annual pollutant loadings to the river for 5
alternatives.
It is evident that the implementation of the proposed Real Time
Control System and use of stormwater management techniques in new
developments will substantially reduce the discharge of pollutant
loads to the river. By stormwater management in new developments
alone, a reduction of some 29% in total annual suspended loads can be
231
-------
achieved (see Figures 3 and 5). Total coliforms loadings to the
river can be reduced up to 87% (compared to Alternative 1) by
disinfecting the combined sewer overflows in addition to the
implementation of the proposed in-system storage programme as
suggested in Alternative 5.
A comparison of the simulation results with and without the proposed
in-system storage indicates that a significant reduction in pollution
will result by using the proposed storage. For example, BOD, SS and
coliform loads will decrease by approximately 55%, 58% and 63%
respectively, after implementation of the proposed management system.
Similarly, the average annual number of overflows from combined
areas, which occur mainly during the months of May to September, will
drop from 24 to 8.
A simple, steady-state model based on the Streeter-Phelps approach
was used to analyze the river water quality. Pollution discharged to
the river was simplified by combining all separated storm sewer
discharges into three major inflows to the river and by combining all
combined sewer discharges into a single inflow point. Modelling
results for the present conditions for BOD indicate that for a river
flow of 4,000 cfs, (exceeded 50% of the time), 50% of the total
increase in BOD concentration is caused by the combined sewer
overflows. However, DO concentration in the river never drops below
7 mg/1. The net effect on the DO deficit of the river due to all
waste sources is less than 1 mg/1 within the City Limits.
River water quality analysis for the 5 alternatives described earlier
was also done based on a stream flow of 4000 cfs and 1989 development
and treamtent plant conditions. Results indicated that virtually no
variation from existing condition is experienced with any of the 5
alternatives for the largest storm in 1969, except for total
coliforms count which are slightly improved with disinfection and
combined sewer in-system storage (Alternative 5). Figure 8 shows a
comparison of the river water quality response for the present and
future conditions at the combined sewer overflow discharge point.
232
-------
The study concluded that high coliforms concentrations are a major
problem in the North Saskatchewan River and the present total
coliforms criterion of 5000 mpn/100 ml is frequently violated even
during dry-weather conditions. It was estimated that for a typical
year the total coliforms loadings from the combined and separated
areas assuming treatment plant expansion as proposed resulted in 27
violations for existing conditions as compared to 13 after
implementation of combined sewer in-system storage and 8 violations
after disinfection along with in-system storage.
After completion of the pollution abatement study, it became apparent
that water quality standards in the receiving waters of the North
Saskatchewan River as set down by Alberta Environment (the Provincial
environmental control authority) would not be met even with the
proposed controls or even with total elimination of combined sewer
overflows. Alberta Environment withdrew their restrictions and water
quality standards pending further study on the limits permitted
therein, and implementation of the pollution abatement program
through real-time control was suspended by the City pending the
outcome of their review. At present there is no indication as to
when this issue may be resolved.
4. Miscellaneous Studies
A number of in-house modelling studies are being conducted by
Edmonton Water and Sanitation to develop master drainage plan for the
separated storm sewer systems located within the City boundaries.
For instance, 8 out of 14 major storm trunk systems in the City which
drain a total area of about 40,000 acres have already been analyzed
using SWMM. Lumped modelling techniques are being employed to
analyze these large basins. The results of these studies are being
used to refine the designs for completion of the development of the
respective basins and to evaluate the potential for expansion of
existing drainage basins using stormwater retention techniques.
In another in-house study [12] completed in 1978, HYMO and STORM
models were used to analyze flooding and water quality problems in
233
-------
predominantly undeveloped Mill Creek basin which has a drainage area
of approximately 36.5 square miles. Hydrographs for the various
design storm conditions were developed and pollutant loading from the
major sources were compared.
At present, the Drainage Engineering Section is conducting a SWMM
calibration and verification study using recorded rainfall and sewer
flow measurements in Norwood, Fulton Drive and Groat Road areas.
Preliminary study results indicate that the simulated flows compare
very well with those recorded and it appears that with some
calibration efforts, SWMM can be adjusted for Edmonton Conditions.
CONCLUSIONS
Successful modelling applications in the City of Edmonton have proven that
large economic benefits can be derived from a substitution of the
traditional approach to drainage design with more sophisticated
computerized methods. Computer models are invaluable tools which provide
aid in the understanding of the complex and large sewer systems and
provide the engineer greater capacity and flexibility to evaluate
alternatives.
ACKNOWLEDGEMENTS
I would like to acknowledge with gratitude the contribution and assistance
of the following persons in developing computer modelling expertise at the
City:
Dr. Paul Wisner, University of Ottawa, Ottawa, Ontario (formerly with
James F. MacLaren Ltd.). who introduced computer modelling to the City of
Edmonton and contributed in many studies for the City.
R.S. Cebryk, Cumming-Cockburn & Associates Limited (formerly with the City
of Edmonton).
J.F. Hugo, Chief - Drainage Engineering, City of Edmonton.
234
-------
TABLE 2
TOTAL AVERAGE ANNUAL
LOADS TO THE RIVER
STP
STP
Storm
Sewer Combined
Out- Sewer
flows Overflows Total
Volume (109 cu. ft.)
Suspended Solids
(106 Ibs.)
BOD (10& Ibs.)
Total Coliforms
(1017 MpN)
2.79
6.03
5.02
11.1
0.14
2.34
1.46
8.57
0.75
24.0
1.12
0.6
0.11
2.8
0.41
7.6
3.79
35.17
8.01
27.87
TABLE 4
COMPARISON OF ANNUAL LOADINGS
TO THE RIVER FOR FIVE ALTERNATIVES
Alt. 1 Alt. 2 Alt. 3 Alt. 4 Alt. 5
Volume (10 cu. ft.)
Suspended Solids
(106 Ibs.)
BOD (106 Ibs.)
Total Coliform
(1015 MPN)
5.2
5.1
5.2
5.1
5.1
43.3 41.7 30.6 29.0 41.7
9.5 9.3 9.2 9.0 9.3
841.2 362.0 841.2 362.0 110.0
235
-------
TABLE 3
LIST OF ALTERNATIVES
Alternative 1
1989 treatment plant effluent conditions.
Existing combined sewer overflows without in-'system storage.
Separated areas without stormwater management.
Alternative 2
1989 treatment plant effluent conditions.
Proposed in-system storage in combined areas.
Separated areas without stormwater management,
Alternative 3
1989 treatment plant effluent conditions.
Existing combined sewer overflows without in-system storage.
Separated areas with stormwater management in new developments,
Alternative 4
1989 treatment plant effluent conditions.
Proposed in-system storage in combined areas.
Separated areas with stormwater management in new developments,
Alternative 5
1989 treatment plant effluent conditions.
Proposed in-system storage in combined areas plus disinfection
of overflows.
Separated areas without stormwater management.
236
-------
LOCATION PLAN
CITY OF EDMONTON
COMBINED SEWER AREA
237
-------
r\>
CO
co
-------
COMPARISON OF POLLUTION FROM
TREATMENT PLANT AND SEWER OVERFLOWS
FIG. 3
FIG. 4
FIG. 5
FIG. 6
FIG. 7
TOTAL ANNUAL AVERA
LOAD TO RECEIVING Wi
FOR FUTURE COrOmC
239
-------
5-7
5-7
5-3
5-5
56
5-7
3-51
8-42 8-42
8-SO 8-48
8-4Z
B.O.D. (nvj/l)
0.0.
121
120
6-20
107
113
113
5-97
3-45
345
!|l|!|:|!|!i
ilil'liWtt
0-43
SUSPENDED SOLIDS (m,/i)
TOTAL COLIFORM di03MPN/iOOmi)
PRESENT CONDITIONS
ALT. 1
ALT. 2
ALT. 3
ALT. 4
ALT. 5
FUTURE CONDITIONS
COMPARISON OF WATER QUALITY
AT COMBINED SEWER OVERFLOW FOR
PRESENT AND FUTURE CONDITIONS
WORST EVENT IN
I969
Figure 8
240
-------
REFERENCES
1. SWMM - Storm Water Management Model - Volumes I to IV, prepared for
the U.S. Environmental Protection Agency, Washington, D.C., July,
1971.
2. Water Resources Engineers, Inc., San Francisco Stormwater Model
User's Manual and program documentation; Department of Public Works,
City and County of San Francisco, California, 1972.
3. STORM - Storage, Treatment, Overflow, Runoff Model - User's Manual ,
the Hydrologic Centre, Corps of Engineers, U.S. Army, Davis,
California, July, 1976.
4. HYMO - Problem Oriented Computer Language for Hydrologic Monitoring,
U.S. Agricultural Research Service, U.S. Department of Agriculture,
May, 1973.
5. ILLUDAS - The Illinois Urban Drainage Area Simulator, Illinois State
Water Survey, Urbana, Illinois, 1974.
6. "Combined Drainage Area Study", report by James F. MacLaren Limited,
Toronto, to the City of Edmonton, 1976.
7. "Combined Trunk Sewer System Analysis Study", In-house study by
Edmonton Water and Sanitation, 1978.
8. "Capital Expenditures and Financing for Upgrading the Combined Sewer
System", In-house study by Edmonton Water and Sanitation, May, 1978.
9. "Report on Flood Relief Requirements for Five Combined Sewer
Districts in the City of Winnipeg", prepared by James F. MacLaren
Limited to the City of Winnipeg, April, 1977.
10. "Feasibility Study for Combined Sewer Pollution Abatement by Real
Time Control", report by James F. MacLaren Limited, Toronto, to the
City of Edmonton, 1976.
241
-------
11. "Pollutant Loadings from Storm and Combined Sewer Discharges - An
Assessment of Stormwater Management Measures for the City of
Edmonton", report by James F. MAcLaren Limited to the City of
Edmonton, January, 1978.
12. "Mill Creek Ravine Hydrologic Study", In-house Study by Edmonton
Water and Sanitation, February, 1979.
242
-------
STORMWATER RUNOFF MODELING OF THE
TAMPA PALMS PROPERTY
By
Kevin Smolenyak1
Introduction
The Tampa Palms development which is located approximately eleven
miles northeast of downtown Tampa is a resident!'ally oriented master
planned community totaling 5,400 acres in size with an ultimate density
of 13,000 units. Stormwater discharges from this property will enter
the Hillsborough River, a potable source of water for the City of Tampa.
In order to assess the effects of the development of the Tampa
Palms Property on the quantity and quality of stormwater runoff entering
the receiving systems, the Storage Treatment Overflow Runoff Model
1 o
(STORM) * was employed. STORM, a widely used and proven hydrologic
model (continuous simulation), was initially applied to the project in
its current undeveloped condition, using water quantity and quality data
collected from several monitoring stations strategically located at
surface runoff sites throughout the property for model calibration
(Figure 1). The proposed changes in land use and drainage characteris-
tics due to development were then incorporated into the model to predict
future stormwater runoff quantity and quality conditions and to analyze
potential impacts to receiving systems.
1. Environmental Engineer (E.I.)
The Deltona Corporation
3250 S.W. 3rd Ave., Miami, Fla. 33129
243
-------
Figure 1
Pre-Development Drainage Basins
244
-------
Model Input Requirements
The STORM model requires hourly precipitation data for simulation
purposes. Accordingly, hourly precipitation data for the years 1948-
1975 were obtained from the U.S. National Weather Service (Ashville,
North Carolina) for the Tampa Airport station, which is approximately
thirteen miles southwest of the Tampa Palms property. In addition, con-
tinuous (hourly) rainfall data from a rain gage at the City of Tampa
Water Treatment Plant adjacent to the property was used in the analysis.
Average daily pan evaporation rates for each month of the year are
used for the recovery of depression storage and soil moisture capacity.
Fourteen years of pan evaporation data from the Lake Alfred experimental
station (closest available long-term record) were analyzed to determine
average daily pan evaporation rates.
An extensive literature review was undertaken to determine appro-
priate average daily pollutant accumulation rates for each land use in
the developed and undeveloped areas. It was determined that studies
conducted by the United States Geological Survey (USGS) in Broward County,
Q
Florida3' 4' 5> 6> 7> and Miami, Florida would provide the most suitable
loading rates for modeling the developed areas of the Tampa Palms prop-
erty for the following reasons.
1. The USGS study areas and Hillsborough County are similar geog-
raphically and climatically (e.g. mild ground slopes, similar
rainfall patterns etc.)
2. The published USGS stormwater pollutant loadings were the
8,
result of an excellent water quality monitoring program
(e.g. well documented study, an extensive number of storms
245
-------
monitored, a large number of water quality samples taken for
each storm resulting in well defined pollutographs, good
data collection and data management system used).
3. Four land use sites were monitored in the study (i.e. single
family residential, transportation, commercial shopping center
for the Broward County sites and a multi-family residential
site in Miami). These land uses encompass the various types
of urban development planned for the Tampa Palms property.
4. The single family residential site in Broward County employs
open channel swale drainage which is planned for the Tampa
Palms site.
Loading rates used for the undeveloped areas were average values
obtained from studies within Florida.
Basins Modeled
Within the Tampa Palms property six basins were defined and surface
water monitoring stations setup at each (Figure 1). Of these six basins
Taylor Slough (MLS) and Outfall Slough (CPS) were modeled by STORM for
the following reasons.
1. Each basin lies entirely within the Tampa Palms property.
2. The boundaries of each basin are well defined.
3. The channels draining these basins are well defined.
4. The Taylor Slough Basin boundaries will only change slightly
after development. Therefore, the basin is ideal for comparing
stormwater quantity and quality before and after development.
5. The Outfall Slough basin (pre-development) serves as the existing
subbasin of a much larger post-development Outfall Slough basin.
246
-------
Since high density housing is planned for this subbasin, the
magnitude of the effects of such development on stormwater
quality is of interest.
6. The other post-development basin, West Branch, would be dif-
ficult to model in pre and post-development phases because
of the discharge of backwash water from the Tampa Water Treat-
ment Plant onto the basin. In addition, accurate comparisons
of the pre and post-development stormwater quality is not
possible because the two basins are different in size and
boundaries.
For modeling purposes the pre and post-development Taylor Slough
and Outfall Slough basins were subdivided into land uses. A comparison
of the land uses of the basins is given in Table 1.
Calibration of Pre-Development Basins
Certain input parameters in the STORM model are highly site speci-
fic and affect such factors as the quantity and time distribution of
stormwater runoff and the magnitudes of the pollutants modeled. Cali-
bration involved the adjustment of these input parameters until reason-
able agreement was achieved between predicted and measured (or actual)
stormwater quantity and quality. Considerable effort was expended in
determining appropriate values for the various parameters required as
input for the model to help insure the validity of the output.
