MODELING OF NON POINT SOURCE POLLUTION FROM RESIDENTAL
AND COMMERCIAL CATCHMENTS IN SKUDAI, JOHOR
SITI NAZAHIYAH BTE RAHMAT
A thesis submitted in fulfilment of the
requirements for the award of the degree of
Master of Engineering (Hydrology and Water Resource)
Faculty of Civil Engineering
Universiti Teknologi Malaysia
DECEMBER 2005
PSZ 19:16 (Pind. 1/97)
UNIVERSITI TEKNOLOGI MALAYSIA
BORANG PENGESAHAN STATUS TESIS
JUDUL:
MODELING OF NON POINT SOURCE POLLUTION FROM
___________________________________________________________________
RESIDENTIAL AND COMMERCIAL CATCHMENTS IN SKUDAI,
___________________________________________________________________
___________________________________________________________________
JOHOR
SESI PENGAJIAN: ___________________
2005/2006
Saya
___________________________________________________________________
SITI NAZAHIYAH BTE RAHMAT
(HURUF BESAR)
mengaku membenarkan tesis (PSM/Sarjana/Doktor Falsafah)* ini disimpan di Perpustakaan
Universiti Teknologi Malaysia dengan syarat-syarat kegunaan seperti berikut:
1.
2.
3.
4.
Tesis adalah hakmilik Universiti Teknologi Malaysia.
Perpustakaan Universiti Teknologi Malaysia dibenarkan membuat salinan untuk tujuan
pengajian sahaja.
Perpustakaan dibenarkan membuat salinan tesis ini sebagai bahan pertukaran antara
institusi pengajian tinggi.
* * Sila tandakan ()
SULIT
TERHAD
(Mengandungi maklumat yang berdarjah keselamatan atau
kepentingan Malaysia seperti yang termaktub di dalam
AKTA RAHSIA RASMI 1972)
(Mengandungi maklumat TERHAD yang telah ditentukan
oleh organisasi/badan di mana penyelidikan dijalankan)
TIDAK TERHAD
Disahkan oleh
________________________________
(TANDATANGAN PENULIS)
________________________________
(TANDATANGAN PENYELIA)
Alamat tetap:
4057, TAMAN TUNKU PUAN CHIK
____________________________________
72100 BAHAU
____________________________________
NEGERI SEMBILAN
____________________________________
PM DR ZULKIFLI BIN YUSOP
___________________________________
DISEMBER 2005
Tarikh: ______________________________
DISEMBER 2005
Tarikh: ____________________________
CATATAN:
Nama Penyelia
* Potong yang tidak berkenaan.
** Jika tesis ini SULIT atau TERHAD, sila lampirkan surat daripada pihak
berkuasa/organisasi berkenaan dengan menyatakan sekali sebab dan tempoh
tesis ini perlu dikelaskan sebagai SULIT atau TERHAD.
Tesis dimaksudkan sebagai tesis bagi Ijazah Doktor Falsafah dan Sarjana
secara penyelidikan, atau disertasi bagi pengajian secara kerja kursus dan
penyelidikan,
atau
Laporan
Projek
Sarjana
Muda
(PSM).
“I/We* hereby declare that I/we* have read through this thesis and in my/our*
opinion this thesis is sufficient in terms of scope and quality for the award of the
degree of Master of Engineering (Hydrology and Water Resource)”
Signature
:
……………………………………..
Supervisor I
:
Assoc. Prof. Dr. Zulkifli bin Yusop
Date
:
December 2005
Signature
:
……………………………………..
Supervisor II
:
Assoc. Prof. Dr. Maketab bin Mohamed
Date
:
December 2005
* Delete as necessary
BAHAGIAN A – Pengesahan Kerjasama*
Adalah disahkan bahawa projek penyelidikan tesis ini telah dilaksanakan melalui
kerjasama antara _____________________ dengan _______________________
Disahkan oleh: Tandatangan : …………………………….
Nama
Tarikh : ……………………..
:……………………………..
Jawatan
:……………………………..
(Cop rasmi)
* Jika penyediaan tesis/projek melibatkan kerjasama
BAHAGIAN B – Untuk Kegunaan Pejabat Sekolah Pengajian
Siswazah (SPS)
Tesis ini telah diperiksa dan diakui oleh:
Nama dan Alamat Pemeriksa Luar
: Prof. Madya Dr Ismail Bin Abustan
School Of Civil Engineering,
Engineering Campus,
Universiti Sains Malaysia,
14300 Nibong Tebal,
Pulau Pinang.
Nama dan Alamat Pemeriksa Dalam I: Dr Supiah Bte Shamsudin
Fakulti Kejuruteraan Awam,
Universiti Teknologi Malaysia,
81310, Skudai, Johor.
Nama Penyelia lain
(jika ada)
: Prof Madya Dr Maketab Bin Mohamed
Fakulti Kejuruteraan Kimia dan Sumber
Asli,
Universiti Teknologi Malaysia,
81300, Skudai, Johor.
Disahkan oleh Timbalan Pendaftar di SPS:
Tandatangan : ...............................................
Nama
Tarikh : ..................................
: ..........................................................................................................
I declare that my thesis entitled “Modeling of non point source pollution from
residential and commercial catchments in Skudai, Johor” is the result of my own
research except as cited in references. The thesis has not been accepted for any
degree and is not concurrently submitted in candidature of any other degree.
Signature
:
……………………………………
Name
:
Siti Nazahiyah Bte Rahmat
Date
:
December 2005
Specially dedicated to my beloved parents, brothers and sisters for their love,
encouragement and endless support towards the success of this study
Last but not least, to special one, Ahmad Zurisman, thanks for everything…
ACKNOWLEDGEMENTS
I would like to convey my sincerest thanks to my supervisor Associate
Professor Dr. Zulkifli bin Yusop for his dedicated guidance, valuable assistance and
endless encouragement throughout the accomplishment of this research. I would
also like to thank Associate Professor Dr. Maketab bin Mohamed for his support.
I am grateful to the technicians of the Environment Laboratory in the Faculty
of Civil Engineering, UTM who had been very helpful in providing assistance
throughout the work. My thanks is also extended to Associate Professor Dr. Ismail
bin Abustan for his advice whether directly or indirectly in improving my research.
Also not forgetting my fellow colleagues at Institute of Environmental and Water
Resources (IPASA), UTM for their support.
I am deeply indebted to Kolej Universiti Teknologi Tun Hussien Onn
(KUiTTHO) and Public Services Department (JPA) for the financial aid provided in
pursuing this study. My most heartfelt gratitude to my beloved family for their
unfailing love, care, encouragement and prayers.
Last but not least, my thanks to all those who had helped directly or indirectly
in my research.
ABSTRACT
Sampling of urban runoff was carried out in two small catchments, which
represent residential and commercial areas in Skudai, Johor. Ten storm events for
residential and seven events for commercial catchments were analysed. Runoff
quality showed large variations in concentrations during storms, especially for SS,
BOD5 and COD. Concentrations of NO3-N, NO2-N, NH3-N, and P were also high.
Lead (Pb) was also detected in both catchments but the levels were low (<0.001
mg/l). In general, the water quality was badly polluted and fell in class V of the
Interim National Water Quality Standards. The hydrographs and pollutographs for
both catchments showed rapid increases and decreases equally rapidly.
pollutants were diluted as storm events progress.
Most
In most cases, the peak
concentrations preceded the peak runoff. This suggests that the pollutants were of
short distant sources/origins and the bulk of the pollutant mass arrived at the
catchment’s outlet much faster than the runoff itself. For the hysteresis loop, both
catchments showed most of the parameters were characterized by clockwise
hysteresis.
Only a few plots were exhibited counterclockwise and figure eight
hysteresis loop. The relative strength of the first flush for the commercial catchment
was P> COD>SS> NO3-N> NO2-N> BOD5> NH3-N whereas for the residential
catchment was SS> COD> BOD5> NH3-N> P> NO3-N> NO2-N. The loadings were
higher in the commercial than in the residential catchment and this was attributed to
a greater runoff volume per unit area and higher Event Mean Concentration (EMC)
in the former. Detail calibration and validation of Storm Water Management Model
(SWMM) for modeling water quantity and quality were discussed. The simulation
results, evaluated in terms of runoff depth, peak flow and the hydrograph shapes,
were satisfactorily. For the water quality modeling, the simulation results were
evaluated in terms of total load and peak load.
SWMM can model SS load
reasonably well for the residential catchment, but was not satisfactory for the
commercial catchment.
ABSTRAK
Air larian bandar telah disampel di tadahan perumahan dan tadahan komersil
di Skudai, Johor. Sampel dari 10 kejadian ribut bagi tadahan perumahan dan tujuh
kejadian ribut bagi tadahan komersil telah dianalisis.
Kualiti air larian ribut
menunjukkan pelbagai variasi terutamanya untuk kepekatan SS, BOD5 dan COD.
Kepekatan NO3-N, NO2-N, NH3-N dan P juga mencatatkan nilai yang tinggi.
Sementara kepekatan Pb dalam kedua – dua tadahan adalah rendah (< 0.001 mg/l).
Secara amnya, kualiti air dikategorikan di dalam kelas V berdasarkan Piawai Interim
Kualiti Air Kebangsaan.
Hidrograf dan pollutograf bagi kedua – dua tadahan
menunjukkan kenaikan dan juga penurunan secara mendadak.
Bahan cemar
mengalami pencairan sepanjang kejadian ribut. Kebanyakan ribut, menunjukkan
kepekatan maksimum berlaku lebih awal berbanding air larian puncak.
Ini
mencadangkan bahawa sumber bahan cemar adalah berdekatan dan sejumlah besar
bahan cemar tiba di salur keluar tadahan lebih awal dari air larian itu sendiri. Bagi
analisis‘hysteresis loop’, kebanyakan parameter di kedua- dua kawasan tadahan
menunjukkan ‘clockwise loop’. Kekuatan relatif fenomena ‘first flush’ bagi tadahan
komersil adalah P> COD> SS> NO3-N> NO2-N> BOD5> NH3-N sementara bagi
tadahan perumahan: SS> COD> BOD5> NH3-N> P> NO3-N> NO2-N. Beban bahan
cemar di tadahan komersil adalah lebih tinggi berbanding di tadahan perumahan. Ini
disebabkan isipadu air larian per unit luas dan ‘Event Mean Concentration’ yang
lebih besar di tadahan komersial. Kalibrasi dan validasi menggunakan SWMM bagi
pemodelan kuantiti dan kualiti air dibincangkan secara terperinci.
Keputusan
simulasi dinilai dari segi kedalaman air larian, aliran puncak dan bentuk hidrograf.
SWMM memberi keputusan yang memuaskan untuk proses kalibrasi dan validasi.
Sementara itu, bagi pemodelan kualiti air, keputusan simulasi dinilai dari segi jumlah
beban dan beban puncak.
Beban SS dapat dimodel dengan baik menggunakan
SWMM bagi tadahan perumahan tetapi tidak begitu memuaskan bagi tadahan
komersil.
TABLE OF CONTENTS
CHAPTER
1
2
TITLE
PAGE
DECLARATION
ii
DEDICATION
iii
ACKNOWLEDGEMENT
iv
ABSTRACT
v
ABSTRAK
vi
TABLE OF CONTENTS
vii
LIST OF TABLES
xi
LIST OF FIGURES
xv
LIST OF SYMBOLS / ABBREVIATIONS
xx
LIST OF APPENDICES
xxi
INRTRODUCTION
1.1
Background of the research
1
1.2
Statement of the problem
3
1.3
Objectives
5
1.4
Scope of work
6
LITERATURE REVIEW
2.1
Urban water pollution
7
2.2
Runoff Quality
9
2.3
Control of non point source pollution
13
2.4
Pollution Build up and Wash off Process
14
2.5
Concentrations
16
2.5.1 Event Mean Concentration
16
2.6
Load Estimation
17
2.7
Hysteresis Loop
18
2.8
Concepts of first flush
23
2.8.1 Analysis by difference between
25
dimensionless Cumulative Pollutant
and Volume Mass
2.8.2 Dimensionless cumulative
26
analysis
3
2.8.3 Analysis by cumulative curve ratio
27
2.9
Non point source pollution simulation models
31
2.10
Storm Water Management Model (SWMM)
33
2.11
Sensitivity Analysis
35
METHODOLOGY
3.1
Catchment Description
36
3.2
Rainfall Measurement
39
3.3
Flow Measurement
40
3.4
Sampling Procedure
42
3.4.1 Analysis of water quality
43
3.5
Event mean concentration
45
3.6
Correlation Analysis
46
3.7
Box plot analysis
46
3.8
Pollutant loading
47
3.9
First Flush Concept
48
3.9.1 Dimensionless cumulative analysis
48
Storm Water Management Model (SWMM)
48
3.10
3.11
Water Quantity
49
3.12. Hydrologic Parameter
50
3.12.1 Horton Equation
50
Routing Method
53
3.13.1 Kinematic Wave
53
3.14
Calibration and Validation
54
3.15
Sensitivity Analysis
56
3.16
Goodness of fit
56
3.17
Water Quality
58
3.17.1 Estimation of Buildup Parameters
58
3.17.2 Estimation of Wash-off Parameters
63
3.13
4
5
RUNOFF QUALITY AND POLLUTANT LOADINGS
4.1
Characteristics of Rainfall Events
66
4.2
Event Mean Concentrations (EMC)
67
4.3
Correlation Analysis
70
4.4
Baseflow and Stormflow Concentrations
71
4.5
Hydrograph and Pollutograph Analysis
80
4.6
Hysteresis Loop Analysis
85
4.7
First Flush Phenomenon
89
4.8
Pollutant Loadings
94
STORM WATER MANAGEMENT MODEL (SWMM)
5.1
Hydrologic Parameters
97
5.2
Sensitivity Analysis
99
5.2.1 Sensitivity to Runoff Depth
99
5.2.2 Sensitivity to Peak Flow
102
5.3
Model Calibration and Validation for
104
Commercial Catchment
5.4
5.3.1 Calibration of Runoff Depth
104
5.3.2 Validation of Runoff Depth
105
5.3.3 Calibration of the Peak Flows
106
5.3.4 Validation of the Peak Flows
108
Calibration and Validation for Residential
112
Catchment
5.5
5.6
5.4.1 Calibration of the Runoff Depth
112
5.4.2 Validation of the Runoff Depth
112
5.4.3 Calibration of the Peak Flows
113
5.4.4 Validation of the Peak Flows
115
Water Quality
118
5.5.1 Estimation of Buildup Parameters
118
5.5.2 Estimation of Washoff Parameters
122
Calibration of water quality for commercial
127
catchment
5.7
Calibration of water quality for residential
134
catchment
6
CONCLUSION AND RECOMMENDATIONS
6.1
Conclusion
141
6.2
Recommendations
143
REFERENCES
144
APPENDICES
160
LIST OF TABLES
TABLE
2.1
TITLE
Mean concentrations (mg/l) of stormwater
PAGE
17
runoff for various types of studies
2.2
Ranges of pollutants loads for various
18
types of studies (in kg/ha/year)
2.3
Summary of C/Q relationship
19
2.4
Summary of methodologies for
29
determining First Flush
3.1
Physiographical conditions of sub catchments
39
3.2
The analytical methods used in this study
44
3.3
Independent variables and parameters
50
influencing stormwater pollution washoff
3.4
Initial Infiltration Capacity
51
3.5
Minimum value of infiltration capacity
51
for different hydrologic soil group
3.6
Decay rate of infiltration
52
4.1
Characteristics of monitored storms
67
4.2
Event means concentration (EMC) of water
68
quality parameters
4.3
Water Quality Index Class Standard
68
4.4
Comparison between the results obtained from
69
this study with the other studies
4.5
Statistical analysis of pollutant concentration for
70
both catchments
4.6a
Correlation analysis of pollutant concentration
in the residential catchment
71
4.6b
Correlation analysis of pollutant concentration
71
in the commercial catchment
4.7a
Median concentration of stormflow and
74
baseflow quality in the commercial catchment
4.7b
Median concentration of water quality parameters
78
for residential catchment
4.8
Characterisation of pollutographs for each event
82
4.9
Summary of pollutographs for both catchments
82
4.10
Summary of C/Q hysteresis loop characterization
86
for both catchments
4.11
Cumulative load at 20-30% of the runoff volume
91
in the commercial catchment
4.12
Cumulative load at 20-30% of the runoff volume
93
in the residential catchment
4.13
Event pollutant loadings for residential and
94
commercial catchments
5.1
The Manning’s roughness values
98
5.2
Default parameter coefficient in SWMM
98
5.3
Parameters values used for calibrating SWMM
104
in the commercial and residential catchments
5.4
Observed and simulated runoff depths of the
105
calibrated events in the commercial catchment
5.5
Observed and simulated runoff depths of the
106
validated events in the commercial catchment
5.6
Comparison between observed and simulated
107
peak flows of calibrated events in the
commercial catchment
5.7
Statistical fits of the validated peak flows in the
109
commercial catchment
5.8
Comparison between observed and simulated
runoff depths of calibrated events in the
residential catchment
112
5.9
Comparison between observed and simulated
113
runoff depths of validated in the residential
catchment
5.10
Observed and simulated peak flows of
113
calibrated events in the residential catchment
5.11
Comparison between observed and simulated
115
peak flows of validated hydrographs in the
residential catchment
5.12
Collected dust and dirt in the commercial
120
catchment
5.13
Deposition of dust and dirt in the residential
121
catchment
5.14
Parameters of buildup data in the commercial and
121
residential catchments
5.15
Washoff parameters for the commercial catchment
124
5.16
Washoff parameters for the residential catchment
126
5.17
Calibration of SS loading for 16 March 2004’s storm
128
5.18
Peak load of SS (kg/s) for the event on
128
March 16, 2004
5.19
Calibration of SS loading for March 11, 2004
129
5.20
Peak load (kg/s) for the event on March 11, 2004
129
5.21
Calibration of SS loading for storm on March 19, 2004
130
5.22
Peak load of SS (kg/s) in the commercial
131
catchment for storm on March 19, 2004
5.23
Calibration of SS loading for storm on
132
April 14, 2004
5.24
Peak load of SS (kg/s) in the commercial
132
catchment on April 14, 2004
5.25
Calibration of SS loading in the commercial
133
catchment for storm on September 10, 2004
5.26
Peak load of SS (kg/s) in the commercial
133
catchment for storm on September 10, 2004
5.27
Calibration of SS loading in the residential
catchment on March 4, 2004
135
5.28
Peak load of SS (kg/s) on March 4, 2004
135
5.29
Calibration of SS loading in the residential
136
catchment for storm on 12 July 2004
5.30
Peak load of SS (kg/s) in the residential
137
catchment on 12 July 2004
5.31
Calibration of SS loading from residential
138
catchment on September 8, 2004
5.32
Peak load of SS (kg/s) from the residential
138
catchment on September 8, 2004
5.33
Calibration of SS loading from the residential
139
catchment on November 4, 2004
5.34
Peak load of SS (kg/s) from the residential
catchment on November 4, 2004
139
LIST OF FIGURES
FIGURE
2.1
TITLE
PAGE
Nitrogen cycle
11
2.2
Phosphorus cycle
12
2.3
Pollutant accumulation on impervious surfaces
15
of urban areas
2.4
Three subgroups of single-valued-line class
20
2.5
Clockwise loop C-Q relationship
21
2.6
Counter-clockwise loop C-Q relationship
21
2.7
Single-valued plus a loop C-Q relationship
22
2.8
Figure eight C-Q relationship
22
2.9
Typical first flush of solids
25
2.10
Definition of the first flush
26
2.11
Cumulative mass/volume ratio for determining
27
the presence of first flush
2.12
Calibration procedure for SWMM
34
3.1
Location of the study site
37
3.2a
Residential Catchment
37
3.2b
Commercial Catchment
38
3.3
A rain gauge installed on roof top in the
40
residential catchment
3.4
Water Level – Discharge Rating Curve for
41
both catchments
3.5
Typical stormwater sampling interval
43
employed in this study
3.6
Information that can be derived from a box plot
46
3.7
Decay rate coefficient
52
3.8
Buildup relationship
59
3.9
Test strip for the commercial catchment
61
3.10
Test strip for the residential catchment
62
3.11
Cumulative values of KW and runoff rate (R)
65
4.1
Box plot analysis of stormwater quality in
73
the commercial catchment
4.2
Effects of antecedent dry days on
75
Pollutant concentrations
4.3
Box plot analysis of stormwater quality in
77
the residential catchment
4.4
Effects of antecedent dry day on
79
pollutant concentrations
4.5a
Pollutographs and hydrographs in the residential
83
catchments
4.5b
Pollutographs and hydrographs in the commercial
84
catchments
4.6
Hysteresis loops for residential catchment
87
4.7
Hysteresis loops for commercial catchment
88
4.8
Mass Volume, M(V) ratios of BOD5, COD, SS,
90
NO3-N, NO2-N, NH3-N and P in the commercial
catchment
4.9
Mass Volume, M(V) ratios of BOD5, COD, SS,
92
NO3-N, NO2-N, NH3-N and P in the residential
catchment
5.1
Relationship between observed runoff and rainfall
96
depth for commercial catchment
5.2
Relationship between observed runoff and rainfall
96
depth for residential catchment
5.3
Sensitivity analysis for percentage of impervious
99
area on runoff volume
5.4
Sensitivity analysis of catchment width on runoff
99
volume
5.5
Sensitivity analysis of impervious depression storage
100
on runoff volume
5.6
Sensitivity analysis of percentage of impervious area
101
on runoff volume
5.7
Sensitivity analysis of runoff volume for catchment
101
width
5.8
Sensitivity analysis of runoff volume for impervious
101
depression storage
5.9
Sensitivity analysis of percentage of impervious area
102
on the peak flows
5.10
Sensitivity analysis of catchment width on the
103
peak flows
5.11
Sensitivity analysis of percentage of impervious area
103
on the peak flows
5.12
Sensitivity analysis of peak flows for catchment width
103
5.13
Observed and simulated hydrographs in the
107
commercial catchment for storm on March 16, 2004
5.14
Observed and simulated hydrographs in the
108
commercial catchment for storm on March 11, 2004
5.15
Observed and simulated hydrographs of the
109
commercial catchment for storm on April 14, 2004
5.16
Observed and simulated hydrographs of the
110
commercial catchment for storm on March 19, 2004
5.17
Observed and simulated hydrographs of the
110
commercial catchment for storm on
September 10, 2004
5.18
Observed and simulated runoff depth in the
111
commercial catchment
5.19
Observed and simulated peak flows in the
111
commercial catchment
5.20
Observed and simulated hydrographs of the residential
catchment for storm on March 4, 2004
114
5.21
Observed and simulated hydrographs of the residential
114
catchment for storm on July 12, 2004
5.22
Observed and simulated hydrographs of the residential
115
catchment for storm on November 4, 2004
5.23
Observed and simulated hydrographs of the
116
residential catchment for storm on September 8, 2004
5.24
Observed and simulated hydrographs of the
117
residential catchment for storm on March 6, 2004
5.25
Observed and simulated peak flows for the
117
residential catchment
5.26
Observed and simulated runoff depth for
118
the residential catchment
5.27
Total dust and dirt (DD) buildup rate in the
119
commercial catchment
5.28
Dust and dirt (DD) buildup rate of particulates
119
≤ 150 µm
5.29
Total dust and dirt buildup rate in the
120
residential catchment
5.30
Dust and dirt (DD) buildup rate for particulates
120
< 150 µm
5.31
Cumulative washoff coefficient and cumulative
124
runoff exponent for the commercial catchment
5.32
Cumulative washoff coefficient and cumulative
126
runoff exponent for the residential catchment
5.33
The observed and simulated loadographs of SS on
128
16 March, 2004
5.34
The observed and simulated loadographs of SS on
130
March 11, 2004
5.35
The observed and simulated loadographs of SS on
131
March 19, 2004
5.36
The observed and simulated loadographs of SS on
132
April 14, 2004
5.37
The observed and simulated loadographs of SS on
September 10, 2004
133
5.38
Relationship between observed and simulated
134
loadings of SS in the commercial catchment
5.39
The observed and simulated loadographs of SS in
136
the residential catchment on March 4, 2004
5.40
The observed and simulated loadographs of SS in
137
the residential catchment on 12 July 2004
5.41
The observed and simulated loadographs of SS from
138
the residential catchment on 8 September 2004
5.42
The observed and simulated loadographs of SS from
140
the residential catchment on November 4, 2004
5.43
Relationship between observed and simulated
loadings for the residential catchment
140
LIST OF SYMBOLS/ ABBREVIATIONS
ALD
-
Absolute Load Difference
ARE
-
Absolute Relative Error
EMC
-
Event Mean Concentratotion
F
-
Dimensionless cumulative runoff volume
FF
-
First Flush
Fp
-
Infiltration capacity
Kw
-
washoff coefficient
L
-
Dimensionless cumulative pollutant mass
M
-
Mass
R
-
Runoff rate
RE
-
Relative Error
RLD
-
Relative Load Difference
Qo
-
Observed discharges
Qs
-
Simulated discharges
V
-
Volume
LIST OF APPENDICES
APPENDIX
A
TITLE
Phosphorus, P Analysis (ICP – MS)
PAGE
160
Calibration
B
Sample Analysis
162
C
Residential Catchment
164
D
Commercial Carchment
166
E
Hydrographs and Pollutographs
167
F
Hysteresis Loops
182
G
Sieve Analysis
200
H
Estimation of Washoff Parameters
201
CHAPTER 1
INTRODUCTION
1.1
Background of the research
Pollution has been defined as changes in the physical, chemical or biological
quality of the resources (air, land and or water) that is injurious to the existing,
intended or potential uses of the resource (Novotny and Chesters, 1981). The
sources or causes of pollution can be classified as either point sources (PS) or nonpoint sources (NPS) of pollution. Point sources of pollution are defined as pollutants
that enter the transport routes at discrete, identifiable locations and that can usually
be measured. Non point sources are defined as diffuse, water flows on the surface
dissolving and washing away pollutants and soil sediments along its path and finally
discharging into receiving waters (Stevenson and Wyman, 1991; Taebi and Droste,
2004). There are several general characteristics that describe non point source
pollution;
i)
NPS discharges enter surface waters in a diffuse manner and at
intermittent intervals that are related mostly to that occurrence of
meteorological events
ii)
Pollution arises over an extensive area of land and is in transit
overland before it reaches surface waters
iii)
Generally, NPS cannot be monitored at their point of origin and their
exact source is difficult to trace
iv)
Elimination or control of pollutants must be directed at specific sites
2
v)
NPS pollutants cannot be measured in terms of effluent limitations
Several major factors have severely disrupted the environmental (ecological)
balance, resulting in accelerated increases of nonpoint sources pollution (Novotny
and Olem, 1994). They are
population increase (sometimes termed explosion) especially in developing
countries
land-use transformation and conversion of land to intensive agriculture and
increased use of chemicals to sustain high agricultural yield
urbanization and industrialization
increased living standard, resulting in an increased per capita use of natural
resources and increasing waste generation
There are various types of diffuse sources of pollution, but the ones that are
most common and regarded as having the most significant impact are agriculture
(mainly nutrients and pesticides), transport (road, air, shipping), atmospheric
deposition (especially on lakes and the sea), leaching and corrosion of building
materials and consumer products, urban and industrial site run-off, storm water and
forestry activities (Moxon, 1998). Also, some non-agricultural land use (e.g. golf
courses) can be a significant contribution for nonpoint pollution source (Evans and
Nizeyimana, 1998). Due to complex modes of transport and site-specific
characteristics, NPS pollutants are generally more difficult to control compared to
point sources. Because of the difficulty to quantify and understand the processes that
contribute to pollutant generation, transport and deposition, the effective management
of non-point source control is complex and always involved non-standard local
boundaries. In addition the cost involved is high whereas the benefits are often not
obvious.
Among non point sources, urban stormwater runoff was reported as a major
contributor to the pollution of many receiving waters (Saget et al., 1996; Appel and
Hudak, 2001; Brezonik and Stadelmann, 2002; Buffleben et al., 2002; Lee et al.,
2004). The quantity and quality of stormwater runoff from urban areas are influenced
by many factors including human activities, meteorological variables and catchment
3
characteristics. The meteorological variables include rainfall, temperature, wind and
inter event periods, whereas the catchment characteristic include catchment area,
topography, landuse, soil types and conditions, population density, drainage systems
and waste disposal practices (Driver and Tasker, 1990).
1.2
Statement of the problem
Typically, there are two main impacts of urbanization. First the hydrology is
modified causing more rapid flow path and the second, increase of human activities
that adds pollutants. Construction of roads and buildings reduce the vegetated area
and increase the catchment’s imperviousness, while the groundwater recharge is
reduced (Whipple, 1983; Lazaro, 1990). These often lead to enhancement of
overland flow, greater peakflow with a shorter time to peak and decrease base flow.
In addition, rapid population growth and land disturbance generate significant
sources of contaminants especially from residential and industrial areas. Ineffective
handling of urban wastewater is quite common and results in adverse environmental
problems (Bedient et al., 1978; Lee et al., 1996).
The more rapid hydrological pathway and readily available sources of
pollutants are responsible for the quality degradation of many receiving water
systems (Petry et al., 2002; Pieterse et al., 2003; Taebi and Droste, 2004). Numerous
studies on urban runoff quality conducted in different parts of the world over recent
decades have shown that runoff can carry relatively high concentrations of a variety
of pollutants. In the early stages of runoff, the land surfaces, especially the
impervious surfaces like streets and parking areas, as well as solids accumulated in
the collection system during the antecedent dry weather period, are flushed by
stormwater. Normally, the velocity of the flow is high in urban drainage systems so
that the runoff is able to transport higher volume of sediments. In small catchments,
this can transport large loading of pollutants in the form of a first flush.
4
The loadings and concentrations of suspended solids, nutrients and other
contaminants are much higher in urban stormwater runoff than in runoff from
unimpaired and rural areas (Sartor and Boyd, 1972; Vaze, 2002). These pollutants
are transported into water bodies, such as lakes and rivers, especially during rainy
season and may lead to eutrophication. Nitrogen in the form of ammoniacal-N and
nitrate, and phosphorus as orthophosphates are readily available for plant growth.
This could lead to algal blooms and excessive macrophytic growth and causing
depletion of dissolved oxygen upon death and decay. The long term effects would
include eutrophication, sedimentation of lakes and rivers, threatening habitats, losses
of biodiversity, channel constriction and more frequent flooding.
The existence of the first flush of pollutants provides an opportunity for
stormwater managers and engineers to control water pollution in an economic and
efficient way. If most of the urban-surface pollutant load were transported during the
initial phase of a storm, then a rather small volume of runoff storage would be
needed to treat and remove the bulk of urban-surface pollutants. As a result,
controlling the first flush has become the most practiced criterion for the design of
stormwater treatment facilities; and first flush collection systems are employed to
capture and isolate this most polluted runoff, with subsequent runoff being diverted
directly to the stormwater system and finally into the receiving environment (Deng,
2005).
Concern over continuous degradation of urban runoff quality, emerged only
recently as opposed to quantity aspects of flood mitigation. Unlike in the developed
countries especially the US, Japan and the EC where funding are more readily
available, monitoring of urban runoff in the developing countries generally receive
less priority. This despite the fact that the latter are experiencing much more severe
water quality problems. A comprehensive understanding on the processes of
contaminants transport and loadings are crucial for formulating effective urban water
and waste-water management strategies (Brezonik and Stadelmann, 2002). To date
coordinated and comprehensive study on these aspects, particularly in tropical
region, is still scarce.
