United States Office of Water Planning and Water Quality Management Guidance
Environmental Protection Standards (WH 551) EPA 440/3-79-023
Agency Washington nc 20460 May 1979
Water
vvEPA A Statistical Method
for the Assessment of
Urban Storm water
-------
A STATISTICAL METHOD
FOR ASSESSMENT OF URBAN STORMWATER
LOADS - IMPACTS - CONTROLS
FOR
EPA - NON POINT SOURCES BRANCH
WASHINGTON, D.C.
PROJECT OFFICER
DENNIS N. ATHAYDE
MANAGER
NATIONWIDE URBAN RUNOFF PROGRAM
JANUARY 1979
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
DATE:
191979
SUBJECT:
FROM:
TO:
Transmittal of Document Entitled "A Statistical Method
for Assessment of Urban Runoff" -',
Swep T. Davis, Deputy assistant Administrate]
Office of Water Planning and Standards "
All Regional Water Division Directors
ATTN: All Regional 208 Coordinators
All Regional NPS Coordinators
All Nationwide Urban Runoff Prototype Projects
All State and Areawide Water Quality Management Agencies
Other Concerned Groups
TECHNICAL GUIDANCE MEMORANDUM-TECH- 49
Purpose
This document "A Statistical Method for Assessment of Urban Runoff"
has been prepared to provide technical assistance to the Nationwide
Urban Runoff prototype projects and other interested groups in assessing
the impact of urban stormloads on the quality of receiving waters, and
to evaluate the cost and effectiveness of control measures for reducing
these pollutant loads.
Guidance
The enclosed report is provided in accordance with the Nationwide
Urban Runoff Program established under Section 208 of the Clean Water
Act of 1977. This methodology is appropriate for use at the planning
level where preliminary assessments are made to define problems, establish
the relative significance of contributing sources, assess feasibility of
control, and determine the need for and focus on additional evaluations.
It can also be used effectively in conjunction with detail studies, in
evaluating the most cost-effective alternatives for controlling urban
runoff.
Attachment
EPA FORM 1320-6 (REV. 3-76)
-------
STORMWATER MANUAL
TABLE OF CONTENTS
Chapter Number
Number Page
1 INTRODUCTION 1-1
2 STORMWATER RUNOFF - A REVIEW OF THE PROBLEM 2- 1
2.1 MAJOR WATER QUALITY PROBLEMS IN URBAN AREAS 2- 2
2.2 URBAN RUNOFF 2-3
2.3 CONTAMINANTS IN URBAN RUNOFF 2-3
2.4 WATER QUALITY PROBLEM DEFINITION 2- 6
2.4.1 RELEVANT TIME AND SPACE SCALES 2-11
2.4.1.1 NATURE OF CONTAMINANTS 2-11
2.4.1.2 NATURE OF THE RECEIVING WATER
SYSTEM 2-12
2.4.2 WATER USE OBJECTIVES AND CRITERIA 2-12
2.4.3 CHARACTERISTICS OF PARTICLAR STUDY AREA 2-15
2.5 REFERENCES 2-16
3 THE STATISTICAL METHOD FOR THE ASSESSMENT OF RUNOFF
AND TREATMENT 3-1
3.1 STORM RUNOFF EVENTS AS RANDOM OCCURRENCES 3-2
3.2 CHARACTERIZATION OF RUNOFF EVENTS 3-3
3.2.1 STATISTICAL PROPERTIES OF RUNOFF
PARAMETERS 3-5
3.2.2 LONG TERM RUNOFF PROCESS 3- 6
3.2.3 DETERMINATION OF RUNOFF PARAMETERS 3-10
3.3 RAINFALL, THE DRIVING FORCE 3-11
3.4 DEVELOPMENT OF STORMWATER LOADS 3-16
3.4.1 RUNOFF QUANTITY 3-16
3.4.2 RUNOFF QUALITY AND RESULTING LOADS 3-19
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
3 3.5 IMPACTS IN THE RECEIVING WATER 3-26
3.5.1 PREDICTION OF LONG TERM IMPACTS 3-28
3.5.1.1 STREAMS AND RIVERS 3-28
3.5.1.2 ESTUARIES AND COASTAL WATERS 3-30
3.5.1.3 LAKES AND RESERVOIRS 3-33
3.5.2 PREDICTION OF TRANSIENT IMPACTS 3-36
3.5.2.1 STREAMS AND RIVERS 3-36
3.5.2.2 ESTUARINE SYSTEMS 3-42
3.6 ASSESSMENT OF STORMWATER CONTROL ALTERNATIVES 3-43
3.6.1 STRUCTURAL TREATMENT DEVICES 3-46
3.6.1.1 INTERCEPTION 3-46
3.6.1.2 STORAGE 3-48
3.6.1.2.1 EFFECT OF PREVIOUS STORMS 3-50
3.6.1.2.2 STORAGE EFFECTIVENESS 3-50
3.6.1.2.3 FIRST FLUSH EFFECT 3-53
3.6.1.2.4 TREATMENT OF STORED
RUNOFF 3-57
3.6.1.2.5 EXAMPLE OF STORAGE DEVICE
EVALUATION 3-57
3.6.1.3 INTERCEPTION AND STORAGE 3-59
3.6.1.4 IN-LINE TREATMENT DEVICES WHICH
REDUCE POLLUTANT CONCENTRATION 3-60
3.6.1.4.1 EXAMPLE: ANALYSIS OF
IN-LINE TREATMENT DEVICE 3-63
3.6.1.4.2 CONCENTRATION SENSITIVE
TREATMENT DEVICES 3-65
3.6.1.4.3 DISINFECTION 3*67
11
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
3 3.6.1.5 COMBINED TREATMENT SYSTEMS WHICH
CAPTURE, STORE AND TREAT RUNOFF 3-69
3.6.1.6 THE EFFECTS OF STORMWATER CONTROL
DEVICES ON THE FREQUENCY OF
LOADINGS 3-69
3.6.1.6.1 INTERCEPTION 3-70
3.6.1.6.2 STORAGE 3-70
3.6.1.6.3 IN-LINE TREATMENT DEVICES 3-72
3.6.1.6.4 DISINFECTION 3-72
3.6.1.7 COMPARISON OF STATISTICAL METHOD TO
SIMULATIONS OF TREATMENT 3-76
3.6.1.7.1 EFFECT OF PREVIOUS STORMS 3-76
3.6.1.7.2 STORAGE DEVICE PERFORMANCE 3-76
3.6.1.7.3 INTERCEPTION AND STORAGE 3-78
3.6.1.7.4 IN-LINE TREATMENT DEVICE 3-78
3.6.2 MANAGEMENT PRACTICES FOR STORMWATER CONTROL 3-85
3.6.2.1 SOURCE CONTROLS: STREET SWEEPING
AND CATCH BASIN CLEANING 3-85
3.6.2.2 CONTROL OF HARMFUL MATERIALS 3-86
3.6.2.3 EROSION CONTROL 3-88
3.6.2.4 CONTROL OF SURFACE FLOWS 3-88
3.6.2.5 LAND USE CONTROL 3-90
3.6.2.6 COLLECTION SYSTEM MANAGEMENT:
SEWER SEPARATION 3-90
3.6.2.7 INFILTRATION AND INFLOW CONTROL 3-91
3.6.2.8 SEWER FLUSHING 3-91
3.6.2.9 POLYMER INJECTION 3-92
3.6.2.10 AUTOMATED SYSTEM CONTROL 3-92
3.6.3 BENEFIT AND COST EVALUATION 3-93
3.6.3.1 IMPROVEMENTS IN RECEIVING WATER
QUALITY 3-93
iii
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
3 3.6.3.2 INDIRECT BENEFITS OF STORMWATER CONTROL 3-94
3.5.3.3 EVALUATING COST OF CONTROLS 3-94
3.7 REFERENCES 3-95
4 SIMULATION OF STORMWATER IMPACTS 4-1
4.1 AVAILABLE MODELS 4- 1
4.2 BROADSCALE RECEIVING WATER SIMULTOR 4- 2
4.2.1 INTERNAL CALCULATIONS 4-5
4.2.2 PROGRAM CAPABILITIES AND LIMITATIONS 4- 5
4.2.3 MODEL INPUTS 4-5
4.2.4 MODEL OUTPUT 4-7
4.3 REFERENCES 4-7
5 NUMERICAL ESTIMATES FOR STORMWATER ASSESSMENT METHODOLOGIES 5- 1
5.1 RAINFALL 5-1
5.1.1 PRECIPITATION CHARACTERISTICS IN DIFFERENT
PARTS OF THE UNITED STATES 5-1
5.1.2 DEFINITION OF STORM EVENT 5-20
5.1.3 APPLICABILITY OF THE GAMMA DISTRIBUTION FOR
STORM EVENT CHARACTERISTICS 5-22
5.1.4 AREAL DISTRIBUTION OF RAINFALL 5-37
5.1.4.1 RAINGAGE AGGREGATION 5-39
5.2 RUNOFF QUANTITY 5-46
5.2.1 DETERMINATION OF AVERAGE RUNOFF TO RAINFALL
RATIO 5-46
5.2.2 DETERMINATION OF AVERAGE DURATION OF RUNOFF
EVENT 5-52
5.3 RUNOFF QUALITY 5-60
IV
-------
TABLE OF CONTETS
(Continued)
Chapter Number
Number Page
5 5.3.1 DETERMINATION OF AVERAGE POLLUTANT CONCENTRA-
TIONS 5-60
5.3.1.1 VARIABILITY OF POLLUTANT CONCENTRA-
TIONS 5-61
5.3.2 EFFECT OF LAND USE ON STORMWATER QUALITY 5-61
5.4 RECEIVING WATER 5-78
5.4.1 TRANSPORT PROPERTIES 5-78
5.4.1.1 CHANNEL GEOMETRY 5-80
5.4.1.2 DISPERSION COEFFICIENT ESTIMATES 5-83
5.4.2 REACTION OF POLLUTANTS 5-84
5.4.2.1 SEQUENTIAL REACTIONS 5-86
5.4.3 BACKGROUND RECEIVING WATER CONDITIONS 5-88
5.4.3.1 VARIABILITY OF BACKGROUND CONDITIONS 5-92
5.5 TREATMENT DEVICE PERFORMANCE 5-101
5.5.1 CONSIDERATIONS FOR VARIOUS POLLUTANTS 5-102
5.5.2 TREATMENT DEVICE PERFORMANCE EFFICIENCIES 5-103
5.5.2.1 SEDIMENTATION PERFORMANCE 5-103
5.5.2.2 DISSOLVED AIR FLOTATION PERFORMANCE 5-106
5.5.2.3 SWIRL CONCENTRATOR PERFORMANCE 5-106
5.5.2.4 HIGH RATE, DEEP BED MEDIA FILTRATION
PERFORMANCE 5-109
5.5.2.5 SCREENS AND MICROSCREENS 5-111
5.5.2.6 BIOLOGICAL TREATMENT 5-115
5.5.2.7 DISINFECTION 5-120
5.5.2.8 TREATMENT AT DRY WEATHER PLANTS 5-130
5.6 COST ESTIMATES FOR TREATMENT ALTERNATIVES 5-132
v
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
5 5.6.1 STRUCTURAL TREATMENT DEVICES 5-133
5.6.1.1 STORAGE BASINS 5-133
5.6.1.2 SEDIMENTATION 5-135
5.6.1.3 AIR FLOTATION 5-135
5.6.1.4 SWIRL CONCENTRATOR 5-135
5.6.1.5 HIGH RATE FILTRATION 5-142
5.6.1.6 STATIONARY SCREENS 5-142
5.6.1.7 HORIZONAL SCREENS 5-142
5.6.1.8 CHEMICAL COAGULATION 5-149
5.6.1.9 CHLORINATION FEED EQUIPMENT 5-149
5.6.1.10 HIGH INTENSITY MIXING/CHLORINE
CONTACT BASIN 5-158
5.6.1.11 RAW WASTEWATER PUMPING 5-158
5.6.1.12 SLUDGE PUMPING 5-158
5.6.1.13 BIOLOGICAL TREATMENT 5-158
5.6.1.14 EXAMPLE COST CALCULATION 5-165
5.6.2 MANAGEMENT PRACTICES 5-168
5.7 REFERENCES 5-168
6 MONITORING FOR STORMWATER ASSESSMENT 6- 1
6.1 PURPOSE OF MONITORING PROGRAM 6- 1
6.2 RAINFALL MONITORING 6- 2
6.2.1 USE OF RAINFALL DATA TO DEFINE SITE
CHARACTERISTICS 6- 2
6.2.2 USE OF RAINFALL DATA FOR PROJECTIONS AND
EVALUATION OF IMPACTS 6- 3
6.2.3 RAINGAGE DENSITY 6- 5
6.2.4 OTHER RAINGAGE CONSIDERATIONS 6- 9
6.3 SITE SELECTION AND DRAINAGE BASIN CHARACTERIZATION 6-10
6.4 MONITORING OF RUNOFF AND OVERFLOWS 6-12
vi
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
6 6.4.1 STUDY PERIOD AND SAMPLING FREQUENCY 6-12
6.4.1.1 SAMPLING INTERVAL WITHIN STORMS 6-13
6.4.2 SAMPLING PROCEDURE AND PARAMETERS 6-15
6.5 MONITORING OF RECEIVING WATER 6-17
6.5.1 CONTINUOUS MONITORS 6-18
6.6 REVIEW OF MONITORING LITERATURE 6-19
6.7 REFERENCES 6-20
7 EXAMPLE STORMWATER ANALYSES 7-1
7.1 SALT LAKE CITY, UTAH 7- 1
7.1.1 RAINFALL ANALYSIS 7- 3
7.1.2 DRAINAGE BASIN CHARACTERIZATION 7- 3
7.1.3 RUNOFF QUANTITY 7- 5
7.1.4 RUNOFF POLLUTANT LOADS 7-8
7.1.5 RECEIVING WATER RESPONSE 7- 9
7.1.6 STORMWATER CONTROL ALTERNATIVES 7-19
7.2 EXAMPLE STORMWATER ANALYSIS: KINGSTON, NEW YORK 7-27
7.2.1 RAINFALL ANALYSIS 7-27
7.2.2 DRAINAGE BASIN CHARACTERIZATION 7-30
7.2.3 DETERMINE RUNOFF VOLUMES 7-31
7.2.3.1 CAPTURE BY TREATMENT PLANT 7-32
7.2.4 DETERMINE STORMWATER LOADS 7-32
7.2.4.1 RECEIVING WATER RESPONSE 7-34
7.2.4.2 VARIABILITY IN THE HUDSON 7-36
7.2.4.3 OBSERVED WATER QUALITY 7-37
VII
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
7 7.2.5 SIMULATION OF FECAL COLIFORM RESPONSE 7-40
7.2.6 CONTROL ALTERNATIVES 7-47
7.3 REFERENCES 7-50
APPENDIX A - USERS MANUAL FOR BROAD SCALE SIMULATOR
APPENDIX B - PROGRAM LISTING FOR BROAD SCALE SIMULATOR
Vlll
-------
STORMWATER MANUAL
LIST OF TABLES
Table Number
Number Page
2- 1 CONTAMINANTS IN URBAN RUNOFF 2-4
2-2 SELECTED INSTREAM IMPACTS 2-4
2- 3 EXAMPLE BENEFICIAL WATER USE INVENTORY 2^15
3- 1 STATISTICS FOR STORM CHARACTERIZATION 3- 5
3- 2 MINNEAPOLIS RAINFALL ANALYSIS 3-13
3- 3 EXAMPLE STORMWATER LOADING TABLE FOR ADVECTIVE STREAM 3-27
3- 4 SUMMARY OF STEADY STATE SOLUTIONS FOR POLLUTANT
CONCENTRATIONS IN RECEIVING STREAMS 3-29
3- 5 STEADY STATE EQUATION FOR WASTE CONCENTRATIONS IN
TIDAL RIVERS AND ESTUARIES DUE TO POINT SOURCE 3-32
3- 6 CONCENTRATIONS IN LARGE, COMPLETELY MIXED IMPOUNDMENT 3-34
3- 7 GENERAL GUIDELINES FOR ESTIMATING MAGNITUDE OF FIRST
FLUSH EFFECT 3-56
3-8 SUMMARY OF RUNOFF STATISTICS DENVER, COLORADO,
RAINGAGE 05220, 1960 3-80
3- 9 SUMMARY OF RUNOFF DATA ANALYZED 3-80
3-10 RATIO OF WITHIN STORM FLOW VARIABILITY TO BETWEEN
STORM FLOW VARIABILITY 3-84
3-11 FLOW CONCENTRATION CORRELATION 3-84
3-12 MEASURES FOR REDUCING AND DELAYING URBAN STORM RUNOFF 3-89
4- 1 AVAILABLE COMPUTER MODELS (FROM AAPM, APPENDIX A(l)) 4- 3
4-2 ANALYTICAL SOLUTIONS USED IN BRWS 4-6
ix
-------
LIST OF TABLES
(Continued)
Table Number
Number Page
5- 1 CITIES SELECTED FOR RAINGAGE ANALYSIS 5- 2
5- 2 EFFECT OF ALTERNATIVE STORM DEFINITION RAINGAGE - 286026,
NEWARK AIRPORT, MAY 1948-DECEMBER 1975 (27.58 yr) 5-27
5- 3 (a) SUMMARY OF NEW YORK CITY RAINGAGE AGGREGATION ANALYSIS
PERIOD OF RECORD: 1/1/50-12/31/60 5-43
5- 3(b) RATIO OF AGGREGATED RAINGAGE STATISTICS TO POINT RAIN-
GAGE STATISTICS 5-44
5- 4 RUNOFF TO RAINFALL RATIOS FROM VARIOUS STUDIES 5-51
5- 5 COMPARISON OF QUALITY OF COMBINED SEWAGE FOR VARIOUS CITIES 5-64
5- 6 COMPARISON OF QUALITY OF STORM SEWER DISCHARGES FOR
VARIOUS CITIES 5-65
5- 7 POLLUTANT CONCENTRATIONS IN COMBINED SEWER OVERFLOWS 5-66
5- 8 POLLUTANT CONCENTRATIONS IN STORMWATER RUNOFF 5-67
5- 9 VARIABILITY OF RUNOFF AND OVERFLOW CONCENTRATIONS 5-68
5-10 POLLUTANT LOADING FACTORS FOR DESKTOP ASSESSMENT 5-75
5-11 SAMPLE U.S.G.S. SURFACE WATER RECORD DATA SHEET 5-82
5-12 RANGE OF VALUES OF REACTION COEFFICIENTS IN NATURAL
WATERS (48) 5-86
5-13 SUMMARY OF BACKGROUND CONCENTRATIONS FROM VIRGIN LAND 5-91
5-14 COMPARISON OF TREATMENT ALTERNATIVES 5-105
5-15 EXAMPLE CALCULATION OF STORMWATER TREATMENT COST 5-167
5-16 UNIT COSTS OF STREET SWEEPING 5-169
-------
LIST OF TABLES
(Continued)
Table Number
Number Page
7- 1 SUMMER RAINFALL STATISTICS (JULY-SEPTEMBER) SALT LAKE
CITY 7- 3
7-2 STORMWATER FLOWS AND LOADS GENERATED BY SUMMER STORMS
SALT LAKE CITY 7-11
7- 3 JORDAN RIVER GEOMETRY, TRANSPORT, AND REACTION RATES
DURING SUMMER STORMS 7-13
7- 4 WET WEATHER DISSOLVED OXYGEN RESPONSE OF JODRAN RIVER
TO SUMMER STORMS 7-17
7- 5 WET WEATHER DISSOLVED OXYGEN AFTER IMPLEMENTATION OF
MANAGEMENT PRACTICES 7-26
7- 6 HUDSON RIVER FECAL COLIFORM (MPN/100 ml) SUMMER 1976 7-38
7- 7 HUDSON RIVER FECAL COLIFORM (MPN/100 ml) 7-39
7- 8 INPUT DECK RECEIVING WATER SIMULATION 7-46
XI
-------
STORMWATER MANUAL
LIST OF FIGURES
Figure Number
Number Page
2-1 TYPICAL ANALYSIS OF WET WEATHER DISSOLVED OXYGEN 2-7
2- 2 PROBABILISTIC ANALYSIS OF WET WEATHER DISSOLVED OXYGEN 2- 9
2- 3 MINIMUM DISSOLVED OXYGEN FREQUENCY- CURVES FOR EXISTING
CONDITIONS IN THE DES MOINES RIVER 2-10
2- 4 TIME SCALES STORM RUNOFF WATER QUALITY PROBLEMS 2-13
2- 5 SPACE SCALES STORM RUNOFF WATER QUALITY PROBLEMS 2-14
3- 1 REPRESENTATION OF STORM RUNOFF PROCESS 3- 4
3- 2 (a) CUMULATIVE DISTRIBUTION FUNCTION FOR GAMMA DISTRIBUTION 3- 7
3- 2(b) CUMULATIVE DISTRIBUTION FUNCTION FOR GAMMA DISTRIBUTION 3- 8
3- 3 STORM EVENT AND LONG TERM FREQUENCY DISTRIBUTIONS 3- 9
3- 4 MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
CENTRAL PARK STATION 305801 3-15
3- 5 IDEALIZATION OF FIRST FLUSH EFFECT 3-22
3- 6 CORRECTION IN ESTIMATE OF Mn WHEN FIRST FLUSH IS PRESENT 3-24
K
3- 7 COMPARISON OF SPATIAL DETAIL IN STORMWATER LOADING
CHARACTERIZATION 3-31
3- 8 GRAPHICAL SOLUTION TO THE DILLON APPROACH 3-35
3- 9 WATER QUALITY RESPONSE SIMULATOR BOD-DISSOLVED OXYGEN
EXAMPLE 3-38
3-10 STREAM RESPONSE CHARACTERISTICS TO PULSE LOADS 3-39
3-11 EFFECT OF DISPERSION ON POLLUTANT CONCENTRATION AT MID-
POINT OF STORM PULSE 3-41
xii
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
3-12 GRAPHICAL SOLUTION TO K BESSEL FUNCTION 3-44
o
3-13 COMPARISON OF THEORETICAL AND SIMULATED ESTUARY RESPONSE
TO STORMLOADS 3-45
3-14 REPRESENTATION OF STORM RUNOFF PROCESS, INTERCEPTION AND
STORAGE 3-47
3-15 DETERMINATION OF LONG TERM INTERCEPTOR PERFORMANCE 3-49
3-16 DERIVATION OF SOLUTION FOR EFFECTIVE TANK VOLUME 3-51
3-17 EFFECT OF PREVIOUS STORMS ON LONG TERM EFFECTIVE STORAGE
CAPACITY 3-52
3-18 DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE 3-54
3-19 IMPROVEMENT IN LONG TERM STORAGE DEVICE PERFORMANCE DUE
TO FIRST FLUSH EFFECT 3-55
3-20 DISSOLVED AIR FLOTATION PERFORMANCE RELATED TO DESIGN RATE 3-61
3-21 IDEALIZED REMOVAL EFFICIENCY CURVE FOR A FLOW SENSITIVE
TREATMENT DEVICE 3-62
3-22 LONG TERM PERFORMANCE OF A FLOW SENSITIVE TREATMENT DEVICE 3-64
3-23 COMPARISON OF IDEALIZED AND ACTUAL DISSOLVED AIR FLOTATION
PERFORMANCE 3-66
3-24 EFFECT OF STORMFLOW VARIATION ON PERFORMANCE OF EMPIRICAL
DISINFECTION DEVICE 3-68
3-25 FREQUENCY DISTRIBUTION OF STORM OVERFLOW LOADING RATES
BEFORE AND AFTER INTERCEPTION 3-71
3-26 FREQUENCY DISTRIBUTION OF STORM LOADS BEFORE AND AFTER
STORAGE 3-73
Xlll
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
3-27 FREQUENCY DISTRIBUTION OF STORM LOADING RATES BEFORE
AND AFTER IN-LINE TREATMENT 3-74
3-28 FREQUENCY DISTRIBUTION OF STORM LOADING RATES BEFORE
AND AFTER DISINFECTION 3-75
3-29 EFFECT OF PREVIOUS STORMS ON LONG TERM EFFECTIVE STORAGE
CAPACITY 3-77
3-30 DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE 3-79
3-31 STORAGE/INTERCEPTION ISOQUANTS PERCENT BOD CAPTURED WITH
FIRST FLUSH DENVER, 1960 RAINGAGE 052220 3-81
3-32 COMPARISON OF SIMULATED FLOW SENSITIVE DEVICE WITH
THEORETICAL CURVES 3-83
3-33 LONG-TERM EFFECTIVENESS OF STREET SWEEPING 3-87
4- 1 SCHEMATIC OF BROADSCALE RECEIVING WATER SIMULATOR
OPERATION 4-4
5- 1 DISTRIBUTION OF PRECIPITATION AND LOCATION STUDY CITIES 5- 3
5- 2(a) ANNUAL PRECIPITATION 5- 4
5- 2(b) ANNUAL PRECIPITATION 5- 5
5- 2(c) ANNUAL PRECIPITATION 5-6
5- 2(d) ANNUAL PRECIPITATION 5- 7
5- 3(a) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
CARIBOU, MAINE STATION 17115, (1948-1973) 5- 9
5- 3(b) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
BOSTON, MASSACHUSETTS STATION 190770, (1948-1973) 5-10
XIV
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
5- 3(c) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
COLUMBIA, SOUTH CAROLINA STATION 381939, (1948-50,54-73) 5-11
5- 3(d) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
TAMPA, FLORIDA STATION 088788, (1948-51, 59-73) 5-12
5- 3(e) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
DETROIT, MICHIGAN STATION 202103, (1960-1973) 5-13
5- 3(f) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
DALLAS, TEXAS STATION 412244, (1941-46,48-73) 5-14
5- 3(g) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
DENVER, COLORADO STATION 05220, (1948-1973) 5-15
5- 3(h) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
SALT LAKE CITY, UTAH STATION 427598, (1948-1973) 5-16
5- 3(i) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
PHEONIX, ARIZONA STATION 026481, (1948-1973) 5-17
5- 3(j) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
OAKLAND, CALIFORNIA STATION 046335 (1948-1973) 5-18
5- 3(k) MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
PORTLAND, OREGON STATION 365751, (1948-1973) 5-19
5- 4 LAG-K AUTOCORRELATION FUNCTION OF DES MOINES, IOWA
HOURLY RAINFALL, 1968 5-21
5- 5(a) COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE
DISTRIBUTION OF TIME BETWEEN STORMS NEWARK AIRPORT,
RAINGAGE 286026, 1948-1975 5-23
5- 5(b) COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE
DISTRIBUTION OF TIME BETWEEN STORMS NEWARK AIRPORT,
RAINGAGE 286026, 1948-1975 5-24
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
5- 5(c) COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE
DISTRIBUTION OF TIME BETWEEN STORMS NEWARK AIRPORT,
RAINGAGE 286026, 1948-1975 5-25
5- 6 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE
DISTRIBUTION OF TIME BETWEEN STORMS MINNEAPOLIS/ST.
PAUL AIRPORT, RAINGAGE 215435, 1948-1973 5-26
5- 7 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE
DISTRIBUTION OF DURATION CENTRAL PARK, NEW YORK, RAIN-
GAGE 305801, 1948-1973 5-29
5-8 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE DIS-
TRIBUTION OF INTENSITY CENTRAL PARK, NEW YORK, RAINGAGE
305801, 1948-1975 5-30
5- 9 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE DIS-
TRIBUTION OF UNIT VOLUME CENTRAL PARK, NEW YORK, RAINGAGE
305801, 1948-1975 5-31
5-10 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE DIS-
TRIBUTION OF TIME BETWEEN STORMS CENTRAL PARK, NEW YORK,
RAINGAGE 305801, 1948-1975 5-32
5-11 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE DIS-
TRIBUTION OF DURATION MINNEAPOLIS/ST. PAUL AIRPORT,
RAINGAGE 215435, 1948-1975 5-33
5-12 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE
DISTRIBUTION OF INTENSITY MINNEAPOLIS/ST. PAUL AIRPORT,
RAINGAGE 215435, 1948-1975 5-34
5-13 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE DIS-
TRIBUTION OF UNIT VOLUME MINNEAPOLIS/ST. PAUL AIRPORT,
RAINGAGE 215435, 1948-1975 5-35
5-14 COMPARISON OF OBSERVED AND THEORETICAL CUMULATIVE DIS-
TRIBUTION OF TIME BETWEEN STORMS MINNEAPOLIS/ST. PAUL
AIRPORT, RAINGAGE 215435, 1948-1973 5-36
xv i
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
5-15 CORRELATION OF MONTHLY RAINFALL IN THE NEW YORK CITY AREA 5-38
5-16 METHOD FOR RAINGAGE RECORD AGGREGATION 5-40
5-17 RAINGAGE AGGREGATION LOCATIONS 5-41
5-18 EFFECT OF RAINGAGE AGGREGATION ON STATISTICAL EVENT
PROPERTIES 5-42
5-19 DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE 5-48
5-20 RELATIONSHIP BETWEEN IMPERVIOUS AREA AND RUNOFF-TO-
RAINFALL RATIO 5-49
5-21 IMPERVIOUSNESS AS A FUNCTION OF DEVELOPED POPULATION
DENSITY 5-53
5-22 UNIT HYDROGRAPH DEFINITION SKETCH 5-55
5-23 ESTIMATE OF MEAN DURATION OF RAINFALL EXCESS 5-56
5-24 WIDTH AT 25 PERCENT OF PEAK VERSUS AREA-DESIGN CURVES 5-59
5-25 REPRESENTATIVE STORMWATER DISCHARGE QUALITITIES 5-62
5-26 TIME WEIGHTED NORMALIZATION 5-63
5-27 TIME SINCE LAST CLEANING VERSUS SOLIDS LOADING 5-70
5-28 NORMALIZED BOD LOADINGS VERSUS DEVELOPED POPULATION
DENSITY 5-73
5-29 EFFECT OF STREET SWEEPING FREQUENCY ON ANNUAL BOD
CONTRATION IN URBAN STORMWATER RUNOFF 5-74
5-30 NORMAL DISTRIBUTION OF SURFACE WATER RUNOFF 5-79
5-31 AVERAGE RUNOFF YIELDS IN THE COTERMINOUS UNITED STATES 5-81
XVll
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
5-32 DISPERSION COEFFICIENT FOR DIFFERENT TIDAL RIVERS AND
ESTUARIES 5-85
5-33 DEOXYGENATION COEFFICIENT (Kd) AS A FUNCTION OF DEPTH 5-87
5-34 TRANSFER COEFFICIENT (K ) AS A FUNCTION OF DEPTH 5-89
LJ
5-35 SATURATION-TEMPERATURE-CHLORIDE RELATIONSHIPS 5-90
5-36 RELATIONSHIP BETWEEN TOTAL PHOSPHORUS AND FLOW FOR
GENESEE RIVER (1968-1974) 5-93
5-37 RELATIONSHIP BETWEEN ORGANIC NITROGEN AND FLOW FOR
GENESEE RIVER (1968-1974) 5-94
5-38 RELATIONSHIP BETWEEN AMMONIA AND FLOW FOR GENESEE RIVER
(1968-1974) 5-95
5-39 RELATIONSHIP BETWEEN NITRATE AND NITRITE AND FLOW FOR
GENESEE RIVER (1968-1974) 5-96
5-40 RELATIONSHIP BETWEEN TOTAL PHOSPHORUS AND FLOW FOR
TRENT RIVER (1967-1973) 5-97
5-41 RELATIONSHIP BETWEEN ORGANIC NITROGEN CONCENTRATION AND
FLOW FOR TRENT RIVER (1967-1972) 5-98
5-42 RELATIONSHIP BETWEEN AMMONIA AND FLOW FOR TRENT RIVER
(1967-1972) 5-99
5-43 RELATIONSHIP BETWEEN NITRATE AND NITRITE AND FLOW FOR
TRENT RIVER (1967-1972) 5-100
5-44 SEDIMENTATION TANK PERFORMANCE 5-104
5-45 AIR FLOTATION PERFORMANCE 5-107
5-46 SWIRL CONCENTRATOR PERFORMANCE 5-110
xvi 11
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
5-47 HIGH RATE FILTRATION PERFORMANCE 5-112
5-48 PERFORMANCE OF SCREENS 5-114
5-49 TRICKLING FILTER PERFORMANCE 5-116
5-50 TRICKLING FILTER PLANT PERFORMANCE 5-117
5-51 ROTATING BIOLOGICAL CONTACTORS PERFORMANCE 5-119
5-52 TOP VIEW OF DISINFECTION DEVICE DESIGNED TOWARDS PLUG
FLOW CONDITIONS 5-121
5-53 DISINFECTION RATING CURVE FOR EMPIRICAL DEVICE WITH FOUR
ORDER REDUCTION AT MEAN RUNOFF FLOW 5-125
5-54 EFFECT OF STORMFLOW VARIATION ON PERFORMANCE OF EMPIRICAL
DISINFECTION DEVICE 5-127
5-55 COMPARISON OF DISINFECTION RATING CURVES FOR ALTERNATIVE
MODELS 5-129
5-56 CAPITAL COST - STORAGE BASINS (JUNE 1975 COSTS) 5-134
5-57 ANNUAL OPERATION AND MAINTENANCE COST - STORAGE BASINS
(JUNE 1975 COSTS) 5-136
5-58 CAPITAL COST - SEDIMENTATION BASINS (JUNE 1975 COSTS) 5-137
5-59 ANNUAL OPERATION AND MAINTENANCE COST - SEDIMENTATION
BASINS (JUNE 1975 COSTS) 5-138
5-60 CAPITAL COST - AIR FLOTATION (JUNE 1975 COSTS) 5-139
5-61 ANNUAL OPERATION AND MAINTENANCE COST - AIR FLOTATION
(JUNE 1975 COSTS) 5-140
5-62 CAPITAL COST - SWIRL CONCENTRATOR (JUNE 1975 COSTS) 5-141
XVIX
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
5-63 ANNUAL OPERATION AND MAINTENANCE COST - SWIRL CONCEN-
TRATOR (JUNE 1975 COSTS) 5-143
5-64 CAPITAL COST HIGH RATE FILTRATION CONCRETE GRAVITY FILTERS 5-144
5-65 ANNUAL OPERATION AND MAINTENANCE COST - HIGH RATE
FILTRATION (JUNE 1975 COSTS) 5-145
5-66 CAPITAL COST - STATIONARY SCREEN (JUNE 1975 COSTS) 5-146
5-67 ANNUAL OPERATION AND MAINTENANCE COST - STATIONARY
SCREEN (JUNE 1975 COST) 5-147
5-68 CAPITAL COST - HORIZONTAL SCREENS (JUNE 1975 COSTS) 5-148
5-69 ANNUAL OPERATION AND MAINTENANCE COST HORIZONTAL SCREENS
(JUNE 1975 COSTS) 5-150
5-70 CAPITAL COST - CHEMICAL FEED SYSTEMS (JUNE 1975 COSTS) 5-151
5-71 ANNUAL OPERATION AND MAINTENANCE COST - LIME FEED
(JUNE 1975 COSTS) 5-152
5-72 ANNUAL OPERATION AND MAINTENANCE COST - ALUM FEED
(JUNE 1975 COSTS) 5-153
5-73 ANNUAL OPERATION AND MAINTENANCE COST - FERRIC CHLORIDE
FEED (JUNE 1975 COSTS) 5-154
5-74 ANNUAL OPERATION AND MAINTENANCE COST POLYMER FEED 5-155
5-75 CAPITAL COST - CHLORINE FEED EQUIPMENT (JUNE 1975 COSTS) 5-156
5-76 ANNUAL OPERATION AND MAINTENANCE COST CHLORINE FEED
EQUIPMENT (JUNE 1975 COST) 5-157
5-77 CAPITAL COST - RAPID MIX BASIN (JUNE 1975 COSTS) 5-159
xx
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
5-78 ANNUAL OPERATION AND MAINTENANCE COST - RAPID MIX BASIN
(JUNE 1975 COSTS) 5-160
5-79 CAPITAL COST - RAW WASTE WATER PUMPING (JUNE 1975 COSTS) 5-161
5-80 ANNUAL OPERATION AND MAINTENANCE COST RAW WASTE WATER
PUMPING (JUNE 1975 COSTS) 5-162
5-81 CAPITAL COST - SLUDGE PUMPING (JUNE 1975 COSTS) 5-163
5-82 ANNUAL OPERATION AND MAINTENANCE COST - SLUDGE PUMPING
(JUNE 1975 COSTS) 5-164
5-83 CAPITAL COST - SECONDARY BIOLOGICAL TREATMENT (JUNE 1975
COSTS) 5-166
6- 1 EFFECT OF AREAL DISTRIBUTION OF RAINFALL ON REAL VS.
APPARENT RATIO OF RUNOFF TO RAINFALL 6-4
6- 2 ESTIMATE OF RAINFALL SAMPLING ERROR IN 400 MI2 ILLINOIS
AREA 6-6
6- 3 EFFECT OF RAINGAGE DENSITY ON RAINFALL ESTIMATES 6- 8
6- 4 ERROR IN ESTIMATE OF AVERAGE VERSUS NUMBER OF STORMS
MONITORED 6-14
7- 1 SALT LAKE CITY STUDY AREA 7- 2
7- 2 MONTHLY STATISTICAL RAINFALL CHARACTERIZATION, SALT
LAKE CITY, COMBINED RECORD STATIONS 427598 AND 427603 7- 4
7- 3 ESTIMATE OF MEAN DURATION OF RAINFALL EXCESS 7-6
7- 4 WIDTH AT 25 PERCENT OF PEAK VERSUS AREA-DESIGN CURVES 7- 7
7- 5 CUMULATIVE FREQUENCY FUNCTION OF RUNOFF RATES SALT LAKE
CITY SUMMER STORMS 7-10
xxi
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
7- 6 EFFECT OF DISPERSION ON PULLUTANT CONCENTRATION AT
MIDPOINT OF STORM PULSE 7-14
7- 7 MINIMUM DISSOLVED OXYGEN AT MILEPOINT 15 OF JORDAN RIVER
FOLLOWING SUMMER STORMS 7-18
7- 8 EFFECT OF PREVIOUS STORMS ON LONG TERM EFFECTIVE STORAGE
CAPACITY 7-20
7- 9 DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE 7-21
7-10 IMPROVEMENT IN LONG TERM STORAGE DEVICE PERFORMANCE DUE
TO FIRST FLUSH EFFECT 7-22
7-11 CUMULATIVE FREQUENCY FUNCTION OF RUNOFF VOLUMES
SALT LAKE CITY SUMMER STORMS 7-23
7-12 CAPITAL COST - STORAGE BASINS (JUNE 1975 COSTS! 7-24
7-13 ANNUAL OPERATION AND MAINTENANCE COST - STORAGE BASINS
(JUNE 1975 COSTS) 7-25
7-14 MAJOR FEATURES OF KINGSTON STUDY AREA 7-28
7-15 MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
KINGSTON, NEW YORK STATION 304424 7-29
7-16 DETERMINATION OF LONG TERM INTERCEPTOR PERFORMANCE 7-33
7-17(a) SIMULATOR OUTPUT SPECIFICATION OF INPUT PARAMETERS 7-41
7-17(b) SIMULATION OUTPUT TIME HISTORY PLOTS 7-42
7-17(c) SIMULATOR OUTPUT RESPONSE CONCENTRATIONS 7-43
7-17(d) SIMULATOR OUTPUT LOCATION SUMMARY - KINGSTON POINT 7-44
7-17(e) SIMULATOR OUTPUT LOCATION SUMMARY - PORT EWEK 7-45
xxi i
-------
LIST OF FIGURES
(Continued)
Figure Number
Number Page
7-18 FREQUENCY DISTRIBUTION OF SIMULATED HUDSON RIVER FECAL
COLIFORM RESPONSE (SUMMER, 1965) 7-48
7-19 CAPITAL COST - CHLORINE FEED EQUIPMENT (JUNE 1975 COSTS) 7-49
7-20 ANNUAL OPERATION AND MAINTENANCE COST CHLORINE FEED
EQUIPMENT (JUNE 1975 COST) 7-51
7-21 CAPITAL COST - RAPID MIX BASIN (JUNE 1975 COSTS) 7-52
7-22 ANNUAL OPERATION AND MAINTENANCE COST - RAPID MIX BASIN
(JUNE 1975 COSTS) 7-53
xxi 11
-------
ACKNOWLEDGEMENT
This manual was prepared by Hydroscience, Inc., Westwood, New Jersey,
and is the product of contributions by a number of individuals.
The project management and direction was provided by Eugene D. Driscoll,
who also participated in establishing the scope and extent to which areas
of concern were addressed. The conceptual approach and basic theoretical
development were provided by Dominic M. DiToro. These have been under de-
velopment for a period of ten years, with the initial ideas developed dur-
ing conversations with Robert V. Thomann. Mitchell Small provided support
in the theoretical development, made significant contributions in this area,
and is also responsible for a major share of the effort in preparation of
the report text.
Significant contributions were made by Eugene D. Donovan, Jr., Joseph
Cleary, and Tzu-shiung Hsu, in assembling cost and performance information
on control devices, in a format compatible with the statistical methodology.
Thomas Gallagher, James Fitzpatrick, and Daniel Szumski provided a similar
contribution on receiving water quality aspects. Richard Sheridan assisted
in developing the structure and organization of the manual and provided a
major share of the effort in developing and presenting material in Chapter
5.
Valuable insight, advice, and guidance were provided by Donald J.
O'Connor and Robert V. Thomann.
Important contributions by the following Environmental Protection Agency
personnel are appreciated, and hereby acknowledged. Dennis Athayde (N.P.S.
Branch, Washington, D.C.) recognized the potential value of the type ap-
proach described in this manual, and supported the concept of a manual to
develop and disseminate the methodology. He provided valuable guidance
throughout the program on the scope and the emphasis adopted. Richard Field
(Combined Sewer Section, Edison, New Jersey) provided valuable advice and
guidance on performance of control measures.
-------
CHAPTER 1
INTRODUCTION
This manual describes a simplified methodology which can be used to
assess the impact of urban stormloads on the quality of receiving waters, and
to evaluate the cost and effectiveness of control measures for reducing these
pollutant loads. The methodology is particularly appropriate for use at the
planning level where preliminary assessments are made to define problems,
establish the relative significance of contributing sources, assess feasibil-
ity of control, and determine the need for and focus of additional evalua-
tions. It can also be used effectively in conjunction with detailed studies,
by providing a cost-effective screening of an array of alternatives, so that
the more detailed and sophisticated techniques can examine only the more
attractive alternatives.
The methodology is based on the determination of certain statistical
properties of the rainfall history of an area. From these statistics, the
desired information on loads, performance of controls, and receiving water
impacts is generated directly. Procedures are quite simple to apply, using
charts and graphs which facilitate screening alternate types or levels of
control, testing sensitity to assumptions concerning drainage area character-
istics, stormwater contaminant levels and similar variable factors.
The theoretical basis for the methodology is presented, although the
reader need not be familiar with statistical theory or procedures to utilize
it effectively. The user need not read the manual from cover to cover and
understand and apply each part of it in a rigorous sequence, in order to
benefit from it. While separate chapters are mutually supporting, each
essentially stands on its own for the particular aspects which are addressed.
Chapter 2 - Presents an overview of the urban stormwater problem,
and a perspective in which water quality problems caused
by storm loads can be considered.
Chapter 3 - Presents a description of the statistical methodology,
its theoretical basis, and its application for character-
izing storm loads, receiving water impacts and the per-
formance of selected control measures.
Chapter 4 - Presents a description of a simulation model which
calculates receiving water impacts for streams and
estuaries. It is designed to operate on input consisting
1-1
-------
of time variable storm loads as generated by various load
generating simulators (e.g. STORM). As an alternative to
the statistical calculations, it will provide the user
with both the time history of receiving water concentra-
tions and their statistics for the period analyzed, at a
number of receiving water locations.
Chapter 5 - Presents a condensed summary and analysis of available
information and data on numerical estimates of para-
meters required for performing a stormwater impact
analysis. The information is of value for analyses
utilizing simulation methods as well as for application
of the statistical procedures described in this manual.
Chapter 6 - Presents considerations for the design of effective
monitoring programs, applicable for either preliminary
assessments or more intensive programs.
Chapter 7 - Provides examples which illustrate the applications of
the methodology to specific problem settings.
Bibliography - An extensive bibliography is provided which can direct
the reader to additional sources of information on
aspects which he may wish to pursue in greater depth.
Although the statistical methodology presented is essentially designed
to be an estimating technique, and of maximum applicability and value in
preliminary assessments performed at the planning level, it provides results
quite similar to those generated by simulation techniques (such as STORM)
when similar basic data inputs are used, and comparable levels of spatial
and/or temporal definition are employed. The validity of calculations per-
formed by the statistical methodology has been "established" by comparisons
with simulation results both as reported in this manual, and by others.
The statistical methodology is not proposed as a substitute for simu-
lation techniques, other than for preliminary assessments. For preparation
of formal facility plans and especially for final design, the optimum dis-
tribution, location and individual size of controls will normally require a
definition of spatial and temporal detail which can be more effectively
provided by detailed simulation techniques.
1-2.
-------
CHAPTER 2
STORMWATER RUNOFF - A REVIEW OF THE PROBLEM
The most fundamental issue in urban stormwater control studies is
determining the degree of control which is justifiable from the standpoint
of benefits returned on investment. While this basic issue is quite clear,
the best methods for addressing it within the context of a planning study
are the subject of continuing debate. Numerous analytical approaches for
evaluating urban stormwater needs have been developed and applied (1,2,3).
Most of these tend to be strongly oriented toward load estimation, treatment
performance, or cost estimation, and useful assessment procedures for these
problems have been presented. However, the other side of the equation, that
of evaluating benefits, has generally been approached with subdued enthusiasm.
This is principally due to the difficulty in quantifying the long term water
quality improvements associated with urban stormwater control.
The principal purpose of this chapter is to develop a framework within
which stormwater control requirements can be evaluated from the standpoint
of water quality improvements. This perspective is essential to the planning
study, since stormwater control is a questionable investment if it does not
result in long term enhancements to legitimate water uses.
From this perspective, stormwater control studies must begin with an
evaluation of water quality problems. What are the existing problems in an
area? To what extent does urban stormwater runoff contribute to identifiable
problems? Is stormwater treatment a reasonable alternative to effectively
control existing water quality problems?
Once the potential benefits of stormwater control have been defined,
the planning process can proceed to more specific planning questions such as:
How much treatment is enough? What are the benefits of alternative levels of
treatment? Are there particularly effective treatment devices or other
controls which should be considered? What are the costs of achieving alter-
native water quality objectives?
Chapter 2 will explore problems and opportunities in evaluating the
needs for urban stormwater control. The principal water quality problems
in urban areas will be discussed in terms of the resulting limitation of
beneficial water uses. Various analysis methods for evaluating stormwater
problems and control requirements will also be discussed. A more detailed
presentation of the analysis methods required for effective planning will
be made in subsequent chapters.
2-1
-------
2.1 Major Water Quality Problems in Urban Areas
The nature of water quality problems associated with the receiving
waters in and around metropolitan areas is quite varied. Although a list of
common potential problems can be drawn up readily, local factors have a
predominant influence in determining both the class of problem, its severity,
and the specific source or sources which are most critical. The local
factors which affect water quality include climate, geography, population,
population concentration, the nature and degree of industrialization in the
area, the receiving water system, its nature, size and hydrology; and the
nature of the surrounding area - both upstream and downstream of the urban
area itself.
Water quality problems, either current or potential, will be generated
by waste loads which enter the receiving water from the area in question.
There are a number of different types of sources which contribute waste
loads, including discharges of domestic sewage, either treated or untreated,
industrial waste discharges, storm runoff from urban, agricultural, or un-
developed land areas, surface returns from irrigated agriculture, sub-
surface seepage of groundwater, and leachate from land disposal sites.
Waste loads are generally classified according to their temporal
variability as either continuous or intermittent, and according to their
spatial extent as either point or non-point (distributed) sources. Contin-
uous sources, such as municipal sewage treatment plants or industrial
facilities, produce a relatively constant load over time, although daily or
seasonal variations are often present. Intermittent loads, often associated
with wet weather conditions, occur infrequently and at random intervals.
Some types of pollutant loads, such as those due to groundwater seepage, may
occur with an intermediate degree of temporal variability.
The classification of waste loads as point or distributed sources is
also somewhat ambiguous in certain cases. Storm runoff, for example, is
essentially a non-point source in that it is generated by precipitation
falling over a wide area. In terms of the actual load to the receiving
water, however, the point source classification may be more appropriate for
urban areas where storm runoff is collected and enters the stream at specific
locations. In developing a solution oriented approach to water quality
problems, storm water loads may be treated as either intermittent point
sources, or as distributed loads in order to fit the simplest and most
effective approach to analysis.
Each of the individual source types has distinctive characteristics.
The type of pollutants which predominate can differ radically between
sources, as can the absolute quantity of pollutant. Further, controls
which may be applied, with either treatment facilities or management
practices, usually modify both the total quantity of pollutant in the source
and the predominant type present. The method of analyzing the effect of all
sources on water quality, and the impact of alternative control measures
must be able to handle these variations effectively.
2-2
-------
2.2 Urban Runoff
Human activity as well as natural processes result in conditions which
cause contaminants to be mixed with stormwater. Automobiles cause oils and
other hydrocarbons as well as certain heavy metals to accumulate on streets
and parking lots. Lawn fertilizers and pesticides are applied in a manner
which makes them suseptible to erosive processes. Pets contribute to
organic pollutant buildups in the urban environment. Construction activity
often leaves unconsolidated soils exposed to the elements. Natural processes
such as the decomposition of animal and vegetative materials and wind erosion
similarly cause potential pollutants to buildup on the land surface where
they are susceptable to transport by storm runoff.
Urban runoff may reach receiving waters via storm sewers, overland
runoff, or combined sewer overflows. The quantity of storm related pollut-
ants entering a receiving water is largely influenced by which of these
conveyance systems are present.
Storm sewers are designed to reduce ponding and flooding problems in a
drainage area by conveying runoff away from the area into the receiving
stream. Sources of the runoff flows include street runoff, roof drains,
drainage from large paved areas (such as parking lots and industrial com-
plexes), and runoff from parks and vacant lands. The quantity of storm sewer
flow is influenced by drainage basin characteristics such as the percent
impervious area, soil types, and land slope, which will be discussed in more
detail in Chapters 3 and 5.
Overland runoff has essentially the same quality and flow character-
istics as stormwater discharges with one exception; the flow and load enter
receiving waters as distributed sources rather than at discrete stormwater
discharge points.
Combined sewer systems convey both dry weather sewage flows and storm-
water runoff, normally to a wastewater treatment plant. The systems are
designed to accommodate a design flow which is periodically exceeded. When
a storm runoff volume exceeds the design capacity of the combined sewer
system, the excess flow, a mixture of raw sewage and stormwater, is bypassed
to nearby receiving waters. The quantity of flow bypassed is a function of
interceptor capacity and regulator operational procedures.
Combined sewer overflow is generally higher than storm runoff in its
concentration of most pollutants, due to the higher concentration of these
materials (BOD, nutrients, bacteria, etc,) in raw sewage. This generaliza-
tion should, however, be borne out by site specific measurements since
certain materials, such as suspended solids, are frequently higher in storm
runoff than in domestic sewage. Monitoring programs for making evaluations
of this type are discussed in Chapter 6.
2.3 Contaminants in Urban Runoff
The previous discussion of urban runoff sources describes some of the
factors which influence the types and concentrations of contaminants which
2-3
-------
are present in storm related loads from an urban area. Materials transported
by runoff constitute a problem when their discharge into receiving streams
cause violations of water quality standards or limit legitimate water use.
The definition of a problem may be either quantitative, as in the case of
comparing measured or computed water quality with standards, or as is often
the case, a qualitative description. This later type of problem definition
often focuses on factors such as aesthetic concerns (floatables, turbidity,
or surface slicks) or on quantitative measurements for which federal, state
and local standards do not exist.
The contaminants in urban runoff generally fall into the seven categor-
ies itemized in Table 2-1.
TABLE 2-1 - CONTAMINANTS IN URBAN RUNOFF
1. Floatables and visual contaminants
2. Degradable organics
3. Suspended solids
4. Nutrients
5. Bacteria, virus
6. Toxicants
7. Dissolved solids
Each of these contaminant categories can, when present in sufficient
quantities, contribute to water quality problems. Although stormwater re-
lated loads generally contain measurable amounts of materials in all seven
classes, the total load may or may not constitute a water quality problem
depending on the magnitude of the instream impact to the load. Table 2-2
lists some typical classes of water quality impacts which can result from
urban runoff loads.
TABLE 2-2 - SELECTED INSTREAM IMPACTS
1. Aesthetic deterioration
2. Dissolved oxygen depression
3. Sediment deposition
4. Excessive algal growth
5. Public health threats
6. Impaired recreational value
7. Ecological damage
8. Reduced commercial value
The relationships between the contaminants generally found in urban
runoff and the resulting instream impacts are discussed in the following
paragraphs.
1. Floatables and Visual Contaminants - Aesthetic deterioration is caused
either by the general appearance of water bodies (dirty, turbid, cloudy), or
the actual presence of specific objectionable conditions, including odors,
floating debris or films, scums or slimes, etc. These conditions may make
the receiving water unattractive or repugnant to those in its proximity.
Ecological problems might also result in fish, water fowl, or lower levels
2-4
-------
of the food chain due to impairments of physiological functions.
2. Degradables Organics - Degradable organic materials stimulate the growth
of bacteria which may consume oxygen more rapidly than it can be replenished
by natural reaeration processes. This condition may or may not be visually
apparent. In its extreme stages, excessive oxygen depressions can cause dis-
coloration and the formation of gas and odors. However, before this extreme
is reached, the environmental stress may be sufficient to cause respiratory
damage to fish and other lower aquatic organisms. Species diversity shifts
may result if conditions prevail for more than a few days. Similar condi-
tions can occur due to the presence of reduced organic and ammonia forms of
nitrogen which utilize oxygen as they are stabilized.
3. Suspended Solids - Particulate matter may contribute to a variety of
problems, such as objectionable aesthetic conditions and the formation of
sediment deposits which smother bottom dwelling aquatic organisms, impede
navigation, and restrict river flows, thus increasing flooding potential.
Organic sediment deposits can also react to form a benthal oxygen demand.
4. Nutrients - The discharge of materials which fertilize or stimulate
excessive or undesirable forms of aquatic growth can create significant
problems in some receiving water systems, particularly lakes and impound-
ments. Overstimulation of aquatic weeds or algae (eutrophication) can be
aesthetically objectionable, cause dissolved oxygen problems, and in extreme
cases, interfere with commercial and recreational uses by impeding small boat
navigation, creating odors, and heavy mats of floating material at shore-
lines.
5. Bacteria/Virus Concentrations - The presence of excessive concentrations
of objectionable microorganisms can impair the ability to utilize the re-
ceiving water for water supply, recreational purposes, or shell-fish harvest-
ing. Excessive bacteria concentrations are generally taken as an indication
of a potential public health problem.
6. Toxicants - Toxicity problems can fall into either of two categories:
chronic bioinhibition or acute toxicity. Chronic effects may be exhibited
by relatively low concentrations of metals, pesticides or persistent
organics which tend to accumulate in the tissue of aquatic organisms over
long periods of time. Their effects can be manifested at all levels of the
food chain and may occur in areas well removed from the point of discharge.
In excessive concentrations these materials, and others such as ammonia
nitrogen and effluent chlorine bi-products, can exhibit acute toxic impacts
in a local area surrounding a discharge. These effects are typified by
fish kills or shifts in biological diversity.
7. Dissolved Solids - A number of beneficial uses can be impaired by ex-
cessive concentrations of dissolved solids. Both domestic and industrial
water uses are sensitive to dissolved solids concentrations. Irrigated
agricultural is quite sensitive to the salt content of the applied water.
On a practical day to day basis, fanners must compensate for high salinity
in irrigation water by increasing the quantity of irrigation flow. This
imposes additional demands on water development and conveyance programs, and
2-5
-------
further contributes to drainage problems. Effective agriculture can be
destroyed if irrigation rates required to compensate for high salinity exceed
the percolation and drainage capacity of the soil.
2.4 Water Quality Problem Definition
The previous section summaries the potential problems which can result
from a variety of storm related water impacts. The severity of each problem
is defined by the degree to which water use interferences occur. From this
perspective, "the stormwater problem" can be defined only in terms of
specific water quality impacts in a study area. Site specific water quality
problems are related to specific contaminants, often originating from
identifiable sources. Each contaminant, and possibly each of its sources,
will respond differently to control measures. Therefore, effective control
practices must concentrate on reducing the specific contaminants which con-
tribute to identified water quality problems, and not those which have neg-
ligible or insignificant water quality impacts. Thus, effective control of
a stormwater related problem requires an accurate definition of the problem.
The definition of a problem should begin with a stream walk for visual
identification of conditions, and be followed by a quantitative of water
quality impacts. This is often a difficult task, since water quality is also
influenced by factors other than stormwater runoff and it is necessary to
separate the existing water quality impact into its component parts. This
can be accomplished using a mathematical analysis of the receiving water
system. An illustration of such an analysis is shown in Figure 2-1.
Figure 2-1(a) shows the result of an analysis of urban point and non-
point source impacts on dissolved oxygen in a hypothetical river. The dis-
solved oxygen example is used to illustrate one type of impact, and to show
an approach for identifying the relative impact of storm loads versus input
from other sources. Similar comparisons can be made for other contaminants,
some of which may have more significant impacts than dissolved oxygen in
specific cases.
The computed dissolved oxygen concentrations are given as a function of
downstream distance. The dissolved oxygen saturation concentration, and a
dissolved oxygen standard are also shown. This analysis is for the impact
of a particular event, the mean summer storm, on the river. It indicates
a violation of the 4.0 mg/1 dissolved oxygen standard between Milepoints 24
and 35. The causes of the dissolved oxygen depression may be seen more
clearly in Figure 2-1(b) which converts the dissolved oxygen concentration
to the corresponding dissolved oxygen deficit value (amount below saturation)
and displays the components of the dissolved oxygen deficit profile that
make up Figure 2-1(a). It is apparent that the most significant factors
contributing to the problem (violation of the water quality standard during
the average summer storm) is the storm related loads from storm sewers and
combined sewer overflows within the urban area. Point sources from two
sewage treatment plants, and various industrial discharges contribute to the
problem to a lesser degree. Similar analyses for other water quality indi-
cators would yield similar insight into the causes of potential or actual
problems.
2-6
-------
,- 12
10
UJ
CD
X
o
Q
UJ
>
O
in
5
(a)
SUMMER STORM PERIODS
TEMPERATURE:25°C
BASE FLOW*23OCFS
P.O. STANDARD
_SATURAT10N_= £ IJ7 mj| /_'
MEAN STORM-
I
I
I
fO 15 20 25
MILES BELOW ROUTE 80 BRIDGE
30
35
o>
E
o
\L
UJ
z
X
o
Q
UJ
o
00
O
10
9
8
7
UN/T RESPONSE ANALYSIS
MEAN STORM LOAD
Mare:
REPRESENTS DISSOLVED OXY6EN DEFICIT
CONCENTRATION WHICH WILL ONLY BE
EXCEEDED 1.2% OF THE TIME.
I BACKGROUND 8 UPSTREAM
10 15 20 25
MILES BELOW ROUTE 80 BRIDGE
SOUTH RIVER
|-INDUSTRY
30
35
NOTE.
REFERENCE [4]
FIGURE 2-1
TYPICAL ANALYSIS OF WET WEATHER DISSOLVED OXYGEN
2-7
-------
This type of analysis may be used to develop certain conclusions regard-
ing storm related dissolved oxygen problems. For example, stormwater appears
to be a major contributor to the problem, where the problem is defined in
terms of a stream standard specifying dissolved oxygen concentrations to
never be less than some value (in this example, 4.0 mg/1). Control of storm,
sewer flow and/or combined sewer overflow will yield the largest single im-
provement to the indicated problem. But there are other important questions
which Figure 2-1 cannot answer, and which can have a significant bearing on
the cost of controls required to protect the beneficial use which the stream
standard is designed to preserve. These relate to: How often will storm-
water events cause violations of standards? This is a final refinement to
the definition of the problem which provides major additional insights.
Figure 2-2 presents these results for the same urban area indicated in
Figure 2-1. In this case, however, the water quality is associated with
storms having different frequencies of occurrence. This analysis provides
additional insights into the problem. It indicates that the summer storm,
shown in Figure 1, is a large storm; one which is exceeded only 35 percent of
the time. The degree to which other storm frequencies impact summer water
quality is also apparent. The analysis indicates that 50 percent of the
storms are small enough not to cause water quality standard violations.
Furthermore, the analysis permits an assessment of how frequently dis-
solved oxygen concentrations fall below prescribed levels due to storm im-
pacts. These frequencies are shown on the right side of the figure and in-
clude both dry and wet weather periods. For example, summer dissolved oxygen
concentrations fall below the 4.0 mg/1 standard 2 percent of the time due to
stormwater impacts. This analysis applies for the conditions selected to
characterize the other significant elements of the analysis, principally the
magnitude of other contributing waste loads and the stream flow assigned.
However, where uncertainty exists in the selection of these factors, ad-
ditional calculations of the same type can be made to provide the desired
understanding and perspective on the problem. For example, the analysis
could be repeated for a different stream flow, selected such that the two
analyses would bound the range of possible responses.
An alternative method of presenting probabilistic water quality results
is shown in Figure 2-3, reproduced from the "Nationwide Evaluation of Com-
bined Sewer Overflows and Urban Stormwater Discharges, Volume II, Cost
Assessment and Impacts" (5), which shows the result of wet weather mathema-
tical modeling studies for the Des Moines River. The computed minimum dis-
solved oxygen concentration during wet weather periods is indicated as a
function of the storm frequency. The components of the dissolved oxygen im-
pact are also indicated. Storm related loads from separate runoff and com-
bined sewer overflows are shown to be the primary source of the dissolved
oxygen deficit during runoff periods. Under existing conditions, the model
indicates that the minimum dissolved oxygen concentration falls below the
desired level C4.0 mg/1) following 42 percent of the runoff events.
The planning question which relates to problem definition at this
point is: does violation of the standard 2 percent of the time (from
Figure 2-2) or following 42 percent of the storms (in the case of Figure
2-8
-------
UJ
I NVMi SS3-I
1
wv
1
u w
e »•
- n o
=8
00
o
CC
Ul
m
oo
UJ
2°
— u
Q ffl
o: w
0 ,
2
o r
<0
o
CO
CO
Q:
UJ
C\J UJ
CVJ
l-
UJ UJ
cr ^
1 &
CO
z
o
CO
I]
m
Ul
o
Ul
(E
Uj ui
b *
ffi
O
cc
Q.
2-9
-------
100 -i
\
PRECIPITATION YEAR OF RECORD • 1968
DWF TREATMENT RATE' 85% (SECONDARY)
WWF TREATMENT RATE ' 0% (NO TREATMENT)
RIVER FLOW: 100% (OF MEASURED FLOW)
COMBINED SEWER AREA 8 16% (OF TOTAL URBAN AREA)
INFLOW COMBINATION
RIVER FLOW + DWF
- - RIVER FLOW + DWF+ SEPARATE FLOW
\ • RIVER FLOW + DWF-t-COMBINED FLOW
RIVER FLOW + SEPARATE FLOW -I- COMBINED FLOW
\ RIVER FLOW + DWF 4- SEPARATE FLOW + COMBINED FLOW
x —_ INDICATES EVENTS EXCEEDING DESIRED DO. LEVEL
40 60 8.0 100 120
DISSOLVED OXYGEN CONCENTRATION, mg/l
140
NOTE:
REPRODUCED FROM REF. [5]
FIGURE 2-3
MINIMUM DISSOLVED OXYGEN FREQUENCY CURVES
FOR EXISTING CONDITIONS IN THE DES MOINES RIVER
1-10
-------
2-3) constitute a water quality problem? This type of question is at the
heart of meaningful analyses of stormwater related problems and will be dis-
cussed in detail in subsequent chapters. Those discussions will focus on
the frequency analysis of stormwater loads, treatment, and receiving water
impacts. For the purposes of this section a more detailed discussion of the
factors influencing the definition of water quality problems is appropriate.
The principal factors involved in defining stormwater problems have,
thus far, been related in a very general way to water quality. There are,
however, a series of factors which must re considered for an adequate review
of the potential for water quality probl is in a site specific urban setting.
These are:
1. Relevant time and space scales of the problem
2. Water use objectives and criteria
3. Characteristics of the particular study area
These factors are discussed below.
2.4.1 Relevant Time and Space Scales
A basic step in identifying potential water quality problems is the
definition of the time and space scale of the problem. How large an area
might be affected by a particular class of contaminant? How long does it
take for the problem to become manifest and over what interval will the
impact be felt?
The definition of proper time and space scales is important in a number
of stormwater problem analysis tasks. The time and space scale of a poten-
tial problem will determine the time and space scale of the models or other
analyses which are used to address it. Short term transient phenomenon are
analyzed using different analysis frameworks than those used to evaluate
longer term steady state buildups of less reactive contaminants.
Time and space scales of receiving water impacts are also important
in the determination of effective monitoring programs and for the character-
ization of waste loads. If a particular problem is due to continuous point
source loads, the monitoring program which addresses it will be entirely
different than a monitoring program evaluating stormwater loadings to the
same river.
2.4.1.1 Nature of Contaminants
Each of the classes of water quality contaminants in Table 2-1 and the
potential problems in Table 2-2 must be viewed in a specific time and space
scale. For example, bacterial contamination is particularly relevant in the
time scale of a few days and generally occurs in a localized area. This is
due to the high rate of decay of coliform bacteria in natural water systems
which typically result in their reduction to background levels within a few
days. The space scale over which they are relevant is similarly small and is
governed by the distance that they are transported before they die-away. By
way of contrast, toxic substances such as pesticides, persistent organics,
2-11
-------
and heavy metals are normally viewed in longer time scales, and space scales
which may extend many hundreds of miles. These substances tend to be per-
sistent (i.e., they do not readily decay in the environment). The time scale
in this case may be as large as decades, and contaminants can have effects
at locations which are remote from their point of introduction into the
environment.
Certain contaminants may fall into multiple time and space scales. For
example, the persistent toxicants discussed above may exhibit acute toxic
effects in the immediate vicinity of discharge if local concentrations exceed
tolerable limits of sensitive aquatic species. The appropriate time and space
scale chosen for the analysis should reflect judgement based on local con-
ditions, the type of water body, particularly sensitive areas, local transport
conditions, and types of waste loads. A guideline in developing the proper
time and space scales for suspected water quality problem is contained in
Figure 2-4 and 2-5.
2.4.1.2 Nature of the Receiving Water System
The classes and appropriate time and space scales of storm related water
quality problems in a particular urban setting is often related to the nature
of the receiving waters. Streams and rivers are particularly sensitive to
intermittent short term discharges (storm loadings) because the mass is trans-
ported as an identifiable pulse with relatively little dispersive mixing.
Sensitive downstream locations, such as water intakes or bathing areas, may
be affected by these identifiable pulses. Other systems such as estuaries
tend to spread and dilute the impacts of storm loads, and short term temporal
definition may not be necessary.
Specific types of contaminants are assimilated differently in different
water bodies. For instance, sediment inputs may contribute to high suspended
solids concentrations in a swiftly moving river. Classes of water quality
problems which are important in this case are normally related to aesthetic
concerns or interferences with physiological functions of aquatic organisms.
In estuaries and lakes where velocities are not as great the sediment problem
is largely due to deposition which causes solids buildup in navigable
channels or increased organic bottom activity. Similarly, nutrient enrichment
is normally more critical in lakes of long detention times than in free flow-
ing streams. In lake systems the problem is often related to discoloration,
floating algal mats, and depressed oxygen levels, while in streams and rivers
aquatic weeds may be the major problem.
2.4.2 Water Use Objectives and Criteria
Water quality problems should be defined in terms of their limitation
of beneficial uses of the water body. Thus, an inventory of present and
planned water use is an integral part of the planning process. An example
inventory of beneficial water uses is presented in Table 2-3.
2-12
-------
IOJ
SECONDS
10
10
10
10"
10'
FLOATABLES
BACTERIA
DISSOLVED OXYGEN
SUSPENDED SOLIDS
NUTRIENTS
DISSOLVED SOLIDS
ACUTE TOXIC EFFECTS
L O/VG TERM
TOXIC EFFECT
I
_L
HOUR
DAY
MONTH
YEAR
WEEK
SEASON
DECADE
FIGURE 2-4
TIME SCALES
STORM RUNOFF WATER QUALITY PROBLEMS
-13
-------
HYDRAULIC DESIGN
fLOATAULES
BACTERIA
SUSPENDED SOL I OS
DISSOLVED OXYGEN
NUTRIENTS
TOX 1C EFFECTS
DISSOLVED SOLIDS
10 '
( 5 FT )
10 '
I SOFT )
10 '
( 5OOFT)
10
10'
10'
10J
EFFECTIVE DISTANCE - MILES
-LOCAL-
•REGION-
-BASIN-
H
FIGURE 2-5
SPACE SCALES
STORM RUNOFF WATER QUALITY PROBLEMS
2-14
-------
TABLE 2-3 EXAMPLE BENEFICIAL WATER USE INVENTORY
1. Water Supply
a. Domestic Water Supply
b. Industrial Water Supply
c. Irrigation
d. Livestock and Wildlife Use
2. Maintenance and Propagation of Fish and Other Aquatic Life
a. Commercial Fisheries
b. Recreational Fishing
c. Aesthetic Value
3. Recreation
a. Water Contact Sports, Swimming
b. Boating
c. Aesthetic Value
4. Navigation
5. Power Generation
6. Transport and Assimilation of Treated Wastes
Each of these should be screened for potential impacts early in the planning
process. The protection and enhancement of these beneficial water uses is
the primary goal of any stormwater control plan.
Water quality criteria and standards should be developed to protect
specific beneficial water uses. Comparing current or projected pollutant
concentrations in the receiving water with regulatory standards thus
provides a basis for defining the stormwater problem in terms of beneficial
water use limitations.
2.4.3 Characteristics of Particular Study Area
Planning studies are particularly meaningful when their scope goes
beyond the "standard" dissolved oxygen and suspended solids problems, and
explores other potential problem classes which are important in terms of
water use in the study area. This does not mean that every potential water
use interference must be explored in detail. But, the potential for signifi-
cant water use limitations should at the very least be screened.
In this regard the planner must think beyond documented problems in
their study area and identify areas where storm loads might contribute to as
yet undocumented problems. These can include concerns related to oils and
greases, heavy metals, pesticides, herbicides, carcinogens, and viruses. In
identifying the potential for water use interferences in any of these classes
two factors are normally evaluated: 1) existing or potential water uses
which might be effected, and 2) potential sources of the contaminant. Local
knowledge of the study area is required for both evaluations. For example,
downstream water supplies suggest a class of potentially harmful contaminants.
2-15
-------
A review of upstream study area characteristics will indicate whether storm
loads could potentially contribute harmful contaminant loads. Agricultural
areas are a potential source of pesticides and herbicides. Landfill leachate
is a potential source of heavy metals and oils and grease. Combined sewer
overflows are sources of viruses. Any one of these areas may be a potentially
important contaminant source. Crude but conservative estimates of loads
from each source, using local data where possible, and simple mass balance
computations are generally sufficient to determine the potential for water
quality problems from each of these sources. Where these simple estimates
indicate a high problem potential, more detailed analyses using sampling data
will serve to further document the problem.
Planning develops sequentially and planners must continually expand
their technical base of information in order to provide input to the water
resource management decision process. The current trend towards higher
water quality standards and increased water use will undoubtedly continue.
Planning studies must therefore be responsive to this increased demand.
2.5 References
1. Field, Richard, Anthony N. Tafuri, Hugh E. Masters, Urban Runoff
Pollution Control Technology Overview, U.S. Environmental Protection
Agency, EPA-600/2-77-047, Cincinnati, Ohio, March 1977.
2. Brandstetter, Albin, Assessment of Mathematical Models for Storm and
Combined Sewer Management, Battelle, Pacific Northwest Laboratories
for U.S. Environmental Protection Agency, EPA-600/ 2-76-175a, Cin-
cinnati, Ohio, 1976.
3. Lager, John A,, W.G. Smith, W.G, Lynard, R,M. Finn, E,J. Finnemore,
Urban Stormwater Management and Technology: Update and Users Guide,
Metcalf Agency, EPA-600/8-77-014, Cincinnati, Ohio, September 1977.
4. Hydroscience, Inc., 208 Areawide Assessment Procedures Manual, Volume
I, Chapter 5, U.S. Environmental Protection Agency, EPA-600/9-76-014,
Cincinnati, Ohio, July, 1976.
5. Heaney, James P., et al., Nationwide Evaluation of Combined Sewer
Overflows and Urban Stormwater Discharges, Volume II: Cost Assessment
and Impacts, University of Florida for U.S. Environmental Protection
Agency, EPA-600/2-77-064, March, 1977.
2-16
-------
CHAPTER 3
THE STATISTICAL METHOD FOR THE ASSESSMENT
OF RUNOFF AND TREATMENT
Once site specific urban stormwater problems are identified, suitable
analysis techniques are required for evaluating the magnitude of the receiv-
ing water impact and the effectiveness of alternative control options. This
task is particularly complex because of the random nature of the problem's
principal forcing function, the rainfall-runoff process.
Chapter 3 explores a particularly effective methodology for evaluating
urban stormwater problems. It is the result of a growing awareness on the
part of individuals involved in stormwater impact analysis that the real
benefits in stormwater control can only be determined by evaluating the long
term response characteristics of treatment devices and receiving water
quality (1,2,3,4,5,6). The techniques discussed in this chapter accommodate
this need- by analyzing storm related problems in terms of the long term
statistical properties of rainfall, runoff, treatment performance and receiv-
ing water quality. The analysis produces results which address critical
urban stormwater questions such as: How often will stormwater runoff cause
water quality and water use objectives to be violated? What is the long term
performance efficiency of various control options? What are the marginal
costs associated with alternative water quality objective sets? Subsequent
chapters will develop analytical techniques for evaluating model coefficients
in a specific urban setting. These incorporate monitoring program design as
well as various numerical estimates.
The predominant analytical tool currently used to evaluate stormwater
problems is rainfall-runoff simulation. The emphasis in simulation models
has evolved from the more detailed description of a particular storm event
(7,8,9,10) towards the more general analysis of long term stormwater impacts
with "continuous versions" (4,5,11,12). Examples of these more general,
long term simulators are discussed in Chapter 4.
Recently, analytical methods which utilize the statistical properties
of rainfall and runoff to investigate stormwater problems have been developed
(2,13,14,15). This chapter presents a general methodology for assessing
stormwater problems analytically, without a computer based simulation model.
The method requires minimal computer use and can be implemented with an
electronic calculator. Both the statistical method and computer simulation
models are analysis tools which involve simplifying assumptions about the
mechanisms of the real stormwater process. Many of the assumptions are the
same in both approaches, and as demonstrated in Section 3.6.7.7, the results
3 - 1
-------
are similar. The primary advantage of the statistical method is that it
allows a relatively quick, simple, and inexpensive screening of stormwater
problems and a wide range of control alternatives. This is particularly
useful in the early stages of the planning process. The use of the tech-
niques and curves of the statistical method gives planners a better under-
standing of their stormwater problems. Insight is gained by following the
problem through, step by step, rather than developing a list of input
factors, running a complex simulation model, and receiving the final output.
The statistical method is particularly well suited to assessment studies
where the principal issues involve broad planning questions such as: What
are the major loads contributing to the problem? What treatment devices are
particularly effective? What are the levels of costs associated with alter-
native controls or alternative water quality objectives? While the statisti-
cal method may also have utility in more detailed siting studies, such as 201
facilities plans, it is often advisable to supplement the statistical method
with additional detailed modeling using one of the more sophisticated storm-
water simulators. Refined simulation is particularly useful at the design
stage, where the planner is interested in the behavior of control alternatives
and the response of the receiving water during a specific sequence of critical
storms. In this respect, the statistical method and simulators should be
viewed as complementary, each appropriate at different stages in the analysis
and each providing insight and direction for the use of the other.