Water quantity and quality information obtained from the surface
water monitoring stations at both Taylor and Outfall Sloughs were used
to calibrate the model for each pre-development basin. The pollutants
sampled at these sites were modeled by STORM and include BOD5, TSS
247
-------
Table 1. Land Use Breakdown (Taylor and Outfall Basins)
Taylor Slough - Pre and Post-Development
Land Uses
Taylor Slough/Pre-Development
Land Use Area (Acres)
Rangeland
Wetland
Upland Forest
188.3
162.8
131.9
Total=483.0
Taylor Slough/Post-Development
Land Use Area (Acres)
Single Family
Residential
Open Space
Golf Courses
Roads
Commercial
294.7
126.2
64.3
54.7
23.7
Total=563.6
% of Basin Area
39.0
33.7
27.3
% of Basin Area
52.3
22.4
11.4
9.7
4.2
Outfall Slough - Pre and Post-Development
Land Uses
Outfall Slough/Pre-Development
Land Use Area (Acres)
Wetland
Range!and
Upland Forest
121.0
70.5
21.5
Total=213.0
Outfall Slough/Post Development (Subbasin No. 5)
Land Use Area (Acres)
141.0
Open Space
Multl-Family
Residential
Roads
Commercial
75.4
10.0
5.1
Total=231.5
% of Basin Area
56.8
33.1
10.1
% of Basin Area
60.9
32.6
4.3
2,2
248
-------
(total suspended solids), TOTN (total nitrogen as N), TOP (total ortho-
phosphate as P and TCOL (total coliforms).
Hourly rainfall data (September, 1978 thru June, 1979) obtained
from a station at the City of Tampa Water Treatment Plant, which lies
adjacent to the Tampa Palms property were used in the model simulation.
Calibration procedures were conducted using information gathered between
December, 1978 and May, 1979 at the appropriate water monitoring stations,
For each pre-development basin the models were calibrated for stormwater
runoff quantity followed by calibration for runoff quality.
For the calibration of stormwater runoff quantity reasonable agree-
ment was achieved between:
1. Total runoff volumes for the period of record.
2. Individual event runoff volumes.
3. Peak runoff rates, runoff durations and timing (i.e. the shape
of the individual hydrograph).
For the calibration of stormwater runoff quality average annual
pollutant concentrations, determined using pollutant accumulation rates
based on water quality studies in Florida , were compared to the mean
concentrations determined via the surface water monitoring program. In
the STORM model a linear relationship exists between the pollutant ac-
cumulation rates and output pollutant loads and concentrations.
249
-------
Accordingly, appropriate changes were made in the loading rates for the
individual pollutants to achieve agreement between the measured and
predicted pollutant concentrations.
Calibration Results-Quantity
Good agreement was achieved between measured and predicted runoff
volumes and individual storm volumes for Taylor Slough. Individual
storm eventsexcept that of May 8, 1979 (Figure 2) were calibrated by
comparing mean daily runoff discharges. Drainage within the basin is
slow and storm event durations are on the order of days, therefore the
use of mean daily discharges for calibration is sufficient. Figure 2
represents the measured and calibrated outflow hydrograph for Taylor
Slough for a 10.6 inch rainfall (10.25 inch, 12 hour total = 100 year
event). Measured and calibrated runoff volumes for this event were
7.47 inches and 7.88 inches respectively. The discrepancy in peak
flows can be attributed to the extreme nature of the event, the com-
plex nature of the basin and the influence of the initial soil mois-
ture conditions chosen for the relatively short simulation period.
The analysis of runoff volume for Outfall Slough indicated a
high baseflow component within the basin. Rainfall infiltrates into
the well drained soils and is transported laterally through the soils
due to a steep groundwater gradient to the outfall point of the basin.
Examination of groundwater wells within the basin indicates that be-
tween the uplands area in the northwest portion of the basin and the
outfall point, a distance of 4,500 feet, a 9.3 foot gradient exists
during the wet season with a slightly higher gradient during the dry
season. STORM does not simulate base flow. Water quality modeling,
however, should not be significantly affected since the major source
250
-------
Figure 2
Taylor Slough - Runoff Hydrograph for
May 8, 1979 Storm Event
300
250
200
^ RAINFALL
STORM EVENT 5/8/79
CO
u_
o
o
LEGEND
• FlOW 30
MONITOHINO
(T) WO fAMPU
FlOW
f KIDICTf D
FLOW (ITOHMI °
5/10
5/M
DATES FROM.^/8/I9 TO'5/11/79 3TATION TAYLOR SLOUGH
-------
of pollutant loads results from surface runoff. Pollutant contribu-
tions from base flow are minor due to filtering capabilities of the
soils and the slow movement of water through the soils. Water quality
samples were also taken at Outfall Slough for the 10.6 inch rainfall
(10.25 inch, 12 hour total=100 year event) which began May 8, 1979.
Backwater from Cypress Creek inhibited flow out of the basin approxi-
mately two days after initial stormwater runoff from the basin.
Comparisons of calibrated values for the soil storage parameters
within the STORM model to those of a study by the US6S using a
modified version of the Georgia Tech watershed model indicate excellent
agreement. The US6S study was conducted to simulate streamflow in the
Hillsborough River. The Tampa Palms property lies within the 160 square
mile basin modeled by the USGS.
Calibration Results-Quality
As mentioned earlier quantity calibration for both Taylor and Out-
fall Slough involved altering pollutant accumulation rates in order
that measured and predicted (annual) mean pollutant concentrations were
identical.
Comparisons of individual storm pollutant loadings were only pos-
sible for the May 8, 1979 event. Measured pollutant loads (determined
using water quality samples taken at various times during the storm
(Figure 2) are compared with those predicted by STORM below.
252
-------
Taylor Slough - Storm Event (5/8/79) Total Loads (Pounds)
BOD TSS TOP TNIT TCOL (MPN)
Measured 1618 3235 33.5 184 74 x 1012
Model 630 4347 8.8 163 28 x 1012
Outfall Slough - Storm Event (5/8/79). Total Loads (Pounds)
BOD TSS TOP TNIT TCOL (MPN)
Measured 844 1628 6.35 171 31 x 1012
Model 257 1653 2.49 39 6.2 x 1012
STORM Application to Post-Development Basin
After calibrating both Taylor and Outfall Sloughs using hourly
rainfall data from a station at the City of Tampa Water Treatment Plant,
the, two basins were modeled using hourly rainfall data from the National
Weather Service (NWS) station at the Tampa Airport (nearest long-term
hourly rainfall record available). The rainfall record chosen for simu-
lation purposes (6/48 - 8/52, a total of 1553 days) represents a period
of average rainfall for the Tampa area. The results of these simulations
will be discussed later when they will be compared with the results for
the post-development basins for the same period of record.
Calibrated input variables obtained from STORM runs of the pre-
development basins were applied to the STORM runs of the post-development
basins when appropriate. For example, input parameters affecting hydro-
graph shapes were changed to account for the decrease in the time of
response of the system (i-e. decreased time of concentration and time
of recession over time to peak).
253
-------
Pollutant accumulation rates for single family residential, com-
mercial and highway (road) areas were derived from extensive stormwater
studies conducted by the USGS in Broward County, Florida3' 4» 5j 6> 7
Multi-family rates were derived from a similar USGS study in Miami,
Q
Florida. For the open space areas calibrated accumulation rates, de-
termined from pre-development STORM simulations for the Woodlands/Wet-
lands areas, were used. Golf courses were assigned the calibrated ac-
cumulation rates for pasture areas to account for increased nitrogen
and phosphorous loadings due to fertilization.
Results and Discussion
Tables 2 and 3 summarize the results of the STORM simulations
(6/48 - 8/52) for the pre and post-development Taylor Slough and Out-
fall Slough basins. The following findings are noteworthy:
1. The number of storm events for the simulation period
(.6/48 - 8/52) increases after development. This is due
to changes in the shapes and durations of individual storm
hydrographs following development. In addition, response
times in the developed basins are shorter and therefore
runoff defined as a single event (an event is defined as
beginning when runoff is initiated and ending when runoff
stops in the pre-development basin might be characterized
as two separate events in the post-development basin.
2. For the pre-development basins 16.0% of the rainfall appeared
as surface runoff. A USGS study, using 10 years of data
(10/64 - 9/74) from a streamflow station at Cypress Creek near
Sulfur Springs, reports a mean annual runoff coefficient of
254
-------
Table 2. Taylor Slough - A Comparison of Pre and Post-development
Water Quantity and Quality Conditions for tne Simulation
Period June, 1948 through August, 1952.
Taylor SIouqh/Pre-Oeve1opment
Number of storm events » 87
Rainfall • 207.34 inches (period of record 6/43-8/52)
Runoff * 33.25 inches
Fraction of rainfall as runoff » 16.05
Area of basin * 483 acres
Average Annual Pollutant Loads
Total Pounds
Pounds/acre
Cone (mg/1 }
B005
2,210
4.58
3.05
TSS
13,547
28.1
18.7
TOP
31
.064
.043
TOTN
570
1.18
.79
TCOL
97 x 1C12 MPN
2 x 1011 MPN/acre
2.95 x 104 MPN/100 ml
Taylor SI ough/Post-Oev^looment
Number of storm events 3 112
Rainfall « 207.34 inches (Period of record 6/48-8/52)
Runoff * 26.75 inches
Fraction of runoff » 12.9*
Area of basin - 563.7 acres
Average Annual Pollutant Loads
B005 TSS TOP TOTN TCOL
Total Pounds 5,513
Pounds/acre 9.78
Cone (mg/1) S.06
* Increase 114
over Pre-de-
velopment loads
20,110 67 1 ,074 324 x 1C1?.MPN
35.7 .119 1.91 5.75 x 101' MPN/acre
29.4 .097 1.57 10.5 x 104 MPN/100 m
27 36 62 188
Table 3. Outfall Slough - A Comparison of Pre and Post-Oevelopment
Water Quantity and Quality Conditions for the Simulation
Period June, 1948 through August, 1952.
Outfall 51ough/Pr°-Oeve1opment
Number of storm events » 91
Rainfall » 207.34 inches (Period of record 6/48-8/52)
Runoff » 33.05 inches
Fraction of rainfall as runoff * 16.0*
Area = 213 acres
Average Annual Pollutant Loads
Total Pounds
Pounds/acre
Cone (mg/1)
TSS
5,262
24.7
16.5
BOD5
911
4.28
2.36
TOP
8.4
.04
.03
TOTN
237
1.11
.74
TCOC
39.3 x 10]2 MPN
1.8 x 10'' MPN/acre
27,200 MPN/100 ml
Outfall Slough (Subbasin No. 5)/Post-Oeve1oament
Number of storm events » 94
Rainfall =• 207.34 inches (Period of record 6/48-8/52)
Runoff » 27.1 inches
Fraction of rainfall as runoff » 13.05
Area * 231.5 acres
Average Annual Pollutant Loads
BOD5 TSS TOP TOTN
Total Pounds
Pounds/acre
Cone (mg/1)
* Increase
over Pre-de-
3,054
13.2
10.8
152
102,748
444
363
1 ,700
55.9
.24
.20
400
1124
4.86
3.S6
338
TCOC
172 x 10]2 MPN
7.4 x 1011 MPN/acrs
133,455 MPN/100 ml
311
velopment loads
255
-------
16.6%. The drainage basin studied by the USGS is 160 square
miles in area and includes the Tampa Palms property.
3. Simulations indHcst& that surface runoff will decrease by
approximately 20% following development. This reduction is
primarily attributed to the improvement of soil hydrologic
/
ratings from pre to post-development conditions as a function
of the enhanced setting resulting from drainage network imple-
mentation. Additional reduction in runoff is expected to
occur as a result of the drainage system's increased potential
for groundwater recharge and evapotranspiration.
4. Without accounting for the significant stormwater treatment
which will be achieved in the post-development, Tampa Palms
water management system simulations indicate stormwater
pollutant loads (lb/acre/year) at the outfall points of the
basins will increase following development. The larger in-
creases in pollutant loadings for Outfall Slough can be
attributed to the high pollutant accumulation rates used in
modeling the high density multi-family residential area found
in the post-development Outfall basin.
The predicted increases in stormwater pollutant loads due to the
development of Taylor Slough and Outfall Sloughs are highly conservative
(i.e. predicted increases in pollutant loads are significantly higher
than expected). The above conclusion is based upon the fact that pol-
lutant removals via stormwater treatment within the post-development
256
-------
drainage system (i.e. treatment due to retention of stormwater in
DRA's) and detention throughout the system cannot be modeled by
STORM. This treatment can only be reflected by decreasing the pollut-
ant accumulation rates, which are linearly related to the stormwater
runoff quality. As mentioned earlier, in the post-development basins
pollutant accumulation rates for single and multifamily residential
sites, commercial sites and highway sites were developed from ex-
tensive stormwater runoff studies conducted by the USGS in Broward
County ' ' ' ' and Miami, Florida . In these studies pollutant
loads were measured at outfalls near the source allowing for little
or no stormwater detention. Only the single family site in Broward
County, Florida provided any type of treatment before water quality
samples were taken. This treatment, which can be considered minimal
compared to that which will be obtained in the Tampa Palms drainage
system, consisted of drainage to grass swales which discharged directly
into a storm sewer where the water quality samples were gathered.
Significant stormwater treatment will be obtained when runoff
from impervious areas enters into the post-development drainage system
of the Tampa Palms property. This drainage system is comprised of
grass swales (roadside and rear yard) which convey stormwater overland
to a system of drainage retention areas (natural wetlands) and lakes
followed by final diffuse sheet flow outfall to naturally vegetated
wetlands prior to introduction to receiving waters. The processes
governing these treatments include grass-soil filtration, nutrient
uptake via cypress roots and underlying sediments, controlled sedi-
mentation in drainage retention areas and emergent macrophytic uptake
257
-------
in the shallow littoral zone of the lakes. None of these removals
were accounted for in the post-development basin simulations (i.e.
unadjusted loading rates developed from the USGS studes were used).
The actual pollutant removals which can be attained in the post-
development drainage system are difficult to quantify. However,
1213
Heaney and Huber ' have developed a relationship between pollutant
removals and stormwater detention times based on studies of the
Kissimmee River basin in Florida. Applying this relationship to the
post-development drainage basins leads to the following treatment
efficiencies.