5
Systematic evaluation of non point source impacts often requires water
quality models. Models provide a predictive ability, which enables potentially
expensive water quality management options to be evaluated and tested prior to their
implementation. This is far more cost-effective by considerably reduce financial
resources required for data collection and provides a systematic and rigorous
framework for examining water quality impacts. In this study the Storm Water
Management Model (SWMM) has been selected. This model performs both
continuous and single event simulation. The model can also simulate backwater,
surcharging, pressure flow, looped connections and has a variety of options for
quality simulations, including traditional buildup and wash off formulations
(Novotny and Olem, 1994).
A major advantage of using simulations model is the insight gained by
gathering and organizing data required as inputs to the mathematical algorithms that
made up the overall model system. Besides, many alternative schemes for
development and flood control can be quickly tested and compared with simulation
models (Huber and Bedient, 1992).
1.3
Objectives
The overall aim of this study is to gain a better understanding on the extent of
non-point source pollution in developed urban catchments. Specifically the study
will:-
i)
Quantify loadings of major pollution from selected urban catchments
ii)
Investigate the influence of hydrological regime (rainfall and runoff) on
the pattern of pollutant loading
iii)
Simulate the NPS pollution loadings using Storm Water Management
Model (SWMM)
6
1.4
Scope of work
In order to archive the above objectives the following tasks were carried out:-
1)
Selecting two small catchments representing residential and
commercial catchment;
2)
Measuring discharge and water level during low flow and storm
flow;
3)
Baseflow and stormflow sampling of water;
4)
Labarotary analysis of Biochemical Oxygen Demand (BOD5),
Chemical Oxygen Demand (COD), Suspended Solids (SS),
Nutrients (NO3-N, NO2-N, NH3-N and P) and heavy metal (Pb);
5)
Data analysis including Event Mean Concentration (EMC),
pollutant loading, statistical analysis, box plot analysis, hysteresis
loops and first flush analysis;
6)
Simulation of NPS using XP-SWMM model in terms of water
quantity and quality
CHAPTER 2
LITERATURE REVIEW
This chapter discusses in detailed stormwater quality in urban areas, as one of
non point source pollutions, and its significant impact to receiving waters. This has
been widely discussed in the literature and still remained a major research topic.
Some findings and values that have been reported from other studies are reviewed in
terms of event mean concentration (EMC), pollution estimation, first flush analysis
and hysteresis loop. Besides, findings of non point source pollution simulation
model using Stormwater Water Management Model (SWMM) are also discussed.
2.1
Urban water pollution
While more attention has been paid in recent years to urban point source
pollution control through the establishment of wastewater treatment plants in many
developing countries, no considerable planning and any serious measure has been
taken to control urban non-point source pollution (urban stormwater runoff) (Taebi
and Droste, 2004). It is difficult to establish strategies to control non point sources
pollution because its source and the route of effluent are not clear (Choe et al.,
2002).
8
Stormwater runoff, a type of non-point source pollution, is identified as one
of the leading threats to water quality. It may contain a broad range of pollutants: (1)
sediments from construction sites and disturbed areas; (2) nutrients from fertilized
lawns and roadsides; (3) bacteria from leaking sewers and septic tanks, and pet
wastes; (4) oxygen demanding substances from leaky sewers and septic tanks,
organic matter, and pet waste; (5) oil and grease from leaky motor vehicles,
industrial areas, and illegal dumping; (6) heavy metals from automobile wear and
tear, auto exhausts, and industrial areas; (7) toxic and synthetic chemicals from
pesticide applications, automobiles, accidental spills, or illegal dumping; (8) thermal
impacts from heated impervious areas and many other constituents that are
detrimental to the environment (Akan, 1993; Deng, 2005).
In urban environments, the most important point source is the discharge from
wastewater collection systems; and where a treatment plant exists, this would be
semi-treated effluent from the plant. Through time, different aspects of urban runoff
have been studied by engineers. The major concern amongst engineers and urban
planners around 1980’s was flood control and immediate transfer of the resulting
runoff to non-urban areas. Today, designers not only consider the quantitative
management and control of urban runoff, but they also place great emphasis on its
quality management and control (Nix, 1994).
Knowledge of characteristics of urban runoff is useful for developing water
quality management plan for urban streams. These aspects have been studied by a
number of investigators and the most comprehensive being the Nationwide Urban
Runoff Program (NURP) in the early 1980s by the United States Environmental
Protection Agency (USEPA). This program was implemented in 28 urban areas
throughout the United States in which samples from 2300 rainfall events were
analyzed. The major output of the NURP studies was the development of urban
runoff pollution loading factors in the form of event mean concentration (EMC)
(Taebi and Droste, 2004).
Due to the interaction of various factors, non-point source pollution is
characterized by random process, complex mechanism, widespread distribution and
latent occurrence. Thus study and control of non-point source pollution remained a
9
major research topic. Although many loading estimates have been reported for
various land use, high variability and inconsistencies exist among reported values.
These differences may represent real variations or differences in sampling and
analytical methods (Mulcahy, 1990).
2.2
Runoff Quality
Urban stormwater discharge during wet weather flow is a major contributor
to the pollution of many receiving waters (Saget et al., 1996; Appel and Hudak,
2001; Brezonik and Stadelmann, 2002; Buffleben et al., 2002; Lee et al., 2004). The
loadings and concentrations of suspended solids, nutrients and other contaminants
are typically much higher in urban stormwater runoff than in runoff from unimpaired
and rural areas (Sartor and Boyd, 1972; Chiew et al., 1997). Urban surface runoff
accounted for most of the negative effects observed in rivers, lakes and other
receiving waters downstream or within urban areas. These negative effects include
accelerated river banks erosion, devastation of river habitats, faster eutrophication
rates in lakes, and a decline in receiving water quality. It was also observed that
discharge of a large storm event may shock the receiving water body many times
greater than an ordinary sanitary effluent load (Loehr, 1974; Bedient et al., 1978;
WEF and ASCE, 1998; Lee and Bang, 2000).
The development and corresponding alteration of natural areas can produce
two major sources of contamination. The first source is associated with land clearing
and the second source is produced by the wide range of human activities that
introduce toxic substances and other contaminants to urban runoff. The runoff event
produces a complex yet reasonably predictable change in water quality conditions,
some of these changes may only last for a matter of seconds or minutes (Herricks,
1995). One of the phenomena that is often found is the first flush. Contaminants
and other material accumulate between storms. These accumulations will be washed
away by the first stormwater flows, creating the possibility for a high concentration
transient of contaminants in the early stage of a runoff event.
10
While quantity control techniques have been well established, a much more
difficult task is the water quality control of urban runoff. As urban runoff relate to
stormwater discharges and combined sewer overflows, it is important to understand
the types of pollutants that are present, or are expected to exist, as well as their
potential impacts on receiving water bodies.
i)
Suspended Solids
The most prevalent form of stormwater pollution is the presence of
suspended matter that is either eroded by stormwater or washed off from paved
surfaces (Novotny and Olem, 1994). Suspended solids cause a number of
environmental impacts such as increase in turbidity of the receiving water, thereby
reducing the penetration of light, and subsequently decreasing the activity and
growth of photosynthetic organisms. It also degrades the aesthetic value of natural
waters. The solids that settle in the receiving water could pose long-term threats due
to increase oxygen demand and gradual accumulation of toxic substances (Moffa,
1990).
ii)
Oxygen Demanding Matter
Sufficient levels of dissolved oxygen (DO) in the water column are necessary
to maintain aerobic conditions to support aquatic life. The influx of stormwater
containing organic and other oxidizable matter may exert substantial oxygen demand
on the water column, thus impairing water quality by depleting the DO level. These
impacts are estimated either by direct measurement of DO or by indirect measures of
biochemical oxygen demand (BOD), chemical oxygen demand (COD) and total
organic carbon (TOC).
iii)
Nutrients
Despite its abundance in the atmosphere, nitrogen is often the most limiting
nutrient for plant growth. This problem occurs because most plants can only take up
nitrogen in two forms: ammonium ion (NH4+) and the ion nitrate (NO3-). Humans
activities have severely altered the nitrogen cycle. Some of the major processes
11
involved in this alteration include: the application of nitrogen fertilizers to crops has
caused increased rates of denitrification and leaching of nitrate into groundwater.
The additional nitrogen entering the groundwater system eventually flows into
streams, rivers, lakes, and estuaries. In these systems, the added nitrogen can lead to
eutrophication (Figure 2.1).
Lightning Fixation
Precipitation
Fossil Fuel Emissions
Gaseous
Atmospheric
Nitrogen Store
Bacteria Fixation
Gaseous
Losses
N2 & N2O
Runoff
Fertilizers
Leaching
Eutrophication
Organic Matter
(R-NH2)
Denitrifcation
Plant
Consumption
Mineralization
Ammonium
(NH4)
Leaching
Nitrification
Nitrates
(NO3)
Nitrification
Nitrites
(NO2)
Figure 2.1: Nitrogen cycle (after Novotny and Olem, 1994)
While for the phosphorus, it originates with the introduction of phosphate
(PO4) into soils from the weathering of rocks (Figure 2.2). Phosphate enters living
ecosystems when plants take up phosphate ions from the soil. Plants take up
waterborne phosphate, which then travels up through successive stages of the aquatic
food chain. Phosphate that is not taken up into the food chain, as well as that in dead
and decomposing organisms, settles on the bottoms of the lake. When these
12
sediments are stirred up, this phosphate may re-enter the biological phosphorus
cycle, but much more of it is buried in the sedimentary rock. Phosphorus also
considered a pollutant when it occurs in excess concentrations in surface waters.
Phosphorus can contribute to over-fertilizing or eutrophication of water bodies.
Feeding by heterotrophs
Cell respiration
Phosphate excreted
Phosphate taken up by
plants. Fixed into organic
phosphate in plant
biomass
Fertilizer
Agriculture
Ecosystem
Mining of phosphate rock
Leaching of
fertilizers
Algae phytoplankton
Sedimentation formation of
phosphate rock
Phosphate in soil
Phosphate dissolved
in water
Millions of years
Figure 2.2: Phosphorus cycle (after Novotny and Olem, 1994)
iv)
Heavy Metals and Other Toxic Constituents
Studies in the United States and Canada indicate that heavy metals were the
most prevalent toxic contaminants found in urban runoff (Marsalek et al., 1997).
Heavy metals that commonly found in urban runoff are lead, zinc and copper. Other
toxic pollutants found in stormwater include phenols and creosols (wood
preservatives), pesticides and herbicides, oils and greases (Wanielista and Yousef,
1993).
13
In general, typical pollutants originating from non-point sources are organic
compounds, measured as COD or BOD, total phosphorus, ammonical-nitrogen,
nitrate- nitrogen, total nitrogen, pesticides, lead, copper, cadmium, zinc and
chromium (Novotny, 1995).
2.3
Control of non point source pollution
For many years, the effort to control storm water discharge has focused on
the quantity and to a limited extent, the quality of storm water. However, the
awareness of the need to improve water quality has increased. At source control
which could prevent the pollutants from coming in contact with runoff is the best
type of control. Source control includes such practices as covering litter control,
street sweeping and preventing illegal dumping (Novotny and Olem, 1994). Besides,
delivery control measures could reduce the pollutant loads which are in transit
between the source area and the receiving water body. The most commonly used
practices are swales, dry and wet detention ponds, constructed wetlands, grassed
water ways and filter trips. To a certain extent, these structures could remove
pollutants in the runoff before they enter the receiving water body.
Detention ponds increase detention times to provide treatment for the
captured first flush runoff to enhance solids settling and the removal of other
pollutants. Constructed wetlands are very popular in stormwater pollution control
and provide very effective treatment for many pollutants. Pollutants are removed in
wetlands by several mechanisms including sedimentation, filtration and vegetative
uptake. Study on wetland harvesting carried out by Supiah (2003) Layang reservoir
in Masai, Johor found that the levels of TP, TDP and Nitrate-N were lower at the
harvested site compared to the control site and inflowing river. Another finding of
this study was the mean percentage removal of Total Phosphorus after harvesting
(66%) was generally higher compared to before harvest (46%). As a conclusion,
wetlands harvesting is a promising technique for reducing nutrient accumulation.
2.4
Pollution Build up and Wash off Process
14
Pollutant buildup on surfaces occurs during dry weather periods and is
subsequently washed off by rainfall. Buildup is a term that represents the entire
complex spectrum of dry weather processes that occur between storms, including
deposition, street cleaning, etc. (Huber, 1986). Wash off is the process of erosion or
dissolution of constituents especially from urban catchments during a runoff event.
Figure 2.3 is a schematic representation of street surface pollutant accumulation
process. Pollutants deposited on road surfaces during a dry period can be carried by
wind and traffic and accumulate near curb or median barrier.
The effects of road surfaces are important for simulating the quality and
quantity of stormwater. Furthermore, many small rainfall events result in runoff
occurring only from those impervious surfaces which are directly linked to the
drainage system with the majority of these surfaces being road surfaces. Road
surfaces, therefore, are an important source of pollutant constituents in stormwater
runoff from urban environments. A number of studies have investigated the links
between pollutants, constituents and vehicular traffic; (e.g., James and Boregowda,
1985; Maestri et al., 1985; Peterson and Batley, 1992). Thus, many urban pollution
studies report street pollution accumulation rates related to the unit length of the curb
instead of an apparently more logical area loading (Ball et al., 1998).
The time elapsed since cleansing by a storm event or street sweeping is
important for simulating urban runoff quality (Sartor and Boyd, 1972; Shaheen,
1975; Pitt, 1979). The procedure used by Sartor and Boyd (1972) required closure of
the road being sampled for a full day to enable sample collection from the road
surface as well as from the gutter adjacent to the road. Sartor and Boyd (1972)
found that 95% of the constituent load was within 1 m of the road immediately
adjacent to the gutter. However, the soluble portion of the constituent load was
minimal in comparison with the particulate fraction (Sartor and Boyd, 1972;
Shaheen, 1975). Ball et al. (1998) studied the build-up of pollutant constituents on
road surfaces in the eastern suburbs of Sydney, Australia in an attempt to improve
this perceived deficiency in the available information.
15
This concept has since been employed in many softwares, such as SWMM
(Huber and Dickinson, 1988), for simulating runoff from urban catchments. In
particular, the time since last cleaning has been used in SWMM as one of the control
parameters for modeling the available constituents on the catchment surface prior to
the commencement of a storm event.
Pollutant build up is usually related to land uses of an urban area and to the
dry periods between the rainfall events. The quantity of pollutants washed off from
impervious area depends primarily on two factors: the amount of pollutant that has
accumulated during the dry period preceding a rainfall event and the characteristics
of the rainfall, especially rainfall volume and intensity. In contrast, the pollutant
buildup on pervious areas is not considered significant, compared to erosion and
dissolution mechanisms by runoff (Adam and Papa, 2000).
Atmospheric Deposition
Pollutants carried away by
wind and traffic
Pollutants emitted from motor vehicles
Litter Deposits
Median
Curb
Pollutants accumulated at road surface
Figure 2.3: Pollutant accumulation on impervious surfaces of urban areas (after
Novotny and Olem, 1994)
2.5
Concentrations
16
2.5.1
Event Mean Concentration
An appropriate and convenient time scale for analysis and evaluation of
urban runoff loads, concentrations and effects is the event duration. To take into
account the hysteretic behaviour of pollutant concentration during a storm event, a
single index known as event mean concentration (EMC) was computed (Adam and
Papa, 2000). The EMC can also be used to compute loadings on the annual time
scale associated with long-term water quality impacts. It should be pointed out that
instead of measuring discrete flow and concentrations throughout the event; the
EMC is represented by the concentration of a flow-weighted composite sample of
the runoff event (Shelley and Gaboury, 1986). Many stormwater monitoring
sampling devices are equipped for flow-weighted composite sampling. Furthermore,
in most cases the total mass and EMC of the pollutant are far more important than
the individual discrete concentrations within the event.
The Nationwide Urban Runoff Program (NURP) EMCs are used widely in
the US for a variety of stormwater management planning purposes. However, a
considerable amount of urban runoff quality monitoring data has been collected in
the US since the completion of NURP studies (Taebi and Droste, 2004). In the US,
an urban stormwater quality data base has been developed by combining different
US data sets including the NURP data, USGS data, urban runoff data collected by
cities and others (Smullen et al., 1999).
A summary of runoff quality parameters for various studies is given in Table
2.1. Taebi and Droste (2004) found highly polluted urban stormwater runoffs in
Isfahan. The mean concentrations in terms of oxygen demanding matter and
suspended solids are much higher than in raw sanitary wastewater. Study by DiazFierros (2002) indicated that the values of total suspended solids and BOD5 are
within the ranges reported previously (Lee and Bang, 2000; Smullen et al., 1999).
The values for TKN and TP are markedly higher, while the values for heavy metals
17
are in general lower. Due to large variation in the water quality, it makes urban
surface runoff quality control more difficult.
Table 2.1: Mean concentrations (mg/l) of stormwater runoff for various types of
studies
Parameters
BOD5
COD
TS
TSS
SS
NO3-N
Tot–KjedalN
Tot -N
PO4-P
Tot -P
Pb
Zn
Fe
EC (µS/ cm)
Droste
and
Hartt
(1975)
150
300
-
NURP
(USEPA
1983)
Smullen
et al.
(1999)
66.1
174
-
Stahre
and
Urbonas
(1990)
10
73
101
1.9
2.98
0.522
300
2.51
0.337
0.175
0.176
-
0.4
-
DiazFierros
(2002)
Taebi
et al.
(2004)
52.8
78.4
-
Lee
and
Bang
(2000)
72.5
172
230
1.7
7.6
123
329
282
23
649
963
149
-
2.39
0.315
0.067
0.162
-
3.4
5.9
0.2
1.8
-
2.2
4
68
-
6.75
0.274
0.314
0.453
507
All parameters in mg/L (Unless otherwise shown)
2.6
Load Estimation
The unit load of contaminants gives a better basis for determining the
management method of the receiving water quality. Reviews and evaluations of
runoff pollution loads and their computational methods were presented by Dolan et
al. (1981), El-Shaarawi et al. (1986) and Brown (1987). These methods included the
direct average method, the flow-weighted-concentration method and the regression
method.
18
In the direct average method, the average load is obtained by multiplying the
measured daily flow mean by the observed mean concentrations. In the flowweighted-concentration (FWC) method, the average load is multiplied by the ratio of
mean-sample and mean-population flows to adjust for potential differences in flow
distributions of the sample and flow data. The regression method follows the FWC
method and an exponent factor is added to account for potential bias associated with
the correlation between concentration and flow and in its modified version, a
variance constant similar to that described for the FWC method is also added
(Brown, 1987).
Pollutants loads of major pollutants in urban domestic wastewater are shown
in Table 2.2. Lee and Bang (2000) based on study at Chongju, Korea, found COD
and BOD5 loads of 695 (kg/ha/yr) and 202.3 (kg/ha/yr), respectively. It is within the
ranges of the other studies. But the values of Total-P showed the highest value. The
TSS and COD loadings tend to be higher with the increasing of rainfall intensity
reported by Taebi and Droste (2004) at Isfahan, Iran.
Table 2.2: Ranges of pollutants loads for various types of studies (in kg/ha/year).
1
1
2
3
3
3
separate combined Residential
Low
Mild
High
sewers
sewers
precipitation precipitation precipitation
TSS
350 - 100 - 3500
1803
97
410
820
2300
COD
22 - 1100 62 - 2000
695
421
1785
3570
BOD5
35 - 210 85 - 800
202.3
NO3- N
1.6
NH4- N
1 - 25
15 - 85
14.8
0.2
1
2
TP
0.5 - 4.9 2.2 - 8.8
TKN
22.4
1
2
3
References: Ellis (1991)
Lee and Bang (2000)
Taebi and Droste (2004)
Parameters
19
2.7
Hysteresis Loop
In order to better understand the effects of urbanization upon pollutant
transport, concentration discharge (C/Q) relationships were investigated. The
hysteresis loop is a method of presenting the sequential relationship between
discharge and concentration through time (Rose, 2003). Williams (1989) proposed
five common classes of such relationships i.e. single valued (straight or curved),
clockwise loop, counter clock wise loop, single-valued plus a loop, and figure eight
(Table 2.3).
Table 2.3: Summary of C/Q relationship (after Williams, 1990)
Class Relationship
C/Q criteria
Reference
I
Single valued line
(C/Q)R ≅ (C/Q)F
Wood (1977)
II
Clockwise loop
(C/Q)R> (C/Q)F for all values
Paustin and
of Q
Beschta (1979)
Counterclockwise
(C/Q)R < (C/Q)F for all values
Axelsson
loop
of Q
(1967)
Figure eight
(C/Q)R > (C/Q)F for one range
Arnborg et al.
of Q values
(1967)
III
IV
(C/Q)R < (C/Q)F for other
range of Q values
Single-valued line is the simplest type of C-Q relation. Its unique
characteristic is that for the same value of Q any C/Q ratio on the hydrograph’s
20
rising limb equal the C/Q ratio on the falling limb. Three subgroups of this singlevalued-line class are the straight line, the curve bending upward and the curve
bending downward (Figures 2.4a, b & c). In all the three subgroups, C increases as
Q increases.
C
CONCENTRATION
CONCENTRATION
DISCHARGE
Q
DISCHARGE
DAYS
(a) Straight line
CONCENTRATION
C
CONCENTRATION
DISCHARGE
Q
DAYS
DISCHARGE
(b) Curve bending upward
21
DISCHARGE
Q
CONCENTRATION
CONCENTRATION
C
DISCHARGE
DAYS
(c) Curve bending downward
Figure 2.4: Three subgroups of single-valued-line class
Clockwise loops occur if the sediment or solute peak arrives at the
catchment’s outlet before the water peak discharge and both graphs have about the
same skewness (Figure 2.5). As such, a C value on the rising limb of the Q-graph is
greater than for the same discharge on the falling limb. Similarly, the C1 / Q1 ratio at
any chosen time on the rising limb of the Q-graph is greater than for the same
discharge on the falling limb.
Q
DAYS
CONCENTRATION
CONCENTRATION
DISCHARGE
C
DISCHARGE
Figure 2.5: Clockwise loop C-Q relationship
22
For the counterclockwise loop, the C peak arrives later than the Q peak (same
skewness for both graphs). Therefore, C-values on the rising limb of the Q- graph
are lower than those on the falling limb (Figure 2.6). Similarly, the C/Q ratio on the
rising limb of the Q-graph is less than the ratio for the same discharge on the falling
limb.
Q
CONCENTRATION
CONCENTRATION
DISCHARGE
C
DAYS
DISCHARGE
Figure 2.6: Counterclockwise loop C-Q relationship
The C-Q relationship for a single hydrologic event can also plot as a singlevalued relationship in one range of water discharges and a sequential loop in the
adjoining range of discharges (Figure 2.7).
CONCENTRATION
Q
CONCENTRATION
DISCHARGE
C
DAYS
DISCHARGE
23
Figure 2.7: Single-valued plus a loop C-Q relationship
Various- skewness experiments revealed that when C peak preceeding Q, a
clockwise loop is produced. In contrast, when Q peak first, a counterclockwise will
occur. Instead, under certain conditions, the hysteresis will form a figure –eight C-Q
loop (Figure 2.8).
CONCENTRATION
Q
CONCENTRATION
DISCHARGE
C
DAYS
DISCHARGE
Figure 2.8: Figure eight C-Q relationship
Many factors affect C-Q relationship such as precipitation intensity and area
distribution, runoff amount and rate, travel rates and distances of floodwaters in the
main channel. C-Q relationship also depends in part on: a) the timing and amount of
sediment on pollutants that arrive at the measuring site, and b) the proximity of the
sources to the measuring site. This involves the travel time or travel rate of the
suspended sediment relative to the travel rate of the water wave (Walling and Web,
1986).
24
Evans and Davies (1998) provide an integrated means of relating C/Q
hysteresis loops to those mixing processes that generate storm flow. It has been long
noted that solute concentrations (C) vary systematically with respect to rising and
falling limb discharge (Q) on the storm hydrograph (Toler, 1965). Such a variation
often results in a non-unique solute concentration for a given value of stream
discharge or hysteresis. Evans and Davies (1998) managed to systematize the
interpretation of C/Q plots by modeling variable contributions of ideal twocomponent (pre-event and event water) and three-component (ground water, soil
water, and event water) mixtures and devised a nomenclature describing the rotation,
curvature and trend of the resulting hysteresis loops.
The above model was used by Evans et al. (1999) to interpret a two-end
member source (shallow and deep saturated subsurface flow) for sodium within
Catskill Mountains (New York) watersheds. They also utilized mixing line plots in
conjunction with hysteresis plots to interpret possible sources of stream flow.
According to Hornberger et al. (2001) it is possible to interpret solute-discharge
hysteresis in terms of ‘non-conservative’ chemical dynamics, when reactions such as
‘leaching’ occur as fast as or faster than the operative hydrological mixing processes.
2.8
Concepts of first flush
The concept of the first flush was first advanced in the early 1970s (Schueler
1994). It assumed that the stored pollutants that had accumulated on paved surfaces
in dry weather quickly washed off during the beginning of the storm. Although the
runoff rate may be greater in the remainder of the storm duration, the stored mass of
pollutants available for washoff was almost depleted, and consequently the
concentration of pollutants declined rapidly (Sansalone and Christina, 2004; Deng,
2005).
25
Modern transportation and urban activities generate suspended and dissolved
solids generally characterized as aggregate parameters [for example suspended solids
concentration (SSC) and total dissolved solids (TDS)] that are of concern not only as
anthropogenic solids but also because such solid matter may serve as a vehicle for
transport of constituents such as metal species and generate an oxygen demand in
receiving waters. Incident of high SS concentrations in the beginning of flood wave
during storm events in combined sewers are common (example in Figure 2.9). This
occur when deposits previously accumulated during dry weather periods are present
and eroded by overland flow (Stotz and Krauth, 1984; 1986; Johansen, 1985;
Pearson et al., 1986; Davies, 1987; Thornton and Saul, 1987; Philippe and Ranchet,
1987; Hvitved-Jacobsen et al., 1987).
First flush phenomenon has been observed for a variety of events including
the removal of heavy metals from rooftops (Forster, 1996; He et al., 2001), the
removal of oil and grease from roadway surfaces (Stenstrom et al., 1984), the
washoff of nitrate from roadway surfaces (Cordery, 1977; Barrett et al., 1997; Lee
and Bang, 2000) and the removal of particulate matter (Maidment, 1993; Wanielista
and Yousef, 1993; Deletic, 1998; Wu et al., 1998; Appel and Hudak, 2001; Lee et
al., 2001; Farm, 2002).
Flow (m3/ s)
SS (mg/ l)
0.3
Flow (m 3/s)
700
2nd Flush peak
First flush peak
600
0.25
500
0.2
400
0.15
300
0.1
200
0.05
100
0
0
4:00
4:20
4:40
5:00
Time
5:20
5:40
6:00
Figure 2.9: Typical first flush of solids
Solids (mg/l)
0.35
26
There are three methodologies commonly found in the literature to quantify a
mass or concentration of first flush (Table 2.4). However, they are conceptually and
mathematically the same.
2.8.1
Analysis by difference between dimensionless Cumulative Pollutant and
Volume Mass
Gupta and Saul (1996) defined the first flush as a portion of the storm up to
the maximum divergence between the dimensionless cumulative percentage of
pollutant mass and the cumulative percentage of volume plotted against the
cumulative percentage of time (Figure 2.10). This definition is more restrictive than
the previous definition of Helsel et al. (1979), proposed to simply define the first
flush when M(V) curve is above the bisector.
Cumulative of mass and flow
1
0.8
0.6
Maximum divergence
0.4
0.2
0
0
0.2
0.4
0.6
0.8
1
Cumulative time/ total time
Figure 2.10: Definition of the first flush (after Gupta & Saul 1996)
27
2.8.2
Dimensionless cumulative analysis
In this technique, a first flush presents when the cumulative M/V ascended
above the 45° line (Figure 2.11). M is normalized cumulative mass and V is
normalized cumulative volume. The 45° line represents the case when the
concentration of pollutants remained constant through out the storm runoff.
Conversely, a dilution will result when the cumulative M/V is below the 45° line
(Geiger, 1987; Lee et al., 2002).
Flush Definition
Cumulative load (%)
100
First Flush
50
45 0
Dilution
0
0
50
100
Cumulative volume (%)
Figure 2.11: Cumulative mass/volume ratio for determining the presence first flush
(Geiger, 1987)
2.8.3
Analysis by cumulative curve ratio
The deviation of the cumulative pollutant mass curve upward from the
diagonal line corresponds to the strength of the first flush. This deviation is inverse
in relation to the first flush coefficient, b. In this variation, M(t) is related to V(t)
through the following expression:
M(t) = [V(t)]b
where b is the fitted exponential parameter.
(2.1)
28
Values of b less than 1 indicate the occurrence of first flush, i.e., a
disproportionately high mass delivery of mass. Values of b greater than 1 indicate
that a first flush is absent (Saget et al., 1996; Bertrand- Krajewski et al., 1998).
Alternatively, common first flush criteria have been defined as the first 20 L
of runoff from elevated bridge scuppers (Drapper et al., 2000); the first 1.27 cm (0.5
in.) of runoff per contributing area (Grisham, 1995); the first 1.27 cm (0.5 in.) runoff
per contributing impervious acre (first 3.14 cm per contributing hectare) or the
volume of runoff produced by a 1 in. storm (Schueler, 1987); the volume of water
obtained by a 1.9 cm (0.75 in.) rainfall event (State of California, 2001). Recently,
the term water quality volume (WQV) has been used to represent the volume of
runoff constituting the first flush (City of Boise, 1998; Barrett, 1999; State of Idaho,
2001).
However, there is a controversy about the exact quantitative definition of the
first flush. The first flush as a design parameter has been variously quantified as: (1)
the first half inch (13 mm) of rainfall; (2) the first half inch of runoff; (3) the first
inch (25 mm) of rainfall; and (4) the first inch of runoff, or other depths of
precipitation or runoff. Definition (2) generally means that during a storm event the
first 1/2 in. of runoff washes off the majority of pollutants accumulated on the
watershed. The half-inch rule simply requires that the one-half inch of runoff be
treated from the total area of the site. The design water quantity volume of the first
flush is calculated by multiplying the depth of 0.5 in. by the total site area (Deng et
al., 2005).
