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

-------