v>EPA
United Slates
Environmental Protection
Agency
Office of Air
Land and Water Use
Washington DC 20460
EPA 600/9-79-026
June 1979
Research anrj Development
Proceedings
Stormwater Management
Model (SWMM)
Users Group Meeting
May 24 - 25, 1979
Miscellaneous Reports Series
-------
DISCLAIMER
This report has been reviewed by the Office of Air, Land and Water Use,
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 Protec-
tion Agency, nor does mention of trade names or cortimerical products con-
stitute endorsement or recommendation for use.
-------
EPA-600/9-79-026
June 1979
PROCEEDINGS
STORMWATER MANAGEMENT MODEL (SWMM)
USERS GROUP MEETING
24-25 MAY 1979
Project Officer
Harry C. Torno
Office of Air, Land, and Water Use
Office of Research and Development (RD-682)
U.S. Environmental Protection Agency
Washington, D,C. 20460
OFFICE OF AIR, LAND, AND WATER USE
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
-------
CONTENTS
Page
FOREWORD
ABSTRACT i
DETERMINATION OF BASIN CHARACTERISTICS
FOR AN URBAN DISTRIBUTED ROUTING
RAINFALL - RUNOFF MODEL, W.M. Alley and
J.E. Veenhuis 1
EVALUATION OF PROPOSED URBAN RUNOFF
CONTROL ALTERNATIVES, G. LeClerc 28
SUMMARY - SEMINAR ON THE DESIGN
STORM CONCEPT, S.G. Walesh 47
USE OF RUNOFF/EXTRAN FOR HYDRAULIC
ANALYSIS OF URBAN DRAINAGE SYSTEMS,
R.E.G. Sprenger and C.E. Bright 55
SIMPLIFIED SEGMENTATION OF AN URBAN
CATCHMENT FOR URBAN RUNOFF SIMULATION
WITH SWMM, P. Purenne, G. Courchesne
and G. LeClerc 88
WATER QUALITY PREDICTION FOR
URBAN RUNOFF - AN ALTERNATIVE
APPROACH, E.V. Diniz 112
WASTEWATER INTERCEPTION ON THE COMMUNAUTE
URBAINE DE MONTREAL TERRITORY, G. Paquin 135
PROCEDURE FOR THE ESTABLISHMENT OF
STATEWIDE WASTELOAD ALLOCATIONS,
R.A. Hawkins, B.F. Malloy and
J.L. Pavoni * 154
CUM'S SIMPLIFIED HYDROGRAPH METHOD FOR
DETERMINING URBAN RUNOFF, C. Mitci,
H.N, Nguyen Van, M. Osseyrane 174
FEASIBILITY OF STORM TRACKING FOR AUTOMATIC
CONTROL OF COMBINED SEWER SYSTEMS, V.T.M. Nguyen,
M.B. McPherson, 201
IMPSWM, P. Wisner and A. Kassem 237
-------
DESCRIPTION AND APPLICATION OF AN
INTERACTIVE MINI-COMPUTER VERSION OF
THE ILLUDAS MODEL, G. Patry, L. Raymond,
G. March! 242
MODELLING URBAN AND AGRICULTURAL
DEVELOPMENT, HYDROLOGY AND EUTROPHICATION
IN A LOUISIANA SWAMP FOREST ECOSYSTEM,
C. Hopkins and J. Day, Jr. 275
DEVELOPMENT AND IMPLEMENTATION OF AN
URBAN-RURAL SUBCATCHMENT HYDROLOGIC
MODEL (SUBHYD) FOR DISCRETE AND CONTINUOUS
SIMULATIONS ON A MICRO COMPUTER, L. Thompson
and J. Sykes 320
RAINFALL - RUNOFF MODELING OF FLOW AND
TOTAL NITROGEN FROM TWO LOCALITIES IN
THE DENVER, COLORADO, METROPOLITAN AREA,
W. Alley and S. Ellis 362
LIST OF ATTENDEES 404
-------
FOREWORD
A major function of the Research and Development program of the
Environmental Protection Agency is to effectively and expeditiously trans-
fer, 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.
Thonas A. Murphy
Deputy Assistant Administrator
Office of Air, Land and Water Use
iii
-------
ABSTRACT
This report includes fourteen papers, on various model-related topics.
Presented at the semi-annual joint U.S.-Canadian Stormwater Management Model
(SWMM) Users Group Meeting, held May 24-25, 1979, in Montreal, Quebec, Canada.
Topics covered included applications of the SWMM to urban and rural catch-
ments, the selection of design rainfall events for model use, a description
of the Montreal Urban Comiunity's wastewater interception program, several
applications of other models in facility planning and design, development of
a mini-computer version of the ILLODAS model, and development of an urban/
rural subcatchment hydrologic simulation model (SUBHYD).
lv
-------
DETERMINATION OF BASIN CHARACTERISTICS
FOR AN URBAN DISTRIBUTED ROUTING RAINFALL-RUNOFF MODEL
By William M. Alley and Jack E. Veenhuis
U.S. Geological Survey
Presented at the
Storm Water Management Model Users
Group Meeting
Montreal, Canada
May 24-25, 1979
-------
DETERMINATION OF BASIN CHARACTERISTICS
FOR AN URBAN DISTRIBUTED ROUTING
RAINFALL-RUNOFF MODEL
BY William M. Alley1 and Jack E. Veenhuis2
INTRODUCTION
During the past decade, there has been a proliferation of urban rainfall-
runoff models. Although many publications describe the mathematical formu-
lation of these models, guidelines for determination of the basin
characteristics to be used as part of the model input are almost nonexistent.
Often, a listing of input and output data for a sample computer run is the
extent of such guidelines.
The purpose of this paper is to discuss some of the procedures involved
in determining basin characteristics for rainfall-runoff modeling. This
discussion will center on the rainfall-runoff model documented by Dawdy and
others (1978). However, transferability of much of the methodology to other
rainfall-runoff models such as the Storm Water Management Model (Huber and
others, 1975) should be possible. The approach described is not meant to
provide hard and fast rules, but rather to serve as a guide. Different study
objectives, study constraints, and models will require deviations from this
guide.
Hydrologist, U.S. Geological Survey, Reston, Virginia
o
Hydrologist, U.S. Geological Survey, Lakewood, Colorado
-------
DRAINAGE BOUNDARIES
A basin characteristic which is of interest in virtually all rainfall-
runoff studies is the total drainage area of the watershed. Determination of
this parameter to the uninitiated may appear trivial—simply outline the
basin boundary on a topographic map and determine the drainage area with a
planimeter. However, considerable errors might result from using such a tech-
nique. This is illustrated in figure 1 which shows actual basin boundaries
as determined after field inspection and the basin boundaries determined
solely from U.S. Geological Survey 7-1/2 minute topographic maps. The ratios
of drainage area determined solely from a topographic map to drainage area
determined from field-verified boundaries for the 12 Denver basins ranged from
0.13 to 4.90. Major factors involved in these discrepancies include the
effects of the street and storm-sewer systems, irrigation ditches, and the
resolution capability of the maps.
All of the drainage basins shown in figure 1 have well-defined boun-
daries. However, this is not the case for all basins. Our experience has
indicated that (1) well-defined urban watersheds with a hydraulically suitable
site to measure discharge and without internal basin complications are diffi-
cult to locate, and (2) basin complications can be overlooked unless a
thorough investigation of basin characteristics is undertaken. For example,
the drainage area of an urban drainage basin may vary between different
storms as a result of an irrigation ditch overflowing into the basin or
exceedance of the capacity of an inlet or storm sewer resulting in conveyance
of an unknown quantity of runoff to another basin. Other problems may occur
such as a large storm sewer which discharges into two smaller storm sewers,
one of which conveys an unknown quantity of water out of the basin. A basin,
-------
Basin
Outlet
Basin
Outlet
Goose Creek at Boulder
Basin
Outlet
South Platte River \
Tributary at Englewood \
Basin
Outlet
Big Dry Creek Tributary
at Littleton
Basin
Outlet
South Platte River Tributary at Denver
Basin
Outlet
N
O
h
Clear Creek Tributary at Arvada
EXPLANATION ]
1 Field-verified Basin Boundary
Basin Boundary from Topographic Map Only
121 - Ratio of Topographic to Field-verified Drainage Area
Harvard Gulch Tributary at Englewood
] 2 Miles
2 Kilometers
Figure 1.—Topographic and field-verified drainage basin boundaries
for 12 Denver area sites.
-------
Basin
Outlet
Basin ILc
Outlet
\
Sand Creek Tributary
at Denver
4.9O
Sand Creek Tributary at Aurora
\
\
Basin
Outlet
Basin
Outlet
Sanderson Gulch Tributary at Lakewood
Kennedy Drive Drain
at Northglenn
.Basin Outlet
Boulder Creek Tributary
at Boulder
Basin
Outlet
O
South Platte River Tributary
No. 2 at Northglenn
IMile
EXPLANATION 1 Kilometer
—— Field-verified Basin Boundary
Basin Broundary from Topographic Map Only
O.13 - Ratio of Topographic to Field-verified Drainage Area
Figure 1.—Continued—Topographic and field-verified drainage basin
boundaries for 12 Denver area sites.
-------
particularly in newer developments, may have a complex, unregulated de-
tention reservoir. All of these problems and others have been identified in
actual field studies of potential sites. Areas with undersized storm sewers,
storm-sewer inlets, or irrigation ditches are particularly susceptible to
problems. Field verification during rainfall and snowmelt runoff is sometimes
necessary to resolve questions about basin response to runoff. Also, consul-
tation with local government officials familiar with urban drainage, zoning,
and/or public works can be invaluable.
These observations indicate that basin characteristics are not only im-
portant for rainfall-runoff modeling. In addition, if a basin is to be
instrumented for a rainfall-runoff data-collection program, a thorough in-
vestigation should be conducted at the outset of the program to insure the
usefulness of the basin. Detailed determination of segment characteristics
for modeling would not be necessary at that time. However, prior to initiating
an expensive data-collection program, it is important to assure that there
are no basin complications which can negate the utility of that basin for
inclusion in the study and to determine general basin characteristics which
can be used to assist in final selection of basins. Occasionally, an adequate
description of the storm-sewer system draining a basin may not be available.
This factor alone may limit the utility of the basin for data collection.
It is important to recognize that urban drainage basins are often
"watersheds in transition" and that the basin characteristics may change
significantly from year to year. Some of these changes, such as in impervious
area, can be monitored from aerial photographs. However, other changes, such
as in the sewer system, often cannot be detected from aerial photographs.
There have been instances when renovations to the storm-sewer system
-------
were made during a rainfall-runoff data-collection program. Upon analysis
of the data, the old storm-sewer maps had been lost or destroyed and the
characteristics of the previous storm-sewer system no longer known.
Usually, basins selected for data collection are not expected to undergo
significant changes. However, this is often difficult to predict. Occasion-
ally, a basin might be selected with the intent of monitoring changes in
runoff characteristics as the basin develops. In either case, an annual
update should be made of the characteristics of all basins included in a
rainfall-runoff data-collection program.
BASIN SEGMENTATION
The objective of basin segmentation is to characterize a basin in terms
of a discrete set of segments which account for the essential basin
properties. Each segment is assumed to be homogeneous in its properties.
The model developed by Dawdy and others (1978) considers four types of
segments:
1. Overland-flow segments
2. Channel segments
3. Reservoir segments
4. Junction segments
Overland-flow segments receive uniformly distributed lateral inflow from
excess precipitation. They represent a rectangular plane of a given overland-
flow length, roughness, slope, and percent imperviousness.
Channel segments receive upstream inflow from as many as three other
segments, including combinations of other channel segments, reservoir
segments, and junction segments. They also receive lateral inflow from
overland—flow segments.
-------
Reservoir segments provide for either linear-storage routing or
modifled-Puls routing (Soil Conservation Service, 1972). They can be used
to describe an on-channel detention reservoir. Alternately, they can be use
to simulate culverts which detain water due to limited capacity and for whic i
outflows are uniquely described as a single-valued function of storage behind
the culvert.
Junction segments are used when more than three segments contribute in-
flow to the upstream end of a segment or as input-hydrograph points where
the user may specify an input hydrograph for each storm.
Most of the discussion in this paper centers on overland-flow and channel
segments. The interrelationships of rainfall, overland-flow segments, and
channel segments are illustrated in figure 2 which shows the discretization
(segmentation) of a simple urban watershed.
Consider the discretization of an urban drainage basin into a set of
model segments; the following steps might be followed:
1. Obtain available information on the basin including storm-sewer maps,
topographic maps, and land-use maps. Obtain 1-inch equals 100- to 1,000-foot
aerial photography. Aerial photographs of this scale should be available
either from local governments or from a local engineering firm. If no aerial
photographs are available, then a contract for obtaining them might be
considered. These services can usually be obtained at $200 to $1,500 per
basin, depending on the number of basins flown, the size of the basins, and
the scale of the photographs, among other factors. Mylar prints should be
obtained so that work copies can be made of each photo.
-------
iBasin
Outlet
EXPLANATION
—*- Direction of Flow
0F02 Overland-flow Segment 2
CH06 Channel Segment 6
—-—- Overland-Flow Segment
Boundary
*-> —HI Channel Segment
(a) PLAN VIEW OF DRAINAGE BASIN
{ 0F02
CH04 ^
| 0F05
" CH°6
0F06
0F04
CH05"
0F03 |
0F01
\ \
CH03 CH02
Inflow to CH01
Equals Sum of
Outflows from
CH04, CH05,
and CH02
CH01
BASIN
OUTFLOW
HYDROGRAPH
(b) SCHEMATIC REPRESENTATIONS OF MODEL SEGMENTS
Segment
0F01
0F02
0F03
0F04
0F05
CHOI
CH02
CH03
CH04
CH05
CH06
Inflow to Segment
Lateral Inflow
Rainfall Excess
it a
a a
a a
a a
0F01
0F01
0F02
0F03,0F04
0F05,0F06
Upstream
Inflow
- —
CH02, CH04
,CH05
CH03
CH06
(c) SEGMENT INTERRELATIONSHIPS
Figure 2.--Discretization of an urban basin into segments.
-------
10
2. Assuming aerial photographs are obtained, mark the location of the
drainage network and inlets on a set of the work photos. This will require
some field verification to assure that storm-sewer maps and other information
obtained for the basin are "as built".
3. Mark overland- and street-flow directions on the aerial photographs,
particularly at street intersections. This step will require some field in-
spection of the basin. Many times, flow directions on streets will be diffi-
cult to determine. Use of hand levels or field verification during periods
of rainfall or snowmelt might be necessary to resolve questions about flow
directions.
4. Using the marked-up photos, the basin can then be segmented.
There are no exact rules on how to proceed and segmentation will be
a function of the model used and the purpose of the study. However, as an
example, some general rules of thumb for basin segmentation for application
of the model documented by Bawdy and others (1978) are as follows:
A. It is often easiest to begin segmentation by starting at the down-
sream end of a basin.
B. Channel segmentation and overland-flow segmentation are generally
done at the same time as the two are highly interrelated.
C. The more highly developed part of the basin generally will require a
more detailed breakdown into segments because of additional complexity of the
drainage and the greater contribution to watershed runoff per unit of land
area.
-------
11
D. This model, as do most urban-runoff models, assumes all overland-
flow segments are rectangular. Therefore, attempts should be made to create
overland-flow segments which approach this shape.
E. Attempts should be made to obtain segments of fairly uniform char-
acteristics throughout the segment since this is assumed by the model.
F. Channel intersections often exert a controlling influence on
overland-flow segmentation because each overland-flow segment must drain to
a channel segment.
G. Channel segmentation is also governed to a certain extent by the
overland-flow segmentation; although, this effect has been minimized. The
lateral inflow rate to the channel segment that picks up the overland flow
is expressed in terms of overland outflow in cubic feet per second per foot
of channel length. It suffices to model the flow from a 1-foot wide strip
of overland flow having a characteristic length, slope, percent imper-
viousness, and roughness coefficient. Therefore, an overland-flow segment
can be specified as contributing lateral inflow to a number of different
channel segments. For example, overland-flow segment 0F01 drains to channel
segments CH02 and CH03 in figure 2. The model routes the flow through segment
0F01 only once and uses outflow from 0F01 (in cubic feet per second per foot
of channel length) as lateral inflow to both channel segments.
H. Segmentation into the fewest number of segments which will preserve
the essential basin hydrologic-response characteristics is desired. Finding
a suitable simplified segmentation is important for derivation of runoff
frequency curves which can be expensive to simulate. One approach might
be to perform various levels of segmentation and to determine the sensitivity
of the model to these different levels. Leclerc and Schaake (1973) have
-------
12
demonstrated that when a detailed segmentation is simplified, the changes
most likely to be observed are a faster response of the rising limb and
recession limb of the simulated hydrograph. Experience with the model has
shown that the "optimum" number of segments may depend more on basin com-
plexity and subbasin hydrograph interest than basin size.
I. Break points between channel segments should occur at channel inter-
sections (for example, the intersection of CHOI, CH02, CH04, and CH05 in
figure 2), at poincs of considerable change in channel characteristics, at
points where sewer surcharge or culvert surcharging may restrict the flow,
and at points of interest for subbasin hydrographs.
J. Modified-Puls reservoir segments can be used to simulate culverts
which detain water due to limited capacity and for which outflows are uniquely
described as a single-valued function of storage behind the culvert. A cul-
vert rating and stage-storage relationship will be needed to determine the
outflow-storage relationship required by the modified-Puls formulation.
SEGMENT CHARACTERISTICS
Some of the watershed characteristics are specified as pertaining to the
entire drainage basin. For example, in general application a single set of
infiltration and soil-moisture parameters are specified for the entire basin.
The model includes an option for calibrating these parameters using measured
rainfall and runoff data and a modified-Rosenbrock optimization technique
(Rosenbrock, 1960; Dawdy and O'Donnell, 1965). Other watershed character-
istics are required for each segment.
-------
13
Overland-Flow Segments
Characteristics required for each overland-flow segment include percent
impervious area, length of overland flow, overland-flow slope, and a
roughness coefficient.
Percent Impervious Area
Two types of impervious surfaces are considered by the model. The
first type, effective impervious surfaces, are those impervious areas that
are directly connected to the channel drainage system, such as streets and
roofs that drain onto driveways and streets. The second type, noneffective
impervious surfaces, are those impervious areas that drain to pervious areas.
An example of a noneffective impervious area is a roof that drains onto
a lawn.
Rain falling on noneffective impervious areas is assumed to run off
onto the surrounding pervious area. The model assumes that this occurs uni-
formly over the entire pervious area. This volume of water is added to the
rain falling on the pervious areas prior to computation of infiltration.
Several studies (Graham and others, 1974; Proctor and Redfern Limited
and James F. MacLaren Limited, 1976) of a similar model, the Storm Water
,*
Management Model II (Huber and others, 1975), have shown simulated runoff
volumes and peak flows to be highly sensitive to the percent effective
impervious area. A number of studies have outlined methods to determine
impervious area; however, few studies have separated effective from non-
effective impervious area. Most studies of this type have related effective
impervious area to the minimum ratio of runoff/precipitation measured for
-------
14
small storms (Crawford, 1971; and Miller, 1978). The rationale for
this approach is that for basins with highly permeable soils, runoff from
small storms originates almost entirely from the effective impervious
area in the watershed. Limitations of this approach are the large scatter
often observed in runoff/precipitation plots, the requirement for rainfall-
runoff data from the watershed, the methodology may not apply to basins
with soils of moderate to low permeability, and the sensitivity of the
method to errors in precipitation and flow measurements.
An alternate approach used in Denver rainfall-runoff studies was as
follows. The lengths and widths of all streets in each overland-flow segment
were measured from aerial photos and the total area of streets for each
overland-flow segment was determined. All roofs, parking lots, and other
impervious areas which could not be identified as either effective or non-
effective were then field inspected and all of these areas which were
effectively impervious were colored in red on the aerial photo. If only
part of a roof or other impervious area was effective, then only that part
which was effective was colored. The areas of several representative roofs
in each overland-flow segment were then measured from the aerial photo-
graphs and the average of these areas multiplied by the number of effective
impervious roofs for each overland-flow segment. A similar approach was
used for driveways. Finally, the area of remaining effective impervious
areas, such as parking lots, was planimetered. The combined field and office
work required for this approach was approximately 4-person days per square
mile of basin for a highly developed watershed.
The above methodology was assumed to be consistent with the sensitivity
of runoff simulations to estimates of percent effective impervious area
-------
15
as well as the inherent subjectivity of the analysis. For example, it was
often difficult to determine which part of a house roof was effective
impervious area, particularly for houses with a downspout close but not
connected to a driveway.
Other approaches might include a random sampling of roofs or a grid
overlay approach. Important requirements appear to be a large-scale aerial
photo to delineate impervious areas and some field inspection to differentiate
effective and noneffective impervious areas. Results from Denver studies
suggest that there may be some transferability of impervious area information
throughout a metropolitan area for selected land-use types.
Table 1 summarizes the analyses of impervious areas for 12 urban basins
in the Denver area, including 91 subbasins (overland-flow segments). As
expected, for single-family residential development the percentage of area
that is impervious appears to decrease as the lot size increases. Despite
this trend, the percentage of impervious area that is effective appears to
be relatively constant for all single-family residential lot sizes.
Remarkably, the five subbasins with trailer courts had essentially the same
mean and range of values for percentage of impervious area that is effective
as did the single-family residential subbasins.
Interesting aspects of the data summarized in table 1 are that five
different people were involved in their collection and that single-family
residential developments of many different ages and parts of the Denver
metropolitan area were sampled. The data suggest that for metropolitan
Denver, in the absence of field data for basins or parts of basins in single-
family residential land use (including trailer courts), it might be assumed
-------
16
Table 1.Total- and effective-impervious areas for urban basins in
metropolitan Denver
Land use
Single-family
residential
Trailer
Courts
Multifamily
residential
Commercial
Industrial
Lot size
(acres)
1-1
6 5
1-1
5 4
1-1
4 2
1
2-1
-
-
-
No. of
subbasins
sampled
59
3
3
3
5
4
3
11
Percentage of area
that is impervious
Mean Range
41
42
26
15
48
72
93
i/
37-46
38-46
24-28
12-18
40-52
61-86
92-94
37-78
Percentage of
impervious area
that is effective
Mean
58
56
59
55
57
75
92
if
Range
54-63
52-61
55-63
52-59
53-64
71-79
80-99
37-100
I/ Mean not reported because of large variations
-------
17
that approximately 57 percent of the total impervious area is effective.
The total impervious area might be measured or the mean percentages shown
in table 1 could be used. Only subbasin sampling in other urban areas
can confirm the values shown in table 1 for use outside the Denver area.
The seven multifamily residential and commercial subbasins that were
sampled had values for the percentage of impervious area that is effective
ranging from 71 to 99 percent. Additional subbasin sampling might be
warranted before using these as representative values. Industrial subbasins
showed large variations in both percent impervious area and the percentage
of impervious area that is effective. These variations result from the large
building sizes and the wide variations of storm-water-management practices,
ranging from connecting directly to storm-sewer lines to draining onto
vacant land.
Length of Overland Flow
The length of overland flow, in feet, can be computed as the area, in
square feet, of the overland-flow segment divided by the length, in feet,
of the channel(s) into which it contributes lateral inflow. Leclerc and
Schaake (1973), and Huber and others (1975) discuss other methods of de-
termining length of overland flow; however, these options are not included
in the present version of this model.
Overland-Flow Slope
Unless a large scale, smaller contour interval, topographic map of the
basin is available, the expense of contracting services for such a map is
generally not warranted. Overland-flow slopes can be estimated from
U.S. Geological Survey 7 1/2-minute topographic maps. One method
-------
18
would be to determine a weighted average slope from representative cross
sections of the overland-flow segment using the following equation:
n
E S, L.
... i i ,- ,
Slope = izi _ (!)
n
T, L.
i-1 X
where
Sj. = the slope of the ith cross section,
L^ = the length of the ith cross-sectional line, and
n = the number of sampling lines.
An alternate would be to use the following equation described by Wisler
and Brater (1959) :
DL
Slope = --£•
(2)
where
D = contour interval, in feet,
LC= total length of contours for segment, in feet, and
A = area of segment, in square feet.
Roughness Coefficients
Roughness coefficients used by the model are Manning's "n" values.
Because of the shallow flows involved, roughness coefficients are not static
but rather vary with the depth of flow and with variations in the properties
of the watershed such as grain size, slope, and density of grass blades. An
additional complication results from overland-flow segments having pervious
and impervious areas dispersed throughout the segment. An alternative pro-
vided by the model is to separate the runoff from an overland-flow segment
-------
19
into a pervious and impervious portion and to route each separately with
different roughness coefficients. Successful simulations have been made
assuming 0.014-0.016 for the impervious-area roughness coefficient and
0*15-0.50 for the pervious-area roughness coefficient. Some guidance on rough-
ness coefficients for pervious areas is given by Chen (1976).
Channel Segments
Segment characteristics required for each channel segment include channel
length, channel slope, Manning's "n", and cross-section geometry.
Channel Length
Channel length is a relatively easy parameter to measure from topographic
maps, storm-sewer maps, or aerial photos. Lengths of storm sewers are often
marked on storm-sewer maps.
Channel Slope
Channel slope is often measured as the difference in elevation at points
10 percent and 85 percent of the distance along the channel, measured from
the downstream end of the channel, divided by the distance between the two
points. Slopes of storm sewers are often marked on the storm-sewer maps.
Manning's "n"
Chow (1959) presents a good discussion of Manning's "n" values
for channels. Typical values might range from 0.02 to 0.12.
Cross-Section Geometry
The model uses kinematic wave formulation whereby a relationship is
specified between the area of flow (A) and the rate of flow (Q) as
Q = aAm
-------
20
The model parameters a and m can be determined by analysis of the cross
section geometry. For example, if for a given channel segment, the wetted
perimeter (P) can be expressed as a power function of the area of flow by:
P = aAb (4)
where a and b are constants, then Manning's equation for the segment
can be expressed as
Q - 1.49VS A(5/3 - (5)
2/3
n a
Therefore, a = 1.49/S" (6)
2/3
n a
and m = 1 _ -L b (7)
where S is the segment slope and n is Manning's "n".
The kinematic-wave parameters a and m are computed internally by the
model for rectangular conduits, circular pipes, and triangular cross sections
based on geometric properties of these cross sections. They may also be
specified externally by the user either based on flow measurements or
from equations 6 and 7.
In natural channels, the question arises as to whether a representative
channel cross section should be surveyed and equations 4, 6, and 7 used to
determine a and m or whether a simple rectangular, circular, or triangular
geometry can be assumed. A sensitivity analysis of rainfall-runoff simula-
tions was performed for South Utoy Creek Tributary at Woodberry Avenue at
East Point, Georgia, (the Woodberry basin) in order to investigate this
subject.
-------
21
The Woodberry basin is 0.21 square miles in area with drainage by a
natural channel down the center of the basin. A schematic of this basin is
shown in figure 3. Basin characteristics for rainfall-runoff modeling were
available (E. J. Inman, written commun., 1979). In addition, representative
cross sections of each of the channel segments had been surveyed from which
the values of a and b in equation 4 could be determined (E. J. Inman, written
comm., 1979). Alpha and m for each channel segment were then computed using
equations 6 and 7. Three computer runs were made for a set of 19 measured
rainfall-runoff periods. The first computer run was made using values for a
and m as computed using equations 6 and 7. The second computer run was made
using values of a and m as computed using "rough" approximations of channel
cross sections as having rectangular geometry. The third computer run was
made using "best" approximations of channel cross sections as having either
rectangular or triangular geometry.
The channel geometries used in the three computer runs are shown in
figure 4. Note that the same cross sections were used for channel segments
CH20 and CH21 for both the "rough" and "best" approximation runs. The
results of the three simulations are shown in table 2. Assuming that
simulated peaks for the first computer run are the "true" values, then the
average errors were 5.3 cubic feet per second for the "rough" approximation
run and 1.0 cubic foot per second for the "best" approximation run. Standard
error of estimate was 17 percent for the "rough" approximation run as compared
to 1.9 percent for the "best" approximation run.
The results of this sensitivity analysis indicate that, in general,
determination of values for a and b in equation 4 is not necessary and that
a "best fit" rectangular, circular, or triangular geometry can be assumed
-------
22
Basin Outlet
N
O
EXPLANATION
Drainage Basin Boundary
l _QI12L J Channel Segment and
"~ Number
0FO1 - Overland-Flow-Segment
Number
General Direction of
Overland Flow
1OOO Feet
Figure T,—Schematic of Woodberry basin showing segmentation for rainfall-runoff
modeling.
CH2O
CH21
CH22
CH23
CH24
EXPLANATION
Representative Channel Cross Section
, . . . "Rough" Approximation of Channel Cross Section with Rectangular
Geometry
"Best" Approximation of Channel Cross Section With Rectangular
and Triangular Geometry
Figure 4-—Actual and approximated channel geometry.
-------
23
Table 2.—Simulated peak flows for various channel-geometry assumptions
Runoff
Period
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Simulated peak
in cubic feet
Run No. I y
105
37.5
10.0
32.7 j
j
139
59.0
29.2
21.7
• • ~ " '
70.5
37.1
104
52.2
24.8
102
36.2
96.4
115
104
110
flow,
per second
i
Run No. 2 ^
100
32.9
8.4
27.7
137
52.7
20.0
15.1
65.8
30.7
99.6
51.0
23.0
99.9
28.0
84.0
103
101
105
Run No. 3 ^
105
37.5
10.2
33.2
144
59.4
30.0
22.1
72.1
37.3
107.7
52.5
24.9
103
36.7
95.2
115
104
114
-^ Alpha and ra values for channel segments based on representative channel
cross sections and equations 4, 6, and 7
2/ Alpha and m values for channel segments based on "rough" approximation
with rectangular geometry
—' Alpha and m values for channel segments based on "best" approximation
with rectangular and triangular geometry
-------
24
for each channel segment. However, care should be exercised In determining
these "best fit" channel geometries and dimensions. An approach used in
Denver studies was to estimate a simple channel geometry in the field without
resorting to detailed surveying of channel cross sections.
The capacity of rectangular-conduit and circular-pipe segments is limited
in the model to nonpressurized-flow capacity. If that capacity is exceeded
during a storm, provision is made to store the excess water arriving at the
upstream end of the segment. The volume stored increases without upper limit
as long as the upstream inflow exceeds segment capacity. After the upstream
inflow drops below segment capacity, the volume stored is released to the
segment. The upstream inflow to the segment remains at the maximum capacity
until the water stored at the upper end of that segment has been released.
In the real world, the capacity of a sewer may be controlled by inlets which
restrict flow in the sewer to less than full-pipe flow. A second possibility
is that a sewer may flow under pressure, thus having more capacity than
predicted using kinematic-wave theory. A third possibility is that once a
sewer is flowing full, additional inflow to the sewer is transferred to
streets parallel to the sewer system. These situations can, at times, be
approximated by adjusting the size of the circular or rectangular segment
appropriately. A modified-Puls reservoir segment can be used to simulate
culverts which detain water due to limited capacity and for which outflows
are uniquely described as a single-valued function of storage behind the
culvert. More complex situations may call for revision of the model or use
of a dynamic-wave model such as that presented by Water Resources Engineers,
Inc., (1972).
-------
25
SUMMARY AND CONCLUSIONS
This paper has described some of the difficulties, procedures, and
limitations of determining basin characteristics for rainfall-runoff model-
ing, A procedure for discretization of an urban drainage basin into a set
of model segments has been described including 10 general rules of thumb for
basin segmentation and approaches for determining segment characteristics.
The discussion has centered on a rainfall-runoff model presented by Dawdy and
others (1978). However, transferability of much of the methodology to other
rainfall-runoff models such as the Storm Water Management Model (Huber and
others, 1975) should be possible.
During investigations to identify urban watersheds for a rainfall-runoff
data-collection program, it was noted that well-defined urban watersheds with
a hydraulically suitable site to measure discharge and without internal basin
complications are difficult to locate. Basin complications can be overlooked
unless a thorough investigation of basin characteristics is undertaken prior to
final site selection.
-------
26
REFERENCES
Chen, C. L., 1976, Flow resistance in broad shallow grassed channels:
American Society Civil Engineers Proc., Journal Hydraulics Division,
v. 102, no. HY3, pp. 307-322.
Chow, V. T., 1959, Open channel hydraulics: New York, McGraw-Hill, 680 p.
Crawford, N. H., 1971, Studies in the application of digital simulation to
urban hydrology: Palo Alto, California, Hydrocomp Internat., 100 p.
Dawdy, D. R., and O'Donnell, Terence, 1965, Mathematical models of catchment
behavior: American Society Civil Engineers Proc., Journal Hydraulics
Division, v. 91, no. HY4, pp. 123-137.
Dawdy, D. R., Schaake, J. C., Jr., and Alley, W. M., 1978, User's guide for
distributed routing rainfall-runoff model: U.S. Geological Survey Water-
Resources Investigations 78-90, 146 p.
Graham, P. H., Costello, L.S., and Mallon, H. J., 1974, Estimation of
imperviousness and specific curb length for forecasting storm water
quality and quantity: Water Pollution Control Federation Journal, v. 46,
no. 4, pp. 717-725.
Huber, W. L.; Heaney, J. P.; Medina, M. A.; Peltz, W. A.; Sheikh, Hasan; and
Smith, G. F., 1975, Storm water management model user's manual, Version
II: Cincinnati, Ohio, U.S. Environmental Protection Agency, National
Environmental Research Center, EPA-67012-75-017, 350 p.
Leclerc, Guy, and Schaake, J. C., Jr., 1973, Methodology for assessing the
potential impact of urban development on urban runoff and the relative
efficiency of runoff control alternatives: Ralph M. Parsons Laboratory
Report No. 167, Massachusetts Institute of Technology, Cambridge, 257 p.
Miller, R. A., 1978, The hydraulically effective impervious area of an urban
basin, Broward County, Florida, in Internat. Symposium on Urban Storm
Water Management, July 24-27, 1978, Proceedings: University of Kentucky,
Lexington, pp. 259-261.
Proctor and Redfern Limited and James F. MacLaren Limited, 1976, Storm Water
Management Model Study, Volume 1: Toronto, Ontario, Canada, Ontario
Ministry of the Environment, Research Report No. 47, 295 p.
Rosenbrock, H. H., 1960, An automatic method of finding the greatest or least
value of a function: Computer Journal, v. 3, pp.175-184.
Soil Conservation Service, 1972, National engineering handbook, sec. 4,
Hydrology, chapt. 17, Flood routing: Department of Agriculture, pp. 17-1
to 17-93.
-------
27
Water Resources Engineers, Inc., 1972, San Francisco stormwater model user's
manual and program documentation: Dept. of Public Works, City and County
of San Francisco, California.
Wisler, C. 0., and Brater, E. F., 1959, Hydrology: New York, John Wiley and
Sons, Inc., 408 p.
-------
28
SWMM USERS CONFEr">!CE
MONTREAL, MAY 24-25, 1979
EVALUATION OF PROPOSED URBAN RUNOFF
CONTROL ALTERNATIVES
by
Guy LECLERC^
(1) Civil Eng. Dept., Ecole Polytechnique de Montreal
-------
29
ABSTRACT
This paper advocates using runoff frequency curves as the basis of
decision in drainage systems planning and design.
A conceptual framework for deriving runoff-related frequency curves
from rainfall probability distributions is presented and illustrated for
the case where the runoff statistic is a function of only two random vari-
ables.
Stochastic simulation is discussed as one solution procedure and ex-
amples of application are given.
Complete derivation of a runoff related frequency curve is probably
too costly for pratical planning and design applications. The stochastic
simulation procedure described suggests that the use of an ensemble of
historical storms selected from observed records can be a good pratical
trade-off between a design storm and complete derivation of frequency
curves from rainfall.
-------
30
1.0 INTRODUCTION
Storm and combined drainage networks are currently designed for
a design rainfall event, a storm with a given duration and mean rain-
fall intensity, both interlinked through intensity-duration curves.
The spatial distribution of the design storm is normally assumed uni-
form whereas its temporal distribution (the hyetograph) may be uni-
form, triangular or of the Chicago type. With intensity-duration
curves available, and with an estimated time of concentration of the
basin, at the upstream end of the element designed, the design rain-
fall exterior variables (intensity and duration) are known once the
recurrence interval for the design has been selected.
The recurrence interval associated with the design rainfall
event is one measure of the risk assumed, and more significantly,
an explicit recognition of the randomness of the rainfall process.
When the rational formula is chosen as transfer function, the recur-
rence interval of its runoff statistic (peak runoff) is equal to the
recurrence interval of the storm; this follows from the linearity
relationship between storm intensity and peak runoff.
Recent advances in urban drainage have been centered around many
aspects of the drainage problem, namely: a) improvement or develop-
ment of rainfall-runoff models with particular focus on urban runoff
applications, b) assessment of urban runoff control alternatives such
as temporary storage of urban runoff, c) assessment of water quality
control facilities for combined sewers and d) description of the rain-
fall process at a scale relevant to urban drainage analysis.
Very little attention has been given to the overall question of
how to assess a given drainage network, of how to compare two or more
drainage alternatives (both network and control facilities). Indeed
urban drainage studies even those using sophisticated rainfall-runoff
models, are still embedded in the design storm framework.
-------
sess
s
31
While it is accepted that the recurrence interval of a given
runoff statistic is not identical to the recurrence interval of the
runoff producing rainfall event, few studies have been concerned with
the problem of selecting the appropriate recurrence interval based on
runoff (quantity or quality) statistic(s),,or of comparing (and as-
ing) various control runoff alternatives. The San Francisco
tudy moved in the right direction with continuous simulation of ur-
ban runoff from a record of observed storms; recurrence interval were
then defined on runoff rather than rainfall statistics.
This paper presents a theoritical framework for deriving runoff
frequency curves from rainfall probability distributions , discusses
numerical methods of solution and assesses the faisability of deriving
urban runoff frequency curves within a drainage project study.
2.0 LITTERATURE REVIEW
Perkins (1970) presented a procedure for computing frequency
curves of flood stages in a flood plain.
Eagleson (1971, 1972) derived analytically the flood frequency
curve of some natural catchments in Connecticut from the rainfall
probability distribution of the average excess intensity and dura-
tion; he found that the analytic expression for the cumulative dis-
tribution of peak runoff reproduces well the essential features of
the observed curves.
Fleming and Franz (1971) compared four methods for deriving
flood frequency curves for small watersheds; in a test of eleven
watersheds, the simulation approach proved superior to the other
methods.
Leclerc & Schaake (1972) derived.by stochastic simulation the
flood frequency curve of a natural catchment; they concluded that
the simulation approach is feasible and reliable and that future
-------
32
work should be devoted to improve the computational efficiency of the
procedure developed.
In 1973, they used this approach to derive frequency curves of
urban runoff statistics, and proposed the use of these curves for
assessing the potential impact of urban development on urban runoff
and for estimating the relative efficiency of proposed runoff control
alternati ves.
Howard (1976) proposed an analytical model of storage and treat-
ment-plant overflows alternatives based on probability distributions
of rainfall variables; this model develops statistics of untreated
discharge through mathematically derived probability functions. This
model can be a screening model for directing more detailed simulation
analysis of the most interesting alternatives.
Bras and Chan (1978) derived an analytical expression of the
volume of water above certain threshold discharge rates resulting
from runoff. This expression is for an overland flow segment, whose
runoff is estimated through the kinematic wave model, and for expo-
nentially distributed storm duration and effective intensity.
Dendrou and Delleur (1978) argued for introducing the notion
of risk at the planning stage of ur urban storm drainage study. They
proposed a complete set of reliability measures, associated with the
three basic function of protection against local street-floods, con-
tribution to downhill river floods and pollution from runoff.
Marsalek (1978) reported on the results of research on the
implications in the use of the design storm concept in urban drain-
age design and planning. Several statistics were considered; based
on his results, he suggested that design storms could be replaced by
historical storms in future drainage design.
-------
33
3.0 THEORETICAL FRAMEWORK
Deriving the frequency curve of a runoff statistic from the prob-
ability distribution of relevant rainfall variables is essentially an
exercise in derived distributions; this topic is discussed among others
by Benjamin and Cornell (1970). In this section a general theoretical
framework is first presented, followed by a graphical illustration for
the particular case where the runoff statistic is assumed to be a func-
tion of only two rainfall random variables.
J.1 GENERAL CASE
Let Q be a runoff statistic and B_ be a vector of variables such
that
Q = Q(6) = Q(0r 82) (1)
The mathematical relationship Q,(£) between Q and 0 Is equivalent to
a mathematical model of the rainfal 1- runoff process. Q_ is made of
the parameters of the model and of the input variables which are as-
sumed random variables, Qy. Then the derived distribution of Q is
obtained through
FQ(q) = ///-. /R(q)
where ffi (9.) is the joint probability function of 9_.
£. — — "
F_(q) is the cumulative density function of Q, and R(q) is the
region where the vector 8_ leads to values of Q that are less than or
equal to q. The task then consists mainly in establishing the bound-
aries of the region R(q), once fQ (67) is known.
2
The rainfall-runoff model is considered here a deterministic
transformation function, the random components being accounted for
in the inputs to the model.
-------
34
F_(q) is the derived probability distribution of all rainfall-
runoff events. To obtain the frequency curve, say on an annual basis,
additional effort is needed to identify the maximum Q. that occured
in a given year, or the ensemble of the r largest Q's. The calendar
time must then be introduced to obtain an estimate of the recurrence
i nterval.
3.2 CASE OF TWO RANDOM VARIABLES
Equation 2 is a double integral when 6^ contains only two random
variables; in this case a graphical illustration of the derivation
mechanisms can be made.
Let tr be the storm duration and i be rainfal 1 excess mean
e e
intensity. Their joint probability density function, f(92) = f(l , tr )
is known and is shown in figure 1 for the double exponential case.
The integration region R(q) is defined by the three planes
trg)' (f(Te> ^ » Te> and (f^e' tre) ' tre)» the surface f(Te> trj
and the vertical plane through the contour line Q. = q. In figure 1,
Q = 200 ft'/s. The volume in this region is the probability that
Q is smaller than or equal to 200 ft3/s, that is F (200). GQ(200)
is the volume taken-off (GQ(q) = 1 - FQ(q)).
To obtain the frequency curve, the following equation, derived
by Eagleson (1972) is used:
= nGQ(q)
where n is the mean number of runoff producing rainfall events in a year
and T is the recurrence interval on an annual exceedance basis (years).
The role of the transformation function, thus of the rainfall-
runoff model is emphasized: the model defines the region of inte-
gration Rn(q) as !t locates the contour where 0_ is equal to q. All
-------
_ 1
tre
(-0
VI
CONTOUR Q =200 C.F.S,
Figure 1: Graphical representation of
-------
36
possible combinations of i and tr are accounted for; thus in de-
e re _
riving F (q), the probability of any combination i ~^re' Y'e'd'n9
Q. < q, is explicitly accounted for.
3.3 SUMMARY OF THE CONCEPTUAL MODEL
Note, before studying its practical implications, that this
model is very general.
- No assumptions are written relative to the probabil-
ity density function f~ (60.
- No assumptions are imposed on the maximum number of
random variables, J9_, governing the rainfall-runoff
process.
- No constraints are given to the selection of the
rainfall-runoff model.
The conceptual model is also applicable to the derivation of
frequency curves of other runoff statistics. For instance of the
number and magnitude of overflows at any point in the drainage net-
work, or of urban runoff quality indicators, provided a known and
reliable transformation function, Q(6|) is available.
It is also interesting to discuss the design storm concept In
locating this event in the i -tr plane. Normally its duration is
equal to the time of concentration of the basin, t*, and its mean
intensity is then determined from intensity-duration frequency (IDF)
curves. This storm shown in figure 2 is a single point in the i -tr
e ' e
plane; its does not defines a contour Q = q nor does it yield any
information about the frequency of the runoff parameter.
The design storm approach is a design method which can be accept-
able for design purposes; it cannot however rationalise any decisions
or future interventions on the drainage system.
-------
X
X,
52
,
-------
38
METHODS OF S(3LUTI 0N
Several numerical strategies can be designed to derive the fre-
quency curves T (Q) from random variables representative of the rainfall-
runoff process. Leclerc and Schaake (1973) compared four possibilities
including an analytical approach developed by Eagleson (1972) and a
stochastic simulation method.
The stochastic simulation procedure is presented here but this
does not imply that it is better than other approaches.
Figure 3 presents the solution procedure. Once selected the
runoff statistic for which the frequency curve is sought, a transfor-
mation function, a rainfall-runoff model, is chosen.
A stochastic rainfall generator (the vector Q^) is then selec-
ted. This choice is governed by several factors namely the availabil-
ity of pertinent data, the presence of a regional meteorological sta-
tion to be used as a pivot station, the size of the basin and the
objectives of the study. For the revision of the drainage master-
plan of Ville de Laval, rainfall statistics have been estimated for
operationally independant rainfall events at the Dorval Airport mete-
orological station. In this the approach adopted by Eagleson was
fcllowed.
The rainfall generator then is used to simulate a trace of
rainfall events; it contains a large number of events, most of them
not of any interest for the frequency curve, since they yield values
of q too frequent to be pertinent to the problem.
The purpose of the fifth step is the identification of those
rainfall events whose runoff will be simulated. Design of the screen-
ing procedure of storm events is crucial and special attention must
be given to it since any potentially interesting event not selected
could never be included in the derived frequency curve. In its design
consideration must be given to the nature of the runoff statistic and
-------
39
DETERMINE APPROPRIATE
RUNOFF PARAMETER
SELECT A CATCHMENT
RESPONSE MODEL
SELECT A STOCHASTIC
RAINFALL MODEL
SIMULATE A TRACE OF
RAINFALL EVENTS
DEVISE A SCREENING
PROCEDURE TO SELECT
STORMS TO BE SIMULATED
BY THE CATCHMENT MODEL
SIMULATE SELECTED
STORMS WITH THE
CATCHMENT MODEL
CONSTRUCT TE (Qmax)
Figure 3- Simulation solution procedure for
.}-
-------
40
to the physiographic characteristics of the catchment. Upon com-
plition ofthis step, a bank of rainfall events pertinent to the fre-
quency curve sought is available. The next step consists in the
simulation of these selected events with the rainfal1-runoff model
to determine the magnitude of the runoff statistic.
The final step, the construction of the frequency curve is done
by assigning to each Q a plotting position.
Stochastic simulation was used to derive the flood frequency
curve of natural catchments; figure *f compares the analytical ap-
proach of Eagleson with the simulated flood frequency curve.
Stochastic simulation was also used to assess the relative effi-
ciency of urban runoff control alternatives. Figure 5 compares peak
runoff frequency curves for uncontroled and controled conditions.
The control alternative is a small reservoir for which two outlet
designs were studied.
Stochastic simulation was also used to estimate the size of a
retention basin, storing runoff when discharges exceed a given thresh-
old, QTH. Figure 6 relates the overflow storage and threshold dis-
charge at constant T .
5.0 CONCLUSION
Frequency curves of a runoff statistic describe globally its
range of variation and the probability of occurence of a given mag-
nitude taken by the statistic. Frequency curves also allow relative
assessment of the efficiency of control alternatives.
Additional effort is needed a) to develop numerically efficient
methods of solution, b) to assess the component of the probability
density function fn(9.) f°r urban drainage studies, as a function of
the objective of each study, the size of the study area, etc., and
-------
1100 -,
PARTIAL DURATION SERIES
100 -
5 tO 20
RECURRENCE INTERVAL, Tg (YEARS)
100
Figure 4: Analytical.and simulated runoff frequency curves for the east branch
of the..Elghtm!le-Rlver near North Lyme, Connecticut,
-------
200
ISO
160
140
I20h
UJ
g 100
<
I 80
< 60
at
a.
20
GRAY HAVEN*
(UNCONTROLEO)
DESIGN A «
DESIGN B »
>d— ^-
..y
10 20 50 100 200
RECURRENCE INTERVAL,TE (YEARS)
500
Figure 5: Peak runoff frequency curves for reservoir outlet designs A and 8,
-------
43
YEARS
YEARS
6O
80
100
120
THRESHOLD DISCHARGE (C.F.S)
Figure 6: Overflow storage and maximum allowable outflow
pate at constant T_.
-------
44
to c) establish Q(6) when Q is a water quality variable of urban
runoff or of the receiving stream.
The design storm plots as a single point in the i ~tr plane and
& t-
does not yield any information about the frequency of occurence of the
statistics of interest.
Stochastic simulation and the method suggesting the use of se-
lected historical storms are closely related. In both cases a screen-
ing device is required to select the rainfall events of interest.
Eventhough historical storms may appear more credible than stochastic
simulation each approach has its own merits; they are indeed comple-
mentary.
It appears that the design storm concept will eventually be
replaced by an ensemble of historical design storms; assessment of
any runoff control alternatives or drainage designs will then be
more complete than an evaluation based on a single storm event with-
out calling for the formal derivation of a frequency curve. Such an
approach will soon be proposed to Ville de Laval as a future design
procedure.
-------
45
REFERENCES
BENJAMIN, J.R. and CORNELL, C.A., Probability, Statistics and Decision
for Civi1 Engineers, McGraw-Hill Book Co., 1970.
BRAS, R.L. and CHAN, S.O., "Theoretical Models of Hydrologic Parameters
Using Derived Distribution Techniques", International Symposium on Risk
and Reliability in Water Resources, University of Waterloo, Ont. Canada,
June 26-28, 1978.
DENDROU, S.A. and DELLEUR, J.W., "Reliability Concepts in Planning Storm-
Drainage Systems", International Symposium on Risk and Reliability in
Water Resources, University of Waterloo, Ont. Canada, June 26-28, 1978.
EAGLESON, P.S., The Stochastic Kinematic Wave in Systems Approach to
Hydrology, Proc. First Bilateral U.S.-Japan Seminar in Hydrology Honolulu,
1971-
EAGLESON, P.S., "Dynamic of Flood Frequency",, Wate£_Resou_rces Research,
Vol. 8, no. k, 1972.
FLEMING, G. and FRANZ, D.D., "Flood Frequency Estimating Techniques for
Small Watersheds", J. Hydraul, Div., Amer. Soc. of Civil Eng., 97 (HY9)
197'..
HOWARD, C.D.D., "Theory of Storage and Treatment-Plant Overflows", J. of
Env. Div. , Amer. Soc. of Civil Eng. , 102 (EE*0 1976.
LECLERC, G. and SCHAAKE, J.C., Jr., "Derivation fo Hydrologic Frequency
Curves", H.I.T. Dept. of Civil Eng., Ralph M. Parsons Laboratory, Report
no. 1^2, Cambridge Ma. 1972.
LECLERC, G. and SCHAAKE, J.D., Jr., "Methodology for Assessing the
Potential Impact of Urban Development on Urban Runoff and the Relative
Efficiency of Runoff Control Alternatives", M.I.T. Dept. of Civil Eng.,
Ralph M. Parsons Laboratory, Report no. 167, Cambridge Ma. 1973-
-------
46
MARSALEK, J., "Research on the Design Storm Concept", ASCE Urban Water
Resources Research Program, Tech. Memorandum no. 33, ASCE, New York, N.Y.
1978.
PERKINS, F.E., "Flood Plain Modeling", Water Resources Bulletin, 6 (3),
1970.
-------
47
SUMMARY
SEMINAR ON THE DESIGN STORM CONCEPT
ECOLE POLYTECHNIQUE, UNIVERSITY OF MONTREAL
MAY 23, 1979a
PURPOSE OF SEMINAR
1. To sensitize civil engineers to the positive and negative
features of the design storm concept so as to discourage its
blind use.
2. To provide a forum for exchange of information and ideas among
researchers in the United States, Canada, and elsewhere.
PANEL AND ATTENDEES
The nine person panel represented a variety of experience, interests,
and employment situations. Panel members were:
M.B. McPherson, ASCE Urban Water Resources Research Program
Jiri Marsalek, Environment - Canada
Harry Torno, US EPA
Ben Urbonos, Urban Drainage and Flood Control District, Denver
Gilles Patry, Ecole Polytechnique, University of Montreal
Jacques Delluer, Purdue University, Indiana
Harry Wenzel, University of Illinois
Michael Terstriep, Illinois State Water Survey
Stuart G. Walesh, Donohue & Associates, Inc., Wisconsin
The seminar was chaired by M. McPherson. Approximately 40 people, in
addition to the panel, participated in the seminar. Included in this
group were foreign guests, students and staff of Ecole Polytechnique,
and local engineers.
Presented at the May 24-25, 1979 SWMM Users Group Meeting in Montreal
by Stuart G. Walesh, Head - Water Resources Section, Donohue & Associates,
Inc., Misconsin.
-------
48
PROCEDURE
The seminar was inflated by providing each pane! member with an
opportunity to rna>e n statement of ten minutes duration or less. Panel
members used i ~'s as ar> oppor' vritv to express concerns, present ideas,
and comment br.efly on suoe ol their research and development efforts.
The remainder >./; the seminar, !-?hich Issued for approximately four
hours, consisted primarily of response, questions and contributions from
the audience. Th-? seminar ws characterized by an open exchange of
ideas and information tetjee-n the participants.
STRUCTURE OF THIS SUMMARY
Rather than attempt a sequential account of the ideas and information
exchanged during t> course of the seminar, this summary is structured
so as to collect t'.u^a ideas and information into related categories.
More specifically, the following categories are described in detail
below:
1. Definition of Design Storm,
2. Perceived or Demonstrated Short-Comings of the Design Storm
Concept.
3. Perceived or Demonstrated Positive Aspects of the Design Storm
Concept.
4. Alternatives to the Traditional Use of Design Storms.
5. Research Underway or Proposed.
6. Miscellaneous Considerations.
DEFINITION OF DESIGN STORM
The overall idea of a design storm is to provide a means of estimating a
discharge or volume of specified recurrence interval for planning or
design purposes. That is, a recurrence interval is assigned to the design
storm, a rainfall-runoff procedure is used to convert the rainfall to run-
off, and the recurrence interval of the design storm is transferred to
the resulting runoff discharge or volume.
The general concensus of seminar participants was that there are two
types of design stern.
-------
49
Design storms are typically constructed so as to generate peak
discharges. Therefore, design storms usually do not include
the long duration-high volume rainfall events likely to be
important in determining required sizes of detention/retention
facilities. Design storms are even less likely to be suitable
for determining non-point source pollutant loads to and concen-
trations in streams.
6. Area! variability of rainfall may be an imporant factor in
determining discharge and volume at the catchment outlet. This
variability is typically ignored in applying design storms.
7. Finally, engineers and other professionals involved in planning
and design may expose themselves to criticism and even liability
through use of an "unproven technique."
PERCEIVED OR DEMONSTRATED POSITIVE ASPECTS OF THE DESIGN STORM CONCEPT
The design storm concept enjoys widespread use. This is apparently in
response to the following positive aspects of the concept.
1. Use of design storms requires minimal resources in terms of
time and money.
2. Design storms appear to give conservative results, that is,
high discharges and volumes.
3. Application of the design storm approach is generally accepted
in practice and one can argue for maintenance of consistency
of methodology in a given jurisdiction or geographic area.
ALTERNATIVES TO THE TRADITIONAL USE OF DESIGN STORMS
Seminar participants identified the following available or potential alter-
natives to use of design storms:
* 1. Continuous Simulation - Although this approach completely
eliminates the need to use the design storm approach, it may
prove to be costly primarily because of long computer runs.
In addition, a continuous model may not be available to match
the physical situation under study. Furthermore, precipitation
data may not be available in the study area or may be of in-
sufficient duration requiring excessive extrapolation of the
resulting discharge-probability or volume-probability relation-
ships.
2. Continuous simulation with cost-reducing techniques.
-------
50
1. Direct use of intensity-duration-frequency (IDF) relationships
to obtain a uniform intensity for a given duration and recurrence
interval, such as is commonly done in applying the rational
method of storm sewer design.
2. Development of a synthetic hyetograph (or storm profile) from
IDF relationships, historic storms, or other means. Examples
of design storms of this type are: the Chicago hyetograph
and variations on it, the U.S. Soil Conservation Service Type
II hyetograph, and the quartile hyetographs developed by the
Illinois State Water Survey.
PERCEIVED OR DEMONSTRATED SHORT-COMINGS OF THE DESIGN STORM CONCEPT
1. A fundamental assumption associated with the design storm
concept as it is usually applied is that the recurrence interval
of the design storm may be transferred to the discharge or
volume produced with the storm. This equivalency of recurrence
interval has not been confirmed.
2. Casual observations of rainfall runoff data for watersheds
reveal instances in which strikingly similar hyetographs
produce markedly different hydrographs. Also, very
similar hydrographs can be shown to have been generated by
markedly different hyetographs. This suggests the important
role of factors such as antecedent moisture conditions and
area! distribution of rainfall over the catchment. Recent
modeling studies (i.e., Marsalek-Canada, Wenzel and Voorhees-
Illinois, and Urbonos-Denver) using historic storm series
suggest that design storms may produce very large errors in
calculated discharges.
3. Some of the methods used to construct design storms may be
incorrect uses of IDF relationships, (i.e., possibility of
incorrectly interpreting a Chicago-type hyetograph).
4. Design storms don't usually yield all the probability informa-
tion desired for planning and design studies. For example,
whereas a design storm might provide a flood flow of specified
recurrence interval, it is not suitable for developing a flow
duration curve. The need for additional probability-based
hydro!ogic-hydraulic data is expected to become more pressing
with increased work in the area of non-point source pollution.
5. If the design storm concept is generally less than adequate
for determination of discharges of specified recurrence interval,
chen existing design storms are even less likely to be suitable
for determination of volumes of specified recurrence interval.
-------
51
3. Use of a series of "major" historic storms as input to an
event model followed by discharge-probability or volume-
probability analyses of the model output.
4. Improved, rationally developed and statistically meaningful
design storms. One possibility is use of modeling with
historic storm sequences to determine by "trial and error"
combinations of hyetograph shape or peakedness (peak of
hyetograph relative to duration), antecedent moisture condi-
tion, and duration that may be expected to yield accurate
discharges or volumes of specified recurrence interval for a
given geographic area and type of catchment. Another approach
is based on the premise that a hyetograph is defined by the
following five random, but not necessarily independent, variables:
duration, mean intensity, temporal distribution, area! distribution,
and antecedent moisture conditions. It may be possible to use
probability theory to develop probability distributions for
each of the five variables, to determine their interdependence,
and to ultimately combine the distributions into design hyetographs.
5. Use of one or a few readily remembered historic storms for a
study area without explicit assignment of a probability or
recurrence interval. This may be an effective way to relate
the carrying capacity of a plan or design to the public.
RESEARCH UNDERWAY OR PROPOSED
Given the variety of available alternatives to design storm concept,
some of the seminar discussion was devoted to research that is underway
or proposed. Research efforts identified during the seminar are as
follows:
1. Use of a series of historic storms with an event model followed
by subsequent discharge-probability or volume-probability
analyses of the output, (i.e., Wenzel and Voorhees-Illinois,
Marsalek-Canada, Urbonos-Denver and Walesh.lau and Liebman-
Wisconsin).
2. Reducing the cost of continuous simulation, (i.e., Geiger-
Germany, Huber-Florida, Alley-USGS, and Walesh and Snyder-
Wisconsin).
3. probablistic development of a design storm, (i.e., Hydroscience
for US EPA and Laurenson-Australia).
-------
52
4. Use of modeling to identify appropriate combinations of
hyetographs shape or peakedness, antecedent moisture conditions,
duration, and area! availability. (For example, Urbonos-
Denver, Lowing-United Kingdom, Falk-Sweden, and Geiger-Germany).
5. Development of computer programs intended to efficiently and
effectively screen long-term historic or precipitation data to
select subsets of storms suitable for simulation, with event
models^of discharges, volumes, or water quality. (For example,
Terstriep-Illinois State Water Survey, 4»Jenzel and Voorhees-
Illinois, and Lowing-United Kingdom).
MISCELLANEOUS
Some of the ideas and information offered during the seminar do not
fit into the above categories but are important and are presented
as follows.
1. There seemed to be a general consensus that there never will
be one methodology to replace the design storm concept but
rather the goal should be an improved set of methodologies.
Assuming the availability of such a set, practitioners could
match methodologies to the characteristics of each particular
planning or design project. In attempting to match methodologies
to projects, questions such as the following should be raised:
How complicated is the system?
How sensitive is the size and cost of the planned system
or the designed facility to the methodology?
What are the likely consequences of failure?
2. It was suggested that we should perhaps separate our science
from our engineering. Although many engineers are also scientists,
it is important to recognize that a scientifically unsatisfying
method may yield acceptable results for engineering purposes.