29
Table 2.4: Summary of methodologies for determining first flush (after Sansalone and Christina, 2004)
Method
Mass based
Concentration
based
Empirically
based
Expression
M(t) > V(t)
1. High initial concentration
2. Rapid concentration decline
3. Relatively low and constant
concentration for duration of event
Land use
T
T
Mixed
Mixed
T
T
T
R&I
Mixed
RR
CM & R
T
Mixed
T
Mixed
U
RR
Mixed
R
All
All
All
Drainage area (ha)
0.03–0.054
0.6
0.02–0.03
94
0.03
0.03
NR
4.3
0.7–190
233–609
NR
87–558
0.03–11.5
0.02–0.03
0.15–0.45
NR
NR
NR
658.12
55–131
Variable
Variable
Variable
Investigators
Cristina and Sansalone (2003)
Barbosa and Hvitved-Jacobsen (1999)
Deletic (1998)
Larsen et al. (1998)
Sansalone et al. (1998)
Sansalone and Buchberger (1997)
Stahre and Urbonas (1990)
Farm (2002)
Lee et al. (2001)
Appel and Hudak (2001)
He et al. (2001)
Lee and Bang (2000)
Barrett et al. (1997)
Deletic (1998)
Wu et al. (1998)
Maidment (1993)
Wanielista and Yousef (1993)
Forster (1996)
Stenstrom et al. (1984)
Cordery (1977)
Schueler (1987)
Grisham (1995)
State of California (2001)
First 1.27 cm of runoff per impervious acre
First 1.27 cm of runoff per contributing area
Runoff volume produced by 1.9 cm of
rainfall
Exponential decline
T
Variable
Sartor and Boyd (1972)
Runoff volume produced by 2.54 cm of
All
Variable
Schueler (1987)
rainfall
Linear/multiple linear regression
U
41–121
Gupta and Saul (1996)
Note: T=transportation; RR=roof runoff; I=industrial; U=urban; CM=commercial; R=residential; NR=not reported; M(t) =mass; and V(t)=volume.
30
There are many factors that affect the occurrence and impact of a first-flush
event, such as the size and drainage characteristics of the catchment area, the
imperviousness of the basin, the mobility and property of the pollutants. Four
important factors may be responsible for the first-flush effect in a drainage system as
follows (Wanielista and Yousef, 1993; Gupta and Saul, 1996):
i)
Volume and intensity of rainfall: These are predominant factors in the
generation of runoff rate. A high-intensity rain may produce a first-flush either by
dislodging deposited pollutants from the catchment surface or forcing the
accumulated solids out of the sewer system.
ii)
Catchment characteristics: Catchment surface conditions and
gradients have significant effects on controlling and delivering pollutants to the
sewer system. The first-flush may be more distinct for impervious areas, where the
supply of pollutant is limited. For example, sediment washing off the pervious area
from erosion, may not show a first-flush because the supply of soil particles is not
limited. First flush is most readily observed on small catchments or individual
premises, particularly if a high proportion of the catchment is impervious (such as
paved surfaces and roads) (Deng, 2005).
iii)
Drainage system: Characteristics of drainage system may prevent or
accelerate the first-flush effect. For example, in large catchment the initial runoff
from remote parts of the catchment requires a longer time to reach the catchment
outlet and a first-flush may not be distinct. Moreover, drainage system characteristic
such as effective transport capacity and roughness may influence the transport of
pollutants.
iv)
Pollutant mobility: Runoff may not remove certain pollutant such as
oil and grease, and fine particles, as easily as others. Soluble pollutants exhibit the
least tendency toward a first-flush effect, and delayed flushes of those pollutants are
common. As such, first flush may not happen in some cases. However, a first-flush
is more distinct on small catchments with high impervious areas. In this case, the
existence of first-flush could provide an opportunity for controlling runoff pollution.
31
Sansalone and Cristina (2004), at Baton Rouge, Los Angeles, observed frequent first
flush in small watersheds but are less evident for complex watersheds, especially
when there is a multitude of sub watersheds, each possess unique travel times to the
outlet.
The existence of first flush of pollutants provides an opportunity for
stormwater managers and engineers to control water pollution in an economic and
efficient way. If most of the urban-surface pollutant load were transported during
the initial phase of a storm, then a rather small volume of runoff storage would be
needed to treat and remove the bulk of urban-surface pollutants.
As a result, controlling the first flush has become the most practiced criterion
for the design of stormwater treatment facilities. First flush collection systems are
employed to capture and isolate this most polluted runoff, with subsequent runoff
being diverted directly to the stormwater system and finally into the receiving
environment (Deng, 2005).
2.9 Non point source pollution simulation models
Management of the quantity and quality of stormwater runoff from urban
areas is a complex task which has become an increasingly important environmental
issue for urban communities (Choi and Ball, 2002). Two standard approaches for
obtaining the desired information are:
• monitoring of the system for both water quantity and quality, or
• implementation of catchment modelling systems which simulate the
important processes influencing the quantity and quality of stormwater
runoff in the urban environment
Difficulties in gathering storm runoff data and constraint of time and
financial for an effective pollution control and management make a complete
characterization of urban stormwater runoff for every catchment merely impossible.
32
The computer models for predicting quantity and quality of urban stormwater runoff
provides an economic yet reliable choice (Jewell et al., 1978). Non point source
pollution models are used to meet several objectives, including:
•
Characterize runoff quantity and quality as to temporal and spatial
detail (concentration/load ranges, etc.)
•
Provide input to a receiving water quality analysis (e.g. drive a
receiving water quality model)
•
Determine effects, magnitudes and optimum locations and
combinations of control options
•
Perform frequency analyses on selected quality parameters (e.g. to
determine return periods of concentrations/ loads)
•
Provide input to cost-benefit analyses.
Other methods, such as regression models are easier to implement but only
suitable to predict annual pollutant loads (Driver and Lynstrom, 1986; Huber, 1988)
and not suitable for a detailed analysis of the runoff. Computer models are cost
effective and reasonably accurate substitutes for extensive field data gathering
programs, provided that the models are calibrated and verified for conditions of the
catchment being studied.
There are many well known models in the literature such as U.S. Geological
Survey (USGS), Storage, Treatment, Overflow, Runoff Model (STORM), Storm
Water Management Model (SWMM), Hydrological Simulation Program- Fortran
(HSPF) etc. These models are appropriate for full-scale simulation models of urban
areas. Others like Hydrologic Modeling System (HEC-HMS) and Soil Conservation
Service (SCS) might also useful in the hydrologic aspect of water quality studies but
do not simulate water quality directly. In USGS, runoff generation and subsequent
routing use the kinematic wave method and parameter estimation assistance is
included in the model. Water quality is simulated using buildup and washoff
functions with settling of solids in storage units dependent on a particle size
distribution (Alley, 1986). STORM uses simple runoff coefficient, SCS and unit
hydrograph methods for the generation runoff depths and rainfall inputs. The
buildup and washoff formulations are used for the simulations of six prespecified
33
pollutants. However, the model can be manipulated to provide loads for arbitrary
conservative pollutants (Najarian et al., 1986). In this study, SWMM was used and a
detail about SWMM is discussed in sections 2.9 and 2.10.
2.10
Storm Water Management Model (SWMM)
Storm Water Management Model (SWMM) was developed in 1971 for the
US EPA to address deal with the quantity and quality variations of urban runoff
(Metcalf and Eddy et al., 1971; Huber, 1995). The model can be used for storm
event or continuous simulation and has undergone a number of updates and
improvements (Huber and Dickinson, 1988). Many mathematical formulations
proposed in SWMM, especially those concerning water quality simulations, have
been re-used in more recently developed models. The latest version of SWMM
benefited greatly from experienced garnered from many years of use, allowing it to
remain a reference in the field of urban runoff quality modeling (Gaume et al.,
1998).
Calibration of a catchment modeling system requires that the control
parameters (input data) for each of these conceptual components must be
determined. The resultant system of process models and input information must be
able to mimics the real response of the catchment. Since surface runoff varies with
the catchment characteristics, calibration of a catchment modeling system usually
requires adjustment of the model control parameters to minimize prediction errors
(Choi and Ball, 2002).
In the calibration process, one of the alternatives can be described as a ‘trial
and error’ basis whereby the values of the control parameters are modified in a
systematic manner to achieve correlations between the monitored parameters and the
predicted parameters describing the catchment response. This approach is used
when calibrating a catchment modeling system with monitored information as shown
in Figure 2.12.
34
START
CATCHMENT INFORMATION
Modeller
Interpretation
INPUT INFORMATION TO
CATCHMENT MODELING SYSTEM
Modeller
Interpretation
ADJUST INPUT
INFORMATION
SIMULATION
OUTPUT SIMILAR TO
RECORDED DATA
No
Yes
APPLICATION
STOP
Figure 2.12: Calibration procedure for SWMM (after Choi and Ball, 2002)
Some calibration procedures have been proposed to improve the model
performance. Jewell et al. (1978) showed how the quantity and the quality of runoff
can be calibrated for an urban catchment using measured data from several single
storm events. Maalel and Huber (1984) proposed a calibration procedure using
multiple single events and a continuous simulation. Their findings support Jewell et
al. (1978), that a calibration using conditions of several storms improves the
reliability of the results and reduce the predictive errors. Continuous simulation can
be used to compare statistics between observed and predicted events. Warwick and
Tadepolli (1991) and Gaume et al. (1998) found good agreements between the
simulated and observed storm water quality using SWMM. Calibration and
validation of SWMM for the exponential washoff studied by Sriananthakumar and
Codner (1995) in Giralang, Canberra was carried out separately for low and high
runoff events to increase prediction capability.
35
2.10
Sensitivity Analysis
A model sensitivity analysis is a valuable tool for setting up, improving,
testing and calibrating hydrological models. According to Lane and Richards (2001)
and Rabitz (1989) a sensitivity analysis can:-
(i) identify the parameters the model reacts most sensitively to and thus
simplify and accelerate the calibration or enable a more focused planning for
future research and field measurement,
(ii) show whether the model’s reaction to representative variations of
parameter values and boundary conditions is realistic,
(iii) prove the model concept is sufficiently sensitive to represent the natural
system’s behaviour, and
(iv) reduce a model to its essential structures.
Thus, sensitivity analysis can prove the suitability of a model concept and as
Saltelli (2000) pointed out, strengthen trust in a model and its predictions. In
SWMM, the results of the sensitivity analysis can be used for estimating the default
calibration parameters as well as evaluating the comparative performance of
parameters using the default values. By doing so, this procedure was expected to
reduce the amount of effort and time required to calibrate a model (Abustan, 1997).
CHAPTER 3
METHODOLOGY
This chapter describes the study sites characteristics and methods of data
collection in terms of water quantity and quality. Flow measurement is very
important for developing rating curves. The sampling procedure and analytical
analysis to get reliable data are discussed in detail. In addition, formula for
calculating pollutant concentration as well as pollutant loading and first flush
analysis are presented. Finally this chapter describes the modeling approaches
including estimation of local build up data and washoff parameters.
3.1
Catchment Description
Sampling of urban runoff was carried out in two small catchments, namely
residential and commercial in Skudai, Johor. These two catchments are located in a
bigger Melana catchment (Figure 3.1). The residential catchment is 3.34 ha and the
commercial catchment is 0.75 ha. Each catchment is dominated by only one type of
urban land use. Development in the residential catchment is mainly single storey
houses with 147 houses. The catchment represents a typical high density low cost
residential area (Figure 3.2a). It was estimated that about 15% of the catchment is
still pervious while the rest are roofs, pavement and road surfaces. Open spaces are
located at the outlet of the catchment.
For the commercial catchment, the buildings include restaurants and shops.
Based on the layout plan, about 20% of the catchment is still pervious mainly
covered by grass land while the rest are roofs, pavement and road surfaces (Figure
3.2b). Delineation of the catchment boundaries were based on topo-map and on site
37
observations of overland flow pathways at several locations during storms. Field
observation was useful in determining the runoff contributing areas to the catchment
outlets for such a flat area.
Taman Universiti
Figure 3.1: Location of the study site
TAMAN UNIVERSITI – JALAN
PERTANIAN
N
Catchment outlet
N = 1º 32’ 24”
E = 103º 37’ 6”
Legend
Houses
Pervious area
Pavement roads
Monsoon drain
Rain gauge
Catchment boundary
Drainage network
Scale 1: 4,200
Figure 3.2a: Residential Catchment
38
TAMAN UNIVERSITI – JALAN
KEBUDAYAAN
N
Catchment outlet
N = 1º 32’ 15”
E = 103º 37’ 20”
Legend
Building
Pervious area
Residential area
Pavement roads
Scale
Rain gauge
Catchment boundary
Drainage network
1: 7,000
Figure 3.2b: Commercial Catchment
The catchment drainage system separates stormwater system from the sewer
system. The separated system comprises separate system of pipes to handle sanitary
sewage and stormwater. Total lengths of the drain in the residential and the
commercial catchment were 2614 m and 509 m, respectively. The catchments
physiographical conditions are shown in Table 3.1.
39
Table 3.1: Physiographical conditions of sub catchments
Unit
Residential
Low cost residential:
high density singlestorey houses
3.34
Commercial: shops;
restaurants
%
66
45.5
N-E
2.53
53.5
47.5
S-W
1.8
%
1.08
0.66
km/km2
78.3
65
m
2614
509
Land use
Area
Elevation
Maximum
Minimum
Aspect
Average slope
Average drain
gradient
Drainage density
Total drainage
length
3.2
ha
m.a.s.1
m.a.s. l
Commercial
0.75
Rainfall Measurement
Rainfall data are very important to describe the runoff behaviour. Besides,
this information is required for modeling purposes. For example, SWMM requires
spatial information on rainfall and flowrate for water quantity analysis. The
collection of urban stormwater data must be done in a proper way due to the
extremely variable rainfall, rapid response times and varying water levels. Thus,
rainfall data should be available at 5-min or shorter increments to predict the runoff
hydrograph adequately (Bedient and Huber, 1986).
The study site was equipped to monitor rainfall data continuously. Rainfall
data was measured and digitally recorded by ISCO™ tipping bucket rain gauge, with
a resolution of 0.1 mm per tip. The rain gauge was installed on a leveled platform on
a house roof to make sure there is no surrounding obstacle. Two gauges were used
in this study, located at about 0.2 km from the residential catchment’s outlet and 1
km from the commercial catchment’s outlet. The gauges were maintain and serviced
weekly. Site maintenance includes checking the rain gauges and downloading the
data from the data loggers. Fine litters and dust trapped in the collecting rain gauges
were removed.
40
Figure 3.3: A rain gauge installed on roof top in the residential catchment
3.3
Flow measurement
The objectives of the streamflow gauging are to measure discharge at study
the catchment’s outlet and to develop discharge-level rating curves. The standard
velocity-area method of gauging was used in this study. A current meter (Model
SWOFFER 2100) was used to measure the flow velocities. The stream discharges
were then calculated as the product of the flow velocities and the drain cross section
area for various water levels. During each sampling, the water level was monitored
and recorded manually. The channel water level-discharge rating curve was
developed at the outlet of the catchment to convert the water level into discharge.
Stormwater volume was measured between November 2003 and November 2004 for
41
both catchments. A total of 16 storm events for residential and commercial
catchments were investigated.
Velocities and water level were taken on two separate events for the purpose
of calibration and verification as well as to develop rating curve. For the residential
catchment, the first event was on November 4, 2003 and followed by the second
event on November 8, 2003. As for the commercial catchment, the first and the
second observations were on March 11 & March 16, 2004. Figures 3.4a & b show
the water level – discharge rating curve for both catchments.
Water Level-Discharge Rating Curve
0.40
0.35
y = 2.8128x + 0.0932x
0.30
R = 0.9966
2
2
Q (m 3/s)
0.25
0.20
0.15
0.10
0.05
0.00
0
0.1
0.2
H (m)
0.3
0.4
a) Residential
Water Level-Discharge Rating Curve
0.2
0.16
2
Q (m 3/s)
y = 1.1444x - 0.0531x
2
0.12
R = 0.9808
0.08
0.04
0
0
0.1
0.2
H (m)
0.3
0.4
b) Commercial
Figure 3.4: Water Level - Discharge Rating Curve for both catchments
a) Residential catchment and b) Commercial catchment
42
3.4
Sampling Procedure
Samples are collected manually instead of using automatic water samplers.
This is due to safety factor where both sites are very open to public. Manual
sampling also provide opportunities to collect more samples at preferred time
intervals. Since this method require researcher to be at site during storm, important
events happening at site can be recorded.
There are some procedures which must be carefully followed during
sampling and laboratory analysis of water samples. If the data are not consistent,
there are many potential causes that need to be checked. These include problems in
sampling (e.g., bias, lack of representativeness, interferences), problems in analyses
(e.g., bad preservation and storage of samples, inadequate analytical procedures,
human or technical errors) and specific local conditions which may explain the
results (e.g., scouring of contaminated deposits). All bottles used for collecting
samples were washed with dilute acid (sulfuric or hydrochloric) and thoroughly
rinsed with deionized-distilled water. It is very important to keep these bottles clean
to ensure they were free of any contamination.
Stormwaters were grab-sampled using a 1-litre polyethylene bottle. To
define pollutant loads from urban catchments with reasonable degree of accuracy,
Bedient et al. (1980) suggested collecting a minimum of four samples during the
period increasing flow and six samples on the falling time of storm hydrographs.
Samples were taken when there was noticeable change in water level. For every
sample taken, the sampling time and water level were recorded. About 10 to 15
samples were collected during each storm on both the rising and the falling limb of
the hydrograph (Figure 3.5). In the small catchment, more samples were collected
on the rising limb than on the falling limbs. This is important to get a better
description of the first flush phenomenon.
43
Discharge/ Water Level
Collected Samples
Time
Figure 3.5: Typical stormwater sampling interval employed in this study
3.4.1
Analysis of water quality
Water samples were sent for analysis within 24 hours after sampling.
Electrical conductivity, turbidity, TDS and pH were analysed immediately upon
arriving at the Environment Laboratory in the Faculty of Civil Engineering, UTM.
The EPA (1984) listed five categories of pollutants which are commonly used to
characterise non-point source pollutants i.e., total suspended solids, oxygenconsuming constituents, nutrients, metals and bacterial constituents.
Water samples were analysed for Biochemical Oxygen Demand (BOD5), Chemical
Oxygen Demand (COD), Suspended Solids (SS), Nutrients (NO3-N, NO2-N, NH3-N
and P) and heavy metal (Pb). The laboratory analysis followed the Standard
Methods for the examination of water and wastewater (APHA et al., 1995). The
analysis of phosphorus was carried out using Inductively Coupled Plasma -Mass
Spectrometry (ICP - MS). Because the response of the mass spectrometer in counts
per second is directly proportional to the concentration of a given element in a
sample, it is relatively easy to calibrate the system using a series of external
standards at least for three different concentrations. The concentrations of standards
must cover the expected P concentrations in the samples. For calibration purposes,
reagent H2NaO4P.2H2O was used to prepare the standard solution. The calculations
of calibration are shown in Appendix A. Based on formula M1V1 = M2V2, three
samples of 40 µg/l, 20 µg/l and 10 µg/l were prepared and tested for the intensity
level. Graph of calibration with r2 = 0.998 is plotted. Any sample entered into the
44
mass spectrometer under exactly the same conditions will provide a count rate,
which can be converted directly to P concentration from the calibration curve.
Table 3.2: The analytical methods used in this study
Parameters
Biochemical Oxygen Demand (BOD)
Analysis
5– Day BOD test
(Standard Method, Section 5219 B)
Chemical Oxygen Demand
Open Reflux Method (Standard Method,
Sec. 5220B)
Suspended Solids
(Standard Method, Section 2540 D)
pH
HACH SensIon 156 (Sension 156
Manual)
Electric Conductivity
HACH SensIon 156 (Sension 156
Manual)
Turbidity
HACH DR/4000 Spectrophotometer
Phosphorus
Inductively Coupled Plasma -Mass
Spectrometry (ICP – MS)
Nitrate-Nitrogen
Screening Method (Standard Method,
Section 4500 B) is used by measuring
UV absorption at 400 nm in the range of
0 to 5 mg/l. (HACH DR/4000 Manual,
Method 8171) using Ultraviolet Visible
Spectrophotometric (UV-Vis) Shimazu,
Model UV-1601 PC
Nitrogen-Ammonia
Screening Method is used by measuring
UV absorption at 425 nm in the range of
0 to 0.25 mg/l. (HACH DR/4000
Manual, Method 8038) using Ultraviolet
Visible Spectrophotometric (UV-Vis)
Shimazu, Model UV-1601 PC
3.5
Event Mean Concentration
45
The use of an event mean concentration (EMC) is appropriate for evaluating
the effects of stormwater runoff on receiving waters. Receiving water bodies
respond relatively slowly to storm inflows compared to the rate at which constituent
concentrations change during a storm event. Thus, EMC is an important analytical
parameter (Lee et al., 2002).
The event mean concentration (EMC) is defined as the total pollutant load
(mass) divided by total runoff volume, for an event of duration t (Sansalone and
Buchberger, 1997; Charbeneau and Barretti, 1998). Mathematically:
EMC =
M
=
Σ Qi Ci ∆t / Σ Qi ∆t
(3.1)
V
where EMC is event mean concentration (mg/l), M is total mass of pollutant over the
entire event duration (g), V is total volume of flow over the entire event duration
(m3), t is time (min), Qi (t) is the time-variable flow (m3/min), Ci is the time-variable
concentration (mg/l) and ∆t is the discrete time interval (min) measured during the
runoff event.
The EMC is computed for the entire runoff duration. For a time less than the
full runoff duration, a partial event mean concentration (PEMC) can be defined as
PEMC = m(t) =
v(t)
Σ Ct Qt ∆t
Σ Qt ∆t
(3.2)
where m(t) is pollutant mass transported up to time t (g); v(t) is flow volume up to
time t (m3); From Eqs. (1) and (2) it follows that M = m(t), V= v(t); and PEMC(t) =
EMC (Sansalone and Buchberger, 1997; Lee et al., 2002).
3.6
Correlation Analysis
46
The correlation between pollutants is used to identify possible relationships
between individual pollutants as well as between different categories of pollutants.
This would be useful when studying areas where not all quality parameters are
measured. When two pollutants have a high correlation, a high value of EMC of one
pollutant may indicate that other unmeasured pollutant would also have a
respectively high value.
3.7
Box plot Analysis
Outliers
Maximum value
The 75th quartile
Median
The 25th quartile
Minimum value
Figure 3.6: Information that can be derived from a box plot
A box plot is a way of summarizing a set of data measured on an interval
scale. It is often used in exploratory data analysis. It is a type of graph which is
used to show the shape of the distribution of a continuous data set, its central value,
and variability. A box and whisker plot provides a 5 point summary of the data
(Figure 3.6).
47
1) The box represents the middle 50% of the data.
2) The median is the point where 50% of the data is above it and 50% below
it (or left and right depending on orientation).
3) The 25th quartile is where; at most, 25% of the data fall below it.
4) The 75th quartile is where, at most, 25% of the data is above it.
5) The whiskers cannot extend any further than 1.5 times the length of the
inner quartiles. Data points that fall outside this range will show up as
outliers.
3.8
Pollutant loading
A unit pollutant loading was calculated by multiplying the nutrient
concentrations of the individual composite samples with the total volume of flow
during the corresponding sampling period. The mathematical formula was given by
Lee (1998):L = EMC x V
where L
= pollutant loading (mg)
EMC = event mean concentration (mg/l)
V
3.9
= total volume of flow over entire event duration (l)
First Flush Concept
(3.3)
48
3.9.1
Dimensionless cumulative analysis
The dimensionless normalized mass and flow volumes are useful for
characterising first flush of storm runoff. Equations (3.4) and (3.5) represent the first
flush phenomenon;
L = m(t)/ M
(3.4)
F = v(t)/ V
(3.5)
where L is dimensionless cumulative pollutant mass; m(t) is pollutant mass transport
up to time t (g); M is total mass of pollutant over the entire event duration (g); F is
dimensionless cumulative runoff volume; v(t) is flow volume up to time t (m3) and V
is total volume of flow over the entire event duration (m3).
A first flush exists at time t if the dimensionless cumulative pollutant mass L
exceeds the dimensionless cumulative runoff volume F at all instances during the
storm events. Values on 45° line, when plotting L vs. F, indicates that pollutants are
uniformly distributed throughout the storm events. If the data for a particular storm
falls above the 45° line, a first flush is suggested and PEMC (tr) > EMC.
Conversely, dilution is assumed to occur when the data falls below the 45° line and a
first flush fails to occur.
3.10
Storm Water Management Model (SWMM)
SWMM is capable of simulating various aspects of urban hydrologic cycle in
terms of the quantity and quality of surface runoff. This model consists of several
blocks: i.e., the Runoff block for runoff estimation, the Transport and Extended
Transport for the routing of this runoff using a kinematic wave technique, the
Storage and Treatment block routes through storage units using the Puls (storage
indication) method and the Receive block for its mixing in a receiving water body
49
(Huber and Dickinson, 1988; Roesner et al., 1988). Hence, there are many
advantages in using it for planning, design and operational purposes.
3.11
Water Quantity
SWMM requires physical and hydrologic parameters to simulate
hydrographs. The physical parameters include the area of the subcatchment, its
slope and the diameter and length of pipes. The hydrologic parameters include the
width of the subcatchments, Manning’s coefficients, depression storages and
infiltration rates or coefficients. These parameters are more difficult to estimate than
the physical ones. Nevertheless, the shapes of the predicted hydrographs are very
sensitive to input values of the hydrologic parameters (Baffaut et al., 1990).
There are many factors affecting the hydrograph such as rainfall intensity,
rainfall duration, the catchment size, slope, storage and morphology, channel type,
land use and percent impervious. Rainfall intensity and duration are the major
causes of the rainfall-runoff process, followed by catchments characteristics. Size,
slope, shape and storage capacity are all important parameters in catchment
geomorphology. Land use parameters can significantly alter the natural hydrologic
response through increases in impervious cover, altered slopes and improved
drainage channel networks. Field data gathering should include all of the dynamic
independent variables and the parameters shown in Table 3.3.
Table 3.3: Independent variables and parameters influencing stormwater pollution
washoff
50
Dynamic Variables
Storm event totals
Instantaneous flux
• time since last storm
event (dry days)
• street cleaning
practices
• total volume of runoff
• storm duration
• total volume of
rainfall
• average rainfall
• average runoff
Parameters
• runoff intensity
•
land use
• cumulative volume of
runoff
• time from start of storm
• rainfall intensity
• cumulative volume of
rainfall
•
area
•
•
% impervious area
length of overland
flow
% street and parking
area
length of streets/ unit
area
population density
particulate fallout rate/
unit area
number of catch –
basins/ unit area
climatological data
temperature
rainfall
•
•
•
•
•
•
•
•
3.12
Hydrologic Parameter
3.12.1 Horton Equation
Horton's model is one of the most commonly used infiltration equation.
Horton (1940) showed that when rainfall exceeded the infiltration rate, water infiltrates
surface soils at a rate that decreases with time. Horton derived infiltration capacity as a
function of time as:
Fp = Fc + (Fo - Fc) e -kt
(3.6)
51
where
Fp
= infiltration capacity into soil, mm/hr
Fc
= minimum or ultimate value of Fp, mm/hr
Fo
= maximum or initial value of Fp, mm/hr
t
= time from beginning of storm, sec
k
= decay coefficient, 1/sec
The maximum or initial infiltration capacity, mm/hr depends primarily on
soil type, initial moisture content and surface vegetation conditions. For single event
simulation, the initial moisture content is important. Typical values of infiltration
capacity are listed in Table 3.4.
Table 3.4: Initial Infiltration Capacity
A.
Representative values of Initial Infiltration Capacity
DRY soils (with little or no vegetation)
127 mm/hr
Sandy soils:
76.2 mm/hr
Loam soils:
25.4 mm/hr
Clay soils:
B.
DRY soils (with dense vegetation)
Multiply values given in A by 2
C.
MOIST soils (for single event simulation)
Soils which have drained but not dried out:
divide values from A and B by 3
Soils close to saturation:
choose value close to saturated
hydraulic conductivity
Soils partially dried out:
divide values from A and B by 1.5-2.5
The minimum or ultimate value of infiltration capacity, mm/hr is essentially
the saturated hydraulic conductivity, or permeability, of soils. Table 3.5 lists ranges
of this parameter for various soil groups.
Table 3.5: Minimum value of infiltration capacity for different hydrologic soil group
52
Hydrologic Soil
Group
A
B
C
D
Ultimate Infiltration
Rate(in/hr)
0.45 - 0.30
0.30 - 0.15
0.15 - 0.05
0.05 - 0
Figure 3.7: Decay rate coefficient
Equation 3.6 is dimensionally homogeneous and the unit used for k should be
the inverse of that used for t. Decay coefficient is a constant that describes the decay
of the capacity with time and reflects the soil-cover complex. The rate of decrease of
infiltration capacity, 1/sec is independent of initial moisture content. Table 3.6
shows the rate of decay of infiltration for a range of parameter values.
Table 3.6: Decay rate of infiltration
Decay Rate
1/hr
(1/sec)
2
(0.00056)
3
(0.00083)
4
(0.00115)
5
(0.00139)
3.13
Routing Method
Infiltration capacity towards
limiting value after 1 hour
76
95
98
99
53
SWMM provides five major types of Hydrograph Generation techniques as
follows:
i)
SWMM Runoff Non-linear Reservoir Method
ii)
Kinematic Wave Method
iii)
Laurenson Non-linear Method/Rafts
iv)
SCS Unit Hydrograph Method
v)
Other Unit Hydrograph methods, Nash, Snyder (Alameda), Snyder,
Rational Formula, Time/area, and Santa Barbara Urban Hydrograph.
This study used kinematic wave method for routing the hydrograph.
3.13.1 Kinematic Wave
The kinematic wave method for overland flow applies only the kinematic
wave component of the flow equations for momentum and continuity. Kinematic
wave procedures apply either the Horton or Green Ampt loss to the pervious
percentage of the subarea, as defined by the % impervious data item.
All flows are assumed to agree with the equations of continuity and
momentum. The continuity and momentum equations for overland kinematic waves
can be reduced to the following two equations, respectively:
i)
Continuity Equation –
∂y + ∂q = i - f = ie
∂t
∂x
and
Momentum Equation =
(3.7)
54
q = aym = 1.49/N So½ yo 5/3
(3.8)
where
y = y(x,t) = depth of overland flow (m),
q = q(x,t) = rate of overland flow/unit width,
i-f = ie = net rainfall rate,
a = conveyance factor = (1/N) So½ when obtained from Manning’s equation,
m = 5/3 when obtained from Manning’s equation,
N = effective roughness coefficient,
So = average overland slope,
yo = mean depth of overland flow.
3.14
Calibration and Validation
Most models for predicting quality of stormwater runoff are coupled with a
quantity simulation model. The calibration of these models necessitate calibrating
the quantity and quality models sequentially, using the output from the calibrated
quantity model as input to the quality model.
Calibration is the adjustment of model parameters using one set of data.
Control parameter values for a catchment modeling system typically are determined
by one of two alternative methods; these alternative methods are:
• modification of control parameter values until the simulated and observed
hydrographs, or other catchment response measure, are similar; and
• selection of control parameter values based on some hydrological, hydraulic
or other characteristics of the catchment.
55
Verification is about testing the established parameter using an independent
data set. Calibration and verification data are usually in the form of measured flows
and concentrations at outfalls. However, it is important to note that detailed shorttime-increment pollutographs during a storm are seldom needed for analysis of
receiving water quality. Hence, total storm event loading or event mean
concentrations are usually sufficient for quality calibration and verification.
Nix (1994) suggested the following technique for calibrating a model and
these steps can be applied for calibrating SWMM.
i)
It is important to adjust the runoff volume parameters. However,
without an accurate representation of those parameters, the process
of adjusting will be much more difficult.
ii)
The hydrograph peak and shape can be adjusted but after the
simulated runoff depths are found compatibly fit with the observed
data.
iii)
A model that is not calibrated for quantity cannot produce reliable
quality results. Therefore, parameters responsible for generating
pollutant loads and their transport through the watershed should be
adjusted after calibrating the quantity parameters.