Chapter 3 presents a sequential development of the statistical method.
The chapter begins with a discussion of the random nature of storm events and
provides the basic framework for the characterization of the rainfall-runoff
process. Stormwater loads are developed through the analysis of rainfall,
runoff quantity, and quality. Methods for determining receiving water im-
pacts are presented, and techniques for assessing the benefits of stormwater
control alternatives are developed.
3.1 Storm Runoff Events as Random Occurrences
The most basic source of variability in the rainfall-runoff process is
the fact that sometimes it is raining, and sometimes (usually) it is not.
The rainfall-runoff process consists of a series of events occurring randomly
in time. A more precise definition of the random nature of storm event
occurrences is presented in this section.
Consider the rainfall event process and let T be a fixed length of time,
e.g., one month. For each successive period of length T, let n be the number
of events which occur in that period. Since the occurrence of rainfall is a
probabilistic event, the number of occurrences during a fixed length of time
is a random variable. The critical assumption which allows a comparatively
straightforward analysis is that n is a Poisson random variable with discrete
probability density function:
rT//nn -T/A
fn(n) =^V n= 0, 1, ... (3-1)
3 * 2
-------
with parameter A. Further, n is assumed to be an independent random
variable, that is, the number of rainfall events in any period of length T is
independent of the number that occurred in any other period. A direct con-
sequence of this assumption is that the storm interval, 6 (measured from the
temporal midpoints of the events), is a random variable with an exponential
probability density function. To see this, consider the probability that
6 _< T, that is, that at least one rainfall event occurs during the period T:
Pr (6 <_ T) = 1 Pr (6 > T) (3-2)
and
Pr(<5 > T) = Pr(n = 0) = eT/A (3-3)
since if 6 > T no rainfall event has occurred within the period T and n = 0.
Therefore, the cumulative distribution function of the random variable 5 is
Pr(5 <_ T) = 1 e~T/A (3-4)
so that 6 has an exponential probability density function:
P6(6) = - e~6/A 6 >_ 0 (3-5)
Further, the average value of 6 is:
6 = 7 | e"6/A d6 = A (3-6)
o
The parameter of the Poisson density function is thus seen to be the average
time between storms.
This characterization of the rainfall events as a Poisson process is the
most convenient available and is also surprisingly realistic. Analysis of
rainfall records (Chapter 5) indicate that the storm intervals is well de-
scribed by the exponential distribution and that the storm event definition
(number of consecutive dry hours needed to terminate an event) may be chosen
to more accurately approximate an exponential distribution for 6. The
assumption that the interval between two storms is independent of the interval
between any other two storms is also quite reasonable.
3.2 Characterization of Runoff Events
The storm runoff process may now be characterized as in Figure 3-1 as a
series of independent events occurring randomly in time. The intrastorm
variability depicted in Figure 3-1(a) is ignored for the time being and each
event is characterized in Figure 3-1(b) by its duration (d ), runoff volume
(v ), time since the previous storm (<5) (defined from the midpoints of the
successive storms), and the average runoff flow (q = v /d ) . The transforma-
tion from rainfall to runoff is not yet directly addressee. Rather it is
assumed that the relevant runoff characteristics are available either from
direct observations, or from suitable modifications of the rainfall record.
This is discussed in more detail in Section 3.2.3.
3-3
-------
a) VARIATION WITHIN EVENTS
k o
b) VARIATION BETWEEN EVENTS
n
I
TIME
TIME
FIGURE 3-1
REPRESENTATION OF STORM RUNOFF PROCESS
3-4
-------
3.2.1 Statistical Properties of Runoff Parameters
The runoff event characteristics may be treated as random variables,
each with an associated probability density function. For a given period of
record, the mean and coefficient of variation (standard deviation divided by
the mean) of each variable are calculated. The required statistics are
summarized in Table 3-1.
TABLE 3-1
STATISTICS FOR STORM CHARACTERIZATION
Coefficient
Parameter
Runoff Flow
Duration
Runoff Volume
Time Between Storms
For Each Storm
q
d
r
v
r
6
Mean
QR
Dn
R
V_
R
A
of Variation
vq
d
Vr>
R
v „
6
Storm flows and durations are assumed to be independent and gamma dis-
tributed. A gamma distribution is defined by a mean, in thjs case, Q and
DR; and a coefficient of variation v and v, . The probability density
function for runoff flows is then: "
Vq) = -)T
K
2
in which K = 1/v . A similar expression describes the probability density
function of storm durations. The gamma distribution is effectively a more
generalized version of the exponential distribution. The gamma distribution
allows the coefficient of variation to vary; the exponential distribution is
a special case of a gamma distribution in which the coefficient of variation
is equal to 1.
As described in the previous section, the probability density function
of the time between storms (6) is well described by an exponential distri-
bution; that is, the coefficient of variation is approximately equal to one.
If this is not the case, the gamma distribution may also be used to describe
the time between storms.
The runoff volume of a particular event is equal to the product of the
runoff flow and the duration. The runoff volumes may thus be represented by
3-5
-------
the product of two independent, gamma distributed, random variables. If this
were completely true, the mean runoff volume, V , would equal the product of
the mean runoff flow and mean runoff duration (VR = D Q ). This may not al-
ways be valid, however, particularly in areas where snorter, intense summer
storms and longer, less intense winter and spring showers are typical and
runoff flows and durations are not completely independent. Thus the statis-
tics of runoff volumes must be obtained in addition to those on flows and
durations. In fact, the assumption that storm runoff volumes are themselves
gamma distributed is probably quite reasonable.
The cumulative distribution function for the gamma distribution is
shown in Figure 3-2. Figure 3-2 is used to determine the percent of storms
with runoff characteristics less than or equal to the given value. For
example, if the variation of storm runoff flows is v =1.25, from Figure
3-2(a) the 90th percentile runoff flow is 2.6 times rhe mean runoff flow,
Q . In other words, ten percent of the storms have average runoff flows
greater than 2.6 Q . Interpolation may be used for intermediate values of
the coefficient of variation. Note that when the coefficient of variation
(v) is large, events with very small and very large values relative to the
mean become more likely. That is, there is more spread around the mean in
the probability density function.
Once the percent of storms larger than a given value is determined, the
expected number of storms greater than a given value may be estimated. The
average number of storms occurring during a given period is first calculated:
, , - . Length of Period ,_ 0,
Average number of storms = — (3-8)
Then, for example, if the period of interest is one year and the statistical
analysis indicates that the average time between storms, A, is 70 hours:
. , . . 1 year * 8766 hr/year 10t.
Average number of storms = — =-^—, = 125
70 hr
The expected number of storms greater than a given value is then the fraction
of storms greater than the given value times the average number of storms.
From the previous example, there will be (on the average) (0.10)125 = 12.5
storms per year with average runoff flows greater than 2.6 Q .
R
3.2.2 Long Term Runoff Process
Once the expected number of events exceeding a given criterion is
estimated, the final step in the statistical characterization is the deter-
mination of how often this will occur during the entire interval of interest,
including both the rain and nonrain periods. This estimate is most strongly
influenced by the fact that it is usually not raining.
The transformation from a probability distribution function for storm
events to a probability distribution function for both rain and non-rain
periods is depicted in Fi-gure 3-3. The fraction of the time during which
storm runoff is occurring is estimated as D /A. The remaining portion of
the time (generally between 70 and 98 percent) there is no storm runoff. The
3-6
-------
\ I I
(0
O)
f-
O)
ID
O)
o
CD
J
3n~IVA N3AI3 01 ~IVD03tlO NVHJ.
o o oo
rO cv) —
3mVA N3AI9 OJ. ~IVr>O3UO NVHJ. SS31 J.N30U3d
oc.
O
U.
H
O
z
~ ^>-»
S"S
"§S
L rrcD
UJ
o
3-7
-------
A
\
X
O)
0)
CO
h-
0">
ic in »fr to (vj — O
CTl <7> O 01 O 0>
O
00
O
h-
MULTIPLES OF THE MEAN
o o o o ooo
(0 in **• 10 cj —
3ITIVA N3AIS OJ. ~IVn03 UO NVHl SS31 J.N30H3d
\
N
\
\
I I I I I I
r- to in
u.
z
g
H
m
(T
I-
O5
UJ
>
fe
3
2
O
o
3-0
-------
3 a) STORM EVENT DISTRIBUTION
u_
100
o
X
\-
UJ
_J
CO
oa
o
(-
CO
u_
o
UJ
o
a:
UJ
a.
2
UJ
CO
UJ
UJ
UJ
o
or
UJ
Q.
RUNOFF FLOW
b) LONG TERM DISTRIBUTION
lOO-r
O-1-
TIME PERIODS WITH
NO STORM RUNOFF FLOW
RUNOFF FLOW
FIGURE 3-3
STORM EVENT AND LONG TERM FREQUENCY DISTRIBUTIONS
-------
probability of a given runoff flow being exceeded is the probability that a
storm will have an average runoff equal to or greater than the given flow
times the fraction of the time that runoff occurs. Given the assumption that
storm runoff flows are gamma distributed, the percent of time (including rain
and nonrain periods) when runoff flows exceed a given flow may be determined
from Figure 3-2 by knowing Q , v , and D /A. For example, assuming D = 8
hours, A = 70 hours, Q = 4.U cfi, and v = 1.25; Figure 3-2(a) indicates
that 10 percent of the storms have average runoff flows greater than 2.6 Q
= 10.4 cfs. The fraction of time when average runoff flows exceed 10.4 cfs
then equals:
0.10 (D_/A) = 0.10 (8/70) = 0.0114
R
In other words, considering the entire time history, runoff flows are estima-
ted to exceed 10.4 cfs about 1.1 percent of the time.
The mean and variance of the long term runoff process shown in Figure
3-3(b) may be calculated from the generalized Cambell's Theorem for random
(i.e., Poisson) pulse processes (17,18). The long term average runoff flow
and its coefficient of variation (including rain and non-rain periods) are:
Q0 - QR CDR/A) t3"9)
V
Note that the long term average runoff flow is calculated by assuming that
the storm runoff is effectively spread out over the entire time history, and
Q is thus smaller than Q . The coefficient of variation of the long term
process (v ) is larger than the variation of storm flows (\> ) due to the
many increments of time when runoff flows equal zero. The smaller the
fraction of the time runoff is occurring (D /A), the smaller is the long term
average flow (Q ), and the larger is the variability of the long term process
Cv ). °
qo;
3.2.3 Determination of Runoff Parameters
The basic theoretical framework for analyzing the statistical properties
of runoff events has been presented. The actual method for determining the
runoff characteristics (i.e., the runoff volume, duration, etc., for a series
of storms) has not yet been addressed. A number of approaches may be taken
to determine runoff event characteristics. The most direct method is to
monitor the runoff flows each storm for a period of interest. Although a
monitoring program is useful and necessary for certain aspects of the study
plan, time and budgetary constraints limit the applicability of monitoring
for a complete long term characterization, and estimating techniques are
required.
Rainfall is the primary driving force in the generation of stormwater
runoff. Analytical models may be used to transform rain event characteristics
3 ^ 10
-------
to runoff event parameters. These may range from simple models using a direct
linear conversion of the precipitation falling over an area to its runoff; to
sophisticated models employing varying infiltration rates, depression storage,
overland flow and flow routing through the conveyance system.
A simple method for converting rainfall records to runoff records is
presented in the following sections. The techniques for evaluating the long
term statistical properties of the stormwater induced receiving water re-
sponses and the long term efficiency of control alternatives are not depen-
dent upon the use of these simple rainfall - runoff conversions. More
sophisticated models may be appropriate, particularly during the later stages
of the planning study. The general statistical methodology is flexible, and
more refined estimates may be used at different stages in the analysis, de-
pending upon the specific problem setting and the data available.
3.3 Rainfall, the Driving Force
Precipitation is the driving force in the generation of stormwater run-
off and its associated pollutant loadings. Precipitation may occur as rain-
fall, snow, or hail; and the characteristics of the resulting runoff are
very different, depending upon which of these types of precipitation has
occurred. Stormwater impacts generated by rainfall are the primary focus of
this manual. The techniques presented in this manual may be suitable for
evaluating impacts caused by snowmelt on a long term (i.e., yearly) time
scale, however, more sophisticated modeling techniques are required for a
more detailed analysis (19,20,21). Rainfall and snow impacts may be sep-
arated by a seasonal assessment, aided by the monthly characterization of
precipitation described later in this section.
Rainfall occurs as a series of random events, with the amount of pre-
cipitation varying spatially as well as temporally. The approach for ana-
lyzing rainfall presented in this section is directed towards a long term
characterization of rainfall at a point, i.e., a raingage. A discussion of
the areal variability of rainfall, and techniques for transforming point to
areal rainfall are presented in Section 5.1,4,
The method used to describe point rainfall is based on a statistical
characterization of event properties for a long period hourly rainfall
record. Such records are collected by the U. S. Weather Bureau at weather
stations through the United States, and are available on cards or magnetic
tape through:
The National Climatic Center
National Oceanic and Atmospheric Administration
Federal Building
Asheville, North Carolina 28801
The length of rainfall record required to adequately define the statistical
variability of the rainfall record is generally 20 to 30 years. Shorter
records may contain atypical numbers of dry or wet years, but may be used if
longer data records are not available.
3 T 11
-------
The development of rainfall event statistics begins with a grouping of
the discrete hourly records into a series of storm events. Table 3-2 demon-
strates the implied transformation. Table 3-2(a) shows one month of hourly
rainfall data from Minneapolis, Minnesota. Only days with some precipitation,
and the first day of the month, are shown. The characteristics of the events
for this period are shown in Table 3-2(b). The storm duration is calculated
from the first hour of rainfall until the last hour of rainfall which is
followed by 6 consecutive dry hours. The number of consecutive dry hours
required to end a storm event is the basic parameter which determines how the
hourly records are grouped into storms. Considerations for selecting this
parameter in a particular study area are presented in Section 5.1.2. The
intensity is the average intensity during the storm and the volume (also re-
ferred to as the depth) is the total storm volume in inches. The interval
between storms (calculated from the temporal midpoint of the previous storm)
is also shown. This analysis thus characterizes each storm for the period of
record.
The event characteristics are statistically analyzed for all storms or
for groups of storms in the period of record. These statistics are used to
evaluate the long term properties of storm events in the urban area. The
mean and coefficient of variation of each of the event characteristics are
calculated. This is accomplished using a synoptic rainfall analysis which
has been described in the Areawide Assessment Procedures Manual (2). A
typical output from such an analysis is presented in Figure 3-4 which dis-
plays the rainfall statistics from Central Park, New York City, summarizing
the mean and coefficient of variation of the four rainfall parameters on a
monthly basis. The statistical summary is for all storms which occurred in
each of the months during the 27 year period (1948-1975).
Some interesting properties of the rainfall process are evident in
Figure 3-4. For example, the mean intensity and mean duration show marked
seasonal variability. The summer months are generally characterized by
short, high intensity storms (an indication of thunderstorm activity), while
the winter months have longer, less intense events. Note that the mean storm
volume is less variable seasonally (0.3 to 0.4 inches) and the mean storms
interval is approximately 80 hours, or slightly more than three days. Storms
occur least frequently in October, when the mean time between events is
approximately 110 hours. Storm intensities and volumes tend to be more
variable in the summer, while the coefficient of variation of duration and
the time between storms (v and v., respectively) are relatively constant
throughout the year and very nearly equal to one.
The rainfall statistics for a particular month or seasonal period may
be estimated from Figure 3-4. For example, for the period of June - August,
the storm characteristics are given below:
3-12
-------
TABLE 3-2
MINNEAPOLIS RAINFALL ANALYSIS
Date
5/ 1/74
5/ 4/74
5/ 7/74
5/ 9/74
5/10/74
5/11/74
5/13/74
5/14/74
5/15/74
5/16/74
5/21/74
5/30/74
A. HOURLY PRECIPITATION
12345
am
pm
am
pm
am
pm
am
pm
am
pm
am
pm
am
pm
am
pm
am
pm
am
pm
am
pm
am
pm
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
4
0
0
0
0
0
0
0
0
0
1
0
3
0
0
0
0
0
0
0
0
0
5
5
4
(Hundredths of
678
0
0
0
0
0
0
5
0
0
15
6
0
5
1
0
0
0
0
0
0
0
2
0
10
0
0
0
0
0
0
6
0
0
2
0
0
19
0
0
0
0
1
0
0
0
0
0
27
0
0
0
0
0
1
2
0
0
3
0
0
6
0
0
0
0
1
0
0
0
0
0
0
an Inch)
9 10
0
0
0
2
0
0
0
0
0
8
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
12
0
0
0
3
0
0
0
0
0
0
0
0
0
0
11
0
0
0
0
0
1
1
0
0
23
0
0
0
0
0
0
0
0
0
0
0
0
0
0
12
0
0
0
0
0
1
0
0
0
13
0
0
0
0
0
0
0
0
0
0
0
J
0
0
3 T 13
-------
TABLE 3-2
MINNEAPOLIS RAINFALL ANALYSIS (Continued)
B. CHARACTERISTICS OF STORM EVENTS
Storm
No.
2688
2689
2690
2691
2692
2693
2584
2695
2696
2697
2698
2699
2700
Date
05/04/74
05/07/74
05/09/74
05/10/74
05/11/74
05/13/74
05/13/74
05/14/74
05/15/74
05/16/74
05/21/74
05/30/74
05/30/74
Beginning
Hour
21
19
6
14
17
6
18
13
19
4
14
5
16
Duration
(hr)
1
6
6
17
1
4
5
2
2
1
5
1
4
Intensity
Cin/hr)
0.020005
0.005000
0.025000
0.050000
0.030005
0.077501
0.010001
0.010002
0.010002
0.020005
0.016000
0.050005
0.112501
Volume
(in)
0.02
0.03
0.15
0.85
0.03
0.31
0.05
0.02
0.02
0.02
0.08
0.05
0.45
Interval
Between
Storms
(hr)
186.5
72.5
35.0
37.5
19.0
38.5
12.5
17.5
30.0
8.5
132.0
205.0
12.5
3 - 14
-------
MEAN VARIATION
0. 12
a: 0. 10
* 0 08
2 0.06
*~ 0. 04
H 0.02
_ STORM INTENSITY
..
J/^\
s^ *L-
M' ^^^*.
^^__ArfH*W ^>v^
,"- ' "»
1 1 1 1 1 1 1 1 1 1 1 1
2 5
2 0
._ 1 5
^
1 0
05
-
-
^^
s^f^^***— 1 5
J^
1 0
0.5
-
-
J~-+—*-*^'^\--
• — •— •^*J * ^T»
—
-
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
120
100
rr
X 80
< 60
40
20
n
- TIME BETWEEN STORMS
.A
~~ *^* *V- mr-*'*'^ *~~+
^^r-*-^
.
i i i I l i I I I i i i
2 5
2 0
ao 1.5
1 0
0 5
0
-
-
-
-*-«- •_
•—• — » »-*^"^ ••••••
1 1 1 1 1 1 1 1 1 1 1 1
12345678 9 10 II 12
MONTH
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
FIGURE 3-4
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
CENTRAL PARK
STATION 305801
3-15
-------
Estimated Central Park Storm Characteristics
June - August
Coefficient
Mean of Variation
Storm Intensity I = 0,08 in/hr v. = 1.4
Duration D=4.5hr v, = 1.1
d
Unit Volume (Depth) V=0.35in v = 1.5
Time Between Storms A = 70 hr v = 1.0
o
(Note that some error is introduced by simply averaging monthly statistics
for a given season, particularly for the coefficient of variation where a
pooled variance calculation is necessary. This error is usually quite small,
however, and may be ignored for the purposes of this manual.)
After the mean and coefficient of variation of the rainfall properties
are determined, the frequency distribution is compared to the gamma distri-
bution in Figure 3-2. Examples are presented in Section 5.1.3, along with a
more detailed discussion of the applicability of the gamma distribution to
storm characteristics.
Once the rainfall properties are adequately characterized, the estima-
tion of stormwater loads may proceed. This is outlined in the following
sections.
3.4 Development of Stormwater Loads
This section discusses estimating techniques for stormwater loads of
various time scales, consistent with the problem definition time scales
presented in Chapter 2. These include the average annual load, various
seasonal loads, and transient loads occurring from individual events. The
technique for making these estimates utilizes a direct and simple transforma-
tion from rainfall to runoff quantity and quality. It is a basic and very
effective method for long term load characterizations when applied with
reasonable estimates of the various model coefficients.
The specific techniques for estimating loads are presented here.
Detailed discussions of model coefficients and useful numerical estimates
are presented in Chapter 5.
3.4.1 Runoff Quantity
Stormwater runoff is generated by the rainfall on an area, which flows
across the surface and gravitates toward, the natural outlet from the area
either through natural drainage courses or through sewers or other collection
systems. The amount of runoff is related to the quantity of rain. Not all
3 -r 16
-------
rainfall reaches the outlet of the drainage area. Some percolates into the
ground or is retained in natural depressions. The net result is that only a
fraction of the rain falling on an area ultimately becomes runoff.
An important influence on the fraction of rainfall which runs off is the
degree of imperviousness of the area. The major effect is due to artificial
surface cover such as pavement, roof tops, sidewalks, and street surfaces.
Soil types in the other areas (i.e., sandy vs. clay type) also exert an in-
fluence, as do ground slopes and soil moisture. Regardless of the signifi-
cance of these factors, it is well to bear in mind that the predominant in-
fluence on the amount of runoff to be expected at any particular time is the
amount of rainfall. A convenient approach is to define the average runoff
to rainfall ratio for a study area:
Volume of Runoff ,,
~ Volume of Rainfall
This assumes a linear relationship between rainfall and runoff volume, and
assumes that this relationship is the same for all storms in the period of
record. The fraction of the rainfall volume which becomes runoff actually
varies from storm to storm, depending upon the antecedent conditions, the
storm intensity and patterns, etc. In using a single ratio of runoff volume
to rainfall (R,7) , the runoff volume is overestimated for some storms and
underestimated for others. The long term characterization of the storm
runoff properties (i.e., mean, variability, and frequency distribution) is,
however, well estimated, and the simple runoff to rainfall ratio is used in
other long term, initial assessment methodologies employing simulators (4) .
The best way to determine the runoff coefficient for a particular
study area is to compare raingage data with the runoff monitored during cor-
responding storms. Sufficient data of this type are often not available, and
in such cases estimates must be made based upon land use characteristics,
either from land use surveys of the drainage area, or inferred from the pop-
ulation density. Methods for estimating the average volumetric ratio of
runoff to rainfall (R^) based on drainage basin characteristics are presented
in Section 5.2.1. Ov
o
The average volumetric ratio of runoff to rainfall is used to convert
the rainfall statistics to runoff statistics. The mean runoff volume (V )
is determined:
VR = 0.027(RyVA) (3-12)
where :
V = mean runoff volume (million gallons, MG)
K
V = mean storm volume (in)
A = drainage area (acres)
0.027 = conversion factor (MG/acre-in) .
3-17
-------
To estimate the mean runoff flow (QR), a transformation from rainfall
intensity is required:
Q = RlACD/D (3-13)
R
where :
QR = mean runoff volume (cfs)
I = mean storm intensity (in/hr)
D = mean storm event duration (hr)
D = mean runoff event duration (hr)
R
The conversion factor from (acrein/hr) to (cfs) equals one. The term (D/DR)
is a correction factor to account for the attenuation of runoff beyond the
end of a rain event. This is particularly necessary for a large catchment
area with a long equilibrium time (the time it takes for the entire basin to
contribute runoff at a particular point in the receiving water) where D
will be somewhat larger than D. A method for estimating DR as a function of
drainage area size and the degree of urbanization, using unit hydrograph
analysis, is presented in Section 5.2.2.
The storm to storm variation in runoff volumes and flows is due primar-
ily to differences in the amount of rainfall, although variability in the
ratio of runoff to rainfall and the amount of runoff attenuation experienced
during storms, caused by changing antecedent conditions, may also contribute
variability. However, for most planning purposes, the estimate of runoff
variation may be based solely upon the variation of the measured rainfall
parameters :
vyR = vv (3-14)
vq = v. (3-15)
The procedure described does not specifically correct for runoff
attenuation due to depression storage in the drainage basin. Where the user
considers it important to take this into account in his analysis, the long
term effect of depression storage on runoff loads reaching the receiving
water can be estimated by using the treatment curves presented later (Figure
3-18) which define the level of control provided by storage basins.
Since the statistical analysis does not depend directly on the method
used to estimate runoff statistical properties more sophisticated models or
data for estimating the basic runoff parameters may be incorporated where
appropriate in the later stages of the planning process, without loss of
continuity.
3-18
-------
3.4.2 Runoff Quality and Resulting Loads
Runoff flows may be translated into stormwater loads by multiplying by
the appropriate pollutant concentration, c (mg/1). The concentration of
contaminants in stormwater runoff and overflows varies within storms, between
storms, and from location to location as a function of drainage basin char-
acteristics, land use, conveyance type, season, storm type. Since the
variation is often, random and unpredictable, it is useful to begin the load-
ing assessment in a particular area using a simple average pollutant concen-
tration (c, mg/1) in the runoff or overflows. The significance of concentra-
tion variation between and within storms will be presented as the methodology
is developed.
The best way to estimate average runoff pollutant concentrations is with
a wet weather monitoring program. When sufficient data of this type are
unavailable, however, estimates may be made based on drainage area character-
istics, although significant errors are possible due to the wide variability
observed in stormwater concentrations. Further guidance and estimates for
determining runoff quality are presented in Sections 5.3.1 and 5.3.2.
Once appropriate estimates of the average runoff concentration are made,
storm loads may be determined. Equations for estimating the mean runoff
loading rate during storms (W , Ibs/day), the mean load per storm (M , Ibs),
and the long term average mass discharge rate (W , Ibs/day) are presented in
this section. The relevance of each of these loading characterizations is
discussed in Chapter 2. Estimates for the variability and frequency distri-
bution of stormwater loads are also presented.
Advective streams and rivers are usually sensitive to instantaneous
loading rates of pollutants, and the analysis of transient stormwater impacts
thus requires a characterization at this time scale. If storm runoff flows
and concentrations are statistically independent, the mean runoff loading
rate, (W , Ibs/day), will simply equal the product of the mean concentration,
(c, mg/lj, and the mean runoff flow, (Q , cfs).
K
WR = 5.4 E QR (3-16)
where 5.4 is a conversion factor to make units consistent (Ibs/day)/(cfs-
mg/1) 1. It is reasonable to begin by assuming c and q are independent;
if data collected for specific pollutants indicate they are not, the
following refinement can be employed:
WR = 5.4 cQR (1 + vcvqpcq) (3-17)
where:
v = the variation of the pollutant concentration (between storms)
W
v = the variation of the runoff flow (between storms)
3-19
-------
v = the linear correlation coefficient between the pollutant
" concentration and the storm runoff flow (ranging from -1
to +1)
Positive correlation between the pollutant concentration and the storm
runoff flow will yield a higher average loading rate, while negative correla-
tion will yield a lower average loading rate. Positive correlations general-
ly occur where higher flows are necessary to cause a significant scouring of
suspended solids in the sewer system or to dislodge and transport solids
from the ground surface. Negative correlations may occur when the diluting
effects of the larger runoff flows dominate.
Equation 3-16 calculates mass loadings based on mean runoff concentration
(c), when there is no flow-concentration correlation. Where a flow-concentra-
tion correlation does exist for a particular study area, the correct mass
loading will be calculated either by equation 3-17 (in which c is the un-
weighted mean concentration), or by equation 3-16 when c is taken to be a
flow-weighted concentration.
The variation of the runoff loading rate at a particular site is due to
variation in both the flow and the concentration. An equation is available
for estimating loading rate variability as a function of v and v by
assuming flow and concentration are independent (2). Because the technique
for evaluating instream responses to runoff loads requires information on
both the pollutant load and the flow associated with it (except in large
rivers where the flow contributed by the runoff is insignificant compared to
the base flow already present), a consistent probabilistic analysis requires
that all the variability in the runoff load be associated with flow variation.
This is discussed further in section 3.4.2.1 on impacts in streams and rivers.
The coefficient of variation of the loading rate during storms (v ) is thus
estimated to be equal to the coefficient of variation of the runoff flow:
v = v (3-18)
w q
This is equivalent to assuming the runoff concentration is constant, as if
often done with simplified simulation models (4).
Highly dispersive receiving waters such as estuarine rivers and bays
require only that the total mass entering the water due to the storm runoff
be properly identified. The high degree of mixing makes them insensitive
to the actual loading rate which occurs during the storm. For these systems,
an estimate of the total mass discharged per storm and its variability from
event to event is sufficient. If storm runoff volumes and concentrations are
independent, the mean storm runoff load (MR, Ibs) is estimated as the product
of the mean concentration (c, mg/1) and the mean runoff volume (V , MG).
K
MR = 8.34 cVR (3-19)
where 8.34 is a conversion factor to make the units consistent [lb/(MG-
mg/1}].
3-20
-------
If storm runoff volumes and concentrations are not independent, a
correction similar in form to Equation 3-17 may be used to adjust the
estimate of MK. This correction requires large amounts of data to estimate
the correlation coefficient between storm volumes and concentrations with
statistical significance. When a first flush effect is present, however,
there is a negative correlation inherent between runoff concentrations and
storm volumes as shown below.
The first flush phenomenon is the condition, often occurring in storm-
water discharges, in which a disproportionately high fraction of the runoff
load is carried in the first portion of the discharge or overflow. The
pollutant concentration in the runoff at the beginning of a rainfall event is
relatively high, and as the rainfall continues, the subsequent runoff con-
centration decreases. The temporal profile of the pollutant concentration
during storms is approximated by an exponential decrease, as depicted in
Figure 3-5. Given this assumption, and the assumption that runoff durations
and flows are independent and exponentially distributed (v = v, = 1, and
v = /3) the correction in the estimate of M due to the first flush effect
can be calculated. The average runoff concentration (c) is calculated as
the expected value of time integral c(t) for the varying storm durations:
00 d , ,
c = E c(t) = / / £-^i Pd(d) dt dd (3-20)
d=o t=o
The result is:
c = CQ + |- (c c ) ln(l + DR/0) (3-21)
R P °
Where the calculated average concentration is between c and c , depending
upon the rate of subsidence of the first flush peak. This provides an
estimate of the time averaged runoff concentration observed at a site
(without flow weighting). The average runoff volume is the product of the
mean duration and runoff flow:
VR = D Q C3-22)
r\ i\ r\
The average runoff load is calculated as the expected value of the storm
load for the varying flows and durations from event to event; and the
decaying concentration profile within storms:
CO CO £[
M = E c(t)q d = / / / z±U- q d p (d) p (q) dt dd dq
q=o d=o t=o q (3-23)
This calculation accounts for the fact that storms of longer duration have a
lower average runoff concentration.
The result is: r~
MR -
c
o (DR/
(3-24)
3-21
-------
POLLUTANT
CONCENTRATION
Co
-C(t) = C0+ (Cp-C0)
TIME
FIGURE 3-5
IDEALIZATION OF FIRST FLUSH EFFECT
3-22
-------
The quantity of interest is M /cVR, which is the ratio of the actual
mean storm load to the storm load predicted by using c and VR. MR/cVR is
plotted in Figure 3-6 as a function of c /c , the ratio of the peak concen-
tration to the concentration after the fSrst flush has subsided, which in-
dicates the magnitude of the first flush; and DR/B, the ratio of the average
runoff duration to the first flush decay time, which indicates the rate of
first flush subsidence. The effect is seen to be generally small, with the
correction factor for M near 1.0 although it can approach 0.7 for large,
rapidly subsiding first flush effects. Note that these curves will be some-
what different when durations and flows are not independent and exponentially
distributed, i.e., v, / 1, v ? 1, and v R ^ vT), However the results in
Figure 3-6 are reasonable for a first estimate, and may be used to adjust the
estimate of M when a first flush effect is present.
K
Theoretically, a first flush should always exist, since materials will
accumulate on land surfaces and in sewer lines during the periods between
storms. Local factors, such as the size of the drainage basin and the
staggered time interval during which first flushes from different parts of
the basin reach the overflow or monitoring location, may suppress this effect.
Generally, local monitoring of a sufficiently large number of storm events
will be necessary to reliably characterize the first flush effect actually
present in a study area. However, as shown in Figure 3-6, the influence on
the estimate of the mean runoff load (M ) may not be of major importance.
The first flush effect also has an impact on the performance of storage
devices for stormwater control, as discussed in Section 3.6.1.2.3.
The variability of storm runoff loads is due to variation in both the
volume and the concentration. Because of the negative correlation sometimes
present between storm volumes and concentrations, it is reasonable to
associate all of the load variation with volume. The coefficient of varia-
tion of the runoff load (v ) is thus estimated to equal the coefficient
of variation of the runoff volume:
\ - VvR <3
The probability distribution of storm loads and loading rates is
expected to be similar to the distribution of volumes and flows. A gamma
distribution describes the frequency characteristics and Figure 32 may be
used to predict the fraction of storms and the expected number of times per
year, month, or season that a given storm load or loading rate is exceeded,
as outlined previously.
The average storm load (MR) and loading rate (WR) are representative of
pollutant loads during storm periods. The long term average mass discharge
rate, W , is calculated by determining the total storm load during the year
(pounds;, and assuming that it occurs continuously (during both rain and non
rain periods). If the period of interest is a particular month or season,
rather than the entire year, W may be calculated by determining the total
storm load during the particular month or season, and by assuming that the
storm load occurs continuously. W is also estimated from the mean storm
load (M ) as follows:
K
3-23
-------
>
It)
I 0
0.9
0.8
0.7
0.6
0.5
RATE OF FIRST FLUSH SUBSIDENCE
I
I
I
I
3456
Cp/C0
10
SMALL MODERATE LARGE
MAGNITUDE OF FIRST FLUSH
FIGURE 3-6
CORRECTION IN ESTIMATE OF MR WHEN FIRST FLUSH IS PRESENT
3-24
-------
W = WD/A (3-26)
O K
The total pollutant mass occurring over a long term period is the product
of the length of the period and W . For example, the average yearly load
(Y , Ibs/year) is:
Y = 365.25 (day/year) x WQ (Ibs/day) (3-27)
W is the loading rate which is used to assess the cumulative long term
stormwater effects and may be compared with the continuous municipal and in-
dustrial point source loadings to determine the relative magnitude of each
source. At least five years of raingage data, either for the entire year, or
during the particular month or season of interest, should be used to provide
adequate confidence in W .
For pollutants which impact the receiving water in a transient fashion,
such as coliforms or BOD5, the long term loading rate, W , may not fully in-
dicate the severity of the problem. For example, stormwater loads may con-
tribute only a small part of the total yearly BOD5 load entering a receiving
water in a particular area, but the occurrence of this load only during storm
periods may lead to violations of dissolved oxygen standards during or immed-
iately following a number of rain events. It is for such cases that the
actual mean storm load (M ), mean loading rate during storms (W ), and their
variations, v and v , respectively, become the important indicators of storm-
water loadings.
To demonstrate a typical loading table developed for a study area, assume
the following factors are determined for stormwater BOD5 loadings in the
summer (July - September).
Q_ = 10 cfs, v = 1.20, Vn = 2.2 MG
K q K
A = 85 hr = 3.54 days, c = 40 mg/1 BOD , Moderate first flush
The mean storm loading rate (W ) is
K
WR= 5.4 EQR
=5.4 (Ibs/day)/(cfs-mg/1) 40(mg/l) 10 (cfs)
= 2,160 Ib BOD5/day
The coefficient of variation of the loading rate (v ) is estimated to be:
v = v = 1.20
w q
The mean storm load (M ) calculated from equation 3-19 is:
R
MR = 8.34 cVR
= 8.34 lb/(MG-mg/l) 40(mg/l) 2.2(MG)
3 - 25
-------
= 734 Ibs BOD5
Correcting the estimate of M for a moderate first flush ef assuming D /g =
1.0, Figure 3-6 indicates M /cVR equals about 0.85. The corrected estimate
of MD is therefore:
K
M_ = 0.85(734)
K
= 624 Ibs BOD5
To calculate the frequency of occurrence of different storm flows and loading
rates, Figure 3-2 is used. These are translated into the expected number of
summer storms which will exceed a given flow and loading rate by knowing the
average number of storms per summer:
A *r u j: f* Length of Period 92 days _,
Average Number of Storms = = • _ . / = 26
6 A 3.54 days
The results are summarized in Table 3-3, together with the long term average
loading information. The probabilistic loading rate information (W, Ibs/day)
is appropriate for an advective stream or river, whereas a table constructed
for an estuarine area would include frequencies of storm loads (M, Ibs) in
multiples of the mean load (M ). The long terra average loading rate during
the summer (W ) is:
Wo - MR/A
= 624 (lb)/3.54 (days)
= 176 Ib BOD5/day
Table 3-3 summarizes the essential loading information necessary for
the instream response estimate.
3.5 Impacts in the Receiving Water
The quantification of stormwater related loads in an urban area provides
insight into the importance of runoff and overflows relative to other waste
loads. It is not, however, the final step in the assessment since contami-
nant loadings from urban runoff are evaluated in terms of their impact on
the adjacent receiving water. The range of water quality problems which may
be due to stormwater runoff is quite large. This section presents methods
and equations for making quantitative assessments of receiving water impacts
including both long term and transient effects. Different estimating
techniques are developed for streams, rivers, estuaries, lakes and impound-
ments. Guidance for selecting particular model coefficients and parameters
is presented in Section 5.4 of Numerical Estimates. Note that a basic
understanding of receiving water modeling is assumed for the users of this
manual. A more detailed description of mathematical water quality models
and their important components may be obtained from the following references:
1. Areowide Assessment Procedures Manual, Chapters 2 and 5 (2)
2. Simplified Mathematical Modeling of Water Quality (22)
3 -n 26
-------
TABLE 3-3
EXAMPLE STORMWATER LOADING TABLE FOR ADVECTIVE STREAM
(JULY - SEPTEMBER)
1. Loading during storms:
Flows
Q
(cfs)
5.5
12.0
17.0
25.0
35.0
QR = 10 cfs
v = 1.20
q
Q/QR
0.55
1.20
1.70
2.50
3.50
W_ = 2,160 Ibs BOD5/day
K
Loading Rate
(Ibs/day) W/WR
1190 0.55
2590 1.20
3670 1.70
5400 2.50
7560 3.50
v = 1.20
w
Percent
of Storms
Greater Than
(%)
50%
30%
20%
10%
5%
Expected Number
Summer Storms
Greater Than
(No.)
13.0
7.8
5.2
2.6
1.3
2. Long term loading rate
W =176 Ibs BOD5/day
3-27
-------
3. Mathematical Modeling of Natural Systems (23)
4. Systems Analysis and Water Quality Management (24)
3.5.1 Prediction of Long Term Impacts
Although stormwater loads occur intermittently in the order of hours,
their important receiving water impacts may be manifest over longer time
periods. As discussed in Chapter 2, this is particularly true for problems
such as sediment or nutrient buildups in lakes and impoundments, and the in-
crease in levels of persistent organics or metals. These problems may be
addressed by determining the portion of the average, steady state receiving
water concentration which is due to stormwater loads. The long term average
loading rate (W ) is the appropriate input for such an analysis. In addition
to these long term problems, some understanding of transient effects, such as
coliform bacteria increases and dissolved oxygen depressions, may be gained
by examining the average, long term contributions of stormwater runoff.
Again, a steady state analysis using long term average loads is appropriate.
Finally, results from steady state mathematical models may also be used to
infer information about the transient nature of these impacts, as will be
demonstrated in the following sections. Steady state representations are
particularly useful at the assessment stage because of the relative simplici-
ty of the calculation and the ability to respond rapidly and relatively in-
expensively to specific planning questions.
3.5.1.1 Streams and Rivers
The simplest type of receiving water is a one-dimensional flowing
stream or river where the mixing characteristics are such that the dispersion
of the mass of material can be neglected in comparison to the flow. In this
case, the river flow is the major mass transport mechanism. This simplifi-
cation is significant in terms of computational complexity and the amount
of information required for water quality analysis.
For a complete specification of stream responses to pollutant loads,
the initial concentration, the reaction rate, and the river flow and cross-
sectional areas are required. For some variables, there may be a coupling
effect where the solution of one equation feeds forward into a second equa-
tion and acts as an input. For example, the interaction between the bio-
chemical oxygen demand and dissolved oxygen is represented by a coupled set
of equations. A summary of the basic Streeter-Phelps solutions for steady
state pollutant concentrations in a stream is presented in Table 3-4.
Critical seasonal effects are estimated by assuming constant waste and stream
characteristics for the particular season. Concentrations are assumed to be
constant throughout the depth and across the width of the receiving water.
The receiving water geometry is, therefore, approximated by a series of con-
stant geometry and constant flow segments. The governing differential
equations for the receiving water concentrations are linear so that the
effects of the individual waste sources can be calculated separately and, at
a given location, added together to give the total instream concentration.
It is recommended for the first assessment that a single segment model
be used with a spatially aggregated stormwater load (W ). If more spatial
3-28
-------
•p
ol o
(T U
0)
z o
g z
i- >
o y
ro
UJ z
it
CD O *j
< < o:
f— |jj |
(f) UJ
o
u. z
o o
O
a.
-------
detail is required, it is recommended that the stream be segmented into a
maximum of five reaches. The purpose of limiting the segmentation of the
stream is to simplify the number of calculations required in the impact
analysis and to keep the level of detail of the impact analysis consistent
with the accuracy of the load estimation. In general, stream segments are
constructed for areas of approximately constant flow, cross-sectional areas,
depths, and velocities. Additional segments are formed at the location of
important point source load inputs. If less than five stream segments are
required for the particular basin, then the analysis is more manageable.
An important factor in the segmentation of the model is the effect of
the spatial detail of the stormwater load characterization on the accuracy
of the predicted instream response. Various levels of spatial detail which
may be employed are demonstrated in Figure 3-7. To limit the error in the
predicted downstream water quality concentration to 5 percent, loading, aggre-
gations should be limited to a distance X (miles), such that X = 0.05 -g,
where U is the stream velocity (miles/day) and K is the reaction rate (per
day). The resulting load is the sum of the individual loads.
Note that as the reactivity of the water quality parameter increases,
the distance over which load aggregation can take place is reduced. There-
fore, an analysis of coliform bacteria loads will usually result in con-
siderably smaller aggregation distances than an analysis of BOD or suspended
solids loads. Judgment should be used in aggregating loadings of conserva-
tive materials as there is a practical limit to the aggregation distance.
3.5.1.2 Estuaries and Coastal Waters
An estuary is that portion of a coastal river where the tidal action
from the ocean is a significant hydrodynamic parameter. There are two broad
sections of estuaries, the tidal river portion where the water body ebbs and
floods, but is entirely freshwater; and the lower estuarine portion where,
in addition to the ebbing and flooding of the tide, a significant intrusion
of sea salts occurs. One or two spatial dimensions (e.g., the longitudinal
and vertical dimensions) may be of importance in estuaries, although initial
assessments may simplify the problem to a one-dimensional analysis. The
primary difference between estuaries and the one-dimensional river flow
situation is the dispersive mass transport due to the tidal mixing. This
forms an important transport phenomena in addition to the net freshwater flow
through the estuary and is included in the analysis. The steady state
equations for pollutant concentrations in one dimensional estuaries are pre-
sented in Table 3-5. The additional parameter of interest is the dispersion
coefficient (E) due to the tidal action. The long term stormwater loading
rate (W ) is used to determine the average response of the estuary. The
variability of the response for transient impacts is addressed in Section
3.5.2.2.
Coastal waters such as tidal embayments and near-shore areas usually
require more complex analysis with two or three-dimensional specification of
geometry, hydraulic regimes, circulation, etc. Some simplified analysis
techniques for evaluating ocean outfalls and localized near-shore areas are
presented in Chapter 5 of the Areawide Assessment Procedures Manual (2).
3-30
-------
AGGREGATED
STORMWATER LOAD
RECEIVING WATER
STORM AND /
COMBINED /[
SEWER / L
OVERFLOWS-/
©
M ?
i
i!
ji -
i
i
i
i
i
i
i
i
L
©
3-1
r
r- /
1
©
L
\
3
i
" t
— ^>
"MUNICIP/I
1
TREATMENT
PLANT
LfGfHO:
Q SU8AREA NUMBER
TOTAL URBAN DRAINAGE
AREA BOUNDARY
IN
FIGURE 3-7
COMPARISON OF SPATIAL DETAIL
STORMWATER LOADING CHARACTERIZATION
3-31
-------
X
rt
T3
i
C
X
UJ
_J
m
g
^ o
o °-
o o
UJ
I-
co
UJ
CO
o £
u. <
CO
UJ
CO
H OC.
CO ,
Q ^
< H
UJ
CO
3-32
-------
3.5.1.3 Lakes and Reservoirs
Lakes and reservoirs can involve either two or three spatial dimensions.
The flow regime in these bodies of water can be quite complex since there are
usually no dominant mechanisms which determine the advective flow and mixing,
in contrast to the case of estuaries and rivers. The stratification which can
occur due to the absence of intense advective or mixing forces, complicates
the distribution of water quality constituents in a vertical direction. Thus,
lakes and reservoirs can encompass a broad spectrum of complexity, ranging
from completely mixed water bodies to highly stratified water bodies.
Initial assessments of long term stormwater impacts in lakes and
reservoirs may be made with a few simplifying assumptions. To determine the
average concentration of conservative or slowly reactive constituents (i.e.,
dissolved solids, persistent organics, etc.), the water body may be assumed
to be completely mixed. Equations for a large, completely mixed impoundment
are presented in Table 3-6. Because lakes with long detention times require
a long time period to reach steady state, equations for estimating the
pollutant buildup (or reduction after treatment) over time are also presented.
Note that these equations are appropriate only for an initial assessment, and
not for more detailed planning.
The eutrophication of lakes and impoundments due to excessive nutrient
load is a problem which has received a considerable amount of attention (25) .
Simplified techniques for estimating the potential for lake eutrophication
have been recently developed (26, 27) . These techniques are particularly
applicable for estimating the long term impact of stormwater related nutrient
loads. Because eutrophication of a water body is very complex, simplified
analysis techniques should be used with extreme caution.
The equation used for the model developed by Dillon (27) considers the
hydraulic flushing time, the nutrient loading, the nutrient retention ratio,
the mean depth, and the nutrient concentration of the impoundment:
L (1" R) = HN (3-28)
where :
L = nutrient loading divided by the surface area of the lake
(gm/m /yr);
R = fraction of nutrient retained;
p = hydraulic flushing rate (1/yr) ;
H = mean depth (m) ; and
N = nutrient concentration (mg/1)
The nutrient loading due to stormwater runoff is calculated from the total
yearly load (Y ). The graphical solution to this equation is a log-log plot
of L (lR)/p versus H. Figure 3-8 is a reproduction of Dillon's work on
3-33
-------
CONCENTRATIONS IN LARGE, COMPLETELY MIXED IMPOUNDMENT
Constant Load
W = W
Linearly Increasing
or Decreasing Load
W
W = W +wt
o—
Conservative
- Concentration
vs.
Time
C e
-fit
"*
- Steady State
Slowly Reactive
Note: a = ^ + K
- Concentration
vs.
Time
Loe
-at
W
-at o ., -at,
Loe + a7 C1 - e 5
±Ji- (at + e'at -
- Steady State
o
aV
Note
Q = Flow through Impoundment
V = Volume of Impoundment
TABLE 3-6
CONCENTRATIONS IN LARGE,
COMPLETELY MIXED IMPOUNDMENT
3-34
-------
100
001
10
I
50 IOO
MEAN DEPTH (METERS)
500
80
40
8.0
4 5
a:
2
in
a:
to
a:
i
0.8
045
0.08
1000
FIGURE 3-8
GRAPHICAL SOLUTION TO THE DILLON APPROACH
3-35
-------
several lakes in Canada, based on phosphorous loadings. The nitrogen axis
has been added to the graph based on a stoichiometric relationship of the
mass of nitrogen to phosphorous in algae. This relationship does vary; a
range of 3 to 15 mgN/mgP has been reported (28) with an average of 8 being
typically used. Lakes or impoundments which fall above the 20 vgP/1 or 160
ygN/1 concentration for total phosphorous or total inorganic nitrogen re-
spectively tend to be eutrophic while those below the 10 pgP/1 or 80 PgN/1
concentration line for total phosphorous or total inorganic nitrogen respec-
tively tend to be oligotrophic. Dillon's work should be referred to for a
more detailed description and development of the method.
3.5.2 Prediction of Transient Impacts
Steady state mathematical models are useful for determining the long
term average concentration of pollutants in the receiving water. For many
water quality problems, however, this information is insufficient for a
complete evaluation. Significant stormwater impacts leading to violations of
receiving water standards and criteria may only occur during, or immediately
following, storms. A method is needed for estimating the variability of the
receiving water response and the frequency with which stormwater related
problems occur.
The most direct method for evaluating the variation of receiving water
quality is with a time variable simulation model. The hourly (or any other
suitable time interval) stormwater flows and loads are input into the model,
and the resulting pollutant concentration is calculated for each hour during
the period of interest, such as a season or a given year. The next step is
to statistically evaluate the continuous temporal concentration profile cal-
culated by the model to determine its mean, variability and frequency
characteristics. Because time variable receiving water simulations are
complex and costly, methods have been developed for directly estimating the
pertinent characteristics of the receiving water response from steady state
models using information on the mean and variability of the stormwater loads.
3.5.2.1 Streams and Rivers
Streams and rivers are characterized by a predominantly advective
transport. Storm loads from an urban area enter the river and are transport-
ed downstream. In the idealized case, there is no interaction between storm
events in the river, and the response to each storm may be calculated indepen-
dently of any other event. The frequency distribution of instream concentra-
tions is thus directly related to the frequency characteristics of the storm-
water loadings.
In the special case of constant flow advective systems, the variability
characteristics of the response function as a function of load variability
have been investigated (24). In particular, it can be demonstrated that the
coefficient of variation of the water quality response at any location is
equal to the coefficient of variation of the input loads. Thus, knowing the
mean load and its variability, one can compute the mean response using a
steady state water quality model and then calculate the variability of the
water quality response based on the variability of the load. This is a valid
3-36
-------
and recommended approach for analyzing variable load impacts on streams where
the constant flow assumption is reasonable. However, in situations where
intermittent storm related loads are important, the impact of the runoff on
the advective flow is often a major factor. A calculation of the impact of
each storm event, with a defined frequency of occurrence, is thus required.
The concept is shown diagrammatically in Figure 3-9. Two loads are
considered: a continuous steady state load and an intermittent load. The
continuous loading rate is characterized completely by its mean, W, the in-
termittent loading rate is characterized completely by its mean, W , its
coefficient of variation, v , and its probability distribution function.
The runoff flow associated with each of the loadings is also required. The
instream concentrations calculated by the water quality model are peak con-
centrations which pass a particular location during or following a given
storm. The 90th percentile receiving water response is induced by the 90th
percentile storm loading and flow, determined by a loading table such as that
shown in Table 3-3.
This is a considerably simplified representation of the probabilistic
nature of pollutant concentrations in rivers and streams. The frequency
distribution of instream concentrations is in reality also affected by
variations in the base flow (which may be partially correlated with storm
events, depending on the size of the upstream drainage area and areal rain-
fall patterns) and temperature; both of which are assumed to be constant in
this simplified analysis. Variations in these and other factors may be
included in a sophisticated continuous simulation by incorporating them as
stochastic inputs. For initial planning studies, however, it is felt that a
simplified representation based on the frequency characteristics of storm
loads and flows provides an adequate basis for estimating and assessing
stormwater impacts.
There are two steps in determining the impacts of a particular storm.
First, the steady state Streeter-Phelps equations are used to estimate the
spatial profile which would result if the given storm load and flow occurred
continuously. If there is no dispersion in the river, this would be the
concentration observed at each location as the storm pulse passes (23). The
second step is to adjust the result to account for attenuation of the pulse
due to dispersion.
To understand the effect of dispersion, one may first view the stream
as a purely advective, plug flow system. If such a system is loaded with a
series of pulse loads of a conservative tracer as indicated in Figure 3-10(a),
measurements of a downstream point would yield a series of pulse responses as
indicated in Figure 3-10(b). The time between the measured pulses and their
magnitude would be directly related to the characteristics of the input load-
ing function and the pertinent stream characteristics such as river flow,
channel characteristics, etc.
In natural water systems, there is normally some longitudinal mixing
taking place as the pulses move downstream. The effect of such mixing, or
dispersion as it is commonly called, is to spread the pulses out, as indica-
ted in Figure 3-10(c). The effects of longitudinal dispersion on wet weather
3 r 37
-------
90%
TIME
SAMMA DISTRIBUTED]
10% LOAD
50% LOAD
90% LOAD
DISTANCE
10% LOAD
X
o
o
Ul
o
-------
1 4°
V)
2
D
0
CX. c. 0
Q
z
^ 1 0
(rt
o
Ir\
(a)
—
t- 0
o> 4
E
2 ^
o 3
h-
h-
2
0 1
2
0
/"v
Dill <
~C/t-*3
r-
C
f S) Af)
.tsftts
A
!
z
B
1
3
1
4
c
1
5
D
6
7
TIME-DAYS
ADVECTIVE STREAM
-
0
(NO
U= 2
D
1
I
@ T
= 7
DAYS
DISPERSION)
MILES/DAY
2 3
C
1
4
|
5
I
6
1
7
B
1
8
1
9
1
10
1
A
12
1
13
4
DISTANCE-MILES
en 4
E
0 3
-------
receiving water quality can be quite dramatic. Under certain conditions
peak concentrations in a storm related pulse can be attenuated by 30-60%
within 15 miles of the point of stormwater discharge. Figure 3-11 presents
a simple graphical solution for determining the degree to which model results
should be corrected to account for dispersion. Figure 3-11 was developed
based on simulation calculations with and without dispersion (29) . The
figure indicates the reduction in peak concentration as a function of a
dispersive transport factor, a:
C3-29)
va u
r
where :
E = longitudinal dispersion coefficient (milesz/day)
t = time of travel from the discharge point (days)
d = duration of the runoff event (hours)
U = river velocity (miles/day)
An a is computed for discrete distances downstream using the time of
travel to that point from the discharge location, the average stream
velocity through the river segment, and an estimate of the instream long-
itudinal dispersion. Typically, dispersion coefficients for streams and
rivers vary between 0.01 and 1.0 miles2/day. The site specific value is
dependent upon a number of factors which influence velocity gradients. For
example, the existence of impoundments or dense aquatic weed growths lead to
high dispersion coefficients while narrow, free flowing streams generally
have low dispersion coefficients. The best method for determining the dis-
persion characteristics of a specific stream is through the analysis of dye
study results (30) . A technique for estimating E in the absence of dye
studies is presented in Section 5.4.1.2. As an example of the use of
Figure 3-11, assume the following stream characteristics:
E = 0.3 miles2/day
t = 10 hours = 0.42 days
d = 3 hours = 0.125 days
U = 10 miles/day
2Et
- 2(0.3) (0.42) _
" ~ U-
From Figure 3-11, a = 0.16 corresponds to a reduction in the peak concentra
tion of a 3 hour loading pulse to 78% of its initial value at the point of
discharge. This will occur a distance:
3-40
-------
10
0.9
bf 0.8
0.7
0.6
05
04
03
02
01
NOTE:
E=DISPERSION COEFFICIENT (mlz/DAY)
U=STREAM VELOCITY (ml/DAY)
dr=DURATION OF STORM (DAYS)
t = TIME SINCE BEGININ6 OF STORM (DAYS)
X = U(t-dr/2)
= DISTANCE DOWNSTREAM FROM STORM LOAD (ml )
I I I I i
I I I I I I
I I I I I
0.05
01
02
0.5
1.0
DIMENSIONLESS FACTOR =
2.0
2E1
d2 U2
5.0
10.0
FIGURE 3-11
EFFECT OF DISPERSION ON POLLUTANT
CONCENTRATION AT MIDPOINT OF STORM PULSE
3-41
-------
x = U(t - d /2)
= 10 (0.42 - 0.125/2)
= 3.6 miles downstream of the loading point.
The concentrations calculated represent the maximum concentration of
each water quality indicator that will occur as the diluted and dispersed
pulse load moves downstream. This minimum concentration is calculated for
each storm with a given frequency of occurrence. Note that for a direct
mapping of the storm frequency onto the stream imput frequency, the resultant
pollutant concentration in the river must increase monotonically with larger
storm sizes. The applicability and limitations of this assumption are dis-
cussed in Chapter 7 where a detailed example of the computation of probabil-
istic water quality in streams using the statistical method is presented for
Salt Lake City.
3.5.2.2 Estuarine Systems
The high degree of mixing in estuarine systems makes the separate
analysis of individual storm events inappropriate. The effects of previous
storms may still be prevalent when the current storm occurs, and the impact
of each of the storms must be superimposed to determine the total stormwater
response. This may be accomplished with a time variable simulation of the
loadings and system response.
It is possible however, to estimate the mean and variability of pol-
lutant concentrations in an estuary directly, without continuous simulation
(18). The equations are derived from the response shape of a single loading
pulse to an advective-dispersive system, and the assumption that these pulses
occur as a Poissen process. The mean and standard deviation of the concen-
tration are estimated as:
2 2
CTm R Ux vh Uxm
a = exp -^= K —=—
c , /X „. zh o h
where:
m = /I + 4KE/UZ
M = mean storm load (Ib/storm)
K
a = standard deviation of storm load
m
U = freshwater velocity = (freshwater flow)/ a
A = cross sectional area
3-42
-------
E = dispersion coefficient
K = reaction rate
x = distance from storm load (The sign in Equation 3-30 is
negative when x >_ 0, positive when x < 0. Positive x
is in the direction of freshwater flow).
A = mean time between storms
K = modified Bessel function (Note K (-b) = K (b) ). Figure
3-12 or tables of modified Bessel functions may be used
for evaluating this term (31).
The solution for the mean concentration is the same as the steady state
solution which would be calculated using the long term average loading
rate, W = M /A. This is reasonable given the fact that the total mass of
pollutant entering the estuary is the same, whether it occurs as discrete
pulses, or evenly distributed in time. The standard deviation of the re-
ceiving water response increases as the storm loads increase (increasing M )
and more variable (increasing o ). The variability within tidal cycles is
not included in this analysis.
The simplicity of Equations 3-30 and 3-31 is remarkable. The equations
allow the simulation procedure to be by-passed in the calculation of c and
To illustrate the results of this calculation, an example using storm-
water coliform loads into a simple, one dimensional model of the lower Hudson
River Estuary is summarized in Figure 3-13. The theoretical results are
calculated using Equations 3-30 and 3-31. The simulator results are from an
hourly simulation using Central Park raingage data transformed into runoff
loads. Note the close agreement between the theoretical and the simulated
results. The highly variable nature of the response is reflected in the fact
that the standard deviation is from one to two times the mean concentration.
The final step in the analysis is to determine the frequency distribution of
the estuary response to predict, for example, what the 90th percentile
coliform concentration is at a given location. Further research is currently
under way to examine this problem.
3.6 Assessment of Stormwater Control Alternatives
Once the magnitude of stormwater impacts on receiving water quality are
estimated, control strategies for the reduction of these impacts may be
analyzed. A variety of stormwater control alternatives are available.
These are generally grouped into two types of approaches:
1. Structural, end-of-pipe treatment devices.
2. Management practices.
Structural, end-of-pipe alternatives include devices which capture and store
runoff, such as interceptors and retention basins, and devices which reduce
the pollutant concentration of runoff or overflows, such as screens, filters,
concentrators, and disinfection systems. Management practices include source
3-43
-------
10
5x10"
10
-2
5x10
- icT'
o
X
5x I 0":
ICT
5x10"
10"
5x10"'
10"'
I I
NOTE:
K0(0)=
\
\
23456789 10
b OR-b
FIGURE 3-12
GRAPHICAL SOLUTION TO K0 BESSEL FUNCTION
-------
100,000
50,000
E
O
O
10,000
< 5,000
cc
Ul
O
2
O
o:
o
u_
o
o
<
o
1,000
500
100
0 - IO.OOOCFS
o = 150,000 ft2
k = 2/day
E = 15 mi 2/day
A = 1.55 days
VR=103 Ibs
crm= 106 Ibs
(10* MPN/IOOml = I mg/l )
THEORETICAL
MEAN
STANDARD DEVIATION
SIMULATED
•-MEAN
D-STANDARD DEVIATION
LOCATION OF STORMLOAD
I
-12
-10
-8
-6
-202
MILEPOINT
10
12
FIGURE 3-13
COMPARISON OF THEORETICAL AND SIMULATED
ESTUARY RESPONSE TO STORMLOADS
HUDSON RIVER, RAINGAGE 305801, JULY-AUGUST 1969, HOURLY RECORDS
3-45
-------
controls, such as street sweeping and erosion control; and collection system
management techniques, such as sewer flushing and polymer injection to in-
crease the flow capacity of the sewerage system.
A brief description of each control alternative is presented in this
section together with quantitative methods for estimating their effective-
ness. The basic statistical properties of the runoff loads and the charac-
teristics of the treatment alternatives are used to determine modified
stormwater loads to the receiving water. The projected improvement in re-
ceiving water quality due to the modification of stormw?ter loads then rep-
resents the benefit of potential stormwater control actions.
3.6.1 Structural Treatment Devices
Stormwater control devices may be constructed to provide a given level
of treatment for a fixed runoff flow, storm duration and influent concentra-
tion. Treatment performance will change, however, as the storm runoff
characteristics vary from storm to storm. A statistical method of analysis
is described in this section, which focuses on the determination of the long
term performance efficiency of devices subjected to the varying rainfall-
runoff process.
The structural control devices considered are grouped into two basic
categories: (1) those which capture and store runoff, and (2) those which
reduce the pollutant concentration of the stormwater. The first group is
typified by interceptors and retention basins or tanks.
The operation of interceptors and storage devices is depicted in Figure
3-14. The storm runoff process is represented as a series of independent
events, as shown in Figure 3-14(b). The interceptor removes a constant flow
rate, Figure 3-14(c), the storage device captures a fixed volume, Figures
3-14(d), and the combination of interception and storage removes a constant
flow rate and captures a fixed volume of the interceptor overflow (Figure
3-14(e)). The unshaded areas in Figure 3-14 represent the uncaptured portion
of the storm runoff. This provides the basic theoretical framework for the
analysis of the long term performance of devices which capture and store
runoff.
3.6.1.1 Interception
An interceptor captures up to a flow rate, Q , the available treatment
plant capacity. Thus, Qt is the total capacity minus the dry weather flow.
The portion of the runoff in excess of Q overflows into the receiving water
through established relief points in the system. The performance of the
interceptor is important because it captures a portion of the runoff which
may subsequently receive treatment; either at the municipal treatment plant
of a combined sewer system, or at special stormwater treatment facilities for
separate storm sewer systems.
The long term fraction of the runoff load, f , not captured by an inter-
ceptor is calculated as the expectation of the runoff load that overflows
[cd(q-Q )] divided by the total runoff load:
3-46
-------
a) VARIATION WITHIN EVENTS
n
rrnfUi
TIME
b) VARIATION BETWEEN EVENTS
n n
dr |
>
^^-vr
n
i — i
TIME
c) INTERCEPTION
TIME
d) STORAGE
.n
TIME
e) INTERCEPTION AND STORAGE
TIME
FIGURE 3- 14
REPRESENTATION OF
STORM RUNOFF PROCESS, INTERCEPTION AND STORAGE
3-47
-------
VQ
/ (q-QT)d pr
. d=o L c
j(d) p (q) dd dq
5QRD
where M is the mean overflow load per storm, and p, (d) and p (q) are the
probability distribution functions of storm durations and flSws respectively,
which are assumed to be independent and gamma distributed. The integrals in
Equation 3-32 are evaluated numerically, and the results are shown in
Figure 3-15. The fraction of the runoff load which is uncaptured, f , is a
function of the normalized interceptor size, Q /Q , and the coefficient of
variation of the runoff flows v . The greater the variation in runoff flow,
the more poorly the intcrceptor^performs on average. Note the diminishing
increases in the amount of runoff captured for each increment in interceptor
size.
The use of Figure 3-15 may be demonstrated with an example. Assume
a drainage area with an interceptor has the following characteristics:
Mean runoff flow = Q = 10 cfs
K
Variation of runoff flow = v =1.15
q
Available interceptor capacity = Q = 12.5 cfs
For Q-r/Qp - 1.25, and interpolating between the curves for v =1.00 and
v =1.25 for the case when v = 1.15, Figure 3-15 indicates^that the
fraction of the runoff load n8t captured is: f = 0.33.
The analysis has assumed that the variation of flow within storms is
small compared to the variation of flow between storms. When this is not
the case, the performance level of the interceptor will be further reduced
beyond that shown in Figure 3-15. The analysis has also assumed that the
concentration of the runoff is independent of the flow. If this is not the
case, and higher flows tend to have higher concentrations, the interceptor
will perform more poorly than predicted in Figure 3-15. If higher flows
tend to have lower concentrations, the device will perform better than pre-
dicted in Figure 3-15. Analysis of runoff data have thus far indicated that
these effects are not of major importance for a first estimate.
3.6.1.2 Storage
A storage device captures up to a capacity volume, V , and the remaining
flow from the storm is by-passed. The captured runoff may then be discharged
at greatly reduced flow rates, with or without treatment, to the receiving
water, or pumped to the interceptor for treatment at the municipal or indus-
trial treatment plant. The storage capacity allows a significant reduction
in the size of the treatment facilities required and provides a reduction in
the magnitude of the shock load delivered to the receiving water.
3-48
-------
i o
05
I 0
.
OR
20 30 40
EXCESS INTERCEPTOR CAPACITY
MEAN RUNOFF FLOW
- 90
100
50
tE
O
Q.
UJ
O
o:
UJ
CO
Q
Id
CE
CL
<
O
z
UJ
o
-------
3.6.1.2.1 Effect of Previous Storms
The total storage may not always be available at the beginning of a
particular storm. The basin may still have leftover stored runoff from pre-
vious storms. The storage that is available on the average, termed the
effective storage capacity, Vp, will determine the long term performance of
the retention basin. The effective storage capacity, V , is a function of
the actual size of the basin, V , the mean runoff volume, V , the rate at
which the basin is emptied, ft, and the average time between storms, A.
The derivation of the solution for the effective storage capacity is
outlined in Figure 3-16. Storm 1 is assumed to begin with the long term
effective storage capacity available. It rains and a volume, v, further
fills the basin. Between storm 1 and storm 2 the tank is emptied at a rate
fi. The basin then has an available storage capacity of V at the beginning
of storm 2. The problem is to find the expectation of V eover the possible
values of v and 6. This expectation of V is the long term effective volume
of the basin, V e
The integral of Figure 3-16 is solved for the special case when v = v,
= v. = 1 (i.e., runoff flows durations, and the time between storms are ex-
ponentially distributed and independent). The results, normalized by the
mean runoff volume, V , are displayed in Figure 3-17.
R
This analysis provides a useful guideline for estimating the effect of
previous storms and determining an adequate treatment rate for the storage
volume. The expression Afi may be thought of as the average captured volume
processed (i.e., emptied from the tank) between storms. The smaller this
value is relative to V , the more likely the basin will still contain left-
over runoff with a storm begins, and the effective storage capacity, V , is
lower. When Aft/VR > 2, there is very little loss in effective storage
volume. When Aft/V < 2, however, the effective storage drops rapidly. The
drop is more pronounced in large basins where large storms may be accumulated
rather than by-passed or overflowed.
Note that some error in the estimate of the effect of previous storms
may be introduced due to deviations from the assumptions used to derive the
curves of Figure 3-17; for example, the coefficient of variation of storm
volumes may not equal /3~ as assumed in the calculations, or the actual
storage device may be operated with a variable emptying rate. The estimate
is still useful, however, for an initial assessment and screening of storage
treatment.
3.6.1.2.2 Storage Effectiveness
Once the effective storage capacity has been determined, the analysis
of basin performance may proceed. The long term fraction of the runoff load,
fv, not captured by a storage device is calculated as the expectation of the
by-passed runoff load, cq(d-V /q), divided by the total runoff load:
3-50
-------
VOLUME
Ve
STORM
STORM
2
TIME
FIGURE 3-16
DERIVATION OF SOLUTION FOR
EFFECTIVE TANK VOLUME
-------
UJ
•5
=>
-1
O
UJ
IE
O
1-
co
Ul
I-
o
UJ
u.
u.
UJ
LU
(E
Z
Ul
JK
1.0
2.0 3.0 4.0
STORAGE VOLUME (EMPTY)"
MEAN RUNOFF VOLUME
0.75
5.0
IE
O
UJ
UJ
ui
00
Ul
en
UJ
O
O
IE
Q.
u.
u.
O
z
O
>
(E
UJ
FIGURE 3-17
EFFECT OF PREVIOUS STORMS
ON LONG TERM EFFECTIVE STORAGE CAPACITY
-------
c 1 1 q(d-V /q) p (d) p (q) dd dq
. M q=0 d=V_/n q
fV = M~ = — ^ (3-33)
5QRD
where M is the mean by-passed load per storm. Equation 3-33 is evaluated
numerically, and the results shown in Figure 3-18. The fraction of the run-
off load which is not captured, fv, is a function of the normalized storage
capacity, V /V , and the coefficient of variation of the runoff volumes, v
The use of volume statistics makes the curves applicable to situations in
which runoff flows and durations are not independent, as is often the case.
Note that for a given coefficient of variation, the curves in Figure 3-18 are
very similar to the curves in Figure 3-15. This suggests that there is a
similarity between a random variable which is gamma distributed (runoff flows)
and a random variable which is the product of two independent, gamma dis-
tributed, random variables (runoff volumes). The assumption that storm run-
off volumes are themselves gamma distributed is probably quite adequate (16).
3.6.1.2.3 First Flush Effect
The curves in Figure 3-18 are developed without consideration of a first
flush effect. Therefore, the curves actually represent the fraction of the
runoff volume captured, rather than the runoff load. Adjustments should be
made to account for the first flush effect when it exists. Because a first
flush effect results in a disproportionately high fraction of the runoff
load in the first portion of the discharge or overflow, a correspondingly
high fraction of the runoff load is captured by a storage device.
The improvement in storage device performance associated with the first
flush effect is evaluated by assuming that the temporal concentration profile
of the runoff has an exponential shape as shown in Figure 3-5. The results
of this analysis are simplified and presented in Figure 3-19, which may be
used to adjust the performance curve selected on Figure 3-18, with the cor-
responding new values of f . The magnitude of the first flush effect to be
used in Figure 3-19 may be estimated from the ratio of the peak pollutant
concentrations generally found at the beginning of storms, c , to pollutant
concentrations observed after the first flush subsides, c . HTable 3-7 pro-
vides general guidelines for estimating the magnitude of ?he first flush
effect. As previously stated, local monitoring of a sufficiently large
number of storm events will be necessary to reliably characterize the first
flush effect actually present in a study area. However, as shown in Figure
3-19, the influence on device performance may not be of major importance.
The analysis of the first flush effect assumes that the storage device
is operated in a by-pass mode, that is, by-passing the later storm runoff
flows after the device is full. The device may also be operated in an over-
flow mode; accepting all storm runoff flows in one end, and overflowing from
the other when the basin becomes full. This will negate the storage improve-
ments related to the first flush effect, however, some treatment may be pro-
vided within the storage device, such as sedimentation, which may make the
overflow mode more favorable. These factors should be considered and weighed
when designing a storage-treatment system.
3-53
-------
— 08 -
Q
UJ
CC
:D
t-
Q.
<
CC
£ CGEND
O-CHIPPEW4 FALLS,
D - DURHAM, J/-- I 41
0 I -
0 5
I 0
Ib 2O 25 3O 35 40
EFFECTIVE STORAGE CAPACITY
MEAN RUNOFF VOLUME
4.5
50
FIGURE 3-18
DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE
3-51
-------
MAGNITUDE OF FIRST FLUSH EFFECT:
NONE
SMALL
MODERATE
LARGE
03 OA O5 Q6 a?
fv (NO FIRST FLUSH)
[FRACTION OF LOAD NOT CAPTURED]
[(ASSUME NO FIRST FLUSH EFFECT)J
FIGURE 3-19
IMPROVEMENT IN LONG TERM STORAGE DEVICE PERFOMANCE
DUE TO FIRST FLUSH EFFECT
3-55
-------
TABLE 3-7
GENERAL GUIDELINES FOR ESTIMATING MAGNITUDE
OF FIRST FLUSH EFFECT
Ratio of Peak to
Final Concentrations
(c /c )
v p' o^
Magnitude of
First Flush Effect
1.0 - 1.5
Small
1.5 - 4.0
Moderate
> 4.0
Large
Assumed D /B =1.0
K.
3-56
-------
3.6.1.2.4 Treatment of Stored Runoff
Once the fraction of the runoff load captured by a storage device has
been determined, the effects of treating the stored portion of the runoff
(off-line treatment) may be incorporated. Treatment may occur either through
settling or reaction in the basin itself or by pumping the stored stormwater
through a treatment device. The treatment rate should be controlled to maxi-
mize the overall treatment benefits. A lower treatment rate is desirable
to improve the pollutant removal and to attenuate the release of stormwater
into the receiving stream. If the rate is too low, however, the effective
storage capacity may be reduced, as described previously. Assuming the
captured runoff is treated with a percent removal (r), the modified average
runoff load (M*), will be:
K
M* = fyMR + (1 - fv)MR(l - r) (3-34)
Again, note that this modified load will enter the receiving water over a
longer time period than the original storm runoff.
3.6.1.2.5 Example of Storage Device Evaluation
The determination of storage device performance may be demonstrated with
an example. Assume a drainage area served by a storage device has the fol-
lowing characteristics:
Mean runoff volume = Vn = 4 MG (.536 x 103 ft3)
K
Variation of runoff volume = v _= 1.75
vR
Mean runoff load = MD = 2,000 Ib BOD
K
First flush effect = moderate
Average time between storms = A = 84 hr = 3.5 days
Storage volume (empty) = V_ = 6 MG (.804 x 10b ft3)
D
Emptying rate from storage = tt = 3 MGD (4.6 cfs)
Percent removal of BOD5 in stored runoff = r = 50%
To determine the effective storage capacity, the ratio of the average volume
processed between storms to the mean runoff volume is calculated:
/„ = (3.5 days) (3 MGD)/(4 MG) =2.63
K
The ratio of the empty storage volume to the mean runoff volume is:
VB/VR = (6 MG)/(4 MG) = 1.5
Figure 3-17 is then used to determine the effective storage capacity:
3-57
-------
VVR s
VE = 5.4 MG
The long term performance, assuming no first flush effect, is determined
from Figure 3-18. For V /V = 1.35 and v R = 1.75, Figure 3-18 indicates
the long term fraction of the runoff not captured:
fv = 0.45
The improved treatment due to the moderate first flush effect is then incor-
porated, using Figure 3-19. Adjusting fv = 0.45 for a moderate first flush
effect, Figure 3-19 indicates the revised fraction of the runoff load not
captured is:
fv = 0.35
Given the average runoff load M = 2,000 Ib BOD5, and the percent BOD
removal of the captured runoff load r = 50%, the modified average runoff
load is:
MR =
= 0.35(2,000) + (0.65)(2,000)(0.50)
= 1,350 Ib BOD5
To demonstrate the long term average pollutant reduction achieved by the
storage treatment system, the yearly stormwater BOD5 load (Y ) may be calcu-
lated with and without the control:
Before storage:
Y = (MD/A)(365.25 days/year)
HI R
= (2,000 Ib BOD5/3.5 days)(365.25 days/year)
= 208,700 Ib BOD5/year
After storage:
Y = (MS/A) (365.25 days/year)
m K
= (1350 Ib BOD5/3.5 days) (365.25 days/year)
= 140,900 Ib BOD5/year
A 32% reduction in the long term average BOD5 load is indicated. In addition,
because of the storage device, part of the load discharged to the receiving
water can be spread over an extended period, rather than just during or
immediately following storm events.