3. Land use changes may be expected to have a significant impact
on the hydrologic response of a watershed. Expected impacts
of urbanization should be accounted for in our hydrology
either by conservatisms in hydrologic analyses or by forecasts
of future land use.
4. Two apparently important characteristics of historic design
storm hyetographs, as revealed by more recent investigations,
are peakedness and areal distribution of precipitation over the
catchment. Hyetograph peakedness should be a consideration in
developing the hyetograph and in selecting the time increment
for the hydrologic analysis. In a similar manner, possible areal
-------
53
distribution of rainfall should be considerecis particularly
as larger and larger basins are analyzed.
5. Regardless of how the design storm issue is resolved, it is
often necessary to determine discharges or volumes for two or
more recurrence intervals for a given study catchment. For
example, the design of a "minor" underground drainage system
for a residential development, which may be sized to accommodate
flows of a five year recurrence interval, should be fallowed
by the design of a^major^above ground system, sized so as to
accommodate discharges having a recurrence of perhaps 100 years.
6. Another important issue is the selection of the recurrence
interval used in planning and design of storm and flood water
facilities. A related issue is consistency of recurrence
interval. For example, if residents of flood prone areas along major
streams are provided with protection up to the 100-year recurrence
interval flood stage, shouldn't similar levels of protection
be provided to residential areas away from major streams that
are subject to localized storm water inundation?
7. Although antecedent moisture conditions generally appear to
have a significant influence on runoff discharges and volumes,
there is some indication that this may not be »s important a
factor in arid areas.
8. As we proceed with our studies of the design storm concept, we
should be cognizant of the ideal way to test the design storm
concept and to develop alternatives to it. This will require
long-term rainfall-runoff data on relatively smell homogeneous
catchments. If such data were available, it would be possible
to do a parallel-type analysis. On one hands historic rainfall
data could be used to determine or develop design storm hyetographs
which could, in turn, be used to develop a discharge-probability
or volume-probability relationship for the catchment. Parallel-
ing this, the long-term runoff data could be statistically
analyzed resulting in a second discharge-probability or volume-
probability relationship. The two discharge-probability o
volume-probability relationships could then be compared and used
to assess the validity of the design storm concept and to make
adjustments to it.
CONCLUSION
Onaway to summarize the present status of the unproven design storm
concept and of replacments for it is to say, as one participant did that
"urban hydrology is a paradise for research." Perhap? that ^ z,i over-
statement of the problem and the situation is not oirKe thct ' oomy.
-------
54
On the positive side, the seminar revealed that both researchers and
practitioners in the United States, Canada, and elsewhere are very much
aware of the shortcomings of the design storm concept and are concerned
about the problem. Equally important, they are exploring alternatives
to and refinements of the design storm concept. Given that kind of
atmosphere, we can be optimistic that much will be accomplished in the
next few years.
-------
55
USE OF RUNOFF/EXTRAN FOR
HYDRAULIC ANALYSIS OF URBAN
DRAINAGE SYSTEMS
For presentation at
SWMM USERS GROUP MEETING
Montreal, Canada
May 1979
R.E.G. Sprenger, P.Eng.
C.E. Bright, P.Eng.
I.D. ENGINEERING COMPANY
(formerly
Templeton Engineering Company)
966 Waverley Street,
Winnipeg, Manitoba
R3T 4M5
(204) 477-1270
-------
56
INTRODUCTION
This paper outlines some of our experiences in
using the RUNOFF block and the EXTRAN block of the SWMM
program. Primarily we have employed these programs for
detailed hydraulic analysis of urban drainage systems.
We have based our mathematical modelling capabi-
lities on RUNOFF and EXTRAN for two reasons. First, their
basic approach to simulating the runoff and hydraulic trans-
port processes are sound. Secondly, they are widely avail-
able and are non-proprietary. The last point is important
in our opinion as the user has unrestricted access to the
programs calculation methods.
Some readers may question our choice of the
EXTRAN block instead of the TRANSPORT block. Most of our
work is in the City of Winnipeg, which is located on ex-
tremely flat terrain. In general the ground level is about
20-feet higher than the normal water level of the Red River.
Our sewer systems extend from 2 to 5 miles from the river
and that sewer systems usually operate under surcharged
conditions is a significant factor.
These programs have been criticized for their
complexity. However, the processes they attempt to simu-
lation are highly complex, and logically such programs will
be complex also.
These programs are not totally comprehensive. In
general, our philosophy has been both to adapt the programs
-------
57
to model the problem adequately where required and to
simplify the problem to fit the program wherever possible.
The expression equivalent pipe is frequently used. Conse-
quently we have had to develop at least a working knowledge
of computer programming and the operating system of the
computer facility we use, in addition to an understanding of
modelling techniques.
The SWMM programs were developed as a planning
aid, specifically for pollution abatement analysis. When
these programs are used for detailed hydraulic analysis in
the design process, the user often faces both technical and
logistical problems. As this is our primary interest in
SWMM, we have found ourselves in the position of debugging,
making minor corrections, and often modifying many of the
quantity routines.
Computer development has made mathematical model-
ling techniques available to most hydraulic engineers.
However, a modelling program cannot be used as a "Black
Box", a pitfall which has caught most EXTRAN users to some
degree. By employing a computer program the engineer cannot
shift his professional responsibility to the program's
authors or sponsors. It is the engineer's responsibility to
understand how the program computes its answers, and their
significance.
-------
58
TECHNICAL PROBLEMS
We have had relatively few technical problems with
the RUNOFF block. However, the magnitude and shape of the
RUNOFF hydrographs it develops are critical in the design of
sewer systems. Unless calibration data is available, an
understanding of the method the program uses to compute
hydrographs is essential to achieve reasonable results.
Through a literature review and detailed examina-
tion of the program listing, we were able to develop the
sketch shown in Figure I which illustrates the program's:
methodology.
In contrast we have experienced numerous techni-
cal difficulties with EXTRAN. This program is extremely
sensitive to the model configuration, stable results are
often difficult to achieve. Problems with surcharge flow
calculations and the resulting "Artificial Storage" require
proper management. Again, after considerable research a
conceptual sketch of Artificial Storage was developed as
shown in Figure 2.
We have attempted to use the following special
features: in line-storage nodes, weirs, orifices and tidal
gates. These routines are probably adequate for qualitative
work where hydraulic accuracy is not critical. However, in
each case modifications to the program have been required to
obtain correct hydraulic results for our purposes.
Our solutions to these problems and additional
features we have developed are outlined later in the context
of projects to which they apply.
-------
59
NOTE: SUBCATCHMENT TREATED AS THREE SHEETS.
SHEET I - IMPERVIOUS WITH
SURFACE RETENTION.
2 - PERVIOUS WITH
ALL FLOW SEEN-AS POINT
SOURCE TO GUTTER
i
SURFACE RETENTION.?
SHEET 3 - IMPERVIOUS
WITHOUT SURFACE RETENTION.
INLET
/
TEMPLETON ENGINEERING COMPANY
PROFESSIONAL ENGINEERS
WINNIPEG
MANITOBA
CONCEPTUALISED VIEW OF SWMM
RUNOFF BLOCK COMPUTATIONS
DATE:
MAY / 79
SCALE: NONE
DWG. NO.
-------
60
CONDUIT
E.RA. SWMM77
TEMPLETON ENGINEERING COMPANY
PROFESSIONAL ENGINEERS
WINNIPEG
MANITOBA
CONCEPTUALIZED VIEW OF SWMM EXTRAN BLOCK
SURCHARGED FLOW COMPUTATIONS
DATE: FEBRUARY 8,1978
SCALE:
NONE
FIG. NO.
-------
61
LOGISTICAL PROBLEMS
In our experience, the major logistical problems
are:
1. The high cost of operating the programs, parti-
cularly EXTRAN. Our runs are generally in the
$40 to $150 range.
2. The slow turn-around time associated with large
programs, especially when using time-sharing
facilities, a 24-hour turn-around is common for
large runs.
3. The potential for errors to occur in the manipu-
lation of larger volumes of input and output data
Points 1 and 2 above are primarily caused by the
size of these programs. For example, EXTRAN in SWMM 77
required in excess of 300K of computer core to operate.
For hydraulic analysis we use versions of RUNOFF
and EXTRAN set up as individual programs, from which all
quality routines have been removed. In addition, we have
removed print-out associated routines and variable arrays
which require core for storage, from EXTRAN. Our version
writes the hydraulic results (flow and heads) for each cycle
to a disk file, from which selected results are retrieved by
a post-processor (a separate program—REPORT). As a result
we operate RUNOFF at 145K and EXTRAN at 75K (100 pipes & 100
nodes). Turn-around times for EXTRAN average 20 minutes
including a REPORT run. REPORT requires 101K to operate.
-------
62
COMPARATIVE TESTING
EXTRAN - TECTRAN/REPORT
« NAME * CM
TlNE
bKUi> ubtu »
»_„___«„__—_»-
oASE HUN * 220.067 *
* « &
TKST 1 o
7.363 *
19.696 *
0.00*
TEST 2
7.BOO »
I7.«W1 *
51.75 *
- «•
TEST 3
7.401 *
«
5.1.54-
66
« THE JObCOST FOR A FULL REPORT RUN FOR THE ABOVE TESTS IS Si.80.
NOTES:
(.1) All versions handle 100 pipes & 100 nodes
(2) Base Run - Independent Program, no quality subroutines
(3) Test 1 - Printout Routines blocked
(4)- Test 2 - Printout Subroutines S common block arrays
removed (101 K)
(5) Test 3 - All remaining quality routines & arrays
removed (SI K)
COMPARISON
Computation
Base
§1.99
Tectran/Report
$1.54
Savinc
•23%
Cgmments
Fast Turnaround
Printing
$2.52
$4.53
$1.80
$3.34
29%
26%
All Computed
Results Available
Fig 3
-------
63
Figure 3 shows comparative core requirements and operating
costs for various stripped down versions of EXTRAN.
The problems associated with the management of
data, point 3 above, are two-fold. The output capabilities
of EXTRAN are limited to 100 printed cycles, with Time
Histories limited to 15 pipes and nodes. Therefore, the
user has to select the output prior to making the computer
run, which is a disadvantage since it is difficult to decide
in advance which are the most important elements. Addi-
tional EXTRAN runs are often required to obtain complete
results, especially when the model is unstable. Our post-
processor, REPORT, has greatly improved this situation,
through its ability Of retrieving any computed data quickly
from the disk file. Time histories can be obtained for any
or all pipes and nodes and at intervals as close as every
cycle, if desired. Examples of the new output are shown on
Figures 4-6.
Input data manipulation problems were not so
easily resolved as they involved an indefinable factor, the
human element. Our approach has been to automate the
creation of input files using interactive pre-processor
programs for both RUNOFF and EXTRAN.
It is rarely practical to simulate large systems
in a single run, a segmental approach is generally more
efficient. For example, separate models for individual
subtrunks and a model of the trunk alone. This results in
-------
1 r E A t. V f, L U i S R t I i;
iT, JOHN'S C«LI6R«riON - SjHicUr,N 103,
OCVERMNu TCP OF WATER OF ELtl/ATION 42,00 fll TRUNK REPLACED M FREELY-
PEAK HOUAL INFLOW IM CFS, FOR
MODE /
i 30 i i 0/02
i 6JOJO/03
i 80200/02
( 802JO/02
( 60310/02
( 60401/02
( 80421/02
< 30510/02
i 30600/02
(900000/03
TIIIE
.00,10)
,00.00)
,00.10)
.00,10)
,00,10)
.00.10)
.05,00)
,00,10)
.00.10)
.00,00)
IMF LOU'
5,63
0.00
1.33
4,39
2,81
3.26
1.23
3.20
3.34
0,00
NOSE /
( 00120/02
( 60100/02
( 80201/02
( 30240/03
t 80320/02
( 80-102/03
( 80430/02
( 80520/02
( 80700/03
(
TIME
,00,10)
,00.10;
.00.10)
.00,00)
.Ob. CO)
.00,00)
.00,10)
.05.00)
.00,00)
INFLOW
1.20
1.43
4,10
0.00
1,37
0,00
3.13
.95
0.00
NODE /
( 80121/02
i 80000/03
( 80202/03
( 30300/02
( 80330/02
( 60403/02
! 30440/03
( 80521/02
( 90000/03
TiH£
.00,10;
.00,00)
.00.00)
,05,00)
.00.10)
.00,10)
,00.00)
,05,00)
.00,00)
INFLOW
1.77
0,00
0.00
1.95
2,74
1,34
0.00
.7?
0.00
H3l.£ i
v UJiJw'OJ
< 30100/02
( 60210/02
i aojji/o:
i aoj4i>/03
( 80410/0.'
i 80500/02
i 80530/02
(800000/03
UK
.00.10)
.05.00;
.00,10)
, 00 , 1 J )
.00.00)
.00. 10)
.05.00)
.00,10)
.00.00)
INFLOW
4.8o (
4.58 <
4.02 I
3.01 1
0.00 (
3.16 (
2,24 (
3,29 (
0,00 (
NOSE / flHt IHFLOU
00140/03,00,00) 0.00
( 80101/02.00,10) 3,14
I 60220/02.05.00) 1.01
I 80302/03,00.00) 0,00
( 80400/02,05,00) 2,24
0420/02,05.00) 2.31
( 80501/02,05.00) ,72
80540/03,00,00) 0.00
1600000/03,00.00) 0,00
MAXIMUM HOWL WATER ELEVATION IN FEET FOR 46 NODES.
NODE /
( iOOOO/02
( 30110/02
i 602CO/02
( 30230/02
< 50310/02
( 80401/02
( B0421/J2
< 80510/02
< 80aOQ/02
(900000/02
riHE WATER E.
.05.50)
.07.40)
.08.20)
.06.40)
.07,40)
.08.20)
.08.50)
.08.40)
.03.00)
,08,40)
37,66
46.33
45.13
5!. 47
48.69
49,74
50,07
51.22
50,70
38.11
NOUE /
( oO 100/01
( 80120/02
( 30201/02
( 80240/02
( 80320/02
( 80402/02
( 80430/02
( 80520/02
( 80700/02
(
TIME UAIER E.
.59.20)
.07.40)
,06.40)
.06.40)
.07.40)
.03.401
.03.40)
.03.50)
.08,00)
42.98
47,43
46.130
51.50
49,25
49.89
49.99
51,72
50,72
NODE / TIKE HATER E.
( 80000/02.
( 80121/02.
( 80202/02.
( 80300/02.
( 80330/02,
( 80403/02.
( 60440/02,
( B0521/02.
( 90000/02,
08,30)
07.30)
07.40)
08.30)
07.10;
03.40)
08,20)
08,50)
08.40)
41.89
43.51
46.81
47,36
49.68
49,97
50,07
51.74
41.96
KOBE /
( 60100/02
( 80130/02
( 30210/02
( S0301/02
( 80340/02
( 80410/02
( 30500/02
( 30530/02
(800000/02
TIHE WATER E.
.08.20)
.04.20)
.07.20)
.08,40)
.08.10)
,03.40)
.08,30)
.03,501
.08,40)
43,12
54 .'00
47.94
47,83
49,69
49,39
50,04
52,24
38.01
NODE /
( 80101/02
( 80140/02
< 90220/02
( 80302/02
( 60400/02
( 80420/02
( 80501/02
( 80540/02
(400000/02
TIHE UATER E.
.05.10)
,02,50)
,07.30)
,08.40)
.08.30)
.08,40)
,08,30)
.08,40)
,05.50)
46,03
54.06
48,91
47,86
48,43
49.76
50,08
52. 2i
35.13
PEAK CUNl'UIT FLOWS IN CFS FOR SI CONDUITS.
CONDUIT/
( 80130/02
( 80122/02
i 30202/02
( 80240/02
< 80320/01
( 80401/02
( 80421/02
i 80510/02
,' 30600/02
,800000/02
( 51/02
--EOR--
--EOF —
1
HUE
,08.00)
.00,40!
.07,00)
.02.00)
.59.10)
.00.10)
,01.50)
.00,40)
,06,30)
,03.40)
,08,40)
FLOW
40.73
2,01
-.22
-.66
4. 65
3.42
1.28
3,?i
1,91
40. c6
4.50
CONDim/
( 80101/02
( 30130/02
( 30210/02
( 60300/02
( 80321/02
( 30402/02
( 304JO/02
( 80520/01
< 80700/02
(600000/02
(
TIME
.00.50)
.06.50)
.01.50)
.09,10)
,10.00)
,01.00)
.01.00)
,59,30)
,02.20)
,05,50)
FLOU
2.57
5.09
6.33
20.81
1.68
1.19
3.11
3.13
-1,33
1,43
CONDUIT/ TINE
( 80110/02.04,00)
( 80140/02. 00, 30>
( 80220/02,00.00)
( 80301/02.00.10)
( 80330/02.00.20)
( 80403/02,02,20)
( 60440/02.06.50)
( 80521/02.01.00)
( 90100/02.07.10)
(900000/02. 08. 40)
FLOW
6,55
-1,60
4,35
2.13
2.45
1.44
-.96
,95
4,79
4,50
COHIiUlT/ TIME
( 80120/02,07,30)
i 50200/02.09.00)
( 80-21/02.10.20)
( 80302/02. 05, SO)
( 80340/02. 07. 30)
i 30410/01.39,30)
i. 30500/02,09.00)
i t/0630/02 .OO.Oy )
i oOiOO/02.01.20;
( 49/02.05.50)
FL.CU
3,10
29.84
-1.54
-.81
-.28
4,61
7.23
2.74
2.04
1,43
CONDUIT/ TIHE
( 80121/02,07,30)
•' 80201/02,01.00)
< 80230/02,00,50)
( 80310/02.00.50)
( 80400/02.09.00)
( 80420/02,00,10)
( 00501/01.57.50)
< 80540/02.04.20)
( 60110/02.01.00)
1 50/02.08.40)
FLOU
-2.64
3.73
3,54
5,13
14,33
3,47
.94
-1,1/3
.42
40.66
-------
KU.V tESCRIPTIOt;
si. JOHN'S CALIBRATION - SUBTRUNK *oc,
GOVERNING TOP OF WATER OF ELEVATION 43.00 AT TftlJlIK REFLACct' M ffEEU •
DISCHARGING PIPES DESIGNED TO CAUSE BACKUP FO ELEVATION -12.0- iiT i'ril ,riiu>...
S.n. 6- ir />.
INTEGRATION CYCLES 720
TOTAL NUMBER OF CONDUITS IS 51
TOTAL NUMBER OF NODES IS 4i
LENGTH OF INTEGRATION STEP IS 10. SECONDS.
TINE ZERO IS 01.00.00. TIHE ENUEB IS 03.00.00,
TIME HISTORY OF FLOW AND VELOCITY
B(CFS), VEL(FPS)i HLOS(FT) PACE! i.
ST. JOHN'S CALIBRATION - SUBTRUNK »OB. FILE CREATION DATE/HUE 7V/OS/OB. 11.55.25.
GOVERNING TOP OF MATER OF ELEVATION 42.00 AT TRUNK REPLACED BY FREELY-
(T>
Ul
TIME CONDUIT 80100 CONDUIT 80101 CONDUIT B0110 CONDUIT 80120 CONDUIT 80121
FLOW VEL ENT. EXT. FLOU MEL EHT. EXT. FLOU VEL ENT. EXI. FLOW VEL ENT. EXT. FLOU VEL ENT. EXT.
01. 05. 00
01.10,00
01,15.00
oi.;o.oo
01.25.00
Cl.30,00
C1.3S.OO
01. ,0,00
01.4S.OO
01.50.00
01.55,00
02.00,00
02.05.00
02.lu.00
02.15,00
02.20.00
02.25.00
02.30.00
02.35,00
02,40,00
02,45.00
02.50.00
02,55.00
03,00.00
—EOS —
--c JF--
1
0.00
0.00
.00
.00
.01
.07
.31
.72
.93
2.12
3.89
31.53
38.71
39. S7
34.64
30.72
27.18
21,37
14,82
9.54
7,32
4.U
5,44
4.92
0,0 0.00 0.00
0.0 0,00 0.00
.2 0.00 0.00
.3 0.00 0.00
.5 0.00 0.00
,1 0.00 0.00
1.5 0.00 0.00
1.9 0.00 0.00
2.0 0.00 0.00
2.7 0.00 0.00
4.1 0,00 0.00
5,1 0.00 0,00
5,4 0.00 0.00
5,7 0.00 0.00
4,9 0.00 0.00
4.4 0.00 0.00
4.1 0.00 0.00
4.0 0.00 0.00
3.7 0,00 0.00
3.4 0.00 0.00
3,3 0.00 0.00
3,2 0.00 0,00
3.1 0,00 0.00
3.0 0.00 0.00
T
0.00
0.00
,00
.00
.01
.02
.02
.03
.04
.19
.97
2.50
2.35
l.-U
0.00
1.73
0.00
0.00
1.13
.03
.52
.34
.24
.14
I H E H
0,0 0.00
0,0 0,00
.5 0.00
1.0 0.00
1.6 0.00
1.7 0.00
1.9 0,00
2.0 0,00
2.2 0,00
3.5 0.00
4.1 0,00
4.1 0.00
3.7 0.00
2.4 0,00
2,o 0,00
2.4 0.00
2.4 0.00
3.0 0,00
0.0 0.00
2.9 0,00
2.4 0,00
2.9 0.00
2.8 0,00
0,0 0,00
I S T 0 R
(I(CFS)
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Y o F
0.00
0.00
.00
.01
.05
.07
.09
.12
.14
.58
2.14
5,82
4,29
5.94
2.42
1.81
1.53
1.41
0.00
.79
.47
.35
.34
.32
FLO
0.0 0.00
0.0 0,00
.4 0,00
1,1 0.00
1,8 0.00
2.1 0.00
2.3 0.00
2.5 0,00
2,4 0,00
4.0 0,00
5.3 0,00
5,4 0.00
5.2 0.00
4.9 0.00
2.7 0,00
1.4 0.00
0.0 0.00
0.0 0.00
0.0 0.00
2,8 0.00
2,7 0.00
2.4 0.00
2.5 0.00
2.7 0,00
U A N (i U
i VEUFPSli HLOS(FT) )
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0.00
0.00
0.00
0,00
0.00
0.00
0.00
0.00
o.co
0.00
0.00
0.00
0.00
o.co
0.00
E 1 0 u
J
0.00
0,00
,00
,00
,01
,02
.03
.03
.03
.10
.34
1.03
2.74
2.79
.55
.00
.01
.01
.00
,00
,00
.01
.01
.01
I T Y
0.0
0.0
.2
.4
.4
.8
,9
,9
.9
1.2
1.7
2.4
2.9
2,8
2,4
,9
.5
.4
0.0
.3
0.0
0.0
.7
.7
0.00
0.00
0,00
0.00
0.00
0.00
0.00
0,00
0.00
0.00
0.00
0.00
0.00
0,00
0,00
0.00
0.00
0,00
o.os
0,00
0.00
0.00
0,00
0.00
0.00
0.00
0.00
0,00
0.00
0,00
0.00
0,00
0,00
0.00
c.oo
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0.00
0,00
0.00
0,00
0.00
0.00
0.00
,00
.00
.01
.01
.02
.02
.01
.04
.17
.38
-1.97
-2,19
-1,03
.07
.03
.02
,02
.02
.01
.01
,01
.01
0.0 0.00
0.0 0.00
.1 0.00
,3 0,00
,4 0,00
,5 0,00
,4 0,00
,4 0,00
,5 0,00
.7 0,00
1,0 0.00
,9 0.00
-2.7 0.00
-3.1 0,00
-1.9 0.00
.4 0,00
,5 0.00
,5 0.00
0.0 0.00
.4 0.00
0.0 0,00
0.0 0.00
.5 0.00
.5 0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0,00
0,00
0.00
0,00
0.00
0.00
0,00
0.00
0.00
-------
f I rt E H ! S I u It i i. ," .i. J. ,
IMBLUES in Hi!) f.10
S CALIBRATION - SUSTRUttt. 108. :,_E CSE.5TIGN i.Al£/TJr.£ 7J/05/03.
TOP Of MIES OK ELtVATIGU '12,00 fil TRUNK KEPLflCEr. 11 fSEEU-
TI.1E
Mri.MD.SS
01.55.00
01. Si. JO
01.57.00
01. S3, 00
01. £9. 00
02.00.00
02.01,00
02.02.00
02.03,00
o:,04.oo
02.05.00
02.04.00
02.07,00
02.08.00
02,09.00
02.10.00
02.11.00
02.12,00
02.13.00
02.14.00
02.15.00
02.14.00
02.17.00
02.18.00
02.H.OO
02.30.00
02.21,00
02.22,00
02. 23.00
02.24.00
02.25.00
02.24.00
02.27.00
02.23.00
02.29.00
02.30.00
02.31,00
02.32.00
02.33.50
02.34,00
02.35.00
02.34.00
02.37.00
02.33,00
02.3J.OO
02. -40. CO
02.41.00
02.42.00
C2.43.00
02.44,00
1
BKM
JUNCTION
ELEV
35.96
34.00
36.05
34.12
34.25
34.48
34.63
37.14
37,44
37,58
37.44
37.44
37.48
37.27
37.08
34.34
34.55
34.35
34.13
34.05
35.99
35.94
35.90
35.37
35.85
35.84
35,83
35. S3
35. 82
35,82
35,81
35,81
35.90
35.7*
35.79
35.79
35.73
35.78
35.77
35.77
35.74
35.74
35.74
35.74
35.75
35.75
35.75
35.75
35.74
35.74
52.80
40090
DEPTH
.39
.43
.40
.55
,48
.91
1.24
1.59
1.87
2.01
2.07
2.09
1.91
1.70
1.51
1.29
.98
,78
,41
.48
,42
.37
.33
.30
.23
.27
.24
.24
.25
.25
.24
.24
,23
.22
.22
,22
.21
.21
.20
.20
.19
.19
.19
.19
.18
,13
.18
.13
.17
.17
GRUB
JUNCTION
ELEV
37. 87
38.04
38.24
33.55
38.93
39.44
39.94
40.23
40,71
41,0?
41.38
41.43
41.80
41.88
11,63
41.81
41.48
41,53
41.33
41,13
40.94
40.80
40.44
40,53
40.41
40.29
40.19
40.10
39.99
39.69
39.74
39.59
39.51
39.40
39.23
39.18
39.07
38.97
38.84
33.77
33.47
38.57
3B.48
38,39
38.31
3S.24
36.18
33.13
38,09
38.05
52.00
80000
DEPTH
1,00
1.17
1,39
1.48
2,04
2,57
3,0V
3,41
3,84
4.20
4.51
4,74
4.93
5.01
5.01
4.94
4.61
4,44
4,44
4.24
4.09
3.93
3.79
3,44
3.54
3.42
3.32
3.23
3,12
3,02
2.87
2.72
2.44
2,53
2,41
2,31
2.20
2,10
1.99
1,90
i.ao
1.70
1,41
1,52
1.44
1.37
1.31
1.24
1.22
1.18
tine.
GRNB
JUNCTION
ELEV
38,09
33.17
36.20
38.44
38.45
38.84
39.11
39.34
39.54
39.91
40.38
40.91
41.48
41. S?
41.93
41,44
41,15
40,44
40.20
39,84
3?. 57
39.35
39,17
38,97
38.45
36.50
38,47
38.41
38.38
38.35
38.33
38.30
33.29
33,24
38,25
33,24
38,21
38,21
38.19
38.18
ill, 17
38,14
38.15
38.14
38.11
33,12
38,13
38,11
38.10
38,10
HIST
52.00 GftttD O.CO GRNIi 0.00 GRHP 0.00
90000 JUNCTION 0 JUNCTION 0 JUNCTION 0
DEPTH [LEY DEPTH ELEY SEPTH ELEV DEPTH
,50
.53
.4,9
.85
1,04
1.25
1.52
1.75
1.97
2.32
2.7S-
3.32
3,89
4,30
4.34
4,05
3.54
3.05
2,41
2. 25
1.98
1,74
1.50
1.38
1.04
.91
. (Jb
,82
,79
,74
.74
.71
,70
,47
,44
.45
,42
.42
.40
.59
.S3
.57
.54
.55
.52
.53
,54
,52
,51
.51
OKI Of H. 0. L.
(VALUES IN FEJET)
-------
67
multiple models and their associated input decks for a
single system. Identifying and filing this data becomes a
major problem. Our pre-processors are designed to retrieve
input data from master files which contain all pertinent
data, such as all pipes and nodes for the entire system.
Only the master files are saved as any input deck can be
recreated quickly and accurately without any re-coding at
all. The EXTRAN pre-processor will also accept new or
modified data, which can be used to create or update master
files using ancillary routes.
A flow chart of our operating system is shown in
Figure 7. It appears to be complex but it eliminates a
great deal of the routine "bull work". However, it does
demand a higher level of user competence than is required
to fill out coding sheets by following the SWMM manual.
-------
68
-FLOW CHART-
Fig.7
-------
69
INLINE STORAGE NODES, ORIFICES/WEIRS AND LOOPED
PIPES — LEILA NORTH LAND DRAINAGE SYSTEM
In 1978 we carried out a preliminary engineering
study, and the detailed design for the first phase, of a
land-drainage system for a 700-acre residential development
in the north west sector of the City of Winnipeg. The site
is located a little over two miles from the Red River, the
nearest discharge point, with a fully developed residential
area inbetween.
A planning study, by Templeton Engineering Company,
in 1974, established the possibility of using detention
storage lakes which will discharge, off-peak, into an adja-
cent land drainage trunk sewer.
The system will be quite complex for a land
drainage system. There will be three or more large storage
lakes (design storages totalling 4.5 million ft ), reversing
pipe flows and an interconnected pipe system with two dis-
charge points. It was this complexity that encouraged us to
use EXTRAN for the hydraulic analysis of the proposed
system. However, we were involved in a considerable re-
search effort to develop the use of in-line storage nodes
and their required weirs and orifices.
A schematic of the initial model of the system is
shown in Figure 8). The model will become more complex as
the system develops.
-------
CP.R, \ V aesjEM CUT off j i
LtIL* NOHTH DIUINAOC ARCA
Fig. 8
-------
71
The pipe system is sized to carry a 5-year return
rainfall, while the storage facilities are designed to
contain a 25-year-rainfall within a 4-foot rise above normal
water level. Initial EXTRAN simulations were highly un-
stable. We found that the instabilities originated in the
flow transfer devices, weirs or orifices, required to link
the storage node to the pipe systems. Stable runs were
achieved by using very large orifices with their inverts set
at the normal water levels of the storage lakes. Under
partial flow conditions EXTRAN computes orifice flows by a
weir equation. Although this produced stable results they
were not entirely correct. Under conditions where the lakes
were discharging, it appeared that the mass discharge
through the orifice ( Q x t) did not match the rate of
change of water level in the storage nodes as computed from
the printed results. By getting print out cycle-by-cycle we
found that what was actually happening was that discharges
were pulsating.
Our solution was to develop our own transfer
device, a Control Weir, which simulates the operation of the
actual system that will control the discharge from the
storage lakes. The lake will discharge over a weir, which
maintains the normal water level of the lake. The down-
stream discharge pipe is installed such that its crown is at
or close to the weir crest elevation. This allows rela-
tively high discharges at low storage levels in the lakes,
-------
72
when flows are low in the downstream pipe system. At high
lake levels the discharge is controlled by the discharge
capacity of the outfall pipe or backwater effects of the
downstream pipe system. The latter can reverse flow into
the storage lake. Except for very low flows the discharge
is controlled by the discharge pipe and the weir is in
effect only a minor restriction.
Our Control Weir works as shown in Figure 9. The
weir discharge is computed using the weir equation, h. the
flow transfer to the downstream node is accomplished u: ing
the simpler dummy link concept used by the existing orifice
routine. However, if the storage node is discharging and
the discharge pipe is surcharged, the flow in the dummy link
(Q ) is not allowed to exceed the flow in the discharge pipe
LI
(Q ). If this occurs Q is reset equal to Q or a iser
p ii p
specified percentage of Q . Since QT is reset before the
elevation changes are computed at the nodes, mass balances
within the model are maintained. This aids in ensuring that
the simulation will not become unstable. Once we inserted
this routine into the BOUND subroutine, our model simula-
tions worked satisfactorily.
This is an example of what we call a "site speci-
fic modification". We know it worked for this particular
problem, but it may not be generally applicable to all such
problems. However, the basic approach used could be employed
to resolve similar modelling problems.
-------
STORAGE NODE
ORIFICE
STORAGE NODE
PIPE
DUMMY LINK
IF
Qp +VE
QL>o
H>ZCROWN
QL>Cf)QP
CONTROL WEIR
-------
74
Also of interest is the method used to insert this
modification into the program, which reduced the amount of
reprogramming required. However, this was at the expense of
making the program a little more complicated to operate
while using this routine. In brief, control weirs are
coded as orifices in the input data deck; a single card
entered in the EXECUTIVE block enables the program to
differentiate between orifices and control weirs. This
limited the modifications to the EXECUTIVE and BOUND sub-
routines and avoided changes to the INDAT 1 subroutine and
the common block arrays.
-------
75
UNCALIBRATED SWMM AS A PLANNING AID —
KILDARE LAND DRAINAGE TRUNK SEWER ANALYSIS
The purpose of this study was to determine the
effect on the level of protection provided by an existing
land drainage trunk sewer if its drainage area was increased
by 550 acres. The system, was at that time, servicing
approximately 1450 acres of typical urban development. The
trunk ranges in size from 54" to 114" over a 2 1/2 mile
length and discharges to the Red River Floodway. It was
proposed to connect the new development drainage at the very
downstream end of the trunk as shown on Figure 10.
In addition to the directly connected system this
trunk carries pumped discharge from a detention pond system
which will ultimately service 2800 acres. The design and
operation of this detention system is such that it did not
affect the problem that we were attempting to resolve in
this particular analysis.
The existing system was designed by the rational
method using a rainfall intensity formula which is now
obsolete. There was no data available with which to cali-
brate a simulation model and, also, time and financial
restraints did not permit a detailed analysis.
We employed the SWMM program as a planning aid by
using it to determine the hydraulic capacity of the trunk
when servicing only the existing area and computing the
effect on this capacity if the new area were also serviced.
-------
76
TEMPLETON ENGINEERING COMPANY
PROFESSIONAL ENGINEERS
KILDARE LAND DRAINAGE SYSTEM •
COARSE MODEL 1
DATE: MAY, 1978
-------
77
It was not possible, and fortunately not necessary, to
determine the level of service, in terms of return period,
represented by the hydraulic capacity.
We decretized the drainage area into coarse
(lumped) subcatchments as shown on Figure 10. The RUNOFF
block was used to generate inlet hydrographs of reasonable
shape and peak values while the EXTRAN block was used
to route these flows through the trunk sewers. Since
only the trunk pipes were analyzed in EXTRAN, the model
was balanced by checking to ensure that the subtrunks
could actually deliver their computed peak flows to the
trunk against the water levels computed for their inlet
points. An iterative procedure was used in which runoff
rates were increased until computed inflows could just be
delivered to the trunk by the subtrunks.
It was found that, although inflows from the
proposed new areas were added very close to the outfall,
backwater effects did extend for a considerable distance
upstream as shown in Figure 11. However, these backwater
effects were dissipated before reaching the high end of
the trunk and did not restrict the capacity of the high end
subtrunks. For the length of trunk in which backwater
effects were significant their effect on subtrunk capacity
was minimal. These comparative results were checked by
subsequently reducing and increasing the magnitudes of the
inlet hydrographs through reasonable ranges and repeating
the EXTRAN computations. In each case, backwater effects
-------
so_
56 _
52 _
48_
K>
44.
40_
36 _
EXISTING AREA ONLY
EXISTING AREA PLUS AREA I 8 2
EXISTING AREA PLUS AREA 1,2,3,4
KILDARE TRUNK SEWER PROFILE
FIG, II
-------
79
did not reach the high end of the trunk and their influence
on affected subtrunks was not significant. We, therefore,
concluded that the trunk could safely service the new
development without significantly reducing the level of
service to the existing area.
This study was a good example of how SWMM can
be efficiently and economically used in the planning/
decision-making process.
-------
80
INITIALIZATION, MASS BALANCED STREET/BASEMENT FLOODING AND
FLAP GATES — THE ST. JOHN'S-PQLSON COMBINED SEWER SYSTEM
This project involves the design of relief mea-
sures for two adjacent combined sewer systems/ St. John's
and Poison, and probably part of a third system, Jefferson,
which borders them on their I>ig:; ends. Tne total catchment
area of the three systems is in the order of 1800-1900
acres. These areas suffer extreme basement flooding in
response to the intense thunderstorms common during the
summer months. The normal summer river level is elevation
734 (geodetic), with most: basement elevations slightly over
750 (some as low as 747), the system has only a 16-foot
operating head.
This is by far ,the biggest and most complicated
project which we have tackled so far. A great deal of our
development work and most of our attempts to improve the
accuracy of the program have been pursued with projects like
this in mind. Our pre-processor and post-processor programs
were designed with the objective of making multiple simula-
tions for just such a project practical.
A reasonably refined model of the St. John's
system alone employs more than 350 pipes. Possible relief
system integrating all three systems will involve 700-900
pipes to be analyzed. Because of the size and complexity of
this sytem, it will be impractical to model the complete
system in a single run. Basically, we are analyzing the
individual subtrunks and the trunk separately using EXTRAN.
-------
81
Initially the trunk systems are being analyzed using coarse
subcatchment models. Subsequently, these will be refined
using card hydrographs produced from detailed subcatchment
models. These subcatchments are run with either an equiva-
lent outfall pipe simulating the trunk response or a con-
stant water level where appropriate. The labour and time
required for this procedure is greatly reduced by automating
the input data including the card hydrographs.
Two purely technical routines which we have added
to our program will play an important role in this project.
One is an initialization routine through which we will be
able to tackle the problem of "Artificial Storage" affecting
the computation of surcharge flows. Our basic approach here
is to reduce the amount of artificial storage available in
the model. Figure 12 illustrates the artificial storage
curve in SWMM 77. This results in a more unstable model,
which can only be stabilized by reducing the time step.
However, reducing the time step, and in some cases this will
be a drastic reduction, increases the costs of operating the
model. These operating costs are almost directly propor-
tional to the number of cycles in each simulation. By
employing our initialization routine, we are able to change
the time step (BELT) by stopping and restarting a simul-
ation, and only using small time steps when necessary to
achieve stable results.
This initialization is achieved by directing the
program to write the values of certain internal variables to
-------
82
DEPTH
ABOVE
NODE
CROWN
(FT.)
2.CRQWN L
(J)
GRELEV
DEUOM i 2OO
SURFACE AREA OF SURCHARGED
REDUCTION OF A$(J) UNDER SURCHARGE
IN SWMM 77
-------
83
a disk file at a predetermined cycle. A subsequent run is
then started at this cycle by reading in the disk file,
which resets the variables for the initial cycle, prior to
the execution loop in TRANQT. The program and input deck
are loaded in the normal manner, only the number of cycles
(NTCYC) and the time step (BELT) being changed. Apart from
choosing the correct variables to initialize, the most
difficult problem was to make subroutine INFLOW read the
correct hydrograph values to match the new start time.
The other technical innovation is a routine
developed so that we can simulate street or basement flood-
ing. In conjunction with this routine, we also compute and
print out the internal storage volumes employed in a partic-
ular simulation. We recognize three types of internal
storage; real storage in the pipes, the artificial storage,
which is part of the program methodology for surcharge
conditions, and the real storage which is available either
in the basements of combined sewer systems or on the streets
of separate storm sewer systems. This facility makes it
possible to calibrate a model to a runoff event which causes
basement flooding in the system.
These flooding storage calculations were imple-
mented by modifying the surcharge calculations of EXTRAN to
prevent water loss from the system when the hydraulic grade-
line reaches the ground or basement levels that are
simulated.
-------
84
This capability is achieved by adding another
deflection point to the Artificial Storage reduction curve
at the ground level, GRELEV, as shown in Figure 13. The
2
surface area DENOM is made large (50,000 ft ) above this
point and the water depth Y is allowed to rise freely. This
prevents the loss of water caused by the previous re-set of
water level to the ground. A damping routine is required to
negotiate the rapid transition between DENOM values at this
point.
The downstream sections of the trunks in these
sewer systems are well below river levels and, therefore, we
tried to use the tidal gate feature to model flap gates.
The BOUND subroutine had a re-set of depth to the river
level even though the gate was supposed to be closed. This
caused the program to ignore the flap gates. We corrected
this problem by inserting a statement to re-set the pipe
depth of water to the correctly computed value. This
modification has only been tested for gates located at the
outfall node and to this extent it is a site specific
modification.
-------
85
DEPTH
ABOVE
CROWN
(FT.)
h,
ASM) _AS(J)_ AS(J)
4- 2
DENOM KNOM4 DEMOM3
SUBFACE AREA OF SURCHARGED MODE (SQ.FT.)
REDUCTION OF AS(JJ UNDER SURCHARGE
MODIFICATION TO CORPKJ LOSS OF mm PROBLEM AT FLOODING
50,0i
-------
86
SUMMARY
In summary, the modifications we have made to our
version of the program are, or can readily be made, compatible
with any existing versions of the RUNOFF, EXTRAN, or W.R.E.
TRANSPORT blocks. We have not radically altered the prog-
ram's computational methods. Our changes are generally
expansions and adaptations of the original authors' techni-
ques to meet particular requirements we have encountered.
Such changes were required in greater part due to
our use of the Programs out of the context for which they
were intended, rather than any inherent short-comings of
the programs themselves.
After we had developed an understanding of the
hydraulic aspects of the SWMM program we have found it to be
an excellent planning and design tool.
In conclusion, we have found the SWMM Users Group
meetings to be an important forum for the exchange of ideas
and appreciate this opportunity to make a contribution
through this presentation.
-------
87
ACKNOWLEDGEMENTS
The three projects; Leila North, Kildare Avenue
and St. John's/Poison were all undertaken for the City of
Winnipeg, Waterworks, Waste and Disposal Division, whose
engineering personnel have actively encouraged the use of
mathematical modelling technology in their projects.
We wish to acknowledge the encouragement and
help extended by the U.S. Environmental Agency through
Harry Torno, The University of Florida through Wayne Huber
and Water Resources Engineers, Inc. through Larry Roesner.
-------
88
SWMM USERS MEETING-MONTREAL, QUE.
MAY 24-25, 1979
SIMPLIFIED SEGMENTATION OF AN URBAN
CATCHMENT FOR URBAN RUNOFF SIMULATION WITH SWMM
by
Pierre Purenne *
Guy Courchesne **
Guy Leclerc ***
* Desjardins, Sauriol & Associes
** Service du G€nie, Ville de Laval
*** Ecole Polytechnique de Montreal
-------
89
TABLE OF CONTENTS
gage
1.0 INTRODUCTION 1
1.1 Data bank 2
1.2 Simplified segmentation of urban catchment 2
1.3 Articulation of the conference „. 3
2 . 0 LAND USE DATA BANK „ 4
2.1 The node data bank 4
2.2 Estimation of runoff model parameters .... 5
2.3 Land use data to runoff parameters ....... 9
3.0 SIMPLIFIED SEGMENTATION OF AN URBAN CATCHMENT .. 11
3.1 Simplified segmentation criteria ......... 11
3.2 Development of the segmentation procedure. 12
4 . 0 DISCUSSION AND CONCLUSION 21
REFERENCES 22
-------
90
1. INTRODUCTION
Numerous rainfall-runoff models have been developed and
discussed in the last ten years to simulate urban runoff.
Carefull thoughts should be given, in a drainage study,
to the selection of a model. Ville de Laval, second lar-
gest city in the province of Quebec, had to select a
rainfall-runoff model for the revision of its drainage mas-
ter plan. SWMM was selected tentively for several reasons,
namely:
the model is readily available
the model benefits from an important technical back-up
both at EPA and at Environment Canada
the computer program of the model has a modular structure
the model can simulate water quality and snowmelt
the model, in its EXTRAN version can simulate specific
flow conditions such as pressure flows.
Selecting SWMM for a predominantly planning study was sur-
prising because it was not conceived for planning purposes
and because SWMM uses a fairly large data bank.
Before final selection can be made, a study was conducted
to assess if the data needed for SWMM were available and
to determine a strategy for segmenting urban catchments.
-------
91
1.1 Data bank
Available to the drainage analyst group was an up-to-date
land use data bank, developed primarily for the revision
of the water supply masterplan. The land use data bank
provides information on the actual spatial distribution
and forecasts future land use patterns for several date
in the next thirty years. Land use is expressed in terms
of residential, industrial and commercial areas, informa-
tion that has to be transformed into physiographic parameters.
1-^ Simplified segmentation of urban catchment
A search for a successful strategy for simplified segmenta-
tion of an urban catchment was undertaken. Such a strategy
is determinant in the selection of SWMM for planning purposes;
firstly it allows the simulation of urban runoff at a lower
computationnal cost than a detailed segmentation of the catch-
ment; secondly it reduces the level of spatial resolution in
the physiographic data needed to represent the urban catchment
thus the size of the data bank.
Final selection of SWMM for the Laval study is conditionnal
on the identification of a simplified segmentation procedure.
-------
92
1.3 Articulation of the conference
In this conference the translation of the land use data
bank into information for SWMM, transformation adopted
for the revision study of the Ville de Laval drainage
masterplan is first presented. This is followed by the
description of the simplified segmentation strategy:
comparisons between detailed and simplified hydrographs
are illustrated.
-------
93
2.0 LAND USE DATA BANK
Ville de Laval shares a common border line, the Riviere
des Prairies with the Montreal Urban Community. Laval
has a total area close to a hundred square miles (259 Km2) ,
the main drainage basin has an area of sixty square miles
(155 Km2). This basin is experiencing rapid urbanization
namely in its southern region.
For land use purposes, the city is divided into 1200 zones
each associated with a predominant type of land use; they
are: residential, industrial, commercial, recreation area,
transportation, institutional, agrucultural.
Residential zones are characterized by their population
density (persons/acre), which has been normalized to des-
cribe the type of residence found in the zone, specifically
single unit house, duplex, cottages, apartment buildings,
row houses, etc...
2.1 The node data bank
For drainage purposes, the city is divided into subbasins
of approximately 425 acres area (this number corresponds to
the largest operationnal unit area of the simplified segmen-
tation procedure). Each subbasin is associated to a node,
a point of the future main drainage network. The ensemble
-------
94
2.1 The node data bank (Cont'd)
of nodes constitutes the node bank; which contains the in-
formation used by SWMM as transformed from the land use
information. Thus the node data bank is the interface between
the land use data bank and the runoff model.
2.2 Estimation of runoff model parameters
Two parameters of the runoff process are directly obtained
from land use: the percentage of impervious area and the
drainage density. Both parameters strongly influence the
overland flow runoff hydrograph.
The percentage of impervious area corresponding to a given
land use, has been assessed through field investigations
in Laval. Table 1 shows the percentage associated to each
category of residential zones; table 2 presents this infor-
mation for non-residential zones.
The drainage density is a function of the length of overland
flow; it is essential to model correctly this variable as
it influences significantly the peakedness of the simulated
hydrograph. The width of overland flow (W) accounts in
SWMM for drainage density and is estimated for each zone as
the ratio of total area and mean depth of each lot. The
latter in urban residential area is fairly uniform and in
Laval was found to be about 130 feet; for other land use zones
a model value was adopted, as shown in table 3.
-------
95
TABLE 1
PERCENTAGE OF IMPERVIOUSNESS
IN LAVAL'S RESIDENTIAL AREAS
TYPE
POPULATION
DENSITY
PERS/ACRE
DESCRIPTION
IMPERVIOUSNESS
From
To
1 12 Single-family habitation 20
with lots over 10,000 sq.
feet of area
13 25 Single-family habitation 38
with lots between 5000 and
10,000 sq. feet of area
26 40 Town houses, duplex, triplex 50
adjoining single-family
habitations
41 60 2 or 3 stories multi-family 68
habitation
61 80 2 or 3 stories adjoining 68
multi-family habitations
81 -v- High-rise multi-family 60
habitations
-------
96
TABLE 2
PERCENTAGE OF IMPERVIOUSNESS
IN AREA OF DIFFERENT LAND USES (NON-RESIDENTIAL AREAS)
TYPE OF
LAND USE
DESCRIPTION
IMPERVIOUSNESS
Industrial
Commercial
Institution
Parks
Transport
Services
Agricultural
Industries, manufactures,
warehouses
Commerces, offices buildings,
hotels, motels, restaurants,
service stations
Educational establishments,
detention areas, churches,
hospitals...
Parks, playing ground,
golfs, wooded area, cemetery
Provincial large roads
High tension electric trans-
portation area, railways...
Agricultural area and other
zones not urbanised
80 to 95
95
70
55
10
-------
97
TABLE 3
MEAN DEPTH OF LOT (NON-RESIDENTIAL AREAS)
TYPE OF
LAND USE
Industrial
Commercial
Institution
Parks
Transport
Services
Agricultural
MEAN DEPTH
(ft)
120
150
130
300
50
100
300
-------
98
2.3 Land use_data to runoff parameters
A computer model has been written (Desjardins & Sauriol
(1977)) to estimate the parameters of the runoff model
for each subbasin (each node of the node data bank) from
the land use data bank.
The model identifies the zones or fraction of zones inclu-
ded in the subbasin. Figure 1 illustrates a typical case
where some zones overlap one or more subbasins. It estimates
the percentage of impervious areas in each subbasin as the
mean of each zone, or fraction of, included in the subbasin
and the width of overland flow as the sum of their respective
W; the subbasin area is also computed from the land use data
bank.
The model also computes the dry weather flow in each subbasin
from its total estimated population, or its land use if not
residential.
Other parameters, namely roughness, overland flow slope,
and infiltration constants are presently assumed constant.
Transport elements are not estimated from land use information
but are defined from network on topographic maps.
-------
99
FIGURE 1
TYPICAL LAND USE AND DRAINAGE BASIN OVERLAP.
LEGEND
SUBBASIN BOUNDARY
LAND USE ZONE
PRIMARY NODE
SECONDARY NODE
TRANSPORT ELEMENT
-------
100
3.0 SIMPLIFIED SEGMENTATION OF AN URBAN CATCHMENT
3.1 Simplified segmentation criteria
A catchment may be modelled or segmented in various ways
depending upon the objective of the simulation, the charac-
teristics of the problem analyzed and the level of resolu-
tion sought.
For planning purposes, the segmentation criteria are:
- minimization of the number of segments while preserving
the essential features at the hydrographs derived with
a detailed segmentation
- description of the main drainage parameters which signifi-
cantly influence the runoff, such as land use, topography, etc.
- description of internal (within the transport network) con-
trols (quality and quantity) and of the main transport
elements
- description of external(controls those modifying the inlet
hydrograph). External controls may be structural or non
structural.
A good strategy for simplified seer *a tat ion must yield
results whose resolution is comp cible with the objectives
of the study; the results should also be reliable in order
to select the drainage elements of the future network and the
(quality, quantity) control alternatives.
-------
101
3.2 Development of the segmentation procedure
A simplified segmentation contains as few drainage elements
as needed to reproduce the basic runoff characteristics of the
site.
Aggregating several small sub-catchments into a single one
whose area is the sum of the subcatchment areas and whose
parameters are averaged over these subcatchments is a priori
interesting strategy for planning purposes, both from a tech-
nical and practical stand-points.
There were three phases in the development of the segmentation
procedure. Firstly a simplified procedure was tested for the
23.3 acres Gray Haven catchment, an experimental basin of the
Johns Hopkins University Storm Drainage Research Project,
documented by Tecker (1969). Leclerc and Schaake (1973)
for MITCAT (Harley (1967)), proposed a simplified segmentation
for Gray Haven. Secondly the proposed procedure was applied
to a 53.15 acre pilot basin in Laval to test it to local
conditions. Thirdly a search was conducted to identify the
largest size of a unit catchment compatible with the objectives
of the study and the numerical capabilities of SWMM.
-------
102
3.2 Development of the segmentation procedure (Con't)
Gray Haven segmentation
Figures 2 and 3 show the detailed and simplified segmentations
of the 23.3 acre catchment. The former is made of 26 subbasins,
26 gutters, 14 pipes and 14 nodes, the latter of one basin
and one gutter whose length is equal to the longest transport
distance. The runoff parameters in the simplified version
is equal to the weighted areal average of the parameters of
each subbasin, with the exception of the width of overland
flow, W, which is equal to the summation of each variable.
Figure 4 illustrates the goodness of fit getween the detailed
and simplified hydrographs.
53.15 acreLavaltest basin
Using the procedure tested on Gray Haven, this test basin
was represented by a single catchment and a single gutter
Comparisons between simplified and detailed hydrographs
showed that the procedure can be applied to a catchment of
this size; this test pilot basin is refered to as the unit
module.
-------
LEGENDE =
NUMERO DU SOUS-BASSIN
NUMERO DU NOEUD
CONDUITE OU CANIVEAU
LIMITES DU BASSIN
LIMITES DES SOUS- BASSINS
M02)
-12
-,-.-.-.• z/|02CZi t
^==^=====^ rjii/
„ « ^mtl
PtUVIOMETRE^,^
^^=^W
DEBITMETRE ENREQISTREUR
I H DE TYPE PARSHAUU.
(I O2,
211
N 310
)3O9
N 409
14
13
12
/
\ 308 'I
)
\ 307
)
\ 206
)
\ 205
/
\ 314
;
\ 304
;
\ 2O3
;
\ 202
1
v 217
;
)20I
10
9
a
7
6
5
4
3
2
'•
<
an
810
908
508
SOT
006
SO3
S04
,602
601
303
30Z
603
\
2IO
209
1
5O9
(
208 j
207 /
(
212 y
\
213 /
(
214 /
.. (
204 1
(
215 /
V
316 /
1
316 /
MM
91
©
PLAN DU BASSIN
ECHELLE : I"«30O'
(61
/5I
42
&
&
Q
CONNECTIVITY DES SOUS-BASSINS.CANIVEAUX,
REGARDS ET CONDUITES DANS LA SIMULATION
DU RUISSELLEMENT DU BASSIN GRAY- HAVEN
GRAY HAVEN DETAILED SEGMENTATION
FIGURE: 2
-------
-a-
o
LE6ENDE- NUMERO OU SOUS-BASSIN
NUMERO OU NOEUD
©
IOO
LUNUUIIl UU liANIVCMU f
LtMlTES OU BASSIN . i. ......
r
j
.••;••$
r"
i
,•/"', ; . . . /?l\%£'-& •'£•''
f 1
!
1
I'1 ' '•'• '' "'' '' ' ''"' " ' •'•' 1"-J
I J
l^«»^ ,,'^,^,.:,-^ftf^,.l,^,:,^
•
iniF
J/j
_^xi'
r
co 1
•••;:••:"--• ^••:-^-V:v;;o-:^j-.;V^;--\::
.--;• . .-:-.. : :-.••:•'.;-.•...•-:.•:-;. \ _.-;-•--;-•.--.-.
r1....
1
^^ ••:,.:-:• :••:••• -\!::-:::>^.
1 '
1
•
PLAN DU BASSIN
CCNCU.E: I'* SOO'
-------
105
TIME (MINUTES)
0 10 20 SO 4O
e-
:
t
"T
1
,
I
I
:
:
-
.„+
::j
:
•i
I
'"i
:
. . . .'
•i- 1 , i
i '
r
u
1»M
r
30 60 70
]
H
1
r
1
u
«m 1
|
!
: |
!
*— « , j
1 i
L i
H Y
JUN
ETOGRAPH
E 14, 1963
! i i • .
L_L
_^^-
__u_^__L_
1
1
z
o
u
Ml
IT
&,
DETAILED, OBSERVED AND SIMPLIFIED
HYDROGRAPHS, GRAY HAVEN, JUNE 14, 1963
FIGURED
-------
106
3.2 Development of the segmentation procedure (Con't)
Largest unit basin
Pursuing the aggregation, two unit modules were lumped and
represented by a single basin and a single gutter; then four,
then eight. A practical limit of 425 acres was reached
when eight unit modules were modeled with a single basin and
a unique gutter; a non-convergence in the runoff block was
experienced when intense storms were simulated. The volume
of water to be routed in the gutter was indeed too large to
satisfy the convergence criteria
-------
107
UNIT
SEGMENT NO. I
GUTTER NO. I
BASIN
SESMENT NO.Z
SUTTER MO. Z
MANHOLE
UNIT BASIN REPRESENTATION
FIGURE: 5
-------
108
3.2 Development of the segmentation procedure (Con't)
Finally this simplified procedure was applied on the Vimont-
Nord catchment, a 2225 acre basin in Laval. The simplified
segmentation consisted of 6 subbasins while the semi-detailed
version contained 23 subbasins. Figure 6 illustrates the
goodness of fit of the resulting simulated hydrographs.
Based on these overall results, the procedure was adopted for
the revision of the drainage masterplan of Laval.
-------
109
TIME (MINUTES)
10 20 30 4O 50 60 70
z
o
K
<
J-
g
«.. 30 •
Ik
u
o
n«j
*
"y
s
i
v :
V
1
:
f*
I
•It'
1
j
i
1 /
!lf
u
-7
J
r
If
U
/
/
t
1
r
i
/
/
j
-
\
\
i
'j
/
j
^
^
p
i
/
^
1
i"=
i
Cj
1
Nj
Ml
I'X
J
mi
^
^1
FS
^
A
V
^
JUN
HYE
1
JkVy
1 \
/
<
E 14, 1963
TOSRAPH
j
^
1
NX
i
>M
S,
1
, /
v^N
/^
\2
V
k
)
K
1
1
i
""l"^*
~»~
.u
*"
O IO 2O 3O 4O SO eo TO 80 9O KX3 110 IZO
TIME (MINUTES!
O
DETAILED SEGMENTATION
2 } SIMPLIFIED SEGMENTATION
VIMONT NORD DETAILED AND
SIMPLIFIED HYDROGRAPHS
FIGURED
-------
110
4.0 DISCUSSION AND CONCLUSION
The simplified segmentation procedure identified for
urban catchment in Laval makes the selection of SWMM for
planning purposes, feasible.
Additional work should be devoted to this question namely
a systematic analysis of the interdependance of At and the
largest unit basin, of the magnitude of At with respect to
the shape of the hydrograph (averaging effect of long At ),
and eventually of the influence of simplified segmentation
upon simulation of urban runoff quality indicators.
The use of the land use data bank in conjunction with the
node data bank makes assessment of a modification in the drai-
nage network or the land use, upon urban runoff a relatively
simple task.
The methodology adopted in Laval uses the available land use
information efficiently and provides flexibility to plan the
drainage network and the urban runoff control alternatives,
at various stages of urbanization of the city of Laval.
-------
Ill
REFERENCES
1. Desjardins, Sauriol & Associes, Note tehcnique, No. 3.
Manuel de 1'Usager du modele DOMOR, schema directeur
d'assainissement, Ville de Laval, octobre 1977, revise
raai 1979.
2. Harley, B.M., Perkins, F.E., and Eagleson, P.S.,
A Modular Distributed Model of Catchment Dynamics, M.I.T.
Dept of Civil Engineering, Ralph M. Parsons Laboratory,
Report no. 133, December 1970.
3. Leclerc, G., Schaake, J.C. Jr., MethodologyFor Assessing
the Potential Import on Urban Development on Urban Runoff
and the Relative Efficiency of Runoff Control Alternatives,
M.I.T. Dept. of Civil Engineering, Ralph M. Parsons
Laboratory, Report no. 167, March 1973.
4. Tucker, L.S., Availability of Rainfall-Runoff Data
for Sewered Drainage Catchments, ASCE Urban Water
Ressources Research Program, Technical Memorandum
No. 8, ASCE, 1969.
5. Desjardins, Sauriol & Associes, Note technique No. 4,
Simulation du Ruissellement, Etude de SWMM, Schema
Directeur d1Assainissement, Ville de Laval, Mai 1978.
-------
112
PREDICTION DE LA QUALITE^DES EAUX DE RUISSELLEMENT
UNE APPROCHE DIFFERENTE
WATER QUALITY PREDICTION FOR URBAN RUNOFF
AN ALTERNATIVE APPROACH
By
Elvidio V. Diniz
Espey, Huston and Associates, Incorporated
2500 Louisiana Boulevard, N.E., Suite 225
Albuquerque, New Mexico 87110
Presented at
Colloque SWMM Users Grouc
Montreal, Quebec, Canada
May 24-25, 1979
-------
113
INTRODUCTION
The accurate prediction of storm water quality
resulting from non-point sources of pollution has recently
become a significant area of interest. In general, storm
water quality is a function of the total pollutant accumul-
ation on all surfaces exposed to rainfall. However, the
rainfall intensity and wash-off capacity of the pollutants
are also important factors. After pollutants have been
dislodged from exposed surfaces, overland flow and channel
hydraulics determine the pollutant concentrations evidenced
at a downstream observation or sampling site. Recent studies1
have indicated that pollution accumulation rates are affected
by several factors including land use, population density,
impervious area, traffic intensity, condition of surface
area, total overland flow length time from last rain, street
sweeping frequency, climate, and season.