During calibration it is not necessary to achieve close agreement of measured
and predicted data for individual storm events. The emphasis was on integrated
results over the entire calibration of the data set. If the assumption of representative
data was valid, the calibrated model would give accurate predictions of average
annual runoff and pollutant loadings. Default values found in SWMM were used for
resistance factors, depression storage depths and infiltration rates. If necessary, these
parameters were adjusted during calibration. Before performing calibration process,
it is necessary to ascertain the model output sensitivity to changes in input
parameters.
56
3.15
Sensitivity Analysis
Since models and optimization formulations are mathematical representations
of real systems, there is inherent uncertainty in the output. Moreover, because the
mathematical model cannot respond exactly to the actual system, satisfaction is not
achieved by just obtaining the optimum solution of the mathematical problem (de
Neufville, 1990). A practical approach to treat potential uncertainties is to test the
sensitivity of the analysis to variation in the magnitude of key parameters (James and
Lee, 1971). In the context of modeling in general, the issue of uncertainty in model
output is often not adequately addressed. This uncertainty arises from the model
being an abstraction of reality and the fact that the input parameters are not known
with certainty (Nix, 1994).
Sensitivity analysis can provide estimated parameter values of calibration
processes as well as to observe the sensitivity of the peak flow and time to peak.
This procedure will reduce the time required to calibrate a model. For this study, the
sensitivity analysis was carried out by determining the effect of changing percentage
of the impervious areas, catchment width and impervious depression storage. By
varying one of this input values and holding the other inputs constant, the percentage
changes in response were plotted against the selected input values.
3.16
Goodness of fit
Several statistical methods can be used to compare the shape of observed and
simulated hyrdographs. A number of goodness of fit criteria are available (Fleming,
1975), but the methods used in this study are:
Relative Error (RE)
RE = Observed – Simulated
Observed
(3.9)
57
Absolute Relative Error (ARE)
ARE = Observed – Simulated
Observed
(3.10)
For spatial time series analysis, Root Mean Square Error (RMSE) was used.
n
RMSE = 1
n
∑ | Qs (i) – Qo (i)|
(3.11)
i=1
where
Qs is simulated discharges (m3/s),
Qo is observed discharges (m3/s), and,
n is number of observations in the time series.
For water quality, two parameters were used for assessing the performance of
calibration. These parameters are defined as Relative Load Difference (RLD) and
Absolute Load Difference (ALD) (Baffaut and Delleur, 1990).
N
RLD = 1
N
∑
1
Mi-Pi
Mi
(3.12)
N
ALD = 1
N
∑ Mi-Pi
1
Mi
where
N – number of events used for calibration
Mi – measured event total load
Pi – predicted event total load
(3.13)
58
3.17
Water Quality
The pollutant deposition and wash-off model used in SWMM requires the
calibration of several parameters (Baffaut and Delleur, 1990). Different models
offer different options for conceptual buildup and wash off mechanisms, with
SWMM having the greatest flexibility. In fact, with calibration, a good agreement
can be produced between predicted and measured concentrations and loads
(Novotny, 1995).
3.17.1 Estimation of Buildup Parameters
Deposition and removal of pollutant constituents are a continuous process
and, hence, the loads obtained from any sampling program will show a temporal
variability. These loads would reach a maximum when equilibrium conditions were
achieved. An equilibrium occurs when the rate at which a constituent being
deposited is equal to the rate at which the constituent being removed. In a similar
manner, the loads would be at a minimum immediately after cleansing of the
catchment surface. Consequently, the definition of a cleansing event is important for
the analysis of the temporal pattern of the constituent load (Ball et al., 1998).
For buildup, normalized loadings (e.g., mass/day/area or mass /day/curb
length, or just mass/day) are required, along with an assumed functional form for
buildup versus time (e.g., linear, exponential, Michaelis-Menton, etc.). Ammon
(1979) categorised the buildup relationship into one of four functional forms. These
forms are linear, power function, exponential or Michaelis-Menton.
Pollutant mass with a linear function:
P = ax t
(3.14)
where P is the pollutant load, ax is a coefficient of the linear deposition rate in weight
per day and t is the number of dry days preceding storm runoff
59
An exponential function:
P = PL (1-ekt)
(3.15)
where PL is the maximum amount of pollutant that can be deposited, and k is the
removal rate of pollutant.
A power function:
P = a2 t b
(3.16)
where a2 is a coefficient of the linear deposition rate in weight per day, and b is
exponential coefficient.
A Michaelis-Menton function:
(3.17)
P = PL t/(c+t)
where PL is the maximum amount of pollutant that can be deposited, and c is the
number of days when half of the maximum load has been deposited.
120
Linear
Power
Dust and Dirt (kg)
100
DD Limit = 100
Exponential
MichaelisMenton
80
60
40
20
0
0
5
Time (days)
10
Figure 3.8: Buildup relationship
15
60
Build up parameters were determined using the technique described by Alley
and Smith (1981) and Baffaut and Delleur (1990). Dry deposition (dust and dirt)
were collected and their particle size determined. The dry loadings were than plotted
against duration (day) and the relationships were fitted using equations 3.14 to 3.17
(Figure 3.8). Dry deposition in the commercial catchment was collected from
impervious surfaces (test strip) measuring 5 m x 2 m. The plot is located near a car
park (Figure 3.9). Dry deposition in the residential catchment was collected at the
back lane of the houses. The site of test strip is also 5 m x 2 m. Dust and dirt
deposited on the test strip were carefully extracted using stiff broom and followed by
a vacuum cleaner.
According to Aubourg (1994), the bulk of the pollutants are attached to small
particles < 150 µm. The portion of particles < 150 µm was determined by carrying
out sieve analysis (Appendix G).
The particle size analysis involved determining the percentage by mass of
particles within the different size ranges. The samples were passed through a series
of standard test sieves. In this study, the size used were 4.75 mm, 1.18 mm, 600 µm,
300µm, 150µm 75µm. The mass of soil retained in each sieve was determined and
the cumulative percentage by mass passing each sieve calculated. From these
results, the buildup limit and buildup exponent can be estimated for both catchments.
61
SITE PLAN
COMMERCIAL CATCHMENT
TEST STRIP
600
DRAIN
2500
2500
2000
All dimensions in
mm
CAR PARK
40000
Traffic Direction
JALAN KEBUDAYAAN
Not to scale
Figure 3.9: Test strip for the commercial catchment
62
SITE PLAN
RESIDENTIAL CATCHMENT
HOUSES
TEST STRIP
600
DRAIN
2500
2500
All dimensions in
mm
2000
JALAN PERTANIAN
4000
Not to scale
Figure 3.10: Test strip for the residential catchment
For the purposes of modeling, the main parameters which need to be adjusted
in the buildup relationship during calibration are:
i.
Buildup limit
ii.
Buildup coefficient
iii.
Buildup exponential
A minimum value of the buildup limit is the largest wash-off load that has
been measured. The buildup coefficient is a function of the pollutant deposition rate.
This rate depends on several factors, and no real upper limits and lower limit exist.
However, 90% of the buildup limit is usually deposited in a period that varies
between a few days and 15 days (Baffaut et al., 1990).
3.17.2 Estimation of Wash-off Parameters
The wash-off parameters have an impact on the wash-off rate and therefore
on the shape of the pollutographs. The wash-off coefficient varies typically between
one and 10 but can be much larger, whereas the wash-off exponent is between one
and five (Baffaut et al., 1990).
Generalized data for buildup and washoff are scarse (Manning et al., 1977)
and such measurements are almost never been conducted as part of a routine
monitoring program. For washoff, the relationship of washoff (mass/time) versus
runoff rate must be assumed, usually in the form of a power equation (Novotny,
1995). When end-of-pipe concentration and load data are all that are available, all
buildup and washoff coefficients end up being calibration parameters.
During the calibration process, pollutant washoff parameters that need to be
adjusted are washoff coefficient and washoff exponent. Pollutant washoff was
simulated using an exponential washoff equation as follows:
64
P(t) = KwRnP
(3.18)
where
P(t) is the washoff load rate at time t;
Kw is the washoff coefficient;
R is the runoff rate (mm/hr)
n is the power of runoff rate
P is the amount of pollutant remaining on the catchment
From the basic equation Equation 3.18, the washoff parameters, washoff coefficient
and exponent are determined from a finite difference approximation (Nix, 1994)
which produces:
Po (t + ∆t) = Po(t)exp{-Kw ∆t * 0.5[R(t)n + R(t + ∆t)n]}
(3.19)
where:
- Po (t + ∆t) is the amount of pollutant washoff during simulation time step (t + ∆t)
- Po (t) is the amount of pollutant on land surface during a time step (t)
- Kw is the washoff or decay coefficient
- ∆t is the time step
- 0.5[R(t)n + R(t + ∆t)n] is average runoff rate over a time step and
- n is the power function of runoff rate
The washoff coefficient can be calculated using Equation 3.19 provided the
following assumptions are met:
i.
the amount of pollutant available on the catchment at the start of
simulation is equal to the total pollutant load of the storm event, and,
ii.
the amount of pollutant at the start of every time steps is equal to the
amount of pollutant load of a particular time step.
The value of KW was determined using trial and errors method and assumed
variable value of n. The washoff coefficient and exponential can be determined by
plotting the cumulative values of KW and runoff rate (R) (Figure 3.11).
Cumulative Washoff (1/mm)
65
Kw = R n
Cumulative Runoff Rate (mm/hr)
Figure 3.11: Cumulative values of KW and runoff rate (R)
CHAPTER 4
RUNOFF QUALITY AND POLLUTANT LOADINGS
This chapter discusses the results of the analysis in terms of stormwater
runoff quantity and quality. The results of stormwater quality are discussed and
compared with other studies. The scope covered in this chapter include Event Mean
Concentration (EMC), statistical analysis, box plot analysis, hysteresis loops, first
flush analysis and pollutant loading.
4.1
Characteristics of Rainfall Events
Characteristics of rainfall and runoff data collected from November 2003 to
December 2004 are summarised in Table 4.1. The hydrologic and pollutant
concentration data were used to plot pollutograph and to calculate event mean
concentration and pollutant loading. Ten storm events for residential and seven
events for commercial catchments were sampled and analysed. During the
monitoring periods, observed antecedent dry days ranged from 21 to 120 hours and
the rainfall size ranged from 1.52 to 65 mm. The storm depth and intensity are
important for the transport of particulates whereas the length of dry periods indicate
the amount of accumulated pollutant since the preceding event (Lee et al., 2004).
67
Site
Residential
catchment
Commercial
catchment
Table 4.1: Characteristics of monitored storms
Event
Rainfall Intensity Days since last
measured
(mm)
(mm/hr)
storm (hr)
* 8-Nov-03
* 11-Nov-03
11-Jan-04
1.52
3.5
120
2-Mar-04
1.52
2.5
23
4-Mar-04
18.8
16
45
6-Mar-04
2.8
9.3
46.4
12-July-04
19
16
24
8-Sept-04
6.8
8.8
50
4-Nov-04
65
27
21
27-Dec-04
8
9.6
8
11-Mar-04
20.1
24
21
16-Mar-04
31.3
25
15
18-Mar-04
3
47.7
10
19-Mar-04
7
23.8
11
14-Apr-04
11
48
** 20-Aug-04
7
10-Sept-04
4.9
49
Note: * rain gauge was not installed yet
**no rainfall data recorded due to technical problem
4.2
Event Mean Concentration (EMC)
The event mean concentration (EMC) of contaminants can be represented by
the total constituent mass discharged, during an event, divided by the total runoff
volume of the event (Adam and Papa, 2000). Table 4.2 presents the EMC values for
various parameters together with their site mean concentrations (SMC) or the
average EMC. Pollutant concentrations vary considerably from storm to storm and
between the two catchments.
By comparison with the existing Interim National Water Quality Standards
for Malaysia (INWQS) (Table 4.3), the runoff is highly polluted especially in terms
of SS, BOD and COD with maximum EMCs for the residential catchment of 1024
mg/l, 190 mg/l and 728 mg/l, respectively. The corresponding values for the
commercial catchment were 341 mg/l, 237 mg/l and 781 mg/l. Storm size didn’t
seem to be a single factor influencing the EMC values. For example the highest SS
was registered on 11 January 2004 with total rainfall of only 1.52 mm. Similarly, the
highest COD value was recorded on 6 March 2004 following 2.8 mm rainfall.
68
The concentration of SS was considerably higher in the residential compared
to commercial catchment. This is due to the fact that most of the area is covered and
therefore surface erosion is not happening in the commercial catchment. In contrast,
the residential catchment has more open area (e.g. playing ground) which is subject
to erosion when the soil landscape is not properly maintained. Concentrations of
NH3-N and P were also high in both catchments and could be associated with
domestic wastes, sullage and discharge from restaurants. In general, the storm
quality can be considered very polluted with major contaminants exceeded the limits
for class V water.
Table 4.2: Event means concentration (EMC) of water quality parameters.
Events
BOD5
(mg/l)
COD
(mg/l)
8-Nov-03
11-Nov-03
11-Jan-04
2-Mar-04
4-Mar-04
6-Mar-04
12-July-04
8-Sept-04
4-Nov-04
27-Dec-04
SMC
73
54
85
190
123
156
68
39
47
113
95
238
136
296
675
316
728
181
98
118
320
311
11-Mar-04
16-Mar-04
18-Mar-04
19-Mar-04
14-Apr-04
20-Aug-04
10 Sept-04
144
59
237
180
121
97
108
135
495
158
585
781
422
251
714
487
SMC
SS
NO3-N NO2-N
(mg/l) (mg/l) (mg/l)
Residential Catchment
274
0.69
0.02
259
0.07
0.03
1024
0.91
0.13
742
3.9
0.06
259
3.3
0.07
374
1.7
0.82
85
6
152
0.17
21
3.7
405
3.1
364
2.4
0.1
Commercial Catchment
124
1
0.03
69
0.71
0.02
166
1.9
0.06
274
2.8
0.05
162
12.3
1.1
232
0.23
0.12
341
0.52
0.21
195
2.8
0.43
NH3-N
(mg/l)
P
(mg/l)
Pb
(mg/l)
0.66
0.73
8.7
7
9.12
3.6
0.37
0.37
0.8
3.7
3.5
1.11
0.6
7.3
7.8
1.24
3.44
0.93
4.52
0.01
1.6
3
0.05
0.07
0.03
0.001
0.002
0.0004
0.02
1.9
2.83
6.59
8.7
5.7
0.79
0.23
3.8
1.1
0.3
2.59
3.0
0.9
0.55
1.97
1.5
0.0001
0.002
.001
0.001
0.0001
0.001
Table 4.3: Water Quality Index Class Standard
Parameter
Unit
NH3-N
BOD
COD
DO
pH
SS
mg/l
mg/l
mg/l
mg/l
mg/l
After, DOE (1986)
I
< 0.1
<1
< 10
>7
>7
< 25
II
0.1 – 0.3
1-3
10 - 25
5–7
6–7
25 – 50
Class
III
0.3 – 0.9
3-6
25 - 50
3-5
5–6
50 - 150
IV
0.9 – 2.7
6 - 12
50 – 100
1-3
<5
150 - 300
V
> 2.7
> 12
> 100
<1
>5
> 300
69
As shown in Table 4.4, the EMCs of this study are compared with those of
Chongju, South Korea (Choe et al., 2002), Isfahan, Iran (Taebi and Droste, 2004),
Spain (Diaz-Fierros et al., 2002), Centennial Park Catchment, Australia (Abustan,
1995), Sembulan River, Sabah, Malaysia (Faridah, 2004) and Zulkifli et al. (2002) of
Sg. Pandan, Johor, Malaysia. For both catchments, concentration of COD is within
the ranges of the reported values. For BOD5, the residential catchment showed the
second lowest but the commercial catchment showed the highest value. In the
residential catchment, the values of SS are found to be relatively higher than the
other studies but in the commercial catchment the value is still within the range.
Table 4.4: Comparison between the results obtained from this study with the
other studies
Site
Landuse
Area
(ha)
Residential
• Multi – family
• Single –family
• Commercial
Average
4.8
3.2
0.7
2. Isfahan, Iran
• River
200
3. Spain
• Combined
Sewer System
4. Centennial Park
Catchment,
Australia
5. Sembulan river
1. Chongju,
Korea
6. Sg. Pandan
Catchment,
Johor
7. This study
Rainfall
(mm/yr)
1225
COD
EMC (mg/l)
BOD5 SS
Pb
NO3-N
211
226
501
313
76
125
169
123
146
414
276
279
0.4
0.2
0.1
0.24
-
118
649
-
-
0.31
-
-
1600
329
123
282
4
-
• Residential
(Separated
Sewer System)
1.27
km2
1581
-
-
129
-
-
• Mix landuse
(Residential,
Commercial)
43.68
km2
-
146.8
37
177
0.03
0.62
• Mix landuse
(Agricultural,
Residential,
Commercial)
17.14
km2
20002005
132
39
184
-
0.5
3.34
0.75
2000
311
487
95
135
364
195
0.02
0.001
2.4
2.8
• Residential
• Commercial
70
Descriptive statistics of the stormwater quality for both catchments are
summarised in Table 4.5. High standard deviations in both catchments indicate that
the concentrations tend to be widely scattered about the mean, which is an indicator
of variability from storm-to-storm and between samples of the same event.
Table 4.5: Statistical analysis of pollutant concentration for both catchments
4.3
Pollutant
(mg/l)
Mean
BOD5
COD
SS
NO3-N
NO2-N
NH3-N
P
pH
86
299
281
1.95
0.21
4.4
3.3
6.8
Residential
Median Standard
Deviation
56
72
158
325
97
386
1.1
2.5
0.02
0.58
1.6
6.8
2.7
3.4
6.9
0.4
Mean
126
489
229
3.1
0.3
5.4
2
6
Commercial
Median Standard
Deviation
105
94
346
437
117
269
0.9
8.1
0.05
0.9
2.7
6.6
1.2
2.4
6
0.4
Correlation Analysis
Correlations amongst various pollutants are presented in Tables 4.6a and
4.6b. For theresidential catchment, moderate positive correlations (p<0.01) between
SS with P, COD, BOD5 and NH3-N were observed. However, the correlations
between SS and NO3-N and NO2-N were not significant (p>0.05). These
relationships are not consistent between catchments. For commercial catchment, SS
concentration was positively correlated with NH3-N and P (p<0.01) but not with
other pollutants.
71
Table 4.6a: Correlation analysis of pollutant concentration in the residential
catchment
SS
SS
BOD
COD
NO3-N
NO2-N
NH3-N
P
1
0.542**
0.606**
0.081
-0.224
0.539**
0.729**
BOD
COD
1
0.897**
0.152
-0.155
0.383**
0.514**
1
0.049
-0.160
0.370**
0.445**
NO3-N
NO2-N
1
-0.087
-0.006
0.098
1
-0.088
-0.055
NH3-N
P
1
0.547**
1
** Correlation is significant at p◦≤ 0.01 level (2-tailed).
Table 4.6b: Correlation analysis of pollutant concentration in the commercial
catchment
SS
SS
BOD
COD
NO3-N
NO2-N
NH3-N
P
1
0.011
0.127
-0.100
-0.110
0.279*
0.335**
BOD
COD
1
0.596**
0.117
0.059
0.195
0.399**
1
0.108
0.065
0.126
0.501**
NO3-N
1
0.963**
0.174
0.062
NO2-N
1
0.060
0.002
NH3-N
P
1
0.572**
1
** Correlation is significant at p ≤ 0.01 level (2-tailed).
* Correlation is significant at p ≤ 0.05 level (2-tailed).
4.4
Baseflow and Stormflow Concentrations
Comparisons between stormflow and baseflow concentrations in the
commercial catchment are shown in Figure 4.1. With the exceptions of NH3-N and
NO3-N, the other pollutants in the commercial catchment exhibit higher
concentrations in baseflow than in stormflow. The higher baseflow concentration
could be associated with the limited sources of the pollutants that were accumulated
in the drains during dry days and became diluted during storms. This is in contrast to
when a large portion of a catchment is cleared and the soil is exposed to raindrops.
In such a condition the source of sediment is unlimited thus sediment concentration
is expected to be higher during storm events. However, when mean values for all
events in this study were considered, stormflow concentrations were higher than the
baseflow. Various pollutants might be transported from the impervious catchment
72
surfaces by overland flow during storms. SS showed considerable variation between
events with obvious dilution during storms. SS could be evacuated from exposed
land surfaces in the catchment. The next highest concentrations are BOD5 and COD,
ranging from 22.5 to 180 mg/l and 35 to 1094 mg/l, respectively. This may suggest
large supplies of organic matter during storm events.
Figure 4.1a shows the box plot of storm water quality (see section 3.7 for
interpretation). The median values of stormflow and baseflow concentrations are
summarized in Table 4.7a. The percentage differences between median stormflow
and baseflow concentrations of BOD, COD and SS are -40%, 124% and 44%,
respectively. P produced the biggest difference of -275%.
73
BOD (mg/l) - Commercial
200
COD (mg/l) - Commercial
1200
SS (mg/l) - Commercial
1400
1200
1000
1000
800
800
100
600
600
400
400
200
0
N=
200
0
N=
66 4
SF BF
SF BF
NO3-N (mg/l) - Commercial
30
0
N=
71 4
NO2-N (mg/l) - Commercial
80 4
SF BF
NH3-N (mg/l) - Commercial
5
50
4
40
3
30
2
20
1
10
20
10
0
N=
79 4
0
N=
SF BF
0
N=
80 4
SF BF
SF BF
P (mg/l) - Commercial
14
78 4
pH - Commercial
7.5
Conductivity - Commercial
600
12
500
7.0
10
400
6.5
8
300
6
6.0
200
4
5.5
100
2
0
N=
80 4
SF BF
5.0
N=
80 4
SF BF
0
N=
80 4
SF BF
Figure 4.1: Box plot of stormwater quality in the commercial catchment
74
Table 4.7a: Median concentration of stormflow and baseflow quality in the
commercial catchment (mg/l)
Stormflow
Baseflow
%
difference
BOD5
COD
SS
100
140
317
710
117
169
-124%
-44%
-40%
NO3N
NO2- NH3N
N
Commercial
0.9
0.05
2.7
0.4
0.07
1.1
56%
-40% 59%
P
pH
EC
1.2
4.5
5.9
6.7
60
230
-275%
-14%
-283%
Concentrations of SS, BOD5, SS, NO3-N, NO2-N, NH3-N and P are plotted
against the antecedent dry days (Figure 4.2). For the commercial catchment, the
observed antecedent dry days ranged from 24 to 49 hours and the rainfall size ranged
from 3 to 31.3 mm. Interestingly, even with less than 1 dry day period, the event on
March 19, 2004 produced the highest SS concentration compared to the other events.
Storm size also didn’t seem to be a single factor influencing the SS concentration.
No correlation was found between the concentration of SS and the antecedent dry
day period. COD shows the highest value of median concentration with 49 hours
(2.04 days) antecedent dry day. But again there was no correlation between COD
and dry day period. The same case was found for the other parameters.
For the commercial catchment, there was no correlation between pollutant
concentration and the antecedent dry day period. This finding agrees with Gupta and
Saul (1996). Since some kitchens in the residential catchment, especially after
renovation, directly discharge sullage into drains, the pollutant transport behaviour is
less predictable.
75
SS - Commercial
BOD - Commercial
1400
COD - Commercial
200
1200
1200
1000
1000
600
Concentration (mg/l)
Concentration (mg/l)
Concentration (mg/l)
800
800
100
600
400
400
200
200
0
0
0.99
1.04
1
2
1.99
0
0.99
2.04
1.04
1
Dry Days
2
1.99
0.99
2.04
Dry Days
NO3-N - Commercial
2
1.99
2.04
Dry Days
NO2-N - Commercial
30
1.04
1
NH3-N - Commercial
5
30
4
Concentration (mg/l)
Concentration (mg/l)
20
Concentration (mg/l)
20
3
2
10
10
1
0
0
0.99
1.04
1
2
1.99
0
0.99
2.04
1.04
1
1.99
Dry Days
Dry Days
0.99
2
2.04
1.04
1
2
1.99
2.04
Dry Days
P - Commercial
12
10
Concentration (mg/l)
8
6
4
2
0
0.99
1.04
1
2
1.99
2.04
Dry Days
Figure 4.2: Effects of antecedent dry days on pollutant concentrations
76
For the residential catchment, with the exceptions of BOD5, COD, P, and
Electrical conductivity, the other pollutants show much higher concentration during
stormflow than in baseflow (Figure 4.3). Concentrations of BOD5, COD, SS, NO3N, NO2-N, NH3-N and P were diluted during storms. The range of pH in stormflow
was not so different from the baseflow, ranging from 5.7 to 7. At this range, the pH
values are highly buffered thus a large change in chemical compositions would cause
small changes in pH value (Kohlmann, 2003).
Median values of stormflow and baseflow pollutants for the residential
catchment are summarized in Table 4.7b. The percentage differences between
median stormflow and baseflow concentrations of BOD, COD and SS are -48%, 203% and -23%, respectively. COD produced the biggest percentage difference and
could be due to the sources of pollutants from the kitchens (sullage) during baseflow.
77
BOD (mg/l) - Residential
300
COD (mg/l) - Residential
1600
SS (mg/l) - Residential
2000
1400
1200
200
1000
800
1000
600
100
400
200
0
N=
0
N=
99 9
SF BF
SF BF
NO3-N (mg/l) - Residential
20
0
N=
110 8
108 10
SF BF
NH3-N (mg/l) - Residential
NO2-N (mg/l) - Residential
30
3.0
2.5
20
2.0
10
1.5
10
1.0
.5
0
N=
91 9
0.0
N=
SF BF
P (mg/l) - Residential
14
62 9
SF BF
0
N=
pH - Residential
8.0
74 9
SF BF
Conductivity - Residential
600
12
500
7.5
10
400
7.0
8
300
6
6.5
200
4
6.0
100
2
0
N=
73 7
SF BF
5.5
N=
91 10
SF BF
0
N=
91 10
SF BF
Figure 4.3: Box plot of baseflow and stormflow quality in the residential catchment
78
Table 4.7b: Median concentration of water quality parameters for residential
catchment
Stormflow
Baseflow
%
difference
BOD5
COD
SS
52
77
155
470
93
72
-48%
-203%
-23%
NO3- NO2- NH3N
N
N
Residential
1.1
0.02
1.5
0.8
0.02
1.4
27%
0%
7%
P
pH
EC
2.7
3.6
6.9
6.9
75
231
-33%
-0%
-208%
Concentrations of various pollutants of the residential catchment are plotted
against the antecedent dry days (Figure 4.4). The observed antecedent dry days
ranged from 21 to 120 hours and the rainfall size ranged from 1.52 to 19 mm.
Similar to the commercial catchment, no discernible trend was found between
pollutant concentration and the antecedent dry day period.
79
SS - Residential
BOD - Residential
1600
COD - Residential
300
1600
1400
1400
1200
1000
800
600
200
Concentration (mg/l)
Concentrations (mg/l)
Concentration (mg/l)
1200
1000
800
600
100
400
400
200
200
0
0
0.88
1
1.93
5
0
0.88
0.96 1.88 2.08
1
1.93
0.96 1.88
Dry Days
5
0.88
2.08
Dry Days
NO3-N - Residential
1.93
5
Dry Days
NO2-N - Residential
10
1
0.96 1.88 2.08
NH3-N - Residential
3.0
30
2.5
20
Concentration (mg/l)
Concentration (mg/l)
Concentration (mg/l)
2.0
1.5
10
1.0
.5
0
0
0.88
1
0.96
1.93
1.88
5
2.08
0.88
0.0
0.96 1.88 1.93
Dry Days
Dry days
5
0.96
1
1.93
1.88
5
2.08
Dry Days
P - Residential
30
Concentration (mg/l)
20
10
0
0.88
0.96
1
1.93
1.88
5
2.08
Dry Days
Figure 4.4: Effects of antecedent dry day on pollutant concentrations
80
4.5
Hydrograph and Pollutograph Analysis
The pollutographs and hydrographs of the residential catchment are presented
in Figure 4.2a. Most of the observed hydrographs were of single peaked types. The
hydrographs show rapid response to rainfall with short time to peak, averaging at 9
minutes for the residential and 11 minutes for the commercial catchment. This
flashy characteristic of small urban catchment was also observed in South Korea by
Lee et al. (1996). There are two important runoff parameters for assessing flood,
namely peak runoff rate and runoff volume. The peak runoff rate for a given
recurrence period can be used in designing storm-water runoff facilities such as
storm inlets and also in hydrologic planning. The maximum flow rate (0.25 m3/s) in
the commercial catchment was recorded on March 16, 2004’s event whereas in the
residential catchment was 0.9 m3/s on November 4, 2004. In a catchment where
storage is large, design work is usually based on total volume rather than the peak
discharge rates. Total water volume from a storm of a given recurrence period is
necessary in the design of flood control basins especially when a large portion of a
catchment is cleared such as for urbanization.
The solutions to some engineering problems require models that use a time
dependent variable such as discharge as a predictor variable (e.g., concentration –
flow relationship). This is usually demonstrated using pollutograph which describe
changes in solute/pollutant concentrations over time. Pollutographs can be important
in cases such as accidential spills of toxic chemical into streams (McCuen, 2004).
From this study, pollutant concentration show rapid increases and decreases
as the storm event progress. In most cases, the peak concentrations preceded the
peak runoff. This suggests that the pollutants were of short distant sources/origin so
as the pollutant mass arrived at the catchment’s outlet much faster than the runoff
itself. Such behaviour of SS transport is quite common for small catchments (Lee
and Bang, 2000).
The observed pollutographs exhibit two kinds of response; i) rapid increase
(flushing) and followed by rapid decrease, ii) dilution as presented in Table 4.8. For
81
the residential catchment, SS shows strong flushing with six out of seven events
exhibit first flush effect whereas NO2-N and P have dilution effects. The overall
responses are summarized in Table 4.9. Both catchments showed quite different
response with respect to pollutants that exhibit first flush or dilution effect. Dilution
effects were observed for NO2-N and P in the residential catchment and NH3-N in
the commercial catchment. First flush effects were observed for SS in the residential
catchment whereas COD and P in the commercial catchment. On the other hand,
BOD5 and NO3-N showed mix patterns in both catchments.
For commercial catchment, storm on the March 16, 2004 showed dilution for
all parameters. This is expected because March 16, 2004 registered the highest
rainfall depth and intensity. NH3-N showed dilution for all events. This can be
related to the first flush analysis (section 4.9), where NH3-N showed the weakest
flushing. From the results it can be concluded that COD and P showed strong first
flush and the other parameters had mix pattern.
The pollutographs for both catchments vary from one event to another and
this could be associated with the rainfall intensity and the antecedent dry days. In
addition, lowflow conditions in the residential and commercial catchments could
affect the concentrations on the early stage of storm events.
The first flush refers to the delivery of a disproportionately large load of
pollutants during the early portion of a storm runoff event. If most of the urbansurface pollutant load were transported during the initial phase of a storm, then a
small volume of runoff storage would be needed to treat and remove the bulk of
urban-surface pollutants. As such by installing first flush collection, the bulk of the
pollutant (such as SS) could be captured and isolated, with the subsequent runoff
which is cleaner being diverted directly to the stormwater system and finally into the
receiving water body.