3-58
-------
3.6.1.3 Interception and Storage
Interceptors and storage devices may be operated in combination. The
interceptor captures up to a flow rate, CL. The overflow from the interceptor
is stored until the basin capacity is reached, after which overflows from the
interceptor are by-passed. This is equivalent to the "storage-treatment"
system commonly analyzed with simulators. The treatment component can be
thought of as available interceptor capacity.
The long term fraction of the runoff load, f , not captured by an
interceptor in combination with a storage device is calculated as the expec-
tation of the runoff load both overflowed and by-passed, divided by the
total runoff load:
CO OO
c / / (q-QT) d-V /(q-Q ) p (d)p (q)dd dq
M I=QI d=V /(q-Q ) q
fiv = r = l : (3-35)
R c QRD
Equation 3-35 is evaluated numerically and the fraction of the runoff load
not captured by both the interceptor and the storage device is well estimated
by the product of the individual fractions remaining for each device:
fiy = fjfv (3-36)
Equation 3-36 is exact when v = 1, slightly underestimates fTV when v > 1,
and slightly overestimates f ^ when v < 1. The error of the estimate"is
small compared to the overall uncertainties involved in stormwater treatment
analyses.
The effect of treating the captured portion of the runoff may again be
incorporated for the combination interceptor-storage device system. This is
typified by a combined sewer system where the interceptor routes runoff to
the municipal treatment plant and storage is added to capture overflows which
are subsequently pumped back to the interceptor and the treatment plant.
Assuming the treatment plant provides a percent removal, r, for both flows
captured by the interceptor and for stormwater retained by the storage device
and subsequently returned, the modified runoff load, M* is:
K
MR = fIVMR + (1 - fIV)MR(1 - r) (3-3?)
This assumes that the percent removal (r) obtained at the treatment facility
is similar for both the storm flows which reach it during the rainfall event
and for the captured stormwater returned later at controlled rates. This is
not usually the case, however, and it may be prefereable to analyze such a
system as an in-line treatment device whose percent removal decreases with
flow, as will be described in the following section.
3 - 59
-------
3.6.1.4 In-Line Treatment Devices Which Reduce Pollutant Concentration
A number of treatment devices have been applied as control measures to
reduce the concentration of stormwater overflows and runoff. The pollutant
removal efficiency will vary from one treatment device to another, and may
also vary in any one unit operation due to variations in the flow rate and
the influent waste load characteristics. In general, as more flow passes
through a device of a given size, the efficiency of removal will decrease.
Furthermore, the removal efficiency (expressed as percent removal) of some
treatment devices, in particular screens and sedimentation tanks, will be en-
hanced when the influent suspended solids concentration increases.
The performance of most treatment devices may be approximated by an ex-
ponential decrease in the percent removal as the flow increases. Most treat-
ment devices cannot be subjected to an excessive large flow rate, however,
and runoff beyond a certain flow rate must be by-passed. Hydraulic capacity
limitations as well as process considerations will determine the maximum flow
accepted. The effect of this by-pass may be approximated as a continued
exponential decrease in the overall treatment efficiency (including processed
and by-passed stormwater). This is demonstrated for dissolved air flotation
in Figure 3-20. The assumption that the percent removal decreases exponent-
ially as the flow increases is an approximation and may not describe the
performance of some devices as well as it does others; however, this approach
is a useful simplification and is reasonable for initial assessments.
For an exponential performance curve the relationship between flow and
percent removal is described analytically by the following equation:
r(q) = Z exp(q ln(|)/QR) (3-38)
where:
r(q) = percent removal as a function of q
q = influent flow
Qn = mean runoff flow
K
Z = best percent removal obtainable at very low flows
F = percent removal at the mean runoff flow, Q ,
K
The relationship is shown graphically in Figure 3-21.
The long term average reduction, P , obtained by a treatment device
which is subject to varying runoff flows is calculated from the expectation
of the portion of the load remaining:
OO CO OO
1-P_=— / / / l-r(q) qdc p (q) p, (d) p (d) dq dd dc (3-39)
r M_ , qdc
R q=o d=o c=o
3-60
-------
100
UJ
a:
40 -
20 -
J I
NOMINAL DESIGN FLOW RATE 1000 GPD/ft*
SS WITH CHEMICALS
BOD5 WITH CHEMICALS
SS
BOD5
REMOVALS BASED ON
BYPASS AT 0^>45 0 DESIGN
10
FLOW/DESIGN FLOW
FIGURE 3-20
DISSOLVED AIR FLOTATION PERFORMANCE
RELATED TO DESIGN RATE
3-61
-------
UJ
cr
UJ
o
cc
UJ
OL
INFLUENT FLOW, q
FIGURE 3-21
IDEALIZED REMOVAL EFFICIENCY CURVE
FOR A FLOW SENSITIVE TREATMENT DEVICE
3-62
-------
where storm flows are assumed to be gamma distributed and independent of c
and d, and r(q) is expressed as a fraction. The result is:
K+1 (3-40)
K-ln(F/Z)
—• •—
where: K = 1/v 2
Equation 3-40 is plotted in Figure 3-22 which demonstrates the long term
performance which can be achieved by in-line treatment devices. Note that
the greater the variation in the runoff flow (v ), the more poorly the device
performs on average.
To use Equation 3-40 or Figure 3-22, an efficiency curve for the partic-
ular device is required. These are presented for many devices in Figures
5-43 - 5-54 in Section 5.5.2 of the Numerical Estimates Chapter. Also in-
cluded in Section 5.5 is a discussion of the amount of field and experimental
data used to develop the curves, their applicability to separate versus com-
bined runoff, and the extrapolation of information on suspended solids re-
moval to predictions for the removal of other pollutants. If other informa-
tion is available to develop an efficiency curve for a particular treatment
device, this curve may be used in the same manner. Once the percent removal
versus flow curve is chosen, the following procedure is employed to determine
the average long term performance:
Step 1. From the efficiency curve, determine F, the percent removal
of the given size device at the mean runoff flow (Q ).
K
Step 2. From the efficiency curve, determine Z, the largest percent
removal of the given size device at very low flows. The
idealized removal curve (from Equation 3-38) should be
drawn over the actual curve to insure a reasonable
representation.
Step 3. Given the variation of the runoff flow, v , use Equation
3-40 or Figure 3-22 to determine ?„. ^
r
3.6.1.4.1 Example: Analysis of In-Line Treatment Device
To demonstrate the evaluation of the long term performance of a treat-
ment device, assume a drainage area served by an in-line dissolved air flo-
tation system with chemicals (see Figure 3-20 for efficiency curve) has the
following characteristics:
Mean runoff flow = Q = 4 MGD = 4,000,000 GPD (6.2 cfs)
K
Variation of runoff flow = v =1.15
q
Mean runoff load = MD = 2,000 Ib BOD5
K
Surface area of air flotation device = 1,000 sq. ft.
3-63
-------
1.00
COEFFICIENT
OF
VAR I ATION = Z/q
O.IO
O.2O O 3O O40 0.5O O 60 070 0.80 090
PERCENT REMOVAL AT MEAN RUNOFF FLOW
PERCENT REMOVAL AT VERY LOW FLOW
I OO
FIGURE 3-22
LONG TERM PERFORMANCE OF
A FLOW SENSITIVE TREATMENT DEVICE
3-64
-------
The surface loading rate at the mean runoff flow is (4, 000, 000 GPD)/ (1,000
ft2) = 4,000 GPD/ft2, which is 4 times the nominal design flow. From Figure
3-20 (BOD5 with chemicals), the values of F and Z are selected. The first
inclination is to select F = 57 and Z = 80. Because of the flat shape of the
actual performance curve at low flows, however, the idealized exponential
curve with F = 57 and Z = 80 does not provide a good match to the actual
performance. Selecting F = 55 and Z = 100 provides a much better overall
match. The idealized curves are compared to the actual performance curve in
Figure 3-23. For these values the long term average reduction in the runoff
load is calculated from Equation 3-40.
K = 1/v 2 = 1/1. 152 = 0.76
P - zr i - 100 r °-76
^LJ ~ 1U 1
F K-ln(F/Z) ~ 0.76 - ln(55/100)
= 36 percent reduction of long term BOD5 load.
Using Figure 3-22, F/Z = 55/100 = 0.55. Given that v =1.15, Pp/Z =
0.36, and P is determined as P = 100(0.36) = 36 percent. q Note that either
the numerical or the graphical method may be used, yielding the same result.
The modified runoff load, M* is then:
K
M* = (1 - 36/100) Mn = 0.64 (2,000) = 1,280 Ib BOD5
R R
As with the interceptor, the long term performance of the flow sensitive
treatment device is adversely affected by large within storm flow variations
and positive flow-concentration correlations. Again, these effects are
usually not large enough to be significant over the long term (see Section
3.6.1.7.4).
3.6.1.4.2 Concentration Sensitive Treatment Devices
The percent pollutant removal obtained by a number of treatment devices,
including screens and sedimentation tanks, increases with higher influent
concentrations. The relationship between the influent concentration and the
percent removal is assumed to be increasing to a limit. An analysis similar
to the one for the flow sensitive treatment device has been performed to
determine the long term load reduction as a function of the percent removal
at the mean runoff concentration, the best percent removal obtainable at very
high concentrations, and the variability of the runoff concentration (32).
The long term performance curves are similar in form to those of Figure 3-22,
except the performance improves with a higher coefficient of variation of
runoff concentration. Similarly, high within storm concentration variations
and positive flow concentration correlations improve the long term treatment .
Most treatment devices which are sensitive to influent concentration
are also sensitive to influent flow. These dually sensitive devices have
also been analyzed (32), and an intermediate long term performance is
obtained, depending upon the degree of sensitivity to either flow or
concentration. The improved treatment at higher concentrations generally
3-65
-------
<
>
o
100
90
80
70
60
50
UJ 40
o:
30
20
10
NOMINAL DESIGN FLOW RATS IOOO GPD/ft *
BOO, WITH CHEMICALS (ACTUAL)
FLOW=0,
I
REMOVALS BASED ON
BYPASS AT Q>4.5 0 DESIGN
J I
34567
FLOW/DESIGN FLOW
10
LEGEND
IDEALIZED CURVE'. F= 57, Z = 80
IDEALIZED CURVE! F= 55, Z= 100
FIGURE 3-23
COMPARISON OF IDEALIZED
AND ACTUAL DISSOLVED AIR FLOTATION PERFORMANCE
3-66
-------
to an improvement in the long term performance of the device subjected to
varying influent concentrations. This improvement may be ignored, however,
for an initial, conservative assessment, and the methodology developed for
the flow sensitive treatment device may be used, with the average influent
pollutant concentration determining the appropriate curve selected in Section
5.5.2 to define the treatment device performance at the mean runoff flow
and low flows.
3.6.1.4.3 Disinfection
Disinfection of pathogenic organisms in stormwater overflows is often
necessary to protect public health, protect water supplies, bathing beaches,
and other water uses. Conventional disinfection of wastewater generally uses
chlorine and chlorine compounds, and most of the investigations on the
disinfection of combined sewer overflows have been conducted with chlorine
disinfectants. These and other disinfection systems are discussed in more
detail in Section 5.5.2.7 of the Numerical Estimates Chapter.
The disinfection process is quite complicated and the approach outlined
in Figure 3-22 is not readily applicable to the analysis of disinfection. A
unique methodology is required and the basic assumptions and mathematical
development of the analysis are presented in Section 5.5.2.7.
The purpose of the analysis is to provide a method of transforming in-
formation about the bacterial removal of disinfection systems under controlled
conditions to estimates of the long term performance of systems subjected to
varying runoff or overflow rates. The results are displayed in Figure 3-24.
The abscissa is the product of the effective kill rate (k), the mean contact
time (t), each determined with the influent flow equal to the mean runoff
flow, (Q ). The long term bacteria reduction is shown for a device treating
a constant influent flow (equal to Q ) and devices subjected to varying in-
fluent flow (v = 1); one with a constant disinfectant feed rate and one with
a feed rate directly proportional to the influent flow.
T6 demonstrate how Figure 3-24 is used, assume a proposed disinfection
system is designed to achieve nearly plug flow and gives a 99.99 percent
bacteria removal (ICT1* remaining) at the mean runoff flow. This is equiva-
lent to (k t) = 20. The long term average reduction for the proportional
feed device is 99.85 percent (1.5 x 10"3 remaining) and only 97 percent (3 x
10 2 remaining) for the constant feed device. Changes in the size of the
device may be estimated by changing the contact time (t) while changes in the
amount of disinfectant used may be estimated by changing the kill rate (k).
Improvements in long term performance may be compared to the cost of flow
metering and proportional feed equipment, larger disinfection systems, and
the use of more disinfectant.
The analysis of disinfection systems, as well as other types of storm-
water treatment, has been directed towards the determination of the long
term pollutant load reduction. While this provides much insight as to the
effectiveness of the treatment alternative, and may be adequate for many
planning purposes, it may provide an incomplete picture. This is particularly
true for disinfection controls, where the fact that most storms are treated
3-67
-------
o
2
O
o:
UJ
a.
z
o
99.9999
99 9998
99.9996
99.9994
99.999
99 998
99.996
99.994
99.99
99.98
99.96
99.94
99.9
99.8
99.6
99.4
99
98
96
94
90
80
60
4O
VARYING FLOW, PROPORTIONAL
CHLORINE FEED
VARYING FLOW, CONSTANT
CHLORINE FEED
I
I
I
I
I
I
10"
10"
10"
10"
10"
0 5 10 15 20 25 30 35 40 45 50
(kt)R= DEVICE SPECIFICATIONS AT MEAN RUNOFF FLOW (QR)
Z
o
K
O
<
OL
QL
UJ
l-
co
z
O
FIGURE 3-24
EFFECT OF STORMFLOW VARIATION
ON PERFORMANCE OF EMPIRICAL DISINFECTION DEVICE
3-68
-------
adequately is more important than the fact that a few large storms are over-
flowing with little treatment, thereby having an adverse effect on the long
term average reduction. The important consideration is the frequency of storm
load occurrences after treatment, rather than the long term average. This is
discussed in more detail for disinfection and other stormwater control al-
ternatives in Section 3.6.1.6, The Effect of Stormwater Control Devices on
the Frequency of Loadings.
3.6.1.5 Combined Treatment Systems Which Capture, Store, and Treat
Runoff
To estimate the long term performance of a combined treatment system
which captures, stores, and provides in-line treatment for stormwater over-
flows, the approximation developed for a combination of interceptors and
storage devices may be employed: the fraction of the load remaining for the
combined system equals the product of the fractions remaining from each of
the individual components.
To illustrate the evaluation of combinations of control strategies,
assume the storage and the in-line facility previously presented are both
used. The storage device captures runoff until it is full, when the by-
passed flows are treated by the in-line dissolved air flotation system. The
runoff captured by the storage device is then treated at a controlled rate
such that there is 50 percent treatment (r = 50%). For these examples, the
following loads were determined:
M = Mean runoff load = 2,000 Ib BOD5
K
M* predicted after storage and treatment of captured runoff alone
= 1,350 Ib BOD5
M* predicted after in-line treatment alone = 1,280 Ib BOD5
K
The resulting runoff load with both the storage and in-line treatment is
then estimated as :
MR = ( * > * 2'000 • 86° lb
3.6.1.6 The Effects of Stormwater Control Devices on the Frequency
of Loadings
The techniques, curves, and equations presented in the previous
sections for estimating the reduction in the long term average stormwater
load due to various control devices provide useful information to the
planner; and for stormwater impacts which are long term in nature, such as
sediment desposition and eutrophication due to nutrient runoff into an im-
poundment, the knowledge of the long term removal is completely adequate.
For transient stormwater impacts which occur primarily during, or immediately
following storm events, however, planning decisions may require information
on how treatment devices modify the frequency of stormwater loadings.
3 T. 69
-------
Research has been conducted to determine the effect of treatment on
the variance of stormwater loads (33). A general observation is that control
devices reduce the variance of the runoff load less than they reduce the
mean. The majority of the smaller storms are treated very well, and in the
case of storage or interception, they may be completely captured. The larger
storm loads, however, may only be marginally reduced by the treatment system.
The long term output from the treatment system thus has more variation around
the mean than does the input.
Assuming one can estimate the effect of treatment on the variance of the
runoff load, problems still arise because the modified storm loads may no
longer be gamma distributed. Different control devices affect the frequency
distribution of storm loads in different ways. A few simple examples of this
are presented for different control devices. When a single, simple device
is used some estimates may be made of the resulting frequency of modified
storm loads. More complex cases involving combinations of control devices
require simulation for reliable estimation.
3.6.1.6.1 Interception
Interceptors capture a portion of the flow from a storm runoff event
and their effect on the frequency distribution of storm loading rates (i.e.,
pounds per hour during the storm) may therefore be approximated. All storms
with flows less than or equal to the available interceptor capacity (QT) are
completely captured while larger storms (higher average flows) have their
flow rates reduced by Q . Assuming a constant runoff concentration, this
effect on the flow rate may be transformed to an equivalent effect on the
loading rate.
The effect of an interceptor on the frequency distribution of storm
loading rates is demonstrated in Figure 3-25. The original runoff loading
rates are assumed to be gamma distributed, and Figure 3-2 is used to draw
the frequency distribution given the coefficient of variation of the runoff
flows, v = 1.15. An interceptor with QT/QR = 1-25 completely captures 71
percent of the storms. The overflow rates from the remaining 29 percent of
the storms are reduced as indicated. For example the loading rate during
the 90th percentile storm (with a loading rate exceeded by only 10 percent of
the storms) is reduced from about 2.4 W to 1.15 W . Knowing the average
number of storms per year or season, this result may be transferred to the
expected number of occurrences during the year or during the particular
season.
3.6.1.6.2 Storage
Storage devices capture a portion of the volume from a storm runoff
event, and their effect on the frequency distribution of storm loads (i.e.,
total pounds per storm) may therefore be approximated. All storms with
volumes less than or equal to the effective storage capacity (V ) are com-
pletely captured while larger storms (higher total runoff volumes) have their
runoff volumes reduced by V . Assuming a constant runoff concentration, this
effect on the runoff volume may be transferred to an equivalent effect on the
runoff load.
3-70
-------
99
Ul
3 98
Z 97
UJ
5 96
O 95
^ 92
o:
o
CO
M
UJ
s:
80
70
50
30
20
10
1.25
MULTIPLES OF THE MEAN RUNOFF
LOADING RATE, WR
LEGEND:
—— WITHOUT INTERCEPTOR, l/q = I 15
OVERFLOW FROM INTERCEPTOR, QI/0R= 1.25
FIGURE 3-25
FREQUENCY DISTRIBUTION OF STORM OVERFLOW LOADING RATES
BEFORE AND AFTER INTERCEPTION
3-71
-------
The effect of a storage device on the frequency distribution of storm
loads is demonstrated in Figure 3-26. The original runoff loads are assumed
to be gamma distributed and Figure 3-2 is used to draw the frequency distri-
bution given the coefficient of variation of the runoff volumes, v R = 1.75.
A storage device with V /V =1.35 completely captures 77 percent of the
storms. The by-pass loads from the remaining 23 percent of the storms are
reduced as indicated. For example, the total runoff load from the 90th per-
centile storm (with a total load exceeded by only 10 percent of the storms)
is reduced from about 2.9 M to 1.55 M .
K R
3.6.1.6.3 In-Line Treatment Devices
In-line treatment devices reduce the concentration of stormwater runoff
and overflows. Assuming the removal efficiency is sensitive to influent flow,
as described in Equation 3-38, and assuming that the influent concentration
is constant and equal for all storms, the effluent loading rate during a
particular storm (i.e., pounds per hour) is a function only of the flow rate.
Therefore, the effect of in-line treatment on the frequency distribution of
storm loading rates may be approximated.
Larger storms (higher average flows) have lower percent removals, as
shown in Figure 3-22 and Equation 3-38. Therefore, if a particular storm
has a higher loading rate than another storm before treatment, it will also
have a higher loading rate after treatment (though both are reduced). The
nth percentile storm before treatment is thus the nth percentile storm after
treatment, allowing a simple transformation from the untreated frequency
distribution to the treated frequency distribution with Equation 3-38. This
is demonstrated in Figure 3-27 for a device which gives 60 percent removal
at the mean runoff flow (F = 60) and 80 percent removal at low flows (Z = 80).
The untreated frequency distribution is drawn for v = 1.15, as in Figure
3-25. For example the 66th percentile (mean) storm"which has a loading rate
of W before treatment, has a loading rate of 0.4 W after treatment, cor-
responding to a 60 percent reduction. Note that for larger storms the un-
treated and treated curves become approximately parallel. This indicates
that none of the additional runoff is receiving treatment, as is the case
when the hydraulic capacity of the device is reached and the additional flows
are merely by-passed.
3.6.1.6.4 Disinfection
Disinfection systems are assumed to operate in a similar fashion to in-
line treatment devices. Given the basic assumptions presented in Section
5.5.2.7, the effluent coliform bacteria loading rate is a function of only
the runoff flow. Larger storms (higher average flows) have lower percent re-
movals due to the decreased contact time in the device and, in the case of a
constant chlorine feed system, lower disinfectant concentrations.
Assuming the empirical disinfection device designed to achieve nearly
plug flow (described in Section 5.5.2.7) is used, Figure 3-28 shows the
frequency distribution of storm loading rates before and after disinfection.
The untreated loading rate distribution is estimated by assuming that v =
1.00. The treated loading rate distribution is calculated from Equations
3-72
-------
LU
ID
IS
o
QL
O
X
I-
00
CO
LJ
UJ
O
o:
UJ
a.
99
98
97
96
95
94
93
92
91
90
80
X
1.35
MULTIPLES OF THE MEAN RUNOFF LOAD, MR
LEGEND;
WITHOUT STORAGE, ^VR = 1.75
BYPASS FROM STORAGE, VE/VR= I 35
FIGURE 3-26
FREQUENCY DISTRIBUTION OF STORM LOADS
BEFORE AND AFTER STORAGE
3-73
-------
UJ
ID
o
UJ
(£.
O
<
X
CO
CO
UJ
CJ
or
UJ
a.
98
97
96
95
94
93
92
91
90
80
70
60
50
40
30
20
10
°c
1
1
1
1 /
//
//
\l
/
f
/
1
1
1
1
1 /
/
/
/
/
/
/
f
/
/
/
/
/
/
/
t
/
/
/
/
/
/
/
234567
MULTIPLES OF THE MEAN RUNOFF
LOADING RATE, WR
LEGEND:
WITHOUT IN-LINE TREATMENT, Z/q = 1.15
WiTH FLOW SENSITIVE IN-LINE TREATMENT
F = 60% Z = 80%
FIGURE 3-27
FREQUENCY DISTRIBUTION OF STORM LOADING RATES
BEFORE AND AFTER IN-LINE TREATMENT
3-74
-------
UJ
>
e>
o
i-
oc.
o
2
co
co
2
UJ
98
97
96
»
94
93
92
91
90
80
70
60
50
^O
30
20
10
10"
io"
10
10
MULTIPLES OF THE MEAN RUNOFF
LOADING RATE, WR
LfGEND:
WITHOUT TREATMENT, I/q = I.OO
WITH DISINFECTION, FLOW PROPORTIONED CHLORINE FEED
WITH DISINFECTION,CONSTANT CHLORINE FEED
(FOR BOTH DISINFECTION CASES, 99.99% REMOVAL AT
MEAN RUNOFF FLOW)
FIGURE 3-28
FREQUENCY DISTRIBUTION OF STORM LOADING RATES
BEFORE AFTER DISINFECTION
3-75
-------
5^-25 and 5-26 in Numerical Estimates and shown for systems with a flow pro-
portioned and a constant chlorine feed rate. Both systems are designed to
give 99.99 percent removal (lO"4 remaining) at the mean runoff flow. If it
has been determined, for example, that coliform loading rates greater than
0.01 W result in unacceptable receiving water impacts, the frequency of
occurrence of these impacts may now be estimated for each of the disinfection
alternatives. The constant chlorine feed system results in about 13 percent
of the storms having loading rates greater than 0.01 W while the flow
proportioned feed system results in about 3.5 percent of the storms having
loading rates greater than 0.01 WD.
K
A few simple examples of frequency estimates have been presented.
However, estimates of the frequency distribution of stormwater loads result-
ing from more complex combinations of treatment systems may require a more
sophisticated analysis with simulation.
3.6.1.7 Comparison of Statistical Method to Simulations of Treatment
This section presents comparisons between the treatment performances
predicted by the statistical method and results obtained with simulation
studies. The comparisons support the validity of the theoretical curves and
indicate that estimates based on these curves are likely to be similar to
those obtained with simulation modeling techniques,
3.6.1.7.1 Effect of Previous Storms
To check the approximation for the effect of previous storms on the long
term effective storage capacity given in Figure 3-17, the results of a
simulation of an 8 square mile drainage basin in Dallas, Texas, are used.
A simple rainfall/runoff ratio is used to convert the 1968 hourly rainfall
record to runoff flows. A storage device is simulated and the amount of
storage available at the beginning of each storm is recorded. Nine simu-
lations are made with different size basins and different emptying rates.
The results of these simulations are summarized in Figure 3-29. The curves
for AQ/V = 0.5, 1.0, and 2.0 reasonably approximate the simulation results.
However these curves are intended only for an initial assessment of storage
device operation and not for actual design since they depend on an idealized
representation of the basin operation.
3.6.1.7.2 Storage Device Performance
In a report on combined sewer overflow problems in the City of Trenton,
New Jersey, Kaufman analyzes the potential impact of a detention basin at
the storm by-pass of the City's sewage treatment plant (34, 35). The Trenton
rainfall is characterized by ten years of United States Weather Bureau
Records (1963-1973). Assumed relationships between the intensity of rainfall
and the amount of runoff lost to infiltration and overland flow (not entering
the combined sewer) are used to determine the volume of combined sewage
overflowed during each storm at the treatment plant by-pass. These volumes
are used to calculate the percent of the total overflow which would be cap-
tured by a detention basin of various sizes (35, Plate 14).
3-76
-------
5 0
O
UJ
cc
O
h-
co
UJ
CJ
UJ
5
O
_l
O
>
4.0 -
BROADSCALE SIMULATOR RESULTS
DALLAS, 1968, RAINGAGE 412244
2.0 30 40
STORAGE VOLUME (EMPTY)l
0.75
cc
O
v-
c/i
UJ
co
UJ
-------
To compare the Trenton results with the statistical method, the mean
and the coefficient of variation of the overflow volumes were determined
(V = 3.48 MG, v = 1.49). The results plotted in Figure 3-30 show a very
close agreement between the theoretical and calculated long term performance.
Note that storm overflow volumes are calculated in a more sophisticated
manner than with a simple rainfall to runoff ratio. This demonstrates the
flexibility of the general methodology, where more refined estimates may be
used to determine the runoff statistics, depending upon the specific problem
setting and the data available.
3.6.1.7.3 Interception and Storage
As part of a nationwide evaluation of combined sewer overflows and
urban stormwater discharges, Heaney and Huber et al. use the STORM simulation
model to develop storage-treatment isoquants for five cities in the United
States (3, 36). A relatively simple transformation from rainfall to runoff
is used to generate one year of hourly runoff flows and loads, and a storage-
treatment system is employed to simulate the capture of these flows and
loads. A number of these STORM simulations are executed with different
combinations of storage and interception, and the percent of the yearly
runoff load captured is noted for each simulation. Curves are fitted to
storage-interception combinations with equal percentages of the runoff load
captured to form the isoquants (3, Figures 12-16).
The curves and relationships of the statistical method presented in this
Chapter can be used to generate similar isoquants for one of the cities,
Denver, Colorado; and the results are compared to those generated by the
STORM simulation. The STORM simulations were made with Raingage 052220 in
Denver for the year 1960, and this record is therefore used to generate the
appropriate runoff characteristics, summarized in Table 3-8, for use in the
comparison. The mean runoff volume and flow are given per unit area, in
inches per storm and inches per hour respectively. The effects of depression
storage and evaporation are considered negligible and not included. Accumu-
lation rates for the BOD and the effects of street sweeping which are used in
the STORM simulations are also not incorporated. The storage-interception
isoquants are generated using the statistics in Table 3-8, Figures 3-15,
3-17, 3-18, and 3-19, and Equation 3-36. The results are compared to the
STORM simulations in Figure 3-31. The amounts of BOD captured as predicted
using the curves and relationships of the statistical method are very
similar to those simulated with the STORM model.
3.6.1.7.4 In-Line Treatment Device
The performance of an idealized flow sensitive treatment device is
simulated using actual runoff quality (chemical oxygen demand, COD, and
suspended solids, (SS) and flow data from Durham, North Carolina (37),
Milwaukee, Wisconsin (38), Washington, D. C. (39), and Lubbock, Texas
(40). A summary of the data is presented in Table 3-9. For each data set,
the performance of a flow sensitive treatment device is simulated by assum-
ing that Equation 3-38 applies exactly. The actual observed concentrations
and flows are then subjected to a removal consistent with Equation 3-38.
The within storm observations are processed sequentially for all storms.
3 r 78
-------
TRENTON,NEW JERSEY, 1963-1973
LEGEND:
- KAUFMAN (15,16)
STATISTICAL METHOD, V =1.49
VARIATION = V.
0.5
1.0
VR
1.5 2.0 2.5 3.0 3.5 4.0
EFFECTIVE STORAGE CAPACITY 1
MEAN RUNOFF VOLUME
4.5
FIGURE 3-30
DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE
3-79
-------
TABLE 3-8
SUMMARY OF RUNOFF STATISTICS
DENVER, COLORADO, RAINGAGE 052220, 1960
Runoff to Rainfall Ratio = 0.39
QD = 0.0131 in/hr v = 1.38
K q
Vn = 0.078 in/storm v _ = 1.49
K VK
A = 119 hrs
Assume Moderate First Flush
TABLE 3-9
SUMMARY OF RUNOFF DATA ANALYZED
City
Durham (37)
Milwaukee (38)
Washington, D.C. (39)
Lubbock (40)
insufficient data.
Coefficient
of Variation
of Flow = v
q
(Between Storms)
1.59
1.05
9) 0.70
0.68
Constituent Pollutants
COD
No. of
Storms
26
13
6
11
Total No.
Samples
398
144
45
93
No. of
Storms
26
12
*
11
SS
Total No.
Samples
354
130
*
93
3-80
-------
.60
.50
UJ
cc
o
CO
- STATISTICAL METHOD
(ASSUME MODERATE
FIRST FLUSH)
.10
.00
.000
.010 .020
INTERCEPTION, Qj/A, ( in./hr.)
.030
FIGURE 3-31
STORAGE/ INTERCEPTION ISOQUANTS
PERCENT BOD CAPTURED WITH FIRST FLUSH
DENVER, I960 RAINGAGE 052220
3-31
-------
The results are analyzed for the average percent removal. The calculation
is made with F/Z equal to 0.2, 0.4, 0.6, 0.8, and 0.9. The resulting removal
over all the storms is compared to the theoretical reduction in Figure 3-32.
In general, the comparison between the theoretical and simulated re-
duction is quite good. Stipulation of the relative removal at the mean
runoff flow (F/Z), and a knowledge of the coefficient of variation of the
runoff flow (v ), allows a reasonable prediction of the long term perfor-
mance. Furthermore, the differences between the theoretical and the simu-
lated values are explainable in terms of two mechanisms previously mentioned:
high within storm flow variation and flow-concentration correlation.
Lubbock, which has very high flow variability within storms, has simu-
lated overall reductions smaller than those predicted by Equation 3-40.
The relative magnitude of within storm flow variation is depicted in Table
3-10 as the average standard deviation of flow within storms (a ) divided by
the standard deviation of flow between storms (a ). For all the cities
except Lubbock, the within storm flow variation is small compared to the
between storm variation.
The other effect, flow-concentration correlation, is depicted in Table
3-11 as the ratio of the flow-weighted average concentration (c-) to the
standard, time average concentration (c). If flow and concentration are
independent, cJc is nearly equal to one. If flow and concentration are
positively_correlated, c-,/c is greater than one; and if the correlation is
negative, c-/c is less tnan one. Note that the SS data tend to show a
positive correlation (scour effect), particularly in Durham; while the COD
data tend to show a negative correlation (dilution effect), with the excep-
tion of Durham. The ratios shown in Table 3-11 are consistent with the
observation that:
1. The simulated COD reductions for Durham are very nearly equal to
the theoretical reductions, while the SS reductions are less than
predicted.
2. The simulated SS reductions for both Milwaukee and Lubbock are
less than their respective COD reductions.
3. The simulated COD reductions for Washington, D. C., are greater
than the theoretical reductions.
Despite the influence of within storm flow variation and flow-concen-
tration correlation, Equation 3-40 provides a reasonable first estimate of
the long term performance of flow sensitive stormwater treatment devices.
A number of other checks of the statistical method have been made using
results from Atlanta, Georgia, Chippewa Falls, Wisconsin, and the four
cities used in the comparison of in-line treatment devices. These compari-
sons together with the results presented in this section, demonstrate the
utility of the statistical method for initial assessments of stormwater
control systems.
3-82
-------
1.00
DURHAM, N.C
MILWAUKEE, WIS.
LUBBOCK, TEX
WASHINGTON, D.C.
0.10 020 030 040 0.50 060 070 0.80 0.90 100
_[ PERCENT REMOVAL AT MEAN RUNOFF FLOW
PERCENT REMOVAL AT VERY LOW FLOW
FIGURE 3-32
COMPARISON OF SIMULATED FLOW SENSITIVE DEVICE
WITH THEORETICAL CURVES
3-83
-------
TABLE 3-10
RATIO OF WITHIN STORM FLOW VARIABILITY
TO BETWEEN STORM FLOW VARIABILITY
City
Durham
Milwaukee
Washington, D.C.
Lubbock
a /a
qw q
0.48
0,59
0.64
1.43
TABLE 3-11
FLOW CONCENTRATION CORRELATION
City Constituent Cf/C
Durham COD 1.07
SS 1.50
Milwaukee COD 0.97
SS 1.15
Washington, D.C. COD 0.65
Lubbock COD 0.82
SS 1.17
84
-------
3.6.2 Management Practices for Stormwater Control
Controls can be instituted at several stages of the stormwater pollution
process. The previous section was directed towards controls instituted at
the "end of the pipe," after the loads have been generated and conveyed.
Because of the magnitude and intermittent nature of stormwater loads, the
structural requirements to provide adequate end-of-pipe treatment may be
significant. Management practices designed to reduce stormwater pollution
before it is generated and conveyed provide an alternative to these large
structural controls.
Management practices include source controls, such as street sweeping,
catch basin cleaning programs, controls on the use and transport of harmful
or hazardous materials, erosion control, control of surface flows, and for
long range planning programs, land use control. Management techniques are
also available for reducing pollutant concentrations and overflows from the
sewer system. These include sewer separation, infiltration-inflow control,
sewer flushing, polymer injection, and automated system controls. The
effectiveness of each of these practices can be assessed by estimating their
impact on the quality and quantity of stormwater runoff and overflows. A
brief discussion of each of these management practices is presented, to-
gether with guidelines for quantitatively estimating their effectiveness.
3.6.2.1 Source Controls: Street Sweeping and Catch Basin Cleaning
Studies have been conducted to characterize the effect of street
sweeping and catch basin cleanout on the removal of various contaminants
(41,42,43,44). Important factors to consider are the type of sweeper used,
the frequency of sweeping, the frequency of rain events, the rate at which
contaminants accumulate in the drainage basin, and the street surface type
and condition.
Current street sweeping practices in urban areas are estimated to be
between 35 and 65 percent effective, averaging about 50 percent (42). In-
creased efficiencies can be achieved by reducing the speed of a sweeping
pass to less than five m.p.h., and by increasing the frequency of passes
(41). Enforced bans on parking along sweeping routes is necessary to insure
effective removal. Utilization of more efficient machines and the adjustment
of schedules to sweep more frequently near areas of high solids production
can increase the total effectiveness of a street cleaning program (43).
Street sweeping approaches its maximum performance in terms of reducing
the total yearly stormwater load when its frequency is much greater than the
frequency of storm events. For example, street sweeping in areas with very
long periods between storms, such as Phoenix, Arizona, would be quite
effective at reducing the yearly stormwater load. One sweeping per week
might be sufficient in these areas. However, in areas where it rains more
frequently, such as Portland, Oregon, more frequent sweeping in the order of
once per day may be necessary to significantly reduce the long term storm-
water load.
85
-------
To quantitatively assess the effect of street sweeping, an estimate may
be made of the reduction in the average runoff concentration, c. Experiments
have demonstrated the accumulation of solids on street surfaces as a function
of the elapsed time since the last cleaning by sweeping or rain (45), how-
ever, this is not the same as the concentration of contaminants in the run-
off. Recently, numerous runs of the STORM simulation model were made by
Heaney and Nix (46) for the City of Minneapolis, Minnesota, with varying
street sweeping frequencies. From these runs, a relationship was developed
for the long term fraction of street surface BOD removed as a function of
the fraction of days streets are swept and the efficiency of the sweeper.
This relationship is shown in Figure 3-33. Typical "pick-up" efficiency for
a common brush-type sweeper is e = 0.50 while more expensive vacuum sweepers
may yield a higher efficiency of about e = 0.90 (46). Note that Figure 3-33
was developed with one year's simulation in one location (A = 3-3.5 days),
and as previously discussed, may not be applicable to areas where the
average time between storms (A) is significantly different. Ideally, a
relationship similar to that shown in Figure 3-33 should be developed in-
corporating both the sweeping frequency and the average rainfall frequency.
Until this analysis has been performed, however, Figure 3-33 may be used
with caution for a first estimate. Also, note that Figure 3-33 describes
the fraction of available street surface pollutants (BOD) removed, while
nothing is said about pollutant runoff from other portions of the drainage
area. Heany and Nix estimate that in a typical separate or unsewered area,
70 percent of the pollutant runoff is from the street surface, however, this
number may vary (46). Area specific land use information may be used to
refine the estimate of the fraction of the runoff load actually treatable by
street sweeping. Finally, in combined sewer areas where overflow quality is
largely influenced by the mix of runoff with sanitary sewage, street sweeping
is considerably less effective at reducing stormwater loads.
3.6.2.2 Control of Harmful Materials
An effective method of decreasing the levels of toxic materials in
stormwater runoff is to restrict the usage of those materials. Lead, zinc,
antimony and asbestos are examples of toxic materials which are currently
introduced to the environment through wear of automobile brakes and clutches
and the breakdown of fuel and lubricants. In a long term program, these
substances could be replaced by others which are less harmful. More
immediate results could be expected through restrictions on the local use of
pesticides and herbicides. The elimination of harmful compounds from
deicing materials and a general improvement in the efficiency of deicing
programs can reduce the pollution from this activity.
Very little is known about the relationship between the amount of a
particular material introduced into the environment and its eventual concen-
tration in stormwater runoff. Preliminary assessments may assume a direct
relationship to estimate an order of magnitude concentration. Sophisticated
simulation techniques are available for materials such as pesticides (2,47,
48), however, the projected improvement in runoff quality due to the re-
stricted use of materials will probably only be a general estimate.
3-86
-------
1.0
o
<
L_
OL
^
OT O
\- Z
OT S
CO
2s
< 0
-"£
Q
*0
O OD
o:
u.
STREET SWEEPING SIMULATION,
MINNEAPOLIS, MINN. ( REF.
€= EFFICIENCY OF SWEEPER
(1971 SIMULATION
0.8
0.6
0.4
0.2
02 04 06 0.8
FRACTION OF DAYS STREETS ARE SWEPT
I 0
FIGURE 3-33
LONG-TERM EFFECTIVENESS OF STREET SWEEPING
3-87
-------
3.6.2.3 Erosion Control
The following activities are suggested to improve control over erosion,
which can introduce large amounts of nutrients and suspended solids to
stormwater 042,44).
proper selection of building and highway sites
maintenance and protection of native vegetation
use of mulches
drainage channel protection modification
careful backfilling after laying pipes
protection of stockpiles for removed earth
sediment retention basins
scheduling of clearing and grading during season when erosion
is less
traffic control for construction and earth hauling equipment
seeding areas with high erosion potential
Additional information for assessing the effect of erosion controls is pre-
sented in Chapter 4 of the Areawide Assessment Procedures Manual (2).
3.6.2.4 Control of Surface Flows
Several methods are available to reduce or delay stormwater runoff in
urban areas. The increased attention these methods have received in recent
years mark a new philosophy in the design of stormwater collection facilities
which seek to retain stormwater within a drainage basin rather than trans-
porting it as quickly as possible from the area.
A reduction of runoff can be brought about by increasing the period
over which stormwater can percolate through permeable soil layers. Further-
more, stormwater which is not able to percolate into the soil can at least
be delayed in order to reduce the surge or "slug" effects of the runoff.
Methods for reducing or delaying runoff are listed in Table 3-12 (49). Al-
though many of these methods actually involve structural modifications of
the drainage basin, they are included in this section because they generally
involve decentralized measures rather than larger, end-of-pipe facilities.
To assess the effectiveness of surface flow controls, estimates may be
made of the increase in infiltration rates or the effective storage capacity
provided throughout the watershed. Once these are determined, techniques
developed for the analysis of long term interceptor and storage device per-
formance can be applied. For example, the increased infiltration rate
3-88
-------
TABLE 3-12
MEASURES FOR REDUCING AND DELAYING URBAN STORM RUNOFF
Area
Large flat roof
Parking lots
Residential
General
(ref. 49)
Reducing Runoff
1. Cistern storage
2. Rooftop gardens
3. Pool storage or
fountain storage
4. Sod roof cover
Porous pavement
a. Gravel parking
lots
b. Porous or punc-
tured asphalt
Concrete vaults and
cisterns beneath
parking lots in
high value areas
Vegetated ponding
areas around parking
lots
Gravel trenches
4.
3.
4.
Cisterns for indi-
vidual homes or
groups of homes
Gravel driveways
(porous)
Contoured landscape
Grandwater recharge
a. Perforated pipe
b. Gravel (sand)
c. Trench
d. Porous pipe
e. Drywells
5. Vegetated depressions
1. Gravel alleys
2. Porous sidewalks
3. Mulched planters
3.
4.
Delaying Runoff
Ponding on roof by
constricted downspouts
Increasing roof
roughness
a. Rippled roof
b. Gravelled roof
Grassy strips on
parking lots
Grassed waterways
draining parking lot
Ponding and detention
measures for imper-
vious areas
a. Rippled pavement
b. Depressions
c. Basins
Reservoir or detention
basin
Planting a high delay-
ing grass (high
roughness)
Gravel driveways
Grassy gutters or
channels
Increased length of
travel of runoff by
means of gutters,
diversions, etc.
Gravel alleys
3 * 89
-------
provided by porous pavement is equivalent to an available interceptor capac-
ity, Q ; with the captured runoff entering the groundwater regime. The
amount of depression storage provided by rippled pavement or ponding on
roofs is equivalent to an effective storage capacity, V , in the watershed.
C
3.6.2.5 Land Use Control
Increases in the degree of development and urbanization generally result
in more severe stormwater impacts. The relationships between land use and
stormwater loads are important for predicting future changes in stormwater
loadings and for investigating the efficacy of land use control as a means
of stormwater management. The Numerical Estimate sections of this manual
discuss the impact of land use on both runoff quantity and quality. The
relationships presented in these sections may be used to estimate the effec-
tiveness of land use control.
A relationship between the percent impervious area and the average ratio
of runoff to rainfall (R.J is shown in Figure 5-20. Land use modifications
which change the percent impervious area may be evaluated by calculating the
new ratio of runoff to rainfall, the resulting change in stormwater loads,
and the subsequent impact on receiving stream concentrations.
Urbanization often results in an even greater increase in runoff rates
than in the total volume of runoff. This is due to a decrease in the
attenuation time of the runoff event. The relationship between the attenua-
tion of runoff events and urbanization is depicted in Figure 5-23 and
Equations 3-13 and 5-9. Note that population density is used as a general
indicator of land use conditions. Although the assessment of land use
changes is thus somewhat indirect, estimates may still be made of the new
mean runoff event duration (DR), the resulting mean runoff flow (QR) and
loading rate (W ), and the subsequent change in stream concentrations.
R
Land use changes may also affect the quality of stormwater runoff. As
discussed in Section 5.3.2 of Numerical Estimates, the ability to assess the
effect of land use changes in runoff quality is dependent upon the establish-
ment of a significant relationship for the particular study area. Assuming
a satisfactory relationship exists, the new land use proposed for an area
will result in a new estimate of the average pollutant runoff concentration
(c).
3.6.2.6 Collection System Management: Sewer Separation
It is recognized that routing stormwater through the same sewer system
as sanitary sewage, i.e., a combined sewer system, can cause a reduction
in municipal treatment plant efficiency during storm events, Furthermore,
the surges in flow force the collection system to by-pass wastewater and
discharge it directly to the receiving water body. Separate conveyance
systems for the stormwater and the wastewater help to eliminate this problem,
however, the cost of sewer separation in the older urban areas where combined
sewers are prevalent may be prohibitive. Furthermore, if provisions are not
made to treat the stormwater, direct stormwater discharges to a receiving
water body from a separate system may contribute a greater pollution load
3 - 90
-------
than would be caused by occasional by-passes from a combined system (50).
To quantitatively estimate the effect of sewer separation on stormwater
loads three steps are required. First, the stormwater concentration is
changed in the now separately sewered area as indicated in Section 5.3.1 of
Numerical Estimates. Secondly, the fraction of the runoff load formerly
captured by the interceptor and treated at the municipal treatment plant now
enters the receiving stream directly (though with a different pollutant
concentration). Finally, the average sewage treatment efficiency at the
municipal plant may improve, thereby decreasing the continuous municipal
loading rate.
3.6.2.7 Infiltration and Inflow Control
"Infiltration" is the introduction of additional flow into a sewer
through leaky joints and broken pipes. "Inflow" is the introduction of
additional flow through deliberate or accidental sewer connections from
water users. Both intrusions utilize a portion of the sewer capacity.
Infiltration can be prevented in new pipe systems through adequate
design and testing; it can be eliminated in existing systems by survey and
correction. Elimination systems can help reduce the necessity for by-passing
during storm events. Roof leaders can either be reconnected to the storm
sewer system,, where the runoff will be treated or discharged to the receiving
water body, or can be allowed to drain onto pervious areas. The reduction of
the infiltration rates into a combined sewer system will increase the avail-
able interceptor capacity for stormwater capture, Q_, This may be evaluated
using Figure 3-15 or with the technique demonstrates in Figure 3-25. The
control of infiltration and inflow into a separate sewer system should
ideally eliminate wet weather overflows,
3.6.2.8 Sewer Flushing
The deposition of solids during dry weather in slow flowing portions of
combined sewers and the trapping of sediments in portions of separate sewers
provide a source of pollutants for resuspension during subsequent stormwater
flows. Recent studies of this phenomena and the potential of flushing to
reduce the resulting wet weather load provide some basis for evaluating the
effectiveness of sewer flushing (51,52,46),
The study of deposition and flushing in the Boston area (52) presents
general relationships for estimating the total mass of solids deposited in
a combined sewer system as a function of the total collection system pipe
length, drainage area, average pipe slope, and the average wastewater flow.
Guidelines are also presented for estimating the extent of the sewer system
over which significant deposition occurs. Finally, current research on
sewer flushing operations is described. The relationships presented are
appropriate for an initial assessment of the potential effectiveness of sewer
flushing in a study area. Hendy and Nix (46) provide a further simplifica-
tion of this approach which allows an estimate of the reduction in the
deposition related stormwater load. This assessment may be used to modify
3-91
-------
the average overflow concentration (c) to relate flushing programs to
eventual changes in receiving water quality.
3.6.2.9 Polymer Injection
The flow capacity of certain pipes can be increased by adding polymers
to the water flowing through them. Injection of a polymer-gelled slurry
reduces wall friction and allows a temporary increase in line capacity.
Automatic polymer-injecting units can be positioned along key sewer trunk
lines and directed to inject the chemical at critical points in a storm
event. The polymers selected for this use are similar to those developed
for treatment plant clarifiers, and have been shown not to be disruptive to
bacterial growth or to provide nutritive value to algae.
The overall combined sewer flow capacity is dictated by the critical
points in the system, where the polymer injection should be directed. The
long term effectivness of the polymer inj ection program is thus evaluated by
estimating the resulting increase in the available interceptor capacity for
storm runoff (QT).
3.6.2.10 Automated System Control
In combined sewer systems, interceptor lines which carry both dry
weather sewage flows and storm runoff during rainfall events may provide a
significant measure of control by virtue of their ability to retain a portion
of the storm flow and route it to the sewage treatment plant. Consistent
with the ability of the sewage treatment plant to handle increased flows,
control may be improved by maximizing retention of storm flows in existing
collection systems.
Both separate and combined sewer systems can be made more effective by
the utilization of remote monitoring and control systems. These systems,
which might include level sensors, tide gates, raingage networks, sewage
and receiving water quality monitors, overflow detectors, and flow meters,
can effectively regulate sewer flows and provide a controlled flow to the
treatment facility. This approach can be useful in avoiding overflowing and
by-passing by making a more effective use of the line capacities (53,54).
To assess the effectiveness of automated controls on a preliminary
basis, an estimate may be made of the increase in the effective storage
capacity (V ) provided by the sewer system. A decrease in the average over-
flow concentration (c) may also be appropriate for systems with selective,
quality sensitive overflows. Simulation should be used for more detailed
studies of automated control systems.
An overview of management practices available for the control of storm-
water runoff and overflows has been presented. Guidelines for evaluating
their effectiveness within the framework of the statistical assessment
methodology have also been provided. The final step is to relate the im-
provements in receiving water quality due to the various treatment alterna-
tives to the cost of each plan.
3-92
-------
3.6.3 Benefit and Cost Evaluation
Stormwater controls in an urban area are implemented to protect or im-
prove the quality of adjacent waterways. As discussed in Chapter 2, benefits
are measured in terms of the value of beneficial uses of the water which
are created or preserved. The costs are measured in terms of the resources
required to develop, implement, and maintain the stormwater control strategy.
The selection of a stormwater control plan requires a favorable balance
between these benefits and costs.
Methods are available for quantitatively estimating the value of water
uses, such a water supply, recreation, navigation, etc. If this can be done
for a particular area, a rigorous cost-benefit analysis may be performed
(55). Often, however, the benefits may be difficult to quantify, and the
assessment framework becomes less direct. Receiving water benefits are
measured in terms of compliance with water quality standards or criteria,
which in turn are established by responsible planning agencies to protect
or enhance specific beneficial uses. The problem then becomes one of meeting
receiving water standards in a cost-effective manner. The following sections
develop this approach and provide guidelines for implementing it in a
planning study.
3.6.3.1 Improvements in Receiving Water Quality
Methods for estimating the impact of stormwater loads in the receiving
water are presented in Section 3.5. These methods may again be used with the
modified loads determined after treatment, and a new receiving water response
is predicted. Different combinations of types of controls and sizes or
amounts for each may be tested.
If a particular water quality standard is used as the basis for anal-
usis, the load reductions necessary to meet the standard may be determined.
For standards directed towards long term average water quality, reductions
in the long term average loading rate (W ) are determined. For transient
impacts, load reductions necessary to violate a standard less frequently
(i.e., only 3 times per summer rather than 15 times) are examined. In some
areas, standards are now being written in a probabilistic manner, identifying
the number or frequency of exceedances which are permitted (56). These types
of standards are more appropriate for dealing with intermittent, stormwater
related water quality problems. The primary problem in developing these
standards is identifying the ecological effects of intermittent periods of
high (or in the case of dissolved oxygen, low) instream concentrations.
Research is needed to obtain a better understanding of the severity of these
impacts in terms of water use limitations. Once these relationships are
better understood, more meaningful probabilistic standards may be developed.
Changes in the mean, variability, and frequency distribution of stormwater
loads are then examined and translated into receiving water responses to test
for compliance.
Analysis tools have been presented for quantitatively assessing pollut-
ant concentrations in the receiving water. These tools are appropriate for
addressing many of the standard water quality problems encountered in a
3-93
-------
planning study: dissolved oxygen, coliform bacteria, solids, toxicants, etc.
Many of the stormwater problems found in an urban area, however, are not as
readily addressed with these models. These problems include such things as
floatables, and oils and grease in the vicinity of combined sewer overflows,
particularly near residential or recreational areas, the deposition of
organic matter near outfalls, and the erosion of natural stream channels.
These problems are more difficult to analyze with generalized models, and a
good understanding of local conditions and particular problems in the study
area is required. The planner should visit the area during and following
storms, and significant interaction with the public should be encouraged. In
certain areas, these less easily modeled problems may be the most important
for improving the quality of urban waterways (57).
3.6.3.2 Indirect Benefits of Stormwater Control
Although the most direct impact of stormwater control is the benefit on
receiving water quality, there are also a number of other impacts which may
result from the implementation of the plan. These indirect impacts are par-
ticularly important when considering stormwater control as one aspect of an
overall, integrated plan for improving the quality of life in urban areas.
Flood control is one of the most obvious programs which may be integrat-
ed with a stormwater control plan. Storage ponds or tanks which reduce and
attenuate pollutant loads may also serve to alleviate downstream flooding
problems. Indeed, flood control may be the primary benefit of detention
systems. This has important applications for the costing and funding of
stormwater control plans. By integrating the controls with other aspects of
urban planning and development, costs may be shared and an overall increase
in the efficiency of community efforts may be obtained. Other examples of
auxiliary benefits which may be integrated with stormwater control include
land preservation through erosion control, improved drainage, cleaner streets
due to street sweeping or catch basin cleaning, and improvements in street,
utility, and conveyance system layouts for the purpose of maintenance and
control. Finally, potentially harmful aspects of a stormwater control plan
should also be identified and evaluated. These include factors such as
relocation for construction, the contamination of surface or groundwaters
from system failure or operational accidents such as chemical spills, and
requirements for sludge or residual disposal after treatment.
3.6.3.3 Evaluating Cost of Controls
Planning assessments require an estimate of the cost of alternative
stormwater control strategies. Curves for estimating the capital, operation-
al and maintenance costs of stormwater treatment processes of various sizes
are presented in Section 5.5.3 of the Numerical Estimates Chapter. Standard
engineering costing procedures incorporating factors such as inflation and
interest should be used.
Costs may be determined for a number of strategies, and a procedure
developed for comparing alternatives. This procedure may involve a simple
listing of alternative plans, their benefits and costs; or it may involve
more sophisticated attempts to optimize with tools such as linear programming
3-94
-------
(58). The planner is cautioned, however, not to let the decision making
process become lost in a complex computer program or a routinized scheme.
These may serve as useful aides, however, they are no substitute for sound,
informed, and practical planning and engineering judgment for the solution
of stormwater problems. The planning process should be flexible and
developmental; cognizant of shortcomings in the analysis and data upon which
it is based and seeking to improve the understanding of stormwater problems
to develop effective and workable solutions.
3.7 References
1. Field, Richard, Anthony N. Tufuri, Hugh E. Masters, Urban Runoff
Pollution Control Technology Overview, U. S. Environmental Protection
Agency, EPA-600/2-77-047, Cincinnati, Ohio, March 1977.
2. 208 Areawide Assessment Procedures Manual, Volume I, Chapter 2-5,
Hydroscience, Inc. for U. S. Environmental Protection Agency, APA-
600/9-76-014, Cincinnati, Ohio, July 1976.
3. Heaney, James P., Wayne C. Huber, Stephan J. Nix, Storm Water Management
Model, Level I, Preliminary Screening Procedures, EPA-600/2-76-75,
Cincinnati, Ohio, October, 1976.
4. Lager, John A., Theodor Didriksson, George B, Otte, Development and
Application of a Simplified Stormwater Management Model, Metcalf and
Eddy, Inc., for U. S. Environmental Protection Agency, EPA-600/2-76-218,
Cincinnati, Ohio, 1976.
5. Brandstetter, Albin, Assessment of Mathematical Models for Storm and
Combined Sewer Management, Battelle, Pacific Northwest Laboratories for
U. S. Environmental Protection Agency, EPA-600/2-76-175a, Cincinnati,
Ohio, 1976.
6. Short Course Proceedings, Applications of Stormwater Management Models;
1976, Edited by Francis A. Digiano, Donald A. Adrian, and Peter A.
Mangarella, University of Massachusetts for Environmental Protection
Agency, EPA-600/2-77-065, Cincinnati, Ohio, March 1977.
7. Storm Water Management Model, Volume I - Final Report, Metcalf and Eddy,
Inc., University of Florida, and Water Resources Engineers for U. S.
Environmental Protection Agency, Water Quality Office, Report No.
11024DOC07/71, September, 1971.
8. Storm Water Management Model, Volume II - Verification and Testing.
Metcalf and Eddy, Inc., University of Florida, and Water Resources
Engineers, Inc. for U. S. Environmental Protection Agency, Water Quality
Office, report No. 11024DOC08/71, September, 1971.
9. Storm Water Management Model, Volume III - User's Manual, Metcalf and
Eddy, Inc., University of Florida, and Water Resources Engineers, Inc.,
for U. S. Environmental Protection Agency, Water Quality Office, Report
No. 11024DOC09/71, September, 1971.
3-95
-------
10. Storm Water Management Model, Volume TV - Program Listing, Metcalf and
Eddy, Inc., University of Florida, and Water Resources Engineers, Inc.,
for U. S. Environmental Protection Agency, Report No. 11024DOC10/71,
September, 1971.
11. Smith, G. F., Adaptation of the EPA Storm Water Management Model for Use
in Preliminary Planning for Control of Urban Storm Runoff, M.E. Thesis,
Department of Environmental Engineering Sciences, University of Florida,
Gainesville, May, 1975.
12. Hydrologic Engineering Center, Corps of Engineers, Urban Storm Water
Runoff: STORM, Generalized Computer Program 723-58-L2520, Davis,
California, January, 1975.
13. Howard, Charles D. D., Theory of Storage and Treatment Plant Overflows,
Journal of the Environmental Engineering Division, ASCE, Vol. 102, No.
EE4, August, 1976.
14. Di Toro, Dominic M., and Mitchell J. Small, Discussion of Theory of
Storage and Treatment Plant Overflows, Journal of the Environmental
Engineering Division, ASCE, Vol. 102, No. EE3, June, 1977.
15. Water Quality Management Planning Methodology for Urban and Industrial
Stormaater Needs, Hydroscience, Inc., for Texas Water Quality Board,
Arlington, Texas, 1976.
16. Chow, Ven Te, Ben Chie Yen, Urban Stormwater Runoff: Determination of
Volumes and Flowrates, University of Illinois for U. S. Environmental
Protection Agency, EPA-600/2-76-116, Cincinnati, Ohio, May, 1976.
17. Parzens, E., Stochastic Processes, Holden Day, San Francisco, 1964.
18. DiToro, Dominic M., Statistical Analysis of Intermittent Runoff and
Receiving Water Responses. In press 1978.
19. Snow Hydrology, Summary of Report of the Snow Investigations, U. S. Army
Corps of Engineers, North Pacific Division, Portland, Oregon, 1956.
20. Donigian, Anthony S., Jr., and Norman H. Crawford, Modeling Nonpoint
Pollution from the Land Surface, Environmental Protection Agency,
Research Grant No. R803315-01-0, February, 1976.
21. Dunne, Thomas, Anthony G. Price, Samuel C. Colbreck, The Generation of
Runoff from Subarctic Snowpacks, Water Resources Research, Vol. 12, No.
4, August, 1976.
22. Simplified Mathematical Modeling of Water Quality, Hydroscience, Inc.,
and Mitre Corporation, Environmental Protection Agency Publication,
March 1971.
3 - 96
-------
A STATISTICAL METHOD
FOR ASSESSMENT OF URBAN STORMWATER
LOADS - IMPACTS - CONTROLS
FOR
EPA - NON POINT SOURCES BRANCH
WASHINGTON, D.C.
PROJECT OFFICER
DFNNIS N. ATHAYOE
MANAGER
NATIONWIDE URBAN RUNOFF PROGRAM
JANUARY 1979
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
DATE:
191979
SUBJECT:
FROM:
TO:
Transmittal of Document Entitled "A Statistical Method
for Assessment of Urban Runoff"
Swep T. Davis, Deputy assistant Administratq
Office of Water Planning and Standards
All Regional Water Division Directors
ATTN: All Regional 208 Coordinators
All Regional NPS Coordinators
All Nationwide Urban Runoff Prototype Projects
All State and Areawide Water Quality Management Agencies
Other Concerned Groups
TECHNICAL GUIDANCE MEMORANDUM-TECH- 49
Purpose
This document "A Statistical Method for Assessment of Urban Runoff"
has been prepared to provide technical assistance to the Nationwide
Urban Runoff prototype projects and other interested groups in assessing
the impact of urban stormloads on the quality of receiving waters, and
to evaluate the cost and effectiveness of control measures for reducing
these pollutant loads.
Guidance
The enclosed report is provided in accordance with the Nationwide
Urban Runoff Program established under Section 208 of the Clean Water
Act of 1977. This methodology is appropriate for use at the planning
level where preliminary assessments are made to define problems, establish
the relative significance of contributing sources, assess feasibility of
control, and determine the need for and focus on additional evaluations.
It can also be used effectively in conjunction with detail studies, in
evaluating the most cost-effective alternatives for controlling urban
runoff.
Attachment
EPA FORM 1320-6 (REV, 3-76)
-------
STORMWATER MANUAL
TABLE OF CONTENTS
Chapter Number
Number Page
1 INTRODUCTION 1-1
2 STORMWATER RUNOFF - A REVIEW OF THE PROBLEM 2-1
2.1 MAJOR WATER QUALITY PROBLEMS IN URBAN AREAS 2- 2
2.2 URBAN RUNOFF 2-3
2.3 CONTAMINANTS IN URBAN RUNOFF 2-3
2.4 WATER QUALITY PROBLEM DEFINITION 2- 6
2.4.1 RELEVANT TIME AND SPACE SCALES 2-11
2.4.1.1 NATURE OF CONTAMINANTS 2-11
2.4.1.2 NATURE OF THE RECEIVING WATER
SYSTEM 2-12
2.4.2 WATER USE OBJECTIVES AND CRITERIA 2-12
2.4.3 CHARACTERISTICS OF PARTICLAR STUDY AREA 2-15
2.5 REFERENCES 2-16
3 THE STATISTICAL METHOD FOR THE ASSESSMENT OF RUNOFF
AND TREATMENT 3-1
3.1 STORM RUNOFF EVENTS AS RANDOM OCCURRENCES 3-2
3.2 CHARACTERIZATION OF RUNOFF EVENTS 3- 3
3.2.1 STATISTICAL PROPERTIES OF RUNOFF
PARAMETERS 3-5
3.2.2 LONG TERM RUNOFF PROCESS 3- 6
3.2.3 DETERMINATION OF RUNOFF PARAMETERS 3-10
3.3 RAINFALL, THE DRIVING FORCE 3-11
3.4 DEVELOPMENT OF STORMWATER LOADS 3-16
3.4.1 RUNOFF QUANTITY 3-16
3.4.2 RUNOFF QUALITY AND RESULTING LOADS 3-19
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
3 3.5 IMPACTS IN THE RECEIVING WATER 3-26
3.5.1 PREDICTION OF LONG TERM IMPACTS 3-28
3.5.1.1 STREAMS AND RIVERS 3-28
3.5.1.2 ESTUARIES AND COASTAL WATERS 3-30
3.5.1.3 LAKES AND RESERVOIRS 3-33
3.5.2 PREDICTION OF TRANSIENT IMPACTS 3-36
3.5.2.1 STREAMS AND RIVERS 3-36
3.5.2.2 ESTUARINE SYSTEMS 3-42
3.6 ASSESSMENT OF STORMWATER CONTROL ALTERNATIVES 3-43
3.6.1 STRUCTURAL TREATMENT DEVICES 3-46
3.6.1.1 INTERCEPTION 3-46
3.6.1.2 STORAGE 3-48
3.6.1.2.1 EFFECT OF PREVIOUS STORMS 3-50
3.6.1.2.2 STORAGE EFFECTIVENESS 3-50
3.6.1.2.3 FIRST FLUSH EFFECT 3-53
3.6.1.2.4 TREATMENT OF STORED
RUNOFF 3-57
3.6.1.2.5 EXAMPLE OF STORAGE DEVICE
EVALUATION 3-57
3.6.1.3 INTERCEPTION AND STORAGE 3-59
3.6.1.4 IN-LINE TREATMENT DEVICES WHICH
REDUCE POLLUTANT CONCENTRATION 3-60
3.6.1.4.1 EXAMPLE: ANALYSIS OF
IN-LINE TREATMENT DEVICE 3-63
3.6.1.4.2 CONCENTRATION SENSITIVE
TREATMENT DEVICES 3-65
3.6.1.4.3 DISINFECTION 3i-67
11
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
3 3.6.1.5 COMBINED TREATMENT SYSTEMS WHICH
CAPTURE, STORE AND TREAT RUNOFF 3-69
3.6.1.6 THE EFFECTS OF STORMWATER CONTROL
DEVICES ON THE FREQUENCY OF
LOADINGS 3-69
3.6.1.6.1 INTERCEPTION 3-70
3.6.1.6.2 STORAGE 3-70
3.6.1.6.3 IN-LINE TREATMENT DEVICES 3-72
3.6.1.6.4 DISINFECTION 3-72
3.6.1.7 COMPARISON OF STATISTICAL METHOD TO
SIMULATIONS OF TREATMENT 3-76
3.6.1.7.1 EFFECT OF PREVIOUS STORMS 3-76
3.6.1.7.2 STORAGE DEVICE PERFORMANCE 3-76
3.6.1.7.3 INTERCEPTION AND STORAGE 3-78
3.6.1.7.4 IN-LINE TREATMENT DEVICE 3-78
3.6.2 MANAGEMENT PRACTICES FOR STORMWATER CONTROL 3-85
3.6.2.1 SOURCE CONTROLS: STREET SWEEPING
AND CATCH BASIN CLEANING 3-85
3.6.2.2 CONTROL OF HARMFUL MATERIALS 3-86
3.6.2.3 EROSION CONTROL 3-88
3.6.2.4 CONTROL OF SURFACE FLOWS 3-88
3.6.2.5 LAND USE CONTROL 3-90
3.6.2.6 COLLECTION SYSTEM MANAGEMENT:
SEWER SEPARATION 3-90
3.6.2.7 INFILTRATION AND INFLOW CONTROL 3-91
3.6.2.8 SEWER FLUSHING 3-91
3.6.2.9 POLYMER INJECTION 3-92
3.6.2.10 AUTOMATED SYSTEM CONTROL 3-92
3.6.3 BENEFIT AND COST EVALUATION 3-93
3.6.3.1 IMPROVEMENTS IN RECEIVING WATER
QUALITY 3-93
iii
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
3 3.6.3.2 INDIRECT BENEFITS OF STORMWATER CONTROL 3-94
3.5.3.3 EVALUATING COST OF CONTROLS 3-94
3.7 REFERENCES 3-95
4 SIMULATION OF STORMWATER IMPACTS 4-1
4.1 AVAILABLE MODELS 4- 1
4.2 BROADSCALE RECEIVING WATER SIMULTOR 4- 2
4.2.1 INTERNAL CALCULATIONS 4-5
4.2.2 PROGRAM CAPABILITIES AND LIMITATIONS 4- 5
4.2.3 MODEL INPUTS 4-5
4.2.4 MODEL OUTPUT 4-7
4.3 REFERENCES 4-7
5 NUMERICAL ESTIMATES FOR STORMWATER ASSESSMENT METHODOLOGIES 5-1
5.1 RAINFALL 5-1
5.1.1 PRECIPITATION CHARACTERISTICS IN DIFFERENT
PARTS OF THE UNITED STATES 5-1
5.1.2 DEFINITION OF STORM EVENT 5-20
5.1.3 APPLICABILITY OF THE GAMMA DISTRIBUTION FOR
STORM EVENT CHARACTERISTICS 5-22
5.1.4 AREAL DISTRIBUTION OF RAINFALL 5-37
5.1.4.1 RAINGAGE AGGREGATION 5-39
5.2 RUNOFF QUANTITY 5-46
5.2.1 DETERMINATION OF AVERAGE RUNOFF TO RAINFALL
RATIO 5-46
5.2.2 DETERMINATION OF AVERAGE DURATION OF RUNOFF
EVENT 5-52
5.3 RUNOFF QUALITY 5-60
IV
-------
TABLE OF CONTETS
(Continued)
Chapter Number
Number Page
5 5.3.1 DETERMINATION OF AVERAGE POLLUTANT CONCENTRA-
TIONS 5-60
5.3.1.1 VARIABILITY OF POLLUTANT CONCENTRA-
TIONS 5-61
5.3.2 EFFECT OF LAND USE ON STORMWATER QUALITY 5-61
5.4 RECEIVING WATER 5^78
5.4.1 TRANSPORT PROPERTIES 5-78
5.4.1.1 CHANNEL GEOMETRY 5-80
5.4.1.2 DISPERSION COEFFICIENT ESTIMATES 5-83
5.4.2 REACTION OF POLLUTANTS 5-84
5.4.2.1 SEQUENTIAL REACTIONS 5-86
5.4.3 BACKGROUND RECEIVING WATER CONDITIONS 5-88
5.4.3.1 VARIABILITY OF BACKGROUND CONDITIONS 5-92
5.5 TREATMENT DEVICE PERFORMANCE 5-101
5.5.1 CONSIDERATIONS FOR VARIOUS POLLUTANTS 5-102
5.5.2 TREATMENT DEVICE PERFORMANCE EFFICIENCIES 5-103
5.5.2.1 SEDIMENTATION PERFORMANCE 5-103
5.5.2.2 DISSOLVED AIR FLOTATION PERFORMANCE 5-106
5.5.2.3 SWIRL CONCENTRATOR PERFORMANCE 5-106
5.5.2.4 HIGH RATE, DEEP BED MEDIA FILTRATION
PERFORMANCE 5-109
5.5.2.5 SCREENS AND MICROSCREENS 5-111
5.5.2.6 BIOLOGICAL TREATMENT 5-115
5.5.2.7 DISINFECTION 5-120
5.5.2.8 TREATMENT AT DRY WEATHER PLANTS 5-130
5.6 COST ESTIMATES FOR TREATMENT ALTERNATIVES 5-132
v
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
5 5.6.1 STRUCTURAL TREATMENT DEVICES 5-133
5.6.1.1 STORAGE BASINS 5-133
5.6.1.2 SEDIMENTATION 5-135
5.6.1.3 AIR FLOTATION 5-135
5.6.1.4 SWIRL CONCENTRATOR 5-135
5.6.1.5 HIGH RATE FILTRATION 5-142
5.6.1.6 STATIONARY SCREENS 5-142
5.6.1.7 HORIZONAL SCREENS 5-142
5.6.1.8 CHEMICAL COAGULATION 5-149
5.6.1.9 CHLORINATION FEED EQUIPMENT 5-149
5.6.1.10 HIGH INTENSITY MIXING/CHLORINE
CONTACT BASIN 5-158
5.6.1.11 RAW WASTEWATER PUMPING 5^-158
5.6.1.12 SLUDGE PUMPING 5-158
5.6.1.13 BIOLOGICAL TREATMENT 5-158
5.6.1.14 EXAMPLE COST CALCULATION 5-165
5.6.2 MANAGEMENT PRACTICES 5-168
5.7 REFERENCES 5-168
6 MONITORING FOR STORMWATER ASSESSMENT 6- 1
6.1 PURPOSE OF MONITORING PROGRAM 6- 1
6.2 RAINFALL MONITORING 6- 2
6.2.1 USE OF RAINFALL DATA TO DEFINE SITE
CHARACTERISTICS 6- 2
6.2.2 USE OF RAINFALL DATA FOR PROJECTIONS AND
EVALUATION OF IMPACTS 6-3
6.2.3 RAINGAGE DENSITY 6- 5
6.2.4 OTHER RAINGAGE CONSIDERATIONS 6- 9
6.3 SITE SELECTION AND DRAINAGE BASIN CHARACTERIZATION 6-10
6.4 MONITORING OF RUNOFF AND OVERFLOWS 6-12
VI
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
6 6.4.1 STUDY PERIOD AND SAMPLING FREQUENCY 6-12
6.4.1.1 SAMPLING INTERVAL WITHIN STORMS 6-13
6.4.2 SAMPLING PROCEDURE AND PARAMETERS 6-15
6.5 MONITORING OF RECEIVING WATER 6-17
6.5.1 CONTINUOUS MONITORS 6-18
6.6 REVIEW OF MONITORING LITERATURE 6-19
6.7 REFERENCES 6-20
7 EXAMPLE STORMWATER ANALYSES 7-1
7.1 SALT LAKE CITY, UTAH 7- 1
7.1.1 RAINFALL ANALYSIS 7-3
7.1.2 DRAINAGE BASIN CHARACTERIZATION 7- 3
7.1.3 RUNOFF QUANTITY 7-5
7.1.4 RUNOFF POLLUTANT LOADS 7-8
7.1.5 RECEIVING WATER RESPONSE 7- 9
7.1.6 STORMWATER CONTROL ALTERNATIVES 7-19
7.2 EXAMPLE STORMWATER ANALYSIS: KINGSTON, NEW YORK 7-27
7.2.1 RAINFALL ANALYSIS 7-27
7.2.2 DRAINAGE BASIN CHARACTERIZATION 7-30
7.2.3 DETERMINE RUNOFF VOLUMES 7-31
7.2.3.1 CAPTURE BY TREATMENT PLANT 7-32
7.2.4 DETERMINE STORMWATER LOADS 7-32
7.2.4.1 RECEIVING WATER RESPONSE 7-34
7.2.4.2 VARIABILITY IN THE HUDSON 7-36
7.2.4.3 OBSERVED WATER QUALITY 7-37
VII
-------
TABLE OF CONTENTS
(Continued)
Chapter Number
Number Page
7 7.2.5 SIMULATION OF FECAL COLIFORM RESPONSE 7-40
7.2.6 CONTROL ALTERNATIVES 7-47
7.3 REFERENCES 7-50
APPENDIX A - USERS MANUAL FOR BROAD SCALE SIMULATOR
APPENDIX B - PROGRAM LISTING FOR BROAD SCALE SIMULATOR
Vlll
-------
STORMWATER MANUAL
LIST OF TABLES
Table Number
Number Page
2- 1 CONTAMINANTS IN URBAN RUNOFF 2-4
2- 2 SELECTED INSTREAM IMPACTS 2- 4
2- 3 EXAMPLE BENEFICIAL WATER USE INVENTORY 2.15
3- 1 STATISTICS FOR STORM CHARACTERIZATION 3- 5
3- 2 MINNEAPOLIS RAINFALL ANALYSIS 3-13
3- 3 EXAMPLE STORMWATER LOADING TABLE FOR ADVECTIVE STREAM 3-27
3- 4 SUMMARY OF STEADY STATE SOLUTIONS FOR POLLUTANT
CONCENTRATIONS IN RECEIVING STREAMS 3-29
3- 5 STEADY STATE EQUATION FOR WASTE CONCENTRATIONS IN
TIDAL RIVERS AND ESTUARIES DUE TO POINT SOURCE 3-32
3- 6 CONCENTRATIONS IN LARGE, COMPLETELY MIXED IMPOUNDMENT 3-34
3- 7 GENERAL GUIDELINES FOR ESTIMATING MAGNITUDE OF FIRST
FLUSH EFFECT 3-56
3- 8 SUMMARY OF RUNOFF STATISTICS DENVER, COLORADO,
RAINGAGE 05220, 1960 3-80
3- 9 SUMMARY OF RUNOFF DATA ANALYZED 3-80
3-10 RATIO OF WITHIN STORM FLOW VARIABILITY TO BETWEEN
STORM FLOW VARIABILITY 3-84
3-11 FLOW CONCENTRATION CORRELATION 3-84
3-12 MEASURES FOR REDUCING AND DELAYING URBAN STORM RUNOFF 3-89
4- 1 AVAILABLE COMPUTER MODELS (FROM AAPM, APPENDIX A(l)) 4- 3
4- 2 ANALYTICAL SOLUTIONS USED IN BRWS 4- 6
ix
-------
LIST OF TABLES
(Continued)
Table Number
Number Page
5- 1 CITIES SELECTED FOR RAINGAGE ANALYSIS 5- 2
5- 2 EFFECT OF ALTERNATIVE STORM DEFINITION RAINGAGE - 286026,
NEWARK AIRPORT, MAY 1948-DECEMBER 1975 (27.58 yr) 5-27
5- 3 (a) SUMMARY OF NEW YORK CITY RAINGAGE AGGREGATION ANALYSIS
PERIOD OF RECORD: 1/1/50-12/31/60 5-43
5- 3(b) RATIO OF AGGREGATED RAINGAGE STATISTICS TO POINT RAIN-
GAGE STATISTICS 5-44
5- 4 RUNOFF TO RAINFALL RATIOS FROM VARIOUS STUDIES 5-51
5- 5 COMPARISON OF QUALITY OF COMBINED SEWAGE FOR VARIOUS CITIES 5-64
5- 6 COMPARISON OF QUALITY OF STORM SEWER DISCHARGES FOR
VARIOUS CITIES 5-65
5- 7 POLLUTANT CONCENTRATIONS IN COMBINED SEWER OVERFLOWS 5-66
5- 8 POLLUTANT CONCENTRATIONS IN STORMWATER RUNOFF 5-67
5- 9 VARIABILITY OF RUNOFF AND OVERFLOW CONCENTRATIONS 5-68
5-10 POLLUTANT LOADING FACTORS FOR DESKTOP ASSESSMENT 5-75
5-11 SAMPLE U.S.G.S. SURFACE WATER RECORD DATA SHEET 5-82
5-12 RANGE OF VALUES OF REACTION COEFFICIENTS IN NATURAL
WATERS (48) 5-86
5-13 SUMMARY OF BACKGROUND CONCENTRATIONS FROM VIRGIN LAND 5-91
5-14 COMPARISON OF TREATMENT ALTERNATIVES 5-105
5-15 EXAMPLE CALCULATION OF STORMWATER TREATMENT COST 5-167
5-16 UNIT COSTS OF STREET SWEEPING 5-169
-------
LIST OF TABLES
(Continued)
Table Number
Number Page
7- 1 SUMMER RAINFALL STATISTICS (JULY-SEPTEMBER) SALT LAKE
CITY 7- 3
7- 2 STORMWATER FLOWS AND LOADS GENERATED BY SUMMER STORMS
SALT LAKE CITY 7-11
7- 3 JORDAN RIVER GEOMETRY, TRANSPORT, AND REACTION RATES
DURING SUMMER STORMS 7-13
7- 4 WET WEATHER DISSOLVED OXYGEN RESPONSE OF JODRAN RIVER
TO SUMMER STORMS 7-17
7- 5 WET WEATHER DISSOLVED OXYGEN AFTER IMPLEMENTATION OF
MANAGEMENT PRACTICES 7-26
7- 6 HUDSON RIVER FECAL COLIFORM (MPN/100 ml) SUMMER 1976 7-38
7- 7 HUDSON RIVER FECAL COLIFORM (MPN/100 ml) 7-39
7- 8 INPUT DECK RECEIVING WATER SIMULATION 7-46
xi
-------
STORMWATER MANUAL
LIST OF FIGURES
Figure Number
Number Page
2- 1 TYPICAL ANALYSIS OF WET WEATHER DISSOLVED OXYGEN 2- 7
2- 2 PROBABILISTIC ANALYSIS OF WET WEATHER DISSOLVED OXYGEN 2- 9
2- 3 MINIMUM DISSOLVED OXYGEN FREQUENCY CURVES FOR EXISTING
CONDITIONS IN THE DES MOINES RIVER 2-10
2- 4 TIME SCALES STORM RUNOFF WATER QUALITY PROBLEMS 2-13
2-5 SPACE SCALES STORM RUNOFF WATER QUALITY PROBLEMS 2-14
3- 1 REPRESENTATION OF STORM RUNOFF PROCESS 3- 4
3- 2(a) CUMULATIVE DISTRIBUTION FUNCTION FOR GAMMA DISTRIBUTION 3- 7
3- 2(b) CUMULATIVE DISTRIBUTION FUNCTION FOR GAMMA DISTRIBUTION 3- 8
3- 3 STORM EVENT AND LONG TERM FREQUENCY DISTRIBUTIONS 3- 9
3- 4 MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
CENTRAL PARK STATION 305801 3-15
3- 5 IDEALIZATION OF FIRST FLUSH EFFECT 3-22
3- 6 CORRECTION IN ESTIMATE OF Mn WHEN FIRST FLUSH IS PRESENT 3-24
R
3- 7 COMPARISON OF SPATIAL DETAIL IN STORMWATER LOADING
CHARACTERIZATION 3-31
3- 8 GRAPHICAL SOLUTION TO THE DILLON APPROACH 3-35
3- 9 WATER QUALITY RESPONSE SIMULATOR BOD-DISSOLVED OXYGEN
EXAMPLE 3-38
3-10 STREAM RESPONSE CHARACTERISTICS TO PULSE LOADS 3-39
3-11 EFFECT OF DISPERSION ON POLLUTANT CONCENTRATION AT MID-
POINT OF STORM PULSE 3-41
xii
-------
23. Mathematical Modeling of Natural Systems, Summer Institute Notes,
Environmental Engineering and Science Division, Manhattan College,
Bronx, New York, May 1976.