A water quality prediciton strategy, as normally
formulated, requires the compilation of a data base, cal-
ibration of a predictive model, simulation of existing and
future conditions, analysis of these predictions, and use of
the results in planning decisions2. Consequently, the data
base provides a foundation on which both the analysis and
subsequent decisions will be carried out. Because of the
importance of valid data to the overall strategy, numerous
investigators have attempted to compile data which could be
used to develop predictive correlations between land use,
watershed characteristics, runoff, and water quality. The
result of these efforts is a very comprehensive data, base
-------
114
for areas throughout the United States. Unfortunately, the
data collection and analysis procedures utilized by each
investigator were rarely uniform and consequently different
sets of data have to be carefully processed to be comparable.
-------
115
INITIAL DEVELOPMENTS
The original attempts to predict stormwater quality
concentrated on regression analyses between watershed aspects
and pollutant concentrations or loads1>3>k. The resulting
equations were site specific and frequently not adaptable to
different areas. A typical example of this type of equation
is presented in Table 1.
The most extensively used deterministic model for
predicting stormwater quality is represented by the following
equation:
where:
= p e "KRAt
4- At Ft8
P. = Total pounds of pollutants on
t
surface of the watershed at the
begining of the time interval
t = the measure of time from the
start of the storm
At * the time interval over which
the integration is occuring
K = the decay constant or rate of
pollutant washoff
R = the actual runoff rate measured
in inches per hour
This equation was originally developed as part of the
EPA Storm Water Management Model (SWMM) and has been sub-
sequently introduced into numerous other water quality
predictive capability models including the U. S. Army Corps
-------
116
TABLE I
EQUATIONS OF BEST FIT FOR THE
THIRTEEN WATER QUALITY CONSTITUENTS*
1) Suspended Solids, mg/£ =
21.55 + 4.36 LOG (Q/A) for Q/A ^0.75
Suspended Solids, mg/£ =
37.83 + 134.7 LOG (Q/A) for Q/A > 0.75
2) Dissolved Solids, mg/£ = 155.02 - 40.25 LOG (Q/A)
3) Ammonia, rug/£ = 0.465 - 0.078 LOG (Q/A)
4) Organic Nitrogen, mg/£ = 0.306 + 0.071 LOG (Q/A)
5) Nitrates, mg/£ = 0.188 + 0.148 LOG (Q/A)
6) Total Phosphorus, mg/£ =
0.0366 - 0.957 LOG (Q/A) for Q/A ^ 0.305
Total Phosphorus, mg/£ =
0.508 - 0.042 LOG (Q/A) for Q/A > 0.305
7) BOD, mg/£ = 4.11 - 0.282 LOG (Q/A)
8) COD, mg/£ = 34.43 + 10.12 LOG (Q/A) for Q/A ^5.6
COD, mg/£ = 46.32 - 5.77 LOG (Q/A) for Q/A > 5.6
9) Fecal Streptococci, 1000 counts/100 ml =
1010 (Q/A)3"21+ for Q/A ^0.22
Fecal Streptococci, 1000 counts/100 ml =
15.35 (Q/A)0'"2 for Q/A > 0.22
10) Total Coliform, 1000 counts/100 ml =17.4 (Fecal Strep)1-US3
11) Fecal Coliform, 1000 counts/100 ml -
0.152 (Total Coliform)0•7S7
12) Total Insecticides, yg/£ = 0.269 +0.11 LOG (Q/A)
13) Total Herbicides, yg/£ = 0.158 + 0.038 LOG (Q/A)
Vc
"All logarithms are base 10.
-------
117
of Engineers Storage Treatment Overflow Runoff Model (STORM)
and the Illinois State Water Survey QUAL - ILLUDAS Model.
In application, this equation is reliable if the initial
concentration of pollutants and the washoff rate is ade-
quately known5. One approach to evaluating the initial
concentration is to use a literature value, most notably
from an American Public Works Association Study of the
Chicago Area6 . The values from that work are based on sur-
face samples taken from various streets in and around the
south side of Chicago. Because only street surfaces were
sampled, the data neglects pollutant accumulation on non-
street areas. In general, experience has shown the Chicago
values rarely give accurate results in SWMM modeling, how-
ever, they are still used when estimating the relative
storm water quality impacts in areas where little or no
recorded quality data exists.
Another method for estimating the initial concen-
tration is by calibration against recorded storm water
quality data. This method requires estimating the initial
concentration and generating a pollutograph for an actual
rainfall and runoff event. The computed pollutograph is
then compared to a recorded pollutograph. If the two vary
significantly, the initial concentration is revised and a
new pollutograph generated. This trial and error procedure
is continued until a satisfactory version has been achieved.
Unfortunately these results are only site specific and there-
fore not easily transferable.
In view of the previous discussion, any alternative
approach would have to be generally applicable in at least a
regional scope, as well as, easily determined from a minimum
data collection and analysis effort. Consequently the new
method, which is the subject of this presentation, was developed.
-------
118
DATA SELECTION AND ANALYSIS
Ten watersheds from 7 different regions of the
United States were chosen for analysis. The data for a
total of 112 runoff events was compiled. Table 2 presents
brief descriptions of the selected watersheds. The analysis
of the data base was specifically directed towards the
development of a predictive model which, with given parameters
of runoff quantity and watershed characteristics, will
predict storm water quality.
To compare the quantity - quality relationships
between different watersheds, and hence different surface
and land use characteristics being studied, it was necessary
to reduce the data to a common format. Because each water-
shed is different, in terms of the drainage area, ground slope,
soil type, and vegetation, it is obvious that different
runoff volume and pollutant quantities would be produced
even if other factors such as land use and rainfall were
equal. For this reason, runoff volumes and pounds of pollu-
tants were converted to inches of runoff and average pounds
of pollutant per acre generated during each storm event.
A previous study7 had identified significant
functional relationships between pollutant loads and runoff
volume. The two quantity-quality relationships which showed
reasonable linear association were logarithmic transforms of:
1. Total runoff volume in inches versus the
total unit load in pounds per acre.
2. Cumulative runoff volume in cubic feet
versus the cumulative pollutant load in
pounds.
-------
119
TABLE 2
DESCRIPTIONS OF WATERSHEDS USED FOR ANALYSIS
Watershed
Area
(ac.)
Percent
Impervious
Number of
Events
Description of Land
1. 39th and Holdrege
Lincoln, Neb. 79.
2. 63rd and Holdrege
Lincoln, Neb.
3. 73th and A
Lincoln, Neb.
4. Third Fork
Durham, N.C.
85.
375.
24.
22.5*
15.0*
1069. 29.0
18 Older family dwellings and
a limited number of multi-
family dwellings. Assume
90% S.F.R. and 10% M.F.R.
14 10 thru 15 year old single
family residential. Assume
1007. S.F.R.
10 Area under development for
single family residential
and consisting of homes less
than 10 years old, homes
under construction, unde-
veloped lots and farm land.
35 Consists of 59% residential
development; 19% commercial
and industrial; 12% public
and institutional lands; and
10% vacant.
5. Panther Branch
Woodlands, Tex.
16050.
1.0*
Primarily undeveloped wood-
lands.
6. Panther Branch
Woodlands, Tex.
21007.
10.0*
Area under development for
single and multi-family res-
idential with related commerc-
ial developments.
7. Hunting Bayou
Houston, Tex.
1976.
23.0
Residential and commercial
with some industrial
developments.
8. K.:i. Clapp Basin
Lubbock, Tex.
223.
30.0
11 Essentially residential de-
velopment with no industrial
development, although one
commercial shipping center with
accompanying parking lot and
two schools are present.
9. Selby Street
San Francisco. Ca. 3400. 22.0
Consists of 77% residential;
2% commercial; 6% industrial;
and 15% vacant land.
10. Alameda Flood-
way. Albq., M.M. 65920
23.0*
Residential and commercial with
undeveloped upper reaches.
*Estimated based on land-use.
-------
120
The reliability of these relationships was statistically
verified by use of the correlation coefficient and its
standard deviation.
For all storm events and pollutants examined, a
reasonably straight line relationship was found to exist.
This relationship was found to hold true on both the total
and cumulative runoff versus pollutant relationships. As a
result of this analysis, it was decided that the total load
runoff relationship would be used as a basis for more
detailed statistical studies.
-------
121
ALTERNATE METHOD DEVELOPMENT
The correlation analysis results indicated that
the logarithmic transforms of total runoff volume in inches
and total pollutant load in pounds per acre yielded the
highest degree of linear correlation with correlation co-
efficients ranging from 0.77 to 1.00 and having a median
value of 0.93. Therefore, best fit equations were generated
for total suspended solids, biochemical oxygen demand, chemical
oxygen demand, Kjeldahl-nitrogen, ammonia nitrogen, total
phosphates and dissolved oxygen for each watershed. This
analysis consisted of first transforming the variables, total
runoff in inches, and total pounds per acre, to their loga-
rithmic values and then performing a linear regression on
the variables. The second correlation between logarithmic
transforms of cumulative pounds of pollutant and cumulative
runoff in cubic feet also proved to be successful. Sartrele
results are presented in Figures 1, 2, and 3.
Both correlation equations are of the form.:
where:
Y = axb
Y = total pounds of pollutant uer acre or
cumulative pounds at time, t
x = total runoff in inches or cumulative runoff
in cubic feet at time, t
a,b = regression coefficients
"a"
Table 3 presents the generated regression coefficients
and "b" for the total volume and Load correlation for the ten
study watersheds. It is postulated that in the case of
-------
122
of Area 10, Alameda Floodway, Albuquerque, New Mexico, the
definition of runoff volume in inches may be inaccurate
because of the non-uniform rainfall patterns typical for
this area. Higher rainfall volumes would elevate these
curves and perhaps reduce the extreme differences shown here.
To evaluate the statistical significance of the
regression analysis, a test utilizing the standard deviation
of the correlation coefficients (s ) was -performed. The
results of this analysis indicate a generally higher degree
of significance, with standard deviations ranging from 0.17
to 1.00 and a median value of 0.85.
In attempting to relate these regression equations
to different surface characteristics found on each watershed,
a number of previously identified parameters were examined.
The only parameter which appeared to produce rational
association with runoff quantity and quality was percent im-
perviousness -of each watershed. In general, as the percent
imperviousness of the watershed increases, the pollutant load
generated by a specific runoff event was found to increase.
This relationship is illustrated in Figure 4, for total sus-
pended solids. Therefore, using percent imperviousness, a
statistical analysis was conducted in an attempt to relate
this parameter to the quantity-quality equations previously
developed. The results of this analysis indicate a strong
correlation between variations in the imperviousness of a
basin and the previously defined coefficient "a". Coefficient
"b", on the other hand, exhibits no such correlation, and
seems to cluster around a value approaching 1.0. From this
analysis, a final set of total runoff quantity and total
quality functions is presented in Table 4. Given the percent
-------
FIGURE 1
TOTAL POLLUTANT LOADINGS AS A FUNCTION OF RUNOFF
100.0
TOTAL SUSPENDED
SOLIDS
o
-i
10.0
1.0
.0!
0.1 1.0
RUNOFF , ins.
x I. 63rd and Holdredge, Lincoln, NB
• 2. Third Fork, Durham, NC
o 3. Selby Street, San Francisco, CA
A 4. Hunting Bayou, Houston, TX
100.0-
10.0
10.0
O.I
COD
U>
O.I 1.0
RUNOFF, ins.
-------
FIGURE 2
WATER QUALITY RELATIONSHIP - COD
60
t/i
a
4.0
0
a.
u>
o
0
STATION P-K>
COD
2.0
3/13/75
12/5/74
4/7/75
4.0 6.0
LOG RUNOFF
8.0
2.0
4.0 6.0
LOG RUNOFF
40
in
Q
2 2.0
o
O
0
_m
— . i . .
STATION M-20
CO!
„ . _
)
y
= l.33x-Z
]
6/30/75,xX:
3/26/74^^'
—
- x
/
S.75 V7'
,/,'
js.
^^t* 5/8/75 "
/ ^4/
11/74
1
,
^
•• -•
.-_.
40
20
0
STATION WS. - i
C
^
3D
5/8/75 x
v^> -
-^**
x^
^ _.
- ~ 1
.02x-2.66 A/
^4
JIS^'
^S^S'
s's
^VV 6/30/75
**
. — _ .
\
2.0
2.0
LOG RUNOFF
4.0 6.0
LOG RUNOFF
-------
FIGURE 3
WATER QUALITY RELATIONSHIP - SUSPENDED SOLIDS
60
§ 4.0
— •)
0
°-
0 2.0
_i
o
STATION P-O
- SuapandoJ Solids
' - ~
....
.. . . .
-
^
— _ _ _
-
*?.
^>/
*,*?//,'
^*
— -
-----
4/7/75 ^
<>•*
<&^
^^A''** 12/5/74
f 3/13/75
V y=.84x-l.42
20 4.0
^«-
'<£-^
>-*"^
~
1
''^^
-^•~~~~
- ..
60
Q
§4.0
O
O-
(JU
y?o
n
STATION P-30
- ~ Suspended Solids
—
— '•
1/18/74 j, ^^
4/22/74 ^'-'^
10/28/74
s
/
^^ ^
/^ ^
,•£,
^/i''
jr t*
^ S
' 3^*
•/
f
S^
S^
._
'<' y=l.046x- .83
x-V 4/7/75
i
--
>^-
6.0 8.0 2.0 4.0 6.0
LOG RUNOFF
6.0
in
Q
0 4.0
Q.
O
O
_J 20
A
STATIC
Suspent
N H-20
lad Solid
s
y
6/30/75 , ^
—
~"/''\
/ '/?'
'*/,S
p'
' 5/8/7
20 40
= l.47x-
-5.45 -
/
2?
svs
&''
' •
jS^.^
4/11/74
•5 J-
L
x.
" '
.. _ _
40
tn
Q
0
Q.
2.0
o
_J
0
STATION
Suspend*!
^
WS.
d Solids
LOG RUNOFF
6/30/75 \s'
^^
^ ^'
6.0 80 20
—
/
/
/ ^^
,' -^S^ '
y - .yox - <
-""""""" ^^^
^\,
S
/^/ 5/5/75
^•^
.. .. ...
I2>;
,-^
. ..
4O 6.0
8.0
LOG RUNOFF
LOG RUNOFF
-------
126
TABLE 3
SUMMARY OF REGRESSION COEFFICIENTS
DERIVED FROM THE QUANTITY/QUALITY ANALYSES
1.
2.
3.
A .
5.
6.
7.
a.
9.
0,
WATERSHED
39ch and Holdrege
Lincoln, Neb.
63rd and HoLdrege
Lincoln, Neb.
78ch and A
Lincoln, Seb.
Third Fork
Durham, M.C.
Pancher Branch
Woodlands, Tex.
Pancher Branch
Woodlands, Tex.
Hunting Bayou
Sous con, Tex.
K..N. Clapp Baa In
Lubbock, Tex.
Selby Street
San Francisco, Ca.
Alaneda Floodway
Albuquerque, N.M.
WMBER
OF TSS
EVENTS a b
18 185.88 1.19
14 156.26 1.14
10 155.75 .97
35 222.66 1.05
4 7.16 .75
5 38. 94 .77
4 49 . 77 1.16
11 763.63 1.54
7 51.19 .72
4
B.O.D.C5) C.O.D, KJELDAHL-N NH.-N TOTAL PO,.
a b ab ab ab ab"
3.69 .55 38,81 .34 —
2.89 ,35 26.24 .64
1.08 .71 15.46 .95
13.15 1.18 35.78 .99 .08 .73 .19 1.08
18.51 1.14 .25 .92 .02 1.01 .02 .99
12 49 89 39 ^9 02 77 02 67
IQQC 07 '17 1TQ 7Q 1£T ? iQ2
27.65 1.70 — — .06 1.41
4.77 .31 26.52 .54 .63 .47 .12 .38 .11 -19
14454 2.56 32.4 1.94 234.4 2.98 30.9 1.94
-------
127
of imperviousness of a basin and the runoff volume in inches,
this function can be used to generate total pollutant
load in pounds per acre for a given runoff event.
The correlation between watershed parameters and
the coefficients of the cumulative runoff-water quality
equations has not been sufficiently investigated at this
time to define acceptable modifications as those conducted
for the total load relationships.
-------
128
FIGURE 4
CHANGE IN THE QUANTITY/QUALITY OF
STORM WATER RUNOFF AND PERCENT IMPERVIOUS
cj
<
CO
o
o
en
a
o
z
Cd
O4
C/3
350
300
250
200
150
100
25 50 75 100 125
TOTAL RUNOFF VOLUME (INCHES)
150
175
@ Percent Imperviousness of Watershed
9 Watershed Number
-------
129
TABLE 4
SUMMARY OF POLLUTANT YIELD (LB/AC)
QUANTITY/QUALITY EQUATIONS
GENERAL EQUATION: Load - a * xb
1) TSS 0
Ibs TSS/acre = I//.1362 + .00384(1) (inches runoff)1*
2) BOD(5) 87
Ibs BOD(5)/acre - 7.3971 (83.0334/1) (inches runoff)'8-
3) COD
Ibs COD/acre = 12. 71e'0315(-I) (inches runoff)'86
4) KJELDAHL-N naaa^n qn
Ibs KJD-N/acre = . 01369e'Ubaau; (inches runoff)'yu
5) NH--N n/1 oo/T\ qs
3 Ibs NH3-N/acre = .2437e'U4Jau; (inches runoff)'"
6) Total Phosphates 89
Ibs Total Phosphates/acre = .09677(1) (inches runoff)'
-------
130
APPLICATION OF ALTERNATE METHOD
This method can be very simply applied to determine
both pollutographs and total loads from a given rainfall run-
off event, using the following procedure:
1. Initially, runoff quantity and rate will
have to be determined by other hydrologic
methods generally available (Rational Equation,
Unit Hydrograph, Kinematic Wave Routing, etc.).
2. The hydrograph generated for either a design
storm or an observed storm can then be con-
verted to a mass flow accumulation curve.
3. The data from this curve can be used in con-
junction with the cumulative runoff versus
cumulative pollutant load curves (Figures 2
and 3) and the corresponding pollutant load
accumulations may be determined. At present,
the site specific or comparable data is required
to determine the pollutant load accumulations;
however, as more data become available and the
regionalization of the data is verified, depend-
ency upon data collection will be reduced.
4. The pollutant load accumulations may then be
converted to pollutant concentrations by means
of the flow volume and the imcremental pollutant
loads. The result of this exercise will be a
pollutograph an example of which is presented in
Figure 5.
-------
131
FIGURE 5
MASS TRANSPORT OF NITRATES AT STATION H-20
STORM OF 5/08/75, HUNTING BAYOU
21
24 3 6
TIME (hrs.)
-------
132
5. If only total pollutant loads are desired,
or to verify the previously computed pollutant
loads, the alternate set of correlations between
total pollutant load and total runoff may be used.
-------
133
FUTURE DIRECTIONS
The relatively simple and facile method to predict
stormwater quality as presented herein should prove most
effective in determing total pollutant loads and associated
pollutographs in most streams and rivers. The linear re-
lationships between runoff and pollutant transport indicate a
direct proportionality between pollutant load and urbanization
as measured by percent imperviousness of the area.
This storm water quality prediction method has re-
ceived only limited testing and verification. Furthermore
only the total load and pollutant relationships have been
correlated to- impervious area on a multi-regional basis. A
similar analysis has to be conducted for the accumulative
runoff and pollutant relationships which currently are only
available for specific regions.
Also, as more data become available in other areas
not previously considered, the correlations developed thus far
must be re-evaluated and verified or revised as neccessary.
However, major revisions are not anticipated.
-------
134
REFERENCES
"Water Quality Management Planning for Urban Runoff,"
EPA Report No. 440/9-75-004, Environmental Protection
Agency, 1975.
Young, G. K. and Terney, C-. F. , "Environmental Data Manage-
ment," ASCE Journal of the Water Resources Planning and
Management Division, 1976.
Colston, N. V., "Characterization and Treatment of Urban
Land Runoff," EPA Report No. 670/2-74-096, Environmental
Protection Agency, 1974.
Winslow, D. E. and Espey, W. H. Jr., "Storm Runoff Analysis
for Residential Settings at the Woodlands, Texas," Tracor,
Inc., 1972.
Huber, ¥. C., Heaney, J, P., Medina, M. A., Peltz, W. A.,
Sheikh, H., and Smith G. F., "Storm Water Management Model,
User's Manual" EPA Report No. 670/2-75-017, Environmental
Protection Agency, 1975.
American Public Works Association, "Water Pollution Aspects
of Urban Runoff," FWPCA Publication No. WP-20-15, United
States Department of the Interior, 1969.
Diniz, E. V. , and Espey, W. H. Jr., "Maximum Utilization of
Water Resources in a Planned Community - Application of the
Storm Water Management Model," Final Report, Environmental
Protection Agency, 1976.
-------
135
WASTEWATER INTERCEPTION ON THE
COMMUNAUTE URBAINE DE MONTREAL TERRITORY
by: Guy Paquin
Group leader engineer
Service d1assainissement des eaux.
CommunautS urbaine de Montreal
Translated from original text in french
by Denis Burroughs, P.E.
-------
136
WASTEWATER INTERCEPTION ON THE
COMMUNAUTE URBAINE DE MONTREAL (CUM) TERRITORY
1- INTRODUCTION
In early 1975, the "Communaute' urbaine de Montreal" started the
construction of a system of interceptors as it engaged itself
in a vast program for the treatment of its wastewaters. The
twenty-nine (29) municipalities which form the CUM regroup
approximately 1,900/000 inhabitants on a territory that totals
about 123,000 acres.
Presently, the developped portion of the CUM territory amounts
to about 90,000 acres, 75% of which are drained by a combined
sewer system, while only 25% of the area is served by a sepa-
rate sewer network.
In the ultimate stage of development, 63% of the area will be
drained via a combined sewer system, and 37% via the separate
or pseudo-separate sewer system.
The different areas of the CUM served by combined, separate or
pseudo-separate sewer systems are shown on figure 1, which
also indicates the location of existent and proposed intercep-
tors .
With the exception of the Ste-Helene and Notre-Dame islands
(which have their own treatment plants) and the industrial sec-
tor of the eastern portion of the island (whose wastewaters
will be treated locally), the North and South Interceptors will
concentrate the wastewater generated by the CUM's 120,000 acre
drainage area, towards one single and gigantic wastewater
treatment plant located at the eastern tip of the island.
-------
TREATMENT PLANT
SCALE 1: 200.OOO
km 2 I 0 2 4 6 8 iO IZ km
TNEATMENT PLANT LOCATION
EXISTING STRUCTURES
PROPOSED STRUCTURES
DRAINAGE AREA LIMITS
HEA SESVEO BT HQTftt OAME-TREAT WENT PLANT
MASTER PUN FOR INTERCEPTORS
-------
138
The existing portion of the North Interceptor, with an approxi-
mate length of 120,000 ft., was rock-tunneled at depths varying
from 50 to 100 ft. At the entrance of the treatment plant's
pumping station, the invert of the sixteen (16) foot diameter horse-
shoe shape interceptor is located at the -50' elevation (M.S.L.),
while at its exit, the water level is at the -4-49* level mark.
The proposed South Interceptor will also be constructed using
the tunneling method. Its diameter will be about nineteen (19)
feet at the entrance of the pumping station, while its invert
will be set at about -100' (M.S.L.).
The maximum flow at the treatment plant's pumping station is
evaluated at 3520 cfs, that is, 1560 cfs from the North Inter-
ceptor,, and 1960 cfs from the South Interceptor.
The North Interceptor is expected to be in service in 1982
when the treatment plant's pumping station is to be completed
and made operational.
For the time being, the construction schedule of the South
Interceptor has not yet been finalized.
The cost of this CUM project, including both interceptors and
the sewage treatment plant, is estimated at 1.2 billion dollars.
2- DESIGN CRITERIA FOR THE NORTH INTERCEPTOR
2.1 Mean Dry-weather flows (M.D.W.F.)
The values chosen for design are given in Table 1. These
values were derived from a water consumption study and
a flow measurement program carried out over the CUM's
entire territory.
-------
TABLE 1
NORTH INTERCEPTOR - MEAN DRY WEATHER PLOW (M.D.W.F.) CALCULATIONS
Land Use
Residential
Industrial
Commercial
Institutional
Parks
Transport
Infiltration
Area
Unit flow (acres)
75 Imp. gal/day - person 30689
0.008 cfs/acre (sector 1)
0.012 cfs/acre (sectors 2 &3)
0.007 cfs/acre (sector 1) 2227
0.009 cfs/acre (sectors 2 & 3)
0.007 cfs/acre (sector 1) -?n?4
0.010 cfs/acre (sectors 2 & 3)
5129
915
0.0006 to 0.015 cfs/acre for 1972 53640
developped areas
0.0016 to 0.004 cfs/acre for 1972
non-developed areas
Ultimate mean
drv weather flow
(cfs)
130.2
132.2
18.6
28.4
-
-
199.0
1 558-4
* Ultimate population: 1,297,839 inhabitants
-------
140
2.2 Maximum flows
The total maximum flow carried by the interceptors is
equal to the sum of the maximum flow from the sanitary
sewer system (separate) and the maximum intercepted flow
from the combined or pseudo-separate sewer system.
The maximum flow from the different sanitary sewer drainage
areas is function of the average flow cumulated along
the length of the interceptor. It can be computed using
the following equation derived from the CUM's flow measure-
ment program.
Q max = 6.40 Qmean °
The maximum flow intercepted from a combined or pseudo-separate
sewer -drainage area is equal to the mean flow of the drain-
age area multiplied by one of the following factors:
3
4.0 x M.D.W.F. for all drainage basins served by pseudo-
separate systems, as well as for all the combined-sewer
basins located west of the Laurentian Autoroute, plus
the small basins which will not be equipped with auto-
matic control regulators.
2.5 x M.D.W.F. for all combined-sewer basins which are
connected to the interceptor east of the Laurentian
Autoroute, and equipped with automatic control regula-
tors (permitting maximum use of the interceptor's
capacity).
The M.D.W.F. multiple of 4.0, chosen for the small combined-
sewer basins and the pseudo-separate basins, corresponds
approximately to the average peak factor for dry weather
flows observed throughout the CUM for drainage basins of
these dimensions. This value was thus chosen to preclude
any overflow during dry-weather periods.
-------
141
This value of 4.0 was also adopted for the remaining com-
bined sewer drainage areas west of the Laurentian Autoroute
to insure adequate protection of the existing water in-
takes of Ville de Laval, located on Riviere des Prairies.
The intercepted flow for the larger combined sewer drain-
age areas equipped with automatic control regulation
devices is 2.5 times the M.D-W.F. This value should
prevent any overflows during the dry-weather periods and
especially during the spring thaw, when the flow reaches
a value of 2 and 3 times the Mean Dry Weather Flow (M.D.W.F,
This M.D.W.F. multiple also considerably reduces
the volume and number of overflow events and their duration
during wet weather periods.
Table 2 shows the influence of the ratio of intercepted
flow to M.D.W.F. on the characteristics of the overflows.
However, due to the nature of regulator control adopted
by the CUM, to the vastness of the territory covered, and
to the off-setting of inlet flows to the interceptor, the
number and importance of the overflows should be consider-
ably less than those expected on a single collector.
A computer simulation study showed that there was no need
to further increase the M.D.W.F. multiple for
combined-sewer drainage areas. As illustrated in Table 3,
any increase in this value has but a negligeable effect
on the volume of the overflow during wet weather.
It is to be clearly noted that these simulations were
made considering a uniform storm over the entire CUM
territory., For isolated storms however, due to the use
of automatic control regulators, the volume of overflow
to the river would be considerably less than that shown in
Table 3.
-------
142
TABLE
OVERFLOW CHARACTERISTICS OF THE MEILLEUR-ATLANTIQUE COLLECTOR
AS A FUNCTION OF INTERCEPTION CAPACITY
Interception capacity Overflow duration
as "n" times
M - D . W . F .
2
2.5
3
i
4
i
5
6
hours
514
265
163
87
0
39
25
% time
5.8
3.0
1.9
1.0
0,4
0.3
number
of events
157
82
62
38
26
19
%• of
wastewater
as overflow
6.
4.
3.
2.
1.
1.
0
2
1
0
4
0
Note: Study made with the measured flow of the Meilleur-Atlantique
collector for the december '67 to december '68 period.
(developped area: 4600 acres)
-------
TABLE
VOLUME OF OVERFLOWS FOR THE ULTIMATE STAGE OF DEVELOPMENT
(million cubic feet)
Storm
Frequency
4/year
2/year
I/year
1/2 years
1/5 years
1/10 years
With regulator control
Interception capacity as "n" times M.D.W.F.
1-.5
1.8
7.8
18.4
37.0
52.2
77.5
2.0
1.2
6.5
15.9
34.1
49.2
74.2
2.5
0.8
5.6
13.9
31.5
46.4
71.3
3..0
0.5
4.8
12.5
29.2
44.0
68.6
Without regulator control
Interception capacity as "n" times M.D.W.F.
1..5
5.9
14.3
25.5
44.6
60.0
85.8
2.0
5.0
13.3
24.4
43.3
58.7
84.9
2.5
4.1
12.1
22.8
41.3
56.4
81.7
3.p
3.. 5
11.1
21.6
39.9
55.6
80.0
Note: Simulations made considering a sanitary flow of 1.0 x M.D.W.F.
-------
144
3- DESIGN OF INTERCEPTION STRUCTURES
Presently, nearly all of the sanitary collectors are discharged
directly,via pumping stations, to the waters surrounding the
island, without any treatment. When the interceptors will be in
operation, the sanitary sewer collectors will be connected directly
to the drop-structures of the interceptor and the present pumping
stations will be abandoned.
Every combined sewer collector will be connected to the intercep-
tor via a regulation structure that will ensure maximum utilization
of the interceptor capacity.
Figure 2 depicts the method of regulation that is to be adopted.
This derivation structure is essentially a chamber built below
the invert of the existing or proposed combined sewer collector.
The dimensions of the chamber were chosen to deviate a certain
portion of the flow of the collector towards the interceptor via
a pipe whose crown is located slightly below the invert of the
collector.
Sluice gates, in the regulation structure, control the flow to
the interceptor. They are actuated by the command signals coming
from the central console located at the treatment plant or from
the local controller. Water then flows into the drop well to
* the interceptor and towards the treatment plant.
To ensure maximum utilization of interceptor capacity, when
available, the regulator structures are individually dimensioned
to allow, on the whole, more flow into the interceptors than the
interception capacity designed for, particularly to cope with iso-
lated or localized storms.
To obtain a certain uniformity in the design and in the construc-
tion details of the interception structures, their dimensions
were established, after scale studies undertaken at the Labora-
toire d'Hydraulique LaSalle, for regulator capacities of 150, 300,
450, 600 and 750 cfs.
-------
145
FIGURE 2.
OPERATING SCHEME FOR THE REGULATION STRUCTURES
-------
146
The nominal capacity for each interception structure was chosen
with regard to the theoretical flows for different storm frequen-
cies the theoretical capacity of the collectors, the capacity
of the interceptor at the point of junction with the collector
and finally the receiving water quality to be maintained
at the point of overflew.
The storm which can be intercepted at a given point of junction,
if the capacity of the interceptor is available, will generally
fluctuate from a frequency of four (4) times a year to one of
once in two (2) years.
Table 4 gives the characteristics of the existing regulators or
those to be constructed before the North Interceptor is to be
made operational.
It should be noted that all. regulators designed with hydraulic
actuators and the Curotte regulator will be equipped with an
auxiliary power plant to assure their continuous functioning
even during power failures.
A study by Dr. Paul Edouard Brunelle, ing., from 1 "Universite" de
Sherbrooke, showed that even when the interceptor flows at maximum
capacity at the plant's pumping station, in the case of a power
failure, the resuming of service with the use of 2 pumps of
223 cfs each, equipped with auxiliary power plants, ensures
adequate control of the system with no excessive surge of water
in the North Interceptor, assuming the gates of the regulation
chambers can be closed within 2 to 3 minutes.
The O'Brien, St-Laurent, 25e Avenue and 54e Avenue regulators
will not be equipped with auxiliary power plants because of their
temporary nature: their collectors are to be rebuilt eventually.
4- CONTROL SYSTEM FOR THE REGULATOR STRUCTURES
gj.--' - ^ *~' "•c r"' ""' 'on of the control system
-------
TABLE
CHARACTERISTICS OF REGULATORS
TO BE IN OPERATION WHEN THE
NORTH INTERCEPTOR GOES INTO SERVICE
Regulator
identification
Bitfield *.*
Autoroute des
Laurentides *
O'Brien **
Meilleur **
St-Laurent **
Auteuil **
Curotte **
Lauzanne *
HSnault *
Salk *
Lanthier *
Langelier *
25e avenue *
54e avenue *
Area
drained
(acres)
.926
9655
1823
3455
816
1643
2883
2005
631
116
362
1781
798
1107
Collector
capacity
(cfs)
proposed
2624
335
2234
251
955
1096
600
400
150
400
950
560
850
Ultimate
M . D . W . F .
(cfs)
8
119
20
46
10
23
43
25
8
1
4
22
9
14
.9
.6
.0
.5
.6
.5
.4
.6
.3
.2
.9
.7
.2
.3
Regulator
Capacity
(cfs)
200
600
150
600
150
300
300
300
300
150
150
450
150
150
Number
and size
of gates
2-4
2-6
1-4
2-6
1-5
2-4
2-4
2-4
2-4
1-4
1-4
2-5
1-5
1-5
'x
'X
' X
'x
1 Y.
1 X
1 X
'x
'X
' X
1 X
'x
'X
'X
4'
6'
4'
6'
5'
4'
4'
4'
4'
4'
4'
5'
5 '
5'
Actuator
type
hydraulic
hydraulic
electric
hydraulic
electric
hydraulic
electric
hydraulic
hydraulic
hydraulic
hydraulic
hydraulic
electric
electric
existing
** proposed 1982
-------
148
The goals to be achieved by the control system advocated
by the CUM are as follows:
to minimize the storm overflows with the optimal use
of the interceptor's capacity and the choice of inter-
ception sites as function of time and wastewater
quality.
to control the flow in the interceptor to ensure the
optimal functioning of the pumping station.
to avoid any backwater flow in the sewer system directly
connected to the interceptor and especially in the case
of a power failure at the pumping station.
The supervision and operation of this control system requires
the acquisition and processing of a large quantity of data.
Table 5 gives the list of the different-types of measures
and points of control needed for each regulator of the
hydraulic type. To these must be added measures of the
water level in the interceptor at different points of
junction with the sanitary sewer collectors, water level
in the river, surveillance of pumping stations and of the
rain gauge network.
A schematic representation of the mode of operation to be
provided for the regulating structures' control system is
presented in figure 3.
A "remote intelligent multiplexer" located at each regula-
tor structure will ensure the regulation-system's opera-
tion, and the transmission by telephone lines of the
different measurement and control signals, to the central
console located at the treatment plant.
The process computer in the treatment plant will maintain
supervision over the whole control system of the North
-------
TABLE 5
MEASURE AND CONTROL ITEMS FOR HYDRAULIC TYPE REGULATORS
Digital Measures
1. Hydraulic Actuators
3. Miscellaneous
cylinder low-pres'sure
reservoir-oil high-temperature
pump filter high differential pressure
. master feeding pump high negative pressure'
start-stop of auxiliary feeding pump engine
oil low-level in reservoir
actuator control mode
2. Generator
start-stop of generator
fuel low-level in reservoir
battery charge
state of power supply
Analog measures
water level in interceptor
water level in collector
water level upstream of gates
positioning of gates
hydraulic fluid pressure in actuators
actuators' hydraulic fluid temperature
local power failure
start-stop of ventilation system
presence of noxious gases in regulation chamber
gas detection system's failure
heating system malfunction
heat detector
presence of water in regulation chamber
no entry in building
manual flushing of bubbler tube
instrumentation air low-pressure
back water gate at the submerged outfall
Commands
. gates* commands
start-stop of generator (emergency power supply)
command of flushing of bubbler tube
-------
150
LOCAL CONTROL
FIGURE 3
SCHEMATIC DIAGRAM OF THE CONTROL
SYSTEM'S OPERATION MODES.
REGULATION
SYSTEM
OPERATOR
CENTRAL COMMAND CENTER
REGULATOR
OPERATOR CONTROL
>RATOR
AUTOMATIC CONTROL
AUTOMATIC
CONTROL
MODEL
SUPERVISORY
COMPUTER
PROCESS
COMPUTER
CENTRAL
CONTROL
TO
REGULATOR
OPERATOR
LEGEND: TELEPHONE LINE
! ! OPTIONAL
-------
151
4 . 2
4.3
Interceptor regulators and will be able if required to
carry out the execution operations depending on the mcde
chosen . A similar system will be provided for the South
Interceptor regulators ,
will assure the surveil-
lance of all the different process computers, and will carry
out memory banking of data, preparation of periodical re-
ports, maintenance schedules, etc.
The first stage of regulator control that will be imple-
mented after the interceptor is made operational will be
•che automatic local control of the regulators. "he mul-
tiplexer will be programmed so as to maintain the gate po
sitioning as a function of the water level in the collec-
tor and in the interceptor to ensure that the permissible
water level in the interceptor is not exceeded.
Eventually, to obviate any difficulties stemming from
failure of the central console, of the computer, of the
communication system, or any other emergency situation ,
when the interceptor is functioning in another mode of
control the commands will automatically be returned to
the regulators local control to avoid forming a pressure
head in the interceptor.
Once the central console and the computer are
installed, and the computer program controlling the opening
and closing of the gates prepared, the operator will be
able to control the regulators by giving the appropriate
instructions to the process computer.
-------
152
Evidently, in the beginning of the operations, this action
will be very limited due to the restricted knowledge of
the system. But after a certain breaking-in period, with
the acquisition of data collected during the operation of
the interceptor, a better knowledge of the behavior of the
system will aid in the preparation of a simulation model
which will facilitate the operator's work.
The model is to anticipate the flow to be intercepted at
each derivation structure, the overflow to the river and
the water level in the interceptor. With these informations,
the operator will then be able to control the gates as a
function of the flow admissible to the interceptor. A
subsequent verification will be made to compare the effec-
tive water levels to the anticipated levels, and corrective
actions will be taken if necessary.
Automatic control of regulators
The simulation model will serve as a basis for the prepara-
tion of the automatic control and operation model.
The automatic control model incorporates in its calculations,
the real values of the water levels in the collectors,
interceptors and regulation chambers. These calculations
will permit the adjustment to real conditions in the inter-
ceptor of the values forecasted by the simulation model,
to obtain finally an automatic real-time control of the
whole system.
The process computer will then be able to operate the gates
depending on the intercepted flow value submitted by the
automatic control model.
Different 'studies now in progress at the CUM and 1'Ecole
Polytechnique de Montreal should help set up this control
-------
153
and operation program during the coming months.
4.5 Raingauge network
To optimize this wastewater interception system, there has
to be a real time reading of the rainfall data so as to
anticipate the flows in the collectors.
The CUM will eventually install fifty (50) raingauges that
will transmit their data by telephone lines to the central
console. Ten (10) tipping bucket raingauges are already
installed on the roofs of different buildings throughout
the CUM and are now operating and transmitting data to the
Service d'assainissement des eaux. The data thus collected
is presently processed and stored in the compute1' and will
help in the study of storm characteristics in the Montreal
Island Area and in the formulation of precise strategies for
the regulation system's control.
5- CONCLUSION
With the help of a control and supervision system based on the use
of computers located at the treatment plant, the CUM will be able
to assure continuous monitoring and control of the interceptors,
regulation structures and the treatment plant pumping station.
The control system advocated will also help to minimize storm
overflows by assuring optimal use of interceptor capacity and by
the choice of interception sites depending on wastewater quality
to be intercepted and the priorities of the overflow points as a
function of water use.
The work undertaken by the CUM will greatly help in improving
the quality of the Montreal Island surrounding waterways.
-------
154
PROCEDURE FOR THE
ESTABLISHMENT OF STATEWIDE
WASTELOAD ALLOCATIONS
By:
Robert A. Hawkins, Bernard F. MaToy, and Joseph L. Pavoni
TenEch Environmental Consultants, Inc.
515 Park Avenue
Louisville, Kentucky 40208
Presented at
SWMM User's Group Meeting
Montreal, Quebec, Canada
May 25, 1979
-------
155
INTRODUCTION
State regulatory agencies are faced with the problem of developing
effluent limitations for industrial, semi-public and municipal dis-
chargers which assure that water quality standards are met and that
the economic burden of environmental protection is shared as equitably
as possible among all dischargers to a stream segment. In the past,
emphasis has been placed on the development of sound technical water
quality modeling procedures and the calibration and verification of
the water quality models.
The purpose of this paper is to present the minimum treatment
requirements which have been established for dischargers by the
United States Environmental Protection Agency and a procedure which
allows regulatory agencies to decide where limited resources for water
quality sampling and field investigation should be placed in developing
defensible waste load allocations.
INDUSTRIAL POINT SOURCE ALLOCATION PROCEDURES
Waste load allocation analysis procedures for industrial dis-
chargers must be developed in accordance with the Clean Water Act
of 1977, Public Law 95-217. This law establishes revised criteria
for the determination of effluent limitations for industrial dis-
chargers. Under the Clean Water Act, pollutants discharged by in-
dustrial point sources are divided into three broad classifications:
conventional pollutants, toxic pollutants, and nonconventional
pollutants. The act defines conventional pollutants as biological
oxygen demanding substances, suspended solids, fecal coliforms, and
pH. All industrial treatment works are required to implement best
-------
156
conventional pollutant control technology (BCT) for conventional
pollutants by July 1, 1984. Effluent limitations in excess of BCT
are required in situations where BCT controls will result in con-
tinued violations of water quality standards.
Effluent limitations for toxic substances are determined by
the implementation of best available technology economically
achievable (BAT). Two separate lists of toxic substances with dif-
ferent implementation schedules are specified by the Act. All toxic
pollutants referred to in Table 1 of Committee Print Number 95-30
of the Committee on Public Works and Transportation of the House of
Representatives must meet BAT effluent limitations by July 1, 1984.
This list is presented in Table 1 of this report. Any pollutants
added to this list by the Administrator of the U.S. Environmental
Protection Agency require the implementation of BAT controls no later
than three years after the time these pollutants are added to the
list. Any pollutants which are not considered to be conventional
pollutants or which do not appear on the toxic substances list, are
referred to as nonconventional pollutants and are required to meet
BAT effluent limitations within three years of the time limitations
are established, or July 1, 1984 whichever is later.
The Environmental Protection Agency is currently conducting a
review of BAT effluent limitations for those industries where BAT
effluent limitations have previously been established, to determine
what changes are necessary to conform to the BCT criteria. Table 2
lists the schedule which EPA has established to publish BAT effluent
guidelines for toxic substances in individual industries. Any in-
dustries which currently lack BAT guidelines for conventional
pollutants will receive BCT guidelines according to the same schedule.
-------
157
TABLE I: TOXIC POLLUTANTS LIST PURSUANT TO THE
CLEAN WATER ACT OF 1977 (PUBLIC LAW 95-217}
1. Acenaphthene.
2. Acrolein.
3. Acrylonitrile.
4. Aldrin/Dieldrin.1
5. Antimony and compounds.2
6. Arsenic and compounds.
7. Asbestos.
8. Benzene.
9. Benzidine.1
10. Beryllium and compounds.
11. Cadmium and compounds.
12. Carbon tetrachloride.
13. Chlordane (technical mixture and
metabolities).
14. Chlorinated benzenes (other than
dichlorobenzenes).
15. Chlorinated ethanes (including 1,2-
dichloroethane, 1,1,1-trichloroethane, and
hexachloroethane).
16. Chloroalkyl ethers (chloromethyl,
chloroethyl, and mixed ethers).
17. Chlorinated naphthalene.
18. Chlorinated phenols (other than those
listed elsewhere; includes trichlorophenols
and chlorinated cresols).
19. Chloroform.
20. 2-chlorophenol.
21. Chromium and compounds.
22. Copper and compounds.
23. Cyanides.
24. DOT and metabolites.1
25. Oichlorobenzenes (1,2-,1,3-, and 1,4-
dichlorobenzenes).
26. Oichlorobenzidine.
27. Dichloroethylenes (1,1-, and 1,2-dich-
loroethylene).
28. 2,4-dichlorophenol.
29. Oichloropropane and dichloropropene.
30. 2,4-dimethylphenol.
31. Dinitrotoluene.
32. Diphenylhydrazine.
33. Endosulfan and metabolites.
34. Endrin and metabolites.1
35. Ethylbenzene.
36. Fluoranthene.
40.
41.
42.
43.
37. Haloethers (other than those listed
elsewhere; includes chlorophenylphenyl
ethers, bromophenylphenyl ether, bis
(dichloroisopropyl) ether, bis-(chloroeth-
oxy) methane and polychlorinated diphenyl
ethers).
38. Halomethanes (other than those
listed elsewhere; includes methylene
chloride methylchloride, methylbromide,
bromoform, dichlorobromomethane, tri-
chlorofluoromethane, dichlorodifluro-
me thane).
39. Heptachlor and metabolites.
Hexachlorobutadiene.
Hexachlorocyclohexane (all isomers).
Hexachlorocyclopentadi ene.
Isophorone.
44. Lead and compounds.
45. Mercury and compounds.
46. Naphthalene.
47. Nickel and compounds.
48. Nitrobenzene.
49. Nitrophenols (including 2,4-dinitro-
phenol, dinitrocresol).
50. Nitrosamines.
51. Pentachlorophenol
Phenol.
Phthalate esters.
Polychlorinated biphenyls (PCBs).1
55. Polynuclear aromatic hydrocarbons
(including benzanthracenes, benzopyrenes,
benzofluoranthene, chrysenes, dibenzanth-
racenes, and indenopyrenes).
56. Selenium and compounds.
57. Silver and compounds.
58. 2,3,7,8-tetrachlorodibenzo-p-dioxin
(TCDD).
59. Tetrachloroethylene.
Thallium and compounds.
Toluene.
Toxaphene.1
Trichloroethylene.
Vinyl chloride.
65. Zinc and compounds.
52.
53.
54.
60.
61.
62.
63.
64.
'Effluent standard promulgated (40 CFR Part 129).
2The term "compounds" shall include organic and inorganic compounds.
-------
15£
TABLE 2: PROPOSED DATES FOR PROMULGATION
OF BAT EFFLUENT LIMITATIONS
Group 1 (September 30, 1978 (proposed). March 31, 1979 (final})
Timber Products
Steam Electric Power Plants
Leather Tanning & Finishing
Iron & Steel Manufacturing
Petroleum Refining
Group 2 (December 31, 1979 (proposed), June 30. 1979 (final})
Nonferrous Metals Manufacturing
Paving & Roofing Materials (Tars and Asphalt)
Paint & Ink Formulation and Printing
Ore Mining & Dressing
Coal Mining
Group 3 (March 31, 1979 {proposed}, September 30, 1979 {final})
Organic Chemicals Manufacturing
Inorganic Chemicals Manufacturing
Textile Mills
Plastics & Synthetic Materials Manufacturing
Pulp & Paperboard Products
Rubber Processing
Group 4 (June 30, 1979 (proposed}, December 31, 1979 {final})
Soaps & Detergent Manufacturing
Auto & Other Laundries (Industrial Laundries)
Miscellaneous Chemicals
Pesticide Manufacturing
Photographic Products
Gums & Woods (Glue Making)
Pharmaceuticals
Explosives
Adhesives & Sealants
Machinery & Mechanical Products Manufacturing
Battery Manufacturing
Plastics Manufacturing
Foundries
Coil Coatings
Porcelain & Enameling
Aluminum
Copper
Electronics
Shipbuilding
Mechanical Products (Metal Fabrication)
Electroplating
-------
159
In some cases, BAT guidelines will be published for nonconventiona'
pollutants at the same time that BAT guidelines are published for toxic
pollutants. In other cases, BAT guidelines for nonconventional
pollutants will be postponed until a later time. However, as was the
case when BPT guidelines were published, water quality standards must
be met, even if effluent limitations more stringent than the appro-
priate BCT or BAT effluent limitations are required.
Until the BCT and BAT guidelines are issued, state regulatory
agencies are faced with the problem of issuing NPDES permits for
industries whose existing permits expire during 1978 or 1979. For
those industries for which BCT or BAT guidelines have not been pub-
lished, it is recommended that waste load allocations be developed
based on meeting in-stream water quality standards. As a minimum,
all industrial dischargers should be required to comply with BPT
(current permit) effluent limitations. For those parameters for
which no water quality standards have been established, and for which
BAT or BCT guidelines have not been published, waste load allocation
analysis should not be attempted.
SOURCE ALLOCATION PROCEDURES.
Municipal and semi -public dischargers are required to implement,
as a minimum, secondary treatment for all discharges. Where more
stringent effluent limitations are required to insure compliance
with water quality standards, we have recommended that discrete treat-
ment levels be utilized, similar to those presented in Table 3. These
treatment levels, listed in order of increasing stringency, reflect
current wastewater treatment technology, are applicable throughout
the planning period (i.e., 20 years), and are capable of being achieved
with common unit process combinations. The use of specific treatment
-------
160
TABLE 3: RECOMMENDED TREATMENT LEVELS FOR MUNICIPAL AND
SEMI-PUBLIC DISCHARGERS
Treatment Removal Requirements
Effluent
Quality
A
B
C*
D*
E
f
G*
H*
I*
J*
K*
L*
M
N
0
P
Q
R
BOD
(* Removal )
85-90
85-90
85-90
85-90
95
95
85-90
85-90
85-90
85-90
95
95
85-90
85-90
95
95
97.5
97.5
Nitrification
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Effluent Concentration (Thirty Day Mean)
BODc
(mg/T)
30
30
20
20
10
10
25
25
20
20
10
10
20
20
10
10
5
5
NHVN
mill
15
15
15
15
15
15
8.5
8.5
8.5
8.5
8.5
8.5
2
2
2
2
2
2
Dissolved
Oxygen
(rnq/1 )
4.0
6.0
4.0
6.0
4.0
6.0
4.0
6.0
4.0
6.0
4.0
6.0
4.0
6.0
4.0
6.0
4.0
6.0
Effluent
Reaeration
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
UOD**
(mg/1 )
113
113
98
98
83
83
76
76
68
68
53
53
39
39
24
24
17
17
*These treatment levels represent a split flow treatment scheme.
**UOD = (1.5 x BOD5) + (4.5 x NH3-N)
-------
161
levels does not dictate a particular treatment process or combination
of processes since the treatment levels can be achieved with a variety
of commonly used unit processes as indicated in Table 4.
The following sections of this paper incorporate the industrial
point source allocation requirements and discrete treatment levels for
municipal and semi-public dischargers into an overall waste load allo-
cation procedure.
WASTE LOAD ALLOCATION METHODOLOGY
The development of a methodology for establishing waste load alloca-
tions involves consideration of several aspects of river basin manage-
ment. A flow diagram of a general waste load allocation process is
provided in Figure 1. Considerations related to this process which
must be addressed include:
• Discharger flow regimes to be used.
0 Allocation of reserve assimilative capacity for future
discharges.
• Treatment levels to be examined for municipal and semi-
public dischargers located on water quality limited
segments.
• Industrial effluent limitations
• Methods for allocating assimilative capacity among multiple
dischargers.
• Techniques for estimating model input data.
Based upon an evaluation of the results of a recent study, the general
waste load allocation process has been adapted into a specific metho-
dology designed to meet water quality standards through implementation
of the least stringent treatment levels by industrial, municipal, and
senri-public dischargers. Two major objectives to be met in achieving
this goal are to:
• Provide an equitable allocation of assimilative capacity
among existing dischargers.
-------
162
TABLE 4 ALTERNATIVE WASTEWATER TREATMENT SCHEMES
BODs NHs-N
Removal Removal Alternative Treatment Schemes
85-90? No 1. AS + CHL
85-90% Yes 1. AS + BN + CHL
2. EA + CHL
952 No 1. AS + MMF + CHL
95% No 1. AS + LCI + MMF + CHL
2. LCI + MMF + AC + CHL
95% Yes 1. AS + BN + MMF + CHL
2. AS + LC2 + AST + MMF + CHL
3. LC2 + AST + MMF + AC + CHL
95% Yes 1. EA + MMF + CHL
2. AS + B.ND + AL + CHL
99% Yes 1. AS + BND + AL + MMF + AC + CHL
2. AS + LC2 -i- AST + MMF + AC + CHL
3. LC2 + AST + MMF + AC + CHL
Effluent Concentrations (mg/1'
BODs NH3-N UOD
30
20
20
10
5
5
5
5
5
10
10
2
2
2
15
2
2
15
15
15
2
2
2
2
2
2
2
2
113
39
39
83
75
75
17
17
17
24
24
12
12
12
Abbreviations:
AC - Activated Carbon
AL - Alum in Act. Sludge
AS - Activated Sludge
AST - Ammonia Stripping
BN - Biological Nitrification
BND - Biological Nitrification and Denitrification
CHL - Chlorination
EA - Extended Aeration
LCI - Lime Coagulation - 1 stage
LC2 - Lime Coagulation - 2 stage
MMF - Mixed Media Filtration
-------
163
FIGURE 1 : WASTE LOAD ALLOCATION PROCESS
Define Hatershed
I
Establish Water Quality
Standards
I
Examine Existing Water Duality
Data
I
Locate Point Source Dischargers
Determine Pollution Loadings
at BAT/BCT and BPT
for Specified Discharge Rates
Run Model at Critical Low Flow
Conditions With BAT/BCT and Bpf
Violations?-
Reserve _
Capacity?
_Set Treatment
Requirements
Select Water Quality Model
I
Establish Critical Low Flow
Conditions and Determine
Model Input Data
Calibrate X Verify
Model
Collect Water
• Quality Data
if Necessary
Sensitivitv Analysis
of Selected Model
Yes Define Water Quality
—»- Limited Segments
Select Higher Treatment
Levels For Municioal &
Semi-Public Dischargers
j
Reserve Run Model Using Various
Capacity?""Treatment Alternatives
Select Equitable and/or Cost
Effective Alternative Which
Insures Compliance with Water
Quality Standards
-------
164
• Provide a means for accommodating new dischargers and
future flow increases from existing dischargers.
The recommended waste load allocation procedure is outlined in
Figure 2. This procedure was synthesized from a recent study in
which model input parameters related to reaeration and deoxygenation
were determined for stream segments on several streams.
Several significant aspects of the recommended procedure warrant
further discussion. Using stream sampling results obtained in a re-
cent study, an analysis of the initial recommended values for the
deoxygenation rate constants in relation to other values determined
through model calibration indicated that:
* Ninety (90) percent of the KI values determined for stream
reaches downstream from dischargers were less than or equal
to 0.30 (at 200C).
t Seventy-five (75) percent of the Kh values determined for
stream reaches downstream from dischargers were less than
or equal to 0.40 (at 2QOC).
Since these values appear to be conservative, they may result in
overly restrictive effluent limitations. Therefore, consideration
of model output using lower values may be warranted as indicated
in the recommended waste load allocation procedure.
The minimum dissolved oxygen criteria associated with path A
(i.e., 6.0 mg/1) was selected on the basis of the results of a sen-
sitivity analysis of a calibrated model and past experience in water
quality modeling. It appears that the achievement of this minimum
dissolved oxygen level, using the initial values of the deoxygenation
rate constants, will insure that water quality standards are main-
tained at the discharge levels in question.
The primary incentive for following path B is to justify the
necessity of requiring higher treatment levels. The deoxygenation
rate constants associated with this path were selected on the basis
-------
FIGURE 2: RECOMMENDED PROCEDURE FOR DEVELOPING WASTE LOAD ALLOCATIONS ON UNCALIBRATED STREAM SEGMENTS
Estimate Year 2000 Flows and
Assume Minimum Required Treatment Levels:
BAT/BCT for Industries
BPT for Municipal/Semi-Public Dischargers
0
•Select Next Higher Treatment Level
For Municipal/Semi-Public
Dischargers and Reduce Industrial
Loadings Proportionately
•Simulate Water Quality Using*:
KI = 0.30 (@ 20°C)
Kn = 0.40 (@ 200C)
D.O. min > 6.0 mg/1
Selected Treatment
Level is Adequate
0
D.O. min < 5.0 mg/1
-3
Yes
Simulate water Quality Using:
K! = 0.20 (e 20°c)
Kn = 0.25 (@ 20°C)
©
D.O. min < 5.0 mg/1?
No
5.0 mg/1 < D.O. min < 6.0 mg/1
Simulate Water Quality Using:
Kj = 0.60 (@ 20°C)
Kn = 0.50 (@ 20°C)
T No
D.O. min < 5.0 mg/1? —»-
- Sampling Program is Justified
for Model Calibration Purposes
Treatment Level(s) May Or Hay
Not Be Used To Develop Waste
Load Allocations.
Selected Treatment
Level is Adequate
Yes
*Use annual Qjg-7 f1ow for winter conditions and seasonal Qjo-7 flow f°r summer conditions.
-------
166
of an evaluation of calibration results and other reported values.
These values are representative of the lowest values which are likely
to occur on the streams which were sampled. Although lower values
were found to occur on some streams, primarily during summer con-
ditions, these values appear to represent a net change in carbonaceous
oxygen demand resulting from oxidation and algal contributions.
The primary incentive for following path C is to insure that the
selected treatment levels are adequate to maintain water quality
standards. The debxygenation rate constants associated with this
path were selected on the basis of an evaluation of calibration re-
sults and other reported values. Although higher values were de-
termined for streams during winter-ice cover conditions, these values
may not be reliable since simulated travel times may have contained
significant errors under these conditions. It appears that the de-
oxygenation rate constant values associated with this path are the
highest values which are likely to occur on the streams which were
studied.
It is recommended that waste load allocations be based on
estimated year 2000 flows for all dischargers to provide reserve
assimilative capacity for projected future growth in the planning
period. These estimated flows should include all municipalities
which may construct collection and disposal systems during the
twenty year planning period. This will insure reserve assimilative
capacity in the receiving streams. Although waste load allocations
can only be predicted and implemented for the next permit period
due to possible future technology changes, water quality standards
changes, etc., they are acceptable from a facilities planning stand-
point. In addition to providing reserve capacity for projected
-------
growth, the use of year 2000 flows rather than 1983 flows is recom-
mended for several additional reasons. First, since any plant con-
structed or upgraded in the near future will be required to treat
year 2000 waste flows (i.e., the twenty year planning period), plant
design should be based upon year 2000 population projections.
Secondly, our experience in waste load allocation studies has indi-
cated that treatment requirements based upon 1983 flows seldom differ
from treatment requirements based upon year 2000 flows when using a
treatment level approach. In those instances when treatment levels
may vary, it has usually been found that it is more cost-effective
to design the plant for the more stringent treatment level rather
than upgrade the facility at some point during the twenty year plan-
ning period. The recommended municipal/semi-public discharger treat-
ment levels were previously presented in Table 3.
Initially, a computer simulation of water quality should be made
using BAT/BCT for industrial dischargers and BPT (i.e., Treatment
Level A) for municipal and semi-public dischargers at projected year
2000 flows as inidcated in Figure 2. If no water quality violations
occur (i.e., path A) then treatment requirements should be set accord-
ing to BAT/BCT and BPT guidelines.
If a violation of water quality standards occurs (i.e., path B)
in the case of a single discharger (municipal or semi-public), suc-
cessively higher treatment levels (i.e., Treatment Levels B through
R from Table 3) should be modeled using the deoxygenation rate con-
stants indicated in Figure 2 until no violations occur (path C or A).
Waste load allocations are then established based on the appropriate
treatment level required to maintain water quality standards. In
such cases, land application and/or controlled discharge of
-------
168
secondary effluent (i.e., Treatment Level A) should be evaluated on
a cost-effective basis as an alternative to providing more stringent
treatment requirements.
If implementation of the controlled discharge option appears
feasible, it will be necessary to determine the streamflow conditions
during which discharges will be permitted. Assuming that a detention
lagoon with six months of storage capacity is provided, the minimum
rate of discharge to the stream will be twice the average daily
wastewater flow. Since it may be possible to discharge over a shorter
time period (three months for example) a correspondingly higher mini-
mum rate of discharge would be required. In order to insure that
water quality standards are maintained during the period of discharge,
the Q]Q-7 flow will have to be determined for that period. These
streamflow and wastewater discharge rates, along with the effluent
characteristics associated with Treatment Level A should then be used
to simulate water quality with the initial values of the deoxygenation
rate constants (i.e., Kj = 0.30 and Kn = 0.40 at 20°C). If no water
quality violations occur, then the controlled discharge option is a
feasible alternative for maintaining water quality standards. Since
this option will usually be considered only for very small dischargers,
it does not appear to be necessary to repeat the entire recommended
waste load allocation procedure depicted in Figure 4 when considering
the controlled discharge option.