Maximum
Runoff Detention
Detention time I
Basin
Taylor Slough
Outfall Slough
West Branch
The reported detention times assume that the drainage system is
not overloaded (i.e. individual basin storage is capable of handling
the amount of stormwater via slow release through discharge pipes
without overflows of the discharge weirs). Overflows (high rate dis-
charges) would lead to reduced detention times and corresponding de-
creased pollutant removals. However, STORM simulations (period of
rainfall 6/48 - 8/52) of the post-development Taylor and Outfall
Sloughs, assuming a single conservative drain-down rate for each de-
tention basin, indicate that discharges over the weirs are infrequent,
thereby satisfying the assumption criteria.
258
Runoff
Detention
(inches)
1.03
2.81
1.47
Detention
time
(days)
16
24
30
%
Pollutant
Removal
86
95
98
-------
Applying these removals to the STORM simulation results for the post-
development Taylor Slough and Outfall Slough (Subbasin No. 5) basins,
the following results are obtained (See Table 4).
The reported removals apply to treatment within the basins. Further
removals are possible as the stormwater runoff discharges as sheetflow
across naturally vegetated wetlands (Hillsborough River Conservation
Area) prior to introduction to receiving waters (Hillsborough River).
The net impacts of stormwater discharges from the Tampa Palms prop-
erty on the Hillsborough River should be minimal, because pollutant
loads (mass) from the developed property will only constitute a small
percentage of the loads carried by the river. Comparisons of calculated
yearly pollutant loads for the post-development Taylor and Outfall
(Subbasin No. 5) basins and the Hillsborough River at Fowler Avenue
(approximately 1.5 miles south of the property) are shown on Table 5.
Loads for the Tampa Palms property were obtained from STORM simulations
(6/48 - 8/52) and are reported for prior to treatment and after treat-
ment conditions (See previous discussion). Calculated yearly loads for
the Hillsborough River at Fowler Avenue are based on water quality studies
14
by Hatcher and Courtney and a conservative average discharge of 450
cfs.
For all the pollutants, total loads for the post-development basins
represent a minor to very minor percentage of the total loads for the
Hillsborough River. Impacts, therefore are expected to be minimal. In
addition, stormwater discharges from the developed drainage system will
be attenuated (via controlled release) which will increase the ability
of the Hillsborough River to assimilate the loads.
259
-------
Table 4
Average Annual Pollutant Loads (Pounds/Acre)
For Taylor and Outfall Sloughs (Simulation Period June, 1948 through August, 1952)
30D5 TSS TOP
TOTN TCOL
Taylor Slough
(Pre-Oevelopment)
4.58
28.1
35.7
Tay-lor Slough
(Post-Development 9.78
prior to treatment
within Tampa Palms Water Management System)
% Increase over
Pre-Oevelopment
Loads
(114)
2.62
Taylor Slough
(Post-Development
following treatment
within Tampa Palms Water Management System)
(27)
12.2
% Increase over
Pre-Oevelopment
Loads
(-43)
(-57)
Outfall Slough
(Pre-Oevelopment) 4.28 24.7
Outfall Slough
(Post-Development 13.2 444
prior to treatment
within Tampa Palms Water Management System)
* Increase over
Pre-Oevelopment
Loads (152)
(1700)
26.0
Outfall Slough
(Post-Development 2.38
following treatment
within Tampa Palms Water Management System)
°i Increase over
Pre-Oevelopment
Loads (-44) (+5)
0.064
.119
(86)
.033
(-48)
1.18
2xlOU MPN/Acre
1.91 5.75x1O11 MPN/Acre
(62) (188)
.60 1.37x10a MPN/Acre
(-49) (-31)
.04 1.11 I.SxlO11 MPN/Acre
.24 4.36 7.4x10n MPN/Acre
(400) (338) (311)
.03 .77 I.SxlO11 MPN/Acre
(-25) (-31) (-17)
260
-------
Table 5
A Comparison of Average Annual Pollutant Loads (Ibs)
For the Hillsborough River at Fowler Avenue, Taylor Slough and Outfall Slough
ro
LOCATION
Hillsborough River
at Fowler Avenue
Taylor Slough
(prior to treatment)
Taylor Slough
(after treatment)
Outfall Slough
(subbasin No. 5
prior to treatment)
Outfall Slough
(subbasin No. 5
after treatment)
% of Total
Basin Area
100
.14
.14
.06
.06
BOD5
Load
7.96xl05
5510
1480
3050
551
X
.69
.19
.38
.07
TOTN
Load
7. 53x1 O5
1070
338
1120
178
%
.14
.04
.15
.024
TOP
Load
2.80xl05
67
19
56
6.9
%
0.24
.007
.02
.002
TCOL
Load
6.18xl015
• m
3. 24x10*'
1 O
7.72xlOIJ
1.72xl014
3.5xl013
X
5.2
1.3
2.8
.57
-------
REFERENCES
' 'Hydrologic Engineering Center, Corps of Engineers, "Storage
Treatment Overflow Runoff Model: STORM," Generalized Computer Program
723-58-L7520, Users Manual, Davis, California, August, 1977.
'2'Hydrologic Engineering Center, Corps of Engineers, "Guidelines
for Calibration and Application of STORM", Training Document No. 8,
Davis, California, December, 1977.
/0\
v ' Matt raw, H.C., Jr. and Sherwood, C.B., "Quantity of Stormwater
from a Residential Area", Journal of Research of the U.S. Geological
Survey, Vol. 5, No. 6, pp. 823-834, Nov. - Dec. 1977.
(^Mattraw, H.C., Jr., "Quality and Quantity of Stormwater Runoff
from Three Land Use Areas, Broward County, Florida", Proceedings Inter-
national Symposium on Urban Stormwater Management, University of Kentucky,
Lexington, Kentucky, July, 1978.
^Mattraw, H.C., Jr., Hardee, J. and Miller R.A., "Urban Stormwater
Runoff Data for a Residential Area, Pompano Beach, Florida", U.S.G.S.
Open File Report 78-324, Tallahassee, Florida, 1978.
^Miller, R.A., Mattraw, H.C. and Hardee, J., "Stormwater-Runoff
Data for a Commercial Area, Broward County, Florida", U.S.G.S. Open-
File Report 79-982, Tallahassee, Florida 1979-
(^Hardee, J., Miller, R.A., Mattraw, H.C. Jr., "Stormwater-Runoff
Data for a Highway Area", U.S.G.S. Open-File Report 78-612, Tallahassee,
Florida, June, 1978.
(^Hardee, J., Miller, R.A., and Mattraw, H.C., Jr., "Stormwater-
Runoff Data for a Multi-family Residential Area, Dade County, Florida",
U.S.G.S. Provisional Report, Tallahassee, Florida, 1979.
, W.C. and Heaney, J.P., "Urban Rainfall-Runoff Quality
Data Base, "U.S.E.P.A. Report EPA-600/8-77-009, July, 1977.
00)wanielista, M.P., "Non-point Source Effects", Florida Techno-
logical University, Report No. ESEI-76-1, Orlando, Florida, Jan., 1976.
^ 'Turner, J.F. Jr., "Streamflow Simulation Studies of the Hills-
borough, Alatia, and Anclote Rivers, West-Central Florida", U.S.G.S.
Water-Resources Investigations 78-102, February, 1979.
"'Heaney, J.P., W.C. Huber et al., Environmental Resources Man-
agement Studies in the Kissimmee River Basin, Final Report to Central
and Southern Florida Flood Control District, 1975.
^ 'General Development Corporation, "East Port St. Lucie Phase I -
Application for a Permit for the Construction of a Surface Water Manage-
ment System", October 31, 1975.
^4'Hatcher, J. and C.M. Courtney. 1979. Water Quality Analysis -
Tampa Palms, Appendix A15B, Tampa Palms, A.D.A., Deltona Corporation,
Miami, Florida.
262
-------
Drainage System Design and Analysis of the
Tampa Palms Property
By
Robert James Motchkavitz, P.E.
Introduction
This paper provides a review of the design objectives and analysis
procedures as well as a summary of selected simulation results for an
investigation of the drainage system proposed for the Tampa Palms prop-
erty, Hillsborough County, Florida. The proposed drainage system is of
current interest since it incorporates numerous "non-structural" Best
Management Practices including the use, as detention and conveyance net-
works, of natural and man-made wetland systems. In addition, a simple
procedure developed by the author referred to as the Conservative Weir
Approach, employing the Soil Conservation Service unit hydrograph method
and the TR-20 Hydrology Computer program was used in the design/analysis
investigations and is reported on herein.
Whereas such investigations are considered only as preliminary
studies in regards to structure design, they have proven to be of sub-
stantial value in obtaining regulatory agency acceptance. In addition,
the results have been used as input data and as qualitative support for
continuous water quality modeling (STORM) where simulation results require
detention area relationships of stage and discharge.
1. Director, Department of Environmental Services
The Deltona Corporation
3250 SW 3rd Avenue, Miami, Florida 33129
263
-------
General Project Description
The study site is located in northwestern Hillsborough County,
Florida, approximately 12 miles from downtown Tampa. The site is
bounded on the west by Cypress Creek and its floodplain; on the south by
Cypress Creek and Hillsborough River and their respective floodplains;
and on the east by Levee 112N (Lower Hillsborough Flood Detention Area)
which lies west of and adjacent to Trout Creek. An area of about 5400
acres included within the property boundaries is owned by Deltona.
Existing vegetative cover has been described pursuant to Florida
Land Use and Classification System by DuBois and Courtney and soil
mapping of the property has been performed through the Hillsborough
County Extension - Soil Conservation Service. Topography for the study
area has been reported through the auspices of the Southwest Florida
Water Management District (SWFWMD).
A review of climatological data and summaries provided through the
National Oceanic and Atmospheric Administration reveal several salient
facts regarding the study vicinity. Firstly, the area is subject to fre-
quent thundershowers during the months of June through September. During
this period about 60 percent of the annual rainfall can be anticipated.
Average annual rainfall for the Tampa station for 79 years of record
is 47.72 inches. The wettest year recorded was 1959 with 76.57 inches
of precipitation and the driest year occurred in 1956 when 28.89 inches
was recorded. An evaluation of recent rainfall for Tampa indicates an
upturn trend in 16 year moving average rainfalls after continual de-
creases for the past 9 years. Finally, annual evaporation as measured
at the Lake Alfred Experimental Station (closest established site) is
264
-------
reported to be 68.74 inches based on 13 years of data.
Drainage System Design/Performance Objectives
Meaningful storm water drainage design addresses both the quantity
and quality aspects of excess precipitation. In particular, of paramount
concerns with the quantity of discharge are the protection of the project
against flooding, and the attenuation of peak runoff rates and the re-
duction of total volumes from within the developed property to approximate
natural discharge conditions and thus prevent flooding and erosion prob-
lems downstream, or upstream as in the case of backwater flooding, The
quality of the drainage discharge must be such that it will be compatible
with basin management objectives and regulatory requirements and not cause
significant degradation to intermediate receiving bodies.
In order to most efficiently and cost effectively meet the quantity
and quality objectives, natural drainage systems of depressions and
sloughs and the accompanying natural vegetation should be maintained and
utilized to the fullest extent practical. This will result in a system
that is not only visually pleasing and effective in meeting design ob-
jectives, but one which has low initial costs and requires a minimum of
skilled maintenance for upkeep. It has been the goal to design such a
system for Tampa Palms and the approach and results of those design efforts
are herein presented.
Drai nage
Numerous considerations and objectives have been incorporated into
the design of the subject project drainage system. The following list
includes, not necessarily in order of importance, several of those items
considered in the drainage design process:
265
-------
1. • Provide adequate drainage to ensure protection to private property
and drainage works in the development.
2. Design system to prevent flooding impacts on upstream and downstream
properties.
3. Design system to minimize fill requirements thereby preserving natu-
ral setting and reducing development costs (energy and resources).
4. Establish sub-basin control water levels to be consistent with re-
corded pre-development low ground water table levels.
5. Maintain surficial ground water table regime to insure continued
ground water recharge contribution to Cypress Creek and Hillsborough
River.
6. Incorporate and preserve function and hydro-period of wetland areas
included in development plan.
7. Preserve existing drainage basins and patterns of flow to the greatest
degree possible.
8. Maximize system's storm runoff retention and detention capabilities.
9. Attenuate peak discharges and total quantities to approximate pre-
development conditions.
10. Provide a system that is consistent with design standards and minimizes
the need for maintenance and/or operational procedures.
11. Insure that discharge velocities will not result in scour of contri-
butory or receiving systems.
12. Provide for maximum use of grassed conveyance swales and non-structural
controls in lieu of storm sewer systems.
13. Allow for low flow discharge (approximating natural conditions) at
two locations through the proposed Cypress Creek levee to be conveyed
266
-------
via existing natural outfalls at times when Creek stage is low.
14. Insure that the ultimate outfalls of the drainage sub-basins are
spatially separated from the open water of the receiving body by
a wetland marsh "buffer" system.
15. Incorporate as many best management techniques (BMT's) as possible
into the drainage designs so as to reduce impacts (quality and quan-
tity) of project on receiving water bodies.
16. Maintain and provide for "base flow" discharges from the property to
replenish receiving bodies during low flow conditions.
Design Concepts
The major obstacle to designing the drainage system for this project
was not the ability of the system to adequately drain the site of excess
rainfall. Sufficient topographic relief and existing hydrologic condi-
tions are more than adequate to allow for a standard design to meet local
flooding objectives. The burden of the task proved to be a result of the
ambitious determination to incorporate and maintain natural wetland sys-
tems into the drainage plan while maintaining the hydraulic viability of
the conveyance network so as to prevent both internal and external flood-
ing. As a result of experiences in similar situations, review of the lit-
erature, and discussions with knowledgeable agency personnel, the design
concepts necessary to achieve the design/performance objectives were able
to be defined.
The proposed drainage system can best be described as a network of
grassy swales, natural wetland areas and lakes which have the potential
to store, treat and, subsequently, slowly release excess rainfall from
the site to naturally vegetated sheet flow areas prior to introduction
267
-------
to receiving waters. The system is designed to be positively drained
(i.e., by gravity) and will not be adversely impacted by receiving body
flood stages up to and including the 25 year flood. Beyond that, re-
ceiving water flood stage, localized internal flooding (minimum elevation
streets) of short duration (as a result of secondary drainage structure
inadequacies) could occur depending upon the severity of the accompanying
storm event. However, the statistical likelihood of simultaneous flood
and design storms is highly remote and standard engineering practice and
financial practicality do not support the investigation and design for
such extreme events. We would, however, note that the system could still
positively drain rainfall from the property even in the case of the 100
year flood in the receiving body.