82
Table 4.8: Characterisation of pollutographs for each event
Events
BOD5
(mg/l)
11-Jan-04
2-Mar-04
4-Mar-04
6-Mar-04
12-July-04
8-Sept-04
4-Nov-04
27-Dec-04
FF
FF
D
D
D
FF
D
FF
11-Mar-04
16-Mar-04
18-Mar-04
19-Mar-04
14-Apr-04
20-Aug-04
10 Sept-04
FF
D
D
FF
FF
D
FF
COD
SS
NO3-N NO2-N
(mg/l) (mg/l) (mg/l) (mg/l)
Residential Catchment
FF
FF
D
D
FF
D
FF
D
D
FF
D
D
D
FF
FF
D
D
FF
FF
FF
FF
FF
D
D
D
FF
FF
D
Commercial Catchment
FF
D
FF
D
D
D
D
D
D
D
D
D
FF
D
FF
FF
FF
FF
FF
FF
FF
FF
FF
FF
FF
FF
D
D
NH3-N
(mg/l)
P
(mg/l)
D
D
D
FF
FF
D
D
FF
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
FF
FF
FF
FF
FF
*D = Dilution
FF = First Flush
Table 4.9: Summary of pollutographs for both catchments
Sites
Residential
Commercial
Parameters
Dilution
NO2-N, P
NH3-N
First Flush
SS
COD, P
Mix Pattern BOD5, COD, NH3-N, NO3-N
SS, NO2-N, NO3-N,
BOD5
Figures 4.5a and 4.5b show selected pollutographs and hydrographs for both
catchments. The pollutographs for both catchments, produced by single-peaked
storms, generally exhibit rather simple patterns, characterized by rapid increase and
followed by gradual decreases. The pollutographs of other pollutants are presented
in Appendix E.
83
Residential - 2 March 2004
Residential - 6 March 2004
Flow
15
0.0
0.2
0.4
Storm Duration (hr)
400
20
0
0
0.0
0.6
0.5
1.0
Storm Duration (hr)
Residential - 8 September 2004
Residential - 6 March 2004
40
Flow
2
20
0
Flow (l/s)
BOD
100
75
0
0
0.0
1.0
Residential - 8 September 2004
SS
Flow
300
75
0
0
0.8
Flow (l/s)
150
2
4
C oncentration (m g/L)
4
C oncentration (m g/L)
0.8
0
Rainfall (m m )
2
0.4
Storm Duration (hr)
0.4
Storm Duration (hr)
Residential - 8 September 2004
0
0.0
150
Flow
0
0.5
Storm Duration (hr)
R ainfall (m m )
200
R ainfall (m m )
Concentration (mg/L)
NO3-N
4
F low (l/s )
1
4
2
C onc entration (m g/L)
0.5
Rainfall (mm)
0
0
0.0
Rainfall (m m)
40
Flow
0
0.0
SS
COD
600
150
Flow
300
75
0
F low (l/s )
0.1
1
800
Flow (l/s)
NO2-N
Flow (l/s)
C oncentration (m g/l)
0.3
0.5
Concentration (m g/L)
0.1
0
R ainfall (m m )
0
0
0.0
0.4
Storm Duration (hr)
0.8
Figure 4.5a: Pollutographs and hydrographs in the residential catchments
84
4
4
C oncentration (m g/L)
70
Flow
6
Flow (l/s)
35
0
Flow
35
150
0
0.0
0.6
Flow
70
0.05
35
0.00
0
0.3
Storm Duration (hr)
400
35
0
0.0
0.6
10
35
0
0
0.5
1.0
Storm Duration (hr)
F low (l/s )
C onc entration (m g/L)
70
Flow
0.7
1.4
1000
C onc entration (m g/L)
4
0
R ainfall (m m )
2
0.0
0.5
1.0
Storm Duration (hr)
Commercial - 10 September 2004
0
NO2-N
R ainfall (m m )
0
Commercial - 14 April 2004
20
70
Flow
F low (l/s )
NO2-N
4
800
SS
F low (l/s)
C oncentration (m g/L)
4
2
C onc entration (m g/L)
2
0
R ainfall (m m )
0
0.0
0.6
Commercial - 14 April 2004
Commercial - 19 March 2004
0.10
0.3
Storm Duration (hr)
SS
R ainfall (m m )
0.3
Storm Duration (hr)
70
BOD
0
0
0.0
300
40
Flow
500
20
0
F low (l/s )
Concentration (m g/L)
P
2
F low (l/s)
2
12
0
Rainfall (m m)
0
R ainfall (m m )
Commercial - 19 March 2004
Commercial - 19 March 2004
0
0.0
0.5
Storm Duration (hr)
1.0
Figure 4.5b: Pollutographs and hydrographs in the commercial catchments
85
4.6
Hysteresis Loop Analysis
Concentration-discharge (C/Q) relationships are investigated to better
understand the effects of urbanization on stream flow. This hysteresis loop patterns
can be used, often in conjunction with supporting evidence, to delineate source area
contributions to stream flow, to infer geochemical processes that affect storm water
quality and to discern mixing processes as they occur before, during and after storm
events.
A hysteresis loop demonstrates the occurrence of defined extremes rainfall,
the hydrograph recession conditions and could reflect the hydrogeological
characteristics of a catchment. Five classes of C/Q relationship have been diccussed
and summarised in Chapter 2 (section 2.6). Table 4.10 shows the result of C/Q
relationships for both catchments. For the residential catchment, 32 out of 42 C/Q
plots were interpreted as clockwise loops. Only eight plots were characterized by a
counterclockwise loop and five plots were figure eight loop. NH3-N – Discharge
loop showed the most consistent pattern with all events produced clockwise
hysteresis. For commercial catchment, most of the parameters exhibit both
clockwise and anticlockwise hysteresis. BOD5, P and COD are best characterized by
a clockwise loop. Only two parameters exhibited figure eight hysteresis which were
SS and NO3-N.
The clockwise loop occurs when the pollutants peak arrived at the
catchment’s outlet before the peak discharges (Figures 4.6a & 4.6b). It was likely
that the drain continue to receive large volumes of event water resulting in low solute
concentrations during the onset of the recession. Concentrations increase on the
rising limb of the hydrograph and than decrease on the falling limb. Rising limbs
dilutions are quite pronounced in both catchments. Counterclockwise loop occur
when the C peak arrives later than the Q peak (Figures 4.4b & 4.5b). It is
characterized by dilution at the beginning of the storm event, near static
concentrations during the relatively prolonged period of rising water levels and then
slightly increased concentrations during the long recession period (Rose, 2003). The
figure-eight hysteresis (Figures 4.4c & 4.5c) combines parts of clockwise and
86
counterclockwise loop. It consists of a clockwise loop at high flows and a
counterclockwise loop at low flows.
For both catchments, the C/Q plots were almost uniformly characterized by
hysteresis loops that approximated the ideal clockwise types. This finding agrees
with Evans and Davies (1998). Although more than one set of mixing dynamics can
generate such plots, the most likely explanation is that this hysteresis resulted from
two components mixing which was baseflow and stormflow (Rose, 2003). The
clockwise hysteresis suggests that much of the water mixture during the storm brief
recession period (particularly during the early stages of recession) was contributed by
storm event water.
Table 4.10: Summary of C/Q hysteresis loop characterization for both catchments
Residential (7 events)
C
CC
BOD5 (mg/l)
5
1
COD (mg/l)
4
1
SS
(mg/l)
6
1
NO3-N (mg/l)
4
2
NO2-N (mg/l)
1
2
NH3-N (mg/l)
7
P
(mg/l)
5
1
F8
1
2
1
1
-
Commercial (7 events)
C
CC
F8
6
1
6
1
4
2
1
5
1
1
5
1
5
2
6
1
-
C = Clockwise
CC = Counter-clockwise
F8 = Figure Eight
The representative clockwise, counter-clockwise and figure eight hysteresis
loops for both catchments are presented in Figures 4.5a, b and c & 4.6a, b and c.
Other results are presented in Appendix F.
87
Residential - 8 November 2003
Residential - 10 November 2003
400
BOD
Concentration (mg/l)
Concentration (mg/l)
150
100
50
SS
200
0
0
50
100
150
0
200
0
50
100
Discharge (l/s)
Discharge (l/s)
150
a) Clockwise loop
Residential - 2 March 2004
Residential - 10 November 2003
1600
SS
0.04
Concentration (mg/l)
Concentration (mg/l)
NO2-N
0.02
1200
800
400
0.00
0
0
50
100
Discharge (l/s)
150
0
5
10
15
Discharge (l/s)
20
25
b) Counter-clockwise loop
Residential - 11 January 2004
Residential - 12 March 2004
500
NO3-N
Concentration (mg/l)
Concentration (mg/l)
16
COD
8
0
0
0
5
10
Discharge (l/s)
15
0
100
200
Discharge (l/s)
c) Figure Eight
Figure 4.6: Hysteresis loops for the residential catchment
a) Clockwise loop
b) Counter-clockwise loop
300
88
c) Figure Eight
Commercial - 18 March 2004
Commercial - 11 March 2004
1600
Concentrations (mg/l)
Concentrations (mg/l)
0.2
P
0.1
COD
1200
800
400
0.0
0
0
50
100
150
0
20
40
60
Discharge (l/s)
Discharge (l/s)
a) Clockwise loop
Commercial - 20 August 2004
Commercial - 20 August 2004
5
Concentrations (mg/l)
Concentrations (mg/l)
NH3- N
0
P
5
0
40
80
0
40
Discharge (l/s)
Discharge (l/s)
b) Counterclockwise loop
Commercial - 14 April 2004
80
Concentrations (mg/l)
0
NO3-N
40
0
0
40
80
Discharge (l/s)
c) Figure Eight
Figure 4.7: Hysteresis loops for the commercial catchment
80
89
a) Clockwise loop
b) Counter-clockwise loop
c) Figure Eight
4.7
First Flush Phenomenon
For a given catchment, the pollutographs and hydrographs vary from one
storm event to another depending on the rainfall intensity, the antecedent dry weather
period, the condition of the sewer system, the dry deposits and the accumulation of
pollutants. To resolve this discrepancy, Bertrand-Krajewski (1980) proposed a
dimensionless representation of cumulative loading ratio curve against cumulative
runoff ratio or M(V) curve. The diagonal line represents the loading of a hypothetical
pollutant with constant concentration. The loadings for BOD, COD, SS, NO3-N,
NO2-N, NH3-N and P are plotted in Figures 4.7 & 4.8 for the commercial and the
residential catchments, respectively.
Data points above the diagonal line represent a higher loading and associated
with pollutants that exhibit a seasonal first flush (Lee et al., 2004). For the
commercial catchment, most data points within the initial 30% of the runoff volume
fall above the diagonal line. Again, this indicates that first flush is important. Table
4.11 shows averages cumulative pollutant loadings that fall in the 20-30% range of
runoff volume.
90
COD - Commercial
BOD - Commercial
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
Dimensio nless Cumulative Flo w rate
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flo w rate
NO3-N - Commercial
SS - Commercial
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.2
0.4
0.6
0.8
0.0
1.0
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flo w rate
Dimensio nless Cumulative Flo w rate
NO2-N - Commercial
NH3 -N - Commercial
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flo w rate
0.0
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flo w rate
P - Commercial
1.0
0.8
0.6
0.4
0.2
0.0
0.0
0.2
0.4
0.6
0.8
1.0
Dimensionless Cumulative Flo w rate
Figure 4.8: Mass Volume, M(V) ratios of BOD, COD, SS, NO3-N, NO2-N, NH3N and P in the commercial catchment
91
As shown in Table 4.11, the mean values ranged from 0.38 to 0.55, with a
maximum of 0. 87. Since the average cumulative pollutant load that fall in the 2030% range of runoff volume are considered, hence, there is no first flush exist if the
mean values was 0.25. It can be concluded that all contaminants exhibited strong
first flush.
Table 4.11: Cumulative load at 20-30% of the runoff volume in the commercial
catchment
Commercial
Minimum
Maximum
Mean
BOD
COD
SS
NO3-N
NO2-N
NH3-N
P
0.2
0.7
0.45
0.25
0.82
0.54
0.22
0.74
0.48
0.24
0.68
0.46
0.2
0.7
0.46
0.07
0.69
0.38
0.22
0.87
0.55
92
COD - Residential
BOD - Residential
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
Dimensio nless Cumulative Flo w rate
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flow rate
NO3-N - Residential
SS - Residential
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.2
0.4
0.6
0.8
0.0
1.0
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flo w rate
Dimensio nless Cumulative Flo w rate
NH3-N - Residential
NO2-N - Residential
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flo w rate
Dimensio nless Cumulative Flo w rate
P - Residential
1.0
0.8
0.6
0.4
0.2
0.0
0.0
0.2
0.4
0.6
0.8
1.0
Dimensio nless Cumulative Flo w rate
Figure 4.9: Mass Volume, M(V) ratios of BOD, COD, SS, NO3-N, NO2-N, NH3N and P in the residential catchment
93
For the residential catchment, the mean values ranged from 0.17 to 0.47 as shown
in Table 4.12. SS showed the strongest first flush with mean mass to volume ratio of 0.47
and NO2-N the weakest with a mean ratio of 0.17.
Table 4.12: Cumulative load at 20-30% of the runoff volume in the residential
catchment
Residential
Min
Max
Mean
BOD
COD
SS
NO3-N
NO2-N
NH3-N
P
0.2
0.59
0.4
0.15
0.69
0.42
0.15
0.78
0.47
0.14
0.49
0.3
0.14
0.19
0.17
0.23
0.53
0.38
0.23
0.43
0.33
The degree of flushing, however, was higher in the commercial catchment as
reflected by more points fall above the bisector line. The first 20 - 30% of the flow
evacuated between 20-70% BOD, 25-82% COD, 22-74% SS, 24-68% NO3-N, 2070% NO2-N, 7-69% NH3-N and 22-87% P. For the residential catchment, the first
20 - 30% of the discharges evacuated between 20-59% BOD, 15-69% COD, 15-78%
SS, 14-49% NO3-N, 14-19% NO2-N, 23-53% NH3-N and 23-43% P. For the
commercial catchment, the relative strength of the first flush was P> COD>SS>
NO3-N> NO2-N> BOD5> NH3-N but slightly different sequence in the residential
catchment: SS> COD> BOD5> NH3-N> P> NO3-N> NO2-N.
Study carried out by Lee et al., (2002) in Chongju, South Korea, found that
the first 20 – 30% of the flow evacuated between 5 – 60% COD, 5 – 75% SS, 5 –
65% TKN, 5 – 75% TP, 5 – 50% PO4-P and 5 – 65% Fe, which are lower compared
to this study. The relative strength of a first flush was TKN> PO4-P> TP> COD>
Fe> SS. Diaz – Fierros et al. (2002) in their study highlighted that frequent first
flush occur for TKN, TP, NH4+-N, COD and SS. Further, study carried out by Lee et
al. (2004) in California defined that the first 20 – 30% of the flow evacuated between
21 – 59% COD, 18 – 63% TSS, 15 – 56% Pb, 18 – 48% Ni, 15 – 44% Cu and 20 –
53% Zn.
94
It can be concluded that the first flush was found in this study. The degree of
first flush was determined by the storm characteristics and the antecedent dry day.
Large storms tend to cause dilution on pollutant concentration.
4.8
Pollutant Loadings
Pollutant loading obtained during a storm event represents the total local point
source (PS) and non point source (NPS) pollution. To compare pollutant loads from
the PS and NPS, the pollutant loadings were calculated as the product of EMC and
runoff volume, expressed in per unit area (kg/ha) (Table 4.13). PS load which was
determined from water sample during lowflow (non-rainy days and after mid-night)
was subtracted from the total load (during storm) to estimate the contribution of
NPS. Despite the apparently higher concentrations of SS and P in the residential
catchment, the loadings were higher in the commercial catchment than in the
residential catchment. This was attributed to a greater runoff volume per unit area in
the commercial catchment than in the residential catchment.
Table 4.13: Event pollutant loadings for residential and commercial catchments
Parameters
Loading (kg/ha)
Residential
BOD5
COD
SS
NO3-N
NO2-N
NH3-N
P
Pb
Range
0.5-16.5
1.2-27.5
2.2-22
0.001-1.5
0.0003-0.01
0.013-1.2
0.002-0.14
undetactable-0.001
Mean
4.9
9
7.5
0.35
0.004
0.2
0.05
0.001
Commercial
Range
5.9-25
28-38
7.4-26.3
0.03-2
0.002-0.4
0.13-0.93
0.05-0.19
undetectable -0.001
Mean
14
32
17
0.4
0.1
0.4
0.12
0.0003
95
Averages pollutant loading resulted from three storm events reported by
Zulkifli et al. (2002) at Sg. Pandan, Johor, were BOD5 (1.2 kg/ha), COD (7 kg/ha),
NO3-N (0.1 kg/ha) and NH3-N (0.1 kg/ha). These values were markedly smaller
compared to the present. Sg Pandan is a much bigger catchment (17.1 km2). Due to
high spatial variation of rainfall in the tropics, the observed storms may not fall on
the entire catchment. As such the storms tend to produce smaller loadings when the
values are expressed as per unit area basis.
CHAPTER 5
STORM WATER MANAGEMENT MODEL (SWMM)
This chapter discusses the application of Storm Water Management Model
(SWMM). The model calibration and validation processes for simulating water
quantity and quality are described in detail. Sensitivity analysis of the model was
carried out prior to the calibration process. The observed and simulated data are
compared using goodness of fit tests. The results are evaluated in terms of peak
flow, runoff depth and time to peak (shape of hydrograph). For water quality
modeling, SWMM was used to simulate suspended solids using local buildup and
washoff data collected at the study sites.
5.1
Hydrologic Parameters
Catchment characteristics required for the modeling of water quality using
SWMM were derived from topography and landuse maps. Total impervious areas in
the commercial and residential catchments were 90% and 85%, respectively. The
catchments slopes were 0.001 for the commercial and 0.01 for residential
catchments. The depression storage was determined from the relationship between
the observed runoff and rainfall depth as suggested by Zaman and Ball (1994). The
estimated values of the depression storages, obtained from the x-axis when runoff
equal to zero, were 0.6 mm for the commercial and 0.3 mm for the residential
catchments (Figures 5.1 and 5.2).
97
Runoff vs Rainfall - Commercial
Runoff (mm)
50
25
Depression storage
y = 1.4082x - 0.8546
R 2 = 0.9927
0.6 mm
0
-10
10
30
-25
Rainfall (mm)
Figure 5.1: Relationship between observed runoff and rainfall depth
for commercial catchment
Runoff vs Rainfall - Residential
50
y = 0.5858x - 0.2042
Runoff (mm)
R 2 = 0.9897
25
Depression storage
0.3 mm
0
-20
20
60
-25
Rainfall (mm)
Figure 5.2: Relationship between observed runoff and rainfall depth for
residential catchment
Manning’s roughness values proposed by Huber and Dickinson (1988) were
used for estimating runoff from pervious and impervious areas in both commercial
and residential catchments (see Table 5.1).
98
Table 5.1: The Manning’s roughness values
Ground Cover
Manning's n
Range
Smooth Asphalt
Asphalt or concrete paving
Packed clay
Light turf
Dense turf
Concrete or asphalt
Bare sand
Gravelled surface
Bare clay-loam (eroded)
Range (natural)
0.012
0.014
0.03
0.025
0.06
0.011
0.01
0.02
0.02
0.13
0.010-0.013
0.010-0.016
0.012-0.033
0.012-0.033
0.010-0.32
For the infiltration, Horton with maximum and minimum infiltration rate and
decay rate parameters was used (Equation 3.6). In this study, kinematic wave
equation was selected for the routing method (Equation 3.7). From this default
values as shown in Table 5.2, the sensitivity analysis was performed for both
catchments.
Table 5.2: Default parameter coefficient in SWMM
Parameters
Impervious Areas
Impervious Depression Storage
Pervious Depression Storage
Impervious Manning’s n
Pervious Manning’s n
Horton’s Max. Infiltration Rate
Horton’s Min. Infiltration Rate
Horton’s Decay Rate
Default values
Residential
Commercial
85
90
0.6
0.3
1.5
2.5
0.012
0.012
0.03
0.03
150
150
15
15
0.00115
0.00115
99
5.2
Sensitivity Analysis
Sensitivity analysis of the model was carried out prior to the calibration
process. Data from three storm events were used for the sensitivity analysis i.e. on
March 11, March 16, and April 14, 2004 for the commercial catchment and on
March 4, July 12, and November 4, 2004 for the residential catchment. The model
sensitivity was first examined by looking at the effects of varying the percentage of
impervious areas on the model performance. Then, the model sensitivity was tested
on the effect of changing the catchment width and the impervious depression storage.
5.2.1
Sensitivity to Runoff Depth
For the same values of catchment imperviousness, depression storage and
catchment width, the sensitivity analysis produced very similar runoff pattern for
both catchments. This suggests that it is possible to use similar input values for
calibrating SWMM to other catchments with similar basic characteristics. The
percentage of imperviousness was the most sensitive parameter as it shows positive
response to runoff depth (Figures 5.3 and 5.6). Even a small change in
imperviousness can greatly influenced the runoff volume. The slope depicts the
response of the model to a 1% change in the initial value of one parameter while
holding the other parameter values constant. The second most sensitive parameter is
impervious depression storage and followed by catchment width. The catchment
width showed a positive response (Figures 5.4 and 5.7) but has small affect on the
runoff depth. The percentage change in runoff depth was inversely related to the
change in impervious depression storage with changes in the output of less than 5%
for the commercial catchment (Figures 5.5 and 5.8). It suggests that depression
storage produced small changes in the runoff depth. The results of this sensitivity
analysis are very useful for selecting the most important parameters for modeling
hydrograph. By knowing this, the less important parameters can be ignored and the
modeling exercise becomes more cost effective.
100
Commercial
Percentage Change in Volume
80
40
imp16/03/04
imp11/03/04
0
-60
-40
-20
0
20
40
60
80
imp14/04/04
-40
-80
Percentage Change in imperv iousness
Figure 5.3: Sensitivity analysis of impervious area on runoff volume
Percentage Change in Volume
Commercial
10
w 16/03/04
w 11/03/04
w 14/04/04
0
-100
-50
0
50
100
-10
Percentage Change in w idth
Figure 5.4: Sensitivity analysis of catchment width on runoff volume
Commercial
Percentage Change in Volume
5
dp16/03/04
dp11/03/04
0
-100
-50
dp14/04/04
0
50
100
-5
Percentage Change in depression storage
Figure 5.5: Sensitivity analysis of impervious depression storage on runoff volume
101
Residential
Percentage Change in Volume
40
20
imp04/03/04
imp12/07/04
0
-40
imp4/11/04
-20
0
20
40
-20
-40
Percentage Change in imperv iousness
Figure 5.6: Sensitivity Analysis of percentage of impervious area on runoff volume
Percentage Change in Volume
Residential
0.5
w 04/03/04
w 12/07/04
w 4/11/04
0.0
-100
-50
0
50
100
-0.5
Percentage Change in w idth
Figure 5.7: Sensitivity analysis of runoff volume for catchment width
Percentage Change in Volume
Residential
20
dp12/07/04
dp04/03/04
10
dp04/11/04
0
-100
-50
0
50
100
-10
Percentage Change in depression storage
Figure 5.8: Sensitivity analysis of runoff volume for impervious depression storage
102
5.2.2
Sensitivity to Peak Flow
Similar to the runoff depth, the sensitivity analysis on the peak flow produced
very similar response for each input parameter in both catchments. As shown in
Figures 5.9 to 5.12, both the impervious percentage and catchment width showed
positive correlations with peak flow. Again it shows that the peak flows was
sensitive to the changes in the catchment imperviousness. For the catchment width,
percentages of change between -60% and +60% for the residential and commercial
catchments were tested. It suggests that the catchment width influenced the peak
flows during the calibration but with smaller effect.
Commercial
Percentage Change in Peak Flows
80
40
imp16/03/04
0
-60
-40
-20
0
20
40
60
80
imp11/03/04
imp14/04/04
-40
-80
Percentage Change in imperv iousness
Figure 5.9: Sensitivity analysis of percentage of impervious area on the
peak flows
103
Commercial
Percentage Change in Peak Flows
15
w 16/03/04
w 11/03/04
5
-100
-50
w 14/04/04
0
50
100
-5
Percentage Change in w idth
Figure 5.10: Sensitivity analysis of catchment width on the peak flows
Percentage Change in Peak Flows
Residential
60
imp04/03/04
20
imp12/07/04
imp4/11/04
-40
-10-20
20
50
-60
Percentage Change in imperv iousness
Figure 5.11: Sensitivity analysis of percentage of impervious area on the peak flows
Residential
Percentage Change in Peak Flows
0.6
w 4/11/04
0.2
w 12/07/04
w 04/3/04
-100
-50
-0.2
0
50
100
-0.6
Percentage Change in w idth
Figure 5.12: Sensitivity analysis of peak flows for catchment width
104
The parameter values in Table 5.3 were used for calibrating and validating on
SWMM model in both catchments. These values were adopted from results of the
sensitivity analysis.
Table 5.3: Parameters values used for calibrating SWMM in the commercial and
residential catchments
Parameters
Imperviousness of Subcatchment
Impervious Depression Storage
0.2
0.2
Pervious Depression Storage
2.5
2.5
Subcatchment Slope
0.01
0.001
Impervious Manning’s n
0.014
0.014
Pervious Manning’s n
0.03
0.03
Horton’s Max Infiltration Rate
175
150
Horton’s Min Infiltration Rate
15
15
0.00115
0.00115
Horton’s Decay Rate
5.3
Default Values
Residential
Commercial
94
98
Model Calibration and Validation for Commercial Catchment
The simulations of SWMM were evaluated in terms of runoff depths, peak
flows and the shape of the hydrograph.
5.3.1
Calibration of Runoff Depth
This analysis used larger storm size in view of the fact that the ultimate use of
hydrological modeling is for designing hydraulic structures, for which calibration of
large storms is more appropriate. The model was calibrated based on two storm
events, observed on March 11and March 16, 2004. As for the validation exercise,
three events were used, i.e. on March 19, April 14 and September 10, 2004. The
105
calibration results suggest that runoff depth is very sensitive to changes in
imperviousness depression storage.
Since the model is very sensitive to changes in impervious depression
storage, this value was first adjusted and refined. In the first trial, an impervious area
of 90% was used for both catchments. The depression storage was set between 0.15
and 0.25 mm. After several simulations, the best values for depression storage and
impervious area were 0.2 mm and 94% (residential); 98% (commercial),
respectively.
The resulted RE and ARE during the calibration process were between 17.5% and 17.5% (see Table 5.4). The highest RMSE of 4.7 mm was obtained from
event on March 16, 2004. Abustan (1997) found lower RE and ARE values of -5.2
and 6.7%, respectively. However, the modeling can be considered adequate because
the average ARE for all events is less than 20% (Sivakumar and Codner, 1995).
Table 5.4: Observed and simulated runoff depths of the calibrated events in the
commercial catchment
5.3.2
Simulated
(mm)
RMSE
(mm)
Relative
Errors (%)
Abs.
Relative
Errors
(%)
18
Events
Observed
(mm)
16-Mar-04
25
29.7
4.7
-18
11-Mar-04
14
16.5
2.5
-17
17
Average
-
-
3.6
-17.5
17.5
Validation of Runoff Depth
During the validation process, runoff depth also resulted in quite high values
of RE (-8%) and ARE (8.9%). Storm event on March 19, 2004 gave the highest
RMSE of 0.17 mm. This despite the fact that the peakflow was satisfactorily fitted
during the validation process (Table 5.5). Validation of runoff depth by Abustan
(1995) in Australia resulted in RE of -1.1% and ARE of 5.2%. The errors were
106
slightly lower than the errors in this study. However, the errors are acceptable since
the average RE was less than ± 10% and the average ARE was less than 15%
(Baffaut and Delleur, 1990; Sriananthakumar 1992).
Table 5.5: Observed and simulated runoff depth of validated events in the
commercial catchment
Simulated
(mm)
RMSE
(mm)
Relative
Errors (%)
Abs.
Relative
Errors (%)
4.4
5
0.6
-14
14
10.15
9.98
0.17
1.7
1.7
3.5
3.9
0.4
-11
11
-
-
0.4
-8.2
8.9
Events
Observed
(mm)
19-Mar-04
14-Apr-04
10-Sep-04
Average
5.3.3
Calibration of the Peak Flows
A comparison between the observed and simulated events is shown in Table
5.6. The storm on March 11, 2004 gave a higher RMSE than event on March 16,
2004 with a difference between observed and simulated of 0.02 m3/s. Absolute
relative error for this event was 7.4%. Based on the small values of RE and ARE of
3.1% and 4.3%, respectively, it can be concluded that the calibration process for peak
flows was successful. Abustan (1997) based on study at Centennial Park catchment,
Australia, found RE value of -2% and ARE of 10%. Besides, this calibration was
acceptable since the average RE was ± 10 percent and the average ARE was less than
15% (Sri Ananthakumar, 1992). Comparisons of the observed and simulated
hydrograph shape are presented in Figures 5.13 and 5.14 for both events.
107
Table 5.6: Comparison between observed and simulated peak flows of calibrated
events in the commercial catchment
Simulated
(m3/s)
RMSE
(m3/s)
Relative
Errors (%)
Abs.
Relative
Errors (%)
0.23
0.24
0.01
-1.18
1.18
0.17
0.15
0.02
7.4
7.4
-
-
0.015
3.1
4.3
Events
Observed
(m3/s)
16-Mar-04
11-Mar-04
Average
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.13: Observed and simulated hydrographs in the commercial catchment for
storm on March 16, 2004
108
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.14: Observed and simulated hydrographs in the commercial catchment for
storm on March 11, 2004
5.3.4
Validation of the Peak Flows
The model validation for the commercial catchment was performed using
observed stormflow data on March 19, April 14 and September 10, 2004 (Table 5.7).
All events produced very low RMSE ranging from 0.001 to 0.01. In terms of
absolute relative error, the highest was 10.1% for event on March 19. This validation
was acceptable since the average RE was less than ± 10% and the average ARE was
less than 15% (Sri Ananthakumar, 1992). Again studied by Abustan (1997), found
values of RE and ARE of -6.4 and 14%, respectively. It can be concluded, that the
validation process for peak flows was successful. The validated hydrographs of
these three events are presented in Figures 5.15 to 5.17.
109
Table 5.7: Statistical fits of the validated peakflow in the commercial catchment
Simulated
(m3/s)
RMSE
(m3/s)
Relative Errors
(%)
Abs.
Relative
Errors
(%)
Events
Observed
(m3/s)
19-Mar-04
0.075
0.067
0.01
10.1
10.1
0.075
0.082
0.001
-9.3
9.3
0.043
0.048
0.001
-7
7
-
-
0.004
-2.1
8.8
14-Apr-04
10-Sep-04
Average
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.15: Observed and simulated hydrographs of the commercial catchment for
storm on April 14, 2004
110
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.16: Observed and simulated hydrographs of the commercial catchment for
storm on March 19, 2004
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.17: Observed and simulated hydrographs of the commercial catchment for
storm on September 10, 2004
111
Runoff depth exhibits marked deviations especially for the higher runoff
where SWWM tended to overestimate the observed values (Figure 5.18). Runoff
also produced a weaker correlation (r2 = 0.98). On the other hand, the simulated
peakflows were very similar with the observed values. As shown in Figure 5.19, the
data points fall close along the diagonal line with a coefficient of determination, r2 of
0.99. SWMM can predict the peakflow better than the runoff depth.
Commercial
30
15
0
0
15
30
Observed Runo ff Depth (mm)
Figure 5.18: Observed and simulated runoff depth in the commercial catchment
Commercial
0.3
0.15
0
0
0.15
0.3
Observed Peak Flo ws (m 3/s)
Figure 5.19: Observed and simulated peak flows in the commercial catchment
112
5.4
Calibration and Validation for Residential Catchment
5.4.1
Calibration of the Runoff Depth
For the residential catchment, the model was calibrated using three storm
events; on March 4, July 12 and November 4, 2004. As for the validation, two storm
events on March 6 and September 8, 2004 were used. Similar to the commercial
catchment, the simulated runoff obtained from the calibration and validation
processes resulted in higher values of RE (-12%) and ARE (12%) compared to the
peak flows (Table 5.8). However, this calibration is still acceptable as the average
ARE is less than 15%.