If a single industrial discharger causes water quality standards
violations with BAT/BCT, then treatment must be upgraded as necessary
to comply with water quality standards. It is recommended that this
be accomplished by reducing the BODU and N^-N levels by the same
proportion as indicated by the municipal/semi-pub!ic discharger treat-
ment levels contained in Table 3.
-------
169
In situations where the recommended procedure indicates that a
water quality sampling program is justified, it should be left up
to the discretion of the regulatory agency as to whether the treat-
ment levels resulting in this destination should be used to establish
waste load allocations. In such situations consideration should be
given to the magnitude of the minimum dissolved oxygen levels ob-
tained. For example, if in following path C, a minimum dissolved
oxygen level of 4.90 mg/1 is obtained with the second simulation
(i.e., at the higher deoxygenation rate constants), the treatment
levels may be considered adequate for waste load allocation purposes.
Particular attention should be given to situations in which
modeling indicates that implementation of BAT/BCT and BPT is not
adequate to maintain water quality standards on stream segments
containing multiple dischargers. It may be possible to establish
equitable waste load allocations by requiring one or more dischargers
to provide more stringent treatment than other dischargers. It is
recommended that this possibility be investigated only in cases
where the recommended procedure leads to either path A or path C-l
with uniform treatment levels for all dischargers. In these cases,
the procedure outlined in Figure 3 should be followed. This pro-
cedure which is similar to that in Figure 2, is intended to indicate
those treatment level combinations which are adequate to maintain
water quality standards.
As indicated in Figure 3, it is recommended that treatment
levels be downgraded for the smaller dischargers rather than the
larger dischargers. The objective of this approach is to require
the largest dischargers to provide the most stringent treatment,
since this usually results in the most equitalbe, though not
-------
FIGURE 3: MULTIPLE DISCHARGER PROCEDURE
Reduce Treatment Requirements For
Smaller Discharqer(s)
Simulate Water Quality Using:
KI = 0.30 (@ 20°C)
Kn = 0.40 (@ 200C)
D.O. min > 6.0 mg/1
Selected Treatment
Levels Are Adeauate
D.O. min < 5.0 mg/1
Simulate water Quality Using:-*-
KI = 0.20 (@ 20°C)
Kn = 0.25 ((? 20°C)
D.O. min < 5.0 mg/1?
5.0 mg/1 <_ D.O. min < 6.0 mg/1
Simulate Water Quality Using:
K]_ = 0.60 (@ 20°C)
Kn = 0.50 (@ 20°C)
Yes
No
D.O. min < 5.0 mg/1?
Yes
No
Selected Treatment
Levels are Not
Adequate
Sampling Program is Justified
for Model Calibration Purposes
Treatment Levels May or May
Not Be Used To Develop Waste
Load Allocations.
Selected Treatment
Levels Are Adequate
-------
171
always the most cost-effective alternative. In cases where water
quality standards will be maintained by requiring either the largest
discharger, or one or more of the smaller dischargers to provide a
higher level of treatment, consideration must be given to the greater
operation and maintenance capabilities of the largest discharger.
In cases such as this, where multiple municipal dischargers are in
close proximity and it is not cost-effective and/or equitable to
provide more stringent treatment for the largest discharger, con-
sideration should be given to cost sharing between the large and
small dischargers to obtain an equitable solution. Again, land
application and/or controlled discharge should be evaluated on a
cost-effective basis as an alternative to requiring higher treat-
ment levels.
Implementation of the recommended procedure for multiple dis-
chargers in many cases will result in industrial dischargers having
less stringent treatment requirements than municipalities since they
are generally smaller dischargers. However, in most situations the
industrial discharges of deoxygenating wastes have a much less
significant impact on the dissolved oxygen levels of receiving
streams than the municipal discharges. Since this may sometimes
represent an undesirable situation, alternative treatment level
combinations may be evaluated according to the procedure out-
lined in Figure 3.
As in the case of single dischargers, it should be left to
the discretion of the state regulatory agency as to whether treat-
ment levels which indicate that a sampling program is justified
should be used to establish waste load allocations.
-------
172
With regard to new dischargers, the recommended waste load allo-
cation procedure provides reserve assimilative capacity in two ways:
• The use of treatment levels provides reserve capacity since
they are usually more stringent than waht are actually re-
quired to maintain water quality standards.
• The use of year 2000 flows for existing and anticipated
dischargers provides reserve capacity for projected future
growth in the planning period.
In cases of proposed new dischargers, the recommended waste load
allocation procedure should be repeated to determine the treatment
level required for the new discharger to maintain water quality
standards. New municipal and semi-public dischargers will not
usually be significant since flows are usually small. These types
of discharges should be accounted for in the year 2000 flow pro-
jections. In most cases, the only significant new dischargers in
a watershed will be industries, which should consider the avail-
ability of assimilative capacity in the evaluation of site selection
alternatives. The waste load allocations for existing dischargers
should not be changed during a planning period as a result of new
dischargers to a stream segment. Waste load allocations for new
dischargers should be based on whatever treatment level is re-
quired to maintain water quality standards for the duration of the
planning period. This is the most suitable and cost-effective
method for accommodating new dischargers for waste load allocation
purposes.
CONCLUSION
Implementation of the recommended approach will enable a state
regulatory agency to reliably determine the least stringent treat-
ment levels which are capable of maintaining water quality standards.
-------
173
In addition, this approach enables the systematic identification
of stream segments on which water quality sampling programs should
be conducted. In consideration of the state-of-the-art of water
quality modeling, the recommended approach will enable state regu-
latory agencies to make the best possible use of available data
and appropriate water quality models for the purpose of developing
waste load allocations.
-------
174
GUM'S SIMPLIFIED HYDROGRAPH METHOD
FOR DETERMINING URBAN RUNOFF
by C. Mitci*, H.N. Nguyen Van*, M. Osseyrane*
INTRODUCTION
The "Communaute urbaine de Montreal" (CUM) is a metropolitan ser-
vices special district under whose jurisdiction the "Service d'as-
sainissement des eaux" is responsible for the design, construction
and operation of the water pollution control works on the island of
Montreal, which consists mainly of two major interceptors feeding
a single sewage treatment plant.
Since 1972 CUM started to investigate various methods of runoff
determination capable of providing outflow hydrographs from main
sewage collectors, which could be used in the rational design of
the intercepting system and the real-time control of combined
sewer overflows.
The main objective of the conducted study was to find a simple
method of runoff determination which would:
simulate the rainfall - runoff process;
be sufficiently accurate and yet feasible for operating
within the computational limitations of real-time con-
trol systems;
be easily implemented on a minicomputer;
require few and easily obtainable input data;
not be significantly affected by the lumping of
subbasins;
become a reliable tool for urban drainage studies after
calibration and verification.
Respectively, engineer in charge of hydraulics, project engineer
and design engineer, Service d1assainissement des eaux, Com-
munaute urbaine de Montreal.
-------
175
REVIEW OF THE CHARACTERISTICS OF EXISTING MODELS
A brief review of the best known methods and computer-based
mathematical models developed in the last 15 years for computing
storm runoff from urban areas (RRL, SWMM, ILLUDAS...).[. 1~\ , and
used in practice in their original, modified or improved form
revealed the following common features:
They are based on a "micro-approach" of the runoff phenomenon,
which means that the main hydrological and hydraulic components
of the rainfall-runoff process are analysed in detail, in partic-
ular the abstractions from rainfall and the hydraulic stages of
overland, gutter and conduit flows. The process often includes
the synchronizing, routing and combining of hydrographs from
upstream points of entry to downstream points of concentration,
both in the lateral and main sewers of the drainage subbasins.
Such an approach,however, requires considerable detail and sub-
sequent effort in the preparation of input data,, unless large
lumping in catchment discretization is applied, in which case,
the lumped hydraulic components (overland width, slope, gutter
and conduits) do not truly represent the physical aspects of the
catchment and thus, the sophisticated approach is somewhat al-
tered by the rough representation of the basin.
On the other hand, the recent conceptual models (linear systems,
kinematic models) Q , £J and the somewhat older but simpler
synthetic unit hydrograph methods [2] derived mainly from natural
watersheds and in which the physiographic properties of the
basin are related to the characteristics of the resulting runoff
hydrograph, generally do not have a well established antecedent
in urban drainage practice.
GUM'S APPROACH
The above considerations and the main objective of the CUM to
-------
176
find a simple hydrograph method led to the investigation of the
possibility of using a "macro-approach" to the study of the rainfall-
runoff events.
In this approach, all the hydrological components of runoff (infil-
tration and depression storage) are grouped together through the
use of a single parameter, while the hydraulic components (over-
land, gutter and conduit flow) are lumped together by using a direct
correlation between excess rainfall and runoff as defined by the
synthetic unit hydrograph theory |2 .*
However, some preliminary attempts to apply such an approach to
the hydrological components, by employing an overall "runoff
coefficient" varying with time 5 , 7]and accounting not only for
the type of the surface but also for the observed time delay between
the start of rainfall and the beginning of runoff 6 I, did not give
satisfactory results when applied to a selected pilot basin. This
was apparently due to the fact that the effects of storm pattern
on depression storage and on infiltration are difficult to be
represented by an overall coefficient.
Detailed investigation of models using the "micro-approach" reveals
that the complicated and time consuming part of the calculations of
the runoff process lies mainly in the hydraulic phase, while the
hydrological phase can be easily modelled. Consequently, a "macro-
approach" was adopted for the hydraulic phase and a "micro-approach"
maintained for the hydrological one.
Abstractions from rainfall to determine the rainfall excess of
overland flow supply rate are assumed to be mainly due to infiltra-
tion, depression storage and some minor losses resulting from
evaporation and. interception in the case of pervious surfaces and
from evaporation, pools of water due to false slopes, losses due
to surface cracks in the case of impervious areas.
A similar approach is used in the "Battelle Urban Wastewater
Management Model" [^IJ as outlined in an article by Brand-
Stetter et al., published after the drafting of the CUM's
first article on the same subject m.
-------
177
Depression storage supply is disregarded in the case of a signifi-
cant antecedent precipitation.
Overland flow and depression storage supply from pervious surfaces
is assumed to start when the antecedent mass rainfall equals the
antecedent mass of infiltration jV] .
The rainfall excess obtained by subtracting the above losses from
the recorded rainfall is directly transformed to a hydrograph by
using the synthetic unitgraph concept developed by, among others,
the USSCS [2] .
The extensive technical literature on the subject of the time of
occurrence of peak discharge Tp [JLOJ and the overall duration of
the hydrograph reveals that this time depends on all the hydro-
meteorological characteristics of the rainstorm (areal and temporal
distribution) and the physiographic properties of the drainage basin
such as impervious surface, size, shape, a representative length,
slope, surface drainage pattern, storage factor and a representative
runoff resistance.
Since many of the existing empirical formulas I10J diverge conside-
rably on the effects of the various influential parameters, it appears
justified to use a simple triangular hydrograph, such as the one
developed by USSCS J_2J , and illustrated in Fig. 1. In this hydro-
graph, the time of peak surface runoff depends on the location of
the centroid of the excess rainfall hyetograph and on a characteris-
tic factor of the basin which is expressed by £Tc. In this expres-
sion, the components of the familiar time of concentration Tc (inlet
and travel time) reflect mainly the characteristics of the drainage
system, such as length and slope. The factor/8 accounts for the
other physical properties of the basin.
Ourotion
Fig. 1 Geometric elements of triangular hydrograph
-------
178
In the routing of the hydrographs through the conduits the "time-
offset" procedure is used as being consistent with the overall
simplification sought 18.
MODELLING
The model requires the following physical data of the catchment:
area, percentage of imperviousness, time of concentration and, if
discretization is applied, time offset of subcatchment hydrograph
to basin outlet. Values of parameters of the unit tri-
angular hydrograph should also be determined in each particular
case.
The model computes runoff based on specified rainfall as follows*:
1. Determination of the beginning of overland flow and
depression storage supply rate from pervious surfaces, by
equalizing the antecedent mass of rainfall and the ante-
cedent mass of infiltration.
Infiltration is expressed by Horton's exponential function:
f - fc + (f0 - fc) e ~K fcf (l)
Where: f = infiltration capacity rate,
t_ = time from the beginning of the infiltration
capacity curve;
f = the constant rate at which f is approached
c
asymptotically as time increases, or final
sustained rate;
fQ = infiltration capacity rate when tf=0, or
initial sustained rate;
* Linear interpolation is used when rainfall interval exceeds
one minute.
-------
179
K » decay rate of infiltration related to soil
characteristics (I/time unit);
e = base of the system of natural logarithms.
(Values of fc and K vary with the antecedent moisture
conditions).
2. Calculation of the overland flow supply (rainfall excess)
from pervious surfaces by subtracting from the rainfall
depth losses due to infiltration, depressional storage
and other minor losses:
ie » Kx (i - f - s) (2)
Where ie = rainfall excess;
i = rainfall intensity;
f = infiltration capacity rate
P-F
s = depression storage supply = (i - f) e S
-------
180
ie =. K2 (i - s) (4)
P
Where: s - i e~ S^ (5)
d
K- = dimensionless coefficients! which accounts
for minor losses.
4. Transformation of the rainfall excess hyetographs determined
in steps 2 and 3, to runoff•hydrographs at the outlet of
each subbasin by applying the triangular, hydrograph concept
to elemental excess rainfall hyetographs of short duration
(one minute).
5. Combination of the hydrographs obtained in step no 4 at the
outlet of each subbasin.
6. Routing of the combined hydrographs of step no 5, by the
"time-offset" method, through the main conduits of the
catchment in order to obtain ultimately a computed outflow
hydrograph at the outfall of the basin.
7. In case of combined sewers, addition of a specified base-
flow (dry weather flow) to the calculated runoff to obtain
the total outflow of the catchment.
Fig. 2 illustrates the above steps.
PROGRAM DESCRIPTION
The program is written in FORTRAN IV language and is implemented
on a Digital PDF 11/34 minicomputer. It can also be executed on
an IBM 370 without any modification.
It has about 300 statements and takes 14 K of memory without
overlays. The time of execution is less than half a minute to
simulate an event of five hours.
-------
181
Excess rainfall from
pervious areas
Infill ration
iprosaion storage
depression storage
Total hydrograph
{from pervious and
impervious areas)
Offset hydrograph
FIG. 2
ILLUSTRATIONS OF THE VARIOUS CALCULATION
STEPS OF THE C.U.M. MODEL
-------
182
DESCRIPTION OF THE SELECTED TEST BASINS
Three urban drainage basins of different size, location and charac-
teristics were selected in order to verify the proposed method of
runoff determination. (Fig. 3, 4 and 5)-.
No
1
Basin
Ball-De 1'Epee
Size
Location (acres!
Montreal 70
Approxi- % of im-
mate po- pervious-
pulation ness
6,000 75
2 Curotte-Papineau Montreal 2214 103,000 47.5
Malvern
Burling-
ton (On-
tario)
58
1,000 34
Time of
concentra-
tion
(min)
21
50
12
The general surface topography of basin nos 1 and 2 is characterized
by a gentle slope. In basin no 1, land use is predominantly
residential and consists mainly of row duplexes with flat roofs,
narrow front strips of grassed surfaces and wider grassed back-
yards draining into sewer inlets located in paved alleys. Insti-
tutional and commercial use together with open spaces and parks
represent less than 15% of the total area of the catchment. All
roof drains are directly connected to the sewers.
In basin no 2, residential land use represents approximately 60%
and industrial use 20% of the total catchment area. 85% of the
residential zone consists entirely of row duplexes, the rest
being occupied by single-family units. The surface drainage pattern
is similar to basin no 1.
In basin no 3 |_12J , the entire area is single family residential
with no vacant lots or parks. The average size of the lots is
rather large with the front and backyards alone representing
approximately 65% of the total catchment area. Backyards drain
into gulleys which in turn drain into streets or into sewer inlets.
-------
183
LIMITS OF SUBBASINS
SCALE
FIG. 3
BASIN
(BALL DE L'EPEE)
-------
184
LIMITS OF SUBBASINS
2400'
SCALE
FIG. 4
BASIN N2 2 (CUROTTE PAPiNEAU)
-------
185
SCALE
FIG. 5
BASIN N*3 (MALVERN)
-------
186
The general surface topography is characterized by a gentle slope.
Directly connected impervious areas represent 31% of the total.
Estimated time of concentration at the outlet of each of the three
basins was based on full pipe travel time. The difference found
between this time and travel time calculated on the basis of a
flow proportional to the contributing sub-areas of the catchment
is quite small.
Inlet time for impervious surfaces was evaluated according to the
location of the first upstream sewer inlet and some field obser-
vations .
RECORDED RAINFALL AND RUNOFF DA'TA
For basin no 1, precipitation data were provided by the CUM's
raingauge network consisting of tipping bucket type raingauges
connected directly to a mini computer (Digital PDF 11/34) which
also does the data processing.
Runoff data were obtained from stage-discharge curves established
from direct measurements at the basin outlet.
For basin no 2, precipitation data were provided by the City of
Montreal's own tipping bucket raingauges. Flow was measured at
the basin outlet by a permanently installed weir previously cali-
brated in a laboratory.
Rainfall and runoff data for basin no 3 were extracted from various
studies on this catchment [ 12, 13j
For basins no 1 and 2, input hyetograph of five-minute intervals
was used for modelling. For basin no 3, this interval varies
from one to ten minutes.
DISCRETIZATION OF SELECTED TEST BASINS
It is common practice in urban runoff modelling to discretize the
catchment, i.e., to divide the area and the related drainage
system into discrete elements.
-------
187
Although no general rules for such a procedure exist, the most
common assumption is the finer the discretization, the more
accurate the representation of runoff phenomenon.
One of the main objectives of the CUM's method was, however,
to obtain acceptable results by large lumping of the subbasins
and ultimately by considering the entire basin as a single catchment,
Comparison between outflow hydrographs calculated by dividing the
three test basins into 6, 8 and 10 subbasins respectively and by
considering the entire basin as a single catchment, showed no
significant improvement in the results obtained by discretization
(Fig. 6). Therefore, a single basin approach is adopted for the
CUM model.
MODEL CALIBRATION
Through calibration, which normally constitutes a prerequisite to
the verification of the model, an adjustment of input parameters
is made until a good fit is obtained between measured and computed
hydrographs [_14J .
It must be noted, however, that the adjusted values of some input
parameters should not normally deviate too much from the corres-
ponding extreme values reported in the literature, otherwise one
may be left with an input that is partly man-made.
The intensity of the recorded precipitations in all three test
basins was not sufficiently high to produce significant runoff
from the pervious areas. Thus, the model was calibrated for the
impervious areas directly connected to the sewer system.
Basin no 1 was calibrated with storm nos 2, 5, 7 and 8 (Table 1),
basin no 2 with storm nos 1 to 10 (Table 2), and basin no 3 with
storm nos 3, 7 and 8 (Table 3) .
The following parameters of the model have to be calibrated:
-------
00
co
o 08
BASIN N2 1
storm r»o 2
73-06-E9
——observed
— simulated with
6 subcoichmenfs
— simulated with
! subcatchmuM
0 I
10 •£
1900 2000 21-00 2200 23-00 000
I OO 2 00
FIG. 6
COMPARISONS BETWEEN SIMULATIONS
WITH DISCRETIZED SUBBASINS AND
ONE SINGLE CATCHMENT
-------
189
Percentage of imperviousness
This physical characteristic of the basin is particularly important
since it affects directly the runoff volume. It can be obtained by
direct site measurements (Basin nos land 3) or estimated from
typical land use (Basin no 2). In this latter case, it should be
verified by comparing the total excess rainfall with the measured
runoff volume.
Since the values of this parameter used for all three basins gave
good results, no further ajustments have been necessary.
Depression storage capacity (3d)
The effects of this parameter on computed peak flow and runoff
volume depend on the storm pattern and the total rainfall, especially
in the case of advanced storm pattern and low precipitation event.
The value used for basins no 1 and 2 is 0.0625 in.as suggested in
the literature [8J and 0.020 in. for basin no 3, as found in previous
studies on this catchment 112, 13J .
Coefficient of minor losses (K-^, K?)
The value of this coefficient should be close to unity. A value
of 0.98 was used for all test basins.
Geometric elements of the triangular hydrograph ( & and ff )
Various tests showed that values of
-------
190
Therefore, (Xand ft should be calibrated.
Good results were obtained with (X =1.5 for the three test basins
and with values of /B-0.57, 0.54 and 0.45 respectively for basins
no 1, 2 and 3. These values are fairly clos-e to the average ones
recommended by USSCS ( K =* 1.67 and ft =* 0.6) for ungaged catchments,
RESULTS OF SIMULATION
Simulations of 9, 10 and 23 rainfall-runoff events respectively
for the three basins were made.
A summary of these events is shown in Tables 1, 2 and 3, where the
mean and standard deviation of recorded/simulated peak flows
and flow volumes indicate a good fit between observed and calcu-
lated hydrographs.
Some typical results are shown in Fig. 7, 8 and 9.
Using the results of various studies on basin no 3 1 12, 13J , a
comparison of simulation with SWMM, ILLUDAS and CUM models is
presented in Table 4. It should be noted that with SWMM and
ILLUDAS, discretization of basin no 3 in 10 and 40 subcatchments
respectively was applied, while with the CUM model only one lumped
single catchment was considered. The results show that the three
models compare fairly well.
SUMMARY AND CONCLUSIONS
Recorded precipitation and runoff data in three test basins were
used to calibrate and verify the proposed CUM model for urban
runoff determination.
Because of the relatively low intensities of the precipitations
recorded, runoff from pervious areas could not be verified.
Simulations made with a lumped single catchment for each basin
showed that the CUM model, due to its nature, is not very sensitive
-------
191
to discretization. This suggests that it can be used with some
confidence with a minimum amount of input data.
The results obtained indicate that, within the scope and purpose
of this utilization, this simple, quick and inexpensive computer
model could become a fairly accurate tool for predicting the
quantity of runoff from urban catchments and for enabling the
initiation of the real-time control of intercepting systems.
-------
TABLE 1 SUMMARY OF SIMULATIONS ON BASIN NO. 1 (BALL-DE L'EPEE)
.p
G
-------
Table 2 SUMMARY OF SIMULATION ON BASIN NO. 2 (CUROTTE-PAPINEAU)
• 4J
c
0)
w
No
1
2
3
4
5
6
7
8
9
10
11
12
13
14
,15
16
17
18
19
20
21
22
23
t:
Date
76-07-24
75-07-27
75-09-02
75-10-13
76-07-16
76-07-27
76-09-23
76-05-11
76-06-11
76-09-18
75-07-25
75-07-28
75-08-21
75-09-02
75-09-06
75-10-15
75-11-21
76-05-31
76-06-25
76-07-07
76-09-20
76-09-22
76-09-26
Total
rainfall
(in)
0.35
0.63
0.95
0.57
0.83
0.81
0.13
0. 38
0.44
0.47
0.27
0. 30
0.10
0.21
0.31
0.17
0.20
0.24
0.90
0.40
0.60
0.21
0.64
Peak flow (cfs)
0
353.1
469.6
54.1.7
402.0
646.2
776.9
132.0
203.4
399.0
512.0
487
472
104
215
236
153
229
130
519
485
258
243
194
S
364.3
404.9
761.0
376.6
679.8
823.8
127.9
190.4
402.3
440.5
452
441
122
204
259
135
214
154
500
584
255
267
187
0/S
0.9692
1.1600
0.7118
1.0674
0.9505
0.9430
1.0320
1.0682
0.9917
1.1623
1.0774
1.0729
0.8524
1.0539
0.9111
1.1333
1.0700
0.8480
1.0380
0.830
1.0117
0.9101
1.0374
Volume (ft3)
0
1 646 183
3 317 960
3 914 314
2 456 859
3 805 813
5 069 316
1 197 352
3 041 563
2 832 870
2 231 221
1 948 227
1 869 861
640 450
1 207 197
1 828 083
1 360 432
1 346 539
2 390 457
5 389 108
1 762 713
3 815 687
1 794 179
4 180 745
S
1 752 979
3 203 335
3 847 365
2 870 634
3 790 725
4 354 208
1 258 724
3 017 141
2 632 679
2 134 785
1 771 269
1 801 165
697 670
1 126 555
1 902 732
1 362 865
1 160 982
2 449 839
5 103 147
1 777 832
3 908 103
1 757 534
4 339 330
0/S
0.9390
1.0357
1.0174
0.8558
1.0039
1.1642
0.9512
1.0080
1.0760
1.0451
1.0999
1.0381
0.9179
1.0715
1.0408
0.9982
1.1598
0.9757
1.0560
1.0085
0.9763
1.0208
1.0367
Time to peak (min.)
0
95
130
230
140
80
50
135
150
45
40
55
55
135
100
55
85
90
105
55
50
100
62
300
S
86
116
242
152
75
52
127
141
44
53
46
47
105
80
64
73
101
97
58
43
147
53
315
0/S
9
14.
-12
-12
5
-2
8
9
1
-13
9
8
30
20
9
12
-11
8
_ •}
7
13
9
-15
0 •= Observed values
S = Simulated values
x = 0.9957
s = 0.1130
1.0215
0.0705
x" = 11.4
s = 37.15
-------
TABLE 3 SUMMARY OF SIMULATIONS ON BASIN NO. 3 (MALVERN)
-M
q
01
w
No
1
2
3
6
7
8
9
10
11
12
Date
73-09-22
73-09-23
73-10-13
73-11-14
73-11-15
73-11-28
#4 1974
#7 1974
#12 1974
#15 1974
Total
rainfall
(in)
0.71
0. 36
0.31
0.60
0.71
0.47
0.63
1.20
0.63
0.30
Peak Flow (cfs)
0
32.4
25.2
8.4
10.5
6.5
9.5
31.8
27.2
15.1
8.8
S
29.7
30.0
8.4
9.4
6.8
9.9
34.9
24.0
20.1
8.4
0/S
1.0909
1.2600
1.0000
1.1170
0.9558
0.9595
0.9111
1.1333
0.7572
1.0476
Volume (ft3)
0
54600
25400
19900
44200
46100
32800
44717
15925
36183
20283
S
44792
23474
19822
39912
47494
34249
41295
15893
43541
19309
0/S
1.2189
1.0820
1.0039
1.1074
0.9706
0.9576
1.0827
1.0020
0.8310
1.0504
Time to peak (min.)
0
42
117
112
344
142
27
34
13
13
54
S
44
120
111
356
143
45
32
11
11
54
0-S
-2
-3
1
-12
-1
-18
2
2
2
0
0 = Observed values
S = Simulated values
x = 1.0232
s - 0.1392
s
1.0306
0.1040
x = -2.545
s — 6.5
-------
Table 4: COMPARISONS BETWEEN SWMM, ILLUDAS & CUM SIMULATIONS
ON BASIN NO 3 (MALVERN)
-M
C
0)
W
No
1
2
3
6
7
8
9
10
11
12
Peak Flow (cfs)
0
32.4
25.2
8.4
10.5
6.5
9.5
31.8
27.2
15.1
8.8
0/SWMM
0.9391
1.1277
1.0181
1.1138
0.9377
0.9938
0.8231
1.1574
0.8146
1.1794
0/ILLUDAE
0.9672
1.1749
1.0696
1.1258
0.9377
1.1778
0.8508
1.1628
0.8168
1.2408
0/CUM
1.0909
1.2600
1.0000
1.1170
0.9558
0.9595
0.9111
1.1333
0.7572
1.0476
Volume (ft3)
0
54600
25400
19900
44200
46100
32900
44717
15925
36183
20283
0/SWMM
1.1030
1.0410
0.9522
1.0676
0.9370
1.000
1.1284
1.1045
0.9068
1.0982
0/ILLUDAS
1.1294
1.0686
0.9809
1.0869
0.9525
0.9724
1.1463
1.1356
0.9203
1.1297
0/CUM
1.2189
1.0820
1.0039
1.1074
0.9706
0.9576
1.0828
1.0020
0.8310
1.0504
Time to peak (min)
0
42
117
112
344
142
27
34
13
13
54
0-SWMM
2
-5
2
-10
-6
-17
6
4
4
0
0-ILLUDAS
0
-5
2
-11
-3
-8
4
4
4
2
0-CUM
-2
-3
1
-12
-1
-18
2
2
2
0
Ui
observed values
x - 1.0104 1.0524 1.0232
s = 0.1328 0.1493 0.1392
1.0331 1.0523 1.0306
0.0798 0.0869 0.1040
-2 -1.10 -2.54
7.34 5.40 6.51
-------
10.~04
06
20 |
10 i
6
10 £
S.
OS
t'JOO HOOO 2100 2200 23-00 O-OO 1-00 2-00
4 'XI 5-00 6'OQ 7:OO S-OO SCO 10 00
IIOO
time (t*Ot'r:}
FIG. 7
BASIN Mei (BALL-DE L'EPEE)
OBSERVED AND SIMULATED HYDROGRAPHS
-------
1S7
2-
700-
6OO-
10
-20
•30
•40
•SO
20
1300 1500 17 CO iS CO 2100
FIG. 8
BASIN H-2 (CUROTTE-PAPiNEAU)
OBSERVED AND SIMULATED HYDRCGRAPHS
-------
U U LI U U H B U D UUUU|]UUf\\]UUU U U D
i-D
CO
time (hours)
FIG. 9
BASIN N23 (MALVERN)
OBSERVED AND SIMULATED HYDROGRAPHS
-------
193
REFERENCES
1. U.S. Environmental Protection Agency, "Assessment of
Mathematical Models for Storm and Combined Sewer Mana-
gement; Section V", EPA-600/2-76-175a (1976).
2. Gray, M.D., "Handbook on the Principles of Hydrology",
Secretariat, Canadian National Committee for the Inter-
national Hydrological Decade, VIII.5 (1970).
3. Willeke, E.G., "Time in Urban Hydrology", Jnl. of Hy-
draulic Div., ASCE, Vo.. 92, Jan. 1966, pp. 13 - 29 and
Discussions by Viessman, W., and Vician, B.E., ibid.
Sept. 1966.
4. Rovey, W.E., Woolhiser, A.D., "Urban Storm Runoff Model",
Jnl. of Hydraulic Div., ASCE, Nov. 1977, pp. 13369 - 1351.
5. "Design and Construction of Sanitary and Storm Sewers",
WPCF Manual of Practice no 9 (1970).
6. Escritt, B.L., "Design of Surface Water Sewers", C.R.
Books Ltd., London, U.K. (19-64).
7. Mitci, C., "Sur une nouvelle methode de calcul des di-
bits d'orage et des hydrogrammes de ruissellement dans
les bassins de drainage urbain", T.S.M. - L'Eau, Febr. 1974
8. U.S. Environmental Protection Agency, "Urban Runoff
Characteristics", Section VI, EPA 11024 DQU 10/70 (1970).
9. "The Illinois Urban Drainage Area Simulator (ILLUDAS)",
Illinois State Water Survey, Bulletin 58 (1974).
-------
200
REFERENCES
10. Yen, C.B., Shen, Y.Y., Chow, T.V., "Time of Peak Surface
Runoff from Rainstorms", Surface and Subsurface Hydrology,
Proceedings, Fort Collins Third International Hydrology
Symposium on Theoretical and Applied Hydrology, Water
Resources Publications, Colorado (1979).
11. Marsalek, J., "Malvern Urban Test Catchment", Volume I,
Research Report no. 57, Environment Canada, (1977).
12. Patry, G., Raymond, L., "Malvern Catchment, Illudas model
study". Progress report no 1, Bessette, Crevier, Parent,
Tanguay et Ass. (1978).
13. Jewell, K.T., Nunno, J.T., Adrian, D.D., "Methodology for
Calibrating Stormwater Models", U.S. Environmental Protec-
tion Agency, Proceedings. Stormwater Management Model
(SWMM), Users Group Meeting, EPA-600/9-78-019 (1978).
-------
201
Feasibility of Storm Tracking
For Automatic Control
of Combined Sewer Systems
by
Van-Thanh-Van Nguyen
M.B. McPherson
Jean Rousselle
Presented at the
Stormwater Management Model (SWMM)
Users Group Meeting
Montreal, Quebec
May 25, 1979
-------
202
PREFACE
by Mn Bo McPherson
Background
The following Technical Memorandum is Addendum 6 of a 1977 ASCE Program
report on "Urban Runoff Control Planning",,(1) Addendum 1, "Metropolitan
Inventories," and Addendum 2, "The Design Storm Concept," were appended to the
latter report,, Addenda 3,(2) 4(3) and 5(^) were the first of several additional,
individual additions planned for release over the period 1977-1979 „
The principal intended audience of the ASCE Program's June, 1977, report
was the agencies and their agents that are participating in the preparation of
areawide plans for water pollution abatement management pursuant to Section 208 of
the Federal Water Pollution Control Act Amendments of 1972 (P0Lo 92-500),, While
the presentation which follows is also directed to areawide agencies and their
agents, it is expected that it will be of interest and use to many others,
particularly local governments,
ASCE Program
The American Society of Civil Engineers' Urban Water Resources Research
Program was initiated and developed by the ASCE Urban Water Resources Research
Council, The basic purpose of the Program is to promote needed research and
facilitate the transfer of findings from research to users on a. national scale,,
Abstracts of the twenty-eight technical memoranda and reports of the
Program for the 1967-1974 period are included in a readily available paper0(5)
Two reports and six technical memoranda of the regular series completed since are
identified in a companion publication»(6) Also included in the latter is a
listing of all but one of the twelve national reports in the special technical
memorandum series for the International Hydrological Programme; and the last
national report(^) and an international summary(S) have been released since„
A Steering Committee designated by the ASCE Council gives general
direction to the Program: S0 W0 Jens (Chairman); W0 C0 Ackermann; J0 C0 Geyer;
N0 So Grigg; C0 F0 Izzard; D. E, Jones, Jr0; and L, S0 Tucker,, M. B. McPherson
is Program Director (23 Watson Street, Marblehead, Mass,, 01945). Administrative
support is provided by ASCE Headquarters in New York City0
Automatic Control
Since its inception in 1967 the ASCE Program has been monitoring leading-
edge projects of local governments in an effort to stimulate state-of-the-art
advances elsewhere. For example, the master plans for combined sewer overflow
pollution abatement in San Francisco and Chicago,(9) which call for eventual
incorporation of some form of automatic control,
Installation of new storage and treatment facilities in combined sewer
systems is expected to increase the complexity of operations substantially,,
Automatic control is viewed by many as a means for maximizing the operational
efficiency of these more complicated systems, Should storage and treatment be
adopted as a means for abating pollution from separate storm sewer runoff, there
will be an even greater interest in using automatic control.
-------
203
Anticipation of rainfall over combined sewer systems in advance of its
occurrence is a goal sought by all developers of automatic control capabilities.
The exploratory study report which follows is offered as a modest contribution
towards the reaching of that goal,,
Acknowledgments
This report covers the introductory phase of the doctoral research of
Mr. Van-Thanh-Van Nguyen on storm tracking and storm characterization, one of
several components of a new Urban Water Resources Program in the Department of
Civil Engineering, Ecole Polytechnique of Montreal, The other two authors were
his advisors on the study,, (The writer is privileged to serve the Department as
a quarter-time Invited Professor under an agreement between ASCE and Ecole
Polytechnique, and thus has filled two roles in connection with this report).
The ASCE Urban Water Resources Research Council is indebted to the
Department of Civil Engineering of Ecole Polytechnique of Montreal for its generous
contribution of this report as a public service.
Processing, duplication and distribution of this Technical Memorandum
was supported by Grant No0 ENV77-15668 awarded to ASCE by the U.S. National Science
Foundation,, However, any opinions, findings, and conclusions or recommendations
expressed herein are those of the authors, including this writer, and do not
necessarily reflect the views of the National Science Foundation,,
References
lo McPherson, M. B., "Urban Runoff Control Planning," ASCE, New York, N.Y.,
118 pp., June, 1977. (Available as PB 271 548, at $5.50 per copy, from the
National Technical Information Service, 5285 Port Royal Road, Springfield,
Virginia 22161).
2. Espey, W. H., Jr., D0 G. Altman and C. B. Graves, Jr., "Nomographs for
Ten-Minute Unit Hydrographs for Small Urban Watersheds," ASCE UWRR Program
Technical Memorandum No. 32, ASCE, New York, N.Y., 22 pp., December, 1977.
(NTIS: PB 282 158).
3. Marsalek, Jiri, "Research on the Design Storm Concept," ASCE UWRR Program
Technical Memorandum No. 33, ASCE, New York, N.Y., 27 pp., September, 1978=
(NTIS:
4. Abbott, Jess, "Testing of Several Runoff Models on an Urban Watershed," ASCE
UWRR Program Technical Memorandum No. 34, ASCE, New York, N.Y., 42 pp., October,
1978. (NTIS:
5o McPherson, M. B., and G. F. Mangan, Jr., "ASCE Urban Water Resources Research
Program," J.Hyd.Djv., ASCE Proc., Vol. 101, No. HY7, pp. 847-855, July, 1975.
6. McPherson, M. B., and G. F. Mangan, Jr., Closure to Discussion of "ASCE
Urban Water Resources Research Program," J.Hyd.Div., ASCE Proc,,, Vol. 103,
No. HY6, pp. 661-663, May, 1977.
7. Ramaseshan, S., and P. B. S. Sarma, "Urban Hydrological Modeling and
Catchment Research in India," ASCE UWRR Program Technical Memorandum
No. IHP-12, ASCE, New York, N.Y., 21 pp., May, 1977. (NTIS: PB 271 300).
-------
204
McPherson, M. B., and F. C. Zuidema, "Urban Hydrological Modeling and
Catchment Research: International Summary," ASCE UWRR Program Technical
Memorandum No0 IHP-13, ASCE, New York, N.Y., 48 pp., November, 1977«
(NTIS: PB 280 754).
McPherson, M0 B0, "innovation: A Case Study," ASCE UWRR Program Technical
Memorandum No0 21, ASCE, New York, N.Y., 59 pp», February, 1974,, (NTIS:
PB 232 166).,
Footnote
The above references Numbers 1, 2 and 3 are being published
collectively in a single volume by the U.S. EPA. as Urban Runoff Control Planning
EPA.-600/9-78-035, 187 pp., October, 1978.
-------
205
FEASIBILITY OF STORM TRACKING
FOR AUTOMATIC CONTROL
OF COMBINED SEWER SYSTEMS
(Addendum 6 of the ASCE Program report "Urban
Runoff Control Planning," June, 1977)
by
Van-Thanh-Van Nguyen, Etudiant au Doctorat
M. B. McPherson, Professeur Invit£, and
Jean Rousselle, Professeur Titulaire,
Departement de Genie Civil
Ecole Polytechnique de Montreal
Campus de 1'Universite de Montreal
Case Postale 6079, Succursale "A"
Montreal, Quebec H3C 3A7
Canada
November, 1978
-------
206
TABLE OF CONTENTS
SECTION 1 - INTRODUCTION
Figure 1 - Montreal Region ot,oo»oo<,o»0o
Automatic Control „ „ „ „ „ » o „ o o o » » o » . o o o o
Figure 2 - Automatic Operational Control Components
SECTION 2 - RAINGAGE DATA „ „ . „ . , , , . . .
Raingage Network „ o » „ » 0 0 o o o o . o
Figure 3 - Recording Raingage Network
Rainfall Record „ . „ „ „ „ „ . . . . „ „ „
Storm Selection 0 0 ° » 0 o » <, » « o » o »
SECTION 3 - STORM ANALYSIS „„„„„„„„«,
Network Plotting ooo°°oooooo<
Storm Center Correlation „ „ „ „ . „ „ ,
Mean Storm-Center Paths „ „ . „ . „ „ „ <
Figure 4 - Mean Storm-Center Paths
Reference Depths „ « „ » o „ » o « . . <
Figure 5 - Reference Depths „ . „ <
1
1
1
2
1
3
5
5
6
5
5
8
8
8
8
9
10
11
SECTION 4 - CORRELATION ANALYSIS „ „ . . „ . „ „ . . „ . . „ „ . „ „ „ „ . 12
Storm Data Used oooooo0.»0ooooo<,ooooooooo<,o 12
Table 1 - Event Characteristics „ „ , „ „ „ „ . . „ „ „ . „ „ 13
Regression Analyses oooo.<,oo<,.ooooooooooo0o»o 12
9 -
Correlation Coefficients for Individual Events . . . 16
Observed and Fitted at Dorval oo0o...o,>oo 17
Observed and Fitted at Dorval „ 0 „ . „ „ „ „ , , , 18
Observed and Fitted at Dorval „„.„.. <> . « . , 19
Observed and Fitted at Dorval „ „ „ „ „ „ „ 0 „ „ „ 20
Observed and Fitted at Dorval ... e o ...... 21
Observed and Fitted at Dorval ........... 22
Standard Errors for Individual Events o.o.oo. 23
Radar Check <,oooo<,oooooooo°oooooooooo.oo. 15
Parameter Exploration oooooooooooooo.ooo.ooooo 24
Table
Figure
Figure
Figure
Figure
Figure 10
Figure 11
Table 3
SECTION 5 - CONCLUSIONS AND RECOMMENDATIONS „ . . . . . . „ „ . . „ . „ . 25
Conclusions ooooooooooooooo................ 25
Recommendations oooo<,oooo0ooooo00.ooooo..oo 25
Encouraging Developments o0o<,ooooooo.oo<,«».0oo. 26
Acknowledgments „ . .00000. ,, 0 » „ . o « . « . .00.0000 27
SECTION 6 - REFERENCES . . . . . „ . „ . . „ . . . . . . . . . . . . . . . 28
-------
207
SECTION 1
INTRODUCTION
The objective of the empirical study reported herein was to explore the
feasibility of tracking storms for combined sewer system automatic control
applications„ Data used were for the largest depth events in a twenty-seven month
record of hourly rainfall from a network of eighteen gages in the Montreal n^ion.
Simplistic hypotheses have been exploited to extract the maximum amount of
information from the limited data base0 Despite reservations about the adequacy
of the data base, it is concluded that the feasibility of tracking storms for
automatic control applications is suggested by the findings„ However, all that
has been demonstrated is that it is possible to estimate expected rainfalls one
observation interval in advance, one interval at a time. Much more work will be
required, with more extensive data bases, before desirable tracking capabilities
can be attained„ Noted are encouraging advances that are taking place in
hydrometeorological research,,
To insure attribution of opinion, a narrative style has been employed in
most of the text.
Background
Ecole Polytechnique is affiliated with the University of Montreal and is
located on the campus of the University. Under development by the Department of
Civil Engineering of Ecole Polytechnique (E0P») is an Urban Water Resources Program
to serve metropolitan Montreal. Greatly facilitating the organization of the
Program is a written agreement between the Montreal Urban Community (M0U0C0) and
Ecole Polytechnique for cooperation and collaboration in urban, water resources
research, development and training. The M0U0C0 is a metropolitan maltiple-purpose
special district that includes among its responsibilities, mainly for the island of
Montreal, the collection of all wastewaters and their treatment by the design,
construction and eventual operation of the required interceptors and treatment plant„
While the M.U0C0 is committed to some degree of supervisory control of its interceptor
system, it has felt obliged to investigate the feasibility of automatic control for
later consideration. As a service to the M0U0C0 in this connection, E0P0 embarked
on an exploratory study, reported herein, as one of several projects in its new
Program,,
Figure 1 shows the locations of the islands of Montreal and Laval. Tha
City of Montreal occupies tha central portion of the ISO-square mile island of
Montreal, the remainder being under the jurisdiction of several other cities and
towns,, The City of Laval occupies the 90-square miles of the island of Laval, (In
a separate project, E0Po is assisting the City of Laval in precipitation data
analysis). Most of the three million people in. the Montreal metropolitan area live
on the islands of Montreal and Laval.
Automatic Control
An observation made in 1974 with regard to urban water subsystems, such
as combined sewer systems, remains valid today: "The big step to be taken for most
subsystems is tha third modeling step — the control model — to bring to operational
fruition the ultimate: fully automatic operational control".(I) The components
for the latter are graphed in Figure 2»(2) The developmental bottleneck in
Figure 2 is "automatic control signals generation" (lower left box).
-------
208
' W^K
-0/0
\
\
-m? ^
.r.v
AIRPORT
CANADA
"U.S.A"
PLATTSBURGHoE?:
LAKE
"CHAMPLAIN
FIGURE 1- MONTREAL REGION
-------
209
ON-SITE
VARIABLES
DISPLAY
FIELD
INSTRUMENT
SENSING
INSTRUMENT
SIGNALS
TELEMETRY
V
SIGNAL PROCESSING AT CONTROL CENTER
v
LOGGING
FIELD
PARAMETERS
DATA
ANALYSIS
DISPLAYS
CHARTING
FIELD
PARAMETERS
WARNING
AND ALARM
DISPLAYS
SURVEILLANCE BY HUMAN SUPERVISOR
I
t
±
MANUAL
OVER-RIDE
CONTROL
SIGNALS
AUTOMATIC
CONTROL
SIGNALS
GENERATION
CONTROL
SIGNALS
TELEMETRY
ON-SITE
DISPLAY OF
RESPONSES
FIELD
CONTROL
ELEMENTS
ACTUATION
RESPONSE
SIGNALS
TELEMETRY
CONTROL
ELEMENTS
RESPONSE
SENSING
FIGURE 2 AUTOMATIC OPERATIONAL CONTROL COMPONENTS
(2)
-------
210
The most advanced conceptual development for total automatic control of
combined sewer systems has been in connection with a plan for overflow pollution
abatement being implemented by the City and County of San Francisco, California,
General strategies were developed at Colorado State University^) in a cooperative
project with the City* Initial detailed efforts by the City have favored what might
be called "reactive" control, whereby the combined sewer system would be manipulated
in response to a storm as it occurred over the City,, The ultimate scheme would be
what might be termed "anticipatory" control, where the sewer system would still be
manipulated according to the actual incidence of a storm over the system, but
tempered by the expected rainfall at the start of the storm and in subsequent
periods of time throughout the storm. This requires development of capabilities
for storm tracking and for estimation of expected storm patterns on a probabilistic
basis. Achievement of anticipatory control is particularly attractive where
manipulation of large volumes of storage is involved,,
San Francisco has embarked on the development of an anticipatory
capability.(A) Storm-tracking and pattern-anticipation research is also a part
of a project in a three-State sector centered about Chicago, where special radar
equipment is being used in conjunction with the deployment of a network of 320
raingages.(5) in contrast to these efforts, the Ecole Polytechnique study reported
herein is minuscule, being merely the introductory phase of the doctoral thesis
research of the first author„
Returning to "automatic control signals generation" in Figure 2 (lower
left box), it is important to realize that storm tracking and pattern anticipation
is only one of the capabilities that have to be developed to achieve "hands-off,"
or complete, automatic control of combined sewer systems„ Storm characterization
models have next to be interfaced with models that simulate the important hydrologic
and hydraulic behavioral characteristics of catchments, in groups and collectively,
and the whole integrated with decision algorithms reflecting system performance
criteria,, All this must be accomplished in a real-time context, with shifts in
tactics during a storm required in as short an interval as fifteen minutes„
However, while storm tracking and pattern anticipation capability is only one of
the above three elements, it is the one that is needed to advance from a reactive
mode to an anticipatory mode,,
Considerable background is contained in a report from Colorado State
University,(6) which includes a discussion of the feasibility of real-time
prediction of urban rainstorms and an extensive bibliography on urban rainfall
variation. Also discussed are questions of sensitivity in the interfacing of
storm models with system behavior models„
-------
211
SECTION 2
RAINGAGE DATA
Rajngage Network
Shown, in Figure 3 are the locations of the eighteen .recording raingages
that formed the network for this study,, /The gage at Dorval Airport on the island
of Mantreal was chosen, as'the reference or target gage because it is at a first-
order weather station. While there are five gages in the vicinity of Ottawa
only the Ottawa Airport gage was used because it is at a first-order weather
station. The reason that all recording gages in the immediate vicinity of
Montreal were used was that part of the research, not reported here, was
involved with point versus areal rainfall in that vicinity,,
Data for the network had to be obtained from three sources: two provinces
and the national government. In retrospect, it would have been better if two gages
had also been included that are located in Ontario just off the lower left corner
of Figure 3, If there had been time, it would have been helpful to include a gage
or two in New York and possibly a gage in Vermont, The U,S, border is only a
little more than thirty miles below Montreal,
St0 Jerome (Figure 3) is on the edge of the Laurentian Mountains and Ste0
Clothilde (Figure 3) is in the foothills of the Adirondack Mountains, The broad
and comparatively flat St0 Lawrence Valley passes in a southwest-northeast
direction between these ranges in the Montreal region,
Rainfall Record
Rainfall data from the eighteen recording gages was readily available only
at an hourly interval, a rather severe limitation. Records were made available to
Ecole Polytechnique on magnetic tapes.
The longest concurrent record for the eighteen gages was for the period
August 1, 1972, through October 31, 1974, Therefore, our study was restricted to
this twenty-seven month record.
Storm Selection
The record for the target or reference gage, Dorval Airport, was first
examined in terms of the twenty-four hour periods with the largest total depths,
From a total of 275 days with rain at Dorval, the about one-sixth with the largest
depths were segregated. The smallest twenty-four hour total depth for these cases
was 0020-in, The reason the largest cases were emphasized was because we assumed
that these would be the most challenging to an automatic control system. From
these top twenty-four hour cases, forty-two distinct storms were identified, some
of which carried over from one calendar day into the next.
-------
FIGURE 3-RECORDING RAINGAGE NETWORK
-------
213
We reasoned that low-intensity, long-duration cyclonic-type storms
would be. the easier to anticipate for automatic control applications because of
the extensive continental weather forecasting that is made of frontal movements„
Also, the hourly depths associated with such mild cyclonic storms are rather
uniform and comparatively small, further facilitating their anticipation for
automatic control,, By eliminating the obvious low-intensity, long-duration
cyclonic occurrences (5 cases) and those for which there were too many missing
data for the network (8 cases) or there were anomalies in the data or other
suspicious implications (8 cases), the original forty-two candidate storms were
reduced to twenty-one.
-------
214
SECTION 3
STORM ANALYSIS
Network Plotting
The records for the segregated twenty-one storms were plotted using the
SYMAp(7) grid interpolation and computer plotting program,, With this program,
contours for a given range of values are demarcated by the use of a specific symbol
for that range„
Storm Center Correlation
In the San Francisco work, a good lagged correlation had been obtained
for certain hourly data between gages in the City and other gages to the north-
northwest along the coast,,(8,9) Inspired by these indications, from among the
twenty-one storms that remained for study we identified six for which the mean
storm centers appeared to pass first near the most remote gage at Ottawa Airport
and then near the Dorval Airport gage. These six events had been selected by
studying the network tabulated record, augmented by visual inspection of the SYMAP
plots.
For the six storms, a correlation analysis was made of the Ottawa versus
Dorval hourly depths for time lags of one, two, three and four hours„ For time
lags of two, three and four hours, the associated individual hourly depths for
the Ottawa and Dorval gages were acceptably correlated, with a correlation
coefficient of about 0.7 or more for five of the six candidate events„ Thus, a
fairly strong persistence in the occurrence of maximum depths along the mean path
of the center of a storm was suggested,,
Mean Storm-Center Paths
For most of the record used, the mean storm center passed away from
Dorval Airport or on a path different from the Ottawa-Dorval axis,, The estimated
mean storm-center paths for all the twenty-one events discussed subsequently are
indicated in Figure 40 These paths were selected by studying the tabulated records
for the network, augmented by visual inspection of the SYMAP plots. We feel
reasonably confident of the mean storm-center compass bearings indicated, but the
distance by which these storm-centers passed beyond Dorval is a much more subjective
estimate,, However, our goal was to extract as much information as possible from
the limited data available,,
Our after-the-fact appraisal of storm path is obviously contrary to the
need in automatic control applications to anticipate storm paths. Extensive work
by the Illinois State Water Survey^, 10) kas demonstrated that a special 10-cm
radar augmented by perhaps a dozen telemetry-connected raingages is capable of
tracking storms over a range of at least one hundred miles and of measuring
rainfall amounts in a storm to an acceptable precision,, We could not conceive of a
control system that would be driven exclusively by raingages because of the
extensive network of telemetry-connected gages that would be required and the
massive amount of data processing in real time that would be involved.
We therefore postulated the position that, in future control applications,
mean storm-center paths would necessarily be determined via special radars
continuously calibrated against a few telemetry-connected raingages. For the
-------
FIGURE 4-MEAN STORM-CENTER PATHS
-------
216
purposes of our study we somewhat arbitrarily chose a fifty mile radius from
Dorval as the reference circle for the succeeding analyses. We reasoned that
mean storm-center paths could be determined adequately outside this periphery
with special radars. As a matter of convenient coincidence, we had found that a
two-hour lag time from Ottawa to Dorval was the hourly multiple that provided the
best correlation, on the average, for storm centers that passed approximately
along the Ottawa-Dorval axis. Ottawa Airport is very close to one-hundred air
miles from Dorval Airport. Thus, we would assume initially that the appropriate
time lag from a fifty mile circle around Dorval would be one hour. We had been
advised that storm movement was generally around half that speed, so we
investigated a two hour and three hour lag time as well. The results of these
analyses are presented in the next Section.
Reference Depths
Returning to Figure 4, only a few of the twenty-one events had estimated
mean storm-center paths that passed over Dorval. Some sort of relationship had
to be hypothesized for the other cases. Shown schematically in Figure 5 is a
representative "mean path of center of storm," together with a parallel line
passing through Dorval. Where the parallel line passing through Dorval crosses
the fifty-mile circle of Figure 4 we have designated the interpolated rainfall depth
as d^Q. The observed rainfall depth at Dorval is designated as dp. We have
assumed that d^Q and dp are correlatable. This is an extrapolation of our meager
findings for the Ottawa-Dorval axis in that we have assumed that a degree of
proportionality exists between depths along a line parallel with the mean storm-
center path as was inferred between depths along the mean storm-center path itself.
This is to say that the lateral distribution of rainfall amounts is assumed to be
similar as a storm traverses a fifty mile distance. Or, putting it another way,
we assumed that not only does the magnitude of hourly depth tend to persist at the
storm center as it moves along its path, but that the hourly depth along a line
parallel with the mean storm-center axis also tends to persist along that slice
of the storm.
Before leaving the question of storm path, some mention should be made of
our interpretation of it. We were aware of the fact that convectional storms are
made up of cells that grow and decay in a mostly random fashion and that cyclonic
storms have pulsating bands of higher intensity rainfall. The center of mass of
the aggregate of such cells or bands would thus be expected to wander rather than
follow a straight line or even a continuous curve, and these tendencies could be
observed even with the limited data at our disposal. However, in an engineering
application, rigor must give way to simplification to arrive at applicable tools.
We feel fairly comfortable about the use of a mean storm-center path as it was
derived from our data. Only more extensive and detailed evidence would provide a
greater assurance.
Two reviewers of a draft of this report felt that the hypothesized
lateral variations in rainfall depths schematically portrayed in Figure 5
represented a too idealized or too deterministic simplification of complex
processes. We are confident that succeeding studies elsewhere using more generous
networks will lead to greatly improved hypotheses.
-------
LARGEST
LATERAL
RAINFALL
DEPTH
LARGEST
LATERAL
RAINFALL
DEPTH
DORVAL
FIGURE 5-REFERENCE DEPTHS
-------
218
SECTION 4
CORRELATION ANALYSIS
Storm Data Used
Because of an automatic control orientation, our attention was focused
on the largest total depth events and on the greatest rainfall intensities within
these events. To reduce the data processing burden we restricted analysis to the
continuous portions containing the larger depths„ We reasoned that small depths,
such as in antecedent rainfall, could be handled with relative ease in control
applications, but if included in our data sets would distort correlations because
of the disproportionately large errors in small values0 For example, we dismissed
values of one-hundredth to five-hundredths of an inch occurring over a period of
several hours before or between periods of heavy rainfall over the raingage
network^ Thus, our definition of an event was of a fairly continuous occurrence
of larger depths. Some events continued from one calendar day into the next»
Parenthetically, in characterizing a sixty-seven year record of hourly
values for a gage in San Francisco, events were defined in terms of the major
rainfall portions of storms,,^'
Characteristics of the twenty-one events used in the analyses to be
reported are listed in Table 10 Indicated therein are: the event total duration
at Dorval, TD; the total event depth interpolated at the reference point, 059, and
recorded at Dorval, DD; and the maximum hourly depths, (d5o)m and (dj))m, and
average hourly depths, d^Q and "3"^, at the two points,, Hourly and other depths at
the reference point (e0go, d^Q and D^Q) were all obtained from interpolated values
as read from the SYMAP plottings,, As noted earlier, the mean storm center paths
for the twenty-one events are shown in Figure 40
Regression Analyses
We had hypothesized that because a two-hour time lag had produced the
best correlation, on the average, between Ottawa and Dorval hourly depths, over a
distance close to 100-miles, a one-hour lag time should be appropriate in relating
the interpolated hourly depths 50-miles distant from Dorval, d^Q, with those
observed at Dorval, dDo Simple linear regression analysis of the observed hourly
depths at Dorval at time t+1, dD , against the interpolated hourly depths at the
reference point an hour earlier, d^ , resulted in the following equation of fit
for all twenty-one events:
dDt+1 = 0..7246 d50t + 0<>0214 (Equation 1),
* t+1
where dD is the hourly depth predicted by the equation. For Equation 1, the
simple correlation coefficient was 006417 and the standard error of estimate was
Oo0671-in.
-------
219
TABLE 1 - EVENT CHARACTERISTICS
Event
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Event Date
(yr/mo/dy)
720803
720807
720808
720827
721012
721028
730428
730518
730519
730611
730629
731002
731005
740304
740404
740414
740430
740531
740804
740817
741014
Event
Duration
(hr.)
TD
6
10
10
11
3
10
4
12
6
3
5
6
10
14
5
4
8
3
5
4
7
Total
Event Depth
(in.)
D50
,,522
1,170
.800
,685
.210
,935
,070
,580
,235
,420
,520
,300
1,270
1.110
0460
.210
,300
,,200
,450
,245
.180
DD
1,170
,,720
,660
,340
,210
.840
,220
,440
.350
,470
,270
,260
1,040
1,210
,630
,220
.430
0200
,330
,250
,300
Maximum
Hourly Depth
(in,)
(d50^m
,24
,26
,39
,22
,14
,25
,03
,16
,06
.25
.32
,12
,41
.16
,46
,15
,06
.11
,16
,12
,04
(dD>m
,47
.25
,,34
,19
,16
,16
,08
,13
.10
,41
,16
,06
,36
,17
,23
,20
,10
,14
,18
.18
,09
Average
Hourly Depth
(in,)
d50
.087
.117
,080
,062
,070
,093
,018
.048
,039
,140
,104
,050
,127
,079
,092
,052
,037
,066
,090
,061
,026
dD
,196
,072
,066
,031
.070
.084
,055
,036
,058
.156
,054
,043
,104
,086
,126
,055
,054
,066
,066
,062
,043
-------
220
Because we had been advised that storm movement generally required closer
to two hours to traverse a distance of 50-miles, we next made a simple linear
regression analysis for a two-hour time lag.. The following equation of fit was
obtained for all twenty-one events, for the observed hourly depths at Dorval at
time t+1, dpt+ , against the interpolated hourly depths at the reference point two
hours earlier, 1"
dpt+1 = 0.1246 d50t"1 + 0.0663 (Equation 2),
For Equation 2 the simple correlation coefficient was 0.0122 and the standard error
of estimate was 0.0870-in.
Simple linear regression analysis for a three-hour time lag resulted in
a simple correlation coefficient of only 0.0021 and a standard error of estimate
of 0.0874-in.
Of the three time lags tested, the one-hour time lag was obviously
superior. However, several researchers have noted a degree of dependence between
the hourly rainfall at a given station and the consecutive hourly rainfall at that
station. In an attempt to incorporate some accounting for any serial dependence of
hourly values, a multiple stepwise linear regression analysis of six variables was
performed. The dependent variable was the hourly depth at Dorval, dp , one hour
after a given depth had occurred at the fifty-mile distant reference point,
Four independent variables in addition to d^Q were included among the total of
six random variables: d$Q , d$Q , dp and dp ,r For a lack of significance,
two of the independent variables were rejected, d50C" and dp"-" „ The resulting
expression for the twenty-one events was as follows:
dpt+1 = 0.7788 d50t - 0.2850 d50t"1 + 0.1474 dpfc + 0.0274 (Equation 3).