Stormwater within the drainage system is generally conveyed frpm
upland development sites to wetland areas, to lakes and other conveyance
ways. In all instances where the wetlands are part of the conveyanceway
the system is designed so that velocity of flows will be low enough to
«
prevent scour or other damage to it.
The preserved wetland areas within the drainage retention areas
(DRA's) will be allowed to maintain their natural hydroperiod by means
of either stabilized earthen berms or control structures (weirs) which
will prevent severe unnatural water level draw-downs.
To aid the reader in understanding the concepts described above, at-
tention is directed to Figure 1. The typical flow of runoff from the
system is portrayed as traveling first through roadside or rearyard swales,
thence into wetlands or lakes, and then through the storm water conveyance
system to the ultimate outfall. From the point of ultimate outfall storm
water flows in an overland fashion through extensive areas of natural
268
-------
Figure 1
Drainage Design Concepts
DRAINAGE SYSTEM-JliSIM DETAILS
WETIAHO* PBOCE3JIHO-CONVEVANCE
ro
cr>
BUY. vr Lin UKMiita . ^u ft ur t.ui uftni
TO flCAR YARD SWALE j TO STREET SWALE
NOTEiDRAWING NOT TO SCALE
-------
vegetation (on Deltona property) before entering open receiving waters.
The major drainage control structures are designed to maximize stor-
age and detention while both assuring flood protection to the upland prop-
erties and allowing for positive drainage (by means of small diameter pipes)
to regain storage potential after storm events.
The structure design typically prevents the lowering of ground water
levels below historic low readings. The structures themselves are to
function as weirs but'their ultimate shape and characteristics (construction
material) can vary so long as they provide the discharge characteristics
the design was based on. The structural designer may find it aesthetically
pleasing and cost effective to construct stabilized and protected earthen
weirs or may elect to adhere to a more conventional solution by erecting
concrete structures. Such decisions will be made at the time of platting
and submittal of detailed drainage plans.
The grassed swales in street right-of-ways, rear yards and conveyance
networks are to be shallow (typically 6" deep) with minimum slopes (iden-
tical to street slopes) to maximize contact time and facilitate particulate
removal and nutrient scrubbing. The placement of water and sewer (force)
mains within the swale areas along the streets will effectively penetrate
any "hard-pan" soil layer that might exist, thereby promoting additional
infiltration and percolation.
Lake areas will be constructed with gentle side slopes (5:1) extending
to a minimum of 3 feet below storage levels (see Figure I) to pre-
vent bank sloughing and to increase littoral zone so as to enhance nutrient
removal efficiencies. All lake areas shall be deeper than 6 feet as mea-
sured from storage level to prevent the establishment of cattails and other
270
-------
rooted aquatics in open water areas.
Procedures
The analysis of the drainage system was performed using the Soil
Conservation Service's (SCS) computer program TR-20. The TR-20 program
is an event model capable of generating basin runoff hydrographs, adding
hydrographs and routing hydrographs through reservoirs and reaches. ATI
computer work was performed on a CHI 2020 with graphic output generated
on a Calcomp 633 Roll Plotter.
The TR-20 employs the unit hydrograph method to generate design
stormwater runoff characteristics. Output from the program includes time
and quantity at peak discharge and volume of runoff on the basin.
Input data requirements include the designation of both a dimension-
less hydrograph representative of the study area, and a dimensionless
rainfall distribution curve. Due to the characteristics of the property
(gentle topographic relief, large wetland slough system) a dimensionless
hydrograph with a peak rate factor of 325 was selected.
The selected rainfall pattern is described by the SCS as being a
"B" type distribution appropriate for use with design storms of 24 hour
duration. Additional input requirements include the designation of the
design storm, antecedent soil moisture condition, time of concentration of
the basin, area of basin, and composite Complex Number (CN) value. Design
storm event quantities were taken from Technical Paper-40 (TP-40).
Antecedent soil moisture was assumed to be
2 (average) under design conditions but was raised to 3 (wet) when the
system was loaded with a 2-year - 24 hour event at a time when all system
storage was depleted. The time of concentration, i.e., time it takes for
a drop of water furthest away from the basin outfall to reach the outfall,
271
-------
was computed using assumed velocities for each type of travel segment.
The area of each basin was determined by planimeter. The composite basin
CN value was computed by proportionately weighing the CN values assigned
to the various basin land uses or was taken from TR-55. Soil mapping of
the property was performed by the Soil Conservation Service,
Soil hydrologic
ratings were also taken from Technical Paper-55 (TR-55) and for the de-
veloped condition the well drained rating for each soil type was employed.
The TR-20 program will route inflow hydrographs through a reservoir
if a stage-storage-discharge relationship for the reservoir can be de-
scribed. Output from the procedure includes a stage-discharge (vs. time)
hydrograph and the designation of the time and quantity at peak discharge
as well as the volume of runoff on the basin.
Discharge data from the TR-20 reservoir routing procedure can be in-
corporated into the HYDRA program, developed by the author, which generates
a graphic output on the Calcomp Plotter.
The author has also developed a computer program named TR-20D, which,
when given basin and drainage control structure characteristics, will de-
scribe the stage-discharge-stage relationship for the area in a format
suitable for introduction into the TR-20. In general, only lake storage
was considered in the design condition. However, a storage credit of 3-
inches was applied to each contiguous drainage retention area (DRA) greater
than 50 acres in size, and 1-inch of runoff storage was assigned to all
commercial, industrial and multi-family areas, (to be enforced by deed
restriction)
An iterative approach employing the TR-20 program, known as the
272
-------
Conservative Weir Analysis (CWA), was developed by the author for use in
the design of the subject drainage works. The benefits enjoyed by employ-
ing the CWA approach include:
1. Compatibility with TR-20 which generates discharge hydrographs and
elevation profiles.
2. Conservative solutions since it assumes free flow hydraulic conditions
(over weirs and through pipes) but is designed to allow headwater,
tailwater equalizations (with pipes) yielding lower actual flows.
3. Conservative solution since the peak elevation resulting from the 25
year - 24 hour design storm will be well below the lowest street in
each watershed, and for less frequent or multi-event storms one can
allow an additional rise in head (above design highwater) of up to
6" over the storage area (where flow over the weir is increasing at
IV ' ') before the upstream weir becomes submerged.
4. Simple and accurate discharge relationship can be defined for a weir.
5. Provides for maximum storage and detention yielding water at a lower
rate of flow and higher quality:
a. Promotes settlement of particulate matter.
b. Increases recharge capacity of the system.
c. Provides for nutrient absorption.
d. Prevents "shock" loadings to receiving water bodies.
6. Computer solution allows for the evaluationof numerous alternatives
quickly in order to achieve a "best" overall design at a relatively
inexpensive cost to user.
The CWA approach requires that each major drainage control structure
be a weir with minimum pipe(s) (18-inch) placed through it. Major
273
-------
conveyance structures within each basin were assumed to be sized so as to
prevent significant increases (greater than 0.2') in water surface profile
through them. Secondary drainage works are assumed to be sized so as to
prevent local flooding for the 10 year - 24 hour event.
The CWA approach requires that each structure be capable of conveying
the maximum flow for the design storm over the weir alone (no pipe dis-
charge contributions) and at no time shall water elevations exceed design
high water. Simply stated the TR-20 analysis is run twice for each struc-
ture, once with no pipe in the weir to assure that water elevation will not
rise above design level, and again with a pipe to compute total possible
flow to downstream structure.
The steps involved in the CWA design approach are as follows:
1. Select pipe invert elevation, weir elevation and length, and design
high water, specify basin storage characteristics and input into TR-
20D. Output from TR-200 includes stage-discharge-storage relation-
ships for the structure with and without pipe(s).
2. Insert structure data (without pipe) generated in step 1 into TR-20.
3. Using TR-20 generate design inflow runoff hydrograph and route through
•j-
structure described in step 2.
4. Evaluate results from step 3 and determine whether peak elevation
exceeds design high water. If peak elevation is greater than design
high water go back to step 1 and redesign. Likewise, if peak elevation
is significantly lower than design consider starting again at step 1.
If acceptable, continue.
5. Insert, structure data (with pipe) generated in step 1 into TR-20.
6. Go to step 1 and design downstream structure, then repeat procedures
274
-------
2 through 6 until all downstream structures have been designed.
Once the design of each basin is completed the evaluation of other
storm events is easily accomplished by use of TR-20. In this study storms
of 50, 10, 5 and 2 year frequency and 24 hour durations were also inves-
tigated assuming design storage was available and soil moisture conditions
were average (2). In addition, a 2 year - 24 hour storm was applied to
the system for a wet soil condition (3) with no storage available. It is
also of interest to note that we have investigated the 25 year - 5 day
storm (14.5 inches distributed on a "B"-type SCS rain curve) and have found
that peak discharges fall within, and typically 10% below those resulting
from the 25 year - 24 hour event, thereby supporting the selection of
critical design storm duration.
Basin Description
In its developed condition, Tampa Palms will be composed of 6 drain-
age basins as illustrated on Figure 2. All areas of the property will
maintain their historic hydrologic relationship and patterns of flow
except for those areas north of SR-581, east of the proposed Cypress Creek
Levee which have historically drained to Cypress Creek. However, steps
have been taken to prevent hydrologic impacts to Cypress Creek by pro-
posing the incorporation of pipe structures through the levee at several
historic outfall points. These structures will allow the maintained up-
land flows to Cypress Creek during low stage period of the Creek. In
addition, the maintenance of relative historic ground water levels (by
staging storage areas) within the areas adjacent to Cypress Creek will
allow for continuance of the ground water contribution from the property.
Two of the developed drainage basins (HB-1 and CC-2) require no major
275
-------
Figure 2. Master Drainage Plan
ro
•-a
o>
Scale:
-------
drainage works or internal retention areas as a result of their limited
breadth and setting in the project. The four remaining basins ranged in
size from a minimum of 28-0 acres for CC-1 to a maximum of 1660.4 acres
for the West Branch drainage basin (WB-1 through WB-7). In the interest
of space only results for the West Branch basin are reported herein.
Only limited areas of property not owned by Deltona contribute to
the proposed drainage system. It is of importance to note that the land
uses of the offsite property are well defined and not subject to significant
changes, which might impact the design of the drainage system. Offsite
areas contributing to the drainage system are classified as: roads (SR 581,
1-75 extension); power easements (TECO and FPC); utility sites (City of
Tampa Water Plant); and open space adjacent to S.W.F.W.M.D. Levee-112 N.
Lands north of the Deltona property boundary may have historically
drained through portions of Deltoha property to Cypress Creek. Drainage
from that small area of offsite lands that could be deprived of their
natural outfall as a result of Cypress Creek Levee construction, will be
conveyed (via shallow swales on Deltona property) to low lying wetland
areas within Deltona property outside of the levee. This will prevent
unnatural flooding of the offsite property and relieve Deltona's
"controlled" (development restrictions and operation) drainage system
from unquantifiable externalities that would otherwise be imposed upon
it.
Results for the West Branch Basin
The developed West Branch drainage basin contains 1660.4 acres
and is composed of 7 sub-basins, the composite areas of which are in-
cluded in its existing undeveloped drainage basin. A breakdown of the
basin land uses and a description of the control structures and design
277
-------
storm results are included on Figure 3. Historically these areas drained
to Trout Creek; however, with the implementation of the Corps of Engineers
Lower Hillsborough Flood Detention Area project this natural outfall will
be eliminated (by Levee 112N construction) and future flows will be directed
to the Hillsborough River via a conveyance ditch. It will be necessary for
Deltona to provide for additional outfall capacity for this basin since the
design of the present Levee conveyance ditch will only be capable of hand-
ling a portion of the projected discharge from the design event. Graphic
results of the analysis of the design storm at the outfall structures (25
years - 24 hour event) are included on Figure 4. This figure depicts:
estimated pre-development discharge vs. time; simulated post development
discharge vs. time assuming no control structures (inflow); design post
development discharge vs. time with control structures (out flow); and
cumulative percent discharge vs. time for the post-development design storm
with control structures. Design development control discharges are con-
sidered to be within reasonable accord when compared with pre-developed
conditions in regard to both total quantities and peak flows. Any discre-
pancies can be explained as resulting from reinforcing conservative analysis
approaches for pre and post-development studies.
Review of the curve illustrating the percentage of cumulative dis-
charge reveals that only.26.8% of the total discharge quantity is released
after the first 20 hours of the storm event. In the same regard, 52.0%,
81.4% and 91.5% of the total discharge volumes are released after 25, 35,
and 45 hours, respectively, into the storm event. These values indicate
desirable retention capabilities which will provide for increased storm
water quality enhancement and recharge potential.
278
-------
Figure 3. West Branch Drainage Characteristics
WEST BRANCH
HIM. STREET ELEV. ; 37.8
WETLANDS CONTROL : 36.«
OESION HIGH WATER : 3S.3O
CONTROL WATER LEVEL : 38.0
25 WEIR
STORAGE 11.6 AC./FT,
WB-1
DItlON PEAK
DISCHARGE t JO CFS
MIN. STREET ELBV. l S«.8
WETLANDS CONTROL I 38.•
DESION HIOH WATCH I 34.»0
CONTROL WATER LEVEL ! 34.OO
1 - 1» f PIPE
INVERT AT 34.00
40 WEIR
STORAGE 1S.O AC./FT.
OESION PEAK
• ••CHANGE l 1O8 CFi
• 1 - IS"* PIPE
INVERT AT 31.«
WB-2
MIN. STREET ELEV. i 3S.S
WETLANDS CONTROL l 37.S
BASIN LAND USE
USE
SINGLE FAMILY
RESIDENTIAL
MULTI FAMILY
RESIDENTIAL
PARKS
DRA
LAKES
GOLF
SHOPPING
MISC. DEVEL.
OTHER
TOTAL
«m
354.4
240
34
287.6
1Q.4
77.
83.6
1»1.6
283.6
1860.4
PERCENT
COVERAGE
22
IS
2
18
6
4
4
12
17
100
ro
^j
vo
PESIGN HIGH.WATER :3T.O»
CONTROL WATER LEVEL : 3S.OC
30' WEIR
STORAGE : 17.0 AC./PT.
OESION PEAK
DISCHARGE : SS CPS
«1 - 18"f PIPE
INVERT AT ss.s
MIH. STREET ELEV. : 3s.s
WETLANDS CONTROL l34.8
DESIGN HIOH WATER i 33.OS
CONTROL WATER LEVEL l sT.8
WB—3
MIN. STREET BLEV. iSS.O
WETLANDS CONTROL l 38.0
DESIGN HIOH WATER : 34.00
CONTROL WATER LEVEL : S4.C
SO WEIR
S.uRAOE i «I.S AC./PT.