24. Thomann, R. V., System Analysis and Water Quality Management, Environ-
mental Research and Applications, Inc., New York, 1972.
25. Water Quality Management Planning Methodology for Hydrographio Modi-
fication Activities, Hydroscience, Inc. for the Texas Water Quality
Board, 1976.
26. Vollenweider, R. A., Scientific Fundamentals of the Eutrophication of
Lakes and Flowing Waters—With Particular Reference to Nitrogen and
Phosphorus as a Factor in Eutrophication, Organization for Economic
Cooperation and Development, DAS/CSI/68.27, Paris, France, 1968.
27. Dillon, P. J., and F. H. Rigler, A test of a Simple Nutrient Budget
Model Predicting the Phosphorus Concentrations in Lake Water, Canada
Fishery Research Board Journal, Vol. 31, 1974.
28. Thomann, R. V., D. M. Di Toro, R. P. Winfield, and D. J. O'Connor,
Mathematical Model of Phytoplankton in Lake Ontario, U. S. Environmental
Protection Agency, EPA-660/3-75-005, March 1975.
29. Hydroscience, Inc., Effects of Dispersion Upon Storm Pulses in a Stream,
Internal Technical Memorandum, December 1976.
30. Proceedings of Symposium on Diffusion in Oceans and Fresh Waters,
Columbia University, 1964.
31. Handbook of Mathematical Functions with Formulas, Graphs, and Mathemati-
cal Tables, U. S. Department of Commerce, National Bureau of Standards,
Applied Math Series No. 55, 1964.
32. Analytical Method: Further Analysis of Filters with Performance
Dependent Upon Influent Concentration, Internal Memorandum, Hydro-
science, Inc., April 1976,
33. Di Toro, Dominic M., Statistical Analysis of Urban Runoff Treatment
Devices, Presented at U. S. Environmental Protection Agency National
Conference on 208 Planning and Implementation, Reston, Virginia,
March 1977.
34. Kaufman, Herbert L., Combined Sewage Overflows - Frequency, Environmen-
tal Effects, and a Corrective Program, National Symposium on Urban
Hydrology, Runoff, and Sediment Management, University of Kentucky,
July 1974.
35. Control of Combined Sewage Flooding and Overflows, Clinton Bogert
Associates for the City of Trenton, New Jersey, January, 1974.
3-97
-------
36. Heaney, J. P., et al., Natiorajn.de Evaluation of Combined Sewer Overflows
and Urban Stormwater Discharges, Volume II: Cost Assessment and Impacts,
University of Florida for U. S. Environmental Protection Agency, EPA-
600/2-77-064, Cincinnati, Ohio, March 1977.
37. Colston, N. V., Characterization and Treatment of Urban Land Runoff,
North Carolina State University for U. S. Environmental Protection
Agency, EPA-670/2-74-096, Cincinnati, Ohio, December 1974.
38. Consoer, Townsend and Associates, Eumbolt Avenue Pollution Abatement
Demonstration Project for the City of Milwaukee, Engineering Report,
September 1974.
39. Roy F. Weston, Inc., Combined Sewer Overflow Abatement Alternatives,
Washington, D. C., U. S. Environmental Protection Agency, EPA-11024,
EXF 08/70, August 1970.
40. Texas Tech University, Water Resources Center, Variation of Urban
Runoff Quality with Duration and Intensity of Storms, WRC-73-2, 1973.
41. Sartor, J. D. and G. B. Boyd, Water Pollution Aspects of Street Surface
Contaminants, U. S. Environmental Protection Agency, EPA-R2-72-081,
November 1972.
42. Lager, J. A. and W. G. Smith, Urban Stormwater Management and Technol-
ogy: An Assessment, U. S. Environmental Protection Agency, EPA-670/2-74-
040, December, 1974.
43. Shaheen, D. G., Contributions of Urban Roadway Usage to Water Pollution,
U. S. Environmental Protection Agency, Washington, D. C., EPA-600/2-75-
004, March 1975.
44. Guidelines for Erosion and Sediment Control Planning and Implementation,
Maryland Department of Water Resources and Hittman Associates, Inc. for
U. S. Environmental Protection Agency, 15030 FMZ, August 1972.
45. Water Quality Management Planning for Urban Runoff, URS Research
Company, for U. S. Environmental Protection Agency, EPA-440/9-75-004,
December 1974.
46. Heaney, James P., Stephen J. Nix, Storm Water Management Model: Level
I - Comparative Evaluation of Storage Treatment and Other Management
Practices, U. S. Environmental Protection Agency, EPA-600/2-77-083,
April 1977.
47. Frere, M. H., C. A. Onstad, and H. N. Holtan, ACTMO-An Agricultural
Chemical Transport Model, Agricultural Research Service, U. S.
Department of Agriculture, Hyattsville, Maryland, ARS-H-3, June 1975.
48. Donigian, A. A., Jr., and N. H. Crawford, Modeling Pesticides and
Nutrients on Agricultural Lands, U. S. Environmental Protection Agency,
EPA-600/2-76-043, February 1976.
3-98
-------
49. Urban Hydrology for Small Watersheds3 U. S. Conservation Service, Tech.
Release No. 55, NTIS PB-244 531, Washington, D. C., January, 1975.
50. Burm, R. J., D. F. Krawczyk, and G. L, Harlow, Chemical and Physical
Comparison of Combined and Separate Sewer Discharges, Journal of the
Water Pollution Control Federation, Vol. 40, No. 1, January 1968.
51. A Flushing System for Combined Sewer Cleaning, FMC Corporation for U. S.
Environmental Protection Agency, 11020 DNO, March 1972.
52. Pisano, William C., Celso S. Queiroy, Procedures for Estimating Dry
Weather Pollutant Deposition in Sewerage Systems, U. S. Environmental
Protection Agency, EPA-600/2-77-120, July 1977.
53. Leiser, E.P., Computer Management of a Combined Sewer System, U. S.
Environmental Protection Agency, EPA 670/2-74-022, Cincinnati, Ohio,
July 1974.
54. 'Labadie, J. W., N. S. Grigg, B, H. Bradford, Automatic Control of Large-
Scale Combined Sewer Systems, Journal of the Environmental Engineering
Division, ASCE, Vol. 101, No. EE1, February 1975.
55. Dorfman, Robert, and Nancy S. Dorfman, Economics of the Environment,
Selected Readings, W. W. Norton and Company, Inc., New York, 1972.
56. Evaluation of Discharge Alternatives for South (San Francisco) Bay
Dischargers Authority, Hydroscience, Inc., December 1975.
57. Girtys Run, A Study in Urban Watershed Management, Carnegie-Mellon
University, Pittsburgh, Pennsylvania, May 1974.
58. Kirshen, Paul H., David H. Marks, and John C. Schaake, Mathematical
Model for Screening Stormwater Control Alternatives, Department of
Civil Engineering, Massachusetts Institute of Technology, Report No.
157, October 1972.
3-99
-------
CHAPTER 4
SIMULATION OF STORMWATER IMPACTS
In addition to the statistical methods presented in Chapter 3, storm
loads and their impacts may be estimated by the use of computer simulation
techniques. Simulators provide a representation of the actual temporal
sequence of storm events, their loads, and the time sequence of the receiving
water response. In this sense, simulation is a "brute-force" method for
evaluating wet weather impacts and their potential control. Like the
statistical method which develops direct, analytical solutions for pertinent
problems, simulators require a degree of simplification of the real rainfall-
runoff-receiving water process. However, computer simulators often allow
a more detailed representation of the spatial and temporal complexities of
the actual urban drainage system, and are thus particularly useful for the
later, more refined levels of stormwater planning. If simulators are used
in the initial assessment stage of runoff analysis, simpler versions, able
to process relatively long periods of rainfall records, at the sacrifice of
detail for individual events, are preferable.
4.1 Available Models
Models for evaluating wet weather impacts are generally divided into
two classes: (1) land-side drainage simulators, and (2) receiving water
models; though some packages include both aspects. The land-side simulators
utilize rainfall records and drainage area characteristics to develop storm
flows and loads. They incorporate factors such as the catchment hydrology,
sewer hydraulics, runoff quality, and control alternatives (storage, treat-
ment, etc.)- Receiving water models may be used generally for a range of
problems, or developed specifically for analyzing wet weather quality
impacts. They include factors such as the physical characteristics of the
water body, pollutant transport, and the chemical and biological reaction
rates of pollutants. A wide range of complexity, spatial and temporal
detail are found in available models.
Three recent reports provide useful summaries of computer simulation
models currently available to planners. These reports are:
1. "Areawide Assessment Procedures Manual, Appendix A" (1),
2. "Evaluation of Water Quality Models: A Management Guide for
Planners" (2), and
3. "Assessment of Mathematical Models for Storm and Combined
Sewer Management" (3).
4-1
-------
Appendix A of the Areawide Assessment Procedures Manual discusses models
which are generally available, well-documented, and have been tested and
applied. The models covered are listed in Table 4-1. The three reports
listed above, as well as the specific model documentations, should be
referenced for details on the program capabilities, limitations, and use.
It is noteworthy that while a wide range of land-side simulators are
available specifically for stormwater assessments, including simplified
models appropriate for the preliminary evaluation procedures presented in
this manual (4,5), less effort has been made to make receiving water models
particularly applicable to broad scale stormwater assessments. A relatively
simple receiving water model useful for initial wet weather evaluations has
been developed for this manual. This receiving water model, described in
the following sections, is designed to accept as input, the time variable
storm loads generated by the available land-side simulators.
4.2 Broadscale Receiving Water Simulator
The Broadscale Receiving Water Simulator (BRWS) provides the engineer
or planner with an analytical tool to determine the response of a river
(stream) or estuary to coliform bacteria, BOD, and dissolved oxygen deficit
stormwater discharges and combined sewer overflows. The model is geared to
use with available land side simulators for preliminary assessments. It
can provide the user with an added level of temporal detail of loads and
impacts, and greater spatial definition. Like the land side simulators
which are used in such assessments, BRWS adopts a simplified approach.
A one-dimensional representation of the receiving water stream or
estuary, and simple input parameters are employed. A single storm load
discharge location is used, with input flows and loads which vary with time
according to rainfall records and the output of a land side simulator, with
or without reductions due to applied control measures.
The model calculates the time history of concentrations which result in
the receiving water due to such intermittent loads, at up to six different
locations. Results are displayed in a matrix showing concentration as a
function of location and time, and also in a plot of concentration vs. time.
A separate run is required for each contaminant analyzed. The model can
accomodate conservative contaminants (Dissolved Solids), simple reactive
contaminants (coliform bacteria), or coupled reactions (BOD-D.O.).
At the completion of each run, BRWS calculates and outputs the frequency
distribution, the mean concentration and the standard deviation of the
contaminant analyzed at each of the six locations in the receiving water.
Concentration history is compared with a water quality standard, and the
percentage of time this criterion is violated at a particular location within
the river or estuary, is calculated and output. The model is thus designed
to address directly the significant planning questions relevant to wet
weather impact evaluations. The procedure is outlined schematically in
Figure 4-1.
4-2
-------
TABLE 4-1
AVAILABLE COMPUTER MODELS
(FROM AAPM, APPENDIX A (1))
Land-Side Simulators:
Acronym
STORM
SWMM
HSP
MITCAT
HVM-QQS
(Simplified Simulator)
AGRUN
ILLUDAS
Model Origin
Corps of Engineers
EPA
Hydrocomp
Mass. Inst. of Technology
Dorsch Consult
Metcalf $ Eddy
Water Resources Eng.
111. State Water Survey
Receiving Water Quality Models;
DOSAG-I
QUAL-II
RECEIV
RECEIV-II
SRMSCI
WRECEV
HWQM
LAKECO
LEVEL III
iPA
EPA
EPA
Raytheon (EPA)
Systems Control Inc.
Water Resources Eng. (EPA)
Hydrocomp, Inc.
Office of Water Resources
Research
EPA (by Medina, in press 1979)
4-3
-------
_ HOURLY STORM LOADS= FUNCTION (RAINFALL, BASIN PROPERTIES)
INPUT
a
\
CD
1
CALCULATIONS
TIME
RECEIVING WATER
OUTPUT —
o
SIMULATED CONCENTRATION PROFILE
STANDARD
FREQUENCY
DISTRIBUTION
{% LESS THAN)
— 90
— 80
L
-SO
-0 •
TIME
ZERO OR BACKGROUND
"NONSTORM"CONCENTRATION
FIGURE 4-1
SCHEMATIC OF BROADSCALE
RECEIVING WATER SIMULATOR OPERATION
4-4
-------
4.2.1 Internal Calculations
The BRWS model treats each recorded time interval of rainfall, flow,
and load as a separate, discrete event. The response for each is calculated
from the appropriate analytical solution for an impulse function (of
specified duration 6, e.g. 1 hour, 24 hours) shown in Table 4-2. The water
quality response for the entire rainfall-loading sequence is generated via
superposition of the discrete response. This is based on the assumption
that the receiving water concentration responds linearly to pollutant load-
ings. The program is written in Fortran IV, and a complete listing is pro-
vided in Appendix B.
4.2.2 Program Capabilities and Limitations
The basic capabilities and limitations of the BRWS model are outlined
as follows:
1. Capabilities:
a. Receiving water bodies - rivers (streams) and estuaries;
b. Water quality constituents - coliform bacteria, BOD, and dissolved
oxygen; Note: conservative constituents can be handled using the
BOD input routine, by setting reaction rate (K) = 0.
c. Maximum length of rainfall sequence - approximately 1,500 records
(two months) if hourly intervals are selected; four years if daily
interval is selected;
d. Number of discharge locations (per run) - one (1);
e. Maximum number of locations within river or estuary for evaluating
water quality response - six (6);
f. Ability to include stormwater flow in calculation of concentration
if significant;
g. Ability to include background conditions in rivers; and
h. Input either from disk files (generated by a drainage basin
simulator) or from punched cards.
2. Limitations:
a. Geophysical parameters (cross sectional area, temperature, reaction
rates) are assumed to be constant throughout the river or estuary;
b. BOD removal will be calculated even if river or estuary becomes
anaerobic;
c. Inability to specify background (boundary) conditions for
estuaries;
d. Inability to correct velocity (time-of-travel) if stormwater flow
is significant in relation to base stream flow;
e. Only net dispersion is modeled in estuaries, not flow reversals
within tidal cycles.
4.2.3 Model Inputs
The following inputs are required for the BRWS model:
4-5
-------
TABLE 4-2
ANALYTICAL SOLUTIONS
USED IN BRWS
Stream BOD Response: „
, ... *,„ x.. M u
c(x,t) = «Ct - -) Q e
Stream Dissolved Oxygen Deficit Response:
K x K x K x
M If -£- -2- M —
n/- *•» *r* x^ BOD r_d_w u u , .,. X, POD u
D(x,t) = 6(t - -) — =- C-jr-^HCe - e ) + «(t - -) — «- e
x a r
K x
a
* D e" U
o
Dissolved Oxygen Response:
c(x,t) - Cs - DCx,t)
fa")
Estuary BOD Response v J :
Cx - ut)2
, .. M 4Et " r
c(x,t) = — — — e
A/4irEt
Estuary Dissolved Oxygen Deficit Response *-aJ :
2
(x-Ut) „ Qc-U
" "
„
DOD -- 4Et
e
fa")
'Note that there is no finite solution for t = 0, so an arbitrarily small
time was selected for the first calculation at the discharge point. This
causes some error in the calculation, and the exact discharge location,
x = 0, should be avoided (i.e. chose a location a litte upstream or
downstream of the discharge for simulation).
4-6
-------
1. Specification of water body type (river or estuary);
2. Water quality constituent type;
3. Location of evaluation points relative to the discharge;
4. Geophysical parameters including flow, dispersion, cross
sectional area, reaction rates, background concentrations, and
dissolved oxygen saturation value;
5. Water quality standards to be checked; and
6. Rainfall and discharge record [storm flows and loads).
The input structure for the BRWS model is presented in Appendix A. An
example input deck is shown in Table 7-8 of the Kingston, New York applica-
tion of the simulation model (Section 7.2.5).
4.2.4 Model Output
The following output is produced from a BRWS model run:
1. Listing of input parameters used;
2. Time history plot of the rainfall, load, and concentration
calculated at each location specified;
3. Time history listing of the concentrations calculated at each
location specified; and
4. Summary for each location including the cumulative distribution
function, the number of occurances and the duration of standard
violations, and the average, standard deviation, and maximum
concentration simulated.
A complete application of the receiving water simulator for the Kingston,
New York example, is presented in Section 7.2.5. This includes portions of
the simulator output (Figures 7-17 (a)-(e)).
4.3 References
1. Areawide Assessment Procedures Manual, Volume II, Appendix A3 U.S.
Environmental Protection Agency, EPA-600/9-76-014, July 1976.
2. Grimsrud, G. Paul, E.J. Finnemore, and H.J. Oen, Evaluation of Water
Quality Models: A Management Guide for Planners3 Systems Control Inc.
for U.S. Environmental Protection Agency, EPA 600/5-76-004, July 1976.
3. Brandstetter, Albin, Assessment of Mathematical Models for Storm and
Combined Sewer Management, Battelle, Pacific Northwest Laboratories
for U.S. Environmental Protection Agency, EPA-600/2-76-175a, July 1976.
4. Lager, John A., T. Didriksson, G.B. Otte, Development and Application
of a Simplified Stormwater Management Model, Metcalf and Eddy, Inc.,
for U.S. Environmental Protection Agency, EPA-600/2-76-218, August 1976.
5. Hydroscience, Inc., and Consoer, Townsend and Associates, City of
Milwaukee, Wisconsin, Humboldt Avenue Pollution Abatement Demonstration
Project, Appendix I - Description of Detention Tank Model, September
1974.
4-7
-------
CHAPTER 5
NUMERICAL ESTIMATES FOR STORMWATER ASSESSMENT METHODOLOGIES
A variety of parameters are required for the calculation of stormwater
loads and impacts, by either the statistical method described in Chapter 3
or by simulators of the type discussed in Chapter 4 of this manual. These
include site specific characterizations of rainfall, runoff quantity and
quality, receiving water properties, and the performance and cost of treat-
ment alternatives. Guidelines for estimating these factors are presented
in this Chapter. While these guidelines are useful for initial assessments
and as a means of comparison, they are no substitute for site specific in-
formation. Methods for obtaining site specific information through a local
monitoring program are described in Chapter 6 of this manual.
5.1 Rainfall
A method for statistically characterizing rainfall is discussed in
Section 3.3. The following sections describe considerations for the use and
application of this method. Examples of precipitation characteristics in
different parts of the United States are presented. Various procedures for
defining the end of a storm event are examined. This is necessary for both
the statistical method and simulators. An examination of the applicability
of the gamma distribution for storm properties is included and methods are
presented for addressing the areal distribution of rainfall.
5.1.1 Precipitation Characteristics in Different Parts of the United States
Rainfall properties vary markedly from one part of the United States to
another. This section presents the results of analyses of raingage data
from eleven cities in the coterminous United States. The purpose of this
section is simply to indicate the general magnitude and range of storm
characteristics which occur in different areas. It does not replace the
need to obtain and analyze raingage data in a specific study location.
The eleven cities selected are located in major geographical subdivi-
sions of the country, as indicated in Table 5-1 and in Figure 5-1, which
also shows the distribution of average annual rainfall. The study cities
can be ordered in terms of increasing average annual precipitation: Pheonix
(10 in), Salt Lake City (15 in), Denver (15 in), Oakland (20 in), Detroit
(30 in), Dallas (35 in), Caribou (35 in), Portland (40 in), Boston (45 in),
Columbia (50 in), and Tampa (50 in). For a given year, however, the total
precipitation may vary considerably from the long term average. This is
demonstrated in Figures 5-2 (a, b, c, d).
5-1
-------
TABLE 5-1
CITIES SELECTED FOR RAINGAGE ANALYSIS
1. Caribou, Maine
2. Boston, Massachusetts
3. Columbia, South Carolina
4. Tampa, Florida
5. Detroit, Michigan
6. Dallas, Texas
7. Denver, Colorado
8. Salt Lake City, Utah
9. Pheonix, Arizona
10. Oakland, California
11. Portland, Oregon
Northeast
Northeast
Southeast
Southeast
Midwest
South
Rocky Mountain
Rocky Mountain
Southwest
Far West
Northwest
5-2
-------
CO
UJ
o
o
I-
co
o
*-
"
5 2
O K
"• t
£
o
UJ
o:
Q.
O
i
CO
o
i o
ll
5-3
-------
.
rsCVJCSI
HJ ki
r
o
o
10 O
"o:
O
IO
o
-------
»»
iQ
«
**
.
QIOIO
>I010
HJ ' I
J1
i"
(O*
O O O OO O O O O
« CM - « « «• 10 CM
OO
O
CM
ac.
o
o
w O
" o:
n >-
75
10 O
UJ SJ
S"-
® -I
Z
Z
(S3HONI) NOIlV.LIdlD3Wd 1VONNV
5-5
-------
9^
X "
§ <•>
r
o o o o o oo o o o
10 » 10 CM — co r-
-------
r
5 -
* "
«> k-2
O
u>
O
in
O
O O
V)
-------
For a few of the cities, the yearly rainfall is shown for two weather
stations, and differences are evident. For example, at Denver more rain
falls at the airport than at the city station. At Portland, this phenomenon
is reversed. Differences may be due to orographic effects or locations
relative to predominant storm paths in the area.
The synoptic analysis which separates the rainfall record into a series
of discrete storm events is described in Section 3.3. The statistical prop-
erties of the four event characteristics (duration, volume (depth), average
intensity, and storms interval) are determined. Seasonal patterns are
identified by grouping storms according to month.
The data from the eleven cities presented below are all analyzed by
requiring three consecutive dry hours to end a storm event. As discussed
in the Section 5.1.2, various criteria may indicate that a different storm
definition is more appropriate for certain cities. The statistical proper-
ties will change somewhat given a different storm definition, however, the
basic trends and seasonal patterns will remain basically unchanged.
The results of the monthly rainfall characterization are shown in
Figures 5-3 (a-k). A number of patterns are evident from the plots. The
mean event duration decreases during the summer months at each of the eleven
cities while the average intensity increases everywhere during the summer
except Oakland and Portland. The highest summer storm intensities occur in
Columbia and Tampa, followed by Dallas, Detroit, and Pheonix.
Longer storms tend to occur in the Northeastern and Northwestern por-
tions of the United States with mean durations of from 4 to 8 hours in
Boston, Caribou and Portland. The average duration in the rest of the
country generally varies from 2 to 6 hours. The cities with the highest
mean precipitation depth per storm are Columbia, Tampa, Dallas and Boston
where the average volume is generally greater than 0.25 inches throughout
the year.
Considerable geographic differences are indicated in the seasonal
patterns of the mean time between storms. The most frequent storms occur in
Portland during the non-summer months where the mean time between storms (A)
ranges from 30 to 40 hours. The most consistently frequent events occur in
Caribou where A ranges from 40 to 60 hours throughout the year. The least
frequent storms occur in Oakland and Phoenix during the summer, where A is
very high. Note the different plotting scale used for A in these cities.
Events occur much more frequently during the winter in Oakland, Phoenix, Salt
Lake City, and Portland; while more events occur during the summer in Tampa
and Denver.
The coefficient of variation of the time between storms (v ) is
generally near one. For those cities where v is significantly greater than
one, in particular, in the western portion of the country, the rainfall
analysis should also be examined with a minimum dry interval that defines a
storm of greater than 3 hours. The coefficient of variation of intensities
(v.) generally ranges from 0.8 to 1.6, though v. is somewhat higher in
Columbia, and lower in Oakland during the summer. The coefficient of
5-8
-------
MEAN VARIATION
— 0.18
fj: 0.15
\ 0.12
z 0.09
0.06
H
0.03
Q
STORM INTENSITY
-
-
_
— ^S^**.^
- _ ^j+^*^ ^>*~~»-«^
t ? T i i i i i i i i i
2.5
20
— 1.5
^
1 0
0 5
n
-
-
_
y^«s^ ^M
~" ~S ^f ^^ ^V
0_^_~ •— W ^ ^
-
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
8
X
^- 4
0
2
Q
-DURATION
- N^ J>
— ^"W**^
—
1 1 1 1 1 1 1 1 1 1 II
2.5
2.0
Su° L5
1.0
0.5
n
—
—
*- • -*- »
-
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
0.6
~ 0.5
* 0.4
" 0.3
> 02
O.I
Q
DEPTH
-
-
^Hf^f^^Hm^~~9t—
— • * *^"*~"'^ ^U">*
•*""^—^
—
1 1 1 1 1 1 1 1 1 1 II
2.5
2 0
-^ 1 5
1 0
0.5
n
-
JiL
_• •~»>s>»_^>y^ ^v^*
—
_
1 1 1 L i I 1 1 I 1 1 1
123456789 10 M 12 123456789 10 II 12
MONTH MONTH
180
— 150
-------
MEAN VARIATION
— 0.18
°E 0.15
I
\ 0.12
5 0.09
" 0.06
H
f\ ft"!
O.U3
o
STORM INTENSITY
_
—
_
_ ^
_ jy^^"^^***m.
. _ M' ^k"» • j.
1 1 1 1 1 I 1 1 1 1 1 1
2.5
2 0
1.5
^
1 0
0 5
n
-
_
-X^SSfLl
™" * ^J^~^&*+i^^^Q^^^ ^^^^"^^^^*—^
qf^ "' • ^
1 1 1 1 1 1 1 1 t 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
8
~~- 6
or
0
2
Q
-•s^^ DURATION /
N. -y^^
: ^-^
—
1 1 1 1 1 1 1 1 1 1 1 1
2.5
2.0
-0 15
1 0
0.5
r»
K—
__
-»^-^-« •-» * *~*^>-»
-
i i I 1 1 1 I I 1 1 1 i
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
0 6
-. 05
5 04
"•" 0.3
> 02
O.I
Q
DEPTH
-
- ^jr*
_•• *" *"-•<—, _^r~*~*^
*^™-^r
—
—
1 1 1 1 1 1 1 1 1 1 1 1
2 5
2.0
> 1 5
1.0
0.5
n
-
Ji ^"~*NL
- ~-+**~* »— • ^^^s
»~^
—
—
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
180
— 150
cc
X 120
~ 90
"^ 60
30
Q
TIME BETWEEN STORMS
_
—
- _-^*v
* > • 4 * • * T~*^^V« ~*
_
1 1 1 1 1 1 1 1 1 1 II
2 5
2 0
^° ' '5
'
0.5
n
-
-
-
* •- » i" «^_
* * ^^+~ "^-»
—
i i i i I I I 1 I I 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
FIGURE 5-3 (b)
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
BOSTON,MASSACHUSETTS
STATION 190770, (1948-1973)
5--10
-------
MEAN
VARIATION
H
0.18
0.15
0.12
0.09
0.06
0.03
0
STORM INTENSITY
1 234 5678 9 10 II 12
MONTH
2.5
2 0
1.5
I 0
0 5
0
2 3 4 5 6 7 8 9 10 II 12
MONTH
I
—
Q
- DURATION
1 1 1
1 1 1
2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
'-5
1.0
0.5
I I I 1 1 1 1 1 1 1 1 1
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
0.6
0.5
04
0.3
02
O.I
0
DEPTH
I I I I I I I I 1 I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
'•*
1.0
0.5
1 1 1 1 1 1 1 1 1 1 1 1
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
cc
T
180
150
120
90
60
30
0
TIME BETWEEN STORMS
1 1 I 1 1 1 1
1 1 1 1
1234 56 789 10 II 12
MONTH
2.5
2.0
l5
1.0
0.5
t i l i i l l 1 1
1 2 345 6 78 9 10 II 12
MONTH
FIGURE 5-3 (c)
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
COLUMBIA, SOUTH CAROLINA
STATION 381939, (1948-50, 54-73)
5-11
-------
MEAN
VARIATION
cc
i
H
0.18
0.15
0.12
0.09
006
0.03
0
STORM INTENSITY
J I
I I I I
2 34 5 6 78 9 10 II 12
MONTH
2.5
2.0
'-5
1.0
0.5
I I I I I I I I I I I I
123456789 10 II
MONTH
a:
X
- DURATION
I I I
I I
2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
I I I I I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
0.6
0 5
04
0.3
0.2
O.I
0
DEPTH
I
J 1
I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
I .5
1.0
0.5
I J _ I 1
1 I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
a:
x
ISO
150
120
90
60
30
0
TIME BETWEEN STORMS
I i
2 34 56 78 9 10 1112
MONTH
2 5
2.0
I .5
I 0
0.5
I I I
I I I I I
I 2 345 6 78 9 10 II 12
MONTH
FIGURE 5-3 (d)
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
TAMPA, FLORIDA
STATION 088788, ( 1948-51, 59-73)
5-12
-------
MEAN
VARIATION
— o.ie
§: o.is
> 0.12
? 0.09
"~ 0.06
0.03
0
STORM INTENSITY
1 I I I I I I I I I
2 34 5 6 78 9 10 II 12
MONTH
2.5
2.0
'•*
1.0
0 5
I I I I I I I I I I I I
I 2 3 4 5 6 7 6 9 10 II 12
MONTH
I
Q
-DURATION
I I
I l I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
I I I I I I I I I
123456 789 10
MONTH
12
0.6
0.5
0.4
0.3
0.2
O.I
0
DEPTH
I I I I I I I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
I .5
1.0
0.5
I I I I I I I
I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
180
T 150
I 120
~~" 90
« 60
30
0
TIME BETWEEN STORMS
J I
J_
J I
I 234 56 789 10 1112
MONTH
2 5
2.0
I .5
1.0
0.5
I I I I
I I I
I I I
12345678 9 10
MONTH
12
FIGURE 5-3 (e)
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
DETROIT, MICHIGAN
STATION 202103, (1960-1973)
5-13
-------
-~ 0.18
": 0.15
^ 0.12
5 0.09
""* 0.06
0.03
* n
8
7 6
X
~- 4
0
2
O.6
— 0.3
5 0.4
*•* 0.3
> 0.2
O.I
ISO
*•: 150
cr
i 120
~" 90
< 60
30
MEAN
_STORM INTENSITY
1 1 1 1 1 1 1 1 1 1 1 1
1 234 5678 9 10 II 12
MONTH
-DURATION
1 1 1 i I i 1 1 1 1 1 1
123456789 10 II 12
MONTH
DEPTH
1 I 1 1 1 1 1 1 1 1 It
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
_TIME BETWEEN STORMS
1234 56 789 10 II 12
MONTH
2.5
2.0
a7 '-5
1.0
0.5
2.5
2.0
a? '-5
.0
0.5
2.5
2.0
af ' 5
1.0
0.5
2 5
2 0
a.* l5
l.O
0.5
VARIATION
i i i i i i i i i l l i
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
^ . . , ,/\ ^*^^.
• • • +* \^ • • —
i t l l I I 1 I 1 I 1 i
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
I 1 I 1 1 1 1 1 1 1 1 1
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
i i i i i i i i i i l l
12 345 6 78 9 10 II 12
MONTH
FIGURE 5-3 (f )
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
DALLAS, TEXAS
STATION 412244, ( 1941-46, 48-73)
5-14
-------
H
CC
I
o:
x
MEAN
0.18
0.15
0.12
0.09
0.06
0.03
0
STORM INTENSITY
23456789
MONTH
-DURATION
I I I
J I
I l 1
I 23456789 l(
MONTH
0.6
0.5
0.4
0.3
0.2
O.I
0
180
ISO
120
90
60
30
0
DEPTtf
I I I I I
lilt
234567891
MONTH
TIME BETWEEN STORMS
I I
1 I
I 2
3456789
MONTH
II 12
j I
II 12
J I
) II 12
II 12
VARIATION
2.5
2.0
'• =
1.0
0.5
I I I I I I I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
I I I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2 5
2.0
I 5
1.0
0.5
I I I I I I I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2 5
20
I .5
1.0
0.5
I I I I I I I I I
I 2 345 6 78 9 10 II 12
MONTH
FIGURE 5-3 (g)
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
DENVER,COLORADO
STATION 052220, (1948-1973)
5-15
-------
MEAN VARIATION
— 0.18
^ 0.15
X
> 0.12
^. 0.09
~" 0.06
H
0.03
o
STORM INTENSITY
_
—
_
"~ ^ -fc.
-^ , _L , t -*/ >»-^^ ^
1 1 1 1 1 1 1 1 1 1 1 1
2.5
2 0
<- 1.5
1 0
0.5
n
-
_
_
^^"K^,
~" *J^^^f ^^^(L
••••^..l^^.^^ * ^ ^MrtA
-
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
8
~~- 6
tr
X
— 4
O
2
o
-DURATION
^ ^
~ • • • ^^ jf** *
— 1|L _/^
«k>^
~
1 1 1 1 1 1 1 1 1 1 1 1
2.5
2.0
x° 1.5
1.0
05
n
1
—
- *~~***+'~*^*^J*^~*—+-*
^ir
i i i i I i i i I I I i
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
0.6
— 0.5
Z 0.4
~" 0.3
> 0.2
O.I
Q
DEPTH
-
-
_
-^
_ *r**~~~ **" **" * ~"
1 1 1 1 1 1 1 1 1 1 1 1
2.5
2.0
> 1 5
1.0
0.5
n
-
-
^^^ y^^9^\^ _ ^
*^r ^""*w * *
"
i i i i i i i 1 i 1 I i
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
180
—. 150
CC
X 120
"" 90
o
60
30
n
TIME BETWEEN .
STORMS A
/\
/ \
/ *^S^,
1 ^^^
^^•— •^^_-*-* »
_•--• ^>^,-»^
1 1 1 1 1 1 1 1 1 1 II
2.5
2.0
-» 1 .5
1 0
0.5
n
-
-
JJL
~ »^ J>~*^*^-/*~~*~~ *^^»
^ ir*~ ~~ ^w
—
—
i i i i i i i i i i i i
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
FIGURE 5-3 (h)
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
SALT LAKE CITY, UTAH
STATION 427598, (1948-1973)
5-16
-------
MEAN
VARIATION
— • 0.18
* 0.15
^ 0.12
5 0.09
0.06
0.03
0
H
STORM INTENSITY
I I I I I I I I I
J I
I 234 5678 9 10 II 12
MONTH
2.5
2.0
'•»
1.0
0.5
I I I I I I I I I I I
I 2 3 4 5 6 7 8 9 IO 1112
MONTH
-DURATION
I I
2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
I I I I I I I I
I 2 3 4 5 6 7 6 9 10 II 12
MONTH
0.6
0.5
0.4
0.3
0.2
O.I
0
1050
900
~ 750
X 600
" 450
*" 300
150
0
DEPTH
I I
I 2 3 4 S 6 7 8 9 10 II 12
MONTH
TIME BETWEEN STORMS
I I
I 234 56 789 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
I I I I I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
I I I
I 2 345 6 78 9 10 II 12
MONTH
FIGURE 5-3 ( i )
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
PHEONIX, ARIZONA
STATION 026481, (1948-1973)
5-17
-------
MEAN
VARIATION
*—
ad
X
^
2
*-*
H
0.18
0.15
0.12
O.O9
0.06
0.03
0
STORM INTENSITY
—
-
-
-
_
A • • tt atttB
— ''^k.^ ./" ' ~ ^
i i i l i iT~T i i i i
2 34 5678 9 10 II 12
MONTH
2.5
2.0
'•*
1.0
0.5
I I I I I I I I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
ac.
I
-DURATION
l I I i I l
I l I
2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
I I I
II I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
0.6
0.5
0.4
0.3
0.2
O.I
0
DEPTH
I I I I I I
I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
1050
9OO
~ 750
I 600
""' 450
^ 3OO
150
0
TIME BETWEEN
- STORMS
I 234 56 789 10 II 12
MONTH
2.5
2.0
I .5
1.0
0.5
I I I 1 1 J L I I I I I
I 2 3 4 5 6 7 8 9 10 II 12
MONTH
2.5
2.0
1.5
1.0
0.5
i i i J l
I 2345 678 9 10 II 12
MONTH
FIGURE 5-3 (j )
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
OAKLAND, CALIFORNIA
STATION 046335, (1948-1973)
5-18
-------
-~ 0.18
* 0.15
>; 0.12
5 0.09
""" O.O6
H 003
8
d 6
X
~- 4
Q
a
0.6
~ 0.5
Z 0.4
~ 0.3
> 0.2
O.I
180
T 150
o:
I 120
"~" 90
* 60
30
MEAN
STORM INTENSITY
1 1 I T 1 1 1 1 1 1 1 1
1 234 5678 9 IO II 12
MONTH
DURATION
1 1 I i 1 1 i 1 1 1 I 1
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
DEPTH
1 I I 1 1 I 1 1 1 I II
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
TIME BETWEEN
STORMS
LA.
i t i i i i i i i i i i
I 234 56 78 9 IOM 12
MONTH
2.5
2.0
*- '•»
1.0
0.5
2.5
2.0
^ '•»
1.0
0.5
2.5
2.0
^> IS
^ ' -3
1.0
0.5
2.5
2.0
5* 1.5
1.0
0.5
VARIATION
|.~~A-
i i i i i i i i i i i i
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
t • • . • -^*-— . - ^
I I I I 1 I I I 1 I 1 I
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
"A-.^.-.— -
i i i i i i i i i i i i
1 2 3 4 5 6 7 8 9 10 II 12
MONTH
i i i i I I I I I I 1 1
J 2 345 6 78 9 10 II 12
MONTH
FIGURE 5-3 (k)
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
PORTLAND, OREGON
STATION 356751, (1948-1973)
5-19
-------
variation of durations (v,) is generally near one while the coefficient of
variation of unit volumes (v ) ranges from 1.0 to 2.4.
5.1.2 Definition of Storm Event
The grouping of hourly rainfall records into discrete events requires
a definition of the end of a storm. This is conveniently specified by some
minimum number of consecutive dry hours that marks the end of a storm event.
A number of procedures for selecting the minimum inter-event time have been
proposed.
Heaney, et al (1) and Howard (2) suggest an autocorrelation analysis of
the hourly rainfall records to determine the time lag at which the hourly
rainfall records become uncorrelated. An example of this type of analysis
is presented in Figure 5-4. The autocorrelation coefficient of hourly rain-
fall is shown as a function of the lag time. At a lag time of zero the
autocorrelation equals one, as each rainfall hour is completely correlated
with itself. The other points on the curve indicate the correlation of data
from a given hour with data of another hour at the indicated interval. The
95 percent tolerance limits for significant nonzero correlation are also
shown. The first minimum in the autocorrelation function occurs at 10 hours,
although insignificant correlation (at the 95 percent probability level) is
reached at about 7 hours. This is then used to indicate the appropriate
interevent time to separate the rainfall record into independent events.
Heaney et al (1) and Howard (2) suggest using the first minimum (in this
case, 10 hours), although the point where the autocorrelation is no longer
statistically significant may also be considered.
Although this method is properly directed towards defining when
storm events may be considered independent, there are a number of difficul-
ties in the approach. For example, the appropriate tolerance limits for
considering records uncorrelated must be chosen. More fundamentally, how-
ever, the correlation procedure does not limit itself to the correlation of
rainfall records which are separated by a given number of consecutive dry
hours. Since the analysis is made for the entire rainfall record, the
autocorrelation coefficient includes correlation within storms as well as
between storms. For example, during a nine hour period of continuous rain-
fall, there are five occurrences of rainfall at a given hour and its four
hour lag. These occurrences inflate the autocorrelation coefficient for
short lag times, while indicating nothing about the independence of data
after a given number of consecutive dry hours. A more appropriate analysis
would isolate those periods separated by a given number of dry hours, and
check for correlation between these.
Another approach is to plot the number of events per year versus the
minimum inter-event time. The point at which an increase in the minimum
inter-event time does not result in a correspondingly significant reduction
in the number of storm events is considered adequate. As indicated by
Heaney, et al (1), however, this graphical approach may not always indicate
a well-defined transition point.
5-20
-------
1.0
0.8
H 0.6
0.4
-0.2
M% T.L.
»»% T.L.
I
0 10 20 30 40 SO 60 70 80 90 100
LAG K,HOURS
REPRODUCED FROM
REF.(|)
FIGURE 5-4
LAG-K AUTOCORRELATION FUNCTION OF DES MOINES, IOWA
HOURLY RAINFALL, 1968
5-21
-------
A third approach has been examined in the development of this manual.
The approach is based upon the characterization of the rainfall process as a
random, Poisson process, as described in Section 3.1. If storm events occur
in this fashion, there are a number of implications. One is that the time
between events (6) is an exponentially distributed random variable (equiva-
lent to a gamma distributed variable with coefficient of variation (v ) equal
to one). Grouping the storm events (i.e. selecting the number of dry^ours)
such that this property holds, provides a useful criteria for selecting the
minimum inter-event time.
It is observed that when the statistical rainfall analysis is performed
with a short minimum inter-event time, the coefficient of variation (v ) is
greater than one. As the minimum inter-event time is increased, v decreases
and eventually becomes less than one. This is demonstrated in Table 5-2
using rainfall records from Newark, New Jersey. Note that greater inter-
event times also result in fewer storms, greater average durations (D),
greater average depths (V), and a greater average time between storms (A).
The important point, however, is that at an inter-event time of about 4
hours, v is equal to one.
To demonstrate that 6 is exponentially distributed with the minimum 4
dry hour storm definition, the cumulative density function for 6 is plotted
in Figures 5-5 (a, b, and c) with a 3, 4 and 6 dry hour storm definition
respectively. The gamma distribution fits the observed data quite well, and
in particular, the case when v. = 1.02 (minimum inter-event period of 4
hours) closely corresponds to the special case of the gamma distribution when
v = 1, which is the exponential distribution. A minimum inter-event time of
4 hours is chosen for the storm definition. A similar analysis was conducted
for the Minneapolis-St. Paul raingage and a minimum inter-event time of 6
hours was found to yield v, nearly equal to one. The cumulative density
function for 6 given this definition is shown in Figure 5-6.
It is important to note that while the guideline suggested for choosing
the storm definition is useful, it is by no means absolute. While a random
storm process will have 6 exponentially distributed, this does not guarantee
that the events occur randomly. It is felt, however, that barring marked
seasonal changes in 6, the time between two given storms should be indepen-
dent of the time between any other two storms. Seasonal patterns should be
noted and the storm definition may be chosen to have v. nearly equal to one
during a particular critical or design period. Finally, the effects of
alternative storm definition should be included as part of the sensitivity
analysis in the overall assessment. For example, the differences in mean
unit storm volume (V) indicated in Table 5-2 may change the predicted per-
formance of different size storage units in the Newark area.
5.1.3 Applicability of the Gamma Distribution for Storm Event
Characteristics
The two parameter gamma distribution has been used in this manual for
the probabilistic analysis of rainfall events and their impacts, with little
discussion or background as to why it was selected. This section presents a
more detailed presentation of the applicability of the gamma distribution
5- 22
-------
LJ
3
>
o
o
UJ
cc
O
V)
CO
o
a:
99
98
97
96
95
94
93
92
9 I
90
80
70
60
50
4O
30
20
•I/=l.25
J L
"01 2345
MULTIPLES OF THE MEAN
TIME BETWEEN STORMS, A
LEGEND:
THEORETICAL GAMMA DISTRIBUTION
-•- OBSERVED DISTRIBUTION
NOTE
MINIMUM 3 DRY HOURS BETWEEN STORM.S
(A = 65.6HR, l/s= 1.09)
FIGURE 5-5 ( a)
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF TIME BETWEEN STORMS
NEWARK AIRPORT, RAIN6A6E 286026, 1948-1975
5-23
-------
"C I 2345
MULTIPLES OF THE MEAN
TIME BETWEEN STORMS,A
LEGEND
THEORETICAL GAMMA DISTRIBUTION
—•— OBSERVED DISTRIBUTION
NOTE
MINIMUM 4 DRY HOURS BETWEEN STORM.S
(A -71.5 HR,I/g= 1.02)
FIGURE 5-5 (b)
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF TIME BETWEEN STORMS
NEWARK AIRPORT, RAINGAGE 286026, 1948-1975
5-24
-------
12345
MULTIPLES OF THE MEAN
TIME BETWEEN STORMS,A
IfGf/VO.
THEORETICAL GAMMA DISTRIBUTION
—•— OBSERVED DISTRIBUTION
NOTE
MINIMUM 6 DRY HOURS BETWEEN STORMS
(A = 80.8 HR, ZVS = 0.9I)
FIGURE 5-5 (c )
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF TIME BETWEEN STORMS
NEWARK AIRPORT, RAINGAGE 286026, 1948-1975
5-25
-------
12345
MULTIPLES OF THE MEAN
TIME BETWEEN STORMS, A
LEGEND.
THEORETICAL GAMMA DISTRIBUTION
—•— OBSERVED DISTRIBUTION
NOTE
MINIMUM 6 DRY HOURS BETWEEN STORMS
( A = 84HR,l/= 1.02)
FIGURE 5-6
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF TIME BETWEEN STORMS
MINNEAPOLIS/ST PAUL AIRPORT, RAINGAGE 215435, 1948-1973
5-26
-------
CM
1
LO
CQ
H
P^
O
H
p^l
»H
P-,
OJ
O
-J.
0
CO
PJ
rH
H
2
ct
PJ
J
p.
o
H
U
PJ
PL,
p.,
PJ
•0
0) w
6 S ^
•H +-> O
H 0) +j
DQ CO -77
E £
00
LO
Ci ?>
0)
LO e
£> ,H
rH o ' — <
CH >> -H^
CQ
•z.
PJ
u
w
Q
1
I "
•-1 x
>* -H
S (/)
" s ^
rv- C1"1 ^-~
0 ^ -H
B. ' — '
Oi
HH
<
^
Di TJ
< P
^ C
m §
2 — '
CM
LU
CJ
U • S $H fn
2 O S H> cj
I-H 2 o a. >
c2 co
o
1— 1
LO
CM
to
1— 1
to
o
f —
to
1-1
LO
0
o
O)
o
.
1— 1
o
LO
•
LO
to
to
i-H
CM
o
I— 1
LO
1— 1
LO
1— 1
r^-
to
o
CM
to
1-1
LO
0
0
Ol
o
•
rH
vO
CM
•
^o
CM
CM
iH
O)
o
LO
to
LO
V«O
to
o
to
to
1-1
LO
o
0
t^.
o
1— 1
00
•
vO
,-,-
1— 1
1— 1
1— 1
01
o
00
o
00
0
LO
LO
oo
to
0
to
to
1 — 1
LO
o
o
t-^
o
•
I-H
LO
vO
•
r —
00
o
rH
00
OO
o
to
00
rj-
CT>
to
o
^3-
to
1-1
LO
0
o
LO
o
•
1— 1
^*
rH
•
00
LO
0
1 — t
*2
LO
O
00
to
OO
to
to
LO
tH
to
O)
CM
00
00
CM
g X
•H f-l
C Q
C w
O CO
CQ
to
LO
5- 27
-------
for storm event characterization and its use in the statistical method for
the assessment of runoff and treatment.
The distributions of rainfall event characteristics (duration, unit
volume, average intensity, and time between storms) are quite skewed. There
are a few very large observations, with many relatively smaller ones. Dis-
tributions which are commonly used to represent such highly skewed random
variables include the lognormal, Weibull, exponential, and gamma probability
density functions.
The gamma distribution has traditionally found application in the
analysis of precipitation data (3,4). Howard bases his analysis of storm-
water storage and treatment on the assumption that event durations, intensi-
ties, and the time between storms are exponentially distributed, that is, a
special case of the gamma distribution with the coefficient of variation
equal to one (2). Chow and Yen use the gamma distribution to describe the
basic rainfall event characteristics (5,6).
The Weibull distribution is put forth by Eagleson for the time between
independent storm events and event durations; and the gamma distribution is
suggested for storm depths (unit volumes) (7). Note that the Weibull dis-
tribution for the special case when the coefficient of variation (v) equals
one, is also equivalent to the exponential distribution and that the gamma
and Weibull distributions are very similar when v is nearly equal to one,
as is usually the case for the time between storms and event durations.
Smith has recently indicated that the lognormal distribution may also be
appropriate for rainfall analysis in some cases (8).
To illustrate comparisons of observed cumulative density functions and
the gamma distribution chosen for use in this manual, the hourly rainfall
data from Central Park, New York, and the Minneapolis-St. Paul Airport are
analyzed. The resulting empirical distributions are shown in Figures 5-7
through 5-14. Note that the observed distributions are well represented by
the theoretical gamma curves in all cases except for the intensity at the
Minneapolis-St. Paul raingage, where a lognormal distribution results in a
better fit. This type of comparison should be conducted for at least one
gage in a particular study area, to check the applicability of the basic
probabilistic assumptions. Note also that the comparisons shown in Figures
5-7 through 5-14 accentuate the less frequent end of the distribution
function, as most of the stormwater analyses presented in this manual are
predominantly influenced by this range. Comparisons in the more frequent
range may not be as favorable, particularly for the time between storms
which has a practical minimum determined by the inter-event time selected
for storm definition. A small shift in the distribution to represent this
minimum may be appropriate.
The gamma distribution is selected for general application in this
manual because it is sufficiently accurate in most cases, while simple
enough for determining analytical solutions for loading, receiving water,
and treatment device evaluations. There is nothing absolute about the
gamma distribution, however, and other functions may fit observed distribu-
tions more closely in selected areas. Given the level of uncertainty in
5-28
-------
2345
MULTIPLES OF THE MEAN
DURATION, D
LEGEND.
THEORETICAL GAMMA DISTRIBUTION
—• — OBSERVED DISTRIBUTION
NOTE
MINIMUM 4 DRY HOURS BETWEEN STORM.S
( 0 = 6 62 HR , Z/d = 1.02 )
FIGURE 5-7
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF DURATION
CENTRAL PARK, NEW YORK, RAINGAGE 305801, 1948-1973
5-29
-------
MULTIPLES OF THE MEAN
INTENSITY, I
- THEORETICAL GAMMA DISTRIBUTION
-•— OBSERVED DISTRIBUTION
NOTE
MINIMUM 4 DRY HOURS BETWEEN STORMS
(I = .055 IN./HR., I/ = 1.38)
FIGURE 5-8
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF INTENSITY
CENTRAL PARK, NEW YORK, RAINGAGE 305801, 1948-1975
5-30
-------
Ul
uo
O
Ul
a:
o
Z
<
X
to
I/)
uj
o
a:
ui
a.
MULTIPLES OF THE MEAN
UNIT VOLUME, V
LEGEND:
THEORETICAL GAMMA DISTRIBUTION
-•— OBSERVED DISTRIBUTION
NOTE
MINIMUM 4 DRY HOURS BETWEEN STORMS
( V = 0. 37 IN., i/v = 1.51 )
FIGURE 5-9
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF UNIT VOLUME
CENTRAL PARK, NEW YORK, RAINGAGE 305801, 1948-1975
5-31
-------
13345
MULTIPLES OF THE MEAN
TIME BETWEEN STORMS,A
LEGEND
THEORETICAL GAMMA DISTRIBUTION
—•— OBSERVED DISTRIBUTION
NOTE
MINIMUM 4 DRY HOURS BETWEEN STORM.S
(A = 75.6 HR, I/ = 0 97)
FIGURE 5-10
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF TIME BETWEEN STORMS
CENTRAL PARK, NEW YORK, RAINGAGE 305801,1948-1975
5-32
-------
99
12345
MULTIPLES OF THE MEAN
DURATION,D
LEGEND
THEORFTICAL GAMMA DISTRIBUTION
—•- OBSERVED DISTRIBUTION
NOTE
MINIMUM 6 DRY HOURS BETWEEN STORM.S
( 0 = 6 30HR, Z/d = I. 14 )
FIGURE 5-11
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF DURATION
MINNEAPOLIS/ST PAUL AIRPORT, RAINGAGE 215435, 1948-1975
5-33
-------
MULTIPLES OF THE MEAN
INTENSITY, I
THEORETICAL GAMMA DISTRIBUTION
—•— OBSERVED DISTRIBUTION
NOTE.
MINIMUM 6 DRY HOURS BETWEEN STORMS
IN /HR., Z/;=l.73)
FIGURE 5-12
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF INTENSITY
MINNEAPOLIS/ST PAUL AIRPORT, RAINGAGE 215435, 1948-1975
5-34
-------
UJ
UJ
<
r>
o
UJ
a:
O
CO
00
a:
UJ
MULTIPLES OF THE MEAN
UNIT VOLUME, V
£ EG END.
THEORETICAL GAMMA DISTRIBUTION
-•- OBSERVED DISTRIBUTION
NOTE
MINIMUM 6 DRY HOURS BETWEEN STORMS
( V = 0 25 IN., Z/v = I. 56 )
FIGURE 5-13
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRI BUTION OF UNIT VOLUME
MINNEAPOLIS/ST PAUL AIRPORT, RAINGAGE 215435,1948-1975
5-35
-------
CL
MULTIPLES OF THE MEAN
TIME BETWEEN STORMS, A
L EGENO
THEORETICAL GAMMA DISTRIBUTION
-•-- OBSERVED DISTRIBUTION
NOTE
MINIMUM 6 DRY HOURS BETWEEN STORM.S
(A = 84HR,Z/g=l 02)
FIGURE 5-14
COMPARISON OF OBSERVED AND THEORETICAL
CUMULATIVE DISTRIBUTION OF TIME BETWEEN STORMS
MINNEAPOLIS/ST PAUL AIRPORT, RAINGAGF 2154^5, 1948-1973
5-36
-------
stormwater analyses, the error introduced by the misrepresentation of the
distribution function is not expected to be of major importance in the over-
all assessment.
Another consideration that is appropriate for more detailed probabilis-
tic analyses is the relationship between the different event characteristics.
Eagleson describes joint probability density functions for event durations
and storm depth (7). Chow and Yen use conditional probability relationships
in their analysis, and demonstrate an inverse relationship between average
intensities and durations in Urbana, Illinois. A portion of this relation-
ship may be due to the seasonal effects discussed previously, and the
seasonal segregation of records may result in a reasonable level of indepen-
dence between i and d. Their approach is useful, however, particularly when
more than one storm characteristic must be specified, such as in the analysis
of instream concentrations presented in Section 3.5.2.1, where both runoff
flows and event durations are required to evaluate the dispersion of the
storm pulse.
5.1.4 Areal Distribution of Rainfall
The methods presented for the statistical characterization of storm
properties are thus far limited to point rainfall records. Although the
rainfall occurring at any location may have relatively similar long-term
properties throughout a study area, any method of calculating runoff and
pollutant loads caused by storm events require rainfall depths over portions
(and for some of the analyses, all) of the study area. Because precipitation
does not occur simultaneously over the entire study area, but varies both in
time and space, areal rainfall properties differ somewhat from those measured
at a single location.
Rodriguez-Iturbe and Mejia provide a good review of studies dealing with
the transformation of point to areal rainfall (9). These studies concentrate
primarily on estimating the average rainfall occurring over an area during a
specified interval of time such as a month, a day, an event, an hour, etc.
The rainfall volume recorded at a single gage is reduced as a function of the
size of the area to estimate the average areal rainfall. The estimating
procedures have developed from empirically based curves, to theoretical
estimates based on the spatial and temporal correlation properties of pre-
cipitation in a particular region.
An example of the type of correlation analysis performed is demonstrated
in Figure 5-15 which shows the correlation coefficient of monthly rainfall
totals between raingages in the New York City Metropolitan area. Four years
(48 observations) of data from 12 raingages are analyzed. The 12 gages
result in 66 possible combinations (without duplication) for which correla-
tions are performed. The results indicate the expected general decrease in
the correlation coefficient as a function of the distance between gages.
Note, however, that even at the most distant gages, the correlation coeffic-
ients are very high. This shows the areal uniformity of monthly rainfall
totals in this region. Correlation coefficients for shorter time intervals,
however, such as daily or hourly rainfall, can be expected to fall off much
5-37
-------
*.!••• .
UJ
or
8(0
UJ
v>
UJ
O
O <
O
UJ
UJ
UJ
o
or
o
ui
z
UJ
X
h-
V)
8
to
10 ->
UJ <
or j£
o <
X
I-
o
I-
_l
UJ
00
d
W
o
CM
o
1N3IOUJ300
5-38
-------
ore rapidly with distance, although lag correlation computations may be
ppropriate if storms tend to move across the area in a favored direction.
While the principal application of methods to transform point to areal
•ainfall has been to correct the average depth of particular events, the
lethods applied in this manual are based primarily on long-term rainfall
statistics; in particular, the mean and coefficient of variation of storm
Intensities, durations, depths, and the time between events. A raingage
aggregation method for estimating these statistical event properties over an
area, rather than at a single point, is developed and applied in the follow-
ing section.
5.1.4.1 Raingage Aggregation
Rainfall records from different gages in a study area may be aggregated
to form a single record representative of the area. The procedure is out-
lined in Figure 5-16. The hourly rainfall of the aggregated record is cal-
culated as a weighted average (in this case, with equal weights) of the three
individual records. A computer program which combines raingage records over
a selected time period, with selected weighting factors for each gage, is
used. The synoptic analysis program which computes event statistics then
analyzes the combined record.
The raingage aggregation analysis is performed using records from
January 1, 1950 - December 31, 1960 (11 years) at nine gages in the New York
Metropolitan area, shown in Figure 5-17. Six pairwise gage combinations are
examined as well as a combination of all nine gages. Equal weights are used
in all the analyses. The results of the synoptic event analysis for each of
the nine original records, and the seven combined records, are shown in
Table 5-3(a). Table 5-3(b) lists the ratio of the event statistics of the
combined record (D', v,', etc.) to the event statistics of the individual,
point records. The demoninator used to represent the point records (D, v,,
etc.) is the average of the statistics of the individual gages in the combin-
ed record (Table 5-3). The ratios are plotted in Figure 5-18 as a function
of the distance, x, between the gages. The combination of all nine gages is
shown with a triangular symbol at the edge of the plot, since distance
between gages is not appropriate for a nine gage aggregation.
Clear patterns are evident in the modification of event statistics as
the distance between gages increases. The average storm intensity decreases
with greater gage separation, indicating the smoothing effect of the areal
processing. There are more storms, as indicated by the drop in A'/A, but
the average unit volume (depth) per storm is smaller as indicated by the
decrease in V'/V. The consistency of the drop in time between storms (A)
and the drop in unit volume (V) is due to the uniformity of total rainfall
quantities when considering a long-term period. The average duration is
shown to decrease by a factor of about 0.94 and is relatively independent of
the distance between gages. This decrease in average duration seems to be
due to the fact that most of the additional storms added to the record
(storms which occur at one gage but not at the other), are of short duration.
5-39
-------
0.3
V. 0.2
Z O.I
CO
z
UJ
<
u_
z
<
a:
GAGE 1
TIME
0.3
0.2
O.I
n
~GAGE 2
:
boRl
1
^\
£5 lN>
1
vl R3
TIME
0.3
!|: 0.2
I °''
>•
i 0
CO
z
UJ
z
- 0.3
_l
Z
5 0 I
GAGE 3
TIME
AGGREGATED RECORD= (GAGE I -f GAGE 2 + GAGE 3 ) -=- 3
^
TIME
FIGURE 5-16
METHOD FOR RAINGAGE RECORD AGGREGATION
5-40
-------
•QUID tnocw
JERSEY
NEW MILFORD
( H ) •
\ 4 \ / 'fj>
~'"T » \ «—A,Y «"<>" ?VK
«? f fr X f /C, /'.*r»l I I V
/f^Af1^
CENTR/&./PK.,;
/ , ill /
**""I
NEWARK A.P "^,^.. ,,J j(F)j^
"'","-/ *-ffA°VTERY
RAH WAY
STATEN
.A ,.
f ' —- / ^ ,
x '
,* -----
NE* JESSEY
FIGURE 5-17
RAINGAGE AGGREGATION LOCATIONS
5-41
-------
1 0
H °'9
H 0.8
0.7
0 6
INTENSITY
-.
- * * •
* •
- LEGEND
_ »-PAIRS OF GAGES
A-NINE GAGE COMBINATION t
1 1 1 1 1 1 1 1
1 .0
> 0.9
0.8
L
n c
_
—
-•• .
— t
—
—
I i i i i i i i
k
0 5 10 15 20 25 30 35 40 05 10 15 20 25 30 35 40
X (MILES) X (MILES)
l.O
O
^ 0.9
0
0.8
0 7
DURATION
—
1.0
— • • • • A -o
•• * * ^
—
—
~ 1 1 1 1 1 1 1 1
\ 0.9
^
0.8
• • |
— • — • —
—
_
-
i i i i i i ii
0 5 10 15 20 25 30 35 40 O 5 10 15 20 25 30 35 40
X (MILES) X (MILES)
_\ 0.9
0.8
_ DEPTH
-\ • .
~~ i
\ \ 1 1 1 1 1 1
> 0.9
0 8
-
-
i i i i i i ii
*
0 5 10 15 20 25 30 35 4O 05 10 15 20 25 30 35 40
X (MILES ) X (MILES)
_> 0.9
**
0.8
_TIME BETWEEN STORMS
*• 0
—
* .
i
~ 1 1 1 1 1 1 1 1
«>
_\ 0.9
SJ°
^ 08
-
_ • * '
-
—
—
i i i i i i ii
k
05 10 15 20 25 30 35 40 05 10 15 20 25 30 35 40
X ( MILES ) X (MILES)
FIGURE 5-18
EFFECT OF RAINGAGE AGGREGATION
ON STATISTICAL EVENT PROPERTIES
5-42
-------
TABLE 5-3 (a)
SUMMARY OF NEW YORK CITY RAINGAGE AGGREGATION ANALYSIS
PERIOD OF RECORD: 1/1/50 - 12/31/60
(Minimum 4 Dry Hours Between Storms)
Time
Between
Duration Intensity Depth Storms
D
Gage
(hr) vd Cin/hr)
V
(in)
V /-I. -v V,
v (hr) 5
A.
B.
C.
D.
E.
F.
G.
H.
I.
Ave. V, Brooklyn 6.58 1.01 0.059 1.57 0.384 1.52 76.0 0.98
Central Park
La Guardia
Airport
Staten Island
Scarsdale
Battery Park
Newark Airport
New Milford
Rahway
6.62 1.05 0.052
6.52 1.09 0.051
6.69 1.09 0.052
7.64 0.99 0.056
6.72 1.09 0.052
6.62 1.06 0.051
7.39 0.99 0.058
7.07 0.96 0.060
1.32 0.344 1.46 74.9 0.92
1.23 0.364 1.73 69,3 0.95
1.39 0.379 1.61 71.1 0.92
1.42 0.436 1.62 76.5 0.94
1.22 0.373 1.63 76.0 0.93
1.22 0.348 1.50 70.9 0.99
1.21 0.420 1.47 82.9 0.92
1.45 0.415 1.46 85.4 1.00
Number
Storms
1234
1277
1231
1336
1213
1259
1348
1154
1105
a. Combination of
B and C
b. Combination of
B and F
c. Combination of
A and G
d. Combination of
B and D
e. Combination of
H and I
f. Combination of
E and I
6.11 1.05 0.050 1.13 0.335 1.59 71.4 0.91 1350
6.13 1.07 0.049 1.15 0.330 1.58 72.9 0.93 1322
6.25 1.05 0.051 1.22 0.345 1.51 70.8 0.99 1362
6.27 1.10 0.047 1.17 0.331 1.57 69.7 0.90 1384
6.77 1.03 0.049 1.12 0.355 1.54 72.7 0.95 1325
6.97 1,05 0.048 1.45 0.360 1.63 67.7 0.96 1423
g. Combination of
A through I
6.54 1.09 0.037 1.07 0.299 1.70 60.0 0.93 1606
-------
TABLE 5-3Cb)
RATIO OF AGGREGATED RAINGAGE STATISTICS
TO POINT RAINGAGE STATISTICS
Distance ,_. _ .
„ ,. ^. „ ^ Time Between
Combination Between r> *• -r + • *. ^ i. ^^.
,, r Duration Intensity Depth Storms
of Gages x ' r
Gages (miles) D'/D V/Vd I'/I Vj'/Vi V'/V V/Vv A'/A V5f/V6
a.
b.
c.
d.
e.
£.
g.
B and
B and
A and
B and
H and
E and
A thru
C
F
G
D
I
I
I
4.8
6.5
12.0
16.8
28.2
37.0
_
0.93
0.92
0.95
0.94
0.94
0.95
0.95
0.98
1.00
1.01
1.03
1.06
1.08
1.05
0.97
0.94
0.93
0.90
0.83
0.83
0.68
0.89
0.91
0.87
0.86
0.84
1.01
0.80
0.95
0.92
0.94
0.92
0.85
0.85
0.78
1.00
1.02
1.00
1.02
1.05
1.06
1.09
0.99
0.97
0.96
0.95
0.86
0.84
0.79
0.97
1.01
1.01
0.98
0.99
0.99
0.98
Note: Point raingage statistics (D, v,, I, etc.) calculated as average of
contributing gages from Table 5-3.
5-'44
-------
This apparently overrides the increased duration of storms which are lagged
from one gage to the other, an effect illustrated in Figure 5-16.
The coefficient of variation of intensity (v.) decreases with distance
between gages, again demonstrating a smoothing effect. An exception to this
is noted for one of the combinations (c) where missing records from one gage
correspond to an intense storm at the other. The aggregation program bases
the created record on only the available records, so an intense storm is
included with the other smoothed storms, resulting in an overestimation of
v.. The coefficient of variation of durations (v,) and depths (v ) increase
slightly with the distance between gages, while tne coefficient of variation
of the time between storms (v.) remains relatively unchanged.
To use the results shown in Figure 5-18 to modify the event statistics
for drainage areas in the New York area, a transformation is needed from
the distance between two gages to the size of the area they represent.
Methods suggested for gage correlation analysis (9) based on the average
distance between randomly selected pairs of points in an area, do not appear
appropriate for this approach, as the aggregated record clearly represents
only the area between the gages. Estimates could be based on the assumption
that the gages are on opposite points of a regular geometric-figure: sides
of a square (A = x ); of the diagonal of a square (A = 0.5 x ); the diameter
of a circle (A = 0.79 x ); the bisecting segment of an equilateral triangle
(A = 0.5 x ); etc. Some guidance is gained by examining the results of the
nine gage combination, which may be roughly outlined to cover an area of
about 1,000 square miles. Based on the shape of the curves generated by the
two gage combinations in Figure 5-18, the results generated by the nine
gages (triangular symbol) correspond to a distance of about 45 miles between
gages. An estimate of A = 0.5 x would then yield A = 0.5 (45) = 1,000
square miles. A factor of one half therefore seems reasonable for converting
x to A in the New York study area.
To illustrate the application of the areal rainfall modification,
consider a 30,000 acre (47 square mile) drainage area in the New York study
region. The appropriate distance between gages is determined:
A = 0.5 x2
x
/2A
/2 x 47
9.7 - 10 miles
Figure 5-18 indicates the following approximate transformations for x = 10
miles: I'/I = 0.94, v.f/vi = 0.89, D'/D = 0.94, ^d'/vd = 1.01, V'/V = 0.94,
v '/v = 1.01, A'/A = 6.96, v5'/v~ = 1.00. For example, if the mean storm
volume is estimated as V = 0.57 in. at a point, the areal estimate is V =
0.94(0.37) = 0.35 in. The difference is negligible, particularly when
considering the overall uncertainties in any rainfall-runoff model. While
greater modifications may be expected during summer periods due to the
occurrence of more localized thunderstorms (rather than larger frontal
storms), the use of long-term event statistics to characterize the New York
area precipitation appears to reduce some of the need for point to areal
5-45
-------
rainfall transformations, except when very large drainage areas are consid-
ered.
The method presented for point to areal rainfall transformations is em-
pirically based. Curves such as those shown in Figure 5-18 will vary from
location to location, depending upon the rainfall correlation properties of
each study area. The advantage of the method is that it directly addresses
long-term areal rainfall properties, rather than individual events. The
raingage aggregation is useful not only for statistical analyses but when
developing input to simulation models which accept records from only one
gage, even though a large basin is being modeled.
5.2 Runoff Quantity
Runoff flows are generated when rainfall exceeds the storage and infil-
tration capacity of the drainage basin. Surface runoff begins as thin sheet-
flows which are consequently gathered into natural stream channels or man-
made conveyance systems. At the beginning of the storm event, little or no
runoff is generated as the initial precipitation is captured by the available
depression storage or intercepted by the vegetation throughout the basin.
This capture is often referred to as initial abstraction. As the storm con-
tinues and the available depression storage is filled, runoff occurs as the
difference between the rainfall rate and the infiltration rate. While
depression storage is available in both pervious and impervious areas, in-
filtration is confined to pervious land surfaces (although water may runoff
from impervious to pervious areas). The rate of runoff at various times
during the event is affected by time lags of contributing flow from different
portions of the catchment and the means by which the flow is routed through
the conveyance system. These processes are the basic mechanisms that deter-
mine modeling the quantity of stormwater runoff.
While a variety of detailed techniques are available for modeling the
journey from raindrop to stormwater runoff or overflow, the methods pre-
sented in this manual rely on simple transformations from rainfall to runoff.
The basic parameters required include the average ratio of runoff to rainfall
(R ) for use in the statistical method and broadscale drainage basin simula-
tors, and the average duration of runoff events (DR) for use in the statisti-
cal method. Methods for estimating these factors Based on drainage basin
characteristics are presented in this section. As discussed previously, this
simplified approach for evaluating runoff quantity is appropriate only for
the assessment of a long term series of rainfall events and their impacts,
and not for the accurate representation of individual storms.
5.2.1. Determination of Average Runoff to Rainfall Ratio
The average ratio of runoff to rainfall (R^) is used to convert rainfall
volumes to runoff volumes, as described in Equation 3-12. It is a composite
coefficient incorporating the effects of depression storage and infiltration
into a single ratio. It should not be confused with the runoff coefficient
traditionally employed in the Rational Method to calculate the peak rate of
runoff after the initial abstraction has occurred and all portions of the
drainage catchment are contributing (10). Rather it provides an estimate of
5- 46
-------
the fraction of rainfall which becomes runoff over a long term period. In
this sense R^ is a volume weighted average ratio, accounting for correlations
between storm depth and the runoff fraction of individual storms, when they
exist. (There is often a positive correlation between storm depth and the
runoff fraction, as larger storms are relatively less affected by depression
storage, and may also result in saturated soil conditions).
Drainage basin characteristics which tend to increase the average ratio
of runoff to rainfall include a high percent of impervious surfaces (streets,
sidewalks, rooftops, parking lots, etc.), tight, clay-type soils, and steep
land slopes. Impervious surfaces are much less effective at generating run-
off when they drain onto pervious areas (such as roof gutters onto lawns),
than when they are directly linked to the conveyance system. During wet
years or seasons, R^ is also higher because of the more saturated state of
the soils and the more frequent reduction of available depression storage
due to previous storms.
The primary factor used to predict R. in urban runoff studies is the
percent impervious area. Miller and Viessman investigate the effect of im-
perviousness on the rainfall-runoff relationship in four urban drainage
catchments (11). A linear relationship is found between the rainfall which
remains after the initial abstraction is subtracted, and the amount of
runoff:
R = 1.165 (I - 0.17) (P - I ) (5-1)
3.
where R is the runoff in inches, I is the fraction impervious area, P is the
rainfall in inches, and I is the initial abstraction in inches. The value
of I is estimated to be Between 0.10 and 0.15 inches, and the equation is
applicable to individual storms in areas where I ranges from 0.35 to 0.80.