That is, with this expression, given the rainfall depth at the reference point
fifty miles from Dorval at time t (d^Q^) and t-1 (d5Qt ), and the depth at Dorval
at time t (dp1"), the depth at Dorval can be predicted at time t+1 (dp). The
multiple correlation coefficient for the above equation was 0.6643 and the
standard error of estimate was 0.0649 -in.
Equation 3 seemingly represents only a very modest improvement over
Equation 1. After comparing the standard errors of estimate for the individual
events, the overall standard errors of estimate for these two equations, and the
predicted versus the observed values of dp for each of the twenty-one events, we
concluded that the two equations were of about equal validity, although Equation 3
held a slight edge. We decided to use Equation 3 exclusively in the remainder of
this presentation because we suspect that the inclusion of two more independent
variables, d5Q and dp , in this equation might tend to compensate, respectively,
for the variations overlooked in taking one hour as the average lag time for a
slice of all the twenty-one storms as it moved from a point 50-miles distant from
Dorval to Dorval, and the effects of dependence of rainfall amounts in consecutive
rainy hours.
Because the lag time between the reference point and Dorval was taken
as one hour for Equation 1 and is the lag time of the dominant independent
-------
221
variable in Equation 3, listed in Table 2 are the individual event simple
correlation coefficients for the observed hourly depths at Dorval, dDt+1, versus
the interpolated reference depths, d50t. In Table 2, the correlation coefficients
for fifteen of the twenty-one events are 00756 or larger. Considering the errors
inherent in the crude means that were employed to estimate d^Q, we have no basis
for concluding that the interpolated values of d5Q were more dependable for these
fifteen cases, or that they were less dependable for the six cases with correlation
coefficients ranging between 0,589 and Oo670«
Plotted in Figures 6 through 11 are cumulative curves for all twenty-one
events of the observed rainfall at Dorval at t+1 and of the values regenerated
using Equation 3, termed "fitted" in these graphs. The dispersion for each event
displayed in Figures 6 through 11 has been quantified as the standard error of
estimate (SE)« Listed in Table 3 are the SEs for each of the twenty-one events,
for values of hourly depths of Dorval that satisfy Equation 3, do*" , versus
those actually observed at Dorval, dj)*- . Because all event parameters — such as
total event depth,, hourly event depths, total storm duration — are of importance
in automatic control applications, it is difficult to interpret the physical
meaning of the relative magnitudes of the SEs8
Returning to Figures 6 through 11, we wish to draw attention to the
fact that, because of the self-corrective nature of Equation 3, the errors in
individual time steps are smoothed out over the total event,, That is, overall,
the equation does track the correspondence between d^Qfc and djf . Of course, the
acceptability of errors in individual hour projections would be a function of the
control system catchment size and the degree of control precision sought, as well
as the state of any storage that might be involved at that time.
We are satisfied that the results described in this subsection show
considerable promise for further development of storm tracking capability where
more generous data bases are available,,
It is important to remember that our objective was to explore the
possibility of developing an anticipatory model, not to develop an operational
model. Therefore, it was not necessarily appropriate for us to split the
twenty-one events into calibration and verification components. Furthermore, the
very small number of usable events, and the loss of supporting information from
the equal number of events rejected, militated against viewing the test model as
being more important than it should be regarded,,
Radar Check
After the mean storm-center paths had been estimated on the basis of
listed hourly depths for all eighteen raingages and reference to the SYMA.P plots,
radar photos were checked for further confirmation,, However, radar records were
available for only thirteen of the twenty-one events involved* The weather radar
installation at issue is located in the western part of the island of Montreal at
Ste. Anne de B.ellevue, Figure 3U However, it was possible to discriminate only
the mean directions of the storms but not their mean paths because gradations in
intensity were not discernable on. the photos. As a result of this check, the
direction of two of the storms was revised somewhat,, As noted earlier, the mean
storm-center paths graphed in Figure 4 are the ones that were used in arriving at
the values of d used in the correlation analyses„
-------
222
TABLE 2
CORRELATION COEFFICIENTS
FOR INDIVIDUAL EVENTS
Event
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Event Date
(yr/mo/dy)
720803
720807
720808
720827
721012
721028
730428
730518
730519
730611
730629
731002
731005
740304
740404
740414
740430
740531
740804
740817
741014
Correlation
Coefficient
,, t+1 , t%
(dD vs. d50 )
r
.911
.883
,792
.,620
.998
.756
o870
.589
U589
.,994
.992
.823
,,641
o889
0888
,968
,670
,997
.881
,924
.648
-------
223
234
HOURS
1 2 3
HOURS
o.
UJ
a
2
f: 0.3
a:
o 0-2
LLJ
t-
«* 0.1
234
HOURS
0
No. 7
(730428)
O
O
-------
224
0.4
0.3
0.2
0
No. 4
•(720827) / I
/ i
I
I
I
_L
_L
567
HOURS
8
10 11
0
10 11 12 13 14
FIGURE 7- OBSERVED (—) AND FITTED ( o) AT DORVAL
-------
22 5
0
0 1
23456
HOURS
7 8
FIGURE 8-OBSERVED (—) AND FITTED (-- ©)
AT OORVAL
-------
226
3456
HOURS
0.9
0.8
0.7
QC
Q
Ld
< 0.6
o
0.5
0.4
0.3
0.2
0.1
0
No. 6
(721028)
0 1
456
HOURS
7 8
10
10
FIGURE 9- OBSERVED (
AT DORVAL
) AND FITTED ( o)
-------
227
No. 19
0.31- (740804)
h-
< 1.2
tl.l
UJ
D
_, 1.0
1?0.9
z
Q
0.8
0.7
(-
<
^ 0.6
o 0.5
o
0.4
0.3
0.2
0,1
0
No. 1
-(720803)
i
0 I
FIGURE 10- OBSERVED (
234
HOURS
0.6
0.5
0.4
0.3
0.2
0.1
0
No. 10
-(730611)
0
1 2 3
HOURS
0
234
HOURS
5 6
) AND FITTED ( ©) AT DORVAL
-------
228
0.3
0.2
0.1
0
- No. 21
(741014)
0
345
HOURS
7
< 0.7
cr
§0-6
Q_
IJL)
Q
0.5
0.4
0.3
< 0.2
< 0.1
cr
o n
LJ U
No. 3
"(720808)
I
0
-------
229
TABLE 3
STANDARD ERRORS
FOR INDIVIDUAL EVENTS
Event
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Event Date
(yr/mo/dy)
720803
720807
720808
720827
721012
721028
730428
730518
730519
730611
730629
731002
731005
740304
740404
740414
740430
740531
740804
740817
741014
Standard Error of
Estimate, Equation 3
f" t+1 , t+1,
(dD vs. dD )
S.E. (in.)
.170
.046
.061
,063
o045
o037
o035
.040
o030
,226
o072
,,033
.094
.034
.082
.060
.028
.053
.059
.016
.027
-------
2310
Of interest in passing is that members of the Stormy Weather Group,
McGill University, Montreal, had used records from a special radar at Ste0 Anne
de Bellevue several years ago to study summer storms that occurred between 1969
and 1972 over Ottawa, some 85 miles away0( ' Their interest centered on a
study of events which had led to the flooding of house basements in Ottawa,
Parameter Exploration
An attempt was made to correlate such variables as storm compass
bearing, offset of mean storm path from Dorval, total event depth, total event
duration, etc0 Because of the inexactness of storm path estimates and the small
number of events involved, no responsible conclusions could be drawn. However,
in future more extensive studies elsewhere it may be possible to arrive at the
type of generalizations soughto
-------
231
SECTION 5
CONCLUSIONS AND RECOMMENDATIONS
Conclusions
The objective of this study was to explore the feasibility of tracking
storms for combined sewer system automatic control applications. We attempted
to extract the maximum amount of information from a sparse network of recording
rainfall gages for which only hourly data were readily available. As a
consequence of our limited data base, we have been obliged to make simplistic
hypotheses on the character and movement of storms. However, this should not be
taken as a disparagement of the network, which serves other, quite different
purposes.
Despite the preceding reservations, we feel that this exploratory study
has suggested that the tracking of storms for automatic control applications is
feasible. However, all that has been demonstrated is that it is possible to
estimate expected rainfalls one observation interval in advance, one interval at
a time. The ultimate in predictive tools would also estimate expected total
event duration, total event depth and total event pattern. We are confident that
analyses from the present and future use of special radars^1^) W£H make it
possible to predict, say, half-hourly values of rainfall on a specific sector of
a jurisdiction at least two hours in advance.
With respect to storm pattern characteristics, advances continue to be
made on statistical characterization of rainfall records for a single point for
intervals of an hour or more,(13) Storm total durations and storm total depths
seem to be highly independent of each other, as do event total durations and
event total depths. The small number of event cases in Table 1 reflect this
common observation. Perhaps no realistic total storm prediction tool can be
developed. That is, perhaps the best that can be done will be evaluation of
storms as they occur, in a real-time sense, with only about a two-hour lead time,
We are nevertheless optimistic because of indications from San Francisco of a
banding of total event durations and total event depths("' and because of research
under way in the U,S,(12) and the U,K,(14) where special radars are being used
to develop physical and climatological models of storms over metropolitan areas
for local jurisdiction drainage system applications.
To reiterate, storm tracking is only part of the capability required
for automatic control. Also needed are compatable hydrologic-hydraulic system
models and the incorporation of operating rules and stratagems in a computer
format.(15,16) These three elements must be operable in a real-time mode to be
fully effective,
Recommendat ions
We noted at the outset that the study reported here was merely the
introductory phase of the doctoral research of the first author. His theoretical
attention was devoted to statistical analysis of point and areal rainfall. In
that connection, we obtained considerable insights from the storm tracking study.
We are of the opinion that local government operating agencies would find the
undertaking of a similar exploratory study to be very illuminating, whether or not
the attainment of automatic control was a serious objective. There are some
metropolitan areas in the U.S.(17) and Canada that have more extensive raingage
-------
232
networks than the one in the Montreal area. Data from these networks could be
exploited to considerable advantage. Our experience with Montreal data has been
that the storm tracking study has led to a much better grasp of rainfall data
information potentialities and shortcomings for planning and design applications
than would otherwise have been possible,,
For the operating agency committed to the achievement of complete
combined sewer system automatic control there appear to be only two options: wait
for the technology to be perfected for San Francisco or Chicago or London; or
invest in a suitable radar installation and commence a minimum of two years
duration investigation and calibration phase to develop their own technology,,
We sincerely hope that the findings from our study will provide
encouragement and yet serve as a caution to those interested in automatic control
of combined sewer systems. In particular, we hope that our study will help to
facilitate communication between engineers and meteorologists.
Encouraging Developments
A radar installation backed with calibrating telemetry-connected
raingages could also serve as part of a metropolitan flood-warning system,. Flash
floods are associated with thunderstorm rainfall, which occurs over fairly small
areas. "Radar with a scanning radius in excess of 100 miles can provide information
over data void areas. ..... The ultimate solution is 'real-time1 processing and
analysis of radar data by automatic computer,, <,„„<,„ Automatic processing
techniques are currently undergoing operational testing by the NWS0 A prime
objective of these tests has been the improvement of hydrological forecasting
techniques„"C-*-"^ A metropolitan flood warning system is under development in
Melbourne, Australia,,'-'-"' Feasibility of linking a weather radar to a computer
in which an optimum reservoir operating program is stored, as well as a program
of a hydrologic rainfall-runoff model producing the reservoir inflow hydrographs
from radar-measured rainfall data, has been explored in the Federal Republic of
9 ( 9n ^
Germany for two small adjacent river basins 13- and 22-mi, in size)
Lastly, special new radars of the type alluded to earlier in this Section
have been developed in conjunction with weather research of the Illinois State
Water Survey,,(-'-2) A recent phase of that research has included the deployment of
two radar stations and 320 raingages in a three State sector centered about
Chicago, Illinois. Although the thrust of the current research is on inadvertent
weather modification, in an ancillary part a radar-computer system was linked to
the Waterways Control Center of the Metropolitan Sanitary District of Greater
Chicago to demonstrate its possible potential for use in real-time control0(5)
One of the several objectives of the research is to "demonstrate the utility of
a sophisticated weather radar system in the operation of the large urban,
semi-automated water resource system of Chicago and environs".(21) Also, important
contributions have been made to the understanding of characteristics of seveve
storms associated with flash flooding.(22)
-------
233
A£knpwledgments
Reported in this Technical Memorandum are the findings from the
introductory phase of the doctoral research of the first author. His
dissertation,(23) scheduled for completion early in 1979, features a probability
analysis of rainfall temporal patterns for a thirty-year record of hourly depths
at Dorval Airport, and a probability analysis of the relationships between hourly
point rainfalls at Dorval Airport and the associated mean rainfalls over the
islands of Montreal and Laval,
This study was supported in part by a grant from the Ministry of
Education of Quebec (No» CRP-388-77) and by a development grant from the National
Research Council of Canada (No,, RD-812)0
Dr» Gaston Paulin, Director of the Meteorology Department of the
Ministry of Natural Resources of Quebec, made valuable comments on the
meteorological aspects of the study with respect to the data used. Messrs„
Constantin Mitci, Guy Paquin and Nguyen Nam of the Water Purification Department
of the Montreal Urban Community made valuable comments on the methodology
developed with respect to practical applications. Professor Andre Leclerc,
Head of the Civil Engineering Department at Ecole Polytechnique, gave
encouragement, to the study and facilitated its administrative support.
-------
234
SECTION 6
REFERENCES
1» Sonnen, M. B0, L. A0 Roesner and R. P0 Shubinski, "Urban Water Management
Models," pp0 89-97 in Urban Runoff., Quantity and Quality, ASCE, New York,
N.Yo, 19750
20 McPherson, M0 B0, "Feasibility of the Metropolitan Water Intelligence System
Concept (Integrated Automatic Operational Control)," ASCE UWRR Program
Technical Memorandum No0 15, ASCE, New York, N.Y0, 110 pp», December, 1971=
(NTIS: PB 207 301).
3. Trotta, Paul D0, John W0 Labadie and Neil S0 Grigg, "Automatic Control
Strategies for Urban Stormwater," Journal of the Hydraulics Division, ASCE
Free., Vol0 103, No. HY12, pp0 1443-1459, December, 19770
4. Bureau of Sanitary Engineering, City and County of San Francisco, and Water
Resources Engineers, IncOJ "Demonstrate Real-Time Automatic Control in
Combined Sewer Systems, Progress Report Number 1," U.S., EPA Demonstration
Grant No. S-803745-010, San Francisco and Walnut Creek, California, 83 pp.,
May, 1976,
5. Changnon, Stanley A_, Jr., and Floyd A0 Huff, "Chicago Hydrometeorological
Area Project: A Comprehensive New Study of Urban Hydrometeorology," Annual
and Interim Report to the National Science Foundation, Illinois State Water
Survey, Urbana, 56 pp., December, 1977.
6» Grigg, N« S0, J« W0 Labadie and H0 G, Wenzel, "Metropolitan Water
Intelligence Systems —Completion Report, Phase HI," Department of Civil
Engineering, Colorado State University, Fort Collins, 258 pp., June, 1974,
(NTIS: PB 235 532).
7o Laboratory for Computer Graphics & Spatial Analysis, "IAB-LOB," Graduate
School of Design, Harvard University, Cambridge, Mass., 32 pp., January,
1977.
80 Bureau of Sanitary Engineering, City and County of San Francisco, and Water
Resources Engineers, Inc0, "Demonstrate Real-Time Automatic Control in
Combined Sewer Systems, Progress Report No,, 2," U0S0 EPA Demonstration Grant
No, S-803743-02Q, San Francisco and Walnut Creek, California, 49 pp.,
October, 19760
9, Bureau of Sanitary Engineering, City and County of San Francisco, and Water
Resources Engineers, Inc., "Demonstrate Real-Time Automatic Control in
Combined Sewer Systems, Progress Report No. 3," U0S. EPA Demonstration Grant
No. S-803743-020, San Francisco and Walnut Creek, California, 76 pp., April,
1977.
10o Huff, P. A., and N. GD Towery, "Hydrometeorological Characteristics of Severe
Rainstorms in Chicago Metropolitan Area," pp. 214-219 in Preprints, Second
Conference on Hydrometeorology, October 25-27, 1977, Toronto, American
Meteorological Society-Canadian Meteorological and Oceanographic Society.
-------
21S
11. Austin, Go L,,, and L. B. Austin, "The Use of Radar in Urban Hydrology "
Journal of Hydrology, 22, pp0 131-142, 1974,
12. Huff, F. A0, and S. A0 Changnon, "A Hydrometeorological Research Program "
Water Resources Bulletin, Vol0 13, No0 3, pp, 573-581, June, 1977. '
13, Corotis, Ross B0, "Stochastic Considerations in Thunderstorm Modeling,"
Journal of the Hydraulics Division, ASCE Proc,, Vol0 102, No. HY7, pp[ 865-
879, July, 1976o Author's closure to discussion: Vol0 103, No0 HY12
pp. 1479-1480, December, 1977, '
14. Collier, C0 Go, and K0 A, Browning, "The Short-Period Weather Forecasting
Pilot Project and Its Relevance to Storm Sewer Operation," pp, 42-50 in
Preprints, International Conference on Urban Storm Drainage, 11-15 April
1978, University of Southampton, England, U0K0
15. Grigg, Neil So, and John W. Labadie, "Computing the Big Picture," Water &
Wastes Engineering, Vol. 12, No0 5, pp. 37-39 & 86, May, 1975,
16. Grigg, No So, J. W0 Labadie, G. Ro Trimble and D, A0 Wismer, "Computerized
City-Wide Control of Urban Stormwater," ASCE UWRR Program Technical
Memorandum No0 29, ASCE, New York, N.Y., 81 pp., February, 1976. (Available
from Department of Civil Engineering, Colorado State University, Fort
Collins, Colorado 80521)„
17. Tucker, L. S., "Raingage Networks in the Largest Cities," ASCE UWRR Program
Technical Memorandum No0 9, ASCE, New York, N0Y0, 90 pp., March 17, 1969,
(NTIS: PB 184 704)o
18. Clark, Robert A0, "The National Weather Service Program in Urban Hydrology,"
pp. 5-10 in Proceedings, National Symposium on Urban Hydrology, Hydraulics
and Sediment Control, Report UKY BU111, University of Kentucky, Lexington,
386 pp0, December, 1976»
19. Earl, C0 T0, et al., "Urban Flood Warning and Watershed Management Advances
in Metropolitan Melbourne," ASCE UWRR Program Technical Memorandum No, 30,
ASCE, New York, N.Y., 72 pp., June, 1976„ (NTIS: PB 259 549).
20, Anderl, Bernard, Walter Attmannspacher and Gert A» Schultz, "Accuracy of
Reservoir Inflow Forecasts Based on Radar Rainfall Measurements," Water
Resources Research, Vol. 12, No0 2, pp. 217-223, April, 1976.
21. Changnon, S. A., and R. G0 Semonin, "Chicago Area Program: A Major New
Atmospheric Effort," Bulletin of the American Meteorological Society,
Vol., 59, No0 2, pp, 153-160, February, 1978 0
22. Huff, Fo A0, "The Synoptic Environment of Flash Floods," pp. 10-16; R» C.
Grosh, "A Local Severe Rainstorm,," pp0 70-79; and S0 A0 Changnon, "The
Meteorological Aspects of a Severe Urban-Centered Rainstorm," pp. 152-157;
all in Preprint Volume, Conference on Flash Floods, Hydrometeorological
Aspects, May 2-5, 1978, Los Angeles, Gal., published by the American
Meteorological Society, Boston, Mass0
23. Nguyen, Van-Thanh-Van, "Caracterisation hydrologique de la pluie pour le
controle et la planification des syst"emes de drainage urbain" (Hydrologic
Characterization of Rainfall for the Control and Planning of Urban Drainage
Systems), doctorate in applied science theei-, Ecole Polytechnique of
Montreal, Quebec, 1979.
-------
236
IMPSWM
A Research and Services Project
for the Implementation of
Storm Water Management Modeling
by
Dr. Paul Wisner
Dept. of Civil Engineering
University of Ottawa
Presented at the
Stormwater Management Model (SWMM)
Users Group Meeting
Montreal, Quebec
May 25, 1979
-------
237
I M P S W M
A RESEARCH AND SERVICES PROJECT FOR THE IMPLEMENTATION
OF STORM WATER MANAGEMENT MODELLING
Storm Water Management requires new drainage
alternatives and new tools for the analysis of drainage
systems. Continuous efforts are required from all those
involved in drainage studies and in particular from munici-
palities and consulting firms to keep abreast with new deve-
lopments in this field.
In the last few years, the Rational Formula is
being gradually replaced by various hydrologic models such
as SWMM, ILLUDAS, HYMO, etc. Many municipalities and consul-
tants have already invested time, efforts and money in acquiring
various aspects of the storm water modelling technology.
Modelling of zero runoff increase has become a key issue in some
new developments and its results are presented to nonspecialists.
Conflicting results and delays are sometimes caused by the fact
that storm water management modelling is still a developing
technique.
Developers and taxpayers are interested in money
saving procedures, municipal engineers need simple methods
with adequately safe design. One of the key problems in urban
drainage design is the standardization and simplification of
modelling procedures.
After consultation with many major direct and in-
direct model users, the Department of Civil Engineering,
Faculty of Science and Engineering of the University of Ottawa,
has started in February 1979, a project coordinated by
Professor Paul Wisner, with the main goal of assisting efforts
of various organizations in the implementation and application
of practical modelling in urban drainage.
-------
238
The main tasks of the program are
1) to assess the use of some existing applications of SWM
models in practical problems;
2) to improve and test some parts of SWM models used by
participants in the project;
3) to coordinate modelling experience between a group of
interested researchers, municipal engineers and consul-
tants in view of standardization and simplification of
modelling procedures;
4) to organize and exchange of views between modellers,
developers and regulatory agencies on the best use of
models and in particular on how to design more economic
drainage systems in new developments;
5) to organize information and training programs on modelling
and storm water management procedures.
Participants in the project have access to the
following services:
1) INFORMATION ON NEW STORM WATER MANAGEMENT
TECHNIQUES AND MODELLING APPLICATIONS
2) PARTICIPATION IN TRAINING PROGRAMS INCLUDING
A SHORT-COURSE PER YEAR, AND FORUMS FOR THE
DISCUSSION OF STORM WATER MANAGEMENT ISSUES
3) DISTRIBUTION OF THE UPDATED OR IMPROVED
COMPUTER PROGRAMS AND TESTING RESULTS
4) DISTRIBUTION OF REPORTS REGARDING THE PERFORMANCE
OF MODELS AND APPLICABILITY FOR DIFFERENT TYPES
OF ENGINEERING PROBLEMS.
By May 15, 1979, fourteen organizations including
provincial government agencies, municipalities, a major
conservation authority, consulting firms and a large developer
have joined the project. Organizations interested in the implemen-
tation of practical modelling, reducing their direct or indirect
expenses for the development of modelling capabilities, the
-------
239
standardization of modelling procedures and reducing the cost
of drainage by adequate application of new techniques are
invited to inquire regarding their participation.
A preliminary sponsoring committee was formed by
representatives of organizations which have already committed
financial support to the program.
Members of this committee are:
Mr. John Anderson, P.Eng.
Mr. R. Cebryk, P.Eng.
Mr. Martin Fereday, P.Eng.
Mr. W.J. Ford, P.Eng.
Mr. Ron Hicks, P.Eng.
Mr. Ivan Lorant, P.Eng.
Mr. Craig Mather, P.Eng.
Mr. Dipen Mukherjee, P.Eng.
Mr. Z. Novak, P.Eng.
Mr. Alan Perks, P.Eng.
Mr. Harold Reinthaler, P.Eng.
Manager, Ottawa Branch,
Gore & Storrie Limited
Senior Drainage Engineer,
City of Edmonton
Project Manager,
Genstar Development,
Edmonton
Associate,
Cole, Sherman & Assoc. Ltd.
Coordinator,
Ministry of Natural Resources,
Central Region
Chief Water Resources Engineer,
M.M. Dillon Limited,
Toronto
Manager, Flood Control,
Metropolitan Toronto,
Conservation Authority
Town Engineer,
Town of Markham
Drainage Engineer,
Water Modelling Group,
Ontario Ministry of Environment
Manager, Hydro technical Group,
Proctor & Redfern Limited,
Toronto
Project Manager,
Fred Schaeffer & Associates,
Toronto
-------
240
Mr. George Walker, P.Eng. President,
Walker, Newby & Associates,
Edmonton
Mr. Ron Wigle, P.Eng. Manager, Hydrotechnical Division,
James F. MacLaren Limited,
Willowdale
Mr. F.T. Wilton., P.Eng. Manager, McElhanney,
Surveying & Engineering,
Vancouver.
An advisory committee is also being formed. Based on
a preliminary activity diagram, the research team has produced
a first report on an update of the TRANSPORT model developed
by Camp Dresser and McKee Inc. - Water Resources Engineers.
This version has several features such as:
a) Improved surcharge computation. The new version no longer
considers a surge tank analogy, and modified Hardy-Cross
method is applied instead, thus eliminating unrealistic
manhole storage.
b) Weir transfers are node to node, which improved stability.
c) Improved scheme for orifice simulation.
d) Print-out summary for junctions, which shows:
Maximum computed depth and time
of occurance
Maximum computed surcharge
Minimum depth below ground elevation
Duration of surcharge.
e) Print-out summary for conduits which shows:
Design flows
Maximum computed flows and velocities
and their time of occurance
Ratio of maximum to design flows
Maximum depth above inverts at conduit ends.
The updated version of the program was distributed
to interested participating organizations. A second report on
the applicability of the Rational Method is scheduled for July.
-------
241
Work also continues on testing the TRANSPORT program for
various specific problems such as the simulation of orifices
and weirs, superpipes and drop shafts,
A first forum on the WRE TRANSPORT was held on
March 8 and 9, 1979, with Dr. Larry Roesner from WRE, author
of the new WRE - TRANSPORT. Another forum with Mr. Terstriep,
author of ILLUDAS, is scheduled on May 26.
The first short course on Storm Water Management for
Runoff Control in New Developments will also be held on
October 16 to 18, 1979.
The course will consist of five workshops:
Analysis of runoff control regulations.
Design aspects of runoff control facilities by examples.
Incorporation of runoff control in master planning.
Traditional versus new techniques in drainage
computations (methodology).
Traditional versus new techniques in drainage
computations (examples).
Each workshop will start with a general report
followed by presentation and discussion of case studies.
Reports will be prepared by the IMPSWM team and reviewed by
members of the IMPSWM steering committee.
Two lectures and a panel will present viewpoints of
regulatory agencies, municipalities and developers.
For information regarding the IMPSWM project, please
contact Dr. Paul Wisner, IMPSWM Coordinator, phone (613) 231-7022
or Mr. Atef Kassem, Assistant Coordinator, phone (613) 231-3911,
at the Department of Civil Engineering, University of Ottawa,
Ottawa, Ontario, Canada KIN 9B4.
-------
242
DESCRIPTION AND APPLICATION OF
AN INTERACTIVE MINI-COMPUTER VERSION
OF THE ILLUDAS MODEL
BY
G, Patry , L. Raymond ' and G. March!
PAPER PRESENTED AT THE SWMM USERS GROUP MEETING
MAY 24 & 25, 1979 Montreal (Quebec)
(7) AAAlAtant Pno^eA&oi, Eco£e Potyte.c.hnlqu.n, P.O. Box. 6079, Station A,
Montteat (Quebec) H3C 3A7.
(2) Be^^ette, C/teu-teA, Potent, Tangaay £ k&&., 110 Cftmcizln Blvd.
Bureau 1001, Mon&Lzal (Quebec) H2P 7B9.
[3) Rue.(Vtdh MtiAtant, Ecole. Polytzchniqu.?,, P.O. Box. 6079, Station A,
(Quebec) H3C 3A7.
-------
243
DESCRIPTION AND APPLICATION OF
AN INTERACTIVE MINI COMPUTER VERSION
OF THE ILLUDAS MODEL
1. INTRODUCTION
Computer simulation of the urban runoff process has received
a great deal of attention over the last few years; in fact
computer models have been developped and released at an
impressive rate since the early 70's. However practically
all of the now available rainfall/runoff models of interest
have been developped for large computers. With the upcoming
development of automatic control algorithms and with the
proliferation of mini-computers and/or micro-processors in
design offices there seems to be increasing interest in
providing some tools that might be adapted to these small,
but yet powerful computing systems.
The object of this paper is not to present yet another
computer model but rather to describe the modifications and
improvements introduced in an existing popular model, namely
the ILLUDAS model, in order to make it compatible with a
mini-computer system.
The first section of this paper deals with the description
of this particular version of the ILLUDAS model (interactive
mini-computer version) while the second section briefly
describes the application of the model to two different
catchments, namely the Malvern Catchment in Burlington
(Ontario) and the Curotte-Papineau Catchment in Montreal
(Quebec). Finally, a simple example presented in the Appendix
illustrates the interactive features of the model.
2. DESCRIPTION OF THE MODEL
2.1 HARDWARE DESCRIPTION
The basic features of the ILLUDAS model have been described
expensively by Terstriep and Stall (1974) and no attempt
will be made here to present the model again.
-------
244
Adapting a model to a particular computer configuration
usually requires some modification of the algorithm. The
importance of such modifications depends on the differences
between the two computing systems. In this particular case,
the original ILLUDAS model had to be rewritten completely in
order to make it compatible with the mini-computer configu-
ration available. This modified version of the ILLUDAS
model, written in BASIC has been adapted to a HP 9830 16 k
byte programmable calculator. The particular hardware
configuration is shown in Figure 1.
The core of the system is the HP 9830 calculator. Program
files are read sequentially from the internal cassette drive
and loaded into the 16 k byte memory block. Matrix and
string variable manipulations are handled on two external
ROMs (Read Only Memory) shown in Figure 1. Once a program
file has been loaded into the system, the file is executed,
statement by statement, in a sequential manner. A common
memory block saves data and/or results between loading
operations. Data and results can also be stored or loaded
from an external cassette memory. Simulation results are
printed progressively on a 132-character line printer.
2.2 INTERACTIVE FEATURES
In rewriting into BASIC the ILLUDAS model, several modi-
fications and improvements have been implemented and will be
described more extensively in the following section. However
the most significant difference of this version with the
original model lies in the interactive operation of the
program. While most rainfall/runoff models handle simula-
tion in a batch-type process, this version of the model
establishes a 'dialogue1 with the user prior and after the
simulation. It is not the intent of this paper to discuss
the pros and cons of interactive computer programming,
however among some of the interesting aspects of such an
approach are:
minimization of input data errors
increase in user's confidence with the model
- accelerate the training period with the model
excellent teaching capabilities, etc.
-------
245
FIGURE I
HARDWARE CONFIGURATION
OUTPUT
LINE PRINTER
HP 260IA
HP 9830
16k BYTES
(INTERNAL CASSETTE
DRIVE)
EXTERNAL CASSETTE
HP 9865A
ROM
MATRIX OPERATIONS
STRING VARIABLES
SYSTEM COMPONENTS
PROGRAMMABLE CALCULATOR
EXTENDED MEMORY 16k BYTES
MATRIX OPERATIONS ROM
STRING VARIABLES ROM
EXTERNAL CASSETTE DRIVE
LINE PRINTER
HP 9830A
HP II28IA
HP I I270B a OPTION 270
HP II274B a OPTION 274
HP 9865A
HP 2607A
BESSETTE,CREVIER, PARENT,TANGUAY 8
30CIATES
-------
246
Through a visual display the program asks the user to supply
the necessary information required by each subprogram block
in the logical programmed sequence. Data is entered in a
free-format form and syntax error are automatically pointed
out in the process. The user is given the opportunity to
enter or leave any of the initial sub-program blocks as he
wishes prior to entering the simulation block.
In the initial stages of program execution, control is
returned to the user after each block of operations through
the use of the 'special function key1. As shown in Figure
2, 10 sub-program blocks are available in the initial stage
of program execution prior to entering the simulation block.
These subprograms are:
- "Enter catchment data"
- "Read catchment data from file"
- "Print catchment data as read"
- "Print catchment data on forms"
- "Correct catchment data"
- "input hyetographs"
- "Print hyetographs"
- "Store catchment data"
- "Print forms"
- "Plot hydrograph from files"
Control over the program is temporarily lost after the
'Simulation' key is pressed. However, once the simulation
is complete, the following options or subprogram blocks are
available:
- "Store final hydrograph on file"
- "Plot hydrographs: simulated and observed (if available)
- "Statistical analysis on simulated and observed hydrographs",
2.3 PROGRAM DESCRIPTION
In adapting and rewriting into BASIC the original ILLUDAS
model major programming modifications were necessary. New
subroutines were created by either segmenting portions of
the original model or through the addition of new features.
-------
figure 2
INTERACTIVE FEATURES OF THE MODIFIED VERSION OF THE ILLUDAS MODEL
ENTER
CATCHMENT DATA
STORE CATCHMENT
DATA ON FILE
READ CATCHMENT
DATA FROM FILE
PRINT DATA
AS READ
PRINT DATA
ON FORMS
KEEP FINAL
HYOROGRAPHS
ON
FILE
>
PLC
HV
OB
OUTPUT
i
STATISTICAL
ANALYSIS
BETWEEN
SIMULATED i
OBSERVED
>
CORRECT
CATCHMENT DATA
INPUT
HYETOGRAPHS
PRINT
HYETOGRAPHS
PRINT FORMS
PLOT HYOROGRAPHS
FROM FILE
-------
248
This version of the model contains a total of 23 files (or
subroutines) as shown in Figure 3, that can be loaded and
executed sequentially by the system. The package now contains
close to 2 000 statements.
As mentionned previously the basic features of the original
ILLUDAS model have been presented elsewhere (Terstriep and
Stall, 1974; Patry and Raymond, 1978) and will not be consi-
dered here. However, in rewriting the program many new
features have been introduced in this particular version and
are interesting to examine. It might also be of interest to
point out that most of these new features have also been
introduced in a recent FORTRAN (French) version of the
program (Patry & Marchi, 1979).
2-3.1 Multiple rainfall hyetographs
Because of the interest and importance of spatial distri-
bution of rainfall over large urbanized catchments, this
version of the model has been designed to handle as many
rainfall hyetographs as there are reaches in the catchment
In fact, every reach of the catchment can have its own
specific rainfall hyetograph.
2.3.2 Inlet hydrograph
Because of the limited storage memory of the computing
system it is sometimes necessary and also interesting to
simulate the performance of large catchments by subdividing
them into a number of smaller and more discretized subcatch-
ments having up to a maximum of 150 reaches. This version
allows generated hydrographs from upstream catchments to be
fed into specific downstream branches.
2.3.3 Infiltration parameters
In the original version of the model, Horton's soil infil-
tration parameters are limited to four particular soil
groups. This version of the model has been designed to
handle specific soil infiltration characteristics simply
through the definition of Horton's infiltration parameters
f , f and k.
-------
249
FIGURE 3
ILLUDAS MODEL
FLOW CHART
MODIFIED VERSJON 1978
/ DAT
L
V
A
^*^ 10
* *
51MUL ATION DATA
- STARTING TIME
.T.)
-j-X
,
[PSP' ^
OfiV W£*THEH FLOW
-tMPERVtOUS DIRECTLY CONNECTEDJ
-IMPERVIOUS IMOIRtCTLY CONNECTEDf
-PERVIOUS
[0,3= [si] E
>«i
-------
250
2.3.4 Inle t time
Hodgson (1975) reported that one of the weaknesses of the
ILLUDAS model lied in the inlet time calculation from imper-
vious surfaces. The technique described by Terstriep and
Stall (1974) was reported to generate in some instances,
extremely short lag times. The kinematic wave equation was
therefore introduced as an optional means to evaluate both
pervious and impervious surfaces inlet times, thereby yield-
ing a minimum time base of about 5 minutes from impervious
surfaces.
2.3.5 D e s i g n d i a m e t e r
Even though traditionnally downstream pipe diameters are
equal to or greater than upstream diameters, it was decided
to add as an option a "hydraulic design mode" where such a
constraint on pipe diameter selection is lifted.
2.3.6 Ponding on street surfaces
Under the EVAL mode, runoff in excess of the reach capacity
is theoretically stored onto street surfaces and routed
through the reach only when the capacity of the reach
allows. Because of the difficulty in judging the importance
of such street ponding, a new subroutine has been added to
allow for a rough evaluation of the ponding depth on street
surfaces. Ponding depths are evaluated using relationships
such as those described in Figure 4. Should the ponding
depth be greater than the curb height, the full right-of-way
width is then taken into consideration in the flooding depth
calculation.
FIGURED 4
PONDING ON STREET SURFACES
„ ,. . Reach length
L = Maximum of , -,„ ,. . ,- . / , . ,
. 170 feet of street/acre x subcatchment area
V = Ponding volume available on street surface per acre (h max = 6 in.)
V = 9 ft3/ft when h = 6 in.
-------
251
2.3. 7
Hydraulic gradelineanalys is
Very often the flow that must be routed through a reach Is
larger than the pipe capacity. Hydrograph truncation through
upstream storage is not always realistic. In order to
appreciate the importance of such surcharged conditions a
third simulation mode has been introduced allowing for a
rough estimation of the hydraulic grade line slope necessary
to carry the total peak flow through the reach. While
theoretically weak this simulation mode has been found to be
very helpfull in identifying problem areas within a particular
catchment.
2.3,
Base flow (dry weather flow)
In dealing with combined sewer systems it is necessary to be
able to handle dry weather flow conditions. This version of
the model allows for a very crude way of handling dry wea-
ther flow routing. The user must first specify the total
dry weather flow to be generated by the whole catchment.
This base flow is then generated automatically at each reach
in direct proportion to the cumulative drained area at the
reach. A much more sophisticated base flow handling routine
could be implemented with little additional effort, however
such a refinement has not been found necessary for our
applications.
2.3.9
S t a t i s
ana ly s i s
A statistical analysis package has been added to evaluate
the goodness of fit of the simulated hydrographs when ob-
served runoff values are available. The following statis-
tical parameters are thus reported:
QOBS/QSIM
VOLOBS/V°LSIM
- R : Correlation coefficient
- R : Special correlation coefficient
3 coefficient
- Time to peakOBS/Time to peaks][M
- ISE: Integral square error
- Required time shift to minimize Rg
along with corresponding values of
R, R and ISE.
-------
252
2.3.10 Output and plotting features
Major changes were introduced in the presentation of the
results in order to give more readily available information
to the user. A sample output is shown in the Appendix.
Simulated and observed hydrographs can through the use of
the plotting routine also be represented on the same graph
immediately after the simulation, to allow for a quick
evaluation of the results.
3. APPLICATION OF THE MODEL
This version of the model has already been applied and
tested on several catchments. Results from two different
size catchments will be presented here. These two catch-
ment s are the:
a) Malvern Catchment in Burlington (Ontario)
b) Curotte-Papineau Catchment in Montreal (Quebec).
3.1 MALVERN CATCHMENT, BURLINGTON (ONTARIO)
The Malvern Catchment is a typical residential, 57.6 acre
catchment whose land use characteristics are described in
Table 1.
A total of 12 rainfall events were selected to evaluate the
performance of the model. Table 2 offers a summary of the
selected events. Of the twelve events selected, seven had a
maximum rainfall intensity less than or equal to 0.6 in/hr,
corresponding approximately to the final minimum estimated
infiltration rate from pervious surfaces. Furthermore, as
reported by Marsalek (1978) the maximum runoff generated by
the selected events corresponds to a once per year recur-
rence interval based on the simulation of historical rain-
fall events.
-------
253
TABLE 1
CATCHMENT SURFACE CHARACTERISTICS
SURFACE SUB-TYPE
AREA
IMPERVIOUS
acres
ha
PERVIOUS
acres
ha
FERQENT_OF
CATCHMENT AREA
Frontyards
Backyards
8.00
30.16
3.24
12.21
13.87
52.31
Sub-total
38.16
15.45
66.18
Roofs
Streets
Driveways
Sidewalks
8.09
6.68
3.10
1.63
3.27
2.70
1.25
0.66
14,03
11,58
5.38
2.83
Sub-total
Total
19.50
19.50
7.88
7.88 38.16
15.45
33.82
100.00
-------
TABLE 2
EVENT
NUMBER DATE
1
2
3
4
5
6
7
8
o
10
11
12
22 sept. 1973
23 sept. 1973
13 oct. 1973
28 oct. 1973
29 oct. 1973
14 nov. 1973
15 nov. 1973
28 nov. 1973
#4^ 1974
#7(1) 1974
/m1-15 1974
#15(1) 1974
Mean
STARTING
TIME
6:56
9:46
16:14
12.10
11:20
0:30
11:30
7:20
19:00
19:05
19:06
14:30
SELECTED RAINFALL EVENTS - MALVERN
DURATION TOTAL TOTAL ANTECEDENT TIME INTERVAL
PRECIPITATION PRECIPITATION MOISTURE RECORDED SIMU-
5 DAYS BEFORE COEFFICIENT LATED
(min)
138
130
158
665
710
590
260
245
41
21
91
142
266
(in)
0.71
0.36
0.31
1.16
1.48
0.60
0.71
0.47
0.63
1.20
0.63
0.30
0.71
(mm)
18
9
8
29
38
15
18
12
16
30
16
8
18
(in)
0.75
1.02
0
0.13
1.66
0
0.64
0.48
1.40
0.89
0.08
0.59
0.65
(mm)
19
26
0
3
42
0
16
12
36
23
2
15
16
(AMC)
3
4
1
2
4
1
3
2
4
3
2
3
3
(min)
2
2
2
10
10
10
5
10
1
1
1
2
(min)
2
2
2
5
5
5
5
5
1
1
1
2
MAXIMUM INTENSr
DURING ONE RECOI
TIME 1NTERVA1
(in/h)
2.10
1.65
0.60
0.30
0.48
0.54
0.36
0.54
3.48
2.76
2.03
0.60
(tnm/h
53
42
15
8
12
14
9
14
88
70'
52
15
ISJ
Ui
(1) Rainfall supplied by J. Marsalek without date.
-------
255
3.1.1 Results
Table 3 and Figure 5 offer a summary of the simulation
results using this version of the model. These results are
compared in Table 4 with those obtained with SWMM (Marsalek,
1977). Figures 6 through 9 represent observed and simulated
hydrographs for the first four events.
Peak flows
As shown in Table 3 the ILLUDAS model tends to underestimate
the peak flow by about 11.5% on the average, with a coeffi-
cient of variation of 18.7%. Such results are similar to
those reported by Marsalek (1977) using the SWMM model as
shown in Table 4. While no general pattern could explain
the difference between observed and simulated results, it
was observed that the worst simulations were associated with
the two longest rainfall events, events #4 and #5, which
lasted 11.1 and 11.8 hours respectively. Careful examina-
tion of the observed rainfall/runoff data for these two
events further indicated that the recorded rainfall could
hardly have generated the observed flows. It has therefore
been suggested that the actual rainfall over the catchment
might have been different from that recorded at the gauging
station (Patry & Raymond, 1979). Neglecting events #4 and #5
from the rainfall record brings the peak flow mean (x) to
5.2% and the coefficient of variation to 15%.
Peak flow simulation results using SWMM are described under
Table 4. While peak flows are underestimated by 7.6% on the
average (using all 12 events), the coefficient of variation
is slightly larger than with ILLUDAS (19.40% versus 18.7%).
Runoff volumes
Simulated runoff volumes with ILLUDAS were underestimated on
the average by about 5.6% with a coefficient of variation of
7.5%. Because of the relatively low rainfall intensities,
runoff from pervious surfaces did not contribute signifi-
cantly to the flow for any of the events considered. Simula-
ted runoff volumes using SWMM are on the average about 4%
lower than the observed values.
-------
TABLE 3
EVENT RAINFALL
NUMBER DEPTH
(in)
RESULTS
1
2
3
4
5
6
7
8
9
10
11
12
STATISTICS
0.71
0.36
0.31
1.16
1.48
0.60
0.71
0.47
0.63
1.20
0.63
0.30
MAXIMUM
RAINFALL
INTENSITY
(in/h)
2.
1.
0,
0.
0.
0.
0.
0.
3.
2.
2.
0.
10
65
60
30
48
54
36
54
48
76
04
60
SUMMARY OF RUNOFF SIMULATIONS WITH ILLUDAS (AMC = 1)
PEAK FLOWS (CFS) RUNOFF VOLUMES (ft3)
0 I 0/1 01 0/1
32.
25.
8.
8,
10.
10.
6.
9.
31.
27.
15.
8.
40
26
45
52
86
47
47
54
82
21
11
81
33.5
21.5
7.9
5.4
8.5
9.3
6.9
8.1
37.4
23.4
18.5
7.1
0
1
1
1
1
1
0
1
0
1
0
1
3?
s
c.v.
.9672
.1749
.0696
.5778
.2776
.1258
.9377
.1778
.8508
.1628
.8168
.2408
= 1.1150
- 0.2090
- 18.74%
54 600
25 400
19 900
86 100
110 300
44 200
46 100
32 800
44 717
15 925
36 183
20 283
48
23
20
79
102
40
48
33
39
14
39
17
343
770
288
969
411
667
398
732
Oil
023
316
955
1
1
0
1
1
1
0
0
1
1
0
1
X
S
c.v.
.1294
.0686
.9809
.0767
.0770
.0869
.9525
.9724
,1463
.1356
.9203
.1297
= 1.0564
= 0.0792
= 7.50%
TIME TO
0 I
42
117
112
316
437
344
142
27
34
13
13
54
42
122
110
310
425
355
145
35
30
9
9
52
PEAKS (MIN)
0-1
0
_ 5
+ 2
+ 6
+ 12
- 11
- 3
- 8
+ 4
+ 4
+ 4
+ 2
x - 0.5833
S = 6.4025
0 = OBSERVED
I = ILLUDAS SIMULATION
-------
257
Q
UJ
H-
-30
•20
Ho
20
30
OBSERVED PEAKS (cfs )
20 40 60 80 100
OBSERVED RUNOFF VOLUMES (x10~ft3)
: 1-400
>300
h200
MOO
400
OBSERVED TIMES TO PEAK(mln.)
OBSERVED TIMES TO PEAK (m\n )
flgire
-------
TABLE 4
EVENT RAINFALL
NUMBER DEPTH
(in)
I
2
3
4
5
6
7
8
9
10
11
12
STATISTICS
0.71
0.36
0.31
1.16
1.48
0.60
0.71
0.47
0.63
1.20
0.63
0.30
MAXIMUM
RAINFALL
INTENSITY
(in/h)
2.10
1.65
0.60
0.30
0.48
0.54
0.36
0.54
3.48
2.76
2.04
0.60
SUMMARY OF RUNOFF SIMULATIONS WITH SWMM (MARSALEK, 1977)
PEAK FLOWS (CFS) RUNOFF VOLUMES (ft )
0 S 0/S 0 3 0/S
32.40
25.26
8.45
8.52
10.86
10.47
6.47
9.54
31.82
27.21
15.11
8,81
34.50
22.40
8.30
5.40
8.80
9.40
6.90
9.60
38.66
23,51
18.55
7.47
0.
1.
1.
1.
1.
1.
0.
0.
0.
1.
0.
1,
If
S
c.v.
9391
1277
0181
5778
2341
1138
9377
9938
8231
1574
8146
1794
= 1.0764
= 0.2086
= 19.38%
54 600
25 400
19 900
86 100
110 300
44 200
46 100
32 800
44 717
15 925
36 183
20 283
49 500
24 400
20 900
80 200
103 700
41 400
49 200
32 800
39 629
14 418
39 901
18 470
1.
1.
0.
I.
1.
1.
0.
1.
1.
1.
0.
1.
"x
S
C.V.
1030
0410
9522
0736
0636
0676
9370
0000
1284
1045
9068
0982
- 1.0397
= 0.0735
= 7.07%
TIME TO PEAKS (MIN)
0 S 0- S
42
117
112
316
437
344
142
27
34
13
13
54
40
122
110
321
412
354
148
44
28
9
9
54
+ 2
- 5
+ 2
- 5
+ 25
- 10
- 6"
- 17
+ 6
+ 4
+ 4
0
IT » O.COOO
S «= 10.3397
K>
U1
00
0 = OBSERVED
S = SWMM SIMULATION
Several figures contained in R.R. No.57 have been corrected
to reflect simulated record flows as discussed uith Marsalek.
-------
fr ou UIKIIS— frou aoe jo
1N3HH01VO NH3A1VW
NH3A1VW NISSV8
EOU UJJOIS — EOU 86BJO
J.N3WHO1VO NH3A1VH
NIBSVB
JLMLJ I , I II H II I
I os
3 SI*
3
? Q
\ ou iuJO]S i ou aBeio
1N3WH01VD NH3A1VH
N93A1VH NISSV8
..,..., IIJ
(svamii) J*
-------
260
More details on the results of these simulations are con-
tained in Patry and Raymond (1979) and BCPTA (1978, 1979)
3.2 CUROTTE-PAPINEAU CATCHMENT, MONTREAL (QUEBEC)
While the Malvern Catchment was a relatively small residen-
tial catchment, the Curotte-Papineau Catchment , tributary
to the new 35 mile long north shore Montreal interceptor, is
a much larger urbanized catchment typical of those tributary
to the interceptor. A little over 2 100 acres are drained
by the combined sewer system. Land use characteristics of
the catchment are described in Table 5.
TABLE 5
LAND USE CHARACTERISTICS
CUROTTE-PAPINEAU CATCHMENT
MONTREAL, (QUEBEC)
BASIC DATA
Drainage area 2 130 acres
Population 103 000
Density 48.6 pers/acre
Imperviousness 49%
LAND USE
Residential 59%
Commercial 20%
Other 21%
Because of our interest in using the ILLUDAS model on much
larger catchments, it was decided to test and calibrate the
model on this type of catchment using a total of 10 rainfall-
runoff events gathered by MUG during 1975 and 1976.
The catchment was divided into 64 subcatchments having an
average area of 33 acres per subcatchment. The largest
single subcatchment covered over 340 acres. Table 6 offers a
description of the selected rainfall events along with a
summary of the ILLUDAS simulation results.
-------
TABU 6
EVENT DATE TOTAL
# PRECIPITATION
(In)
1. 75/7/27 0.62
2. 75/9/2 0.88
3. ?V 10/13 0.58
A. 76/7/16 0.84
5. 76/7/24 0.36
6. 76/5/11 0.38
7. 76/6/11 0.44
8. 76/7/27 0.81
9. 76/9/18 0.48
10. 76/9/23 0.14
0 = Observed Values
I = Simulated ILLUDAS values
CUROTTE
MAXIMUM
INTENSITY
( In/hr)
0.64
1.98
0.40
1.91
0.96
0.23
1.18
2.32
0.90
0.19
- PAPMEAU CATCHMENT -
PEAK
OBSERVED
430.8
541.7
317.1
610.9
345.4
203.4
374.3
759.2
512 .0
132.1
- SIMULATION
FLOW (CFS)
ILLUDAS (I/I
400.1
723.2
297.4
650.0
332.3
186.8
391.4
763.1
433.9
125.5
S
C.V.
1.0766
0.7491
1.0663
0.9398
1.0393
1.0887
0.9564
0.9949
1.1799
1.0523
= 1.0143
« 0.1161
=11.44*
RESULTS USING ILLUDAS
VOLUME
OBSERVED
2 880.6
3 914.3
2 400.9
3 605.8
1 646.2
3 041.6
2 832.9
5 069.3
2 231.2
1 197.4
(1000 ft3
ILLUDAS
3 150.0
4 411.6
3 086.5
3 773.7
1 804.6
3 069.6
3 002.4
4 312.5
2 317.8
1 205.0
x =
S
C.V.-
)
0/1
0.9145
0.8873
0.7779
1.0085
0.9122
0.9909
0.9435
1.1755
0.9627
0.9937
• 0.9567
= 0.1021
=10.67*
TIME
OBSERVED
130
50
140
80
95
150
45
50
40
135
TO PEAK
ILLUDAS
110
60
145
65
75
145
40
45
45
130
X
S
0-1
20
-10
-5
15
20
5
5
5
-5
5
= 5.5
• 10.39
-------
262
As seen from Table 7, peak flows are particularly well simu-
lated being underestimated on the average by only 1.4%, with
a coefficient of variation of 11.4%. Such results are par-
ticularly encouraging since they tend to show the consisten-
cy of the model for different type events and different size
subcatchments. It is also interesting to note that while
this catchment is about 20 times less discretized than the
Malvern Catchment (33 acres per subcatchment versus 1.44 acres
per subcatchment) peak flow simulation results are still
very good. Even though peak flow recurrence intervals have
not been estimated, it might be of interest to note that
flows on three occasions were larger than 50% of the full
pipe capacity of the main collector and on one occasion even
reached 75% of the pipe capacity estimated at 1 000 cfs.
TABLE 7
CUROTTE-PAPINEAU CATCHMENT
64 SUBCATCHMENTS
Peak Flows
x =
s
CV
'OBS
1.0143
0.1161
11.44%
Volumes of Runoff
x =
CV
VOL
OBS
VOL
SIM
0.9567
0.1021
10.67%
Time to Peak
x = OBS-SIM = 5.5 min
s =10.4 min
-------
263
Simulated runoff volumes are on the average overestimated by
about 5%, with a coefficient of variation of 10.7%. Such a
difference might be associated with the differences in
simulated and observed dry weather flows. Figures 10 and 11
show observed and simulated hydrographs for events #6 and #8
using the FORTRAN plotting option (Patry and Marchi, 1979).
Because of our interest in minimizing the number of subcatchments
to be simulated this 2 100 acre catchment was also simulated
using a single subcatchment representation, i.e. a totally
lumped catchment. Results, while not reported completely
herein were also very encouraging:
Peak flows Volumes
x = 1.0046 x = 0.9198
s = 0.1505 s = 0.1121
I
CV = 15.02% CV = 12.18%
4. EXAMPLE ON THEUSE OF THE INTERACTIVE VERSION OF
THE ILLUDAS MODEL
Appendix I shows a typical output in "printall" mode, illus-
trating a few features of the interactive version of the
ILLUDAS model.
5. CONCLUSION
While most rainfall/runoff programs have been developped for
large computers it is possible to adapt such models to small
micro processors and/or mini computers. These small, but
yet powerful computing systems are becoming increasingly
popular in design offices, thereby creating a need for
adapted rainfall/runoff models. The ILLUDAS model has been
adapted to a HP 9830 - 16 k byte programmable calculator
without loss in flexibility. In fact, many new features
were implemented in this new version of the model.
The model has been successfully applied to two different
size catchments:
a) Malvern catchment (57 acres), Burlington (Ontario)
b) Curotte-Papineau catchment (2 130 acres), Montreal
(Quebec).
-------
264 TEMPS
15. SO 17.05 IB. HO 19.35 20.50 23.05 23
"DP I
>— t
m
G>
rc-°-
3D§
rn
1 ! 1
P-HJ
.1
pJ
r
T.
cn
FIGURE 10
CUROTTE-PAPINEAU CATCHMENT
JUNE 11, 1976
15.50
17.05
18. 2D
19.35
20.50
TEMPS
22.05
—i
23.20
-------
265 TEMPS
^.20 14.35 15.50 17.05 18.20 19.35 20.50
, • " ~t FT " " J"( m ^"»>«>=>i™«i»«™J«»=a««i««-»ra. »««A-_, I
i a
i — i
m
~D
Q
TD"°"
!_,
_
T-
LjJ— - - -J • T-T"1 •" 1
J 1
n
a
o
FIGURE 11
CUROTTE-PAPINEAU CATCHMENT
JULY 27, 1976
1U.35
15.50
17.05
18.20
TEMPS
19.35
Z0.50
-------
266
REFERENCES
Bessette, Crevier, Parent, Tanguay & Associates (1978) -
"ILLUPAS Mode.1 Study", Ministry of Fisheries and Environ-
ment, Contract No l-SS-78-00049 , Progress Report No 1,
October.
Bessette, Crevier, Parent, Tanguay & Associates (1979) -
"ILLUPAS Model. Study", Ministry of Fisheries and Envi-
ronment, Contract No l-SS-78-00049, Progress Reports
No 2 and 3, March.
Hodgson, J. (1975) - "A compa^-i on o£ Jktue.
S-imutdtton ModdJtLA" , Greater Vancouver Sewerage and
Drainage District, February.
Patry, G. and Raymond, L. (1978) - " S-tmuldtton da
&e.ll&me.nt e.n mtltau unbatn, le. moditi ILLUPAS: Pte.6 e.ntat4.on
da. module.", Eau du Quebec, Vol. 11, No 4, Novembre.
Patry, G, and Raymond, L. (1979) - "VzAAton modt^t^d du.
mod^tt 1LLUVAS app£^tQuee aa baAAtn. Matvitin", Eau du Quebec,
Vol. 12, No 2, April.
Patry, G. and Marchi, G. (1979) - "Ve.tit>ton mocUf^ee du
module. 1LLUVAS, Manue£ d ' uttltAatzul" , in preparation,
to be released in August 1979.
Marsalek, J. (1977) - "Malvztin Jn&t Catchment, Volume, 1" ,
Canada-Ontario Agreement on Great Lakes Water Quality
RR-57.
Marsalek, J. (1978) - "Re^ea^cfc on the, Ve.Ai.gn Stonm
Co n ce.pt" , ASCE Urban Water Resources Research Program,
Technical Memorandum No 33, June.
Terstriep, M.L. and Stall, J.B. (1974) - "The.
Ufiban Vfia-inage. A/z.ea S4.muta.tox. •• 1LLUVAS", Illinois State
Water Survey, Bulletin 58, Urbana.
-------
267
APPENDIX A
EXAMPLE - ONTHEUSE OF THE INTERACTIVE
VERSION OF THE ILLUDAS MODEL
-------
268
BESSETTE, CREVIER, PARENT, TANGUAY & ASSOCIES
CONSULTING ENGINEERS
ILLUDAS
PAGE,..OF,
PREPARED BY! , DATE.
TITLE; ,
IF DATA NOT ON FILE
A) CONTRACT NUMBER!
B) CATCHMENT AREA'. , . , , , ACRES
MINIMUM DIAMETER IN DESIGN,' INCHES
MANNING COEFFICIENT IN DESIGN!,..
C) DEPRESSION STORAGE ON IMPERVIOUS AREA! INCH
DEPRESSION STORAGE ON PERVIOUS AREA! INCH
D) ENTRY TIME ILLUDAS=0 K,W,E.=1
IF K.W.E. MANNING N IMP,...,,,N PER.......
IF ILLUDAS MANNING N IMP,
E) TIME INTERVAL .MINUTES
RAINFALL DURATION, , , MINUTES
RAINFALL FREQUENCY! YEARS
ANTECEDENT MOISTURE COEFFICIENT (AMC 1, 1, 3 OR 4)
F) DESIGN! HYDRAULIC=0 STATE OF ART=1
G) ROUTING OPTION (0 OR l) ...,.(lN DESIGN MODE)
H) CHANGE ONE INFILTRATION CURVE (YES OR NO)
IF YES1SOIL GROUP!
MAXIMUM RATE: IN/HR
MINIMUM RATE; IN/HR
DECAY CONSTANT /HR
l) TOTAL DRY WEATHER FLOW CFS
j) NUMBER OF PIPES + JUNCTIONS
INPUT NETWORK DATA
IF DATA ON FILE
IN 9865A FILE NUMBER
TIME STEP, DURATION, FREQUENCY, AMC
WIN, MIN, , , , .YEARS,
TOTAL AREA .ACRES
-------
269
BESSETTE/ CREVIER, PARENT, TANGUAY & ASSOCIATES FM2
CONSULTING ENGINEERS
ILLUDAS
PAGE.,.OF ...
PREPARED BY: nATC.
UAIt, i . i i . . , ,
SIMULATION IDENTIFICATION
A) PRINT HYDROGRAPHS (YES OR NO) ......
B) CITY OF , , (15 )
C) CATCHMENT DESCRIPTION....
D) TOTAL MODE CHANGE (YES OR NO)
IF YES NEW MODE OR ALL PIPES
E) STARTING TIME (l.E. 06:56)
F) HYETOGRAPH (d,l OR 2)
= 0 TO INPUT HYETOGRAPH
= 1 IF SPECIFIED AT EACH PIPE
= 2 SINGLE ON FILE
F) IF SINGLE ON FILE(= 2)
IN 9865A FILE NO
IF TO INPUT (= 0)
NUMBER OF RAINFALL STEPS
HYETOGRAPH (IN.)
.0.00.1 ,,,,,,.,,..,.,.,,,..,.....,.