DESIGN PEAR
DISCHARGE i 300 CPS
r3 - IS"* PIPES
INVERT At 30.0
WB-5
WB-6
NULL
STRUCTURE
40 WEIR
STORAGE : SO.4 AC./FT.
DESION PEAK
DISCHARGE ! 110 CFS
-1 - IS"* PIPE
INVERT AT 31.8
MIN. STREET ELEV. : 38.6
WETLANDS CONTROL : 31.S
DESION HIGH WATER : 30.BO
CONTROL WATER LEVEL ! 36.0
WB-4
60 WEIR
STORAGE : 31.3 AC./FT.
DESION.PEAK.
DISCHARGE ! 4S8 CFS
0
SWFYVMD
RIM DITCH
Drainage Basin
Description
WB-7 \
3 - Wff PIPES \
INVERT AT 2S.S A
PS ALTERNATE
OITTFALL
-------
Figure 4
West Branch Discharge Hydrograph
RAINFALL DISTRIBUTION
ABSTRACTION RUNOFF
DISCHARGE HYDROGRAPHS
CONTROL STRUCTURE WB-7
ro
00
O
SYSTEM
STORAdE
DESIQN CUMULATIVE PERCENT
DISCHARGE HVDROQRAPH
UNCONTROLLED SYSTEM
DISCHARGE HYDROOKAPH
DESIGN CONTROLLED
SYSTEM DISCHAROE HYDROORAPH
ESTIMATED UNDEVELOPED
DISCHAROE HVDROORAPH
1O.OO 20.00 30.OO 40. OO
50.0O
60.00
TIME
70.00
HOURS
80.OO
90.OO 1OO.OO
110.00
-------
Results of hydraulic analysis for each sub-basin for various storm
events are presented on Table 1. It is of interest to note that even under
severely adverse drainage conditions (storage completely depleted, soil
moisture condition = 3) the introduction of a 2 year 24 hour storm (5.5
inches) does not cause increases in discharges of greater than 10% for
selected basins and in no cases would these increased levels impose adverse
impacts to the receiving areas, development or drainage network. With a
total 203.1 acre-feet of storage available in design conditions, this basin
can retain (conservatively estimated) 1.47 inches of runoff, 3.70 inches
of rainfall), before ultimate discharge over the outfall weir.
In conjunction with additional DRA storage, heretofore not considered
in design, the system will experience reductions in peak flows under design
conditions with an improved ability to manage storm water qualities. It
should also be noted that the practices of draining 60% of the lots abut-
ting open space towards the open space, and requiring on-site retention
for the first inch of runoff on all multi-family, commercial and industrial
areas will further reduce the direct hydraulic effectiveness of impervious
areas thereby reducing flows and pollutant loads.
Quantitative estimates have been made of the time it would take to
lower storage levels within the system from the weir control levels to
the normal lake storage levels (design storage level). Excluding evapo-
transpiration and seepage losses, levels are. expected to return to normal
within 30 days. Sub-basin WB-5 is the governing area in this regard in
comparison to up-gradient basins which will return to normal within 10
to 14 days.
281
-------
Table 1
West Branch Drainage Analysis Summary
Storm Event
Structure Amount of Rain
Soil Condition
25 »r.-5 day
14.5"
25 »r.-24 Mr.
9.5"
(2)
WEST BRANCH
10 »r.-24 Hr.
8.5"
5 »r.-24 Mr.
7.5"
(2)
2 Yr.-24 Mr.
5.5*
2 »r.-24 Mr.
5.5"
lakes Full (3)
WB-I Pk. Dlsch. (Tim) 20.20(53.55) 18.97(18.06) 12.54(21.40) 8.78(23.47) 5.42(24.39) 20.35(13.41)
Pt. Elev. 36.22 36.20 36.05 35.84 35.32 36.43
In. On Basin 10.40 5.39 4.99 4.15 2.52 3.S6
WB-2 Pk. Dlsch. (T1*e) 101.20 (53.36) 103.67 (17.30) 84.87 (17.00) 67.81 (17.64) 36.09 (21.05) 82.60 (15.14)
Pk. Elev. 34.86 34.88 34.75 34.64 34.40 34.81
In. On Basin 10.16 5.56 4.69 3.85 2.24 3.06
N8-3
Pk. 01sch. (Tin
Pk. Elev.
In. On Basin
>) 55.29 (49.46)
36.72
11.44
85.10 (12.58)
37.04
6-66
71.82 (12.85)
36.90
5.72
59.09 (13.13)
36.76
4.83
30.73 (14.5)
36.44
3.03
62.48 (11.58)
36.82
4.43
WB-4 Pk. Dlsch. (Time) 94.19(50.63) 108.24(15.31) 79.48(16.36) 60.19(17.42) 22.13(23.36) 92.34(12.80)
Pk. Elev. 34.84 34.91 34.74 34.56 34.21 34.87
In. On Basin 10.90 6.13 5.22 4.32 2.54 4.40
WB-5 Pk. Olsch. (Tine) 320.30 (55.03) 298.58(21.26) 238.55(22.42) 182.73(24.20) 75.06(29.72) 237.13(17.49)
Pk. Elev. 33.14 33.06 32.83 32.62 32.10 32.87
In. On Basin 9.64 4.93 4.05 3.18 1.53 4.00
VB-6
Null Structure
W8-7 PK. Olsch. (Time) 501.04 (54.97)
Pk. Elev. 31.00
In. On Basin 3.82 (67.71)
463.60 (21.24) 372.58 (22.37) 286.33 (24.13) 116.97 (29.32) 365.61 (17.69)
29.88 29.59 29.31 28.76 29.55
5.15 (54.2W) 4.27 (S0.24J) 3.39 (45.2J) 1.73 (31.45*) 4.02 (73.10*)
Notes:
Pk. Dlsch.• Peak discharge, values In cubic feet per second
Tine* Time of peak discharge from beginning of rain event
Pk. Elev.3 Peak elevation, values 1n feet« NGVD
In. On Basin* Amount of runoff on basin In Inches.
282
-------
LITERATURE CITED
Courtney, C.M. 1979. Wildlife of the Tan-pa Palms Site, Appendix
A1CC, Tampa Pains A.D.A., Deltona Corporaiton, Miami, Florida.
DuEois, S. and C.M. Courtney. 1979. Vegetation Analysis of the
Tampa Pains Site, Appendix A18A, Tampa Palms, A.D.A., Deltona Corporation,
Miami, Florida.
Hatcher, J. and C.M. Courtney. 1979. Water Quality Analysis -
Tampa Palms, Appendix A15B, Tampa Palms, A.D.A., Deltona Corporation,
Miami, Florida.
Hillsborough County Planning Department. 1978. Horizon 2000,
Tampa, Florida.
Kent, K.H. 1973. A Method for Estimating Volume and Rate of Runoff
in Small Watersheds. U.S. Department of Agriculture, Soil Conservation
Service Technical Paper -' 149.
Metchkavitz, R.J. 1979. A Compendium and Analysis of the Ground
and Surface Water Resources of Tampa Palms. Appendix A15A, Tampa Palms,
A.D.A., Deltona Corporation, Miami, Florida.
Motchkavitz, R.3. 1979. Cypress Creek Flooc Study, Appendix A17A,
Tampa Pains, A.D.A., Deltona Corporation, Miami, Florida.
Murphy, W.R., Jr. 1978. Flood Profiles for Cypress Creek, West
Central Florida. U.S. Geological Survey, Tallahassee, Florida.
National Oceanic and Atmospheric Administration. Monthly Clima-
tological Data and Annual Summaries. National Climatic Center, Asheville,
North Carolina.
Odum, E.P. 1966. Ecology, Holt, Rinehart and Winston, New York,
Hew York.
283
-------
Srrolenyak, K. 1979. Water Quality Modeling of the Tampa Palms
Property. Appendix A22C, Tampa Palms, A.D.A., Deltona Corporation,
Miami, Florida.
South Florida Water Management District. 1978. Management and
Storage of Surface Waters, Permitting Information Manual, Volume IV,
West Palm Beach, Florida.
Tourbier, J. and R.W. Dierscn, Jr. 1976. Biological Control of
Water Pollution, University of Pennsylvania Press, Philadelphia, Pa.
U.S. Army Corps of Engineers. 1974. General and Detail Design
Memorandum - Lower Hillsborough Flood Detention Area, Jacksonville, Flori
U.S. Department of Agriculture. Rainfall Frequency/Duration Kaps,
Soil Conservation Service Technical Paper - 40.
U.S. Department of Agriculture. 1960. Rainfall-Runoff Tables for
Selected Runoff Curve (lumbers, Soil Conservation Service Technical Releasi
No. 16, Washington, D.C.
U.S. Department of Agriculture.^ 1969. Computer Program for Project
Formulation - Hydrology, Soil Conservation Service, Technical Release No.!
Fort Worth, Texas.
U.S. Department of Agriculture. 1975. Urban Hydrology for Small
Watersheds, Soil Conservation Service Technical Release No. 55.
U.S, Department of Agriculture, Soil Conservation Service. 1958.
Soil Survey, Hillsbcrough County, Florida. Florida AgricuturaV Experiment
Station, Gainesville, Florida.
U.S. Department of Agriculture, Soil Conservation Service. 1969.
Supolerr.ent to the Soil Survey, Hi 11 sborough County, Florida. Florida
Agricultural Experiment Station, Gainsville, Florida.
284
-------
U.S. Gteclogicla Survey. Water Resources Data for Florida. Published
annually, Tallahassee, Florida.
Whipple, VI., Jr. 1375. Urban Run-off - Quantity and Quality,
American Society-of Civil Engineers, New York, Mew York.
285
-------
WATER QUALITY IN THE FIRST SECOND AND
THIRD ORDER STREAMS OF AN UPLAND AND
FORESTED WETLAND WATERSHED
by
Charles M. Courtney1
INTRODUCTION
As part of a multidisciplinary approach to the private de-
velopment of the 2200 hectare Tampa Palms, Florida, land
use and the ambient background water quality of adjacent
streams are described in this paper. The Tampa Palms site
(Figures 1 and 2) is located between the confluence of two,
interim'ttant, second order streams and the Hillsborough
River, which is a potable water source for the city of
Tampa. As is typical of central west coastal Florida there
is gentle relief to the terrain which slopes from 44.0 feet
to 24.5 feet from north to south across the property. Du-
Bois and Courtney (1979) have described land use on the site
according to the Florida Land Use and Cover Classification
System (Anon., 1976). This system not only divides the land
among several broad categories (i.e. agricultural, rangeland,
forested uplands, and wetlands) but it also subdivides these
categories based on specific uses or dominant species asso-
1 Director, Applied Environmental Services,
990 North Barfield Drive, Marco Island, FL 33937
286
-------
Figure 1. A Map Showing the Location of the Tampa Palms Study Site
ro
oo
Scale:
-------
Figure 2
ro
CD
CO
E j-y uj-t; e.
A Land Use Description* for the Tampa Palms Development Site
1«7« I*KB U*C CLA9StFtCATION« BASED
ON FIELD SUHVCY AND OMOUHD TBOTHEO
DuBois and Courtney (1979)
-------
clations (e.g. pine flatwoods, xeric oak hardwoods and cy-
press forested wetlands). The forested uplands of the site
(Figure 2) were shown to comprise 44.6% of the land space,
and they were dominated by Quercus laurifolia with other
hardwoods (22.2%); (slash) pine flatwoods (18.4%); some
mixed forest (3.8%) where live oak and slash pine were in-
termixed; xeric oak (Quercus geminata) (.1%); and clearcut
areas (.1%). Additional uplands were in use as rangeland
(15.1% of land space) and these were dominated by palmetto
prairies (11.2%) with minor components (2.6%) of other
shrub or brush species and grasslands (1.3%). Little, of
the land was in urban use (4.4%) and that consisted of a
powerline right of way (2.6%) and. the space occupied by
State Road (SR) 581 (1.8%). The wetlands that drain these
uplands occupied another 35.5% of the land space. A mix-
ture of Fraxinus, Quercus, Acer and Taxodium dominated
communities made up 17.3% of the wetlands; cypress domi-
nated forests made up 15.5%; while freshwater marsh made up
2.3% and freshwater swamps only made up 0.4%.
Motchkavitz (1979) defined six major internal drainage ba-
sins which discharged from the property at well defined
points (Figure 3) with the remainder of the property out-
side of these areas draining overland to either Cypress
Creek or to the Hillsborough River.
289
-------
Figure 3
A Map Showing the Location of the Six Pre-Development
Basins at Tampa Palms
2000 4000
Motchkavitz (1979)
290
-------
The water quality of Cypress Creek was described by the
Southwest Florida Water Management District (Anon., 1975)
and Briley Wild Associates, Inc. (Anon., 1978). This se-
cond order stream and another, Trout Creek, flow south from
Pasco County, Florida to the Hillsborough River but resem-
ble riverine swamps adjacent to the property. The property
is separated from Trout Creek by a dike that SWFWMD is using
to retain Hillsborough River floodwaters and thereby pro-
tect the city of Tampa.
The Hillsborough River is a first order stream which ori-
ginates in central Florida and flows westward through rich
phosphate deposits (Jones et al., 1973) to Hillsborough Bay.
291
-------
METHODS AND MATERIALS
Ambient receiving water chemical and microbiological data
were collected at regular monthly intervals at five sta-
tions (Figure 4) over the period October 1978 to December
1979.
Station 1 was located at the intersection of Trout Creek
and SR 581 and described the water quality of Trout Creek
upstream of the property while station 4 was located at
the mouth of the West Branch of Trout Creek (an intermit-
tant tertiary stream) where it discharged from the pro-
perty. Station 2 was in Cypress Creek at the northwest
corner of the site and represented the water quality of
this creek as it entered the property while station 5 at
the intersection of the creek and SR 581 was representative
of water quality in the creek as it exited the property.
Station 3 was located in the Hillsborough River upstream
of the property. Monthly collections were also made at
station 6 in the Hillsborough River over the period Febru-
ary - December 1979 to represent water in the River down-
stream of the property.
Since the discharge points for the internal drainage basins
of the property were well defined, a second series of sta-
tions were designated to monitor the quality of runoff from
selected storm events. Each storm event sample series con-
292
-------
Figure 4. A Map Showing the Water Quality Monitoring Sites
(X)
-------
sisted of samples which were collected symetrically on the
discharge hydrograph. Storm runoff was differentiated by
a period of no flow at the station for seven days prior to
the storm event and the occurrence of no rainfall on the
property during that period.