To apply these results to the simpler representation where the initial ab-
straction is lumped together with other factors into a single ratio (R ), a
correction must be made for the effect of I on R^. Assuming the initial
abstraction is equivalent to a storage device (see Sections 3.6.1.2 and
3.6.2.4), the long term effect of an I = 0.10 inches may be estimated with
Figure 3-18 (reproduced as Figure 5-19J. If V = 0.40 inches and v =1.5
(typical values), a depression storage of about 0.10 inches reduces the
long-term runoff about 20 percent. Equation (5-1) is then modified to
estimate R as follows:
R^ = (1-0.20)(1.165)(I - 0.17)
(5-2)
= 0.93 (I - 0.17)
where I is the fraction impervious area. The equation is applicable for
0.35 <_ I <_ 0.80. This relationship is plotted in Figure 5-20 and referred
to as the extension of Miller and Viessman.
The STORM simulation model computes a runoff coefficient equal to 0.15 +
0.75 I, where I is the fraction impervious area. Again this value represents
the runoff to rainfall ratio after depression storage has been removed.
Heaney, et al found however, that the long term runoff is overpredicted by
5-47
-------
XL [~_£F
VD I
2O ? 5 3 O i5 4 O
EFFECTIVE STORAGE CAPACITY
4 5
MEAN RUNOFF VOLUME
50
FIGURE 5-19
DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE
5-48
-------
>
oc.
o"
I-
<
OL
I
o
1.0
.9
.8
.7
.6
.5
.4
.3
.2,
UJ
LOW DEPRESSION STORAGE, TIGHT SOILS,
STEEP LAND, WET YE:ARS, LITTLE
RUNOFF FROM IMPERVIOUS TO
PERVIOUS AREAS
HIGH DEPRESSION STORAGE,
LOOSE SOILS, FLAT LAND, DRY
YEARS, RUNOFF FROM IMPERVIOUS
TO PERVIOUS AREAS
I
I
I
I
I
I
10 20 30 40 50 60 70 80
PERCENT IMPERVIOUS AREA
90
100
STUDY LOCATION CODE:
(T) UPPER WHITE ROCK, DALLAS
@ LOWER WHITE ROCK,DALLAS
(3) BACHMAN BRANCH, DALLAS
(4) TURTLE CREEK, DALLAS
@ HENDRIX CREEK, NEW YORK CITY
(i) SPRING CREEK EAST, NEW YORK CITY @ MADISON, WISCONSIN
(?) THURSTON BASIN, NEW YORK CITY @ TULSA, OKLAHOMA
(g) FOURTH CREEK, KNOXVILLE @ DURHAM, NORTH CAROLINA
(5) THIRD CREEK, KNOXVILLE @ TROUT RUN, ROANOKE, VIRGINIA
LEGEND:
(lO) FIRST CREEK, KNOXVILLE
(H) PLANTATION HILLS, KNOXVILLE
@ NORTHAMPTON, ENGLAND
@ BRIGHOUSE,ENGLAND
(l4) BRADFORD, ENGLAND
"STORM" EQUATION ESTIMATE (la )
--- —— EXTENSION OF MILLER AND VIESSMAN ( II )
-SUSPECT ERROR, DIRECTION OF PROBABLE CORRECTION
FIGURE 5-20
RELATIONSHIP BETWEEN
IMPERVIOUS AREA AND RUNOFF-TO-RAINFALL RATIO
5-49
-------
only about 0.3 equivalent inches per year by ignoring the depression storage
correction in the STORM calculation (1). The assumption that this correction
is minimal is also made implicitly by Metcalf and Eddy in their simplified
stormwater simulation model (12). Rainfall is converted to runoff with a
runoff coefficient (K), which is equivalent to the R^ used in this manual,
and the ratio of runoff to rainfall is referenced by Metcalf and Eddy from
the STORM model as:
K = Ry = 0.15 + 0.75 I (5-3)
This relationship is shown in Figure 5-20, and identified as the STORM
equation estimate.
To supplement the Miller and Viessman study and the STORM equation,
data from 8 studies on 18 different catchments are analyzed. The results
of these studies are summarized in Table 5-4, and the average ratio of runoff
to rainfall (R..) is plotted for each as a function of the percent impervious
area in Figure 5-20. A number of the studies include information which in-
dicate possible errors, and these are noted in the comments. For example,
considerable transmission losses are observed in the Plantation Hills and
possibly the Fourth Creek stream channel in the Knoxville, Tennessee study.
While Figure 5-20 indicates the expected positive relationship between
the percent impervious area and R , a great deal of variability is observed,
and other factors are clearly important in determining the rainfall-runoff
relationship. In choosing an R^ for a particular study area., engineering
and planning judgement should be used to estimate the effect of the other
factors indicated in Figure 5-20. If no other information is available, the
STORM equation appears to provide a reasonably conservative estimate and may
be used until local information is collected. Local data and monitoring
are clearly needed to better refine the estimate of R^ , and the effects of
drainage basin characteristics on this estimate.
In order to use the percent impervious area to estimate R.., the percent-
age of impervious area in a drainage basin may be determined by examining
aerial photographs or detailed land use plans. This may be a long and
tedious task, however, particularly for large drainage areas. Satelite
photographs (LANDSAT) may be used to help evaluate the land cover of larger
basins (23). Estimates may also be made on the basis of the general fraction
of land in different land use categories. Specific land use classifications
are assigned an average percent impervious area as shown below. The indica-
ted values or other handbook estimates (24,25) may be modified or refined
on the basis of local information, past experience or field inspection
surveys.
Land Use Category Percent Impervious Area
Residential
Low Density 20
Medium Density 40
High Density 60
Commercial 80
5-50
-------
to
u:
rH
0
w
to
0
rH
p-*
^
2:
i a:
1 U.
LO
la o
1-2 rH
CO t-
s s
_]
*J
u.
z
I—I
s
e
u.
o:
c
-H
< o re
4-1
rt o
<£ p
in
3
P 0
C -H
0 >
0 r<
rl CD
(U El
c- E
o
M
cs re
•H rl
CO <
rl
Q
O
-rl
p
re
o
o
X
•o
j_>
to
m
CD
c-
o
in
o\°
LO
CM
in
•H
O
in
p
X
M
•n
p
X
rH
CD
P
M
rl
O
•a
o
i — .
CM
O>
00
rH
O
S? rH
in o
o o
rl OO
O -
rt co
*— ' i-H
in
n
o
n
O
o
a:
a>
•H
r"
S
fH
P-
3
i-H
t-1
tn
CD tn
OH 4.
o CL,
-H O
en •-•
1/1
o10
LT> o\°
rH LO
1 1
CM CM
in (fl
•H -H
O O
tn in
•P +-1
,C X
DO 00
•H -H
4-> 4-*
X X
0)
•P 4-*
rt rt
H (H
(U O
T3 T3
SI
0 O
h~- ["•••
(N rt
en en
O) OO
CM CN
o o
"3- O
r-l tO
0 0
o o
^- \o
V)
A
i-H in
rH rt
rt <~H
Q rH
* o
o -
O rC
a: o
M
O
^
0
0
LO
in
«
rH
re
rl
CJ
CD
P
rl
3
t— .
^
sents ratio of overflow (rather
off) to rainfall, though inter- 1 ,.
CD C
rl 3
Cl, rl
CD
rl C
C8
C^S
en in
o O
in in
CM \O
S 5
o o
o o
CM CM
Ol CO
-* to
rH
• U
• >-'
z •
n
O .^
Cj CJ
rl CD
U
X
•H &£
Vi C
•a -H
C rl
CD CL,
X to
LO VO
capture little. Suspected flow i •'-'••••-
rston from adjacent area.
3
w X
(H f-
O
P O
O, P
o c
CJ .H
E
o
in
to
3
o
o
o
CO
rH
U
>H'
2;
C
•H
in
C
O
3
JC
E-,
j^
d transmission losses in
e portions of stream channels riAi
CD rH
P X
CJ P3
CD CD
P. E
in EH
3 CD
W D,
ro to
i^> ^
i i
rH rH
(71 Cl
r^ CM
CM LO
o o
LO 00
LO LO
CN CM
LO O
rH
CD
rH CD
rH rH
> -H
X >
O X
c o
^
CD M
CD CD
rl CD
U rl
U
P "O
3 -H
O X
U, (-
oo en
>
H
X
X
• H
in
in
O
P.
"O
c
CO
in
rH
-H
I
C
O X
•H CD
P CD
re rl
P O
c
CO X
rH p
O. rl
3
P O
re u.
to to
1 1
rH rH
r-~ r>-
d C7)
O *£>
O 0
O O
*£) tO
rH CM
o to
CM LO
tO rH
CD
rH
rH
•H
X -
o in
C rH
^ rH
•H
- I
o c
CD O CD
rl -H *-l
U P rH
CO -H
in c x
ri re o
•rl rH C
U, 0. ^.
O rH
rH rH
CD
O.
O
rl
CD
3
CD
in
c
re
•H
E
to
o
o
rH
VO
1
0
vO
00
to
o
o
LO
Ol
CN
C
re
on
K
c
0
P
P.
E
re
rl
O
Z
CN
0> HI
0. 0.
O O
rH rH
in tn
rl rl
CD CJ
3 3
D O
tn (/)
c c
rt rt
•H -H
i= £
«3- rH
O O
o o
rH -^f
sD O
I I
CO t-t
Oi O^
00 rH
O CM
0 0
rH CO
t-t CM
^f r^>
CT> \O
Ln T— i
C TJ
rt C
rH (0
00 rH
C W)
03 C
UJ
in T3
3 r«
0 £
0013
•H rt
fn FH
CO CQ
ro ^f
CO
tn
E
0
in
'J
00
o
r-
CM
to
CM
cr
•H
in
C
0
u
•H
2
.,
C
0
in
*H
T3
CD
E
LO
fs drain to pervious areas (19)
0
o
rl
P
n
O
•z.
tf)
g
0
in
CM
1-1
o>
I-H
O
^.
rf
r-
Ol
re
e
o
CO
•J
0
*,
tn
rH
3
H
\O
rH
possibly underestimated by ,_„,
rt CM
M-t ^-^
C
-H eX*
rt o
a: -^t
tn
g
0
tn
ro
(N
L/>
O
^
?— <
o
f^
o
i-H
rt
C
•H
i-H
O
r<
U
rH
O
z.
E
rC
r^
3
Q
^
rH
fN
CM
VI
E
O
tf)
"*
*"*
o
1— 1
o
o
to
f^
01
CTl
rt
^
o
c
a!
o
c£
C
3
C£
3
o
rl
H
00
5-51
-------
Industrial 70
Institutional, Public 30
Open, Undeveloped 0
The percent impervious area for the entire drainage area is calculated by
taking a weighted average of the individual components of the area. For
example, assume an area has the following land use characteristics:
Low Density Residential 30%
Medium Density Residential 20%
Commercial 10%
Industrial 10%
Institutional, Public 5%
Open, Undeveloped 25%
Total 100%
The overall percent impervious area is:
Percent Impervious Area = (0.30)(20) + (0.20)(40)
+ (0.10)(80) + (0.10)(70)
+ (0.05)(30) + (0.25)(0)
= 30.5%
This value is used in Figure 5-20 to determine the average ratio of runoff
to rainfall, R...
If information on the land use categories cannot be obtained, the
portion of developed areas which is impervious may be estimated from the
population density. Graham et al. (Washington, DC) (26), the American
Public Works Association (27) and Stankowski (New Jersey) (24) have developed
equations to predict imperviousness as a function of population density. The
imperviousness is estimated for the developed portion of the urbanized area
only. The weighted average imperviousness and population density have been
calculated for nine Ontario cities (28). These results are plotted on Figure
5-21 together with the three estimating curves (29). If the New Jersey data,
which is based on 567 municipalities, is selected as a reasonable guideline,
the equation used to estimate imperviousness is:
n + T - * n ^ nn (0.573-0.0391 log, JPD,) fc ..
Percent Impervious Area =9.6 PD, ^ 610 d (5-4)
where: PD, = population density in developed portion of the
urbanized area (persons/acre).
This provides a convenient estimate for an initial estimate of the runoff
coefficient.
5.2.2 Determination of Average Duration of Runoff Event
In large, unurbanized drainage basins the runoff flow may continue well
beyond the end of the rainfall event. If the average storm runoff flow is
calculated directly from the average intensity and the rainfall to runoff
5- 52
-------
100
90
80
70
i .
1
I 50
.
I
(J! 40
30
20
10
10 20
PERSONS/HECTARE
30 40 50
60 TO 80
I I
NEW JERSEY (24 )
ONTARIO (28)
LEGCND
A WASHINGTON, D.C.
D ONTARIO
10 15 20 25 30 35
DEVELOPED POPULATION DENSITY, PDd, persons/acre
KCTBtCHCC (23)
FIGURE 5-21
IMPERV10USNESS AS A FUNCTION OF
DEVELOPED POPULATION DENSITY
5-53
-------
ratio (Q = R..IA) , it is assumed that the duration of runoff events t is
essentially equal to the duration of rainfall events. When the runoff dura-
tion is significantly modified by drainage basin characteristics, however,
the average runoff flow can be considerably overestimated. To account for
this, the correction factor (Equation 13 of Chapter 3) is suggested:
QR = RylA (D/DR) CS-5)
where D/D is the ratio of the average rainfall event duration to the
average runoff event duration. This section provides a method for estimating
the average runoff event duration (D ) based on the hydrologic properties
of the drainage basin.
The approach for estimating D is based on the analysis of unit hydro-
graphs. The basic unit hydrograph and its shape parameters are shown in
Figure 5-22. A profile of discharge versus time at the outlet of the catch-
ment or basin is determined from a unit volume of precipitation excess over
some time interval (t ), which should be no larger than one fourth (10) to
one half (30) the lag time to peak (t ) . To estimate the average duration
of runoff events, two steps are suggested: (1) the average duration of
precipitation excess (D ) is calculated; (2) the width of the hydrograph is
determined. e
The early portions of the storm generate negligible runoff until the
initial abstraction is removed. The duration of precipitation excess is
thus shorter than the rainfall duration. The average duration of precipita-
tion excess (D ) is approximated by assuming that the initial abstraction
occurs as storage, as shown in Figure 3-14 (d), The duration of precipitation
excess is equal to the widths of the unshaded portions of Figure 3-14(d) .
The average duration of precipitation excess is equal to the expected value
of the unshaded durations :
CO CO
D = / / (d - V /i) p (d) p (i) dd di (5-6)
6 i=0 d=V./i d d i
d
where V, is the storage due to the initial abstraction. Equation (5-6) is
solved for the special case where storm durations and intensities are in-
dependent and exponentially distributed and, dividing by the mean rainfall
duration (D) , the solution is :
where K is a modified Bessel function, D and I are the average intensity
and duration. Equation (5-7) is plotted in Figure 5-23. The initial ab-
straction (V.) is estimated by Miller and Viessman to be generally between
0.1 and 0.15 inches in urbanized catchments. V, may also be estimated as
equal to the depression storage (though all the depression storage may not
be available at the beginning of every storm) . The STORM model estimates
the depression storage for pervious and impervious areas as follows:
5-54
-------
CENTROID OF PRECIPITATION EXCESS
PRECIPITATION EXCESS
DURATION OF PRECIPITATION EXCESS
PERIOD OF RISE
SIGNIFICANT PERIOD OF RISE
O.IOq
TIME
FROM BRATER AND SHERRILL (30)
FIGURE 5-22
UNIT HYDROGRAPH DEFINITION SKETCH
5-55
-------
0 0.2 04 06 0.8 1.0 12 14 16 1.8 2.0 2.2 2.4
INITIAL ABSTRACTION
DI
MEAN STORM DEPTH
FIGURE 5-23
ESTIMATE OF MEAN DURATION OF RAINFALL EXCESS
5-56
-------
Land Use Depression Storage (in)
Impervious 0.0625
Pervious 0.25
For a given land use, the area weighted depression storage, (DS), in inches,
is:
DS = 0.25 - 0.1875 (Percent Impervious Area/100) (5-8)
Using these guidelines, the estimate of V, is generally in the range found
by Miller and Viessman. For an area with a mean storm depth of 0.3 inches
and an estimated V, of 0.1 inches, Figure 5-23 indicates that D /D is about
0.54. d 6
Once the mean duration of precipitation excess (D ) is estimated, the
period of runoff attenuation must be added. Brater and Sherrill (30) present
relationships between unit hydrograph parameters and drainage area character-
istics :
The unit hydrograph peaks as well as their time characteristics such as
their periods of rise and widths at various fractions of the peak dis-
charge can be correlated with watershed areas and population density to
provide statistically significant relations which enable hydrologists
to estimate the runoff characteristics of ungaged areas. These rela-
tionships were derived from the analysis of hundreds of flood hydro-
graphs from 53 drainage basins from five states. The areas of these
basins vary from 0.02 to 743 sq. mi. and the population densities cover
a range from less than 100 to more than 14,000 persons/sq. mi. (30,
p.l.).
Note that population density is used as a general indicator of the percent
impervious and channel condition effects. Modifications for intensive
commercial or industrial areas can be made by using a larger-than-actual
population density to reflect the level of urbanization in the basin.
Assuming that the time period (t ) used to generate each unit hydrograph
is small (each of the resulting hydrographs are superimposed to generate the
total runoff), the runoff duration may be estimated as the duration of the
rainfall excess plus the time necessary for the runoff generated by the final
rainfall excess to reach some arbitrarily small value. Using the time of
the base (W ) assumes that the runoff flow must return to zero for the
runoff event to be considered complete. This will yield a very large esti-
mate of D , which will lead to a very small estimate of Q . A more reason-
able estimate of the period of effective runoff, W^,., is suggested since a
large majority of the runoff volume has been accounted for by then, and D
will not be biased by the long tail of the hydrograph. These considerations
lead to an estimate of the runoff event duration as:
DR = De + W25 (5-9)
5-57
-------
Based on the work of Brater and Sherrill, estimates of W9 may be made
from Figure 5-24, using the drainage area (sq. miles) and the population
density (people/sq. mile). As indicated in Figure 5-24, W 5 increases with
increasing drainage area and decreasing population density. Note that in
small catchments the reduction in the period of runoff due to the initial
abstraction may be larger than the attenuation reflected in W , and D/D is
estimated to be greater than one.
As an example consider a basin which has the following characteristics:
A = 20,000 acres = 31.25 sq. miles
Ry = 0.50
PD = 8000 people/sq. mile
V, = 0.10 inches
d
I = 0.06 in/hr
D = 6 hr.
The uncorrected mean runoff is calculated as :
0.50 (0.06) 20,000
600 cfs
Using the correction, however:
Vd/DI= 0.10/(6 x 0.06)
0.28
D /D = 0.58 (from Figure 5-23)
G
D = 0.58 (6)
3.5 hr
W = 5 hr (from Figure 5-24)
DR - De+W25
3.5 + 5
8.5 hr
QR = VA (D/V
5-58
-------
1
_l
_l
s
i
u.
O)
O
CD
(O
UJ
O
I
111
£C
(O
^ 3
CM w
I «
10 ^
UJ ^
K <
O UJ
e> o.
UJ
u
o:
UJ
Q.
tf)
CM
Z
O
SUH'(SZM) >«V3d JO %SZ iV H1QIM HdV«90dOAH UNO
5-59
-------
600 (6/8.5)
425 cfs
This estimate is presumably more representative of the mean runoff flow and
more appropriate for initial load estimates and treatment device screenings.
Of course if local runoff monitoring data are available, they are preferable
to the estimating technique outlined in this section.
5.3 Runoff Quality
A variety of factors influence the quality of stormwater runoff and
overflows. Precipitation falling through the atmosphere washes out both
natural and man-made contaminants. While the contaminant concentration of
rainfall may be important for specific pollutants in some situations (31,32,
33,34), the most significant portion of stormwater loadings are generally
collected from the land surface .and the conveyance system. The amount of
pollutant washoff from the land surface is affected by the land use, vegeta-
tion cover, population density, activities which generate pollutants such as
automobile traffic, litter, animal wastes, construction, transportation of
materials, and fertilization. Pollutants discharged will also be influenced
by activities which reduce pollutants such as street cleaning, road mainten-
ance, erosion control, and general upkeep. The conveyance system may con-
tribute a significant mass, particularly in combined sewer systems where
the contaminants in the sanitary sewage mix with the runoff. The type of
conveyance system '(natural, separate, or combined) and its characteristics
(slope, state of repair, tendency for deposition or stratification, channel
stability, etc.) are important factors in influencing the quality of wet
weather flows. Finally, storm properties may also affect the pollutant
concentration of stormwater flows. Long intervals between storms allow
greater dirt and dust accumulation on the land surface and larger depositions
in the sewerage system. High intensity storms may dislodge and transport
large quantities of solids and cause significant scour in the sewerage system
while high volume storms may demonstrate a dilution effect if the source of
contaminants is limited. Combinations of effects may occur due to variations
in storm properties.
With the myriad of complex factors which influence the quality of storm-
water runoff and overflows, it is very difficult to establish useful deter-
ministic relationships for predicting appropriate pollutant concentrations.
Furthermore, the great variability in observed concentrations, both within
and between storms, and from one location to another, make any estimate
highly uncertain. This section attempts to provide some guidance for
evaluating stormwater pollutant concentrations in a study area. Reported
values for average concentrations, its variability, and attempts to correlate
runoff and overflow quality with land use are discussed.
5.3.1 Determination of Average Pollutant Concentrations
The most recent reviews of stormwater quality surveys throughout the
country are presented in the Urban Stormwater Management and Technology
reports prepared by Metcalf and Eddy, Inc. for the U.S. Environmental
5-60
-------
Protection Agency (35,36). The results of the 1973 assessment (35) are
shown in Table 5-5 and 5-6 for combined sewer overflows and separate storm-
water runoff respectively. The results from the 1977 update (36) are shown
in Table 5-7 and 5-8, for combined and separate runoff respectively. The
studies are from diverse locations with different land uses and rainfall
patterns. The results from both studies are summarized in Figure 5-25. For
some of the pollutants, only a few study locations are represented. Figure
5-25 indicates the wide range of possible concentrations which may be en-
countered. Typical or conservative values for the average concentration, c,
may be chosen, but local monitoring and data are clearly needed to better
refine estimates of stormwater quality in a particular area. The Urban
Rainfall-Runoff Data Base compiled for the U.S. Environmental Protection
Agency by the University of Florida (37) may be referred to for a detailed
source of stormwater quality data from a number of studies.
5.3.1.1 Variability of Pollutant Concentrations
Figure 5-25 demonstrates the wide variation in stormwater quality from
one location to another. Pollutant concentrations also vary considerably
from storm to storm and within storms at a given site. To illustrate the
typical level of variability between storms, data from five sites are
analyzed for chemical oxygen demand (COD) and suspended solids (SS) concen-
trations. The average concentration for each storm is calculated and these
are used to determine the overall average concentration, the standard
deviation (reflecting between storm variability), and the coefficient of
variation between storms (v ) at each site. The results are shown in Table
5-9. The coefficient of variation of stormwater quality (v ) is seen to
range from about 0.50 to 1.00. For measurements of bacteria organisms,
somewhat higher between storm variability is expected, although a portion of
this may be due to the sampling error.
Runoff and overflow concentrations also vary markedly within a storm.
The primary descriptive tool used to represent this variation is the first
flush profile, although this describes only a portion of the within storm
variability. To demonstrate the general applicability of the first flush
profile, the results of the quality data normalization from the 1977 update
of the Urban Stormwater Management and Technology report are reproduced in
Figure 5-26. On the average, overflow and runoff concentrations of BOD are
about three times higher in the first half hour of storms as compared to the
later portions of events (c /c = 3) and the ratio for SS is about two to
one (c /c = 2) . "
5.3.2 Effect of Land Use On Stormwater Quality
A number of conflicting claims have been made concerning the relation-
ship between the land use of a drainage area and the concentration of pol-
lutants in its stormwater runoff and overflows. Some maintain that strong
correlations exist, while others conclude that little information about
runoff quality may be inferred from land use. While most seem to feel that
different land uses yield different runoff quality, the primary problems
appear to be:
5-61
-------
g
1,000
e
*— ' o
LU
t~ 4
2
^ 2
t-
a- 100
LU g
0
X 6
LU
4
cn
c ->
O i n
1- 8
a: 6
^ 4
uj
o
Zo
O
O
1 0
.8
.6
.4
.2
0
L EG END
I I «
kA/sJ c
NOTE: ^^
REPRODUCED FRO
;
A
<
X
X
X
X
X
_
»v
V
"*
' — '
—
-
:•:
— •:•
ji
3
- y>
f
±ld
)? .
^
§
X
r—
_
r:
¥:
-
—
\
\
1 ^
\/
— j>
V
^
g
r ^
TYPICIAL MUN
SEWAGE AFTEf
SECONDARY TR
L
n
^
Sl
:M
s
ICIPAL
EATMENT
gj
n 1
A
X I 5?
x "V v.
i % £
—
rr-
,».•
•,•.
;I;1
X;
jx
X
^
\
x
xxvxxxxxxxx>
x^
x
b
i b3 ^
TSS BOD COD TOTAL TOTAL LEAD FECAL
NITROGEN PHOSPHORUS COL FORMS
IOOO ORGANISMS/ IOO mL
PARAMETER
>TORMWATER RUNOFF FROM DECEMBER, 1973 ASSESSMENT
5TORMWATER RUNOFF FROM THIS UPDATE
:OMBINED SEWER OVERFLOWS FROM DECEMBER, 1973 ASSESSMENT
:OMBINED SEWER OVERFLOWS FROM THIS UPDATE
M REF. ( 36)
FIGURE 5-25
REPRESENTATIVE STORMWATER DISCHARGE QUALITY C2J
5-62
-------
I 0
rn!
LEGEND
ss coneiN EO Sf.it»
OVERFLOWS (CSO)
BOO CSO
SS STO«M»«TE R BUMOf F ( S« )
100 S «
TIME HFTER START OF RUNOFF OR OVERFLOW, h
NOtE
REPRODUCED FROM FIGURE 4, REF. 36
FIGURE 5-26
TIME WEIGHTED NORMALIZATION
5-63
-------
location, year.
Ref. No.
municipal
•urucipal
Secondary effluent
Selected combined
Atlanta, Ca ,
1969
Berkeley. Calif C
1968-69
Brooklyn, N Y..
1972
196S-69
Cine innatt , Ohio,
1970
Des Homes , Iowa,
1968-69
Detroit, Mich. ,
196S
Kenosha, Wis .
19?0
MiUaukce, his. ,
1969
Northampton, U.K. ,
1960-62
Rac me , his. ,
Roanoke. V*. ,
1969
Sacramento, Calif. ,
1966-69
1969-70
Kashington, D.C. ,
1969
a. Data presented here
procedures were used
b. Only orthophosphate.
c. Infiltrated sanitary
Total lot.il Jutul
mg/I mg/1 mg/1 mg/1 MPN/100 «1 ">fi/l ..s N my/I i- 1'
ZOO 100-300 &OQ 2 SO -7 SO -• 200 100-350 S'}07 1«107-1«109 40 30
6 8
25 15-45 55 25-80 -- 15 10-30 1'JC3 1- 10' • 1-10* to i.C
100 48-540 -- -- fl.5 -- -- 1"107 -- -- 1.2b
60 18-300 ZOO 20-600 • • 100 40-150
180 86-428 -- -- -- 7,051 132-8,759 -• •- -- J.«b
:20 11-560 400 13-920 - - 4?0 20-2.440 1*1 0? 2"10S-5«107 J3 S.S
200 80-580 250 190-410 -- / , 100 500- 1 ,800
IIS 29-158 -• -- -- 285 155 -1,166 -- -- 12. 7 1 1.6
1S3 74-68S 775 -• -- 2?4 120-804 -• -• 7«. JJ 4.9
1Z9 -- 464 - - •• 4i>8 -• 2*70* -- 10.4 5.*
56 26-182 777 118-765 -- S44 113-848 -• 2«iOS-3-107 3-Z4 0.8*
ISO 80-350-- -- •- 400 200-800 -- -- 10*
ns .. .. .. .. 79 .. 7*107
. < 7f
MS 70-328 239 59-513 -- 725 56-502 5" 10 7-10 -9"10
49 l.S-202 I6i 17-616 -- 88 4-42b J-IO* 2»10*-2»107
71 10-470 382 80 -1,760 -- 622 35 -2, 000 3*10* 4- 10S -6«106 3.6 1.0
are for general comparisons only- Since different soup ling Methods, number of samples, and other
. the reader should consult the references before using the data for specific planning purposes.
sewer overf low.
e. Only ammonia.
f. Only fecal
NOTE:
REPRODUCED FROM TABLE II, REF.35, 1973 ASSESSMENT.
TABLE 5-5
COMPARISON OF QUALITY OF COMBINED
SEWAGE FOR VARIOUS CITIES0
5-64
-------
Total Total Total
P°ns. COD, HO. SS. conforms. nitrogen, phosphorus,
Type of wastPfcJter, m^/I n^/l mg/1 mg/l MPN/100 ml mg/1 35 S rag/1 as P
local ion, year,
Ref No. -,'j; Range Avj Range Avg Aug Range ,4u0 Range
Typical untreated
municipal "CC 100-300 .V0 2 SO-750 -- :oo 100 350 £-70 1-10 -1-10
Tvp ical t reared
municipal
Primary effluent US 70-200 3fO 105-500 -- 81 40-130 2*10* 5'10*-S-108
Secondary effluent s$ 15-45 SS 25-80 -- If. 10-30
Storm se*er
discharges
Ann Arbor, Mich ,
1965 se 11-62 - -- -- 2,OS0 650-11,900
Castro \al lev,
Calif ,19"1--Z 14 4 57 -- 8.4
1969 ' Jf i;-100 -- -- -- SOS 95 1,053
Durham. N C ,
1968 j: 2 - 232 IK 40-660
Los \nKeles. .
(alif .lifcT-ftS J 4 -- -- f * *,-;. -- -- 3«103*2«10°
Madison, his.,
15-0-71 -- -- •- " '- SI 10-1 ,000
Sew Orleans. La , , .
19tT-69J 12 '- -- -- 4 ' "e -- I'lO 7-10-7-10
Roanoke , \a ,
1969 ? -- -- -- SO
Sac ra"iento .Calif, .
1^68-69 ICB 24-281 !S 21-176 -- 71 3-211 8*10* 2-10 -1-10'
TuJba, Okla ,
1968-69 11 1-39 Si 12-405
Hashington, D.C.,
1969 2$ 3-90 33i 29-1 ,514
a Data presented here are for £cncral comparisons only Since different sampltng methods , number of samples, and
other procedures were used, the reader shouId consult the references before us ing the data for specific planning
purposes
b Onlv ammonia plus nitrate
(. Only fc^al
e Only organic (KjeldahlJ nitrogen.
f. Only soluble orthophosphate
NOTE:
REPRODUCED FROM TABLE IZ, REF 35, 1973 ASSESSMENT.
TABLE 5-6
COMPARISON OF QUALITY OF STORM SEWER
DISCHARGES FOR VARIOUS CITIES0
5-65
-------
Average pollutant concentration,
rjpldahl Total Fecal
TSS VSS BOO COD nitro<3°n nitrogen PQ«-P OPGa-P Lead cotlformsa
Des Opines,
Iowa 413 117 64 . . 4.3 1 B6 1 31
MiIwaukee,
Wisconsin 321 109 59 264 49 63 1 23 0 £6
New York City,
New York
fiewtown Creek 306 182 222 481 ... .... ... 0 60
Spring Creek 347 ... Ill 358 . 16.6 i &
Poissy, France0 751 387 279 1005 .. 43 17°
Racine,
Wisconsin 55] 154 158 ... ... 2.78 092 ... 201
Rochester .
New York 273 . 65 ... 2.6 ... 0.88 014 1140
Average (not
weighted) 370 140 115 367 3.8 9.1 1 95 1 00 0 37 670
Range 273-551 109-182 59-222 264-481 2.6-49 43-16.6 1.23-278 086-1 31 0.14-0.60 201-1140
a. 1000 organisms/100 nl.
b. Total P (not included in average).
c. Not included in average because of high strength of municipal sewage when compared to the United States.
NOTE:
REPRODUCED FROM TABLE 33, REF. 36 , 1977 ASSESSMENT.
TABLE 5-7
POLLUTANT CONCENTRATIONS IN COMBINED SEWER OVERFLOWS
5-66
-------
Avera^i" jolHtant concentrations, ng/L
Kjelrfahl Total Phos- npn _ Fecal
City TSS VSS BOD COD ni trosen nitrogen phorus "™,- LM() coli'ontu*
287 9 48 0 57 0.62 0 33 .... 0 15 6 300
419 104 56 2 09 3 19 0 56 0 15
C,rhjn. 'o'th Carolina 1223 122 .. 170 096 ... 082 .... 046 230
K-o
-------
TABLE 5^9
VARIABILITY OF RUNOFF AND OVERFLOW CONCENTRATIONS
(Between Storms]
COD SS
Conveyance Number
Location Type of Storms Mean Std. Dev. vc Mean Std. Dev. vc
Cmg/1) (mg/1) (mg/1) (mg/1)
Durham, separate 26 182 89 0.49 1,198 902 0.75 (20)
' Separate 11 365 283 0.78 1,056 630 0.60 (38)
Combined 13 339 184 0.54 338 225 0.67 (39)
j '(!} Combined 11 395 338 0.86 251 275 1.10 (40)
Combined 11 269 205 0.76 137 63 0.46 (40)
City Dock, Newark
Ivy Street, Kearny
5-68
-------
1. Isolating the effects of land use from other sources of
variability in the runoff concentration,
2. Establishing a statistically significant relationship.
This section presents a review of some of the work that has appeared in the
literature relative to these problems.
Study of Street Surface Contaminants:
Sartor and Boyd present the results of an extensive runoff surveying
program in "Water Pollution Aspects of Street Surface Contaminants" (41).
Devices were used to simulate rainfall and samples were collected and ana-
lyzed at a large number of test sites. Their results indicate differences
between land use types:
Contaminant loading intensities were found to vary with respect to
land-use patterns in the surrounding locale. In general, industrial
areas have substantially heavier than average loadings. All industrial
test sites (20 of them) taken together have an average loading of some
2800 Ib/curb mile; twice the mean for cities on the whole. This is
probably because industrial areas tend to be swept less often and be-
cause generation rates of dust and dirt tend to be high (e.g., "fall-
out", spillage from vehicles, unpaved dirt areas, streets in poor
conditions, etc.). Of these, heavy industrial areas showed the heaviest
loadings, medium industrial the lightest. The loadings varied so
widely between individual sites that it would be speculative to state
why one type of industrial area is dirtier than another.
Commercial areas have substantially lighter loading intensities than
the mean for cities on the whole (290 Ib/curb mile average vs. 1400).
This is probably because they are swept so often; typically several
times weekly, daily in prime areas.
Residential areas were found to have an average loading intensity com-
parable to the average for all land uses of all cities taken together:
1200 Ib/curb mile. Here again, the loadings varied widely from site to
site, and it would be speculative to state why one city is more heavily
loaded than another or why one type of residential neighborhood is
cleaner than another. The data implies, however, that there is some
tendency for newer, more affluent neighborhoods to be cleaner; possibly
because they are better maintained by residents and/or are further from
sources of contamination (41, p. 6).
Since the effects of street sweeping frequency are hypothesized to be a
factor in the differences in loading rates between the land use types an
attempt was made to separate the land use and street sweeping effects by
calculating the accumulation rate since the last rain or sweeping at each
site. The results summarized in Figure 5-27, show there are still some
differences due to land use. The writers conclude that, "In general, indus-
trial land-use areas tend to accumulate contaminants faster than commercial
5-69
-------
o
z
Q
O ~
_l UJ
_l
to —
Q 2
3 en
o
o
Q >*
UJ O
1200 -
1000 -
800 -
600 -
<
-J
Z)
o
o
— 400 -
200
I
I
I
I
LEGEND:
• INDUSTRIAL
• RESIDENTIAL
ACOMMERCI AL
I I I
345 6789
ELAPSED TIME SINCE LAST CLEANING
BY SWEEPING OR RAIN (DAYS)
12
REPRODUCED FROM REF (41)
FIGURE 5-27
TIME SINCE LAST CLEANING VERSUS SOLIDS LOADING
5-70
-------
or residential areas." (41, p. 4). However, the large scatter of the data
upon which the curves in Figure 5-27 are based weakens the conclusion.
URS Research Company Study:
In studies stemming from the previous report, the URS Research Company
analyzed storm runoff data from fifteen cities in the United States in their
report, "Water Quality Management Planning for Urban Runoff" (42). The data
were gathered from published reports and analyzed to investigate the effects
of climate (as determined by the geographic sector in which the city is
located), land use, type of street surface, traffic density, and the type of
surrounding landscape.
The writers analyzed the data by grouping the records according to the
independent parameters (climate, land use, etc.), calculating the mean and
other basic statistics of the dependent variables (runoff quality parameters)
and testing the differences between these means for statistical significance
using the Student's t test. The results are summarized in Table A-24 of
Appendix A of the URS report. Loading rates are given in terms of the Ibs/
curb mile/day of total solids. Loading rates for particular pollutants are
then determined from the ratio of pollutant mass to mass of dry solid.
Examples of the writers' conclusions concerning the relationship between land
use and the solids loading rate include
... (1) commercial and residential area means are different at the 97
percent confidence level, (2) commercial and light industrial area means
are different at only the 73 percent confidence level, (3) commercial
and heavy industrial area means are different at the 95 percent confi-
dence level, and (4) the commercial mean differs from the mean of the
set of all data at the 98 percent confidence level (42, P. A-32).
As found by Sartor and Boyd, industrial land uses have the highest solids
loading rate, followed by residential, commercial, and open space, in that
order.
Although the URS report does find that the grouping of the data indica-
tes differences in the street surface runoff from different land uses, the
writers temper their conclusions about the nature of this relationship in
their summary:
The URS staff has assembled all presently available data on the rates
of accumulation of solids and the concentrations of various constituents
in those solids that collect on street surfaces. The range and scatter
in the available data are extreme. Both the sampling variability and
the complexity of natural systems contributed to this extreme variance.
Attempts to test conceptual models with the available data have met with
absolutely no success because the models are much too simple to ade-
quately describe the conditions in the real world (42, p. A-43).
5-71
-------
University of Florida Study:
A study team from the University of Florida presents a general procedure
for predicting runoff quality from land use and population density in "Storm
Water Management Model, Level I, Preliminary Screening Procedures" (29).
The writers comment on the variability of the data encountered in formulating
their procedure and present the methodology as follows:
It is unfortunate that perhaps the only consistent remark about urban
runoff quality analysis in general is that data and results of previous
studies are so remarkably inconsistent. Few studies have been made of
characteristics of street liter, and they offer a wide range of values
of concentrations and loads. Effluent data show a similar scatter.
However, it is necessary that a decision be made regarding actual values
for use in the analysis. Table 8 (shown as Table 5-10) presents a pre-
dictive equation developed after a review of available stormwater
pollutant loading and effluent concentration data (1). The equation
permits one to estimate BOD-, SS, VS, PO. and N loads as a function of
land use, type of sewer system, precipitation, population density, and
street sweeping frequency. Loading in combined sewer areas are assumed
to be 4.12 times as large as loadings in separate areas. They are
assumed to vary as a function of developed population density as shown
in Figure 4 (shown as Figure 5-28). The intercept (0.142) was deter-
mined based on data for open space. The exponent (0.54) is based on
the exponent of the imperviousness equation at a population density of
8 persons per acre (20 persons per ha) such that pollutant concentration
increases as a function of population density. Lastly, the coefficient
(0.218) is based on an average of data points with a PD, ranging from
5 to 15 persons per acre (12 to 37 persons per ha) to yield a value of
f (PD.) of 0.895 at 10 persons per acre (25 persons per ha). (Note
tne large spread in the observed data). The street sweeping relation-
ship was derived by making numerous runs of STORM with varying street
sweeping frequencies. The results are shown in Figure 5 (shown as
Figure 5-29) (29, p. 16).
The results and relationships presented in the University of Florida
study are directed towards the effect of land use on runoff quality. Due to
the wide variability in observed data, however, the results are only appli-
cable for preliminary estimates when local data are not available.
Atlanta Study:
In stormwater studies conducted in Atlanta, "Storm and Combined Sewer
Pollution Sources and Abatement, Atlanta, Georgia" (43), attempts are made
to correlate both combined and separate sewer runoff to local land use and
population indicators. Initially, six land use classifications are employed
with a linear regression analysis to predict the quality from six drainage
areas. To achieve significant results, the dependent variables are simpli-
fied to include only population density and the percent undeveloped land.
Their results are summarized as follows:
5-72
-------
20
PERSONS/HECTARE
60 80 IOO
120
140
2.0 -
<
a:
0 10 20 30 40 50 60
DEVELOPED POPULATION DENSITY, PDd (PERSONS/ACRE)
FROM REF (29)
WASHINGTON, D.C.
DES MOINES
MILWAUKEE
TULSA
ROANOKE
CINNCINNATI
SEATTLE
WINDSOR
CALIBRATION POINT
SEPARATE
(5)
(R)
COMBINED
FIGURE 5-28
NORMALIZED BOD LOADINGS
VERSUS DEVELOPED POPULATION DENSITY
5-73
-------
sanos a3QN3dsns
o
o
ID
O
o
in
O
O
1-
o
o
O
O
CO
O
O
in
a
8
I
3
-------
The following equations may be used to predict annual average
loading rates as a function of land use, precipitation and population
density.
separate Areas:
Combined Areas:
where
Land Uses: 1 •
i -
i -
MS - au.j; • r - I2vrv . , ^tf.yr
M • 6(1, j) • P • *2^"d^ ' Y acre-yr
M - pounds of pollutant j generated per acre of
land use 1 per year,
F - annual precipitation. Inches per year,
PD - developed population density, persons per acre
a,B - factors given in table below,
r * street sweeping effectiveness factor, and
f-(PD.) - population density function.
i a
1 Residential
2 Commercial
3 Industrial
4 Other Developed, e.g., parks, cemeteries, schools
(assume PD • 0)
Pollutants:
Population
j - 1 BOD , Total
j - 2 Suspended Solids (SS)
J - 3 Volatile Solids, Total (VS)
j - 4 Total PO, (as PO )
j - 5 Total N
Function: i - 1 £-|(pIV " 0.142
i - 2,3 f^PD.) - 1.0
i - 4 ff(PDj) - 0.142
+ 0.218
PD
0.54
Factors
and 6 for Equations: Separate factors, a, and combined factors,
B, have units Ib/acre-in. To convert to kg/ha-cm, multiply
by 0.442.
Pollutant. J
Land Use, 1 1. BOD 2. SS 3. VS 4.
5. N
Separate
Areas, a
Combined
Areas, 6
1.
2.
3.
4.
1.
2.
3.
4.
Residential
Commerc ial
Industrial
Other
Residential
Commercial
Industrial
Other
0.
3.
1.
0,
3,
13,
5,
0
.799
20
,21
.113
.29
.2
.00
.467
16
22
29
2
67
91
120
11
.3
.2
.1
.70
.2
.8
.0
.1
9.
14.
14.
2.
38.
57.
59.
10.
45
0
3
6
9
9
2
8
0
0
0
0
0
0
0
0
.0336
.0757
.0705
.00994
.139
.312
.291
.0411
0.131
0.296
0.277
0.0605
0.540
1.22
1.14
0.250
Street Sleeping: Factor y is a function of street sweeping Interval,
N . (days):
f.; /20 if 0 <^ Ng <_ 20
( 1.0 If N > 20 days
days
NOTE:
REPRODUCED FROM TABLE 8, REF. (29 )
TABLE 5-10
POLLUTANT LOADING FACTORS FOR DESKTOP ASSESSMENT
5-75
-------
For combined sewer areas, land use may not be the appropriate primary
variable upon which to base pollution load estimates. For all but
initial scouring flows and high flow rates, pollution characteristics
depend primarily upon dilution of domestic and industrial waste entering
the system. The strength of this waste determines the BOD concentration
in the overflow. The pertinent parameter for correlation would there-
fore be the population equivalent of the domestic and industrial waste,
which would entail a detailed analysis of industrial waste discharges
within each combined area. Storm sewer areas and combined sewer areas
at high flow rates yield BOD concentrations that may well depend upon
land use more than upon population; however, further data will be re-
quired to determine the nature of the relationship (43, p. 110).
The differences in approach that is necessary for combined and separate
sewer areas are demonstrated by these findings.
Lubbock Study:
In a study in Lubbock, Texas, "Variation of Urban Runoff Quality and
Quantity with Duration and Intensity of Storms -- Phase III" (44), research-
ers from Texas Tech University examined the effect of storm characteristics
and antecedent conditions on runoff quality in a single drainage area using
multiple regression and lag regression techniques. Although they did not
examine land use effects, the difficulties they encountered shed light on
this problem as well:
... the extreme range in concentration of constituents contained in
runoff makes it unlikely that anyone will ever be able to predict urban
runoff quality with a high degree of accuracy. Variations in concen-
trations of one to two orders of magnitude are common, and analysis of
all data that was available in August, 1971 from the watershed in
Lubbock showed standard deviations in concentrations of all constituents
normally being in the range of two-thirds to three-quarters of the mean
concentration (41, p. 12).
Rochester Study:
Runoff quality from combined sewer overflows in Rochester, New York, is
examined by Metcalf and Eddy, Inc. in "Development and Application of a
Simplified Stormwater Management Model" (12). Logarithmic regression ana-
lysis is first used to relate storm and antecedent conditions to runoff
quality during 29 storms with two to seven subareas averaged for each storm.
The results are probably not statistically significant with a correlation
coefficient of 0.257 for the equation predicting the chemical oxygen demand
concentration of the overflows, 0.181 for the equation predicting total
suspended solids, and 0.487 for the equation predicting nitrogenous oxygen
demand (where the NOD concentration was found to be negatively correlated
with the number of days since the last rain). The F statistic and the
resulting level of significance are not presented. The writers state that,
"The correlation coefficients are not high because stormwater quality is
highly variable, and the data that were developed from the sampling program
had some major irregularities" (12, p. 65).
5-76
-------
The Rochester data are also analyzed to examine the effects of land use:
Attempts were also made to develop equations including population
density to reflect the impact of land use patterns on stormwater qual-
ity. These equations were correlated with Data Set 2. The correlation
coefficient indicated that there was essentially no correlation between
the measured values and the predicted values from these equations. This
lack of correlation may be because of irregularities in the data and
because of the particular blend of land uses in the City of Rochester
(12, p. 66).
The difficulties encountered in obtaining significant correlations lead
the writers to suggest a conceptually simpler approach of analyzing the
average quality of subareas:
These subarea averages can be ranked to indicate trends. The signifi-
cant land use or surface characteristics can also be ranked. If these
rankings are indicated on a simplified map of the study area, areal
trends in overflow quality can be noted (12, p. 66).
The complexity and variability of the quality-land use relationships, partic-
ularly in a combined sewer area such as Rochester appear to justify these
simpler attempts to gain some level of insight, rather than sophisticated
efforts to predict actual concentrations.
Urban Stormwater Management and Technology Update:
A number of studies are cited in the recent update of the Urban Storm-
water Management and Technology report (36) which are directed towards
identifying correlations between runoff quality and land use or storm prop-
erties. These include analyses of separate runoff in Atlanta, Georgia,
Tulsa and Oklahoma City, Oklahoma, and Santa Clara County, California. Some
water quality parameters are found to be correlated with land use or storm
properties.
Land Use - Receiving Water Studies:
The concern for the water quality effects of land use has also led to
studies relating average, instream quality to land use and watershed activity
indicators. Haith found, "... significant correlations between land uses
and average nitrogen, phosphorus, and suspended solids concentrations..."
(45, p. 9) in his study of New York State streams.
A Carnegie-Mellon University study of 52 streams in the Pittsburgh,
Pennsylvania area (46) employed discriminant analysis to predict whether a
stream would be likely to violate applicable water quality standards based
on indirect indicators of watershed activities such as land use, the fraction
of the developed land unsewered, population, municipal and industrial
activity, inactive mines, etc. The writers conclude that, "... watershed
activity indicators not only provide a measure of overall water quality, but
also a reliable mechanism for predicting which water quality parameters are
likely to be problematic" (46, p. 249).
5- 11
-------
Although these studies deal with receiving stream quality, rather than
runoff or overflow concentrations, the techniques employed and their results
are of interest.
Conclusions and Implications:
A survey of the available information and a range of perspectives on the
relationship between land use and stormwater runoff quality has been present-
ed. A number of conclusions emerge from this review:
1. It would be desirable to establish some relationship between land
use ard stormwater quality to permit evaluation of the effect of future
land use changes and certain management practices on runoff quality.
2. Although nationwide studies have found significant differences in
runoff quality from different land uses, the variability of the data
precludes the establishment of a good general characterization of
quality as a function of land use, and makes the application of the
results to any particular area quite tenuous. Perhaps the clearest
implication is that local data is necessary.
3. When investigating land use effects in a particular study area,
conceptually simple approaches, such as looking for grouped averages
or discriminant analysis of groupings, may yield more insight than
attempts to predict actual runoff concentrations. Land use effects in
combined sewer areas should be particularly difficult to identify be-
cause of the predominant influence of sanitary sewage on the quality of
overflows. Finally, the analyst should anticipate that due to the com-
plexity and uncertainty involved in the stormwater runoff process, that
even for separate sewers and with a large local database, significant
correlations between land use and stormwater runoff quality may not be
evident.
5.4 Receiving Water
The methodologies suggested for initial assessments of receiving water
quality require a number of estimates for parameters, coefficients, and
rates. These include transport properties such as flow and dispersion,
coefficients to describe the reaction of pollutants, and estimates of the
background quality of the receiving water. Guidelines for estimating these
parameters are presented in the following sections.
5.4.1 Transport Properties
To assess the impact of stormwater loads, an estimate of the average
flow upstream of the loading area is required. In any drainage basin, stream
flow will vary over the year depending on the rainfall and/orzsnowmeIt.
Figure 5-30 shows the average monthly distribution of runoff as a percent of
the total runoff over the year for 16 river basins in the United States. It
is interesting to note that in the Kissimmee River Basin in Florida, the
stream flow is relatively constant over the year. However, in the Yellow-
stone River Basin, Montana, the maximum monthly stream flows are about 10
5-78
-------
i —
O
1C
(9
5-79
-------
times as great as the minimum monthly stream flows. In addition, the
Yellowstone River has a minimum stream flow which occurs in the month of
February by contrast to the majority of streams in the country, in which low
flow occurs in the late summer or early fall.
Because of the variation in runoff across the country, it is usually
necessary to collect site specific stream flow data. In general, there will
be one or more flow gaging stations within the 208 planning boundaries.
These records are published as annual surface water reports by the U.S.G.S.
Table 5-11 is a sample data sheet from the U.S.G.S. surface water records.
The data sheet summarizes the drainage area, daily flows for the year,
monthly average flows, and monthly maximum and minimum flows at the gaging
station.
For ungaged streams, an estimate of the average stream flow can be made
using Figure 5-31, which shows the distribution of average annual runoff
yield. Preliminary high and low flow runoff estimates can be made with the
use of the monthly runoff distributions presented in Figure 5-30.
In tidal rivers and estuaries, the freshwater flow is estimated by
adding the flows orginating from tributaries, municipal and industrial point
sources, and surface and groundwater inflow. While the freshwater flow in
tidal rivers and estuaries results in a net seaward flow the freshwater
velocity in tidal waters is generally small compared to the average tidal
velocity. Tidal velocity information is available from the National Ocean
Survey, which is a branch of the National Oceanic and Atmospheric Adminis-
tration. Tidal velocity information is used to calculate the atmospheric
reaeration coefficient and the average tidal translation which are required
for the water quality impact analysis.
5.4.1.1 Channel Geometry
Stream channel depth and cross-sectional area data are also required
for the main receiving water body. This information may be available from
the U.S.G.S., state agencies, EPA Coast and Geodetic Surveys, or from
previous studies. If no depth or cross-sectional area data are available,
it is usually necessary to make in-stream measurements.
In some study areas, time of passage (travel time) information will be
available from the U.S.G.S. or from previous studies. Time of passage data
along with stream cross-sectional area and depth data are used for computing
the average velocity. The time of passage data and the channel cross-
sectional area data should be related to the stream flow. If insufficient
data are available to establish these relationships, then Equations 5-10
to 5-13 can be used to estimate the changes in river characteristics under
different flow regimes.
a? Q? n <;
= C--) °'5 (5-10)
5-t80
-------
EXPLANATION
IN CUBIC FEET
PER SEC. PER SQUARE MILE
.8 to I 6
1.6 to 31
3.1 to 6.2
<.08
.08 to .4
.4 to .8
REF. (48)
FIGURE 5-31
AVERAGE RUNOFF YIELDS IN THE COTERMINOUS UNITED STATES
5-81
-------
LOCATION —Lat 43*59'08", lonR 75*55' JO". Icfferton County.
3.5 ml (5.6 km) u]>stre.im f mm I'hl I nan: I l'n.-L'k.
i)RAINA(,E AKEA.--1,«7<> ml1 (4,859 km2).
PERIOD OF RECORD.--July 1920 to current year.
STKKAMS TRIBUTARY TO I_AKt OVIAK1"
OAlftO^OO BLACK klVFR AT WATHtTOWN, N.Y.
downstrt^m side of rl>'lit abutmt-nt of V mduzr
H.imr wltu ,itiil ilnlum.
AVKKAf.f DISCHARGE.--54 yearn, ),M91 ftj/n (110.2 tnVa)
can.,1 season, water
1934 (RjKe h< l^ht, -0.1
1 17 ft3/:. ( » «* m'/s) Sept. 4, 1'iJ1*.
6500). tulttm ili.iln "f Ukis (sec station 04253500), and
,^Klv.r U,n,l m.,wlnK south). ,i t station UWS2000.
RFV!SI()Nb.--WSP
D15CHAHGE. IN CUBIC FEET
DAY OCT
1 1.410
2 1 .520
3 1.B20
4 2.200
5 2.320
6 2.470
7 2.710
6 2. 750
9 2.660
10 2.180
11 1.730
12 ,850
13 ,650
14 ,930
15 ,990
16 ,660
17 ,950
18 1,660
19 1,660
20 2.070
21 1,990
22 2.050
23 2.050
24 1,910
25 1,700
27 l.bOO
28 1.600
29 1.600
30 1.700
TOTAL 60.650
MEAN 1,956
MAX 2,750
MIN 1,410
NOV
2,020
3,890
5,310
5.490
5,170
4,470
1,760
3, JOO
3,010
2,650
2,560
2,460
2,530
2,580
2,530
4,360
5,770
6,150
6.100
5.560
4.490
3.560
3.400
3.500
3.440
4 .640
4.600
5.790
6.580
124,530
4,151
6,580
2,020
CAL Yd 1973 TOTAL 1,865,
WTH Yk 1974 TOTAL 1,H2S,
DtC
6,8'0
6.4'0
5.5JO
4,600
4,6UO
5,400
6, 7 )0
7,930
8,790
10,800
11,000
11,000
10,200
6,650
6,990
5,510
3,090
2, '60
3, 5BO
3,820
4,820
5,890
6,560
B, 290
6,990
13.500
15. 100
1 7.600
17.600
JAN
11,900
9,840
8,160
6,930
6,200
5.580
.780
,470
,490
,450
.510
.620
.340
3.740
3.170
3.420
3.740
3. '60
3.480
3.320
2.890
3.120
4.450
5.600
6.050
7.050
9.380
9.380
9.700
n - J J n
259.010 178.540
8.355
17.600
2. '80
3S9 MEAN
650 MEAN
5.759
11.900
2.890
5.111
S.002
etc SECONO. KATE"
FtB
6.690
B.290
6.640
5.790
4,700
4,490
3.740
3.580
3.520
3.340
3.120
3.070
3.070
3.100
3.250
2.990
2.710
2.700
2.670
2.760
3.300
3.500
6.130
6.250
6,670
7.14I>
6.730
......
129,610
4,629
8,690
2.700
MAX 16
MAX 20
MA8
6.130
5,720
5,100
5.720
9.910
1 1.300
16,700
IB. TOO
15,400
12,000
10.400
8.750
'.020
5.680
5.220
4.5'0
4.440
4.220
4, 170
4.240
4.900
3.660
3.760
3.660
4. 150
3. 720
3,200
3. 460
3,440
207,840
6,705
10, '00
3,200
,600 M1N
,400 M I N
YEAU OCTOUtH 1973
APR
4,090
4,620
7,520
16, 700
15,700
18,900
20,400
15.600
12.600
1 1 .200
9.030
8,250
6,690
8,440
11,700
13,700
17,200
13,900
12,000
10,400
9,470
8,530
9,380
6.640
9,600
8,610
7,580
6,400
6,160
324,670
10,830
20,400
4,090
859
1,060
MAY
6,230
6,600
6,600
6,400
6,180
5,660
6, 150
6,650
7,040
6,670
10,200
12,200
13.300
14,400
15,300
13,700
12.000
10,400
9.140
8,390
7,520
6,620
5,760
5.270
5,100
4,880
4.570
4.130
3,880
3,780
242,000
7,806
15,300
3,700
10 StUE*
JUN
3,600
3,540
3.500
J.500
3.280
3,020
2.000
2.270
2. 140
2.U80
3.040
4,530
4,400
J.720
3,160
1.100
3,800
4,310
3,760
3,260
2,960
2,690
2,500
2,400
2,500
2,540
2.000
2.530
2.160
92.580
3,086
4,530
2,060
UEM 1974
JUL
2.080
2.620
2.420
3.940
4.260
4.420
4.130
3.2BO
2.660
2.930
2.690
3.000
2.530
1 .900
1.860
1 .860
1 .370
1.330
1 .440
1.300
1.260
1.760
1.660
1 .630
1.650
1 .520
1 .920
2.500
3.560
75.080
2.422
4.420
1.260
AUO
4.350
3.860
2.960
2,560
3,960
5,570
5.760
5,240
4.070
2,890
2.450
1 .630
1 .920
1.710
1 .450
1.320
1 .330
1.200
1.710
2,130
1 ,810
1,440
1.250
1 ,100
I.OBO
1 , 1 '0
1 .360
1.2)0
1 .300
1.570
1.840
73,380
2,367
5,760
I.OBO
SEP
1.870
1.420
1. '60
1.890
2.710
2.610
2.190
2.130
2.220
1.360
1.220
1.130
.100
.160
.730
.710
.510
.360
.660
.920
l.'OO
2.050
3,060
3.240
2.540
2 . 240
2.180
2.060
1.890
2.080
57,760
1.92S
3,240
1.100
IBISSE, 17,000 CFSI
12-29 2200
3-08 0430
19,000
19,400
SOURCE: uses,SURFACE WATER HECOKO, NEW TOUK STATE, i»7« VOL
TABLE 5-11
SAMPLE U.S.G.S. SURFACE WATER RECORD DATA SHEET
5-82
-------
H2 ^204
l
U Q
0 5
0-5 (5-12)
Time of Travel Q9 n
_ £ = f_i ~u>i> f5
Time of Travel 1Q:J l
where Q is the flow rate, a is the cross-sectional area, H is the mean
stream depth, and U is the average velocity.
It should be recognized that these relationships are true only for
free flowing rivers and that the exponents may actually vary by 50% for any
river. Therefore, if possible, site specific exponents should be established
from available data.
5.4.1.2 Dispersion Coefficient Estimates
Estimates of dispersion coefficients are required to predict the atten-
uation of storm pulses in primarily advective streams (Section 3.5.2.1) and
to describe mixing in tidal rivers and estuaries (Section 3.5.1.2). The
longitudinal dispersion which attenuates concentrations in natural streams
(appropriate for the first analysis) depends on factors such as flow, shear
velocity, and channel characteristics. Considerable research has been con-
ducted into methods for estimating the longitudinal dispersion coefficient
(49,50,51,52). A recent study by Liu (52) indicates that the relationship:
(5-14)
is particularly successful for estimating applicable longitudinal dispersion
coefficients. In the above formulation,
Q = stream flow
U* = shear velocity = / gRs
where g = gravitational acceleration
and s = channel slope
R = hydraulic radius, or mean depth in large, wide rivers
and g = dimensionless coefficient
Typical values of g range from 0.001 to 0.1. To define a likely value for
B, the relationship
g = 0.18 (U^/U)1'5 (5-15)
5- 83
-------
where U = mean velocity is suggested (52). Equations (5-14) and (5-15) are
appropriate for an order of magnitude estimate of E, with errors expected to
be no greater than a factor of six. Sensitivity analyses should be conducted
to identify the effect on predicted maximum concentrations.
The dispersive properties of tidal rivers and estuaries are primarily
influenced by the density-induced enculations and the tidal velocities. The
effect in many estuaries is sufficient to reduce significantly the effect of
freshwater flow on the transport of pollutants, although the mixing effect is
a function of the density difference between saline water from the ocean and
the freshwater from the rivers. Thus, the magnitude of the dispersion
coefficient is relatively large in the vicinity of the mouth of the estuary
where both salinity and tidal effects are great. It decreases in the up-
stream direction with decreasing salinity and tides. It is further reduced
in the non-saline but tidal section of the estuary but is still of suffic-
ient magnitude to be taken into account in conjunction with the freshwater
flow. Typical values of E in tidal rivers and estuaries are shown in
Figure 5-32 (48).
5.4.2 Reaction of Pollutants
Reactive constituents are subject to change within the receiving water
due to physical, chemical and biological reactions. Examples of pollutants
that fall into this category are BOD, coliform bacteria and nutrients. Al-
though total nitrogen and phosphrous can often by treated as conservative on
an annual average basis, they can be reactive during the summer low flow
period due to algal uptake of the nutrients and subsequent removal by
settling.
Decay mechanisms occur for each of these constituents and first order
kinetics are assumed to be applicable. Representative reaction rates for
these constituents are indicated in Table 5-12. The coliform death rate
tends to increase with increasing salinity and exposure to light (53). The
BOD reaction rates are particularly applicable to the carbonaceous fraction,
but in a preliminary analysis, the rate is also considered appropriate for
the nitrogenous oxygen demand, though analyses to confirm the occurrence
of nitrification should be performed. The nutrient removal rates are gen-
erally applicable to conversion to other nutrient forms, but the lower range
also applies to an estimate of the first order removal coefficient due to
algal settling.
5-84
-------
IUU
Z>-
0°
ul Q. 10
DISPERSION
QUARE MILE
o
UJ —
n [
-
-
:
_ jm
i 1 1 1 1 1 1, 1
•
v
-
ef
A o e
c
-
i
i
O V
1 «ff f
&
) t
X
Sami - Dturnol
)
i i i i i i i i
IOO 1,000
NET NON-TIDAL FLOW
(CFS)
10,000
100,000
O Hudson River
A Delowore River
CD Cooper River
O Cope Fear River
^7 Savannah River
+ Waccosossa River
X Rhine River
• Houston Ship Channel
LEGEND
O Elms River
O Potomac River
• Wappmger and Fishkil
O Cornpton Creek
A Lower Roritan River
V South River
C) River Folye Estuary
Creeks
FIGURE 5-32
DISPERSION COEFFICIENT FOR DIFFERENT
TIDAL RIVERS AND ESTUARIES
5-85
-------
TABLE 5-12
RANGE OF VALUES OF REACTION COEFFICIENTS IN NATURAL WATERS (48)
Kr (per day)Ca)
Freshwater Tidal Rivers
Substance Streams and Estuaries
Coliform Bacteria 1-3 2-4
BODS 0.2 - 2.0 0.2 - 0.5
Nutrients 0.1-1.0 0.1-0.25
(a)Base e, 20°C
The coefficients in Table 5-12 are for water temperatures near 20 C. For
preliminary estimates, temperature corrections can generally be ignored.
Where appropriate, conversion to other temperatures can be made by:
T - 20
Kr(T) = Kr(20)(9)1 (5-16)
where K (T) is the reaction coefficient at temperature T(°C), K (20) is
the reaction coefficient at 20 C, and 6 ranges from 1.02 to l.lT
5.4.2.1 Sequential Reactions
Sequential reactions occur if the growth and removal of the initial
consitutent causes changes in a second constituent. For a preliminary
dissolved oxygen analysis, the initial constituent is ultimate oxygen
demand (UOD) and the dissolved oxygen deficit is the second constituent.
The loadings are expressed as UOD by scaling the carbonaceous BOD,, by
a factor (an estimate of the ratio of ultimate carbonaceous BOD to BOD-)
which ranges from 1.3 to 2.5. This ratio varies depending on the components
in the wastewater. Therefore, if a measured value is available for the
different waste loads in the system, this ratio should be used in the input
analysis. The nitrogenous oxygen demand can be approximated by multiplying
the reduced nitrogenous constituents (organic nitrogen and ammonia) by 4.57,
which is the mass of oxygen in pounds required to completely oxidize one
pound of ammonia. Total kjeldahl nitrogen measures both ammonia-N and
organic-N. The organic-N fraction is assumed to oxidize as ammonia does
although sequential reactions can be important.
The removal of UOD causes an uptake of oxygen and an increase in the
DO deficit of the stream. In a preliminary analysis, the UOD oxidation
rate (K.) is assumed equal to the UOD removal rate (K ). This assumes there
is no settling of the UOD. Where significant settling occurs, such as in
the vicinity of combined sewer overflows, K is somewhat higher than K,.
(Note that the settling of organic material often leads to bottom deposits
which exert a subsequent uptake of oxygen. An estimate of this "bottom
demand" should be included in more detailed modeling studies (54). An
estimate of K, can be made using the information in Figure 5-33. This
5-86
-------
4 0
0 I
005
O
O
-SlabJe,Rocky Bed
Moderate Treatment
Some Ammonia
Unstable,Sandy Channel
Highly Treated Effluent
with Nitrification
DEPTH IN FEET
LEGEND
©Shallow Streams (l-3Ft)
B Medium Streams (3-l5Ft)
A Deep Rivers ( > 15 Ft)
SOUHCf: (48 )
FIGURE 5-33
DEOXYGENATION COEFFICIENT (Kd) AS A FUNCTION OF DEPTH
5- 87
-------
Figure relates the oxidation rate to the stream depth and is based on data
collected during many stream studies (48).
The deficit caused by the oxidation is itself reduced through reaera-
tion. Reaeration coefficients (K ) may be calculated from a number of
formulae (55,56,57) such as the Cr Connor-Dobbins equation (55):
K = 12.96 U1/2/ H3/2 @ 20°C (5-17)
where K is per day, U is the average stream velocity in ft/sec and H is the
water depth in feet. When necessary, temperature corrections can be made
using:
T-70
K (T) = K (20) (9)1 ZU (5-18)
d d
Figure 5-34 is an alternate method useful for estimating the reaeration
coefficient, where K = K /H and K is the surface transfer rate (ft/day).
This Figure shows the range of measured values together with the curves
representing the theoretical formula.
Stream dissolved oxygen concentrations are calculated by deducting
the DO deficits from the temperature-dependent saturation concentration.
Curves of dissolved oxygen saturation concentrations versus temperature are
shown in Figure 5-35.
5.4.3 Background Receiving Water Conditions
The primary emphasis of this manual is the analysis of urban runoff and
overflows. Depending, however, on the characteristics of the study region,
such as the relative size of the upstream drainage area, upstream activities,
and areal storm patterns, the water quality of the receiving stream may be
dominated by the background conditions. Because the background quality of
the receiving stream entering the urban area is usually taken as a boundary
condition in the analysis, some guidance is needed to estimate pollutant
concentrations in the incoming flows.
Background water quality concentrations are usually available for most
drainage basins, and data sources should be reviewed. In general, back-
ground water quality is affected by land uses and water use practices
upstream of the urban area, and concentrations are therefore in excess of
natural levels. A number of approaches have been suggested for estimating
average background concentrations based on land use, geographic location,
drainage basin size, precipitation patterns and other factors (58,59,60,61),
however the variability of receiving water conditions, even from adjacent
and similar basins, indicates the need for site specific measurements of
water quality. If no data are available, background water originating from
relatively undisturbed drainage areas can be estimated using the values in
Table 5-13 (62).
-------
AVd H3d 133J-1N3IOIJJ30D d3JSNVdl
u
UJ
Q_
UJ
Q
CO
a.
>-
H
" !
3
2
in _•,
o '
10 ro
Q.
U
O
U.
O
o
I CO
in <
o
u.
u.
UJ
o
o
QC.
CO
z
<
or
uj
Q0OOO
5-89
-------
I/Bui- NOI1V81N30NOO
N39XXO
UJ
•n
in
^£
o.
2
UJ
I-
5-90
-------
TABLE 5-13
SUMMARY OF BACKGROUND CONCENTRATIONS^ FROM VIRGIN LAND
(a]
Parameter
Comments
Concentration
Range
Cmg/D
Nitrogen (inorganic) 0.05 - 0.50 highest concentrations: Iowa, Illinois,
Indiana
lowest concentrations: South, East,West
coasts
Phosphorus (total) 0.0 - 0.20 highest concentrations: Iowa, Nebraska,
Dakotas
lowest concentrations: South, East,
West coasts
BODr
0.50 - 3.0 highest concentrations: Iowa, Illinois
lowest concentrations: South, East
West coasts
Coliform (total)
(b)
100 - 2,000 highest concentrations: west of
Mississippi River
lowest concentrations: Northeast,
Southwest
Sediment (TSS)
2 - 100 highest concentrations: Montana, South
Dakota, Nebraska
lowest concentrations: East, West
coasts
(a)
(b)
See Midwest Research Institute (59) for iso-concentration maps of
virgin land runoff concentrations.
Number/100 ml.
5-91
-------
5.4.3.1 Variability of Background Conditions
For any upstream location in the receiving water the concentration of
pollutants may vary considerably with time. The amount of variability is a
function of waste load variations and drainage basin characteristics. Con-
centrations in larger rivers should tend to be somewhat less variable because
of dilution and the temporal smoothing of contributing runoff. In small
watersheds the receiving water variability is comparable to that of the storm
related input, because it is largely a direct result of runoff from individ-
ual, localized storm events. Analyses of data from a number of Northeastern
streams and rivers has indicated a trend towards greater variability in
smaller basins, though the results are not entirely conclusive (63).
Another useful approach to investigate the variability of background
quality is through flow-concentration correlations. If concentration is
plotted as a function of flow a number of possibilities exist. A concen-
tration-flow rating curve may be used to describe the relationship:
c - a (|)b (5-19)
where: c = constituent concentration
Q = streamflow
A = drainage area
a,b = coefficients
When concentrations tend to decrease at higher flow rates, this is commonly
referred to as a dilution effect. This occurs when lower concentration
runoff mixes with relatively constant flow sources of higher concentration
such as municipal or industrial point sources. If concentrations tend to
increase at higher flow rates, this is referred to as a runoff effect, where
storm runoff is a major source of pollutants in the stream. An alternative
cause of the runoff effect is the scouring and resuspension (during high
flow) of particles which have settled during the lower flow periods. Other
factors may also contribute, such as relationships between flow and reaction
rates, and seasonal flow patterns. Finally, concentrations may be observed
to be independent of flow. This occurs where a variety of factors affect the
quality-flow relationship, resulting in no net correlation.
A number of studies have investigated the relationship between flow and
concentration (64,65,66,67). In addition to the studies referenced, Hem
found that specific conductance (an indicator of dissolved solids) decreased
with flow in the San Francisco River at Clifton, Arizona, and was basically
independent of flow in the Gila River at Bylas, Arizona, though very high
flows did exhibit a lower specific conductance (68). A study of the Sacra-
mento River found suspended solids concentrations positively correlated
with flow (69).
Background sources are usually a relatively significant factor in
nutrient water quality, and the relationship between flow and nutrient con-
centrations has received considerable attention in the literature. Wang and
Evans found that orthophosphate concentrations demonstrate a dilution effect
in the Illinois River, a tendency to decrease with flow, while nitrogen forms
5-92
-------
2.00
I 00
0 80
060
— 040
0>
E
CO
z>
cc
o
-X
Q.
CO
O
X
Q.
0 20
0 10
008
006
004
002
OOl
200
O O
O O
o 8 o
-------
2
2 00
I 00
0.80
0.60
040
0 20
0 10
008
006
004
002
0°
00° O
o 01
200
O O O O O Qp OO
O ° Co OO
O O
(p
o o o
o o
t I I I I I I I
I I I 1 1 I I
I .1 I I
500 1,000
5,000 10,000
FLOW (cfs)
50,000
FIGURE 5-37
RELATIONSHIP BETWEEN ORGANIC NITROGEN
AND FLOW FOR GENESEE RIVER
(1968- 1974)
5-94
-------
2 00
I 00
080
060
040
--• 020
\
o>
E
Z 008
006
004
002
001
o o <* o <
o
00 °
o o
00 OO o
° O O °
00
o o
O OD O
08 o
0 o
0 °0 °0
o o
I I I
I I I I I I I
200
500 1,000
5,000 10,000
FLOW (cfs)
50,000
FIGURE 5-38
RELATIONSHIP BETWEEN AMMONIA
AND FLOW FORGENESEE RIVER
(1968-1974)
5-95
-------
io
O
200
100
0.80
0 60
0 40-
3 0 20
Q 010
z
"* 0 08
N
O
0 06
0 04
002
0 01
200
Q> o
O O Cfc C
o
o
00°
o o
o o
o o
o
o ° oo
o
I 1 I I I 1
i I I I I I
j I
500
1,000
5,000
10,000
50,000
FLOW (cfs)
FIGURE 5-39
RELATIONSHIP BETWEEN NITRATE AND NITRITE
AND FLOW FOR GENESEE RIVER
(1968-1974)
5-96
-------
,-
Q.
6
—
to
cr
o
X
Q.
to
O
X
Q.
_J
h-
f. UU
1 00
080
060
040
020
0 10
008
006
004
002
O 0 1
-
-
-
-
_
o
o
0
o o
o o o o
- o ° o
I *" ° oo o °
O O O f\ f\ Q
O OO O OO O r\
°° 0^° o o
0 O OO OO O
o
0 0 °
oooo
o
1 1 1 1 1 1 1 1 1 1 1 ll 1 1 1 1 1 1 1 1
20O 500 1,000 5,000 10,000 50,000
FLOW (cfs)
FIGURE 5-40
RELATIONSHIP BETWEEN TOTAL PHOSPHORUS
AND FLOW FOR TRENT RIVER
(1967-1973)
5-97
-------
2 00
I 00
0 80
0 60 -
040
en 020
O
r
£9 0.10
cr
O 0 08
0 06
0.04
002
0.01
200
0 ° o
o 0
-------
0>
E
300
200
100
080
060
040
020
< 010
008
006
004
OOO
O O
00
O O OOO O O
O O O OO
O O O
O OOO
0,02
O (D
O O
001
o ho 00)00*0'' 'od b
II
I I
200
5OO
1,000
5,000 (OpOO
FLOW (cfs)
50,000
FIGURE 5-42
RELATIONSHIP BETWEEN AMMONIA
AND FLOW FOR TRENT RIVER
(1967-1972)
5- 99
-------
300
200 -
I 00
080
060
040
UJ
CC 0 20
LJ 010
E 008
K
006
O
o
O
o o
o
o
oo
o
004
002
001
200
O O
ooo
o o
I
500 1,000
5,000 10,000
FLOW (cfs!
J 1
50,000
FIGURE 5-43
RELATIONSHIP BETWEEN NITRATE AND NITRITE
AND FLOW FOR TRENT RIVER
(1967-1972)
5-100
-------
tend to demonstrate more of a runoff effect (70). An Enviro Control report
on many drainage basins found that a runoff effect is more common for total
phosphate than for orthophosphate (71). Cahill, Imperato, and Verhoff found
that phosphate tends to be lower at higher flows in the Brandywine River when
considering steady state conditions, however, during unsteady stage flow,
phosphate tends to increase with increasing water flow rate (72). Two pos-
sible explanations are the increased scouring of sediments high in adsorbed
phosphate during increasing flow periods and the contribution of phosphate in
runoff from only a limited area near the waterway. A study of tributaries
to Lake Ontario found that total phosphorus and nitrogen components are
generally independent of flow in the Genessee and Trent Rivers (73). Figures
5-36 through 5-43 illustrate the relationships observed in these waterways.