Table 5.8: Comparison between observed and simulated runoff depths of
calibrated events in the residential catchment
RMSE
(mm)
Relative
Errors (%)
16
2
-14
Abs.
Relative
Errors (%)
14
11
13
2
-18
18
4-Nov-04
56
59
3
-5
5
Average
-
-
-2.3
-12
12
Events
Observed
(mm)
Simulated
(mm)
4-Mar-04
14
12-July-04
5.4.2
Validation of the Runoff Depth
Table 5.9 shows comparison between the observed and simulated
hydrographs for the validation process. Runoff depth resulted in higher values of RE
and ARE of -23% and 23%, respectively. The highest RE (25%) occurred on
September 8, 2004.
113
Table 5.9: Comparison between observed and simulated runoff depths of
validated events in the residential catchment
5.4.3
Events
Observed
(mm)
Simulated
(mm)
5
RMSE
(mm)
1
Relative
Errors (%)
-25
Abs. Relative
Errors (%)
25
8-Sept-04
4
6-Mar-04
1
1.2
0.2
-20
20
Average
-
-
0.6
-23
23
Calibration of the Peak Flows
The highest error was observed from storm on November 4, 2004 with RMSE
and RE of 0.13m3/s and 14.4%, respectively. The degree of error seems to increase
with the magnitude of the peakflow with the smallest error was observed for the
smallest storm on July 12 and the largest on November 4, 2004 (Table 5.10). This
calibration is considered good with average ARE of 6.8%, which was less than 15%
the allowable limit.
Table 5.10: Observed and simulated peak flows of calibrated events in the
residential catchment
RMSE
(m3/s)
Relative
Errors (%)
Events
Observed
(m3/s)
Simulated
(m3/s)
4-Mar-04
0.49
0.47
0.02
3.3
Abs.
Relative
Errors
(%)
3.3
12-July-04
0.28
0.29
0.01
-2.8
2.8
4-Nov-04
0.93
0.8
0.13
14
14
Average
-
-
0.05
5
6.7
Comparisons of the observed and simulated hydrograph shape are presented
in Figures 5.20, 5.21 and 5.22. Except toward the end of the storm event, the rising
and falling limbs of the simulated hydrographs generally followed closely the
observed hydrographs.
114
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.20: Observed and simulated hydrographs of the residential catchment for
storm on March 4, 2004
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.21: Observed and simulated hydrographs of the residential catchment for
storm on July 12, 2004
115
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.22: Observed and simulated hydrographs of the residential catchment for
storm on November 4, 2004
5.4.4
Validation of the Peak Flows
The validated hydrographs on September 8, 2004 and March 6, 2004 showed
RE and ARE values slightly higher than the calibrated ones (Table 5.11). The
average RE and ARE were -15% and 15%, respectively. The simulated hydrograph
was overestimated during validation. The differences between the observed and
simulated peak flows are summarised in Table 5.11.
Table 5.11: Comparison between observed and simulated peak flows of validated
hydrographs in the residential catchment
Events
Observed
(m3/s)
Simulated
(m3/s)
0.2
RMSE
(m3/s)
0.01
Relative
Errors (%)
-5
Abs. Relative
Errors (%)
5
8-Sept-04
0.19
6-Mar-04
0.04
0.05
0.01
-25
25
Average
-
-
0.01
-15
15
116
The validated hydrographs are presented in Figures 5.23 and 5.24. The
validated hydrograph shows a smooth shape on the rising and falling limbs compared
to the observed one. The simulated peakflows were also higher than the observed
values.
 Observed
-⋅⋅-⋅⋅-⋅⋅- Simulated
Figure 5.23: Observed and simulated hydrographs of the residential catchment for
storm on September 8, 2004
117
Figure 5.24: Observed and simulated hydrographs of the residential catchment for
storm on March 6, 2004
The SWMM model gives good accuracies for predicting peak flows and
runoff because the observed and simulated values generally fall close to 1:1 line
(Figures 5.25 and 5.26). Regression coefficients, r of 0.998 and 0.992 were obtained.
Residential
1
0.5
0
0
0.5
1
Observed P eaks Flo ws (m 3/s)
Figure 5.25: Observed and simulated peak flows for the residential catchment
118
Residential
60
30
0
0
30
60
Observed Runo ff Depth (mm)
Figure 5.26: Observed and simulated runoff depth for the residential catchment
5.5
Water Quality
For purposes of modeling water quality, local buildup data and washoff
process are needed in order to calibrate the model. Buildup data was collected at
both cathments and the data were then used for calculating washoff parameters. The
calibration was carried out to obtain the closest agreement between measured and
predicted event pollutant loadings. Similar to the runoff modeling, the observed and
simulated data were compared using statistical fits analysis in order to achieved a
good calibration. The simulation results are evaluated in terms of total load and peak
load.
5.5.1
Estimation of Buildup Parameters
The buildup data for both sites were developed and shown in Figures 5.27
and 5.29. The methodology for this estimation has been discussed in Chapter 3.
Dust and dirt (DD) buildups of particulates ≤150 µm are also plotted (Figures 5.28
and 5.30)
119
Commercial
Buildup DD (g/m of curb)
100
y = 303.93x
0.57
2
R = 0.7
50
0
0
4
8
Time (day s)
Figure 5.27: Total dust and dirt (DD) buildup rate in the commercial catchment
Particulates < 150 µm (Commercial)
Buildup DD (g/m of curb)
30
y = 4.3241x
0.70
2
15
R = 0.45
0
0
2
4
6
8
Time (day s)
Figure 5.28: Dust and dirt (DD) buildup rate of particulates ≤ 150 µm
From the data collected (Table 5.12), it was found that commercial catchment
showed a wide range of DD accumulation. For four days of dry period, the variation
was from 44.7 g/m to 103.1 g/m. But for DD <150 µm the variations were smaller
from 9.5 g/m and 14.1 g/m. The largest DD was obtained for five days dry period.
120
Table 5.12: Collected dust and dirt in the commercial catchment
Date
Dry periods
(days)
4
4
6
5
1
4
1
2 Dec 2004
23 Dec 2004
30 Jan 2005
8 Jan 2005
13 Jan 2005
14 Jan 2005
19 Jan 2005
DD
(g/m)
103.1
52.2
66.1
89.6
30.4
44.7
37.9
DD < 150 µm
(g/m)
14.1
6.86
16.9
27.8
5.0
9.5
1.7
Buildup DD (g/m of curb)
Residential
30
y = 65.5x
0.74
2
R = 0.49
20
10
0
0
2
4
6
Time (day s)
Figure 5.29: Total dust and dirt buildup rate in the residential catchment
Particulates < 150 µm (Residential)
Buildup DD (g/m of curb)
0.6
0.3
y = 1.0908x
0.9
2
R = 0.46
0
0
3
6
Time (Day s)
Figure 5.30: Dust and dirt (DD) buildup rate for particulates < 150 µm
121
Similar to the commercial catchment, the residential catchment shows a wide
range of DD accumulation. For one day dry period, the variation was from 2.4 g/m
to 12.4 g/m. For DD <150 µm the variations were smaller, from 0.08 g/m and 0.1
g/m. The DD data for the residential catchment was much smaller compared to the
commercial catchment. The test strip in the commercial catchment adjoins a car park
which is quite heavily used. This may contribute additional source of DD. In
contrast, the residential catchment was not so busy compared to the commercial
catchment.
Table 5.13: Deposition of dust and dirt in the residential catchment
Date
2 Dec 2004
23 Dec 2004
8 Jan 2005
13 Jan 2005
14 Jan 2005
19 Jan 2005
Dry periods
(days)
4
4
5
1
4
1
DD
(g/m)
31.7
11.9
18.1
2.4
6.6
12.4
DD < 150 µm
(g/m)
0.36
0.51
0.55
0.1
0.3
0.08
Table 5.14 shows the buildup limit and buildup exponent for both
catchments. These values were used in the calibration and validation of SWMM.
Table 5.14: Parameters of buildup data in the commercial and residential catchments
Total Dust and Dirt
Fine DD (< 150 µm)
Commercial
Buildup
Buildup
Limit
Exponent
80 (g/m)
0.57
15 (g/m)
0.7
Residential
Buildup
Buildup
Limit
Exponent
17 (g/m)
0.7
0.5 (g/m)
0.9
122
5.5.2
Estimation of Washoff Parameters
Plotting of both cumulative values of washoff and runoff rate could indicate
the washoff processes (using Equation 3.19). Figure 5.31 shows the calibrated
values of washoff coefficients and washoff exponents used in this analysis.
Commercial - 11/3/2004
Cumulative washoff
0.3
0.2
y = 0.001x
0.1
0.71
2
R = 0.87
0
0
750
1500
2250
Cumulative Runoff Rate (mm/hr)
Commercial - 16/3/2004
Cumulative washoff
0.47
y = 0.4423x
0.01
2
R = 0.76
0.46
0.45
0
500
1000
1500
Cumulative Runoff Rate (mm/hr)
2000
123
Commercial - 19/3/2004
Cumulative washoff
3
y = 1.0272x
0.13
2
R = 0.82
1.5
0
500
1000
1500
2000
2500
Cumulative Runoff Rate (mm/hr)
Commercial - 14/4/2004
Cumulative washoff
4
y = 0.0075x
2
0.81
2
R = 0.93
0
0
1000
2000
Cumulative Runoff Rate (mm/hr)
Commercial - 20/8/2004
Cumulative washoff
2
y = 0.0653x
0.4
2
R = 0.96
0
0
1000
Cumulative Runoff Rate (mm/hr)
2000
124
Commercial - 10/9/2004
Cumulative washoff
4
y = 0.987x
2
0.14
2
R = 0.94
0
0
1500
3000
Cumulative Runoff Rate (mm/hr)
Figure 5.31: Cumulative washoff coefficient and cumulative runoff exponent for the
commercial catchment
Different values of n, ranging from 0.5 to 2, were tested for various events
during the calibration to get the best estimate of the washoff parameters. The
resulted washoff coefficients varied markedly from 0.001 to 1.0 (Table 5.15).
Table 5.15: Washoff parameters for the commercial catchment
Event
11-Mar-04
16-Mar-04
19-Mar-04
14-Apr-04
20-Aug-04
10-Sept-04
Washoff coefficient
0.001
0.4
1
0.01
0.1
0.9
Washoff exponent
0.7
0.01
0.1
0.8
0.4
0.1
n values
1.5
2
1
1
1
1
125
Cumulative washoff and runoff rates for residential catchment are plotted in
Figure 5.32.
Cumulative washoff
Residential - 04/03/04
y = 0.0479x
8
0.7
2
R = 0.82
4
0
0
500
1000
1500
Cumulative Runoff Rate (mm/hr)
Residential - 06/03/04
Cumulative washoff
2
y = 0.0479x
0.55
2
R = 0.87
1
0
0
300
600
Cumulative Runoff Rate (mm/hr)
Residential - 12/07/04
Cumulative washoff
y = 0.3826x
0.3
2
R = 0.79
3
0
0
1000
Cumulative Runoff Rate (mm/hr)
2000
126
Residential - 08/09/04
0.48
Cumulative washoff
y = 0.4654x
0.002
2
R = 0.54
0.46
0
1000
2000
3000
Cumulative Runoff Rate (mm/hr)
Residential - 04/11/04
Cumulative washoff
0.28
y = 0.1227x
0.11
2
R = 0.52
0.08
0
1000
Cumulative Runoff Rate (mm/hr)
Figure 5.32: Cumulative washoff coefficient and cumulative runoff exponent for
the residential catchment
The values of washoff coefficients for the residential catchment were much
smaller compared to commercial catchment (Table 5.16). The washoff coefficient
range from 0.01 to 0.5 and for n values range from 1 to 2.5. Each value was used for
calibrating and validating the water quality model in SWMM.
Table 5.16: Washoff parameters for residential catchment
Event
Washoff coefficient
Washoff exponent
n values
04-Mar-04
0.05
0.7
1
06-Mar-04
0.05
0.5
1
12-July-04
0.4
0.3
1
14-Apr-04
0.01
0.8
1
08-Sept-04
0.5
0.002
2.5
04-Nov-04
0.12
0.1
2.5
127
5.6
Calibration of water quality for commercial catchment
Using the buildup and washoff data obtained in the preceding sections, five
events were calibrated and the results are presented in Tables 5.17 to 5.21 and
Figures 5.33 to 5.37. The calibration was carried out to obtain the closest agreement
between the measured and the predicted event pollutant loadings and at the same
time attempt to match the measured and predicted loadographs as closely as possible
(Sivakumar and Codner, 1995).
Two parameters, relative load difference (RLD) and absolute load difference
(ALD) (Equations 3.12 and 3.13) were used. The calibration is considered adequate
when the average ALD for all events is less than 20%. If this criterion is not
achieved, the average RLD was considered. The acceptable RLD values should be
between -15% and +15% for all events and the difference between the number of
events with under predicted values should be balances with the number of events
with over predicted values (Baffaut and Delleur, 1990; Sivakumar and Codner,
1995). Further, Sivakumar and Boroumand-Nasab (1995) suggest that a calibration
can be considered good when the average value of ALD is less than 30% and
satisfactory when the average ALD is less than 60%.
On the above basis, the calibration of SS loading for event on March 16,
2004, with an average ALD of 59% is considered satisfactory (Table 5.17). Studied
by Abustan (1995) gave much smaller values of RLD (8%) and ALD (36%). The
simulated peak load of SS was very similar with the observed value (0.182 vs. 0.18
kg/s) as shown in Table 5.18. Figure 5.33 shows the observed and simulated
loadographs. The observed values slightly preceded the simulated values.
Generally, the simulated loading followed the shape of the observed values well.
128
Table 5.17: Calibration of SS loading for 16 March 2004’s storm
Sampling
time
3:01:30 PM
3:22:23 PM
3:36:35 PM
3:42:35 PM
3:46:25 PM
3:51:20 PM
3:53:37 PM
3:55:45 PM
3:57:30 PM
4:01:18 PM
4:27:07 PM
5:00:00 PM
Observed
(kg/s)
0.0004
0.001
0.002
0.005
0.008
0.01
0.014
0.018
0.01
0.007
0.001
0.0003
Simulated
(kg/s)
0.0003
0.0003
0.003
0.004
0.007
0.011
0.013
0.016
0.017
0.018
0.004
0.001
Mean
RLD
ALD
0.27
0.7
-0.15
0.32
0.14
-0.1
0.04
0.11
-0.24
-1.51
-3
-0.67
-0.34
0.27
0.7
0.15
0.32
0.14
0.1
0.04
0.11
0.24
1.51
3
0.67
0.59
Table 5.18: Peak load of SS (kg/s) for the event on March 16, 2004
Load (kg/s)
Observed
Simulated
RLD
0.018
0.0182
-0.01
Observed and Simulated Loadograph
March 16, 2004
0.02
Observed
Simulated
Load (kg/s)
0.015
0.01
0.005
0
2:52 PM 3:21 PM 3:50 PM 4:19 PM 4:48 PM 5:16 PM
Time
Figure 5.33: The observed and simulated loadographs of SS on 16 March, 2004
129
For event on March 11, 2004, the calibration results of SS loadings and the
differences between the observed and simulated were presented in Table 5.19. The
average RLD was -33% under predicted. However, the average ALD was not good
with value of 82% which is greater than the acceptable limit of 60%. This is
probably due to the small values of the observed data. For the peak loading of SS
(Table 5.20), the calculated RLD of 2% is considered good.
Table 5.19: Calibration of SS loading for March 11, 2004
Sampling
time
3:47:00 PM
3:58:00 PM
4:07:40 PM
4:17:56 PM
4:22:30 PM
4:33:20 PM
4:36:12 PM
Observed
(kg/s)
0.019
0.025
0.012
0.012
0.013
0.006
0.003
Simulated
(kg/s)
0.002
0.005
0.012
0.025
0.02
0.01
0.008
Mean
RLD
ALD
0.89
0.8
0.01
-1.1
-0.5
-0.73
-1.5
-0.33
0.89
0.8
0.01
1.1
0.5
0.73
1.5
0.82
Table 5.20: Peak load (kg/s) for the event on March 11, 2004
Load (kg/s)
Observed
Simulated
RLD
0.025
0.0254
-0.02
130
Similar with the March 16 calibration, the observed values preceded the
simulated values.
Observed and Simulated Loadograph
March 11, 2004
0.03
Observed
Load (kg/s)
Simulated
0.02
0.01
0
3:36 PM
3:50 PM
4:04 PM
4:19 PM
4:33 PM
4:48 PM
Time
Figure 5.34: The observed and simulated loadographs of SS on March 11,
2004
The third event was on March 19, 2004. Similar to the earlier event on
March 16, the calibration is considered satisfactory. From Table 5.21, the averages
of RLD and ALD were 38 % and 60 %, respectively. The calibration of SS peak
loading was good with RLD of 29 % (Table 5.22).
Table 5.21: Calibration of SS loading for storm on March 19, 2004
Sampling
time
4:48:45
4:53:26
4:56:18
4:56:50
5:09:55
5:13:41
Observed
(kg/s)
0.009
0.016
0.024
0.019
0.005
0.002
Simulated
(kg/s)
0.007
0.014
0.016
0.017
0.009
0.007
RLD
ALD
0.26
0.1
0.33
0.1
-0.94
-2.5
0.26
0.1
0.33
0.1
0.94
2.5
Mean
-0.38
0.6
131
Table 5.22: Peak load of SS (kg/s) in the commercial catchment for storm on March
19, 2004
Observed
Simulated
RLD
0.024
0.017
0.29
Load (kg/s)
It is obvious that the shape of the loadograph showed some differences. In
terms of peak flow, the observed value is higher than the simulated one.
Observed and Simulated Loadograph
March 19, 2004
0.03
Observed
Load (kg/s)
Simulated
0.02
0.01
0
4:40 PM
4:48 PM
4:55 PM
5:02 PM
5:09 PM
5:16 PM
Time
Figure 5.35: The observed and simulated loadographs of SS on March 19, 2004
The calibration result for 14 April 2004’s storm is presented in Table 5.23.
Even though the average RLD value was good but the ALD (66%) exceeded the
acceptable value of 60%. However, the model can simulate peak load of SS well
with a RLD of 16% (Table 5.24).
132
Table 5.23: Calibration of SS loading for storm on April 14, 2004
Sampling
time
3:06:17
3:09:00
3:14:31
3:20:21
3:31:00
3:31:20
3:33:19
3:43:03
3:49:00
3:57:00
Observed
(kg/s)
0.007
0.018
0.011
0.01
0.011
0.011
0.007
0.003
0.001
0.001
Simulated
(kg/s)
0.003
0.006
0.011
0.014
0.014
0.013
0.013
0.007
0.004
0.002
Mean
RLD
ALD
0.6
0.66
0.001
-0.27
-0.24
-0.17
-0.95
-1.23
-1.44
-1
-0.4
0.6
0.66
1.14
0.27
0.24
0.17
0.95
1.23
1.44
1
0.66
Table 5.24: Peak load of SS (kg/s) in the commercial catchment on April 14, 2004
Load (kg/s)
Observed
Simulated
RLD
0.018
0.015
0.16
Observed and Simulated Loadograph
April 14, 2004
0.02
Observed
Load (kg/s)
Simulated
0.01
0
2:52 PM
3:07 PM
3:21 PM
3:36 PM
3:50 PM
4:04 PM
Time
Figure 5.36: The observed and simulated loadographs of SS on April 14, 2004
133
The last event used for the calibration was on September 10, 2004. SWMM
model provides good agreement between the observed and simulated SS loading with
RLD and ALD of 0.15 and 0.48, respectively. Table 5.26 shows peak loading with
average RLD of 29%.
Table 5.25: Calibration of SS loading in the commercial catchment for storm on
September 10, 2004
Sampling
time
4:15:00
4:19:20
4:20:50
4:23:00
4:25:57
4:38:25
Observed
(kg/s)
0.002
0.012
0.017
0.014
0.016
0.005
Simulated
(kg/s)
0.001
0.009
0.009
0.01
0.011
0.01
Mean
RLD
ALD
0.5
0.43
0.48
0.27
0.31
-1
0.15
0.5
0.43
0.48
0.27
0.31
1
0.48
Table 5.26: Peak load of SS (kg/s) in the commercial catchment for storm on
September 10, 2004
Load (kg/s)
Observed
Simulated
RLD
0.017
0.012
0.29
Observed and Simulated Loadograph
0.02
September 10, 2004
Observed
Load (kg/s)
Simulated
0.01
0
4:12 PM
4:19 PM
4:26 PM
4:33 PM
4:40 PM
Time
Figure 5.37: The observed and simulated loadographs of SS on September 10, 2004
134
The performance of SWMM in predicting SS loading was further examined
by plotting the observed and simulated values on 1:1 line (Figure 5.38). It can be
concluded that calibration processes was not satisfactory for the commercial
catchment.
Observed and Simulated SS Load
Simulated Load (kg/s)
0.03
0.02
0.00
0.00
0.02
0.03
Observ ed Load (kg/s)
Figure 5.38: Relationship between observed and simulated loadings of SS in the
commercial catchment
5.7
Calibration of water quality for residential catchment
Similar with the commercial catchment, four events were calibrated and the
results are presented in Tables 5.26 to 5.32 and Figures 5.39 to 5.42. The first event
was on March 4, 2004. Calibration of SS load was considered satisfied with ALD
value of 50%.
135
Table 5.27: Calibration of SS loading in the residential catchment on March 4, 2004
Sampling
time
3:21:30 PM
3:22:50 PM
3:23:36 PM
3:24:18 PM
3:26:00 PM
3:27:41 PM
3:36:30 PM
3:40:52 PM
3:43:25 PM
4:51:00 PM
Observed
(kg/s)
0.035
0.084
0.105
0.087
0.071
0.071
0.041
0.025
0.01
0.0001
Simulated
(kg/s)
0.052
0.071
0.071
0.077
0.088
0.097
0.064
0.036
0.031
0.0001
Mean
RLD
ALD
-0.47
0.16
0.32
0.12
-0.23
-0.35
-0.55
-0.39
-2.10
0
-0.38
0.47
0.16
0.32
0.12
0.23
0.35
0.55
0.39
2.10
0
0.50
For peak loading of SS, the difference in RLD was less than 2 % and this is
considered good (Table 5.28). Figure 5.39 shows the simulated and observed
loadographs. The observed SS peaked faster and higher than the simulated values.
However, the simulated loadings exhibit quite similar pattern with the observed
values except for the last one hour of the loadographs.
Table 5.28: Peak load of SS (kg/s) on March 4, 2004
Load (kg/s)
Observed
Simulated
RLD
0.105
0.1034
0.02
136
Observed and Simulated Loadograph
March 4, 2004
Load (kg/s)
0.1
Observed
Simulated
0.05
0
3:07
PM
3:21
PM
3:36
PM
3:50
PM
4:04
PM
4:19
PM
4:33
PM
4:48
PM
5:02
PM
Time
Figure 5.39: The observed and simulated loadographs of SS in the residential
catchment on March 4, 2004
The calibration results for the event on July 12, 2004 were presented in Table
5.29. Both RLD and ALD were satisfied with averages of 8% and 54%,
respectively.
Table 5.29: Calibration of SS loading in the residential catchment for storm on July
12, 2004
Sampling
time
2:22:00 PM
2:29:00 PM
2:47:00 PM
2:48:50 PM
2:50:20 PM
2:51:00 PM
2:52:00 PM
2:52:50 PM
2:54:00 PM
4:50:00 PM
Observed
(kg/s)
0.0025
0.0167
0.0252
0.0130
0.0136
0.0153
0.0155
0.0140
0.0072
0.0001
Simulated
(kg/s)
0.0001
0.0001
0.016
0.017
0.018
0.018
0.018
0.019
0.019
0.00012
Mean
RLD
ALD
0.96
0.98
0.37
-0.3
-0.32
-0.2
-0.16
-0.33
-1.6
-0.2
-0.08
0.96
0.98
0.37
0.3
0.32
0.2
0.16
0.33
1.6
0.2
0.54
137
The peak loading as shown in Table 5.30 was also good with RLD values of
20%. The observed and simulated loadographs are shown in Figure 5.40. The
observed loadograph declined very rapidly compared to the simulated loadographs.
Table 5.30: Peak load of SS (kg/s) in the residential catchment on July 12, 2004
Observed
Simulated
RLD
0.025
0.02
0.2
Load (kg/s)
Observed and Simulated Loadograph
July 12, 2004
0.03
Load (kg/s)
Observed
Simulated
0.02
0.01
0
1:55
PM
2:24
PM
2:52
PM
3:21
PM
3:50
PM
4:19
PM
4:48
PM
5:16
PM
Time
Figure 5.40: The observed and simulated loadographs of SS in the residential
catchment on July 12, 2004
From Table 5.31, both the averages RLD and ALD were not satisfied as they
exceeded the maximum acceptable values. SWMM seems to under predict the SS
loading. Table 5.32 shows the peak loading with average ALD of 17%. Similar to
the March 4, 2004 simulation, the observed loadograph decrease rapidly after the
peak load.
138
Table 5.31: Calibration of SS loading from residential catchment on September 8,
2004
Sampling
time
2:49:31
2:52:17
2:55:27
2:57:20
2:58:15
2:58:58
3:01:05
3:03:03
3:05:38
3:10:08
3:14:10
3:29:27
Observed
(kg/s)
0.0001
0.0006
0.006
0.024
0.035
0.032
0.025
0.011
0.005
0.003
0.002
0.0008
Simulated
(kg/s)
0.0
0.0002
0.005
0.019
0.025
0.028
0.029
0.026
0.019
0.013
0.007
0.004
Mean
RLD
ALD
1.0
0.66
0.08
0.22
0.28
0.12
-0.15
-1.33
-2.36
-3.24
-2.5
-3.63
-0.98
1.0
0.66
0.08
0.22
0.28
0.12
0.15
1.33
2.36
3.24
2.5
3.63
1.2
Table 5.32: Peak load of SS (kg/s) from the residential catchment on September 8,
2004
Observed
Simulated
RLD
0.035
0.029
0.17
Load (kg/s)
Observed and Simulated Loadograph
September 8, 2004
0.04
Observed
Load (kg/s)
0.03
Simulated
0.02
0.01
0
2:45
PM
2:52
PM
3:00
PM
3:07
PM
3:14
PM
3:21
PM
3:28
PM
3:36
PM
Time
Figure 5.41: The observed and simulated loadographs of SS from the residential
catchment on 8 September 2004
139
The last event was on November 4, 2004. Here the calibration was not
satisfactory as the average ALD was so large (111%). However, the model performs
reasonably well in predicting the peak load with ALD of 23% (Table 5.33). The
value, however, was slightly under predicted (Figure 5.42).
Table 5.33: Calibration of SS loading from the residential catchment on November
4, 2004
Sampling
time
2:27:41 PM
2:29:02 PM
2:52:31 PM
2:59:54 PM
3:02:33 PM
3:03:47 PM
3:07:12 PM
3:08:30 PM
3:13:13 PM
3:19:02 PM
3:22:20 PM
4:50:50 PM
6:30:32 PM
Observed
(kg/s)
0.001
0.001
0.001
0.004
0.008
0.009
0.015
0.025
0.013
0.006
0.003
0.0003
0.00001
Simulated
(kg/s)
0.0
0.0
0.004
0.01
0.01
0.011
0.016
0.017
0.019
0.012
0.011
0.001
0.000005
Mean
RLD
ALD
1.0
1.0
-2.18
-1.57
-0.79
-0.15
-0.04
0.28
-0.48
-0.9
-2.53
-2.41
1.0
-0.6
1.0
1.0
2.18
1.57
0.79
0.15
0.04
0.28
0.48
0.9
2.53
2.41
1.0
1.1
Table 5.34: Peak load of SS (kg/s) from residential catchment on November 4, 2004
Load (kg/s)
Observed
Simulated
RLD
0.0248
0.019
0.23
140
Observed and Simulated Loadograph
November 4, 2004
0.03
Observed
Load (kg/s)
Simulated
0.02
0.01
0
2:09 PM
3:21 PM
4:33 PM
5:45 PM
6:57 PM
Time
Figure 5.42: The observed and simulated loadographs of SS from the residential
catchment on November 4, 2004
The performance of SWMM in predicting SS loading was further examined
by plotting the observed and simulated values on 1:1 line (Figure 5.43). It can be
concluded that SWMM can model the SS load reasonably well for the residential
catchment.
Observed and Simulated SS Loadings
Predicted Load (kg/s)
0.12
0.06
0.00
0.00
0.06
0.12
Observ ed Load (kg/s)
Figure 5.43: Relationship between observed and simulated loadings for the
residential catchment
CHAPTER 6
CONCLUSION AND RECOMMENDATIONS
6.1
Conclusion
This concluding chapter presents important findings on the behaviour of
stormwater quality in typical residential and urban catchments in the tropical region.
The findings were able to support the objectives of the study especially in
quantifying loadings of pollutant for both catchments and the potential use of XPSWMM to simulate NPS for local conditions. In addition, some analyses that have
been discussed including statistical analysis, box plot analysis, hysteresis loops and
first flush analysis also support the objectives of this study for a better understanding
of non point source pollution. This study has made significant contribution in
improving strategies to control and manage urban water pollution by providing new
information/data and better describe the pollutant transport process in the urban
areas. Recommendations to improve future studies are also highlighted.
The following conclusions have been drawn from this study.
i)
By comparison to the Interim National Water Quality Standards for Malaysia,
the stormwater quality from residential and commercial catchments was
severely polluted with major parameters generally exceeded the acceptable
limits for class V water;
ii)
Event mean concentrations for all parameters were found to vary greatly
between storms. For the residential and commercial catchments, the values
142
(mg/l) were BOD5 (72, 94), COD (325, 437), SS (386, 269), NO3-N (2.5,
8.1), NO2-N (0.58, 0.9), NH3-N (6.8, 6.6), P (3.4, 2.4), respectively.
iii)
Both, hydrographs and pollutographs in the studied catchments showed rapid
increases and rapid decreases or dilution in pollutant concentration as the
storm event progress. For the residential catchment, SS showed strong
flushing whereas NO2-N and P showed dilution effects. For the commercial
catchment, NH3-N was diluted in all events.
iv)
In general, the peak concentrations preceded the peak runoff. This resulted in
clockwise hysteresis loops, suggesting a short distant sources of pollutants
into the water course. For both catchments, only a few events exhibit a
counterclockwise loop which indicates dilution at the beginning of the storm
event and then slightly increased during the long recession period. Figure
eight loop was also observed which is associated with a clockwise and
followed by consisting of a clockwise and counterclockwise loop.
v)
First flush phenomena were detected in this study. For the commercial
catchment, the relative strength of the first flush was P> COD>SS> NO3-N>
NO2-N> BOD5> NH3-N but slightly different sequence in the residential
catchment: SS> COD> BOD5> NH3-N> P> NO3-N> NO2-N.
vi)
In terms of water quantity and quality modeling, the calibration and
validation processes for both catchments were quite successful. Storm Water
Management Model (SWMM) can satisfactorily predict storm runoff but is
less promissing to predict SS load at the present catchments. However, it is
possible to improve the model adequacy by developing more quantitative
SWMM parameter identification based on the catchment characteristics.