(60)
-------
270
BESSETTE, CREVIER, PARENT, TANGUAY & ASSOCIATES
CONSULTING ENGINEERS
ILLUDAS
CONTRACT; PAGE..,OF,,.
PREPARED BY: DATE
REACH NUMBER:
BRANCH IREACH I TERMI- I CONTI- I DES =1 I HYETO- I SOILI AREA OF
NUMBER [NUMBER I NATED I NUING I EVAL=2 I GRAPH I 1,2 I SUBCAT-
I I BRANCH I BRANCH I GRAD=3 I NUMBER I 3,4 I CHMENT
II I I I I I
II I ,1 I II
I I I I I II
LENGTH ISLOPE I N I DIAMETER IQ ALLOW I STORAGEI
OF FEET I % I MANNING I INCHES I CFS I 1000FMI
III III
III III
III III
IMPERVIOUS AREAS
DIRECTLY I SUPPLEMENTAL I ENTRY TIME
CONNECTED I I
ACRES I % I ACRES I % I MINUTESI LENGTH FEET I SLOPE %
III II I
I I , I II I
I I I I I I
— — _ _. _ _ _ ________________ -ft*— — — — —.— — —.— __„ _____ _ _ _ —. ___ _ — — ,........i...™,.«..._-.,__.-.—.„»___•„.„„_.__•»_
PERVIOUS AREA
I ENTRY TIME
ACRES I % I I MINUTESI LENGTH FEET I SLOPE %
II II I
II II I
II II I
-------
271
PROGRAM RUNNING IN PRINTALL FORM
52,4 14
LGADO
RUN
Si!U;-Q F0RMSM GRfiPH=2 HYETC=3?0
C' ft T H ON F J L E ? 0
CONTRACT HC'i
AREA , HIH . DI A . . t>:fiNH!HG ' S hi?57 65, 12, .013
DEFRES . STORAGE IRF & PE!?V 7 02, 184
I L L U C A S < 0 ; K . ki . E < 1 > " 0
MANNING'S FOP- IMF SURFACES?.02
TIhE STSF-,OUf<6 . FREG, AHC72, 13S, -1, 1
H'ryS DESIGN (OiSTATt OF AP-T ( 1 )?0
ROUTING 0 OS. 1?0
CHAKCE INFILTRATION CURVE?!
INPUT SCIL # .MAX,MIHi,DECAY R: A T E S ? 4, 3 ,
TCTftL CF-V kJtFTHEK FLOW (CFS)?25
7 PIPES *• JUKCTIONS76
INPUT DPTA LINE 1 FOR PIPE 1
FOR PIPE 1
FOR PIPE 1
FOS PIPE 1
0
FOR PIPE 2
FOR PIPE i
FOR PIPE 2
FOR PIPE 2
FOR PIPE 3
FOR PIPE 3
FOR PIPE 3
FOR PIPE 3
FOS PIPE 4
FOR PIPE 4
PIPE 4
INPUT DPTP LINE 4 FOR PIPE 4
I HF o1 D At S LINE 1 FOR PIPE 5
INPUT Dftlft LINE 1 FOS PIPE £
INPUT DfHfi LINE 2 FOR PIPE 6
INPUT DATA LINE 3 FOR PIPE 6
IH F U T D £ T F LINE 4 FOS PIPE s
PRINT HYDROCRAPHS71
CITY OF 7BURLIHGTON
CATCHMENT OEC-CRIPTION7PALVERN TEST BASE FLOW
TCTftL MCDE CHSNGE70
STARTING TIME XX'XX ?05i56
Ht . IN?U1=C>, MULT HY=1,SINGLE HY*2?0
NUMBER CF RPINFftLL STEPS ? ? 9
1 T H S T E P ' ...
i TH STEP?..
3 T i-i S T £??...
INPUT
INPUT
INPUT
INLET
INPUT
INPUT
INPUT
INPUT
INPUT
INPUT
INPUT
I H F U T
INPUT
I H F U T
INPUT
DATA
0
-------
272
CITY Or E (J * L I K G T C N
STORM WATER NETWORK
CONTRACT 3
MALVERr- TEST WITH DRV isEATKER FLOW
CATCHftEHT AREA ! 57.85 ACRES
TIME STEP : 1.999998 MINUTES
RAINFALL DURATION : 138 MINUTES
ANTECEDENT MOISTURE COEFFICIENT '• I
DEPRESSION STORAGE < I MFES V I OUS'' '• O.CS INCH
DEPRESSION STORAGE (PERVIOUS; .' 0 184 INCH
HftKHING'S CCEFICIENT s 0.013
HYDRAULIC DESIGN < 0 ; ; DOySTREAK D I ft >= UFST5£3r< DIft i.
ROUTING OPTION : 0
INFILTRATION CURVE PfiFiMETESS CHANGED (SOIL GROUP : -4
HfiX RfiTE =3 IN/H
ftlh RATE = 0 52 IN/H
DECAY RftTE = 4 14 ,'H
TOTAL CRY WEATHER FLOW: 25 CFS
RAINFA
F
6.
6
6
6 .
6
6
7 .
? .
7
7
?
7 .
8
S .
8
8 .
8
C
9.
56
06
16
26
36
46
56
06
16
26
36
46
56
06
16
26
36
46
54
06
v
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
TOTAL
HYETOG
RAPH
LL IN IHCH
COC
COO
coo
coo
coo
coo
. coo
010
. 020
000
065
coo
coo
000
coo
coc
000
020
coo
010
RAINFALL
CORRECTI
-_
6
B
6 .
6.
6
fe.
7
?
?
7.
7.
7
8
c
8
g
8.
8
9
0
OH
ES
38
08
18
28
38
48
58
OS
IS
23
38
48
58
08
18
28
38
48
58
08
71
70
HYETCGRAPH
0
0
0
o
0 .
0 .
0 .
0
0
0
0 .
0 .
0
.0
0
0 .
0 .
0 .
0
0
. 0 0 0
000
. 000
. 000
. 000
. 000
. 010
010
0 1 0
010
. 000
.010
000
000
000
010
010
030
040
010
INCHES
6
6
6 .
6
6
6
7
7.
7 .
7 .
7 ,
7 .
8.
8,
8
8.
8,
8
9,
9.
NO:
00
10
. 20
30
40
50
. 00
. 10
. 20
, 30
. 40
. 50
. 00
10
20
30
40
50
00
10
0
l)
0
0
0
0
o
0
0
0
0
0
0
0
0
0
0
0
0
0
0 .
. 000
. 000
000
000
. 000
. 000
. 000
000
010
065
000
000
. 000
000
000
000
,010
. 030
000
000
6
6
&
6
6,
6
7.
7
7 .
7
7.
7
$
S .
8
8.
8.
8
9
9.
02
. 1 2
22
. 32
. 42
52
. 02
12
. 22
32
. 42
52
02
, 1 2
22
32
42
52
02
12
0
0
0
0
0
0
0
0
i)
0
0
0
0
0
0
0
0
0
0
0
0 0 0
.000
000
0 0 0
.000
. 000
.010
010
. OvO
. 065
0 0 0
. 070
. 000
. 000
000
OOv
0 0 0
000
0 1 0
.010
6
6
6
6
S.
6.
7
7.
7.
7
7.
7
C
8
8
8
8 .
8
9
04
14
24
34
44
54
04
14
24
34
44
54
04
14
24
34
44
54
04
0
0
0
0
0 .
0 ,
0 ,
0
0
0
0
0
0
0
0
0
0
0
0
, 0 0 0
.000
.000
. 000
, 000
. 000
. 020
010
. 00*
065
.010
.010
. 000
.000
000
000
010
010
010
-------
273
FIFE N: 1
* * .-I *
IRIS
IMFE
IMFE
PERV
NOT
TCTft
IRFE
I H r £
5 56
6 Of
6 16
€ 26
t 36
6 4 1
6. 5s
7 06
7 16
7 26
7.36
7 5 6
R 0 £
S 16
S. 26
S 3 6
8 46-
S 56
906
^ 1 ^
PERV
r cftV
5 56
e . C>t
6 16
i: £6
;: j;
6 4 6
6 56
7 Of
7 1 i
? S i
1 "i'i
"f 4 6
7 c»:
. i 6
t:T£RY PREP IN *
SVICU: D.C
svicis r.c
DSn I NED
L
SVIOUi ARtH IKL
RV i 0-US PSfcn HYC
v
v
C:
0
t
0
C
J
"'
1
1 ?
c
6
0
<:
0
0
3
4 .
^-
i
IGl'J
lOvi
0
c,
o
i.
(.
>;.
i'
<.
\
^
v
;
C
v
. C
V
(.:
. 0
c
1
3
1
0
i
v
0
. c
4
7
0
7
1
PREP
HK£^
0
,.
0
v
*;
V
•'f
y
V
^_
-^
R ~ C
60S
6 IS
6 . 2S
i 38
6 48
i • 3 v
7 OS
7 18
726
7 43
7 ^ C
i OS
S 18
6 28
3 3S
S. 48
8 56
9 08
9 IS
1 H L E T
h t yftO
5 58
B 0 6
6 IS
i- 28
':. 3 S
6 4S
t . 58
< CS
7 1 6
7 2S
~< ~* •"•
? ^ i~-
•j '. -3
c; E x
ET II
R 0 G S s
e
0
0
c
0
0
0
e." .
J
i
1
s:
e
0
0
4
4
2 .
f
TI hF
G S ft f n
0
v
0
v
(;
0
v
v
e
i^
<:•
C
FH
v
0
C-
0
0
0
9
6
1
4
7
0
0
7
7
7
ft
5
7
=
j,,
0
0
0
c,
u
0
0
<>
o
0
2
1
TO
Ci
IM C
B
6.
6.
6.
6
6
7 .
7
7
7
7 .
s.
s .
8.
8
c
8
9
9.
q
22 .
Cf S
6
6 .
6
6
6 .
6
7
7 .
7
7
7
"3
S
8 -
10
4
15.
31.
5 M
F S
0 0
10
20
30
4 0
50
00
10
20
30
40
50
00
10
20
30
40
50
00
i 0
20
7 MI
00
10
20
30
40
50
00
10
20
30
40
r- ...
0 V
00
10
13
61
I H 1 T ~.
<;
0
f:
0
i)
o
2
4
c
1 1
T
0
0.
0
1
6
-
j
0
N U 1 E S
0
u
0
0
0
0
v
v
i)
0
V
0
'S
a
i)
v
0
0
9
0
4
9
4
c
0
0
•7
4
4
6
E:
4
( IHFUT )
i.
4
j
6
t.
g
7
•7
7
7
7
S
s
S
c
c
s
?
a
«2
j 2
2 2
3 2
. 4 2
52
0 2
12
22
. 32
4 2
. 0 2
. 12
. 22
32
42
c 2
02
12
r- ; ; -.- . . ; .^ r
<) 0
0 0
0 C
o o
0 0
d <•-
~ c.
7 ~
10 0
6 1
0 4
0 0
ft o
0 7
1 4
:- 1
4 0
2 . 5
S 9
£
'-,
z
;.
£
7
7
7
7
7
3
S
8
5
g
c-
9
?
04
1 4
2 4
3-i
44
54
0 4
1 i
24
34
4 4
57 ^
0J
14
. 2*
"7 4
44
5-s
04
1 ^
,;,
(j
0
0
^
2
2
14
f
>:•
0
0
2
=
4
1
( ILLUOAS)
0
(J
A
0
0
0
0
0
0
v
2
.-
0
6
6 .
6
S ,
5 .
<•
7 .
7 .
7
7
7
7
S
8
02
. 12
22
32
42
-_' C.
v S.
12
(t C
32
42
2T -*.
02
1 2
0 0
v w
A A
0 0
0 0
v 0
0 0
(- 0
0 v
v v
•x i
0 2
0 0
6
6 .
6
S
5
•i
7
7
7
7
•?
S
P.
04
1 *
24
34
si 4
;»
i'- 4
14
24
3»
4 4
54
04
1 -*
0
u
0
0
0
DRV tEhlrES FLO'- "T
-------
274
5
f
6
d.
i
t
,-
^
7
-
~r
~,
~-
\
':
."•
;r
c.
?CLTi
5£
. 06
. 16
. 2 t
3 £
•ft
. 56
vi
If
2f
. ;• i
4t
° f
C':
i i
2 >•
36
46
56
C t
1 V
: D K '
1 0
14
1
-------
275
MODELLING URBAN AND AGRICULTURAL DEVELOPMENT, HYDROLOGY AND
EUTROPHICATION IN A LOUISIANA SWAMP FOREST ECOSYSTEM
C. S. Hopkinson*
J. W. Day Jr.
In the past 30 years there have been rapid changes in the structure
and function of the ecosystems located in the headwater regions of
Louisiana's coastal estuaries (Craig and Day 1977, Butler 1975, Gael and
Hopkinson 1979) . Freshwater lakes that were once clear and coffee-
colored and that had abundant game fish populations (Lantz 1970) are now
characterized as being green and odoriferous, having frequent algal
blooms, occasional fish kills, and gizzard shad and catfish populations.
Large portions of swamp and marsh have been reclaimed and marginally dry
natural levee land drained for agricultural and urban use. An extensive
canal network and channelized bayous lead to rapid storm water runoff
from uplands to marsh lakes. This study investigates the relation
between land use, hydrology, nutrient loading and biological productivity
in the des Allemands swamp system.
DESCRIPTION OF AREA
The des Allemands swamp forest system is located in the headwaters
of the Barataria Basin Hydrologic Unit. The roughly triangularly shaped
unit is an interdistributary basin bordered by the Mississippi River,
Bayou Lafourche, and the Gulf of Mexico (Figure 1). Natural levees form
the perimeter of the basin and range in elevation from 9 m along the
artificial levees of the Mississippi River to 0.5 m at the junction with
the basin wetlands. Since the artificial leveeing of the Mississippi
*Coastal Ecology Lab Publication No. LSU-CEL-79-07, Louisiana State Univ,
/
-------
276
River overbank flooding no longer occurs, the only source of fresh water
is rainfall, which averages 152 cm/yr. Water flow is from the natural
levees into the interior wetlands and from the Mississippi River - Bayou
Lafourche basin apex to the Gulf of Mexico. The Gulfward three-quarters
of the basin is tidally influenced.
The des Allemands swamp forest system occupies the quarter of the
Barataria basin most remote from the Gulf of Mexico (Figure 1) . The
embankment of U.S. Highway 90 forms a barrier between the swamp forest
and the rest of the basin. The only significant connection is Bayou des
Allemands. The swamp forest system is a freshwater system without any
direct tidal effects. However, water levels are controlled to some
extent by lower estuary conditions. For example, prolonged southeasterly
winds raise water levels in the upper estuary.
An obvious feature of the area is the enormous number of linear
canals (Figure 1) , constructed primarily for agricultural drainage,
petroleum activity (pipeline and rig access canals), and navigation.
Drainage densities range from 2.9 in relatively natural areas of the
swamp to more than 800 in sugarcane fields on the natural levee uplands
(Gael and Hopkinson 1979).
The effects of the increased drainage network are manifested two
ways, (1) increased eutrophication of water bodies (Gael and Hopkinson
1979, Seaton 1979, Day et al. 1977, Butler 1975, Lantz 1970, Craig and
Day 1977), and (2) lowered swamp forest productivity (Conner and Day
1976, Conner unpublished). Gael and Hopkinson (1979) found a direct
linear correlation between drainage density and trophic state condition
of water bodies. As drainage density increased the trophic state also
increased. The reason for this is that nutrient-laden runoff water from
-------
277
the uplands is rapidly discharged to swamp lakes without overland flow
and cleansing through the swamp forest. The effects are also seen in
decreased biomass and productivity and species changes of trees (Conner
unpublished) . Aerial infrared photographs of the region vividly portray
this phenomenon. The reason is related to hydrologic changes. Canal
construction requires the placement of spoil material that invariably
ends up in the form of linear spoil banks running parallel to canals.
This prevents overbank flooding, retards water and material^exchange,
and causes prolonged ponding and stagnation.
OBJECTIVES
In this section the following subjects are investigated: (1)
urbanization and changing land use patterns in the des Allemands eco-
system in relation to storm water and nutrient runoff to the swamp, and
(2) the effects of several hydrologic regimes in the wetlands on nutrient
loading and eutrophication of Lac des Allemands. This is done using
Stormwater Management Model (SWMM; EPA 1971). Specifically the following
are considered: (1) present (1975) and future (1995) quantitative rates
of nutrient and water runoff from the rapidly developing natural levee
lands along the perimeter of the swamp as a result of changing land use;
and (2) hydrologic changes from the present such as water levels and
discharge rates that would result from (a) the removal of most spoil
banks, and (b) the removal of all spoil banks and wiering of several
major agricultural drainage canals; and (3) the problem of relating and
synthesizing data suggestive of increased nutrient uptake in the swamp
forest in a moving water overland flow regime. In particular this
section deals with the question of by what what percentage nutrient
loading to an already hypereutrophic Lac des Allemands can be reduced if
swamp hydrology is modified.
-------
278
THE SWMM MODEL
The SWMM is applied in this study to examine runoff and associated
nitrogen and phosphorus nutrients from the natural levee uplands into
the swamp and then to follow it through the swamp, through Lac des
Allemands, and out of the system via Bayou des Allemands. Runoff from
uplands is modelled with the RUNOFF block of SWMM and receiving effects
in the swamp forest are modelled in the RECEIVE block (see Figure 2).
Nutrient accounting is an integral part of RUNOFF and RECEIVE blocks.
In RUNOFF the availability of various nutrients is determined for the
land surface. Nutrients are transported to the swamp in conjunction
with water runoff from uplands. In contrast to RUNOFF, nutrients are
treated nonconservatively in RECEIVE. Water flow advects nutrients from
node to node, and depending on concentration and location are either
added to or removed from the water column.
DES ALLEMANDS SYSTEM MODEL
LAND USE
Geometrical conceptualization of flow patterns and estimation of
model coefficients and parameters are based on the area! extent of the
various land use categories existing in the des Allemands system pre-
sently and in the future. The relationship of land use to hydrology is
shown in the conceptual model (Figure 2).
Present and future areal extents of the major land uses in the des
Allemands system are presented in Table 1. Of particular interest are
the areas of various upland habitats. At present the majority of land
on the natural levee is devoted to agriculture with forest land also
being significant. Land devoted to urban activities such as residential,
industrial, etc. comprises only 7 percent of the total. By 1995,
-------
279
however, based on growth projections of the Louisiana State Planning
Office, land devoted to urban categories will increase by 9708 ha (321
percent). These projections vividly portray the imminent impact of the
LOOP superport development. Present plans call for onshore storage
facilities to be located in St. James Parish, which is within the study
area. The majority of the increased urban area will be at the expense
of agriculture land (70 percent). Smaller percentages will come from
reclaimed swamp forest (20 percent) and upland forest (9 percent).
CONCEPTUALIZATION
Physical abstraction of the system is the first step in imple-
menting the model (Figure 3). Separate procedures were followed for
RUNOFF and RECEIVE. For RUNOFF, natural levee uplands were conceptual-
ized into 11 different subcatchment areas, each based on all existing
input locations of storm water runoff into the swamp. A subcatchment
represents the drainage basin of an inlet to the swamp. Each subcatchment
is unique in size, slope, and drainage density. Abstraction of the
swamp, for RECEIVE, was more complicated. The swamp system was logically
divided into three types of physical elements: lake, bayou or canal,
and back swamp. In this finite element model each element is treated as
a node(s) (or junction) that is (are) characterized by area, and ele-
vation. The connection between nodes is represented as a channel that
has a characteristic elevation, width, length, and roughness coeffi-
cient. The swamp is conceived as a few major channels, representing
major agricultural drainage canals and channelized bayous, which lead to
Lac des Allemands. The lake is represented by five nodes. Swamp area
adjacent to these major channels is conceptualized as backswamp nodes
that are separated from the channels by levees. Channels connecting
-------
280
backswamp to bayou or canal are represented in two ways. The major
channel represents over-levee flow and has long characteristic widths
and depths that are above the level of backswamp areas. Small breaks in
levees are represented by channels with narrow widths and depths similar
to backswamp elevations. This conceptualization reflects the effect of
the enormous canal network: swamp impoundment. Currently most back-
swamps serve only as water storage areas. Water does not flow down-
stream at backswamp junctions. Rather water flows perpendicular to the
main bayou. If several canals, levees or roads that run perpendicular
to the downstream axis of the bayous were removed so that water could
flow downstream from backswamp node to backswamp node. The conceptual-
ization of this regime is shown in Figures 2 and 3.
PARAMETERIZATION PROCESS AND RESULTS
RUNOFF
Estimates were made of the coefficients and parameters that are
necessary to characterize the hydraulic properties of a subcatchment
such as surface area, drainage density, slope, infiltration rates
(maximum, minimum, decay), detention depths, roughness coefficients,
percent imperviousness, and canal morphology. Each subcatchment repre-
sents an idealized runoff area with uniform slope. It is assumed that
each subcatchment has a unique area, drainage density, and slope. The
other parameters are similar for each subcatchment and change only as
land use does.
Hydraulic properties for the subcatchments were determined by first
calculating them for each land use category and then by calculating an
average based on the relative area of all land use categories on the
subcatchments. Hydraulic properties assigned to each land use category
-------
281
are seen in Table 2. Parameters that vary the most for the different
land uses are percent imperviousness of the land and the roughness
coefficient for pervious land. Whereas about 100 percent of agricultural
and forest lands are pervious only 30 percent of industrial lands are.
Likewise pasture land has a high roughness factor while for urban and
industrial areas it is lower. The overall picture that emerges is that
urban land has hydraulic properties that will lead to rapid storm water
runoff. Pasture land will have the least rapid runoff rate.
In Table 3 subcatchment parameters that represent an average of the
various land use categories are listed. These were calculated using
land use ratios calculated for 1975 and projected for 1995. Land use
was not determined on a subcatchment basis but on a total upland basis.
Therefore each subcatchment is treated identically. Most runoff param-
eters change little or none from 1975 to 1995- Minor differences were
calculated for infiltration rates, detention depths on pervious areas,
and roughness coefficients on pervious areas. The most significant
hydraulic change will be the area of impervious land which increases
from 4 to 13 percent.
Runoff from each subcatchment is collected in separate canals and
then routed to the swamp. Output from the 11 canals serves as input to
the RECEIVE block. Table 4 presents hydraulic information of each
canal.
RECEIVE
Estimates for the RECEIVE block of swamp forest topography, junction
areas, water depths, channel lengths and widths, and roughness coefficients
were obtained by three methods. First, in the field the water slope
from node 6 to 45 was measured using USGS benchmarks at nodes 18 and 6
-------
282
(Figure 4) USACOE stage recorders at nodes 18 and 45. Surveyed levee
heights along stream channels were referenced to MSL using the calculated
water level of each node. Backswamp elevations and channel depths were
calculated likewise. Backswamp slopes were calculated by surveying a
transect from the uplands north of junction 20 to junction 20- It was
determined that at the swamp-upland interface land slope is essentially
0. (2) Junction surface areas were calculated by digitizing USGS 15'
quadrangle maps. Third, additional topographic and channel morphologic
information was gathered from Louisiana Department of Public Works
surveying and blue print drawings of more than a dozen major canal
network systems that encompass most of the swamp forest.
To simulate water flow through occasional spoil bank breaks a
parallel system of channels was simaltaneously used to connect two
junctions. One channel is characterized by large negative depths
(negative implies above MSL) and wide widths, while the other channel is
characterized by small negative depths and small widths. These channels
represent, in the former case, flow over the top of the spoil banks
(possible only when water reached that elevation), and in the latter
case, flow through bank breaks.
Water levels, areas, and bottom depths that describe each of the 45
junctions modelled are presented in Table 5. In Table 6 descriptive
channel parameters are listed. Manning's roughness coefficients were
estimated from SCS (1972). Backswamp channels had coefficients of 0.2,
bayous 0.06, lake 0.014, and drainage canals 0.028. Channels changed
for the simulation representing spoil bank and drainage canal removal
are described in Table 7. Fewer parallel channels were used to abstract
this condition and, therefore, there are only 73 channels versus 82 in
-------
283
the original setup. To represent the case of levee removal only, all
parallel channels were dropped. Instead the wide channel widths
associated with channels going over spoil banks were combined with
depths of the spoil bank break channels.
NUTRIENT ACCOUNTING
Nutrient parameterization was quite involved. First, nutrient
availability was determined by land use and then weighted with respect
to total area to estimate total loading per year to the swamp. This was
done using 1975 and 1995 land use areas. Loading was then divided into
storm events and the coefficients relating to runoff determined. Last,
uptake rates and equilibrium concentrations of N and P were determined
for each junction in both stagnant and flowing backswamp hydrologic
regimes.
Uplands
Nutrient availability on the upland subcatchments is a function of
land use. The quantities of nitrogen and phosphorus expected to be
available on various upland sources are listed in Table 8. With the
exception of agricultural nutrient runoff all values are from the
literature. Estimates from local studies, e.g. Craig and Day (1977)
were used when available. Local estimates were available for P associ-
ated with sewage and urban runoff. In the former case the Louisiana
value is considerably lower than that measured elsewhere (214 vs 795 g
P/person/yr), while in the other the estimate is slightly higher. On an
areal basis agricultural land contributes more N and P than any other
land use. Forest and cleared lands contribute the least. The N to P
ratio of these various sources varies from 155:1 for pastureland to
4.5:1 for cropland. Sewage, cropland, and urban lands are relatively
-------
284
rich with respect to potential P sources while pasture, cleared, and
forest lands are richer with respect to N.
Agricultural nutrient runoff estimates were calculated by con-
structing N and P budgets for each major crop and pasture. Budgets were
constructed assuming steady state conditions by solving material balances
for unknown inputs or outputs. Considered sources and sinks were
fertilizer, crop harvest, fire (volatilization), sediment erosion, and
i
runoff. Runoff of nutrients was in all cases the unknown flow.
Figure 4 portrays nutrient budgets of cropland and pasture. N
2
fertilizer inputs range from 0 to 209 g/m /yr for soybeans and vege-
table-orchard crops respectively. These same crops also had the lowest
and highest P application rates. As sugarcane and pasture together
represent more than two-thirds of total farmed agricultural land, it
should be mentioned that both receive high loadings of N fertilizer.
N and P that are removed from the land via harvest is highest for soy-
beans, a crop that receives no fertilizer. Nutrient loss is also high
for sugarcane.
Nutrient loss associated with sediment erosion was assumed to be
equal for all crops except pasture for which no sediment erosion was
assumed. Nutrient loss was calculated by multiplying the weight of
sediment lost with the percentage N and P of an average local soil type.
The loss of sediment lost with the percentage N and P of an average
local soil type. The loss ofsediment associated N and P was calculated
2
to be 12 g N and 4.8 g P per m per year. Sediment erosion rates were
assumed to be the same as that measured for sugarcane land in the Lake
Verret watershed (SCS 1976). Lake Verret is adjacent to Lake des
Allemands and has similar soil types and agricultural practices. The
SCS study reported that 2.55 kg sediment erodes from an average m of
-------
285
agricultural land. Eighty-two percent of all sediment erosion in the
watershed was attributed to agricultural land. Of this total 68 percent
was deposited in field ditches. The remainder reaches the swamp. Of
that, it was further calculated that 6.2 percent reaches the lake. Soil
nutrient analyses of Golden and Ricaud (1965) for Mhoon, Commerce, and
Sharkey soils which are found in the des Allemands area have N and P
contents ranging from .055 percent to 0.175 percent for N and 243 to 362
ppm for P.
At this point solving the materials balance for nutrient content of
each soil leads to substantial quantities of N and P in the soil which
are unaccounted for. If soil fertility is assumed not to be increasing
then the balance indicates that the surplus must be running off the
fields either in dissolved or particulate form. This precludes denitri-
fication. Patrick (personal communication) has done extensive agronomic
work in Louisiana coastal soils and suggests, that to account for
refractory N and P buildup in soils and denitrification, 10 percent of
the applied N and 1 percent of the applied P be routed from cropland to
the swamp. The range of values in Table 8 reflects what combination of
estimates is used. I believe that the best estimate for cropland includes
measured erosion plus 10 percent and 1 perdent of the applied N and P
fertilizer, respectively. For pastureland I use 10 percent of excess
fertilizer plus Shannon and Brezonik's (1972) estimate for phosphorus.
In Table 9 calculated loadings to the swamp in 1975 and 1995 are
shown. These figures were obtained by multiplying the expected loadings
per unit area by the ared of each land use category. Note that in 1975
about 75 percent of the N loading and 95 percent of the P loading comes
from agricultural land. Urban- and sewage-derived nutrients are minor
in comparison. In 1995 total loadings are projected to increase 28
-------
286
percent for N and 16 percent for P. The most dramatic differences in
1995 will be with respect to N. Agricultural loading will decrease,
while sewage related loading will increase. N loading related to sewage
in 1995 is projected to be almost triple that of 1975. P loading in
1995 shows the same trend. P runoff from croplands will constitute only
61 percent of the total in 1995 versus 93 percent in 1975, while loading
from sewage and urban areas will constitute 36 percent versus 14 percent
of the 1975-1995 totals respectively.
These nutrient budget calculations represent only a first step in
quantifying nutrient runoff to the swamp. Some sources and sinks, e.g.
nitrogen fixation, rain associated nutrients, groundwater, and denitri-
fication require elaborate sampling to be adequately evaluated. Esti-
mates from the literature are unavailable for this area. Consequently,
the budgets presented here are partial budgets; they estimate only gross
supply.
I assume that the loading rate of the above calculated annual
loadings is directly proportional to annual rainfall. If 60" of rain
falls annually, then 1" will transport 1/60 of the annual nutrient
loading budgeted. A value of 4.6 was used for the constant to determine
how much and how fast a percentage of nutrients exit per unit runoff.
This value dictates that 1/2 inch of runoff per hour will wash away 90
percent of the nutrients available per hour (EPA 1971).
Swamp
The fate of nutrients after reaching the swamp is a function of
concentrations and locations. Nutrients are either released from the
sediments to the overlying water or removed from the overlying water and
sorbed to sediments or incorporated into biomass.
-------
287
Nutrient uptake is considerably different in the swamp than it is
in the aquatic system. For simplification it's assumed that uptake
rates in the lake are similar to those in other aquatic areas such as
canals and bayous. Both N and P are retained at 0.0014 per day in water
bodies whereas only N is retained in the swamp (0.005 per day). Swamp
uptake was calculated from a nutrient accounting model and measurements
of surface water flux for the des Allemands Swamp (Kemp 1978). The N
uptake rate he found at a "natural" location (.01/day) was decreased by
an amount corresponding to the difference in surface water flux between
3
that site and the impounded swamp (901/1851 m /month). The rationale
operating is that nutrient retention is positively related to the
development and extent of an aerobic surface layer at the soil-water
interface. The rate of supply of oxygen to the interface is a function
of turbulence or water movement (Howeler and Bouldin 1971). Hence with
reduced water flux in the impounded area there is also a reduced aerobic
zone.
Uptake in water bodies was taken as the average of estimates based
on Butler's (1975) research in Lac des Allemands and empirically defined
relationships of Kirchner and Dillon (1975). In a nutrient budget
analysis of Lac des Allemands Butler found that 45 percent of incoming
nutrients to the lake were removed from the water column. Kirchner and
Dillon predict nutrient retention from a relationship between discharge
and lake surface area (areal water load). For Lac des Allemands 55
percent of incoming nutrients would be retained annually.
The above information indicated that if the present impounded
regime of the swamp forest system were replaced by one characterized by
significant overland flow, nutrient uptake rates would increase in the
-------
288
swamp. A crawfish pond hydrologically managed to mimic overland flow
was investigated by Kemp (1978), and his data has been used to estimate
nutrient uptake rates in an overland flow regime. Water levels and
constant sheet flow are controlled via pumps, levees, and drainage
culverts. In April 1977 Kemp analyzed crawfish farm water for N and P
along a transect from the input pump to an outlet culvert. Uptake rates
for N and P were calculated by reconstructing the hydrologic regime.
The owner provided information concerning farm area, water depth,
drainage rates, and pumping rates (1.38 x 106 m2, 45 cm, entire volume/5
days, or 1.24 x 10 m3/day, 1.42 x 10 m /day respectively). If the
5 3
farm can drain at 1.24 x 10 m /day without maintenance of head differen-
tials then with head maintenance drainage would be greater. It was assumed
5 3
that the owner was pumping an average of 1.33 x 10 m /day. The average
distance a parcel of water would have traveled from the pump to the
culvert was measured from a USGS 7 1/2" map to be 1466 m. Velocity
would then average 0.0034 m/s. Kemp's stations were a total of 1200 m
apart. The concentration of P decreased 33 percent, and of N 34 percent
as water moved those 1200 m. In the pond, the uptake rate of N and P
per day was 0.083 and 0.080 respectfully.
Swamp sediments have a buffering effect on water column P. The
amount of P in flood water and soils depends on the capacity of the soil
to release P to a solution low in P and to remove it from a solution
high in P (Patrick and Khalid, 1974). No direct research has been done
in the swamp to date that indicates the equilibrium concentration where P
is removed and added to the column at equal rates. However, the equi-
librium concentration estimate was based on Seaton's (1979) extensive
sampling in a backswamp location that receives little upland runoff.
-------
289
The average lowest concentrations of N and P found in the flood water of
this location were 0.41 and 0.07 mg/1 respectively. As concentrations
never go below this level and as allochthonous inputs are minimal, we
can assume that at these levels nutrients are released to the water
column. These concentrations are used in the model as minimal concen-
trations at all junctions. There was insufficient data available upon
which to estimate what these concentrations would be in an overland flow
regime with aerobic surface layers. Patrick and Khalid (1974) found
that under anaerobic conditions more P is released from the soil into
solution than under aerobic conditions. By assuming no difference I
underestimate true nutrient uptake in an overland flow situation.
Initial concentrations are from Butler (1975), Kemp (1978), and
Seaton (1979). Total-N and total-P concentrations were set at 1.54-0.25
mg/1 for stream junctions, 1.24-0.32 mg/1 for backswamp junctions, and
1.60-0.27 mg/1 for lake junctions, respectively.
SIMULATION RESULTS
Simulations were made for validation and predictive purposes.
Validation was done mainly by sensitivity analysis in RUNOFF and by
comparing simulation results to field observations in RECEIVE. Twenty-
four simulations of RUNOFF were made to assess the effect of (1) 1975-
1995 land use changes on water and nutrient runoff and (2) differing
initial soil water contents on runoff. Simulations were made of RECEIVE
to analyze water stages and discharges under four hydrologic conditions:
(1) present conditions using low initial water levels (stages below
elevation of channel spoil banks), (2) present conditions using initial
water levels 30 cm higher than 1 (above spoil banks) , (3) conditions
representative of spoil bank removal along Bayou Chevreuil, and (4)
-------
290
conditions representative of complete spoil bank removal and of upland
discharge into backswamp areas rather than into drainage canals. Simu-
lations were also made to compare storm water and nutrient runoff using
1975 and 1995 land use areas and to compare nutrient loading to Lac des
Allemands using swamp hydrologic conditions as in 1 and 4 above.
VALIDATION
Runoff
To properly calibrate and verify a model with large data inputs
such as RUNOFF, some amount of field measurements are necessary. Lack-
ing extensive field measurements I can partially validate this model by
showing that total runoff as calculated in the model for a single storm
event falls within ±10 percent of the total runoff calculated using a
static determination procedure (SCS 1972) that also relies on many
estimates of area hydraulic properties. Additionally, by doing a
sensitivity analysis those parameters that most affect the results can
be identified. If changes in unmeasured parameters do not affect
results too much, then one can have some degree of confidence in model
predictions.
Parameters that are adjustable and need field evaluation are
subcatchment width, area, slope, percent imperviousness, Manning's
roughness, surface storage, and infiltration rates. Of the 11 para-
meters high confidence is had in 6: subcatchment width, area, slope, and
infiltration rates.
Sensitivity analyses proved that with the exception of area and
slope (which must be considered as accurate), the percent imperviousness
and the minimum infiltration rate were the only variables that when
raised or lowered tenfold had major effects on runoff. By contrast,
-------
291
resistance factors and surface storage depths did not change the total
percent running off more than 1 percent for 2.8 cm rainfalls.
Changing subcatchment width an order of magnitude resulted in
increasing runoff slightly less than a factor of two. Changing the
maximum infiltration rate from an estimated maximum rate (signifying a
dry soil) to the estimated minimum rate (signifying a saturated soil)
resulted in a 1.9 fold increase in total runoff with a 2.8 cm rainfall.
However, for a 3.8 cm rainfall, runoff increased only 1.1 fold. This
implies that to accurately predict runoff from small rainfalls great
precision is needed in both determination of the maximum rate for a
particular soil and in determination of the present moisture content of
a soil.
Lowering the minimum infiltration rate by three-quarters caused
major changes in runoff volumes for 2.8 through 11.0 cm rainfalls.
Although the estimated true value was measured in a similar adjacent
system, future effort should be directed towards gaining accurate
measurements from several locations in the des Allemands basin.
Of all parameters the percent imperviousness proved to be the most
critical. However, as the primary goal of this study is to predict
runoff changes associated with changing land use areas, comparative
results will be valuable, as both are based on the same assumptions
regarding the imperviousness of various land use categories.
As the parameters whose estimations I have the least confidence in
are the least sensitive components of the model, simulation results can
be considered to be reasonably accurate.
Receive
Calibration and verification of RECEIVE results proved difficult,
-------
292
as field measurements were few and crude. The following information was
available against which model simulation results could be compared: (1)
Monthly current measurements taken for a one year period at junctions
18, 37, and 42 (Butler 1975), (2) water levels in Bayou Chevreuil
extrapolated from measurements at junctions 18 and 6 (this study), (3)
one time measurements of velocity and discharge at channels 11-14 and
11-30 (this study), (4) yearly average discharge predicted by Light et
al. (1973) in channels 18-16, 37-16, and 44-45, and (5) measurements of
water level drop at junction 20 during a period in which the level was
just below spoil bank crest (Whitehurst 1977).
RECEIVE is started from a standstill condition using measured or
estimated water level values at each junction. The first day of simu-
lation represents a transient state where water levels and velocities
are approaching a dynamic steady state. The acceptance of hydraulic
inputs from RUNOFF begins on the second day.
Model verification is made using present hydrologic conditions and
low initial water levels, no inputs from RUNOFF, and results from day
two. Butler (1975) recorded surface currents at junctions 18, 37, and
45 monthly for a one-year period. The range and average velocity at
each location was as follows: 1-41 cm/sec - 22 cm/sec average; 5-30
cm/sec - 16 cm/sec average; 4-41 cm/sec - 20 cm/sec average, respectively.
Current measurements taken at several depths did not consistently show
logarithmic profiles with depth. Speeds predicted in the model were 13,
6, and 12 cm/sec at junctions 18, 37, and 45, respectively. Model
results assume cross-sectional homogeneity in flow hence are expected to
be less than surface measurements. Simulated speeds do agree favorably
with the range observed in the field.
-------
293
After two days of simulation water levels in Bayou Chevreuil
(channel running from node 6 to 16) rose an average of 16 cm (SD = 8 cm)
at each junction. Although water levels rose the overall water level
slope from junction 6 to 18 fell only slightly (2.39 vs 2.46 cm-km'1).
These results suggest that initial water levels used in the model were
somewhat in error, but that the parameterization of the hydraulic
character of this portion of the swamp is reasonably accurate. Otherwise
slopes would have changed more.
Several cross-sectional and vertical measurements of water currents
were measured at the intersection of channels 11-14 and 11-30 to deter-
mine what percentage of flow followed each channel and to check model
predictions. Average integrated velocities and discharges for the
channels were 7.1 cm/sec - 6.4 m/sec and 10 cm/sec - 5.6 m3/sec, respec-
2
tively. Model results show 8.2 cm/sec, 7.3 m /sec, and 12.2 cm/sec -
3
6.8 m /sec, respectively. The high degree of similarity among observed
and predicted currents, discharges, and percentage of discharge taking
each route lends confidence in the accuracy of the model. Roughly two-
thirds of the flow goes down Bayou Chevreuil, while the remainder goes
toward Bayou Boeuf.
Light et al. (1973) developed computer programs to evaluate the
flow regime in the Lac des Allemands system. Mean annual discharge was
plotted at various locations in the swamp. Results indicated that
discharge into Lac des Allemands from Bayou Chevreuil (channel 19-16)
averaged 11.0 m3/sec. Discharge south from junction 45 was estimated at
27.0 m3/sec. Although Light's results were not verified, agreement with
present results should lend credence to both models. Results from day
two of RECEIVE indicate the discharge from Bayou Chevreuil to be 9.6
-------
294
-i 3
m /sec. From junction 45 discharge was 7.1 in /sec. Agreement is
excellent between discharge from Bayou Chevreuil. Slightly lower flow
in RECEIVE is expected as low water conditions were simulated. Flow
from junction 45 is controlled by the head differential between Lac des
Allemands and lower Barataria estuary. A tidal function with zero
amplitude is used in the model to represent the downstream head. The
downstream head causes water to pile up in Lac des Allemands. Conse-
quently, discharge is lower than that shown by Light.
Due to the complexity of factors controlling water levels at
junction 45 (des Allemands) this analysis of hydrologic and nutrient
RECEIVE simulation results is restricted to the Bayou Chevreuil basin.
This includes backswamp and stream junctions upstream of channel 16-19.
For this reason, developing an accurate representation df forcing
functions acting at junction 45 can be avoided. The remainder of the
des Allemands system can be used as a forcing function acting on Bayou
Chevreuil.
Whitehurst (1977) measured water levels at junction 20 during April
1977. To see if RECEIVE accurately represents field observations, his
measurements of water level decrease when initial water levels were just
below streamside spoil bank crests were reviewed. Whitehurst's obser-
vations were separated into two components: (1) Evapotranspiration, and
(2) water flow through spoil bank breaks. Evapotranspiration was calcu-
lated using Thornthwaite and Mather (1955) techniques and local newspaper
(Morning Advocate, Baton Rouge, LA) listings of temperature. Total
water level drop not due to evaporation was attributed to flow. In
RECEIVE the same evaporation rate was used. Water levels decreased 1.1
times faster in RECEIVE than as measured by Whitehurst. This close
-------
295
agreement tends to verify model results. It is assumed that if there is
close representation of junction 20 hydrology, other backswamp junctions
will also be adequately modeled.
Agreement among simulation results of RECEIVE with available field
data suggests that this system has been suitably abstracted and modelled.
The verification process showed that flow from junction 45 is not
necessarily accurate, and for that reason analysis of simulation results
is restricted to the upper portion of the swamp system only: Bayou
Chevreuil basin.
SIMULATIONS
Runoff
RUNOFF simulations were made to assess runoff and nutrient loading
using three different rainfall rates (2.8, 5.6, 11.2 cm/storm) and five
different infiltration rates under 1975 and projected 1995 land use
makeups. These results are shown in Figures 6 and 7. The 2.8 cm rain-
fall represents actual hourly rates measured at Gonzales, LA, 1 March
1977.
Runs 1-15 represent present land use areas while 16-30 represent
1995 land use areas. Runoff increases with increasing rainfall using
/
present and future land use areas. As the maximum infiltration rate is
lowered from 4.1 cm/hr to 0.2 cm/hr total runoff increases. Runoff
increases more with the 2.8 cm rain than with the 5.6 and 11.2 cm rains.
Lowering the minimum infiltration rate from 0.2 cm/hr to 0.05 cm/hr
leads to a much greater increase in runoff than a comparable decrease in
the maximum infiltration rate.
The increase in runoff under a 1995 land use scenario is the most
important result of these simulations. The 1975-1995 runoff difference
-------
296
is greatest with dry soil conditions and low rainfall. Runoff increases
4.2 times over what it was with 1975 conditions. With high rainfall,
however, the increase is a more modest 7 percent over 1975 runoff
volumes. There is less difference found between dry and wet antecedent
moisture conditions using 1995 land use areas than when using 1975
areas.
As 85 percent of all rainfall episodes have 3.2 cm or less of rain
per day the increase in impervious area by 1995 will have dramatic
overall effects on runoff volumes to the swamp. Standing, stagnant,
impounded backswamp areas are already experiencing productivity de-
creases. Additional runoff to the swamp can only exacerbate the problem.
Receive
Four RECEIVE simulations were made to assess storm water flow under
different hydrologic regimes using RUNOFF as the driving force. Regime
1 (RUN 1) is based on a physical abstraction of present hydrologic
conditions. Initial water levels are those measured in February 1978.
Levels are low. Regime 2 (RUN 2) uses the same hydraulic properties as
RUN 1 but starts with water levels 30 cm higher. In Run 1 water levels
in the main streams are lower than spoil bank elevations. In Run 2
water levels are high enough to allow flow over many spoil banks.
Regime 3 (RUN 3) describes conditions that simulate spoil bank removal
in many places. Water flow directions and initial levels are the same
as Run 1. Backswamp flow remains at right angles to major stream flow.
Regime 4 (RUN 4) has been physically abstracted to represent runoff
directly into backswamp areas. Major agricultural drainage canals are
lacking, which allows possibility of downstream overland flow. All
spoil banks are removed. Water levels begin similar to Run 1.
-------
297
RUNOFF uses 1975 land use areas, a 5.6 cm rainfall, and a maximum
infiltration rate of 0.81 cm/hr. This infiltration rate corresponds to
the annual average moisture content of soils in the area.
Simulations 1 and 2 (RUN 1 - RUN 2) compare water levels and currents
for ten days during and after a storm. Rainfall occurs during day 1.
In Figure 7a water levels are plotted for three junctions along Bayou
Chevreuil. Levels rise more at the upper portion of the basin (junction
4 and 11) than at the lower portion (junction 18). Bayou Chevreuil
water levels rise more in RUN 1 than RUN 2. The reason for this is that
in RUN 1 water levels are initially considerably below spoil bank
crests. All water input from the uplands is confined to canals and
streams. A large volume of water added to the small volume capacity of
the streams causes levels to rise significantly. In RUN 2, however,
stream levels are already above levee crests, and the large runoff
volume is not confined to streams but is spread over the entire swamp.
In both simulations the maximum water level rise occurs on the same day
as the storm. An important difference between these runs is that most
water levels recede faster in RUN 2 than RUN 1. This is because water
that is impounded in the backswamp in RUN 1 is released slowly, while in
RUN 2 water is released rapidly until levels approach that of spoil bank
heights. This simulation shows that impoundment buffers stream water
levels.
Figure 7b shows water levels at backswamp locations adjacent to the
above stream junctions. Compared to water levels in the stream, less
difference is seen between these runs. Water levels at junctions 12,
13, and 20 rise less in RUN 1 than the rise measured at adjacent stream
junctions (Figure 7a) . In RUN 2 the backswamp and the stream rise
-------
298
comparably. This shows that in RUN 1 water is confined for the most
part to the streams while in RUN 2 water is rising everywhere. In both
runs water levels at junction 13 reach a maximum at a day other than day
1. Stage recessions are more rapid in RUN 2 than RUN 1. This agrees
with the previous observation concerning stream recession rates. The
buffering effect of impoundment is shown. Slow release from the back-
swamp causes slow stage recession in streams. In RUN 2 discharge to Lac
des Allemands is about 80 percent higher than RUN 1. Flow through
backswamp channels, e.g., 18-20, 20-16, 18-21, 21-16 is much greater in
RUN 2 than RUN 1. This is as expected as in RUN 1 water levels are not
high enough to flow into and through the backswamp.
Two other simulations examine the effect of levee and/or upland
drainage canal removal. RUN 3 has the same channel grid as RUN 1 but
all parallel channels are omitted. Instead depths of channels into the
backswamp are the same as the backswamp elevation itself. RUN 4 also
simulates levee removal, but upland dishcarge to the swamp is via
overland flow and not via canals. Canal removal along with associated
levees allows water to flow parallel to original canal axes, thus
allowing considerably more downstream overland flow in the backswamp.
RUN 3 and 4 use RUN 1 initial water levels. These are two alternatives
that land owners or the state could examine if an attempt to manage the
swamp were deemed beneficial.
RUN 1, 3, and 4 hydrographs for Bayou Chevreuil junctions 4, 11,
and 18 are shown in Figure 7c. Hydrographs for RUN 1 and 3 rise more in
the upper basin than they do at junction 18. At junction 4 RUN 1 and 3
hydrographs rise more than two times higher than the RUN 4 hydrograph.
At junction 11 all runs show about the same water level rise on day 1,
-------
299
but while they then fall in RUN 1 and 4 they continue to rise through
day 10 in RUN 3. At the lower end of Bayou Chevreuil initial hydro-
graphs are comparable, but then water levels continue to rise in RUN 4
while they fall in the other two runs.
Backswamp hydrographs for RUN 1, 3, and 4 exhibit a different
pattern than that exhibited in the stream (Figure 7d). As mentioned
previously backswamp water levels in RUN 1 do not rise as much as the
adjacent bayou levels. In RUN 3 and 4, however, levels in the backswamp
rise considerably. RUN 4 contrasts the pattern of RUN 1 the most.
Water levels at junctions 12 and 13 rise more than they did in the
bayou. This demonstrates the flux of water directly to the backswamp
from the uplands. At node 13 in RUN 3 water levels rise for 9 days, as
long as water levels rise in the adjacent bayou. At junction 20 hydro-
graphs decline from day 1 maximums for RUN 1 and 3 and climb through day
9 for RUN 4. Hydrographs show that upper backswamp recession rates are
greatest in RUN 4. Analysis of RUNS 1, 3, and 4 indicates that upstream
water levels fluctuate most in RUN 4. In RUN 4 the upper Chevreuil
basin drains rapidly and water piles up in the lower basin.
Discharge rates past junction 18 reveal several changes that occur
when swamp geometry is changed. RUN 3 has 1.4 times the discharge of
RUN 1 and RUN 4 is greater still. In RUN 4 70 percent of the total flow
to junction 18 is via the bayou. The remainder flows through the back-
swamp as overland flow. By way of contrast in RUN 1 only ten percent
of total flow to junction 18 is through the backswamp. Spoil banks
effectively prevent overbank flooding and significant downstream,
backswamp flow. Water drains rapidly from the upstream regions of Bayou
Chevreuil in RUN 4 and piles up in the lower region. Heads at junction
-------
300
18 are higher in RUN 4 than in RUN 3 or RUN 1. The higher head at 18 is
partially responsible for maintaining greater discharge to the lake in
RUN 4 than in the other simulations.
Nutrient Model
The effect of land use area changes on nutrient runoff from the des
Allemands uplands was shown in Figure 6. With higher rainfall episodes
runoff of N and P is close to that calculated statistically in Table 9.
With rainfalls less than or equal to 2.8 cm/day, low soil moisture
content leads to nutrient runoff that is up to 2.1 times greater under
1995 conditions than under 1975 conditions. Soil conditions for 1995
lead to increased water runoff and consequently causes increased nutrient
runoff.
Nutrient loading to Lac des Allemands from the Chevreuil basin was
determined using RUN 1 and RUN 4. The latter was run with projected
1995 land use areas and nutrient runoff estimates while the former uses
1975 conditions. RUN 4 was chosen because unlike RUN 3 it does ex-
perience considerable downstream overland flow. Many backswamp areas in
RUN 3 experience no or very slow back and forth flow from the bayou.
Backswamp current speeds averaged 0 in RUN 1, 0.49 in RUN 3, and 0.81
cm/sec in RUN 4. As nutrient uptake is proportional to the development
of an aerobic surface layer, RUN 4 is most likely to lead to such
development. Sparling (1966) showed that at velocities <0.4 cm/sec 07
was depleted in a bog. At velocities <1.0 cm/sec waters remained
agitated and saturated with 0-. RUN 4 has the highest backswamp
velocities and comes closest to resembling conditions under which
nutrient dynamics were determined in the crawfish pond.
The nutrient load to Lac des Allemands is calculated as the flux
-------
301
volume of water times concentration. Figure 8 shows total flux from day
I through day 10 to Lac des Allemands from the Chevreuil basin. Not
only is discharge greater in RUN 4 but it continues to increase through
day 9. RUN 1 has peak discharge one day after the storm. In Figure 9
cumulative discharge for 10 days is plotted for RUN 1 and RUN 4. Total
discharge is 22 percent higher in RUN 4 versus 1, and a larger percentage
of floodwater flows through the backswamp than through the bayou. Al-
though nutrient runoff into the swamp was projected to increase 25
percent and 18 percent respectively for N and P between 1975 and 1995,
loading to the lake in 1995 under RUN 4 hydraulics is projected to
actually decrease 23 percent for N and 28 percent for P. This readily
shows the effect of increased uptake rates in backswamp areas. In RUN 1
23 percent of total flow passes along channel 16-21 while in RUN 4 28
percent does. Although more water flows through this channel in RUN 4
considerably fewer nutrients reach the lake. A larger portion of
nutrients entering this backswamp junction are removed from the water
column in RUN 4 than RUN 1. Consequently loading is lowered. The same
occurs with P. Water flow increases in RUN 4 but because of enhanced
nutrient uptake in an overland flow regime nutrient loading is lowered.
MANAGEMENT IMPLICATIONS AND CONCLUSIONS
The ramifications of RUN 4 results are immense. The estimated
decrease in nutrient loading to Lac des Allemands warrants additional
scientific research to further verify model results. Contrary to
commonly held beliefs, simulation results suggest that removal of levees
and agricultural drainage canals could actually increase the rate of
water runoff to Lac des Allemands with no increase in flooding of
agricultural fields.
-------
302
Nutrient loading to La;- des Allemands is presently 3.0 gN/m2/yr
and 4.0 gP/m2/yr (Craig and Day 1977). RUN 4 results suggest that
2 2
loading can be reduced to at least 2.3 gN/m /yr and 3.2 gP/m /yr if
reduction rates modeled in the Chevreuil basin can be extrapolated to
the remaining des Allemands system. As RUN 4 results are based on
higher than present loading rates, the above reductions should be
considered as conservative estimates. Shannon and Brezonik (1972)
studied the relationships between lake trophic state and N and P loading
rates. If loading can be reduced to RUN 4 levels then the lake trophic
state could be reduced 43 percent. This kind of change would help
return the lake to former conditions.
In addition to lowering the trophic state in Lac des Allemands,
hydrologic modifications similar to those modelled in RUN 4 could also
increase swamp forest productivity. In addition to fertilization the
Ji •-•
improved drainage regime would lead to enhanced forest growth. In the
crawfish farm that has an artificially controlled hydrologic regime with
drained periods, Conner (unpublished) found forest productivity to be
twice that in uncontrolled and Impounded areas of the swamp. As RUN 4
conditions could lead to conditions similar to those found on the craw-
fish farm one could expect similar productivity increases throughout the
swamp.
The benefits realized by managing swamp hydrology are not restricted
to the forest and lake but extend to the entire des Allemands system.
Natural energy inputs associated with primary production could be
increased approximately twofold. As primary production accounts for
half of all natural energy inputs a doubling would increase total
natural energy 1.3 times. With the level of industrial and urban growth
-------
303
predicted for the region this area will need the potentially available
increase in natural energies to match the increase of purchased, human
system energies. As the investment ratio between these two energies is
already beyond the optimal value of 2.5 (Odum and Odum 1976, Hopkinson
1979) any increases in natural energies to match the higher quality
purchased energies will enable more productive use of human system
energies.
In summary this modelling exercise has shown that the projected
urban and industrial expansion in the des Allemands system will lead to
increased nutrient loading and increased storm water runoff to the
wetlands. Increased runoff is attributed mainly to the increase in
impervious areas. It will be most pronounced with the predominant
rainfalls of under 2.5 cm. Modelling swamp hydrology showed that the
present system of canals and spoil banks causes impoundment of swamp
areas and does not optimize discharge rates to Lac des Allemands.
Backswamp areas act as buffers, storing water for slow release. Simu-
lations show that the hydrology could be managed to increase discharge
rates to the lower estuary, to increase productivity of the swamp forest
and to decrease lake eutrophication. This could be done by removing
spoil banks and allowing upland runoff to pass through the backswamp in
place of through drainage canals. Such management would be welcomed by
nature lovers, sugar cane farmers, hunters, fishers, and those who
appreciate the aesthetic appeal of healthy wetland ecosystems.
We wish to acknowledge and thank several individuals who greatly
assisted in the construction and performance of this modelling exercise,
B. T. Gael, Tan Poh Chuan, and Wayne Huber. This research was supported
by the Louisiana Sea Grant College Program, a part of the National Sea
-------
304
Grant College Program maintained by the National Oceanic and Atmospheric
Administration, U.S. Department of Commerce. The U.S. government is
authorized to produce and distribute reprints for governmental purposes
notwithstanding any copyright notation that may appear hereon.
-------
305
REFERENCES CITED
Butler, T. 1975. Aquatic metabolism and nutrient flux in a south
Louisiana swamp and lake system. M.S. thesis, Louisiana State
University, Baton Rouge, 58 pp.
Chow, V. T. 1964. Handbook of Applied Hydrology. McGraw-Hill Book
Co., New York.
Conner, W. H. , and J. W. Day. 1976. Productivity and composition of a
baldcypress water tupelo site and a bottomland hardwood site in a
Louisiana swamp. Amer. J. Bot. 63(10):1354-1364.
Craig, N. J., and J. W. Day. 1977. Cumulative impact studies in the
Louisiana coastal zone: Eutrophication; Land loss. Final report
to Louisiana State Planning Office, Baton Rouge.
Day, J. W., T. J. Butler, and W. H. Conner. 1977. Productivity and
nutrient export studies in a cypress swamp and lake system in
Louisiana. Pages 255-269 in Estuarine Processes. Vol. II: Circu-
lation, Sediments, and Transfer of Material in the Estuary.
Academic Press, New York.
Environmental Protection Agency. 1971. Storm water management model.
U.S. EPA Water Pollution Control Research Series, 11024DOC07/71.
Gael, B. T., and C. S. Hopkinson. 1979. Drainage density, land use and
eutrophication in Barataria Bay, Louisiana. Pages 147-165 in J. W.
Day, Jr., D. Culley, R. E. Turner, and A. Mumphrey (eds.), Third
Coastal Marsh and Estuary Management Symposium. Louisiana State
University, Div- of Cont. Ed., Baton Rouge.
Golden, S., and J. Ricaud. 1965. Foliar analysis of sugarcane in
Louisiana. Louisiana State University Ag. Exp. Sta. Bull. #588.
Hopkinson, C. S. 1979. The relation of man and nature in Barataria
Basin, Louisiana. Ph.D. diss., Louisiana State University, Baton
Rouge. 236 pp.
Howeler, R. H., and D. H. Bouldin. 1971. The diffusion and consumption
of oxygen in submerged soils. Soil Sci. Soc. Am. Proc. 35:202-208.
Kemp, G. P. 1978. Agricultural runoff and nutrient dynamics of a swamp
forest in Louisiana. M.S. thesis, Louisiana State University,
Baton Rouge. 57 pp.
Kirchner, W. B. , and P. J. Dillon. 1975. An empirical method of esti-
mating the retention of phosphorus in lakes. Water Resources
Research 2(1), 1975.
Lantz, K. 1970. An ecological survey of factors affecting fish prod-
duction in a Louisiana natural lake and river. La. Wildlife and
Fisheries Comm. Bull. #6, New Orleans. 92 pp.
-------
306
Light, P., R. J. Sehlemon, P. T. Culley, and N. A. Roques. 1973.
Hydrologic models for the Barataria-Terrebonne area, southcentral
Louisiana. Hydrologic and Geologic Studies of Coastal Louisiana,
Rept. #16. Louisiana State University Center for Wetland Resources,
Baton Rouge. 43 pp.
Nix, C. 1978. Personal communication. Soil Conserv. Service, USDA,
Alexandria, La.
Odum, H. T., and E. C. Odum. 1976. Energy Basis for Man and Nature.
McGraw-Hill, New York. 287 pp.
Patrick, W. H., and R. A. Khalid. 1974. Phosphate release and absorption
by soils and sediments: Effect of aerobic and anaerobic conditions.
Sci. 186:53-55.
Soil Conservation Service. 1972. SCS Natl. Engineering Handbook, Sec.
4: Hydrology. U.S. Dept. Agric.
Soil Conservation Service. 1976. Environmental Impact Statement, Lake
Verret Watershed. (Draft) U.S. Dept. Agri., Alexandria, La.
Seaton, A. M. 1979. Nutrient chemistry in the Barataria Basin: A
multivariate approach. M.S. thesis, Louisiana State University,
Baton Rouge.
Shannon, E. E., and P. L. Brezonik. 1972. Relationships between lake
trophic state and nitrogen and phosphorus loading rates. Envir.
Sci. and Techn. 6(8):719-725.
Sparling, J. H. 1966. Studies on the relationship between water
movement and water chemistry in Mirex. Can. J. Bot. 44:747-758.