Mean concentration and mean flow are indicated by arrows on the plots. The
exceptions to the independence finding are that nitrate-nitrite concentra-
tions increase with increased flow in the Genessee River, while ammonia
concentrations decrease with the flow. Analyses of other Northeastern water-
sheds indicates a general independence of flow rates and pollutant concentra-
tions. Local analyses are clearly needed to establish flow concentration
relationships for a particular study area.
5.5 Treatment Device Performance
A number of wastewater treatment devices have been developed or modified
for application to combined sewer overflows. Section 3.6.1.4. presents the
general methodology for estimating the long term efficiency of treatment
devices which operate on varying runoff flows. To apply this methodology to
a particular device of a given size, information is needed on the removal
.efficiency of the device at the mean runoff flow and the removal efficiency
at very low flows. This information may be determined by examining the
removal efficiency curve for each device. Performance curves developed in
this study are based on available information from laboratory, pilot scale,
and prototype scale treatment studies reported in the literature. Numerous
reports are available which summarize the results of pilot scale and proto-
type studies on the treatment of combined sewer overflows. Two reports (35,
36) have been completed by Metcalf and Eddy on the State of the Art of com-
bined sewer overflow treatment technology. The performance equations for
the various treatment alternatives in the EPA Stormwater Management Model
(SWMM) (74) were also reviewed and incorporated in the development of this
section. The information used to develop the performance curves is based
primarily on experience with dry weather flow and combined sewage. Some of
the treatment methods may be applicable to separate storm runoff, although
experience in this area is very limited, and care should be taken when
applying results obtained from combined sewage studies to separate runoff
analyses. This will be discussed in more detail in following sections.
As additional performance data becomes available thru literature
sources, or from local pilot studies, new performance curves can be prepared
and utilized in the methodology described in this manual.
5-101
-------
5.5.1 Considerations for Various Pollutants
Most of the treatment devices examined for stormwater control operate by
removing a portion of the suspended solids from the waste stream. These
devices separate the solids physically and may achieve additional separation
through chemical coagulation. The removal efficiency curves are thus pre-
sented in terms of the percent removal of suspended solids. The reduction
in the concentration of other contaminants is determined by the portion of
the pollutant which is associated with the suspended solids and the portion
which is in the soluble form. This breakdown depends upon the composition
of the stormwater and will vary in different locations. Separate runoff will
tend to have a very small portion of the BOD present in the soluble form,
while the soluble fraction may be somewhat higher in combined sewage.
General guidelines are suggested in the following section, and may be modi-
fied for particular study areas where sampling results indicate a different
relationship between a given pollutant and suspended solids.
The relationships for determining BOD, nutrients, and heavy metal
removal are expressed so that the percent removal of the substance equals
the percent removal of suspended solids times a factor. The range of report-
ed values in the literature reviewed for the ratio of BOD removal to SS
removal is 35 to 80 percent without chemical treatment, and 50 to 100 percent
with chemical treatment. As discussed, these ratios will be affected by
variation in the composition of combined sewage or runoff in different
locations and also the relative association of BOD with different size par-
ticles and the treatment process's capability to remove each particle size.
To approximate the BOD removal for each physical treatment device using the
performance curves for suspended solids removal, the following general re-
lationships can be used.
% BOD = .50 (% SS ) - no chemicals
% BODr = .60 (% SSr) - with chemicals
r r'
Relationships developed for nitrogen, phosphorus and heavy metals
removals as a function of suspended solids removal are based on a number of
studies dealing with the characterization and treatment of sanitary and com-
bined sewage (35,41,90,94,97,103,104). The factors developed for nitrogen
removal as a function of suspended solids removal for combined sewage are:
% TN = .30 (% SS ) - no chemicals
% TNr = .40 (% SSr) - with chemicals
r r'
The relationships developed for phosphorus removal are:
% TP = .15 (% SS ) - no chemicals
% TPr = .60 (% SSr) - filtration with polymer
% TPr = 1.0 (% SS1) - lime coagulation
Since one study (41) shows that street surface heavy metals were almost
completely insoluble in water after 25 days, they are assumed to be complete-
ly associated with the suspended solids and therefore removed at the same
rate as suspended solids:
5-102
-------
% Heavy metals = 1.0 (% SS )
5.5.2 Treatment Device Performance Efficiencies
The stormwater treatment devices analyzed, which rely primarily upon
the physical separation of solids, include sedimentation basins, air flota-
tion units, swirl concentrators, high rate filtration units, and screens and
microscreens. For each of these devices, a brief description of the process
operation is presented. A removal efficiency curve is presented for each
device, together with the documentation of the studies, tests, and analyses
used to develop the curve. When chemica addition may be used to enhance the
performance of the device, this is discussed and demonstrated. Typical pol-
lutant removal for BOD and suspended solids, hydraulic loading rates, and
detention times for these unit operations are summarized in Table 5-14.
Sections are also included on the applicability and analysis of bio-
logical treatment methods and disinfection for stormwater control. Finally,
the suitability of stormwater treatment at dry weather municipal treatment
plants is discussed.
5.5.2.1 Sedimentation Performance
In sedimentation, suspended solids are removed by gravity settling and
the removal efficiency is related to the settling velocity of the particles
contained in the wastewater and the design overflow rate of the settling
tank. If the settling velocity of a particle is greater than the overflow
rate or upflow velocity in the settling tank, the solids particle is removed.
Chemical addition can be used to increase particle agglomeration and settling
rates. Higher influent suspended solids concentration has also been shown to
increase removal efficiency.
The removal efficiency curve developed for sedimentation tanks, which
relates the percent removal of suspended solids as a function of the hydrau-
lic overflow rate (gpd/ft ) for various levels of influent suspended solids,
is developed by examining the original EPA SWMM model performance curves
(74), the revised SWMM curve based on work by the University of Florida (75),
performance data from a small treatment plant treating combined sewage in
New Providence, New Jersey (98), data on sedimenation tank performance pre-
sented by Eckenfelder and O'Connor (79) and the constant hydraulic loading
performance of settling basins in Rochester, New York treating combined
sewage (106). The Rochester study utilized a floculation basin with 6 to
14 minutes detention time for polymer addition and floe development prior to
the sedimentation basin. The sedimentation performance was evaluated for 19
storms. Polymer addition of 1 mg/1 was added during several tests and three
constant hydraulic loading conditions of 800, 1500, and 2000 gpd/ft were
evaluated. These Rochester data provided the basis for the performance
curve presented in Figure 5-44.
The removal curve is based on suspended solids removal efficiency at
constant hydraulic loading conditions. Fluctuation of the hydraulic loading
rate of settling basins will result in reduced removal efficiencies by
creating flow surges and turbulence in the unit. Performance data for
-------
100
5 90
_1 80
O 70
UJ
* 6°
Q
H 50
40
O
V)
O
UJ
o
z
LU
O.
CO
30
20
I 0
600
LEGEND:
NO CHEMICALS
WITH POLYMER
I
200 INFLUENTSS
(mg/I)
I
1000
2000
3000
OVERFLOW RATE ( GPD / FTC)
FIGURE 5-44
SEDIMENTATION TANK PERFORMANCE
5--104
-------
TABLE 5-14
COMPARISON OF TREATMENT ALTERNATIVES
fa\
Typical Pollutant Removals (%) *• J
Physical Treatment
1. Storage
2. Screens
Static
Rotary
Microscreen
3. Sedimentation
4. Air Flotation
BOD
No With
Chemicals Chemicals
4-8
8-28
16-72
20-40 40-50
30-50 48-80
SS
No With
Chemicals Chemicals
5-10
10-35
20-90
30-70 40-75
40-70 58-90
Hydraulic
Loading
Rate Detention
(1,0002 Time
gpd/ft ) (min.)
Range Range
70-260
20-70
20-70
1-3
2-4.5
-
-
-
40-120
15-20
5. Swirl Concentrator 0-35
6. High-rate
Filtration
40~60
20-70
40~80
80~96
10-250 0.4-10
10~60
l'8
Biological Treatment
2. High-rate
Trickling
Filtration
3. Lagoons
4. Rotating
Biological
Contactors
65-80
50-90
50-85
90
65-85
50-95
75-85
fa]
Minimum removal is at maximum hydraulic loading rates
'
4o-.7otWi5.i2oW'
0.9-2.7 40-1201 J
(-C'd-)
1 J
0. 1-2.3 2-20
.002-. 036 15-20
fcl
Lbs BOD^/1,000 cubic feet volume of contact with biological mass/day
Expressed in days
Time of flow in contact with biological mass, does not include
sedimentation.
5-105
-------
Rochester showed results similar to those from a full scale system in
Toronto treating combined sewer overflow, based on performance reported by
O'Brien and Gere (106). A study by Beak Consultants Limited (105) reported
that settling velocities of solids in stormwater runoff are an order of
magnitude lower than those in similar settling studies conducted on sanitary
sewage. Limited settling test data show that median settling rates are 1100
and 385 gpd/ft for sanitary sewage and urban stormwater respectively. No
tests were conducted on combined sewage. The relatively low settling veloci-
ties found for the stormwater tests are attributed to coarse clays and silt
in the samples. The study demonstrates that testing should be conducted for
specific areas and soil types. It is also observed that storage prior to
sedimentation may increase the settling rates of particles due to agglomera-
tion of small particles during storage.
The removal efficiency for sedimentation can be increased with the use
of chemicals such as alum, polymer and ferric chloride. Pilot scale per-
formance data from Rochester, New York, utilizing polymer, are used as the
basis for the performance curves for polymer addition shown in Figure 5-44.
5.5.2.2 Dissolved Air Flotation Performance
Dissolved air flotation (DAF) removes suspended solids by the release
of fine air bubbles from a pressurized, saturated recycle stream of waste-
water. Particles rise to the surface and are removed by skimming. A typical
DAF installation consists of pre-screening, a saturation or retention tank
to dissolve air into the recycle flow, an air compressor, a small mixing
chamber, a flotation cell, recycle pumps, a solids handling system to remove
floated solids, and chemical and polymer feed equipment.
The principal factors that affect the removal efficiency are the
hydraulic overflow rate, the pressurized recycle flowrate, and chemical
addition. The most critical design parameter is the rise velocity of the
combined air-solid particle expressed as a surface overflow rate. Metcalf
and Eddy (35) report maximum recommended air flotation design overflow rates
for combined sewage of 4,500 gpd/ft2.
Figure 5-45 shows the percent removal of suspended solids, both with
and without chemicals, as a function of the surface loading rate. The DAF
performance is calculated from the SWMM model equations using an average
solids concentration of 410 mg/1 and an average BOD concentration of 120
mg/1, based on data for combined overflows for several cities (35). The
calculated performance curve compared fairly well with the limited real data
that were available from pilot scale and prototype studies in Milwaukee,
Wisconsin (78), Fort Smith, Arkansas (84), San Francisco, California (83)
and Racine, Wisconsin (36).
5.5.2.3 Swirl Concentrator Performance
This device consists of a circular chamber in which the kinetic energy
of the combined sewage is used to impart a rotary motion. The principle of
the device is based on centrifugal force, created by the rotary motion of
combined sewage which follows a spiral path through the circular chamber.
5-106
-------
100
90
80
70
60
50
~ 40
<
>
o
tr
V)
O
V)
LJ
O.
V)
•3
V)
30
10
SS WITH CHEMICALS
SS
NOTE
MAXIMUM RECOMENDEO
HYDRAULIC LOADING RATE = 4500 GPD/FT2
I
I
1000 2000 3OOO 4000
SURFACE LOADING RATE (GPD/FT2)
5OOO
FIGURE 5-45
AIR FLOTATION PERFORMANCE
5-107
-------
This centrifugal force increases the removal of suspended solids by providing
an additional force on suspended particles causing them to settle at higher
rates than in conventional quiescent settling tanks. Solids tend to concen-
trate along the bottom and are discharged from the swirl concentrator through
a foul sewer underflow. This flow can be directed to the dry weather sewer
flow and subsequently treated. Effluent flow discharges over a circular
weir and can be conveyed to further treatment, and/or the receiving water.
Suspended solids removal in a swirl concentrator is due to the reduction
in suspended solids concentration in the overflow and a reduction in the
flowrate due to the foul sewer underflow fraction captured. The higher the
ratio of the influent flowrate to the foul sewer underflow, the less signifi-
cant is the additional mass removal caused by the foul sewer underflow. As
the hydraulic loading increases through a swirl concentrator, suspended
solids removal decreases. As with conventional sedimentation, higher in-
fluent suspended solids concentrations have been shown (89,106) to result in
increased removals at similar hydraulic loading rates.
The American Public Works Association (APWA) has completed laboratory
simulation and mathematical modeling studies of the swirl concentrator (85).
Combined sewer overflow characteristics such as the particle distribution and
specific gravity were simulated using synthetic materials. The body of the
report details the basis of the assumptions used to establish the character
and amount of flow to be treated, and the design of the swirl concentrator
based upon the hydraulic and mathematical studies.
The APWA recommended (85) that three flow rates be considered in the
design of a swirl concentrator:
1. the peak dry weather flow;
2. the design flow, i.e., the flow for which the optimum treat-
ment is desired; and
3. the maximum flow likely to pass through the chamber.
The peak dry weather flow should pass through the unit to the foul sewer
underflow and thence to the dry weather interceptor. The recommended
design overflow rate for the swirl concentration based on these studies
(85) was 13,300 gallons per day per foot diameter to the 5/2 power (gpd/
ft ). This power function of 5/2 is used to scale up the laboratory unit
performance to full scale based on Froude's Law.
The swirl concentrator treatment option has been added recently to the
original EPA SWMM model, based on work by Florida University (75). The unit
is modeled in such a manner that given the flow, the size of the swirl con-
centrator, the particle sizes and specific gravities, and the fraction of
particles by weight of each size, the performance can be computed based on
particle settling velocities and Stoke's Law, for each particle size and
corresponding specific gravity.
Two swirl concentrators have been evaluated (106) in Rochester, New York
for 19 storm events. A 3 foot diameter unit operating at a constant .hydrau-
lic loading per storm event which ranged from 4200 to 19,500 gpd/ft was
5-108
-------
sed to remove heavier particles or grit. Suspended solids removals were
0-80 percent at the low loading rates and 20-30 percent at the higher load-
ng rates. Removals increased with higher influent suspended solids concen-
rations at similar hydraulic loading conditions. A second swirl concentra-
or of 6 foot diameter received the effluent from the first swirl concentra-
or. Loading rates^were also kept constant for each storm and varied from
500 to 7000 gpd/ft ' . Removals ranged from 60-80 percent at the lower
.oading rate to 35 to 45 percent at the higher loading rate.
Storm runoff data have been reviewed for a twelve foot diameter swirl
concentrator in Syracuse, New York (109). The design flow was 6.8 mgd.
Total mass loading removals for suspended solids ranged from 44% to 65%
dth concentration reductions ranging from 18% to 55%. The foul sewer dis-
charge was approximately 10%. The higher mass loading reduction through the
swirl concentrator is due to both flow and concentration reduction.
The data performance from the Syracuse prototype unit indicates good
removals at the beginning of the storm, when concentrations were high, and
at the end of the storm when flow rates were low such that a high percent of
the flow discharged to the foul sewers. The data based on 11 storm,.events
showed average storm hydraulic loading rates as high as 6600 gpd/ft .
Figure 5-46 shows the performance of a swirl concentrator for mass sus-
pended solids removal as a function of flow/ diameter . In defining the
performance of different size swirl concentrators as a function of the flow
rate, the loading rate of gpd/ft is used, based on the scale-up relation-
ship developed in the initial study. Performance curves are shown for
three, ten, and twenty percent foul sewer underflows. These underflows are,,
expressed as percent of the design flow or peak storm flow (13,300 gpd/ft )
as used in the APWA report (85). A comparison of this performance curve
developed from the APWA study results with actual operating data from the
Rochester and Syracuse units, based on average per storm hydraulic loading
conditions, showed good agreement.
5.5.2.4 High Rate, Deep Bed Media Filtration Performance
In the high rate filtration process, wastewater is passed through a
dual media filter bed typically composed of anthracite and sand. Loading
rates are significantly higher than those commonly used in water treatment.
The filter bed is usually deeper and coarser to permit greater penetration
of solids into the media at high rates. Some screening must be provided
before the filters to remove coarser solids, to extend filter runs, reduce
head losses, and provide a more efficient operation.
The most important factors which affect the suspended solids removal
efficiency of high rate filters are the flux or filtration rate, the type or
size of the particlates, and the media size. Removal efficiency is inversely
proportional to the flux rate, and media size. Chemical treatment using
polyelectrolytes and/or alum can be used to increase suspended solids
removal.
5-109
-------
100
90
80
O
LD
o:
70
CO 60
Q
CO 50
a
z
a.
CO
CO
CO
40
30
20
10
100 %
REMOVAL
NOTE
•20% FOUL SEWER UNDERFLOW
-10% FOUL SEWER UNDERFLOW
-3% FOUL SEWER UNDERFLOW
DESIGN FLOW =13,300 GPD/FT5/2 (APWA STUDY)
I I I
IO 15
FLOW/DIAMETER5/2 (1000 GPD/FT5/2 \
20
25
FIGURE 5-46
SWIRL CONCENTRATOR PERFORMANCE
5-110
-------
Pilot scale filtration studies were conducted by Hydrotechnic, Inc.,
(90) to investigate high rate filtration techniques for combined sewer over-
flows in Cleveland, Ohio. The most successful results were obtained using a
dual media filter, consisting of anthracite coal over sand, with a fine 420
micron screen as a pretreatment device. The use of the screen reduced the
solids loading on the filter to yield acceptable filter run lengths. The
screen was operated at a hydraulic loading of 100 gallons per minute per
square foot, for the duration of the pilot testing and the filters were
operated at constant hydraulic loading rates from 10 to 40 gpm/ft . The dual
media consisting of five feet of No. 3 anthracite (effective size 4.0 mm)
over three feet of No. 12 sand (effective size 2.0 mm) was determined to
achieve the optimum results of the several alternatives evaluated. Poly-
electrolyte feed was found to be an essential part of the system to achieve
maximum solids removal efficiency. A maximum design loading rate of 24
gpm/ft was recommended based on a deterioration in filtration performance at
higher loadings.
Pilot scale high rate filter studies were conducted in Rochester, New
York by O'Brien and Gere Engineers (106). Performance of the filters was
evaluated at hydraulic loading rates of 10 to 25 gpm/ft both with and with-
out chemical pretreatment. A swirl concentrator was used as a pretreatment
device for solids removal. The filters consisted of 3 feet of No. 12
anthracite over 5 feet of No. 1220 sand. Chemical addition in the swirl
concentrator was found to achieve higher removal efficiencies than chemical
addition directly before the filter. The importance of contact time for
chemical conditioning was identified in this study.
A performance curve for high rate filtration is presented in Figure
5-47. This curve was developed using the Cleveland and Rochester pilot study
results (90,106). The suspended solids removal efficiencies are for the
filtration step only and do not include the solids removal in the pretreat-
ment step. The addition of polymer at dosages of approximately 1-2 mg/1
was shown to enhance removal efficiency by 20 to 30 percent.
5.5.2.5 Screens and Microscreens
Screens remove suspended material from combined sewer overflow by
physically straining the solids from the liquid. They range in size from 3
inch clear openings (bar screens) to openings as small as 15 micron (stain-
less steel woven microscreens). Screens have been divided into four classi-
fications by Metcalf and Eddy (35) 1, bar screens; 2, coarse screens; 3,
fine screens; and 4, microscreens.
1. Bar screens are typically installed prior to storage treatment
facilities and pump stations to protect downstream equipment.
2. Coarse screens or static screens are usually used as pretreat-
ment devices prior to other treatment units such as dissolved
air flotation, or high rate filtration to remove the coarse
solids and increase the operational efficiency of the subsequent
treatment steps.
5-111
-------
100
90
80
7°
liJ 60
o:
CO
Q
- 50
o
CO
0 40
LU
Q
CO
-D
(f)
30
20
10
POLY
NO CHEMICALS
10
20
FLUX RATE
30
4O
50
;GPM
FIGURE 5-47
HIGH RATE FILTRATION PERFORMANCE
5-112
-------
3. Fine screens and microscreens are similar in operation but
differ in the size of the aperatures. Fine screen devices
commonly referred to as drum screens, have a range of apera-
ture sizes from 100 to 600 microns. These screens may rotate
on either a vertical, or horizontal axis. Several 105 micron
units have been tested.
4. Microscreens, commonly called microstrainers, have aperature
sizes ranging from 15 to 65 microns. They provide versatility
since they can be designed for specific applications by changing
the aperature size of the screen. A typical microstrainer
unit consists of a rotating drum fitted with a fine screen
operating partly submerged in a tank. The stormwater enters
the interior of the drum, through the open end, passes out
through the screen and into the outlet chamber. The suspended
solids are retained on the screen and develop a mat of screened
particles that acts as a strainer, retaining particles even
smaller than the screen aperature. As it rotates, the screen
and its mat of retained solids passes under a row of backwash
jets which wash the solids into a trough for disposal. The
backwash water stream is small and is usually sent to a dry
weather treatment plant. The backwash water source is usually
microstrained effluent. In microstraining applications, the
head loss is about_6 inches for a 23 micron screen with a flow
rate of 6-8 gpm/ft of gross submerged screen area. Micro-
straining mav be operated at 24 inches head differential and
15-25 gpm/ft hydraulic loading. With polyelectrolyte hydraulic
loading rates of 30-40 gpm/ft are reported to be possible
with similar effluent quality. Microstrainers should be
preceeded by a coarse bar screen and have a bypass arrangement
to divert flow in excess of the peak capacity of the treatment
equipment.
Several pilot scale and prototype studies (36,76,77,78,82,106,107,108,
109) have been conducted to evaluate the performance of screening devices to
remove suspended solids from combined sewer overflows. Detailed description
of the studies and results are contained in these reports. In general,
these studies demonstrate that the suspended solids removal efficiency of
screening devices, especially the fine screens and microscreens, is related
to the influent suspended solids concentration and not the hydraulic loading
rate. Screens tend to produce a consistent effluent suspended solids con-
centration independent of the influent solids concentration and hence have
higher removal efficiencies at higher influent suspended solids concentra-
tions. Maximum hydraulic loading rates are defined in these studies and vary
from a low 16-36 gpm/ft for a 23 micron microscreen (with the higher value
possible with polymer addition), to as high as 180 gpm/ft for static
screens.
Figure 5-48 shows the performance of various size screens and micro-
screens as a function of influent suspended solids concentrations. These
curves have been developed from performance data from a 23 micron micro-
strainer tested by Glover (76) on combined sewage in Callowhill, Philadel-
5- 113
-------
100
80
60
UJ
o:
o
_i
o
S 40
a.
(f>
13
20
ROTARY SCREEN, 105 MICRON
STATIC SCREENS, 250-760 MICRON
0 100 200 300 400 500
INFLUENT SUSPENDED SOLIDS CONCENTRATION (mg/l)
FIGURE 5-48
PERFORMANCE OF SCREENS
5-114
-------
phia; 105 micron rotating screens in Belleville, Ontario, (36) Fort Wayne,
Indiana (107), and Syracuse, New York (109); and hydrosieves and static
screens in Fort Wayne, Indiana (107), and Belleville, Ontario (36) respectiv-
ely. These curves represents an average performance or removal efficiency at
various influent suspended solids concentration. Actual performance data
from these pilot and prototype screening studies show a range of +_ 15-20
percent at influent suspended solids concentration greater than 200 mg/1
and +^ 30-40 percent at lower influent suspended solids concentrations.
5.5.2.6 Biological Treatment
Biological treatment systems investigated to treat combined sewer
overflows include contact stabilization, high rate trickling filtration,
lagoons, and rotating biological contactors. Typical removals, loading
rates and detention times are shown for each system in Table 5-14.
Contact stabilization is a modification of the conventional activated
sludge process. Since less tank volume is required to yield carbonaceous
BOD and suspended solids removals in the range of 75-90 percent, the contact
stabilization process is considered more applicable to treat combined sewer
overflows than conventional activated sludge (35). In the treatment process,
the combined sever overflow is mixed with return activated sludge for twenty
minutes detention at the design flow (35). Removal of BOD is accomplished
by adsorption onto the biological sludge. Following the contact period, the
flow is settled in a clarifier. The concentrated activated sludge is re-
turned from the bottom of the clarifier to a stabilization basin, where it
is aerated for several hours. During this period, the organics from the
wastewater are utilized for growth and respiration of the organisms. The
overflow from the clarifier may then be chlorinated and discharged. The
stabilized sludge is returned to the contact tank to mix with the incoming
wastewater, and excess sludge is wasted.
A study at a 20 MGD combined sewer overflow facility in Kenosha,
Wisconsin showed overall removal efficiencies of 92% S.S., 83% BOD, 50%
Total-N and 50% Total P (35). No correlations are made between performance
and parameters such as sludge age, food to microorganism ratio or detention
time, because overflows did not last long enough for the plant to stabilize.
High rate trickling filtration is used at a plant in New Providence,
New Jersey (98) to treat both domestic sewage and combined runoff. The
plant was designed to treat a DWF of 0.6 MGD and a maximum wet weather flow
of 6.0 MGD. Both rock and plastic media are used in each of the two filters.
Performance of this facility is presented in terms of both hydraulic and
organic loading rates and as a percentage of dry weather flow, in Figures
5-49 and 5-50. Under storm conditions, a trickling filter must handle
highly varying flows. Applying a varying organic load to a filter does not
produce optimum removals. It is generally thought that only sufficient
biomass adhers to the supporting medium to oxidize the normal organic load,
although, some excess biomass always adheres to the medium and can accept
some of the organic load.
5- 115
-------
AON3 I
G08
UJ
o
cr
o
0) U-
UJ
o
a:
5-116
-------
CO
a
0
CO
a
Ul
0
z
Ul
¥1 fl
C\ Q CO
2 0 3
kj m OT
•
/
/
/
.
,
}/
'
• • ,
/
/
/'"
•I • •
"^ l^^^M
a:
UJ
X
K
S
y
1
I
/
' / '
'
1
I
M
^
"
•i - —
1
/
/
•
CO
b_
LJ
a:
^
(O
tf>
Q
^" ^
O
U.
to
CVJ
O O O O O *""
O 00 IO 1- oj
(%) AON3IDIdd3 1VAOW3H
SQI~IOS Q3QN3dSnS ONV SQ08 "l~IVy3AO
UJ
o
z
5
cr
o
li-
ar
Ul
O 0-
mf*
?H
i
iii — •
ID OC
e> uj
ul ^
u
UB
O
2
O
a:
I-
5--117
-------
A varying hydraulic load also affects removals. The increased shearing
action at high flows causes excess sloughing or washing off of the biomass.
To help dampen this effect, filters operating in series under dry-weather
conditions can be operated in parallel during storm flows, thereby reducing
some of the increased hydraulic load on each filter.
In designing a trickling filter to treat overflows, it must be remember-
ed that dry-weather flow is needed to keep the biomass active between storms.
Generally, two or more units should be used to provide high removals by
operating in series during dry weather and in parallel during storm events
to accommodate the flow variation required.
A unique feature of the New Providence plant is the method of filter
operation designed to keep a live biomass available on both filters at all
times. To do so, the filters operate in series with the plastic medium
filter in a lead position for treating all flows up to 2.8 mgd. At this
point, an automatic transfer to parallel operation is accomplished and
maintained until flows again drop within the series range. In parallel
operation (the normal combined sewer overflow treatment mode), both filters
receive equal flow, resulting in a much higher (3 to 1) unit loading on the
smaller plastic medium filter.
Lagoons can be used to both store and treat wastewater. Stormwater
lagoons can be of several types: oxidation ponds in which oxygen is supplied
by natural means from the atmosphere and algae growth, aerated lagoons which
use mechanical aeration for mixing and oxygen supply; and facultative lagoons
which have an upper aerated zone and a lower sludge decomposition anaerobic
zone. A major problem of lagoon treatment is solids and algae carryover
into the effluent. Solids removal devices such as microscreens and filtra-
tion have been used in some locations to treat lagoon effluents.
Rotating biological discs, a recent development in the biological
treatment field, were tested on combined sewer overflows in Milwaukee,
Wisconsin, because of their reported ability to handle highly varying flows
(99).
The rotating biological contactor consists of a set of rotating discs
upon which a biomass is grown. The rotating discs are partially submerged
in the wastewater, and baffles are used to prevent shortcircuiting. The
waste flow enters the contact tank at one end and is allowed to flow either
perpendicular to, or parallel to, one or more units in series for treatment.
The removal of organic matter is accomplished by adsorption at the surface
of the biological growth covering the rotating discs. Rotational shearing
forces cause sloughing of excess biomass. Secondary clarification is nor-
mally provided to remove any sloughed biomass from the wastewater. The
demonstration pilot plant exhibited an ability to treat storm flows of 14
to 20 times dry weather flow. The performance of the Milwaukee facility for
BOD, COD, total nitrogen, and phosphorus is presented in Figure 5-51.
Biological treatment dependent on living organisms, and efficient
performance requires that these organisms be sustained by continual applica-
tion of wastewater. Storm overflows are random in their occurrence, and the
5-'11-8
-------
100
MAX1 VUM
LOW MG
RATE
LOSS OF B CVASS AT
>26gpd/S'J FT
NOTE
DESIGNFLOW FOR OPTIMUM REMOVALS = 5gpd/SO FT
10 20 30 40
HYDRAULIC LOADING RATE (GPD / ft2 )
FIGURE 5-51
ROTATING BIOLOGICAL CONTACTORS PERFORMANCE
5-119
-------
time between overflows can at times be quite long. This situation limits the
appropriateness of biological treatment for stormwater runoff to situations
where such facilities are operated at the dry weather treatment plant, or
sustained by dry weather flow diverted from the principal waste stream. A
possible exception to this limitation are lagoons which are less dependent
on a constant waste application may go for long periods without discharge.
5.5.2.7 Disinfection
A number of modifications in both the design and analysis of disinfec-
tion systems are required when they are applied to the treatment of storm-
water runoff or overflows. Nevertheless, disinfection is often considered
for wet weather control either by itself or in combination with other treat-
ment devices, particularly when treating combined sewer overflows. Field
(110) presents a brief review of recent developments in stormwater disin-
fection. The methods currently available incorporate either chlorine, sodium
and calcium hypochlorite, chlorine dioxide, or ozone. Recent research has
emphasized high rate disinfection with the more rapid oxidants (i.e.,
chlorine dioxide and ozone) two stage disinfection, and methods of obtaining
rapid, adequate mixing. The possibility of residual toxicity effects must
be considered when large dosages of disinfectant are used for combined sewer
overflows.
In this section, the fundamental factors which affect disinfection
performance are analyzed to estimate the effect of storm flow variation on
treatment efficiency. A generalized approach is developed and may be used to
estimate the long term performance of disinfection systems which operate at
variable rates, when removals are specified at a fixed flow rate.
The basic mechanisms of disinfection are reviewed by Collins, Selleck,
and White (111) where they compare the bacteria removal obtained by an ideal
plug flow device (where each element of water has the same contact time, t)
and an ideal completely mixed tank (where contact times are exponentially
distributed with mean t). Each element of water is assumed to behave accord-
ing to Chicks Law Q12), which states that the survival fraction of bacteria
organisms equals e , where k is the kill rate and t is the contact time of
that element. The-overall fraction of bacteria remaining for the plug flow
system is thus e~ ; while for the complete mix device the fraction remaining
is (1 + kt)~ . Evaluation of the expected value of these removals over
varying influent flows (using the method developed further in this section)
confirms the conclusion of Glover and Herbert (76) that wet weather disin-
fection systems should be designed to approach plug flow rather than complete
mix conditions. Glover and Herbert suggest a flow-through device with
corrugated baffles to increase the mixing intensity without causing a
significant deviation from plug flow conditions. Their design is depicted in
Figure 5-52.
In an actual treatment device such as the one depicted in Figure 5-52,
the ideal plug flow removals (e~ ) may not be obtained. There are two
primary reasons for the deviation:
i. Chicks Law for the bacteria decay may not be valid;
5-120
-------
o
h-
o
o
u
o
u.
(O
o
or
<
o
(D
z
10
CM
10
I
in
UJ (O
o: u
Z) O
« UJ
u. o
UJ
o
z
o
U
CD
oe
UJ
o
UJ
u.
I
O
4
oe
o
z
o
u.
o
UJ
Q
Q.
o
5-121
-------
ii. The distribution of contact time will be non-ideal, and have
some spread around the mean, t. '
These issues are discussed by Collins, Selleck, and White (111). Their
empirical analysis of devices designed towards plug flow indicates that the
fraction remaining may be expressed as (1 + kt)~ , where k is an effective
kill rate proportional to the concentration of disinfectant, in this case
the combined amperometric chlorine residual. This performance may be
thought of as being intermediate between the ideal plug flow performance and
that obtained in a completely mixed tank.
To determine the performance of the empirical disinfection device
treating varying influent flow, the analysis of Collins et al is expanded
with a number of simplifying assumptions. It is felt that these assumptions
allow for useful solutions which are consistant with the order-of-magnitude
accuracy of the overall analysis.
The first of these assumptions concern the chlorine dosage and the
effective kill rate.
1. The kill rate, k,, is proportion to the chlorine concentration,
(Cl), raised to the n power.
k a (Cl)n (5-20)
2. The chlorine concentration is equal to the chlorine dosage,
W , divided by the flow through the device, q.
LJ -L
(Cl) = Wcl/q (5,21)
This is a gross simplification of the chemical process of chlorination,
ignoring the interaction of free available and combined available chlorine,
chlorine demand and chlorine residual, and the temperature and pH. The
exponent n is referred to as a coefficient of dilution, with reported values
ranging from 0.8 to 1.5 (113). In the current analysis, n is assumed equal
to 1.0. If n is greater than 1.0, the kill rate is more sensitive to the
concentrations of disinfectant, and subsequently more sensitive to the in-
fluent flow rate when the chlorine dosage rate is constant and not metered
in proportion to flow. Note that while chlorine is assumed to be the dis-
infectant, the analysis is also applicable to other disinfectants, as dis-
cussed later in the development of an alternative disinfection model.
Assumptions 1 and 2 allow a determination of the kill rate given the in-
fluent flow and rate of chlorine dosage. If W,,, is constant (i.e., the
dosage mechanism is triggered when a storm begins, but dosage is not pro-
portioned to flow) the kill rate will be high during small storms and de-
crease as the flow increases. If the chlorine dosage is proportioned to the
flow, a constant kill rate may be maintained. Note that the improved mixing
of the chlorine (resulting in better kills) which may be caused by the higher
velocity gradients associated with larger flows is not incorporated. This
may make the analysis somewhat conservative, as is discussed further later
in this section.
5- 122
-------
The relationship between the contact time (residence time), the size'
of the device, and the influent flow is also important.
3. The mean contact time, t, equals the volume of the disinfection
device, V , divided by the influent flow.
t = VD/q (5-22)
This says that if the flow doubles, the runoff will pass through the system
twice as fast. This does not allow for the backup of runoff, which should
be analyzed separately as a storage device. The disinfection is considered
to be completely in line with the varying runoff flow. This assumption is
reasonable for devices which must treat the large surges associated with
storm runoff.
A storm is assumed to be defined by an average flow. The variation
between storms is assumed to be much bigger than the variation within storms
The probability distribution function of average storm flows will determine
the long term performance.
4. Runoff flows are assumed to be exponentially distributed, with
a mean Qp. The probability density function of average storm flows
is thus :
Pq(q) = - e (5-23)
This would indicate a coefficient of variation for the flow, v , equals
to one. When v is greater than one, or if the within storm variation is
large, the disinfection device will perform more poorly than predicted in
these analyses.
The final assumption concerns the relationship between the runoff flow
and coliform counts.
5. The concentration of bacteria in the runoff (c) is independent
of the flow.
If bacteria concentrations are positively correlated with flow (i.e.,
greater flows have higher bacteria concentrations) , the disinfection device
will perform more poorly than predicted here. If higher bacteria concentra-
tions are associated with lower flows, performance will be underestimated.
Given the assumptions and the empirical performance equation for the
disinfection system designed towards plug flow, the fraction remaining, f(q),
of bacteria as a function of flow is determined.
The basic equation for the empirical device is:
f(q) = (1 + kt)"3 (5-24)
W V
Cl - C
where k a - , and t = —
q q
5- 123
-------
Note that k is constant when the chlorine dose is regulated proportional to
the influent flow.
The following terms are now defined:
(kt)D = the value of (kt") at the mean runoff flow, Q_
K K
fR = the fraction remaining at the mean runoff flow
= 1 * (kt)R -3
The expression for f(q) for the flow proportioned dosage (constant
k) can be derived as:
3
f (q) = —3— T (5-25)
where h = QR(fR1/3 - 1)
The fraction remaining for the case where W is constant is:
6
f (q) = —j-* ^ (5-26)
(q + hQRr
Equations (5-25) and (5-26) .are plotted in normalized form in Figure 5-53
for the case where f_= 10 . Note that as expected, the device with the
flow regulated chlorine dosage is less sensitive to flow. In both cases,
however, the disinfection performance is considerably reduced at high flows
(intense storms) where a large fraction of the total bacteria load is dis-
charged.
The long term fraction of the stormwater bacteria load remaining after
disinfection is calculated as an expected value:
oo oo oo
/ / / f(q) cdq p (c) p (d) p (q) dc dd dq
M = c=0 d=0 q=0 4 f- .
M l J
R cDQR
Assuming q is independent of c and d:
_ oo
_M 1 .
MR = QR q=0 f(q) q PqCq) dq (5'28)
For the flow regulated chlorine dosage:
oo 4 -a/0
M 1 , q q ^ A ft on-,
— = —=• / —^—r- e dq (5-29)
MR Q Z q=0 (q+h)-5
5- 124
-------
bJ
o:
LU
o
a:
99 9999
999998
999996
99 9994
99 999
99 998
99 996
99 994
99 99
99.98
99.96
99 94
99 9
99 8
99.6
99.4
99
98
96
94
90
80
60
40
0
0 0
FLOW PROPORTIONAL
CHLORINE FEED
CONSTANT CHLORINE FEED
1.0
2 0
30
4 0
5 0
10'
10 '
ID'4
10
-j a:
t-
o
or
u.
10"'
10"
I
60
- NORMALIZED INFLUENT FLOW
FIGURE 5-53
DISINFECTION RATING CURVE FOR EMPIRICAL DEVICE
WITH FOUR ORDER REDUCTION AT MEAN RUNOFF FLOW
5-125
-------
For the constant chlorine dosage:
~q/Q
/ — - - 3 e dq (5-30)
R QR q=0 (q +h
Equations (5-29) and (5-30) are evaluated with numerical integration and the
results are displayed in Figure 3-24. reproduced as Figure 5-54. As discus-
sed in Section 3.6.1.4.3, Figure 5-54 demonstrates the effect of flow varia-
bility on the average performance of disinfection. The long term performance
of a system with a proportionally regulated chlorine dosage subjected to
varying storm flows is about one order of magnitude poorer than a system re-
ceiving a constant flow rate (i.e., the mean runoff flow, Q ) . The long term
performance of the system treating varying flows with a constant chlorine
dosage is about two to three orders of magnitude poorer than the constant
flow device.
Alternative Disinfection Model
In a recent study of high-rate disinfection in Rochester, New York,
Geisser and Carver (114) perform multiple regression analysis on disin-
fection performance data. Their analysis for high-rate chlorine disinfection
yields the following equation:
N , . -0.0043C -0.00456C , .
log (I) = 0.04223 C°'66245 a0'28090 T0'45611 10 2 3. ..(5-31)
N2 l
N
, , ,1, .. , . -, , , influent F. Coli.
where log (_) = log kill = log (effluent F. Coli)
C1 = concentration of chlorine (mg/1)
G = velocity gradient (minutes )
T = nominal contact time (minutes)
C? = concentration of TKN (mg/1)
C, = concentration of BOD,, (mg/1)
The disinfection performance improves with larger chlorine concentration
[C = (Cl)], longer detention time, and higher velocity gradients (improved
mixing). Higher concentrations of TKN and BOD reduce disinfection
efficiency, reportedly because they react with chlorine, reducing the disin-
fectant available for bacteria reduction. The study of Geisser and Carver
should be referred to for a more detailed discussion of the conditions of
the experiment, limitations in the range of the independent variables, and
the implications of the results.
To compare Equation (5-31) with the previously developed model, the
sensitivity to influent flow of each term in the equation is evaluated.
5- 126
-------
o
S
t-
z
IU
o
CE
UJ
£L
5
K
UJ
I-
o
z
o
99.9999
99.9998
99 9996
99.9994
99.999
99 998
99.996
99.994
99.99
99.98
99.96
99.94
99.9
99.8
99.6
99.4
99
98
96
94
90
80
60
40
VARYING FLOW, PROPORTIONAL
CHLORINE FEED
•VARYING FLOW, CONSTANT
CHLORINE FEED
I
I
I
I
I
10"
10
10"
10"
10"
10"
0 5 10 15 20 25 30 35 40 45 50
(kt)R= DEVICE SPECIFICATIONS AT MEAN RUNOFF FLOW (QR)
<
LJ
(T
Z
O
t-
0
o:
LU
FIGURE 5-54
EFFECT OF STORMFLOW VARIATION
ON PERFORMANCE OF EMPIRICAL DISINFECTION DEVICE
5-127
-------
The chlorine concentration is assumed constant when a flow proportional
chlorine feed is used, and inversely proportional to the flow when W is
constant:
GI = constant, for proportional chlorine feed
! (5-32)
C..a —, for constant chlorine feed
The contact time is inversely proportional to the flow:
T a i (5-33)
The velocity gradient is proportional to the square root of the velocity,
and thus proportional to the square root of the flow.
1/2
G a qix (5-34)
The BOD- and TKN concentrations are assumed independent of the flow, Com-
bining these assumptions with Equation (5-31) yields the following relation-
ships :
For the device with a flow proportional chlorine feed:
log (-) a q~°'3 (5-35)
For the device with a constant chlorine dosage rate:
log (rp) a q-1-0 (5-36)
N2
This allows a determination of the removal flow relationship given the re-
moval at the mean runoff flow, Q . This is demonstrated and compared to the
previous model in Figure 5-55, for a device which yields a 4 order bacteria
reduction at the mean runoff flow (10" remaining). The relationships are
quite similar and expected values of the long term average performance
should likewise be similar (i.e., Figure 5-54 would be similar using the
alternative model) . This provides further support for the findings of the
current study.
Carver and Geisser also evaluated the effectiveness of chlorine
dioxide with the following results:
Ni n 62897 n nso?2 n 07*12 0.00314C -0.00719C f5 _71
log (J-) = 0.95229 c°'62897 G0'05022 T°-07812 10 2 3. . . (5-37)
N2
where the terms are the same as in Equation (5-31), except C is the concen-
tration of CIO- (mg/1) rather than Cl,,. Note the low level of dependency
on contact time, T. The flow relationships are thus:
5- 128
-------
UJ
o:
o
o:
99.9999
99.9998
99.9996
99 9994
99.999
99.998
99 996
99.994
99.99
99.98
99.96
99.94
99.9
99 8
99.6
99.4
99
98
96
94
90
80
60
40
0
0
CONSTANT CHLORINE FEED
FLOW PROPORTIONAL CHLORINE FEED
Eq. (5-35), ALTERNATIVE MODEL
Fig 5-53, ORIGINAL MODEL
Eq. (5-36), ALTERNATIVE MODEL
Fig. 5-53, ORIGINAL MODEL
J I
10"
10"'
iff4
10
UJ
3 o:
o
<
at
10"
10'
1.0
2 0
30
4.0
50
6.0
= NORMALIZED INFLUENT FLOW
FIGURE 5-55
COMPARISON OF DISINFECTION RATING CURVES
FOR ALTERNATIVE MODELS
(ASSUME FOUR ORDER REDUCTION AT MEAN RUNOFF FLOW)
5-129
-------
For a device with a flow proportional chlorine dioxide feed:
log (-I) a q-°-05 (5-38)
2
For a device with a constant chlorine dioxide dosage rate:
log (J-) a q-°-7 (5_39)
N2
The sensitivity to flow is lower in both cases when C10_ is used rather than
Cl~. This would yield flatter performance curves in Figure 5-55 and less
subsequent deterioration in the long term average performance due to varying
influent flows. This type of conclusion may also apply to ozone due to its
rapid oxidation properties.
Carver and Geisser perform a cost/benefit analysis which shows that
chlorine is more cost effective than chlorine dioxide for the treatment of
Rochester combined sewage. Their analysis assumes a constant, design in-
fluent flow rate at the treatment plant. It would be interesting to see if
this conclusion is still upheld if an analysis allowing for varying influent
flows is performed.
Observed Disinfection Reductions
The primary conclusion of the current disinfection analysis is that long
term reductions of bacteria organisms in variable stormwater overflow will
probably be about one to two orders of magnitude, rather than the four order
reductions usually obtainable under fixed influent conditions. A recent
study of a New Orleans wet weather disinfection program tends to support
this assertion Pontius, Pavia, and Crowder (115) report on the decrease in
coliform bacteria levels in New Orleans stormwater outfall channels after
the installation of a sodium hypochlorite (NaOCl) disinfection system. They
report that tests of the treated water indicated 99.99% or greater removals
of coliform when a 0.5 mg/1 chlorine residual (total available) was present.
After actual system operation, however, "...long term fecal coliform levels
were reduced by one order of magnitude in each outfall canal." (115,
Abstract) Total coliform levels were not reduced due to aftergrowth-
recovery effects. While aftergrowth and other sources of coliform bacteria
(i.e., sediments) may have affected the results, the one order fecal coli-
form reductions in the outfall canals (as opposed to the four order reduc-
tions under test conditions) are consistent with the results presented in
this manual.
5.5.2.8 Treatment at Dry Weather Plants
Municipal treatment plants in areas served by a combined sewer system
receive a portion of the stormwater runoff. The amount of wet weather flow
reaching the plant is dependent on regulator settings in the sewerage system
and the influent capacity at the plant. While a treatment plant receiving
combined sewage may be modeled by calculating the portion of the runoff
captured and conveyed to the plant with DWF, an alternative approach for an
5-130
-------
initial assessment is to consider the entire system as a flow sensitive
treatment device. The overall percent removal decreases with increasing
runoff flows due to both the deteriorated performance at the plant and the
bypassing of overflows. This section presents considerations for this type
of analysis at dry weather treatment plants.
The quantity of wet weather flow which can be handled in a treatment
unit operation is dependent on the hydraulic design. Dry weather treatment
plants are typically sized for a peak flow rate of about three times the
average dry weather flow. Flows in excess of the design result in flooding
and overflow of the. tanks. For treatment plants in a combined sewer area,
hydraulic design flows might be substantially higher (ie. 10-15 times dry
weather flow). For treatment plants capable of accepting these excess flows,
the performance of the treatment system deteriorates as the flow through
the plant increases. There are many types of dry weather plants, including
primary or physical treatment plants, secondary in the form of activated
sludge and its several modifications, and tertiary treatment plants with
biological nutrient removal. Conventional activated sludge treatment is
discussed here for simplicity, however, the effect of storm flows on other
treatment systems is similar and is governed by the same factors.
Increased storm flows result in an increased loading rate on primary
settling tanks, resulting in higher suspended solids concentrations in the
primary effluent. This results in increased loadings (BOD and SS) to the
secondary or biological system. The increased organic loading on the bio-
logical system, in conjunction with the shorter detention time in the
aeration tanks, result in a reduction in removals of BOD, If the increased
organic loading is sustained for a sufficient period of time (long duration
storms), the settling characteristics of the activated sludge may be altered
due to floe dispersion. This results in poorer settling of the sludge in
the secondary clarifiers. For shorter durations of increased flowrates, the
increased loading may be absorbed by the micro-organisms, which have a
retention time in the system of one to two days.
Increased organic loadings can also cause decreased dissolved oxygen
concentrations in the aeration tanks again resulting in lower BOD removals,
deterioration in settling characteristics, and odor. Sufficient oxygen
must be supplied to the biological organisms for good treatment performance.
The increased organic loading rates result in more biological solids
production and together with the increased primary solids from the storm
flow, result in an increase in the amount of solids requiring disposal.
Solids disposal facilities must be capable of handling these additional
loads.
The major problem associated with excess storm flows is the increased
loading rates on the secondary clarifier. The increased overflow into the
secondary clarifier results in a carryover of biomass in the effluent, a
serious problem since if sufficient biomass is washed out, the treatment ef-
fectiveness will deteriorate. Although treatment effectiveness deteriorates
during storms, some degree of treatment will be maintained if sufficient
secondary clarification capacity is provided for these excess storm flows.
5-131
-------
Since there are many parameters that effect the performance of activated
sludge and other dry weather treatment plants, no attempt has been made to
relate overall performance to flowrate. This relationship would be a specif-
ic one for each plant evaluated. Many factors should be considered in
evaluating the amount of flow a dry weather plant can accomodate. An evalua-
tion would also suggest possible modifications to the treatment facilities to
increase the capacity to handle stormwater flow. Some of the possible modi-
fications are discussed in the remainder of this section.
Flow equalization by storage would dampen the variation in hydraulic
and organic loading rates and minimize some of the problems discussed pre-
viously. Modification of the activated sludge system to a contact stabiliza-
tion process is another possibility. A portion of the existing system
could be used for a contact tank in which a good portion of the influent BOD
is adsorbed onto the biological floe. The sludge would then be settled and
sent to a stabilization tank. The overflow from the clarifier could be
chlorinated and discharged. Sufficient clarification and recycle capacity
must be provided to prevent solids carryover at peak flowrates, and to pass
the settled biomass from the clarifier to the sludge stabilization tank.
The typical operating parameters for contact stabilization should be checked
to determine if additional tank volume, recycle pumping, or additional oxygen
capacity are required. Additional units could be provided next to the dry
weather plant as done in Kenosha, Wisconsin (35).
Modification to a Step Aeration system is another possibility for in-
creasing the capacity of an existing plant. Piping arrangements can be made
to introduce the wastewater to a number of points along the aeration tank
to reduce the initial oxygen demand at the head end and make more efficient
utilization of the biomass. Equivalent performance could be obtained by
treating more flow in the same tank volume.
5.6 Cost Estimates for Treatment Alternatives
This section presents a basis for estimating the capital, operation and
maintenance costs of stormwater control systems. Most of the costs presented
are based on the work of Culp/Wesner/Culp (CWC) in a report entitled,
"Estimating Construction Costs and Operating and Maintenance Requirments for
Combined Sewer Overflow, Storage and Treatment Facilities", published by
the EPA in May 1976 (116). In the CWC report, average construction, opera-
tion and maintenance costs are presented for stormwater treatment facilities
ranging from 5 to 200 million gallons per day in capacity, and storage
facilities ranging in size from 1 to 240 million gallons. Physcial arid
chemical treatment costs are presented using June 1975 prices as a basis.
The report suggests a careful review of the methodology employed if the data
is to be used for specific project planning. The cost data are useful for
general planning and evaluation of alternatives, however they do not reduce
the need for an understanding of local conditions or a recognition of
design requirements for specific applications.
This section presents a simplified presentation of the CWC cost work
which can be used to evaluate treatment alternatives for an initial
assessment. The CWC report presents the construction cost of individual
5-132
-------
treatment processes as a function of plant size and flow capacity. To deter-
mine the total capital cost, engineering design, site preparation, legal and
fiscal adminstration, and loan interest costs during construction are added.
In the CWC report, operation and maintenance requirements are presented
in several categories: operation and maintenance labor, power, chemical
and miscellaneous supplies, administrative costs, laboratory sampling and
yardwork. To obtain the total annual operation and maintenance expenditures
presented in this section, these individual cost components were determined
and added together for each of the control methods. The operation and main-
tenance costs consist of a fixed cost and a variable cost, both of which are
a function of the plant capacity, the number of storms and the number of
hours of operation per year. The fixed cost consists of labor for supplies,
routine maintenance and repairs, yardwork, laboratory and administrative
costs. The major variable costs are the power, chemical requirements, and
labor for plant operation and cleanup.
The labor requirements in the CWC report are based on the number of
storm events per year. For this presentation, labor costs are shown as a
function of plant capacity and the number of hours of operation per year.
Labor costs are estimated as $10 per man-hour and power costs as $0.02 per
kilowatt-hour. Chemical costs are obtained from a 1975 EPA Technical Report
(117). All the costs presented are based on June 1975 prices. The use of
any cost estimating technique requires careful consideration of inflation,
which recently has been averaging about nine percent per year. In the
construction industry, the most frequently used indices are the Engineering-
News Record's (ENR) Construction Cost Index and the Building Cost Index
(118). A sewage treatment cost index was developed by the EPA to provide
a more specific index (119). Cost data must be adjusted by the EPA Sewage
Treatment Cost Index or the ENR Building Cost Index projected to the
appropriate construction period.
r * n * • * in-7c r * c 4. • * Index for construction period
Cost Estimate = 1975 Cost Estimate x , n_r • , r
June 1975 Index
June 1975 ENR Construction Index: 1306.7
June 1975 EPA Index: 246.5
5.6.1 Structural Treatment Devices
5.6.1.1 Storage Basins
Capital costs for storage basins are presented in Figure 5-56 as a
function of basin volume. Costs for earthen basins, and covered and
uncovered concrete basins are presented. Costs for earthen basins include
earthwork, liner, paving and fencing. The costs assume on-site embankment
soil, no rock excavation or groundwater problem. Provisions for mechanical
residue collection are not included in these cost estimates. The capital
costs for the concrete basins include the concrete forming, reinforcing
steel, and in the case of covered basins, the precast concrete members and
roofing material.
5-133
-------
5,000
v>
ac
< 1,000
_l
o
o
-
com
o<
o
-------
The operation and maintenace costs for storage basins are shown in
Figure 5-57 for covered concrete basins. These costs include labor at one
man-day per storm event, material and supply costs, power for automatic spray
cleaning systems, administration, yardwork and laboratory cost.
5.6.1.2 Sedimentation
Capital costs for circular sedimentation basins or settling tanks are
presented in Figure 5-58. The costs include manufactured equipment, mater-
ials, and sludge collection equipment applicable to circular shaped tanks.
The cost should be increased by 15 to 20 percent for straight line sludge
collection equipment in rectangular basins. A basin having a 12 foot side
water depth and 1.5 foot freeboard were used as a basis for the cost
estimates.
The operation and maintenance costs for sedimentation basins are pre-
sented in Figure 5-59. The costs include labor requirements for routine
visits, plant cleanup and maintenance, materials and supplies, administra-
tive, yardwork and laboratory costs.
5.6.1.3 Air Flotation
Air flotation is a unit process designed primarily on the basis of
surface area. Air flotation equipment is furnished in package units which
include the tanks and equipment. For field erected systems, the largest
practical individual unit is about 20 feet wide and 100 feet long. The
capital cost for air flotation units is presented in Figure 5-60 as a
function of surface area and design flow capacity. These costs include
manufactured equipment, construction materials, piping and equipment gallery,
recycle pumping, air retention tanks and flow measurement.
Other unit process typically associated with air flotation for storm-
water treatment include chemical feed, rapid mix, flocculation, prescreening,
chlorination, and raw wastewater and sludge pumping. Costs for these
processes should be added where applicable.
The operation and maintenance costs for air flotation without sludge
pumping are presented in Figure 5-61. These costs include labor for routine
visits twice per month at two hours each, plant cleanup at 0.004 man-hour
per square foot of tank area per storm event, two hours travel time per storm
event, an operator at the plant for four hours per event and maintenance at
0.9 man hours per square foot of tank area per storm. Administrative, yard-
work, laboratory, power, material, and supply costs are also included.
5.6.1.4 Swirl Concentrator
The capital costs for swirl concentrators ranging from 100 to 2000
square feet of surface area are presented in Figure 5-62. The corresponding
design flow for a swirl concentrator is also plotted on the horizontal scale.
The costs presented included construction of the basin itself and flow
measurement. If influent pumping is required, this cost should be added to
the capital cost of Figure 5-62.
5-.135
-------
200,000
100,000
50,000
o
o
(0
|
10,000
5,000
1,000
COVERED BASIN
IflOO
100
NO. OF HOURS OPERATION PER YEAR
I I L I I I I I I
I
I
I I I
I I I
5 10 60
STORAGE VOLUME (MILLION GALLONS)
100
200
FIGURE 5-57
ANNUAL OPERATION AND MAINTENANCE COST - STORAGE BASINS
(JUNE 1975 COSTS)
5-136
-------
10,000
5,000
CO
o
O
O
O
1,000
CO
o
o
-I 500
o_
<
o
100
ipoo
• CIRCULAR UNITS
RECTANGULAR UNITS ARE ESTIMATED TO COST IS TO 2O PERCENT MORE
I I I I I I I I I I I I I I I I I
10,000
SURFACE AREA ( ft2
100,000
I I I I 1 I
I I I I I I I I
10
50
100
RECOMMENDED DESIGN FLOW (mgd)
I III
200 300
FIGURE 5-58
CAPITAL COST - SEDIMENTATION BASINS
(JUNE 1975 COSTS)
5-137
-------
500
co
o:
o
Q
o
o
o
co
en
o
o
05
_i
z>
•z.
•z.
<
100
10
ZflOO
1,000
500
100
NO OF HOURS OPERATION PER YEAR
J | | I I I I I I
I I 1 1 I I I
1,000 5,000 10,000
50,000 100,000
SURFACE AREA ( ft2 )
I I I I I I I I I I I I I
I I I
10
50
100
200
3OO
DESIGN FLOW (mgd)
FIGURE 5-59
ANNUAL OPERATION AND MAINTENANCE COST - SEDIMENTATION BASINS
( JUNE 1975 COSTS )
5-138
-------
10,000
5,000
-------
500
(ft
cc
v_^
a
100
co
O
O 50
(0
Ci
_l
13
2
Z
<
10
WITHOUT SLUDGE PUMPING
COSTS DON'T INCLUDE SLUDGE PUMPINQ
1,000
1,000
IOO
I I I I
NO OF HOURS OPERATION PER YEAR
I I I I I I I |
500
1,000
5,000
10,000
50,000
SURFACE AREA (ft2)
I I I
I I I I I I I I
10
50
100 200
DESIGN FLOW (mgd )
FIGURE 5-61
ANNUAL OPERATION AND MAINTENANCE COST
(JUNE 1975 COSTS )
AIR FLOTATION
5-140
-------
1,000
C/5
8
o
o
o
CO
o
o
Q_
5
100
10
100
I I I I I I I I
500 1,000
SURFACE AREA (ft2 )
I I
5.00O
I I I i I I I
I I
10
50
100
DESIGN FLOW (mgd)
200
400
FIGURE 5-62
CAPITAL COST - SWIRL CONCENTRATOR
( JUNE 1975 COSTS )
5-141
-------
The annual operation and maintenance costs for a swirl concentrator are
shown in Figure 5-63. The costs are presented to include various operational
levels and include manpower for routine visits approximately once every other
week, cleanup after storms, administrative, laboratory, material and supply
costs.
5.6.1.5 High Rate Filtration
The capital costs for high-rate filtration are shown in Figure 5-64
as a function of the surface area. The cost is based on concrete gravity
filters and includes dual filter media, construction materials, housing,
underdrains, pipes and valves, and instrumentation for automatic filter
backwashing. Consideration should be given to pressure filters for opera-
tions in excess of 12 to 15 feet of headloss. Other design features should
include backwash supply storage and influent flow equalization. This could
be incorporated into storage basins preceding the filter.
The operation and maintenance costs for high rate filtration are shown
in Figure 5-65. These costs are based on labor costs for routine visits to
the facility twice per month for two hours each, plant cleanup after each
storm event based on one man-hour per filter, one hour for set-up and shut-
down, an operator at the facility for eight man-hours per storm event and
plant maintenance of 12 hours per year per filter. Costs for miscellaneous
supplies, power, yardwork, laboratory, and admininstration are also included.
5.6.1.6 Stationary Screens
A stationary (static) screen is a wedgewire screen where waste is
discharged along a sloping section causing solids to be removed and dis-
charged by gravity, while the screened wastewater flows through the screen to
a collector flume below. The capital cost for stationary screens is shown
in Figure 5-66 as a function of design flow capacity. A six-foot wide screen
having two screen faces is rated at four MGD capacity. The cost includes
manufactured equipment, construction materials, collection flumes for sludge
and screened effluent, housing, oversized open channels and metal weirs at
each screen for flow splitting.
The annual operation and maintenance costs for stationary screens at
various operational levels are presented in Figure 5-67. The operation and
maintenance costs include: labor for routine visits to the facility of 24
hours per year; labor for two hours of cleanup after each storm event, plus
one hour per screen; and costs for materials, supplies, administration, lab-
oratory and yardwork.
5.6.1.7 Horizonal Screens
Horizonal screens, often referred to as microscreens or microstrainers,
are installed in chambers designed to permit entry of wastewater to the
interior of a drum and discharge of the screened or filtered wastewater from
the exterior side of the drum. Construction costs are the same for any of
the screen aperature sizes used. The capital cost for horizonal screens is
presented in Figure 5-68. The cost includes manufactured equipment,
5-132
-------
100,000
V)
tr.
o
Q
w
o
o
5
0(5
C3
_l
^
z
z
<
10,000
1,000
t,ooo
IJOOO
soo
IOO
NO OF HOURS OPERATION PER YEAR
I I I I I I I I I
100
1,000
SURFACE AREA (ft2)
J I
5,000
10
I I 1 I I I 11
J I
50
100
200
400
DESIGN FLOW (mgd)
FIGURE 5-63
ANNUAL OPERATION AND MAINTENANCE COST-SWIRL CONCENTRATOR
(JUNE 1975 COSTS)
5-143
-------
1001
CO
(T
_
o
o
o
-i 10
<
E
o
I
I I I I I I
I
I I 1 I I
1000
lopoo
FILTER SURFACE AREA (SQUARE FEET)
100,000
FIGURE 5-64
CAPITAL COST HIGH RATE FILTRATION
CONCRETE GRAVITY FILTERS
5-144
-------
500
V)
O
Q
8
O
O
100
50
10
175
10
1,000
ipoo
50O
100
* NO. OF HOURS OPERATION PER YEAR
I I I I I I I I
870 1,735 8/575
FILTER SURFACE AREA ( FT2)
I
17,350
50 100
PLANT CAPACITY (mgd)
500
1,000
FIGURE 5-65
ANNUAL OPERATION AND MAINTENANCE COST
(JUNE 1975 COSTS)
- HIGH RATE FILTRATION
145
-------
5,000
cc
<
o
Q
O
O
O
1,000
10
o
o
50O
O
ID
a:
z
o
o
10
I I I I I I I I
I I I I I I
10
100
DESIGN CAPACITY (MGD)
1,000
FIGURE 5-66
CAPITAL COST - STATIONARY SCREEN
(JUNE 1975 COSTS)
5-146
-------
O
o
100
50
O ,0
as
-i
ID
z
z
<
NO OF HOURS OPERATION PER YEAR
I I I I I I I I
10 50 100
PLANT CAPACITY (mgd)
2,000
1,000
soo
I I I
500
FIGURE 5-67
ANNUAL OPERATION AND MAINTENANCE COST - STATIONARY SCREEN
( JUNE 1975 COST )
5-147
-------
10,000
5,000
CO
(T
O
O
O
O
O
•— 1,000
V)
O
O
_1
<
Q.
<
O
500
100
100
I I I I I I I.
500 1000
SCREEN AREA (FT2)
5,000
10,000
10
50 100
DESIGN FLOW(mgd)
500
FIGURE 5-68
CAPITAL COST-HORIZONTAL SCREENS
(JUNE 1975 COSTS)
5-148
-------
construction material and flow measurement. Concrete construction with a
center pipe gallery and housing for the pipe gallery and screen are used to
estimate the cost. Ultraviolet light slime growth control, a backwash
sprayer, and backwash storage and pumping facilities are also included in
the cost estimate.
The operation and maintenance costs for horizonal.screens are shown in
Figure 5-69. These costs are a function of plant capacity and the number
of hours of operation per year. Included are labor for routine visits to the
facility twice per month for two hours each, plant cleanup after storms, an
operator present at the facility eight hours per day for each event, and
maintenance of 24 man-hours per year. Administration, yardwork, laboratory,
power, material and supply costs are also included.
5.6.1.8 Chemical Coagulation
The use of chemicals in combined sewer overflow treatment facilities
has been employed in conjunction with air flotation, high rate filtration
and sedimentation. To estimate the capital cost of chemical treatment, the
cost of chemical feed systems should be added to the cost of the unit treat-
ment process which is being utilized. A rapid mix tank should also be in-
cluded in a chemical coagulation system and the appropriate cost should be
added. Capital costs for chemical feed systems are presented in Figure
5-70. These costs include feeder and storage equipment, housing, electrical
and instrumentation costs and miscellaneous items. Costs are shown for lime,
ferric chloride, alum and polyelectrolyte feed systems.
The operation and maintenance costs for chemical feed systems are shown
in Figures 5-71 to 5-74. These costs include chemicals, labor for the un-
loading of chemicals, operation of chemical feeding equipment, power,
materials and supplies. The following chemical costs are used: $25/ton
lime, $100/ton alum, $80/ton ferric chloride, and $2/lb polymer.
5.6.1.9 Chlorination Feed Equipment
The capital cost for chlorination feed equipment is presented in Figure
5-75 as a. function of the feed capacity. This cost includes manufactured
equipment, construction materials, housing, distribution panels, a standby
chlorinator, monorail trolley and hoist equipment. These costs are based on
the use of gaseous chlorine feed equipment and do not include a mixing or
contact basin.
The operation and maintenance costs for chlorination are shown in Figure
5-76 as a function of chlorine usage. These costs include power, chlorine,
labor, materials and supplies. Generation and feed costs (per pound of
active disinfectant) are approximately:
(a) for chlorine dioxide - 2 times costs for chlorine gas
(b) for hypochlorite - 10 times costs for chlorine gas.
5-1 49
-------
100,000
50,000
o
o
o
O 10,000
CD
o
spoo
1000
100
900
-*--NO. OF HOURS OPERATION PER YEAR
I I I 1 I I I
500 1000
SCREEN AREA (FT2)
5000
10
50 100
DESIGN FLOW (mgd)
500
FIGURE 5-69
ANNUAL OPERATION AND MAINTENANCE COST
HORIZONTAL SCREENS
(JUNE 1975 COSTS)
5-150
-------
2,000
1,000
CO
oc.
<
o
Q
8
O
O
<
5Z
o
100
20
30
0.5
POLYMER FEED RATE (LBS/HR)
5 10
T T i i i ri
LIME
FERRIC CHLORIDE
POLYMER
I I I I I I
I I I I I I I I
100
ipoo
FEED RATE ( LBS/HR
50
I I I
i i i
5,000
FIGURE 5-70
CAPITAL COST — CHEMICAL FEED SYSTEMS
( JUNE 1975 COSTS )
5-151
-------
500,000
100,000
CO
-------
500,000
co
rr.
o
Q
CO
O
O
OB
0'
_l
:D
z
z
<
100,000
10,000
1,000
f.OOO '
I I I I I
NO OF HOURS OPERATION PER YEAR
I I I I I I I I
50
100
1,000
FEED CAPACITY ( LBS/HR)
5,000
10,000
FIGURE 5-72
ANNUAL OPERATION AND MAINTENANCE COST -ALUM FEED
( JUNE 1975 COSTS)
5-153
-------
500,000
100,000
If)
tr
o
Q
V)
O
O
CD
d
< 10,000
z
1,000
2,000
1,000
I I I I I I I
50 100 1,000
FEED CAPACITY (LBS/HR)
10,000
FIGURE 5-73
ANNUAL OPERATION AND MAINTENANCE COST - FERRIC CHLORIDE FEED
( JUNE 1975 COSTS )
5-154
-------
100,000
50,000
^ I 0,000
CO
OL
5,000
CO
o
o
00
o
1,000
500
100
0.3
10OO
*-NO OF HOURS OPERATION PER YEAR
I I I I I I I I I |
1.0 50
FEED CAPAC!TY(LBS/HR)
10.0
30.0
FIGURE 5-74
ANNUAL OPERATION AND MAINTENANCE COST
POLYMER FEED
5-155
-------
500,000
CE
o
a
^ I 00,000
w
o
o
o
£ 50,000
o
z
o
o
I 0,000
100
500 1000 5000
CHLORINE FEED CAPACITY (LBSXDAY)
10,000
FIGURE 5-75
CAPITAL COST-CHLORINE FEED EQUIPMENT
(JUNE 1975 COSTS)
5-156
-------
500,000
co
or
o
a
,_ 100000
CO
o
o
2
05 50,000
O
z
z
<
10,000
10
50 100
CHLORINE USAGE (T/YR)
500
1000
FIGURE 5-76
ANNUAL OPERATION AND MAINTENANCE COST
CHLORINE FEED EQUIPMENT
(JUNE 1975 COST)
5-157
-------
5.6.1.10 High Intensity Mixing/Chlorine Contact Basin
For application of chlorination facilities to the treatment of storm-
water, consideration should be given to detention times shorter than the
conventional 15 minutes used in sewage treatment. Utilizing high mixing
intensities will improve chlorine utilization and effectiveness of treatment
at lower detention times. For high intensity mixing, a rapid mix basin
similar to that used in water treatment for chemical coagulation can be used.
The capital cost of a rapid mix basin for chlorination and chemical flash
mixing is presented in Figure 5-77 as a function of volume. This cost
includes manufactured equipment, a concrete basin and stainless steel mixers.
The operation and maintenance costs for rapid mix basins are presented
in Figure 5-78. These costs include labor at one man-day per storm event,
material and supply costs, and power for the mixer.
5.6.1.11 Raw Wastewater Pumping
The capital cost of a raw wastewater pumping system is presented in
Figure 5-79 and is based on a facility which includes rough screening, dry
well arrangement, housing for the pumps, piping, electrical control equip-
ment, auxiliary power and surface access.
The operation and maintenance costs for raw wastewater pumping are pre-
sented in Figure 5-80. These costs include labor for clean-up after storms
at eight man-hours per storm event, 24 hours per year inspection and equip-
ment checking, materials, supplies and power. Power costs are based on an
average flow of 45 percent design capacity, 35 feet of head and 65 percent
efficiency.
5.6.1.12 Sludge Pumping
Sludge pumping stations are required in specific stormwater treatment
facilities where gravity return to the dry weather sewer is not possible.
The capital cost for sludge pumping is presented in Figure 5-81. This cost
is based on positive displacement pumps appropriate for high solids concen-
tration and intermittent use. The facility cost is based on underground
installation and piping constructed in conjunction with the unit process
for solids separation. A superstructure is included to provide access to
the station from the ground level and to house electrical control equipment.
The operation and maintenance costs for sludge pumping are presented in
Figure 5-82. These costs include labor, materials, supplies, and power
costs based on a pumping head of 25 feet.
5.6.1.13 Biological Treatment
It is unlikely that biological treatment of stormwater overflows will
be found to be cost effective in many applications. While biological
treatment can be applied to the treatment of stormwater, the variability of
flow and waste characteristics make this method of treatment expensive
and difficult to operate. Also the bacterial population must be sustained
5-1 58
-------
500
(rt
OL
o
Q
O
o
100
O
50
Q.
<
O
10
I
NOTE
DESIGN FLOW (¥00) BASED
ON 2 MINUTES DETENTION
I
10 50
DESIGN FLOW (mgd)
100
300
1
I
I
4,200 10,500 37.5OO 140,000
BASIN VOLUME ( GAL)
4(^000
FIGURE 5-77
CAPITAL COST - RAPID MIX BASIN
(JUNE 1975 COSTS )
5-159
-------
10,000
5,000
CO
en
O
O
cC
C)
_J
z
z
<
1,000
500
IOO
2,000
1,000
I \t\\\\
NO OF HOURS OPERATION PER YEAR
I I I I I I I I 1 I
10 50 100
DESIGN FLOW (mgd)
300
FIGURE 5-78
ANNUAL OPERATION AND MAINTENANCE COST-RAPID MIX BASIN
( JUNE 1975 COSTS )
5-160
-------
20,000
10,000
CO
IT
O
o
O
o
o
CO
o
o
<
H
Q.
O
1,000
200
I I I I I I
5 10
I I I I I I
50
100
300
PUMP STATION CAPACITY (mgd)
FIGURE 5-79
CAPITAL COST — RAW WASTE WATER PUMPING
( JUNE 1975 COSTS )
5-161
-------
700
500
CO 100
co
O
o
50
10
IOO
too
I I I I I I I
I
NO OF HOURS OPERATION PER YEAR
I I I I I I I I I
10 50 100
PUMP STATION CAPACITY (mgd)
300
FIGURE 5-80
ANNUAL OPERATION AND MAINTENANCE COST
RAW WASTE WATER PUMPING
( JUNE 1975 COSTS )
5-162
-------
1000
500
O
Q
O
O
O
V)
O
O
100
Q.
<
O
50
30
30
100 500 1000
SLUDGE PUMPING CAPACITY (GPM)
5000
FIGURE 5-81
CAPITAL COST — SLUDGE PUMPING
(JUNE 1975 COSTS)
5-163
-------
30,000
10,000
CO
cc
O 5,000
CO
O
o
5
-------
during dry weather periods so that it will be available for the intermittent
stormwater loads.
The CWC report did not provide cost information for biological treat-
ment systems and more general guidelines for estimating these costs will be
presented.
The economics of biological methods for the treatment of stormwater
should be evaluated on a case by case basis. Figure 5-83 provides a general
guide for estimating the cost of biological treatment. Capital costs related
to plant capacity are based on the activated sludge process. This curve is
based on information taken from Water and Wastewater Engineering (113) and
provides a rough estimate for building a new biological treatment plant
which includes primary and secondary clarification aeration, chlorination,
and sludge disposal. The costs of adding individual unit processes to a
dry weather treatment plant would involve a more detailed study on a case by
case basis.
A rough estimate of operation and maintenance costs for biological
treatment range from five to 12 cents per 1,000 gallon treated for a trick-
ling filter and eight to 21 cents per 1,000 gallons treated by activated
sludge. These general costs could be associated with the total volume of
stormwater treated annually to roughly estimate the annual operation and
maintenance costs.
5.6.1.14 Example Cost Calculation
To illustrate the use of the cost estimating procedure, assume the
analysis of treatment device performances has indicated that the following
combination of facilities would be beneficial:
1. Wastewater pumping with capacity of 50 MGD (77 cfs),
2. Horizonal screen with a maximum flow accepted of 50 MGD, and
3. Chlorination Facilities: a. Rapid mix tank with two minute
detention time and maximum flow accepted of 50 MGD; b. Chlorine
feed equipment with a feed rate proportional to flow. Average
feed rate during storms of 2,085 Ib/day chlorine.
Assume there is an average of 500 hours of rain (and runoff) per year. The
capital costs as of June 1975 are shown in Table 5-15. The total capital
cost estimate includes land, site work, engineering design, legal, fiscal
and administrative costs and interest during construction. The operation
and maintenance costs are also estimated in Table 5-15. Results of the
sample calculation show that the capital costs for the treatment system are
about $4,274,000 and the annual operation and maintenance costs are
$128,400. These are for June 1975 prices and should be adjusted for known or
predicted inflation. The comparison to other treatment processes may then be
based on either the total present worth or the equivalent annual cost. A
thorough comparison requires some knowledge of the principles of engineering
economics.