143
6.2
Recommendations
In view of the large temporal and spatial variations of EMC value, a more
intensive stormwater monitoring program is recommended. Preferably, the sampling
design must include various storm size and to be replicated for different land-use
types. Consideration on the antecedent conditions of the catchment is also crucial for
a better understanding of the pollutant transport mechanism.
An important issue to be addressed is the influence of length of dry period
and rainfall intensity on the water quality and pollution loading. Continuous water
quality monitoring programme with reliable rainfall data, though expensive, is
crucial in getting reliable data for estimating pollutant loading.
Data on pollutant buildup on catchment surfaces is extremely lacking in the
tropics. More work is necessary in this line to improve water quality modeling using
SWMM. In addition, data of washoff processes are also essential during calibration
of the pollutographs.
REFERENCES
Abustan, I. (1997). Modelling of phosphorous transport in urban stormwater runoff,
Ph.D dissertation, School of Civil and Environmental Engineering, The
University of New South Wales, Australia, Sydney, Australia.
Adams, B.J. and Papa, F. (2000). Urban Stormwater management planning with
analytical probabilistic models, John Wiley & Son, New York.
Akan, O. (1993). Urban stormwater hydrology: A guide to engineering calculations,
Technomic, Lancaster, Pa.
Alley, W.M. and Smith, P.E. (1981). Estimation of accumulation parameters urban
runoff quality modeling. Water Resourses Research,17(6): 1657-1664.
Ammon, D.C. (1979). Urban Stormwater Pollutant Buildup and Washoff
Relationship, Master of Engineering Thesis, Department of Environmental
Engineering Sciences, University of Florida, Florida.
Appel, P.L., Hudak, P.F. (2001). Automated sampling of stormwater runoff in an
urban watershed, North-Central Texas. Journal Environmental Science
Health, A36 (6), 897–907.
APHA, AWWA and WEF. (1995). Standard methods for the examination of water
and wastewater, Washington, DC: American Public Health Association, New
York, no 19.
Ball, J.E. and Luk, K.C. (1998). Modeling the spatial variability of rainfall over a
catchment. ASCE, Journal of Hydrologic Engineering, 3(2), 122–130.
145
Baffaut, C. and Delleur, J.W. (1990). Calibration of SWMM runoff quality model
with expert system, Journal of Water Resources Planning and Management,
116: 247-261.
Barbosa, A. E. and Hvitved-Jacobsen, T. (1999). Highway runoff and potential for
removal of heavy metals in an infiltration pond in Portugal. The science of
the total environment, 235: 151–159.
Barrett, M. E. (1999). “Complying with the Edwards aquifer rules: Technical
guidance on best management practices.”
(http://www.tnrcc.state.tx.us/admin/topdoc/rg/348.pdf)
Barrett, M. E., Irish, L. B., Malina, J. F., and Charbeneau, R. J. (1997).
Characterization of highway runoff in Austin, Texas, area. Journal of
Environmental Engineering, 124(2), 131–137.
Bedient, P.B., Harned, D.A. and Charaklis, W.G. (1980). Stormwater analysis and
prediction in Houston. Journal of Environmental Engineering., 104
(EE6):1087 -1100.
Bedient, P.B., Harned, D.A. and Characklis, W.G. (1978). Stormwater analysis and
prediction in Houston. Journal of Environmental Engineering. Div. ASCE.
104 (EE6), 1087-1100.
Bertrand, J.L., Chebbo, G., and Saget, A. (1998). Distribution of pollutant mass vs
volume in stormwater discharges and the first flush phenomenon. Water
Research, 32: 2341-2356.
Brezonik, P.L. and Stadelmann, T.H. (2002). Analysis and predictive models of
stromwater runoff volumes, loads, and pollutants concentrations from
watersheds in the Twin Cities metropolitan area, Minnesota, USA. Water
Research, 36, 1743 - 1757.
146
Brown, L.C and Barnwell, T.O. (1987). The Enhanced Stream Water Quality
Models QUAL2E and QUAL2E-UNCAS : Documentation and User Manual.
Report No. EPA/600/3-87/007. U.S Environmental Protection Agency,
Athens, Georgia , USA.
Buffleben, M.S., Zayeed, K., Kimbrough, D., Stenstrom, M.K. and Suffet, I.H.
(2002). Evaluation of urban non-point source runoff of hazardous metals
entering Santa Monica Bay, California. Water Science Technology, 45 (9),
263–268.
Charbeneau, R. J. and Barretti, M. (1998). Evaluation of methods for estimating
stormwater pollutants loads. Water Environmental Research, 70:1295-1302.
Chiew, F. H. S., Mudgway, L. B. and McMahon, T. A. (1997). Urban stormwater
pollution. Cooperative Research Centre for Catchment Hydrology, Industry
Rep., Melbourne, Australia.
Choe, J.S., Bang, K.W. and Lee, J.H. (2002). Characterization of Surface Runoff in
Urban Areas. Water Science and Technology, 45 (9). 249-254.
Choi, K.S and Ball, J.E. (2002). Parameter estimation for urban runoff modeling.
Urban Water, 4 :31–41.
City of Boise. (1998). Boise storm water management design manual,
(http://www.cityofboise.org/publicIworks/services/water/stormIwater/manual
.pdf).
Cordery, I. (1977). Quality characteristics of urban storm water in Sydney,
Australia. Water Resources Research, 13(1), 197–202.
Cristina, C. M. and Sansalone, J. J. (2003). First flush, power law and particle
separation diagrams for urban storm-water suspended particulates. Journal
of Environmental Engineering, 129(4), 298–307.
147
Davies, J.W. (1987). Laboratory studies related to the first foul flush in combined
sewers.Proceedings of the 4th International Conferences on Urban Storm
Drainage, Lausanne, Switzerland, 47-48.
Deletic, A. (1998). The first flush load of urban surface runoff. Water Research, 32:
2462-2470.
Deletic, A. and Maksumovic, C.T. (1998). Evaluation of water quality factors in
storm runoff from paved areas. Journal of Environmental Engineering,
124: 869-879.
Deng, Z-Q, de Lima J.L.M.P. and Singh V. P. (2005). Fractional Kinetic Model for
First Flush of Stormwater Pollutants. Journal of Environmental Engineering.
131 (2), 232–241.
de Neufville, R. (1990). Applied Systems Analysis: Engineering Planning and
Technology Management, McGraw-Hill, New York.
Diaz-Fierros T, F., Suarez, J., Puerta, J. and Diaz-Fierros, V, F. (2002). Contaminant
loads of CSOs at the wastewater treatment plant of a city in NW Spain.
Urban Water, 4:291-299.
Dolan, D.M., Yui, A.K. and Geist, R.D. (1981). Evaluation of river load estimation
methods for total phosphorus. Journal Great Lakes Research, 7:207-214.
Drapper, D., Tomlimson, R. and Williams, P. (2000). Pollutant concentrations in
road runoff: Southeast Queensland case study. Journal of Environmental
Engineering, 126(4), 313–320.
Driver, NE and Tasker GD. (1990). Techniques for estimation of storm runoff loads,
volumes and selected constituent concentrations in urban watersheds in the
United States. Water-Supply Paper 2363. Denver, CO: US Geological
Survey.
148
Driver, N.E. and Lynstrom, D.J. (1986). “Estimation of Urban storm runoff loads.”
Urban runoff quality –Impact and quality enhancement technology,
B.Urbonas and L.A.Roesner, ed., ASCE, New York.
Droste, RL. and Hartt, JP. (1975). Quality and variation of pollutant loads in urban
stormwater runoff. Canadian Journal of Civil Engineering, 2:418-429.
Ellis, J.B. (1991). Measures for the control and treatment of urban runoff quality.
Report DT/PL/FV/JB Etude des Moyens de Reduction de la Pollution
Pluviale. Agence de I’Eau Seine Normandie.
El-Shaarawi A.H., Kuntz, K.W. and Sylvestre, A. (1986). Estimation of loading by
numerical integration. In: Statistical Aspects of Water Quality Monitoring,
A.H. El-Shaarawi and R.E. Kwiatkowski (Eds)., Development in Water
Science 27, Elsevier, Amsterdam, pp. 469-478.
U. S. Environmental Protection Agency (1984). Report to Congress: Nonpoint
Source Pollution in the U.S., Office of Water Program Operations, Synectics
Group, Inc., U.S. EPA, Washington, DC.
Evans, C. and Davies, T.D. (1998). Causes of concentration/ discharge hysteresis
and its potential as a tool for analysis of episode hydrochemistry. Journal
Water Resources Research, 34: 129-137.
Evans, C., Davies, T.D. and Murdoch, P.S. (1999). Component of flow processes at
four streams in the Catskill Mountain New York, analysed using episodic
concentration/ discharge relationships. Hydrological Processes, 13: 563-575.
Evans, B.M. and Nizeyimana, E. (1998). GIS-based quantification of regional
nutrient loads. In: Proceedings of the 3rd International Conference on Diffuse
Pollution, Edinburgh, Scotland, IAWQ, pp. 48-55 Venue 1.
149
Faridah Ang Mei Ling, Mohd Harun Abdullah and Maketab Mohamed. (2004).
Urban river pollution in Sabah- A case study of Sembulan river. Proceedings
of 2nd Bangi World Conference on Environmental Management: Facing
Changing Conditions. Univesiti Kebangsaan Malaysia, Malaysia.
Farm, C. (2002). Evaluation of the accumulation of sediment and heavy metals in a
storm-water detention pond. Water Science Technology, 45(7), 105–112.
Fleming, G. (1975). Computer Simulation Techniques in Hydrology, Elserver, New
York.
Forster, J. (1996). Patterns of roof runoff contamination and their potential
implications on practice and regulation of treatment and local infiltration.
Water Science Technology, 33(6), 39–48.
Gaume, E., Villeneuve, J.-P. and Desbordes, M. (1998). Uncertainty assessment and
analysis of the calibrated parameter values of an urban storm water quality
model. Journal of Hydrology, 210, 38-50.
Geiger, W. (1987). Flushing effects in combined sewer systems. In prod. 4th Int.
Conf. on Urban Storm Drainage, Lausanne, Switzerland, pp 40-46.
Grisham, D.M. (1995). Designing for the first flush. Civil Engineering (N.Y.), 65
(11), 67-69.
Gupta, K. and Saul, A. J. (1996). Specific relationships for the first flush load in
combined sewer flows. Water Research, 30 (5): 1244-1252.
He, W., Wallinder, I. O. and Leygraf, C. (2001). A laboratory study of copper and
zinc runoff during first flush and steady-state conditions. Corrosion Science,
43, 127–146.
150
Helsel, D., Kim, J., Grizzard, T., Randall, C. and Hoehn, R. (1979). Land use
influences on metals in storm drainage. Journal of Water Pollution Control
Federation, 51(4): 709-717.
Herricks, E.E. (Ed.) (1995). Stormwater Runoff and Receiving Systems : Impact,
Monitoring, and Assessment, CRC Lewis Publishers, Boca Raton.
Hornberger, G.M., Scanlon, T.M. and Raffensperger, J.P. (2001). Modelling
transport of dissolved silica in a forested headwater catchment: the effect of
hydrological and chemical time scales on hysteresis in the concentrationdischarge relationship. Hydrological Processes. 15, 2029-2038.
Huber, W.C. (1995). EPA Storm Water Management Model – SWMM. In Singh
V.P (ed.) Computer Models of Watershed Hydrology. Water Resource
Publications, Highlands Ranch, CO, pp. 783-808.
Huber, W.C and Dickinson, R.E. (1988). Storm Water Management Model User’s
Manual, Version 4. EPA/600/3-88/00Ia (NTIS PB88-236641/AS), U.S.
Environmental Protection Agency, Athens, GA.
Huber, W.C. (1986). “Deterministic Modeling of Urban Runoff Quality,” in
Proceedings, Urban Runoff Pollution, H.C. Torno et al. (Eds.), Springer –
Verlag, Berlin, pp. 167-242.
Hvitved-Jacobsen, T., Keiding, K.and Yousef, Y.A. (1987). Urban runoff pollutant
removal in wet detention ponds. Proceedings of the 4th International
Conferences on urban Storm Drainage,Lausanne, Switzerland, 137-142.
James, W. and Boregowda, S. (1986). “Continuous Mass-Balance of Pollutant Build
Up Processes,” in Proceedings, Urban Runoff Pollution , H.C. Torno et al
(Eds.), Springer Verlag, Berlin, pp. 243-272.
James, L.D. and Lee, R. (1971). Economics of Water Resources Planning, McGrawHill, New York.
151
Jewel, T.K., Nunno, T.J and Andrian, D.D. (1978). “Methodology for Calibrating
Stormwater Models.” Journal of the Environmental Engineering Division,
ASCE, 104( EE5): 485-501.
Johansen, N.B. (1985). Discharge to receiving waters from sewer systems during
rain. PhD Thesis: Technical University of Denmark, Lyngby, Denmark, 287
pp.
Kohlmann, J. F. (2003). What is pH, and how is it measured?. Technical Handbook
for Industry, Hach Company.
Lane, S.N., Richards, K.S. (2001). The ‘validation’ of hydrodynamic models: some
critical perspectives, in: Anderson, M.G., Bates, P.D. (Eds.), Model
Validation: Perspectives in Hydrological Science. Wiley, Chichester, pp.
413–438.
Larsen, T., Broch, K. and Andersen, M. R. (1998). First flush effects in an urban
catchment area in Aalborg. Water Science Technology, 37(1), 251–257.
Lazaro, T.R. (1990). Urban Hydrology - A Multidisciplinary Perspective. Technomic
Publishing Co. Inc., Lancaster, Pa.
Lee, J. H., Bang, K. W., Ketchum, L.H., Choe, J.S. and Yu, M.J. (2002). First flush
analysis of urban storm runoff. The Science of the Total Environment,
293:163-175.
Lee, J. H., Bang, K. W., Ketchum, L. H., Choe, J. S. and Yu, M. J. (2001). First
flush analysis of urban storm runoff. The Science of the Total
Environment, 293: 163-175.
Lee, J. H. and Bang, K.W. (2000). Characterization of urban stormwater runoff. J.
Water Research, 34 (6): 1773-1780.
152
Lee, J. H. (1998). Estimation of storm water runoff unit loading in urban area,
Proceeding paper in Korean Society of Environmental Engineering, 615-621.
Lee, J. H., Bang, K.W. and Lee, J.K. (1996). A study of runoff characteristics of
pollutants in combined sewer overflow. Journal of Korean Society
Environmental Engineering, 18 (10), 1147-1160.
Lee, H., Lau, S.-L., Kayhanian, M. and Stenstrom, M.K. (2004). Seasonal first flush
phenomenon of urban stormwater discharges. Water Research, 38, 41534163.
Loehr, R.C. (1974). Characteristics and comparative magnitude of non point
sources. Journal of Water Pollution Control Federation, 46 (8), 1849-1872.
Maalel, K. and Huber, W.C. (1984). SWMM calibration using conditions and
multiple events simulation. Proc. 3rd Int. Conf. on Urban Storm Drainage,
Goteborg, Sweeden.
Maestri, B., Johnson, F., Burch, CW. and Dawson, BL. (1985). Management
practices for mitigation of highway stormwater runoff pollution; stormwater
runoff pollution; vol. III. Research Report FHWA/RD-85/003, Federal
Highways Authority, US Dept. of Transportation.
Maidment, D. R. (1993). Handbook of Hydrology, McGraw-Hill, New York.
Manning, M. J., Sullivan, R. H. and Kipp, T. M. (1977). Nationwide Evaluation of
Combined Sewer Overflows and Urban Stormwater Discharges – Vol. III:
Characteristics of Discharges. EPA-600/2-77-064c (NTIS PB-272107), U.S.
Environmental Protection Agency, Cincinnati, OH.
Marsalek , J., Brownlee, B., Mayer , T., Lawal , S. and Larkin, G.A. (1997). Heavy
Metals and PAHs in Stormwater Runoff from the Skyway Bridge,
Burlington, Ontario. Water Quality Research Journal of Canada. 32 (4),
pp.815-827.
153
Metcalf, L. and Eddy, H. (1916). American Sewerage Practice, Disposal of Sewage,
Vol. III, McGraw-Hill, New York.
Metcalf and Eddy, Inc. (1971). University of Florida, and Water Resources
Engineers, Inc., Storm Water Management Model, Volume I- Final Report.
EPA Report 11024DOC07/71 (NTIS PB-203289), U.S. Environmental
Protection Agency, Washington, D.C.
Moffa, P.E. (1990). Control and Treatment of Combined- Sewer Overflows, Van
Nostrand Reinhold, New York.
Moxon, R. (1998). The control of diffuse pollution in the marine environment. In:
Proceedings of the 3rd International Conference on Diffuse Pollution,
Edinburgh, Scotland, 31 August 1998, IAWQ, pp. 2-20 Venue 1.
Mulcahy, J.P. (1990). Phosphorus export in the Twin Cities metropolitan area.
Technical Report Metropolitan Council, St. Paul, MN.
Najarian, T. O., Gunawardana, V. K., Ram, R. V. and Lai, D. (1986). Development
Impacts on Water Quality: A Case Study, Journal Water Resources Planning
and Management, ASCE. 112 (1), 20-35.
Nix, S.J. (1994). Urban Stormwater Modeling and Simulation, Lewis Publishers,
Boca Raton.
Novotny, V. (Ed.) (1995). Non point Pollution and Urban Stormwater Management.
Lancaster: Technomic.
Novotny, V. and Olem, H. (1994). Water Quality Prevention, Identification and
Management of Diffuse Pollution. Van Nostrand Reinhold, New York.
Novotny, V. and Chesters, G. (1981). Handbook of Non point Pollution: Sources
and Management. Van Nostrand Reinhold Publishers, New York.
154
Pearson, L.G.,Thornton, R.C., Saul, A.J. and Howard, K. (1986). An introductory
analysis of the factors affecting the concentration of pollutants in the first
foul flush of a combined storm sewer system. Proceedings of the
International Symposium on Urban Stormwater Quality and effects upon
receiving waters, Wageningen, the Netherlands, 93-102.
Petry, J., Soulsby, C. , Malcolm, I.A., Youngson A.F. (2002). Hydrological controls
on nutrient concentrations and fluxes in agricultural catchments. The Science
of the Total Environment. 294, 95–110.
Peterson, FM. and Batley, GE. (1992). Road runoff and its impact on the aquatic
environment: a review. Investigation Report No CET/LH/IR076. Lucas
Heights, Australia: Centre for Analytical Chemistry, CSIRO.
Philippe, J.P. and Ranchet, J. (1987). Pollution of runoff water in urban areas.
Synthesis of data from ten catchment around Paris. LCPC Research Report
nr. 142, Paris.
Pieterse, N.M., Bleuten, W. and Jorgensen, S.E. (2003). Contribution of point
sources and diffuse sources to nitrogen and phosphorus loads in lowland river
tributaries. Journal of Hydrology. 271, 213–225.
Pitt, R.E (1979). Demonstration of non-Point Abatement through Improved Street
Cleaning Practices. EPA-600/2-79-161 (NTIS PB80-108988), U.S.
Environmental Protection Agency,Chincinati, OH.
Rabitz, H. (1989). System analysis at molecular scale. Science, 246: 221–226.
Roesner , L. A., Aldrich, J. A. and Dickinson, R. E. (1988). Storm Water
Management Model User’s Manual Version 4 : Addendum I, EXTRAN.
EPA-600/2/3-88/001b (NTIS PB88-236658/AS), U.S. Environmental
Protection Agency, Athens, GA.
155
Rose, S. (2003). Comparative solute-discharge hysteresis analysis for an urbanized
and a ‘control basin’ in the Georgia (USA) Piedmont. Journal of Hydrology,
284: 45-56.
Saget, A., Chebbo, G. and Bertrand - Krajewski, J.L. (1996). The first flush in sewer
systems. Water Science Technology, 33 (9), 101-108.
Saltelli, A. (2000). Fortune and future of sensitivity analysis, in: Saltelli, A., Chan,
K., Scott, E.M. (Eds.), Sensitivity Analysis. Wiley, Chichester, pp. 421–426.
Sansalone, J. J. and Buchberger, S. G. (1997). Partitioning and first flush of urban
roadway storm water. Journal of Environmental Engineering, 123 (2):134143.
Sansalone, J.J. and Christina, C.M. (2004). First Flush Concepts for Suspended and
Dissolved Solids in Small Impervious Watersheds. Journal of
Environmental Engineering, 130 (11): 1301–1314.
Sansalone, J. J., Koran, J. M., Smithson, J. A. and Buchberger, S. G. (1998).
Physical characteristics of urban roadway solids transported during rain
events. Journal of Environmental Engineering, 124:427-440.
Sartor, J.D., Boyd, G.B. and Agardy, F.J. (1974). Water Pollution Aspects of Street
Surface Contaminations. Journal of the Water Pollution Control Federation,
46(3): 458-465.
Sartor, J.D. and Boyd, G.B. (1972). Water Pollution Aspects of Street Surface
Contaminants. United States Environmental Protection Agency, Washington,
DC, EPA-R2-72-081.
Schueler, T. R. (1994). First flush of stormwater pollutants investigated in Texas.
Watershed Protection Technology, 1(2), 88–90.
156
Schueler, T. R. (1987). Controlling urban runoff: A practical manual for planning
and designing urban BMPs, MetropolitanWashington Council of
Governments.
Shaheen, DG. (1975). Contributions of urban roadway usage to water pollution.
Report No. EPA-600/2-75/004, Municipal Pollution Control Division, Office
of Research and Development, US Environmental Protection Agency,
Washington, DC, USA.
Shelley, P.E. and Gaboury, D.R. (1986). “Estimation of Pollution from Highway
Runoff: Initial Results,” in Proceedings, Urban Runoff Quality : Impact and
Quality Enhancement Technology, B.Urbonas and L.A.Roesner (Eds.),
ASCE, New York, pp. 459-473.
Smullen, J.T., Shallcross, A.L. and Cave, K.A. (1999). Updating the U.S.
Nationwide Urban Runoff Quality database. Water Science and Technology,
39 (12), 9-16.
Sriananthakumar, K. and Codner, G. P. (1995). Urban Storm Water Pollutant Load
Estimation Using Overland Flow Depth. Proceedings of the International
Symposium on Urban Stormwater Management, Melbourne, Australia. pp.
229-234.
Stahre, P. and Urbonas, B. (1990). Stormwater Detention. Prentice Hall, Inc.,
Englewood Cliffs, New Jersey: 338 p.
State of California. (2001). “Waste discharge requirements for municipal storm
water and urban runoff discharges within the county of Los Angeles and the
incorporated cities therein, except the city of Long Beach.” Order No. 01182, California Regional Water Quality Control Board, Los Angeles.
State of Idaho. (2001). Catalog of stormwater best management practices for Idaho
cities and counties,
(http://www.deq.state.id.us/water/stormwater_catalog/doc_bmp50.asp).
157
Stevenson, H. and Wyman, B. (1991). The facts on the file dictionary of
environmental science. Facts on File Limited, Oxford, UK.
Stenstrom, M. K., Silverman, G. S. and Buraztynsky, T. A. (1984). Oil and grease in
urban stormwaters. Journal of Environmental Engineering, 110(1), 58–72.
Stotz, G., and Krauth, K.L. (1984). Factors affecting first flushes in combined
sewers. In Proc.,111 International Conference On Urban Storm Drainage,
Sweden, June 4-8, pp. 869-878.
Stotz, G and Krauth, K.L. (1986). Depositions in combined sewers and their
flushing behaviour. In Urban Drainage Modeling, C. Maksimovic (Ed.),
London (UK), Pergamon Press, 443-453.
Supiah Shamsudin (2003). Non – point phosphorus loadings and reduction in
Layang Reservoir Systems. Universiti Teknologi Malaysia: Phd Thesis.
Taebi, A and Droste, RL. (2004). Pollution loads in urban runoff and sanitary
wastewater. The Science of the Total Environment, 327: 175-184.
Thornton, R.C. and Saul, A.J. (1987). Temporal variation of pollutants in two
combined sewer systems. In Proceedings of the 4th International Conferences
On Urban Storm Drainage, Lausanne, Switzerland, 51-52.
Toler, L.G. (1965). Relation between chemical quality and water discharge in spring
Creek, Southwestern Georgia. US Geological Survey Professional Paper
525-C, 209-213.
USEPA. (1983). Results of the nationwide urban runoff program , volume I – final
report. NTIS PB84-185552. Washington, DC: US Environmental Protection
Agency.
158
Vaze, J. and Chiew, F.H.S. (2002). Experimental study of pollutant accumulation on
an urban road surface. Journal of Environmental Engineering. 130 (4), 379389.
Walling, D.E. and Webb, B.W. (1986). Solutes in river systems, in Solute
Processes, S.T. Trudgill (Ed.), John Wiley, New York, pp. 251-327.
Wanielista, M.P. and Yousef, Y.A. (1993). Stormwater Management. New York,
USA John Wiley and Sons, Inc.
Warwick, J.J. and Tadepolli, P. (1991). Efficacity of SWMM application. Journal
of Water Resources Planning and Management, 117 (3), 352-366.
WEF and ASCE. (1998). Urban runoff quality management. WEF Manual of
Practice No.23 and ASCE Manual Report on Engineering Practice No.87.
Alexandria, VA: Water Environment Federation and American Society of
Civil Engineers.
Whipple, D.J., Grigg N.S., Grizzard T., Randall C.W., Shubinski R.P. and Tucker
L.S. (1983). Stormwater Management in Urbanizing Areas. Prentice-Hall,
Englewood Cliffs, NJ.
Williams, G.P. (1989). Sediment concentration versus water discharge during single
hydrologic events in rivers. Journal of Hydrology, 111: 89-106.
Wu, J. S., Allan, C. J., Saunders, W. L. and Evett, J. B. (1998). Characterization and
pollutant loading estimation for highway runoff. Journal of Environmental
Engineering, 124(7), 584-592.
Zaman, S. and Ball, J.E. (1994). Simulation of Small Events in an Urban
Catchment, Proceedings of Hydrology and Water Resources Symposium
Water Down Under ‘94, Adelaide, Australia, pp. 353- 358.
159
Zulkifli Yusop, Tan Lai Wai, Kamarul Azlan Mohd Nasir and Maketab Mohamed.
(2002). Stormwater quality and pollutant loadings from an urban catchment
in Johor Bahru. Proceedings of Water Environmental Planning Seminar “Technologies of Water Resource Management”, International Islamic
University, Kuala Lumpur.