Water quality at station 4 represented storm runoff from
the West Branch basin which was the largest internal basin
on the property (609 hectares). This basin contains 41%
forested uplands, 32% wetlands and 27% rangelands and con-
tributes ^ 10% of Trout Creek's overall flow (Motchkavitz,
1979). Station 9 water quality represents storm runoff
from the (195 hectare) Taylor Slough basin which is evenly
divided between rangeland (39%), wetland (33.7%) and for-
ested uplands (27.3%). Water quality at station 10 repre-
sents storm runoff from the Outfall Slough basin which con-
sists of 56.8% wetlands, 33.1% rangeland, and 10.1% for-
ested uplands. The North Finger Slough Basin is the small-
est basin on the property (42 hectares) and is dominated by
forested uplands (63.6%) with small but equal portions of
rangeland and wetlands (18.6% and 17.8%, respectively).
Water samples from station 12 represent storm runoff from
this basin. The Outflow Slough basin in the northwest cor-
ner of the property is dominated by wetlands (45.9%), with
34.4% forested uplands and 19.7% rangeland. Station 15 wa-
ter collections represent storm runoff from this basin, but
it should be noted that the boundaries of this basin are
294
-------
questionable because some flow has been observed entering
its northwest corner from the Inflow Slough Basin. Because
some of the regular monitoring stations in second order
streams (e = g. Cypress Creek) were also intermittant some
storm event sampling also took place at these locations
during runoff pulses.
For regular monthly monitoring and storm event stations j_n_
situ measurements of temperature, conductivity and dissolved
oxygen were made using a Hydrolab System 2000 while pH was
determined in the field using an Orion Model 401 Specific
Ion meter. Turbidity was determined by the Nephelometric
Method (APHA, 1976) on a Hach Model 1860A laboratory turbid-
imeter. Total filtrable residue, nonfiltrable residue, BOD5,
Color, ammonia-nitrogen, total phosphate, chlorophyll a_,
phaeophytin j§_, and total, fecal and fecal streptococcal bac-
teria were determined by the methods of APHA (1976). Ni-
trate-nitrogen, nitrite-nitrogen, and total organic carbon
were measured by the methods of EPA (1974) while total kjel-
dahl nitrogen was analyzed by the soluable organic nitrogen
digestion of Strickland and Parsons (1972). Data from back-
ground and storm event sampling were combined for each sta-
tion for between station tests of significant difference for
the respective parameters using the BMDP 3D program of Dixon
and Brown (1979).
295
-------
RESULTS
A summary of the chemical and physical characteristics of
background and storm event monitoring at Tampa Palms is
shown in Table 1. Four storm events were analyzed and they
included 1.2, 1.4, 2.8 and 10.5 inch events. Total fecal
and fecal strep counts were always highest in Third Order
streams and lowest in Second Order streams. Counts in the
Hillsborough River averaged 33% lower at the upstream sta-
tion 3. Combined nitrate and nitrite nitrogen levels de-
creased from first to second to third order streams. Am-
monia nitrogen was not measured in the tertiary streams,
but it was higher in second order streams than in the Hills-
borough River. Total kjeldahl nitrogen levels were highest
at downstream stations in first and second order streams and
lowest in third order streams. Total phosphate was highest
in the Hillsborough River as was reactive phosphate. High-
est average BODs levels were found in the third order streams
while TOC concentrations were highest in second order streams.
The water was also generally more colored in second order
streams than in the Hillsborough River.
296
-------
TABLE 1 A SUMMARY OF BACKGROUND AND STORM RUNOFF WATER QUALITY IN FIRST. SECOND AND THIRD ORDER STREAMS
ro
VO
COLIFORM («x104/100 ml)
10IAL'
FIRST ORScR STREAM-MONTHLY
STATIC;; NO.
3-HILLSSOROUGH n 9
RIVER 7 1.06
6-H:LLSBOROUGH n 8
RT.'ER 3 7 .11
6-h!LLSSOROUGH n
RIVER S 7
SECOND ORDER' STREAM-MONTHLY
STATION NO.
2-CVFRESS n 9
CREEK B 7 .06
2-cyr>-'tSS n
CRE-K S 7
S-OP'JESS n 9
CREEK 8 7 .08
5-CYPP-SS n
CREEK S 7
1 -TROUT CREEK n 9
7 .11
4-XEST BRANCH n 9
TROUT CREEK 7 .08
FECAL F. STREP
NO?N+* NH3N
NOiN
BACKGROUND SAMPLING <• STORM
10 10
.02 .02
8 ' B
.04 .35
16 16
.426 .024
11 11
.217 .025
BACKGROUND SAMPLING * STORM
9 9
.005 .02
9 9
.004 .02
9 9
.007 .03
10 10
.008 .05
16 IS
.118 .037
16 16
.046 .048
16 16
.123 .132
14 14
.022 .019
TKN
EVENT
17
.63
11
1.20
3
1.35
EVENT
16
.79
3
.77
16
.89
3
1.02
16
1.03
15
1.23
* TP04*
SAMPLING
17
.375
11
.383
3' •
.308
SAMPLING
16
.047
3
.019
16
.071
3
.020
16
.131
15
.028
OP04*
16
.358
VI
.369
16
.028
16
.032
16
.072
14
.016
BOD5*
13
1.3
11
1.0
1
.2
11
1.4
1
.7
13
1.3
1
.1
13
3.3
11
1.4
TOC*
12
12.6
8
14.1
3
IS
10
25.4
3
20.3
9
24.4
3
23
9
22.9
10
22.3
CHLa.**PHA£a"
10 10
8.6 2.3
10 10
3.0 3.7
10 10
2.6 2.1
10 10
5.4 2.8
9 10
5.3 3.6
8. 8
2.3 1.8
SOLIDS*
T.DIS. SUSP. COLOR
(Apt-Co)
14 14
182 107
10 11
J67 138
14 14
164 193
14 14
171 162
14
168
12 12
144 146
TURB.
(NTU)
14
2.5
'11
2.2
14.
1.7
13
2.7
14
8.7
12
1.9
pH TEKP.
CO
' 6 5
7.42 19.7
5 6
7.1 18.4
8 7
6.91 17.7
8 7
6.6 17.6
8 6
7.2 17.9
6 5
6.8 18.2
CO»D.
(uahos/cm)
4
266
S
232
6
192
6
240
4
140
4
100
THIRD OSDER STREAM-STOSK EVENT SAMPLING
STATION NO.
4-W-ST BRANCH n 8
TROUT CREEK 5f 2.29
9-TAVLOR n 6
SLOUGH T 7.58
10-C'JTFALL S
SLOUGH 2.42
12-N. FINGER 5
SLO'JSH 1.71
15-OL'TFLOW TO 5
POWERLIXE 7 1.68
• ms/l3
n number of samples
B Background
S Storm Event
8 8
.26 .75
6 6
.31 1.22
8 3
.16 .98
5 5
.17 .88
5 5
.10 1.45
10
.047
8
.094
10
.070
6
.035
7
.038
1-1
.50
9
.45
11
.43
7
.47
8
.24
1
.010
1
.040
10
.043
8
.033
11
.019
6
.197
7
.028
11
1.7
9
2.1
11
1.9
7
2.5
8
2.3
11
15.5
9
18.1
11
22.8
7
19.4
8
15.8
10
23.8
8
12.6
9
10.9
6
4.0
7
5.6
-------
DISCUSSION
The diluting effect of increased flow in lower order
streams was quite evident for certain parameters such as
the Coliforms, BOD5, TOC and Color. The bacterial para-
meters (total coliform, fecal coliform and fecal strep-
tococci) were used to indicate possible concentrations of
human pathogenic organisms. One problem with these indi-
cators is that they are present in the feces of all warm-
blooded animals to a certain extent. Recent studies (Do-
ran and Linn, 1979) have shown that when these bacterial
indicators are present in storm runoff from grazed .and un-
grazed pasture their levels can naturally exceed the cri-
teria of the Department of Environmental Regulation. Doran
and Linn (1979) also state that the fecal coliform to fecal
streptococcus ratios for humans is 4.3; for cattle and other
livestock and poultry it is .104 to .421 while for rabbits,
birds and mice it is 0.0008 to 0.043. Some mixed pollution
was indicated upstream in the Hillsborough River with the
majority of the other stations showing livestock and poul-
try bacterial sources. The lowest bacterial ratios occur-
red in storm runoff from basins where the wetland to range-
land ratio was approximately 2:1.
The denitrification of nitrate to atmospheric nitrogen oc-
curs in areas like the swamps and third order streams of
the property where there is little or no dissolved oxygen
298
-------
concentration, pH >5.5, and abundant sources of Carbon
(Dierberg and Brezonik, 1976) as is indicated by second
and third order stream TOC levels.
Hunt and Lee (1976) suggested that when there is shallow
sheet flow over organic substrates (such as exists in the
sloughs of the property) conditions would be favorable for
nitrification and denitrification to take place simultan-
eously. Ammonia is oxidized to nitrate in the aerobic wa-
ter column, and conditions are favorable for denitrifica-
tion to nitrogen gas a mere 5 mm below the surface of the
sediment. This might not only explain why nitrate and ni-
trite are progressively lower as one moves from first to
third order streams but also why ammonia and total kjeldahl
nitrogen are not higher. For these reasons and because of
the probable rapid utilization of nitrates by plants, ni-
trate was significantly lower at stations 4 and 5 than at '
station 3 on the Hillsborough River. There were no signi-
ficant differences for either TKN or NHs among any of the
baseline station pairs.
Unlike nitrogen, phosphate enters the water primarily from
the dissolution of geological deposits. Total phosphate in-
cludes phosphate that is available in the water for plant
use while the organically bound phosphate has already been
incorporated and requires bacterial decomposition to be re-
utilized. As mentioned earlier, the Hillsborough River flows
299
-------
through a region that contains rich deposits of phosphate.
For this reason and because of the heavy vegetative growth
bordering third and second order streams the River had sig-
nificantly higher concentrations of total phosphate than
stations 2, 4 and 5. Total organic carbon concentrations
were significantly higher at station 2 in Cypress Creek than
at station 3 on the Hillsborough River.
Riverine swamps have the ability to effectively filter out
suspended solids such as organic detritus. Larger litter-
fall, such as branches, twigs, etc. settle out when they be-
come water-logged or caught in other vegetation. This is
indicated in the storm event data which show lowest average
suspended solids concentrations coming from basins with the
lowest percentage of rangeland. The decomposition of these
materials places an additional demand on the dissolved oxy-
gen of the stream that is not measured by the BODs test.
For this reason diel measurements were performed at some
stations permitting an overall determination of the commun-
ity metabolism rates. This is not a precise method but it
does show the order of magnitude of the respiration and pro-
duction (Table 2). The BOD5 values are less than twenty
percent of any of the respiration rates and in most cases
much less. Odum and Hoskins (1958) stated that if the dis-
solved oxygen concentration remains mostly below 100% sat-
uration the community is heterotrophic, that is, it is re-
300
-------
TABLE 2. COMMUNITY METABOLISM VALUES COMPUTED BY THE DIURNAL CURVE METHOD.
CO
o
Stations 3/29-30/79
Respiration
gO/m3/day
1
2 2.4
3 53.
4 43.
5 72.
Production
gO/m3/day
.4
3.8
3.1
12.
6/25-26/79
Respiration Production
gO/m3/day gO/m3/day
14.
20.
4.6
11.
14.
7.3
1.4
5.5
<.l
5.6
-------
ceiving more organic matter than it is producing. The
Hillsborough River at station 3 was the only station that
ever had values above 100% saturation. The baseline sta-
tions in second and third order streams and at station 3
on the Hillsborough River were heterotrophic tending to-
ward dystrophy (Odum, 1971) because of the heavy litterfall
in those types of swamps (Deghi, 1975). For these reasons
the minimum dissolved oxygen concentrations were observed
on occasion to be below the state criteria of 4 mg/1 at
all stations.
The pH of second order streams was lower than at the Hills-
borough River stations. This difference is associated with
the fact that the tannins that were leached into this water
(cf. discussion of color) are weak organic acids which low-
er the pH in the absence of buffering carbonates.
302
-------
ACKNOWLEDGEMENTS
I want to thank the staff of the Applied Environmental
Services, particularly Dr. Douglas Armstrong, Mr. James
Hatcher, and Douglas Finan for their chemical and micro-
biological analytical support; Robert Motchkavitz, Shaw
Hamilton for assisting in data collection and processing;
and The Deltona Corporation for its financial support.
303
-------
LITERATURE CITED
Anon. 1975. Environmental Assessment for the Cypress
Creek flood detention and well field project. South-
west Florida Water Management District, Brooksville,
Florida pp. I to XII-3.
Anon. 1976. The Florida Land Use and Cover Classification
System: A technical report. Florida Department of Ad-
ministration, Tallahassee. 50 pp.
Anon. 1978. Briley Wild and Associates, Inc. Brooker
Creek Water Management Plan. Southwest Florida Water
Management District. Brooksville, Florida pp. 1-1 to
11-5.
APHA. 1976. Standard Methods for the Examination of Water
and Wastewater. Fourteenth edition. American Public
Health Association, Washington, D.C. 1193 pp.
Deghi, G. 1976. Litterfall and decomposition in four cy-
press domes. In Odum, H.T., K.C. Ewel, J.W. Ordway, M.K.
Johnston, editors. 1976. Cypress wetlands for water ma-
nagement, recycling and conservation. Third annual re-
port to National Science Foundation and the Rockefeller
Foundation Center for Wetlands, Univ. of Florida, Gaines-
ville, Florida, pp. 197-231.
Dierberg, F. , and P.L. Brezonik. 1976. Routine sampling. J_n
Odum, H.T., K.C. Ewel, J.W. Ordway, M.K. Johnston, editors.
1976. Cypress wetlands for water management, recycling
and conservation. Third annual report to National Science
Foundation and the Rockefeller Foundation Center for Wet-
lands, Univ. of Florida, Gainesville, Florida, pp. 337-371.
Dixon, W.J. and M.B. Brown. 1979. Biomedical Computer Programs
P-series. University of California Press. Berkley, Cali-
fornia. 880 pp.
Doran, J.W. and D.M. Linn. 1979. Bacteriological Quality of
Runoff Water from Pasture Land. Applied and Environmental
Microbiology. 37(5):985-991.