5-165
-------
50
CO
CC
o
a
z
o
10
CO
o
o
<
H
0.
o
0.5
FROM WATER AND WASTE ENGINEERING
MARCH 1974
10
50 10 0
PLANT SIZE (MGD)
50.0
FIGURE 5-83
CAPITAL COST - SECONDARY BIOLOGICAL TREATMENT
(JUNE 1975 COSTS)
5-166
-------
TABLE 5-15
EXAMPLE CALCULATION OF STORMWATER TREATMENT COST
(June 1975 Costs)
CAPITAL COST:
Process Unit
Wastewater Pumping
Hori zonal Screen
Rapid Mix Tank
Chlorine Feed Equipment
ANNUAL OPERATION AND
MAINTENANCE COSTS:
Process Unit
Wastewater Pumping
Hori zonal Screen
Rapid Mix Tank
Chlorine Feed Equipment
Unit
Parameter
50 MGD
50 MGD
50 MGD
2,085 Ib/day
Unit
Parameter
50 MGD
50 MGD
50 MGD
C70,000 Gal)
22 tons/year
Reference
Figure
5-79
5-68
5-77
5-75
Total
Reference
Figure
5-80
5-69
5-78
5-76
Total
Cost
Estimate
(dollars)
3,400,000
720,000
54,000
100,000
$4,274,000
Cost
Estimate
(dollars)
84,000
15,500
900
28,000
$ 128,400
5-167
-------
5.6.2 Management Practices
A number of numerical estimates for the cost of implementing stormwater
management practices have been developed. Heaney and Nix present a summary
of unit costs for street sweeping (120), reproduced as Table 5-16. Note the
wide variation in the cost per curb mile from one location to another.
Koplan (121) estimates the capital cost required for a street sweeping
program to be in the order of $10 to $15/curb mile cleaned per year, for the
sweepers alone (ENR 2000). The cost of various methods of catch basin clean
ing are estimated on the order of $3 to $4 per catch basin (122).
Costs for combined sewer flushing are reviewed by Heaney and Nix (120)
based on a study in the Boston area. The annual cost of flushing per unit
cost of sewer line is calculated as $ll,78/ft or $62,200/mile.
The cost of other management practices will vary depending upon the
particular application, materials used, frequency of maintenance, etc.
Wanielista (123) estimates the construction cost of small ponds in Florida
at about $1.00/cu. yd. and the cost of swales (drainage channels with
vegetative stabilization) at about $1.00 per foot of 4 foot wide swale.
Local factors should be considered for particular studies.
5.7 References
1. Heane, James P., et al., Nationwide Evaluation of Combined Sewer
Overflows and Urban Stormwater Discharges, Volume II: Cost Assessment
and Impacts, University of Florida for U.S. Environmental Protection
Agency, EPA-600/2-77-064, March 1977.
2. Howard, Charles D.D., Theory of Storage and Treatment Plant Overflows,
Journal of the Environmental Engineering Division, ASCE Proc. Paper
12310, August 1976.
3. Thorn, H.C.S., A Frequency Distribution for Precipitation, Abstract,
Bulletin American Meteorology Society, Vol. 32, No. 10, December
1951.
4. Thorn, H.C.S., A Note on the Gamma Distribution, Monthly Weather
Review, Vol. 86, No. 4, April 1958.
5. Chow, Ven Te, Ben Chie Yen, Urban Stormwater Runoff: Determination
of Volumes and Flowrates, University of Illinois for U.S. Environmental
Protection Agency, EPA-600/2-76-116, Cincinnati, Ohio, May 1976.
6. Yen, Ben Chie, Discussion of Theory of Storage and Treatment Plant
Overflows, Journal of the Environmental Engineering Division, ASCE,
Vol. 103, No. EE3, June 1977.
7. Eagleson, Peter S., Dynamic Hydrology, McGraw-Hill Book Company,
New York, 1970.
5-1 68
-------
TABLE 5-16
UNIT COSTS OF STREET SWEEPING
Street Sweeping Costs
Mean
Median
$/ Curb -mile
($ /curb- km)
86.61
(53.82)
7.00
( 4.35)
($/m3)
22.14
(28.96)
13.79
(18.04)
$/ton
($/metric ton)
31.31
(34.51)
14.28
(15.74)
$/capita/year
1.54
1.23
Source: Unpublished data from American Public Works Association, 1976.
5-169
-------
8. Smith, Robert, Use of the Gamma Distribution Function for Fitting
Rainfall Data, Memorandum to Record, U.S. Environmental Protection
Agency, Systems and Economic Analysis Section, Cincinnati, Ohio,
March 24, 1977.
9. Rodriguez-Iturbe, Mejia, On the Transformation of Point Rainfall to
Areal Rainfall, Water Resources Research, Vol. 10, No. 4, August
1974.
10. Linsley, R.K., M.A. Kohler, J.L.H. Paulhus, Hydrology for Engineers,
McGraw-Hill, New York, 1958.
11. Miller, Clayton R. and Warren Viessman, Jr., Runoff Volumes from
Small Urban Watersheds, Water Resources Research, Vol. 8, No. 2,
April 1972.
12. Lager, John A., Theodor Didriksson, George B. Otte, Development and
Application of a Simplified Stormwater Management Model, Metcalf
and Eddy, Inc., for U.S. Environmental Protection Agency, EPA-
600/2-76-218, Cincinnati, Ohio, 1976.
13. Nonpoint Sources, An Assessment of Pollutant Loadings to Lakes and
Rivers in North Central Texas, Hydroscience, Inc. for North Central
Texas Council of Governments, Arlington, Texas, 1977.
14. Feuerstein, Donald L. and William 0. Maddaus, Wastewater Management
Program, Jamaica Bay, New York, Volume I, Summary Report, H. F.
Ludwig and Associates and Engineering-Science Inc. for U.S. Environ-
mental Protection Agency, EPA-600/2-76-222a, September 1976.
15. Final Report - lear 1, Spring Creek Auxiliary Water Pollution
Control, H.F. Ludwig and Associates and Engineering-Science, Inc.
for City of New York, May 1970.
16. Betson, Roger, Urban Hydrology, A Systems Study in Knoxville,
Tennessee, Tennessee Valley Authority Division of Water Management,
Knoxville, Tennessee, June 1976.
17. Gameson, A.L.H., R.N. Davidson, J.M. Threlfall, Storm Flows From
Combined Sewerage Systems In Three Areas, Journal of the Institution
of Public Health Engineers, London, July 1965.
18. Kluesener, John W. and G. Fred Lee, Nutrient Loading From a Separate
Storm Sewer in Madison, Wisconsin, Journal of the Water Pollution
Control Federation, Vol. 46, No. 5, May 1974.
19. Storm Water Pollution from Urban Land Activity, AVCO Economic
Systems Corporation for FWQA, U.S. Department of Interior, Report
11034FKL07/70, July 1970.
5-170
-------
20. Colston, Newton V., Characterization and Treatment of Urban Land
Runoff, North Carolina State University for U.S. Environmental
Protection Agency, EPA-670/2-74-046, December 1974.
21. Donigian, A.S. Jr. and Norman H. Crawford, Modeling Non-Point
Pollution from the Land Surface, U.S. Environmental Protection
Agency, 1976.
22. Engineering Investigation of Sewer Overflow Problems, Roanoke,
Virginia, Hayes, Seay, Mattern and Mattern, U.S. Environmental
Protection Agency, Report 11024DMS05/70.
23. Ragan, R.M. and T.J. Jackson, Use of Satellite Data in Urban Hydrologic
Models, Journal of the Hydraulics Division, ASCE, Vol. 101, No.
HY12, December 1975.
24. Stankowski, S.J., Magnitude and Frequency of Floods In New Jersey
with Effects of. Urbanization, Special Report 38, US Geological
Survey, Water Resources Division, Trenton, New Jersey, 1974.
25. Reiner, Franzini, Urbanizations Drainage Consequence, Journal of
the Urban Planning and Development Division, American Society of
Civil Engineers, Vol. 1, December 1971.
26. Graham, Costello, Mallon, Estimation of Imperviousness and Specific
Curb Length for Forecasting Stormwater Quality and Quantity, Journal
Water Pollution Control Federation, Vol. 46, No. 4, April 1974.
27. American Public Works Association, Nationwide Evaluation of Combined
Sewer Overflows and Urban Stormwater Discharges, Volume III:
Characterization, US Environmental Protection Agency Publication,
EPA-600/2-77-064, March 1977.
28. American Public Works Association and University of Florida, Evaluation
of the Magnitude and Significance of Pollution Loading From Urban
Stormwater Runoff, Ontario, Department of Environment, Toronto,
1976.
29. Heaney, Huber, Nix (University of Florida), Storm Water Management
Model, Level I, Preliminary Screening Procedures, US Environmental
Protection Agency Publication, EPA-600/2-76-75, Cincinnati, Ohio,
October 1976.
30. Brater, E.F., J.D. Sherrill, Rainfall-Runoff Relations on Urban and
Rural Areas, University of Michigan for U.S. Environmental Protection
Agency, EPA-670/2-75-046, May 1975.
31. Whitehead, H.C., and J.H. Feth, Chemical Composition of Rain, Dry
Fallout, and Bulk Precipitation at Menlo Park, California, 1957-
1959, Journal of Geophysical Research, Vol. 69, No. 16, August
1964.
5-171
-------
32. Pearson, F.J. and D.W. Fisher, Chemical Composition of Atmospheric
Precipitation in the Northeastern United States, U.S. Geological
Survey, WSP 1535-P, 1971.
33. Cogbill, C.V. and G.E. Likens, Acid Precipitation in the Northeastern
United States, Water Resources Research, V. 10, No. 6, December
1974.
34. Ruppert, H., Geochemical Investigations on Atmospheric Precipitation
in a Medium-Sized City, Water, Air and Soil Pollution (4) 1975.
35. Lager, J.A. and W. E. Smith, Urban Stormwater Management and Technology^
An Assessment, Metcalf and Eddy, Inc. for U.S. Environmental Protection
Agency, EPA-670/2-74-040, December 1974.
36. Lager, John A., W.G. Smith, W.G. Lynard, R.M. Finn and E.J. Finnemore,
Urban Stormwater Management and Technology: Update and Users
Guide, Metcalf and Eddy, Inc. for U.S. Environmental Protection
Agency, EPA-600/8-77-014, September 1977.
37. Huber, Wayne C., James P. Heany, Urban Rainfall-Runoff-Quality Data
Base, University of Florida for the U.S. Environmental Protection
Agency, EPA-600/8-77-009, July 1977.
38. Texas Tech University, Water Resources Center, Variation of Urban
Runoff Quality With Duration and Intensity of Storms, WRC-73-2,
1973.
39. Ewnboldt Avenue Pollution Abatement Demonstration Project, Consoer,
Townsend and Associates and Hydroscience, Inc., for the City of
Milwaukee and the U.S. Environmental Protection Agency, September
1974.
40. Report Upon Overflow Analysis to Passaic Valley Sewerage Commissioners,
Passaic River Overflows, Elson T. Killam Associates, Inc., 1976.
41. Sartor, James D., and Gail B. Boyd, Water Pollution Aspects of
Street Surface Contaminants, U.S. Environmental Protection Agency,
EPA-R2-72-081, November, 1972.
42. URS Research Company, Water Quality Management Planning for Urban
Runoff, U.S. Environmental Protection Agency, EPA-440/9-75-004,
December, 1974.
43. Vilaret, M.R., et al, Storm and Combined Sewer Pollution Sources
and Abatement, Black, Crow, and Eidsness, Inc., NTIS, PB 201 725,
Gainesville, Florida, January 1971.
44. Wells, Dan M. et al, Variation of Urban Runoff Quality with Duration
and Intensity of Storms-Phase III, Volume 4 - Project Summary, OWRT
and The City of Lubbock, Project No. B-177-TEX, December 1975.
5-172
-------
45. Haith, Douglas A., Land Use and Water Quality in New York Rivers,
ASCE, Journal of Environmental Engineering Division, Vol. 102, No.
EE1, February 1976.
46. Darby, William P., Francis Clay McMichael, Robert W. Dunlop, Urban
Watershed Management Using Activity Indicators to Predict Water
Quality, Water Resources Research, Vol. 12, No. 2, April 1976.
47. Gerghty, Miller, Van Der Leeden, Troise, Water Atlas of the United
States, A Water Information Center Publication, 1973.
48. Simplified Mathematical Modeling of Water Quality, Hydroscience,
Inc., and Mitre Corporation, Environmental Protection Agency Publica-
tion, March 1971.
49. Fischer, Hugo B., Dispersion Predictions in Natural Streams, Journal
of the Sanitary Engineering Division, ASCE, Vol. 94, No, SA5,
October 1968.
50. Bansal, Mahendra K., Dispersion in Natural Streams, Journal of the
Hydraulics Division, ASCE, Vol. 97, No. HY11, November 1971.
51. McQuivey, R.S. and T.N. Keefer, Simple Method for Predicting Dispersion
in Streams, Journal of the Environmental Engineering Division,
ASCE, Vol. 100, No. EE4, August 1974.
52. Lice, Henry, Predicting Dispersion Coefficient of Streams, Journal
of the Environmental Engineering Division, ASCE, Vol. 103, No. EE1,
February 1977.
53. Mancini, John, Numerical Estimates of Coliform Mortality Rates
Under Various Conditions, Research and Development Report, Hydro-
science, Inc., May 1977.
54. Mathematical Modeling of Natural Systems, Summer Institute Notes,
Environmental Engineering and Science Division, Manhattan College,
Bronx, N.Y., May 1976.
55. O'Connor, D.J. and W.E. Dobbins, Mechanism of Reaeration in Natural
Streams, Transactions, ASCE, Vol. 123, Paper No. 2934, 1958.
56. Churchill, M.A., H.L. Elmore and R.A. Buckingham, Prediction of
Stream Reaeration Rates, Journal of the Sanitary Engineering Division,
ASCE, Vol. 88, No. SA4, July 1962.
57. Isaacs, W.P. and A.F. Gandry Jr., Atmospheric Oxygenation in a
Simulated Stream, Journal of the Sanitary Engineering Division,
ASCE, Vol. 94, No. SA2, April 1968.
58. Steele, Timothy D. and Marshall E. Jennings, Regional Analysis of
Streamflow Chemical Quality in Texas, Water Resource Research, Vol.
8, No. 2, April 1972.
5-173
-------
59. McElroy, Chiu, Nebgen, Aleti, Bennet, Load-ing Functions for Assessment
of Water Pollution From Non-point Sources, Midwest Research Institute
for U.S. Environmental Protection Agency, EPA-600/2-76-151, May
1976.
60. Influence of Land Use on Stream Nutrient Levels, U.S. Environmental
Protection Agency, EPA-600/3-76-014.
61. Betson, R.P. and W.M. McMaster, A First Generation Nonpoint Source
Mineral Water-Quality Model, Journal of the Water Pollution Control
Federation, Vol. 47, No. 10, October 1975.
62. Hydroscience, Inc., 208 Areawide Assessment Procedures Manual,
Volume I, Chapter 2, U.S. Environmental Protection Agency, EPA-
600/9-76-014, Cincinnati, Ohio, July 1976.
63. Analysis of Upstream Background Loadings, Internal Memorandum,
Hydroscience, Inc., 1976.
64. Ledbetter, J.E. and E.F. Gloyna, Productive Techniques for Water
Quality Inorganics, Journal of the Sanitary Engineering Division,
ASCE, Vol. 90, No. SA-1, February 1964.
65. Gunnerson, C.G., Streamflow and Quality in the Columbia River
Basin, Journal of the Sanitary Engineering Division, ASCE, Vol. 93,
No. SA6, 1967.
66. Hall, F.R., Dissolved Solids - Discharge Relationships, 1, Mixing
Models, Water Resources Research, Vol. 6, No. 3, 1970.
67. Hall, F.R., Dissolved Solids - Discharge Relationships, 2, Applications
to Field Data, Water Resources Research, Vol. 7, No. 3, 1971.
68. Hem, John D., Study and Interpretation of the Chemical Characteristics
of Natural Water, U.S. Geological Survey Water-Supply Paper 1473,
1959.
69. Urban Storm Runoff and Combined Sewer Overflow Pollution, Sacramento,
California, Envirogenics Company for U.S. Environmental Protection
Agency, Report 11024 FKM 12/71, December 1971.
70. Wang, Wun-Cheng and Ralph L. Evans, Dynamics of Nutrient Concentrations
in the Illinois River, Journal of the Water Pollution Control
Federation, Vol. 42, No. 12, December 1970.
71. National Assessment of Trends in Water Quality, Enviro Control,
Inc., National Technical Information Service, June 1972.
72. Cahill, T.H., P. Imperato, and Francis H. Verhoff, Evaluation of
Phosphorus Dynamics in a Watershed, Journal of the Environmental
Engineering Division, ASCE, Vol. 100, No. EE2, April 1974.
5-174
-------
73. Assessment of the Effects of Nutrient Load-ings on Lake Ontario
Using a Mathematical Model of the Phytoplankton, Hydroscience, Inc.
for the International Joint Commission, Great Lakes Water Quality
Board, March 1976.
74. Metcalf and Eddy, University of Florida and Water Resources Engineers,
Storm Water Management Model (SWMM) - Volume I Final Report, Environ-
mental Protection Agency, 11024 DOC 07/71, July 1971.
75. University of Florida, Urban Stormwater Management Modeling and
Decisiormdking , prepared for National Environmental Research Center,
May 1975.
76. Glover, G.E., and G.R. Herbert, Microstaining and Disinfection of
Combined Sewer Overflows - Phase II, Environmental Protection
Agency, EPA-R2-73-124, January 1973.
77. Glover, G.R., Application of Micro straining to Combined Sewer
Overflow, contained in Combined Sewer Overflow Seminar Papers,
Environmental Protection Agency, EPA-670/2-73-077, November 1973.
78. Ecology Division, Rex Chainbelt, Inc., Milwaukee, Wisconsin, Screening/
Flotation of Combined Sewer Overflows, Environmental Protection
Agency, 11020 FDC 01/72, January 1972.
79. Eckenfelder, W.W., and D.J. O'Connor, Biological Waste Treatment,
Pergamon Press, 1961.
80. American Society of Civil Engineers, Sewage Treatment Plant Design,
ASCE Manual No. 36, 1959.
81. Envirogenics Company, Urban Storm Runoff and Combined Sewer Over-
flow Pollution, Environmental Protection Agency, 11024 FKM 12/71,
December 1971.
82. Gupta, M.K., and R.W. Agnew, Screening/Dissolved Air Flotation
Treatment of Combined Sewer Overflows, contained in Combined Sewer
Overflow Seminar Papers, Environmental Protection Agency EPA-670/2-
73-077, November 1973.
83. White, Robert L., and T.G. Cole, Dissolved Air Flotation for
Combined Sewer Overflows, Public Works, February 1973.
84. Rhodes Technology Corporation, Houston, Texas, Dissolved-Air Treatment
of Combined Sewer Overflows, Environmental Protection Agency, 11020
FKI, January 1970.
85. American Public Works Association, The Swirl Concentrator as a
Combined Sewer Overflow Regulator Facility, Environmental Protection
Agency EPA-R2-72-008, September 1972.
-------
86. Sullivan, Richard H., et al, The Swirl Concentrator as a Grit Separator
Device, Environmental Protection Agency, EPA-670/2-74-026, June
1974.
87. Sullivan, Richard H., The Swirl Concentrator as a Combined Sewer
Overflow Regulator, contained in Combined Sewer Overflow Seminar
Papers, EPA-670/2-73-077, November 1973.
88. Dalrymple, R.J., S.L. Hodd, D.C. Morin, Physical and Settling
Characteristics of Particulates in Storm and Sanitary Wastewater,
Beak Consultants Limited for U.S. Environmental Protection Agency,
EPA-670/2-75-011, April 1975.
89. Field, R.I., and P.E. Moffa, Treatdbility Determinations for a
Prototype Swirl Combined Sewer Overflow Regulator/Solids-Separator,
paper for Environmental Protection Agency, 1975.
90. Nebolsine, R., P.J. Harvey and C.Y. Fan, High-Pate Filtration of
Combined Sewer Overflows, Hydrotechnic Corporation for U.S. En-
vironmental Protection Agency 11023 EYI 04/72, April 1972.
91. Harvey, P.J., High-Rate Multi-Media Filtration, contained in Combined
Sewer Overflow Seminar Papers, EPA-670/2-73-077, November 1973.
92. Glover, G.E., High Rate Disinfection of Combined Sewer Overflow,
contained in Combined Sewer Overflow Seminar Papers, EPA-670/2-73-
077, November 1973.
93. Shuckrow, A.J., G.W. Dawson and W.F. Bonner, Physical-Chemical
Treatment of Combined and Municipal Sewage, Environmental Protection
Agency, EPA-R2-73-149, February 1973.
94. Black § Veatch, Consulting Engineers, Process Design Manual for
Phosphorus Removal, Environmental Protection Agency, October 1974.
95. Process Design Manual for Upgrading Existing Wastewater Treatment
Plants, Environmental Protection Agency, October 1974.
96. Pitt, R.E., and G. Amy, Toxic Materials Analysis of Street Surface
Contaminants, Environmental Protection Agency, EPA-R2-73-283,
August 1973.
97. Argo, D.G., and G.L. Gulp, Heavy Metals Removal in Wastewater
Treatment Processes: Part I and Heavy Metals Removal in Wastewater
Treatment Process: Part II - Pilot Plant Operation, Water and
Sewage Works, August-September 1972.
98. Hommack, P., K.L. Zippier and E.G. Herbert, Utilization of
Trickling 'Filters for Dual Treatment of Dry and Wjt Weather Flows,
Environmental Protection Agency, EPA-670/2/73-071, September 1973.
5-176
-------
99. Autotrol Corporation, Milwaukee, Wisconsin, Combined Sewer Overflow
Treatment by The Rotating Biological Contactor Process, Environmental
Protection Agency, EPA-670/2-24-050, June 1974.
100. Process Design Manual for Suspended Solids Removal, Environmental
Protection Agency, January 1975.
101. Schmid, L.A., and R.R. McKinney, Phosphate Removal by a Lime-
Biologioal Treatment Scheme, JWPCF, July 1969.
102. Gulp, R.L., and G.L. Gulp, Advanced Wastewater Treatment, Van
Nostrand Reinhold Company, 1971.
103. Fckenfelder, W.W., Water Quality Engineering for Practicing Engineers,
Barnes $ Noble, Inc., 1970.
104. Advanced Waste Treatment Design for Wantagh, Long Island Groundwater
Recharge Study, Consoer, Townsend £ Associates, Chicago, Illinois,
June 1973.
105. Beak Consultants Limited, Physical and Settling Characteristics of
Particulates in Storm and Sanitary Wastewater, Environmental Protection
Agency, EPA-670-2-75-011, April 1975.
106. O'Brien and Gere Engineers, Pilot Plant Studies, Combined Sewer
Overflow Abatement Programs, EPA Grant New York 005141, Draft
Report, November 1976.
107. Prah, D.H., P.L. Brunner, Combined Sewer Stormwater Overflow
Treatment By Screening and Terminal Ponding at Fort Wayne, Indiana,
Municipal Environmental Research Laboratory, Office of Research and
Development, U.S. Environmental Protection, Draft Report, Volume I,
June 1976.
108. Maher, M.B., Micro straining and Disinfection of Combined Sewer
Overflow - Phase III, U.S. Environmental Protection Agency, EPA-
670/2-74-049, 1974.
109. O'Brien and Gere Engineers, Inc., Syracuse CSO Demonstration Study,
Draft Report, Sections III-VIII, March 15, 1977.
110. Field, Richard, et al., Proceedings of Workshop on Microorganisms
in Urban Stormwater, U.S. Environmental Protection Agency, EPA-
600/2-76-244, November 1976.
111. Collins, Harvey F., R.E. Selleck, G.C. White, Problems in Obtaining
Adequate Sewage Disinfection, Journal of the Sanitary Engineering
Division, ASCE, SA5, October 1971.
112. Chick, H., An Investigation of the Laws of Disinfection, Journal of
Hygiene, Vol. 8, Cambridge, England, No. 92, 1908.
5-1 77
-------
113. Fair, Gordon M., J.C. Geyer, D.A. Okun, Water and Wastewater Engineering,
Volume I and 2, John Wiley and Sons, Inc., New York, 1968.
114. Geisser, Donald F., and S.R. Carver, High-Rate Disinfection:
Chlorine Versus Chlorine Dioxide, Journal of the Environmental
Engineering Division, ASCE, Vol. 103, EE6, December 1977.
115. Pontius, Uwe R., E.H. Pavia, D,G. Crowder, Hypochlorination of
Polluted Stormwater Pumpage at New Orleans, U.S. Environmental
Protection Agency, EPA-670/2-73-067, September 1973,
116. Benjes, H.H., Cost Estimating Manual-Combined Sewer Overflow Storage
Treatment, Gulp, Weaner, and Gulp, Inc. for U.S. Environmental
Protection Agency, EPA-600/2-76-286, 1976.
117. A Guide to the Selection of Cost-Effective Wastewater Treatment
Systems, U.S. Environmental Protection Agency, EPA-730/9-75-002,
July 1975.
118. Engineering News-Record, New York, Monthly.
119. Construction Cost Indexes, Office of Water Program Operations, U.S.
Environmental Protection Agency, Quarterly.
120. Heaney, James P., Stephen J. Nix, Storm Water Management Model:
Level I -Comparative Evaluation of Storage Treatment and Other
Management Practices, U.S. Environmental Protection Agency, EPA-
600/2-77-083, April 1977.
121. Koplan, Michael, Preventive Approaches to Stormwater Management,
Chapter 3, U.S. Environmental Protection Agency, EPA-440/9-77-001,
January 1977.
122. American Public Works Association, Water Pollution Aspects of Urban
Runoff, Federal Water Pollution Control Administration, WP-20-15,
1969.
123. Wanielista, Martin P., Nonpoint Source Effects, Florida Technological
University, Orlando, Florida, January 1976.
5-178
-------
CHAPTER 6
MONITORING FOR STORMWATER ASSESSMENT
A monitoring program is an integral part of any stormwater assessment
study. The attributes and characteristics of urban runoff vary significant-
ly from one location to another, and handbook or extrapolated values from
sources such as Chapter 5 of this manual, or so called "default values"
employed in some mathematical storm runoff models, should be utilized with
caution and judgement. Normally, site specific data will be required for an
adequate understanding and characterization of local conditions, and to
permit published values or ranges to be utilized in a sensible manner.
Chapter 6 presents considerations and guidelines for developing a storm-
water monitoring program. Aspects of the program discussed include rain-
fall monitoring, drainage basin characterization, monitoring of runoff and
overflows, and receiving water monitoring. The emphasis is on the identifi-
cation of the important components of a stormwater monitoring program, and
on discussing some relevant aspects, which should aid in structuring an
effective overall program. Information important to the actual execution of
a monitoring program is available from other sources and is not covered
here. A number of such reference sources which describe monitoring, sampl-
ing techniques, and sampling equipment in detail, are identified and briefly
reviewed at the end of this chapter.
6.1 Purpose of Monitoring Program
The collection of data is not in itself a goal in a stormwater assess-
ment program. Large data bases are of little value unless they aid in the
planning and decision making process. The purpose of any monitoring pro-
gram is to provide the data necessary for application of the methods of
analysis which will be used to study the defined set of water quality prob-
lems. The data needs will vary depending on the problem, method of analysis,
and the degree of confidence desired or achievable. Generally, the more
sensitive a parameter is to the solution, the more accurately it should be
defined.
The design and implementation of a program which will maximize the
effectiveness of an effort with specific time and budgetary constraints is a
challenging task. Because of the substantial variability of the data which
will be obtained, and the physical problems associated with securing and
analyzing the samples which the program will specify, it is most important
that the monitoring program be carefully designed. It will be much more
effective to the overall program to implement a monitoring effort which is
6-1
-------
relatively modest in scope, but which provides complete data sets, reliable
results, and the type of data which improves estimates of the important
parameters in the analytical procedure subsequently used. A program which
generates large quantities of data may at the same time do little to contrib-
ute to an understanding of the system, or improve on the reliability of, or
confidence in, projections to be made.
As a basic principle, the design of a stormwater monitoring program
should reflect how the information obtained will be used in the analysis to
follow. It should be developed by "working backwards" from a prior identific-
ation of the problem to be corrected, and the analytical procedure which will
be used to characterize loads and impacts and examine the effectiveness of
control measures.
Existing water quality data for the area should be collected and review-
ed. Data from dry and wet weather periods should be compared for indications
of stormwater effects. A visual inspection of the study area may indicate the
existance of some water quality problems which normal data bases will not re-
flect. Shortcomings in the existing information and data base should be
noted and the significant time and space scales of the problems should be
determined.
Initial screening using methods outlined in Chapter 3 may be helpful in
providing an overall perspective and in defining any parameter inputs that
are particularly important to the problem solution. The basic elements in
any stormwater analysis include the rainfall characteristics of the area, the
amount of rainfall which reaches the receiving stream as surface runoff, the
contaminants carried by the runoff together with a measure of the amount
associated with runoff events, and the impacts of runoff loads on receiving
water quality. Some pertinent scientific and technical criteria are presented
for each of these elements in the sections which follow.
6.2 Rainfall Monitoring
Rainfall data are utilized in stormwater analyses in two significantly
different ways.
1. To determine runoff characteristics of the study area, rainfall
data for specific storm events are used in conjunction with
measured runoff flows.
2. To project runoff loads from the study area for evaluation of
water quality impacts and the effectiveness of control measures,
a characterization of the overall rainfall patterns for the
area is required. Long term characteristics and statistical
properties of a record representative of the area are important
rather than single event data for subcatchments.
6.2.1 Use of Rainfall Data to Define Site Characteristics
An important requirement of stormwater studies is that rainfall and
runoff measurements be made by a monitoring program to characterize the
6-2
-------
runoff ratio for the area, i.e., the proportion of rain which falls on an
area which will result in surface runoff. A limited number of site events
will be monitored in this manner, even in a relatively comprehensive program;
so long term rainfall characteristics will not be defined by such data. Rain-
fall data used for defining site runoff characteristics must reflect as
accurately as possible the amount of rain falling on monitored catchments
during specific storm events. This almost invariably requires local gages,
or a network of gages in close proximity to the monitored catchment, with
relatively small distances between gages.
The ratio of runoff to rainfall volume determined from such monitoring
efforts is used in the statistical method or simplified simulation models,
such as those discussed in Chapter 4, or as direct input to more sophisticat-
ed, event oriented simulations.
Since rainfall depths vary spatially, there is some error in the areal
estimate of rainfall volume for individual events, even for small areas and
local gages. This error is related to the size of the drainage area, the
rainfall pattern, and the density of raingages. One of the practical effects
of the error is to cause variability in the measured ratio of runoff to rain-
fall. This is illustrated schematically in Figure 6-1 for two hypothetical
storms. Although the same amount of rainfall occurs over the drainage area
and the measured runoff volume is the same, the apparent ratio of runoff to
rainfall is 0.8 in the first case, and 0.33 in the second case. A monitor-
ing program consisting of a number of storms (for example, 10 or 20) may
find considerable variation in the ratio of runoff to rainfall from storm to
storm. While a portion of this variation due, for example, to changing
antecedent conditions is real, in the sense that it would be present even if
the average areal rainfall depth during a storm was exactly determined, a
major portion may also be due to inaccurate areal rainfall estimates. The
larger the number of events which are monitored and averaged, the better will
be the estimate of the mean ratio. When it is possible to monitor only a
limited number of events, the error in the estimate of areal storm depths,
and hence the runoff ratio, may be reduced by increasing the raingage density.
6.2.2 Use of Rainfall Data for Projections and Evaluation of Impacts
Different criteria are important when the purpose is to project runoff
loads under the entire range of storm events which can occur in the study
area. Such determinations require reliable long-term records of precipita-
tion, and a record of adequate duration for definition of the statistical
properties of the storm events. U.S. Weather Bureau records normally provide
both the desired reliability and period of record, although they are often
some distance removed from specific study sites. For any but quite closely
spaced gages (or a gage within a relatively small drainage area) such records
are normally unsuitable for defining site characteristics due to poor corre-
lation on an individual event basis. On the other hand, when monthly, annual,
or particularly, when long period records are compared, quite satisfactory
correlations may exist even for gages spaced on the order of 25 to 50 miles
apart. Thus, barring the presence of defined local influences such as
orographic effects caused by local topography, U.S. Weather Bureau records
6-3
-------
VOLUME RUNOFF/VOLUME RAINFALL =
REAL --05
APPARENT =0 8
(5VRAINGAGE.
DRAINAGE AREA
VOLUME RUNOFF/VOLUME RAINFALL--
REAL = 05
APPARENT -0 33
(GVRAINGAGE
REAL (AREALLY AVERAGED)
RAINFALL VOLUME = 2.00 IN.
APPARENT RAINFALL
VOLUME - 1.25 IN
MEASURED RUNOFF
VOLUME = I 00 IN.
REAL (AREALLY
AVERAGED)
RAINFALL
VOLUME = 2.00 IN.
APPARENT RAINFALL
VOLUME = 3.00 IN.
MEASURED RUNOFF
VOLUME = 1.00 IN.
FIGURE 6-1
EFFECT OF AREAL DISTRIBUTION OF RAINFALL
ON REAL VS APPARENT RATIO OF RUNOFF TO RAINFALL
6-4
-------
can be used effectively for projections, even where gage locations are not
within the study area.
Guidelines have been given to enhance the accuracy of such determina-
tions by integrating the data from a network of gages, or by adjusting point
rainfall records to better reflect areal distributions (see Section 5.1.4).
Such refinements increase in importance with the size of the study area and
the remoteness of reliable long term gaging stations.
The characterization of the long term event statistical properties is
thus usually based upon existing recording raingages which report to the
National Weather Service. If none, or too few of these gages are available,
their records may be supplemented by placing additional continuously record-
ing raingages in the study area. These data will be of little use in the
long term characterization of precipitation until at least one year of
records are accumulated. After one year, the data from the new gages may
be compared to the existing gages to identify differences within the study
area and to study the areal variability of rainfall by aggregating the
records, as described in Section 5.1.4.1. Factors determined for converting
point rainfall to areal rainfall during the study year may be applied to the
longer records of the previously existing raingage(s), until more years of
data are collected at the new gage(s). Therefore, there may be considerable
value in having additional raingages for the statistical characterization of
precipitation.
6.2.3 Raingage Density
A brief discussion of some basic relationships between the density of
raingages in the study area and the accuracy of rainfall estimates is pro-
vided below. As may be expected, at higher raingage densities, the rainfall
on a study area is determined with greater accuracy. However, the intent
is not to specify a minimum raingage density, or the design of a network
which would result in a given maximum error of estimate. At this stage of
the examination of stormwater loads and impacts, it is neither practical
nor desirable to commit resources to the design, establishment and operation
of elaborate or extensive networks of raingages. By and large, stormwater
studies will work with what is available. The value of an understanding of
the raingage density relationships discussed here is that they help provide
an appreciation of the limitations of the existing rainfall record.
It is always preferable to use data from several raingages to determine
the average areal storm depth during a particular storm, rather than data
from a single gage. Standardized techniques, such as the averaging method,
Theissen method, and isohyetal method (1,2) are available for this type of
determination.
A useful method for evaluating the ability of a network of raingages
to represent the rainfall over a given area is to compare the areal rainfall
depth estimated by a subset of the network to that estimated by the entire
network. The rainfall depth estimated by the entire network is assumed to
be the true storm depth. Figure 6-2 summarizes the results of such an
analysis using data from a dense raingage network in Illinois, as reported by
6-5
-------
1.0
0.5
CO
Ul 0.2
I
u
z
~ O.I
(t
o
a:
K 0.05
UJ
0:02
10
UJ °'01
CD
<
QC
0005
0.002
0.001
[100%
AVERAGE
PERCENT < 20%
ERROR -
1%
I I I I I I 1 I
I I I I I I I I
I
I I
400 mi /GAGE
I - HOUR STORM
2
6
12
24
lOOmiVGAGE
I-HOUR STORM
2
6
12
24
25 miZ/GAGE
I-HOUR STORM
2
6
12
24
0.01 OO2 OO5 01 O2 O5 10 2.O
STORM MEAN RAINFALL (INCHES)
5.O
FIGURE 6-2
ESTIMATE OF RAINFALL SAMPLING ERROR IN 400mi
ILLINOIS AREA
.2
6-6
-------
Huff (3). The average sampling error of the storm depth is seen to decrease
with larger volume, longer duration storms, and with a more dense network
(the curves were determined with regression analysis). The calculated
average percent error has been superimposed upon Huff's curves. To provide a
perspective, we can assume that we are most interested in storm events with
mean rainfall volumes between 0.1 and 1.0 inches, because smaller storms
contribute less significantly to waste loads and impacts, and larger storms
occur relatively infrequently. For this range of storm events, in the
Illinois study area, the average percent error is approximately nill for a
raingage density (RGD) of 25 square miles per gage, nill to 20 percent for a
RGD of 100 square miles per gage, and 10 to 60 percent for an RGD of 400
square miles per gage. Note, however, that this is the average percent
error, and a few of the storms may still be estimated with a significantly
larger error.
A similar type of analysis was conducted by Hydroscience, Inc. (4) using
data collected by the U.S. Geological Survey and the Texas Water Development
Board (5,6,7) on seven small watersheds in Texas.
The results of this analysis are summarized by Figure 6-3, indicating
the effect of raingage density on the error which can be expected in deter-
mining the distribution of rainfall on an area, based on data from a gage or
a network of gages. These results of the Texas study are displayed different-
ly than those from Illinois, and use a different basis for defining accuracy.
In the Texas area, raingage densities in the order of 10 to 25 square miles
per gage produce a standard error of estimate in the order of 10 percent.
This is related to, though not the same measure as the average percent error
which describes the Illinois results. A standard error of 10 percent signi-
fies that the estimated rainfall for a storm in the study area will be
within +_ 10 percent of the true rainfall about 2/3 of the time.
A conclusion which appears appropriate based on a review of published
literature on the subject (8), and as illustrated by the two examples pre-
sented, is the following. The number of raingages in relation to the size
of the study area (the raingage density) can have an appreciable effect
upon the estimates of rainfall on a study area by using one or a series of
raingage records. As indicated by the difference between the Illinois and
the Texas results, and further by the differences between individual basins
in the Texas study, the geographic location will influence the relationship
between the raingage density and the accuracy of estimates which are made.
Seasonal effects in an area which, for example, may be prone to summer
thunderstorm activity can also be expected to influence this relationship.
It is, accordingly, possible to provide only some general guidelines con-
cerning raingage density in tnis discussion.
Raingage densities in the order of 10 to 25 square miles per gage
(gages spaced four to five miles apart), would generally appear to provide
a very satisfactory degree of accuracy for utilizing long term records, or
for estimating the distribution of rainfall over a study area, if one is
content to accept the fact that the error might be great on any one specific
storm, but that over a period of time results average out such that the
6-7
-------
(O
UJ
tn
UJ
<
u.
z
CO
UJ
o
o
UJ
u.
u.
UJ
o
10
O
(%
«o_
m"
-------
overall analysis has a high level of accuracy. On this basis, errors in the
order of 10 percent can be considered quite acceptable.
At a density of about 100 square miles per gage (about 10 miles spacing
between gages), accuracy in most cases will probably prove acceptable for
the type of analysis being performed, recognizing the accuracy of other
elements of the analysis, and particularly given the sensitivity of water
quality impacts to uncertainties in the waste loads in the order of 10 or
perhaps 20 percent. It would appear, from available information, that when
raingage densities reach or exceed on the order of 400 square miles per gage
(20 miles spacing between gages), that the accuracy of estimates of the areal
distribution of rainfall on a study area become more uncertain and are likely
to result in a large error for many of the rainfall events.
The guidelines just presented refer to the use of a long term records,
or the application of rainfall records in a manner which takes advantage of
the fact that errors on individual storm events will average out in the long
term. It is quite a different matter when such an "averaging out of results"
cannot be tolerated. This is the case when the purpose of the rainfall
measurement is for comparison with measured runoff from a specific catchment
to determine the characteristics of that site. Here, the requirement is for
the best estimate of how much rainfall occurred on the monitored area during
the particular storm event for which runoff flow and quality are measured.
Considerations such as the average error or the standard error of the esti-
mate are less meaningful in this case, since even acceptably low average
errors provide no confidence in the error of a specific individual event.
For this reason, it is strongly recommended that for the determination
of site characteristics, the monitoring program include at least one gage
located in the catchment being monitored for quality and flow. For this
purpose, relatively simple gages which provide only the total volume during
the storm event are appropriate. Since monitoring teams will be active in
the area for measuring flows and collecting quality samples, the servicing
of such simple gages will not usually add a significant burden to the effort.
If and when the stormwater analyses make use of sophisticated, event oriented
simulation models which require more temporal detail than the methods pre-
sented in this manual, continuous recording gages would be required rather
than the simple volumetric gages suggested above.
Marsalek (14) cites Linsley (26) as recommending the use of two or more
raingages for even the smallest watershed, with an absolute minimum of one
gage located within the watershed. He considers two gages sufficient for
areas up to 4 square miles, and three for an area up to 20 square miles.
6.2.4 Other Raingage Considerations
The selection of sites for raingage locations is important, particlarly
in highly developed urban areas. Ground level locations are preferable,
however, roof tops may be used as long as the gage is not placed near the
edge, or near obstacles on the roof (9). It is generally agreed that gages
not be placed within a distance of three or four times the height of the
nearest obstacle (i.e., building, tree, etc.). However in densely settled
6-9
-------
urban areas, distances of at least one height may serve as a minimal yet
reasonable criterion (9). Although shielding is used at most first order
stations, it is generally not required (9).
There are several types of recording raingages available (10). The
tipping bucket gage is recommended because of its commercial availability
and proven ability. Recordings are made for each 0.01 inch of rain. The
tipping bucket raingage may underestimate very high rainfall intensities,
however, and is not appropriate for snowfall measurements. As previously
discussed, budgetary constraints may limit the use of recording gages in each
area, and less expensive manual funnels, which measure only the total storm
depth at the end of an event, may be appropriate.
6.3 Site Selection and Drainage Basin Characterization
Ideally the runoff and overflows from the entire study area should be
monitored for a large enough number of events to accurately define the prop-
erties of stormwater pollution loads. Since this may not be economically
feasible, the selection of several representative areas is often necessary.
There are a number of criteria for selecting monitoring sites. One approach
is to select small drainage areas, each having one predominant characteristic
or land use. Data collected from each monitored area are then synthesized
and extrapolated to represent similar areas (i.e. land use) in the rest of
the community. This approach is directed towards identifying relationships
between drainage basin characteristics and runoff properties. Character-
istics which should be considered include land use, type of conveyance system
(i.e. combined vs. separate), conveyance system characteristics such as
sewer slope, and geographic location. Sites are selected to represent the
range of conditions found in the planning area.
Another approach to site selection is to monitor larger areas with
mixed characteristics. The runoff properties are more representative of the
average, typical conditions found throughout the area and the total load
from the urban area may be better estimated as a larger portion of the region
is monitored.
One promising method for determining the quality of combined sewage
in larger, integrated areas is to monitor the influent to combined municipal
treatment plants during storms (11). The disadvantage of sampling large,
mixed areas is that it is impossible to isolate the effects of pertinent
drainage basin characteristics (such as land use) on runoff quantity and
quality.
There are advantages and disadvantages to each of the approaches to site
selection described above. A comprehensive monitoring program can incorpo-
rate both types of sampling (12): a number of small, homogeneous drainage
areas can be monitored to attempt to identify and isolate the effects of
land use, sewer characteristics, etc.; and a few larger areas should be
monitored to better represent the total loadings from the study region.
A number of other factors should be considered when selecting monitor-
ing sites. These include:
6-10
-------
1. A complete flow balance must be established in areas where the
ratio of runoff to rainfall (R.J is to be calculated. This limits
the possible sites to hydraulically tight systems for which complete
flow balances are possible. These tend to be smaller, less complex
basins.
2. Sites need to be suitable for the placement of necessary sampling
equipment. Prospective sites can be observed during storm periods
to check for possible problems.
3. Sites that are readily accessible, particularly during adverse
weather conditions are preferred. The availability of electrical
power is an important consideration.
4. The potential for vandalism is also a consideration. Positive
public awareness of the program should be encouraged.
5. Written legal permission should be obtained for land rights and
easements.
A more detailed discussion and methodology for site selection is included
in the Guide for Collection, Analysis, and Use of Urban Stormwater Data
(10).
Characterizations of basins and catchments within the urban area are
necessary for both the site selection and the data analysis procedures. A
comprehensive listing of the elements of a basin characterization is present-
ed below. The level of detail with which the various elements are addressed
in a study is subject to flexibility, based on the scope and depth of analy-
sis in the overall effort. This is particularly true of the detail with
which the collection system is characterized.
The size of each area and the predominant land characteristics are
determined. The percent of each land use type and population density are
noted, and may be used to estimate the percent impervious area in each
catchment. Land use information and a direct estimate of the percent im-
pervious area may be obtained from areal photographs, satelite photographs
(Landsat)(13), or local inspection, though local inspection is impractical
for large study areas. Some information on local soil types, permeability,
and ground slopes should be obtained. If a site specific relationship be-
tween the percent impervious area and the ratio of runoff to rainfall
is being determined to further refine the general range shown in Figure
5-20, a wide range of imperviousness should be obtained throughout the city.
Information on the collection system is also required. Tributary
areas are delineated and overflow points located. The capacity of major
interceptors is catalogued and an estimate of internal storage is made. The
potential for deposition in the system should be estimated using factors
such as average sewer slopes and dry weather flow velocities. An estimate
of the dry weather flow in combined systems (or in separate storm systems
with illegal connections or infiltration) is necessary. Certain situations
where the sewer maps do not provide an adequate representation of actual
6-11
-------
conditions, or where there are faulty or inoperative regulators, illegal
connections, or cross connections between systems are best avoided. Local
sewer maintenance personnel can provide guidance.
Management practices which are currently employed in the drainage areas
are of interest, such as the street sweeping frequency and catch basin
cleaning programs. Construction sites, residual disposal sites, highway
deicing practices, and special industrial facilities (i.e. storage ponds,
transfer sites, etc.) are of particular note. If significant activities
occur on a seasonal basis, samples collected during the different seasons
can be checked for this influence.
6.4. Monitoring of Runoff and Overflows
Stormwater load determinations are based on estimates of quantity and
quality, and both aspects must be monitored in an assessment program. As
discussed previously, the monitoring program is tailored to collect data for
use in a specific modeling procedure for a specific set of water quality
problems, although some consideration can be given to possible future uses
of the data. The guidelines presented in the following sections are primari-
ly directed towards programs collecting information for studies using assess-
ment models such as the statistical method (Chapter 3) and broad scale
simulators (Chapter 4). The required information is the total storm runoff
volume, average flow rate, average concentration, general temporal concentra-
tion profiles (average duration and magnitude of the first flush), and some
indication of flow-concentration correlations. Requirements for programs
collecting data for more sophisticated models, which predict hydrographs and
pollutographs at finer time scales, are also briefly discussed. Factors con-
sidered include the length of the study period, sampling frequency, and the
sampling procedure.
6.4.1. Study Period and Sampling Frequency
Monitoring programs can vary from a few typical events during a few
critical months, to many events over one or more years. The length of the
study period varies depending upon the purpose of the study, availability
of budget and manpower, and the magnitude and complexity of the study area.
It is important to plan a program to obtain needed data within the
constraints of the study circumstances and to yield a maximum of useful
information for the monies spent. In a statistical evaluation of rain
events performed by the National Oceanographic and Atmospheric Administration
(NOAA), one of the findings is that there is a 90 percent confidence level
in the expectation that 85 percent of the various intensity and duration
rain events for a given location will be experienced within a 2.8 year
period (14). It would therefore be a rather long term and costly project to
study even most of the possible rainfall events. Some decision must be made
on how long a study period should last and how many events should be monitor-
ed.
For assessment models, the first purpose of the monitoring program is to
determine the average ratio of runoff to rainfall (R ) and average values for
6-12
-------
pollutant concentrations (c). A consequence of statistical sampling theory,
based on the Central Limit Theorem, is that the sampling distribution of the
average tends to become normal as the number of samples increases, no matter
what the distribution of the variable being sampled (15). This property may
be used to estimate the number of storms necessary for an adequate estimate.
The number is dependent upon the variability of the parameter being
measured (in this case, R.. and c), the maximum desired error in the estimate
of the average, and the confidence that one has that the sampled average
falls within the particular range. Figure 6-4 shows the relationship between
the number of site events monitored and the resulting level of accuracy. The
curves presented are based on a coefficient of variation of 0.75 for the
sampled parameters, which appears to be a reasonably conservative estimate
for both R and pollutant concentrations based on observed data (16,17,18).
In general, 20 to 40 events will provide a very adequate representation.
When time and budget constraints restrict the number of storms which can be
sampled, programs may be designed with as few as 10 storms, although much
below this number may leave little confidence in the estimated averages, as
indicated by the curves in Figure 6-4. It is better to accurately define
the properties of fewer sites than to collect an insufficient number of
storms in many locations (19). Seasonal or temporary conditions (seasonal
industrial operations, leaf fall, construction activity, etc.) may require
that a larger number of storms be sampled to adequately characterize a
given site. At least one dry weather, diurnal sampling of each combined
sewer site should be conducted to identify base conditions and its variation
over the day, to better evaluate storm runoff quantity and quality during
monitored storms.
6.4.1.1. Sampling Interval Within Storms
The variability of pollutant concentrations within storms tends to be
about equal in magnitude to the variability between storms. Therefore,
Figure 6-4 may also be used to estimate the number of samples needed within
a storm to adequately estimate the mean runoff concentration during the
event. This type of characterization is appropriate for the assessment
procedures used in this manual. Ideally, shorter storms should be sampled
more frequently than longer ones, however, there is no way of knowing the
duration of an event until after it is over. An intermediate frequency may
be chosen based on the accuracy desired, the average duration of runoff
events, and the practical limitations of the sampling equipment and col-
lection procedure. If, for example, runoff durations are typically 5 to 10
hours and at least 10 samples per storm are desired, a maximum sampling
interval of one half hour to one hour is indicated. A few short storms may
be undersampled and a few longer events sampled more than necessary.
Sampling within a storm may also be directed towards an examination of
the first flush effect, and more frequent sampling is required in the early
portions of the storm. This more frequent, initial sampling may be routinely
incorporated in the monitoring program, or only conducted on a limited number
of special surveys, depending on the significance of the first flush effect
for the planning study. The frequency required during early portions of the
storm is the same as is generally prescribed for sampling programs collecting
input for more refined analyses which attempt to accurately reproduce the
6-13
-------
MO If
V -. 75 FOR t^TH CURVtb
90% CONFIutNCE LtVLL
68% CON^IUENCt LtVtt.
A) i'l 1
;-NUV:'L« OF STuKMD MOfJlTOHEU
FIGURE 6-4
ERROR IN ESTIMATE OF AVERAGE
VERSUS NUMBER OF STORMS MONITORED
6-14
-------
storm pollutograph. A review of stormwater runoff and overflow monitoring
programs yields the following suggested sampling intervals for this type of
program (14):
SAMPLING INTERVALS FOR ACCURATE FIRST FLUSH OR
POLLUTOGRAPH DEFINITION
Basin Size Sampling Interval
(acres) (minutes)
10 5
100 5-10
500 10-15
1000 10-15
2000 15
3000 20
5000 25-30
The size of the catchment is important because runoff from smaller areas
occurs more rapidly with more strongly defined first flush peaks. However,
this short interval sampling is not necessary for all storms when the
analysis procedure is based upon the type of assessment methods described in
this manual.
6.4.2 Sampling Procedure and Parameters
Sampling procedures can differ, depending on whether the objective is to
identify the within-storm variations in concentration, or to determine the
total mass of pollutant discharged per storm event.
A sampling program designed to define the mass per storm event will
result in a single sample being analyzed for each parameter of interest,
representing the particular site and storm event. The pollutant concentra-
tion so determined will, when multiplied by the total volume discharged
(determined from flow monitoring records) yield the mass discharged. In
order that the single sample provide an appropriate representation of the
entire storm, a careful compositing procedure must be employed.
Obviously, flow-weighted composites are most desirable. This requires
that a series of discrete grab samples collected throughout the duration of
the storm event, be combined using a different volume from each discrete
sample, the volume of each being determined by the amount of storm flow
associated with it. Equal volume samples are taken at regular intervals
during the storm event, either manually or with an automatic sampler; moni-
toring of flow is required. One practical problem with the successful use
of this approach, is the delay sometimes encountered in retrieving and an-
alyzing the flow record. To use this approach, flow records must be examined
on a sufficiently timely basis to comply with guidelines for acceptable time
delays in processing samples for chemical and biological analysis.
6-15
-------
In cases where flow records are not taken, where they can not be re-
trieved rapidly enough, or where they are not available due to malfunction,
then simple time compositing of individual samples must be employed. Al-
though less desirable than flow-weighted samples, simple time composited
samples can provide useful information on mean concentrations and mean
storm loads, provided that:
(a) A sufficiently large number of storm events are analyzed. The
variability of within-storm flows and concentrations would tend
to average out errors introduced by lack of flow-weighting.
(b) The analyst recognizes that the best estimate provided will be
for long-term effects, and that the estimate for an individual
storm event may be poor.
The most appropriate approach for securing a flow-weighted storm
composite, will involve the use of automatic equipment with sampler operation
being controlled by a flow meter. Two approaches can be employed based on
the type of automatic equipment selected. The first results in the collect-
ion of samples of variable volume, but at regular time intervals. The
volume of sample collected is proportioned to the rate of flow at the time it
is taken. The other approach employs a flow meter with an integrator
(totalizer), so that the sampler is activated on the passage of a specific
volume of flow. Thus, uniform sample volumes are collected, but at variable
intervals of time.
Programs, or components of programs, designed to secure information
which will permit generation of Pollutographs, are simpler to address,
since they eliminate the question of compositing. They do however, utilize
a much higher level of laboratory analytical resources, because of the large
number of samples analyzed for each storm. A series of sequential grab
samples are collected, either manually or with automatic samplers, at either
constant or variable time intervals related to flow changes. A modification
of this approach, which reduces the number of samples to be subjected to
laboratory analysis, composites samples for certain periods during the event,
so that a series of sequential composite samples is provided for analysis.
A question typically faced in design of a sampling program, is the
interval at which discrete samples should be secured. An approach which
recommends itself is to adopt an objective of securing some minimum number
of samples for each event, which will span the entire event. Samples are
taken relatively frequently, such that the desired number will be secured
during relatively short storms. During the longer storms, extraneous
samples are discarded.
Whether the preferred approach to storm monitoring for a particular
project should involve the use of manual or automatic methods of sample
collection, is something best decided on an individual basis. Automatic
sampling is theoretically superior, however, automatic equipment is subject
to malfunction and to physical damage in this application. The financial and
manpower resources will be influential in the decision to be made. Either
6-16
-------
manual or automatic methods"can be successfully employed. Both, however, re-
quire care and attention in the design and execution of the program.
The water quality constituents that are of interest in stormwater runoff
and overflows are dependent upon the particular problems present in the study
area. Analytical costs may be high and discretion should be used. Routinely
sampled are indicators of oxygen demand, particulate concentration, patho-
genic microorganisms, nutrients, and toxic substances. Oxygen demand is
measured with the BOD^ test, however, some long term oxygen demand tests
(i.e., 40 days) are run to better estimate the oxidation rate, check for
nitrification, and check for possible toxicity effects which may suppress
the BOD,.. Long term BOD's may be run with and without agents to suppress
nitrification. Measurements may also be made on filtered and unfiltered
samples to estimate the fraction of the biochemical oxygen demand in the
particulate or dissolved states. Filtered and unfiltered tests are also
suggested for other pollutants to aid in the treatability assessment (Section
5.5.1). The particulate concentration is usually determined by the suspended
solids test (non-filterable residue), however, selected tests for settleable
solids may also be performed to examine deposition and treatability. Total
and fecal coliforms are common indicators of pathogenic organisms, and fecal
streptococcus may also be measured to examine sanitary quality. Suggested
nutrient measurements include nitrate and nitrite, total kjeldahl nitrogen,
and total phosphorus. Toxic substances may include heavy metals, pesticides,
or herbicides; depending upon local problems and conditions. Chloride
measurements are made in coastal areas to check for seawater intrusion due
to faulty tide gates. Because the alkalinity of rainwater is very low (near
0 mg/1) and the alkalinity of sewage in a given location is often relatively
constant (between 50 and 200 mg/1), alkalinity measurements may serve as a
useful tracer for distinguishing between domestic wastewater and stormwater
runoff in the sewer system.
6.5. Monitoring of Receiving Water
If the review of existing data and a general knowledge of the problems
in an area are insufficient to adequately define the water quality problems
and their relevant time and space scales, an initial survey of the receiving
water should be conducted. Water quality samples are collected in the re-
ceiving water during representative wet and dry weather periods. A water
quality survey should be conducted during a dry weather period in the summer
months when flows are low and temperatures are high. During these dry
weather periods, the small streams and rivers will often approximate steady
state conditions and can be sampled spatially at an appropriate time to
determine the steady state profiles of various pollutants. Grab samples are
collected at a number of locations with more refined spatial sampling near
major waste loads where steeper spatial concentration gradients are expected.
Composite samples of significant waste loads should also be taken. In
estuaries, the water quality varies over the tidal cycle, but may be at
steady state intertidally, and samples should be collected at both high and
low slack. One or two dry weather surveys should suffice to estimate the
magnitude of water quality problems, and the analysis of flow records will
normally indicate the likelihood of these problems occurring during other
periods of the year.
6-17
-------
A wet weather water quality survey should be conducted by collecting
water quality samples at various locations and at a number of times during
and following a storm event. The sampling interval in the receiving water
is determined based on an analysis of predicted responses. In a stream
where little dispersion or mixing occurs, storm impacts are pronounced, but
are transported rapidly form the point of discharge. In an estuary, the
stormwater load is mixed and diluted upon entering the receiving water, but
the effects of the storm remain well after the end of the event, and less
frequent sampling for a longer period is required.
A preliminary calctilation using a receiving water model or equivalent
calculation, and utilizing preliminary estimates of pertinent input factors
will often prove quite helpful in the design of an effective receiving water
monitoring program. Even relatively crude calculated responses for the para-
meter of interest will aid in establishing an effective time and space
framework for the monitoring effort. The samples collected from the dry and
wet weather survey should be analyzed for dissolved oxygen and pollutant
indicators such as the biochemical oxygen demand, suspended solids, dissolved
solids, total and fecal coliforms, phosphorus and nitrogen, heavy metals,
and other pollutants which may be important in the study area. Areas with
suspected algal problems should be sampled throughout the day for dissolved
oxygen, particularly in the early morning and the late afternoon. Chloro-
phyll 'a' measurements are often made as an indicator of phytoplankton
levels. Areas where oxygen depressions are not readily explainable, particu-
larly near combined sewer overflows or major waste discharges, should be
checked for sediment oxygen demand. Chloride measurements are taken in
estuarine and coastal areas to help identify the transport properties of the
water body.
A review of available information on the physical and hydrological
characteristics of the receiving water should also be made. If certain in-
formation necessary for the modeling effort are lacking, these data should
be collected as part of the monitoring program. These include the geometry
of the water body (cross-sectional area and depth) in a number of locations,
flow and velocity, temperature, and dispersive properties (dye studies) if
these are important. For small streams and rivers, the cross-sectional area
varies throughout the year depending on the flow, and a number of measure-
ments during different periods are necessary.
After the initial assessment of water quality, further sampling surveys
may be appropriate to resolve problems identified by the early efforts, to
improve the calibration and verification of receiving water models, and as
part of the ongoing monitoring of possible problems. These surveys should be
directed towards a better defined set of specific water quality problems.
6.5.1 Continuous Monitors
As methods which predict the long term characteristics of water quality
(mean, variability, and frequency distribution) become more advanced, data
to verify these predictions become more useful. Continuous monitoring
systems for water quality can provide these data. The entire range of water
quality is observed, without missing rare or extreme events.
6-18
-------
The state-of-the-art of continuous water quality monitors is rapidly
changing. Systems range from electrode monitors to automated chemistry
units. Although a number of parameters are measurable, currently the most
reliable monitors are for temperature, dissolved oxygen, pH, and conductiv-
ity. Monitors are still susceptible to malfunction and should be checked
and calibrated frequently. Interest has also been recently generated in
techniques for estimating water quality from satelite photographs, though
this method does not yet appear very promising (20).
6.6 Review of Monitoring Literature
The monitoring section of this manual presents a broad overview of con-
siderations for an effective stormwater monitoring program. A number of
reports are available which provide more detail on specific aspects of
monitoring. A number of particularly valuable publications are briefly dis-
cussed in this section and these or similar reports should be read before
initiating a stormwater monitoring program.
A general, comprehensive review of monitoring requirements, methods, and
costs is included in Appendix D of the Areawide Assessment Procedures
Manual (21). This report discusses sources of available data, types of
monitoring activities, and the selection of sites, parameters, and frequen-
cies for sampling. Flow measurement and sampling equipment and procedures
are analyzed, and cost estimates are provided for various phases of the
monitoring program.
The report, Guide for Collection, Analysis, and Use of Urban Stormuater
Data (10), presents the findings of a unique conference sponsored by the
Engineering Foundation, U.S. Geological Survey, and the American Society of
Civil Engineers. The final report was rewritten by the approximately 70
attending participant specialists assigned to different work groups. The
report discusses stormwater data utilization and modeling analysis, network
planning and design, catchment selection, instrumentation, and sampling
procedures.
Section VI of Urban Stormwater Management and Technology - An Assess-
ment (22) examines various data collection and sampling programs. Guide-
lines are presented for selecting sampling sites and sampling equipment.
Factors such as intake design, collection method, sample transport and
storage, and controls and power are discussed. Case histories of large scale
sampling systems with specific problems and recommendations are presented.
Various types of flow measuring devices are assessed for performance and
cost. Research efforts in flow measuring technology are described and
suggestions are made for data analysis.
The report, Methodology for the Study of Urban Storm Generated Pollution
and Control (23), contains recommendations for standard procedures to follow
in projects dealing with pollution assessment and abatement of storm genera-
ted discharges. The purpose of the project is to develop standard procedures
needed to insure that all dischargers and treatment processes are evaluated
with the same methods. Issues addressed include methods for sampling and
sample preservation, available instrumentation, the choice of water quality
6-19
-------
parameters, analytical procedures, and methods for evaluating treatment
processes.
A study conducted for the Canada Center for Inland Water, Instrumental-
tion for Field Studies of Urban Runoff (14], examines monitoring techniques
for field studies of urban runoff. Recording precipitation gages, sewer flow
measurement instruments and automated wastewater samplers are examined.
Comments and discussions are also presented for sampling runoff loads to
obtain total storm pollutant loadings and to define the magnitude of the first
flush effect. A review of sampling methods is made and comparisons are pre-
sented for data during one storm in Washington D.C.. Sampling intervals used
in a number of urban runoff studies are summarized.
A section on the "Collection of Field Data for Stormwater Model Calibra-
tion" in the 1976 Short Course on Applications of the Stormwater Management
Model (SWMM) (24) reviews the types of field data and the factors involved
in wastewater characterization. Flow measurement, sampling, and laboratory
analyses are discussed in detail and a review of recent field experience is
presented.
The Environmental Protection Agency report, An Assessment of Automatic
Sewer Flow Samplers (25), presents a general discussion of the purposes and
requirements of a sampling program. The favorable characteristics of auto-
matic sampling equipment are set forth and problem areas are outlined. A
compendium of over 60 models of commercially available and custom designed
automatic samplers is included. A description and characterization of each
unit is presented along with an evaluation of its suitability for stormwater
applications. A review of field experience with automatic sampling equipment
is given and a design guide for the development of an improved automatic
sampler for use in storm and combined sewers is presented.
6.7 References
1. Linsley, R.K., Jr., M.A. Kohler, and J.L.H. Paulus, Hydrology for
Engineers, McGraw-Hill, New York, 1958.
2. Chow, V.T., Handbook of Applied Hydrology, McGraw-Hill, New York,
1970.
3. Huff, F.A., Sampling Errors in Measurement of Mean Precipitation,
Journal of Applied Meteorology, Volume 9, February 1970.
4. Water Quality Management Planning Methodology for Urban and Industrial
Stormwater Needs, Hydroscience, Inc. for Texas Water Quality Board,
Arlington, Texas, 1976.
5. Hydrologia Studies of Small Watersheds, Pin Oak Creek, Trinity
River Basin, Texas, 1956-62, Texas Water Development Board, Report
54, August 1967.
6-20
-------
6. Hydrologic Studies of Small Watersheds, Cow Bayon, Brazos River
Basin, Texas, 1955-64, Texas Water Development Board, Report 99,
October 1969.
7. Hydrologic Studies of Small Watersheds, Calaveras Creek, San
Antonio River Basin, Texas, 1959-69, Texas Water Development Board,
Report 154, August 1972.
8. Rodriguez-Iturbe, I., and J.M. Mijia, The Design of Rainfall Networks
in Time and Space, Water Resources Research, Volume 10, No. 4,
August 1974.
9. Huff, Floyd A., Sampling, Analysis, and Interpretation of Rainfall
for Hydrologic Applications, Presentation at MSDGOAMSA Workshop on
Water Quality Surveys for 208: Data Acquisition and Interpretation
of Non-Point Runoff, Chicago, April 21, 1977.
10. Guide for Collection, Analysis, and Use of Urban Stormwater Data,
American Society of Civil Engineers, New York, 1977.
11. Mueller, J.A. and A.R. Anderson, A Mass Balance Method for Estimating
Combined Sewer Runoff and Overflow Quality From Sewage Treatment
Plant Data, Environmental Eng. and Science Program, Manhattan
College, Bronx, New York, 1977.
12. Grizzard, T.J., J.P. Hartigan, et. al., Characterizing Runoff
Pollution - Land Use Relationships in Northern Virginia's Ocoquan
Watershed, Presented at MSDGC-AMSA Workshop on Water Quality Surveys
for 208: Data Acquisition and Interpretation of Non-Point Runoff,
Chicago, April 20-22, 1977.
13. Jackson, T.J., R.M. Ragan, W.N. Fitch, Test of Landsat-Based Urban
Hydrologic Modeling, Journal of the Water Resources Planning and
Management Division, ASCE, Volume 103, No. WR1, May 1977.
14. Marsalek, J., Instrumentation for Field Studies of Urban Runoff,
Environmental Management Service, Canada Center for Inland Waters,
Project 73-3-12, Burlington, Ontario.
15. Au, Shane, Hoel, Fundamentals of System Engineering, Probabilistic
Models, Addison-Wesley, Reading, Massachusetts, 1972.
16. Wells, Dan M., et al., Variation of Urban Runoff Quality and
Quantity with Duration and Intensity of Storms - Phase III, Volume
4 - Project Summary, Texas Tech University, Water Resource Center,
WRC-75-4, Lubbock, Texas, December 1975.
17. Urban-Runoff Data Base, University of Florida for U.S. Environmental
Protection Agency, Project No. 68-03-0496, In Progress.
6-21
-------
18. Summary of Runoff Data from Durham, N.C., Milwaukee, Wis., Lubbock,
Texas, and Washington D.C., Internal Memorandum, Hydroscience,
Inc., 1976.
19. Shelly, P.E., Data Collection to Evaluate 'Management Technique,
Presentation at U.S. Environmental Protection Agency Conference on
208 Planning and Implementation, Reston, Virginia, March 16-17,
1977.
20. Shelly, P.E., Sediment Measurement in Estuarine and Coastal Areas,
EG£G Washington Analytical Services Center, Inc., for National
Aeronautics and Space Administration, NASA CR-2769, December 1976.
21. 208 Areawide Assessment Procedures Manual, Volume II, Appendix D,
EG£G Washington Analytical Services Center for U.S. Environmental
Protection Agency, EPA-600/9-76-014, Cincinnati, July 1976.
22. Lager, J.A., W.G. Smith, Urban Stormwater Management and Technology
An Assessment, U.S. Environmental Protection Agency, EPA-670/2-74-
040, 1974.
23. Wullschleger, R.E., A.E. Zanoni, C.A. Hansen, Methodology for the
Study of Urban Storm Generated Pollution and Control, U.S. Envir-
onmental Protection Agency, EPA-600/2-76-145, August 1976.
24. Short Course Proceedings Applications of Stormwater Management
Models, 1976, University of Massachusetts for U.S. Environmental
Protection Agency, EPA-600/2-77-065, March 1977.
25. Shelley, P.E., and G.A. Kirkpatrick, An Assessment of Automatic
Sewer Flow Samplers, U.S. Environmental Protection Agency, EPA-R2-
73-261, June 1973.
26. Linsley, R.K., A Manual on Collection of Hydrologic Data for Urban
Drainage Design, Hydrocomp, Inc., Palo Alto, California, 1973.
6-22
-------
CHAPTER 7
EXAMPLE STORMWATER ANALYSES
A variety of analytical tools and techniques for evaluating stormwater
runoff and overflow pollution problems are developed in this manual. In
this section, these tools are applied in two example problems: Salt Lake
City, Utah and Kingston, New York. The physical and problem settings are
real, however some liberties have been taken to enhance the basic objective
of illustrating the practical application of the methodology described in
this manual.
The examples selected provide geographic diversity and a range of water
quality problems to be addressed. Salt Lake City is located on a river while
Kingston is adjacent to an estuary. Salt Lake City is served by separate
sewers or natural conveyance while Kingston has a combined sewer system. The
principle problem analyzed in the Salt Lake City example is the impact on
dissolved oxygen levels necessary for aquatic life, while the problem ad-
dressed in Kingston relates to bacteria organisms as they affect bathing
beaches. Both examples represent a preliminary assessment. Not all the
questions raised are fully answered, and further studies may be appropriate.
The examples do demonstrate, however, the utility of the methodologies pre-
sented for obtaining a first evaluation of the stormwater problem in urban
areas.
7.1 Salt Lake City, Utah
Salt Lake City is located on the Jordan River just upstream of the
Great Salt Lake. The principal features of the study area are outlined in
Figure 7-1. The Jordan River flows north from its origin at Utah Lake
through predominantly agricultural and suburban land until it reaches the
Salt Lake City area. A surplus canal in the southern portion of the city
was constructed to divert flood flows away from the city, and route them
more directly to the Great Salt Lake. It serves to regulate the flow in the
lower Jordan River. About 17 miles below the city the Jordan River ends in
an undeveloped marsh area near the Great Salt Lake. The principal use of
this undeveloped area is as a wildlife and waterfowl habitat. The Jordan
River in this reach presently contains a number of rough fish, but there are
proposals to stock the river with game fish to provide additional recreation-
al benefits. The analysis presented in this section is directed towards
evaluating the effect of urban runoff from Salt Lake City on the dissolved
oxygen concentration in the Jordan River, and the possible subsequent limi-
tation of beneficial uses of the waterway. The system is represented in a
7-1
-------
SALT LAKE CITY
DRAINAGE AREA
MODELED
MILEPOINT 0-
OQUIRRH
MOUNTAINS
-JORDAN RIVER
DRAINAGE BASIN
WASATCH
MOUNTAINS
LEGEND:
• SOUTH SALT LAKE CITY STP
A AIRPORT RAINGAGE
A SALT LAKE CITY RAINGAGE
LOCATION MAP
SCALE:
01 2345
MILES
FIGURE 7-1
SALT LAKE CITY STUDY AREA
7-2
-------
simplified manner by assuming that all stormwater runoff from Salt Lake City
enters the Jordan River just downstream of the surplus canal. This example
assessment demonstrates techniques useful for areas with separate or unsewer-
ed runoff, dissolved oxygen problems, and river or stream receiving water
systems.
7.1.1 Rainfall Analysis
Hourly rainfall records for the Salt Lake City study area are generated
by combining the records from the main station in Salt Lake City (Gage
427598) and the Airport (Gage 427603). Equal weights are used to produce a
record from 1950 through 1965 which is essentially the average of the two
individual gages (See Section 5.1.4.1). This procedure helps account for
the areal variation of rainfall over the study area.
A synoptic rainfall analysis (using program SYNOP) is performed on the
combined record to calculate the storm statistics. A minimum interevent
time of 13 hours is found to yield a coefficient of variation of time between
storms (v.) nearly equal to 1.0 throughout most of the year and is thus
chosen for event definition. The results are shown in Figure 7-2. Figure
7-2 is used to determine the appropriate rainfall statistics for the pre-
liminary period of interest, July-September. These are shown in Table 7-1.
TABLE 7-1
SUMMER RAINFALL STATISTICS (JULY-SEPTEMBER)
SALT LAKE CITY
Coefficient of
Mean Variation
Intensity I = 0.030 in/hr v. = 1.10
Duration D = 6.5 hr v* = 1.10
Unit Volume V = 0.16 in v = 1.60
Time Between Storms A = 175 hr vY = 1.10
o
The predominant characteristic of Salt Lake City summer rainfall is its
infrequency. The mean time between storms (measured from storm midpoints),
A = 175 hours (7.3 days) indicates that an average of 12.6 storms occur
during the three month summer period.
7.1.2 Drainage Basin Characterization
Previous studies (1,2) have characterized the predominantly urbanized
portion of Salt Lake City which drains into the Jordan River near, or down-
stream of, the surplus canal. The area, outlined in Figure 7-1, is estima-
ted to have a total area of 13,830 acres (21.6 square miles) and an average
ratio of runoff to rainfall, Ry = 0.46. The drainage basin is served en-
tirely by separate or unsewered runoff conveyance. The population density
in the area is approximately 6,000 persons/square mile.
7-3
-------
MEAN VARIATION
— 0.06
2: 0.05
X
^ 004
5 003
" 0.02
H
001
Q
STORM INTENSITY
_
- -_
_ _/^ ~\
^-^to—Of^^ •" ^^^t
— ••"•"^"""^^^ ^^^^^""^
—
1 1 1 1 1 1 1 1 1 1 II
2.5
2 0
1 5
^
1.0
0 5
n
-
_
m *- m _^Av >^s«
i — ^ ^ ^•r^^'**^^ ^^"^ ^^ 0
«T
—
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
12
— 10
(T
X 8
— 6
0 4
2
/-\
DURATION
-^-A. A*
•-"•"^ V^ A /
A /V
H- \^y
•*• «
—
1 1 1 1 1 1 1 1 1 1 ii
2.5
20
^•o 1.5
1.0
0 5
O
—
—
_^
~ » • • ^r-^^"*^^ r- •
—
i i i i i i i i l l l i
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
0 3
—
? 0.2
> 0 1
Q
DEPTH
l\
~ J ^-^^S*\
-•-*r ^•r~^f
—
1 1 1 1 1 1 1 1 1 1 1 1
2 5
2 0
v* ' 5
1 0
0 5
n
-
-
-,, ,^,>>^*V ^
_ ^k>>«^ ^* ^^>^*
-
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
300
~ 25O
IT
X 200
""" 150
100
5O
/-.
TIME BETWEEN STORMS
_
«
A ^^
y V^^^^
m 1 t t i^ ^
i i i l i 1 1 1 l 1 1 1
2 5
2 0
-.« 1 5
1 0
0 5
O
-
—
-
^ y^Su
— ^"^T^^ t •• • »
-
1 1 1 1 1 1 1 1 1 1 1 1
123456789 10 II 12 123456789 10 II 12
MONTH MONTH
FIGURE 7-2
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
SALT LAKE CITY
COMBINED RECORD, STATIONS 427598 AND 427603
7-4
-------
7.1.3 Runoff Quantity
The mean runoff volume from the Salt Lake City area (V ) is calculated
from Equation 3-12 as follows:
VR = 0.027 RyVA
= (0.027 MG/acre-in)(0.46)(0.16 in) (13,830 acre)
= 27.5 MG
The mean runoff flow (Q_) is estimated from Equation 3-13:
K
QR = RVIA(D/D )
All of the variables in this equation are known at this point, except D ,
the mean duration of runoff events. Using the estimating procedure out-
lined in Section 5.2.2, D is determined as follows:
K
D = mean rainfall event duration =6.5 hr.
V, = estimated initial abstract = 0.10 inches
d
Vd/DI= 0.10 in/(6.5 hr x 0.030 in/hr) = 0.51
D /D = 0.45 (from Figure 5-23, reproduced as Figure 7-3)
D = 0.45 x 6.5 hr = 2.9 = 3.0 hr.
e
2
PD = population density = 6,000 people/mi
W« = width of unit hydrograph at 25 percent of peak
= 7 hr (from Figure 5-24, reproduced as Figure 7-4)
Dn = D + Woc = 3.0 + 7.0 = 10.0 hr
R e 25
Therefore, Q is estimated as:
K
Q_ = (0.46)(0.030 in/hr)(13830 acre)(6.5 hr/10.0 hr) = 124 cfs
K
Note that the correction factor for the time of travel flow attenuation
(D/D ) is equal to 0.65. The flow rates estimated to occur from different
size storms are tabulated in conjunction with the runoff loading rates in
the following section, and are adopted for subsequent analysis because they
are consistent with the general level of flow changes observed to occur in
the Jordan River. The average summer contribution of runoff flow from the
study area (including both rain and non-rain periods) is estimated as:
7-5
-------
CO
CO
UJ
o
X
UJ
cc
<
u.