APPENDIX A
PHOSPHORUS, P ANALYSIS (ICP – MS) CALIBRATION
Preparing P from H2NaO4P.2H2O
1000 ppm standard solution
2H = 2 x 1.00794
= 2.01588
1Na = 1 x 22.989770 = 22.989770
6O = 6 x 15.9994
= 95.9964
1P
= 1 x 30.973761 = 30.973761
4H
= 4 x 1.00794
= 4.03176
= 156.007571
Mol reagent = 156.007571
%P = 30.973761 / 156.007571
= 0.1985401%
1000 ppm P = 1000/0.1985401
= 5036.7659 ppm H2NaO4P.2H2O
5036.7659 ppm = 5036.7659 mg/l
1000 ml = 5.0367659 g/l
100 ml, a = 5.0367659 x 100
1000
a = 0.5036766 gram
; where a = reagent mass
161
i) For 1000 µg/l
M1V1 = M2V2
V1 x 1000 ppm = 100 mL x (1000 x 10-3) ppm
V1 = 0.1 mL
Notes: 0.1 mL from standard solution added with deionized water in 100 mL
container
ii) For 100 µg/l
M1V1 = M2V2
V1 x (1000 x 10-3) ppm = 100 mL x (100 x10-3) ppm
V1
= 10 mL
Notes: 10 mL from 1000 µg/l solution added with deionized water in 100 mL
container
ii) For 40 µg/l
M1V1 = M2V2
V1 x (100 x 10-3) ppm = 100 mL x (40 x10-3) ppm
V1
= 40 mL
Notes: 40 mL from 100 µg/l solution added with deionized water in 100 mL
container
iii) For 20 µg/l
M1V1 = M2V2
V1 x (40 x 10-3) ppm = 100 mL x (20 x 10-3) ppm
V1
= 50 mL
Notes: 50 mL from 40 µg/l solution added with deionized water in 100 mL
container
iv) For 10 µg/l
M1V1 = M2V2
V1 x (20 x 10-3) ppm = 100 mL x (10 x 10-3) ppm
V1 = 50 mL
Notes: 50 mL from 20 µg/l solution added with deionized water in 100 mL
container
APPENDIX B
SAMPLE ANALYSIS
Figure B-1: Equipment for BOD5 analysis
Figure B-2: Analysis of BOD5 at the laboratory
163
Figure B-3: Laboratory analysis of Electrical conductivity, turbidity, TDS and pH
APPENDIX C
RESIDENTIAL CATCHMENT
From kitchen
165
Figure C-1: Rubbish and sullage from kitchen (during dry period)
Figure C-2: Outlet for the residential catchment
Figure C-3: Sampling during storm event
APPENDIX D
COMMERCIAL CATCHMENT
Figure D-1: Discharge from restaurants and drains during non rainy day
APPENDIX E
HYDROGRAPHS AND POLLUTOGRAPHS
Residential - 11 January 2004
Residential - 11 January 2004
1800
Flow
7
250
0
900
7
0
0
1.2
0.0
0.2
200
0
0.1
0.2
14
BOD
14
7
0
0
1.5
Flow (l/s)
100
Concentration (mg/l)
NO3-N
Flow
Rainfall (m m)
0.1
Concentration (mg/l)
1.2
Rainfall (mm)
0
0.5
1
Storm Duration (hr)
0.6
Storm Duration (hr)
Residential - 11 January 2004
Residential - 11 January 2004
0
SS
Flow
1.2
7
0.0
0
0
0.5
1
Storm Duration (hr)
1.5
Flow (l/s)
0.4
0.8
Storm Duration (hr)
14
Flow
0
0.0
R ainfall (m m)
0.2
C oncentration (m g/L)
COD
Flow (l/s)
Concentration (mg/L)
0.2
14
0.1
Flow (l/s)
0.1
500
0
R ainfall (mm)
0
168
0.2
14
0
0
0.5
1
Storm Duration (hr)
Concentration (mg/l)
0
F low (l/s)
C oncentration (m g/l)
Flow
7
0.2
0.4
NH3-N
20
0.1
14
NO2-N
Flow
0.2
7
0.0
0
0
1.5
0.5
1
Storm Duration (hr)
Flow (l/s)
0.1
40
0
R ainfall (m m )
0
R ainfall (mm)
Residential - 11 January 2004
Residential - 11 January 2004
1.5
Residential - 11 January 2004
Rainfall (m m )
0
0.1
0.2
14
14
Flow
7
Flow (l/s)
Concentration (m g/l)
P
7
0
0
0
0.5
1
Storm Duration (hr)
1.5
11 January 2004
Residential - 2 March 2004
0.3
0.3
1400
Flow
15
700
0
0
0.0
0.2
0.4
Storm Duration (hr)
0.6
Concentration (m g/l)
COD
Flow (l/s)
Concentration (m g/l)
1400
0.1
SS
Flow
15
700
0
0
0.0
0.2
0.4
Storm Duration (hr)
0.6
Flow (l/s)
0.1
0
Rainfall (m m )
0
Rainfall (m m )
Residential - 2 March 2004
169
Residential - 2 March 2004
0.3
0.1
0.3
300
14
15
150
0
Flow (l/s)
Flow
Concentration (m g/l)
0
0.2
0.4
Storm Duration (hr)
Flow
15
7
0
0.6
0
0.0
0.2
0.4
Storm Duration (hr)
0.6
Residential - 2 March 2004
Residential - 2 March 2004
0
R ainfall (m m )
0
0.1
P
Flow
15
0
Flow (l/s)
NH3-N
10
0.3
Concentration (mg/l)
C oncentration (m g/l)
0.3
0.1
0
0.0
0.2
0.4
Storm Duration (hr)
Rainfall (m m)
0.0
NO3-N
Flow
15
10
0
0.6
Flow (l/s)
C oncentration (m g/l)
BOD
Flow (l/s)
0.1
0
R ainfall (m m )
0
R ainfall (m m )
Residential - 2 March 2004
0
0.0
0.2
0.4
Storm Duration (hr)
0.6
2 March 2004
0
6
9
1500
0
6
9
Rainfall (mm)
Residential - 4 March 2004
R ainfall (m m )
Residential - 4 March 2004
400
200
500
0
0
0
0.5
1
Storm Duration (hr)
1.5
COD
Flow
1000
400
200
500
0
0
0
0.5
1
Storm Duration (hr)
1.5
Flow (l/s)
Flow
Flow (l/s)
C oncentration (m g/l)
SS
1000
Concentration (mg/l)
1500
170
Residential - 4 March 2004
Residential - 4 March 2004
R ainfall (m m )
6
9
0
6
9
8
Flow
100
400
200
0
0
0.5
1
Storm Duration (hr)
400
Flow
4
200
0
1.5
0
0
Residential - 4 March 2004
0.5
1
Storm Duration (hr)
1.5
Residential - 4 March 2004
0
Rainfall (mm )
0
6
9
6
9
28
Rainfall (mm)
0
NO3-N
Flow (l/s)
200
Concentration (mg/l)
BOD
Flow (l/s)
C oncentration (m g/l)
Rainfall (mm)
0
NH3-N
0.2
200
0.0
0
0
0.5
1
Storm Duration (hr)
Flow
14
200
0
1.5
0
0
Residential - 4 March 2004
Rainfall (m m )
0
6
9
400
5.0
200
0.0
F low (l/s)
Concentration (m g/l)
P
Flow
0
0
0.5
1
400
1.5
Storm Duration (hr)
4 March 2004
0.5 Storm Duration
1 (hr)
1.5
Flow (l/s)
400
Concentration (mg/l)
Flow
Flow (l/s)
C oncentration (mg/l)
NO2-N
171
Residential - 6 March 2004
1
0.5
1
700
20
0
C oncentration (m g/L)
40
Flow
Flow (l/s)
120
20
0
0
0.0
1.0
Residential - 6 March 2004
0
20
Flow (l/s)
C oncentration (m g/L)
Flow
NH3-N
40
Flow
4
20
0
0
0.5
1
8
40
NO2-N
0.5
Concentration (mg/L)
1
3
0
R ainfall (m m )
0.5
0.0
1.0
Residential - 6 March 2004
0
1.5
0.5
Storm Duration (hr)
Rainfall (mm)
0.5
Storm Duration (hr)
40
Flow
0
0.0
BOD
240
0
0.0
1.0
Flow (l/s)
C oncentration (m g/L)
1400
COD
R ainfall (m m )
0.5
0
F low (l/s)
0
R ainfall (mm )
Residential - 6 March 2004
0.5
1.0
Storm Duration (hr)
Storm Duration (hr)
6 March 2004
5
Flow
175
400
0
0
0.0
0.5
Storm Duration (hr)
1.0
3
5
400
COD
F low (l/s )
C onc entration (m g/L)
800
0
SS
Flow
175
200
0
0
0.0
0.5
Storm Duration (hr)
1.0
F low (l/s )
3
C onc entration (m g/L)
0
R ainfall (m m )
Residential - 12 July 2004
R ainfall (m m )
Residential - 12 July 2004
172
Residential - 12 July 2004
5
BOD
175
150
0
F low (l/s )
Flow
16
0
0.5
Flow
0
1.0
0.5
1.0
Storm Duration (hr)
Residential - 12 July 2004
0
3
5
10
Flow
175
5
F low (l/s)
NH3-N
0
3
5
5
C oncentration (m g/L)
R ainfall (m m )
Residential - 12 July 2004
P
175
Flow
3
0
0
0
0.0
0.5
R ainfall
0
0.0
Storm Duration (hr)
C oncentration (m g/L)
175
8
R ainfall (m m )
0.0
NO3-N
Flow (l/s)
C onc entration (m g/L)
300
0
3
5
Flow (l/s)
0
3
C oncentration (m g/L)
R ainfall (m m )
Residential - 12 July 2004
0
0.0
1.0
0.5
1.0
Storm Duration (hr)
Storm Duration (hr)
12 July 2004
Residential - 8 September 2004
4
2
4
150
Flow
1.0
75
0.5
0.0
0
0.0
0.4
Storm Duration (hr)
0.8
C oncentration (m g/L)
NO3-N
Flow (l/s)
C oncentration (m g/L)
5.0
1.5
NH3-N
Flow
150
2.5
75
0.0
0
0.0
0.4
Storm Duration (hr)
0.8
Flow (l/s)
2
0
R ainfall (m m )
0
R ainfall (mm )
Residential - 8 September 2004
173
0
P
Flow
450
0.8
0
Flow
75
0.0
0
0
2
150
15.0
0
0.0
4
0.4
0.8
Storm Duration (hr)
Storm Duration (hr)
8 September 2004
Residential - 4 November 2004
150
0
0
2
0
2
4
Storm Duration (hr)
Storm Duration (hr)
Residential - 4 November 2004
Residential - 4 November 2004
450
0
6
9
Flow
5
0
C oncentration (mg/L)
N03-N
10
12
900
NH3-N
12
900
Flow
6
450
3
0
0
2
Storm Duration (hr)
4
Flow (l/s)
0
0
6
C oncentration (m g/L)
450
4
0
12
900
Flow
Rainfall (m m )
0
BOD
Rainfall (mm)
450
170
0
0
0
4 November 2004
2
Storm Duration (hr)
4
Flow (l/s)
300
Flow (l/s)
Flow
6
Concentration (m g/L)
SS
450
12
900
Flow (l/s)
C oncentration (mg/L)
6
Rainfall (mm)
0
0
Rainfall (m m )
Residential - 4 November 2004
Flow (l/s)
1.6
4
Concentration (mg/L)
12
900
P
2
Flow (l/s)
C oncentration (mg/L)
6
R ainfall (mm)
0
Rainfall (mm)
Residential - 8 September 2004
Residential - 4 November 2004
174
COD
Flow
75
450
0
F low (l/s)
1
Storm Duration (hr)
Flow
75
900
0
0
0
0
3
1.5
1.5
3
Flow
75
250
0
0
1
C oncentration (m g/L)
BOD
F low (l/s)
C oncentration (m g/L)
3
0
R ainfall (m m )
0
0
3
Residential - 27 Disember 2004
Residential - 27 Disember 2004
500
1
Storm Duration (hr)
30
NO3-N
Flow
75
15
0
0
0
3
R ainfall (m m )
0
SS
F low (l/s)
C oncentration (m g/L)
900
3
1800
F low (l/s)
3
1.5
C oncentration (m g/L)
1.5
0
R ainfall (m m )
0
R ainfall (m m )
Residential - 27 Disember 2004
Residential - 27 Disember 2004
1
3
Storm Duration (hr)
Storm Duration (hr)
0
1.5
3
P
Flow
75
3.5
0
F low (l/s)
C onc entration (m g/L)
7
R ainfall (m m )
Residential - 27 Disember 2004
0
0
1
3
Storm Duration (hr)
27 Disember 2004
Figure E-1: Pollutographs and hydrographs in the residential catchment
175
Commercial - 11 March 2004
3
800
3
6
80
80
550
0
0
0.0
1.2
Commercial - 11 March 2004
Commercial - 11 March 2004
6
BOD
Flow
80
0
F low (l/s)
C onc entration (m g/L)
6
3
C oncentration (m g/L)
3
0
R ainfall (m m )
0
150
0
0.0
0.5
NO3-N
Flow
2
80
0
1.0
0
0.0
Storm Duration (hr)
0
0
1.0
Flow (l/s)
Concentration (mg/L)
Flow
80
6
20
NO2-N
0.05
3
Concentration (mg/L)
6
0.1
0
Rainfall (mm)
3
Storm Duration (hr)
1.0
Commercial - 11 March 2004
0
0.5
0.5
Storm Duration (hr)
Commercial - 11 March 2004
0.0
0.5
1.0
Storm Duration (hr)
R ainfall (m m )
0.4 0.6 0.8 1.0
Storm Duration (hr)
Rainfall (mm)
0.2
0
NH3-N
Flow
80
10
0
0
0.0
0.5
Storm Duration (hr)
1.0
Flow (l/s)
0.0
Flow
F low (l/s)
0
SS
1100
F low (l/s)
Flow
C onc entration (m g/L)
COD
F low (l/s )
C onc entration (m g/L)
6
0
R ainfall (m m )
0
R ainfall (m m )
Commercial - 11 March 2004
176
0
3
6
P
Flow
5
80
0
0
0.0
0.5
Flow (l/s)
C oncentration (mg/L)
10
Rainfall (mm )
Commercial - 11 March 2004
1.0
Storm Duration (hr)
11 March 2004
Commercial - 16 March 2004
SS
Flow
200
100
0
6
COD
400
100
2.0
2.5
0
0.0
0.5
1.0
1.5
Storm Duration (hr)
2.0
2.5
Commercial - 16 March 2004
Commercial - 16 March 2004
200
200
BOD
Flow
100
100
0
0
0.0
0.5
1.0
1.5
Storm Duration (hr)
2.0
2.5
Flow (l/s)
C oncentration (mg/L)
12
6
12
3
NO3-N
C oncentration (m g/L)
6
Rainfall (m m )
0
0
R ainfall (mm )
1.0
1.5
Storm Duration (hr)
0
200
Flow
2
100
0
0
0.0
0.5
1.0
1.5
Storm Duration (hr)
2.0
2.5
Flow (l/s)
0.5
200
Flow
0
0.0
12
800
200
Flow (l/s)
Concentration (m g/L)
400
Rainfall (m m )
12
0
Flow (l/s)
6
C oncentration (m g/L)
0
R ainfall (m m )
Commercial - 16 March 2004
177
Commercial - 16 March 2004
Commercial - 16 March 2004
12
6
12
22
200
Flow
0.03
100
0.00
0
0.0
0.5
1.0
1.5
Storm Duration (hr)
2.0
Flow
200
11
100
0
2.5
Flow (l/s)
NO2-N
C oncentration (m g/L)
NH3-N
F low (l/s )
C oncentration (m g/L)
0.06
R ainfall (m m )
6
0
R ainfall (m m )
0
0
0.0
0.5
1.0
1.5
Storm Duration (hr)
2.0
2.5
0
6
12
P
200
Flow
4
100
0
F low (l/s)
C oncentration (m g/L)
8
R ainfall (m m )
Commercial - 16 March 2004
0
0.0
0.5
1.0
1.5
Storm Duration (hr)
2.0
2.5
16 March 2004
0
0.5
1
0.5
1
1500
SS
200
25
0
0
0.0
0.3
Storm Duration (hr)
0.6
Flow
1000
50
25
500
0
0
0.0
Storm Duration (hr)
0.5
F low (l/s)
50
C oncentration (m g/L)
COD
Flow
F low (l/s)
C oncentration (m g/L)
400
0
R ainfall (m m )
Commercial - 18 March 2004
R ainfall (m m )
Commercial - 18 March 2004
178
Commercial - 18 March 2004
Flow
50
250
25
0
F low (l/s )
3
25
0
0.0
0.5
Storm Duration (hr)
Commercial - 18 March 2004
0.5
0.5
1
0.2
R ainfall (m m )
Commercial - 18 March 2004
0
0
0.5
1
12
50
0.1
25
0.0
C oncentration (m g/L)
P
Flow
Flow (l/s)
C oncentration (m g/L)
NO2-N
Storm Duration (hr)
Flow
50
6
25
0
0
0.0
50
R ainfall (m m )
Storm Duration (hr)
Flow
0
0
0.0
NO3-N
5
0
0.0
0.5
Flow (l/s)
C oncentration (m g/L)
BOD
1
R ainfall (m m )
1
500
0.5
F low (l/s)
0.5
0
C oncentration (m g/L)
0
R ainfall (m m )
Commercial - 18 March 2004
Storm Duration (hr)
0.5
18 March 2004
Commercial - 19 March 2004
COD
Flow
35
500
0
0
0.0
0.3
Storm Duration (hr)
0.6
C oncentration (m g/L)
1000
4
1400
70
SS
F low (l/s)
C oncentration (m g/L)
4
2
R ainfall (m m )
2
0
70
Flow
700
35
0
0
0.0
0.3
Storm Duration (hr)
0.6
F low (l/s)
0
R ainfall (m m )
Commercial - 19 March 2004
179
Commercial - 19 March 2004
NH3-N
70
Flow
3
35
0
0
0.0
0.3
Storm Duration (hr)
4
70
Flow
F low (l/s)
NO3-N
F low (l/s )
C onc entration (m g/L)
4
2
C oncentration (m g/L)
2
6
0
R ainfall (m m )
0
R ainfall (m m )
Commercial - 19 March 2004
15
35
0
0
0.0
0.6
0.3
Storm Duration (hr)
0.6
19 March 2004
Commercial - 14 April 2004
Commercial - 14 April 2004
800
BOD
35
200
0
0.5
1.0
Storm Duration (hr)
30
35
0
C onc entration (m g/L)
50
0.5 Duration (hr) 1.0
Storm
4
70
F low (l/s )
C onc entration (m g/L)
Flow
2
NH3-N
R ainfall (m m )
4
0
R ainfall (m m )
2
NO3-N
0.5
1.0
Storm Duration (hr)
Commercial - 14 April 2004
0
0.0
0
0.0
Commercial - 14 April 2004
0
35
100
0
0.0
70
Flow
70
Flow
15
35
0
0
0.0
0.5
Storm Duration (hr)
1.0
F low (l/s )
0
C onc entration (m g/L)
400
4
70
Flow
F low (l/s )
C onc entration (m g/L)
COD
R ainfall (m m )
4
2
F low (l/s )
2
0
R ainfall (m m )
0
180
Commercial - 14 April 2004
R ainfall (m m )
0
2
4
4
70
Flow
2
F low (l/s )
C onc entration (m g/L)
P
35
0
0
0.0
0.5
Storm Duration (hr)
1.0
14 April 2004
0.7
1.4
R ainfall (m m )
0
0
0.7
1.4
20
0
0
20
60
1.0
0
0.0
0.7
1.4
1.2
0
R ainfall (m m )
0
1.4
0.6
20
0.0
0
1.0
C oncentration (m g/L)
40
Flow
F low (l/s )
C onc entration (m g/L)
0.7
0.4
NO3-N
0.5
Storm Duration (hr)
1.0
Commercial - 10 September 2004
Commercial - 10 September 2004
0.0
0.5
Storm Duration (hr)
NO2-N
40
Flow
0.2
20
0.0
0
0.0
0.5
Storm Duration (hr)
R ainfall (m m )
0.5
Storm Duration (hr)
110
1.0
F low (l/s )
0.0
40
Flow
F low (l/s )
1000
C onc entration (m g/L)
BOD
40
Flow
F low (l/s )
C onc entration (m g/L)
COD
R ainfall (m m )
Commercial - 10 September 2004
Commercial - 10 September 2004
181
1.4
0.7
1.4
40
Flow
0.3
20
0.0
0
0.0
0.5
Storm Duration (hr)
1.0
C onc entration (m g/L)
4.0
NH3-N
F low (l/s )
C onc entration (m g/L)
0.6
P
40
Flow
2.0
20
0.0
F low (l/s )
0.7
0
R ainfall (m m )
0
R ainfall (m m )
Commercial - 10 September 2004
Commercial - 10 September 2004
0
0.0
0.5
Storm Duration (hr)
1.0
10 September 2004
Figure E-2: Pollutographs and hydrographs in the commercial catchment
APPENDIX F
HYSTERESIS LOOPS
Residential - 8 November 2003
Concentration (mg/l)
Concentration (mg/l)
Residential - 8 November 2003
COD
300
0
SS
500
0
0
50
100
150
200
0
50
100
150
200
Discharge (l/s)
Discharge (l/s)
Residential - 8 November 2003
Residential - 8 November 2003
Concentration (mg/l)
Concentration (mg/l)
0.04
NO3-N
4
2
0
NO2-N
0.02
0.00
0
50
100
Discharge (l/s)
150
200
0
50
100
Discharge (l/s)
150
200
183
Residential - 8 November 2003
Residential - 8 November 2003
Concentration (mg/l)
Concentration (mg/l)
4
NH3-N
2
1
0
P
2
0
0
50
100
150
200
0
50
Discharge (l/s)
100
150
200
Discharge (l/s)
8 November 2003
Residential - 10 November 2003
Residential - 10 November 2003
100
COD
Concentration (mg/l)
Concentration (mg/l)
200
100
0
BOD
50
0
0
50
100
Discharge (l/s)
150
0
Residential - 10 November 2003
50
100
Discharge (l/s)
150
Residential - 10 November 2003
NO3-N
Concentration (mg/l)
Concentration (mg/l)
0.10
0.05
0.00
NO2-N
0.04
0.02
0.00
0
50
100
Discharge (l/s)
150
0
50
100
Discharge (l/s)
150
184
Residential - 10 November 2003
NH3-N
Concentration (mg/l)
Concentration (mg/l)
Residential - 10 November 2003
0.5
0.0
P
1
0
0
50
100
Discharge (l/s)
150
0
50
100
Discharge (l/s)
150
10 November 2003
Residential - 11 January 2004
Residential - 11 January 2004
2000
BOD
Concentration (mg/l)
Concentration (mg/l)
SS
1500
1000
500
0
100
0
0
5
10
Discharge (l/s)
15
0
Residential - 11 January 2004
5
10
Discharge (l/s)
15
Residential - 11 January 2004
0.4
NO2-N
Concentration (mg/l)
Concentration (mg/l)
NO3-N
1
0
0.2
0.0
0
5
10
Discharge (l/s)
15
0
5
10
Discharge (l/s)
15
185
Residential - 11 January 2004
Residential - 11 January 2004
40
P
Concentration (mg/l)
Concentration (mg/l)
NH3-N
20
0
10
0
0
5
10
Discharge (l/s)
15
0
5
10
Discharge (l/s)
15
11 January 2004
Residential - 2 March 2004
Residential - 2 March 2004
1600
BOD
Concentration (mg/l)
Concentration (mg/l)
COD
1200
800
400
0
200
0
0
5
10
15
Discharge (l/s)
20
25
0
Residential - 2 March 2004
5
10
15
Discharge (l/s)
20
Residential - 2 March 2004
NO3-N
NO2-N
10
Concentration (mg/l)
Concentration (mg/l)
25
5
0
0.1
0.0
0
5
10
15
Discharge (l/s)
20
25
0
5
10
15
Discharge (l/s)
20
25
186
Residential - 2 March 2004
Residential - 2 March 2004
P
Concentration (mg/l)
Concentration (mg/l)
NH3-N
10
0
10
0
0
5
10
15
Discharge (l/s)
20
25
0
5
10
15
Discharge (l/s)
20
25
2 March 2004
Residential - 4 March 2004
Residential - 4 March 2004
200
Concentration (mg/l)
Concentration (mg/l)
COD
1000
500
BOD
100
0
0
0
100
200
300
400
0
500
100
Discharge (l/s)
200
300
400
500
Discharge (l/s)
Residential - 4 March 2004
Residential - 4 March 2004
Concentration (mg/l)
Concentration (mg/l)
10
SS
1000
500
0
NO3-N
5
0
0
100
200
300
Discharge (l/s)
400
500
0
100
200
300
Discharge (l/s)
400
500
187
Residential - 4 March 2004
Residential - 4 March 2004
30
NO2-N
Concentration (mg/l)
Concentration (mg/l)
0.2
0.1
NH3-N
15
0
0.0
0
100
200
300
400
0
500
100
Discharge (l/s)
200
300
400
500
Discharge (l/s)
Residential - 4 March 2004
Concentration (mg/l)
10
P
5
0
0
100
200
300
400
500
Discharge (l/s)
4 March 2004
Residential - 6 March 2004
1200
Residential - 6 March 2004
COD
SS
800
600
600
400
200
0
0
0
10
20
30
Discharge (l/s)
40
50
0
10
20
30
Discharge (l/s)
40
50
188
Residential - 6 March 2004
Residential - 6 March 2004
BOD
NO3-N
3
200
0
0
0
10
20
30
40
50
0
10
Discharge (l/s)
20
30
40
50
Discharge (l/s)
Residential - 6 March 2004
Residential - 6 March 2004
10
NH3-N
Concentration (mg/l)
Concentration (mg/l)
NO2-N
2
0
5
0
0
10
20
30
40
50
Discharge (l/s)
P
5
0
10
20
30
10
20
30
Discharge (l/s)
Residential - 6 March 2004
0
0
40
50
Discharge (l/s)
6 March 2004
40
50
189
Residential - 12 March 2004
SS
COD
Concentration (mg/l)
Concentration (mg/l)
800
Residential - 12 March 2004
600
400
200
400
200
0
0
0
100
200
Discharge (l/s)
300
0
100
200
Discharge (l/s)
Residential - 12 March 2004
Residential - 12 March 2004
NH3-N
Concentration (mg/l)
Concentration (mg/l)
BOD
200
0
5
0
0
100
200
Discharge (l/s)
300
0
Residential - 12 March 2004
4
Concentration (mg/l)
P
2
0
0
300
100
200
Discharge (l/s)
300
12 March 2004
100
200
Discharge (l/s)
300
190
Residential - 8 September 2004
Residential - 8 September 2004
SS
COD
Concentration (mg/l)
Concentration (mg/l)
800
600
400
200
400
200
0
0
0
50
100
Discharge (l/s)
150
200
0
Residential - 8 September 2004
50
100
Discharge (l/s)
150
Residential - 8 September 2004
NO3-N
Concentration (mg/l)
Concentration (mg/l)
BOD
200
0
1
0
0
50
100
Discharge (l/s)
150
200
0
50
Residential - 8 September 2004
100
Discharge (l/s)
150
200
Residential - 8 September 2004
NH3-N
P
4
Concentration (mg/l)
Concentration (mg/l)
200
2
20
10
0
0
0
50
100
Discharge (l/s)
150
200
0
8 September 2004
50
100
Discharge (l/s)
150
200
191
Residential - 4 November 2004
Residential - 4 November 2004
200
BOD
SS
Concentration (mg/l)
Concentration (mg/l)
400
200
0
100
0
0
500
1000
0
500
Discharge (l/s)
1000
Discharge (l/s)
Residential - 4 November 2004
Residential - 4 November 2004
10
NO3-N
Concentration (mg/l)
Concentration (mg/l)
COD
400
200
5
0
0
0
500
1000
0
Discharge (l/s)
500
1000
Discharge (l/s)
Concentration (mg/l)
Residential - 4 November 2004
NH3-N
5
0
0
500
1000
Discharge (l/s)
4 November 2004
Figure F-1: Hysteresis loops of BOD, COD, SS, NO3-N, NO2-N, NH3-N and P in
the residential catchment
192
Commercial - 11 March 2004
Commercial - 11 March 2004
1500
Concentrations (mg/l)
Concentrations (mg/l)
1400
SS
700
COD
1000
500
0
0
0
50
100
150
0
50
Discharge (l/s)
Commercial - 11 March 2004
4
Concentrations (mg/l)
600
Concentrations (mg/l)
150
Discharge (l/s)
Commercial - 11 March 2004
BOD
500
400
300
200
100
0
NO3-N
2
0
0
50
100
150
0
Discharge (l/s)
50
100
150
Discharge (l/s)
Commercial - 11 March 2004
Commercial - 11 March 2004
20
Concentrations (mg/l)
0.10
Concentrations (mg/l)
100
NO2-N
0.05
0.00
NH3-N
15
10
5
0
0
50
100
150
Discharge (l/s)
0
50
100
Discharge (l/s)
11 March 2004
150
193
Commercial - 16 March 2004
Commercial - 16 March 2004
300
SS
Concentrations (mg/l)
Concentrations (mg/l)
600
400
200
BOD
200
100
0
0
0
50
100 150 200 250 300 350 400
0
50
Discharge (l/s)
100 150 200 250 300 350 400
Discharge (l/s)
Commercial - 16 March 2004
Commercial - 16 March 2004
COD
Concentrations (mg/l)
Concentrations (mg/l)
1000
500
0
2
0
0
50
0
100 150 200 250 300 350 400
100 150 200 250 300 350 400
Discharge (l/s)
Commercial - 16 March 2004
Commercial - 16 March 2004
NO2-N
0.05
0.00
NH3-N
20
10
0
0
50
100 150 200 250 300 350 400
Discharge (l/s)
Commercial - 16 March 2004
P
5
0
0
50
0
50
100 150 200 250 300 350 400
Discharge (l/s)
10
Concentrations (mg/l)
50
Discharge (l/s)
Concentrations (mg/l)
Concentrations (mg/l)
NO3-N
100 150 200 250 300 350 400
Discharge (l/s)
16 March 2004
194
Commercial - 18 March 2004
Commercial - 18 March 2004
600
SS
Concentrations (mg/l)
Concentrations (mg/l)
400
300
200
100
0
BOD
500
400
300
200
100
0
0
20
40
60
0
20
Discharge (l/s)
60
Discharge (l/s)
Commercial - 18 March 2004
Commercial - 18 March 2004
6
0.2
NO3-N
Concentrations (mg/l)
Concentrations (mg/l)
40
3
0
NO2-N
0.1
0.0
0
20
40
60
0
20
40
60
Discharge (l/s)
Discharge (l/s)
Commercial - 18 March 2004
Commercial - 18 March 2004
NH3
Concentrations (mg/l)
Concentrations (mg/l)
20
10
0
P
10
0
0
20
40
60
Discharge (l/s)
0
20
40
Discharge (l/s)
18 March 2004
60
195
Commercial - 19 March 2004
Commercial - 19 March 2004
1400
SS
Concentrations (mg/l)
Concentrations (mg/l)
1400
700
COD
700
0
0
0
20
40
60
0
20
Discharge (l/s)
60
Discharge (l/s)
Commercial - 19 March 2004
Commercial - 19 March 2004
9
BOD
300
150
Concentrations (mg/l)
450
Concentrations (mg/l)
40
0
NO3-N
6
3
0
0
20
40
60
0
20
Discharge (l/s)
40
60
Discharge (l/s)
Commercial - 19 March 2004
Commercial - 19 March 2004
0.12
30
Concentrations (mg/l)
Concentrations (mg/l)
NO2-N
0.08
0.04
0.00
20
15
10
5
0
0
20
40
60
0
Discharge (l/s)
Commercial - 19 March 2004
Concentrations (mg/l)
P
7
0
20
40
20
40
Discharge (l/s)
14
0
NH3-N
25
60
Discharge (l/s)
19 March 2004
60
196
Commercial - 14 April 2004
Commercial - 14 April 2004
800
Concentrations (mg/l)
Concentrations (mg/l)
800
SS
400
C OD
400
0
0
0
40
80
0
Discharge (l/s)
40
80
Discharge (l/s)
Commercial - 14 April 2004
Commercial - 14 April 2004
B OD
Concentrations (mg/l)
Concentrations (mg/l)
300
200
100
0
N H3 - N
20
0
0
40
80
Discharge (l/s)
Concentrations (mg/l)
P
3
0
40
40
Discharge (l/s)
Commercial - 14 April 2004
0
0
80
Discharge (l/s)
14 April 2004
80
197
Commercial - 20 August 2004
Commercial - 20 August 2004
2000
Concentrations (mg/l)
Concentrations (mg/l)
800
SS
400
C OD
1500
1000
500
0
0
0
40
80
0
Discharge (l/s)
40
80
Discharge (l/s)
Commercial - 20 August 2004
Commercial - 20 August 2004
Concentrations (mg/l)
Concentrations (mg/l)
150
B OD
0
N O3 - N
2
0
0
40
80
0
Discharge (l/s)
Discharge (l/s)
Commercial - 20 August 2004
Concentrations (mg/l)
1.0
N O2 - N
0.5
0.0
0
40
40
80
Discharge (l/s)
20 August 2004
80
198
Commercial - 10 September 2004
Commercial - 10 September 2004
3000
1200
Concentrations (m g/l)
Concentrations (m g/l)
SS
600
COD
1500
0
0
0
10
20
30
0
40
10
Discharge (l/s)
Commercial - 10 September 2004
40
Commercial - 10 September 2004
1.4
BOD
Concentrations (mg/l)
Concentrations (mg/l)
30
Discharge (l/s)
200
100
NO3-N
0.7
0.0
0
0
10
20
30
0
40
10
20
30
40
Discharge (l/s)
Discharge (l/s)
Commercial - 10 September 2004
Commercial - 10 September 2004
0.5
0.9
NO2-N
0.4
Concentrations (mg/l)
Concentrations (mg/l)
20
0.3
0.2
0.1
NH3-N
0.6
0.3
0.0
0.0
0
10
20
Discharge (l/s)
30
40
0
10
20
Discharge (l/s)
30
40
199
Commercial - 10 September 2004
Concentrations (m g/l)
4.0
P
3.0
2.0
1.0
0.0
0
10
20
30
40
Discharge (l/s)
10 September 2004
Figure F-2: Hysteresis loops of BOD, COD, SS, NO3-N, NO2-N, NH3-N and P in
the commercial catchment
APPENDIX G
SIEVE ANALYSIS
Figure G-1: Mechanical sieve shaker with timing device
Figure G-2: Sieve analysis test
APPENDIX H
ESTIMATION OF WASHOFF PARAMETERS
Equation:P0(t+∆t) = Po(t) exp{-KW∆t 0.5[R(t)n + R(t+∆t)n]}
(3.19)
Table G-1: Sample calculation with n = 1.0 (Storm on July 12, 2004)
R
Available SS
The Time step
KW
(gram)
Load (gram)
n=1.0
8.41
42602.15
752.81
1.048
84.05
41849.34
7011.42
0.152
117.67
34837.92
27190.85
0.006
159.7
7647.07
1433.12
0.326
176.51
6213.95
955.00
0.509
201.73
5258.96
916.77
0.509
210.13
4342.19
932.96
0.422
226.94
3409.23
700.68
0.476
252.16
2708.55
506.57
0.395
184.92
2201.98
1657.30
0.014
84.05
544.68
475.30
0.008
50.43
69.38
67.87
0.004
25.22
1.51
1.51
ΣP(t)
42602.15