DuBois, S.J. and C.M. Courtney. 1979. A description of the
flora of the Tampa Palms development site. Appendix 18A to
the Tampa Palms Development of Regional Impact Application.
pp. 91.
EPA. 1974. Methods for chemical analysis of water and wastes,
1974. U.S. Environmental Protection Agency. Cincinnati,
Ohio. 298 pp.
304
-------
Hunt, P.G. and C.R. Lee. 1976. Land treatment of waste-
water by overland flow for improved water quality. In.
Tourbien, J. and R.W. Pierson, Jr. 1976. Biological
Control of Water Pollution. University of Pennsylvania
Press. Philadelphia, Pennsylvania, pp. 151-160.
Jones, J.I., R.E. Ring, M.O. Rinkel, and R.E. Smith, edi-
tors. 1973. A summary of knowledge of the eastern Gulf
of Mexico. State University System of Florida Institute
of Oceanography. St. Petersburg, Florida, pp. 1-1 to
VIII-74.
Motchkavitz, R.J. 1979. A compendium and analysis of the
ground water resources of Tampa Palms, Hillsborough
County, Florida. Appendix A15A to the Tampa Palms De-
velopment of Regional Impact Application.
Odum, E.P. 1971. Fundamentals of Ecology. W.B. Saunders
Company. Philadelphia, Pennsylvania.
Odum, H.T. and C.M. Hoskin. 1958. Comparative studies on
the metabolisms of marine waters. Publications Insti-
tute of Marine Science. 5:16-46.
Strickland, J.D.H. and T.R. Parsons. 1972. A practical
handbook of seawater analysis. Fisheries Research Board
of Canada. 167:310 pp.
305
-------
IMPLEMENTATION' OF STORM v.'ATER MAHAGEMENT MODELS;
PROS AND CONS OF STANDARDIZATION
Dr. Paul Wisner, Professor
Atef Kassera, M.F.rtg.
and
Philip Cheung, B.Eng,
University of Ottawa,
Dept. of Civil Engineering
INTRODUCTION
Replacement of traditional methods for the design of urban drainage
systems, such as the Rational Method, by modelling techniques has been
promoted in the last few years by several governmental agencies in both
Canada and the United States.
A review of the state of the art in urban drainage in Canada,
carried on under the direction of the first writer, indicated that 5 years
ago only one Canadian municipality -r<- Toronto — was using urban drainage
models (1). To our knowledge, at least seventeen municipalities have
in the meantime conducted one or several modelling studies. Host of
them, however, continue to use the Rational Method, or other "simplified"
techniques,
The review by Poertner of methods used for the design of storage
facilities (2) offers examples of a variety of "simplified"-' techniques
leading to triangular or trapezoidal hydrographs, or to mass curves
derived from the Rational Method. SCS has also promoted simplified
graphs giving peak flows for "flat" or "steep" slopes and tabular methods
for hydrographs and routing (3). In contrast, drainage manuals developed
in Canada for various municipalities (4,5), or for the design of airport
306
-------
drainage (6) usually recommend, for larger areas, specific urban
hydrologic models recently tested by U.S. EPA and the Canadian
Drainage sub-committee. Examples of models described in these manuals
are SWMM, ILLUDAS and unit hydrograph models initially developed for
rural hydrology studies, such as HYMO.
Several studies conducted with the Rational Method indicated that, for
the same area, peak flows determined by various engineers varied
because.of arbitrary selections of the inlet time or runoff coefficient
(1, 7, 8). One of the arguments for modelling was to eliminate the
effects of this subjective selection. If similar tests were conducted
today, discrepancies in peak flows and hydrographs would vary with
models, parameters, design storms, and could eventually be more
significant than in the era, of the Rational Method.
Decisions regarding the choice of the method are in most cases
left with the drainage engineers of the municipality or the consulting
firms who, in turn, are influenced by available funds, pressures to
cut down initial estimates, personal inclination, in-house capabilities
and other factors.
Decisions taken as a result of modelling activities may affect
significant investments. In the design of new storm sewers and relief
work they bear on the safety against basement flooding. Problems like
the effects of runoff increase may also result in litigation. Legal
responsibilities of modellers and consequences of differences between
various models should also be considered.
307
-------
Under these circumstances, control of the quality oC ••".od el ling
in urban drainage represents a serious challenge. Is the profession
ready for a large scale application of sophisticated methods? Are
these preferrable to the simple, traditional techniques which are
easy to understand and verify? Is it desirable to standardize the
techniques or should their selection be left entirely to those in
charge of a given project?
There are no general or clear cut answers to these questions.
Options may vary with the importance of the job or according to non-
technical aspects, such as the training of the staff in charge of the
verification of drainage projects.
It is however a reasonable objective that studies within the
same drainage •jurisdiction (municipality) should be based on the same
philosophy regarding modelling, and that drainage-manuals or local
regulations define the position of the jurisdiction with regard to
some of these difficult questions.
The aim of this paper is to present for discussion by the SWM
users group, some modelling aspects to be considered in the development
of drainage manuals or local regulations, and to present some suggestions
regarding a more uniform approach in modelling. The presentation is
limited to the quantitative aspects of runoff control, but some of the
principles may be extended to other areas of storm water management.
ONE OR SEVERAL MODELS?
It seems that regulatory agencies and drainage jurisdictions
have followed two conflicting schools of thought for the implementation
of modelling.
308
-------
The first one, considers that modelling is an art and should
not be "regulated". Some governmental organizations seem to have
itiade a conscientious effort to avoid sponsoring of a particular
model, in order to stimulate research. This avoids criticism from
model-builders or from consultants more conversant with a given
model.
The second school of thought, typical for some flood control or
erosion control agencies, is to require all those involved in a given
type of study to consider the same procedures in some cases with
as simple a method as possible.
The thesis of freedom of choice between "tested" models seemed
to be supported by the results of various studies commissioned for
the testing and comparison of models which did not result in clear
cut advantages for one given model (7, 9, 10). Comparison of
measurements on a number of small urban watersheds with simulated
flows shows that models such as SWMM and ILLUDAS give an acceptable
prediction within an error in the order of 20% (1).
These tests however were carried on by specialized hydrologists
with a very careful selection of model parameters. Model comparisons
have analyzed mainly real, frequent storms. Data on rare, very
intense storms, which are of interest for design purposes, are
practically not available. Typical differences between model compari-
sons, given in Figure 1(1), are not significant for the practitioner.
In routine design applications however, differences under design
conditions may be significant and, would more likely resemble those
in Figure 2, which shows a comparison of ILLUDAS, HYMO, and SWM models
309
-------
for conditions typical for runoff control studies in small watersheds.
Differences of chis nature are not considered acceptable. By reviewing
the parameters in terms of physical data, results in this example
could be brought closer. This however would require specialized
hydrologic work which is usually not carried on in routine projects.
The direction favouring free selection of models by the hydrologist
may be welcomed by scientists and specialized consultants. The argu-
ment for its implementation is that municipal engineers use specialized
assistance and that modelling should become a speciality in urban
drainage.
Real life, however, shows that this is not the case because most
organizations prefer to conduct studies with existing in-house staff
and will select models accordingly. Projects with sophisticated models
are difficult to verify, modify and improve once the Initial modeller
is no more available. The fact that economic advantages of sophisticated
modelling have not yet been clearly demonstrated, combined with the
high cost of training, have encouraged many jurisdictions to sponsor stan-
dardization of simpler techniques.
Review of these policies indicates on the other hand that this
approach does not only discourage the use of hydrologic expertise but
may also give decision makers limited, if not biased information.
It is therefore suggested that drainage jurisdictions should
start with an analysis of their objectives and required informa-
tion, and not with preselection of one or several models. The
next section will give several examples in this direction.
310
-------
SEU'CTIQN OF MODELS IN TERMS OF OBJECTIVES
Most SWM studies are based on surrogate objectives such as
"zero runoff increase at the outlet of a new development". For real
objectives, such as protection against flooding or erosion, it is impor-
tant to have information on flov? changes not only at the outlet but also
along the downstream channel. If this viewpoint is accepted, the
drainage jurisdiction has to recommend modelling on a watershed basis.
The next logical step is that selected models need proper routing routines.
Consideration of simultaneous operation of all storage facilities results
from the requirement of a watershed approach.
The use of "zero runoff increase" regulation also requires that the
agency clarify its position regarding the concept of "predevelcpment
flow". Drainage manuals or regulations do not consider the fact that:
predcvelopment and postdevelopment flows with the same frequency will
not occur for the same storm.
Another problem to be considered is that some urban models such
as SWMM have not been extensively tested for rural conditions. Other
models like IIYMO have not been tested for urban conditions. A
drainage jurisdiction in charge with runoff control may therefore
consider two different models, or preferably conduct in-depth studies
for the determination of px'e-development flows on a wide area
and recommend runoff rates for all those involved in development
projects.
Storm sewers are designed for free surface flow for a "design
frequency" which is also one of the current surrogate goals. According
311
-------
to the characteristics of Che system, a storm with a higher return
frequency may be conveyed with various surcharges. Protection against
basement flooding, which is the real goal, will vary within.the same
jurisdiction. A requirement closer to the real objective should
therefore be the limitation of surcharges. This would requirea.
model which can handle routing under such surcharged conditions and-
the choice of the drainage engineer would be narrox^ed down'to two
or three models. The City of Toronto, for example, has based the
design of its relief sewers on surcharge analysis determined by the HVM
Model. Most SWM studies are conducted for one event models with design
storms. Limitations of these procedures have been accepted as a. result
of cost and time constraints. A complete information on the response of
various facilities is given by continuous simulation. The choice was
made by several jurisdictions in Virginia, Maryland, and others who
conduct SWM studies on the basis of the HSP model.
An intermediate solution is the use of several real storms,
screened by means of the STORM model. This solution was adopted by
the Metropolitan Toronto Conservation Authority for the overall
hydrology of its large area (11).
Even if a jurisdiction works on the basis of single event - a
design storm for each return frequency, the limitations should be
determined by inlet studies. Consideration should be given at least
in some projects to the following aspects which are not included in
many drainage manuals or ordinances:
sequences of storm events
duration of "critical"storm event
underestimation of rural flows and overestimating
312
-------
storage by using storms which are critical for
urbanized conditions
requirements for simulation of .snowtnelt and frozen soil
conditions
effects of change in downstream timing of peak flows
A comprehensive analysis of resulting information against
cost of modelling was presented for several watersheds by the first
writer, for the Metropolitan Toronto Conservation Authority. Results
indicated that the traditional one-storm and one-unit hydrograph
model was estimated at approximately $60K, while an extensive
long term simulation would cost approximately $120K. Based also
on other considerations including time," available data for calibration
and available experience of the jurisdiction, a multi-event simulation
using STORM for screening of events with an estimated cost of |95K was
finally selected. The point of this comparison was that the increase
in cost for a more comprehensive analysis does not justify on a.
pfU-Qfi^UtiJ rejection. Even if jurisdictions may not have the additional
funds required for in-depth studies, a group of agencies could at
least conduct demonstration projects with various models. For most
large developers, an in-depth study may not only represent a negligible
additional expense, but could also result in savings by eliminating
facilities required by a simplistic approach required to meet the
surrogate goals. If regulations for "floodline determinations" and
"zero runoff increase" are simplistic or rather arbitrary, advanced
modelling becomes a fallacy. Promotion of better modelling should
therefore start with the improvement of the decision making process.
313
-------
RESPONSIBILITIES IN MODELLING STUDIES
Responsibilities in the selection and application of modelling
should be defined more specifically for the different groups partici-
pating in the storm water management planning and design processes.
The drainage manuals reviewed by the authors do not deal with this
delicate question.
The responsibilities of the jurisdiction in charge with the^
implementation of the results of storm water management studies is
obviously to define the required information from a modelling process
in order to select the best alternatives. As discussed in the previous
section, presently used surrogate objectives do not coincide with this
kind of information and may be even misleading. On the other hand,
their use is simple and verification of projects carried out on
this basis may become a routine operation.
The nature of the information required to meet the real objectives
may not be very explicit at the beginning of a SWM study. For these
reasons, a staged approach may help the jurisdiction in defining study
objectives. Preliminary SWM regulations of the Town of Markham have
been developed on this basis.
The first stage of a study has to define the main restrictions,
needs for runoff control, and will compare broad alternatives. In
this stage, the selection of models does not affect directly the invest-
ments and is left with the consultant. In the second stage, the choice
of a model is derived from the conclusions and type of alternatives
selected in Stage I. For example, if dual storage is required, a
sophisticated model such as SWM with the WRE TRANSPORT routine as
discussed in another paper (12 ) will be one of the best choices. For
a smaller' area without controls, the Rational Method will be sufficient.
314
-------
The responsibility of those in charge of the design of SWM facilities
is of course, to select a model which has the algorithm appropriate
for giving the required information.
Model selection is quite often guided by available facilities from
governmental agencies, computer companies, or specialized consultants.
The responsibility of those providing the software has not yet been
defined in SWM studies. In'structural engineering however, it is
considered that the developers of the software have the responsi-
bility of providing models which reproduce correctly the assumptions
in the algorithm (13). At the present time, those distributing or
marketing SWM software are not penalized in user problems resulting
from bugs, instability and perhaps incorrect assumptions. For some
proprietory models, detailed algorithms are not available. Drainage
manuals may require limitation in this regard. Those conducting"
modelling studies should also provide, for example, references for
testing and previous applications. The selected model ideally should
be calibrated for the same or similar conditions and have at least a
good documentation with a-presentation of its limitations.
IMMEDIATE STANDARDIZATION NEEDS
Once a model has been selected on the basis of the previous
considerations, a number of factors related to its usage should be
uniformized within a given jurisdiction. Some of the models have
several generations with various assumptions or with different
parameters. On the other hand, results from some models are very
sensitive to schematization of the watershed or selection of
characteristics of the system.
315
-------
These factors are not discussed in most of the existing drainage
manuals. As a result studies with "the same model" may lead to
different results. Examples of modelling; features which require
immediate attention are:
(a) For the HYMO model, the relation between the parameters K
and T . Several relations presently used in Ontario and
P
corresponding results are given in Figure 3.
(b) The RUNOFF model in SWMM is affected by selection of the "width"
parameter. On the other hand, there are two different schools
of thought with regard to the selection of the width indicated
in Figure 4.'
(c) The SWM TRANSPORT models have been used with various assump-
tions for the surcharge conditions. Some of them lead to
unrealistic storages at. the nodes with the result of underesti-
mating the flows with surcharged conditions. Recent versions
have reduced or eliminated these problems (14).
(d) For some models, if the level of discretization or lumping is
done arbitrarily, the effects may be significant (9) (Figure 5).