Whitehurst, C. A. 1977. Continuation of interpretation of remote
sensing data in the Bayou Lafourche delta of south Louisiana.
Final report to NASA, NGL 19/001/105.
-------
307
Table 1. Present and projected land use in the des Allemands swamp
basin.*
Area (hectares)
Land Use ^975 1595
Wetland Total 100,619 98,678
Swamp 89,413 87,569
Water 4,706 4,609
Lake 6,500 6,500
Dryland Total 61,700 63,641
Residential ,
commercial, and
Industrial 228 1 404
Cropland 20,698 17,459
Pasture 14,135 11,930
Field Support 9,086 7,639
Forest 13,392 12,519
*Based on information in Hopkinson (1979) , projection for 1995.
.(From Louisiana State Planning Office)
Table 2. Estimated values of runoff parameters.
LAND USE CATEGORY
Manning's resistance
factor (n)
Impervious land 0.014 0.014
Pervious land 0.075 0.075 0.30
Detention depth (cm)
Impervious land 0.16 0.16
Pervious land 0.38 0.38 0.38 0.48
Morton's infiltration
rate (cm.hr~l)
Pervious - Max* 3.81 3.81 3.81 5.08
Min 0.18 0.18 0.18 0.25
Decay rate 0.008 0.008 0.008 0.008
Percent Impervious Land 50 70 0 0
Percent Pervious Land 50
A
-1,941
-1,844
97
0
+1,914
+8,532
+1,176
-3,239
-2,205
-1,447
- 873
orest e erence
EPA (1971)
Nix (1978)
Chow (1964)
SCS (1976)
0.25
EPA (1971)
Nix (1978)
0.38
Nix (1978)
3.81
0.18
0.008
0 FPA (1971)
la. State
Planning
Office
*Based on soil moisCure content of zero.
-------
Number (m)
'9S 10549
2 1RR33
3 4075
4 18386
5 6-579
6 9933
7 11941
B 8476
9 13290
JO 14724
11 14t,96
J t
(h»)
3434
7456
4015
6937
2561
3495
3141
1333
4745
4368
4971
Irapervloua
13.05 0
0
0
0
0.
0
0
0,
0.
0.
0,
arfl.e[erfl Ufl,d or Sl^UtioB. ,« on .rcent an « „ ot. WftD,P »
Slope (n) Detention Depth (en) Infiltration (c»*hr a) ('ifc-l)
lmpPrvtou9 Pervious Impervious Pervious H^xlmym Minlmua Dscey
.0014 0.014 0. 16? 0.019 0.048 4,06 0.19 0.003
.001
.0005
.0012
.0008
.0007
.0014
.0014
.0013
.0010
.0011
TABLE 4 • Hydraulic Properties of Canals Draining Des Allemands Uplands.
Canal Tributary Width
Depth (m) Subcatchraent (m)
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
10
11
18.3
18.3
14.6
18.3
14.6
14.6
12.2
12.2
9.8
9.8
24.4
Length
(m)
2634
4641
1021
4493
1737
2438
2972
2073
3353
3696
3635
Slope
(m/m)
56915
19418
10044
15000
10044
10004
13057
17410
56915
56915
56915
Manning
Roughness
0.028
0.028
0.028
0.028
0.028
0.028
0.028
0.028
0.028
0.028
0.028
Overflow Depth
(m)
2.4
2.4
2.1
2.6
2.6
2.7
2.4
2'. 4
3.7
4.9
4.9
-------
309
i i.:
2 1.0
3 1.1
4 1.1
5 0.8
6 0.9
7 0.7
3 o.a
9 0.7
1.0 0.6
11 0.9
; i a . c
i •* C . 3
i.5 •'. ^
La 0. <»
17 0.:
;:• o )
1^ 3. ;
;c 3.,
;i o. »
22 0.4
23 0.3
24 5."
:s 0.3
26 0.4
:7 0.3
29 0-3
30 0,4
31 0-7
32 0.3
33 0-4
34 0.3
35 0-4
36 0.3
3? 0.2
38 0.2
39 C.:
40 0.2
41 0-2
42 0.2
41 0.2
44 0.2
45 0.2
5.n o.js
ifl. " 0 ''
5134,9 i'ixj
'500.0 i. 10
19.8 -0.55
4.3 -?.,•»
15.7 -0.61
7193.3 0.5$
13.2 -O.M
32.1 -C.08
13.9 -0.11
3123.0 c!*3
33.3 -1.46
;;,95. i o i;
13.7 -1
1606.9 0. 9
-0, !> -1.2
31,3 -L- ^
4318.8 0, 3
3601. } 0. 0
34.0 -0. 1
6354.1 0. 0
15.9 -0. 1
}816.0 0, 0
18.0 -0. I
6757.4 0. 0
9084.6 0. 0
13.- -0.91
719.3 0.41
2383.8 0.38
2444.2 0.34
1211.9 C.58
4497.: 0.43
1815.1 0.58
3717.5 C.<*J
2958.* 0.27
ic35.3 -i.s:
1164.J -1.52
765.3 -1.52
547.4 -i.52
:0»C.O -1.52
B89.*. -i.s:
38*. ft -I. >i
99.4 -1.52
CVnn«l Coord tru
1 :
1 2
3 2
i J
5 4
6 ]
7 3
S 3
9 4
1C 10
U 11
12 4
13 9
14 !0
15 U
16 1!
i? a
i« u
19 14
20 U
21 14
22 14
24 a
25 a
26 9
27 0
:a 9
29 I
JO 6
3i 22
3! 22
33 23
3* 39
H 39
36 40
37 41
38 42
39 16
40 16
41 24
42 25
43 2S
44 26
4S 41
-4 2;
47 30
48 31
44 33
50 34
51 32
S2 15
S3 36
54 17
ss :«
37 2J
SB 4)
59 44
60 19
61 9
62 3
63 40
u ;
63 :
66 5
67 7
68 S
69 a
70 12
n 14
72 ;4
73 19
74 30
75 31
76 13
77 34
78 32
79 35
30 36
81 37
42 1$
1
1
12
12
10
10
i j
il
20
30
14
38
17
i5
la
19
21
20
16
16
16
}•)
39
23
39
40
42
42
42
43
42
43
40
40
4
i,
4
,
3
32
34
37
37
36
37
16
45
43
44
45
21
38
5
41
3
4
a
8
10
12
13
17
IS
;o
2
Z
4
7
7
b
37
16
21
7^4 8
2747.8 9
2752.6 10
6S45.B
6834.6 7
2967.2
3137.0
5172.2 4
)632.3 3
-J62.3 4
7-6:6.7 4
4)05.3
6527.0
43S8.2
2516.4 9
3126. ft
9648.1
6283.1
5376.7
1313. J :
2165.6 ?
2757.8
6076.8
3373,0 4
1710.3 6
6302.7 6
7139.3
5601 . 3 2
7805.6 1
8593.5
823,2.1
2804.5 9V
6462.1 I
sail.* 3
SSOL.O :
3409.8 3
485C.& J
5291. t 4
6113. 1 2
4166.6
'S94.6 6
2J85.6 7
8450.9
5J06.3 I
5 jo.. 5 1
8177. fl
7647.7
4168.1
7611. 8
10524.0
4727.8
6329.2
6016.9
13.2 0.35B
52.6 1.164
38.0 1.U4
5.2 0.171
3.8 I.Q15
2.0 -o.s:
6.3 -0.53
1.3 I,"
4.4 I.M
0.6 i.i:
9-5 1.3:
9.1 -0.6t
S. 3 -O.JJ
2.7 -0.9S
5.' 3.91
0.5 -i.C2
a. 3 -i.s:
;s.7 -i.?.
u.» -i.;
)4.6 0.99
24.8 ;.;i
45.7 -l.»9
fc5.7 -1.-.3
5.2 0.4
4.3 O.Si
2.9 0.30
5.J ~I.fa8
0.0 0.10
41.7 -1.52
39.6 -0.91
9.1 -C.i-b
81.3 3. 29
50.0 -1.52
91.7 -1.52
33.5 -1.52
16.3 -1.52
33.6 -i.s:
83. b -1.52
25.1 -1,5?
30.5 -0.98
19.4 0.29
35.3 0.40
21.3 -O.SS
06.4 -1.52
u3.j 3.*
18.3 -1.12
5.2 -1.37
1.3 -1.46
1.3 -1,95
7.4 -1.83
1.3 -1.46
1.3 -1.95
45.7 -1.37
3840. S 782 .2 0.305
62 8. 9 2 .3 -0.<*8
42 7.2
56 8.8
15 4.1 6
70 3.1
6444.0 4
4401.9 1
2747. S 1
2752.6 2
S172.2
3632.3
43*2.3
76.26.7
2516.4 1
1911, 3
2165.6
1985.2
6t>03.a i.
66QS.6 3
3627,7 6
7325.3 4
9520.1 2
4572.0 3
5120.6 5
S503.5
1S34.I
9 .5 -1.52
2 .3 -1.52
0 .0 0.43
1 .2 -1.07
4 .2 1.0
6 .6 -i.s:
o .a i.oi
09.4 1.01
7.5 0.35
0.9 0.
3.8 0. 8
S.l 0. 6
5.1 o. a
5.2 0. 8
9.3 0. 1
6,3 0. 05
63.9 0. 7
43.0 0. 6
42.9 0. 9
11. 8 0. 7
75.6 0. 05
23.7 0. 9
46.8 0. ;
11.3 0. 7
23.3 0, 0
O.G6
O.C23
0.06
0.2
Q.Ob
0.2
0.014
o.c:a
0.06
o.:
0,0:4
O.Oli
0.014
o.ou
O.Oli
0,014
o.o*>
0-iO
0,2
0,023
0,014
0,06
C,0fc
O.C28
0.023
0.06
0.06
0.06
0,06
o.;
0.2
-------
310
Table
Channel
2
3
6
8
9
10
11
13
15
17
19
20
21
23
25
26
31
41
44
60
62
64
65
66
67
68
69
70
71
72
73
•
7 S
L Coo
2
2
3
3
4
10
11
9
11
8
13
14
14
18
18
19
24
24
8
19
3
26
25
10
13
11
15
13
17
22
24
refers
rd traces
5
7
6
12
12
12
12
13
13
30
38
17
15
20
21
20
23
40
12
21
5
27
26
13
14
15
18
17
18
23
40
co upstrea
1 C:.arc«s '.or 1
Length (E) W
2747.8
2752.6
2967.2
5172.2
3632.3
4362.3
7626.7
815S.6
2516.4
9648.1
3172.7
1813.3
2165.6
3459.2
3878.0
1710.8
1090.3
4166.6
3925.5
1534.1
6944.0
6019.2
4012.7
2642.0
5032.9
3176.9
3421.4
7209.4
3299.2
7484.4
4166.6
JE downstream
VSb Slim
idLh (m)
9052.6
10838.0
12192.0
4451.3
3584.4
4330.6
4459.5
5275.2
3285.1
18.3
3996.5
1378.9
2608.2
4765.2
4765.2
6894.3
12165.6
30.5
6978.7
7232.3
4943.2
9375.0
5320.6
3285.1
3285.1
2608.2
2608,2
1619.1
1378.9
1955.0
2072.6
teroinuB
lac Ion.
Bottom depth
1.01
1.01
0.88
0.85
0.70
0.58
0.46
0.48
0.48
0.30
0.48
0.38
0.41
0.30
0.30
0.30
0.30
0.76
0.58
0.30
1.01
0.30
0.30
0.58
0.41
0.48
0.34
0.44
0.34
0.40
0.40
Roughness
Coefficient (n)
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.06
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.06
0.014
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.5
1.1
Brtzonlk (1972)
Brt-7on(k (1972)
0,11
•(0.15 Craig and Day
(197S)
nun and Brezonik (1972) estimate for P.
-------
Table 9. Nutrient loading to swamp forest, and percentage contributions from various cultural and
natural sources for the des Allemands Basin. Values in gN and P x 10 .
LAND USE CATEGORY
Year Nutrient
1975 N Load
percentage
P Load
percentage
1995 K Load
percentage
P Load
percentage
Forest
32
2.6
1,07
0.8
30
2.0
1.0
0.7
Cleared
76
1.3
0.55
0.4
13.8
0.9
0.46
0.3
Pasture
396
33.4
2.54
1.9
334
21.8
2.15
1.4
Crop
491
40.9
110
82.8
414
27.1
92.5
71.3
Urban
38.6
3.2
6.59
5.0
124
8.1
21.1
14.0
Sewage
222
18.6
12
9.1
615
40.1
33.4
22.3
Total
1,200
100
130
100
1,530
100
151
100
-------
312
Figure Legends
Figure 1. Map of the des Allemands system showing its position in
the Barataria Basin, the natural levees or uplands, the interior
wetlands, and the enormous canal network.
Figure 2. Conceptual model relating the influence of land use
characteristics on storm water and nutrient runoff from upland
areas to the receiving wetlands.
Figure 3. Physical abstraction of the des Allemands hydrologic regime.
Uplands are portrayed as 11 subcatchments that drain into interior
wetlands. Wetlands are portrayed as either lake, stream or back-
swamp elements. Channels with x's across them were deleted in
simulations representing overland flow.
Figure 4. Mass balance analyses of agricultural crops.
Figure 5. Effects of several factors on storm water runoff from uplands.
Simulations are aggregated in three groups: (1) 1-5 and 16-20, (2)
6-10 and 21-25, (3) 11-15 and 26-30. Rainfall increased from group
1 to 3. Numbers less than 16 are based on 1975 land use, while
those greater than 15 are based on 1975 land use. In each group
the left most figure represents maximum and minimum infiltration
rates of 4.1 and 0.5 cm/hr. The next 2.0 and 0.5, then 1.0 and
0.5, then 0.5 and 0.5, and finally 4.1 and 0.13 cm/hr, respectively.
The first four represent a sequence of increasing initial soil
moisture content.
Figure 6. Projected relative changes in N and P runoff loads to the
des Allemands swamp from 1975 to 1995. See legend of Figure 5 for
explanation.
-------
313
Figure 7. Hydrographs at selected stream and backswamp areas. See
text for explanation of four different runs.
Figure 8. Comparative daily discharge rates from the Bayou
Chevreuil drainage basin to Lac des Allemands in RECEIVE RUNS
1 and 2.
Figure 9. Comparison of storm water and nutrient inputs to Lac des
Allemands following a single storm event using 1975 and 1995
conditions.
-------
Figure 1.
15
Key
EH
Wetland and
aquatic units
Natural levees and
crevasses
GULF OF MEXICO
-------
Figure 2.
UPLANDS
Drainage Density • OD
Detention Depth - Dt»t
Infiltration - I
Roughness Coefficient - RC
Slope - S
Area
WETLANDS
Figure 3. Physical Attraction of Lac D
-------
Figure 4.
316
22 N volatilized
N 135
P 45
• ———
Fertilizer
^v 76
vegetable s\—S
/10.
market
^^
other ^
cronlrsnri >
71
9 1 :
182 ,.
i35"~lsoil
^\
-------
Figure 5. RUNOFF COMPARISONS
Figure 6. PROJECTED RELATIVE CHANGES
IN FUTURE N AND P LOADINGS (1975-1995)
12345
16 17 18 19 20
6 7 8 9 10
21 22 23 24 25
SIMULATION NUMBER
11 12 13 14 15
26 27 28 29 30
240,
210
180
5 120
6?
90
60
30
Present ] 2
Future 16 17
2.8cm Rainfall
-------
:u8
Bayou Chevreuil Stage Changes
Figure 7a.
7c
30 30
.
I I
0) *>
en * "*
« : ""••-.„ »
: f*> ""••*,
y '^ '--•«!.'""••••.
" ' ""• . o
12 345678 9 W
day
30r 30
I , I
f'^ I
"jx-^N^
30
u
V
OS
.'. rc
/ "'"-^
123456 789 10
day
30
'fcC'" t
1 N-v"---...
$ '^,
Comparing
Receive Runs
.•.,
; *"•-- ,^UN2
1234567 S-P
flay
Comparing
Receive Runsl
f^ _
1,2
,3,4
RUNS
'""RUN1
12345 6^7 8 9 10
day
123456 789 10
day
Backswamp Stage Changes
75 30 / 30
12 3456789 10
day
Comparing
Receive Runs 1,2
.
. M. """"-.. RUN1
123456 789 10
day
7d
30
12 3456789 10
day
123456 ^.5^9 10
day -^ ,
Comparing
Receive Runs 1,3,4
123456789 10
day
123456789 10
day
• RUN 3
RUN 1
-------
319
Z.D
2.3
7
1
22.0
»
n
X
o
0
ID
a
jf
I
1.4
• >yvic u. \xV/mr-MnMI IVC UMI!
!
,^'
/*
/
/ ,// — < •,_
/ ^
storm
i i i
2345
-T UIS
-•*'""
""X^
1
6
^MAKUt KATES
—'" RUN4
"'^^^
'"—. ^^^
"''-^^^
^"*
i i i i
7 8 9 10
days
Figure 9.
COMPARATIVE DISCHARGE
22A
— 19.6
D
2 16.8
X
«
1 14-0
CTJ
o
x
- 11.2
D)
0
t 84
a
9,
i 5.6
"5
o
^~
2.8
f\
CHANNEL
H 20-16
nm 21-16
Illillllll 19-16 CUMULATIVE
CUMULATIVE Q —
Bi
RECEIVE RUN I
1975
, H
I
I
RUN 4
1995
COMPARATIVE LOADING
a .
I
i
\ f
N
l
M
!i
,H
s
i j
rl
1 P
Hi
ill
U nTT
mm mm
RUN I
1975
N
1
r
1
,
1
1
i i
1 — B
1
"
P
T iii
i 1 illri
• mmJSiiiiUm^t^
20.0 '2.
a
-o
X
c
15.0 •=
"5
i
O)
o
10.0|
o
"c
.*
z
5-0 -5
4.0 o
3.0
2.0
1.0
0
RUN 4
1995
-------
320
DEVELOPMENT AND IMPLEMENTATION
OF
AN URBAN-RURAL SUBCATCHMENT HYDROLOGIC MODEL (SUBHYD)
FOR
DISCRETE AND CONTINUOUS SIMULATIONS
ON
A MICRO COMPUTER
BY: Larry R. Thompson
Philips Planning and Engineering Limited
Consulting Engineers and Planners
Burlington, Ontario, Canada
and
Jon F. Sykes
Assistant Professor
Department of Civil Engineering
University of Waterloo
Waterloo, Ontario, Canada
SWMM Users Group Meeting
Montreal, Quebec
May 24-25, 1979
-------
321
INTRODUCTION
Hydroiogic scientists have sought to understand and reduce the hydrologic cycle to
a series of quantified physical laws in order to better predict the response of a watershed
to rainfall. To this end a great deal of basic research has been done to identify and
quantify various aspects and components of the hydrologic cycle in order that hydrologic
modellers might be able to simulate watershed response to rainfall.
Unfortunately, the complexity of the processes involved have so far largely eluded
exact quantification. However, reasonably sophisticated conceptual and semi-empirical
methods have been developed to mathematically describe virtually every element of the
hydrologic cycle which is of interest to the practising hydrologist.
Numerous of these techniques of modelling various components of the hydroiogic
cycle have been available for decades, but in the past it was not practical to estimate
watershed response by means of linking mathematical descriptions of each hydrologic
component. This was so because the shear magnitude of the calculations involved was far
too great for manual applications. The advent and widespread use of computers is now
revolutionizing applied hydrology by making it more practical to estimate watershed
response simply by representing and linking each of the components of the hydrologic
cycle mathematically.
The availability of computers has led to the development of a great number of
hydrologic computer models in recent years. These have ranged in complexity from those
that were simply computerizations of previous manual techniques such as the U.S.
Department of Agriculture Soil Conservation Service (S.C.S.) curve number method
(1954)10 to those such as the U.S. Environmental Protection Agency's (E.P.A.) Storm
*?
Water Management Model that attempted to simulate what were felt to be the most
important aspects of the hydrologic cycle. This proliferation of models has been further
added to by the fact that there are many special purpose models and also numerous
versions of many of the models, in use.
-------
32 2
Unfortunately as a result of this boom in model development there are no longer
standard methods in hydrology which are consistently applied to various situations. Not
infrequently two hydrologists have obtained somewhat different results when applying the
same model to the same simulation, a fact which has tended to discredit hydroiogic
modelling in the eyes of various governmental officials that do not have a thorough
understanding of the models themselves. Model builders themselves have been partly to
blame for this lack of consistency by not selecting parameters carefully enough so that
users will fairly readily agree on the value of a given parameter.
Calibration of hydroiogic models in order to better match a given observed runoff
response has been a fairly common practice. Even very sophisticated distributed models
such as that developed by Gupta and Solomon have placed reliance on calibration in
general and sometimes calibration parameters in particular in order to simulate watershed
response to rainfall. Unfortunately the process of calibration has numerous pitfalls,
particularly for models which assume constant travel time. Although many hydrologists
realize the shortcomings of calibration, users of hydroiogic modelling services are very
often not aware of the following resulting pitfalls.
Misleading ideas regarding the accuracy of a given model in a given situation are
obtained.
The general applicability of the model may be endangered. The model may become
valid only for certain geographic locations and conditions.
The model is apt to be calibrated to runoff events of fairly frequent magnitude but
then used to determine the effects of a very large storm that may occur only once
in a hundred years. If a model has had to be calibrated to the relatively small
storms for which data exists there is virtually no guarantee that it will be reliable
for extreme runoff events. This factor is of particular importance to models based
upon unit hydrographs or runoff isochrones if these do not allow for the fact that
speed of runoff and therefore peak flow increases with rainfall intensity.
-------
323
If after checking that input data values are reasonable, a model requires
calibration, it is in general due to one of the following factors:
a fundamental runoff mechanism has been incorrectly assumed to be unimportant
or has been inadequately simulated. Calibration in this case may allow the
modeller to produce more reliable estimates of peak flow but will not improve
knowledge of how runoff arrived at the outfall, a question of vital importance if
water quality modelling is ever to be attempted.
Rather than calibrate in such a case, the model builder should attempt to
identify those underlying mechanisms which were neglected and incorporate a
representation of them if a true increase in model accuracy is to result.
meteorological or flow data was inaccurate or did not account for spatial variation
over the watershed. Calibration in such a case will serve no useful purpose for it is
highly unlikely that the same spatial distribution of rainfall or combination of
measurement inaccuracies will occur consistently in the future.
the model may have certain calibration parameters intended to tailor
mathematical description of processes such as infiltration to a particular
watershed. Once calibrated such a model is only truly valid for that watershed.
This factor limits the strict applicability of such models to watersheds having a
network of stream and rainfall gauges.
A further problem in hydrology which has become more pronounced since the
advent of computer simulation models has been the artificial separation of the study into
urban hydrology and rural hydrology. This has been largely due to the fact that various
governmental agencies funding hydrologic model research and development have specific
areas of responsibility. For example, when the United States Department of Agriculture
(USDA) has supported the development of hydrologic models, the resulting models have
been primarily applicable to agricultural watersheds. Also since urban hydrologic
modellers are primarily concerned with effects of urbanization they have sometimes
expressed the viewpoint that urban simulation models need not be valid for rural
situations.
-------
324
There has thus been a tendency on the part of hydrologic model builders to develop
a rural model or an urban model, rather than just a hydrologic model. However, even to
model the effects of urbanization necessitates the prediction of watershed response when
the watershed was in a rural state before urbanization.
There has been a difference however, in the modelling techniques which have
proven successful in both urban and rural hydrology. Successful urban hydrologic
modelling has been characterized by careful conceptual simulation of impervious runoff,
the dominant urban runoff mechanism. On the other hand, due to the complexity of the
problem and multiplicity of relevant runoff mechanisms rural hydrologic modelling has
tended to place greater reliance on traditional techniques such as unit hydrographs,
isochronal methods, and hydrograph recession constants.
Differences between urban and rural hydrology and even between hydrology in
diverse geographical locations are not however, due to any true fundamental differences,
but rather to the varying relative importance of various runoff mechanisms and
components in the hydrologic cycle.
The component in hydrologic modelling which has the greatest need for improve-
ment is that which quantifies the runoff from an incremental area or subcatchment. It is
generally felt that there are numerous good routing techniques available and that in most
cases the errors introduced in flood routing calculations are minor compared to those in
9
subcatchment hydrologic modelling. A hydrologic model which consistently predicted
1321
watershed response within 20% has been considered good. '
Based upon factors brought out in the preceeding discussion it was felt that work in
hydrologic modelling would be most profitably spent in the development of a better
subcatchment hydrologic model which could become a component in any of the various
comprehensive models currently available.
-------
325
DESIGN CRITERIA FOR A SUBCATCHMENT HYDROLOGIC MODEL
After general reading and discussions with hydrologists and municipal engineers in
the academic, consulting and government communities the following list of criteria, for a
subcatchment hydrologic model were drawn up.
1. The subcatchment model should be applicable to both urban and rural areas.
2. It should be a conceptual simulation model taking advantage of modern
computer capabilities and not just a computerization of an existing manual
technique.
3. The model should be able to simulate runoff from subcatchments ranging in
size from a section of street to at least 20 sq. km.
4. The data required should be readily available from existing data sources.
Parameters should be chosen and methods for their determination developed
in such a way that various users will readily agree on the value of any given
parameter. The data sources which are readily available to the hydrologist
in Ontario are:
topographic maps
county soil maps
official plans and land use maps
aerial photographs
existing tabulations relating hydroiogic parameters to one of the
above sources
rain gauge, stream gauge, pan evaporation and temperature data
Any soil parameters which are chosen should be readily measurable in the
field should a user so desire.
5. The hydrologic cycle should be well enough represented that the model will
give good results while using reasonable values for all parameters, (i.e. if
accurate values of all variables are input, reliable results should be obtained
thout calibration.)
-------
326
6. The model should not assume that the speed of runoff is constant.
Allowance must be made for the fact that runoff speed increases with
•*-
intensity of rainfall and also for the effect of this factor on peak flows.
7. The model should be capable of continuous simulation and not restricted to
discrete storm events.
8. In order to facilitate the acceptance of the model by the hydrologic
modelling community, accepted methods of representing each of the runoff
mechanisms and components of the hydrologic cycle should be used wher-
ever possible.
9. While meeting the preceeding criteria the model must not use so much
computer time and resources that it becomes too expensive for practical
use.
IMPLICATIONS OF VARIOUS OF THE DESIGN CRITERIA
Criterion number ^ concerning the availability of data almost immediately
disqualified the use of highly rigorous finite difference schemes to evaluate subsurface
flow processes by solving the equations of flow in saturated and unsaturated porous media.
There is simply not sufficient high quality data or inexpensive large computer resources
available to the practicing hydrologist to justify the use of such techniques at this time.
The approach has been taken throughout the model development that the sophistication to
be employed on a given mechanism should be correlated to the quality of the associated
portion of the data base.
Criterion number 6 concerning the avoidance of methods which assume constant
travel time precludes the use of unit hydrograph and isochronal methods of overland flow
calculation for these methods are based upon assumptions of constant travel time.
-------
327
Criterion number 9 at first glance appears to be in direct contradiction to the
other criteria but consideration of computer centre costing formulae suggests methods by
which it may be met. At many computer centres such as that at the University of
''(
Waterloo, batch processing costs increase at a rate greater than the square of the core
memory used and a large portion of the run cost of certain programs is simply due to this
factor.
If the use of large data arrays can be avoided it should be possible to develop a
very complex subcatchment hydrologic model and still incur computer costs less than
those experienced by users of large core programs such as the U.S. E.P.A. SWMM model.
If the memory requirements can be kept down to about 30 kilobytes then a further
possibility for meeting criterion number 9 presents itself. A number of micro computers
are currently coming on the market equipped with the standard FORTRAN language and
which are well within the price range of almost any engineering company or government
agency. If the subcatchment hydrologic model could run efficiently on such a machine
then the perceived cost of applying even a very complex model would be essentially
limited to data collection.
SOURCES OF RUNOFF AND RUNOFF MECHANISMS
The study of runoff mechanisms and sources of runoff is fundamental to the
construction of a hydrologic simulation model for such a model simply consists of a series
of mathematical representations of those mechanisms which are felt to be significant.
A literature search revealed that there are numerous viable runoff mechanisms
such as:
impervious runoff
pervious runoff
runoff from channel precipitation
runoff from agricultural tile drains
groundwater flow
-------
328
It was obvious that the relative importance of these would vary from basin to basin
and from storm to storm. Thus in order to be generally applicable to almost any basin, a
subcatchment hydrologic model should not be overly biased towards any of the
mechanisms which are believed to be significant. Rather the model should allow for the
potential occurrence of all significant runoff mechanisms.
In order to reflect this vein of thought, it was felt that the subcatchment
hydrologic model should have the following features:
the subcatchment should immediately be broken into pervious, impervious and
channel areas since each of these three responds differently to rainfall
groundwater flow should not be assumed to be either insignificant or constant
tile drain flow should be included in an explicit manner
expansion and contraction of channel and wetland areas should be allowed for
the model should explicitly allow for the fact that some but not necessarily all of
roof runoff may drain to pervious area and so be subject to infiltration
the model should include the Hortonion concept of overland flow but should also be
capable of modelling the development of a perched zone of saturation within the
soil and even the fact that the water table itself may rise to the ground surface
seasonal variations in the hydrologic characteristics of a catchment should be
allowed for in a consistent manner.
The components of the hydrologic cycle which are included in the subcatchment
hydrologic model are illustrated in Figure 1.
-------
U)
K5
. J. : CO/7?POULTS OFT#£ //XD^O/.OG/C CYC
-------
330
DESCRIPTION OF SUBCATCHMENT HYDROLOGIC MODEL (SUBHYD)
Numerous concepts and methods of modelling the various components of the
hydrologic cycle illustrated in Figure 1 were considered and a description of these may be
found in reference 14. To some degree the resulting subcatchment model (SUBHYD)
represents a marriage of the concepts and methodologies embodied in two previous
2 8
sophisticated hydrologic models, one urban and one rural. (SWMM and USDAHL-74 )
SUBHYD also has numerous features not found in either SWMM or USDAHL-74 such as
tile drain flow, interception, roof drains discharging to lawns, fully separated pervious and
impervious areas, partial area concept and exponential depression storage equations.
The organization of, and flow of calculations within, the subcatchment hydroiogic
model is illustrated in Figure 2 which is a general flow chart for the model.
From Figure 2 it can be seen that within the model a subcatchment is immediately
broken down into three sub-areas, each of which has a somewhat different response to
rainfall. This subdivision is as follows:
impervious area
pervious area
channel area
Each of the areas is generally viewed as a separate entity with the following
exceptions:
A user specified portion of impervious area is allowed to drain onto pervious area,
bypassing interception. (RUNON) This was included so that the effect of roof
drains discharging onto grassed areas could be simulated. This feature also allows
the simulation of a parking lot draining to pervious area rather than to a catch
basin.
Channel area is treated as being dynamic and is allowed to expand and contract at
the expense of pervious area during a storm. This feature allows the inclusion, to
some degree, of the partial area concept which states that runoff comes from
relatively small and constant saturated areas near channels.
-------
331
FLOW CHART: SUBCATCHMENT HYDROLOGIC MODEL (SUBHYD)
FIGURE NO. 2 GENERAL FLOW CHART OF "SUBHYD"
-------
332
The model has been programmed in such a way that variable time steps may be
used for computational economy during continuous simulations. During periods of rainfall
when rapid changes occur in hydrologic processes the time steps should be fairly small in
order to simulate sudden changes but once rainfall ends, time steps may be increased in
order to achieve computational economy.
Throughout the model flow equations (infiltration, seepage, percolation, overland
flow) are solved simultaneously with applicable continuity equations with the result that
average rates of each process for each time step are determined. This means that if a
flow is calculated for a one hour time step, this is an average flow which would be
sustained for one hour. There may well have been a more instantaneous peak within that
hour which was greater, but the model will only be able to simulate this shorter duration,
higher intensity peak if smaller time steps are utilized. This averaging process has been
found to make the model very stable and no convergence problems have arisen even when
time steps somewhat greater than the travel time across a subcatchment were tried. Due
to this fact choice of time step may be at the preference of the modeller and is not
governed by stability of the model equations. For instance, even on a very small
subcatchment, if a modeller is not interested in any peak flow which is not sustained for
at least five minutes there is no need to specify a time step of less than five minutes.
The model calculates and prints out the following flow hydrographs for any
subcatchment:
runoff from channel precipitation
impervious runoff
pervious runoff
tile drain outflow
groundwater flow
These are added to give the total flow from the subcatchment and comparisons are
made to observed flow hydrographs if such are available.
-------
333
The subcatchment hydrologic model (SUBHYD) was programmed in the
FORTRAN IV computer language on an Applied Digital Data Systems System 70
(ADDS-70) micro computer.
The program as written has no practical limits other than disk storage capacity for
data, on either the number of subcatchments or time steps that may be analyzed in a
given run. A typical run for a single subcatchment requires about 2 minutes from
initiation to final printout. The program requires about 28.5 kilobytes of memory to run
unoverlayed using the ADDS ADOS version 1.* FORTRAN software. Since about 7
kilobytes of this is tied up in extensive format literal strings and a bi-level execution
trace it is conceivable that the model could run with somewhat less than 28 kilobytes of
memory without affecting the calculations.
SUBHYD has also been interfaced with the Storm Water Management Model
Transport Block in the following manner.
Subcatchment hydrographs are generated by SUBHYD and then an interface
program reformats these into SWMM TRANSPORT block input form and transmits them
from the micro computer to the Multiple Access computer system in Don Mills, Ontario.
Here they become part of a standard SWMM data set and detailed routing calculations are
performed. This forms part of a sophisticated major-minor drainage analysis methodology
in which inlet control is simulated and simultaneous routing calculations performed for
both the major and minor drainage system for any storm. (Thompson and Dennison,
197915)
VALIDATION OF SUBCATCHMENT HYDROLOGIC MODEL
Introduction
The model was tested at a subcatchment level for both urban and rural discrete
simulations. Both SWMM,2 a well known and accepted urban hydrologic model and
-------
334
HYMO a well known rural hydrologic model were also used on the same simulations for
comparative purposes.
One parameter, which is not known for validation simulations and which would be a
design parameter for design simulations, is initial soil moisture content. This parameter
was varied within reasonable limits for each simulation until the simulated volume of
runoff agreed closely with the observed volume of runoff. This procedure has been
followed by other investigators including Kouwen as described by Gorrie . By reasonable
limits the following is meant:
If the watershed has not received rainfall for several days during the summer
season, it is to be expected that all the soil free water would have drained from the
surface soil layers and some capillary stoarge would be coming available due to
plant evapotranspiration. Thus if a watershed has not received rainfall for a week
or more the initial soil moisture content should be such that all soil freewater
storage and also some capillary storage should be available to infiltrating water.
Urban Validation
It was decided to test the Subcatchment Hydrologic Model (SUBHYD) on a
catchment located in Northwest Waterloo very near to the University of Waterloo. This
catchment was chosen because of its proximity to the University and also because it had
previously been instrumented and the U.S. E.P.A. SWMM (1975) model applied to it.
The area comprises 18.2 hectares (45 acres) of predominantly single family
residential land use. It is a relatively high income area with hilly half acre lots and homes
valued at approximately $100,000 each.
The location of the Northwest Waterloo Test Catchment is shown in Figure 3.
The catchment data required for SUBHYD was abstracted from the report and
background material produced by Vespi. Vespi had broken the area down into 17
subcatchments and done detailed sewer routing using the SWMM (1975) transport block
and time steps of about 5 minutes.
-------
335
UNIVERSITY OP
WflTESlLOO
flPPROXlMflTE
CflTCHMENT BOUHOftRY
NORTHWEST WATERLOO URBAgJTEST CATCHMENT
( after reference 17 )
-------
336
In order to provide a direct test of the overall subcatchment mo^el and of the
averaging type of algorithms used it was decided to simulate the area as a single
subcatchment using SUBHYD with half hour time steps.
The maximum values of depression storage were estimated with the aid of the
SWMM (1975) user's manual and the impervious area slope was determined as the weighted
average sewer slope.
Soil parameters were determined with the aid of the Waterloo county soils report.
It was decided to test SUBHYD on a storm which occurred on November 10, 1977
since this was a well defined storm of 1.96 cm (0.77 in.) of rainfall.
20
For comparative purposes both the SWMM and the standard HYMO model were
also used to simulate the catchment response to the November 10 storm. All models
treated the area as a single subcatchment and the U.S. Soil Conservation Service (S.C.S.)
curve number in HYMO was adjusted so that calculated and observed volumes of runoff
agreed.
A tabled comparison of the various model versus observed results may be found in
Table 1 while graphical comparisons of the SUBHYD, SWMM and HYMO predicted
hydrographs versus the observed hydrograph and each other may be found in Figures '4 to
6.
The SUBHYD simulation of this storm had the following statistics relative to the
observed flows:
error in peak flow = 2.32%
error in volume of runoff = 1.88%
correlation coefficient = 0.918
The previously mentioned detailed calibrated SWMM simulation of this runoff event
performed by Vespi (1977) resulted in an error of about 15% in volume of runoff and about
12% in peak flow.
-------
337
TADLE It HYDROLOCIC MODEL COMPARISON
NORTHWEST WATERLOO TEST CATCHMENT - STORM OF NOVEMBER SO, 1977
1.96 CM RAINFALL (0.77 IN)
MODEL
HYMO
(DT . 0.5 HR)
SWMM
(DT - 0.5 HR)
SU8HYD
(DT > 0.5 HR)
.MEAS. CALC. ERROR
PEAK FLOW PEAK FLOW %
(CMS) (CMS)
0.03«5 0.037S 9.08
0.03*5 0.0370 7.25
0.0345 0.0337 -2.32
MEAS. CALC. ERROR
VOL. VOL. %
(1000 M**3> (1000 M"»3)
0.357 * 0.357 - 0
0.357 0.43 20.4
0.357 OJ63 1.S8
MEAS. CALC. ERROR
TIME TO TIME TO (HR)
PEAK (HR) PEAK (HR)
^0.5 «1.0 0.3
- 0.5 0.5 a 0.5
.0.5 <0.5 0
NOTE! 1) I CMS . 35.3 CFS
2) 1000 M"3 = 0.81 ACRE FT.
3) DRAINAGE AREA » 0.1S2 SQ. KM
= 0.0695 SQ. ML
« »5 ACRES
FtCVRE * : NOVEMEEX 10, 19" SI05M - SV5HTT3 VESSL'S 03SESTCD FLCWS
-------
338
NOV. 10.1977 STBRN
SWUM CSLCUmTEO FLOW *
FU8HS *
16.00 20.00 2H.OO
TIME IN HOURS/2
FIOTKE 5 : NOVEMBER 10, 1977 ST08H - SWffl VERSUS OSSEXVED FLOWS
NOV. 10.1977 3TC=M
SUSKTB TOTFL FLCa X
HTKO TOTSL FIC« »
SNMN TOTRL FLW O
«OSIBED TOTBL FISH X
12.00 JS.OO 20.00 S<4.00
TIME IN HOURS /2
-------
339
Rural Validation
The West Canagagigue Creek watershed above Floradale, Ontario was chosen as a
test rural subcatchment for a number of reasons:
1. its close promitity to the University of Waterloo
2. it has been gauged by the Water Survey of Canada
3. it has been considered to be a very typical agricultural watershed3
*. it was previously chosen as a test catchment by the Pollution from Land Use
Activities Reference Group (PLUARG) so that a great deal of data was
available for it.
Some of the data collected for this basin by PLUARG researchers included a farm
19 19
by farm tile drain survey, cross sections, rain gauge data, and detailed land use
information.
The West Canagagigue catchment of about 19.46 sq. km. (4,800 acres or 7.5 sq. mi.)
is referred to as AG-4 in the various PLUARG reports and represents the watershed of
Canagagigue Creek, a tributary of the Grand River which flows into Lake Erie. The
catchment is located in Peel Township, Wellington County, about 22 km north of
Waterloo. The topography of the area is gently undulating and the main soil type is loam
with small areas of muck and river bottom flat lands.
The main agricultural activity is mixed farming and predominantly dariy with
lesser numbers of beef and swine. Crops grown include corn for silage, small grains and
hay. A large number of the farmers are traditional old order Mennonites, however their
farming practices are very modern and they utilize extensive tile drain networks.
The extent of the West Canagagigue test subcatchment is illustrated in Figure 7.
Due to the large amount of documented data available for the catchment it was
possible to abstract values for most of the various parameters directly from published
sources.
Hourly streamflow values for storms of interest were/1 obtained from the Water
i)
Survey of Canada from records of their station number 02GA03J6.
-------
340
:-\,. >:•- '.
CANAGAGIGUE
CATCHMENT, ONTARIO
-------
341
Rain gauge data in hourly increments was obtained from the PLUARG rain gauge
maintained by Dr. Anderson of the University of Windsor.
Landuse information was abstracted from the PLUARG landuse report on the
catchment prepared by Frank and Ripley, 1977 which was based upon data collected in
1975-76.
Soil data including depths and textural classifications of each soil layer were
determined from the Wellington County Soil Survey.
Tile drain data was determined with the aid of a detailed farm by farm tile drain
survey carried out by Whiteley et al, 1976 for PLUARG. Figure 8 is a composite of the
information gathered for each farm and serves to illustrate the extent and complexity of
the tile drain network in existance as of 1976. Forty percent of the catchment was
systematically tiled and Whiteley et ai (1976) found no strong correlation between tiling
and soil type.
Pan evaporation data was obtained from the University of Guelph for their
evaporation pan at nearby Elora. Temperature data was also obtained from this source
and used as input to the LANUSE routine of USDAHL-75 (a metric version of USDAHL-
74). This routine was used to determine growth indices for each ianduse and week
throughout the period of interest and thereby allow for seasonal variations in vegetation.
The remaining data concerning drainage area, flow lengths and slopes were
estimated from the 1:50,000 scale topographic map of the catchment.
The initial soil moisture content was varied within reasonable limits as described
previously in order to match observed peak flow and volume of runoff. This was done
because this is one parameter which was not known for validation purposes and which
would be a design parameter for design purposes.-
The Subcatchment Hydrologic Model (SUBHYD) was tested on three storm runoff
events which occurred on the West Canagagigue catchment during the period extending
from the beginning of May to the end of June, 1976. These storms differed widely in
rainfall amount, intensity, duration and antecedent moisture conditions.
-------
jlt 1 .^U,--'K/1 <
-] — - / -,.!, I IT1 >N '---I —-r
'
TILE DRAIN SURVEY
OF UPPER
WEST CANAGAGIGUE CREEK
COMPILED BY
UNIVERSITY OF CUELPH 1975
C I TICEO AHEA
FIGURE S
-------
343
All^SUBHYD parameters were held constant for these storms with the following
exceptions:
meterological input (rainfall, pan evaporation)
growth index values (from growth index curves determined by LANUSE
routine of USDAHL-75)
initial soil moisture content (varied within reasonable limits)
Manning's "n" for the channel (this was allowed to increase from 0.025 in the
spring to 0.050 in mid-summer when the channel is somewhat overgrown).
The HYMO and the SWMM hydrologic models were also applied to this catchment
for each storm for comparative purposes. The S.C.S. curve number in HYMO was varied
to match observed volume of runoff since a similar operation was done for SUBHYD by
varying the initial mositure content. The SWMM infiltration parameters were not varied
since this model has not normally responded significantly to changes in these
12
parameters.
Simulation of May 5-7, 1976 Storm
This storm was essentially a drizzel that continued for more than one day.
Although 2.9 cm. (1.1* in.) of rain fell, the maximum one hour intensity was only 0.2
cm/hr. (0.079 in/hr).
SUBHYD simulated the observed runoff hydrograph with the following statistics:
error in peak flow = 2.3%
error in volume of runoff = 7.0%
correlation coefficient = 0.905
A tabled comparison of the various model versus observed results may be found in
Table 2 while graphical comparisons of the SUBHYD, SWMM and HYMO predicted
hydrographs versus the observed hydrograph and each other may be found in
Figures 9 to 12.
-------
344
TABLE a HYDROLOC1C MODEL COMPARISON
VEST CANAGAC1GUE CREEK - STORM OF MAY 5-7, I57«
2.9 CM (U« IN) RAINFALL
MODEL
HYMO
(DT a OJ HR)
SIMM
(DT » 0.3 HR)
SU8HYD
(DT • 1.0 HR)
MEAS. CALC. ERROR
PEAK FLOW PEAK FLOW %
(CMS) (CMS)
0.2! 1.06 1JJ
tat 0.273 -93.6
».2S *.IS -2-3»
MEAS. CALC. ERROR
VOL. VOL. *
(1000 M"J) (1000 M"3)
283 = 2S3 -0
2S3 14.0 -93.1
ZS3 303 7.07
MEAS. CALC. ERROR
TIME TO TIME TO (HR)
PEAK (HR) PEAK (HR)
3 ' 1
332
3 « 3
NOTE.- 1) 1 CMS = 35.3 CFS
2) 1000 M*'3> 0.8! ACRE FT.
3) DRAINAGE AREA i 19.46 SQ. KM
* 7.31 SQ. ML
« »809 ACRES
1.00-
3.20 -
' 2.HO -
0.80 -
(2.9 cm)
5-7.1376 STBflH
SUSHTD BUBSL FLOWS— *
FLCMS *
—I 1 1 1—
?4.00 32.00 10.00 U8.00
TIME IN HOURS
56. CO
1
614.00
72.00
FICL"R£ 9 • VAY 5-7 STOS.M - StBHYD VERSUS OBSERVED ,LOU'S
-------
345
5-7,1976 STCTH
SUHtl CR.CUU5TEB FLOW *
FUCUS i.
0.0
32.00 UO.OO U8.00
TIHE IN HOURS
FIGURE 10 -. MAIT 3-7 ST08H - SWMM SESSUS OBSERVED FLOHS
S.OO-
' 3.00-
O
_J
U_
2.00-
i.ao -
KBT S-7.I9VS ST3W
KTHO CPLCULHTEC FLOW
f LCWS
-\ T
sa.oo 10,00 na.oo
TIME IN HOURS
ss.oo
FIGURE 11 : ".XI 5-7 STC?.« - iiV"0 VtS5'J3 OCSEKVEB :'t.'-o
-------
346
IWT S-7,1976 STCWt
5U8HTO TOTflL FLCH *
WHO TOTBL FLOW »
SWW TOTRL FLCa O
TOTS. FLOH X
5.00-
CD
2:
cj
2.CD-
32.00 10.00
TIME IN HOURS
FIGURE 12 : K*J 5-7 STORM - COMPARISON OF SUBHID, SWW, HMO, AND OBSERVED FLOWS
' 2.10 -
0.80 -
HHT 5-7. 1976 STSB1
TOTBL CSLC. FL-J ¥
GfOUNOHBTEB FLCy A
TILE OHHIN FLOW O
PEHVIOU5 R'JNCfF FL5M 3
IMPERVIOUS FLOW +•
OflSNEL FLCH X
iio.oa
TIME IN HOURS
^sssssaaea—|—
ta.oo ss.co
ncl'RE 13 ><_-.Y ;-; STOK.M - SUHYD brj^TOO~^ OF FLCW
-------
347
The breakdown of the sources of runoff for this runoff event, as determined by
SUBHYD may be found in Figure 13. The SUBHYD results indicate that this storm event
was dominated by groundwater flow with contributions from tile drains, impervious area,
channel precipitation and pervious area in decreasing order of importance.
Simulation of 3une 13, 1976 Storm
This storm occurred after almost a month without rain, and the catchment was so
dry that the creek had been dried up for over a week. This was a very short intense storm
of 3.1 cm (1.22 in.) with a maximum one hour intesity of 1.8 cm/hr. (0.71 in/hr).
SUBHYD simulated this storm as shown in Figure 14 with the following statistics:
error in peak flow = 0.47%
error in volume or runoff = 3.9%
correlation coefficient = 0.873
Tabled and graphical comparisons of the various models to each other and observed
flows may be found in Table 3 and Figures 14 to 17.
The sources of runoff during this storm, as determined by SUBHYD are illustrated
in Figure 18. This runoff event was dominated by tile drain flow with additional
contributions from impervious area, groundwater, channel precipitation and pervious area
in decreasing order of significance. Tile drain flow was so important during this storm
because the high rainfall intensities involved were able to build up a significant head on
the tiles thus causing them to flow at greater rates. The very low intensities experienced
during the earlier May 5-7 storm did not build up a significant head on the tiles with the
result that most of the water during the May storrn was able to percolate through to
groundwater.
Simulation of June 30, 1976 Storm
The characteristics of this storm lay somewhat between those of the two
previously discussed storms, with the exception of total rainfall.
-------
348
TABLE » HYDROLOG1C MODEL COMPARISON
WEST CANAGAGIGUE CREEK - STORM OF JUNE 13, 1976
XI CM (1.22 IN) RAINFALL
MODEL
HYMO
(DT • 0.1 HR)
SVMM
(DT » 0.3 HR)
SUBHYD
(DT > 1.0 HR)
MEAS. CALC. ERROR
PEAK FLOW PEAK FLOW %
(CMS) (CMS)
3.71 1.S7 -49.6
J.71 2M -17*
3.71 3.73 0.47
MEAS. CALC. ERROR
VOL. VOL. *
(1000 M"3) (1000 M"3)
500 a 300 * 0
500 15.1 -SS.98
500 519 3.91
MEAS. CALC. ERRCf
TIME TO TIME TO (HR)
PEAK (HR) PEAK (HR)
2 » 2
220
220
NOTE! 1) 1 CMS » 35.3 CFS
2) 1000 M"3 ^ 0.81 ACRE FT.
3) DRAINAGE AREA * 19.1$ SQ. KM
• 7.51 SQ. ML
» «S09 ACRES
14.80-i >
3.33-
1 3.10 -
0.80 -
0.0
0.0
RAINFALL
(3.1 on)
JUNE I3.1S7S ST;SH
SUBlflO RUfiSL FuSW-
«BSUR£0 FLOWS
12.00
16.00 20.00
TIME IN HOURS
. „"„:;;; S3 S7c:-;vt SCBKYP VER?LS OBSERVED rtovis
-------
349
It. SO-i
n.oo -
3.20-
'2.10-
JUNE 13.1978 STORM
SM1H CR.CULHTEO FLSH *
FU5MS A
0.80-
S.OO
15.00 20.00 25.00 30.00
TIME IN HOURS
FIGURE IS : JUKE 13 SIOR.1 - SWW VERSUS OBSERVED FLOWS
F1CCRE IS : --JHE 1J STOSX - HYKO VERSUS OESUVEO FlUKS
-------
350
JUNE 13.1375 STORM
SUBHYD TOTflL FLOU *
MTHO TOTftL FUCH •»
SWW TDTflL FLOW O
HEflSURH) TOTBL FLOW X
in
2:
CJ
1S.OO -
8.00
16.00 20.00
TIME IN HOURS
28.00
36.00
FICUBE 17 : JW£ 13 STORM - COMPARISON OF SUBHYD, SWKH, HYKO, AND OBSERVED FLOWS
in
s:
1 2.UO-
O.BO -
H.OO
12.00
IS.00 20.00
TIME IN HOURS
JUNE 13.1976 STCRM
TOTBL CBLC. FLOW »
OROUNDWflTER FLOW *
TILE OfWIN FLOW O
PERVIOUS RUNOFF FLCH O
IMPEBVIGUS FLOW +
CMBNNEL FLOW X
FIGURE 18 . JUNE 13 STCRM - SL'SHYD CREAKPOWX OK KtCVJ SOURCES
-------
351
5.2 cm (2.05 in.) of rainfall fell at a maximum one hour intensity of 1.4 cm/hr.
(0.55 in/hr.). The catchment was dry but not as dry as it had been before the June 13
storm since the creek had just barely dried up.
The statistics of the comparison of the SUBHYD simulation and the observed
hydrograph were as follows:
error in peak flow = 0.70%
error in volume or runoff = 0.59%
correlation coefficient = 0.872
Comparisons of the various models to each other and observed flows may be lound
in Table <4- and Figures 19 to 22.
Figure 23 illustrates the sources of runoff during this storm as determined by
SUBHYD. This runoff event was dominated by tile drain and groundwater flow. The tile
drains contributed more to the peak flow but groundwater flow produced a greater volume
of runoff. Smaller amounts of flow were produced by the mechanisms of impervious
runoff, channel precipitation and pervious runoff in decreasing order of importance.
This storm had high intensities of rainfall although not as high as were experienced
on June 13, 1976, and this caused a significant head to be built up on the tiles creating
substantial tile drain flow.
Effect of Tile Drainage on Peak Runoff
In recent years farmers in the Canagagigue and other watersheds have been
installing extensive tile drainage systems. These have been installed so that fields will
dry out more rapidly in the spring thereby allowing farmers to begin working their fields
19
about two weeks earlier than previously.
By intent these tile drainage systems must contribute to spring runoff and at the
outset of this work it was believed possible that they might contribute significantly to
storm runoff as well.
-------
352
TABLE 4.- HYDROLOC1C MODEL COMPARISON
WEST CANAGACKUE CREEK - STORM OF JUNE 30, 197*
U CM (2.0) IN) RAINFALL
MODEL
HYMO
(DT » 0 J HR)
SWMM
(DT » 0.5 HR)
SUBHYD
(DT • 1.0 HR)
MEAS. CALC. ERROR
PEAK FLOW PEAK FLOW *
(CMS) (CMS)
4.33 6.44 42.2
4.53 2.08 -5».l
4.33 4.30 -0.70
MEAS. CALC. ERROR
VOL. VOL. «
(1000 M"3) (1000 M-«3>
174 * 17» » 0
174 24.70 -S3.S
174 173 -0.39
MEAS. CALC. ERROf
TIME TO TIME TO (HR)
PEAK (HR) PEAK (HR)
2 3 1
2 IS 16
2 3 1
NOTES 1) 1 CMS = 33.3 CFS
2) 1000 M*-3 = O.S1 ACRE FT.
3) DRAINAGE AREA » 19.46 5Q. KM
7.31 SO. ML
4S09 ACRES
JUNE 30,1376 5TCFH
SUSfTO flUSHL FLCHS— X
BEfiSUBED
32.00 W.OO W.SO
TIME IN HOURS
FIGURE 19 : JUKE 30 STO:.-; - SCSUSO VEKCS OSS^SVED FU-i
-------
353
JUNE 30.1976 3TOBM
SWWI CTLCUWTED FUJN M
r&sfss FLOWS *
32.00 '40.CQ
TIME IN HOURS
FIGURE 20 : JUNE 30 5TORH - SWIM VERSUS OBSERVED FLOWS
JUNE 33.197S STWW
H1THO Ca.CULflTffl FLCH M
«flSUR£D fLCHS *
in
a:
o
2.00 -
21I.CO
32.00 "40.00 US.00
TIME IN HOURS
-------
354
12.00 -i
JUNE 30.1976 STCBM
SUBMTO TOTFL Fiat X
HTHO TOTfiL FLCH *
SWW TOTBL FLOW O
TOTHL FLOW X
ZU.OO 32.00 UO.OO 18.00
TIME IN HOURS
FIGURE 22 : JOSE 30 STOTM - CCKPASISOH OF SUBHYD, SW«M, HYXO, ASD OBSERVED FLOWS
JUNE 30.137B 5T;HM
TBTBL CBLC. FLC'J *
GROUNOUOTER FLIM "
TILE ORBIN FLCM O
PERVIOUS KJK'ff FLOH O
IMPERV1CU5 FLCK +
CHRHNEL FLC'J X
32.00 40.00
IKg IN
FIGURE 23 Jl'iE 30 STORM SV5HYD SRF.\KDCT':J OF FUW Sl'l'RC^S
-------
355
A farm by farm tile drain survey conducted as part of a PLUARG study in the West
Canagagigue Creek catchment19 determined that 40% of the catchment was
systematically tile drained and that no strong correlation between soil type and tile
drainage existed. Figure 8, introduced earlier is a composite of this survey.
Examination of this figure reveals that these agricultural tile drainage systems,
including laterals, trunks and surface catch basins, are comparable in extent and
complexity to urban storm drainage networks.
Conventional hydrologic models, including complex distributed ones such as those
developed by Gupta (1974)5 and Constable (1974)1 have not explicitly considered the
potential effects of tile drainage on runoff. Flood plain mapping studies in Ontario have
also tended to neglect tile drains simply because there has been a lack of hydrologic
models which include a representation of this mechanism. Hydrology textbooks also
usually fail to mention or describe this runoff mechanism. The development of surface
drains and tile systems incorporating catch basins, also represent an increase in drainage
density, an underlying hydrologic parameter which has been shown to be very important.
Drainage basin development affects the fundamental hydrologic parameters of
imperviousness and drainage density but conventional rural hydrologic models such as
HYMO only allow for changes in imperviousness.
When possible impacts of tile drainage are discussed with farmers and agricultural
engineers, they typically express the opinion that tiles do not increase runoff and may
even decrease it and also that tiles do not flow during the summer.
However concern of individuals and the Ontario Ministry of the Environment
regarding erosion at tile drain outfalls has reached the point where special outfalls
designed to prevent erosion are being proposed. In this regard it would appear that
thinking regarding development of rural drainage systems is parallel to but somewhat
lagging that concerning urban drainage systems. At first urban storm sewers were
developed with little concern for consequences downstream of the outfall. After
-------
356
some time erosion problems became evident and sewer outfalls were then designed to
minimize erosion. Recently concerns regarding increased urban storm water flow and
water quality have reached the point where attempts are now made to prevent increased
runoff and adverse effects upon water quality.
Up to this time the potential impact of agricultural drainage systems regarding
increased runoff has not been assessed. Due to this fact in Ontario governmental agencies
have on the one had justifiably opposed the construction of urban drainage systems which
could contribute to increased runoff; but on the other, encouraged and supported the
development of more efficient agricultural drainage systems.
The Subcatchment Hydrologie Model was developed in such a way that a separate
representation of each runoff mechanism considered significant was included. Tile drains
were modelled by re-arranging a tile drain design methodology developed by Jan Van
Schilfgaarde. Van Schilfgaarde's method represents a slight improvement upon that of
Glover which has been used by the U.S. Bureau of Reclamation and has been shown to
23 24
produce reliable results. ' Therefore it was decided to use SUBHYD to estimate what
the catchment response to each of the three storms would have been if no tile drains had
been present.
This was done by setting the percent tile drained area to zero but retaining the
same initial moisture contents which had been previously determined to give the best
simulation.
Figures 24 to 26 illustrate the SUBHYD simulations of these storm runoff events,
both with and without tile drains.
The predicted reduction in peak flows from each storm were as follows:
May 5-7, 1976 : 18%
June 13, 1976 : S0%
June 30, 1976 : 63%
-------
357
1,60-1
11.00 -
3.20-
IBT S-7.1976 STOW
TILE ORR1NS UQ.LCED *
TILE OflfllNS KrC'l&J *
o
2
o.eo-
32.00 W.OO M8.00
TIHE IN HOURS
FIGURE 24 : MAY 5-7 STORK - EFFECT OF TILE DRAINS
1 2.1)0 -
l.SO -
0.80 -
o.o
JUNE 13,197G STJSM
TILE ORSItC I^C^UCSJ
TILE OPBI^S REHOVB3
12.00
16.03 20.00
TIME IN HOURS
Oi* - EFFECT OF THE
-------
358
M.eo -i
M.OO -
JUNE 30,1976 STORM
TILE ORB]US INCLUDED K
TILE DfWINS REMOVED *
3.20 -
cn
zr
o
2.10 -
1.60 -
o.eo-
o.o
o.o
8.00
16.00
24.00
32.00 110.00 48.00
TIME IN HOURS
FIGURE 26 : JUKE 30 STORM - EFFECT OF TILE DRAINS
-------
359
When the tiles were removed more water was able to percolate down to
groundwater thereby increasing groundwater flow somewhat but still reducing total peak
flows. The effect of removing tile drains was very similar to the damping effect that
could be introduced by effective flood control storage in the system. A sharply peaking
hydrograph resulting from a sudden, intense storm was very sensitive to the removal of
tile drains, whereas a gradual damped hydrograph was not so sensitive. The physical
explanation of this phenomenum is as follows:
An intense storm produces higher rates of infiltration by approaching or exceeding
infiltration capacity, and consequently creates greater hydraulic heads on the tile drains.
This greater head then causes tiles to flow at greater rates than for a less intense storm.
This is borne out by the SUBHYD final results and becomes especially evident when the
respective heads on the tile drains are monitored in SUBHYD.
SUMMARY
A subcatchment hydrologic model (SUBHYD) has been developed and implemented
on a micro computer. This model is a collection of mathematical representations of
various components in the hydrologic cycle and as such should be updated as better
techniques of modelling these components become available.