UJ
UJ
04 06
DI
0.6 1.0 1.2 1.4 I 6 1.8
'INITIAL ABSTRACTION
MEAN STORM DEPTH
2.0 2.2 2.4
FIGURE 7-3
ESTIMATE OF MEAN DURATION OF RAINFALL EXCESS
7-6
-------
o:
cr
UJ
z
o
z
<
-------
Q0 = QR(DR/A)
= 124 cfs (10.0 hr/175 hr)
= 7.1 cfs
7.1.4 Runoff Pollutant Loads
To determine wet weather organic pollutant loading rates from the Salt
Lake City area, the following estimates (1) of average runoff concentrations
are used:
The carbonaceous biochemical oxygen demand,
CBOD5 = 40 mg/1
The total kjedahl nitrogen,
TKN =2.0 mg/1
The CBOD,. is converted to an ultimate oxygen demand by assuming a factor
of 1.5:
CBOD = 1.5 (40 mg/1)
= 60 mg/1
Assuming all of the TKN is oxidizable in the Jordan River, and using the
stochiometric factor of 4.57:
NBOD = 4.57 (2.0 mg/1)
=9.1 mg/1
The mean runoff loading rate (during storms) is calculated as:
WR = 5.4 c QR
For CBOD : W_ = 5.4(lb/day)/(cfs-mg/1) (60 mg/1)(124 cfs)
U K
= 40,200 Ib/day
For NBOD : Wn = 5.4(lb/day)/(cfs-mg/1) (9.1 mg/1)(124 cfs)
U K
= 6,100 Ib/day
The average summer loading rate (including rain and non-rain periods) is
estimated as:
Wo - WR(DR/A)
7-8
-------
or CBOD : W = 40,200 Ib/day (10.0 hr/175 hr)
= 2,300 Ib/day
For NBOD : W = 6,100 Ib/day (10.0 hr/175 hr)
= 350 Ib/day
These values may be used to estimate the total seasonal load.
The loading rate which occurs during storms of various frequencies
is estimated by assuming v = v. = 1.10, and that runoff concentrations are
constant. Figure 7-5 is tnen used to estimate the factor by which the mean
flow or loading rate is multiplied to estimate the flow or loading rate of
storms with different likelihoods of occurance. The result is tabulated in
Table 7-2.
An additional pollutant load considered in this assessment is that gen-
erated by the South Salt Lake City municipal treatment plant. The organic
load from the plant is estimated as (1):
CBOD = 1900 Ib/day
NBOD = 2000 Ib/day
Note that the long term average stormwater loading rate in the summer (W )
is nearly equal to the treatment plant load in the case of CBOD, and about
6 times smaller than the treatment plant load in the case of NBOD. However,
the intermittent nature of the stormwater loads, which occur only about 6
percent of the time (DR/A = 10/175 x 100% = 5.7%), results in the runoff
loading rates being much higher than the municipal loading rate during
storms. The impact of these loads on the dissolved oxygen concentration of
the Jordan River is now evaluated.
7.1.5 Receiving Water Response
The critical step in the evaluation of stormwater problems is the
assessment of impacts in the receiving stream. A simplified model of the
Jordan River is used to estimate the dissolved oxygen response during wet
weather periods. A one segment model with uniform geometry, flow, and
reaction kinetics is applied. The municipal treatment plant discharge and
all of the stormwater runoff are assumed to enter the river segment at the
upstream end. The resulting dissolved oxygen deficit is due to the carbon-
aceous and nitrogenous biochemical oxygen demand loads in the municipal
discharge, the stormwater runoff, and that transported into the river
segment from the upstream boundary condition; as well as the remaining
portions of the initial dissolved oxygen deficit present in the stormwater
and the incoming river flow. As outlined in Section 3.5.2.1, the Streeter-
Phelps stream solution is determined and modified for dispersion to estimate
the peak dissolved oxygen deficit which passes a location following a storm.
7-9
-------
UJ
z
UJ
>
o
o
cr
o
i
i-
cr
o
o
o:
O I 23456
MULTIPLES OF THE MEAN RUNOFF RATE
FIGURE 7-5
CUMULATIVE FREQUENCY FUNCTION OF RUNOFF RATES
SALT LAKE CITY SUMMER STORMS
7-10
-------
0)
+->
rt
OS
W
c
*o
rt
o
tl_)
o
c
OS
3
Q
O
E M
fi Z
O
CO
W)
c
•H
H 3
3 Q
Q 0
PD
u
o o
PH CO
«M (50
^M £
O -H
C M
, — v
X
rt
T3
J2
o o o o
LO O LO LO
*O i~~( l~^ vO
»t *\ »l is
to ^o o> ^f
f— 1
f — N
X
rt
^
f>
r~^
I/)
«W
U
V /
o o o o
o o o o
i— 1 CN tO LO
^^ ^^ ^^ ^D
CN ^" VO CTl
'rt rt 00 00
f**% (NJ O^ O^
t— t i-H CM
CN
I
CO
H
U
PJ
co
i
LO
LO
CN
CO (D
•-I ^= C
•M 4-*
•-I O
o
o
o
o
o
CN
O > -H 4-1
co ci rt
OS
O rt
-------
Parameters and coefficients used in the analysis are taken from the
recent study of the Jordan River (1). The flow entering the segment of in-
terest in the Jordan River is regulated by the surplus canal to an average
value of about 150 cfs during the summer. The total wet weather stream flow
(Q , cfs) is thus 150 cfs plus the runoff flow. The river depth (H, ft) and
velocity (U, ft/sec) are estimated from the flow (Section 5.4.1.1) as
follows :
H = 0.25 (QT)°t5
U = 0.20 (QT)°>4
The water temperature is 20 °C and the oxygen reaeration rate (K , per day)
is estimated using the O'Connor- Dobbins equation:
K = 12.96 U1/2/H3/2
The reaction rate for CBOD is K = 0.30 per day, the CBOD oxidation rate is
K, = 0.25 per day (a little smaller than K to account for some settling of
CBOD without oxidation), and the reaction rate and oxidation rate of NBOD
is K =0. 15 per day.
2
The longitudinal dispersion coefficient (E, mi /day) is derived from
Equations 5-14 and 5-15, and a knowledge of the average channel slope, s =
0.0004:
0.0605
U3/2 H11/4
The Jordan River geometry, transport, and reaction parameters are calculated
and tabulated for each storm size in Table 7-3.
To select a particular location in the River for analysis, the location
of the critical deficit (X , miles) is calculated:
U Kr °
XC = irnr ln (T~]
r a a
The value calculated for X ranges from 28 miles for the 50th percentile
storm to 48 miles for the 90th percentile storm. Since the length of the
River before the confluence with the Great Salt Lake is only 17 miles, it
is clear that the dissolved oxygen deficit is continuing to increase as
the River flows into the Lake. A location 2 miles from the mouth of the
Jordan River (X = 15 miles) is thus chosen for the analysis, to avoid the
particularly marshy area near the Lake.
The final factor required for the dissolved oxygen deficit calculations
is the reduction in the peak concentration due to dispersion. This is
determined from Figure 3-11 (reproduced as Figure 7-6) and the dimensionless
factor :
7-12
-------
rt
rt
o
rt
(N
LO
to
oo
X
rt
T3
HJ
§
I— I
u
<
U4
os
LO
00
a\
o
to
01
00
rt
to
00
o
(N
to
o
to
X
rt
to
f^*
UJ
CO
t$
E- O
OS H
o
a.
M OS
§ w
S I?
H 5
.
X1 CJ
OS Z
£— i hH
UJ OS
o B
UJ
OS
UJ
^»
1— I
OS
z
Q
D—
o
t~}
/— \
X «H
*. — '
t^ 1-1 r-s to
tO rt rt LO
/ — ^
10
E*-* c^_f
^y o
V '
rt rt OO OO
(N t-^ rt rt
(N (N tO rt
U
0
O
a. «H
+-> (H O
o w a> c
« e •M 3
&, C rt os
II
0)
f-t
3
rt
to LO LO to
II
fj
K>
X
rt
(H
p(
LO
i-(
C5
II
C
UJ
C
0)
4-1 C
o rt
•P H
§
tj (T> Oi
f-< V
0) -J
D.
o
LO
C
rt
a>
s
rt
VD
O
OO
o
13
O
(J
rt
o
u,
7-13
-------
NOTE.
E=DISPERSION COEFFICIENT (ml*/DAY)
U=STREAM VELOCITY (ml/DAY)
dr=DURATION OF STORM (DAYS)
= TIME SINCE BEGINING OF STORM (DAYS
= U (t-dr/2)
= DISTANCE DOWNSTREAM FROM STORM LOAD (ml )
COS
02 05 1.0
DIMENSIONLESS FACTOR =
100
FIGURE 7-6
EFFECT OF DISPERSION ON POLLUTANT
CONCENTRATION AT MIDPOINT OF STORM PULSE
7-14
-------
2Et
2 2
cT IT
The time of passage of the pulse midpoint to X = 15 miles is (from the begin
ning of the storm) :
Assuming a 10 hour runoff event (d = 10), t ranges from 0.61 days (90th
percentile storm) to 0.74 days (50th percentile storm). The resulting values
of « range from 1 x 10" to 2 x 10~ , all well below the beginning of the
abscissa in Figure 7-6.
Even if the value of
-------
DO = Cg - DT
The estimated component deficits and resulting dissolved oxygen concen-
trations following summer storms are calculated at milepoint 15 of the
Jordan River and are tabulated in Table 7-4 and shown graphically in Figure
7-7. A recommended Cl) dissolved oxygen guideline of 5.5 mg/1 for this
portion of the River is also indicated. The dissolved oxygen concentration
is seen to fall below the guideline following about 40 percent of the storms
(5 events per summer). The 90th percentile storm results in about a 1 mg/1
violation at milepoint 15.
The components of the dissolved oxygen deficit are also shown in Figure
7-7. The upstream boundary loads and the remains of the initial deficit are
considered basically uncontrollable for the current analysis and are grouped
together. As indicated in Table 7-4, the decrease in the upstream boundary
load impact associated with larger storms (due to the dilution by the larger
storm flows) is balanced by the larger remaining portion of the initial
deficit during the large storms (due to the shorter time of travel and small-
er K during the large storms); and the uncontrollable portion of the deficit
remains relatively constant. The impact of the South Salt Lake City treat-
ment plant is shown to be quite small, particularly during large storms. The
dissolved oxygen deficit resulting from the organic pollutant load in the
stormwater runoff is the principal factor, increasing from about one half the
total deficit during the 50th percentile storm, to about two thirds the total
deficit during the 90th percentile storm. This preliminary assessment of the
Jordan River indicates that stormwater loadings may in fact result in small
or moderate dissolved oxygen problems during large summer storms.
The resulting dissolved oxygen deficit (at milepoint 15) is shown to in-
crease with larger (less frequent) storm runoff rates. However, caution
should be exercised before concluding that the deficit associated with the
90th percentile runoff rate is not exceeded by 90 percent of the summer
storms. This is true only if the resulting deficit increases monotonically
with increasing runoff rates. If larger runoff rates are calculated to re-
sult in less deficit, the direct mapping of probability no longer applies.
Situations where this may occur include:
1. Where the decrease in the dissolved oxygen deficit associated
with dilution of continuous sources (municipal treatment plant
loads, background BOD, etc.) at higher runoff rates is greater
than the deficit increase associated with the larger storm
runoff load. There are in fact situations where larger storms
improve the pollution problem. This usually corresponds to areas
where dry weather, continuous problems are more severe than wet
weather impacts.
2. Where the time of travel decrease associated with larger storms
does not allow the dissolved oxygen sag sufficient time to develop.
This is not a problem for first order pollutants (e.g., coliforms)
where the maximum concentration occurs at the point of discharge.
These problems can be checked by calculating the impact for a wide
range of runoff rate frequencies. When the runoff rate-stream
7-16
-------
U4
_3
CO
<
Q
W
!-J
O
to
to
H
w
to
O
&< 2
tO O2
PJ O
02 H
to
m 02
I I
o
to
02
>
rH
02
Q
02
O
.-H Ot)r~,
0 XO
(f> X Q
10 O ^— '
Q
P
rH .H
cd 0 H
P -H Q
O <4-l
Q
6
O
!H
P
•H
U
•H
0)
Q
C
60
X
o
*"O
CD
^n
I— 1
o
(/)
•H
Q
i-H P
cd -H
•H O
i i ,_j
C CO
I-H Q
CTl rt
• •
LO LO
rH \O
• •
to to
1^ 0
LO t~-
o o
T3 ^D CM
cd
E 0 Z
cd J
rg CM
•
O O
CD
fH X
P fn
10 cd
^D C CT, CTl
3
O U
CO
LO •n-
• •
o o
to oo
2 tp ^
i-H i-H
0 0
cd tp
5 O
E c
O 02
P U
to
to 00
•
i-H i-H
r-* \o
i-H CD
cd 60 Z
C^i JH
0 0
• •
o o
•H Cd
O ^2
•H CJ I-H CTl
C w
3 -H U
S Q
e
cd ^ 02
rO CD ?H
Z cd o
£ O C
•U fi rH 3
(DO)
* o i -H
to
s c
S o>
O > CD
4-* *H -M
(/) £_3 ftj
02
(l 1 ^*
O Cd 14H
P H O
c c
d) w 3
O IO 62
!H 0
0) nJ
CL.
rH 0
. .
O O
to LO
^o ^±
/ X
c
cd
*»•
"— '
0 'tf-
LO \D
OO
•
^3-
-------
cr
E
X
o
Q
UJ
>
_)
O
CO
CO
10 0
9.0
8 0
7 0
6 0
5 0
4.0
3.0
2.0
1 0
0 0
r-
SATURATION
rz
WATER QUALITY
GUIDELINE
xx
L£GEHD.
r~\ UPSTREAM BOUNDARY AND INITIAL DEFICIT
0 MUNICIPAL TREATMENT PLANT
Q STORMWATER RUNOFF
I I I I
0.0
1.0
2.0 1
3.0 Si
4.0
50
60
50 64 80 90
PERCENT SUMMER STORMS LESS THAN RUNOFF RATE
| 1 I I
63 45 25 13
AVERAGE NUMBER SUMMER STORMS
GREATER THAN RUNOFF RATE
LJ
O
UJ
O
O
UJ
>
o
CO
7.0 ^
8 0
9.0
FIGURE 7-7
MINIMUM DISSOLVED OXYGEN AT MILEPOINT 15
OF JORDAN RIVER FOLLOWING SUMMER STORMS
7-18
-------
concentration relationship deviates significantly from the mono-
tonic assumption, long term simulation is required for a complete
probabilistic assessment of wet weather stream impacts.
7.1.6 Stormwater Control Alternatives
A variety of control alternatives are available to reduce the wet
weather impacts identified. Two possibilities: storage and management
practices, are briefly analyzed in this section.
Storage for stormwater runoff could be provided in the Salt Lake City
area with a series of detention ponds, lagoons, or by utilization of existing
irrigation canals. The storage performance may be estimated using the pro-
cedure outlined in Section 3.6.1.2. Assuming the total capacity of the
retention system is V = 40 MG and the average release rate from storage is
n = 25 MGD, the calculations proceed as follows:
VD/V_ = 40 MG/27.5 MG
D K
= 1.45
nA/Vn = (25 MGD)(7.3 days)/27.5 MG
K
= 6.6
Vn/V_. = 1.40 (Figure 3-17, reproduced as Figure 7-8)
b K
fv = 0.40 (From Figure 3-18, reproduced as Figure 7-9. Note that
VVR = VV = 1>60)
Assuming the presence of a small to moderate first flush, the storage
performance is improved such that,
fv = 0.35 (From Figure 3-19, reproduced as Figure 7-10)
The storage system is thus estimated to capture about 65 percent of the
stormwater load. Furthermore, given the coefficient of variation of storm
volumes (v = v = 1.60), the fraction of storms totally captured may be
estimated. As indicated in Figure 7-11, 79 percent of the summer storms are
completely captured. Therefore, only an expected 2.6 storms per summer have
any direct overflow. To develop a modified version of Figure 7-7, indicating
the expected dissolved oxygen condition in the Jordan River after the im-
plementation of storage treatment, more detailed simulation analyses in-
corporating storage, overflow, treatment in the basins, release from storage,
etc., are appropriate.
The cost of the retention system is estimated by assuming that four 10
MG earthen basins are used to provide the storage capacity. Figure 5-56 (re-
produced as Figure 7-12) estimates the cost of each basin at about $200,000
(June, 1975 costs). The annual operation and maintenance cost is estimated
to be similar to the covered basin, and Figure 5-57 (reproduced as Figure
7-13) is used. The hours of operation per year are estimated as:
7-19
-------
5.0
4.0 -
UJ
O
Ul
o
-------
SALT LAKE
CITY EXAMPLE
0 5
I 0
15 20 25 30 55
EFFECTIVE STORAGE CAPACl'
MEAN RUNOFF VOLUME
40
4 5
50
FIGURE 7-9
DETERMINATION OF LONG TERM STORAGE DEVICE PERFORMANCE
7-21
-------
MAGNITUDE OF FIRST FLUSH EFFECT!
NONE
SMALL
MODERATE
LARGE
SALT LAKE CITY EXAMPLE
Q2 O3 OA O5 06 Q7 Q8
fv (NO FIRST FLUSH)
[FRACTION OF LOAD NOT CAPTURED]
[(ASSUME NO FIRST FLUSH EFFECT)J
FIGURE 7-10
IMPROVEMENT IN LONG TERM STORAGE DEVICE PERFOMANCE
DUE TO FIRST FLUSH EFFECT
7-22
-------
UJ
2
UJ
<
O
w
a:
o
(O
5
o:
o:
U
Q.
0 1142 3 4 5 6 7
MULTIPLES OF THE MEAN RUNOFF VOLUME
FIGURE 7-11
CUMULATIVE FREQUENCY FUNCTION OF RUNOFF VOLUMES
SALT LAKE CITY SUMMER STORMS
7-23
-------
5,000
20
SALT LAKE CITY EXAMPLE
1 I I
5 10 50
STORAGE VOLUME (MILLION GALLONS)
50,000
lO.OOOco
5,000
Oui
--»-
— til
"
a.
I,00(X
o
500
ZOO
100 aoo
FIGURE 7-12
CAPITAL COST-STORAGE BASINS
(JUNE 1975 COSTS)
7-24
-------
200,000
100,000 -
CO
tr
o
O
OT
O
o
-------
DR
Hours of operation per year = (—) 8766 hr/year
where D_ and A are chosen to be representative of yearly averages (from
Figure 7-2).
Hours of operation per year = , (8766 hr/year)
=810 hr/year
The annual operation and maintenance cost for each basin is thus estimated
as $25,000 (June, 1975 cost). The total cost of the storage system (June,
1975 cost) is therefore:
Capital Cost: $800,000
Operation and Maintenance: $100,000/year
These are rough, order of magnitude estimates, intended only to provide
perspective on the stormwater treatment issue.
An alternative to the storage system analyzed is the implementation
of management practices to reduce the pollutant concentration of runoff from
the Salt Lake City area. These practices could include erosion control,
localized detention through landscaping or small ponds, modified street
sweeping, etc. If a 35 percent reduction in the CBOD and NBOD concentrations
is achieved, the dissolved oxygen concentration following summer storms is
improved as indicated in Table 7-5.
TABLE 7-5
WET WEATHER DISSOLVED OXYGEN
AFTER IMPLEMENTATION OF MANAGEMENT PRACTICES
Expected Number of
Percent of Storms Storms Per Summer
Less Than Given Greater Than Given DO
Runoff Rate Runoff Rate (mg/1)
50 6.3 6.4
64 4.5 6.1
80 2.5 5.7
90 1.3 5.5
The 90th percentile storm is seen to just meet the water quality guideline.
A determination of the feasibility and cost of management practices requires
a more detailed evaluation of local conditions and current practices in the
Salt Lake City area.
The initial assessment of wet weather impacts in the Jordan River in-
dicates the possible presence of a small but potentially manageable storm-
water related dissolved oxygen problem near the River mouth. All available
7-26
-------
water quality records for the lower Jordan River should be reviewed to deter-
mine whether available data is generally consistent with the assessment. The
analysis which has been performed will help to focus attention on the most
appropriate station to examine. A complete preliminary assessment would
perform a similar estimate of receiving water impact for other contaminants
which may be significant, and perhaps for other seasons where this may be
pertinent.
If the magnitude and frequency at which violations of standards are
estimated to occur are judged to be significant and if control is judged to
be feasible and justifiable both technically and economically, the prelim-
inary assessment would normally be supplemented by more detailed studies.
Such studies can be organized more efficiently based on the perspective
gained from the results of the preliminary assessment.
7.2 Example Stormwater Analysis: Kingston, New York
Kingston, New York is a small city of approximately 28,000 people loca-
ted on the Hudson River Estuary about midway between New York City and
Albany. The City and its principal drainage and receiving water systems are
shown in Figure 7-14. Runoff from the major portion of the city drains
through a combined sewer system into Rondout Creek, a small tributary of the
Hudson River. Separate runoff also drains north into the Esopus River or
west directly into the Hudson.
The Hudson River and Rondout Creek are tidally influenced with about a
three and one half foot range in stage from low to high tide. Bathing
beaches are located at Kingston Point and Point Ewen, upstream and downstream
respectively of the confluence of the Rondout and the Hudson. The analysis
presented in this section is directed towards evaluating the effect of storm-
water overflows on the sanitary quality of these beaches. The water quality
constituent of interest is fecal coliform bacteria. The system is repre-
sented in a simplified manner by assuming that all of the combined storm-
water load from the areas tributary to the Rondout enters the Hudson River
at the Lighthouse indicated in Figure 7-14. The Hudson River is then
modeled as a one-dimensional tidal river or estuary. This may overestimate
the impact somewhat by negating reaction of the fecal coliform in the
Rondout, however, a reasonable estimate appropriate for an initial problem
assessment is expected. This example demonstrates techniques useful for
combined sewer areas, coliform bacteria problems, and estuarine receiving
water systems.
7.2.1 Rainfall Analysis
Hourly rainfall records for Kingston (U.S. Weather Bureau Gage 304424)
from May 1948 through December 1972 were obtained and analyzed using the
synoptic rainfall analysis program with a minimum dry interevent time of 4
hours. The resulting monthly rainfall statistics are shown in Figure 7-15.
Because the summer months are of particular importance for the bathing beach
analysis, the storm statistics for the period of June-September are estimated
as follows:
7-27
-------
LAND RUNOFF
2
/SEPARATE
'RUNOFF
KINGSTON
SANITARY SEWAGE /
ROOF, STREET DRAINS
KINGSTON
POINT
BEACH
+ RAINGAGE
SEWAGE
TREATMENT
D PLANT
COMBINED RUNOFF
LIGHTHOUSE
POUT
EWEN
BEACH
NEW YORK
STATE
NEW YORK
CITY
LOCATION MAP
FIGURE 7-14
MAJOR FEATURES OF KINGSTON STUDY AREA
7-28
-------
MEAN
VARIATION
10.0
(t
I
5.0
o
II
o
I I I I I I I I I I I
JFMAM JJASOND
JFMAV JJA SOND
050
0.
UJ
Q
n
025
JFMAMJJASOND
JFMAMJJASOND
MONTH
2.0
JFMAM JJASON 0
JFMAMJJASOND
JFMAMJJASOND
2 0
JFMAMJJA SONO
MONTH
FIGURE 7-15
MONTHLY STATISTICAL RAINFALL CHARACTERIZATION
KINGSTON, NEW YORK-STATION 304424
7-29
-------
SUMMER STORM STATISTICS
KINGSTON, NEW YORK
Mean Coefficient of Variation
Intensity I = 0.075 in/hr v. = 1.3
Duration D = 5.0 hr. v, = 1.1
a
Unit Volume V = 0.35 in v = 1.5
v
Time Between . 0_ , . ,
04. A = 80 hr. v, = 1.1
Storms 6
Because a relatively small drainage area is considered, analysis of a single
gage is sufficient.
7.2.2 Drainage Basin Characterization
Three tributary drainage areas are considered for the current analysis.
The first is the combined sewer area tributary to the Rondout. The portion
of Kingston draining directly into the Rondout is 1950 acres. In addition,
because sanitary sewage from the northern Esopus area is pumped to the
Kingston treatment plant, and a considerable number of roof and street
drains are connected to this system, an additional 5 percent of the northern
area (5 percent of 2220 acres = 110 acres) is added to the combined Rondout
tributary basin. The land use breakdown and the percent impervious assumed
for each land use category are as follows:
KINGSTON COMBINED SEWER AREA LAND USE
Land Use Percent of Area Assumed Impervious Percent
Residential 70% 40%
Commercial 10% 80%
Open 20
Net = 36%
The area weighted imperviousness is estimated as 36 percent. Figure 5-20
indicates a wide range of possible runoff to rainfall ratios for a 36 percent
impervious basin, however, R.. = 0.30 is chosen as a typical value.
Because the area tributary to the Rondout is serviced by a combined
sewer system and treatment plant, some of the system capacity is available
for stormwater capture. The total capacity of the system is 10.2 MGD while
the average dry weather flow is 4.5 MGD. This leaves 5.7 MGD (Qj = 10.2 -
4.5 = 5.7 MGD) as the estimated available interceptor capacity.
7-30
-------
The 510 acre area directly tributary to the Hudson River has 60 percent
light industrial land use and 40 percent undeveloped land. Assuming the
light industrial area is 25 percent impervious, the net overall impervious-
ness is 15 percent with an estimated runoff to rainfall ratio, R.. = 0.10.
The entire 510 acre area has separate or natural drainage.
7.2.3 Determine Runoff Volumes
For the area triburary to the Rondout, the mean runoff volume (V ) is
calculated from Equation 3-12 as follows:
VR = 0.027 Ry VA
= (0.027 MG/acre-in)(0.30)(0.35 in) (2060 acre)
= 5.84 MG
The mean runoff flow (Q ) is estimated from Equation 3-13:
QR = lyACD/D^
All of the variables in this equation are known at this point, except for
D , the mean duration of runoff events. Using the estimating procedure
outlined in Section 5.2.2, D_ is determined as follows:
K
D = mean rainfall event duration =5.0 hr.
V, = estimated initial abstraction = 0.10 inches
d
Vd/DI= 0.10 in/(5.0 hr x 0.075 in/hr) =0.27
D /D = 0.6 (Figure 5-33, reproduced as Figure 7-3)
D = 0.6 x 5.0 hr = 3.0 hr = mean duration of excess rainfall
e
PD = population density = 27,630 people/(1950 + 2220 acre)
= 6.63 people/acre
2
= 4,250 people/mi
W_r = width of unit hydrograph at 25 percent of peak
= 4.0 hr (Figure 5-24, reproduced as Figure 7-4)
DD = D + W. = 3.0 + 4.0 = 7.0 hr
K Q Zo
Therefore, Q is estimated as:
K
0D = (0.30)(0.075 in/hr)(2060 acre) (5.0 hr/7.0 hr)
K
33 cfs
7-31
-------
21 MGD
Note that the factor for correcting for runoff attenuation (D/D = 5/7) does
not result in a major change in Q .
R
For the area directly tributary to the Hudson the mean runoff volume is:
VD = 0.027 R.. VA
K V
(0.027 MG/acre-in)(0.10)(0.35 in)(510 acres)
0.48 MG
To determine D , assume again that V, = 0.10 inches, D = 3.0 hr, a low PD of
500 people/mi, W25 = 6.5 hr (Figure 7-4) and DR = 3.0e+ 6.5 = 9.5 hr. The
mean runoff flow is thus:
(0.10)(0.075 in/hr)(510 acres)(5.0 hr/9.5 hr)
2.0 cfs
1.3 MGD
7.2.3.1 Capture by Treatment Plant
For the combined sewer area tributary to the Rondout, a portion of the
overflow is captured by the treatment plant. To estimate this portion,
Figure 3-15 (reproduced as Figure 7-16) is used with:
QT/Q_ =5.7 MGD/21 MGD
1 K
= 0.27
and v = v. = 1.3
The estimated percent captured is thus 20 percent (f.. = 0.80), and the
modified mean runoff volume to the Rondout is therefore:
VD = 0.80 (5.84 MG)
K
= 4.67 MG
The runoff directly tributary to the Hudson is unmodified by capture.
7.2.4 Determine Stormwater Loads
To determine wet weather fecal coliform loadings from Kingston the fol-
lowing concentrations are assumed:
7-32
-------
I 0
10 20 3.0 4.0
"EXCESS INTERCEPTOR CAPACITY
MEAN RUNOFF FLOW
5p°
FIGURE 7-16
DETERMINATION OF LONG TERM INTERCEPTOR PERFORMANCE
7-33
-------
Combined Overflow, c = 800,000 MPN/100 ml
Separate Runoff, c = 10,000 MPN/100 ml
o
Assuming 10 MPN/100 ml = 1 mg/1, the concentrations are converted to
standard units, 8 x 10~ and 1 x 10~ mg/1 respecti/ely. (The actual con-
version value used is immaterial relative to the final receiving water
concentration calculated, so long as the same conversion back to MPN/100 ml
is used.) The mean runoff load from the combined sewer area tributary to
the Rondout is calculated using Equation 3-19:
MR = 8.34 cVR
(8.34 Ib/MG - mg/1)(8 x 10"3 mg/1) (4.67 MG)
= 0.31 Ib fecal coliform
For the separate Hudson area:
MR = 8.34 c VR
(8.34 Ib/MG - mg/l)(l x 10"4 mg/1)(0.48 MG)
0.00040 Ib fecal coliform
The variability 'of both loads are assumed equal to the variability of storm
volumes:
v = v =1.5
m v
7.2.4.1 Receiving Water Response
The response of the Hudson River to the intermittent fecal coliform
loadings from Kingston is modeled using the procedure outlined in Section
3.2.2.2. The characteristics of the Hudson River in the reach adjacent
to Kingston are taken from a study of the Hudson (3):
2
a = cross-sectional area = 115,000 ft
H = average depth = 22 ft
2
E = dispersion coefficient = 1.0 mi /day
A typical flow in the Hudson during the summer months is:
Q = 9000 cfs
which indicates an average freshwater velocity:
U = Q/a = 9000 cfs/115,000 ft2
= 0.078 ft/sec =1.28 mi/day
7-34
-------
The reaction rate for fecal coliform is estimated as:
k = 1.5/day
The mean summer fecal coliform concentration due to Kingston stormwater loads
on the Hudson is calculated using Equation 3-30:
At the assumed discharge location at the Lighthouse (x = 0) , the exponential
term equals one and the mean concentration is:
- VA
c =
(Jam
The m term is first calculated:
m 1 + 4kE/U2
*1 + 4(1.5/day)(l mi2/day) /(I.28 mi/day)2
2.16
Therefore:
0.51 lbs/3.25 days
(0.078 ft/sec)(115,000 ft2)(2.16) 5.39(lb/day)/cfs-mg/l)
9.13 x 10"7 mg/1
To convert back to MPN/100 ml use the conversion outlined in the previous
section (10 MPN/100 ml = 1 mg/1):
c = (9.13 x 10~7) x 108
91 MPN/100 ml
The mean concentration at the Kingston Point Beach (x = -0.5 mile) and the
Port Ewen Beach (x = + 1.0 mile) are then estimated.
At Kingston Point:
VA Ux
~" I\ /UA f~ — •* -i
C = exp(2E (1 + m))
91 MPN/100 ml {expt^ ' ^ J (1 + 2.16)]}
2 (1 mi /day)
33 MPN/100 ml
At Port Ewen:
c = 91 MPN/100 ml {exp[(1'28 mi/da^(+1"° mile) (1 2.16)]}
2(1 mi /day)
44 MPN/100 ml
7-35
-------
This represents the portion of the average summer fecal coliform con-
centration in the Hudson River at the given locations which is due to the
combined sewer overflows. To calculate the mean impact of the separate area
runoff the same procedure is followed using an M of 0.00040 Ibs fecal
coliform instead of 0.31 Ibs. As the mean result is linear with M , the
concentrations are 0.0040/0.31 = 0.013 (or about one percent) of the c's
resulting from the combined overflows. The separate runoff is clearly in-
significant relative to the combined overflow.
Another source of fecal coliform bacteria in the Hudson River is the
relatively continuous discharge from the Kingston municipal sewage treatment
plant. Records from the summer of 1976 (4) indicate an average flow rate
of about 4.8 MGD and a fecal coliform density of 115 MPN/100 ml. This ef-
fective level of disinfection leads to an estimated contribution to the mean
fecal coliform concentration of less than 0.1 MPN/100 ml, and combined sewer
overflows remain the primary source of fecal coliform from the Kingston area.
The final sources of fecal coliform to the Hudson River: loads from other
towns and drainage basins, and the portion transported into the area by the
river flow itself are considered background, and no quantitative estimates
of their magnitude are made.
7.2.4.2 Variability in the Hudson
A method for estimating the intertidal variability of pollutant concen-
trations due to stormwater loads to a tidal river or estuary is presented in
Section 3.5.2.2. The standard deviation of the concentration is determined
using Equation 3-31:
/2 .,2
a + M_.
m R
v =
c
The equation is indeterminent at x = 0, but may be solved for x = -0.5 miles
and x = + 1.0 miles, respectively. Considering only the combined sewer load,
the standard deviation of the load is:
v = v M_,
m m R
1.5 (0.31 Ib)
0.47 Ib fecal coliform
The solution for a is therefore:
c
(0.47)2 + (0.31)2 (3.03
Ib/ft -mi ,(1.28)x, „ 1/2,(1.28)x(2 .16),
a = ~ exP C; n nO K^ ( ^-^ )
(115,000)
7-36
-------
where x = - 0.5 miles for Kingston Point and +1.0 miles for Port Ewen. Note
that 3.03 is a conversion factor to make the units consistent. Figure 3-12
is used to solve for the K term, and the results are:
At Kingston Point (x = -0.5 miles),
a = 1.20 x 10"6 mg/1
120 MPN/100 ml
At Port Ewen (x = 1.0 miles),
a = 1.34 x 10"6 mg/1
134 MPN/100 ml
The standard deviations determined are about 3 or 4 times greater than the
means (c = 33 and 44 MPN/100 ml respectively), indicating the high variabili-
ty which can be expected from intermittent loads of a highly reactive pol-
lutant.
7.2.4.3 Observed Water Quality
Water quality data have been collected at the Kingston Lighthouse,
Kingston Point Beach, and the Port Ewen Beach (5). The results of these
weekly surveys for fecal coliform during the summer of 1976 are shown in
Table 7-6.
The purpose of examining these results is not to accurately calibrate
or verify the model. The model used to represent the system is too simple,
the data upon which it is based (particularly the estimate of overflow con-
centrations) are too uncertain, and the receiving water data are too sparse
for rigorous verification. Furthermore, the model is based on general
summer conditions and not those specific to 1976. Nevertheless, the data
are useful for checking whether the model has properly estimated the order
of magnitude of the impact. The 1976 data and the model results are compared
in Table 7-7.
The predicted coliform concentrations at all three locations are seen to
be of the same order as observed data. Furthermore, both actual data and
calculations show a similar pattern, whereby highest concentrations occur at
the lighthouse, and diminish both upstream and downstream. One may conclude
from this that beach coliform concentrations are predominantly influenced by
discharges to the Hudson River from Roundout Creek, rather than from sources
elsewhere on the Hudson. Combined sewer overflow loads are indicated to be
a significant component of the input from Roundout Creek. The analysis fur-
ther suggests that coliform loads from this source are greater than (perhaps
double) the CSO loads utilized in the calculation. If the levels at the
beaches were considered significant enough to constitute a problem which
warrented further effort, the analyst would be directed to identify the
apparently significant source of additional fecal coliform. Both the actual
concentration in the CSO's would be checked, as well as discharges from the
7-37
-------
TABLE 7-6
HUDSON RIVER FECAL COLIFORM (MPN/100 ml]
SUMMER 1976
Date
5/20
6/2
6/9
6/16
6/23
7/7
7/14
7/28
8/5
8/11
8/18
8/25
9/1
Kingston
Lighthouse
50
380
25
180
170
120
310
40
165
50
400
150
1000
c = 234
a = 262
c
Kingston Point
Beach
50
30
500
5
95
65
45
5
15
55
70
45
35
78
129
Port Ewen
Beach
50
90
25
45
15
65
45
35
60
130
330
40
150
83
84
7-38
-------
TABLE 7-7
HUDSON RIVER FECAL COLIFORM (MPN/100 ml)
Model Estimate Summer 1976 Data
a - a
Location c c c c
Kingston Lighthouse 91 - 234 262
Kingston Point Beach 33 120 78 129
Port Ewen Beach 44 134 83 84
7-39
-------
treatment plant, background loads in the Roundout, or the possibility of
other presently unidentified sources.
7.2.5 Simulation of Fecal Coliform Response
To demonstrate an alternative method for estimating the Hudson River
response to fecal coliform loadings from Kingston stormwater, the broadscale
simulator described in Chapter 4 is used. Hourly loads are generated using
a simple landside simulator with a drainage area of 2060 acres, a runoff to
rainfall ratio (R,r) of 0.30, and an overflow fecal coliform concentration of
800,000 MPN/100 ml. The treatment plant capture is simulated by a continuous
excess interceptor capacity (Q ) of 5.7 MGD. The Kingston raingage (Gage
304424) is used for the summer of 1965 (July 3, 1965 - August 29, 1965).
This 57 day period falls within the maximum allowable for hourly simulation
using the broadscale receiving water simulator.
The receiving water response is simulated using the estuary option with
a freshwater flow, Q = 9000 cfs; cross-sectional area a = 115,000 ft ; dis-
persion coefficient, E = 1 mi /day; and fecal coliform death rate, k = 1.5/
day. Again note that though the loading and calculations are hourly, intra-
tidal transport is not modeled. The input deck for the receiving water
simulation is shown in Table 7-8 (see the users manual in Appendix A for a
description of the input cards).
The printed output from the computer simulation is summarized in Figures
7-17 (a-e). Figure 7-17(a) indicates the input parameters used in the run.
Figure 7-17(b) shows a plot of the first 45 hours of the simulation. The
fecal coliform load (in pounds) is calculated by assuming 10 MPN/100 ml =
1 mg/1. The response concentration is shown as MPN/10Q ml with a base 10 log
scale (the maximum point shown as 3 thus represents 10 = 1000 MPN/100 ml).
Symbol 1 represents the Kingston Point Beach (x = -0.5 miles) while Symbol 2
represents the Port Ewen Beach (x = 1.0 mile). Note that a day with rainfall
is chosen to begin the simulation. This is to avoid too long a period of
zero concentrations at the beginning of the simulation, which may not be
correct if previous days had rainfall. The output shown in Figure 7-17(b)
continues for the full 1367 hours simulated. Figure 7-17(c) shows the next
portion of the output which is a listing of the response concentrations.
This listing also continues for the full period simulated.
The final and most useful output from the simulator are the location
summaries for the receiving water points of interest. These are shown in
Figures 7-17(d) and 7-17(e) for Kingston Point and Port Ewen respectively.
The summary includes a tabulation of the cummulative density function, the
number of times that a selected standard (in this case 200 MPN/100 ml) is
violated, and the average, standard deviation and maximum of the observed
concentrations. Note that the means calculated at Kingston Point (c = 39
MPN/100 ml) and Port Ewen (c = 51 MPN/100 ml) are about 17 percent higher
than those calculated for the typical summer using Equation 3-30, as shown
in Table 7-7. This is consistent with the fact that the total rainfall
during the 57 day period in 1965 (7.15 inches) is 19 percent higher than
would have fallen during the "typical" summer. (Typical total summer rain-
fall = 57 days x V/A = 57 (0.35/3.33) = 5.99 inches),
7-40
-------
n
T>
O
(J
CV
3-
O c/5
^O 2
UJ
J
O
O
Q.
s:
u
a.
o
o
o
o
o
II
LJ
V)
(Z
UJ
I-
UJ
oc.
<
or
o
CD ^ t—
— _ <
"- co o
o
Ul
Q.
(O
7-41
-------
10
ui
2
o
ro
UI
(9
4
O
O
u>
I
O
2
in
LU - O)
_J
O
o
0 Uj
H ±5
o
UI
z
o
7-42
-------
Tl^t
(, .
1.
2,
3.
4,
5.
6.
7.
I ,
9.
1C.
11.
12.
li.
1M .
Ib.
16.
17.
18.
19.
2D.
21.
22.
23,
2i«,
2S.
26.
27,
26,
29,
30,
31,
32,
33,
3i»,
3b,
3b,
i7,
Ab,
39,
'H1 ,
i*l,
"»2,
"43,
14 it .
t'j,
to
"47.
MJ
t«,
JO,
Jl ,
52 .
b?,
bit ,
)*>
jV
,1'fl
,CO
,10
,00
,00
,00
,00
,('0
,00
,f,C
,00
,00
,00
,uO
,oO
,00
00
,(,0
,00
, no
,00
,00
,00
,00
,uO
,00
, j 0
,00
, uO
,1,0
,00
,00
.00
. CO
,00
, 00
, 00
.00
.00
.UO
.UO
.00
. ua
, 03
.00
.1 0
.r o
. 5b
, If3u
, 132b
.lies
. 10i7
,9b05
.S650
.7600
.7U'lO
.GiSl
,S7b2
,t20a
. 14 7 1 b
.W7I*
. i !> 7 7
. 3nl 9
, 31 05
,2^03
.263^1
. 2 '< 0 0
.2127SE
,hi473i;
,77'lSE
.7063E
.6147HE
.S919E
.fmiE
,n9i»8£
,M52bC
,M 1 JSE
•J7C5E
. J:«(.2£
.Lit v_
., oy JE
.s^1 3E
. f '> r ,\ £
.^5?2£
.2n3t£
. 1 "U 3£
. 1 711 jE
. ) fco2c
X
Ol)
00
00
00
01,
oo
22
01
02
02
03
la
03
03
03
03
H3
03
04
03
03
Oj
0 J
Oj
03
Do
03
no
03
03
03
03
03
03
03
D3
02
02
02
02
02
0
-------
FFCliL CL/LtFOrtf SIMULATION -riUO&CN R - GAGE JOHtZ*
DL'.SIIT FU^IIGN Ftrt LOCAllON. X s -G.bO
V/AMMLl - Cat-lFOt-fl
LU^LftllVl 0:-'\.'"irr FUNU1GN CPCT .LE. CJMC)
P'_«Ct«T COrtlt
1965
1
p
3
U
5
f.
7
h
C)
10
11
1?
13
It
15
1C,
17
in
19
2d
.
f'O
rc
CO
I'D
rc
rc
rc
ro
ro
ro
CO
no
rc
CO
ru
00
ro
rt>
ro
rn
rc
00
f'O
0
0
0
0
0
c
c
0
0
u
0
u
0
0
0
0
c
u
c
u
0
0
0
.OUOOE
.OOt-OOL
.UlAOOl
.CuUOOE
.ObCOOE
.O^COCL
.GCOOCL
,6jbl Li
.b^JfeBL
.l^'tjf E.
.11131E
. t)^179t
. 1 % P 3 1 E
,5Vi-b5L
.^imct
.5G12^E
.10b97E
.Ifci'fef.L
.21S>6tE
,3iJfc3E
."»173?E
.bymot
.BH^fetE
00
00
00
00
00
00
01
-6<4
-03
-01
OC
UL
'jl
01
0?
02
02
03
03
04
Oi
U3
03
7
1C
It
NO. OF C'CLUHrtrjCLE- VS HLUkS UF DUiND»ROS
S1ANOAKC = 21'U.O
NO.C'C(.LiHANCLb
\
1
1
NO. HOURS
7
L
3?
NO.
'K VJOL»llu b.tt7 PLRCL^T OF
LUC/MIOfl SLMdfhlr
(•.vr^AC^ cone = o.Jsaa^iE 112 hPN
.j'-rc (;LVJ/.1iOH i 0.1?lt6?E UJ
*1! CUl.C = 0.1?999LE Oi»MPN
FIGURE 7-l7(d)
SIMULATOR OUTPUT
LOCATION SUMMARY-KINGSTON POINT
7-44
-------
KlMbSION FfLAL COLllOhfi SIMULATION -HUDSON R - C-f,GE }bHf O.UOUOUE bb
2 l.nn o.ObUOUE Ob
3 eJ.I'P O.bl'OCOL 00
i» i.cn O.bulfiOE 00
5 -j.cii u.biiOOOl 00
f, /.'H O.bOUOOE 00
7 lb.^0 O.UoUOOE OP
6 lo.OO 0.1
9 2u.''0 0.
10 5U.no 0.tV796L-Oi
11 '(O.'-'O 0.2t-b7(>E OH
1? lu.^O U.U'J6/E 01
13 bb.ro o.jyubHL 01
1H 70.CO 0.11460L 0^
1^ HO.00 U.til7<»2L o<(
16 Bb.HO 0."oi7UE OP.
17 90.CO U.l'-J/ObE O.-l
IB 93. CO U.fV^tiME 0-^
19 93.00 0.*t>379E 0.1
L'O 9/.nn O.^ofatSE O'A
21 9b."0 (j,b.il*i9E b-1
2? 9J.no n./nVBHE 0<
23 9^.''(i u,'
MU.JF OCLUHANLL4 VS HOUKS of OjKATION
TH«T CHUJFOKi'l VIOLAiEU
201). 0
inij'(S NO
1 2
l
-------
LO
vO
en
OS
w
§
CO
k
•*
CM
•
1— 1
w
u
z
o
to
Q
S
1-
z
0
M
H
<
,J
oi
O
n.
O
o
o
CTi
r-.
LT)
OS
o
tL,
o
o
o
LO
o
o
CM
o
en
I „
3
w
PJ
o
i
o
o
CM
to
n
Q
5-1 O
rt 2
00
7-.46
-------
The tabulated cumulative density functions are normalized and plotted
for each location in Figure 7-18. It is of significant interest that the
calculated distributions seem to be very well represented by a gamma distri-
bution with the appropriate coefficient of variation. Although no theory
has as yet been developed which makes this prediction, the possibility
appears promising. This would aid greatly in the formulation of predictions
relative to the frequency distribution of water quality constituents and
problems in tidal rivers and estuaries.
7.2.6 Control Alternatives
The basis for determining the need for stormwater controls is the
assessment of the contribution of stormwater loads to the violation of water
quality standards or guidelines. While no formal standards are available for
the Hudson River beaches, a guideline for fecal coliforms of 200 MPN/100 ml
has been suggested (3). The previous analyses indicate that while this
guideline is met most of the time, there are some periods when it is not.
(If the fecal coliform guideline of 200 MPN/100 ml is indicated as a
median, i.e. the 50th percentile concentration, the guideline is not viola-
ted.) This section examines the stormwater control that might be considered
if the problem was of sufficient magnitude of warrant remedial action.
High rate chlorination units may be considered for overflow locations
in the Kingston combined sewerage system. Assuming for the sake of analysis
that four units are located to treat the entire overflow, each with a
chlorine feed rate (assumed to be 10 mg/1) sufficient to yield a disinfectant
concentration which produces a 99.9 percent kill at a contact time associated
with the mean runoff flow, Figure 3-24 indicates that an average long term
removal of about 93 percent may be expected (without flow proportioning of
the chlorine dosage) . This should reduce mean summer stormwater induced
fecal coliform levels a corresponding amount. As the standard deviation of
the fecal coliform loads is reduced much less than the reduction in the mean,
one may still expect occasional violations of the 200 MPN/100 ml criteria,
though these will occur much less frequently than the currently predicted
frequency of about 5 to 10 percent.
To estimate the order of cost for the four units, the following
calculations are made:
Mean runoff flow at each unit = QD/4
K
21 MGD/4
5.25 MGD
Chlorine feeding rate at each
unit = (10 mg/1) (5. 25 MGD) (8. 34 D mg/
438 Ib/day
Assume from this that each device has a chlorine feed capacity of 500 Ibs/
day, Figure 7-19 indicates the capital cost in June 1975 dollars is about
7-47
-------
UJ
_J
>
z
UJ
O
UJ
ec.
o
o
cc
23456789
MULTIPLES OF MEAN CONCENTRATION, C
LEGEND:
• -SIMULATED AT KINGSTON POINT ( X =-0.5 _M I LE, C = 39 MPN / 100m I, I/c = 3 I )
O-SIMULATED AT POINT EWEN (X=I.OMILE, C = 5 I MPN / 100 m I, I/c = 2.8)
THEORETICAL GAMMA DISTRIBUTION
FIGURE 7-18
FREQUENCY DISTRIBUTION OF SIMULATED HUDSON RIVER
FECAL COLIFORM RESPONSE
(SUMMER,1965)
7-48
-------
500,COC
o
0
I 00,000
to
o
o
r
P 50,000
O
ir
i-
w
z
o
o
10,000
100
KINGSTON EXAMPLE
I
500 1000 5000
CHLORINE FEED CAPACITY (LBS/DAY)
10,000
FIGURE 7-19
CAPITAL COST-CHLORINE FEED EQUIPMENT
(JUNE 1975 COSTS)
7-49
-------
$40,000 for each device. The operation and maintenance cost for each unit
is estimated as follows:
Hours of operation = (No. hours/year)(Fraction of time runoff occurs)
8766 (DR/A)
8766 (7 hrs/80 hrs)
767 hrs/year
Chlorine usage = 767 hrs/year x 438 Ib/day x 1 day/24 hrs
14,000 Ibs/year
= 7 tons/year
Figure 7-20 indicates an annual operation and maintenance cost of $18,000/
year for each device. A rapid mixing basin is also required for each
chlorination facility. Assuming each has a design flow capacity of four
times its mean flow rate (4 x 5.25 = 21 MGD) Figure 7-21 indicates a capital
cost of $30,000 for each basin. Finally, Figure 7-22 estimates the operation
and maintenance cost for each basin at $600/year. The total capital cost
for the chlorination system (chlorine feed and rapid mixing basins) is
therefore:
($40,000 + $30,000) x 4 = $280,000
The annual operation and maintenance cost is estimated as:
($18,000 + $600) x 4 = $74,400/year
These are in June 1975 dollars and should be adjusted for other times
according to the appropriate inflation-cost index. These values should
be viewed only as order of magnitude estimates, without any consideration
of particular local conditions.
This concludes the preliminary assessment of stormwater related fecal
coliform problems at beaches in the Kingston area. The next step is to
decide whether the level of the problem, the projected improvement and
associated cost warrant more detailed study of the overflows, their impacts,
and possible treatment.
7.3 References
1. Jordan River Water Quality Projections for Salt Lake County 208,
Hydroscience, Inc., for Salt Lake County Council of Governments,
Walnut Creek, California, February 1977.
2. Nielson, Maxwell, Wansgard, Inc., Memorandum to Hydroscience, Inc.
Regarding Computation of Stormwater Runoff Flow for Salt Lake County,
1976.
7-50
-------
500,000
o
Q
loopoo
V)
O
o
05 50,000
O
z'
<
10,000
— KINGSTON EXAMPLE
I
I 1 I I
10
50 100
CHLORINE USAGE (T/YR)
500
1000
FIGURE 7-20
ANNUAL OPERATION AND MAINTENANCE COST
CHLORINE FEED EQUIPMENT
(JUNE 1975 COST)
7-51
-------
500
o
o
o
o
100
V)
O 50
o.
<
o
10
I I I I
_li
HOTC
DESIGN FLOW (MGD) BASED
ON 2 MINUTES DETENTION
I I | | | | |
10 50
DESIGN FLOW (mgd)
100
300
j_
4,200 IQ500 37,500
140,000
416,000
BASIN VOLUME ( GAL )
FIGURE 7-21
CAPITAL COST - RAPID MIX BASIN
(JUNE 1975 COSTS )
7-52
-------
10,000
5,000
OT
or
o
o
o
o
_J
z
2
<
1,000
500
100
t,ooo
1,000
soo
IOO
I I
1 I I
KINGSTON EXAMPLE
* NO OF HOURS OPERATION PER YEAR
I 1 I I I I I I I
10
50
100
300
DESIGN FLOW (mgd)
FIGURE 7-22
ANNUAL OPERATION AND MAINTENANCE COST-RAPID MIX BASIN
( JUNE 1975 COSTS)
7-53
-------
Hudson River Water Quality and Waste Assimilative Capacity Study, Quirk,
Lawler, and Matusky Engineers for New York State Department of Environ-
mental Conservation, Albany, New York, December 1970.
Monthly Performance Report, Kingston Wastewater Treatment Facility,
Kingston, New York,
Water Quality Survey, Ulster County Health Department, Ulster County
Health Department, Ulster, New York, 1976.
-------
APPENDIX A
BROADSCALE RECEIVING WATER SIMULATOR
INPUT STRUCTURE
AND
PROGRAM LISTING
-------
APPENDIX A
BROADSCALE RECEIVING WATER SIMULATOR (BRWS)
INPUT STRUCTURE
Card 1 - Title Card
1 8_0
Jackson River June-August 1972
FORMAT (20A4)
TITLE - any alphanumeric text may be used to describe the
simulation to be run
Card 2 - Water Body and Water Quality Constituent Types
1 8 11 18
WBTYP WQCON
FORMAT (A8,2X,A8)
WBTYP - RIVER, for river (or streams)
ESTUARY, for estuaries
WQCON - COLIFORM, for coliform bacteria
BOD, for BOD
DO, for dissolved oxygen
Card 3 - Number of Locations to be Monitored
3 5
NLOC
FORMAT (15)
NLOC - number of locations on the river or in the estuary
where concentration profiles are to be computed:
minimum of one (1), maximum of six (6)
A-l
-------
Card 4 - Milepoints, Relative to Discharge., of Locations to be
Monitored
8 16 48_
LOCS(l) LOGS(2) . . . LOCSCNLOC)
FORMAT (6F8.0)
LOCS(I) - milepoint relative to discharge location (miles)
Convention: (-) for upstream location, (+) for
downstream location
Card 5 - Geophysical Parameters
8 16 24 52 40 48 56 64 72
E A KR U Q KD KA BC CS
FORMAT (9F8.0)
E - dispersion coefficient (square miles/day), Needed
for estuaries only (for rivers - 0)
A - cross sectional area (square feet)
KR - temperature corrected coliform death rate (WQCON =
coliform) or BOD removal rate (WQCON = BOD, DO) (day )
U - freshwater velocity (fps): user may have U calculated
by program by inputting Q (cfs) program uses U = Q/A
Q - freshwater flow (cfs): user may have Q calculated by
program by inputting U (fps), program uses Q = U*A
KD - temperature corrected deoxygenation rate (day" ),
needed only if WQCON = DO
KA - temperature corrected reaeration rate (day" ) needed
only if WQCON = DO. NOTE: KR cannot equal KA
BC - upstream background conditions (mg/1 needed only if
WBTYP = RIVER and WQCON = BOD or DO (for DO reach as
dissolved oxygen, not deficit)
CS - dissolved oxygen saturation value (mg/1) needed only if
WQCON = DO
A-2
-------
Card 6 - Number of Water Quality Standards to be Analyzed
NOSTD
FORMAT (15)
NOSTD - number of water quality standards or criteria to be
analyzed: minimum of one (1), maximum of six (6)
Card 7 - Standards
8 16 24_
TMPSTD(l) TMPSTD(2) TMPSTD(NOSTD)
FORMAT (6F8.0)
TMPSTD(I) - water quality standard or criteria for coliform units
are MPN/100 ml; for BOD or DO units are mg/1
Card 8 - Time Parameters
8 16 24
DT TTOT NDAY
FORMAT (3F8.0)
DT - rainfall observation interval or sampling interval (hours)
TTOT - number of hours to be used for response evaluation
(hours): for streams use maximum distance from discharge
location divided by freshwater velocity, i.e., time-of-
travel to maximum location of interest: for estuaries,
use similar approach but remember dispersion coefficient
E (suggest two hundred (200) hours)
NDAY - length of rainfall record (days) NOTE: maximum length
must meet following constraint:
(NDAY*24 -•- TTOT)/DT < 1,500
Card 9 - Plotting Scales
8 16 2£
PSCAL(l) PSCAL(2) PSCAL(3)
FORMAT (3F8.0)
PSCAL(l) - plot maximum for rainfall record (inches)
A-3
-------
PSCAL(2) - plot maximum for waste discharge record (pounds)
PSCAL(3) - plot maximum for water quality constituent response.
The following scales are recommended:
WQCON PSCAL
COLIFORM 8 (log base 10) MPN/100 ml
BOD 24 mg/1
DO 12 mg/1
Card 10 - Rainfall and Discharge Record Input Device
DVICE
FORMAT (A4)
DVICE - DISK, rainfall record and its associated discharge loads
and flows will be read from disk file (generated by a
drainage basin simulator). If the user selects disk
input, the file should be written as an unformatted
file, 8 bytes per record.
DVICE - CARD, rainfall record and its associated discharge will
be read from punched cards.
if DVICE = CARD
Card(s) 11 and as necessary - Rainfall and Discharge Record
8 16 24 32
RAIN M QW MD
FORMAT (4F8.0)
RAIN - rainfall occurring over the interval between this record
and the next (inches)
M - pounds of coliform bacteria (WQCON = COLIFORM) or
pounds of BOD (WQCON = BOD, DO) occurring over interval
(pounds)
QW - stormwater overflow (cfs); needed only if significant
when compared to freshwater flow; used to obtain proper
mass balance, not used to correct time-of-travel
A-4
-------
MD - pounds of DO deficit discharged in runoff or overflow;
important if DO in discharge is below saturation
Mass should be calculated as follows:
Coliform - M = (QCAt) x 10~8 x 5.39 x 1/24
BOD, DO M = (QCAt) x 5.39 x 1/24
Q = discharge flow (cfs)
C = concentration of coliform (MPN/100 ml), or
= concentration of BOD or DO (mg/1)
t = time interval between rainfall records (hours)
A-5
-------
INTEGER HU'OP.DLUUP.SLOT
. , DUT.FREUi(bU),FRU3V(bO).THRS
. ,AS1ER, BLANK,THKSV
,LINEP163> . ICHARt'J) .MAXJP ( 9 ) .SHIFT t * >
IVTEGER JKSORT « MC'A^
. ,SObS(lb5n)iTOTAL(b),/\VEMb)«SL)(G)iVvlli'j(b),<,'!vi,.'. (&)
. ,PCTILU?3> .SCALEl 3) ,P<9)
DEFINE FILE 10 ( 1 500 , B ,U, 1RE.C )
DATA COLIF/'COLI V , BOU/'BGD '/ , U0/«[)0 •/ , Rl Vt-K/• Ri VE • /
, ESTUR/'ESTUV
DATA DISK/'DISKV « CAKb/'CARO1/
. 80. , fib.i90.,9?.,95.,97.,96.,99.,99.5/
DATA JKSO^T/liO/ i IObORT/1,0/
DATA NPCTL/25/
DATA IREC/I/
DATA JW&/0/,JW3/0/
DATA L'MCN1/0/,*1XCI\IT/0/
DATA Pl/3.mi59/
DATA jN/e/.OJT/fi/, THRS/0/
DATA STDKD/c.300./,ASTER/•**•/,BLANK/' •/
DATA NHRSV/l/.THRSV/O/
DATA ICHAK/»R«,>W«t»U»tlJ'tl2l«l3'«"»t«tbl.tb'/
DATA SHIFT/o121,21,42,42 «^2«t2112,'
c
EXPOF(X,T, KATE) = EXP(-((X-U*T1**2./14.*E*T)+ IvATE*!))
C
CALL SETIA(FRE(J3,00,1,0)
CALL SETiA
-------
josrn
RE: AU UN, 1200) < TMPSTDI isi » ,isT = i,ij'it,TO)
C KEAQ RAINFALL IN JLRVf.L { iKS ) , NO, Or HOUNb TO |.C JSEy F 3K
c RESPONSE EVALUATION, ANJ LENGTH 01 KAINFALL i-.tco^nt jrrsi
READUN,12UO) 3T,TTOT,[J|jAY
C hEAO PLOTTING SCALES
REAliUN, 1200 > PSCAL
C HEAD INPUT JLVICE
REAUUN.1000) DVJCE
IC(U.E9.0. ) U = U/A
.E3.0. ) J = U»A
= A/b2UO./b2flO.
= u*i&.<*
DOPbC r CS-BC
WUlEIOuT.l'jOo) T ITLE.riL I YP.WQCOlM
1500 FORMAT! • 1 • ///20X ,20 A4//aO>, , «RECIt-Vl^(i WftTRK TYHL = • , < A', , «,A , • w«TL>.
. DUALITY CONST1TJLNT = ',£Ai»)
W=?IlE(OUT,lt)10) E.A.U.UTiUD
1510 FOKMATl //16A, ••: = »,F6.2,» SQ.MI/OAY ft = «,(-e.l,' SO.pf.
U = '.FS.a,* TPS = '.F'j.l,' MPD'/Xj
IFUWQ.EQ.1) «
A-7
-------
ISTKT = ISTU1+1
C 'BLANK' PHl'MI" LIUE
C6LL SETlA CdSTRT.K) = CS-
IF ESTUARY, AOJUST MASS FOR FRESHWATER
1 1WB.E0.1! bO TO i»
IF(UW.JE.O) 10 = NJ*Qi\V (Ci + QW)
TERM = M/A/SURTit.*pi*E)
TERMD = fiu/A/sjRT(«>.*Pi*t->
CALL TSTRf
H DO jO J=1,HLOOP
I^U.GT.IU) CALl TSTOP
T = (FLOAT!J-l1+.10)*UT
IF(SLOT.GT.DLOOP) GO TO 100
SLOT = iSTRi+j-i
DO 25 Ksl.NLOC
X = LOCS(K)
GO 10 (5,15),I«3
C RIVER
C CHECK TO SEE IF MASS SLUG ARRIVES
C AT -X- DURING THIS TINc. INTERVAL
5 IF! AQSIT-X/U) . i=:.l)T) GO 10 2b
IF! (T-X/U) .GT.ii. ) GO 1C 25
lFdWQ.EO.3) ^0 TO 10
ciSLOT.K) = QBAL*C (SLOT,»\) + M/(CI+UW)*EXPI-KR*X/U)
GO TO 25
1C ETEhM = EXPt-KA*X/U)
CONC = QQAL-MCs-C (SLOT.K) ) + >"./ ( U + Q^ ) *KO/ ( KA-KR I * (
E1ERN) -nlLVlvJ+J'/I) *ETER-ui
C(SL07,K) = CS-COi:C
GO 10 25
C ESTOARY
Ib IF{1WO.EQ.3) GO TO 20
CtSLOI.K) = C (SLOT.K)+TL"i
-------
GO TO 2b
IF(MD.'"JE.O. ) COfiC = CUUt + IEKM)/b IK f i T ) »EXPO(- ( X , f , •
C(SLOTiK) = C ( SI.OT.M-CGIJL
25 CONTINUE
30 CONTINUE
C CHECK FOK TOP OF PAGE
35 IF(LNCNT.LE.O) WHI TL (l/U f , 1 700 > 7 II LE • ,;UC JfJ.I SCrtL
1700 FORhATI'l* // 14X.20A4 / 14X , ' KA11 Ir'nLI ( HC'tES ; • , 1 l)X .
. 'LOADIPOtliJUS ) • ,2?X,2A4 /I 2X , • 0 . ' , 1 Af,, f t,. 2 , ' U.'.llX.f
3t)X,F'».u/liXt X .' TIME • • f> I 4X . F 7 . 2 . • = X',J.X))
DO 250 1=1.OLOUP
T = FLOATd-1 )*OT
250 URnC(OUTil900) Tt ( C ( n J) . J = l t ULOC )
C PERFORM blHPLE STATISTICAL AN/(LyGIS O'J (HL (.UHC V£CIU->
C8LL SETKAISOBi-.'JLOOP,1,1.0 )
C/lLL TALLY (CiSOtlSi TOTAL,AVLK.SO, VII rj,V%AX,ibbU 16)
00 650 NLCOP = 1 ,.MLOC
00 625 IST=1,NOSTD
STDRD = T.hPbTOUST)
C MHR3V = LOUNTEH FOR IJ'J. OF COiviSEC J I i V E >)0;JSi, THAT ^AlfR (JJAuIlT
C CrdTtRIA IS
A-9
-------
C 1HRSV = TOTAL f"0. OF HOURS T.lnT WnTLK 31MLIT1 CRJTt.-d'V LXCLuDLU
C STufJOARDS
C NHRSV = NAX. NO. OF COtJbECUU^E HUo. THAT WMLK CjUwLiTY
C CKxTEKIA WAS EXCEEDED
C MNCUR = M(,X. NO. OF OCUuRENCEb FUf< ANY CONSECUTIVE 3KDiJt'l,JG
IF(IST.GT.I) GO TCI 350
C
C TABULATE DENSITY FUNCTION AND STA IUAROS VIOLATIONS
WRITE ( OUT, 2bOO) 1 IT LE • LOCS ( IH OOP ) . WJJCLTJ
2500 FORMAT! • 1 • /////5X , 20 A4/5X , 'OENSI I Y F.MCTlOU F'JK LOC.Afii>M, X ='i
F7.2/1UX, 'VAKIARLE =',lX,2At, /
lOXi «CUHLAT1\/E Ot_IMSIfY FUNCTION t PC T .LE. '.UUC ) */
20X.'PLKCE^JT CONCV)
00 260 I = 1.L)LOUP
260 SORS(I) = Cll.NLOOP)
CfiLL SHELRlSUBSOLOOP.lbbUil « JKSORT, iuSORf >
00 iOO I=l.uPClL
IPOS = FLOAT(DULJOP)*PCTILLlI ) /10U . +0 . 05
CONC = SOoS(IP'JS)
iOO fcJi»ITE(OUTi32UO) 1 .PCT ILL ( I ) , CUNC
3200 FORMAT(10X«Ib,&X,F5,?,Eltj.L)
350 DO bOO I=liULOOP
CONC = C(I»[JLOrJP)
IF(IWO.NF..3 .A'MD. CONC.bt. .SIORD) Go TO '4'JO
IFUWQ.EG.S .ANiJ. CUNC.LL.STOHJ) GO TC i+L)0
"»00 IFIIJHR3V.EO.U) GO TO 'jOO
MXCI^T = 0
ICNT = 1
tlO IF( FRI13VI ILfJf ) .EQ.MHKSV I TKEuSdCNT) r FREUidCMT) «• 1
!=•< FRQSVt ICNT) .EQ.NHKSV I RO fO ^oO
Iff FR-33V(ILUT) .NE.O ) l»0 TO i*?.0
ICN1 ) = 1
GO TO 430
t20 ICNT = ICNT + 1
ir ( ICNT.G'I .50) GO TO 9bU
IF( ICIMT.UT.bO ) PAUSCI 5050
GO TO 410
430 CONTINUE
IF( ICMT.C.T .MXCNT ) MXCU1 = ICUT
NHRSV = o
GC TO 500
150 NHRSV = NHRSV + 1
THRSV = THRbV+i
500 CONTINUE
IF(i'jHRSV.t-O.U) bO TO 510
ICNT = i
501 IF( FRrj3V(ICiJT) .CO. NHRSV ) FREi)3 ( ICI'JT ) = FREOc. « ICMT )
It( FR J3VI ICMT) .'JE.O ) GO TO 5J2
FRQ3V(ICNT ) = NHRSV
FSEQ3IICN1 > = 1
GO TO b03
502 ICNT = ICNT + 1
IC( ICNT.GT .50) GO TO 9bO
GO TO 501
503 CONTINUE
IT( ICNT.GT.'IXCUT ) llXtlJl = ICNT
NHRSV = 0
C
£**»******«***** + * + ***•»* + *****
510 W^ITEIOUT ,4'->50) WOCUfj , S TURD
A-10
-------
<«b?)0 FORMAT! /// ISX.'NO.Oh
2 'THAT «.2A<», • VIOLATED bT AUJAKDb' /29X, • STANDARD = *iF7.1/)
W=UTElOUT«<4faOO)
tbno C:;RMVT i inx ,21 • vj.t OUKS NO.occuufUjcLb' ,i+x i)
WRITE(OUTi4700) (FROivI 1),hKEu3(I),1 = 1,rtXCUT)
t7no FORMAT(iox,2dbiiox.i3«9x)/E COI«C =', F6.?,« fWL'
. /22X,'STA,MO<\KD DEVIATION =', F6.2,' MG/L '/2b* ' ' ^AXI'lJ'l LOIJC =',
F6.2,'«G/L'//)
3hO CONTINdE
C
CflLL EXIT
900 WRITE! ")UT,gOOO) fJOAY i LIT , ULOOP
9000 FORMATI///20X, ' THt. STOKMrt«TLi< DI!iCHArt(iL Sl^DLATOK IS NOT C'-HABi-L U
,F pROCEssirjb A iiriE KECOKO GKF.ATI:R T'i<» i i^tu ELL;IE'JTS«/?OX« • AU'J T
.HE HEQ'JESTEO RECOKO OF ' , f-7. L>, ' DAYS «f ',Fb.2,' MRS IS '.ID,' ELL
CALL EXIT
9500 FORMAT(///2UX,'FrtEQ3 CURL bl^E EXCtEOiM •/I
9f)0 WRITE(OUT,93UO)
CALL EXIT
9750 FORMAT!///) OX, • IMPROPER KLLIf.VliMli »Arr.K BODY SU-CCTED ( J J7 t_UUAL TO
. 3IVER OR ESTUARf) «//t3X,'-lii<-'//5X,'ilPRol'Ert WATER UUALlT' CUnSTi
.TjENT SELEC fEDINUT EQUAL 10 COLIKOKM, 30D, OR (Jl))«//)
C4LL EXIT
END
A-ll
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. R
3-79-023
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
A STATISTICAL METHOD FOR ASSESSMENT OF URBAN RUNOFF
5. REPORT
EPORT DATE
May 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Eugene D. Driscoll, Dominic M.D. Toro and Robert V. Thbmann
8. PERFORMING ORGANIZATION REPORT NO.
3. PERFORMING ORGANIZATION NAME AND ADDRESS
Hydroscience Inc.
363 Old Hook Road
Westwood, New Jersey 07675
10. PROGRAM ELEMENT NO.
i >
•«
1. CONTRACT/GRANT NO.
68-01-3251
12.
SPONSOB.ING.AGENCY NAME AND.ADDFIESS .
U.S. Environmental Protection Agency
Water Planning Division ,
Nonpoint Sources Branch
Washington, DC 20460
13. TY
MAY
REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
See Item 12
15. SUPPLEMENTARY NOTES
Project Officer: DENNIS N. ATHAYDE
16. ABSTRACT
This manual describes a simplified methodology which can be used to
assess the impact of urban stormloads on the quality of receiving waters, and
to evaluate the cost and effectiveness of control measures for reducing these
pollutant loads. The methodology is particularly appropriate for use at the
planning level where preliminary assessments are made to define problems,
establish the relative significance of contributing sources, assess feasibil-
ity of control, and determine the need for and focus of additional evalua-
tions. It can also be used effectively in conjunction with detailed studies,
by providing a cost-effective screening of an array of alternatives, so that
the more detailed and sophisticated techniques can examine only the more
attractive alternatives.
The methodology is based on the determination of certain statistical
properties of the rainfall history of an area. From these statistics, the
desired information on loads, performance of controls, and receiving water
impacts is generated directly. Procedures are quite simple to apply, using
charts and graphs which facilitate screening alternate types or levels of
control, testing sensitivity to assumptions concerning drainage area character-
istics, stormwater contaminant levels and similar variable factors.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COS AT I I''ield/Group
Runoff
Rainfall
Cost-Effectiveness
Water Quality
Water Pollution
Statistical Analysis
Computerized Simulation
Methodology
Urban Land Use
Urban Runoff
Urban Hydrology
Receiving waters
Urban Stormloads
Pollutant Loads
Control Measures
13 B
18. DISTRIBUTION STATEMENT
Release to Public
19, SECURITY. CLASS (This Report!
Unclassified!
21.
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
-------
Errata Sheets For
A Statistical tMethod For The Assessment Of
Urban Stormwater
(EPA 440/3-79-023)
1. Page 1-2. Change the notation for the bibliography to "An extensive
bibliography is provided at the end of chapters 2,3,4,5,6,
and 7 which can direct the reader..."
2. Page 3-2, equation 3-1. Place a comma before "n = 0,1,..."
3. Page 3-3, equation 3-2. Change to ". . . = 1 - Pr (6 > T) ."
equation 3-3. Change the last term to "...= e~T//A"
equation 3-4, change to "... = 1 - e"T//A"
equation 3-5. Change to "... = I_ e"5/^"
A
Change the first sentence after equation 3-6 to
"The parameter of the Poisson density function, A,
is thus seen to be the average time between storms,
-------
2
8. Page 3-20. Change the symbol at the top of the page from "v " to "pca".
9. Page 3-21, equation 3-21. Change the term "(c0cp)" to (cp- C0)."
Immediately following equation 3-21, add the following.
"The rate of subsidence of the first-flush peak is defined
by the rate coefficient B, with units of hours".
Equation 3-23. Change to the following.
"MR = E[c(t)qd] = ..."
Equation 3-24. Change the denominator of the second term
in the brackets to:
"(DR/eh i"
10. Page 3-25, equation 3-26. Change to:
"W0 = MR/A"
11. Page 3-25. Following the 10th line from the bottom of the page, add
the following. "Length of period' of interest (July-September) =
31 + 31 + 30 = 92 days. The assumptions were also made that
DR/S = 1.0 and cp/c0 = 3."
12. Page 3-26. Change the second line to "... a moderate first flush,
with assumed characteristics as identified earlier in
this example. Figure 3-6 indicates..."
13. Page 3-27, table 3-3. Add' an asterisk to the last two headings of
the table. Also add the following footnote.
"* 'Greater than' indicates the percent and expected number of
storms which will have flows (Q) and loads (W) greater than
the corresponding values in the table."
14. Page 3-33, equation. 3-28. Change the units for the symbol "L" to
"(gm/m2/yr)."
-------
3
15. Page 3-40. After the line "u= 10 miles/day", change the following
two lines to:
"a = 2Et
dr2 U2
= 2(0.3)(0.42) = 0.16"
(0.125)2(10)2
16. Page 3-42. Change the second line from the bottom of the page to
"... = (freshwater flow)/A"
17. Page 3-43. Change "K0 = modified Bessel function..." to
"K0(b) = modified Bessel function..." and add
"b = Uxm, for this example."
E
18. Page 3-59, equation 3-35. Add brackets around the center part of
the equation as follows.
"...(q -
19. Page 3-60, equation 3-39. Add brackets around the term following the
last integration symbol as follows.
"...[I - Kq)]..."
20. Page 3-63. Change the third line of the second paragraph to-
"5-44 to 5-54 in section 5.5.2 of ..."
21. Page 3-86. Ajid the following footnote.
"The text refers to typical pick up efficiencies of about
50 percent for brush-type and about 90 percent for vacuum
type units. Experience from more recent studies suggests
that while vacuum devices have higher efficiency with all
other things being equal, as a practical matter, other
-------
conditions which influence sweeper efficiency are often much
more significant than sweeper type. In practical applications,
efficiencies in the range of 25 to 50 percent can be expected."
22. Page 3-37, Figure 3-33. Change the note on the figure to
"STREET SWEEPING SIMULATION, MINNEAPOLIS, MINN. (REF.46)..."
23. Page 3-91. Change the third line from the bottom of the page from
"...Hendy and Mi*..." to "...Heaney and Nix..."
24. Page 5-124. Change the fourth line from the top of the page to:
"(kt)R = the value of (kt) at the mean runoff flow, QR"
25. Page 5-168, reference 1. Change the author's name to "Heaney, James P."
26. Page 6-14, figure 6-4. The description of the ordinate should be
"ERROR IN ESTIMATE OF AVERAGE (-%)."
The description of the abscissa should be
"N = NUMBER OF STORMS MONITORED"
The scales for the ordinate and abscissa should begin at
"0" and increase in increments of 10.
------- |