(e) The distribution of the design storm has recently been given
increased attention. What is not always considered is that
some models are oversensitive to the discretization of the
design storm, e.g. the RUNOFF model may lead to different flows
for different levels of design storm discretization (15) .
(f) The ILLUDAS model has now several routines for routing. Selection
of the appropriate routine has to be made in terms of the nature
of the problem.
316
-------
FtNAL REMARKS: THE IMPSM PROJECT
The list in the previous section is not exhaustive. Several
papers and reports have already dealt with the sensitivity analysis
and made suggestions regarding these elements (9, 16, 17). These
results must be organized, tested again on the same basis and
accepted by those who work for the same drainage jurisdiction.
While models will be continuously improved, it is suggested that
these improvements be incorporated in current-practice-in an
organized fashion, with testing and users consensus at each step.
Until test data become available, a cautious approach when
applying new or more sophisticated procedures is required. The
time has come to take advantage of collective past experience in
practical modelling, including mistakes, a step towards the
uniformization of modelling procedures has recently started at the
University of Ottawa. A group of governmental agencies, consul-
tants and municipalities cooperate in the IMPSWM program (which
stands for the "implementation of Storm Water Management Models")
of the University of Ottawa directed by the authors has started
after 6 months of discussions with various direct and indirect model
users. The philosophy is that, at present, standardization of the
factors discussed in the previous section and recommendations
regarding, simplified techniques may start by a consensus of users,
based on discussion of research results. It is also hoped that
IMPSVM participants.will use the same programmes updated on a
regular basis and will give the whole group information on good and
317
-------
bad experiences. IMPSWM will also organize training activities for
a broad audience and for users for discussion of specialized problems
with a small group of specialized participants. Most participants
are from Canada, but we have established a.cooperation with U.S.
specialists and already have received significant assistance from
Dr. Larry Roesner from CDM - Water Resources Engineers. Several
contacts have also been established in the U.K. and Switzerland.
This may be a modest step only but it is an old engineering
experience that once a problem is recognized, the solution is
very close. It seems from our discussions with consultants,
municipalities and various agencies that the present confusion in
modelling procedures is not perceived as a problem by some direct
or indirect users. The principal aim of this presentation is to
stimulate discussions on these aspects and to recommend the striking
of a balance between scientific requirements for continuous improvements
and implementation realities.
318
-------
REFEREKCtvS
1. Janes F. MacLaren Limited, Review of Canadian Design Practice
and Comparison of Urban Hydrologic Models, Research Reporc No.
26, Canada.Ontario Agreement on Great Lakes Water Quality, Env.
Canada, October, 1975.
2. Poertner, A.C., "Practices in the detention of urban storm water
runoff", Reporc for Office of Wacar Resources Research, American
Public Works Association, Special Report No. 43, Chicago, 1974.
3. Engineering Division, Soil Conservation Service, Urban Hydrology
for Small Watersheds, Technical Release Mo. 55, January, 1975.
4. James F. MacLaren Limited, Drainage Criteria Manual for the City
of Winnipeg, 1975*
5. M.M. Dillon Lioited, Drainage Criteria Manual for the City of
Burlington, 1977.
6. Proctor & Redfern Limited, Airport Storm Water Management,
Toronto, September, 1977.
7. Pagan, Alfred, The Use of,Che Rational Formula in New Jersey,
The New Jersey Professional Engineer, January, 1973.
8. Ardis, and altera, "Storm Drainage Practices in Thirty-Two Cities",
ASCE Journal of Hydraulics Division, 95, No. HY1, January, 1969.
9. James, F. MacLaren Limited and Proctor and Redfern Limited, Storm
Water Management Model Study, Vol. I, Research Report No. 47,
Environment Canada.
10. Brandstetter, A.B., "Assessment 6f Mathematical Models for Storm
and Combined Sewer Management", EPA, 1975.
11. Wisner, P., Belore, H., Bishop, R., Selection of Models for Flood
Control Studies, Canadian Hydrotechnical Conference, Edmonton, 1977.
12. ' Wisner, P., Kassem, A., Cheung, P., Application of the SVM Model for
the design of dual storage, SWfcM Users Group Meeting, Montreal,
May, 1979.
13 Flachsbart, Barry B., Managing Software Development, Journal of
the Technical Councils of ASCS, Vol. 105, No. TCI, April 1979,
pp. 51-56.
14. Wisner, P., Kassem, A., Cheung, P., IMPSWM Progress Report No. 1
Unpublished report, University of Ottawa, 1979.
319
-------
REFERENCES (continued)
15. Wisner, P., Kassem, A., Cheung, P., Comoarison of Design Peak
Flows Calculated with the Rational Method and the SWMM Model,
SWMM User's Group Meeting, Gainsville, Florida, January 10-11,1980,
16. Pauther, J.L., Delleur, J.W., Calibration and Sensitivity Analysis
of the Continuous Runoff Simulation Model STORM, Proceedings,
Intern. Symposium on Urban Storm Water Management, University of
Kentucky, July 1978.
17. Kibler, F. and Aron, G. Effects of Parameter Sensitivity and
Model Structure in Urban Runoff Simulation, Proceedings, Intern.
Symposium on Urban Water Management, U. of Kentucky, July 1978.
320
-------
to
10.0
20.0 300
TIME IN MINUTES
OCX)
60.O
— SWMW
" RRU
• —UCUR
- RECORDED
COMPARISON of HYDROGRAPH3
CALCULATED BY SV/MM, RRL
AND UCUR MODELS WITH
O\
RECOnDED HYDROGRARI
CALVIH PARK STORM of Au^iti 23, 1972
FJC, ±
-------
CO
rv>
ro
'P'-V. 1:103 W. 24-HI'S. 610tU
FIG. 2 - DIFFERENCES BETWEEN MODELS UNDER DESIGN CONDITIONS
-------
2 3
co
ro
-------
CO
ro
SOBCATCHMeWT,
A
tl) WIDTH » 2Lp
(2J WIPTM a A/2L
I Mi/*/ PZAMJAGs COUOUIl
THZ.GUOH so&cacHiAtnf - tf
FIGURE 4 - Selection of Width parameter in runoff model.
-------
20
INDIVIDUAL
SUBCATChMENTS
XIX
TRANSPORT BLOCK RESULTS
OUTLET HYOROGRAPH
AFTER CONDUIT ROUTING
CASE I
FINE
DISCRETIZATION
4 SU8CATCHMENTS
WIDTH I « 2000 FT.
M 2 = 1500 ••
» 3:2000 »
» 4 « 1500 «
TIME
(8)
OVERLAND PLOW HYDROGRAPH
FROM SINGLE SU3CATCHMENT
CASE 2
LUMPED
CATCHMENT
WITH WIDTH
SUMMATED
i-i n n
i i i 11
1 i i it
i t__j __j i
— 1 M i
'rt~--
OVERLAND
- WIDTH
ECUIVALENT
TRANSPORT
ELEMENT
WIDTH = 7000 FT.
TIME
(A)
FIGURE .5
325
-------
SWMM Users Group Meeting
January 10-11, 1980
Attendees
Maqbool Ahmad
City of Edmonton
12220 Stony Plain Road
Alberta, Canada
Mark P. Allen
Lozier Engineers
50 Chestnut Plaza
Rochester, NY 14604
William M. Alley
U.S. Geological Survey
National Ctr. MS 415
Reston, VA 22170
Douglas Ammon
U.S. EPA
Woodbridge Ave.
Edison, NJ 0881?
Jim Anderson
Urban Science Applications
1027 Fisher Bldg.
Detroit, Michigan 48202
Frank Baldesarra
Multiple Access
885 Don Mills
Don Mills, Ontario
Thomas 0. Barnwell, Jr.
U.S. EPA
College Station Road
Athens, GA 30605
Robert H. Baumgardner
Federal Hwy. Administration
400 7th St., S.W.
Washington, D.C. 20590
Robert Bishop
James F. Maclaren
1210 Sheppard Ave., E.
Toronto, Canada
John B. Blenk
Lozier Engineers
50 Chestnut Plaza
Rocherster, NY 14604
Hugh Blocksidge
Havens & Emerson, Inc.
1300 E 9th St 700 Bond Ct. Bldg.
Cleveland, Ohio 44144
Larry Bodnaruk
N.W. Hydraulic Consultants
4823-99 St.
Edmonton Canada
Hugo A. Bonuccelli
N. VA Plan. Dist. Comm.
7309 Arlington Blvd.
Falls Church, VA 22042
Gary R. Bowman
Anderson & Assoc.
100 Ardmore St.
Blacksburg, VA 24060
A. M. Candaras
Paul Theil Assoc.
700 Balmoral Rd.
Bramalea Ont. L6J 1X2
John Charowsky
Underwood McLellon
1479 Buffalo PI
Winnipeg, Canada
326
-------
Philip Cheung
University of Ottawa
Ottawa, Canada
Philip G. Clark
O'Brien & Gere Engrs., Inc.
648 Beacon St.
Boston, Mass. 02215
Richard DeGuida
Calocerinos & Spina
1020 N. 7th St.
Liverpool, NY 13088
Kevin J. Fay
Ecol Sciences, Inc.
735 N. Water Suite 715
Milwaukee, WI 53225
James E. Ferro
Central IA Local Govts.
104 E. Locust Street
Des Moines, IA 50309
W. F. Geiger
Techn. Univ. of Munich
Strasslacher Str. 2
8000 Munich 71, Germany
Bernand L. Golding
HNTB
600 Lake Ellenore Dr.
Orlando, FL 32809
James A. Hagarman
Calocerinos & Spina
1020 N. 7th St.
Liverpool, NY 13088
James P. Heaney
Environ. Eng. Sciences
422 A. P. Black Hall
University of Florida
Gainesville, Florida 32b1l
J. M. Hillman
AED
1127 Ford Ave.
Tarrant, Alabama 35217
Scott T. Hsieh
NW Ind. Reg. Planning Comm.
8149 Kennedy Ave.
Highland Lake, Ind. 46322
Wayne C. Huber
Environ. Eng. Sciences
420 A. P. Black Hall
University of Florida
Gainesville, Florida 32611
William James
MoMaster University
Civil Engr. Dept.
Hamilton, Canada
Atef M. Kassem
University of Ottawa
770 King Edward Ave.
Ottawa, Canada
Dennis F. Lai
Clinton Bogert Assoc.
2125 Center Ave.
Fort Lee, NJ 07024
David F. Lakatos
Satterthwaite Assoc., Inc.
11 N. Five Points Rd.
West Chester, PA 19380
Johan Larsson
Dept. of Hydraulics
Royal Inst. of Technology
KTH, 10044 Stockholm 70
Sweden
F. I. Lorant
M. M. Dillon LTD
50 Holly Str.
Toronto, Canada
Mario Marques
FHWA
Washington, D.C.
Stephen J. Nix
Environ. Eng. Sciences
418 A. P. Black Hall
University of Florida
Gainesville, Florida 32611
327
-------
Donald E. Overton
Univ. of Tennessee
Dept. of Civil Engr.
Knoxville, TN 37916
William M. Parker III
Harza Engineering Co.
150 S. Wacker Dr.
Chicago, ILL 60606
Gilles Patry
Ecole Polytechnique
P. 0. Box 6079 Station A
Montreal, Quebec H3C3A7
Larry Pond
W.L. Wardrop & Assoc. LTD
77 Main Street
Winnipeg, Canada R3C3H1
Mark Robinson
McMaster University
Main Street West
Hamilton, Canada L8S4L7
Larry A. Roesner
CDM/Water Resources Engrs.
8001 Forbes Place
Springfield, VA 22151
Simsek Sarikelle
University of Akron
Dept. of Civil Engr.
Akron, OH 44325
George Schlossnagle
USAF
234 S. Jan Drive
Panama City, FL 32401
James E. Scholl
CH2M Hill
P. 0. Box 1647
Gainesville, FL 32602
Howard M. Shapiro
Lozier Engineers
50 Chestnut Plaza
Rochester, NY 14604
Lorali Totten
Richard Browne Assoc.
50 Galesi Drive
Wayne, NJ 08876
Ray Tufgar
Proctor & Redfern
865 King Street, E.
Kitchner, Ont.
Bryan Weber
Underwood McLellon
1479 Buffalo PI.
Winnipeg, Canada R3T1L7
Richard Willet
Town of Blacksburg
300 S. Main Street
Blacksburg, VA 24060
Paul E. Wisner
Univ. of Ottawa
Civil Engr. Dept.
Ottawa, Canada
Rodney E. Zielke
URS/Forrest 4 Cotton
8700 Stemmons Fwy
Dallas, TX 75231
328
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
EPA-600/9-80-017
3. RECIPIENT'S ACCESSION NO.
LE
"LE
roceedings
itormwater Management Model (SWMM)
Users Group Meeting January 10-11, 1980
5. REPORT DATE
March 1980
6. PERFORMING ORGANIZATION CODE
AUTHOR(S)
Edited by Harry C. Torno
8. PERFORMING ORGANIZATION REPORT NO.
PERFORMING ORGANIZATION NAME AND ADDRESS
U.S. Environmental Protection Agency - Office of Research
and Development - Office of Environmental Processes and
Effects Research (RD-682)
Washington. D.C. 20460
10. PROGRAM ELEMENT NO.
B 106
11. CONTRACT/GRANT NO.
N/A
2. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency - Office of Research
and Development - Office of Environmental Processes and
Effects Research (RD-682)
Washington, D.C. 20460
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA 600/16
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report includes fifteen papers, on topics related to the development and
application of computer-based mathematical models for water quantity and quality manage-
ment, presented at the semi-annual meeting of the Joint U.S.-Canadian Stormwater Manage-
ment Model (SWMM) Users Group, held 10-11 January 1980 in Gainesville, Florida.
Topics covered include a description of two urban runoff models, an examination
of runoff quality algorithms in the SWMM, a discussion of improvements to the Extended
Transport (EXTRAN) portion of the SWMM, applications-of several urban drainage models
in planning, analysis and design, and a comparison of the Rational Method and the SWMM.
Also included are a paper on suggested methods for standardizing the process of acquir-
ing modeling services, and a paper on urban runoff quality data collection in the
Toronto, Ontario area.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSAT; Field/Group
Mathematical Models
Combined Sewers
Runoff
Hydrology
Urban Hydrologic Models
Urban Hydrology
Combined Sewer Overflows
Stormwater Runoff
13 B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURM
Unclassified
328
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
«U.S. GOVERNMENT PRINTING OFFICE: 1980 311-132/18 1-3
------- |