The model has been tested on an urban subcatchment of 18.2 hectares (45 acres)
and a rural subcatchment of 19.46 sq. km. (4,800 acres).
By varying only initial soil moisture content, the model was able to simulate
observed runoff within an acceptable degree of accuracy, from four storms ranging in size
from 1.96 cm (0.77 in.) to 5.2 cm (2.05 in.). The model's performance for these storms
compared favourably with that of two other hydrologic models.
The model indicates that under intense storm conditions the presence of substantial
agricultural tile drains significantly affects runoff hydrographs, tending to increase peak
flows.
-------
360
REFERENCES
1. Constable, T.W., "A Distributed Quantity-Quality Runoff Model for Assessing
Potential Impacts of Alternative Land Use Configurations", M.A.Sc. Thesis,
University of Waterloo, Waterloo, Canada, December, 1974.
2. Environmental Protection Agency, "Storm Water Management Model, Vol. I-IV",
U.S. Government Printing Office, July, 1971.
3. Frank, R., and Ripley, B.D., "Land Use Activities in Eleven Agricultural Watersheds
in Southern Ontario, Canada, 1975-76", Ontario Ministry of Agriculture
and Food, March, 1977.
4. Gorrie, J.E., "Radar Precipitation Measurements in Hydrologic Modelling for
Reservoir Operations", M.A.Sc. Thesis, University of Waterloo, Waterloo,
Canada, April, 1977.
5. Gupta, Santosh, K., "A Distributed Digital Model for Estimation of Flows and
Sediment Load from Large Ungaged Watersheds", Ph.D. Thesis, University
of Waterloo, Waterloo, Canada, March, 1974.
6. Harley, Brendan, M., "Runoff Model Performance and Interpretation", presented
at the Second Urban Water Resources Workshop, University of Toronto,
Toronto, Canada, March, 1979.
7. Hoffman, D.W., et al, "Soil Survey of Wellington County, Ontario", Report No. 35
of the Ontario Soil Survey, 1963.
8. Holtan, H.N., et al, "USDAHL-74 Revised Model of Watershed Hydrology", Technical
Bulletin No. 1518, Agricultural Research Service, United States Department
of Agriculture, December, 1975.
9. Kouwen, N., personal communication, 1977.
11. Presant, E.W., and Wickland, R.E., "The Soils of Waterloo County", Report No. 44
of the Ontario Soil Survey, 1971.
12. Proctor and Redfern Limited, and 3ames F. MacLaren Limited, "Storm Water
Management Model Study - Volume 1", March, 1976.
13. Terstriep, M.L., and Stall, J.B., "The Illinois Urban Drainage Area Simulator,
ILLUDAS", Bulletin 58, State of Illinois Departmentof Registration and
Education, 1974.
14. Thompson, Larry R., "A Comprehensive Subcatchment Hydrologic Simulation
Model For Urban and Rural Applications", Master of Applied Science Thesis
(not finalized and published as of May, 1979), Department of Civil Engineering,
University of Waterloo, Waterloo, Canada.
15. Thompson, Larry R., and Dennison, Karen D., unpublished work, Philips Planning
and Engineering, Burlington, Canada, 1979.
-------
361
16. United States Department of Agriculture, Soil Conservation Service, "National
Engineering Handbook, Section ^ - Hydrology", 1969.
17. Vespi, F., "Storm Water Runoff Study of the Northwest Waterloo Test Catchment
- Final Report", fourth year undergraduate project, University of Waterloo,
Waterloo, Canada, December, 1977.
18. Whiteley, H.R., and Ghate, S.R., "Hydrologie Model Project Final Report", Pluarg
Task C Phase II, Project 1.15, March, 1978.
19. Whiteley, H.R., unpublished work, 1977.
20. Williams, 3.R., and Hann, R.W. 3r., "HYMO; Problem Oriented Computer Language
for Hydrologie Modelling - Users Manual", Agricultural Research Service,
U.S. Department of Agriculture, May, 1973.
21. Wisner, Paul E., "Requirements for Standardization in Urban Runoff Modelling",
presented at the Second Urban Water Resources Workshop, University
of Toronto, Toronto, Canada, March, 1979.
22. Van Schilf gaarde, 3an, "Design of Tile Drainage For Falling Water Tables", Journal
of the Irrigation and Drainage Division, A.S.C.E., June, 1963.
23. Dumm, Lee D., discussion on paper by Van Schilfgaarde
24. Dumm, Lee D., "Transient-Flow Concept in Subsurface Drainage: Its Validity
and Use", Paper No. 60-717 at the Winter Meeting of the American Society
of Agricultural Engineers at Memphis, fenn., December, 1960.
-------
362
RAINFALL-RUNOFF MODELING OF FLOW AND TOTAL NITROGEN FROM
TWO LOCALITIES IN THE DENVER, COLORADO, METROPOLITAN AREA
By
William M. Alley and Sherman R. Ellis
U.S. Geological Survey
Presented at the
Storm Water Management Model Users
Group Meeting
Annapolis, Maryland
November 13-14, 1978
-------
363
RAINFALL-RUNOFF MODELING OF FLOW AND TOTAL NITROGEN FROM
TWO LOCALITIES IN THE DENVER, COLORADO, METROPOLITAN AREA
By William M. Alley1 and Sherman R. Ellis2
ABSTRACT
Data from two residential basins in the Denver, Colorado,
metropolitan area were used for calibration and verification of
the U.S. Environtaental Protection Agency's Storm Water Management
Model II for rainfall-runoff modeling of flow and total nitrogen.
Using values for the accumulation and washoff parameters as suggested
in the user's manual and basin-characteristic parameters as determined
from the flow calibration, the errors in simulated storm-runoff
loads of total nitrogen ranged from 67 to 1,000 percent. The model
assumption that Land-surface loads are directly proportional to the
number of days prior to the storm during which the a-.cumulated rainfall
was less than 1.0 inch was not substantiated. Results indicate that the
model simulations of total nitrogen are very sensitive to the poorly
defined model parameters specifying the part of the suspended-solids
-''Hydrologist, U.S. Geological. Survey, Reston, Virginia
2Hydrologist, U.S. Geological Survey, Lakewood, Colorado
-------
364
and settleable-solids concentrations added to the total nitrogen
concentration. Based on the results of this study, caution should
be used when applying the Storm Water Management Model II or other
similar models to assist in making water-quality-management decisions,
particularly in the absence of local data for model calibration
and verification.
INTRODUCTION
The effects of urban runoff on water quality are of concern
in most metropolitan areas in the United States. This concern has
been largely a result of the areawide planning requirements of.
Section 208 of Public Law 92-500, 1972 Amendments to the Federal
Water Pollution Control Act.
Concomitant with this rising concern about urban-runoff quality
has been the development of numerous computer models for simulating
the storm-runoff and water-quality processes. Unfortunately, devel-
opment of these models is based on a very small data set. It has
been suggested that development of these models has " * * * seemingly
already greatly outpaced the data base for model verification * * * "
(McPherson, 1975). Additional data appear to be required before
substantial improvements in storm-runoff-quality modeling in urban
areas can be achieved (Whipple, 1974).
-------
365
The purpose of this paper is to present the results of
calibration and verification of the U.S. Environmental Protection
Agency's Storm Water Management Model II (Huber and others, 1975)
for simulating total nitrogen concentrations and loads in rainfall
runoff. Modeling of urban-runoff quality requires conjunctive
consideration of the quantity and quality aspects of urban runoff
(Alley, 1976). Therefore, the results of rainfall-runoff modeling
of flow are also presented. The data used in this study were
collected by the U.S. Geological Survey during 1976 and 1977 as
part of separate but coordinated cooperative agreements with several
local governmental entities—the Denver Board of Water Commissioners,
the Denver Regional Council of Governments, and the Urban Drainage
and Flood Control District. Data collected during the investigation
have been published by Ellis (1978).
-------
366
DESCRIPTIONS OF DRAINAGE BASINS
Data from two drainage basins were utilized in this modeling
study (fig. 1). One drainage basin is located in Littleton, Colo.
(the Littleton site) and has an area of 606 acres. Approximately
75 percent of the basin contains single-family dwellings and
25 percent is parks and open space. About 15 percent of the basin
is covered by effective impervious surfaces1 and 10 percent by
noneffective impervious surfaces2. Street gutters and open ditches
are the predominant drainage types for flow conveyance.
A second drainage basin is located in Lakewood, Colo, (the
Lakewood site) and has an area of 76.7 acres. Approximately
50 percent of the basin is undeveloped land, 30 percent is multifamily
dwellings, and 20 percent is commercial development. About 30 percent
of the basin is covered by effective impervious surfaces and
10 percent by noneffective impervious surfaces. Street gutters
convey storm runoff to a storm sewer crossing the basin.
•'Effective impervious areas are impervious areas which are
connected and, in turn, connect to some means of conveying the
runoff out of the areas, such as roofs which drain onto driveways,
streets, sidewalks, and paved parking lots.
2Noneffeetive impervious areas are impervious areas which
are not connected to other impervious areas and which drain to
pervious areas, such as roofs which drain onto lawns.
-------
367
EXPLANATION
* MONITORING SITE AND NAME
LAKEWOOD
SITE
"pENVER
f 7 . ~
T.5S.
0
1
1
0
1
5
i
I
10
10 l!
I 1
15 KILOMETERS
15 MILES
Figure 1.—Locations of urban-runoff monitoring sites.
-------
368
INSTRUMENTATION AND DATA COLLECTION
The data utilized in this study were collected using automatic
water-quality-sampling and rainfall-runoff monitors. The automatic
monitor and its operation are described by Smoot, Davidian, and
Billings (197*0. Flows were determined and samples were collected
at the outlets of the drainage basins using the automated equipment.
The gage height-discharge relationship for the Lakewood site was
derived from theoretical ratings using a V-notch weir. The gage
height-discharge relationship for the Littleton site was computed
using the step-backwater approach (Shearman, 1976). Samples for
water-quality analysis were collected at intervals ranging from 2
to 5 minutes depending upon basin size and experience gained during
the sampling program. These intervals were occasionally changed
during runoff periods. All samples were analyzed at the U.S.
Geological Survey regional laboratories. Three tipping-bucket rain
gages were installed in the Littleton basin, and one tipping-bucket
rain gage was installed in the Lakewood basin.
MODEL APPLICATION AND RESULTS
The U.S. Environmental Protection Agency's Storm Water Management
Model II (Huber and others, 1975) is a computer model for the
simulation of storm and combined sewerage. Application of the
model requires that the basin be divided into "homogeneous" subbasins.
-------
369
Overland flow from these subbasins is routed to the basin outlet
through a gutter-and-pipe network. Subbasins and the gutter-and-
pipe network are delineated for the Littleton and Lakewood basins
in figures 2 and 3.
Flow
During this study, six periods of rainfall runoff were gaged
at each site for a sufficient duration and at a sufficient frequency
so that storm-runoff volumes and peak flows could be determined
accurately. Three of these periods of rainfall runoff at each site
were selected for model calibration and three at each site for
model verification. Runoff periods selected for flow calibration
were those to be used for total nitrogen calibration.
Prior to calibrating the model, a literature review was used
to obtain an improved understanding of model-output sensitivity to
changes in the input parameters. Three studies have included
sensitivity analysis of the model (Graham and others, 197^; Huber
and others* 1975; and Proctor and Redfern Limited and James F.
MacLaren Limited, 1976). During each of these studies, all model
parameters except one were held constant. The model parameter
being tested was then varied through a range of values greater and
lesser than the accepted mean value. All studies concluded that
model output was most sensitive to the amount of effective impervious
-------
370
EXPLANATION
BASIN BOUNDARY
SUBBASIN BOUNDARY AND
NUMBER
GUTTER OR PIPE AND NUMBER
YV- :
Figure 2.--Features related to use of U.S. Environmental Protection Agency's Storm Water
Management Model II for the Littleton basin.
-------
EXPLANATION
• • • —— BASIN BOUNDARY
•&—— SUBASIN BOUNDARY AND
NUMBER
_...
^^» • *
39'43'30"
13
GUTTER OR PIPE AND NUMBER
\
*
-i
400
800 FEET
j
100
200 METERS
Figure 3.--Features related to use of U.S. Environmental Protection Agency's Storm Water Management
Model II for the Lakewood basin.
-------
372
area. Sensitivity of the other model parameters was somewhat
variable depending on basin characteristics and the range of
parameter values used. In general, runoff volumes were most
sensitive to the amount of effective impervious area and infiltration
rates. Peak flows were most sensitive to the amount of effective
impervious area, infiltration rates, and subbasin widths. Additional
sensitivity analysis made by the authors showed both runoff volumes
and peak flows to be sensitive to assumed errors of +5 percent in
basin area.
The initial simulations for model calibration were made using
the basin-characteristic parameters as reported by Ellis (1978)
and values for the infiltration and retention-storage parameters
as suggested in the user's manual for the model (Huber and others,
1975). The model parameters were then varied during trial-and-
error calibration guided by a knowledge of model sensitivity,
experience gained in previous simulations, and the ranges within
which model parameters realistically may be adjusted.
The model for all initial simulations underestimated storm-
runoff volumes and peak flows. Model calibration was then performed
as follows: During the first stage of calibration, the maximum
infiltration rate was changed from 3.0 to 0.5 in./h and the minimum
infiltration rate was changed from 0-5 to 0.2 in./h. These values
were more consistent with the soil characteristics of the basins
-------
373
(Larsen and Brown, 1971). Effective impervious areas for each
subbasin were then multiplied by a factor as much as 1.5. This
accounted for the effect of errors in determining effective impervious
areas, as Well as the effect of noneffective impervious areas. If
the average ratio of simulated runoff volume to measured runoff
volume could not attain 1-0 using this strategy, then all subbasin
areas were multiplied by a factor as much as 1.05 until this ratio
was 1.0. A "best" fit was then obtained in the timing of hydrograph
peaks and troughs by multiplying all subbasin widths by a factor
as much as 1.25. Upon completion of this step, the impervious-area
multiplication factor was reduced, if necessary, to achieve an
average ratio of simulated runoff volume to measured runoff volume
of 1.0. The model was then considered calibrated.
Littleton Site
Runoff periods of May 30, 1976, July 25, 1976, and June 11,
1977, were selected for model calibration. Runoff volumes for
initial (uncalibrated) simulations averaged 63 percent of the
measured runoff volumes and the peak flows averaged k2 percent of
the measured peak flows.
Model calibration was then performed. The final calibrated
model had maximum and minimum infiltration rates of 0.5 and 0.2 in./h,
respectively. In addition, all subbasin areas were multiplied by
-------
374
1.05, impervious areas for each subbasin were multiplied by 1.A8,
and all subbasin widths were multiplied by 1.25- The results for
the calibrated model are shown in figure 4. The model simulated
peak flows which averaged 79 percent of those measured.
Verification simulations were then made for runoff periods of
September 7, 1976, April 15, 1977, and April 19, 1977, using the
model parameters determined during model calibration. The results
of model verification are shown in table 1 and in figure 5- The
timing of hydrograph peaks and troughs was similar for the measured
and simulated hydrographs. However, the simulated runoff volumes
and peak flows were significantly larger than the measured runoff
volumes and peak flows for the runoff periods of September 7, 1976,
and April 19, 1977- The measured and simulated hydrographs of
April 15, 1977, were similar.
Lakewood Site
Runoff periods of April 12, 1977, April 19-20, 1977, and
May 20, 1977, were selected for model calibration. Runoff volumes
for initial (uncalibrated) simulations averaged 86 percent of the
measured runoff volumes for April 12, 1977, and April 19-20, 1977-
The simulated runoff volume for May 20, 1977, was about 40 percent
of the measured runoff volume. The runoff of May 20, 1977, which
-------
375
1700
1800
1900
100
25 H
1400
1500
1600
TIME (24-HOUR)
1500
2000
2100
1700
1800
Figure 4.—Simulated and measured hydrographs for runoff periods used
in model calibration for the Littleton site.
-------
Table 1 .--Measured and simulated storm-runoff volumes and peak flows
at the Littleton site
Date
Sept. 7, 1976
April 15, 1977
April 19, 1977
Measured
runoff
vo 1 ume
(in.)
0.07
.17
.08
Simul ated
runoff
vol ume
(in.)
0.14
.20
.13
Percent
error
100
18
62
Measured
peak flow
(ftVs)
52
33
U
Simulated
peak flow
(ftVs)
77
30
18
Percent
error
48
-9
29
-------
377
1600
1700
1800
40
a
z
o
UJ
on
30
20
O
s
o
? 10
April 15. 1977
400
600
800
1000
1200
10 -
5 -
1400
1600
1800 2000
TIME (24-HOUR)
2200
2400
Figure 5.-Simulated and measured hydrographs for runoff periods used
in model verification for the Littleton site.
-------
378
was the result of only 0.07 in. of measured rainfall, was deleted
from the calibration process because of its anomalously large
measured runoff volume and peak flow.
Model calibration was then performed. The final calibrated
model had maximum and minimum infiltration rates of 0.5 and 0.2 in./h,
respectively. In addition, impervious areas for each subbasin were
multiplied by 1.10. The results for the calibrated model are shown
in figure 6. The results for the runoff of May 20, 1977, are
included in figure 6, even though this runoff period was eliminated
from the calibration process.
Verification simulations were then made for runoff periods of
April 11, 1977, April 15, 1977, and May 20, 1977 (a different runoff
period than that used for model calibration), using the model
parameters determined during model calibration. The results of
model verification are shown in table 2 and in figure 7. The
measured and simulated hydrographs of April 11, 1977, and April 15,
1977, were similar. The simulated runoff volume and peak flow of
May 20, 1977, were significantly smaller than the measured runoff
volume and peak flow.
-------
379
1400 1500 1600 1700 1800 1900 2000 2100
1 -
1900
0100
2000
TIME (24-HOUR)
2100
Figure 6.—Simulated and measured hydrographs for runoff periods used
in model calibration for the Lakewood site.
-------
Table 2.--Measured and si-mutated storm-runoff volumes and peak flotis
at the Lakewood site
Date
1977
n — — ; i 11
Apri i i i
Apri i 1 3
M-.I. on
Measured
runoff
vo 1 ume
(in.)
On7A
• u 30
1 A
.ID
nna-j
Simulated
runoff Percent
vo 1 ume e r ro r
(in.)
OnLt i Q
. UH J 1 3
1 A n
.ID U
nn-jQ -eft
Measured Simulated
peak flow peak flow
(ftVs) (ft3/s)
3 A 36
h 1 L 7
**• J **• /
1 •? 7ft
Percent
error
^
D
°
-71
UJ
oo
o
-------
381
1800
0200
0400
0600
0800
1000
May 20, 1977
1400
1500
TIME (24-HOUR)
1600
Figure 7--Simulated and measured hydrographs for runoff periods used
in model verification for the Lakewood site.
-------
382
Total Nitrogen
The mathematical algorithms for rainfal1-runoff quality in
the U.S. Environmental Protection Agency's Storm Water Management
Model II are the relationships most widely used in urban storm-
runoff-quality models. The model will simulate settleable solids,
suspended solids, coliform bacteria, biochemical oxygen demand,
chemical oxygen demand, total nitrogen, and total phosphate. Of
these constituents, only total nitrogen was determined in samples
of rainfal1 runoff.
-------
383
There are two main components of the rainfal1-runoff-quality
algorithms in the model: Constituent accumulation and constituent
washoff. Constituent accumulation is estimated as a function of
land use, the number of days prior to the storm during which the
accumulated rainfall is less than 1.0 in., and street-cleaning
frequency and efficiency. The constituent load on the ground
surface (land-surface load) at the start of a storm is modeled as:
LLOAD=0_FACT x QFACT x DRYDAYS x GLEN x 453.6, (1)
where
LLOAD=1and-surface load, in milligrams;
Q_FACT=mi11igram of constituent per gram of "dust and dirt";
DFACT="dust and dirt" loading rate for each land use,
in pounds per day per foot of curb;
DRYDAYS=number of days prior to the storm during which
the accumulated rainfall is less than 1.0 in.
(modified to account for street sweeping); and
GLEN=curb length, in feet.
453.6=conversion factor (grams per pound).
The parameter values suggested in the user's manual for the
model (Huber and others, 1975) for total nitrogen accumulation are
shown in table 3-
-------
384
Table 3•~-Recommended values for total nitrogen
accumulation parameters
[From Huber and others, 1975]
DFACT
Land use
"Dust and dirt"
accumulat ion
(pound per day per
100 feet of curb)
QFACT
Mill igrams of
total nitrogen as N
per gram of
"dust and dirt"
S ingle- fami ly residential
Multifamily residential
• i
nous L r i a i -— —
0.
2.
.
,
i
7
3
^
..
c
0.48
.61
1,1
. H 1
l.o
-43
nr
-------
385
Washoff of constituents is modeled using an exponential decay
equation:
POFF = PO[l-e("BxRxAt)]5 (2)
where
POFF=amount of constituent removed from land surface during
a time step, in milligrams;
P0=amount of constituent on land surface at beginning
of time step, in milligrams;
B=decay coefficient;
R=runoff rate, in inches per hour; and
At=time step, in hours.
The primary assumption for use of this equation is that the amount
of constituent washed off during each time step is proportional to
the amount remaining on the land surface at the beginning of the
time step. The rate of washoff is controlled by the runoff rate
(R) for each time step and the exponent B. The exponent B is set
at a constant value of k.6, assuming that a uniform rainfall of
0.5 in./h would wash off 90 percent of the constituent in 1 hour.
For total nitrogen, the model assumes that ^.5 percent of the
suspended-solids concentration and 1.0 percent of the settleable-
solids concentration simulated by the model should be added to the
nitrogen concentration computed from equation 2.
-------
386
Littleton Site
The same runoff periods used in the flow calibration, May 30,
1976, July 25, 1976, and June 11, 1977, were used for the total
nitrogen calibration. The initial simulations were made using
values for the accumulation and washoff parameters as suggested in
the user's manual and values for the infiltration and basin-
characteristic parameters as determined from the flow calibration.
Comparisons of simulated with measured storm-runoff loads of
total nitrogen for the initial simulations are shown in table 4.
Simulation errors ranged from 1^0 to 1,000 percent. The extent of
this overprediction increases with the value of the model parameter,
DRYDAYS (number of days prior to a storm during which the accumulated
rainfall was less than 1.0 in.). This is consistent with the fact
that the data from the Littleton basin show no discernible effect
of DRYDAYS on storm-runoff loads of total nitrogen (fig. 8); whereas,
the model assumes that the land-surface load of total nitrogen at
the start of a storm is directly proportional to DRYDAYS.
-------
Table k.—Measured and simulated storm-runoff loads of
total nitrogen from the Littleton basin
Date DRYDAYS1
May 30, 1976— 31
July 25, 1976— 56
June 11, 1977— 57
Measured
load
(pounds)
30
30
38
Simulation I2
Simulated
load Percent
(pounds) error
72 140
240 700
420 1,000
Simulat
Simulated
load
(pounds)
3.1
7.3
13
ion 23
Percent
error
-90
-76
-66
1DRYDAYS is a model parameter set at the number of days prior to the storm
during which the accumulated rainfall is less than 1.0 in.
2Model constituent-accumulation and const!tuent-washoff parameters were set
at those values suggested by model documentation (Huber and others, 1975).
3Same as simulation 1 except no part of the suspended-solids or settleable-
solids concentration was added to the total nitrogen concentration.
oo
—I
-------
CO
CO
ro
0.08
0,06
o
cc
t iu
2 OC
,0
IL i/> 0.04
o§
^
o
z
cc
0.02
0.00
0.00
(33)
(56)
A
.(31)
(32)
(7)
(57)
(56)
A
(33)
Littleton site
Lake wood site
Number in parentheses is number
of days prior to storm during which
the accumulated rainfall was less
than 1.0 inch
I
0.05 0.10
STORM-RUN OFF VOLUME, IN INCHES
0.15
Figure 8.--Relation between storm-runoff load of total nitrogen and storm-runoff volume.
-------
389
A similar lack of discernible effects of number of days since
the preceding rainfall on storm-runoff loads has been reported by
Whipple and others (1977). Their results have been challenged by
Field (1978) and Lager (1978) because the storms used in their
analysis were of such small total storm volumes that very little
of the land-surface loads was washed off. However, for the three
runoff periods of table k, the percentage of the land-surface load
washed off (according to the model) varied from 28 to 48 percent.
A second simulation was made assuming that no part of the
suspended-solids or set.t leable-sol i ds concentration was added to
the total nitrogen concentration. The results, shown in table kt
indicate that the model simulations of total nitrogen are very
sensitive to these poorly defined parameters.
Most computer models which simulate the quality of rainfall
runoff from urban areas contain similar algorithms to those used
in the Storm Water Management Model II. Likewise, many of these
models have a similar set of recommended values for the model
parameters used in the runoff-quality algorithms. These models
have been widely used as a source of input for making management
decisions. The large simulation errors obtained for the Littleton
site suggest that caution should be used when applying these models
to assist in making water-quality-management decisions, particularly
in the absence of local data for model calibration and verification.
-------
390
A model option to specify the land-surface load for each
subbasin, in pounds per acre, at the beginning of storm runoff was
then used in place of the constituent-accumulation equation. The
land-surface loads for each of the three runoff periods were adjusted
until the ratio of simulated to measured storm-runoff load was 1.0.
Because of the lack of any other data, the land-surface loads were
allocated between land uses in the same proportions as the values
of DFACT x QFACT from table 3. The results of'the "calibrated"
model for each runoff period are shown in table 5- No relationship
of basin-average land-surface load with number of days since last
street sweeping is displayed by these results. In addition, the
rainfall that occurred before the runoff of May 30, 1976 (0.17 in.
on May 20 and 0.28 in. on May 21), did not appear to have noticeably
affected the land-surface load at the beginning of storm runoff on
May 30.
The results of the three calibration simulations are shown in
figure 9. These indicate that the simulated and measured rates of
decreasing total nitrogen concentrations with elapsed time from
the start of a storm are similar, resulting in some confidence in
the constituent-washoff equation (equation 2).
Three additional runoff periods were sampled for total nitrogen
but an insufficient number of samples were collected to provide
meaningful storm-runoff loads. However, these runoff periods could
be used for model verification of total nitrogen concentrations.
-------
Table 5.—Basin-average land-surface loads of total nitrogen at the beginning of storm runoff
based on model calibration for the Littleton basin
Date
Daily rainfall for previous 10 days1
(inches)
10 9 876 5^32
Bas in-average
Number of days land-surface load of
since last total nitrogen as N
street at the beginning of storm
1 sweeping runoff (pound per acre)
May 30, '
July 25,
June 1 1 ,
I976—
1976—
1977—
0.17
0
0
0.28
0
0
0
0
0
0
0
0
0.02
.06
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
10
66
>10
0.16
.17
.12
•^Refers to number of days prior to date of storm runoff.
-------
392
6 -
May 30, 1976
"•I-
1300
1400
1500
-------
393
Basin-average land-surface load of total nitrogen was assumed to
be equal to the average of the basin-average land-surface loads
shown in table 5- The results of the three verification simulations
are shown in figure 10.
Lakewood Site
The runoff periods of July 19, 1976, April 12, 1977, and
April 19-20, 1977, were used for model calibration. Again, the
initial simulations for model calibration were made using values
for the accumulation and washoff parameters as suggested in the
user's manual and values for the infiltration and basin-characteristic
parameters as determined from the flow calibration.
Comparisons of simulated with measured storm-runoff loads of
total nitrogen for the initial simulations are shown in table 6.
Simulation errors ranged from 67 to 9^0 percent. The extent of
this overprediction again increases with the value of the model
parameter, DRYDAYS. This is consistent with the fact that the data
from the Lakewood basin showed no discernible effect of DRYDAYS on
storm-runoff loads (fig. 8); whereas, the model assumes that the
land-surface load of total nitrogen at the start of a storm is
directly proportional to DRYDAYS. The percentage of land-surface
load washed off (according to the model) varied from 30 to 78 percent
for the three runoff periods of table 6.
-------
394
April 29-30, 1976
obi:
2100
30
20
10
2200
2300
lO
2400
8
cr
K
V)
d
5
£ 4
z
V)
<
ULI ?
O *
August 1, 1976
0200
1500
40
Q
O
CJ
CE
111
0.
20
O
CO
O
10?
O
0300
0
04oa
April 18, 1977
1600
TIME (24-HOUR)
1700
Figure 10.—Simulated and measured changes in total nitrogen with time for
runoff periods used in model verification for the Littleton site.
-------
Table 6.—Measured and simulated storm-runoff loads of
total nitrogen from the Lakewood basin
Date DRYDAYS1
I..1.. 1 Q 1Q"7£ — «__... 39
juiy iy, ly/o — jz
A—,-: 1 -I •> 1G"7"7_ „— •J 7
April \i, iy// 3J
April 19-20, 1977— 7
Simulation 12 Simulat
Measured Simulated Simulated
load load Percent load
(pounds) (pounds) error (pounds)
2.6 27 9*»0
3-7 -30 ~7QA 9 £
• / jj /yu ^. D
12 20 67 1.1
ion 23
Percent
error
-•71
j I
30
-90
DRYDAYS is a model parameter set at the number of days prior to the storm
during which the accumulated rainfall is less than 1.0 in.
2Model constituent-accumulation and constituent-washoff parameters were set
at those values suggested by model documentation (Huber and others, 1975).
3Same as simulation 1 except no part of the suspended-solids or settleable-
solids concentration was added to the total nitrogen concentration.
OJ
>>D
Ul
-------
396
A second simulation was made assuming no part of the suspended-
solids or settleable-solids concentration was added to the total
nitrogen concentration. The results, shown in table 6, again
indicate that the model simulations of total nitrogen are very
sensitive to these poorly defined model parameters. Again, the
large simulation errors suggest that caution should be used when
applying the Storm Water Management Model II or other similar models
to assist in making water-quality-management decisions, particularly
in the absence of local data for model calibration and verification.
The model option to specify the land-surface loads for each
subbasin, in pounds per acre, prior to storm runoff was again used
in place of the constituent-accumulation equation. The land-surface
loads for each of the three runoff periods were adjusted as previously
described for the Littleton basin. The results of the "calibrated"
model for each runoff period are shown in table 7- No relationship
of initial basin-average land-surface load of total nitrogen with
antecedent precipitation or number of days since last street sweeping
is displayed by these results. The initial basin-average land-
surface loads of total nitrogen are similar for the Littleton
(table 5) and Lakewood (table 7) sites.
-------
Table 7---Basin-average land-surface loads of total nitrogen at the beginning of storm runoff
based on model o^ibration for the Lakewood basin
Number Basin-average
Daily rainfall for previous 10 days1 of days land-surface load of
Date (inches) since last total nitrogen as N
street at the beginning of storm
10 9 876 54321 sweeping runoff (pound per acre)
July 19, 1976 0000000000 46 0.11
Apr. 12, 1977— 0 0 0 0 0 0 0 0 .16 .28 5 .11
Apr. 19-20, 1977— 0 .16 .28 .29 .02 .49 .21 0 .01 .03 12 .21
Defers to number of days prior to date of storm runoff.
-------
398
The results of the three calibration simulations are shown in
figure 11. Three additional runoff periods were sampled for total
nitrogen but an insufficient number of samples was collected to pro-
vide meaningful storm-runoff loads. However, these runoff periods could
be used for model verification of total nitrogen concentrations. Basin-
average land-surface load of total nitrogen was assumed to be equal
to the average of the basin-average land surface loads shown in table 7.
The results of the three verification simulations are shown in figure 12.
SUMMARY
Data from the Littleton and Lakewood basins were used for cali-
bration and verification of the U.S. Environmental Protection Agency's
Storm Water Management Model II for rainfall-runoff modeling of , flow
and total nitrogen. Using values for the accumulation and war^hoff
parameters as suggested in the user's manual (Huber and oth.ers, 1975)
and basin-characteristic parameters as determined from th£e flow cali-
bration, errors in simulated storm-runoff loads of totr.il nitrogen ranged
from 140 to 1,000 percent at the Littleton site and r,from 67 to 940 per-
cent at the Lakewood site. The extent of overpred'iction increased with
the value of the model parameter, DRYDAYS. The r.aodel assumption that
land surface loads are directly proportional to the number of days
prior to the storm during which the accumulated rainfall was less than
1.0 in. was not substantiated. The result:-, indicate that the model
simulations of total nitrogen are very sensitive to the poorly defined
-------
399
July 19, 1976
10
1700
1800 1900
TIME (24-HOUR)
2000
Figure 11.— Simulated and measured changes in total nitrogen with time for
runoff periods used in model calibration for the Lakewood site.
-------
400
June 17. 1976 X
1400
20
cc
LU
a.
15
O
g
< 0
5 1400
September 14, 1976
10
4 _
2 -
1000
1500
I I
October 6. 1976
1100
1200 1300
TIME (24-HOUR)
1400
0
1500
a
z
o
5
a.
CO
3
z
2 ~
lO
1600
0
1500
Figure 12.— Simulated and measured changes in total nitrogen with time for
runoff periods used in model verification for the Lakewood site.
-------
401
model parameters specifying the part of the suspended-solids and
settleable-solids concentrations added to the total nitrogen con-
centration. Based on the results of this study, caution should be
used when applying the Storm Water Management Model II or other
similar models to assist in making water-quality-managetaent decisions,
particularly in the absence of local data for model calibration and
verification
-------
402
REFERENCES
Alley, W. M., 1976, Guide for collection, analysis, and use of
urban storm-water data: American Society of Civil Engineers
Conference, Easton, Md., November 28-December 3, 1976, 115 p.
Ellis, S. R. , 1978, Hydrologic data for urban storm runoff from
three localities in the Denver metropolitan area, Colorado:
U.S. Geological Survey Open-File Report 78-410, 135 p.
Field, Richard, 1978, Discussion of effects of storm frequency on
pollution from urban runoff by Whipple, W., Jr., Hunter, J. V.,
and Yu, S. L.: Water Pollution Control Federation Journal,
v. 50, no. 5, P. 974-976.
Graham, P. H., Costello, L. S. , and Malon, H. J., 1974, Estimation
of impervious and specific curb length for forecasting storm-
water quality and quantity: Water Pollution Control Federation
Journal, v. 46, no. 4, p. 717-725.
Huber, W. L., Heaney, J. P., Medina, M. A., Peltz, W. A., Sheikh,
Hasan, and Smith, G. F. , 1975, Storm water management model
user's manual, version II: Cincinnati, Ohio, U.S. Environmental
Protection Agency, National Environmental Research Center,
EPA-67012-75-017, 350 p.
Lager, J. A., 1978, Discussion of effects of storm frequency on
pollution from urban runoff by Whipple, W., Jr., Hunter, J. V.,
and Yu, S. L.: Water Pollution Control Federation Journal,
v. 50, no. 5, p. 977.
-------
403
REFERENCES--Continued
Larsen, L. S., and Brown, J. B., 1971, Soil survey of Arapahoe
County, Colorado: U.S. Soil Conservation Service, 78 p.,
63 maps.
McPherson, M. B. , 1975, Urban hydrological modeling and catchment
research in the U.S.A.: American Society of Civil Engineers,
Urban Water Resources Research Program, Technical Memorandum
no. IHP-1.
Proctor and Redfern Limited and James F. MacLaren Limited, 1976,
Storm water management model study, volume 1: Toronto, Ontario,
Canada, Ontario Ministry of the Environment, Research Report 47,
295 p.
Shearman, J. 0., 1976, Computer applications for step-backwater
and flooding analysis: U.S. Geological Survey Open-File Report
76-499, 103 P-
Sinoot, G. F. , Davidian, Jacob, and Billings, R. H., 1974, Urban
storm rainfal1-runoff-quality instrumentation, in Symposium
on Flash Floods, Paris, September 1976, Proceedings: IAHS-
AISH Publication 112, p. 44-47.
Whipple, W. , Jr., ed., 1974, Urban runoff, quantity and quality:
American Society of Civil Engineers, 272 p.
Whipple, W., Jr., Hunter, J. V., and Yu, S. L., 1977, Effects of
storm frequency on pollution from urban runoff: Water Pollution
Control Federation Journal, v. 49, no. 11, p. 2243-2248.
-------
404
LIST OF ATTENDEES
SWMM USERS GROUP MEETING
MAY 24-25, 1979
AHMAD, Maqbool
Edmonton Water § Sanitation
8th Floor Century Edmonton
Alberta T5J 3A3
ARCHAMBAULT, Bernard
Ville de St-Bruno de Montarville
1585, rue Montarville
St-Bruno, Quebec J3V 3T8
ALLEN, Mark
Lozier Eng.
Rochester, N.Y.
BALDESARRA, Frank
885 Don Mills Rd
Don Mills, Ontario
M3C 3H1
ALLEY, W.
U.S. Geological Survey
Surface Water Branch, MS #415
Reston, Virginia 22092
BASTIEN, Jean H.
Ville de St-Laurent
777, boul. Laurentien
St-Laurent, Quebec H4M 2M7
AMMON, Douglas
Woodbridge Avenue Bldg 10
Storm and Combined Sewer Section
Edison, N.J. 08817
BEAUDOIN, Pierre
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
ANDERSON, James A.
Urban Science Applications Inc,
1027 Fisher Bldg
Detroit, Michigan 48202
BELLEMARE, Guy
Les Consultants Pluritec
585, Avenue des Cedres
Shawinigan, Quebec G9N 1M6
ARBUCKLE, A.
R.E. Clipsham Ltd
16, Mountainview Road South
Georgetown, Ontario
BERGERON, Richard
S.P.E.Q.
2360, Chemin Sainte-Foy
Ste-Foy, Quebec G1V 4H2
-------
405
BERNIER, Richard ing.
Lajoie Pellerin et Associes
635, Marguerite Bourgeois
Quebec, Quebec CIS 3V8
BOURBEAU, Michel ing.
Hamel, Ruel, Beaulieu § Associes
150, rue Marchant, Suite 600
Drummondville, Quebec J2C 4N1
BERON, Patrick
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
BESSE, Andre
'Ville de Montreal (Travaux Publics)
700 est, rue St-Antoine
Montreal, Quebec H2X 1A6
BHARGAVA, Dilip
O'Brien § Gere Engineers Inc.
5000 Overlook Avenue S.W.
Washington D.C. 20032
BODNARUK, Larry
Northwest Hydraulic Consultants Ltd
4823-99 Street
Edmonton, Alberta
BOISCLAIR, Jean
Consultants V.F.G. Inc.
3300 boul. Cavendish Suite 385
Montreal, Quebec H4B 2M8
BOLAND, Bruce
Ministry of the Environment
London, Ontario
N6E 1V3
BONENFANT, Jean Patrick ing.
Ville de Laval
1, Place du Souvenir
Laval, Quebec
BRIGHT, Clive
966, Waverley Street
Winnipeg, Manitoba R3T 4M5
BRUNNETTE, Jean Guy
Ingenieurs-Conseils
4935, Beaubien est
Montreal, Quebec HIT 1V1
BURROUGHS, Denis
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
BUTTON, Charles
Boston Water § Sewer Commission
10 Post Office Square
Boston MA 02109
CANDARAS, A.M.
Paul Theil Associates Ltd
700 Balmoral Drive
Bramalea, Ontario L6T 1X2
CARLEO, David J.
O'Brien Gere Engineers Inc.
1304 Buckley Rd
Syracuse N.Y. 13221
CARTIER, Pierre
Desjardins Sauriol § Associes
1200 boul. Saint-Martin ouest
Laval, Quebec H7S 2E4
-------
406
CART1ER, Richard
Socicte d'ingenierie Cartier
2045, rue Stanley
.Montreal, Quebec
H3C 3Z8
CASSEUS, Airy ing.
Ville de St-Bruno de Montarville
1585, rue Montarville
St-Bruno, Quebec J3V 3T8
CHEUNG, Philip
Department of Civil Engineering
University of Ottawa
Ottawa, Ontario KIN 9B4
CHOINIERE, Maurice
Ville de Montreal
700 est, St-Antoine
Montreal, Quebec H2Y 1A6
CLARK, Philip G.
O'Brien § Gere Engineers, Inc.
1304, Buckley Road
Syracuse, New York
CLARKE, W.G.
James F MacLaren Ltd
435 McNicoll Avenue
Willowdale ont.
CODERE, Martine
Universite de Sherbrooke
Departement de Genie Civil
Sherbrooke, Quebec J1K 2R1
COOK, Donald
P.O. Box 3025
St-Catherines, Ontario
L2R 7E9
COURTEMANCHE, J.L.
Groupe SNC
1 Complexe Desjardins, C.P. 10,
Succursale Desjardins
Montreal, Quebec H5B 1C8
COUTU, Andre
Ecole Polytechnique, Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
CREVIER, Raymond ing.
Besette, Crevier, Parent, Tanguay § Associes
110, boul. Cremazie ouest, bureau 1001
Montreal, Quebec H2P 1B9
DE GUIDA, Richard
Calocerinos § Spina Cons. Eng.
1020 Seventh North Street
Liverpool New York 13088
DELLAH, Abdellatif
Lavalin, Inc.
1130, ouest Sherbrooke
Montreal, Quebec H3A 2R5
DELLEUR, Jacques
School of Civil Engineering
Purdue University
West Lafayette IN 47907
DENNISON, Karren
Philips Planning § Eng.
3215 Queen Elizabeth Way
Burlington, Ontario L7R 3Y2
DESBIENS, R.
COTE, Bertrand
Universite de Sherbrooke
Departement de Genie Civil
Sherbrooke, Quebec J1K 2R1
COTE, Jean ing.
Hamel Ruel Beaulieu § Associes
150 rue Marchant, suite 600
Drummondville, Quebec J2C 4N1
COUILLARD, Denis
INRS - Eau
Universite du Quebec
9S5, de 1'Auntie
Ancienne Lorette Quebec G2E 3R5
DINIZ, Elvidio
Espey Huston § Associates
2500 Louisiana Blvd NE
Suite 225
Albuquerque, New Mexico 87110
DOWNEY, Dale
Marshall Macklin Monaghan Ltd
275, Duncan Mill Road
Don Mills, Ontario M3B 2Y1
DUFORT, Daniel
Chagnon-Ratelle § Associes
140, rue St-Eustache
Saint-Eustache, Quebec
J7R 2K9
-------
407
DUNLEVY, Neal
Datson Dale £ Associates
Utica, New York
DUPUIS, Lucien
Dupuis Morin Routhier § Associes
3333, boul. du Souvenir
Laval, Quebec H7V 1X1
EAMON, Rick
Ministry of the Environment
4, Montreal Road
Cornwall, Ontario
EL-JABI, Nassir
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec. H3C 3A7
ETHIER, Pierre ing.
Bessette, Crevier, Parent, Tanguay § Ass,
110, boul. Cremazie ouest, bureau 1001
Montreal, Quebec H2P 1B9
FALK, Jan
Dept of Water Resources Engineering
University of Lund, Pack •
S022007, Lund
Sweden
FAY, Kerin J.
Ecol Sciences, Inc.
North Central Region
Suite 715
735 N. Water Street
Milwaukee, Wisconsin 53202
GEORGE, Bruce L.
Cole, Sherman £ Associates
2025 Sheppard Avenue
Willowdale, Ontario
GIGNAC, Claude
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
GIRARD, Marc Andre
Roy Bergeron Gariepy Leroux Dupont
545, 12e Avenue, Laval des Rapides
Laval, Quebec H7N 4C5
GUPTA, Swapan
Dept. of Civil Engineering
University of Ottawa
Ottawa, Ontario KIN 9B4
HAGARMAN, Jin
316 Oak Rd
Glenside Pensylvania
19038
HANDFIELD, Jocelyn
Societe Municor Inc.
1470 Chemin Chambly
Longueuil , Quebec J4J 3X3
HARLCW, Carl
Urban Science Applications
1027, Fisher Bldg
Detroit, MI 48202
FORTIN, Robert
Ecole Polytechnique, Genie Civil
2696 Marlborough Court, #3
Saint-Laurent, Quebec H4K 1M1
GAUTHIER, Gilles
Societe Municor, Inc.
1470, Chemin Chambly
Longueuil, Quebec J4J 3X3
GEIGER, Wolfgang F.
Strasslacher Str. 2
SOOOMunich 71
Germany
FECTAU, Pierre ing. Dir. Gen.
Ville de Lac Megantic
5527, rue Frontenac, bureau 200
Lac Megantic, Quebec G6B 1H6
Ass,
HAY, D.J.
Environmental Protection Service
Environment Canada
Ottawa, Ontario
HENSON, Steve
Clyde E.Williams £ Ass. Inc.
1843 Commerce Dr.
South Bend.IN 46628
HOPKINSON, Charles
Center for Wetland Resources
Louisiana State University
Baton Rouge.Louisiana 70803
HUBER, Wayne
University of Florida
Dept of Env. Eng. Sciences
Gainesville,Florida 32611
-------
408
I BACH, ames R.
Milwaukee Metropolitan Sewerage Dist
755 N. Water St.
Milwaukee, Wisconsin 53202
JONES, David B.
O'Brien P, Gere Eng. Inc.
1504 Buckley Rd
Syracuse N.Y. 13221
KAMKDULSKT, Gregory E.
Maicom Pirnie Inc.
2 Corporate Park Dr.
White Plains N.Y. 10602
KASSEM, Atef M.
Civil Engineering Dept
University of Ottawa
Ottawa, Ontario KIN 9B4
KEAY, William L.
Regional Municipality of
Ottawa Carleton
222 Queen St.
Ottawa, Ontario
KENNEDY, Ian
J.L. Richards § Ass. Ltd
864 Lady Ellen PI.
Ottawa, Ontario K1Z 5M2
LABONTE, Roger
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal', Quebec H3C 3A7
LAFONTAINE, Andre
H6tel de Ville de Chateauguay
5, boul. Youville
Chateauguay, Quebec J6J 2P8
LAURENSON, Eric
Monash University
Clayton, Victoria 3168
Australia
LAVALLEE, Pierre
SPEQ - Qualite du milieu
2560, Chemin Sainte-Foy
Sainte-Foy, Quebec G1V 4H2
LECLERC, Guy
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
LEE, Wayne
1479, Buffalo Place
Winnipeg, Manitoba R3T 1L7
LEMIEUX, Henri Julien
James F MacLaren Ltee
2600 est, rue Ontario bureau 200
Montreal, Quebec H2K 4K4
LEMIEUX, Pierre
Universite de Sherbrooke
Departement de Genie Civil
Sherbrooke, Quebec J1K 2R1
LEMIRE, Rene ing. jr.
Lalonde, Girouard, Letendre
1400 ouest , Sauve suite 214
Montreal, Quebec H4N 1C5
Ass. Ltee
LOISELLE, Philippe
Citee de Salaberry de Valleyfield
61 Sainte Cecile, H6tel de ville
Valleyfield, Quebec J6T 1L8
LORANT, F. Ivan P. Eng.
M.M. Dillon Limited
Box 219, Station K
Toronto, Ontario M4P 2G5
LOWING, Dr. M.J.
Whitehouse Lodge
Brightwell, Baldwin
Oxford, U.K.
MACLEOD, Paul
Giffels Associates Ltd
30 International Blvd
Rexdale, Ontario M9W 5P3
MARCHI, Gilles
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
-------
MASSING, Herbert
Landesanstalt ftir Wasser und Abfall
Nordrhein - Westfalen
4 Dtisseldorf 1
BOrnestrabe 10,
Federal Republic of Germany
MAST, Stanley I
HNTB
1345, Avenue of the America
New York, N.Y. 10019
McCULLOCH, James ing.
Ville de Shawinigan Sud
1550 - 118e rue
Shawinigan Sud, Quebec
McEWAN, John
Damas § Smith Ltd
265 Yorkland Blvd
Willowdale, Ontario M2J 1S5
McKENNA, Gerry
District Officer, Ministry of the
Environment
4 Montreal Road
Cornwall, Ontario
McKENZIE, John
Stanley § Associates
11748 Kingsway Avenue
Edmonton, Alberta T56 0X5
McPHERSON, M.B.
ASCE - Ecole Polytechnique
23 Watson St.
Marblehead Mass. 01945
MEINHOLZ, Thomas L.
Ecol Sciences Inc.
735 No Water St.
Suite 715
Milwaukee, Wisconsin 53202
MERCURE, Richard
Ville de Sherbrooke
1255 rue Daniel
Sherbrooke, Quebec J1H 5X3
MITCI, Constant in
Communaute Urbaine de Montreal
2620, boul. Saint-Joseph est
Montreal, Quebec H1Y 2A4
409
MORIN, Professeur Guy
Universite du Quebec
INRS - Eau, 2700 Einsten
Sainte Foy, Quebec G1V 4C7
NAM, Nguyen V.H., ing. M.Sc.A.
Communaute Urbaine de Montreal
2620 est, boul. Saint Joseph
Montreal, Quebec H1Y 2A4
NGUYEN, Van-Thanh-Van, ing. Ph.D.
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
NOVAK, Z.
Water Resources Branch
135 Saint Clair Avenue W.
Toronto, Ontario M4V IPS
OLIVIER, Bruno
Ville de Montreal
700 est, St-Antoine
Montreal, Quebec
OWENS, Emmet
StearnsfWheler
10 Albany Street
Cazenovia, New York 13035
PAQUIN, Guy
Communaute Urbaine de Montreal
2620, boul. Saint-Joseph est
Montreal, Quebec H1Y 2A4
PARENTE, Mario
Gore § Storie Ltd
Suite 700
331 Cooper Street
Ottawa, Ontario
PATEL, Ramji
, 25185 Goddard Rd
Taylor, Michigan 48180
PATRY, Gilles
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
PERKS, Alan P. Eng.
The Proctor § Redfern Group
75 Eglington Avenue East
Toronto, Ontario M4P 1H3
-------
410
PERRAULT, Andre
A.Q.T.E.
6290, rue Perinault
Montreal, Quebec H4K 1K5
SAINT-DENIS, Richard
Barre Pellerin Lemoine Inc.
4274, Avenue Papineau bureau 300
Montreal, Quebec H2H 2P4
POLINSKY, Joseph SAINT- JACQUES, Jean ing.
Calocerinos § Spina Consulting Engineers Lalonde Girouard Letendre
69 Delaware Avenue 1400 ouest, Sauve Suite 214
Buffalo, N.Y. 14202 Montreal, Quebec .H4N 1C5
Associes Ltee
POND, Larry
W.L. Wardrop § Associates Ltd
77 Main Street
Winnipeg, Manitoba R3C 3H1
PURENNE, Pierre
Desjardins Sauriol § Associes
1200 boul. Saint Martin ouest
Laval, Quebec H7S 2E4
RACE, Richard
Milwaukee Metropolitan Sewerage
Dist.
735 N. Water Street
Milwaukee, Wisconsin 53202
RADZIUS, Mr.
Ministry of Transport
Ottawa
RAYMOND, Louise ing. M. Ing.
Bessette Crevier Parent Tanguay § Ass>.
110, boul. Cremazie ouest, bureau 1001
Montreal, Quebec H2P 1B9
ROBITAILLE, Simon ing.
Louis -P. Couture § Associes
1380 Saint Cyrille ouest
Quebec, Quebec CIS 1W6
ROCHON, Jean Pierre
Hotel de Ville de Chateauguay
5, boul. Youville
Chtteauguay, Quebec J6J 2P8
ROSA, Gilles
Ville de Saint-Hubert
5900, boul. Cousineau
Saint-Hubert, Quebec J3Y 7K8
ROUSSELLE, Jean
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
SEGUIN, Fernand
Larocque Samson Guerette & Associes
5245, boul. Cousineau
Saint-Hubert, Quebec J3Y 6J8
SHAPIRO, Howard
Lozier Eng.
Rochester, N.Y.
SHUBINSKI, Robert Dr.
Water Resources Engineers
8001 Forbes Place
Springfield, Virginia 22151
SIMMONDS, J.C.
City of Ottawa
Sussex Drive
Ottawa, Ontario
SIMON, Jacques
Desjardins § Sauriol § Associes
1200 ouest, boul. Saint Martin
Laval, Quebec H7S 2E4
SKRETTEBERG, Rolf
Norwegian Water
Ressurces and Electricity Boorel
Middelthunsgt 29
Oslo 3, Norway
SPRENGER, Reinhard
966 Waverley Street
Winnipeg Manitoba R3T 4M5
STEPHENSON, Ralph
3901, Blvd Industrial
Indianapolis, Indiana 46254
STOM, Lloyd Edward
Gannett Fleming Corddy §
Carpenter Inc.
No 11 Koger Executive Center
Suite 218
Norfolk, Virginia
SWAN, Donald E., P.E.
25185 Goddard Rd.
Taylor, Michigan 48180
-------
411
TERSTRIEP, Michael
Illinois State Water Survey
Box 232
Urbana, Illinois 61801
THODEN, William
Calocerinos § Spina
69 Delaware Avenue
Buffalo, N.Y. 14202
THOMPSON, Larry
Philips Planning § Eng.
3215 Queen Elizabeth Way
Burlington, Ontario L7R 3Y2
TORNO, Harry
U.S. EPA, RD 682
Washington DC 20460
TREMBLAY, Pierre
ABBDL Inc.
85 ouest Saint Catherine
Montreal, Quebec H2X 3P4
TREZOS, Vomvoris Denise
Societe d'ingenierie Shawinigan
620, Ouest Dorchester
Montreal, Quebec „
TROTTIER, Jacques L.
Laval in Inc.
1130, Sherbrooke ouest
Montreal, Quebec H3A 2R5
UPMEYER, David W.
Black § Veatch
200 Renaissance Center Suite 1220
Detroit Michigan 48243
URBONAS, Ben
Urban Drainage and
Flood Control Division
2480 W. 26th Avenue
Suite 156B
Denver Colorado 80211
VALIQUETTE, Luc
Ecole Polytechnique
Departement de Genie Civil
C.P. 6079, Succursale "A"
Montreal, Quebec H3C 3A7
VALOIS, G. Bourbonnais, ing.
Ville de St-Laurent
777, boul. Laurentien
St-Laurent, Quebec H4M 2M7
VISSER, Peter
Gore § Storie Ltd
331 Cooper Ltd, Suite 700
Ottawa, Ontario
WALESH, Stuart G.
Donohue § Associates, Inc.
1915 Mac Arthur Road
P.O. Box 256
Waukesha, Wisconsin 53187
WALKENSHAW, Barry
Burgess § Niple Ltd
5085 Reed Rd
Columbia, Ohio 43220
WARD, Gerry
University of New Brunswick
Department of Civil Engineering
New Brunswick
WARK, Gordon
300-77 Main Street
Winnipeg, Manitoba R3C 3H1
WEATHERBE, Don
Water Resources Branch
135 St Clair Avenue W,
Toronto, Ontario M4V IPS
WENZEL, Harry G., Jr.
Department of Civil Engineering
University of Illinois
Urbana, Illinois 61801
WIGGINS, Neil
A.S. Robinson and Associates
P.O. Box 13130
Kanata Postal Station
Ottawa, Ontario K2K 1X3
WILKINSON, C.
WISNER, Paul
University of Ottawa
Department of Civil Engineering
Ottawa, Ontario
ZUKOVS, G.
Ministry of Environment
135 St. Clair Avenue W.
Toronto, Ontario M4V 1P5
-------
TECHNICAL REPORT DATA
\i tease read Inslrnciiom on the reverse before completing)
REPORT NO.
EPA-600/9-7 9-026
2.
3. RECIPIENT'S ACCESSION NO.
. TITLE AND SUBTITLE
Proceedings
Storm-water Management" Model (SWMM)
Users Group Meeting, ^A-25 May 1979_
5. REPORT DATE
June 1979
6. PERFORMING ORGANIZATION CODE
. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
Edited by Harry C. Torno
i. PERFORMING ORGANIZATION NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Research and Development (RD-682)
Washington, B.C. 20460
12. SPONSORING AGENCY NAME AND ADDRESS
Office of Air, Land and Water U'sO
U.S. Environmental Protection Ag>ei.icy
Office of Research and Development (RD-682)
Washington, D.C. 20460
10. PROGRAM ELEMENT NO.
1 BA 609
11. CONTRACT/GRANT NO.
N/A
13. TYPE OF REPORT AN.D PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/16
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report includes fourteen papt'-i's, on various model-related topics.
Presented at the semi-annual joint U. S,.--Canad:i-an Stormwater Management Model
(SWMM) Users Group Meeting, held May 24--.25, 1979, in Montreal, Quebec, Canada.
Topics covered included applications ,o.f the SWMM to urban and rural catch-
ments, the selection of design rainfall events for model use, a description
of the Montreal Urban Community's wastewate'r' interception program, several
applications of other models in facility planning and design, development of
a mini-computer version of the ILLUDAS model, &nd development of an urban/
rural subcatchment hydrologic simulation model (SUBHYD).
KEY WORDS AND DOCUMENT ANALYSI.?
DESCRIPTORS
Mathematical Models
Combined Sewers
Runoff
Hydrology
Urban Hydrologic Models
Urban Hydrolog y
Combined Sewer Overflows
Stormwater Runoff
c. COSATI 1 ickl/Croup
13 B
'*. OI3TRI8U T i ON S I'ATEMENT
Release to Public
l~ At Cl.iRITY Cl. ASS ,• ,'V;/v KYp.o-;
Uncla ss if ied.. ..
Unclassified
-------
'fySs RUCTIONS
1 REPORT NUMBER
insert the t.PA report number as it appears on the cover of the publication.
2. LEAVE BLANK
3. RECIPIENTS ACCESSION NUMBER
Reserved for use by each report recipient.
4. TITLE AND SUBTITLE
Title should indicate clearly and briefly the subject coverage of the report, apd be displayed prominer.i tly. Set subtitle, if used, in smaller
type or otherwise subordinate it to main title. When a report is prepared in, more than one volume, r epeat the primary title, add volume
number and include subtitle for the specific title.
5. REPORT DATE
Kach report shall carry a date indicating at least month and year. Indicate the basis on which it was selected (e.g., date of issue, date of
approval, date of preparation, etc.).
6. PERFORMING ORGANIZATION CODE
Leave blank.
7. AUTHOR(S)
Give name(s) in conventional order (John R. Doe, J. Robert Doe, etc.). List author's af.f-'jiation if it differs from the performing organi-
zation.
8. PERFORMING ORGANIZATION REPORT NUMBER
Insert if performing organization wishes to assign this number.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Give name, street, city, state, and ZIP code. List no more than two levels of an c/rganizatkmal hirearchy.
10. PROGRAM ELEMENT NUMBER
Use the program element number under which the report was prepared. Sub r^rdinate numbers may be included in par: inheres.
11. CONTRACT/GRANT NUMBER
Insert contract or grant number under which report was prepared.
12. SPONSORING AGENCY NAME AND ADDRESS
Include ZIP code.
13. TYPE OF REPORT AND PERIOD COVERED
Indicate interim final, etc., and if applicable, dates covered;.,
14. SPONSORING AGENCY CODE
Leave blank.
15. SUPPLEMENTARY NOTES
Enter information not included elsewhere but useful, sweji as: prepared in cooperation with, Translation i>f. Fvt-si-ntcd at conference of,
To be published in, Supersedes, Supplements, etc.
16. ABSTRACT
Include a brief (200 words or less) factual summary of the most significant information contained in the report. If the report contains 3
significant bibliography or literature survey, me^icta it h',ire
17. KEY WORDS AND DOCUMENT ANALYSIS
(a) DESCRIPTORS - Select from the Thesaiw* of Enf peering and Scientific Terms the proper authorized terms that identify the major
concept of the research and are sufficiently apcific ?, nd precise to be used as index entries for cataloging.
(b) IDENliriERS AND OPEN-ENDED TERMS - (jse identifiers for project names, code names, equipment designators, etc. Use open-
ended terms written in descriptor form fot t.feose subjects for which no descriptor exists.
(c) COSATI FIELD GROUP-Field andgpup 'assignments are to be taken from the 1965 COSATI Subject Category List. Since the ma-
jority oi documents are multidiscipUnajy in nature, the Primary Field/Group assignmeiit(s) will be specific discipline area of human
endeavor, or type of physical object. The ap-pi-lcation(s) will be cross-referenced with secondary Field/Group assignments that will follow
the primary postmg(s).
18. DISTRIBUTION STATEMENT
Denote reusability to the public sxrUmi cat jon for reasOns other than security for example "Release Unlimited." Cite any availability to
the public, with address and price...
19. &20. SECURITY CLASS!FICAf ION
DO NOT submit classified reports to tl-,e National Technical Information service.
21. NUMBER OF PAGES
Insert the total number of pages, i.needing this one and unnumbered pages, but exclude distribution list, if any.
22. PRICE
Insert the price set by the Nati< jr.al Technical Information Service or the Government Printing Office, if known.
#U.S. GOVERNMENT PRINTING OFFICE: 1979-281-147:112
------- |